Accelerated Development of Phototrophic Bioprocesses:

A Conceptual Framework

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen

University zur Erlangung des akademischen Grades eines Doktors der Ingenieurwissen-

schaften genehmigte Dissertation

vorgelegt von

Master of Science

Holger Mathias Morschett

aus

Püttlingen an der Saar

Berichter: Univ.-Prof. Dr. rer. nat. Marco Oldiges

Univ.-Prof. Dr.-Ing. Lars Blank

Univ.-Prof. Dr. rer. nat. Wolfgang Wiechert

Tag der mündlichen Prüfung: 22.03.2017

Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verfügbar.

„Erfahrung ist,

wenn man anstelle der alten Fehler neue Fehler macht.“

Willy Meurer

Declaration on oath

I hereby affirm that I independently authored this thesis and did not use any source or appli-

ance other than those cited in this study. This thesis has not been submitted in either whole

or part for a degree at this university or any other institution.

Aachen, March 2017

Parts of this thesis have been published in peer-reviewed journals, at conferences and in

supervised students’ theses. Figures adapted from existing own publications are cited within

the text. Any thought, methodology, result and conclusion that is part of an existing own pub-

lication is not explicitly referred to, but considered as properly cited by the following listing:

peer-reviewed journals

Morschett H., Reich S., Wiechert W. & Oldiges M. (2015): Simplified cryopreservation of the

microalga Chlorella vulgaris integrating a novel concept for cell viability estimation. Eng Life

Sci, doi: 10.1002/elsc.201500056.

Morschett H., Wiechert W. & Oldiges M. (2016): Automation of a Nile red staining assay en-

ables high throughput quantification of microalgal lipid production. Microb Cell Fact, doi:

10.1186/s12934-016-0433-7.

Morschett H., Schiprowski D., Müller C, Mertens K., Felden P., Meyer J., Wiechert W. &

Oldiges M. (2016): Design and characterization of a parallelized micro-photobioreactor ena-

bling phototrophic bioprocess development at elevated throughput. Biotechnol Bioeng, doi:

10.1002/bit.26051.

Morschett H., Schiprowski D., Rohde J., Wiechert W. & Oldiges M. (2016): Comparative

evaluation of the scalability of phototrophic microtiter plate cultivation against laboratory-

scale photobioreactors. Bioproc Biosys Eng, doi: 10.1007/s00449-016-1731-5.

Morschett H., Freier L., Rohde J., von Lieres E., Wiechert W. & Oldiges M. (2017): A frame-

work for accelerated phototrophic bioprocess development: integration of parallelized micro-

scale cultivation, laboratory automation and Kriging-assisted experimental design. Biotechnol

Biofuels, doi: 10.1186/s13068-017-0711-6.

conference posters

Morschett H., Müller C., Wiechert W. & Oldiges M. (03.-04.06.2014): Accelerated process

characterization of phototrophic systems using a parallelized micro-scale photo-bioreactor.

7. Bundesalgenstammtisch, Köthen/Anhalt, Germany.

Morschett H., Schiprowski D., Müller C., Wiechert W. & Oldiges M. (07.-08.09.2015): Accele-

rated Phototrophic Process Development: Tools & Devices. 8. Bundesalgenstammtisch,

Garching, Germany.

Morschett H., Wiechert W. & Oldiges M. (20.-22.02.2017): Integrated Framework for the Ac-

celerated Development of Phototrophic Bioprocesses. Bioprocessing Days 2017, Reckling-

hausen, Germany.

conference talks

Morschett H., Ritter D., Müller C., Welters T., Wiechert W. & Oldiges M. (15.-19.06.2015):

Accelerated phototrophic process development: light meets microtiter plates. ACHEMA Con-

gress, Frankfurt am Main, Germany.

Morschett H., Wiechert W. & Oldiges M. (02.-04.05.2016): Parallelized milliliter scale cultiva-

tion accelerates phototrophic bioprocess development. DECHEMA Himmelfahrtstagung,

Koblenz, Germany.

Morschett H., Müller C., Wiechert W. & Oldiges M. (27.-27.09.2016): Beschleunigte Biopro-

zessentwicklung für phototrophe Systeme. 9. Bundesalgenstammtisch, Jülich, Germany.

students‘ theses

Schiprowski D. (2015): Charakterisierung und Validierung eines Mikro-Bioreaktorsystems zur

Kultivierung phototropher Mikroorganismen. RWTH Aachen University. Master thesis.

Rohde J. (2016): Anwendung Kriging-basierter Versuchsplanung für die beschleunigte Bi-

oprozessentwicklung mittels automatisierter Mikrobioreaktortechnologie. RWTH Aachen Uni-

versity. Master thesis.

Acknowledgements

Diese Arbeit entstand während meiner Tätigkeit als wissenschaftlicher Mitarbeiter der AG

„Bioprozesse und Bioanalytik“ im Institut für Bio- und Geowissenschaften am Forschungs-

zentrum Jülich und war dort in das BMWi-geförderte Kooperationsprojekt „Mikro-Photo-

Bioreaktor“ eingebettet. Eine Vielzahl von Personen hat zu ihrem Gelingen beigetragen.

Zunächst gilt mein Dank Prof. Marco Oldiges für die kompetente Betreuung, ein stets offenes

Ohr für im Projektverlauf aufgetretene Herausforderungen, die Bereitstellung hervorragender

Arbeitsbedingungen, sowie die Begutachtung der Arbeit. Ich danke Prof. Wolfgang Wiechert

für die Begutachtung der Arbeit, die wertvollen Anregungen bei der Publikation von Ergeb-

nissen, sowie die Chance dieses Projekt an seinem Institut bearbeiten zu dürfen. Außerdem

danke ich Prof. Lars Blank für die Übernahme einer weiteren Gutachterrolle.

Ich danke meinen Projektpartnern m2p-labs und Premosys für die konstruktive Zusammen-

arbeit, insbesondere genannt seien an dieser Stelle Carsten Müller, Pamela Felden, Kolja

Mertens, Markus Jorissen, Matthias Kuhl, Ünal Dogan und Jörg Meyer.

Sebastian Reich, Danny Schiprowski und Jannis Rohde haben im Rahmen studentischer

Arbeiten mit ihrem Einsatz im Labor einen erheblichen Anteil zu dieser Arbeit beigetragen.

Dafür möchte ich Euch besonders danken - ohne diese Hilfe wäre eine Vielzahl an Versu-

chen nicht durchführbar gewesen. Lars Freier gilt mein Dank für die Unterstützung bei der

Planung, Durchführung und Auswertung der DoE-Studie. Jochem Gätgens danke ich für die

Durchführung und Auswertung von GC-ToF-MS Messungen zur Fettsäuren-Fingerprint-

Analyse. Den Angestellten der Infrastruktur danke ich für die Unterstützung bei sämtlichen

administrativen Angelegenheiten, weiterhin Hubert Ruhrig, Daniel Klein und Jürgen Paulzen

für die Unterstützung bei konstruktiven Aufgaben im Bereich Mechanik und Elektronik.

In den vergangenen drei Jahren habe ich viel Zeit im Institut verbracht. Insbesondere des-

halb möchte ich mich bei allen Arbeitskollegen und Freunden innerhalb der AG „Bioprozesse

und Bioanalytik“ und auch darüber hinaus für die tolle, entspannte und zugleich motivierende

Atmosphäre bedanken. Das gilt besonders für alle Mitstreiter aus dem Brauteam „JuBräu“

und unserem daraus gewachsenen Projekt „RheinCraft“. Ihr seid echt ‘ne klasse Truppe!

Zu guter Letzt danke ich ganz besonders herzlich all jenen, die ich zum Kreis von Familie

und Freunden zählen darf. Ohne Eure stetige Motivation und den bedingungslosen Rückhalt

in schwierigen Phasen hätte ich es nicht geschafft, danke!

Abstract

Phototrophic systems regained interest as feedstocks for bio-economy but their industrial

exploitation mostly fails for economic reasons until today. Due to lacking high throughput

photobioreactors and supporting methodologies, especially early stage screening suffers

from low efficiency. In this context, a framework for accelerated phototrophic process devel-

opment was designed. Targeting lipid production by Chlorella vulgaris as a microalgal model

process, the full spectrum from strain maintenance, via cultivation and analytics to experi-

mental design was addressed, while parallelized cultivation was focused as key technology.

Contrary to well-established serial subculturing, strain maintenance was done by cryopreser-

vation of glucose-adapted cells. Thus, an easy-to-use protocol was set up and optimized.

According to specially developed growth pattern analysis, post-thawing viabilities of 63 ± 2 %

were achieved and phototrophic pre-cultivation lead to highly reproducible adaptation to light.

Enabling elevated throughput cultivation, a 48-well microtiter plate-based micro-photobio-

reactor was designed and growth was precisely and non-invasively monitored via scattered

light. Strictly controlled conditions permitted a high comparability along the wells of a plate

(± 5 %) while small scale and fast mixing ensured excellent light supply of the cultures. The

system was shown to offer good scalability to established laboratory-scale photobioreactors.

To handle samples from microscale cultivation, a dye-based assay was set up and assay

conditions were optimized towards efficient and precise quantification of C. vulgaris’ intra-

cellular lipid accumulation. Assay automation resulted in minimal hands-on-times while errors

due to fluctuating performance of manual operators could be excluded.

These technologies were merged into a framework for the accelerated development of pho-

totrophic processes while Kriging-assisted experimental design was used to take full ad-

vantage of the improved experimental capacity. Within as little as four experimental rounds,

the volumetric productivity of the model lipid production process was approx. tripled solely via

medium optimization and synergistic multi-parameter interaction could be revealed.

Though this framework was suitable to accelerate phototrophic process development, it may

only be seen as an initial blueprint. Future improvement of the micro-photobioreactor and

intensified robotic integration will enable more complex processes and thus extend the range

of application from screening tasks to the acquisition of kinetic data concerning metabolism

of phototrophic microorganisms or even simulation of complex environmental conditions.

Abbreviations and formula symbols

abbreviation denotation

ACP acyl-carrier-protein

ADP adenosine diphosphate

ATP adenosine triphosphate

CA cellulose acetate

CCC central composite face centred design

CCF composite circumscribed design

C. vulgaris Chlorella vulgaris

CoA coenzyme A

CZ Czech Republic

DMSO dimethyl sulfoxide

DoE Design of Experiments

EDTA ethylenediaminetetraacetic acid

EI Expected Improvement

enBBM enriched Bold’s Basal Medium

enBBMopt optimized enriched Bold’s Basal Medium

enBBMopt,min optimized and minimized enriched Bold’s Basal Medium

enBBMref enriched Bold’s Basal Medium with reference composition

FA fatty acid

G Germany

GC-ToF-MS gas chromatography time-of-flight mass spectrometry

G3P 3-phosphoglycerate

KriKit Kriging ToolKit

LED light emitting diode

MCMC Markov Chain Monte Carlo

MES 2-(N-morpholino)ethanesulfonic acid

continued on next page

continuation from previous page

MBR microbioreactor

MTP microtiter plate

NADP nicotinamide adenine dinucleotide phosphate

NADPH nicotinamide adenine dinucleotide phosphate, reduced state

PES polyether sulfone

PS polystyrene

PTFE polytetrafluorethylene

PYR pyruvate

STR stirred tank reactor

SW Sweden

TAGs triacylglycerides

US United States

symbol denotation dimension

acc accuracy %

a.u. arbitrary unit -

CDW cell dry weight g L-1

DOT dissolved oxygen tension %

f dilution factor -

h Planck’s constant 6.626 · 10-34 W s-2

Iav average light intensity µmol m-2 s-1

I0 incident light intensity µmol m-2 s-1

k number of levels per input variable -

KI light excess inhibition constant µmol m-2 s-1

KM light affinity constant µmol m-2 s-1

KS substrate affinity constant g L-1

lagP lag phase coefficient for product formation -

continued on next page

continuation from previous page

lagµ lag phase coefficient -

m average amount of water transferred mg

mb brutto tube weight mg

mB mass of extracted biomass mg

mi amount of transferred water mg

mI maintenance coefficient -

mn netto tube weight mg

mt target amount of water to be transferred mg

n number of replicates -

NA Avogadro’s number 6.022 · 1023 mol-1

nexp number of required experiments -

ngen apparent generation number -

ni number of input variables -

(n)LC (neutral) lipid content % (w w-1)

OD(λ) optical density (at wavelength λ) -

p probability value -

PAR(λ) photosynthetically active radiation (at wavelength λ) µmol m-2 s-1

pH lat. potentia hydrogenii -

Pλ illumination intensity (at wavelength λ) W m-2

Pmax maximal photosynthesis rate -

prec precision %

Prel relative photosynthesis rate -

Pvol volumetric productivity mg L-1 d-1

qP product formation rate % (w w-1) h-1

R2 correlation coefficient -

Ref reference value -

continued on next page

continuation from previous page

S substrate concentration g L-1

t process time h

tacc,P duration of acceleration phase for product formation h

tacc,µ duration of acceleration phase h

tD doubling time h

tlag duration of lag phase h

tlag,P duration of lag phase for product formation h

tlag,µ duration of lag phase h

tprod time from start of production phase h

tstat time to stationary phase h

v viability %

var variance %

VL liquid volume mL

v v-1 volume per volume -

w v-1 weight per volume -

w w-1 weight per weight -

X biomass concentration g L-1

YX/S biomass yield g g-1

Δngen differential apparent generation number -

ε light absorption coefficient g L-1

λnm wavelength nm

µ(max) (maximal) exponential growth rate h-1 or d-1

List of tables and figures

Table 1.1: Cultivation conditions influence C. vulgaris’ biochemical composition ....................3

Table 1.2: Selected physico-chemical properties of microalgal and mineral diesel .................5

Table 1.3: Lipid content and productivity of selected microalgae ............................................8

Table 1.4: Comparison of selected commercial MBR systems ............................................. 18

Table 2.1: Overview of chemicals ......................................................................................... 27

Table 2.2: Overview of devices ............................................................................................. 28

Table 2.3: Preparation of enBBMref from stock solutions ....................................................... 30

Table 3.1: Key design aspects of the photobioreactors used ................................................ 67

Table 3.2: Comparison of cell dry weight, neutral lipid content and volumetric productivity ... 73

Table 3.3: Target molecule addressing by excitation and emission wavelength variation ..... 77

Table 3.4: Initial evaluation of the medium components ....................................................... 89

Table 3.5: Impact of medium optimization by key performance indicating parameters .......... 96

Table 6.1: Experimental design and measured data used in section 3.5.2.1 ....................... 137

Table 6.2: Comparison of the initial composition of enBBMref, enBBMopt and enBBMopt,min .. 137

Table 6.3: Experimental design and measured data used in section 3.5.2.2 ....................... 138

Table 6.4: Experimental design and measured data used in section 3.5.3 ......................... 138

Table 6.5: Experimental design and measured data used in section 3.5.2.4 ....................... 139

Figure 1.1: Microscopic view of C. vulgaris 211-11b fixed to agar slides ................................2

Figure 1.2: Daughter cell formation during vegetative autosporulation of C. vulgaris ..............2

Figure 1.3: Lipid biosynthesis as integral component of microalgal carbon capture ................6

Figure 1.4: Exemplary photobioreactors for large-scale phototrophic cultivation .....................8

Figure 1.5: Kinetic relation between light intensity and growth rate ....................................... 12

Figure 1.6: Course of phototrophic batch cultivation with its characteristic phases ............... 13

Figure 1.7: Comparison of conventional and accelerated bioprocess development .............. 15

Figure 1.8: Automated microscale bioprocess platform ........................................................ 20

Figure 1.9: Schematic process model ................................................................................... 21

Figure 1.10: Schematic representation of central composite designs for nonlinear models .. 23

Figure 1.11: Framework for accelerated phototrophic bioprocess development ................... 26

Figure 2.1: Workflow for the automated quantification of intracellular neutral lipids .............. 37

Figure 3.1: Heterotrophic growth of C. vulgaris after freezing preservation ........................... 45

Figure 3.2: Comparison of cultivation times necessary to reach stationary phase ................ 46

Figure 3.3: Cell count and biovolume before and after cryopreservation procedures ............ 47

Figure 3.4: Growth pattern-deviated assessment of cell viability ........................................... 49

Figure 3.5: Post-thawing viability after cryopreservation of stationary phase cells ................ 52

Figure 3.6: Phototrophic pre-cultivation of C. vulgaris in shake flasks ................................... 53

Figure 3.7: Design of the micro-photobioreactor prototype ................................................... 57

Figure 3.8: Tailoring of the photo module spectrum .............................................................. 59

Figure 3.9: The photo module ensures a homogeneous photon flux density distribution ...... 60

Figure 3.10: Biomass quantification by scattered light .......................................................... 61

Figure 3.11: Phototrophic cultivation at MTP-scale ............................................................... 63

Figure 3.12: Comparison of the reference process in different photobioreactor systems ...... 68

Figure 3.13: Exponential growth during the reference processes ......................................... 69

Figure 3.14: Light-limited growth during the reference processes ......................................... 71

Figure 3.15: Fatty acid fingerprint from neutral lipids from the reference processes ............. 74

Figure 3.16: Chemical structure and selected properties of the lipid probe Nile red .............. 77

Figure 3.17: Optical properties and average cell size at different physiological states .......... 79

Figure 3.18: Typical fluorescence spectra of the lipid assay ................................................. 81

Figure 3.19: Characterization of the Nile red staining ........................................................... 82

Figure 3.20 Gravimetric calibration of the lipid assay ............................................................ 83

Figure 3.21: Assay automation significantly enhances analytical throughput ........................ 84

Figure 3.22: Automated liquid handling increases accuracy and precision ........................... 85

Figure 3.23: Initial fractional factorial screening analysis ...................................................... 91

Figure 3.24: Refined full factorial screening .......................................................................... 92

Figure 3.25: Representation of parameter space .................................................................. 94

Figure 3.26: Contour plots of the final Kriging model prediction ............................................ 95

Figure 3.27: Comparison of the process changes induced by medium optimization ............. 95

Figure 3.28: Relative composition of the fatty acids from the neutral lipid fraction ................ 97

Figure 6.1: Custom-made shaker for phototrophic cultivation ............................................. 132

Figure 6.2: Emission spectra of the individual LEDs used to set up the photo module ........ 132

Figure 6.3: Individual biomass curves of the proof-of-principle cultivation ........................... 133

Figure 6.4: Spectra of the illumination units installed in the used photobioreactors ............. 133

Figure 6.5: Influence of the shaking frequency on liquid distribution in FlowerPlates® ........ 134

Figure 6.6: Estimated effect of two factor interaction with MgSO4 ....................................... 134

Figure 6.7: Screening plot around reference point .............................................................. 135

Figure 6.8: Monitoring of calcium precipitation in enBBM by means of optical density ........ 136

Figure 6.9: Relative composition of the fatty acids from the neutral lipid fraction ................ 136

Contents

1. Background .............................................................................................. 1

1.1 Introduction .............................................................................................1

1.2 Chlorella vulgaris ....................................................................................2

1.3 Microalgae as feedstock for biodiesel production....................................4

1.3.1 Mechanism of lipid biosynthesis .............................................................5

1.3.2 Technical realisation ...............................................................................7

1.3.3 Economic evaluation ............................................................................ 10

1.4 Kinetics of phototrophic batch cultivation .............................................. 11

1.5 Accelerated bioprocess development ................................................... 14

1.5.1 Microbioreactors ................................................................................... 16

1.5.2 Automated microbioreactor platforms ................................................... 19

1.5.3 Experimental design strategies for bioprocess optimization .................. 21

1.6 Aims of this study ................................................................................. 24

2. Material and methods ............................................................................ 27

2.1 Chemicals............................................................................................. 27

2.2 Devices ................................................................................................ 28

2.3 Microorganism ...................................................................................... 30

2.4 Medium ................................................................................................ 30

2.5 Sterilisation ........................................................................................... 31

2.6 Heterotrophic cultivation ....................................................................... 31

2.6.1 Shake flask cultivation .......................................................................... 31

2.6.2 Stirred tank reactor cultivation .............................................................. 31

2.6.3 MTP cultivation ..................................................................................... 32

2.7 Phototrophic cultivation......................................................................... 32

2.7.1 Shake flask cultivation .......................................................................... 32

Pre-cultivation ....................................................................................... 32 2.7.1.1

Reference cultivation ............................................................................ 33 2.7.1.2

2.7.2 Test tube cultivation .............................................................................. 33

2.7.3 Flat panel reactor cultivation ................................................................. 33

2.7.4 Phototrophic MTP cultivation ................................................................ 33

2.8 Cryopreservation .................................................................................. 34

2.9 Biomass detection ................................................................................ 34

2.10 Nitrate quantification ............................................................................. 35

2.11 Gravimetric lipid quantification .............................................................. 35

2.12 Fluorometric lipid quantification ............................................................ 36

2.13 Fatty acid fingerprinting ........................................................................ 37

2.14 Microscopy ........................................................................................... 38

2.15 Offline pH ............................................................................................. 39

2.16 Temperature at single well resolution ................................................... 39

2.17 Illumination intensity ............................................................................. 39

2.18 Evaporation .......................................................................................... 39

2.19 Acquisition of absorption and emission spectra .................................... 39

2.20 Pipetting accuracy and precision .......................................................... 40

2.21 Experimental design ............................................................................. 40

2.22 Statistical analysis ................................................................................ 42

3. Results and discussion ......................................................................... 43

3.1 Strain maintenance ............................................................................... 43

3.1.1 Evaluation of different cryopreservation strategies ................................ 44

3.1.2 Viability calculation ............................................................................... 49

3.1.3 Optimized protocol and viability estimation ........................................... 51

3.1.4 Re-adaptation to phototrophic conditions and pre-cultivation ................ 53

3.1.5 Conclusion............................................................................................ 54

3.2 Parallelized phototrophic microscale cultivation .................................... 55

3.2.1 Non-biological validation ....................................................................... 59

Illumination ........................................................................................... 59 3.2.1.1

Temperature control ............................................................................. 60 3.2.1.2

Evaporation .......................................................................................... 60 3.2.1.3

3.2.2 Online measurement of biomass .......................................................... 61

3.2.3 Biological validation .............................................................................. 62

3.2.4 Conclusion............................................................................................ 64

3.3 Scalability of phototrophic microscale cultivation .................................. 65

3.3.1 Lag phase............................................................................................. 68

3.3.2 Exponential growth phase .................................................................... 69

3.3.3 Light-limited growth phase .................................................................... 71

3.3.4 Production phase.................................................................................. 72

3.3.5 Conclusion............................................................................................ 75

3.4 High throughput lipid quantification ....................................................... 76

3.4.1 Standardization of biomass concentration ............................................ 78

3.4.2 Assay development and validation ....................................................... 80

3.4.3 Automation ........................................................................................... 84

3.4.4 Conclusion............................................................................................ 86

3.5 Integration of experimental design ........................................................ 87

3.5.1 Choice of relevant media components .................................................. 88

3.5.2 Kriging-assisted optimization ................................................................ 90

Fractional factorial ................................................................................ 90 3.5.2.1

Full factorial .......................................................................................... 92 3.5.2.2

Locating optimal medium composition .................................................. 93 3.5.2.3

Refinement of the optimum ................................................................... 94 3.5.2.4

3.5.3 Validation of optimal medium composition ............................................ 95

3.5.4 Final medium simplification ................................................................... 98

3.5.5 Assessment of achieved volumetric productivity ................................... 98

3.5.6 Conclusions .......................................................................................... 99

4. Conclusion and outlook ...................................................................... 100

5. Literature .............................................................................................. 105

6. Appendix .............................................................................................. 129

6.1 Simplified kinetic model for phototrophic batch cultivation .................. 129

6.2 Additional figures ................................................................................ 132

6.3 Additional tables ................................................................................. 137

Background 1

1. Background

1.1 Introduction

Driven by population growth and proceeding industrialization, the shortage of fossil resources

and the contribution to global warming by their combustion are two of the key challenges of

the 21st century [1-5]. Consequently, the transformation of the currently oil-based economy

towards a knowledge-based bio-economy is focused globally [6]. Bio-economy defines a

multidisciplinary concept aiming at a sustainable, highly efficient economy and agriculture

utilizing renewable biological resources to provide products and services. The underlying,

ideally circular, value chains are based on innovative biological and technological processes

providing an environmentally neutral economy [7-9].

Within this context, green biomass, meaning in particular plant and plant waste material, is

rated as the major source of organic carbon for synthesis and fuel applications [10, 11]. In

contrast to the already established use of terrestrial plants, photosynthetic microorganisms,

namely microalgae, offer numerous advantages for bio-economy applications. Most im-

portant are their superior light conversion efficiency and biomass productivity [12, 13], the

possibility to cultivate algae in open or closed photobioreactor systems [14] that can be situ-

ated on non-arable land thereby circumvent concurrence to the food supply chain [15].

Moreover, retention and recovery of water and mineral nutrients, especially phosphate, is

rather simple [16] and herbicides and insecticides as commonly used for crop production are

not required at all [17]. Algae provide access to a broad spectrum of goods ranging from bulk

products like biofuels and synthetic compounds [14, 18, 19] via food and feed additives or

functional food [20] to pharmaceuticals [21, 22]. However, despite some industrial processes

realized, particularly microalgal bulk production remains uneconomic until now [23].

Besides reactor and value chain engineering, especially advances concerning strain con-

struction, screening and process parameter optimization are crucial [24, 25]. These aspects

are typically studied during numerous laboratory-scale experiments as many parameters

(strain, medium composition, process parameters, light characteristics, etc.), dynamic

changes (e.g. day/night cycles) and their potential interactions need to be considered.

However, current generation laboratory-scale photobioreactors are not designed for strongly

parallelized experimentation as being restricted to single vessel systems, shake flasks and

test tubes [26-29]. Additionally, peripheral methods like strain maintenance, analytics and

experimental design remain highly time-inefficient as they mostly rely on the manual opera-

tion of procedures that have not been optimized towards time and material efficiency, yet.

Background 2

1.2 Chlorella vulgaris

The green alga Chlorella vulgaris (Figure 1.1) belonging to the class of Trebouxiophyceae

was first isolated by Beijerinck from an eutrophic pond near Delft (Netherlands) in 1890 [30].

Since then, it became one of the best studied algae and is nowadays ranked as a model or-

ganism in photobiotechnology due to its easy cultivability and relatively fast growth. This ro-

bustness against a wide range of environmental conditions goes along with its ubiquitous

occurrence in fresh as well as in brackish water or on solid surfaces all over the globe [31,

32].

Figure 1.1: Microscopic view of C. vulgaris 211-11b fixed to agar slides.

C. vulgaris is a spherical, unicellular microalga typically sizing from 2 – 10 µm in diameter

[33-35]. The non-motile cells replicate asexually via vegetative autosporulation. At a distinct

physiological trigger, first the chloroplast and the nucleus of the mother cell divide. Subse-

quent protoplast division and maturation lead to transformation of the mother cell into mostly

2 or 4, but sometimes up to 32 daughter cells being released upon burst of the mother cell’s

cell wall [33, 34, 36]. Thereby, the debris of the mechanically robust glucosamine-based cell

wall [37] are left behind (Figure 1.2).

Figure 1.2: Daughter cell formation during vegetative autosporulation of C. vulgaris: (i) early growth phase, (ii) late growth phase, (iii) chloroplast dividing phase, (iv) early protoplast di-viding phase, (v) late protoplast dividing phase, (vi) daughter cells maturation phase, (vii) hatching phase. Modified from [33].

Background 3

In analogy to many other microalgae, C. vulgaris’ cellular composition is largely dependent of

the cultivation conditions [38]. The most significant alteration is induced by growth repression

inducible by nitrogen [39-41], phosphorus [17] or sulphur [42] starvation or high pH induced

cell cycle inhibition [43] for example (Table 1.1).

Table 1.1: Cultivation conditions influence C. vulgaris’ biochemical composition. Modified from [44].

fraction relative fraction [% (w w-1)]

growing cells nitrogen-starved cells

protein 29 ± 2.5 7 ± 1.6

carbohydrate 51 ± 2.1 55 ± 3.2

lipids 18 ± 0 7 40 ± 2.1

Some strains of C. vulgaris are known to be capable of starch accumulation within their chlo-

roplast, turning them into a potential feedstock for the fermentative production of chemical

building blocks or fuels like bioethanol [45, 46]. However, reports differ about the preferred

mode of cultivation for efficient carbohydrate production [44, 47-51]. Currently, it is discussed

that the carbohydrate fraction of microalgal biomass could be used as an additional by-

product within an integrated biorefinery concept rather than being the main focus of an indus-

trial production process [19, 52].

Contrary to plant biomass, growing C. vulgaris cultures feature an extraordinary high content

of protein. Compared to a reference recommended by the World Health Organization, Chlo-

rella protein provides an excellent nutritional quality by means of amino acid composition and

protein efficiency ratio [53, 54]. Though, algal biomass is used as an additive in animal rather

than human nutrition so far, as its suitability to substitute conventional protein sources like

soybean or fish meal has been demonstrated [38, 55]. The supplementation of non-ruminant

feed requires a pretreatment in order to break up the non-digestible cell wall and thus make

the intracellular protein accessible for breakdown [53]. Considering the high cost for pre-

treatment and biomass production, it is easily comprehensible that algal food and feed addi-

tives are still in their economic infancy. Nevertheless, the nutritional exploitation of Chlorella

represents the most profitable way of its industrial use at present [20].

During growth repression (for example due to nitrogen starvation) photosynthesis remains

active and the metabolic flux of fixed carbon is shifted towards the synthesis of storage sub-

stances [41]. Equivalent to many other microalgae, C. vulgaris synthesizes substantial quan-

tities of cytoplasmic lipid droplets under such conditions. These mainly consist of triacyl-

Background 4

glycerides derived from palmitic (16:0), oleic (18:1 Δ9) and α-linoleic acid (18:3 Δ9,12,15) but

additionally contain traces or minor amounts of myristic (14:0), palmitoleic (16:1 Δ7), hexa-

decadienoic (16:2 Δ7,10), hexadecatrienoic (16:3 Δ7,10,13), stearic (18:0) and linoleic

(18:2 Δ9,12) acid glycerol esters [56]. This fatty acid profile turns C. vulgaris into a valuable

source of polyunsaturated fatty acids for nutritional applications [57] as well as into a poten-

tial feedstock for sustainable biodiesel production [58]. A detailed overview about microalgal

biodiesel production is given in section 1.3.

Besides these three main fractions of Chlorella biomass, further components or properties of

the cells can be exploited. Despite being a nutritional supplement, C. vulgaris can be used as

a source for food colouring agents [55], for example. Moreover, the cells can be used as

sources for vitamins and minerals [59], as soil fertilizer [60] or for the synthesis of sustainably

produced isotope-labelled biochemicals [61]. Even medical application seems worth consid-

ering, as several pharmaceutically active ingredients have been identified to be produced by

Chlorella, so far [21, 62, 63].

Alternatively to phototrophic cultivation, C. vulgaris can be cultivated in heterotrophic (dark-

ness, availability of organic carbon source) or mixotrophic (light and availability of organic

carbon source) mode while glucose, glycerol and acetate are the substrates commonly ap-

plied [39, 48, 49, 64, 65]. However, with respect to industrial production, these strategies

may be restricted to high value production as they significantly increase costs compared to

phototrophic processes.

1.3 Microalgae as feedstock for biodiesel production

With the considerable advances in fuel cell and battery technology during the last decade,

electromobility is expected to be on the verge of its global breakthrough [66, 67]. For heavy

duty vehicles combustion engines will - at least at the mid-term - remain indispensable be-

cause of the superior energy density of their fuels. Thus, carbon dioxide neutral alternatives

to fossil diesel fuel are needed to reduce the greenhouse gas emissions of this sector. These

can be obtained, for instance, from algal triacylglycerides (TAGs) and come up with physico-

chemical properties, making them compatible to established combustion engines and fuel

infrastructure (Table 1.2).

Background 5

Table 1.2: Selected physico-chemical properties of microalgal and mineral diesel. Modified from [45, 68].

property algal

biodiesel

mineral

diesel

EN14214

standard

higher heating value [MJ kg-1] 41 45.9 -

kinematic viscosity at 40 °C [mm2 s-1] 5.2 1.2 – 3.5 3.5 – 5.2

density [kg L-1] 0.864 0.83 – 0.84 0.86 – 0.9

cetane number [-] - 51 > 51

flash point [°C] 115 60 – 80 > 101

pour point [°C] -12 -35 – -15 -

carbon [% (w w-1)] 76 87 -

hydrogen [% (w w-1)] ≤ 12.7 13 -

oxygen [% (w w-1)] ≥ 11.3 0 -

sulphur [% (w w-1)] 0 ≤ 0.05 ≤ 0.02

acid value [mgKOH g-1] 0.374 ≤ 0.5 ≤ 0.5

Algal biomass can be obtained from photobioreactors situated on non-arable land. Thereby,

production does not concur with agricultural land used for food and feed production, as is the

case for the current generation of biodiesel deviated from oleaginous crops such as rape-

seed, sunflower, palm and soybeans [15, 69]. Moreover, microalgal oil production is regard-

ed to be more efficient than the use of oilseed crops as a higher oil content as well as higher

productivities in terms of biomass formation and photosynthetic conversion efficiency can be

achieved [13, 70, 71].

1.3.1 Mechanism of lipid biosynthesis

Within their natural habitat, microalgae are mostly exposed and thus evolutionarily adapted

to nutrient deficiency. Thus, the capability of efficiently synthesizing and metabolizing long-

term storage compounds is of exceptional importance [72]. In this context, TAGs are the

most widely spread storage compounds for carbon and chemical energy. Providing a caloric

value in the range of 37 kJ g-1 they are a more efficient storage compound than, for example,

carbohydrates (17 kJ g-1) [73].

In general, the synthesis of TAGs can be induced if carbon (typically CO2) and energy me-

tabolites (ATP and NADPH via the light reaction of photosynthesis) are available in excess,

but the actual metabolic flux of fixed carbon cannot be completely directed towards biomass

formation. Being essential for biomass formation, the most important triggers of TAG synthe-

Background 6

sis are the depletion of nitrogen [39-41], phosphorus [17] or sulphur [42] from both, the culti-

vation medium and endogenous pools. Moreover, it is reported that despite sufficient nutrient

supply, high pH values can induce TAG formation. Strong alkaline conditions can act as a

cell cycle regulator preventing cell division and thus inducing lipid production [43, 74]. How-

ever, lipid synthesis can even be triggered at appropriate pH and sufficient nutrient supply. At

extensive light intensity, the growth capacity may be exhausted, whereas the light reaction

delivers ATP and NADPH in excess. Under these unbalanced conditions, where the pool of

NADP available for electron uptake from the light reaction may be depleted, the formation of

reactive oxygen species sets in. In this case, triggering the synthesis of TAGs provides an

additional sink for the accumulated NADPH and can thus rebalance light reaction and the

central metabolism via a partial decoupling of photosynthesis and biomass formation to pre-

vent photo-oxidative damage [73, 75-77].

Microalgal biosynthesis of fatty acids is located within the chloroplast being closely linked to

photosynthetic carbon capture. Afterwards, the fatty acids are processed inside the endo-

plasmic reticulum to produce membrane phospholipids as well as TAGs (Figure 1.3) [78].

Figure 1.3: Lipid biosynthesis as integral component of microalgal carbon capture. ACP: acyl-carrier-protein, ADP: adenosine diphosphate, ATP: adenosine triphosphate, CoA: coen-zyme A, G3P: 3-phosphoglycerate, FA: fatty acid, PYR: pyruvate, TAG: triacylglycerides. Modified from [79-81].

Background 7

During the light reaction of photosynthesis, pigment-assisted absorption of photons drives

the cleavage of water into molecular oxygen and protons while the corresponding electrons

are transferred to form NADPH from NADP+. ATP is generated simultaneously. These, in

turn, power the so-called Calvin-Benson cycle (dark reaction) fixing inorganic carbon dioxide

via the enzyme ribulose-1,5-bisphosphate carboxylase oxygenase [82]. The intermediate

3-phosphoglycerate (G3P) is deviated from the cycle and processed to acetyl-CoA via py-

ruvate (PYR). In the next step, carbon dioxide, being present as HCO3- within the chloroplast,

is enzymatically fixed to acetyl-CoA and thus malonyl-CoA is formed [83]. The malonyl resi-

due is then transferred to the acyl-carrier-protein (ACP) leading to the release of CoA. Driven

by carbon dioxide release, malonyl-ACP is coupled to acetyl-CoA and a C4-backbone is syn-

thesized. The resulting precursor then acts as acceptor for the iterative coupling of further

acetyl-CoA to form acyl-ACP molecules of differing chain length [84]. These can be further

processed by the introduction of double bonds via specific desaturases [85]. The fatty acid

synthesis is generally interrupted by the cleavage of ACP, resulting in molecules of different

chain lengths with different degrees of saturation. The fatty acids are subsequently trans-

ferred to the endoplasmic reticulum where they are multiply linked to G3P and finally make

up triacylglycerides (TAGs). These are secreted from the endoplasmic reticulum and form

distinct cytoplasmic droplets, the so-called lipid bodies making up the long-term energy and

carbon storage of algal cells. Alternatively to TAG synthesis, the intermediates phosphatidic

acid and 1,2-diacyl glycerol can be branched off into the synthesis of membrane phospholip-

ids as well [84, 86, 87].

1.3.2 Technical realisation

Although the mechanisms of TAG biosynthesis are rather conserved, different microalgal

species largely differ in their capability of storage lipid accumulation. Besides the fatty acid

spectrum [88], especially the lipid content and thus the productivities achievable vary funda-

mentally (Table 1.3) while strain specific process parameter optimization has a major impact

on overall efficiency. Consequently, optimization with regard to reactor design and location,

process control and medium composition for instance needs to be conducted in accordance

with the algal species chosen to obtain satisfactory results [58, 71]. Additionally, genetic en-

gineering of algae to optimize lipid production is recently being taken into account and re-

garded as a substantial tool towards high productivity. However, this approach is still in its

infancy, as only few species are accessible by molecular genetic methods up to now [79, 83,

86, 89].

Background 8

Table 1.3: Lipid content and productivity of selected microalgae. Modified from [13].

species lipid content

[% (w w-1)]

lipid productivity

[mg L-1 d-1]

Botryococcus braunii 24 – 31 –

Chlorella emersonii 25 – 63 10.3 – 50

Chlorella sorokiniana 19 – 22 44.7

Chlorella vulgaris 5 – 58 11.2 – 40

Dunaliella salina 6 – 25 116

Euglena gracilis 14 – 20 –

Nannochloropsis oculata 22.7 – 29.7 84 – 142

Scenedesmus obliquus 11 – 55 –

Spirulina platensis 4 – 16.6 –

A hypothetical process for the production of biodiesel from microalgae typically consists of a

characteristic sequence of individual unit operations: In the initial step, lipid rich algae bio-

mass is produced in either open ponds or closed photobioreactors while the former are sim-

pler in design and thus offer significantly lower investment and operating cost (Figure 1.4).

However, open systems suffer from comparably high evaporation, the lack of sterile condi-

tions and worse light supply by design [90-92]. Moreover, process control is limited in com-

parison to closed systems. The attainable biomass concentrations are typically

0.02 - 0.06 % (w w-1) [93] and thus around one order of magnitude lower than those achieved

using closed photobioreactors.

Figure 1.4: Exemplary photobioreactors for large-scale phototrophic cultivation. A: raceway pond, B: horizontal tubes, C: vertical bags, D: vertical tubes. Modified from [94].

Background 9

During the second unit operation, algal biomass is separated from the liquid phase while the

preferably applied principle is determined by the biomass loading of the broth. For low con-

centrations, as typically obtained from open pond cultivation, either flotation or flocculation is

induced by adding of chemical agents. The cells can then be harvested from the surface of

the reactor and by sedimentation, respectively [95, 96]. At higher biomass loading as gener-

ated in closed photobioreactors, the cells can be separated either by centrifugation or by

crossflow filtration. However, with regard to process economy it has to be taken into account

that centrifugation, while being the more robust method, requires a significantly higher ener-

gy input than filtration [88, 97]. Biomass separation typically causes 20 - 30 % of the total

cost during biodiesel production [98] and thus needs to be addressed during process optimi-

zation. This aspect has been intensively explored during the last years, as reviewed else-

where [99, 100]. So far, there is no universal strategy available [96].

After biomass separation, the extraction of intracellular TAGs is carried out. Depending on

the method used, an additional drying step may be incorporated while natural sunlight (low

energy demand, but high space requirement) or thermal procedures (high energy demand)

can be used [101]. Lipid extraction is mostly carried out using hexane for instance [71, 79,

101, 102]. However, the utilization of organic solvents introduces complications due to their

toxicity so that more sustainable alternatives like the extraction with supercritical carbon di-

oxide are being explored at the moment [103, 104]. Additionally, diverse strategies for cell

disruption may help to increase extraction efficiency [58, 105, 106].

The extracted TAGs are subsequently processed by acidic or alkaline transesterification

while the latter is faster at mild conditions. On the other hand, the alkaline transesterification

is more sensitive against residual water leading to catalyst depletion and the formation of by-

products [107, 108]. Using methanol, preferably generated from renewable resources, fatty

acid methyl esters (namely biodiesel) are generated [14] while the by-product glycerol can be

used as a feedstock for the fermentative production of bioethanol, for example [89].

Alternatively, the integration of extraction and transesterification into one single unit opera-

tion, called in-situ transesterification, shows a high potential to reduce the demand of toxic

agents and energy. Besides, additional potential lies within the processes insensitivity

against residual water, rendering upstream drying superfluous [109-112].

Depending on the fatty acid chain length and degree of saturation distribution of the produc-

tion host and further variables, the quality of the resulting fatty acid methyl ester mix may not

be sufficient according to the applicable provisions (e.g. EN14214, Table 1.2) [113]. In the

Background 10

case of an impermissible high share of polyunsaturated fatty acids, for example, catalytic

hydrogenation can provide relief. Moreover, blending with diesel from alternatives sources

can be used to match the required specifications [114-117].

1.3.3 Economic evaluation

So far, no industrial process for the production of biodiesel from microalgal biomass has

been realized. Consequently, the available economic analyses are based on the extrapola-

tion of datasets generated at laboratory and pilot-scale [80]. However, these differ substan-

tially and are thereby only of limited informative value. Most striking in this context is the fact

that volumetric productivities determined at small scales usually cannot be preserved during

scale-up.

However, the basic demand for successful realization on industrial scale is a competitive

price compared to that of petroleum-based diesel fuel. Reported prices for the production of

algal biomass vary drastically. Many studies claim costs in the range of 4 – 6 € kg-1 [18-20,

23, 45, 46, 108], rendering algal biodiesel uneconomic so far. Moreover, at the current state

of the art, biodiesel cannot be declared sustainable from an ecologic point of view. From bi-

omass production via transportation to its use in combustion engines, life cycle analysis re-

vealed a negative carbon dioxide balance while especially the high demand for auxiliary en-

ergy (e.g. for biomass separation, section 1.3.2) was identified as a major drawback [46].

Production costs need to be reduced to the level of petroleum-based fuels, while simultane-

ously the overall energy balance has to be lifted to positive values.

Depending on location, target product and reactor system, the critical cost drivers may signif-

icantly vary. In order to close the remarkable price gap between petroleum and algae-based

diesel, the available optimization potential at all the different steps of the production chain

need to be addressed. Some options, so far identified and currently addressed are the reduc-

tion of energy need for mixing [91], gassing [118] and temperature control [119] during culti-

vation, improving light supply of the culture [120] by reactor design at the same time. Instead

of atmospheric aeration or the usage of pure carbon dioxide, flue gas from industrial plants

can be used [121, 122]. Thereby, the ecologic footprint can be improved in addition to cost

reduction. Whenever feasible due to the location and algal strain, cultivation could be con-

ducted using waste, brackish or sea water to reduce the amount of fresh water needed [16,

123]. Moreover, the energy need during harvest, making up a significant fraction of the over-

all cost, needs to be addressed by developing more efficient alternatives [45]. Substitute ex-

traction procedures [103, 104] or the coupling of extraction and transesterification via in-situ

transesterification [109-112] as highlighted in section 1.3.2 offer additional potential for cost

Background 11

reduction and especially for improvement of the economic footprint. Nevertheless, there is

strong evidence that processes focusing on biodiesel as the sole product will not become

economically feasible at all. Instead, microalgal processes need to be set up in the context of

integrated biorefinery concepts [19, 52]. Here, possible routes for additional generation of

value are the use of the cellular protein fraction for feed applications [53] and the fermenta-

tive exploitation of residual biomass fractions in biogas plants [124]. Still, most promising

seems the coproduction of high value substances like pigments or antioxidants and feed-

stocks for sustainable chemical synthesis [19, 125, 126].

Besides process and value chain engineering, extraordinary optimization potential lies within

the choice and optimization of the biological system itself. In addition to productivity, candi-

date strain evaluation is to be carried out with respect to the local environmental conditions

like temperature and light intensity as well as their dynamics [127]. Production strains need

to be tailored taking advantage of molecular tools. Thereby, they can be adapted to the pre-

dominant conditions. Furthermore, systems biology approaches should be taken into ac-

count. These offer opportunities for rational metabolic engineering targeting an optimization

of metabolic fluxes towards higher productivity or the introduction of new pathways for the

coproduction of valuable products [128-134]. All these methodologies offer a high potential

for drastic improvements at the very early steps of process design. Nevertheless, they are

currently still impeded by the lack of appropriate high throughput phototrophic cultivation and

analytic capabilities.

1.4 Kinetics of phototrophic batch cultivation

Contrary to heterotrophic nutrition, the metabolism of phototrophic microorganisms is driven

by energy metabolites withdrawn from the light reaction of photosynthesis. The underlying

mechanisms have been studied in great detail and have extensively been portrayed in litera-

ture [135-137]. Moreover, numerous empirical and mechanistic models have been developed

to quantify these relationships, as comprehensively reviewed by Béchet et al. [138]. For bio-

technological application, the dependency of growth from light intensity as shown in Fig-

ure 1.5 is of extraordinary importance during process and reactor engineering.

Background 12

Figure 1.5: Kinetic relation between light intensity and growth rate: (i) light compensation point, (ii) light limitation, (iii) light saturation, (iv) light excess inhibition. Simulated data, the underlying kinetic model is provided in section 6.1 (appendix).

Microalgal growth needs a certain light intensity threshold (i) below which the complete pho-

tosynthetic turnover is required for respiratory cell maintenance. Intensities above this light

compensation point (Figure 1.5 (i)) then induce growth. Nevertheless, the availability of

NADPH and ATP deviated from the light reaction remains rate-limiting and thus, the growth

rate is directly coupled to light intensity (light limitation, Figure 1.5 (ii)). At sufficiently high

intensities, the dark reaction runs at its maximum turnover so that additional energy metabo-

lites may not be converted. Consequently, the growth capacity is exhausted and high illumi-

nation intensities do not promote further metabolic turnover and growth (light saturation,

Figure 1.5 (iii)). Further increasing light intensities lead to the depletion of the NADP pool as

the turnover of the light reaction continues increasing. Despite this misbalance may be han-

dled by the so-called mechanism of cyclic electron transfer to a certain extent, the absorbed

electromagnetic energy may not be passed on via energy metabolites completely. Instead, a

reduction of dissolved molecular oxygen to reactive oxygen species takes place. These

cause oxidative damage to the photosynthesis apparatus, reducing the photosynthetic ca-

pacity and leading to cell death at worst (light excess inhibition, Figure 1.5 (iv)).

However, the correlation between light intensity and cellular growth is known to be highly

dynamic. Depending on environmental conditions, photosynthetically active cells exhibit sig-

nificant restructuration of the photosynthetic apparatus. Particularly, the adjustment of chlo-

0 150 300 450 600

0.00

0.02

0.04

0.06

0.08

[

h-1]

PAR [mol m-2 s

-1]

(i)

(ii)

(iii)

(iv)

Background 13

rophyll and auxiliary pigment content is used by the cell to adapt photosynthesis towards the

prevailing illumination conditions [139-141]. Consequently, the influence of light on metabolic

performance makes up a highly dynamic connection being influenced by both, light supply

and the interconnected biological response.

With respect to reactor engineering, the optimization of light supply by design is a crucial

issue [142, 143]. The influx of natural or artificial light takes place via the necessarily trans-

parent reactor surface or via internal light sources [144]. The photons provided are then ei-

ther absorbed or scattered by the suspended cells. Inevitably, a biomass and optical path

length dependent light intensity gradient forms throughout the reactor volume [145]. Thus,

light is a radiation field superimposed to the multiphasic fermentation broth (medium, cells,

gas bubbles) [91]. Even though photobioreactors are tailored towards optimal light supply,

above certain biomass concentrations dark zones in the culture liquid emerge. These grow

with further increase of the biomass concentration and induce a light limitation of the cells

[143, 146]. This effect can be reduced by intensive mixing to a certain extent [120], but typi-

cally results in the characteristic kinetics of phototrophic batch processes (Figure 1.6).

Figure 1.6: Course of phototrophic batch cultivation with its characteristic phases: (i) lag phase, (ii) exponential growth phase, (iii) light-limited growth phase, (iv) production phase. Simulated data, the underlying kinetic model is provided in section 6.1 (appendix).

0 48 96 144 192 240

0

4

8

12

16

(ii)

(i)

(iv)

bio

ma

ss

[

g L

-1],

su

bs

tra

te [

g L

-1],

ne

utr

al lip

id c

on

ten

t

[% (

w w

-1)]

time [h]

(iii)

0.00

0.02

0.04

0.06

0.08

[h

-1]

Background 14

Analogous to microbial batch fermentation, a typical phototrophic process is characterized by

an initial lag phase during which the adaptation to the prevailing conditions occurs (Fig-

ure 1.6 (i)). In particular, this may incorporate photoacclimation by restructuration of the pho-

tosynthetic apparatus. Afterwards, exponential growth takes place, as long as all necessary

conditions are situated within the optimal windows of operation (Figure 1.6 (ii)). With increas-

ing biomass concentration, the overall light availability becomes rate-limiting, thus reducing

growth (Figure 1.6 (iii)). This phase is characterized by a steady decrease of the growth rate

due to further biomass formation and a respective shift of the substrate consumption from

exponential to approximately linear kinetics. With the depletion of an essential substrate, bi-

omass formation comes to a standstill. If carbon (e.g. CO2) and energy (light) are sufficiently

supplied, yet, these are typically targeted to the formation of intracellular storage compounds

like TAGs (Figure 1.6 (iv), section 1.3.1).

Besides incident light intensity and the specific requirements of the cultivated strain, the reac-

tor design strongly affects the biomass concentration above which light limitation occurs. In

this context, surface reflection and absorption [147], optical path length and surface area

[144, 148], the fraction of non-illuminated reactor compartments [149] and the mixing regime

[150] were identified to have a major influence. The importance of these factors typically

scales with the reactor size, but may already distort cross-system comparisons at laboratory-

scale.

1.5 Accelerated bioprocess development

Due to the extensive parameter space, additionally incorporating dynamic interactions, bio-

processes are characterized by high complexity [151]. Despite a broad knowledge of cellular

biology and process engineering available, predictive process development nowadays re-

mains impossible in most cases. As a result, a need for conducting numerous experiments

for strain and parameter screening arises. This drawback is typically handled by a series of

development stages being characterized by an incremental increase of scale and information

content. In parallel, experimental capacities shrink due to the growing technical expenditure

at larger scale (Figure 1.7).

Background 15

Figure 1.7: Comparison of conventional (grey arrows) and accelerated (blue arrow) biopro-cess development. Modified from [152].

After identification of a target product, the initial stages of bioprocess development are the

so-called primary and secondary screening [153, 154]. While the former typically addresses

productivity obtained from massively parallelized microtiter plate (MTP) cultivation, the latter

usually applies shake flask cultivation to investigate the influence of process conditions (pH,

temperature, medium composition, induction profiling, etc.) on the productivity of the most

promising strains. However, at the cost of throughput these experiments are mostly per-

formed in a black box manner while very little is known about the exact process conditions

and kinetics [155, 156]. During the next stages, process development and optimization in

laboratory- and pilot-scale fermenters offering highly controllable conditions and wide-ranging

possibilities for on- and offline analysis take place [157]. Thereby, high information content as

well as resilience are achieved. However, compared to the previous stages, high invest and

technical expenditure further limit the experimental capacities.

Classical black box screening systems provide only limited process information [158]. Alt-

hough productivities can be obtained for specific examples, the underlying kinetics remain

unclear. Nevertheless, these are crucial for a better understanding of biological processes

from the engineering as well as from the biological point of view. Introducing methodologies

to obtain these data at the early stages of bioprocess development thus contributes to accel-

erating bioprocess development and investigating basic research questions.

To overcome this bottleneck, the concept of “accelerated bioprocess development” is propa-

gated. It aims at elevating experimental throughput compared to conventional screening and

simultaneously providing in-depth process knowledge as typically generated by laboratory-

Background 16

scale fermentation [154, 157, 159-162]. Experimental capacities in the range of a black box

MTP screening cannot be achieved by design, but this drawback is compensated by the

more detailed process characterization reducing the amount of experiments necessary.

This approach is mainly driven by three key technologies, namely microbioreactors (sec-

tion 1.5.1), liquid handling robotics (section 1.5.2), and experimental design (section 1.5.3).

The underlying technologies have mostly been developed within the last 20 years and are

being refined continuously. Meanwhile, the concept has successfully proven its potential in

time-efficient bioprocess development [163-175].

1.5.1 Microbioreactors

The concept of microbioreactors (MBRs) has been established during the last two decades.

However, there is no universal definition of such systems [176], as different authors give in-

terpretations varying from miniaturized stirred tank reactors via shaken bubble columns to

MTP-based systems [154, 156, 177-180].

From the engineering point of view, the scale-down of laboratory stirred tank reactors (STRs)

to parallelized millilitre scale is regarded to be the most straightforward way of designing an

MBR system. Geometric similarity is maintained by design and the relevant capabilities for

monitoring and reactor control (e.g. temperature, pH, dissolved oxygen tension (DOT), feed-

ing) can be maintained to a certain extent [154, 157, 181]. Thereby, scalable flow regimes as

well as shear conditions and mass transfer coefficients can be obtained while either gas-

inducing impellers or magnetic inductive drives with tailored stirrer designs are applied [181-

185]. Typical working volumes range from the sub-millilitre scale up to approx. 100 mL, while

volumes in the range of 10 mL have been proven a good compromise between paralleliza-

tion by miniaturization and technical feasibility [154, 177, 186-190].

Shaken microtiter plates being operated under well-known and defined conditions, thereby

ensuring comparability to shake flask or stirred tank reactor processes, are the most fre-

quently employed MBR concept at present. As is the case for stirred MBRs, mixing and flow

patterns, power input and energy dissipation as well as mass transfer were identified as key

parameters for transferability to the laboratory-scale as extensively reviewed by Marques et

al. [191]. In the context of such systems, the sterile closure of the MTP proved to be a critical

issue as homogeneous gas exchange is needed while evaporation from the individual wells

of a plate is to be minimized [178, 192]. However, this problem has been solved by the de-

velopment of appropriate commercially available sealing foils. Further reduction of liquid loss

can be realized by incubation in a controlled high humidity atmosphere [193]. Sieben et al.

Background 17

[194] provide a comprehensive comparison of currently available MTP sealing foils. If these

and other prerequisites like temperature homogeneity are fulfilled, each well of an MTP can

be regarded as a distinct MBR. In addition to already established slow release systems for

controlled substrate supply [195-197], integrated and fully disposable microfluidics for feed

and pH control currently emerge as a highly flexible state-of-the art alternative [198, 199].

These microfluidic MTPs enable screening under typical production conditions (pH-controlled

fed-batch) rather than in batch mode. As MTPs used in MBR systems are usually pre-

sterilized and designed for single use, especially setup effort and cleaning times are mini-

mized. Most prominent working volumes are in the range of 200 – 2000 µL per well while the

corresponding MTPs range from 4 – 96 wells. By now, a wide variety of different plate de-

signs for specific applications became available, while a major focus has been put on achiev-

ing mass transfer comparable to laboratory-scale stirred tank reactors [200, 201]. However,

further increase of cultivation throughput by volume reduction seems to be not feasible. For

well diameters smaller than 3.5 mm, it was reported that surface tension of the culture liquid

cannot be overcome by orbital shaking, thus resulting in poor mixing and mass transfer [202].

In recent years, further miniaturization led to the development of a completely new microbio-

reactor, the so-called microfluidic reactors. Such systems allow for the cultivation down to

femtolitre scale to investigate growth and production on the single cell level. A comprehen-

sive review of these promising tools is given by Grünberger et al. [203]. However, such sys-

tems remain limited to basic research and strain development so far, as they do not reflect

the conditions found in laboratory reactors [161].

Besides increasing throughput by miniaturization and parallelization, the acquisition of rele-

vant process data is crucial to overcome the currently prominent black box approach during

primary and secondary screening (section 1.5). It needs to be considered that sampling-

based analytics may be limited due to the small culture volume handled in MBRs [158]. Thus,

adequate miniaturized sensors have been established. Among others, DOT and carbon diox-

ide as well as pH became assessable by sensitive sensor spots that can be read out non-

invasively and ideally online by optical methods. These are typically based on immobilized

fluorophores while fluorescence lifetime or a phase shift is correlated against the respective

measurand [204-210]. State-of-the-art in biomass monitoring for MBRs is optical technology

as well. Despite new impedance spectroscopy methods becoming available recently [211],

biomass is typically quantified either by optical density, or by scattered light while the latter

proved more suitable due to its larger kinetic range [212-215]. However, relying on optical

biomass quantification, possible interferences by changing morphology, pigmentation or

light-absorbing metabolites and products has to be considered critically [216, 217].

Background 18

Moreover, the monitoring of fluorescent reporter proteins (e.g. green fluorescent protein) and

intracellular metabolites like NADP(H) [215] has successfully been established by taking ad-

vantage of specific excitation and emission read-out. Thereby, valuable information about the

physiological state and product formation by means of model proteins became assessable at

the microscale. Subsequent to an early prototype system [218], single-well resolved acquisi-

tion of oxygen transfer rates in MTPs has recently been realized [219]. Thereby, the potential

of off-gas analysis to monitor growth and metabolic state was successfully transferred to

MTP-based MBR systems.

The scalability of several MBR systems has been shown throughout various application stud-

ies [156, 169, 182, 187, 190, 220-227] and some have developed from academic prototypes

to commercial products. Table 1.4 gives an overview of selected commercialized MBRs.

Table 1.4: Comparison of selected commercial MBR systems. Modified from [157] and ma-nufacturers’ specifications.

property ambr BioLector bio-

REACTOR Micro-24

micro-

Matrix

distributor Sartorius m2p-labs 2mag Pall Applikon

design STR MTP STR bubble col-umn

MTP

disposable yes yes vessel only yes yes

parallelization 12-48 48 8/48 24 24

volume [mL] 10 – 250 0.5 – 1.5 8 - 15 3 - 7 1 - 5

organisms procaryotes, eucaryotes, cell culture

procaryotes, eucaryotes, cell culture

procaryotes, eucaryotes

procaryotes, eucaryotes, cell culture

procaryotes, eucaryotes

feeding by liquid handler

by liquid handler

by liquid handler

integrated integrated

process data pH, DOT pH, DOT, biomass, fluorescence

pH, DOT pH, DOT pH, DOT

control temperature, atmosphere, pH, DOT, stirring

temperature, atmosphere, shaking

temperature, atmosphere, stirring

temperature, pH, DOT,

temperature, atmosphere, pH, DOT

automation proprietary robotic plat-form

yes yes yes no

reference [228], [223] [169] [156] [229] [230]

Background 19

1.5.2 Automated microbioreactor platforms

Microbioreactors offer elevated experimental throughput while simultaneously providing a

certain degree of process insight via integrated analytics (section 1.5.1). The variety of sig-

nals available is typically limited to optical biomass measure, pH, DOT, and fluorescence.

Further measures like extracellular substrate and product concentrations still need to be ac-

cessed via sampling. Being restricted to reasonable labour, this impedes with taking full ad-

vantage of these highly parallelized systems.

In this context, the integration of MBRs into automated liquid handling platforms offers valua-

ble benefit. In industry, laboratory automation is already routinely used during drug screening

and for molecular biology applications [231-233], but is just recently being established for

automated bioprocess development as visible by the relatively low, but increasing number of

such systems in literature (selected examples are provided in [163, 164, 167, 168, 170]).

Liquid handling platforms generally consist of a deck accessible by robotic manipulators for

the displacement of labware (mostly MTPs in different layouts) as well as pipetting arms for

liquid manipulation. These are either driven by air displacement (e.g. Microlab STAR, Hamil-

ton Robotics) or by a system liquid (e.g. Freedom EVO, Tecan and JANUS, PerkinElmer),

while the operator can typically choose between disposable pipetting tips and washable steel

needles for persistent application.

These platforms enable the automated and standardized processing of typical routine liquid

manipulation operations. Besides elevated experimental throughput (working speed, reduced

hands-on-time and independency from day/night cycles), data reliability may be increased as

automated processing is less prone to stochastic errors than manual operation [234]. How-

ever, pipetting operations like aspiration and dispense speed need to be defined and opti-

mized to the target volume and liquid properties to achieve precise liquid manipulation [235]

whereas an experienced manual operator may intuitively tune his pipetting behaviour to the

target operation.

By integrating a MBR into such a robotic platform, it can be used for the automated execution

of bioprocesses. In particular, this includes the independent dosing of substances into the

discrete reactors of the MBR system (e.g. for induction [165], pulsed feeding [163] or pH cor-

rection [236]) as well as sampling from the reactors. Establishing a continuous communica-

tion between MBR and liquid handler, any pipetting event may be triggered by signals like

biomass concentration, DOT or pH from the MBR system and thus dynamic liquid dosing as

well as sampling is facilitated [164]. In addition, housing of the whole platform with a laminar

Background 20

flow hood was shown to enable sterile operation not only to minimize the risk of contamina-

tion while dosing into or sampling from the MBR, but to conduct on-site medium preparation

and inoculation as well [169, 175].

Besides MBRs, various further instruments have been integrated into liquid handling plat-

forms, thereby generating fully automated platforms for bioprocess development with flexible

multi-modular functionality. This incorporates centrifuges [168] or vacuum stations for bio-

mass separation, chilling stations for reliable sample deposition [169], shakers with heating

capability for sample processing and a wired spectrum of analytic devices like MTP readers

for (kinetic) assays [163, 168], balances for biomass acquisition, light microscopes for mor-

phology analysis or devices like cytometers or mass spectrometers for high quality in-depth

analysis [167, 172, 237].

Samples generated from automated microscale bioprocessing may either be collected and

analyzed offline after cultivation or even in an at-line manner immediately after sampling

while the latter may require dynamic scheduling solutions as presented by Moore et al. [238].

Thereby, in-depth data can be generated from multiple parallelized microscale processes

investing only minimal hands-on-time.

Figure 1.8: Automated microscale bioprocess platform. The system is based on a JANUS® Integrator liquid handling robot with integrated BioLector® microbioreactor system and further devices for off- and at-line sample processing and analysis and has repeatedly proven its applicability for accelerated bioprocess development as well as its scalability to established laboratory-scale bioreactors [168, 169, 175]. Picture with permission from Andreas Radek, Forschungszentrum Jülich GmbH.

Background 21

1.5.3 Experimental design strategies for bioprocess optimization

As highlighted within section 1.5, the intrinsic complexity of bioprocesses renders predictive

development impossible in most cases until today. Thus, a high number of experiments is

needed for screening and parameter optimization [239]. However, the high number of input

variables (temperature, pH, medium composition, induction time and strength, feed profiles,

etc.) that need to be considered leads to a so-called combinatorial explosion as, for example,

the investigation of 10 input variables on three levels requires 310 ≈ 60,000 distinct experi-

ments. Despite significant technological proceedings and a resulting drastic increase of ex-

perimental throughput (e.g. automated MBR platforms for accelerated process development

and strain phenotyping, sections 1.5.1 and 1.5.2), experimental capacities cannot keep up

with the needed throughput in most cases.

Consequently, methodologies that maximize the knowledge gain per experiment and thereby

minimize the number of experiments needed are required. One approach is the so-called

“Design of Experiments” (DoE) having been introduced in the early 20th century by Fisher

[240]. DoE is a systematic approach where functional relations are approximated using em-

pirical modelling strategies, e.g. polynomial fitting [241]. In this context, signal noise (meas-

urement inaccuracy, stochastic fluctuation, etc.) additionally needs to be taken into account

[242]. The investigated process is usually regarded in a black box manner, which means that

the mathematical description is based on empirical relations. Mechanistic knowledge may be

incorporated for model development. In the context of bioprocess engineering, this approach

mostly remains exceptional due to the intrinsic complexity of cellular metabolism [239]. How-

ever, this kind of simplification is at the prize of a reduced knowledge generation about the

investigated system. Despite the influence and interactions of input variables becomes ac-

cessible, the underlying mechanistic relations remain unclear.

Figure 1.9: Schematic process model. A set of input variables is transformed to correlated output variables via a process handled in a black box manner while different sources of sig-nal noise need to be considered.

During an initial screening, the impact of potential input variables on the optimization target

(output variable) is investigated. It is aimed at identifying the most promising parameters and

thereby to reduce both, the experimental space and its dimensionality [241, 243]. In contrast

to “one-factor-at-a-time” experiments typically used in “conventional” optimization studies

Background 22

[244], this is preferably to be done by factorial experimental plans. Here, more than one vari-

able per experiment is varied in order to display potential interactions and synergisms [239].

Yet, for a “full factorial design” the number of experiments needed scales according to formu-

la (1-1):

nexp = kni

(1-1)

nexp number of required experiments [-]

ni number of input variables [-]

k number of levels per input variable [-]

Thus, even relatively simple systems incorporating only few input variables easily generate

experimental spaces that are no longer accessible due to limiting experimental capacities.

This problem can be addressed by “fractional factorial design” where only a subset of exper-

iments is needed [245].

However, it needs to be stated that all these strategies require a certain extent of preliminary

knowledge about the input values as well as their respective ranges. For example, an input

variable may be erroneously declared to have only a minor influence on the target output, if

its parameter range was specified too narrow [243].

Once a subset of important factors has been identified, these are optimized during subse-

quent, typically iterative, experimental studies. Therefore, numerous experimental plans have

been developed while the central composite face centred design (CCF) [246] and the com-

posite circumscribed design (CCC) [247] are probably most prominent for nonlinear models.

During CCF, additional experiments are embedded into the surface centre points between

the corners of the experimental space while during CCC, these are placed outside the exper-

imental space but at the same distance from the centre point as the distance from the centre

point to the corners (Figure 1.10). Results of such optimization studies are typically repre-

sented via contour plots or response surfaces clearly visualizing the optimal constellation of

variables within the experimental space.

Background 23

Figure 1.10: Schematic representation of central composite designs for nonlinear models. A: CCF, B: CCC. Three-variable parameter space. Blue dots indicate corner points, black dots center points and red dots additional points. Modified from [243].

Besides these “classic” DoE approaches, several alternatives have been established and

applied for bioprocess development. One of these is the genetic algorithm relying on iterative

recombination and selection of parameters from a starting population distributed across the

experimental space [248-250]. Although this approach has been shown to be suitable for

optimization studies, it may fail locating global extrema. In this context, randomized placing of

additional experiments across the investigated space may prevent the genetic algorithm from

getting stuck at local extrema [251]. Another example is the recent application of a Kriging

approximation algorithm for medium optimization integrating confidence analysis. This ap-

proach has proven its suitability for bioprocess optimization especially by its high flexibility as

the experimental space can dynamically be adjusted between distinct iterations, if needed

[175]. Moreover, Kriging was shown to be more appropriate for approximating highly nonlin-

ear relations than “classic” DoE strategies [252].

Until today, a variety of commercial products for DoE applications became available and has

widely been applied in bioprocess development. Prominent examples are Modde (Umetrics,

Umeå/Sweden), Minitab (Minitab Inc., State College/USA) or Design-Expert (Stat-Ease, Min-

neapolis/USA). Due to their proprietary nature, the exact working principles remain unclear to

a certain extent, so that numerous researchers have developed DoE toolboxes tailored to

specific needs. These typically base on Excel (Microsoft, Redmond/USA) or Matlab (Math-

Works, Natick/USA) and are often provided open source to the scientific community.

The most prominent application of DoE strategies for bioprocess development is the optimi-

zation of culture media with the analysis of different media components and their respective

optimal concentrations being the major focus. Here, typically productivity or yield is depicted

Background 24

as output variable of interest. Nevertheless, Design of Experiments has proven is applicabil-

ity in various fields of bioprocess development like process operation (temperature, pH, DOT,

feed profiles, etc.), sequence of operational actions (inoculation, harvest, induction, transfec-

tion, etc.) or downstream processing (cell separation, chromatography, etc.). A detailed over-

view of exemplary studies was given by Mandenius & Brundin [243].

1.6 Aims of this study

Despite being regarded as a promising puzzle piece within the concept of a knowledge-

based bio-economy, the industrial use of microalgae remains mostly un-economic today.

Moreover, photobiotechnology suffers from long development cycles. Especially during early

laboratory-scale development stages, where the conceptual basis of future production pro-

cesses is set, improving material and time efficiency represents a crucial issue.

Within this context, a concept for the accelerated early stage development of phototrophic

production processes was set as the major focus of this study. Therefore, the well-

established green microalga C. vulgaris producing neutral lipids as precursors for biofuel

synthesis was chosen as a biological model system. The targeted framework was designed

to improve material and time-efficiency, addressing the following modules of laboratory-scale

bioprocess development:

(i) strain maintenance

Contrary to state-of-the-art procedures in microbial biotechnology, microalgal strain

maintenance is classically conducted by serial sub-culturing of slowly growing cells.

Besides questionable long term genetic stability, these methods involve continuous

labour. To circumvent this fundamental bottleneck, the development of a cryopreser-

vation-based strain maintenance strategy was targeted. The major focus was to set

up a standardized and consequently simplified workflow reducing material and labour

effort to a minimum. In order to monitor the protocol development process, a robust

method for post-thawing viability quantification, which furthermore fits the aforemen-

tioned aspects of simplification, was set up as an intermediate objective.

(ii) parallelized cultivation

Currently, no system for the parallelized cultivation of phototrophic microorganisms is

available, and thus cultivation throughput is strictly limited. This study was embedded

into an industrial cooperation project aiming for the development of a MTP-based mi-

cro-photobioreactor enabling parallelized phototrophic cultivation under strictly con-

trolled conditions. In this context, the prototype of the reactor was to be evaluated

Background 25

with regard to non-biological as well as to biological aspects. The prototype system

integrated a measuring unit for optical monitoring of biomass during cultivation which

was evaluated towards its suitability for non-invasive quasi-online monitoring of bio-

mass formation during cultivation. In parallel, a process for the defined cultivation of

the model organism C. vulgaris in MTPs needed to be developed. Thereby, a plat-

form for parallelized phototrophic cultivation enabling elevated throughput and online

process monitoring was to be created, building up the core component within the

framework for accelerated development of phototrophic bioprocesses.

(iii) high throughput analytics

Next generation devices for parallelized phototrophic cultivation will generate an in-

creased load of low volume samples that needs to be handled during post-cultivation

analytics. Transferred to the biological model process used throughout this study, in-

tracellularly accumulated storage lipids needed to be quantified. Thereby, the aim

was to establish a standardized and comprehensively characterized workflow to meet

this task. Besides maximizing precision and accuracy, laboratory automation was fo-

cused to minimize hands-on-times and simultaneously increase analytical throughput.

Moreover, the required amount of sample material had to be in the microliter range to

effectively access microscale cultivation.

(iv) experimental design

Due to the de-bottlenecking of phototrophic cultivation and subsequent analytics, ex-

perimental throughput is expected to be significantly raised. However, experimental

throughput still does not match the number of cultivations needed for rigorous optimi-

zation of a bioprocess. These are typically influenced by more than 25 different, often

interacting, factors, while phototrophic cultivation additionally incorporates illumina-

tion-deviated parameters like intensities, spectra and dynamic changes of the latter.

In order to handle the resulting combinatorial explosion, the experimental load needs

to be reduced to a level that can be conducted within an appropriate time scale. Yet,

it is often challenging to carry out this task manually, especially if only little is known

about the process to be optimized. As already fruitfully applied, an algorithm-assisted,

iterative experimental design strategy should be utilized to efficiently carry out this

task. The algorithm was tailored to closely integrate into the accelerated bioprocess

development framework via direct interfaces to automated media preparation and

high throughput analytics.

Background 26

Figure 1.11: Framework for accelerated phototrophic bioprocess development. (i) Strain maintenance via cryopreservation coupled to (ii) phototrophic shake flask pre-cultivation en-sures continuously available genetically stable stock cultures with minimal effort. (iii) A paral-lelized MTP-based micro-photobioreactor enables increased cultivation throughput and real-time monitoring of biomass. (iv) A workflow for automated product quantification matches the elevated cultivation throughput. (v) Algorithm-assisted experimental design takes full ad-vantage of the experimental throughput provided by the platform (vi) being additionally close-ly linked to automated media preparation.

Material and methods 27

2. Material and methods

2.1 Chemicals

All chemicals used were of analytical grade and purchased from either Sigma-Aldrich (Stein-

heim/Germany) or Roth (Karlsruhe/Germany). Table 2.1 gives an overview of all compounds.

Table 2.1: Overview of chemicals.

substance formula

boric acid H3BO3

calcium chloride dihydrate CaCl2 · 2 H2O

chloroform CHCl3

chloropentafluorobenzene C6ClF5

chlorophyll a C55H72O5N4Mg

chlorophyll b C55H70O6N4Mg

cobalt(II) sulphate heptahydrate CoSO4 · 7 H2O

copper(II) sulphate pentahydrate CuSO4 · 5 H2O

dimethyl sulfoxide C2H6OS

disodium EDTA dihydrate C10H14N2Na2O8 · 2 H2O

dipotassium hydrogenphosphate K2HPO4

ethylene glycol C2H6O2

glucose monohydrate C6H12O6 · H2O

heptacosafluorotributylamine C12F27N

heptane C7H16

hydrochloric acid HCl(aq)

iron(II) sulphate heptahydrate FeSO4 · 7 H2O

L-proline C5H9NO2

magnesium sulphate heptahydrate MgSO4 · 7 H2O

manganese chloride MnCl2

MES monohydrate C6H13NO4S · H2O

methanol CH3OH

Nile red C20H18N2O2

penicillin-G sodium salt C16H17N2NaO4S

continued on next page

Material and methods 28

continuation from previous page

potassium dihydrogenphosphate KH2PO4

potassium hydroxide KOH

sodium chloride NaCl

sodium hydroxide NaOH

sodium molybdate dihydrate NaMoO4 · 2 H2O

sodium nitrate NaNO3

sulfuric acid H2SO4

zinc sulphate heptahydrate ZnSO4 · 7 H2O

Furthermore, the analytical standards F.A.M.E. Mix RM-6 (Sigma-Aldrich, Stein-

heim/Germany), M-1603, M-1620 and M-1633 (Lipidox, Stockholm/Sweden) and cis-7-

hexadecenoic acid methyl ester (Biomol, Hamburg/Germany), the Spectroquant

1.09713.0002 nitrate test (Merck, Darmstadt/Germany) and CASYton buffer (Schärfe Sys-

tems, Reutlingen/Germany) were used.

2.2 Devices

All devices used during experiments are listed in Table 2.2.

Table 2.2: Overview of devices. CZ: Czech Republic, G: Germany, SW: Sweden, US: United States

device type manufacturer

autoclave VE-40 Systec, Linden/G

FVA3-A1 Fedegari, München/G

autosampler MPS 2 Gerstel, Mühlheim an der Ruhr/G

balance CPA225D Sartorius, Göttingen/G

LA8200S Sartorius, Göttingen/G

centrifuge Biofuge Pico Heraeus, Hanau/G

Labofuge 400R Heraeus, Hanau/G

data logger GL220 midi logger Graphtec, Kelkheim/G

drying cabinet T6120 Heraeus, Hanau/G

flat panel photobioreactor FMT 150/1000 Photon Systems Instruments, Dra-

sov/CZ

continued on next page

Material and methods 29

continuation from previous page

freeze dryer LT-105 Christ Gefriertrocknungsanlagen,

Osterode am Harz/G

gas chromatograph 6890 N Agilent Technologies, Wald-

bronn/G

gas mixer MX4/4 DasGip, Jülich/G

light meter LI-250A LI-COR, Lincoln/US

light microscope DM LB Leica Microsystems, Wetzlar/G

liquid handling arm Varispan PerkinElmer, Rodgau/G

liquid handling robot JANUS Integrator PerkinElmer, Rodgau/G

MTP microbioreactor BioLector® m2p-labs, Baesweiler/G

MTP gripper MTP railgripper PerkinElmer, Rodgau/G

MTP photo-microbioreactor Photo BioLector® m2p-labs, Baesweiler/G

MTP photometer EnSpire PerkinElmer, Rodgau/G

MTP shaker/heater Teleshake95 Inheco, Martinsried/G

particle counter MultiSizer 3 Beckman Coulter, Krefeld/G

pH meter S20 SevenEasy Mettler Toledo, Gießen/G

photometer UV-1800 Shimadzu, Duisburg/G

quantum sensor LI-190 LI-COR, Lincoln/US

shaking incubator Multitron Standard Infors HT, Einsbach/G

Multitron Pro Infors HT, Einsbach/G

stirred tank bioreactor Bioblock stirrer tank

reactor

DasGip, Jülich/G

spectrometer CAS140CT Instrument Systems, München/G

test tube photobioreactor MC 1000 Photon Systems Instruments, Dra-

sov/CZ

thermal mass flow control-

ler

SLA5080 Brooks Instruments, Dresden/G

thermostatic circulator 2219 Multitemp II Bromma/SW

ToF mass spectrometer Micromass GCT

Premier

Waters, Eschborn/G

thermomixer Thermomixer comfort Eppendorf, Hamburg/G

ultra-low temperature

freezer

C660 New Brunswick, Nürtingen/G

Material and methods 30

2.3 Microorganism

Cultivations were conducted using axenic C. vulgaris 211-11b [30]. The initial stock culture

was obtained from the Culture Collection of Algae at the University of Göttingen (Germany).

2.4 Medium

Cultivations were carried out in variants of an enriched Bold’s Basal Medium (enBBM) [253]

prepared by aseptic supplementation of desalted water with stock solutions. The medium

composition was varied by adjusting the applied volumes of stock solutions. Table 2.3 gives

the recipe for the reference medium (enBBMref) which was used if not stated otherwise.

Table 2.3: Preparation of enBBMref from stock solutions. * for heterotrophic cultivation only ** pH adjusted to 6.5 using sodium hydroxide *** storage at -20 °C

stock solution relative fraction

[% (v v-1)] compound

stock concentration

[g L-1]

glucose* 2.5 C6H12O6 · H2O 440.0

MES buffer** 5.0 MES · H2O 213.2

nitrate 1.5 NaNO3 100.0

phosphate** 1.0 K2HPO4 60.0

KH2PO4 140.0

salt 0.25 MgSO4 · 7 H2O 75.0

NaCl 2.5

calcium 0.25 CaCl2 · 2 H2O 50.0

trace elements 2.0

ZnSO4 · 7 H2O 0.882

MnCl2 0.144

NaMoO4 · 2 H2O 0.12

CuSO4 · 5 H2O 0.157

CoSO4 · 7 H2O 0.047

H3BO3 1.14

EDTA 0.2 Na2EDTA · 2 H2O 50.0

KOH 31.0

acidic iron 0.2 FeSO4 · 7 H2O 4.98

H2SO4 1.84

penicillin*** 0.2 sodium penicillin-G 50.0

Material and methods 31

During experimental design studies (section 3.5), medium variants were prepared by an in-

house laboratory robotic platform [168, 169]. The setup consisted of a JANUS Integrator liq-

uid handling platform equipped with a Varispan liquid handling arm and a MTP railgripper.

The whole setup was installed inside a class 2 laminar flow hood to ensure sterile conditions.

The minimal volume to be pipetted was set to 10 µL and the concentrations of all stock solu-

tions were adapted to the pipetting ranges of the system. During continuous shaking on an

integrated Teleshake95 MTP shaker/heater at 750 rpm, 2500 µL per medium was prepared

in a pre-sterilized disposable 48-well MTP-R-48-B Round Well Plate (m2p-labs, Baes-

weiler/Germany). Subsequently, 950 µL of medium per well was transferred into the MTP to

be used for cultivation (section 2.7.4).

2.5 Sterilisation

Thermostable medium components and materials were sterilized using either a VE-40 or an

FVA3-A1 autoclave at 121 °C and 1 bar overpressure in saturated steam atmosphere for

21 min. Thermolabile solutions (cryo additive solution, MES buffer, trace elements, acidic

iron and penicillin-G) were filter sterilized using 0.22 µm PES Millex-GP syringe filter units

(Merck, Darmstadt/Germany) and 0.2 µm PES bottle top vacuum filtration cups (VWR,

Darmstadt/Germany), respectively.

2.6 Heterotrophic cultivation

2.6.1 Shake flask cultivation

Heterotrophic pre-cultivation was carried out in 500 mL Erlenmeyer shake flasks that were

filled with 150 mL enBBMref and inoculated with 800 µL cryoculture (section 2.8). Incubation

took place in a Multitron Standard shaking incubator at 25 °C, 250 rpm and a shaking diame-

ter of 25 mm. The flasks were protected from light by an aluminium foil coating. The culture

was harvested during exponential growth by centrifugation for 5 min at 4 °C and 3939 · g

(Labofuge 400R). The supernatant was discarded and the pellet re-suspended in sterile

0.9 % (w v-1) NaCl solution to OD750 = 50 to be used as an inoculum in further cultivation ex-

periments.

2.6.2 Stirred tank reactor cultivation

Heterotrophic bioreactor cultivations were carried out in 2 L stirred tank bioreactors equipped

with two six-bladed impellers while the culture broth was protected from light by covering the

reactors with aluminium foil. 1 L enBBM (1 g L-1 NaNO3) was inoculated to OD750 = 0.2 using

biomass from exponentially growing heterotrophic shake flask pre-cultures (section 2.6.1).

Cultivation conditions were set to 25 °C, pH control via 405-DPAS-SC-K8S/225 pH probe

Material and methods 32

(Mettler Toledo, Gießen/Germany) at 6.5 with 1 M HCl as correction medium, stirrer speed

500 rpm and gassing with 6 NL h-1 air. During the cultivation, growth and metabolic state

were monitored online via the carbon dioxide transfer rate provided by a GA4 off-gas analys-

er.

2.6.3 MTP cultivation

Heterotrophic cultivation at microscale was conducted in pre-sterilized, disposable 48-well

MTP-48-B FlowerPlates® (m2p-labs, Baesweiler/Germany) incubated in a BioLector® micro-

bioreactor system. Each well was filled to 1 mL with enBBMref and inoculated with 50 µL. In-

oculum was either cryocultures (section 2.8) thawed for 2 min at 40 °C in a water bath or

biomass harvested from bioreactor cultivation (section 2.6.2) that was concentrated by cen-

trifugation (5 min, 4 °C, 3939 · g, Labofuge 400R) and subsequently resuspended in sterile

0.9 % (w v-1) NaCl solution to OD750 = 25. The plates were sealed with a F-GPR48-10 gas-

permeable sealing foil (m2p-labs, Baesweiler/Germany) reducing evaporation to < 2 % per

day. The following cultivation conditions were applied: 25 °C, shaking frequency 1000 rpm,

shaking diameter 3 mm and relative humidity ≥ 85 %. The closed cultivation chamber of the

BioLector® ensured light protected cultivation conditions.

2.7 Phototrophic cultivation

2.7.1 Shake flask cultivation

Phototrophic shake flask cultivation was carried out in a Multitron Pro shaking incubator at

25 °C, 200 rpm and a shaking diameter of 25 mm. The incubation chamber was illuminated

with 7x F25W/30/GRO fluorescent lamps (Sylvania, Erlangen/Germany) resulting in a photo-

synthetically active radiation (PAR) of 90 - 200 µmol m-² s-1 depending on the position on the

shaking tray (Figure 6.1 A, appendix). Thus, the desired light intensity could be controlled by

targeted positioning of the flasks and was always adjusted to 200 µmol m-2 s-1. The culture

liquid was gassed with 9 NL h-1 air enriched to 2.5 % (v v-1) CO2. Therefore, compressed air

and CO2 were mixed via SLA 5080 thermal mass flow controllers, followed by sterile filtration

via a 50 mm 0.2 µm PTFE aeration filter (DIAFIL, Wieliczka/Poland) and a 250 mL humidify-

ing bottle (200 mL liquid volume) before introduction into the shake flask via a sparger (Fig-

ure 6.1 B, appendix).

Pre-cultivation 2.7.1.1

For pre-cultivation 150 mL enBBMref in a 500 mL Erlenmeyer flask was inoculated with

900 µL of cryoculture (section 2.8) and cultivated using the system described in sec-

tion 2.7.1. The cells were harvested during exponential growth (typically after 60 h of cultiva-

Material and methods 33

tion), concentrated by centrifugation (5 min, 4 °C, 3939 · g, Labofuge 400R) and subsequent-

ly resuspended in sterile 0.9 % (w v-1) NaCl solution. The suspension was adjusted to a

biovolume of 2.5 µL mL-1 to obtain a stock solution for the inoculation of subsequent cultiva-

tion experiments. For experimental design studies using MTP cultivation coupled to automat-

ed media preparation, the biomass concentration of this stock was reduced to a biovolume of

2 µL mL-1.

Reference cultivation 2.7.1.2

For reference cultivation 150 mL enBBMref in a 500 mL Erlenmeyer flask was inoculated to a

biovolume of 0.1 µL mL-1 using exponentially growing cells from a phototrophic pre-culture

(section 2.7.1.1) and cultivated using the system described in section 2.7.1.

2.7.2 Test tube cultivation

For cultivation at 100 mL scale, the MC 1000 test tube photobioreactor was used. 250 mL of

enBBMref was inoculated to a biovolume of 0.1 µL mL-1 using biomass from exponentially

growing phototrophic pre-cultures (section 2.7.1.1) and 80 mL was filled into each vessel of

the reactor. The following cultivation conditions were applied: 25 °C, 200 µmol m-2 s-1 ± 7.4 %

PAR and gassing with 4.8 NL h-1 air enriched to 2.5 % (v v-1) CO2 using the mass flow con-

troller setup described in section 2.7.1.

2.7.3 Flat panel reactor cultivation

Cultivation at 1 L scale was carried out in a FMT 150/1000 flat panel photobioreactor. 1 L

enBBMref was inoculated to a biovolume of 0.1 µL mL-1 using biomass from exponentially

growing phototrophic pre-cultures (section 2.7.1.1). The following cultivation conditions were

applied: 25 °C, 200 µmol m-2 s-1 PAR (blue:red = 1:1) and gassing with 60 NL h-1 air enriched

to 2.5 % (v v-1) CO2 using a MX4/4 gas mixer coupled to a 1 L humidifying bottle with a

50 mm 0.2 µm PTFE aeration filter (DIAFIL, Wieliczka/Poland).

2.7.4 Phototrophic MTP cultivation

Phototrophic microscale cultivation was conducted in pre-sterilized, disposable 48-well

MTP-48-B FlowerPlates® (m2p-labs, Baesweiler/Germany). The black corpus of the plates

efficiently avoids well-to-well cross-talk effects. The polystyrene (PS) bottom with its high

transmittance within the spectrum of photosynthetically active radiation (≈ 90 %) made it a

suitable material for illumination. Each well was filled to a total volume of 1 mL enBBMref pre-

viously inoculated to a biovolume of 0.1 µL mL-1 using biomass from exponentially growing

phototrophic pre-cultures (section 2.7.1.1). During experimental design studies, inoculation

took place immediately after automated medium preparation (section 2.4) using the robotic

Material and methods 34

platform as well. Therefore, 50 µL of inoculum was added to the 950 µL of medium, resulting

in a total cultivation volume of 1 mL per well. The plates were sealed with an F-R48-10 perfo-

rated sealing foil for evaporation reduction (m2p-labs, Baesweiler/Germany) pasted over with

an additional F-GP-AB10 gas-permeable sealing foil (m2p-labs, Baes-weiler/Germany). The

MTPs were incubated using a micro-photobioreactor prototype based on the BioLector® II

microbioreactor system. A detailed description of the apparatus is given in section 3.2. The

incubation conditions were set as follows: 25 °C, 1200 rpm shaking frequency,

2.5 % (v v-1) CO2, 200 µmol m-2 s-1 PAR, ≥ 85 % relative humidity.

2.8 Cryopreservation

Either exponentially growing or stationary phase cells were harvested from heterotrophic

shake flask or bioreactor cultivation, centrifuged for 5 min at 4 °C and 3939 · g (Labofuge

400R) and the pellet was re-suspended in sterile 0.9 % (w v-1) NaCl solution to OD750 = 50.

The suspension was vigorously mixed with an equal volume of sterile cryo additive solution

containing 10 % (v v-1) dimethyl sulfoxide (DMSO), 10 % (v v-1) ethylene glycol and

10 % (w v-1) L-proline in desalted water previously reported as advantageous for the cryo-

preservation of C. vulgaris [254]. 1 mL aliquots were dispensed into 2 mL cryotubes (VWR,

Darmstadt/Germany) and incubated for 10 min at room temperature. The vials were then

frozen by 4 different procedures (i–iv). The first method was very rapid freezing by plunging

them into liquid nitrogen (i), while the other methods (ii, iii and iv) used slower freezing veloci-

ty in a -80 °C freezer. For the latter, tubes were placed in a C660 freezer (ii) inside a polysty-

rene box (PS, wall thickness 34 mm), (iii) in a commercial freezing box “Mr. Frosty” (Thermo

Scientific, Waltham/United States), and (iv) without any additional protection, respectively. All

vials were stored at -80 °C for at least 24 h.

2.9 Biomass detection

As an easily accessible indicator for biomass, the optical density (OD) was measured by light

extinction using an UV-1800 photometer. To avoid interference with algal pigments, meas-

urements were taken at a wavelength of 750 nm as recommended by Griffiths et al. [217]. If

needed, the samples were diluted using 0.9 % (w v-1) NaCl solution to match the linear range

of the photometer (0.1 - 0.3). The samples were analysed in 10 mm PS semi-micro cuvettes

(ratiolab, Dreieich/Germany) while desalted water served as a blank.

Cell counts and biovolume were determined using a MultiSizer 3 particle counter [255]

equipped with a 30 µm capillary in volumetric control mode. The cells were diluted to

OD750 ≤ 0.025 in CASYton buffer (Schärfe Systems, Reutlingen/Germany) and only particles

sizing from 1.8 to 14 µm were analysed.

Material and methods 35

During MTP cultivations, biomass was monitored in each well via the integrated scattered

light analyser at either 620 nm (commercial BioLector®) or at 750 nm (Photo BioLector®).

The cell dry weight (CDW) was determined by gravimetry. To maximize measurement accu-

racy, 10 mL cell suspension was centrifuged in pre-dried and weighed 15 mL tubes for 5 min

at 3939 · g (Labofuge 400R) whenever possible. The supernatant was discarded and the

pellet freeze-dried using an LT-105 freeze dryer until at constant weight. After acclimatization

to room temperature in a desiccator, weighing was repeated and the CDW was calculated

from the mass difference according to formula (2-1).

CDW = (mb-mn)∙f (2-1)

CDW cell dry weight [g L-1]

mb brutto tube weight [g]

mn netto tube weight [g]

f dilution factor [-]

For reduced sample volumes, CDW determination was carried out at 2 mL scale in 2 mL

tubes for 5 min at 16060 · g (Biofuge Pico).

2.10 Nitrate quantification

Nitrate was quantified using the Spectroquant 1.09713.0002 nitrate test (Merck, Darm-

stadt/Germany) according to the manufacturer’s specifications, scaled down to one quarter of

the recommended volume. Before measurement, biological samples were filtrated using

0.2 µm CA syringe filters (DIA-Nielsen, Düren/Germany) and the cell-free filtrate was pre-

diluted with desalted water to fit the linear range of the assay, if needed. Measurements were

carried out in UV semi-micro cuvettes (Brand, Wertheim/Germany) using an UV-1800 pho-

tometer.

2.11 Gravimetric lipid quantification

Gravimetric quantification of the cellular lipid content was carried out via a modified single-

step extraction method as reported by Axelsson and Gentili [256]. An aliquot of 40 mL cell

suspension was filled into a 50 mL tube, centrifuged for 5 min at 3939 · g and 4 °C (Labofuge

400R) and the supernatant was discarded. After resuspending the pellet in 40 mL

0.9 % (w v-1) NaCl solution, centrifugation was repeated and the resulting supernatant was

discarded once more. The cell pellet was lyophilized using an LT-105 freeze dryer until con-

stant weight. The freeze-dried biomass was homogenized using a spatula and a defined ali-

Material and methods 36

quot of ≤ 100 mg was filled into a 50 mL tube. After addition of 14 mL of chloroform and 7 mL

of methanol, the lipids were extracted at 60 °C for 1 h in a Multitron Standard shaking incuba-

tor at 250 rpm and a shaking diameter of 25 mm. 5.6 mL of 0.73 % (w v-1) NaCl solution was

added and vigorously mixed. The resulting phases were separated by centrifugation for

2 min at 3939 · g and 4 °C (Labofuge 400R). The lower chloroform-phase was quantitatively

collected and filled into a dried and pre-weighed 15 mL tube. The lipid extract was evapo-

rated at 40 °C until constant weight using a T6120 drying cabinet. After cooling to room tem-

perature in a desiccator and weighing the extract, the lipid content (LC) was calculated ac-

cording to formula (2-2).

LC = mb-mn

mB

∙100 %

(2-2)

LC lipid content [% (w w-1)]

mb brutto tube weight [mg]

mn netto tube weight [mg]

mB mass of extracted biomass [mg]

2.12 Fluorometric lipid quantification

For the quantification of the intracellular neutral lipid content at high throughput, a modified

version of the Nile red-based fluorometric assay described by Chen et al. [257] was used at

96 well MTP format. Enabling elevated throughput, the assay was automated using the in-

house laboratory robotic platform described in the context of automated media preparation

(section 2.4). Besides the components already used for this task, the platform additionally

carries an EnSpire MTP photometer that was used for fluorescence measurements. The as-

say was designed to be capable of handling up to 48 samples per run.

After initial determination of the biovolume of the respective sample (section 2.9), all subse-

quent steps were carried out using the robotic platform. Each sample was individually diluted

to a biovolume of 0.2 µL mL-1 using 0.9 % (w w-1) NaCl in an MTP-48-OFF offline Flower-

Plate® (m2p-labs, Baesweiler/Germany) resulting in a total volume of 1175 µL. 375 µL of a

freshly prepared staining solution containing 4 mg L-1 Nile red in DMSO was added to each

diluted sample, the plate was sealed with a self-gluing aluminium foil (Greiner Bio-One, So-

lingen/Germany) and incubated at 40 °C and 750 rpm on the Teleshake95 for 3 h. After-

wards, the stained samples were subsequently transferred to three black 96 well MTPs with

clear PS F-bottom (Greiner Bio-One, Solingen/Germany) in five 200 µL aliquots per sample.

During this step, the samples were continuously shaken to ensure proper mixing. After the

transfer of samples to a 96 well MTP, an equivalent volume of water was filled into the emp-

Material and methods 37

tied wells of the FlowerPlate® to keep its weight constant. The plates were successively

transferred to the EnSpire photometer and internally shaken for another 15 s at 600 rpm and

a shaking diameter of 1 mm in orbital mode. After excitation at 480 nm the fluorescence

spectrum within 570 - 580 nm was recorded and the average value across the spectrum was

calculated for each individual well. Transferring the samples to the 96 well MTPs, weight bal-

ancing of the FlowerPlate® and measuring with the EnSpire photometer were scheduled to

minimize the workflow runtime. Figure 2.1 gives a schematic overview of the complete assay.

Figure 2.1: Workflow for the automated quantification of intracellular neutral lipids. Modified from [258].

2.13 Fatty acid fingerprinting

The spectrum of fatty acids accumulated by C. vulgaris was analysed by semi-quantitative

gas chromatography time-of-flight mass spectrometry (GC-ToF-MS) of lipid extracts.

An appropriate volume of culture suspension was centrifuged for 5 min at 3939 · g and 4 °C

(Labofuge 400R), the supernatant discarded, the cells washed with 0.9 % (w v-1) NaCl solu-

tion and centrifuged once more. The pellet was lyophilized using an LT-105 freeze dryer until

constant weight and subsequently homogenized in a mortar. An aliquot of 10 mg was filled

into a test tube. 1700 µL of acidic methanol containing 10 % (v v-1) sulphuric acid was added

for the in-situ transesterification of the lipids at 60 °C for 4 h in a Thermomixer comfort at

Material and methods 38

750 rpm. After cooling to room temperature, 300 µL desalted water was added. The fatty acid

methyl esters were extracted by addition of 1200 µL heptane and vigorously mixing for at

least 1 min. The heptane fraction was collected and stored at -20 °C prior to analysis, if

needed.

Sample analysis was carried out as previously established by Paczia et al. [259] with only

slight modifications. A 6890 N gas chromatograph equipped with a 30 m EZ-Guard VF-5 ms

column supported by a 10 m guard column (Agilent Technologies, Waldbronn/Germany) was

used. 1 µL sample was injected either splitless or at varying splits via a MPS 2 autosampler.

A constant flow rate of 1 mL min-1 helium was applied. Each run started at 60 °C with a hold

time of 2 min, followed by a linear ramp of +12 K min-1 up to 300 °C with a hold time of 8 min.

The transferline temperature was set to 300 °C. The gas chromatograph was coupled to a

Micromass GCT Premier time-of-flight mass spectrometer operated in positive electron im-

pact mode at an electron energy of 70 eV and a source temperature of 180 °C. During

measurements, the accurate masses were corrected to a single point lock mass of chloro-

pentafluorobenzene as a reference at a mass/charge-ratio of 201.9609 while tuning and cali-

bration of mass spectrometer were performed with the fragment pattern of heptaco-

safluorotributylamine. Data was acquired at a scan rate of 0.09 s and an interscan delay of

0.01 s resulting in 10 scans s-1.

A baseline noise corrected fragment pattern was used to identify unknown compounds by

comparison to the in-house database “JuPoD”, the commercial database “NIST11” (National

Institute of Standards and Technology, Gaithersburg/United States) and the freely available

database “GMD” (MPI of Molecular Plant Physiology, Potsdam/Germany) [260]. The analyti-

cal standards F.A.M.E. Mix RM-6 (Sigma-Aldrich, Steinheim/Germany), M-1603, M-1620 and

M-1633 (Lipidox, Stockholm/Sweden) and cis-7-hexadecenoic acid methyl ester (Biomol,

Hamburg/Germany) were used for verification. Unknown peaks were identified by a structural

combination of elemental compositions and verified by virtual derivatization and fragmenta-

tion of the predicted structure. The obtained peak areas were normalized to the total peak

area of all detected lipid compounds in the respective sample. Relative fractions of co-eluting

compounds were calculated by their corresponding extracted-ion chromatogram signals.

2.14 Microscopy

For routine contamination controls, selected samples were investigated by light microscopy.

Therefore, 10 µL of the respective samples was controlled using a DM LB light microscope at

1000x magnification and phase contrast 3.

Material and methods 39

2.15 Offline pH

Offline pH was measured without any pretreatment using a S20 SevenEasy pH meter

equipped with a 6.0234.100 micro electrode (Metrohm, Filderstadt/Germany).

2.16 Temperature at single well resolution

Applying standard phototrophic cultivation conditions (section 2.7.4), temperature measure-

ments within the individual wells of an MTP were carried out using a GL220 midi Logger.

2.17 Illumination intensity

Light intensities were measured by means of photosynthetically active radiation

(400 - 700 nm) using a LI-250A light meter equipped with a LI-190 quantum sensor.

2.18 Evaporation

At standard phototrophic cultivation conditions (section 2.7.4), evaporation from the MTP was

determined gravimetrically. A plate containing 1 mL desalted water per well was incubated

for at least one day and liquid loss was measured using a CPA225D semi-micro balance.

2.19 Acquisition of absorption and emission spectra

Absorption characteristics of chlorophyll a and b were investigated using an UV-1800 pho-

tometer. The absorbance spectrum of 10 mg L-1 chlorophyll a and b (dissolved in DMSO),

respectively were acquired within the interval of 400 – 700 nm using 10 mm PS semi-micro

cuvettes (ratiolab, Dreieich/Germany) while pure DMSO as a blank.

The emission spectra of daylight and light emitting diodes (LEDs) were acquired using a

CAS140CT spectrometer. Within the spectrum of photosynthetically active radiation, the ac-

quired wavelength-resolved illumination intensities were translated to photon flux densities

according to formula (2-3).

PARλ = Pλ∙λnm∙10

-3

NA∙h∙c (2-3)

PARλ

photosynthetically active radiation (wavelength λ) [µmol m-2 s-1]

Pλ

illumination intensity (wavelength λ) [W m-2]

λnm wavelength [nm]

NA Avogadro’s number [6.022 · 1023 mol-1]

h Planck’s constant [6.626 · 10-34 W s-2]

c speed of light [2.998 · 108 m s-1]

Material and methods 40

2.20 Pipetting accuracy and precision

Pipetting accuracy (systematic errors) and precision (stochastic errors) were determined

gravimetrically, as previously described in literature [235]. Different amounts of desalted wa-

ter were transferred to pre-dried (48 h, 80 °C, cooled to room temperature in a desiccator)

glass vials and quantified via a CPA225D semi micro balance. For manual pipetting, 100 µL

and 1000 µL Research Plus pipettes (Eppendorf, Hamburg/Germany) were used, while au-

tomated liquid handling was performed by the JANUS Integrator robotic system in single dis-

pense per aspirate mode. All pipetting steps were repeated in 10 individual replicates. Accu-

racy and precision were calculated according to formulas (2-4) to (2-7).

m̅=1

n∑mi

n

i=1

(2-4)

acc = m-mt

mt

∙100%

(2-5)

var = 1

n-1∑ (mi-m)

2

n

i=1

(2-6)

prec = √var

m∙100%

(2-7)

acc accuracy [%]

m average amount of water transferred [mg]

mi amount of transferred water [mg]

mt

corresponding target amount to be transferred [mg]

n number of replicates [-]

prec precision [%]

var variance [mg]

2.21 Experimental design

During DoE studies (section 3.5), experimental plans were generated and analysed using the

open source “KriKit DoE and Analysis Toolkit”, a hybrid method combining “classic” DoE and

Kriging approximation. The toolkit is freely available at github.com/modsim/KriKit. The ap-

plied optimization strategy was adopted from [175]. Initially, fractional and full factorial exper-

imental designs were applied for estimating single component effects and combinatorial in-

Material and methods 41

teractions. Myers et al. [261] provide a comprehensive overview of these “classical” DoE

methods.

Pre- and main cultivations were carried out as described in section 2.7.1.1 and 2.7.4 while

medium preparation according to the respective experimental plan was performed in an au-

tomated manner as described in section 2.4. During every MTP cultivation five culture wells

were run with enBBMref as a reference and all cultivations were randomized across the re-

spective MTP in order to avoid any bias by the operator or positional effects. The reference

cultures were used to monitor and check plate to plate comparability and to minimize batch

to batch effects by normalization of all results to these references.

Based on the initially collected data, the statistically more advanced concept of Kriging was

applied for data analysis, visualization, and for designing further experiments with potentially

improved lipid productivity. Kriging is an interpolation method that provides unbiased approx-

imations of the underlying nonlinear functional relationships between media composition and

lipid productivity with minimal prediction error. This method originates from geostatistics and

has recently been adapted for optimizing biotechnology processes [175]. Further mathemati-

cal details of the Kriging method can be obtained from Cressie [262].

Given a Kriging model of the current data set, further experiments were designed to maxim-

ize the expected improvement (EI). This experimental design strategy seeks a compromise

between maximizing lipid productivity and reducing prediction uncertainty of the Kriging ap-

proximation in relevant regions of the parameter space [263]. In a comparative study, EI has

been found to outperform other sampling strategies in Kriging-based optimization [264].

In sequential optimization, new experiments are typically planned at maximal EI. Parallel ex-

periments, as in the present study, are most efficiently planned by sampling from the EI dis-

tribution. In a non-deterministic sampling process, using the Markov Chain Monte Carlo

(MCMC) method, new experiments are selected with probability proportional to their EI. Nat-

urally, experiments with high EI are preferred over experiments with lower EI, which nonethe-

less have a reduced chance of being selected, whereas experiments with zero EI are strictly

excluded. It has recently been demonstrated that MCMC sampling can significantly reduce

the number of required experiments in process optimization [265]. In the present study, the

Delay Rejection Adaptive Metropolis algorithm [266] was applied with a chain length of

10,000 elements, of which the first 1,000 are discarded (burn in phase of the MCMC meth-

od).

Material and methods 42

2.22 Statistical analysis

For all statistical analyses, two-sided t-tests for unequal variances (either 90 % or 95 % sig-

nificance level) were applied using Origin9.1.0G (OriginLab Corporation, Northampton/United

States).

Results and discussion 43

3. Results and discussion

The development cycles for phototrophic production processes need to be drastically short-

ened to set up economically feasible processes within an appropriate scale of time and mate-

rial (section 1.6). In this context, a concept for the accelerated development of phototrophic

bioprocesses at laboratory-scale was identified to be a promising approach. Thus, the estab-

lishment of a blueprint framework incorporating elevated throughput cultivation and tailored

peripheral methods was focused.

3.1 Strain maintenance

In the context of biological research and production, the reliable maintenance of strains re-

mains a fundamental issue. The most common method for algal strain maintenance is serial

sub-culturing on solid or in liquid media, applying low temperature and illumination conditions

to minimize biological activity and growth [267]. Besides being labour- and material intensive

and holding a risk of contamination, lasting genetic stability is an issue, as phenotypic

changes have been observed applying sub-culturing techniques at the scale of decades

[268]. The resulting material and labour effort is in direct contrast to the demands on any

workflow integrated into an accelerated bioprocess development framework.

As an alternative procedure, cryopreservation of microalgae has been investigated since the

1960s. Cryopreservation is state of the art procedure in the field of microbial biotechnology

and cell culture technology, showing many advantageous aspects in terms of long-term sta-

bility, material effort and convenience [269]. While procedures for cryopreservation of micro-

bial strains are usually rather simple, complex protocols derived from cell culture technology

were used for microalgae implying multi-step cooling and freezing at defined rates, coupled

to storage in liquid nitrogen [270]. Nowadays, a wide variety of protocols for cryopreservation

of algae exists, as reviewed elsewhere [271]. Different parameters have been addressed and

substantial progress has been made. It has been shown that an optimal cooling rate minimiz-

ing damage by either intra- or extracellular ice formation represents one of the key parame-

ters for successful cryopreservation of algae [272, 273]. Furthermore, many protective

agents have been screened and their simultaneous use can result in symbiotic effects im-

proving cell viability [254].

Although it was shown to be a suitable tool to preserve genetic stability [274], cryopreserva-

tion is not routinely applied in algae sciences so far. Many protocols require specialized

equipment to enable for defined cooling rates probably hampering the establishment of cryo-

preservation. Furthermore, a procedure performing rather well for a distinct species may be

entirely unsuitable for another, as reported viabilities greatly differ [271].

Results and discussion 44

A large set of different methods has been applied for post-preservation viability estimation.

Some approaches use techniques like staining [275], chlorophyll [276] or protein [277]

measurement and photosynthetic oxygen formation [278] but most of the time, plating and

counting of the colony forming units is applied. Besides being labour intensive and inconven-

ient, this method is highly error-prone, suffering from low reproducibility due to subjective

manual counting of colony forming units [279].

Facing these limitations, the development of an easy-to-use protocol for the cryopreservation

and maintenance of the model organism C. vulgaris was focussed. To enable for high cell

densities, the cells to be preserved were grown in heterotrophic mode and subsequently re-

adapted to illumination by phototrophic shake flask cultivation. In this context, the time need-

ed for re-adaption and inoculum production was investigated. In addition, an alternative ap-

proach for simple and robust post-thawing viability assessment was set up, which is based

on growth analysis in suspension culture and circumvents the pitfalls of conventional plating

techniques.

3.1.1 Evaluation of different cryopreservation strategies

Based on protocols recently published [254, 280], an easy-to-handle protocol for microalgae

cryopreservation was deduced (section 2.8). The biomass used was obtained from hetero-

trophic cultivation to exclude potential effects of illumination or light limitation and enabled

high cell density inoculum from defined conditions. The use of specialized cooling rate con-

trolling devices was replaced by using readily available liquid nitrogen or -80 °C freezer for

the cooling process. Using the latter also for storage further decreased the effort of strain

maintenance compared to storage using liquid nitrogen.

The different cryopreservation strategies were evaluated using biomass from heterotrophic

cultures of C. vulgaris. Either exponentially growing cells, or cells being stationary for 120 h

due to nitrogen depletion were cryopreserved applying four different methodologies. Samples

were

(i) directly frozen in liquid nitrogen with subsequent storage at -80 °C,

(ii) frozen at -80 °C inside a polystyrene box,

(iii) frozen at -80 °C inside a commercial freezing container (“Mr. Frosty”), or

(iv) frozen at -80 °C without any additional protection.

After thawing of preserved cultures obtained with the different protocols, an aliquot was used

as inoculum for cultivation in MTPs under identical conditions (Figure 3.1).

Results and discussion 45

Figure 3.1: Heterotrophic growth of C. vulgaris after freezing preservation. The cells were cryopreserved according to different methods and used as inoculum immediately after thaw-ing. A: Representative growth curves of cultures inoculated from cryocultures obtained from growing phase cells, B: Representative growth curves of cultures inoculated from cryocul-tures obtained from stationary phase cells; enBBM (10 g L-1 glucose), 25 °C, shaking fre-quency 1000 rpm, VL = 1.05 mL, relative humidity ≥ 85 %. Modified from [281].

The different freezing protocols differentially influenced the post-thawing growth of

C. vulgaris, especially for the cryopreserved cultures derived from growing phase cells. Simi-

lar growth pattern of C. vulgaris was observed for the three methods using -80 °C for freezing

(ii-iv) independent of the fact that biomass from exponential growth or stationary phase (in-

duced by nitrogen limitation) was used for cryopreservation (Figure 3.1 A+B). Also, decreas-

ing the cooling rate by use of a polystyrene box (ii) or a commercial freezing container (iii)

showed only marginal impact on growth pattern after thawing.

In contrast, plunging into liquid nitrogen (i) resulted in a significantly delayed growth pattern

for cells cryopreserved from stationary phase (Figure 3.1 B) and very poor growth for cells

derived from growth phase, which is hardly detectable even after 120 h cultivation time (Fig-

ure 3.1 A). As the cells did not reach the stationary phase within the cultivation time applied,

there is strong evidence that cryopreservation with very high cooling rate massively harmed

the cells and resulted in very low post-thawing viability. This effect is less pronounced when

0 24 48 72 96 120

0

6

12

18

24

30

0 24 48 72 96 120

0

6

12

18

24

30s

catt

ere

d l

igh

t

[-]

time [h]

-196 °C liquid nitrogen (i)

- 80 °C PS box (ii)

- 80 °C freezing container (iii)

- 80 °C direct (iv)

A B

sc

att

ere

d l

igh

t

[-]

time [h]

-196 °C liquid nitrogen (i)

- 80 °C PS box (ii)

- 80 °C freezing container (iii)

- 80 °C direct (iv)

Results and discussion 46

stationary state cells were used for cryopreservation, which seem to be more robust and bet-

ter suited for cryopreservation strategies.

Taking the time necessary for heterotrophic batch cultivation to reach the stationary phase

after thawing and inoculation, a first viability estimation of the different cryopreservation

methods is possible (Figure 3.2).

Figure 3.2: Comparison of cultivation times necessary to reach stationary phase. Cultures were inoculated from cryocultures derived from different protocols. Error bars deviated from biological replicates (n = 3). Modified from [281].

In general, it could be confirmed that cryopreservation using stationary phase cells is advan-

tageous, since with all four cryopreservation methods (i-iv) such cultures reached stationary

phase earlier than cultures cryopreserved from growing phase cells (p < 0.05). Stationary

phase cells cryopreserved at -80 °C (polystyrene box (ii), freezing container (iii) and direct

(iv)) only required 80 h of batch cultivation time after thawing without apparent differences

(p < 0.05). Stationary phase cells cryopreserved in liquid nitrogen reached the stationary

phase within 110 h of cultivation, which is slower than the three other methods applied

(p < 0.05), but proved higher post-thawing viability compared to equivalent cryocultures pre-

pared from growing phase cells.

0

30

60

90

120

(iv)

-80 °C

direct

(iii)

-80 °C

fr. container

(ii)

-80 °C

PS box

tim

e t

o s

tati

on

ary

ph

as

e [

h]

freezing mode

preservation from growing cells

preservation from stationary cells

n.d.

(i)

-196 °C

liquid nitrogen

Results and discussion 47

In accordance to that, all batch cultivation times for cryocultures prepared from growing

phase cells revealed a delayed growth pattern. Cells inoculated from directly frozen cryocul-

tures (iv) reached the stationary phase after approx. 95 h and thereby only little, but signifi-

cantly later than cells previously frozen using the polystyrene box (ii) or the commercial

freezing container (iii) (p < 0.05). Cryocultures prepared from growing phase cells with liquid

nitrogen showed the lowest post-thawing viability and did not reach the end of the batch

phase under the conditions applied.

For further analysis of the observed relation between growth pattern and method of cryo-

preservation, the cell count and biovolume were measured for each culture before and after

cryopreservation to investigate potential changes in cell count, morphology or shape. To al-

low for direct comparison between the types of cells used, i.e. growing or stationary phase

cells, the initial biomass was set to an optical density of 25 (Figure 3.3).

Figure 3.3: Cell count (blue) and biovolume (red) before and after cryopreservation proce-dures. A: exponentially growing cells, B: stationary phase cells; identical biomass concentra-tion was set to OD750 = 25 before cryopreservation in all experiments. Error bars deviated from biological replicates (n = 3). Modified from [281].

0

2

4

6

8

10

cell

co

un

t [1

08 m

L-1]

pre-cryo

(i)

post-cryo

liquid

nitrogen

(ii)

post-cryo

PS box

(iii)

post-cryo

freezing

container

(iv)

post-cryo

direct

A

(iv)

post-cryo

direct

(iii)

post-cryo

freezing

container

(ii)

post-cryo

PS box

(i)

post-cryo

liquid

nitrogen

pre-cryo

B

0

3

6

9

12

15

bio

vo

lum

e [

L m

L-1]

Results and discussion 48

Although the same initial biomass in terms of optical density was used in the experiments,

the resulting cell count of stationary phase cells was more than 3 times higher compared to

growing phase cells. As both cell types showed similar biovolume, stationary cells must show

a significantly lowered average cell size. Accordingly, it seems that both cell types exhibit

significantly different growth phase dependent optical properties in terms of the measure-

ment of optical density and scattered light.

Most strikingly, with respect to the influence of the four cryopreservation procedures, there

was no significant change of the cell count (p < 0.05) observable. Consequently, the struc-

tural integrity of cell membranes seems to be not affected (Figure 3.3 blue bars) and loss of

cells by disruption seems to be not present. The data for the biovolume showed only minor

changes for all methods using -80 °C for freezing with either growing or stationary phase

cells.

The data for the cells thawed after cryopreservation in liquid nitrogen showed surprising re-

sults. Although the viability was significantly reduced, especially for growing phase cells fro-

zen in liquid nitrogen (Figure 3.1 A) no loss of total cell count was observed as shown in Fig-

ure 3.3 (‘post-cryo liquid nitrogen’). Obviously, the liquid nitrogen freezing seems to reduce

the ability to grow, but remains sufficient cell integrity, so that the cells are detected by the

cell counter.

Nevertheless, the biovolume was differentially affected by the 4 methods (Figure 3.3 red

bars). Plunging either growing, or stationary phase cells into liquid nitrogen caused a signifi-

cant decrease in the biovolume of 30 and 18 %, respectively (p < 0.05). Freezing growing

phase cells at -80 °C using all other methods caused a decrease of 13 % on average,

whereas no significant change was observable for the preservation of stationary phase

C. vulgaris (p < 0.05).

In accordance with the cultivation experiments, two major conclusions can be made. On the

one hand, cryopreservation of stationary phase cells seems favourable compared to using

growing phase cultures. They seem to be less robust to the freezing protocols applied. On

the other hand, from the non-controlled -80 °C freezing protocol similar results were obtained

as for the semi-controlled freezing container or the polystyrene box, which delay the freezing

process. Plunging into liquid nitrogen proved to be the most harmful method for C. vulgaris. It

is assumed that the measured changes in optical properties of growing and stationary phase

cells were most likely caused by morphology changes, leading to improved robustness of

stationary phase cells. Probably this is a result either due to a more rigid cell wall [33], due to

Results and discussion 49

nitrogen starvation induced formation of voluminous intracellular lipid droplets [282] or the

accumulation of compatible solutes like trehalose. Compared to growing phase cells, these

maturated cell walls could have contributed to the enhanced resistance of cells against freez-

ing and thawing stress.

3.1.2 Viability calculation

Since the counting of colony forming units as the best established method for viability deter-

mination is known to be error-prone [279], it was aimed to introduce a novel approach which

is based on the comparative evaluation of online monitored growth patterns which are ob-

tained from MTP cultivations. The underlying idea is that cultures with reduced viability need

more time to reproduce compared to a reference cultivation with 100 % viability. The amount

of additional time necessary is representative to the reduced viability. Cultures to be investi-

gated were used as inoculum just before and after cryopreservation. By means of their opti-

cal density, all cultures were inoculated to an identical level of initial total biomass so that

different fractions of viable cells will result in different total growth times (Figure 3.4).

Figure 3.4: Growth pattern-deviated assessment of cell viability. The 100 % curve corre-sponds to fresh culture before cryopreservation as a reference; prolongation of the post-thawing cultivation by means of the doubling time tD corresponds to 50 % reduction of post-thawing viability. Modified from [281].

In order to increase the robustness against fluctuations in growth rate and the duration of the

lag phase, the cultures were not just compared by their respective times needed to reach the

0 24 48 72 96 120

0

2

4

6

tota

l b

iom

as

s [

a.u

.]

time [h]

100 % viability

50 % viability

25 % viabilitytD

tD

Results and discussion 50

stationary phase. Instead, the apparent number of cell generations from inoculation to the

stationary phase ngen was defined according to formula (3-1):

ngen = tstat-tlag

tD (3-1)

ngen apparent generation number [-]

tstat time to stationary phase [h]

tlag duration of lag phase [h]

tD doubling time [h]

The duration of the lag phase tlag was defined by the time point where the increasing biomass

signal exceeded the initially measured value significantly. As the initial signal fluctuated by

± 2 %, the level of detection (3 times the initial standard deviation) was 6 % of the scattered

light intensity immediately after inoculation. The doubling time tD itself was derived from for-

mula (3-2):

tD = ln (2)

µ (3-2)

Combining the values from cultivations inoculated with cryopreserved cells and cells ob-

tained immediately after harvest, the difference in terms of generation number Δngen was

determined according to formula (3-3):

∆ngen = ngen, post-cryo- ngen, pre-cryo (3-3)

Δngen differential apparent generation number [-]

ngen,post-cryo ngen for cultivation of cryopreserved cells [-]

ngen,pre-cryo

ngen for cultivation of fresh cell material [-]

Comparing growth patterns before and after cryopreservation, each increase of Δngen by one

is defined to represent a halving of the culture’s viability (Figure 3.4). Thus, the viability was

accessed according to formula (3-4):

v = 100 ∙ e-0.693 ∙ ∆ngen

(3-4)

v viability [%]

Results and discussion 51

3.1.3 Optimized protocol and viability estimation

As cryopreservation of stationary phase cells of C. vulgaris using either the PS box method

(ii) or the commercial freezing container (iii) was shown to result in the best post-thawing

growth, this approach was further investigated regarding quantitative viability. Therefore,

heterotrophic cultures were grown in a stirred tank bioreactor. After initial biomass production

on glucose as sole carbon source, nitrogen starvation was applied for about 140 h to prevent

further biomass formation. This phase was intended to give the cells time to fully develop

their cell walls metabolizing the residual glucose. As C. vulgaris is known to replicate via

vegetative autospores [33, 59], the daughter cells lack the rigid cell wall of adult cells, which

was thought to enhance resistance against freezing preservation. Moreover, liposome for-

mation was introduced by nitrogen starvation which could be an additional cause for the en-

hanced freezing robustness of stationary phase cells.

The harvested cells from stationary phase were preserved and their viability was determined

using the viability estimation tool introduced in section 3.1.2. Growth pattern analysis was

conducted to access the viability of the stationary phase cells of C. vulgaris cryopreserved

either using the PS box or the freezing container (Figure 3.5). As initially shown, using the

PS box (ii) or the freezing container (iii) did not result in significant differences in the growth

pattern compared to direct freezing at -80 °C. Therefore, the latter was not considered for

further investigation.

Results and discussion 52

Figure 3.5: Post-thawing viability after cryopreservation of stationary phase cells. Cultures were frozen using the PS box and the freezing container method. Each cryopreservation was conducted from three individual batches (biological replicates), resulting in highly reproduci-ble viabilities. Error bars deviated from multiple determinations of the respective pre- and post-cryo growth patterns (n ≥ 3) according to Gaussian error propagation. Modified from [281].

As revealed by the new viability measurement protocol, both preservation methods resulted

in comparable viabilities of 63 ± 2 %. No significant differences between the biological repli-

cates as well as the applied methods were observable (p < 0.05). Thereby, the results of the

initial comparison of the freezing methods (section 3.1.1) were confirmed.

Referring to post-thawing viabilities formerly determined, the efficiency of this simplified cryo-

preservation method was proven. Some studies available for C. vulgaris show significantly

lower viabilities (35 - 51 %) [254, 283] than achieved during this study. Other authors report

significantly fluctuating values from 18 - 95 % [284-286], applying methods that were sub-

stantially more time and material consuming due to a higher number of experimental steps,

the continuous need of liquid nitrogen for storage and/or required specialized equipment for

sophisticated controlled-rate cooling profiles.

Despite the numerous alternatives for viability estimation available, most studies focussing

on the cryopreservation of C. vulgaris use conventional plating and cell counting techniques.

The relative standard errors of these were, if given at all, in the range of 1.6 - 88 % [254,

283-286], giving, as formerly reported [279], rise to the conclusion that direct comparability

1 2 3

0

25

50

75

100

po

st-

tha

win

g v

iab

ilit

y [

%]

biological replicates

PS box

freezing container

Results and discussion 53

between different studies is difficult. In comparison, the newly introduced growth pattern

analysis approach reproducibly achieved < 5 % of relative standard error, thereby providing

robust and rapid access to well-resolved and valid viability data.

3.1.4 Re-adaptation to phototrophic conditions and pre-cultivation

Mostly, microalgae are cultivated applying phototrophic conditions, while heterotrophic culti-

vation induces the degradation of chlorophyll, strongly reducing the photosynthetic capacity

of the cells [287]. Thus, the cryopreserved cultures originating from heterotrophic cultivation

conditions need to be re-adapted to light for phototrophic cultivation. It was investigated how

much cultivation time is required to provide inoculum with full phototrophic adaptation (Fig-

ure 3.6).

Figure 3.6: Phototrophic pre-cultivation of C. vulgaris in shake flasks. Heterotrophic cryocul-tures from the optimized PS box method were used as an inoculum. During an initial phase of growth repression the heterotrophic cells rapidly adapted to phototrophic conditions (dashed line) and showed exponential growth at a rate of 1.63 ± 0.05 d-1 (solid line); enBBMref, 25 °C, shaking frequency 200 rpm, VL = 150 mL, 200 µmol m-² s-1 PAR, gassing rate 9 NL h-1 (2.5 % (v v-1) CO2). Error bars deviated from biological replicates (n = 3).

After inoculation, a lag phase of approx. one day was observed. It was assumed that during

this period, a major part of the chlorophyll was synthesized and the chloroplasts were regen-

erated. This so-called “greening” process is usually coupled to growth depression [288] as

observed for the first day of cultivation. Although there is evidence that complete greening of

C. vulgaris takes around 35 h [289], it seemed that sufficient regeneration of the photosyn-

0 24 48 72 96

0

1

2

3

4

OD

75

0 [

-]

time [h]

Results and discussion 54

thetic capacity occurred during the first day of cultivation. From this point on, exponential

growth at a rate of 1.63 ± 0.05 d-1 was observed which is in the range of published growth

data for non-limited C. vulgaris [290]. With an optical density of approx. 2, the biomass con-

centration achieved after 3 days was sufficient to serve as an inoculum for possible further

phototrophic experiments.

3.1.5 Conclusion

In section 3.1, an easy-to-use protocol for the cryopreservation of C. vulgaris is introduced.

Heterotrophic cultivation was used to generate high amounts of biomass. Applying subse-

quent nitrogen starvation, the cells were given time to fully develop their robust morphology,

most likely either a rigid cell wall or the accumulation of intracellular lipid droplets as protec-

tion against the following freezing preservation stress. The cells were frozen to and stored

at -80 °C, slowing down cooling by either using a commercial freezing container or a simple

polystyrene box achieving comparable viabilities in the range of 63 ± 2 %. Thus, post-

thawing viabilities so far only reported for protocols being rather complex and/or incorporat-

ing specialized cooling devices were achieved applying a much simpler approach.

In contrast to the well-established strain maintenance concept via serial sub-culturing of ac-

tively growing cultures under suboptimal conditions, the presented methodology states out

the potential of simplifying microalgal cryopreservation protocols which reduce the labour and

material need to a minimum. In subsequent phototrophic cultivations, the cells were rapidly

re-adapted to light and reproducibly delivered sufficient amounts of biomass to inoculate me-

dium-sized laboratory-scale photobioreactors.

Besides this highly time and material efficient strain maintenance approach, a new method

for viability estimation of cryopreserved cells was introduced. This method uses growth pat-

tern analysis comparing pre- and post-preservation growth to access the cultures’ viability.

By the use of growth data deviated from online monitoring, error-prone offline evaluation

methods can become superfluous. Comparing to the well-established plating and counting

techniques for viability estimation, significantly lower experimental errors (< 5 %) were

achieved.

Overall, the developed workflow can be routinely applied to efficiently generate vital

C. vulgaris biomass from genetically stable stock cultures as pre-cultures for further photo-

trophic cultivation experiments. Time and material efficiency as well as reliability support the

demands on strain maintenance in an accelerated bioprocess development framework.

Results and discussion 55

3.2 Parallelized phototrophic microscale cultivation

With regard to the targeted industrial exploitation of microalgae, besides reactor and value

chain engineering, especially advances concerning strain construction and screening as well

as process parameter optimization are crucial [24, 25]. These aspects are typically studied

during countless laboratory-scale experiments, as many factors (strain, medium composition,

process parameters, light characteristics, etc.), dynamic changes (e.g. day/night cycles) and

potential interactions need to be considered. In contrast to accelerated microbial process

development [161, 168, 169, 177, 178, 291], commercially available high throughput photo-

bioreactors are lacking at present. As a result, experimental throughput is fairly limited with

scientists being restricted to single vessel photobioreactors, shake flasks and test tubes [26-

29].

A demand for high throughput photobioreactors has recently been identified upon which

some prototype systems emerged. These rely on 24 to 96-well MTPs, offering significant

advances regarding experimental throughput. However, compared to microbial microbioreac-

tors, the systems so far introduced suffer from at least one of several drawbacks. Mostly, the

homogeneity of light and process conditions and thereby cultivation reproducibility across the

discrete wells of an MTP represents a crucial issue. There is either a restriction to use trans-

parent plates allowing for interference by extraneous light and well-to-well cross-talk [166,

167] or suboptimal properties of the illumination unit [292]. In this context, inhomogeneous

illumination patterns were described to display problems. Moreover, growth phenotypes, like-

ly induced by ventilation or temperature effects within the incubation chamber [293] or une-

ven CO2 mass transfer into the wells [294], were observed. With regard to extended cultiva-

tion times, evaporation control needs to be evaluated carefully, as high liquid loss may distort

cultures’ performance [295, 296]. Many of the systems presented use top side illumination of

the MTP potentially impeding the use of appropriate sterile barriers ensuring monoseptic

conditions and restricting the accessibility with respect to integration into laboratory robotic

platforms. Heo et al. [297] recently presented a newly designed low cost high throughput

photobioreactor system, introducing the simultaneous variation of light intensity and tempera-

ture across a single MTP. Thereby, an additional boost of throughput was realized increasing

the number of parameters variable at well-to-well level. However, these proceedings were at

the expense of mixing performance which has to be considered problematic with respect to

cell sedimentation, light/dark cycles and mass transfer.

In terms of process monitoring and control, sophisticated concepts for automated at-line ana-

lytics and integrated screening workflows became available recently [166, 167]. Though,

state-of-the-art non-invasive online analytics remain exclusively available in microbial micro-

Results and discussion 56

bioreactors [215, 298]. Sampling-based at-line methods are typically prone to interfere with

the cultivation process itself and can only provide sparse data.

In the context of a cooperation project between Forschungszentrum Jülich GmbH, m2p-labs

GmbH and Premosys GmbH, a commercial microbioreactor system, the BioLector® II

(m2p-labs, Baesweiler/Germany), was used as the technical basis to develop a parallelized

micro-photobioreactor. The basic device is designed to run up to 48 individual microliter

scale fermentations in an MTP in parallel while conditions (temperature, gas atmosphere,

humidity, shaking) are strictly controlled and cultivation processes can be monitored online.

Therefore, non-invasive optical techniques are applied during continuous shaking providing

high data density without laborious and potentially interfering sampling [212, 215].

This well-established system [168, 169, 225, 299] was extended by an LED-based illumina-

tion unit, the “photo module”. The module was designed to enable defined illumination of

MTP cultures from the bottom-side for the following reasons:

(i) Irradiation through the clear MTP bottom creates a defined light influx area inde-

pendent from the liquid geometry within the wells which itself is strongly influ-

enced by the shaking speed.

(ii) Standard sealing foils can be used as sterile barrier ensuring monoseptic condi-

tions.

(iii) The MTP remains accessible from the top-side by external devices facilitating the

integration into a liquid handling platform for process automation.

The core of the microbioreactor system consists of two chambers separated by a shaking

tray carrying the MTP. The upper chamber incorporates temperature, atmosphere and hu-

midity control while the lower one is equipped with an optical measuring system mounted to

an x-y-displacing unit for monitoring of the individual wells (section 3.2.2 provides further de-

tails). The photo module was integrated below the lower chamber for bottom side illumination

(Figure 3.7).

Results and discussion 57

Figure 3.7: Design of the micro-photobioreactor prototype. A 48 well MTP (i) is mounted on an orbital shaker tray (ii) housed by two separated chambers. The upper chamber (iii) sur-rounding the MTP incorporates temperature, atmosphere and humidity control of the system. The lower chamber (iv) carries a measuring device for optical process monitoring mounted on an x-y-displacing unit (v). The photo module (vi) is integrated at the bottom of the lower chamber and consists of a box carrying different LEDs (vii) attached to the inner walls. The respective angles are designed in a way that emitted light does not directly leave the pho-to module but is scattered in the inside. Scattering is facilitated by a BaSO4 coating (viii). The aperture of the module carries an additional diffusor array (ix) to improve light homogeneity. Modified from [300].

The photo module itself consists of an aluminium-made box with a top side aperture. The

side walls carry 20 individually controllable circuit boards equipped with 120 LEDs of three

different types in total. By tilting these LED-carrying walls, no directly emitted light is allowed

to leave the module through the aperture. Rather, light is scattered in the inside, while this

effect is additionally supported by a BaSO4-based coating. Thereby, spectral mixing of the

different LEDs is ensured. The aperture is equipped with a custom-made diffusor array facili-

tating a homogeneous light efflux towards the MTP.

At the maximal electric input of 250 W, the setup was able to generate 620 µmol m-2 s-1 PAR.

Assuming a light conversion efficiency of 20 % [301], up to 200 W of thermal energy need to

be dissipated. However, besides lifetime [302], the emission spectra and luminous efficien-

cies [303] of LEDs are influenced by temperature, the photo module needs to be run at low

and especially constant temperature. Therefore, circumferential cooling channels were inte-

grated into the sidewalls of the module, allowing to maintain the module at 25 °C using a

2219 Multitemp II Thermostatic Circulator.

Besides absolute intensity, the spectrum of a light source, either natural or artificial, may

strongly impact photosynthetic performance, as distinct species carry different pigments each

having unique absorption capabilities [144, 304]. The presented system was particularly de-

Results and discussion 58

signed for the cultivation of green microalgae, turning chlorophyll a and b into the relevant

pigments. Within the PAR spectrum, these are known to strongly absorb blue and red light,

whereas photons with wavelengths in the intermediate green colour range are hardly ab-

sorbed. By means of relative absorption across the PAR spectrum, approx. 85 % of the pig-

ments’ absorption capacity lies within the intervals of 400 - 500 nm and 630 - 680 nm (Fig-

ure 3.8 A). These intervals were regarded to be essential for photosynthetic performance of

green microalgae and thus need to be part of the emission spectrum in an appropriate man-

ner.

To suffice this demand, the module was equipped with three different LED types:

warm-white, near-UV and blue (emission spectra of the individual LEDs are provided in Fig-

ure 6.2, appendix).

For adjustment of the spectrum emitted by the photo module, the LEDs were grouped on the

aforementioned circuit boards. 14 of the 20 boards were carrying 6 warm-white LEDs each,

while each of the 6 remaining boards was equipped with 2 near-UV and 4 blue LEDs. The

spectrum of a mid-European summer afternoon was taken as a basis, while both intervals

identified to be relevant by the pigment absorption measurement were focused. Resulting in

a ratio of 1:0.65, the relative power of near-UV/blue and white LEDs was adjusted in a way

that the relative photon flux densities of the 400 - 500 nm and the 630 - 680 nm intervals

matched the fractions of the natural daylight as closely as possible (Figure 3.8 B). With re-

spect to the PAR interval, 35.1 % and 14.8 % of the photon flux of the natural daylight were

within the aforementioned intervals, while the distribution was 43.1 % and 18.4 % for the

photo module spectrum. Although the blue fraction of the artificial light is somewhat

overrepresented, daylight was approximated as good as possible taking into account that

only three different types of LEDs were used. Thus, a photosynthetic performance at least

comparable to natural light conditions could be expected for the cultivation of green microal-

gae. However, the light spectrum can easily be modified towards other needs by adjusting

the power ratio between the white and the near-UV/blue LEDs or by equipping the pho-

to module with an alternative LED setup.

Results and discussion 59

Figure 3.8: Tailoring of the photo module spectrum. A: Absorption spectra of chlorophyll a and b, the major photosynthetic pigments of green algae. With respect to the PAR interval, 85 % of photon absorption occurs within the intervals of 400 - 500 nm and 630 - 680 nm; spectra of 10 mg L-1 chlorophyll a and b in dimethyl sulfoxide. B: The flux densities within the relevant intervals were optimized towards natural daylight by adjustment of the power ratio between the white and the blue LEDs of the photo module; illumination intensities by means of relative photon flux densities normalized to the PAR interval. The daylight spectrum was measured at a typical, slightly clouded, mid-European summer afternoon (date: 23.07.2014, coordinates: 50° 20’ 26.46’’ N, 6° 40’ 11.55‘ E). Modified from [300].

3.2.1 Non-biological validation

Homogeneous process conditions across all wells are essential for the reliability of MTP-

based MBRs. Only if these are guaranteed, the discrete wells can be regarded as distinct

bioreactors. Thus, the system was comprehensively checked for its ability to homogeneously

maintain the relevant process parameters.

Illumination 3.2.1.1

Light intensity is considered as one of the most important parameters affecting phototrophic

cultivation. The spatially resolved intensity distribution was monitored by means of relative

PAR (Figure 3.9).

0.0

0.3

0.6

0.9

400 500 600 700

0.0

0.2

0.4

0.6

rel.

ab

so

rpti

on

[

%]

chlorophyll a

chlorophyll bA

rel.

ph

oto

n f

lux

[

%]

wavelength [nm]

daylight

photo moduleB

0.0

Results and discussion 60

Figure 3.9: The photo module ensures a homogeneous photon flux density distribution. Av-erage intensity fluctuation across all 48 wells of the incubated MTP was ± 2.5 %. Light inten-sities represented by relative PAR. Modified from [300].

The illumination intensity was observed to be almost constant across a wide part of the MTP

(rows B – E). In the rows A and F, light intensities were slightly lower with a maximum devia-

tion of 5 %. Average fluctuation of relative PAR was ± 2.5 % across the whole MTP. Under

these conditions, no light intensity induced growth phenotypes were expected for identical

cultures distributed across the individual wells which need to be validated in parallel cultiva-

tion experiments.

Temperature control 3.2.1.2

Temperature control of the MTP was carried out indirectly by temperature control in the up-

per and lower incubation chamber (section 3.2). However, during phototrophic cultivation,

illumination results in additional energy influx which may affect the temperature inside the

wells. Checking these at standard cultivation conditions, temperature was found to be

25.0 ± 0.1 °C (n = 48) and thereby exactly fitting the desired setpoint while no inhomogeneity

like corner effects was observed. Consequently, the micro-photobioreactor is capable of

maintaining the applied process temperature across the complete MTP in the incubation

chamber compensating for the additional energy input by illumination.

Evaporation 3.2.1.3

Extended cultivation times as well as energy influx by illumination turn liquid evaporation into

a crucial issue [305]. Besides osmotic effects [295], volume loss influences the optical path

length distribution inside the wells by decreasing filling height. Thereby, the light regimes are

potentially altered. The evaporation rate at standard cultivation conditions was found to be

0.56 ± 0.04 % d-1 by gravimetric determination. Under these conditions, liquid loss can be

neglected, enabling long term phototrophic cultivation. For example, 10 days of cultivation at

Results and discussion 61

1 mL would result in a liquid loss of only 56 µL (i.e. 5.6 %) which is regarded as non-

interfering.

3.2.2 Online measurement of biomass

Optical online analytics as established for microbial microbioreactors offer elevated data

density and can be non-invasive by design. Thereby, deeper process knowledge and real

time evaluation are facilitated. Here, the optical analysis unit of the microbioreactor system

was used for biomass quantification via scattered light. A detailed description of the meas-

urement principle and its technical realization is given by Samorski et al. [212] and Kensy et

al. [215]. Using an appropriate combination of excitation LED and filters, measurements were

carried out at a wavelength of 750 nm, avoiding interference by algal pigments [217].

Figure 3.10: Biomass quantification by scattered light. A: Within the concentration range typically relevant for phototrophic cultivation, scattered light is linearly correlated to biomass by means of optical density; biomass: dilution series of C. vulgaris from exponentially grow-ing shake flask cultures resuspended in 0.9 % (w v-1) NaCl. B: Illumination of the MTP during cultivation does not interfere with turbidity monitoring and thus does not need to be switched off for measuring; measurements taken with AMCO Clear turbidity standard; 25 °C, 2.5 % (v v-1) CO2, 0 or 200 µmol m-2 s-1 PAR, shaking frequency 1200 rpm, VL = 1 mL. Error bars deviated from technical replicates (n = 3). Modified from [300].

The suitability of the system for algal process monitoring was investigated by correlating

scattered light measurements of a C. vulgaris dilution series against optical density (Fig-

0 5 10 15 20 25

40

50

60

70

80

0 50 100 150 200 250

0

50

100

150

200

250

sca

ttere

d l

igh

t

[-]

OD750

[-]

As

ca

ttere

d l

igh

t ph

oto

mo

du

le O

N

[-]

scattered lightphoto module OFF

[-]

B

Results and discussion 62

ure 3.10 A). Scattered light was linearly correlated (R2 = 0.996) to biomass concentration to

at least OD750 = 20 while the level of detection was found to be ≤ OD750 = 0.4. By means of

technical triplicates, an average error of 1.7 % was determined for manual OD measurement,

whereas scattered light read-outs achieved a significantly reduced (p < 0.05) fluctuation of

only 0.9 %, proving the accuracy of the system. Moreover, the average fluctuation between

analytical replicates (≥ 5 measurement cycles per technical replicate) was at the level of

0.2 %, confirming the high reliability of the system.

The MTP is intensively illuminated to drive photosynthesis. Hence, the influence of illumina-

tion on optical analysis was characterized recording the scattered light signal of a dilution

series of an AMCO Clear turbidity standard (GFS chemicals, Columbus/US) either in the

dark or with illumination at 200 µmol m-2 s-1 (Figure 3.10 B). By direct comparison and linear

regression, a slope of 1.00 was determined. Thus, scattered light measurement is not affect-

ed by illumination, despite the photo module-derived light being orders of magnitude more

intensive than the light signals for optical analysis. This was realized by a sinusoidal modula-

tion of the excitation light intensity at a specific frequency. Like this, the light being scattered

back from the probes could be differentiated from the non-modulated light of the pho-

to module using a Fourier transformation.

As a result, the integrated measuring technique proved its suitability for biomass monitoring

during phototrophic cultivation in MTPs. Within the range typically relevant for microalgal cul-

tivation, the scattered light was linearly correlated to biomass by means of OD750 while a very

high accuracy and reproducibility were achieved. Phase-modulation enabled measuring with

the photo module being switched on. To facilitate the sequential readout of the single wells of

a plate, the measuring unit incorporates an optical fibre bundle displaceable below the MTP

by an x-y-displacement unit (section 3.2). Between measuring cycles, the fibre bundle is

moved to a special parking position outside the cone of light avoiding shading effects. How-

ever, the partial shading of the MTP experienced during the measurement cycle of 2.5 min

needs to be taken into account. The frequency of measurements was adjusted to the slow

algal growth rates. At a periodicity of 3 h, partial shading occurs during just 1.4 % of the pro-

cess time. Under these conditions, the system can be considered as non-invasive.

3.2.3 Biological validation

Based on the comprehensive evaluation given in section 3.2.1, the micro-photobioreactor

system ensures homogeneous and strictly controlled process conditions across the distinct

wells of an MTP. Additionally, biomass can be precisely monitored in a non-invasive manner

(section 3.2.2). For a subsequent biological validation, it was evaluated if homogeneous

Results and discussion 63

growth could be achieved and monitored across the wells of an MTP under these conditions

(Figure 3.11).

Figure 3.11: Phototrophic cultivation at MTP-scale. Non-invasive growth monitoring via scat-tered light reveals homogeneous growth across the individual wells of the MTP while the typ-ical phases of a phototrophic process could be observed; enBBMref, 25 °C, 2.5 % (v v-1) CO2, 200 µmol m-2 s-1 PAR, shaking frequency 1200 rpm, VL = 1 mL, relative humidity ≥ 85 %. Shaded area represents confidence interval deviated from biological replicates (n = 48). Modified from [300].

During cultivation, the typical course of a phototrophic batch process was observed (blue

line). After an initial lag phase, the cells grew exponentially at a rate of µ = 1.42 ± 0.13 d-1 (fit

from 30 – 58 h), hereby reinforcing growth rates already published for non-limited C. vulgaris

[290]. However, the exact duration of the lag phase could not be assessed due to the weak

signal of the scattered light measurement within the first 30 h (section 3.3), rendering it diffi-

cult to decide when exactly the start of the exponential phase took place. After reaching the

scattered light intensity of 6.3 ± 0.3 (at 58 h), the culture shifted from exponential to linear

growth. At this level of biomass (approx. 5 by means of OD750), self-shading of the cells

started to become a growth limiting factor. This comparably long exponential phase and the

fast biomass increase during the subsequent light-limited growth phase prove the excellent

light supply of the cells within the culture wells. Most probably, this is attributable to the high

surface to volume ratio of 121 m-1 as well as to the short mixing times ensuring short

light/dark-cycles induced by rigorous orbital shaking at 1200 rpm and 3 mm shaking diame-

0 24 48 72 96 120 144 168

0

10

20

30

40stationary

phase

sc

att

ere

d lig

ht

[-

]

time [h]

lag & exp.

growth phase

light limited

growth phase

Results and discussion 64

ter. After approx. 114 h, the increase in scattered light intensity significantly slowed down

representing the beginning of the stationary phase most likely induced by nitrogen starvation.

Regarding growth homogeneity, the average standard deviation of scattered light measure-

ments was as little as ± 5.2 % during the whole cultivation process while no patterns were

observed across the MTP (Figure 6.3, appendix). Moreover, offline measurements at the end

of the cultivation revealed fluctuation in biomass concentration of ± 5.1 % by means of OD750.

This gives additional proof for the reliability of the scattered light assisted quantification of

biomass and the minor well-to-well fluctuations between the discrete cultivations. As a con-

sequence, the individual wells may ultimately be regarded as distinct and independent micro-

photobioreactors. Therefore, they can be used for parallelized screening of different strains

or medium compositions, as already established for heterotrophic systems [156, 165]. An-

other valuable opportunity is the cultivation of numerous replicates applying a “harvest well

concept”. In this context, cross-well comparability enables the, preferably automated, har-

vesting of replicate cultivations for time-resolved offline analysis with high data density [168,

169]. Moreover, the application of advanced experimental design strategies for accelerated

bioprocess optimization as recently introduced by Freier et al. [175] is in range for photo-

trophic processes.

3.2.4 Conclusion

Throughout section 3.2, the design and realization of a new, parallelized micro-photobio-

reactor is presented. Therefore, a specially designed, modular light source was integrated

into a commercial microbioreactor system, the BioLector® II. In this context, the adjustable

light spectrum was tailored towards the cultivation of green microalgae offering a maximal

PAR of 620 µmol m-2 s-1.

The system was designed to run up to 48 simultaneous phototrophic cultivations at MTP-

scale. Strict control of the process conditions enabled a high comparability between the dis-

tinct wells of one plate (± 5.2 % fluctuation in biomass formation). Consequently, the single

wells could be handled as discrete micro-photobioreactors. The small scale, resulting in a

beneficial surface to volume ratio as well as the fast mixing due to rigorous orbital shaking,

ensured an excellent light supply of the cultures.

Non-invasive quasi-online quantification of biomass was realized by phase-modulated meas-

urements of scattered light. Thereby, single well-resolved biomass monitoring at high data

density is facilitated without any additional manual interaction or interference with the pro-

Results and discussion 65

cess while the dynamic range covered the concentrations typically occurring during photo-

trophic batch cultivation of microalgae.

The system was shown to be especially qualified for parallelized laboratory screening appli-

cations like media optimization for instance. Moreover, it can be applied to display processes

using extensive offline analysis by a “harvest well concept”. Easy automation via integration

into a liquid handling platform is given by design. Thereby, the presented micro-photobio-

reactor system significantly contributes to increasing the time efficiency during the develop-

ment of phototrophic bioprocesses.

3.3 Scalability of phototrophic microscale cultivation

Besides the usual process parameters of microbial cultivation, i.e. oxygen supply, medium

composition, temperature, etc., the performance of phototrophic processes is influenced by

light supply and its dynamics. This directly relates to the supply of metabolism with reduction

equivalents and energy metabolites derived from photons captured during the light reaction

of photosynthesis [137]. Cells in the culture suspension absorb and scatter incident light,

however, creating a light intensity gradient along the optical path length of a photobioreactor

[144]. As absorption and scattering are functions of particle size and form, but especially

concentration [306], this gradient usually increases drastically over time as biomass is

formed during algal growth. Although specialized reactor concepts have been developed to

tackle this challenge [307], light availability can never be kept completely constant across the

whole reactor and gradually changing light and dark zones emerge across the liquid volume.

These regions are passed by the cells with a frequency that is determined by the mixing

properties of the reactor as well as size and location of the respective zones [308]. Conse-

quently, the actual photosynthetic efficiency of a cell depends on its distinct position within

the culture volume [150] and thus, the design of photobioreactors has a tremendous influ-

ence on the cultivation process itself [91]. Large-scale outdoor reactors are being designed

to meet the general needs of the strain to be cultivated, however local conditions in specific

microenvironments should also be considered [309]. This tailored design approach can now-

adays be assisted by computational fluid dynamics [143, 310, 311]. On the contrary, labora-

tory-scale photobioreactors are mostly designed rather by empirical rules. Even a reduction

of scale does not necessarily mean an enhanced mechanistic knowledge about the predomi-

nant mixing and light supply properties of the respective reactor.

By now, a variety of either empiric or mechanistic models for the quantitative description and

simulation of photosynthetic processes is available. As comprehensively reviewed elsewhere

[138, 312], these can be categorized into three types (“first to third level”) with steadily in-

Results and discussion 66

creasing complexity. Models are categorized by a hierarchy of increasingly detailed descrip-

tions. As first level models express the whole culture’s photosynthesis rate either in depend-

ency of the incident or average light intensity, they are rather simple and do not consider the

light regimes in a reactor. These light gradients can be taken into account to calculate local

photosynthetic rates in second level models. However, for reliable description of photobiore-

actors many more parameters need to be taken into account. Therefore, third level models

integrate particle trajectories on the single cell level at different levels of detail. These are

ideally assessed by computational fluid dynamics, and thus require enormous computational

power. To reflect reality as closely as possible, the light-acclimation state and the overall cel-

lular metabolism would have to be simulated as well.

MTPs, as being used for the micro-photobioreactor system described in section 3.2 signifi-

cantly differ from established laboratory-scale photobioreactors regarding geometry and mix-

ing characteristics. However, only very little is known about the liquid moving and thus optical

path length distribution as well as particle trajectories under shaking conditions. Consequent-

ly, a system comparison is preferably to be conducted by means of standardized cultivations.

Thereby, all relevant process parameters (strain, medium composition, pH, temperature, il-

lumination, process time, pre-cultivation and inoculation, sampling cycles and analytics) can

be kept constant, so that the reactor performance becomes accessible by the biological per-

formance of the individual processes. To comprehensively cover the spectrum of laboratory-

scale photobioreactors, the following systems were chosen for referencing studies:

(i) illuminated 500 mL shake flasks (section 2.7.1) as a blueprint for the perhaps most

simple photobioreactor that can be set up with minimal invest and thus exhibits the

widest spread to date,

(ii) the eightfold parallel MC 1000 test tube cultivation system (section 2.7.2) being the

smallest screening system for phototrophic processes available on the market, and

(iii) the FMT 150/1000 (section 2.7.3), a well-established 1 L scale photobioreactor in

classical flat panel design.

Results and discussion 67

Table 3.1: Key design aspects of the photobioreactors used. * path length distribution

MTP shake

flask MC 1000

FMT

150/1000

type MTP well shake flask test tube flat panel

light source LED white & blue fluorescent

lamps LED white

LED blue &

red

mode of mixing orbital shaking orbital shaking bubbling bubbling &

stirring

mass transfer surface aeration surface aeration bubbling bubbling

applied volume

[mL] 1 150 80 1000

optical pathlength

[mm] 2 – 25* approx. 15 - 55* 0 – 25.4* 40

specific illumina-

tion area [m-1] 121 51 75 20

As far as realizable by design of the different systems, identical conditions were applied for

the reference processes. Thus, different process courses could be traced back to the specific

reactor design. Although the MTP micro-photobioreactor provides integrated biomass moni-

toring via non-invasive scattered light measurement, offline OD750 and biovolume from har-

vested wells were used for biomass monitoring instead to ensure maximal comparability.

Figure 3.12 gives a direct comparison of the individual process courses.

Results and discussion 68

Figure 3.12: Comparison of the reference process in different photobioreactor systems. A: MTP, B: shake flask, C: MC 1000 test tube system, D: FMT 150/1000 flat panel reactor; enBBMref, 25 °C, 2.5 % (v v-1) CO2, 200 µmol m-2 s-1 PAR. Error bars represent min/max from biological replicates (n = 2). Modified from [313].

In general, the reference processes matched the theoretical course of phototrophic batch

fermentations typically being characterized by a sequence of lag phase, exponential and

light-limited growth followed by a production phase (section 1.4). Thus, the individual courses

are not separately discussed in detail. Instead, similarities and differences between the in-

vestigated systems are emphasised with the aid of characteristic performance indicators of

the individual process phases.

3.3.1 Lag phase

The reference cultivations were inoculated from standard phototrophic pre-cultures as estab-

lished in section 3.1.4. The corresponding change of the cultivation system including a

change of the light source, as well as biomass preparation by washing and buffer exchange

prior to inoculation may thus lead to adaptation processes inducing a lag phase.

0 48 96 144 192 240

0

5

10

15

20

25

0 48 96 144 192 240

0

5

10

15

20

25

0 48 96 144 192 240

0

5

10

15

20

25

0 48 96 144 192 240

0

5

10

15

20

25

OD

750 [-

], b

iovo

lum

e [

L m

L-1],

neu

tral lip

id c

on

ten

t [%

w w

-1]

time [h]

A

0.0

0.3

0.6

0.9

1.2

NO

3-

[g

L-1]

OD

750 [-

], b

iovo

lum

e [

L m

L-1],

neu

tral lip

id c

on

ten

t [%

w w

-1]

time [h]

B

0.0

0.3

0.6

0.9

1.2

NO

3-

[g

L-1]

OD

750 [-

], b

iovo

lum

e [

L m

L-1],

neu

tral lip

id c

on

ten

t [%

w w

-1]

time [h]

C

0.0

0.3

0.6

0.9

1.2

NO

3-

[g

L-1]

OD

750 [-

], b

iovo

lum

e [

L m

L-1],

neu

tral lip

id c

on

ten

t [%

w w

-1]

time [h]

D

0.0

0.3

0.6

0.9

1.2

NO

3-

[g

L-1]

Results and discussion 69

For the MTP, the shake flask and the MC 1000, the lag phase was approx. 12 h. In contrast,

the FMT 150/1000 culture showed no visible delay and grew exponentially from inoculation.

Likely, the lag phase observed for the first three systems was due to extensive adaptation

processes necessary when changing the cultivation system including a change of the light

source and a drastic alteration of the mixing characteristics (except for the shake flask) or by

time delay during inoculum preparation. Especially the change of the light source may lead to

a misbalance of both photosystems that may have to be compensated [314, 315]. In case of

the flat panel reactor, it seemed that the cells were already sufficiently adapted to enable

immediate exponential growth, which may be due to the specific excitation of the photosys-

tems by the blue and red LEDs of its illumination panel. However, with respect to the total

cultivation time of 10 days, the observed lag phases at the beginning of the MTP as well as

the shake flask and the MC 1000 cultivation make up only 5 % of total cultivation time.

3.3.2 Exponential growth phase

After the initial lag phase for acclimation (section 3.3.1), exponential growth was observed in

all of the four different systems.

Figure 3.13: Exponential growth during the reference processes. A: growth rates, B: influ-ence of the specific illumination area on the light limiting optical density. Error bars represent min/max from biological replicates (n = 2). Modified from [313].

0 50 100 150

0

2

4

6

MTP shake flask MC 1000 FMT 150/1000

0.0

0.5

1.0

1.5

2.0

[

d-1]

reactor system

A

lig

ht

lim

itin

g O

D750

[-]

specific illumination area [m-1]

MTP

MC 1000

shake flask

FMT150/1000

B

Results and discussion 70

The growth rate of 1.44 ± 0.02 d-1 achieved during MTP cultivation was in exact agreement to

those obtained from the shake flask and the MC 1000 system and in the same range as al-

ready reported for non-limited growing C. vulgaris [290]. Only growth in the FMT 150/1000

flatpanel system (µ = 1.34 ± 0.07 d-1) was slightly, but not significantly (p < 0.05) slower (Fig-

ure 3.13 A).

This observation may likely be attributed to the distinct emission spectra of the light sources

of the different systems. While the comparable systems MTP, shake flask and MC 1000 are

characterized by a wide coverage of the PAR interval (direct comparison provided in Fig-

ure 6.4, appendix), the FMT 150/1000 with its setup containing narrow band blue and red

LEDs achieves only little coverage of the 400 - 500 nm section. Yet, this part of the spectrum

has been identified to be important for photosynthesis as previously shown by chlorophyll

absorption measurements (Figure 3.8). However, longer cycling times between light and dark

zones in this reactor may additionally be involved. The comparison of growth rates clearly

shows that the design of the MTP-based micro-photobioreactor is suitable for the cultivation

of green microalgae delivering results comparable to established light sources.

Contrary to growth rates, the respective duration of the exponential phases (determined by

the endpoint of the exponential fit for the respective non-limited growth phases) drastically

differed across the four systems (Figure 3.13 B). The light-limited growth started at a bio-

mass concentration of 0.77 ± 0.06 by means of OD750 for the FMT 150/1000 and increased

via 3.22 ± 0.18 (shake flask) and 4.52 ± 0.18 (MC 1000) up to 5.74 ± 0.02 for the MTP sys-

tem. Moreover, the level of biomass resulting in light limitation was correlated to the specific

illumination area of the respective cultivation system.

Thus, cultivation using the MTP-based system enabled prolonging the exponential growth

phase compared to the other systems due to an improved light supply of the cells. Especial-

ly, the delivery of light in the FMT 150/1000 system was much worse under the applied con-

ditions. By design, this reactor suffers from a small specific illumination area and comparably

poor mixing. Consequently, the small scale of the MTP cultivation, providing short optical

path lengths is advantageous for the cultivation of phototrophic systems. For the conditions

selected (1200 rpm shaking frequency), the optical path length was in between 2 mm and

25 mm depending on the position in the well (Figure 6.5, appendix for a visualization of the

influence of the shaking frequency on liquid geometry). This effect may be supported by the

short dark/light-cycles realized by rigorous shaking. In this context, Weiss et al. [316] indicat-

ed mixing times to be in the low seconds range for non-baffled 96-well MTPs, thus being

much shorter than in the other systems. Especially during laboratory experiments, this fea-

Results and discussion 71

ture is of substantial benefit as the superior light supply extends the time period during which

cellular processes can be investigated under completely non-limiting conditions. However,

especially the performance of the MC 1000 system being higher than that of the shake flask

process was surprising. The mixing in the test tubes (MC 1000) is solely driven by gassing

while the shaking flasks were operated at 200 rpm and a shaking diameter of 25 mm. Never-

theless, these improved mixing properties could not compensate for the reduced specific

illumination area of -32 % (Figure 3.13 B).

3.3.3 Light-limited growth phase

Passing the biomass threshold discussed in section 3.3.2, overall light supply of the cells

becomes rate limiting for photosynthesis. This typically results in a shift from exponential to

linear growth kinetics [137].

Figure 3.14: Light-limited growth during the reference processes. A: biomass increase by means of optical density (time scale normalized to beginning of the light limiting growth phase), B: influence of the specific illumination area on the biomass formation rate. Error bars represent min/max from biological replicates (n = 2). Modified from [313].

All of the four cultivations were characterized by a linear increase in biomass concentration

during that phase of the process (Figure 3.14 A). In analogy to the duration of the exponen-

tial growth phases, the respective rate of biomass increase was lowest for the FMT 150/1000

(0.56 ± 0.01 OD750 d-1) and increased via 2.92 ± 0.04 OD750 d

-1 (shake flask) to

0 12 24 36 48 60

0

3

6

9

12

0 50 100 150

0

2

4

6 MTP

shake flask

MC 1000

FMT 150/1000

OD

750

[-]

normalized time [h]

A

MTP

MC 1000

shake flask

O

D750 i

n l

igh

t-lim

ited

gro

wth

ph

ase

[

d-1]

specific illumination area [m-1]

B

FMT 150/1000

Results and discussion 72

4.44 ± 0.11 OD750 d-1 (MC 1000) and 4.05 ± 0.33 OD750 d

-1 in the case of the MTP system

(Figure 3.14 B).

Hence, the respective linear rates of biomass increase reflected the light supply to the cells

with the similar pattern already observed during the exponential growth phase. Nevertheless,

during light limitation, the cells undergo a steady restructuration of the photosynthetic appa-

ratus to optimize their photosynthetic capacity to the declining light supply. Thus, the devia-

tion of physiological and metabolic properties of the cells by data from this phase has to be

considered critically for all of the four systems investigated. This underlines the importance of

optimal light supply during laboratory-scale experiments as already mentioned in sec-

tion 3.3.2.

3.3.4 Production phase

The stagnation of biomass formation from extracellular nutrients (here: nitrate) marks the

beginning of the production phase. For the model cultivation process, this event is coupled to

the initiation of neutral lipid accumulation. During this phase, the clearest differences be-

tween the four systems were observed.

During the production phase, biomass by means of biovolume increased by 39 % (shake

flask) and 52 % (MC 1000) while it almost doubled in the case of MTP cultivation (98 %).

However, biomass did not change during the stationary phase of the FMT 150/1000 cultiva-

tion. Despite optical signals (OD750) showed even more significant changes, they were not

reliable at this point as being strictly distorted by morphological changes (section 3.4.1).

As nitrate had already been depleted from the medium, the increase of biomass concentra-

tion observed for the MTP as well as the shake flask and the MC 1000 cultivation was most

probably attributable to the exploitation of intracellular nitrogen pools. In plant biomass, a

significant share of intracellular nitrogen is typically fixed within the proteins and pigments of

the photosynthetic apparatus. Especially the chlorophyll fraction, containing four nitrogen

atoms per molecule, may contribute to up to 5 % of cell dry weight [317, 318]. Thus, its re-

structuration and a coupled increasing availability of intracellular nitrogen during the MTP,

the shake flask and the MC 1000 cultivation could be a trigger for further biomass formation.

During the FMT 150/1000 process, only 46 % of the available extracellular nitrate was incor-

porated by the cells until harvest. As a consequence, the physiological state cannot be clas-

sified as stationary according to the commonly used definition. In the context of phototrophic

growth kinetics and the underlying suboptimal light supply revealed during the previous pro-

cess phases, it could be assumed that upon a distinct biomass concentration, the total light

Results and discussion 73

energy influx was directed into cellular maintenance (respiration). Thus, no net biomass was

gained and potential product synthesis was suppressed.

Regarding the MTP, shake flask and MC 1000 process, the accumulation of intracellular neu-

tral lipids started with some delay after the depletion of nitrate from the medium and steadily

continued until harvest. The lag time observed was most probably attributable to the upregu-

lation of neutral lipid processing enzymes. On the contrary, product formation was not in-

duced during the FMT 150/1000 because nitrate was not limitating and the non-sufficient

energy supply to the cells by the illumination.

The differing biomass concentrations, lipid contents and thus volumetric productivities are

summarized in Table 3.2.

Table 3.2: Comparison of cell dry weight (CDW), neutral lipid content (nLC) and volumetric productivity (Pvol). Error bars represent min/max from biological replicates (n = 2).

measure MTP shake

flask MC 1000

FMT

150/1000

CDW [g L-1] 5.5 ± 0.1 5.8 ± 0.1 5.3 ± 0.4 1.2 ± 0.2

nLC [% (w w-1)] 13.7 ± 0.1 16.8 ± 1.4 14.7 ± 0.4 0.5 ± 0

Pvol [mg L-1 d-1] 75 ± 1 97 ± 8 78 ± 6 1 ± 0

While the MTP (5.5 ± 0.1 g L-1) as well as the shake flask (5.8 ± 0.1 g L-1) and the MC 1000

process (5.3 ± 0.4 g L-1) resulted in a fairly comparable CDW, the FMT 150/1000 cultivation

achieved as little as 1.2 ± 0.2 g L-1. In analogy to biomass concentrations, the neutral lipid

content of 13.7 ± 0.1 % (w w-1) achieved during MTP cultivation was competitive to those

from the shake flask and the MC 1000 process being at a level of 16.8 ± 1.4 % (w w-1) and

14.7 ± 0.4 % (w w-1), respectively. Consequently, similar volumetric productivities were ob-

tained while no lipid production occurred during the FMT 150/1000 cultivation.

Besides absolute quantities, the fatty acid fingerprint from neutral lipids of the harvested cells

was evaluated making use of GC-ToF-MS (Figure 3.15).

Results and discussion 74

Figure 3.15: Fatty acid fingerprint from neutral lipids from the reference processes. Error bars represent min/max from biological replicates (n = 2). Modified from [313].

Analysis revealed two discrete fingerprint patterns. Both had in common that the fatty acid

fractions of 16:0 (25 %) and 18:2 Δ9,12 (8 %) were rather constant across all of the four refer-

ence processes while those of 14:0 (not shown, only trace amounts present), 16:1 Δ7,

16:1 Δ9, 16:2 Δ7,10 and 18:0 constantly contributed only little to the spectrum (≤ 6 % in total).

In the case of non-induced lipid accumulation (FMT 150/1000), the cells contained significant

fractions of the polyunsaturated fatty acids, namely 16:3 Δ7,10,13 (19 %) and 18:3 Δ9,12,15

(42 %). On the contrary, storage lipid accumulation in MTP, shake flask, MC 1000 resulted in

a shrinkage of these fractions to 6 % and 27 %, respectively, and small amounts of 18:0

(6 %) as well as a significant share of 18:1 Δ9 (30 %) were synthesized.

The qualitative fingerprint recorded was in accordance to the fatty acid fingerprint reported to

be universal for the genus of Chlorella [56]. Despite occurring in rather low amounts, the

simultaneous presence of two stereoisomers of hexadecenoic acid, namely 16:1 Δ7 and

16:1 Δ9 could be detected. For neutral lipid accumulating cells in MTP, shake flask, MC 1000,

the observed semi-quantitative fingerprint was well comparable. With regard to biodiesel syn-

thesis as a typical designated use of algal lipids, the occurring reduction of the polyunsatu-

rated fatty acids’ fraction was clearly advantageous. A share beyond specifications would

need to be reduced by catalytic hydrogenation or blending inevitably resulting in rising cost

(section 1.3.2).

MTP shake flask MC 1000 FMT 150/1000

0

25

50

75

100

rela

tiv

e f

rac

tio

n

[%

]

reactor system

16:0

16:1 7

16:1 9

16:2 7,10

16:3 7,10,13

18:0

18:1 9

18:2 9,12

18:3 9,12,15

Results and discussion 75

3.3.5 Conclusion

In section 3.3, the scalability of the micro-photobioreactor system introduced in section 3.2

was evaluated. Therefore, its performance during an exemplary reference process was eval-

uated against a set of established laboratory-scale photobioreactors. In particular, this model

process aimed at the production of intracellular neutral lipids using the unicellular microalga

C. vulgaris. MTP cultivation was compared against shake flask cultivation as well as pro-

cesses using a parallelized test tube system (MC 1000) and a 1 L photobioreactor in classi-

cal flat panel design (FMT 150/1000) by means of biological performance.

During exponential growth, the MTP cultures achieved a growth rate comparable to that ob-

served for the shake flask and the MC 1000 cultivations (1.44 ± 0.02 d-1) while cultures in the

flat panel reactor grew slightly slower (1.34 ± 0.07 d-1). However, cultures could be kept in

the exponential growth phase in MTPs up to higher biomass concentrations

(OD750 = 5.74 ± 0.02) than in the other systems. Moreover, all systems showed a light-limited

linear growth after the exponential phase. The rates of biomass formation during that phase

reflected the pattern observed for the duration of the exponential phases where the

FMT 150/1000 showed the lowest and the shake flask as well as the MC 1000 and the MTP-

based system showed an improved performance. These observations were traced back to

the beneficial light supply in MTPs, which is based on their high specific illumination area as

well as short mixing times due to turbulent particle trajectories and rigorous shaking.

However, quantitative data about spatially resolved optical path length distribution and parti-

cle trajectories inside the MTP wells at different shaking conditions are lacking at present.

Assessing these for example via computational fluid dynamics, could provide valuable infor-

mation to quantitatively explain the observed differences between MTP cultivation and the

established photobioreactor systems.

Regarding overall productivity, the MTP cultivation achieved a biomass concentration and

neutral lipid content comparable to the shake flask and the MC 1000. Moreover, the lipid

fractions’ relative composition was comparable as well. Contrary, the flat panel system

achieved only 1.2 ± 0.2 g L-1 of CDW and no lipid accumulation was induced as the available

light energy was most probably used up completely for cellular maintenance at this biomass

concentration.

Overall, MTP cultivation proved a very good scalability to the investigated laboratory-scale

photobioreactors. Excellent comparability was achieved regarding the exponential growth

phase while the small scale helped prolonging this phase due to superior light supply and

Results and discussion 76

mixing. Regarding productivity and biomass yield, only little variation was observed. Taking

the differing scales into account, the overall differences were rather small taking so that

MTPs may be regarded as a proper tools for laboratory-scale phototrophic cultivation.

3.4 High throughput lipid quantification

Within the framework of this thesis, intracellular neutral lipids, produced by starved and/or

stressed C. vulgaris were chosen as a target for product formation studies. Consequently,

appropriate quantitative analytics had to be established taking care of their applicability for

accelerated bioprocess development. Apart from elaborate but highly expensive methods

like-time domain nuclear magnetic resonance, high-performance liquid chromatography or

gas chromatography mass spectrometry [319, 320], gravimetric measurement remains the

best-established method for lipid quantification. It mostly relies on two-phase chloroform-

methanol/water extraction of biomass and subsequent gravimetric determination of the rela-

tive cellular lipid fraction [321-325]. Despite of significant proceedings regarding protocol

simplification [256], extractive lipid quantification remains unsuitable for microscale applica-

tions for the following reasons:

(i) Solvent extraction and gravimetry are multistep procedures implying significant time

effort for the handling of consumables (drying, weighing) and solvent evaporation.

Even the fastest protocols need at least 3 days, containing a large proportion of

hands-on-time thereby significantly limiting experimental throughput.

(ii) At least 20 mg of biomass are needed for reliable analysis. This amount cannot be

generated from MTP cultivations at microlitre scale. Especially during the early

growth phase, only very limited amounts of biomass (< 0.1 mg) can be harvested

from individual wells of an MTP.

(iii) Quantitative two-phase solvent extraction suffers from high sensitivity against fluctua-

tions of process conditions [326]. Hence, assessment of relevant errors and error

handling is a crucial issue when down-scaling this kind of analytical procedures.

Fluorescent dyes offer an indirect alternative for the detection of intracellular lipids that, on

the one hand requires only simple optical measurement and on the other hand enables for

high throughput applications. Most often, the lipophilic dyes Nile red (9-diethylamino-5-

benzo[α]phenoxazinone) and BODIPY 505/515 (4,4-difluoro-1,3,5,7-tetramethyl-4-bora-

3a,4a-diaza-s-indacene) are used [320]. In general, the fluorescence of stained cells is ac-

cessible by spectrofluorometry, the use of fluorescence microscopy or flow cytometry provid-

ing the possibility to investigate population heterogeneity in lipid content across individual

cells from a distinct culture [327]. Spectrofluorometric methods can easily be transferred to

Results and discussion 77

MTP-scale thereby enabling automation by using liquid handling robots [320]. BODIPY stain-

ing became increasingly widespread within the last decade due to advantages like superior

photostability [282] or fast cell wall permeation and staining kinetics [328]. Nevertheless, Nile

red (Figure 3.16) remains the more suitable dye for spectrofluorometric applications in terms

of quantitative analysis, since BODIPY staining suffers from strong background fluorescence

of the dye, distorting absolute lipid quantification [282].

property value

formula C20H18N2O2

molecular weight 318.37 g mol-1

solubility in water poor

common solvents acetone, DMSO

Figure 3.16: Chemical structure and selected properties of the lipid probe Nile red.

Nile red is used as a probe for intracellular lipids since the 1980ies [329, 330]. Despite show-

ing poor fluorescence in water, it shows strong fluorescence in hydrophobic environments

However, the fluorescence spectrum is highly dependent on the dye’s microenvironment

[331]. Thereby, varying the excitation and emission wavelengths was used to stain different

classes of hydrophobic molecules (Table 3.3).

Table 3.3: Target molecule addressing by excitation and emission wavelength variation [332-337].

target excitation [-] emission [-]

cholesterol in human plasma 450 - 500 > 528

enzyme mechanisms 550 640 - 660

neutral lipids (triacylglycerides) 450 - 500 570 - 600

polar lipids (phospholipids) 515 - 560 ≥ 590

By now, a large number of protocols using Nile red for the quantification of intracellular lipids

have been developed, while especially dye permeation was identified as a major bottleneck.

Although appropriate staining conditions (cell and dye concentration, temperature, improved

permeation via organic solvents or physical pretreatment) have been established for a variety

of different microalgae, their applicability remains strain specific so that individual parameter

optimization and gravimetric calibration are needed to allow for cross-strain comparison [320,

Results and discussion 78

338]. A comprehensive and current review about assessment of microalgal lipid production

can be obtained from [339].

In the context of accelerated bioprocess development using elevated throughput cultivation

technology, it was aimed at revealing the potential of Nile red staining for high throughput

analysis. Therefore, a protocol that uses laboratory automation via a liquid handling robot to

enable reproducibly measuring large sets of samples with small volume available and mini-

mized hands-on-time was developed.

3.4.1 Standardization of biomass concentration

Practically all published methods using spectrofluorometry take place at a constant biomass

concentration as the amount of biomass to be stained per assay is directly correlated to the

corresponding fluorescence signal. Biomass determination and adjustment is done by meas-

urement of either optical density or cell number for reasons of simplicity. The model organism

C. vulgaris used in this study replicates via vegetative autosporulation [59] and thereby un-

dergoes significant morphological changes during a cultivation process. As the optical prop-

erties of suspended particles are highly dependent on their size, shape and refractive indices

[306], the impact on optical density assisted cell quantification needs to be evaluated. Thus,

samples of C. vulgaris at different physiological states were analysed with regard to their

optical properties, measured in terms of simple optical density in a spectrophotometer as well

as cell size and biovolume using particle counting technology (Figure 3.17).

Results and discussion 79

Figure 3.17: Optical properties and average cell size at different physiological states. De-pending on the state observed, the cells significantly differ in their biovolume-specific optical density and cell size. Thereby, biomass determination across different states by either optical signals or cell number is distorted. Cell samples originated from phototrophic shake flask cultivations. Error bars deviated from samples (biological replicates) of the respective fer-mentation phases (n ≥ 3). Modified from [258].

The ratio of optical density to biovolume of C. vulgaris cells at the different physiological

states observed increased significantly (p < 0.05) by more than 60 % from 1.08 ± 0.06 for

exponentially growing via 1.36 ± 0.1 for light-limited growing cells to 1.73 ± 0.06 and

1.81 ± 0.05 for nitrogen-starved cells containing low and high amounts of intracellular lipids,

respectively. In parallel, the average cell size decreased from approx. 6.5 µm during expo-

nential growth down to 3.4 µm during N-starvation.

The changes of the optical signal were likely caused due to morphology dependent light scat-

tering characteristics of the cells. Cell sizes largely differed across the physiological states as

the relative fractions of small autospores, replicating mother cells and starved cells varies

from non-limited growth to light and nutrient starvation. Additionally, the composition of

C. vulgaris’ cell wall is known to undergo major restructuration during cell cycle [33] which

changes the optical properties as well. Probably, even the accumulation of intracellular lipid

droplets (liposomes) [282, 340] might significantly change the optical properties through an

alteration of the cellular refraction index. Consequently, neither optical density, nor cell num-

ber can be applied as a valid measure for biomass adjustment across different morphological

0.0

0.5

1.0

1.5

2.0

exponential

growth

N-starvation,

high lipid

content

OD

75

0 b

iov

olu

me

-1 [

mL

L

-1]

physiological state

light-limited

growth

N-starvation,

low lipid

content

0

2

4

6

8

av

era

ge

ce

ll d

iam

ete

r [

m]

Results and discussion 80

states and lipid contents. Assuming that the optical properties of the cell, determined by opti-

cal density, would show no conflict with exact biomass determination, a constant ratio of

OD750 to biovolume would have been observed in Figure 3.17 (blue bars).

Nevertheless, optical density is frequently used for biomass quantification prior to Nile red

staining [257, 334, 341] which has to be evaluated quite critically. Instead, the biovolume was

chosen as a more reliable measure of biomass to be used for assay development. Its quanti-

fication via a particle counter is independent from the cells’ optical properties (size, scattering

and absorption characteristics) and can be assessed rather quickly. Being equivalent to the

cytoplasmic volume per volume of sample, it represents the amount of biological reaction

space available to catalyse lipid synthesis and can thereby be seen as an appropriate bio-

mass standardization parameter.

3.4.2 Assay development and validation

The general concept of the Nile red assay to be used was adopted from literature. A spectro-

fluorometric MTP-based protocol with high potential for automation via a liquid handling robot

was chosen as a starting point [257]. Following the optimization experiments given, a dye

concentration of 1 mg L-1 and an incubation temperature of 40 °C were applied. Staining was

conducted in 25 % (v v-1) DMSO to facilitate dye penetration. Choosing an excitation wave-

length of 480 nm for specific detection of neutral lipids (Table 3.3), the resulting fluorescence

spectrum was evaluated for possible interference by background originating from assay

compounds, basal fluorescence of the dye and cell autofluorescence (Figure 3.18).

Results and discussion 81

Figure 3.18: Typical fluorescence spectra of the lipid assay. Staining of cells with the lipid probe Nile red and subsequent excitation at 480 nm results in a specific, lipid correlated fluo-rescence with a maximum at 570 - 580 nm. Cells with a biovolume of 0.15 µL mL-1 were stained for 3 h at 40 °C in 25 % (v v-1) DMSO with 1 mg L-1 Nile red, excitation at 480 nm. Modified from [258].

In accordance with literature [257, 333], intracellular neutral lipids could be detected by stain-

ing of lipid producing cells via a fluorescence signal with a maximum at 570 - 580 nm. Nei-

ther significant background fluorescence nor Nile red (600 - 750 nm) or chlorophyll mediated

cellular autofluorescence (650 - 750 nm) were observed to interfere with the lipid signal, so

that the average fluorescence signal within the interval of 570 - 580 nm was chosen as a

marker for intracellular lipids.

As stated previously, conditions for reliable quantitative staining enabling lipid quantification

via Nile red fluorescence are observed to be highly strain specific due to varying dye uptake

characteristics [320]. Consequently, biomass concentration and staining time were optimized

for C. vulgaris at the selected dye (1 mg L-1) and solvent concentration (25 % (v v-1)) and

temperature (40 °C), respectively. Non-growing cells with high as well as growing cells with

low lipid content were used to investigate potentially different staining characteristics (Fig-

ure 3.19).

500 600 700 800

0

5

10

15

20

25

flu

ore

sc

en

ce

[

10

3 a

.u.]

emission wavelength [nm]

DMSO blank

Nile red blank

cell blank

stained cells

Results and discussion 82

Figure 3.19: Characterization of the Nile red staining. A: Biomass specific lipid fluorescence. After an initial linear correlation, fluorescence saturates above a biovolume of 0.2 µL mL-1 (dashed line). B: Time-dependent staining kinetics of cells with a biovolume of 0.15 µL mL-1. In contrast to staining growing cells containing low levels of lipids (0.5 h), non-growing cells with high lipid need significantly more time (2 h) for quantitative staining. Cells were stained in 25 % (v v-1) DMSO containing 1 mg L-1 Nile red at 40 °C, excitation at 480 nm. Error bars deviated from analytical replicates (n = 5). Modified from [258].

The fluorescence signal deviated from staining was observed to be linearly correlated to the

applied cell concentration up to a biovolume of 0.2 µL mL-1 for both cell types investigated,

while staining of higher biomass concentrations resulted in a saturation (Figure 3.19 A). This

could be a consequence of a negative effect in the cell permeation or the staining reaction

pointing to a potential limitation of the dye transport or depletion of Nile red at higher biomass

concentrations. Especially the latter hypothesis could be explained by adsorption of Nile red

to other hydrophobic compartments like membrane lipids, thereby reducing the amount of

dye available for storage lipid staining. Regarding the staining kinetics, a diverse pattern was

observed. Growing cells, containing only low amounts of neutral lipids were quantitatively

stained within 30 min. Complete staining of starved cells with a high content of intracellular

neutral lipids took at least 2 h for a biomass concentration of 0.15 µL mL-1 (Figure 3.19 B).

The slow staining of starved cells can be explained by different hypotheses: First, dye per-

meation into the starved cells could be hampered by their more rigid cell wall. Furthermore,

diffusion of Nile red from the hydrophobic cell membrane via the aqueous cytoplasm towards

0.0 0.5 1.0 1.5 2.0 2.5

0

25

50

75

100

0 2 4 6

0

5

10

15

20

25 low lipid content

high lipid content

low lipid content

high lipid content

flu

ore

scen

ce [

10

3 a

.u.]

biovolume [L mL-1]

A

flu

ore

scen

ce [

10

3 a

.u.]

staining time [h]

B

Results and discussion 83

the liposomes as well as the dye uptake by the liposomes itself, could be rate-limiting [257,

334, 341]. However, the contribution of each possible effect to the slowed staining kinetics

was not further investigated. Instead, a biovolume of 0.15 µL mL-1 and a staining time of 3 h

were chosen as standard parameters to enable quantitative staining of cells with differing

lipid content within the linear range of the assay.

The novel assay was calibrated against gravimetric determination of the lipid content analys-

ing algal samples after different times of nitrogen-starvation, using both orthogonal methods

(Figure 3.20).

Figure 3.20 Gravimetric calibration of the lipid assay. The Nile red staining deviated biomass specific lipid fluorescence is linearly correlated with the fraction of intracellular lipids meas-ured by extraction and subsequent gravimetric analysis. Cells with a biovolume of 0.15 µL mL-1 were stained for 3 h at 40 °C in 25 % (v v-1) DMSO with 1 mg L-1 Nile red. Error bars of the Nile red assay deviated from analytical replicates (n = 5). Error bars of the gravi-metric measurement from technical replicates (n = 3). Modified from [258].

Within 0.6 · 103 – 34.3 · 103 a.u., the obtained fluorescence was found to be linearly correlat-

ed to a lipid content ranging from 13.3 – 44.5 % (w w-1). Average relative errors were ± 2 %

and ± 8 % for lipid fluorescence and gravimetry, respectively.

Consequently, fluorescence data from Nile red staining can be assessed for absolute quanti-

fication of the cellular lipid content, while accuracy is significantly (p < 0.05) enhanced com-

pared to gravimetric measurement. Key to success enabling reliable and valid quantitative

0 10 20 30 40

0

10

20

30

40

50

lip

id c

on

ten

t

[% (

w w

-1)]

lipid fluorescence [103 a.u.]

Results and discussion 84

data obtained from the fluorescence assay is the standardized approach with fixed values for

biomass concentration (i.e. biovolume), Nile red concentration and incubation conditions.

Strikingly, analysing samples with low lipid content generated by harvest of exponentially

growing cultures revealed a fluorescence signal in the range of the background signal, i.e. no

lipids were detected in the cells, whereas gravimetric analysis showed a lipid content of

13.3 % (w w-1). This offset may be explained by the differing specificity of the methods. By

choosing an excitation/emission setup of 480/570-580 nm, Nile red staining was designed to

specifically stain neutral lipids [257] which are typically produced during nutrient starvation.

On the contrary, gravimetric measurement is based on the extraction of the total cellular li-

pids [256]. Besides neutral lipids, these additionally contain e.g. the membrane phospholipids

that are not detected by the Nile red staining. This fraction is represented by the y offset of

the correlation function (Figure 3.20) and the observed lipid content of approx. 13 % (w w-1)

is in exact agreement with previous literature reports about the biomass composition of grow-

ing Chlorella [54].

3.4.3 Automation

With respect to the increasing number of samples that can be generated from current and

next generation parallelized microscale photobioreactors, not only higher accuracy and pre-

cision, but also further acceleration of lipid quantification is needed to prevent analytics from

becoming a bottleneck. For this purpose, the developed assay was transferred to an estab-

lished liquid handling platform [168] (Figure 3.21).

Figure 3.21: Assay automation significantly enhances analytical throughput. Using the de-scribed liquid handling platform, hands-on-time and total runtime can be reduced by 66 and 37 %, respectively. Arrows represent the duration of individual assay steps, whereas the re-spective crosshatched fractions represent hands-on-times. Dark blue: pre-dilution of sam-ples, red: measurement and adjustion of biovolume, light blue: staining, green: sample trans-fer and fluorescence measurement. Modified from [258].

Assay automation resulted in a considerable increase of experimental throughput. The work-

flow was designed to handle up to 48 samples per run. For this, a total runtime of the robotic

Results and discussion 85

assay of approx. 6.5 h, including 2.5 h of hands-on-time is needed. Compared to manual

assay preparation taking 10.5 h, including 7.5 h of hands-on-time, the overall time needed

was reduced by 37 %. In particular, hands-on-time was reduced by 5 h (-66 %) minimizing

the needed human operator interaction. Thereby, the use of a liquid handling robot for assay

automation proved to be valuable for increasing analytical throughput.

As pipetting steps during assay preparation may have a major impact on data reliability, the

effect of routinely running lipid analysis on a laboratory robot was investigated with regard to

accuracy and precision within the relevant volume ranges. Different volumes of desalted wa-

ter were repeatedly transferred either by the liquid handling robot or manually by four differ-

ent operators using the same equipment (Figure 3.22 A).

Figure 3.22: Automated liquid handling increases accuracy and precision. A: Although liquid handling robots suffer from systematic and stochastic errors as well, these are typically con-stant while manual pipetting by different operators (4 persons in this case) can result in sig-nificant differences. B: Daily operator performance can induce significant fluctuations of manual pipetting performance. Automated pipetting conducted by Janus Integrator liquid handling platform and manual pipetting by four different operators using an Eppendorf Re-search Plus pipette. Volumes determined gravimetrically using a semi-micro balance and normalized to the respective target volumes. Error bars deviated from analytical replicates (n = 10). Modified from [258].

0.96

0.98

1.00

1.02

1.04

timepointtarget volume [mL]1000500250

rela

tive v

olu

me [

-]

MPP

operator 1

operator 2

operator 3

operator 4

50

A

0.96

0.98

1.00

1.02

1.04

rela

tive v

olu

me [

-] 50 L

250 L

500 L

1000 L

morning afternoon night

B

Results and discussion 86

At a target volume of 50 µL no significant difference between automated and manual liquid

transfer was observable. On the contrary, for target volumes > 50 µL the robotic platform

ensured an improved performance compared to manual pipetting. With an average systemat-

ic error of 0.3 % a significantly (p < 0.05) higher accuracy was achieved than by manual pi-

petting (0.6 - 1.0 %, depending on the individual operator). Nevertheless, stochastic fluctua-

tions were comparable between automated (0.3 %) and manual liquid transfer (0.2 - 0.3 %).

The daily performance for pipetting accuracy and precision was investigated as well (Fig-

ure 3.22 B). Depending on time (morning, afternoon, night) and target volume the errors of

one individual operator suffered from noticeable fluctuations. In this case, the pipetting per-

formance was best during the morning hours as a relatively constant systemic error of 0.6 %

with stochastic fluctuations of 0.2 % on average were achieved. During progress of the day,

a loss of accuracy and precision was observable as systematic and stochastic errors signifi-

cantly (p < 0.05) grew up to 1.1 % and 0.4 %, respectively.

It became apparent that replacing manual pipetting by a liquid handling robot increases pi-

petting accuracy and precision. Fluctuations by differing pipetting performance of individual

operators performing the assay can be eliminated as well as variations induced by daily fluc-

tuation.

As described in section 3.4.2, the applied biomass concentration has a strong impact on the

measured fluorescence. Therefore, it needs to be adjusted as precisely as possible. For a

final validation of the automated liquid handling, the individual fluctuations across technical

replicates (n = 5) conducting the assay either by hand or running it on the robotic platform

were evaluated. Average standard errors of 4.3 % and 2.1 % were observed for manual and

automated analysis, respectively. Thus, the previously demonstrated superior performance

of the robotic system was proven to affect reproducibility in a positive manner. Besides the

aspect of elevated throughput, this improvement gave further legitimation to the automation

of the Nile red staining assay, especially with focus to running it as routine analytics.

3.4.4 Conclusion

In section 3.4, an automated high throughput assay for absolute quantification of intracellular

neutral lipids is presented. Cells were stained using the lipid probe Nile red with DMSO as

solvent facilitating dye permeation. Staining parameters were optimized for application at

MTP-scale with regard to differentiating staining kinetics of cells at different physiological

states and biomass concentrations. The assay was designed to be run on an automated liq-

uid handling platform. Thereby, up to 48 samples can be measured within 6.5 h, reducing

Results and discussion 87

hands-on-time to a third compared to preparation by hand. The developed assay was cali-

brated against gravimetric lipid measurement to allow for absolute quantification. Here, it was

shown that analytical errors could be reduced from ± 8 % to ± 2 % on average. Thus, preci-

sion was significantly improved compared to extractive methods. Furthermore, evidence is

given that using a laboratory robot can increase accuracy and precision compared to manual

preparation by different operators whose performance may additionally suffer from daily fluc-

tuation.

Contrary to established spectrofluorometric methods relying on optical density or cell num-

ber, here the biovolume was used to adjust for identical biomass concentration across all

conducted measurements. It was shown that optical signals as well as cell number cannot be

applied for algae with complex lifecycle as is the case for C. vulgaris. Due to their mode of

replication via vegetative autosporulation, the cells undergo morphological changes signifi-

cantly altering their optical properties. Even different amounts of accumulated lipids could

have an impact on refractive properties.

Besides improvements regarding accuracy and precision as well as data reliability and ana-

lytical throughput, the newly established assay is especially suited for the analysis of sam-

ples from current and next generation microscale photobioreactor systems. These typically

generate large sets of samples at reduced volume (approx. 1 mL) which cannot be handled

by conventional gravimetric methods. Applying the presented staining technology, these low

volume samples can easily be handled circumventing analytics from becoming a bottleneck.

Thereby, the assay contributes to boosting the time efficiency for setting up algae lipid pro-

duction processes.

3.5 Integration of experimental design

By the de-bottlenecking of phototrophic cultivation (section 3.2) and the development of sup-

porting methodologies (sections 3.1 and 3.4), experimental throughput can be raised signifi-

cantly. On the mid-term, these initial developments and especially further progress in high

throughput technology and laboratory automation will clearly boost the efficiency of photo-

trophic process development. Nevertheless, experimental capacities, mainly cultivation, can-

not keep up with the required throughput. Phototrophic processes are characterized by their

intrinsic complexity induced by a high number of potentially interacting input variables includ-

ing dynamic parameters like light or temperature cycles. Current micro-photobioreactors

mainly rely on standardized microtiter plates. Thus, a further rise of cultivation capacity by

intensifying parallelization would need to base on scale-out and inevitably go along with in-

creasing cost. Hence, alternative strategies need to be considered.

Results and discussion 88

One approach to achieve that efficiently, which is already well-established for microbial bio-

process development [243], is the use of Design of Experiments (DoE) to focus on experi-

ments providing the highest information content in a targeted parameter space. Despite hav-

ing been established during the early 20th century [240], there is still ongoing research into

this methodology [175]. DoE is regarded to be particularly suitable to deal with the combina-

torial explosion typically occurring when investigating multi-parameter relations [245]. Moreo-

ver, it overcomes a critical limitation of “conventional” one-factor-at-a-time experiments, as

such approaches often fail in locating global optima by not taking potentially synergistic inter-

actions of input variables into account [239]. Regarding bioprocess development, the most

prominent application of DoE is culture media optimization [243]. During such tasks, the om-

nipresent interactions between single compounds render locating a global optimum by “con-

ventional” experimental planning to chance.

Consequently, it was aimed at combining the previously reported technologies for parallel-

ized microscale cultivation and analytics to phototrophic microorganisms with elaborate sta-

tistical experimental design as previously fruitfully applied for heterotrophic [175, 222].

Thereby, an integrated framework for the accelerated development of phototrophic biopro-

cesses is to be set up. With respect to the C. vulgaris lipid production process used as a

model throughout this study, the previously established enBBM cultivation medium should be

optimized towards maximized volumetric lipid productivity as an application study.

3.5.1 Choice of relevant media components

The medium targeted for optimization incorporates 17 different components (section 2.4) with

phosphate salts counted as one compound due to pH-dependent equilibrium. This number is

too high to efficiently perform the experimental study with a manageable number of experi-

ments, since a full factorial design with two concentration levels would result in 217 ≈ 130,000

experiments. In order to keep the number of components of interest, preselection was com-

pleted based on literature information. Table 3.4 summarizes the known biological effects of

the individual components. Penicillin-G concentration was kept constant throughout all condi-

tions and all trace elements were clustered to one single input variable as a similar effect on

cultivation was expected. Sulfuric acid and potassium hydroxide had to be varied together

with FeSO4 and Na2EDTA, respectively, as they were needed to keep the latter two compo-

nents dissolved in their stock solutions. Thereby, the addressed medium components were

CaCl2, FeSO4/H2SO4, K2HPO4/KH2PO4, MES, MgSO4, NaCl, Na2EDTA/KOH, NaNO3 and

trace elements (CoSO4, CuSO4, H3BO3, MnCl2, Na2MoO4, ZnSO4) reducing the number of

input variables by almost 50 % from 17 to 9.

Results and discussion 89

Table 3.4: Initial evaluation of the medium components. Literature indicates potential for the optimization of volumetric lipid productivity.

* varied together with FeSO4 as provided in one single stock solution

** varied together with Na2EDTA as provided in one single stock solution *** all trace elements were clustered to one single input variable

component evaluation reference variation

CaCl2 versatile effector in plant cells; essential for induc-

tion of lipid synthesis

[342-344] yes

FeSO4 influence on growth and lipid metabolism reported [343, 345-

347]

yes

H2SO4 sulfur supply ensured by sulfate anions from other

medium components

- yes*

K2HPO4 /

KH2PO4

essential phosphorus source (nucleic acid synthe-

sis)

[346] yes

KOH potassium excess by phosphate salts - yes**

MES trade-off between osmotic stress and buffering; alka-

line pH may inhibit cell cycle

[43] yes

MgSO4 influence on growth and lipid production reported;

effector of acetyl-CoA carboxylase, an essential

enzyme during lipid biosynthesis; central atom of

chlorophyll

[40, 42,

343]

yes

NaCl reported to increase lipid production; excess may

cause metabolic burden (active sodium export

channels) and thus inhibit growth

[348, 349] yes

Na2EDTA common metal chelator; excess may cause growth

repression by ion depletion

[343, 347] yes

NaNO3 essential nitrogen source (protein synthesis) [346] yes

penicillin-G support of long-time sterile conditions; not metabo-

lized (data not shown)

- no

trace ele-

ments

numerous studies about wastewater detoxification,

but limited information concerning metabolism and

lipid production; general pattern: little amounts es-

sential, but high level cytotoxic; thus clustered to

one input variable

[350-355] yes***

Results and discussion 90

3.5.2 Kriging-assisted optimization

Fractional factorial 3.5.2.1

Starting with the nine remaining media components of interest, a full factorial design would

require 29 ≈ 500 experiments. Making full use of 48-fold parallelized microtiter plate cultiva-

tion, this leads to a total of 11 experimental runs, equivalent to 4 months of cultivation time.

In 12-fold parallelized shake flasks, the experiments would even take 14 months. Yet, such

time scales are clearly far from feasible, underlining the necessity to effectively reduce the

experimental effort.

Fractional factorial designs allow the reduction of the number of experiments by estimating

only single component effects and a subset of combinatorial effects [261]. The chosen de-

sign (Table 6.1, appendix) comprises 37 experiments, while five represent the reference

point using the enBBMref medium (medium composition provided in Table 6.2, appendix).

Taking reference points into account allows the investigation of measurement noise and

normalization. The other experiments allowed for a statistical analysis of the effect of single

components as well as the interaction with magnesium ions. The interaction with this divalent

metal ion was analyzed, as it is reported to be an effector of the acetyl-CoA carboxylase, an

enzyme essential for lipid biosynthesis responsible for the initial step of carbon dioxide fixa-

tion to malonyl-CoA (Table 3.4). An overview of the functionality of this enzyme complex and

its regulation is given by Ohlrogge & Browse [84]. Thus, any interactions with this input vari-

able are of special interest with respect to product accumulation in the cells.

Figure 3.23 A shows the resulting statistical analysis of the experiment based on fractional

factorial design. The blue bars indicate the expected effect of varying the medium concentra-

tions between their minimal and maximal values (Table 6.1, appendix). The error bars indi-

cate the uncertainty of the estimations. In the following, the main and combinatorial effects of

the components are checked for significance using a t-test with a significance level of

p = 0.1. Using a lower significance level would increase the risk of false negatives, i.e. ex-

cluding relevant components from the remaining study. The diagram shows that an increase

in the concentration of NaNO3 has a significant negative effect (p < 0.1) on lipid productivity.

On the other hand, an increase in the trace element’s concentrations results in a significant

(p < 0.1) productivity improvement. Furthermore, the analysis indicates a positive tendency

with increasing CaCl2 and lowering EDTA concentration. However, because of measurement

noise and the comparably low number of experiments, the uncertainty of the estimation is

relatively high and allows no reliable statement about the effect of CaCl2 and EDTA, respec-

tively. Similar holds true for MgSO4, but here, the pairwise interaction with another compo-

Results and discussion 91

nent was additionally investigated. As shown in Figure 6.6 (appendix), a significant (p < 0.1)

negative combinatorial effect was identified with the sodium salts NaNO3 and NaCl.

For visual inspection of the negative combinatorial effect, a Kriging model was constructed

based on the given data. The predicted functional relationship between MgSO4, NaNO3 and

the lipid productivity is visualized in Figure 3.23 B. In case of low NaNO3 concentration, the

interpolation reveals a positive correlation between an increase in MgSO4 and the perfor-

mance indicator. With increasing NaNO3 concentration, this positive effect is weakened.

In conclusion, significant effects of NaNO3 and the trace elements were identified as well as

positive tendencies of MgSO4 and CaCl2. Furthermore, the effect of MgSO4 appears to de-

pend on the Na-salts NaNO3 and NaCl. The remaining components have only low potential

to affect the lipid productivity and were thus excluded from further analysis.

Figure 3.23: Initial fractional factorial screening analysis. A: Estimated relative effect of sin-gle components on volumetric lipid productivity. B: Kriging visualization of combinatorial ef-fect of MgSO4 and NaNO3 on volumetric lipid productivity. Remaining medium components are set to their reference value. Modified from [356].

Results and discussion 92

Full factorial 3.5.2.2

In order to verify the observed tendencies and to investigate potential pairwise or higher

combinatorial effects, a full factorial design was constructed for the remaining five input vari-

ables: NaNO3, MgSO4, CaCl2, NaCl, and the clustered trace elements. This design again

comprises five reference points and 32 experiments with minimal/maximal concentration

(Table 6.3, appendix).

Figure 3.24 A shows the updated statistical results after performing the full factorial design.

The previously observed effects of NaNO3 and the trace elements were confirmed. The posi-

tive tendency of CaCl2 turned out to be significant while the effect of NaCl remained insignifi-

cant. However, the interaction of MgSO4 with the Na-salts could be investigated in more de-

tail. Figure 3.24 B shows the opposing effect of MgSO4 in dependency of NaNO3. This inter-

action leads to a non-distinguishable single component effect of MgSO4, as indicated in Fig-

ure 3.24 A. The analysis also revealed a negative interaction between CaCl2 and the trace

elements, as indicated in the screening plot in Figure 6.7 (appendix).

Figure 3.24: Refined full factorial screening. A: Estimated relative effect of single compo-nents on volumetric lipid productivity, B: Kriging visualization of the combinatorial effect of MgSO4 and NaNO3 on volumetric lipid productivity. NaCl was set to its minimum value, CaCl2 and trace elements are set to their respective maxima. Modified from [356].

Results and discussion 93

Locating optimal medium composition 3.5.2.3

In section 3.5.2.2, single and combinatorial effects of the media components were investigat-

ed on the basis of a full factorial design providing a rough estimate about optimal medium.

The goal of the next step was to examine limitations of the particular effects and to identify

potential optimal media compositions. To achieve this, minimum and maximum concentra-

tions were adjusted, and a more complex experimental design scheme was applied, compris-

ing several nested factorial designs (Table 6.4, appendix).

The maximum concentration of NaNO3 was lowered from 1.7xRef to 1xRef. The upper bound

of the clustered trace elements concentration was increased by 50% to 3.75xRef. The con-

centration of CaCl2 could not be increased, as otherwise precipitation effects were observed

that distorted lipid analysis (Figure 6.8, appendix).

However, MgSO4 was varied over three levels, as illustrated in Figure 3.25 A. For each level,

the concentrations of NaNO3, CaCl2, and trace elements were distributed using a full factorial

design. For the intermediate concentration of MgSO4, the remaining components were varied

only over half of their total ranges. A center point was located in each of these full factorial

cubes. An additional nine points were space filling distributed over the edges of the cubes. In

total 39 experiments were performed and analyzed, including four reference replicates.

Figure 3.25 B shows the Kriging interpolation based on all data available after the third round

of experiments. The figure shows three contour plots where the third component was fixed to

the front, bottom left corner of the inner cube in Figure 3.25 A. The contour plots clearly show

an interaction of the trace elements with NaNO3 and CaCl2, whereas MgSO4 influences the

lipid productivity only slightly positively. Moreover, an optimal region for the medium compo-

sition can be identified around MgSO4=3.25xRef, CaCl2=1.5xRef, trace elements=2xRef, and

NaNO3=0.3xRef.

Results and discussion 94

Figure 3.25: Representation of parameter space. A: Illustration of experimental design. Blue dots represent the full factorial designs with respect to NaNO3, CaCl2, and trace elements. Red dots indicate additionally added points. The black dots represent the respective center points. The minus and plus signs symbolize the minimum and maximum component concen-tration, B: Contour plots of the Kriging model prediction. Remaining third component was

fixed at MgSO4=1.7×Ref, CaCl2=0.82×Ref, trace elements=1×Ref, or NaNO3=0.37×Ref. Black dots indicate measurement data. Modified from [356].

Refinement of the optimum 3.5.2.4

In a fourth and last round of experiments, twelve experiments were placed around the opti-

mum predicted by the Kriging interpolation. These experiments were planned by sampling

the EI distribution, as described in section 2.21, for maximizing lipid productivity and minimiz-

ing prediction uncertainty of the Kriging model. Further 23 experiments were uniformly dis-

tributed over the parameter space in a random manner, in order to improve prediction accu-

racy also in non-optimal regions. In total 39 experiments were performed, including four ref-

erence experiments (Table 6.5, appendix).

Figure 3.26 shows predictions of the updated Kriging model in the same fashion as in sec-

tion 3.5.2.3. The location of the optimum shifted towards MgSO4=3.25xRef, CaCl2=1.25xRef,

trace elements=2.5xRef, and NaNO3=0.45xRef. For the optimal medium composition, the

Kriging model predicts an increase by a factor of 3.03 ± 0.81 in lipid productivity compared to

the reference medium.

Results and discussion 95

Figure 3.26: Contour plots of the final Kriging model prediction. Remaining third component

was fixed at MgSO4=1.7×Ref, CaCl2=0.82×Ref, trace elements=1×Ref, or NaNO3=0.37×Ref. Black dots indicate measurement data. Modified from [356].

3.5.3 Validation of optimal medium composition

In order to validate the determined optimal medium composition (section 3.5.2.4) and to high-

light potential changes of process kinetics, cultivations using enBBMref and enBBMopt were

carried out (medium composition provided in Table 6.2, appendix). Both processes were

monitored in-depth by sequential harvest of replicate wells from microtiter plate cultivations

(Figure 3.27). To maximize comparability with literature reports, biomass concentration at

harvest was acquired as cell dry weight rather than biovolume in this context.

Figure 3.27: Comparison of the process changes induced by medium optimization. A: enBBMref, B: enBBMopt; 25 °C, 2.5 % (v v-1) CO2, 200 µmol m-2 s-1 PAR, ≥ 85 % relative hu-midity. Error bars represent min/max from biological replicates (n = 2). Modified from [356].

0 24 48 72 96 120 144 168

0

5

10

15

20

25

0 24 48 72 96 120 144 168

0

5

10

15

20

25

OD

75

0 [

-],

nL

C [

% (

w w

-1)]

time [h]

A

0.0

0.3

0.6

0.9

1.2

1.5

NO

3

- [

g L

-1]

OD

75

0 [

-],

nL

C [

% (

w w

-1)]

time [h]

B

0.0

0.3

0.6

0.9

1.2

1.5

NO

3

- [

g L

-1]

Results and discussion 96

Table 3.5: Impact of medium optimization by key performance indicating parameters.

parameter enBBMref enBBMopt

µ [d-1] 1.49 ± 0.06 1.45 ± 0.1

time to nitrate depletion [h] 84 52

delay to onset of lipid synthesis [h] 36 20

CDW specific lipid accumulation rate [% (w w-1) d-1] 4.9 ± 0.5 6.4 ± 0.2

CDW at harvest [g L-1] 4.95 ± 0.06 4.93 ± 0.01

nLC at harvest [% (w w-1)] 10.55 ± 0.35 23.9 ± 1.2

Pvol [mg L-1 d-1] 74 ± 1 169 ± 7

Medium optimization resulted in a series of significant changes in process performance as

summarized in Table 3.5. While the exponential growth rates in both media did not differ sig-

nificantly (p < 0.05), time to nitrogen depletion was 84 h and 52 h for enBBMref and enBBMopt,

respectively. This was due to the reduction of nitrate concentration during medium optimiza-

tion down to 0.45xRef. In the reference process, exponential growth shifted to linear kinetics

reaching an optical density of 4.94 ± 0.06 typically indicating the onset of light limitation and

in clear accordance with prior experiments (section 3.2). This effect was not observed for the

optimized medium before nitrogen depletion. Neutral lipid accumulation started within 36 h

(enBBMref) and 20 h (enBBMopt) after nitrogen limitation, which corresponds to a reduction of

approx. 45 %. Moreover, the biomass specific lipid accumulation rate (estimated by linear fit)

increased by approx. 32 % from 4.87 ± 0.53 % (w w-1) d-1 to 6.43 ± 0.17 % (w w-1) d-1 due to

medium optimization. Most probably, both effects are attributable to the increased availability

of magnesium and calcium ions as well as trace elements in the medium. This might result in

a boost of the enzymatic turnover of lipid synthesis, especially regarding acetyl-CoA carbox-

ylase (Table 3.4).

Alternatively, a kinetic limitation of ion import into the cells at the low concentrations in the

reference medium could be an explanation. Regarding downstream processing, the in-

creased magnesium concentration offers another positive aspect, as it was previously re-

ported to assist flocculation of the cells at high pH [357]. This mechanism is currently being

investigated as an alternative to comparably costly biomass separation by centrifugation.

Most strikingly, cell dry weight at harvest did not differ significantly (p < 0.05) for both media,

despite the nitrate concentration was reduced to 45 % in enBBMopt. This indicates the nitrate

specific biomass yield as being a function of the initial nitrate availability, a phenomenon that

has recently been recognized and discussed for a fairly comparable Chlorella process [300].

Results and discussion 97

Together with an increase in the neutral lipid content from 10.55 ± 0.35 % (w w-1) to

23.9 ± 1.2 % (w w-1), this translated into a 2.3-fold increase of volumetric productivity up to

169 ± 7 mg L-1 d-1.

Besides the evaluation of productivity related issues, the relative fatty acid composition of the

neutral lipid product fraction was compared by GC-ToF-MS (Figure 3.28).

Figure 3.28: Relative composition of the fatty acids from the neutral lipid fraction. Error bars represent min/max from biological replicates (n = 2). Modified from [356].

The obtained fatty acid fingerprints were in clear agreement with previous literature reports

[56] as palmitic, oleic, linoleic and α-linolenic acid made up the major product fractions of

85 % (enBBMref) and 89 % (enBBMopt). There are indications that the fatty acid fingerprint

largely depends on cultivation conditions such as temperature [358], illumination [359], etc.

However, these results demonstrate that changes in the medium composition can also lead

to differences in the fatty acid fingerprints. The fractions of palmitoleic (16:1 Δ9), hexadecadi-

enoic (16:2 Δ7,10), hexadecatrienoic (16:3 Δ7,10,13), stearic (18:0) and linoleic (18:2 Δ9,12) acid

remained nearly unchanged. On the contrary, the proportions of palmitic (16:0) and α-

linolenic (18:3 Δ9,12,15) acid shrank by 22 % and 42 %, respectively, linoleic (18:1 Δ9) acid

increased by 92 % to a total share of 48 ±1.8 % using enBBMopt. With respect to biodiesel

synthesis, this reduction in the polyunsaturated fatty acids’ fraction is clearly advantageous,

increasing the fuel’s oxidative stability (section 1.3.2).

enBBMopt enBBMopt,min

0

25

50

75

100

enBBMopt

enBBMref

rela

tiv

e f

rac

tio

n [

%]

medium

16:0

16:1 9

16:2 7,10

16:3 7,10,13

18:0

18:1 9

18:2 9,12

18:3 9,12,15

Results and discussion 98

3.5.4 Final medium simplification

In section 3.5.2, several input variables were identified to be ‘non-relevant’ and thus kept at

the respective reference values throughout the whole study. However, for MES and especial-

ly for EDTA a negative, but still non-significant trend was observed. Besides economic as-

pects, culture media should only contain the necessary ingredients in appropriate concentra-

tions to ensure a high nutrient usage efficiency. Thus, an additional variant, in the following

denoted as enBBMopt,min, was investigated. Here, the concentration of all ‘non-relevant’ com-

ponents was set to the respective minimum value during screening analysis. In particular,

this included the complete omission of MES buffer and the chelator EDTA as well as NaCl

(medium composition provided in Table 6.2, appendix) while phosphate availability was re-

duced to 0.125xRef.

In comparison with the results using enBBMref as well as enBBMopt, these adaptations did not

change the overall obtained cell dry weight significantly (p < 0.05) but led to an increase of

the neutral lipid from 23.9 ± 1.2 % (w w-1) to 30.1 ± 1.6 % (w w-1) while the respective fatty

acid fingerprint remained unchanged in comparison to enBBMopt (Figure 6.9, appendix). The

resulting volumetric productivity increased from 169 ± 7 mg L-1 d-1 to 212 ± 11 mg L-1 d-1 and

thus represents in total a 2.9-fold improvement compared to reference. Leaving out EDTA

and especially the MES buffer considerably reduces the medium costs, so that the price per

liter is lowered by 96 %. Most probably, MES is not required as the phosphate salts offer suf-

ficient pH stabilization capacity. Although EDTA is commonly used as a metal chelator to

improve long-term stability of algae cultivation media, the results clearly indicate that its us-

age is not beneficial for this specific application. Moreover, the reduction of phosphate con-

centration to 12.5 % is advantageous for large-scale application where the recovery of ex-

cess nutrients to prevent overfertilization by wastewater is an important economic aspect.

Yet, these results clearly confirm the validity of the initial screening analysis.

3.5.5 Assessment of achieved volumetric productivity

In the last decade, numerous studies addressed the lipid production of diverse C. vulgaris

strains in different laboratory-scale batch processes [17, 28, 39-41, 49, 65, 325, 358-363].

Among these, the average volumetric productivity was approx. 51 ± 36 mg L-1 d-1 and thus

comparable to the achieved value of 74 ± 1 mg L-1 d-1 using the enBBMref. However, the re-

ported values exhibit a wide spread and it has to be assumed that these differences do not

only originate from the different strains used, but from process conditions and reactor design

as well. Some studies report productivities in the range of 130 mg L-1 d-1 when cultivating

C. vulgaris in laboratory-scale batch processes with optimized nitrogen availability [28, 362].

Unfortunately, it is not generally clarified if productivities refer to either neutral lipid or total

Results and discussion 99

lipid content. In the study at hand, the volumetric productivity of neutral lipids of up to

212 ±11 mg L-1 d-1 clearly exceeds previous reports and thereby underlines the importance of

medium optimization not only for nitrate as commonly done, but especially for the concentra-

tions of further salts and trace ions.

3.5.6 Conclusions

Integrating experimental design, the previously developed technologies and methodologies

were merged into a blueprint framework for the accelerated development of phototrophic

bioprocesses. This strategy is very efficient in terms of time and material, by incorporating

state-of-the-art phototrophic cultivation and analytics with higher throughput that is closely

linked to sophisticated experimental design strategies.

Taking neutral lipids production by the unicellular microalga C. vulgaris as a model process,

the cultivation medium was optimized towards volumetric productivity. Fractional and full fac-

torial designs in combination with Kriging-based approaches for data analysis, visualization

and experimental design allowed an efficient and effective optimization. The optimized pro-

cess has an approx. threefold increased lipid productivity of 212 ± 11 mg L-1 d-1, which was

achieved in only four experimental rounds with one microtiter plate each.

Besides the commonly addressed concentration of the nitrogen source, especially magnesi-

um and calcium ions as well as various trace elements were shown to be of crucial im-

portance. Analysis tools furthermore revealed multi-parameter interactions that could have

been overlooked otherwise. On top of that, the concentration of non-relevant medium com-

ponents was successfully minimized contributing to reducing medium cost. Taking all togeth-

er, a smart combination of microscale phototrophic cultivation with sophisticated design of

experiments led to a tremendous improvement of neutral lipid production with C. vulgaris at

the same time reducing cost for media components by 96 %, while all other process perfor-

mance indicators were kept constant.

Conclusion and outlook 100

4. Conclusion and outlook

The study at hand describes the development of a cultivation device and supporting method-

ologies enabling the accelerated development of phototrophic bioprocesses. In this context,

the complete spectrum from strain maintenance via cultivation to analytics and experimental

design is covered. The presented approach relies on parallelized microscale cultivation and

laboratory automation as key technologies while the neutral lipid producing microalga

C. vulgaris was used as a biological model system throughout all experiments. These lipids

are recognized as promising precursors for biodiesel synthesis.

strain maintenance

To overcome limitations of serial sub-culturing as a common practice for microalgal strain

maintenance, an alternative method based on cryopreservation was established. Although

such protocols already exist, they are rather complex and laborious. The presented method-

ology relies on heterotrophically grown cells for easy generation of high biomass concentra-

tions. The preservation protocol was deviated from literature and substantially simplified so

that no special devices are needed at all and storage can be carried out at -80 °C rather than

using liquid nitrogen.

Exposing the cells to nitrogen starvation before freezing preservation gave them time to fully

develop their robust morphology. Thereby, viabilities in the range of 63 ± 2 % could be

achieved, a level that has so far only been reported if much more complex preservation pro-

tocols were applied. Post-thawing viabilities quantified by comparative growth pattern analy-

sis of the cells before and after freezing preservation. The corresponding cultivation experi-

ments were conducted in MTPs while online monitoring of biomass ensured a minimum of

hands-on-time. Comparing to the well-established plating and counting techniques for viabil-

ity estimation, significantly lower experimental errors (< 5 %) were achieved.

In subsequent phototrophic shake flask cultivations, the cells could be rapidly re-adapted to

light to reproducibly deliver sufficient amounts of biomass for subsequent experiments. Thus,

the developed strain maintenance module can be routinely applied to efficiently generate

vital C. vulgaris biomass from genetically stable stock cultures.

parallelized microscale cultivation

Currently, the main bottleneck in phototrophic bioprocess development is the lack of appro-

priate laboratory-scale photobioreactors for time efficient experimentation. Despite some pro-

totypes have been developed, there is no system allowing for sufficiently parallelized photo-

trophic cultivation under strictly controlled conditions available.

Conclusion and outlook 101

Thus, the core part of this study was the design of an appropriate micro-photobioreactor

which has been carried out in a cooperation project with m2p-labs GmbH and Premosys

GmbH. A specially designed, modular light source was integrated into a commercial microbi-

oreactor system, the BioLector® II while its adjustable light spectrum was tailored towards the

cultivation of green microalgae. The reactor can run up to 48 parallel phototrophic cultiva-

tions in a single MTP while strict control of the process conditions enabled a high comparabil-

ity between the distinct wells of one plate (≈ ± 5 %).

Biomass concentration was monitored by scattered light measurements in a quasi-online

manner at single well resolution. As phase modulated excitation light was used for biomass

monitoring, resulting scattered light could be separated from illumination light despite both

were detected simultaneously. Consequently, distorting the cultivation processes by dark

phases during biomass measurement was prevented successfully as biomass measurement

could be performed without switching illumination off temporarily. The scattered light analyser

provides highly time-resolved and reproducible data along a dynamic range matching the

biomass concentrations typically occurring during phototrophic batch cultivation.

On the basis of an exemplary reference process, the micro-photobioreactor system was

evaluated against a set of established laboratory-scale photobioreactors (illuminated shake

flask, test tube and flat panel reactor).

During the exponential phase, a growth rate of 1.44 ± 0.02 d-1 was observed for the MTP

cultivation being in agreement with the other three systems. Meanwhile the duration of the

exponential phase was correlated to the specific illumination area of the respective system

and thus cells could be maintained in this phase in MTPs until the highest biomass concen-

tration (OD750 = 5.74 ± 0.02). It could thereby be concluded that light supply was best at mi-

croscale. Besides the high specific illumination area, this was due to short mixing times and

shaking-induced turbulent particle trajectories in the MTP wells. With respect to productivity,

microscale cultivation resulted in a biomass concentration of 5.5 ± 0.1 g L-1 with a neutral

lipid content of 13.7 ± 0.1 % (w w-1), being comparable to the shake flask and the MC 1000

process so that similar volumetric productivities were achieved while the lipid fractions’ rela-

tive composition was in good agreement.

In total, MTP cultivation proved a very good scalability to the investigated laboratory-scale

photobioreactors. Excellent comparability was achieved regarding the exponential growth

phase. Despite minor variations with respect to biomass yield and volumetric productivity

were observed, the overall differences were rather small taking the differing scales into ac-

Conclusion and outlook 102

count. Yet, MTP cultivation may be regarded as a valid tool for efficient phototrophic cultiva-

tion at laboratory-scale.

automated product analysis

Parallelized cultivation induces the need to handle the generated samples in a time efficient

manner while the reduced sample amounts available from microscale experimentation have

to be considered. In the context of intracellular neutral lipid accumulation of C. vulgaris, a

staining assay was set up. Applying dimethyl sulfoxide to facilitate dye permeation, the lipid

probe Nile red was used to monitor product accumulation. The assay parameters were opti-

mized with regard to differentiating staining kinetics of cells at different physiological states

and biomass concentrations. To allow for absolute quantification, Nile red fluorescence was

calibrated against gravimetric lipid measurement while analytical errors were as little as

± 2 % on average.

Contrary to established protocols relying on optical density or cell number for biomass ad-

justion across conducted measurements, the biovolume was used. It was shown that optical

signals as well as cell number are not applicable for algae with complex lifecycle as is the

case for C. vulgaris. Due to their mode of replication via vegetative autosporulation, the cells

undergo morphological changes significantly altering their optical properties. Biovolume

measurements are not prone to interference by optical properties of the sample material and

are thus highly preferable.

The protocol was designed to be run on an automated liquid handling platform enabling the

analysis of up to 48 samples per batch within 6.5 h. Assay automation reduced hands-on-

time to a third compared to preparation by hand while errors due to daily fluctuations of dif-

ferent manual operators are eliminated.

Besides the aforementioned improvements, the newly established assay is especially suited

for the analysis of samples from microscale experiments. These typically generate large sets

of samples at reduced volume which cannot be handled by conventional gravimetric meth-

ods. Applying the presented staining technology, these small-scaled samples can easily be

handled circumventing analytics from becoming a bottleneck.

Conclusion and outlook 103

experimental design

With respect to parameter optimization tasks, parallelized cultivation and automation of ana-

lytics cannot fully provide the required experimental throughput and a further necessarily

scale-out driven parallelization is rather costly. Hence, experimental design was considered

as an alternative strategy. In an exemplary application study, the cultivation medium was

optimized towards volumetric lipid productivity. Experimental planning and evaluation were

closely linked to automated medium preparation and product analysis for minimized manual

operation.

Fractional and full factorial design in combination with Kriging-assisted analysis and optimi-

zation approaches allowed an efficient and effective optimization. In total, productivity was

approx. threefold increased to up to 212 ± 11 mg L-1 d-1 conducting as little as four MTP culti-

vation experiments. This value relates to the highest productivities reported in literature so

far. Moreover, the obtained product spectrum had a suitable composition for biodiesel syn-

thesis.

Besides the commonly addressed concentration of the nitrogen source, especially magnesi-

um and calcium ions as well as various trace elements were shown to be of crucial im-

portance. Analysis tools revealed multi-parameter interactions significantly contributing to the

overall improvement of productivity. On the contrary, the concentration of non-relevant medi-

um components such as excess buffer agent and phosphate as well as metal chelator was

successfully minimized contributing to reducing medium cost.

future improvements

Although the presented framework was shown to be suitable to boost the efficiency of photo-

trophic bioprocess development, it may only be seen as an initial blueprint for further pro-

gress to access additional widespread applications.

The biggest potential clearly lies within continued development of the micro-photobioreactor

system. Besides minor improvements in light homogeneity, in a next step the photo module

should be equipped with an extended set of different LEDs. Thereby, dynamic light spectra

and intensity profiles could be realized to enable comprehensive simulation of environmental

conditions like day/night cycles with weather dependent light spectra.

The micro-photobioreactor could be fully integrated into a liquid handling platform as already

well-established for microbial microbioreactors. The resulting capabilities for automated dos-

ing and sampling as well as on deck liquid processing broaden the spectrum of possible ap-

Conclusion and outlook 104

plications. Besides simple manipulation of individual cultivations like sampling, dosing of nu-

trients, etc., fully automated at-line analytics come into range. Additional process integration,

e.g. pre-cultures directly in MTPs, repetitive batch, etc. can be forced to minimize hands-on-

time.

However, there is still only rudimentary knowledge concerning spatially resolved optical path

length distribution and particle trajectories inside the MTP wells at different shaking condi-

tions. Thus, light supply is solely accessible by biological performance at present. Computa-

tional fluid dynamics could provide valuable information to link process engineering to evalu-

ation of biological performance.

In a next step, automated data analysis for screening applications needs to be established.

Taking the aforementioned technologies into account, extensive process data can be gener-

ated within short time. These data are not evaluable by hand on a high detail level. There-

fore, they need to be condensed to a few key performance indicators appropriately describ-

ing the respective screening task in order to allow for fast evaluation by the experimenter.

future applications

Realizing the discussed further developments, a wide spectrum of tasks in the context of

accelerated phototrophic bioprocess development can be accessed. However, the presented

modules strain maintenance, cultivation, analytics and experimental design are tailored to-

wards cultivation of C. vulgaris and may need substantial adaptation if other systems and

tasks are focused. Clearly, parameter optimization as presented here can be conducted in a

very time efficient manner. Moreover, this strategy may be adapted to screen strain libraries

of novel green algae or isolates for biotechnological relevance applying diverse process con-

ditions. Besides such tasks, the platform is suitable to acquire detailed kinetic data of micro-

algal photosynthesis, growth and production under various conditions. These may feed com-

plex mechanistic models for the comprehensive simulation of large-scale plants. Here, the

availability of appropriate datasets is currently limiting, and thus microscale cultivation tech-

nology driven kinetic characterization could efficiently serve model development and refine-

ment.

Overall, it has to be stated that the complexity of processes that can be run within such a

cultivation and analytics framework is primary limited by the technical specifications of the

cultivation unit and its integration depth into laboratory robotics. Thus, these both aspects

indicate the most urgent sites to proceed towards a universal platform for accelerated photo-

trophic bioprocess development.

Literature 105

5. Literature

[1] Bentley, R. W. (2002): Global oil & gas depletion: an overview. Energ Policy 30: 189-205. [2] Bentley, R. W., Mannan, S. A. & Wheeler, S. J. (2007): Assessing the date of the global oil peak: The need to use 2P reserves. Energ Policy 35: 6364-6382. [3] Höök, M. & Tang, X. (2013): Depletion of fossil fuels and anthropogenic climate change—A review. Energ Policy 52: 797-809. [4] Shafiee, S. & Topal, E. (2009): When will fossil fuel reserves be diminished? Energ Policy 37: 181-189. [5] King, D. A. (2004): Climate change science: adapt, mitigate, or ignore? Science 303: 176-177. [6] Organisation for Economic Co-operation and Development (2009): The Bioeconomy to 2030: Designing a Policy Agenda. Paris: OECD Publishing. [7] Birch, K., Levidow, L. & Papaioannou, T. (2010): Sustainable capital? The neoliberalization of nature and knowledge in the European “knowledge-based bio-economy”. Sustainability 2: 2898. [8] Carlson, R. (2007): Laying the foundations for a bio-economy. Syst Synth Biol 1: 109-117. [9] de Besi, M. & McCormick, K. (2015): Towards a bioeconomy in Europe: National, regional and industrial strategies. Sustainability 7: 10461. [10] Anastas, P. T. & Warner, J. C. (2000): Green chemistry: theory and practice. Oxford: Oxford University Press. [11] Sheldon, R. A. (2014): Green and sustainable manufacture of chemicals from biomass: state of the art. Green Chem 16: 950-963. [12] Westlake, D. F. (1963): Comparison of plant productivity. Biol Rev 38: 385-425. [13] Mata, T. M., Martins, A. A. & Caetano, N. S. (2010): Microalgae for biodiesel production and other applications: A review. Renew Sust Energ Rev 14: 217-232. [14] Chisti, Y. (2007): Biodiesel from microalgae. Biotechnol Adv 25: 294-306. [15] Gouveia, L. & Oliveira, A. (2009): Microalgae as a raw material for biofuels production. J Ind Microbiol Biotechnol 36: 269-274. [16] Yang, J., Xu, M., Zhang, X., Hu, Q., Sommerfeld, M. & Chen, Y. (2011): Life-cycle analysis on biodiesel production from microalgae: Water footprint and nutrients balance. Bioresource Technol 102: 159-165. [17] Rodolfi, L., Chini Zittelli, G., Bassi, N., Padovani, G., Biondi, N., Bonini, G. & Tredici, M. R. (2009): Microalgae for oil: Strain selection, induction of lipid synthesis and outdoor mass cultivation in a low-cost photobioreactor. Biotechnol Bioeng 102: 100-112. [18] Wijffels, R. H. & Barbosa, M. J. (2010): An outlook on microalgal biofuels. Science 329: 796-799.

Literature 106

[19] Wijffels, R. H., Barbosa, M. J. & Eppink, M. H. M. (2010): Microalgae for the production of bulk chemicals and biofuels. Biofuel Bioprod Bior 4: 287-295. [20] Pulz, O. & Gross, W. (2004): Valuable products from biotechnology of microalgae. Appl Microbiol Biotechnol 65: 635-648. [21] Borowitzka, M. A. (1995): Microalgae as sources of pharmaceuticals and other biologically active compounds. J Appl Phycol 7: 3-15. [22] Cardozo, K. H. M., Guaratini, T., Barros, M. P., Falcão, V. R., Tonon, A. P., Lopes, N. P., Campos, S., Torres, M. A., Souza, A. O., Colepicolo, P. & Pinto, E. (2007): Metabolites from algae with economical impact. Comp Biochem Phys C 146: 60-78. [23] Norsker, N.-H., Barbosa, M. J., Vermuë, M. H. & Wijffels, R. H. (2011): Microalgal production — A close look at the economics. Biotechnol Adv 29: 24-27. [24] Van Wagenen, J., Holdt, S. L., De Francisci, D., Valverde-Pérez, B., Plósz, B. G. & Angelidaki, I. (2014): Microplate-based method for high-throughput screening of microalgae growth potential. Bioresource Technol 169: 566-572. [25] Ojo, E. O., Auta, H., Baganz, F. & Lye, G. J. (2015): Design and parallelisation of a miniature photobioreactor platform for microalgal culture evaluation and optimisation. Biochem Eng J 103: 93-102. [26] Radmann, E. M., Camerini, F. V., Santos, T. D. & Costa, J. A. V. (2011): Isolation and application of SOX and NOX resistant microalgae in biofixation of CO2 from thermoelectricity plants. Energy Convers Manag 52: 3132-3136. [27] de Morais, M. G. & Costa, J. A. V. (2007): Isolation and selection of microalgae from coal fired thermoelectric power plant for biofixation of carbon dioxide. Energy Convers Manag 48: 2169-2173. [28] Breuer, G., Lamers, P. P., Martens, D. E., Draaisma, R. B. & Wijffels, R. H. (2012): The impact of nitrogen starvation on the dynamics of triacylglycerol accumulation in nine microalgae strains. Bioresource Technol 124: 217-226. [29] Debska, D., Potvin, G., Lan, C. & Zhang, Z. (2010): Effects of medium composition on the growth of Chlorella vulgaris during photobioreactor batch cultivations. J Biobased Mater Bio 4: 68-72. [30] Beijerinck, M. W. (1890): Culturversuche mit Zoochlorellen und anderen niederen Algen. Btg Ztg 45: 725-740. [31] Tomaselli, L. (2004): The microalgal cell. In: Richmond, A. Handbook of microalgal culture: Biotechnology and applied phycology. Oxford: Blackwell Science. [32] Iwamoto, H. (2004): Industrial production of microalgal cell-mass and secondary products - Major industrial species Chlorella. In: Richmond, A. Handbook of microalgal culture: Biotechnology and applied phycology. Oxford: Blackwell Science. [33] Yamamoto, M., Fujishita, M., Hirata, A. & Kawano, S. (2004): Regeneration and maturation of daughter cell walls in the autospore-forming green alga Chlorella vulgaris (Chlorophyta, Trebouxiophyceae). J Plant Res 117: 257-264. [34] Staehelin, A. (1966): Die Ultrastruktur der Zellwand und des Chloroplasten von Chlorella. Zeitschrift für Zellforschung 74: 325-350.

Literature 107

[35] Ting, Y. P., Prince, I. G. & Lawson, F. (1991): Uptake of cadmium and zinc by the alga Chlorella vulgaris: II. Multi-ion situation. Biotechnol Bioeng 37: 445-455. [36] Yamamoto, M., Nozaki, H., Miyazawa, Y., Koide, T. & Kawano, S. (2003): Relationship between presence of a mother cell wall and speciation in the unicellular microalga Nannochloris (Chlorophyta). J Phycol 39: 172-184. [37] Němcová, Y. & Kalina, T. (2000): Cell wall development, microfibril and pyrenoid structure in type strains of Chlorella vulgaris, C. kessleri, C. sorokiniana compared with C. luteoviridis (Trebouxiophyceae, Chlorophyta). Algol Stud 100: 95-105. [38] Becker, W. (2004): Microalgae in human and animal nutrition. In: A., R. Handbook of microalgal culture: Biotechnology and applied phycology. Oxford: Blackwell Science. [39] Yeh, K.-L. & Chang, J.-S. (2012): Effects of cultivation conditions and media composition on cell growth and lipid productivity of indigenous microalga Chlorella vulgaris ESP-31. Bioresource Technol 105: 120-127. [40] Lv, J.-M., Cheng, L.-H., Xu, X.-H., Zhang, L. & Chen, H.-L. (2010): Enhanced lipid production of Chlorella vulgaris by adjustment of cultivation conditions. Bioresource Technol 101: 6797-6804. [41] Griffiths, M., Hille, R. & Harrison, S. L. (2014): The effect of nitrogen limitation on lipid productivity and cell composition in Chlorella vulgaris. Appl Microbiol Biotechnol 98: 2345-2356. [42] Deng, X., Fei, X. & Li, Y. (2011): The effects of nutritional restriction on neutral lipid accumulation in Chlamydomonas and Chlorella. Afr J Microbiol Res 5: 260-270. [43] Guckert, J. B. & Cooksey, K. E. (1990): Triglyceride accumulation and fatty acid profile changes in Chlorella (Chlorophyta) during high pH-induced cell cycle inhibition. J Phycol 26: 72-79. [44] Illman, A. M., Scragg, A. H. & Shales, S. W. (2000): Increase in Chlorella strains calorific values when grown in low nitrogen medium. Enzyme Microb Technol 27: 631-635. [45] Brennan, L. & Owende, P. (2010): Biofuels from microalgae—A review of technologies for production, processing, and extractions of biofuels and co-products. Renew Sust Energ Rev 14: 557-577. [46] Slade, R. & Bauen, A. (2013): Micro-algae cultivation for biofuels: Cost, energy balance, environmental impacts and future prospects. Biomass Bioenerg 53: 29-38. [47] Brányiková, I., Maršálková, B., Doucha, J., Brányik, T., Bišová, K., Zachleder, V. & Vítová, M. (2011): Microalgae - Novel highly efficient starch producers. Biotechnol Bioeng 108: 766-776. [48] Griffiths, D. J. (1965): The accumulation of carbohydrate in Chlorella vulgaris under heterotrophic conditions. Ann Bot 29: 347-357. [49] Liang, Y., Sarkany, N. & Cui, Y. (2009): Biomass and lipid productivities of Chlorella vulgaris under autotrophic, heterotrophic and mixotrophic growth conditions. Biotechnol Lett 31: 1043-1049.

Literature 108

[50] Mizuno, Y., Sato, A., Watanabe, K., Hirata, A., Takeshita, T., Ota, S., Sato, N., Zachleder, V., Tsuzuki, M. & Kawano, S. (2013): Sequential accumulation of starch and lipid induced by sulfur deficiency in Chlorella and Parachlorella species. Bioresource Technol 129: 150-155. [51] Choix, F. J., de-Bashan, L. E. & Bashan, Y. (2012): Enhanced accumulation of starch and total carbohydrates in alginate-immobilized Chlorella spp. induced by Azospirillum brasilense: I. Autotrophic conditions. Enzyme Microb Technol 51: 294-299. [52] Hariskos, I. & Posten, C. (2014): Biorefinery of microalgae – opportunities and constraints for different production scenarios. Biotechnol J 9: 739-752. [53] Becker, E. W. (2007): Micro-algae as a source of protein. Biotechnol Adv 25: 207-210. [54] Tokuşoglu, Ö. & Ünal, M. K. (2003): Biomass nutrient profiles of three microalgae: Spirulina platensis, Chlorella vulgaris, and Isochrisis galbana. J Food Sci 68: 1144-1148. [55] Spolaore, P., Joannis-Cassan, C., Duran, E. & Isambert, A. (2006): Commercial applications of microalgae. J Biosci Bioeng 101: 87-96. [56] Petkov, G. & Garcia, G. (2007): Which are fatty acids of the green alga Chlorella? Biochem Syst Ecol 35: 281-285. [57] Ötles, S. & Pire, R. (2001): Fatty acid composition of Chlorella and Spirulina microalgae species. J AOAC Int 84: 1708-1714. [58] Amaro, H. M., Guedes, A. C. & Malcata, F. X. (2011): Advances and perspectives in using microalgae to produce biodiesel. Appl Energ 88: 3402-3410. [59] Safi, C., Zebib, B., Merah, O., Pontalier, P.-Y. & Vaca-Garcia, C. (2014): Morphology, composition, production, processing and applications of Chlorella vulgaris: A review. Renew Sust Energ Rev 35: 265-278. [60] Mulbry, W., Westhead, E. K., Pizarro, C. & Sikora, L. (2005): Recycling of manure nutrients: use of algal biomass from dairy manure treatment as a slow release fertilizer. Bioresource Technol 96: 451-458. [61] Radmer, R. J. (1996): Algal diversity and commercial algal products. BioScience 46: 263-270. [62] Morimoto, T., Nagatsu, A., Murakami, N., Sakakibara, J., Tokuda, H., Nishino, H. & Iwashima, A. (1995): Anti-tumour-promoting glyceroglycolipids from the green alga, Chlorella vulgaris. Phytochemistry 40: 1433-1437. [63] Mendes, R. L., Nobre, B. P., Cardoso, M. T., Pereira, A. P. & Palavra, A. F. (2003): Supercritical carbon dioxide extraction of compounds with pharmaceutical importance from microalgae. Inorg Chim Acta 356: 328-334. [64] Martínez, F., Ascaso, C. & Orús, M. I. (1991): Morphometric and stereologic analysis of Chlorella vulgaris under heterotrophic growth conditions. Ann Bot 67: 239-245. [65] Mallick, N., Mandal, S., Singh, A. K., Bishai, M. & Dash, A. (2012): Green microalga Chlorella vulgaris as a potential feedstock for biodiesel. J Chem Technol Biotechnol 87: 137-145.

Literature 109

[66] Chan, C. C. (2007): The state of the art of electric, hybrid, and fuel cell vehicles. P IEEE 95: 704-718. [67] Pollet, B. G., Staffell, I. & Shang, J. L. (2012): Current status of hybrid, battery and fuel cell electric vehicles: From electrochemistry to market prospects. Electrochim Acta 84: 235-249. [68] Miao, X. & Wu, Q. (2006): Biodiesel production from heterotrophic microalgal oil. Bioresource Technol 97: 841-846. [69] Gui, M. M., Lee, K. T. & Bhatia, S. (2008): Feasibility of edible oil vs. non-edible oil vs. waste edible oil as biodiesel feedstock. Energy 33: 1646-1653. [70] Weyer, K. M., Bush, D. R., Darzins, A. & Willson, B. D. (2010): Theoretical maximum algal oil production. Bioenerg Res 3: 204-213. [71] Ahmad, A. L., Yasin, N. H. M., Derek, C. J. C. & Lim, J. K. (2011): Microalgae as a sustainable energy source for biodiesel production: A review. Renew Sust Energ Rev 15: 584-593. [72] Klein, R. M. & Cronquist, A. (1967): A Consideration of the evolutionary and taxonomic significance of some biochemical, micromorphological, and physiological characters in the Thallophytes. Q Rev Biol 42: 108-296. [73] Packer, A., Li, Y., Andersen, T., Hu, Q., Kuang, Y. & Sommerfeld, M. (2011): Growth and neutral lipid synthesis in green microalgae: A mathematical model. Bioresource Technol 102: 111-117. [74] Roessler, P. G. (1990): Environmental control of glycolipid metabolism in microalgae: Commercial implication and future research directions. J Phycol 26: 393-399. [75] Berman-Frank, I. & Dubinsky, Z. (1999): Balanced growth in aquatic plants: Myth or reality? Phytoplankton use the imbalance between carbon assimilation and biomass production to their strategic advantage. BioScience 49: 29-37. [76] Li, Z., Wakao, S., Fischer, B. B. & Niyogi, K. K. (2009): Sensing and responding to excess light. Annu Rev Plant Biol 60: 239-260. [77] Müller, P., Li, X.-P. & Niyogi, K. K. (2001): Non-photochemical quenching. A response to excess light energy. Plant Physiol 125: 1558-1566. [78] Maréchal, E., Block, M. A., Dorne, A.-J., Douce, R. & Joyard, J. (1997): Lipid synthesis and metabolism in the plastid envelope. Physiol Plantarum 100: 65-77. [79] Scott, S. A., Davey, M. P., Dennis, J. S., Horst, I., Howe, C. J., Lea-Smith, D. J. & Smith, A. G. (2010): Biodiesel from algae: challenges and prospects. Curr Opin Biotech 21: 277-286. [80] Dillschneider, R. (2014): Die Effizienz der Kultivierung von Mikroalgen zur Biodieselgewinnung - Prozessentwicklung auf der Grundlage von Energiebilanzierung, Simulation und Integration modellprädiktiver Regelungskonzepte. Karlsruhe: Karlsruhe Institute of Technology. Dissertation. [81] Radakovits, R., Jinkerson, R. E., Darzins, A. & Posewitz, M. C. (2010): Genetic engineering of algae for enhanced biofuel production. Eukaryot Cell 9: 486-501.

Literature 110

[82] Berg, J. M., Tymoczko, J. L., Gatto, G. J. & Stryer, L. (2015): Biochemistry. New York: W. H. Freeman and Company. [83] Khozin-Goldberg, I. & Cohen, Z. (2011): Unraveling algal lipid metabolism: Recent advances in gene identification. Biochimie 93: 91-100. [84] Ohlrogge, J. & Browse, J. (1995): Lipid biosynthesis. Plant Cell 7: 957-970. [85] Guschina, I. A. & Harwood, J. L. (2006): Lipids and lipid metabolism in eukaryotic algae. Prog Lipid Res 45: 160-186. [86] Liu, B. & Benning, C. (2013): Lipid metabolism in microalgae distinguishes itself. Curr Opin Biotech 24: 300-309. [87] Hu, Q., Sommerfeld, M., Jarvis, E., Ghirardi, M., Posewitz, M., Seibert, M. & Darzins, A. (2008): Microalgal triacylglycerols as feedstocks for biofuel production: perspectives and advances. Plant J 54: 621-639. [88] Schlagermann, P., Göttlicher, G., Dillschneider, R., Rosello-Sastre, R. & Posten, C. (2012): Composition of algal oil and its potential as biofuel. J Combust 2012: 14. [89] Merchant, S. S., Kropat, J., Liu, B., Shaw, J. & Warakanont, J. (2012): TAG, You’re it! Chlamydomonas as a reference organism for understanding algal triacylglycerol accumulation. Curr Opin Biotech 23: 352-363. [90] Carvalho, A. P., Meireles, L. A. & Malcata, F. X. (2006): Microalgal reactors: A review of enclosed system designs and performances. Biotechnol Progr 22: 1490-1506. [91] Posten, C. (2009): Design principles of photo-bioreactors for cultivation of microalgae. Eng Life Sci 9: 165-177. [92] Ugwu, C. U., Aoyagi, H. & Uchiyama, H. (2008): Photobioreactors for mass cultivation of algae. Bioresource Technol 99: 4021-4028. [93] Singh, A., Nigam, P. S. & Murphy, J. D. (2011): Mechanism and challenges in commercialisation of algal biofuels. Bioresource Technol 102: 26-34. [94] Wijffels, R. H., Kruse, O. & Hellingwerf, K. J. (2013): Potential of industrial biotechnology with cyanobacteria and eukaryotic microalgae. Curr Opin Biotech 24: 405-413. [95] Salim, S., Bosma, R., Vermuë, M. H. & Wijffels, R. H. (2010): Harvesting of microalgae by bio-flocculation. J Appl Phycol 23: 849-855. [96] Uduman, N., Qi, Y., Danquah, M. K., Forde, G. M. & Hoadley, A. (2010): Dewatering of microalgal cultures: A major bottleneck to algae-based fuels. IRESR 2: 012701. [97] Sim, T. S., Goh, A. & Becker, E. W. (1988): Comparison of centrifugation, dissolved air flotation and drum filtration techniques for harvesting sewage-grown algae. Biomass 16: 51-62. [98] Rawat, I., Ranjith Kumar, R., Mutanda, T. & Bux, F. (2011): Dual role of microalgae: Phycoremediation of domestic wastewater and biomass production for sustainable biofuels production. Appl Energ 88: 3411-3424.

Literature 111

[99] Chen, C.-Y., Yeh, K.-L., Aisyah, R., Lee, D.-J. & Chang, J.-S. (2011): Cultivation, photobioreactor design and harvesting of microalgae for biodiesel production: A critical review. Bioresource Technol 102: 71-81. [100] Pragya, N., Pandey, K. K. & Sahoo, P. K. (2013): A review on harvesting, oil extraction and biofuels production technologies from microalgae. Renew Sust Energ Rev 24: 159-171. [101] Sander, K. & Murthy, G. S. (2010): Life cycle analysis of algae biodiesel. Int J Life Cycle Assess 15: 704-714. [102] Lardon, L., Hélias, A., Sialve, B., Steyer, J.-P. & Bernard, O. (2009): Life-cycle assessment of biodiesel production from microalgae. Environ Sci Technol 43: 6475-6481. [103] Sahena, F., Zaidul, I. S. M., Jinap, S., Karim, A. A., Abbas, K. A., Norulaini, N. A. N. & Omar, A. K. M. (2009): Application of supercritical CO2 in lipid extraction – A review. J Food Eng 95: 240-253. [104] Halim, R., Gladman, B., Danquah, M. K. & Webley, P. A. (2011): Oil extraction from microalgae for biodiesel production. Bioresource Technol 102: 178-185. [105] Gerken, H. G., Donohoe, B. & Knoshaug, E. P. (2013): Enzymatic cell wall degradation of Chlorella vulgaris and other microalgae for biofuels production. Planta 237: 239-253. [106] Zheng, H., Yin, J., Gao, Z., Huang, H., Ji, X. & Dou, C. (2011): Disruption of Chlorella vulgaris cells for the release of biodiesel-producing lipids: A comparison of grinding, ultrasonication, bead milling, enzymatic lysis, and microwaves. Appl Biochem Biotechnol 164: 1215-1224. [107] Demirbas, A. (2009): Progress and recent trends in biodiesel fuels. Energy Convers Manag 50: 14-34. [108] Ma, F. & Hanna, M. A. (1999): Biodiesel production: A review. Bioresource Technol 70: 1-15. [109] Levine, R. B., Pinnarat, T. & Savage, P. E. (2010): Biodiesel production from wet algal biomass through in situ lipid hydrolysis and supercritical transesterification. Energy Fuels 24: 5235-5243. [110] Ehimen, E. A., Sun, Z. F. & Carrington, C. G. (2010): Variables affecting the in situ transesterification of microalgae lipids. Fuel 89: 677-684. [111] Xu, R. & Mi, Y. (2010): Simplifying the process of microalgal biodiesel production through in situ transesterification technology. J Am Oil Chem Soc 88: 91-99. [112] Haas, M. J. & Wagner, K. (2011): Simplifying biodiesel production: The direct or in situ transesterification of algal biomass. Eur J Lipid Sci Technol 113: 1219-1229. [113] Hoekman, S. K., Broch, A., Robbins, C., Ceniceros, E. & Natarajan, M. (2012): Review of biodiesel composition, properties, and specifications. Renew Sust Energ Rev 16: 143-169. [114] Bouriazos, A., Sotiriou, S., Vangelis, C. & Papadogianakis, G. (2010): Catalytic conversions in green aqueous media: Part 4. Selective hydrogenation of polyunsaturated methyl esters of vegetable oils for upgrading biodiesel. J Organomet Chem 695: 327-337.

Literature 112

[115] Wadumesthrige, K., Salley, S. O. & Ng, K. Y. S. (2009): Effects of partial hydrogenation, epoxidation, and hydroxylation on the fuel properties of fatty acid methyl esters. Fuel Process Technol 90: 1292-1299. [116] Zaccheria, F., Psaro, R. & Ravasio, N. (2009): Selective hydrogenation of alternative oils: a useful tool for the production of biofuels. Green Chem 11: 462-465. [117] Albuquerque, M. C. G., Machado, Y. L., Torres, A. E. B., Azevedo, D. C. S., Cavalcante Jr, C. L., Firmiano, L. R. & Parente Jr, E. J. S. (2009): Properties of biodiesel oils formulated using different biomass sources and their blends. Renew Energ 34: 857-859. [118] Morweiser, M., Kruse, O., Hankamer, B. & Posten, C. (2010): Developments and perspectives of photobioreactors for biofuel production. Appl Microbiol Biotechnol 87: 1291-1301. [119] Uyar, B. & Kapucu, N. (2015): Passive temperature control of an outdoor photobioreactor by phase change materials. J Chem Technol Biotechnol 90: 915-920. [120] Janssen, M., Tramper, J., Mur, L. R. & Wijffels, R. H. (2003): Enclosed outdoor photobioreactors: Light regime, photosynthetic efficiency, scale-up, and future prospects. Biotechnol Bioeng 81: 193-210. [121] Balasubramaniyan, B. & Jayaraman, J. (2012): Integrated dairy plant effluent treatment and production of biomass and lipids using micro algae - "Chlorella vulgaris". Int J Chem React Eng 10: A74. [122] Kumar, A., Ergas, S., Yuan, X., Sahu, A., Zhang, Q., Dewulf, J., Malcata, F. X. & van Langenhove, H. (2010): Enhanced CO2 fixation and biofuel production via microalgae: recent developments and future directions. Trends Biotechnol 28: 371-380. [123] Feng, Y., Li, C. & Zhang, D. (2011): Lipid production of Chlorella vulgaris cultured in artificial wastewater medium. Bioresource Technol 102: 101-105. [124] Mussgnug, J. H., Klassen, V., Schlüter, A. & Kruse, O. (2010): Microalgae as substrates for fermentative biogas production in a combined biorefinery concept. J Biotechnol 150: 51-56. [125] Singh, J. & Gu, S. (2010): Commercialization potential of microalgae for biofuels production. Renew Sust Energ Rev 14: 2596-2610. [126] Foley, P. M., Beach, E. S. & Zimmerman, J. B. (2011): Algae as a source of renewable chemicals: opportunities and challenges. Green Chem 13: 1399-1405. [127] Griffiths, M. J. & Harrison, S. T. L. (2009): Lipid productivity as a key characteristic for choosing algal species for biodiesel production. J Appl Phycol 21: 493-507. [128] Choi, Y.-E., Hwang, H., Kim, H.-S., Ahn, J.-W., Jeong, W.-J. & Yang, J.-W. (2013): Comparative proteomics using lipid over-producing or less-producing mutants unravels lipid metabolisms in Chlamydomonas reinhardtii. Bioresource Technol 145: 108-115. [129] Guarnieri, M. T., Nag, A., Yang, S. & Pienkos, P. T. (2013): Proteomic analysis of Chlorella vulgaris: Potential targets for enhanced lipid accumulation. J Proteomics 93: 245-253.

Literature 113

[130] Lv, H., Qu, G., Qi, X., Lu, L., Tian, C. & Ma, Y. (2013): Transcriptome analysis of Chlamydomonas reinhardtii during the process of lipid accumulation. Genomics 101: 229-237. [131] Msanne, J., Xu, D., Konda, A. R., Casas-Mollano, J. A., Awada, T., Cahoon, E. B. & Cerutti, H. (2012): Metabolic and gene expression changes triggered by nitrogen deprivation in the photoautotrophically grown microalgae Chlamydomonas reinhardtii and Coccomyxa sp. C-169. Phytochemistry 75: 50-59. [132] Ndimba, B. K., Ndimba, R. J., Johnson, T. S., Waditee-Sirisattha, R., Baba, M., Sirisattha, S., Shiraiwa, Y., Agrawal, G. K. & Rakwal, R. (2013): Biofuels as a sustainable energy source: An update of the applications of proteomics in bioenergy crops and algae. J Proteomics 93: 234-244. [133] Baroukh, C., Muñoz-Tamayo, R., Steyer, J.-P. & Bernard, O. (2014): DRUM: A new framework for metabolic modeling under non-balanced growth. Application to the carbon metabolism of unicellular microalgae. PLoS ONE 9: e104499. [134] Stevens, D. R. & Purton, S. (1997): Genetic engineering of eucaryotic algae: Progress and prospects. J Phycol 33: 713-722. [135] Aiba, S. (1982): Growth kinetics of photosynthetic microorganisms. In: Microbial Reactions. Berlin, Heidelberg: Springer Berlin Heidelberg. [136] Grobbelaar, J. U. (2008): Factors governing algal growth in photobioreactors: the “open” versus “closed” debate. J Appl Phycol 21: 489-492. [137] Blankenship, R. E. (2014): Molecular mechanisms of photosynthesis. Hoboken: Wiley-Blackwell. [138] Béchet, Q., Shilton, A. & Guieysse, B. (2013): Modeling the effects of light and temperature on algae growth: State of the art and critical assessment for productivity prediction during outdoor cultivation. Biotechnol Adv 31: 1648-1663. [139] Falkowski, P. G. & LaRoche, J. (1991): Acclimation to spectral irradiance in algae. J Phycol 27: 8-14. [140] Ibelings, B. W., Kroon, B. M. A. & Mur, L. R. (1994): Acclimation of photosystem II in a cyanobacterium and a eukaryotic green alga to high and fluctuating photosynthetic photon flux densities, simulating light regimes induced by mixing in lakes. New Phytol 128: 407-424. [141] Falkowski, P. G. & Chen, Y.-B. (2003): Photoacclimation of light harvesting systems in eukaryotic algae. In: Green, B. R. & Parson, W. W. Light-Harvesting Antennas in Photosynthesis. Dordrecht: Springer Netherlands. [142] Suh, I. S. & Lee, C.-G. (2003): Photobioreactor engineering: Design and performance. Biotechnol Bioprocess Eng 8: 313-321. [143] Bitog, J. P., Lee, I. B., Lee, C. G., Kim, K. S., Hwang, H. S., Hong, S. W., Seo, I. H., Kwon, K. S. & Mostafa, E. (2011): Application of computational fluid dynamics for modeling and designing photobioreactors for microalgae production: A review. Comput Electron Agric 76: 131-147. [144] Wang, S.-K., Stiles, A. R., Guo, C. & Liu, C.-Z. (2014): Microalgae cultivation in photobioreactors: An overview of light characteristics. Eng Life Sci 14: 550-559.

Literature 114

[145] Sastre, R., Csögör, Z., Perner-Nochta, I., Fleck-Schneider, P. & Posten, C. (2007): Scale-down of microalgae cultivations in tubular photo-bioreactors—A conceptual approach. J Biosci Bioeng 132: 127-133. [146] Cuaresma, M., Janssen, M., Vílchez, C. & Wijffels, R. H. (2009): Productivity of Chlorella sorokiniana in a short light-path (SLP) panel photobioreactor under high irradiance. Biotechnol Bioeng 104: 352-359. [147] Tredici, M. R. (1999): Photobioreactors. In: Flickinger, M. C. & Drew, S. W. Encyclopedia of Bioprocess Technology: Fermentation, Biocatalysis and Bioseparation. New York: J. Wiley & Sons. [148] Tredici, M. R. & Zittelli, G. C. (1998): Efficiency of sunlight utilization: Tubular versus flat photobioreactors. Biotechnol Bioeng 57: 187-197. [149] Lee, C.-G. & Palsson, B. Ø. (1995): Light emitting diode-based algal photobioreactor with external gas exchange. J Ferment Bioeng 79: 257-263. [150] Luo, H.-P. & Al-Dahhan, M. H. (2004): Analyzing and modeling of photobioreactors by combining first principles of physiology and hydrodynamics. Biotechnol Bioeng 85: 382-393. [151] Chmiel, H. (2011): Bioprozesstechnik. Heidelberg: Spektrum Akademischer Verlag. [152] Doig, S. D., Baganz, F. & Lye, G. J. (2006): High-throughput screening and process optimisation. In: Ratledge, C. & Kristiansen, B. Basic Biotechnology. Cambridge: Cambridge University Press. [153] Wenk, P., Hemmerich, J., Müller, C. & Kensy, F. (2012): High-throughput bioprocess development in shaken microbioreactors. Chem-Ing-Tech 84: 704-714. [154] Bareither, R. & Pollard, D. (2011): A review of advanced small-scale parallel bioreactor technology for accelerated process development: Current state and future need. Biotechnol Progr 27: 2-14. [155] Büchs, J., Lotter, S. & Milbradt, C. (2001): Out-of-phase operating conditions, a hitherto unknown phenomenon in shaking bioreactors. Biochem Eng J 7: 135-141. [156] Puskeiler, R., Kaufmann, K. & Weuster-Botz, D. (2005): Development, parallelization, and automation of a gas-inducing milliliter-scale bioreactor for high-throughput bioprocess design (HTBD). Biotechnol Bioeng 89: 512-523. [157] Long, Q., Liu, X., Yang, Y., Li, L., Harvey, L., McNeil, B. & Bai, Z. (2014): The development and application of high throughput cultivation technology in bioprocess development. J Biotechnol 192, Part B: 323-338. [158] Marques, M. P. C., Cabral, J. M. S. & Fernandes, P. (2009): High throughput in biotechnology: From shake-flasks to fully instrumented microfermentors. Recent pat Biotechnol 3: 124-140. [159] National Research Council (2015): Industrialization of biology: A roadmap to accelerate the advanced manufacturing of chemicals. Washington D.C.,: The National Academies Press. [160] Xie, D. (2012): Using an advanced microfermentor system for strain screening and fermentation optimization. In: Cheng, Q. Microbial Metabolic Engineering: Methods and Protocols. New York: Springer New York.

Literature 115

[161] Micheletti, M. & Lye, G. J. (2006): Microscale bioprocess optimisation. Curr Opin Biotech 17: 611-618. [162] Neubauer, P., Cruz, N., Glauche, F., Junne, S., Knepper, A. & Raven, M. (2013): Consistent development of bioprocesses from microliter cultures to the industrial scale. Eng Life Sci 13: 224-238. [163] Hemmerich, J., Adelantado, N., Barrigón, J. M., Ponte, X., Hörmann, A., Ferrer, P., Kensy, F. & Valero, F. (2014): Comprehensive clone screening and evaluation of fed-batch strategies in a microbioreactor and lab scale stirred tank bioreactor system: application on Pichia pastoris producing Rhizopus oryzae lipase. Microb Cell Fact 13: 36. [164] Huber, R., Ritter, D., Hering, T., Hillmer, A.-K., Kensy, F., Müller, C., Wang, L. & Büchs, J. (2009): Robo-Lector - a novel platform for automated high-throughput cultivations in microtiter plates with high information content. Microb Cell Fact 8: 42. [165] Huber, R., Roth, S., Rahmen, N. & Büchs, J. (2011): Utilizing high-throughput experimentation to enhance specific productivity of an E.coli T7 expression system by phosphate limitation. BMC Biotechnol 11: 22. [166] Radzun, K. A., Wolf, J., Jakob, G., Zhang, E., Stephens, E., Ross, I. & Hankamer, B. (2015): Automated nutrient screening system enables high-throughput optimisation of microalgae production conditions. Biotechnol Biofuels 8: 1-17. [167] Tillich, U. M., Wiolter, N., Schulze, K., Kramer, D., Brödel, O. & Frohme, M. (2014): High-throughput cultivation and screening platform for unicellular phototrophs. BMC Microbiol 14: 239. [168] Unthan, S., Radek, A., Wiechert, W., Oldiges, M. & Noack, S. (2015): Bioprocess automation on a Mini Pilot Plant enables fast quantitative microbial phenotyping. Microb Cell Fact 14: 32. [169] Rohe, P., Venkanna, D., Kleine, B., Freudl, R. & Oldiges, M. (2012): An automated workflow for enhancing microbial bioprocess optimization on a novel microbioreactor platform. Microb Cell Fact 11: 144. [170] Knepper, A., Heiser, M., Glauche, F. & Neubauer, P. (2014): Robotic platform for parallelized cultivation and monitoring of microbial growth parameters in microwell plates. J Lab Autom 19: 593-601. [171] Kunze, M., Lattermann, C., Diederichs, S., Kroutil, W. & Büchs, J. (2014): Minireactor-based high-throughput temperature profiling for the optimization of microbial and enzymatic processes. J Biol Eng 8: 1-18. [172] Heux, S., Poinot, J., Massou, S., Sokol, S. & Portais, J.-C. (2014): A novel platform for automated high-throughput fluxome profiling of metabolic variants. Metab Eng 25: 8-19. [173] Riedlberger, P. & Weuster-Botz, D. (2012): New miniature stirred-tank bioreactors for parallel study of enzymatic biomass hydrolysis. Bioresource Technol 106: 138-146. [174] Vester, A., Hans, M., Hohmann, H.-P. & Weuster-Botz, D. (2009): Discrimination of riboflavin producing Bacillus subtilis strains based on their fed-batch process performances on a millilitre scale. Appl Microbiol Biotechnol 84: 71-76.

Literature 116

[175] Freier, L., Hemmerich, J., Schöler, K., Wiechert, W., Oldiges, M. & von Lieres, E. (2016): Framework for Kriging-based iterative experimental analysis and design: Optimization of secretory protein production in Corynebacterium glutamicum. Eng Life Sci, doi: 10.1002/elsc.201500171. [176] Gernaey, K. V., Baganz, F., Franco-Lara, E., Kensy, F., Krühne, U., Luebberstedt, M., Marx, U., Palmqvist, E., Schmid, A., Schubert, F. & Mandenius, C.-F. (2012): Monitoring and control of microbioreactors: An expert opinion on development needs. Biotechnol J 7: 1308-1314. [177] Betts, J. I. & Baganz, F. (2006): Miniature bioreactors: current practices and future opportunities. Microb Cell Fact 5: 21. [178] Duetz, W. A. (2007): Microtiter plates as mini-bioreactors: miniaturization of fermentation methods. Trends Microbiol 15: 469-475. [179] Doig, S. D., Diep, A. & Baganz, F. (2005): Characterisation of a novel miniaturised bubble column bioreactor for high throughput cell cultivation. Biochem Eng J 23: 97-105. [180] Girard, P., Jordan, M., Tsao, M. & Wurm, F. M. (2001): Small-scale bioreactor system for process development and optimization. Biochem Eng J 7: 117-119. [181] Weuster-Botz, D. (2005): Parallel reactor systems for bioprocess development. In: Kragl, U. Technology Transfer in Biotechnology: From lab to Industry to Production. Heidelberg: Springer. [182] Hortsch, R., Stratmann, A. & Weuster-Botz, D. (2010): New milliliter-scale stirred tank bioreactors for the cultivation of mycelium forming microorganisms. Biotechnol Bioeng 106: 443-451. [183] Hortsch, R. & Weuster-Botz, D. (2010): Milliliter-scale stirred tank reactors for the cultivation of microorganisms. In: Allen I. Laskin, S. S. & Geoffrey, M. G. Advances in Applied Microbiology. London: Academic Press. [184] Riedlberger, P., Brüning, S. & Weuster-Botz, D. (2013): Characterization of stirrers for screening studies of enzymatic biomass hydrolyses on a milliliter scale. Bioproc Biosyst Eng 36: 927-935. [185] Kusterer, A., Krause, C., Kaufmann, K., Arnold, M. & Weuster-Botz, D. (2008): Fully automated single-use stirred-tank bioreactors for parallel microbial cultivations. Bioproc Biosyst Eng 31: 207-215. [186] Szita, N., Boccazzi, P., Zhang, Z., Boyle, P., Sinskey, A. J. & Jensen, K. F. (2005): Development of a multiplexed microbioreactor system for high-throughput bioprocessing. Lab Chip 5: 819-826. [187] Ali, S., Perez-Pardo, M. A., Aucamp, J. P., Craig, A., Bracewell, D. G. & Baganz, F. (2012): Characterization and feasibility of a miniaturized stirred tank bioreactor to perform E. coli high cell density fed-batch fermentations. Biotechnol Progr 28: 66-75. [188] Ge, X., Hanson, M., Shen, H., Kostov, Y., Brorson, K. A., Frey, D. D., Moreira, A. R. & Rao, G. (2006): Validation of an optical sensor-based high-throughput bioreactor system for mammalian cell culture. J Biotechnol 122: 293-306.

Literature 117

[189] Kondragunta, B., Drew, J. L., Brorson, K. A., Moreira, A. R. & Rao, G. (2010): Advances in clone selection using high-throughput bioreactors. Biotechnol Progr 26: 1095-1103. [190] Gill, N. K., Appleton, M., Baganz, F. & Lye, G. J. (2008): Quantification of power consumption and oxygen transfer characteristics of a stirred miniature bioreactor for predictive fermentation scale-up. Biotechnol Bioeng 100: 1144-1155. [191] Marques, M. P. C., Cabral, J. M. S. & Fernandes, P. (2010): Bioprocess scale-up: quest for the parameters to be used as criterion to move from microreactors to lab-scale. J Chem Technol Biotechnol 85: 1184-1198. [192] Fernandes, P. & Cabral, J. M. S. (2006): Microlitre/millilitre shaken bioreactors in fermentative and biotransformation processes – a review. Biocatal Biotransformation 24: 237-252. [193] Kensy, F., John, G. T., Hofmann, B. & Büchs, J. (2005): Characterisation of operation conditions and online monitoring of physiological culture parameters in shaken 24-well microtiter plates. Bioproc Biosyst Eng 28: 75-81. [194] Sieben, M., Giese, H., Grosch, J.-H., Kauffmann, K. & Büchs, J. (2016): Permeability of currently available microtiter plate sealing tapes fail to fulfil the requirements for aerobic microbial cultivation. Biotechnol J, doi: 10.1002/biot.201600054. [195] Panula-Perälä, J., Šiurkus, J., Vasala, A., Wilmanowski, R., Casteleijn, M. G. & Neubauer, P. (2008): Enzyme controlled glucose auto-delivery for high cell density cultivations in microplates and shake flasks. Microb Cell Fact 7: 31. [196] Wilming, A., Bähr, C., Kamerke, C. & Büchs, J. (2014): Fed-batch operation in special microtiter plates: a new method for screening under production conditions. J Ind Microbiol Biotechnol 41: 513-525. [197] Scheidle, M., Jeude, M., Dittrich, B., Denter, S., Kensy, F., Suckow, M., Klee, D. & Büchs, J. (2010): High-throughput screening of Hansenula polymorpha clones in the batch compared with the controlled-release fed-batch mode on a small scale. FEMS Yeast Res 10: 83-92. [198] Funke, M., Buchenauer, A., Schnakenberg, U., Mokwa, W., Diederichs, S., Mertens, A., Müller, C., Kensy, F. & Büchs, J. (2010): Microfluidic BioLector - Microfluidic bioprocess control in microtiter plates. Biotechnol Bioeng 107: 497-505. [199] Blesken, C., Olfers, T., Grimm, A. & Frische, N. (2016): The microfluidic bioreactor for a new era of bioprocess development. Eng Life Sci 16: 190-193. [200] Funke, M., Diederichs, S., Kensy, F., Müller, C. & Büchs, J. (2009): The baffled microtiter plate: Increased oxygen transfer and improved online monitoring in small scale fermentations. Biotechnol Bioeng 103: 1118-1128. [201] Lattermann, C., Funke, M., Hansen, S., Diederichs, S. & Büchs, J. (2014): Cross-section perimeter is a suitable parameter to describe the effects of different baffle geometries in shaken microtiter plates. J Biol Eng 8: 1-10. [202] Duetz, W. A., Kuhner, M. & Lohser, R. (2006): Microbial and cell growth in microtiter plates. Genet Eng News 26: 44-47.

Literature 118

[203] Grünberger, A., Wiechert, W. & Kohlheyer, D. (2014): Single-cell microfluidics: opportunity for bioprocess development. Curr Opin Biotech 29: 15-23. [204] Bambot, S. B., Holavanahali, R., Lakowicz, J. R., Carter, G. M. & Rao, G. (1994): Phase fluorometric sterilizable optical oxygen sensor. Biotechnol Bioeng 43: 1139-1145. [205] Huber, C., Klimant, I., Krause, C., Werner, T., Mayr, T. & Wolfbeis, S. O. (2000): Optical sensor for seawater salinity. Fresen J Anal Chem 368: 196-202. [206] John, G. T., Klimant, I., Wittmann, C. & Heinzle, E. (2003): Integrated optical sensing of dissolved oxygen in microtiter plates: A novel tool for microbial cultivation. Biotechnol Bioeng 81: 829-836. [207] Lefèvre, N., Ciabrini, J. P., Michard, G., Brient, B., DuChaffaut, M. & Merlivat, L. (1993): A new optical sensor for pCO2 measurements in seawater. Mar Chem 42: 189-198. [208] Liebsch, G., Klimant, I., Frank, B., Holst, G. & Wolfbeis, O. S. (2000): Luminescence lifetime imaging of oxygen, pH, and carbon dioxide distribution using optical sensors. Appl. Spectrosc. 54: 548-559. [209] Kocincová, A. S., Nagl, S., Arain, S., Krause, C., Borisov, S. M., Arnold, M. & Wolfbeis, O. S. (2008): Multiplex bacterial growth monitoring in 24-well microplates using a dual optical sensor for dissolved oxygen and pH. Biotechnol Bioeng 100: 430-438. [210] Arain, S., John, G. T., Krause, C., Gerlach, J., Wolfbeis, O. S. & Klimant, I. (2006): Characterization of microtiterplates with integrated optical sensors for oxygen and pH, and their applications to enzyme activity screening, respirometry, and toxicological assays. Sensors and Actuators B: Chemical 113: 639-648. [211] Luchterhand, B., Nolten, J., Hafizovic, S., Schlepütz, T., Wewetzer, S. J., Pach, E., Meier, K., Wandrey, G. & Büchs, J. (2015): Newly designed and validated impedance spectroscopy setup in microtiter plates successfully monitors viable biomass online. Biotechnol J 10: 1259-1268. [212] Samorski, M., Müller-Newen, G. & Büchs, J. (2005): Quasi-continuous combined scattered light and fluorescence measurements: A novel measurement technique for shaken microtiter plates. Biotechnol Bioeng 92: 61-68. [213] Zimmermann, H. F. & Raebiger, T. (2006): Evaluation of the applicability of backscattered light measurements to the determination of microbial cell densities in microtiter plates. Anal Bioanal Chem 386: 2245-2247. [214] Kirk, T. V. & Szita, N. (2013): Oxygen transfer characteristics of miniaturized bioreactor systems. Biotechnol Bioeng 110: 1005-1019. [215] Kensy, F., Zang, E., Faulhammer, C., Tan, R.-K. & Büchs, J. (2009): Validation of a high-throughput fermentation system based on online monitoring of biomass and fluorescence in continuously shaken microtiter plates. Microb Cell Fact 8: 31. [216] Kunze, M., Roth, S., Gartz, E. & Büchs, J. (2014): Pitfalls in optical on-line monitoring for high-throughput screening of microbial systems. Microb Cell Fact 13: 53. [217] Griffiths, M. J., Garcin, C., van Hille, R. P. & Harrison, S. T. L. (2011): Interference by pigment in the estimation of microalgal biomass concentration by optical density. J Microbiol Meth 85: 119-123.

Literature 119

[218] Kensy, F., Zimmermann, H. F., Knabben, I., Anderlei, T., Trauthwein, H., Dingerdissen, U. & Büchs, J. (2005): Oxygen transfer phenomena in 48-well microtiter plates: Determination by optical monitoring of sulfite oxidation and verification by real-time measurement during microbial growth. Biotechnol Bioeng 89: 698-708. [219] Flitsch, D., Krabbe, S., Ladner, T., Beckers, M., Schilling, J., Mahr, S., Conrath, U., Schomburg, W. K. & Büchs, J. (2016): Respiration activity monitoring system for any individual well of a 48-well microtiter plate. J Biol Eng 10: 14. [220] Moser, F., Broers, N. J., Hartmans, S., Tamsir, A., Kerkman, R., Roubos, J. A., Bovenberg, R. & Voigt, C. A. (2012): Genetic circuit performance under conditions relevant for industrial bioreactors. ACS Synth Biol 1: 555-564. [221] Isett, K., George, H., Herber, W. & Amanullah, A. (2007): Twenty-four-well plate miniature bioreactor high-throughput system: Assessment for microbial cultivations. Biotechnol Bioeng 98: 1017-1028. [222] Islam, R. S., Tisi, D., Levy, M. S. & Lye, G. J. (2008): Scale-up of Escherichia coli growth and recombinant protein expression conditions from microwell to laboratory and pilot scale based on matched kLa. Biotechnol Bioeng 99: 1128-1139. [223] Bareither, R., Bargh, N., Oakeshott, R., Watts, K. & Pollard, D. (2013): Automated disposable small scale reactor for high throughput bioprocess development: A proof of concept study. Biotechnol Bioeng 110: 3126-3138. [224] Barrett, T. A., Wu, A., Zhang, H., Levy, M. S. & Lye, G. J. (2010): Microwell engineering characterization for mammalian cell culture process development. Biotechnol Bioeng 105: 260-275. [225] Hemmerich, J., Adelantado, N., Barrigón, J. M., Ponte, X., Hörmann, A. & Ferrer, P. (2014): Comprehensive clone screening and evaluation of fed-batch strategies in a microbioreactor and lab scale stirred tank bioreactor system: Application on Pichia pastoris producing Rhizopus oryzae lipase. Microb Cell Fact 13: [226] Kensy, F., Engelbrecht, C. & Büchs, J. (2009): Scale-up from microtiter plate to laboratory fermenter: evaluation by online monitoring techniques of growth and protein expression in Escherichia coli and Hansenula polymorpha fermentations. Microb Cell Fact 8: 68. [227] Minas, W., Bailey, J. E. & Duetz, W. (2000): Streptomycetes in micro-cultures: Growth, production of secondary metabolites, and storage and retrieval in the 96–well format. Antonie van Leeuwenhoek 78: 297-305. [228] Hsu, W.-T., Aulakh, R. P. S., Traul, D. L. & Yuk, I. H. (2012): Advanced microscale bioreactor system: a representative scale-down model for bench-top bioreactors. Cytotechnology 64: 667-678. [229] Chen, A., Chitta, R., Chang, D. & Amanullah, A. (2009): Twenty-four well plate miniature bioreactor system as a scale-down model for cell culture process development. Biotechnol Bioeng 102: 148-160. [230] Sani, M. H. (2016): Evaluation of microwell based systems and miniature bioreactors for rapid cell culture bioprocess development and scale-up. London: University College London, Dissertation.

Literature 120

[231] Smith, A. (2002): Screening for drug discovery: The leading question. Nature 418: 453-459. [232] Nettekoven, M. & Thomas, A. W. (2002): Accelerating drug discovery by integrative implementation of laboratory automation in the work flow. Curr Med Chem 9: 2179-2190. [233] Lorenz, M. G. O. (2004): Liquid-handling robotic workstations for functional genomics. JALA 9: 262-267. [234] Sonnleitner, B. (1997): Bioprocess automation and bioprocess design. J Biotechnol 52: 175-179. [235] Nealon, A. J., Willson, K. E., Pickering, S. C. R., Clayton, T. M., O'Kennedy, R. D., Titchener-Hooker, N. J. & Lye, G. J. (2005): Use of operating windows in the assessment of integrated robotic systems for the measurement of bioprocess kinetics. Biotechnol Progr 21: 283-291. [236] Hemmerich, J. & Kensy, F. (2013): Automation of microbioreactors: Operating 48 parallel fed-batch fermentations at microscale. Bioprocess Int 11: 68-76. [237] Zuleta, I. A., Aranda-Díaz, A., Li, H. & El-Samad, H. (2014): Dynamic characterization of growth and gene expression using high-throughput automated flow cytometry. Nature Methods 11: 443-448. [238] Moore, K. W., Newman, R., Chan, G. K. Y., Leech, C., Allison, K., Coulson, J. & Simpson, P. B. (2007): Implementation of a high specification dual-arm robotic platform to meet flexible screening needs. JALA 12: 115-123. [239] Lee, K.-M. & Gilmore, D. F. (2006): Statistical experimental design for bioprocess modeling and optimization analysis. Appl Biochem Biotechnol 135: 101-115. [240] Fisher, R. A. (1935): The Design of Experiments. Edinborough: Oliver and Boyd. [241] Kumar, V., Bhalla, A. & Rathore, A. S. (2014): Design of experiments applications in bioprocessing: Concepts and approach. Biotechnol Progr 30: 86-99. [242] Telen, D., Houska, B., Logist, F., Van Derlinden, E., Diehl, M. & Van Impe, J. (2013): Optimal experiment design under process noise using Riccati differential equations. J Process Contr 23: 613-629. [243] Mandenius, C.-F. & Brundin, A. (2008): Bioprocess optimization using Design-of-Experiments methodology. Biotechnol Progr 24: 1191-1203. [244] Kennedy, M. J., Reader, S. L. & Davies, R. J. (1994): The kinetics of developing fermentation media. Process Biochem 29: 529-534. [245] Montgomery, D. C. (2012): Design and Analysis of Experiments. New York: John Wiley & Sons. [246] Pabari, R. M. & Ramtoola, Z. (2012): Application of face centred central composite design to optimise compression force and tablet diameter for the formulation of mechanically strong and fast disintegrating orodispersible tablets. Int J Pharm 430: 18-25. [247] Rozas, O., Contreras, D., Mondaca, M. A., Pérez-Moya, M. & Mansilla, H. D. (2010): Experimental design of Fenton and photo-Fenton reactions for the treatment of ampicillin solutions. J Hazard Mater 177: 1025-1030.

Literature 121

[248] Weuster-Botz, D. (2000): Experimental design for fermentation media development: Statistical design or global random search? J Biosci Bioeng 90: 473-483. [249] Franco-Lara, E., Link, H. & Weuster-Botz, D. (2006): Evaluation of artificial neural networks for modelling and optimization of medium composition with a genetic algorithm. Process Biochem 41: 2200-2206. [250] Link, H. & Weuster-Botz, D. (2006): Genetic algorithm for multi-objective experimental optimization. Bioproc Biosyst Eng 29: 385-390. [251] Rohe, P. (2012): Prozessnahe Hochdurchsatzoptimierung der heterologen Proteinproduktion in alternativen Wirtsorganismen. Jülich: RWTH Aachen University, Dissertation. [252] De Cock, D. R. (2003): Kriging as an alternative to polynomial regression in response surface analysis. Ames: Iowa State University, Dissertation. [253] Bold, H. C. (1949): The morphology of Chlamydomonas chlamydogama, sp. nov. B Torrey Bot Club 76: 101-108. [254] Nakanishi, K., Deuchi, K. & Kuwano, K. (2012): Cryopreservation of four valuable strains of microalgae, including viability and characteristics during 15 years of cryostorage. J Appl Phycol 24: 1381-1385. [255] Graham, M. D. (2003): The Coulter principle: Foundation of an industry. JALA 8: 72-81. [256] Axelsson, M. & Gentili, F. (2014): A single-step method for rapid extraction of total lipids from green microalgae. PLoS ONE 9: e89643. [257] Chen, W., Zhang, C., Song, L., Sommerfeld, M. & Hu, Q. (2009): A high throughput Nile red method for quantitative measurement of neutral lipids in microalgae. J Microbiol Meth 77: 41-47. [258] Morschett, H., Wiechert, W. & Oldiges, M. (2016): Automation of a Nile red staining assay enables high throughput quantification of microalgal lipid production. Microb Cell Fact 15: 34. [259] Paczia, N., Nilgen, A., Lehmann, T., Gätgens, J., Wiechert, W. & Noack, S. (2012): Extensive exometabolome analysis reveals extended overflow metabolism in various microorganisms. Microb Cell Fact 11: 122. [260] Hummel, J., Strehmel, N., Selbig, J., Walther, D. & Kopka, J. (2010): Decision tree supported substructure prediction of metabolites from GC-MS profiles. Metabolomics 6: 322-333. [261] Myers, R. H., Montgomery, D. C. & Anderson-Cook, C. M. (2016): Response surface methodology: Process and product optimization using designed experiments. Hoboken: John Wiley & Sons. [262] Cressie, N. (2015): Statistics for spatial data. Hoboken: Wiley-Intersciences. [263] Jones, D. R., Schonlau, M. & Welch, W. J. (1998): Efficient global optimization of expensive black-box functions. J Glob Optim 13: 455-492. [264] Jones, D. R. (2001): A taxonomy of global optimization methods based on response surfaces. J Glob Optim 21: 345-383.

Literature 122

[265] Freier, L. & von Lieres, E. (2016): Multi-objective global optimization (MOGO): Algorithm and case study in gradient elution chromatography. Biotechnol J, doi: 10.1002/biot.201600613. [266] Haario, H., Laine, M., Mira, A. & Saksman, E. (2006): DRAM: Efficient adaptive MCMC. Statistics and Computing 16: 339-354. [267] Lorenz, M., Friedl, T. & Day, J. G. (2005): Perpetual maintenance of actively metabolizing microalgal cultures. In: Andersen, R. A. Algal culturing techniques. New York: Academic Press. [268] Day, J. G., Benson, E. E., Harding, K., Knowles, B., Idowu, M., Bremner, D., Santos, L., Santos, F., Friedl, T., Lorenz, M., Lukesova, A., Elster, J., Lukavsky, J., Herdman, M., Rippka, R. & Hall, T. (2005): Cryopreservation and conservation of microalgae: the development of a pan-european scientific and biotechnological resource (the COBRA project). Cryo-Lett 26: 231-238. [269] Prakash, O., Nimonkar, Y. & Shouche, Y. S. (2013): Practice and prospects of microbial preservation. FEMS Microbiol Lett 339: 1-9. [270] Day, J. G. & Harding, K. (2008): Cryopreservation of algae. In: Reed, B. M. Plant cryopreservation: A practical guide. New York: Springer Science & Business Media. [271] Taylor, R. & Fletcher, R. L. (1999): Cryopreservation of eukaryotic algae – a review of methodologies. J Appl Phycol 10: 481-501. [272] Karlsson, J. O. M. & Toner, M. (1996): Long-term storage of tissues by cryopreservation: critical issues. Biomaterials 17: 243-256. [273] Day, J. G. (2004): Cryopreservation: fundamentals, mechanisms of damage on freezing/thawing and application in culture collections. Nova Hedwigia 79: 191-205. [274] Müller, J., Day, J. G., Harding, K., Hepperle, D., Lorenz, M. & Friedl, T. (2007): Assessing genetic stability of a range of terrestrial microalgae after cryopreservation using amplified fragment length polymorphism (AFLP). Am J Bot 94: 799-808. [275] Bodas, K., Brennig, C., Diller, K. R. & Brand, J. J. (1995): Cryopreservation of blue-green and eukaryotic algae in the Culture Collection at the University-of-Texas at Austin. Cryo-Lett 16: 267-274. [276] Hirata, K., Phunchindawan, M., Tukamoto, J., Goda, S. & Miyamoto, K. (1996): Cryopreservation of microalgae using encapsulation-dehydration. Cryo-Lett 17: 321-328. [277] Mortain-Bertrand, A., Etchart, F. & Boucaud, M.-T. d. (1996): A method for the cryopreservation of Dunaliella salina (Chlorophyceae): Effect of glycerol and cold adaption 1. J Phycol 32: 346-352. [278] Gilboa, A. & Ben-Amotz, A. (1979): An improved method for rapid assaying of viability of cryopreserved unicellular algae. Plant Sci Lett 14: 317-320. [279] Armstrong Lumley, M., Burgess, R., Billingham, L. J., McDonald, D. F. & Milligan, D. W. (1997): Colony counting is a major source of variation in CFU-GM results between centres. Brit J Haematol 97: 481-484.

Literature 123

[280] Bui, T. V. L., Ross, I. L., Jakob, G. & Hankamer, B. (2013): Impact of procedural steps and cryopreservation agents in the cryopreservation of Chlorophyte microalgae. PLoS ONE 8: e78668. [281] Morschett, H., Reich, S., Wiechert, W. & Oldiges, M. (2016): Simplified cryopreservation of the microalga Chlorella vulgaris integrating a novel concept for cell viability estimation. Eng Life Sci 16: 36-44. [282] Govender, T., Ramanna, L., Rawat, I. & Bux, F. (2012): BODIPY staining, an alternative to the Nile Red fluorescence method for the evaluation of intracellular lipids in microalgae. Bioresource Technol 114: 507-511. [283] Hwang, S.-W. & Horneland, W. (1965): Survival of algal cultures after freezing by controlled and uncontrolled cooling. Cryobiology 1: 305-311. [284] Guermazi, W., Sellami-Kammoun, A., Elloumi, J., Drira, Z., Aleya, L., Marangoni, R., Ayadi, H. & Maalej, S. (2010): Microalgal cryo-preservation using dimethyl sulfoxide (Me2SO) coupled with two freezing protocols: Influence on the fatty acid profile. J Therm Biol 35: 175-181. [285] Osório, H. C., Laranjeiro, N. C., Santos, L. M. A. & Santos, F. M. (2004): First attempts to cryopreserve strains from the Coimbra Cellection of Algae (ACOI) and the use of image analysis to assess viability. Nova Hedwigia 79: 227-235. [286] Roshani, O., Yap, L., Jeevan, R. & Mohd Syahril, M. (2011): A preliminiary study towards cryopreservation of unicellular algae, Chlorella vulgaris. Terengganu: Universiti Malaysia Terengganu International Annual Symposium On Sustainability Science And Management. [287] Shirihira-Ishikawa, I. & Hase, E. (1964): Nutritional control of cell pigmentation in Chlorella protothecoideswith special reference to the degeneration of chloroplast induced by glucose. Plant Cell Physiol 5: 227-240. [288] Dubertret, G. & Joliot, P. (1974): Structure and organization of system II photosynthetic units during the greening of a dark-grown Chlorella mutant. BBA-Bioenergetics 357: 399-411. [289] Herron, H. A. & Mauzerall, D. (1972): The development of photosynthesis in a greening mutant of Chlorella and an analysis of the light saturation curve. Plant Physiol 50: 141-148. [290] Yeh, K.-L., Chang, J.-S. & chen, W.-m. (2010): Effect of light supply and carbon source on cell growth and cellular composition of a newly isolated microalga Chlorella vulgaris ESP-31. Eng Life Sci 10: 201-208. [291] Lattermann, C. & Büchs, J. (2015): Microscale and miniscale fermentation and screening. Curr Opin Biotech 35: 1-6. [292] Chen, M., Mertiri, T., Holland, T. & Basu, A. S. (2012): Optical microplates for high-throughput screening of photosynthesis in lipid-producing algae. Lab Chip 12: 3870-3874. [293] Pacheco, A., Hernández-Mireles, I., García-Martínez, C. & Álvarez, M. M. (2013): Microplates as a microreactor platform for microalgae research. Biotechnol Progr 29: 638-644. [294] Han, W., Li, C., Miao, X. & Yu, G. (2012): A novel miniature culture system to screen CO2-sequestering microalgae. Energies 5: 4372-4389.

Literature 124

[295] Betts, J. P. J., Warr, S. R. C., Finka, G. B., Uden, M., Town, M., Janda, J. M., Baganz, F. & Lye, G. J. (2014): Impact of aeration strategies on fed-batch cell culture kinetics in a single-use 24-well miniature bioreactor. Biochem Eng J 82: 105-116. [296] Silk, N. J., Denby, S., Lewis, G., Kuiper, M., Hatton, D., Field, R., Baganz, F. & Lye, G. (2010): Fed-batch operation of an industrial cell culture process in shaken microwells. Biotechnol Lett 32: 73-78. [297] Heo, J., Cho, D.-H., Ramanan, R., Oh, H.-M. & Kim, H.-S. (2015): PhotoBiobox: A tablet sized, low-cost, high throughput photobioreactor for microalgal screening and culture optimization for growth, lipid content and CO2 sequestration. Biochem Eng J 103: 193-197. [298] Weuster-Botz, D., Puskeiler, R., Kusterer, A., Kaufmann, K., John, G. & Arnold, M. (2005): Methods and milliliter scale devices for high-throughput bioprocess design. Bioproc Biosyst Eng 28: 109-119. [299] Wewetzer, S. J., Kunze, M., Ladner, T., Luchterhand, B., Roth, S., Rahmen, N., Kloß, R., Costa e Silva, A., Regestein, L. & Büchs, J. (2015): Parallel use of shake flask and microtiter plate online measuring devices (RAMOS and BioLector) reduces the number of experiments in laboratory-scale stirred tank bioreactors. J Biol Eng 9: 1-19. [300] Morschett, H., Schiprowski, D., Müller, C., Mertens, K., Felden, P., Meyer, J., Wiechert, W. & Oldiges, M. (2017): Design and validation of a parallelized micro-photobioreactor enabling phototrophic bioprocess development at elevated throughput. Biotechnol Bioeng 114: 122-131. [301] Cheng, H. H., Huang, D.-S. & Lin, M.-T. (2012): Heat dissipation design and analysis of high power LED array using the finite element method. Microelectron Reliab 52: 905-911. [302] Narendran, N. & Gu, Y. (2005): Life of LED-based white light sources. J. Display Technol. 1: 167. [303] Chhajed, S., Xi, Y., Li, Y.-L., Gessmann, T. & Schubert, E. F. (2005): Influence of junction temperature on chromaticity and color-rendering properties of trichromatic white-light sources based on light-emitting diodes. J Appl Phys 97: 054506. [304] Campbell, N. A., Reece, J. B., Urry, L. A., Cain, M. L., Wasserman, S. A., Minorsky, P. V. & Jackson, R. B. (2015): Campbell Biologie. Hallbergmoos: Pearson Studium. [305] Zimmermann, H. F., John, G. T., Trauthwein, H., Dingerdissen, U. & Huthmacher, K. (2003): Rapid evaluation of oxygen and water permeation through microplate sealing tapes. Biotechnol Progr 19: 1061-1063. [306] Bohren, C. F. & Huffman, D. R. (2008): Absorption and scattering of light by small particles. Weinheim: Wiley VCH. [307] Jacobi, A., Steinweg, C., Sastre, R. R. & Posten, C. (2012): Advanced photobioreactor LED illumination system: Scale-down approach to study microalgal growth kinetics. Eng Life Sci 12: 621-630. [308] Grobbelaar, J. U. (2006): Photosynthetic response and acclimation of microalgae to light fluctuations. In: Subba Rao, D. V. Algal Cultures Analogues of Blooms and Applications. Enfield: Science Publishers. [309] Slegers, P. M., Wijffels, R. H., van Straten, G. & van Boxtel, A. J. B. (2011): Design scenarios for flat panel photobioreactors. Appl Energ 88: 3342-3353.

Literature 125

[310] Merchuk, J., Garcia-Camacho, F. & Molina-Grima, E. (2007): Photobioreactor design and fluid dynamics. Chem Biochem Eng Q 21: 345-355. [311] Perner-Nochta, I. & Posten, C. (2007): Simulations of light intensity variation in photobioreactors. J Biotechnol 131: 276-285. [312] Olivieri, G., Salatino, P. & Marzocchella, A. (2014): Advances in photobioreactors for intensive microalgal production: configurations, operating strategies and applications. J Chem Technol Biotechnol 89: 178-195. [313] Morschett, H., Schiprowski, D., Rohde, J., Wiechert, W. & Oldiges, M. (2017): Comparative evaluation of phototrophic microtiter plate cultivation against laboratory-scale photobioreactors. Bioproc Biosyst Eng, doi: 10.1007/s00449-016-1731-5. [314] Fujita, Y., Murakami, A. & Ohki, K. (1987): Regulation of photosystem composition in the cyanobacterial photosynthetic system: the regulation occurs in response to the redox state of the electron pool located between the two photosystems. Plant Cell Physiol 28: 283-292. [315] Kim, J. H., Glick, R. E. & Melis, A. (1993): Dynamics of photosystem stoichiometry adjustment by light quality in chloroplasts. Plant Physiol 102: 181-190. [316] Weiss, S., John, G. T., Klimant, I. & Heinzle, E. (2002): Modeling of mixing in 96-well microplates observed with fluorescence indicators. Biotechnol Progr 18: 821-830. [317] Myers, J. & Graham, J.-R. (1971): The photosynthetic unit in Chlorella measured by repetitive short flashes. Plant Physiol 48: 282-286. [318] Richardson, B., Orcutt, D. M., Schwertner, H. A., Martinez, C. L. & Wickline, H. E. (1969): Effects of nitrogen limitation on the growth and composition of unicellular algae in continuous culture. Appl Microbiol 18: 245-250. [319] Han, Y., Wen, Q., Chen, Z. & Li, P. (2011): Review of methods used for microalgal lipid-content analysis. Energy Procedia 12: 944-950. [320] Rumin, J., Bonnefond, H., Saint-Jean, B., Rouxel, C., Sciandra, A., Bernard, O., Cadoret, J. P. & Bougaran, G. (2015): The use of fluorescent Nile red and BODIPY for lipid measurement in microalgae. Biotechnol Biofuels 8: 42. [321] Bligh, E. G. & Dyer, W. J. (1959): A rapid method of total lipid extraction and purification. Can J Biochem Phys 37: 911-917. [322] Folch, J., Lees, M. & Stanley, G. H. S. (1957): A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 226: 497-509. [323] Lepage, G. & Roy, C. C. (1984): Improved recovery of fatty acid through direct transesterification without prior extraction or purification. J Lipid Res 25: 1391-1396. [324] Zhu, M., Zhou, P. P. & Yu, L. J. (2002): Extraction of lipids from Mortierella alpina and enrichment of arachidonic acid from the fungal lipids. Bioresource Technol 84: 93-95. [325] Widjaja, A., Chien, C.-C. & Ju, Y.-H. (2009): Study of increasing lipid production from fresh water microalgae Chlorella vulgaris. J Taiwan Inst Chem E 40: 13-20. [326] Schäfer, K. (1998): Accelerated solvent extraction of lipids for determining the fatty acid composition of biological material. Anal Chim Acta 358: 69-77.

Literature 126

[327] Lee, D.-H., Bae, C. Y., Han, J.-I. & Park, J.-K. (2013): In situ analysis of heterogeneity in the lipid content of single green microalgae in alginate hydrogel microcapsules. Anal Chem 85: 8749-8756. [328] Brennan, L., Blanco Fernández, A., Mostaert, A. S. & Owende, P. (2012): Enhancement of BODIPY505/515 lipid fluorescence method for applications in biofuel-directed microalgae production. J Microbiol Meth 90: 137-143. [329] Kirschner, M. W. & Hara, K. (1980): New method for local vital staining of amphibian embryos using ficoll and crystals of Nile red. Mikroskopie 36: 12-15. [330] Fowler, S. D. & Greenspan, P. (1985): Application of Nile red, a fluorescent hydrophobic probe, for the detection of neutral lipid deposits in tissue-sections: Comparison with Oil Red O. J Histochem Cytochem 33: 833-836. [331] Ghoneim, N. (2000): Photophysics of Nile red in solution: Steady state spectroscopy. Spectrochim Acta A 56: 1003-1010. [332] Greenspan, P., Mayer, E. P. & Fowler, S. D. (1985): Nile red: A selective fluorescent stain for intracellular lipid droplets. J Cell Biol 100: 965-973. [333] Cooksey, K. E., Guckert, J. B., Williams, S. A. & Callis, P. R. (1987): Fluorometric determination of the neutral lipid content of microalgal cells using Nile Red. J Microbiol Meth 6: 333-345. [334] Chen, W., Sommerfeld, M. & Hu, Q. (2011): Microwave-assisted Nile red method for in vivo quantification of neutral lipids in microalgae. Bioresource Technol 102: 135-141. [335] Krishnamoorthy, I. & Krishnamoorthy, G. (2001): Probing the link between proton transport and water content in lipid membranes. J Phys Chem B 105: 1484-1488. [336] Greenspan, P. & Fowler, S. D. (1985): Spectrofluorometric studies of the lipid probe, nile red. J Lipid Res 26: 781-789. [337] Ruvinov, S. B., Yang, X.-J., Parris, K. D., Banik, U., Ahmed, S. A., Miles, E. W. & Sackett, D. L. (1995): Ligand-mediated changes in the tryptophan synthase indole tunnel probed by Nile red fluorescence with wild type, mutant, and chemically modified enzymes. J Biol Chem 270: 6357-6369. [338] Balduyck, L., Veryser, C., Goiris, K., Bruneel, C., Muylaert, K. & Foubert, I. (2015): Optimization of a Nile red method for rapid lipid determination in autotrophic, marine microalgae is species dependent. J Microbiol Meth 118: 152-158. [339] Challagulla, V., Nayar, S., Walsh, K. & Fabbro, L. (2016): Advances in techniques for assessment of microalgal lipids. Crit Rev Biotechnol, doi: 10.1080/07388551.2016.1206058. [340] Cooper, M. S., Hardin, W. R., Petersen, T. W. & Cattolico, R. A. (2010): Visualizing "green oil" in live algal cells. J Biosci Bioeng 109: 198-201. [341] Elsey, D., Jameson, D., Raleigh, B. & Cooney, M. J. (2007): Fluorescent measurement of microalgal neutral lipids. J Microbiol Meth 68: 639-642. [342] Chen, H., Zhang, Y., He, C. & Wang, Q. (2014): Ca2+ signal transduction related to neutral lipid synthesis in an oil-producing green alga Chlorella sp. C2. Plant Cell Physiol 55: 634-644.

Literature 127

[343] Ren, H.-Y., Liu, B.-F., Kong, F., Zhao, L., Xie, G.-J. & Ren, N.-Q. (2014): Enhanced lipid accumulation of green microalga Scenedesmus sp. by metal ions and EDTA addition. Bioresource Technol 169: 763-767. [344] Jones, R. G. W. & Lunt, O. R. (1967): The function of calcium in plants. Bot Rev 33: 407-426. [345] Liu, Z.-Y., Wang, G.-C. & Zhou, B.-C. (2008): Effect of iron on growth and lipid accumulation in Chlorella vulgaris. Bioresource Technol 99: 4717-4722. [346] Ruangsomboon, S., Ganmanee, M. & Choochote, S. (2013): Effects of different nitrogen, phosphorus, and iron concentrations and salinity on lipid production in newly isolated strain of the tropical green microalga, Scenedesmus dimorphus KMITL. J Appl Phycol 25: 867-874. [347] Concas, A., Steriti, A., Pisu, M. & Cao, G. (2014): Comprehensive modeling and investigation of the effect of iron on the growth rate and lipid accumulation of Chlorella vulgaris cultured in batch photobioreactors. Bioresource Technol 153: 340-350. [348] Alyabyev, A. J., Loseva, N. L., Gordon, L. K., Andreyeva, I. N., Rachimova, G. G., Tribunskih, V. I., Ponomareva, A. A. & Kemp, R. B. (2007): The effect of changes in salinity on the energy yielding processes of Chlorella vulgaris and Dunaliella maritima cells. Thermochim Acta 458: 65-70. [349] Duan, X., Ren, G. Y., Liu, L. L. & Zhu, W. X. (2012): Salt-induced osmotic stress for lipid overproduction in batch culture of Chlorella vulgaris. Afr J Biotechnol 11: 7072-7078. [350] Mallick, N. (2002): Biotechnological potential of immobilized algae for wastewater N, P and metal removal: A review. Biometals 15: 377-390. [351] Pittman, J. K., Dean, A. P. & Osundeko, O. (2011): The potential of sustainable algal biofuel production using wastewater resources. Bioresource Technol 102: 17-25. [352] Muñoz, R. & Guieysse, B. (2006): Algal–bacterial processes for the treatment of hazardous contaminants: A review. Wat Res 40: 2799-2815. [353] Chen, M., Tang, H., Ma, H., Holland, T. C., Ng, K. Y. S. & Salley, S. O. (2011): Effect of nutrients on growth and lipid accumulation in the green algae Dunaliella tertiolecta. Bioresource Technol 102: 1649-1655. [354] Yang, J., Cao, J., Xing, G. & Yuan, H. (2015): Lipid production combined with biosorption and bioaccumulation of cadmium, copper, manganese and zinc by oleaginous microalgae Chlorella minutissima UTEX2341. Bioresource Technol 175: 537-544. [355] Clijsters, H. & Van Assche, F. (1985): Inhibition of photosynthesis by heavy metals. Photosynth Res 7: 31-40. [356] Morschett, H., Freier, L., Rohde, J., Wiechert, W., von Lieres, E. & Oldiges, M. (2017): A framework for accelerated phototrophic bioprocess development: integration of parallelized microscale cultivation, laboratory automation and Kriging-assisted experimental design. Biotechnol Biofuels 10: 26. [357] Vandamme, D., Foubert, I., Fraeye, I., Meesschaert, B. & Muylaert, K. (2012): Flocculation of Chlorella vulgaris induced by high pH: Role of magnesium and calcium and practical implications. Bioresource Technol 105: 114-119.

Literature 128

[358] Converti, A., Casazza, A. A., Ortiz, E. Y., Perego, P. & Del Borghi, M. (2009): Effect of temperature and nitrogen concentration on the growth and lipid content of Nannochloropsis oculata and Chlorella vulgaris for biodiesel production. Chem Eng Process 48: 1146-1151. [359] Atta, M., Idris, A., Bukhari, A. & Wahidin, S. (2013): Intensity of blue LED light: A potential stimulus for biomass and lipid content in fresh water microalgae Chlorella vulgaris. Bioresource Technol 148: 373-378. [360] Hsieh, C.-H. & Wu, W.-T. (2009): Cultivation of microalgae for oil production with a cultivation strategy of urea limitation. Bioresource Technol 100: 3921-3926. [361] Gorain, P. C., Bagchi, S. K. & Mallick, N. (2013): Effects of calcium, magnesium and sodium chloride in enhancing lipid accumulation in two green microalgae. Environmental Technology 34: 1887-1894. [362] Yeh, K.-L. & Chang, J.-S. (2011): Nitrogen starvation strategies and photobioreactor design for enhancing lipid content and lipid production of a newly isolated microalga Chlorella vulgaris ESP-31: Implications for biofuels. Biotechnol J 6: 1358-1366. [363] Yoo, C., Jun, S.-Y., Lee, J.-Y., Ahn, C.-Y. & Oh, H.-M. (2010): Selection of microalgae for lipid production under high levels carbon dioxide. Bioresource Technol 101: 71-74.

Appendix 129

6. Appendix

6.1 Simplified kinetic model for phototrophic batch cultivation

The subsequently described model was applied to illustrate the theoretical kinetics of photo-

trophic batch cultivation presented in section 1.4 by Figures 1.5 and 1.6. However, it strongly

simplifies the kinetics of such cultivations. Especially the light supply kinetics is reduced to

minimum complexity. The model may thus not suitable for the description of process data,

but rather for exemplary visualisation purposes.

Biomass formation is modelled by an exponential approach:

dX

dt = µ∙X

(6-1)

µ growth rate [h-1]

X biomass concentration [g L-1]

For simplification, it is assumed that the carbon source CO2 is all time available at excess.

Especially during laboratory-scale experiments, this simplification can be made due to the

comparably high mass transfer capacity of small bioreactors. Here, a mineral compound of

the fermentation medium being essential for biomass but not for product formation (e.g. ni-

trate) is considered as the limiting substrate:

dS

dt = -

1

YX/S

∙µ∙X

(6-2)

S concentration of limiting mineral substrate [g L-1]

YX/S biomass yield [g g-1]

The growth rate is modelled via modified Monod kinetics. Besides the concentration of the

limiting medium component, it is additionally dependent of an initial lag phase as well as of

the relative photosynthesis rate:

Appendix 130

µ = µmax

∙S

KS+S∙lag

µ∙Prel

(6-3)

µmax maximal growth rate [h-1]

KS substrate affinity constant [g L-1]

lagµ lag phase coefficient [-]

Prel relative photosynthesis rate [-]

The lag phase itself is composed of a phase with no growth and an acceleration phase

where µ steadily increases from 0 to the actual maximum value:

lagµ =

1

1+e(2+4∙(

tlag,µ-t

tacc,µ))

(6-4)

tlag,µ duration of lag phase [h]

t process time [h]

tacc,µ duration of acceleration phase [h]

The relative photosynthesis rate is modelled according to substrate excess inhibition kinetics

with an additional maintenance coefficient for cell respiration:

Prel = Pmax∙

(

Iav

KM+Iav+Iav2

KI)

-mI (6-5)

Pmax maximal photosynthesis rate [-]

Iav average light intensity [µmol m-2 s-1]

KM light affinity constant [µmol m-2 s-1]

KI light excess inhibition constant [µmol m-2 s-1]

mI maintenance coefficient [-]

Across a photobioreactor, an optical path length and biomass dependent light intensity gra-

dient can be observed. For simplification, it is assumed that all cells at all positions of the

reactor experience the same light intensity. This average light intensity was modelled accord-

ing to Lambert-Beer while the absorption coefficient incorporates the reactor specific impact

of optical path lengths:

Appendix 131

Iav = I0∙e(-ε∙X)

(6-6)

I0 incident light intensity [µmol m-2 s-1]

ε light absorption coefficient [g L-1]

In this model, intracellular accumulation of storage lipids is simulated via the constant pro-

duction rate qP. Analogous to growth, a lag coefficient for production start is incorporated:

nLC = lagP∙q

P∙tprod

(6-7)

nLC neutral lipid content [% (w w-1)]

lagP lag phase coefficient for product formation [-]

qP product formation rate [% (w w-1) h-1]

tprod time from start of production phase [h]

In analogy to growth, the delay factor of the product formation consists of a lag phase and an

additional acceleration phase:

lagP =

1

1+e(2+4∙(

tlag,P-t

tacc,P))

(6-8)

tlag,P duration of lag phase for product formation [h]

tacc,P duration of acceleration phase for product formation [h]

Product formation is repressed in the presence of S. After its depletion (S < 0.0001), produc-

tion is induced by initialization of tprod.

Appendix 132

6.2 Additional figures

Figure 6.1: Custom-made shaker for phototrophic cultivation. A: PAR distribution across the shaking tray of a Multitron Pro incubator equipped with 7x F25W/30/GRO fluorescent lamps. B: A sparger is used for direct gassing of the culture liquid with CO2-enriched air.

Figure 6.2: Emission spectra of the individual LEDs used to set up the photo module. Modi-fied from [300].

300 450 600 750 900

0.00

0.25

0.50

0.75

1.00

1.25

rela

tiv

e p

ow

er

[%

]

wavelength [nm]

near-UV

blue

warm-white

Appendix 133

Figure 6.3: Individual biomass curves of the proof-of-principle cultivation; enBBMref, 25 °C, 2.5 % (v v-1) CO2, 200 µmol m-2 s-1 PAR, shaking frequency 1200 rpm, VL = 1 mL, relative humidity ≥ 85 %. Modified from [300].

Figure 6.4: Spectra of the illumination units installed in the used photobioreactors. MTP mi-cro-bioreactor spectrum from own measurements (section 3.2), others according to manufac-turers’ information.

0 24 48 72 96 120 144 168

0

10

20

30

40 X33

X34

X35

X36

X37

X38

X39

X40

X41

X42

X43

X44

X45

X46

X47

X48

X17

X18

X19

X20

X21

X22

X23

X24

X25

X26

X27

X28

X29

X30

X31

X32

X1

X2

X3

X4

X5

X6

X7

X8

X9

X10

X11

X12

X13

X14

X15

X16

sc

att

ere

rd l

igh

t

[-]

time [h]

400 500 600 700

0

1

2

3

4

5

rela

tiv

e P

AR

[

%]

wavelength [nm]

MTP

shake flask

MC 1000

FMT 150/1000

Appendix 134

Figure 6.5: Influence of the shaking frequency on liquid distribution in FlowerPlates®; orbital shaking in BioLector® device with 3 mm shaking diameter. Modified from [313].

Figure 6.6: Estimated effect of two factor interaction with MgSO4. Estimation is based on the experiments in section 3.5.2.1 using the fractional factorial design given in Table 6.1. Modi-fied from [356].

ME

S

NaN

O3

ph

osph

ate

s

NaC

l

CaC

l2

trace

FeS

O4

ED

TA

-0.8

-0.4

0.0

0.4

0.8

CaC

l 2

co

eff

icie

nt

[-

]

two component interaction with MgSO4

NaN

O3

FeS

O4

Appendix 135

Figure 6.7: Screening plot around reference point. Estimation of the functional relationship between media components and lipid productivity by Kriging approximation. The model is based on the experiments in section 3.5.2.2. Modified from [356].

Appendix 136

Figure 6.8: Monitoring of calcium precipitation in enBBM by means of optical density. Blue shaded area represents the parameter space covered during experimental design. Error bars deviated from technical replicates (n = 3). Modified from [356].

Figure 6.9: Relative composition of the fatty acids from the neutral lipid fraction. Error bars represent min/max from biological replicates (n = 2). Modified from [356].

0 1 2 3 4 5

0.0

0.1

0.2

0.3

0.4

OD

750

[-]

calcium [mM]

reference

concentration

enBBMopt enBBMopt,min

0

25

50

75

100

enBBMopt,min

enBBMopt

rela

tiv

e f

rac

tio

n [

%]

medium

16:0

16:1 9

16:2 7,10

16:3 7,10,13

18:0

18:1 9

18:2 9,12

18:3 9,12,15

Appendix 137

6.3 Additional tables

Table 6.1: Experimental design and measured data used in section 3.5.2.1. Modified from [356].

ID MES

[mmol L-1

]

NaNO3

[g L-1

]

phosphates

[g L-1

]

NaCl

[mmol L-1

]

MgSO4

[mmol L-1

]

CaCl2

[mmol L-1

]

trace

[x fold]

FeSO4

[mmol L-1

]

EDTA

[mmol L-1

]

lipid

productivity

[-]

ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 291.8598 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 316.6211 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 291.6364 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 275.3174 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 255.8972 1 0 0.25 0.25 5 2.5 2.5 2.5 0.004 0 237.5104 2 0 0.25 2.5 0 2.5 0.1 2.5 0.04 0 921.8246 3 75 2.5 0.25 0 2.5 2.5 2.5 0.04 0 565.1934 4 0 2.5 2.5 5 2.5 2.5 0.1 0.04 0 16.0653 5 0 2.5 0.25 5 0.1 2.5 2.5 0.004 1 521.1095 6 75 0.25 2.5 5 0.1 2.5 2.5 0.04 0 295.3840 7 75 0.25 0.25 5 2.5 0.1 0.1 0.04 0 356.8340 8 75 2.5 2.5 0 0.1 2.5 0.1 0.004 1 74.4502 9 75 0.25 2.5 0 2.5 2.5 0.1 0.004 0 607.1565

10 0 0.25 0.25 5 2.5 0.1 2.5 0.04 1 514.5004 11 0 2.5 0.25 5 0.1 0.1 2.5 0.04 0 0.2860 12 75 0.25 0.25 0 0.1 0.1 2.5 0.004 0 110.3478 13 75 2.5 0.25 5 0.1 0.1 0.1 0.04 1 0.1907 14 75 2.5 2.5 5 2.5 2.5 2.5 0.04 1 23.3291 15 0 2.5 2.5 5 2.5 0.1 0.1 0.004 1 0.3658 16 0 0.25 0.25 0 0.1 0.1 0.1 0.004 1 0.3321 17 75 0.25 2.5 0 2.5 0.1 0.1 0.04 1 3.3241 18 75 0.25 2.5 5 0.1 0.1 2.5 0.004 1 0.1810 19 0 2.5 2.5 0 0.1 0.1 2.5 0.04 1 5.6455 20 0 2.5 0.25 0 2.5 0.1 0.1 0.004 0 17.5138 21 0 2.5 2.5 0 0.1 2.5 2.5 0.004 0 0.3712 22 75 2.5 0.25 5 0.1 2.5 0.1 0.004 0 339.3216 23 0 0.25 2.5 0 2.5 2.5 2.5 0.004 1 720.6980 24 0 2.5 0.25 0 2.5 2.5 0.1 0.04 1 26.8543 25 75 2.5 2.5 0 0.1 0.1 0.1 0.04 0 12.8407 26 75 0.25 0.25 5 2.5 2.5 0.1 0.004 1 78.5940 27 0 0.25 2.5 5 0.1 2.5 0.1 0.04 1 0.1639 28 0 0.25 0.25 0 0.1 2.5 0.1 0.04 0 0.0323 29 75 0.25 0.25 0 0.1 2.5 2.5 0.04 1 770.6465 30 0 0.25 2.5 5 0.1 0.1 0.1 0.004 0 670.6747 31 75 2.5 0.25 0 2.5 0.1 2.5 0.004 1 6.4059 32 75 2.5 2.5 5 2.5 0.1 2.5 0.004 0 13.7916

Table 6.2: Comparison of the initial composition of enBBMref, enBBMopt and enBBMopt,min. Modified from [356].

component concentration [mmol L

-1]

enBBMref enBBMopt enBBMopt,min

CaCl2 0.850 1.063 1.063

CoSO4 0.00033 0.000825 0.000825

CuSO4 0.001 0.0025 0.0025

FeSO4 0.004 0.004 0.004

H3BO3 0.037 0.0925 0.0925

K2HPO4 / KH2PO4 13.732 13.732 1.7165

KOH 1.105 1.105 0.0

MES 50.0 50.0 0.0

MgSO4 0.761 2.5 2.5

MnCl2 0.002 0.005 0.005

NaCl 0.107 0.107 0.0

Na2EDTA 0.297 0.297 0.0

Na2MoO4 0.001 0.0025 0.0025

NaNO3 17.649 7.94205 7.94205

penicillin-G 0.281 0.281 0.281

ZnSO4 0.006 0.015 0.015

Appendix 138

Table 6.3: Experimental design and measured data used in section 3.5.2.2. Modified from [356].

ID MES

[mmol L-1

]

NaNO3

[g L-1

]

phosphates

[g L-1

]

NaCl

[mmol L-1

]

MgSO4

[mmol L-1

]

CaCl2

[mmol L-1

]

trace

[x fold]

FeSO4

[mmol L-1

]

EDTA

[mmol L-1

]

lipid

productivity

[-]

ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 356.5913 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 373.2705 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 338.6284 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 365.9636 1 50 2.5 2 0 0.1 2.5 2.5 0.004 0.297 361.1236 2 50 0.25 2 5 2.5 0.1 0.1 0.004 0.297 8.8542 3 50 2.5 2 0 2.5 2.5 2.5 0.004 0.297 44.6547 4 50 2.5 2 0 2.5 2.5 0.1 0.004 0.297 29.5387 5 50 0.25 2 2.5 1.9 2.5 2.5 0.004 0.297 831.9867 6 50 2.5 2 5 0.1 2.5 0.1 0.004 0.297 468.9143 7 50 0.25 2 5 2.5 2.5 0.1 0.004 0.297 329.6299 8 50 0.25 2 0 2.5 2.5 0.1 0.004 0.297 316.2275 9 50 0.25 2 5 2.5 2.5 2.5 0.004 0.297 767.1525

10 50 0.25 2 5 2.5 0.1 2.5 0.004 0.297 817.1550 11 50 2.5 2 5 2.5 2.5 2.5 0.004 0.297 25.0784 12 50 0.25 2 5 0.1 0.1 0.1 0.004 0.297 0.3298 13 50 2.5 2 5 0.1 0.1 0.1 0.004 0.297 5.4212 14 50 0.25 2 0 2.5 0.1 2.5 0.004 0.297 816.2714 15 50 2.5 2 0 2.5 0.1 2.5 0.004 0.297 23.6259 16 50 2.5 2 5 2.5 2.5 0.1 0.004 0.297 23.7400 17 50 2.5 2 0 0.1 2.5 0.1 0.004 0.297 711.3314 18 50 0.25 2 0 0.1 0.1 2.5 0.004 0.297 148.3129 19 50 2.5 2 0 0.1 0.1 0.1 0.004 0.297 0.3315 20 50 2.5 2 0 0.1 0.1 2.5 0.004 0.297 636.4822 21 50 0.25 2 0 0.1 0.1 0.1 0.004 0.297 3.6634 22 50 2.5 2 5 2.5 0.1 0.1 0.004 0.297 7.8702 23 50 0.8125 2 2.5 1.9 1.9 1.9 0.004 0.297 1542.9296 24 50 2.5 2 0 2.5 2.5 2.5 0.004 0.297 796.7885 25 50 2.5 2 0 0.1 2.5 2.5 0.004 0.297 628.5450 26 50 0.8125 2 0 2.5 1.9 1.9 0.004 0.297 1621.5284 27 50 0.25 2 5 0.1 2.5 0.1 0.004 0.297 295.6224 28 50 0.25 2 0 0.1 2.5 0.1 0.004 0.297 358.8521 29 50 0.25 2 5 0.1 2.5 2.5 0.004 0.297 764.7553 30 50 0.25 2 0 2.5 0.1 0.1 0.004 0.297 11.7765 31 50 2.5 2 5 2.5 0.1 2.5 0.004 0.297 42.1237 32 50 2.5 2 0 2.5 0.1 0.1 0.004 0.297 8.2854 33 50 2.5 2 5 0.1 0.1 2.5 0.004 0.297 364.4683 34 50 0.25 2 5 0.1 0.1 2.5 0.004 0.297 326.6296 35 50 2.5 2 5 0.1 2.5 2.5 0.004 0.297 418.0846

Table 6.4: Experimental design and measured data used in section 3.5.3. Modified from [356].

ID MES

[mmol L-1

]

NaNO3

[g L-1

]

phosphates

[g L-1

]

NaCl

[mmol L-1

]

MgSO4

[mmol L-1

]

CaCl2

[mmol L-1

]

trace

[x fold]

FeSO4

[mmol L-1

]

EDTA

[mmol L-1

]

lipid

productivity

[-]

ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 340.9035 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 333.6144 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 312.5471 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 290.1185 1 50 1.5 2 0.107 2.5 1.3 0.1 0.004 0.297 121.2866 2 50 1.5 2 0.107 0.1 0.1 1.925 0.004 0.297 123.8942 3 50 0.5625 2 0.107 1.3 1.9 1.0125 0.004 0.297 1433.2791 4 50 0.875 2 0.107 1.3 0.7 1.0125 0.004 0.297 374.5056 5 50 1.1875 2 0.107 1.3 0.7 2.8375 0.004 0.297 436.9745 6 50 0.875 2 0.107 2.5 0.1 3.75 0.004 0.297 922.6879 7 50 0.5625 2 0.107 1.3 0.7 2.8375 0.004 0.297 1008.9076 8 50 1.1875 2 0.107 1.3 1.9 1.0125 0.004 0.297 678.6265 9 50 1.5 2 0.107 2.5 0.1 3.75 0.004 0.297 413.6182

10 50 0.25 2 0.107 2.5 0.1 0.1 0.004 0.297 4.8716 11 50 1.5 2 0.107 0.1 0.1 0.1 0.004 0.297 0.2557 12 50 0.25 2 0.107 2.5 2.5 1.925 0.004 0.297 708.2691 13 50 0.5625 2 0.107 1.3 1.9 2.8375 0.004 0.297 83.3691 14 50 0.875 2 0.107 0.1 2.5 0.1 0.004 0.297 410.6806 15 50 0.5625 2 0.107 1.3 1.3 2.8375 0.004 0.297 1395.1340 16 50 0.25 2 0.107 2.5 0.1 3.75 0.004 0.297 759.9804 17 50 1.5 2 0.107 2.5 0.1 0.1 0.004 0.297 4.1753 18 50 1.1875 2 0.107 1.3 1.9 1.925 0.004 0.297 829.3965 19 50 0.25 2 0.107 0.1 0.1 0.1 0.004 0.297 0.2938 20 50 0.25 2 0.107 0.1 2.5 3.75 0.004 0.297 127.7109 21 50 1.5 2 0.107 2.5 2.5 0.1 0.004 0.297 306.3595 22 50 0.25 2 0.107 0.1 1.3 3.75 0.004 0.297 426.2241 23 50 1.5 2 0.107 0.1 0.1 3.75 0.004 0.297 229.9717 24 50 1.1875 2 0.107 1.3 0.7 1.0125 0.004 0.297 289.6067 25 50 1.5 2 0.107 2.5 2.5 3.75 0.004 0.297 4.1512 26 50 0.25 2 0.107 2.5 2.5 3.75 0.004 0.297 243.1859 27 50 0.875 2 0.107 0.1 1.06 1.56 0.004 0.297 555.6468 28 50 0.875 2 0.107 2.5 1.06 1.56 0.004 0.297 1248.7792 29 50 0.5625 2 0.107 1.3 0.7 1.0125 0.004 0.297 1022.2337 30 50 1.1875 2 0.107 1.3 1.9 2.8375 0.004 0.297 575.8948 31 50 0.25 2 0.107 0.1 0.1 3.75 0.004 0.297 379.4746 32 50 0.25 2 0.107 0.1 2.5 0.1 0.004 0.297 288.2517 33 50 1.5 2 0.107 0.1 2.5 3.75 0.004 0.297 5.2624 34 50 0.25 2 0.107 2.5 2.5 0.1 0.004 0.297 338.7271 35 50 1.5 2 0.107 0.1 2.5 0.1 0.004 0.297 458.8032

Appendix 139

Table 6.5: Experimental design and measured data used in section 3.5.2.4. Modified from [356].

ID MES

[mmol L-1

]

NaNO3

[g L-1

]

phosphates

[g L-1

]

NaCl

[mmol L-1

]

MgSO4

[mmol L-1

]

CaCl2

[mmol L-1

]

trace

[x fold]

FeSO4

[mmol L-1

]

EDTA

[mmol L-1

]

lipid

productivity

[-]

ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 322.1343 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 408.7014 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 338.9990 ref 50 1.5 2 0.107 0.761 0.85 1 0.004 0.297 377.4233 1 50 0.375 2 0.107 0.7494 0.2499 3.378 0.004 0.297 1319.7094 2 50 0.625 2 0.107 0.6245 2.12415 1.875 0.004 0.297 1281.4145 3 50 0.75 2 0.107 2.3731 1.12455 1.625 0.004 0.297 739.1991 4 50 0.5 2 0.107 0.4996 1.87425 3.5 0.004 0.297 792.7942 5 50 1.125 2 0.107 2.1233 0.7497 1.875 0.004 0.297 1324.6965 6 50 0.375 2 0.107 1.249 1.62435 1.875 0.004 0.297 968.6335 7 50 0.625 2 0.107 2.498 1.62435 1.25 0.004 0.297 649.319 8 50 1.125 2 0.107 1.249 1.37445 1.625 0.004 0.297 796.9940 9 50 0.375 2 0.107 1.249 0.9996 1.875 0.004 0.297 881.7006

10 50 0.25 2 0.107 1.1241 1.62435 1.5 0.004 0.297 294.8120 11 50 0.5 2 0.107 2.2482 1.4994 1.625 0.004 0.297 728.5409 12 50 1.5 2 0.107 0.8743 1.9992 0.875 0.004 0.297 509.3818 13 50 0.625 2 0.107 0.7494 1.37445 2 0.004 0.297 651.2139 14 50 1.375 2 0.107 0.6245 1.2495 3.125 0.004 0.297 604.4187 15 50 1.125 2 0.107 1.8735 0.4998 2 0.004 0.297 1140.2650 16 50 1.375 2 0.107 0.4996 0.37485 2.375 0.004 0.297 1153.2317 17 50 0.375 2 0.107 1.7486 1.4994 2.375 0.004 0.297 699.8038 18 50 0.375 2 0.107 1.1241 1.7493 1.875 0.004 0.297 628.0611 19 50 0.75 2 0.107 0.7494 0.87465 2.375 0.004 0.297 1174.4848 20 50 0.375 2 0.107 2.1233 1.37445 1.625 0.004 0.297 1154.9457 21 50 0.5 2 0.107 0.2498 1.7493 3.625 0.004 0.297 706.8882 22 50 0.5 2 0.107 1.1241 2.2491 1.25 0.004 0.297 804.2811 23 50 0.75 2 0.107 1.6237 0.9996 0.75 0.004 0.297 468.5243 24 50 0.25 2 0.107 1.4988 0.7497 3.375 0.004 0.297 1069.4915 25 50 1.125 2 0.107 2.2482 2.12415 3 0.004 0.297 755.2046 26 50 0.625 2 0.107 1.4988 1.12455 1.625 0.004 0.297 696.8223 27 50 0.625 2 0.107 2.498 1.7493 0.375 0.004 0.297 1335.2759 28 50 0.875 2 0.107 1.9984 0.4998 0.875 0.004 0.297 972.5113 29 50 1 2 0.107 0.4996 0.9996 3.625 0.004 0.297 694.364 30 50 0.25 2 0.107 2.498 1.2495 1.75 0.004 0.297 1218.5364 31 50 0.5 2 0.107 0.3747 2.12415 2.875 0.004 0.297 928.0238 32 50 0.375 2 0.107 2.1233 1.12455 1.25 0.004 0.297 809.5954 33 50 0.5 2 0.107 0.6245 0.12495 2.125 0.004 0.297 709.5202 34 50 1 2 0.107 0.992 0.2499 1.75 0.004 0.297 967.9693 35 50 1.125 2 0.107 1.8735 2.12415 1.375 0.004 0.297 814.2653