microscale methods to rapidly evaluate bioprocess options for increasing bioconversion yields:...

ORIGINAL PAPER

Microscale methods to rapidly evaluate bioprocess optionsfor increasing bioconversion yields: applicationto the x-transaminase synthesis of chiral amines

Murni Halim • Leonardo Rios-Solis •

Martina Micheletti • John M. Ward •

Gary J. Lye

Received: 21 May 2013 / Accepted: 12 September 2013 / Published online: 28 September 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract This work aims to establish microscale meth-

ods to rapidly explore bioprocess options that might be

used to enhance bioconversion reaction yields: either by

shifting unfavourable reaction equilibria or by overcoming

substrate and/or product inhibition. As a typical and

industrially relevant example of the problems faced we

have examined the asymmetric synthesis of (2S,3R)-

2-amino-1,3,4-butanetriol from L-erythrulose using

the x-transaminase from Chromobacterium violaceum

DSM30191 (CV2025 x-TAm) and methylbenzylamine as

the amino donor. The first process option involves the use

of alternative amino donors. The second couples the

CV2025 x-TAm with alcohol dehydrogenase and glucose

dehydrogenase for removal of the acetophenone (AP) by-

product by in situ conversion to (R)-1-phenylethanol. The

final approaches involve physical in-situ product removal

methods. Reduced pressure conditions, attained using a

96-well vacuum manifold were used to selectively increase

evaporation of the volatile AP while polymeric resins were

also utilised for selective adsorption of AP from the bio-

conversion medium. For the particular reaction studied

here the most promising bioprocess options were use of an

alternative amino donor, such as isopropylamine, which

enabled a 2.8-fold increase in reaction yield, or use of a

second enzyme system which achieved a 3.3-fold increase

in yield.

Keywords x-Transaminase � Asymmetric synthesis �Microscale bioprocessing � Equilibrium-controlled

reaction � Product inhibition

Abbreviations

ABT (2S,3R)-2-Amino-1,3,4-butanetriol

ADH Alcohol dehydrogenase

AP Acetophenone

AQC 6-Aminoquinolyl-N-hydroxysuccinimidyl

carbamate

Ery L-Erythrulose

GDH Glucose dehydrogenase

IPA Isopropylamine

IPTG Isopropyl b-D-1-thiogalactopyranoside

ISPR In situ product removal

IScPR In situ co-product removal

MBA S-Methylbenzylamine

PLP Pyridoxal 50 phosphate

TFA Trifluoroacetic acid

U Units of enzyme activity (lmol min-1)

Introduction

Recently, transaminases have received considerable inter-

est due to their potential for the production of natural and

non-natural amino acids as well as other chiral amines,

which are in demand by the pharmaceutical industry [20].

Optically active amines are required for the preparation of

a broad range of biologically active compounds showing

M. Halim � L. Rios-Solis � M. Micheletti �J. M. Ward � G. J. Lye (&)

Department of Biochemical Engineering, The Advanced Centre

for Biochemical Engineering, University College London,

Torrington Place, London WC1E 7JE, UK

e-mail: [email protected]

M. Halim

Department of Bioprocess Technology, Faculty of

Biotechnology and Biomolecular Sciences, Universiti Putra

Malaysia, 43400 UPM Serdang, Selangor Darul Ehsan, Malaysia

e-mail: [email protected]

123

Bioprocess Biosyst Eng (2014) 37:931–941

DOI 10.1007/s00449-013-1065-5

various pharmacological properties [3, 23] such as building

blocks in the synthesis of neurological, immunological,

anti-hypertensive and anti-infective drugs. At present, it is

estimated that approximately 75 % of pharmaceuticals on

the market involve insertion or modification of an amino

group during their synthesis. Enantioenriched amines can

be produced by either asymmetric synthesis [18] using a

chiral ketone (i.e. amination of the ketone) or kinetic res-

olution[1] of racemic amine (i.e. deamination of the amine)

among other methods.

The x-transaminase from Chromobacterium violaceum

DSM30191 (CV2025 x-TAm) has previously been cloned

by us and expressed in Escherichia coli (E. coli) BL21-

Gold [10]. CV2025 x-TAm shows high enantioselectivity

for the (S)-enantiomer of various amine compounds such

as the widely used MBA. This high stereo selectivity

could lead to a useful asymmetric strategy for the prepa-

ration of enantiopure amines and amino alcohols [10, 15,

31]. The asymmetric synthesis of chiral amines with x-

transaminases may offer many potential advantages over

kinetic resolution, such as overcoming the need for a

reductive amination step, to prepare the racemic amine

from a ketone precursor and a twofold higher theoretical

yield [24].

Nevertheless, certain challenges must still be addressed.

These relate to the speed of biocatalytic process develop-

ment (relative to chemical processes) and the need to (1)

shift unfavourable reaction equilibria toward product for-

mation or (2) overcome issues of substrate/product inhi-

bition [19]. These challenges currently hinder attainment of

high substrate conversion yields and the achievement of

high product concentrations [11]. Additionally, choice of

amino donor is of great importance in the asymmetric

reaction system. For instance, it is impractical to use some

chiral amines as amino donor due to their high cost despite

their high reactivity.

A number of recent publications have begun to address

the issue of increasing the yield of x-TAm catalysed bio-

conversions [9, 25, 26, 33] including use of a second enzyme

system for in situ co-product conversion. While benefits

have been reported in each case the implementation of such

methods requires considerable further experimentation to

optimise bioconversion conditions. Figure 1 summarises the

potential process options to overcome low-yielding, equi-

librium-controlled x-TAm reactions or those with inhibitory

substrates or products. In the case of the exemplar synthesis

of ABT using MBA as amino donor we have recently shown

there to be a high-equilibrium constant, 843, but the

Fig. 1 Summary of the microscale methods established in this work to investigate bioprocess options to overcome thermodynamic or kinetic

limitations of bioconversions

932 Bioprocess Biosyst Eng (2014) 37:931–941

123

attainment of a high reaction yield is severely hampered by

substrate (MBA) and by-product (AP) inhibition [16]. The

focus in this work is the creation of novel microscale

experimental methods to enable the rapid and parallel

evaluation of different process options to increase biocon-

version reaction yield in an automated manner. Such

approaches would provide the data necessary for early stage

process design and economic evaluation [12].

Materials and methods

Materials

Nutrient broth and nutrient agar were obtained from Fisher

Scientific (Leicestershire, UK). Transaminase expression

was obtained using plasmid pQR801 transformed into

E. coli BL21-Gold (DE3), which contains the complete C.

violaceum DSM30191 x-transaminase gene [10]. These

strains were stored as glycerol stocks at -80 �C. All other

reagents were obtained from Sigma-Aldrich (Gillingham,

UK) unless otherwise stated.

Enzyme preparation

CV2025 x-TAm in either whole cell or lysate form

For expression of x-transaminase, 10 % (v/v) inoculum of

liquid overnight culture of E. coli BL21-Gold (DE3)-

pQR801 was inoculated into Luria–Bertani-Glycerol

medium (10 g L-1 tryptone, 5 g L-1 yeast extract,

10 g L-1 NaCl, 10 g L-1 glycerol) containing an appro-

priate resistance antibiotic (150 g L-1 kanamycin). Growth

was performed at 37 �C with orbital shaking at 250 rpm

using an SI 50 orbital shaker (Stuart Scientific, Redhill,

UK). When the OD reached 1.8–2.0, 1 mL of 0.1 M iso-

propyl-D-thiogalactopyranoside (IPTG) was added. After

5 h of induction, the cells were harvested and 2 mM of co-

factor, Pyridoxal 50 phosphate (PLP), was added prior to

either storage at -20 �C (following removal of broth by

centrifugation) or immediate use for whole cell biocon-

versions as described in ‘‘ISPR using adsorbent resins’’.

Upon thawing frozen samples of whole cell biocatalyst, the

cell pellet was resuspended in 50 mM HEPES buffer (pH

7.5) and 2 mM PLP was added. When a cell-free lysate

was needed the cell pellet was resuspended in 50 mM

HEPES buffer (pH 7.5) with additional 2 mM PLP and the

cells were disrupted by sonication carried out on ice using a

Soniprep 150 sonicator (MSE, Sanyo, Japan) with ten

cycles of 10 s, 10 lm pulses with 10 s intervals. The

sonicated suspension was centrifuged at 5,000 rpm for

5 min. The resulting supernatant was then filtered through

a 0.2 lm filter to obtain clarified extract (cell free lysate).

Alcohol dehydrogenase from Lactobacillus kefir (ADH)

For preparation of commercially available ADH, 1.2 g L-1

of ADH (0.44 U mg-1 protein where 1 U corresponds to

the amount of enzyme which reduces 1 lmol AP to

phenylethanol per minute at pH 7.0 and 25 �C) was dis-

solved in HEPES buffer (pH 7.5) with concentrations

depending on the required reaction concentrations con-

taining 10 mM MgCl2 since it is known that Mg2? helps to

stabilise the enzyme activity [28].

Glucose dehydrogenase from Pseudomonas sp. (GDH)

For preparation of commercially available GDH, 0.02 g L-1

of GDH (260 U mg-1, where 1 U corresponds to the amount

of enzyme which oxidised 1 lmol b-D-glucose to D-glucono-

d-lactone per minute at pH 8.0 and 37 �C) was dissolved in

HEPES buffer (pH 7.5) with concentrations depending on

the required reaction concentrations.

Enzyme reactions and microscale methods

Standard CV2025 x-TAm bioconversions

Batchwise x-TAm bioconversions were generally carried

using 0.3 g L-1 CV2025 x-TAm in clarified lysate,

0.2 mM PLP, 50 mM MBA and 50 mM Ery of 300 lL

total reaction volume. Reactions were carried out in a glass

96-well, flat-bottomed microtiter plate (Radley Discovery

Technologies, Essex, UK) and orbital shaking was pro-

vided at 300 rpm with a Thermomixer Comfort shaker

(orbital shaking diameter 6 mm, Eppendorf, Cambridge,

UK). The plate was covered with a thermoplastic elastomer

cap designed to work with automated equipment (Micro-

nic, Lelystad, The Netherlands). The solutions were incu-

bated for 20 min at 30 �C prior to the addition of each

amino donor and amino acceptor and were then continued

to be incubated at 30 �C. The reactions were quenched by

adding 0.1 % (v/v) trifluoroacetic acid (TFA) solution prior

to analysis by HPLC.

Evaluation of alternative amino donors

Reactions were carried out similar to the standard CV2025

x-TAm bioconversion with five different amino donors

(isopropylamine, benzylamine, S-(?)-sec-butylamine, S-

(?)-1-aminoindane and L-a-serine) as alternatives to MBA.

Second enzyme reaction

The CV2025 x-TAm/ADH/GDH coupled reactions were

carried out at the same 300 lL total reaction volume with

1 mM NADPH, 50 mM MBA, 50 or 70 mM Ery, 70 mM

Bioprocess Biosyst Eng (2014) 37:931–941 933

123

glucose, 0.3 g L-1 CV2025 x-TAm, 1.5 g L-1 ADH,

0.025 g L-1 GDH, 0.2 mM PLP and HEPES buffer pH 7.5

(0.5, 1.0 or 1.5 M) at 30 �C. Sampling procedures were

performed in the same way as in the standard CV2025 x-

TAm bioconversion.

ISPR using reduced pressure

The 5 torr reduced pressure reactions were carried out at

the same 300 lL total reaction volume with 50 mM MBA,

50 or 70 mM Ery, 1 g L-1 CV2025 x-TAm, 0.2 mM PLP

and 50 mM HEPES buffer (pH 7.5) at 30 �C. To apply the

necessary vacuum, the microwell plate was placed in an

automation-compatible vacuum manifold (Sigma-Aldrich,

Gillingham, UK) which was connected to a vacuum pump

(Millipore, Gillingham, UK) and placed in an incubator

(Gallenkamp Economy Size 2 Fan Incubator, Severn Sales,

Bristol, UK) for temperature control. Sampling procedures

were performed in the same way as in the standard CV2025

x-TAm bioconversion.

ISPR using adsorptive resins

Reactions were carried out similar to the standard CV2025

x-TAm bioconversion with the addition of various mass

loadings (i.e. 30, 60, 100, 200, and 1,000 mg) of adsorbent

resins (Amberlite TM XAD 7, XAD 1180 or XAD 16) and

were shaken at 500 rpm. Prior to each experiment a pre-

treatment procedure was conducted to wet the internal

pores of the resins. The dry resins were soaked with suf-

ficient volume of methanol for 15 min. Then, the methanol

was decanted and the resins soaked with distilled water for

a further 5–10 min before being finally replaced with fresh

distilled water and being left in solution until use. In order

to obtain kinetic profiles, reactions were conducted using a

sacrificial well procedure, where for each reaction several

identical reaction wells were prepared and the entire con-

tent of a single well was used for sampling at each time

interval. At every sampling point, a portion of the reaction

solution was quenched as described in the standard

CV2025 x-TAm bioconversion and was kept for HPLC

analysis. Any solution left in the well was discarded (fil-

tered with 0.2 lm filter (Pall, MI, USA)), and the well was

refilled with isopropyl alcohol to desorb bound compounds

from the resins. The adsorbent in isopropyl alcohol solution

was left shaking on a thermomixer shaker for an hour prior

to sampling for determination of concentrations of adsor-

bed compounds onto the resins by HPLC.

Analytical methods

A Dionex (Camberly, UK) microbore HPLC system con-

trolled by Chromeleon client 6.80 software was employed

for all HPLC analysis. This system is comprised of a P680

gradient pump, ASI-100 automated sampler injector, STH

585 column oven and UVD170U UV detector. MBA, AP

and (R)-1-phenylethanol were analysed using an ACE 5

C18 reverse phase column (150 mm 9 4.6 mm, 5 lm

particle size; Advance Chromatography Technologies,

Aberdeen, UK). A gradient was run from 15 % acetonitrile/

85 % 0.1 % (v/v) TFA to 72 % acetonitrile/28 % TFA

over 8 min, followed by a re-equilibrium step for 2 min

(oven temperature, 30 �C, flow rate 1 mL min-1). Detec-

tion was performed by UV absorption at 210 nm (MBA)

and 250 nm (AP and (R)-1-phenylethanol). The retention

times under these conditions were MBA, 3.4 min, AP,

7.28 min and 1-phenylethanol 6.45 min.

Erythrulose (Ery) was analysed with an Aminex HPX-

87H Ion exchange column (300 mm 9 7.8 mm, Biorad,

Hemel Hampstead, UK). 0.1 % (v/v) TFA was used as

mobile phase for isocratic elution with an oven temperature

of 60 �C and a flow rate of 0.6 mL min-1. Detection was

performed at a wavelength of 210 nm and the retention

time was 11.63 min.

2-Amino-1,3,4-butanetriol (ABT) was analysed using an

ACE 5 C18 reverse-phase column with UV detection at

254 nm and oven temperature of 37 �C. Mobile phases

used were A = 140 mM sodium acetate, pH 5.05 and

B = 100 % acetonitrile. A gradient was run from 85 % A

to 100 % A over 10 min, flow rate 0.5 mL min-1, fol-

lowed by a column wash phase and re-equilibrium step at

1 mL min-1. Samples for analysis were first derivatised

using 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate

(AQC) as the derivatizing agent. This derivatising agent

was synthesised in house according to the protocol detailed

by [2]. ABT samples were mixed with borate buffer and

briefly vortexed. A clean tip was used to reconstituted AQC

reagent, which was first prepared prior to use by mixed

AQC reagent powder with dried acetonitrile as reagent

diluent and heated at 55 �C until the powder dissolves. The

solution was then immediately vortexed for few seconds

and incubated at room temperature for 1 min prior to

transfer to a HPLC vial. The retention time of derivatised

ABT is 8.9 min. In order to confirm the identity of the ABT

produced in the bioconversions, it was purified and ana-

lysed by mass spectrometry [7].

Results and discussion

Standard CV2025 x-TAm bioconversion kinetics

and yield

In order to illustrate the problem of low-yielding biocon-

versions the synthesis of ABT using MBA as amino donor

is used throughout this work as an exemplar reaction.


123

Initial experiments with the CV2025 x-TAm were per-

formed at equimolar substrate concentrations of Ery and

MBA as shown in Fig. 2. Only a 26 % (mol mol-1) con-

version yield was achieved using substrate concentrations

of 50 mM. This low yield was repeatable for reactions

carried out at both microwell (300 lL) and conventional

laboratory scales.

To obtain further insight into the reasons for the low yield,

issues of substrate and product inhibition were investigated

by exogenously adding individual compounds into the

reaction mixture before the biotransformation was initiated.

Substrate inhibition by Ery was not observed at concentra-

tions below 300 mM [16] so it can be considered negligible

under the reaction conditions used here. Significant inhibi-

tion by the amino donor MBA [15] and by-product AP (see

Fig. 3) was found. AP was particularly inhibitory at con-

centrations above 5 mM while MBA was inhibitory above

10 mM [15]. This inhibitory effect of AP has also been

reported in the literature with x-TAm from Bacillus thur-

ingiensis JS64 [21]. We have previously determined the

inhibition constants for MBA and AP in this reaction to be

4 9 10-3 and 1.1 mM, respectively [16]. Based on these

results the main design constraint on the bioconversion is the

need to avoid formation of AP or to remove it from the

reaction medium as soon as it is produced.

To increase the product yield, four bioprocess options

were thus investigated using microscale methodologies as

depicted in Fig. 1. These involve evaluation of alternative

amino donors, implementation of a second enzyme system

for in situ co-product removal (IScPR) and implementation

of two ISPR methods via either bioconversions at reduced

pressure or addition of adsorbent resins.

Evaluation of alternative amino donors

Five amino donors were investigated as an alternative to

MBA under otherwise standard CV2025 x-TAm biocon-

version conditions again using Ery as the amino acceptor.

The objective was to identify amino donors that were not

themselves inhibitory and which led to formation of non-

inhibitory by-products (Fig. 1, Option 1). Furthermore, in

industrial asymmetric synthesis the choice of amino donor

is important as this will impact on the overall process cost.

As shown in Fig. 4, CV2025 x-TAm bioconversions

showed higher conversion yields with all the alternative

amino donors evaluated apart from the aromatic compound

benzylamine.

When MBA was substituted with the cheaper, more

volatile and more water-soluble substrate, isopropylamine

(IPA), a nearly threefold increase in final bioconversion

yield was attained. This is most likely because of reduced

inhibition with the IPA used and/or the corresponding by-

product (acetone in this case) produced. Nevertheless, the

reaction required almost twice the time to reach the final

yield. It is anticipated that this time could be reduced using

a more concentrated CV2025 x-TAm lysate, although this

would ultimately add to the process cost or by use of dif-

ferent substrate ratios, i.e. an increased ratio of amino donor

Fig. 2 Standard bioconversion kinetics using CV2025 x-TAm clar-

ified lysate with Ery as amino acceptor and MBA as amine donor: Ery

consumption (cross symbols); MBA consumption (diamonds); ABT

production (triangles) and AP production (squares). Reaction condi-

tions: 0.3 g L-1 CV2025 x-TAm; 0.2 mM PLP; [MBA] and [Ery]

were 50 mM each; 30 �C; pH 7.5; shaken at 300 rpm in 50 mM

HEPES buffer; total volume of 300 lL in a 96-glass microwell plate.

Bioconversions performed as described in ‘‘Enzyme reactions and

microscale methods’’. Error bars represent one standard deviation

about the mean (n = 3)

Fig. 3 Quantification of product inhibition effects by AP. Reactions

carried out with equilmolar [Ery] and [MBA] at 50 mM each with the

addition of specified initial concentrations of [AP] (0, 10, 20, 30,

40 mM). Reaction conditions: 0.3 g L-1 CV2025 x-TAm in lysate

form; 0.2 mM PLP; 30 �C; pH 7.5; shaken at 300 rpm in 50 mM

HEPES buffer; total volume of 300 lL in a 96-glass microwell plate.


microscale methods’’


123

can be expected to promote higher reaction rates [15, 22]. A

further advantage of IPA, and the corresponding acetone

by-product, is that both could be more easily removed

during downstream processing owing to their high volatil-

ities compared with aromatic amine donors [22].

More limited increases in bioconversion yields of 40 and

46 % (mol mol-1), respectively, were also attained with

aliphatic amines S-(?)-sec-butylamine and L-a-serine,

which are naturally occurring amino acids. The use of

amino acids as an amine donor is noteworthy considering

their availability at relatively low cost and the strategies

available for their selective extraction [10]. MBA was also

substituted with two other aromatic amino donors: S-(?)-1-

aminoindane and benzylamine. CV2025 x-TAm biocon-

versions with S-(?)-1-aminoindane showed nearly an equal

conversion yield to the reaction with S-(?)-1-sec-butyl-

amine while benzylamine showed a yield of approximately

8 % (mol mol-1) considerably lower than with MBA.

Overall, these observations confirm the broad spectrum of

(S)-amines utilised by the CV2025x-TAm and corresponding

increases in reaction yield. The range of potential substrates

available is considered particularly useful for the synthesis of

enantiopure amines and amino alcohols [10, 31, 33].

Second enzyme system for in-situ co-product removal

For in situ removal of the inhibitory AP, another option is

to convert it into the non-inhibitory achiral alcohol (R)-1-

phenylethanol [33] by adding a second enzyme system to

the bioconversion reactor (Fig. 1). Commercially available

ADH from Lactobacillus kefir is an NADPH-dependent

enzyme that can utilise AP as substrate and effective co-

factor regeneration systems have been developed for use

with ADH [28].

In general, there are two methods to recycle the co-

factor: the coupled-enzyme process and the coupled-sub-

strate approach [5]. In the coupled-substrate process, an

additional substrate (i.e. 2-propanol) is added to recycle the

co-factor. In this case the maximum conversion is limited

by the thermodynamics of the system, as there is a com-

petition between substrate, product, co-substrate and co-

product and by inhibition or deactivation of the newly

added co-substrate to the x-TAm. Meanwhile, in the cou-

pled-enzyme process, two independent enzymes are

involved. In addition to the ADH, a third enzyme, glucose

dehydrogenase (GDH) from Pseudomonas sp., was selec-

ted to facilitate recycling of the co-factor NADPH (Fig. 1,

Option 2). This enzyme was chosen as it has high specific

activity [33], favourable equilibrium for NADPH [28], no

known inhibitory effect of the glucose co-substrate on x-

TAm activity [33] and commercial availability.

A further consideration when implementing the NADPH

recycling system is that good control of pH is needed

because the gluconolactone product formed from glucose

(Fig. 1, Option 2) is spontaneously hydrolysed to gluconic

acid causing a decrease in pH [33]. The optimum pH for

CV2025 x-TAm activity in lysate form is pH 7.5 with

activity decreasing rapidly at lower pH values [10]. As

shown in Fig. 5, when bioconversions were carried out

using the second enzyme system in 50 mM HEPES buffer

(initial pH 7.5), the reaction stopped after 5 h at a yield

comparable to the standard batch CV2025 x-TAm bio-

conversion. The production of (R)-1-phenylethanol by

ADH, which was 13 mM, is at the level where AP became

inhibitory to the CV2025 x-TAm (Fig. 3). At this time the

measured pH of the bioconversion medium had declined to

pH 4.5, a value at which the CV2025 x-TAm is expected to

be deactivated and thus prevented the reaction progressing

any further [7]. Nevertheless, this yield using the second

enzyme system was accomplished in approximately one-

third of the time (6 h) required for the standard CV2025 x-

TAm reaction (16 h). This confirmed the second enzyme

system behaved as anticipated in which the AP produced

was successfully converted into the non-inhibitory (R)-1-

phenylethanol, but points to the need to more tightly con-

trol pH values during these microscale reactions mainly to

prevent the enzyme deactivation.

For preparative scale bioconversions and above, pH-stat

devices can easily be used for control of pH and mainte-

nance of the optimum pH as the reaction progresses [25].

However, such approaches are not feasible with the high-

Fig. 4 Evaluation of alternative amino donors on ABT product

formation kinetics at the microwell scale: MBA (filled diamonds);

IPA (filled squares); benzylamine (filled circles); S-(?)-sec-butyl-

amine (open circles); S-(?)-1-aminoindane (open squares) and L-a-

serine (filled triangles). Reaction conditions: 0.3 g L-1 CV2025 x-

TAm in lysate form; 0.2 mM PLP; [amino donor] and [Ery] were

50 mM each; 30 �C; pH 7.5; shaken at 300 rpm in 50 mM HEPES

buffer; total volume of 300 lL in a 96-glass microwell plate.


microscale methods’’. Error bars represent one standard deviation

about the mean (n = 3)


123

throughput, microwell scale methods developed here. In

order to achieve the necessary pH control in this case

higher buffer concentrations were investigated and the

results verified in separate pH-stat devices operated at a

30-mL scale [7]. Consequently, bioconversions employing

the second enzyme system were operated at HEPES buffer

concentrations of 0.5–1.5 M. All the substrates, co-factors

and enzymes (apart from the CV2025 x-TAm lysate) were

prepared at the desired HEPES buffer concentrations and

adjusted to pH 7.5 prior to use.

As depicted in Fig. 6, the reaction with the lowest

HEPES concentration of 0.5 M proceeded to only 50 %

(mol mol-1) conversion and the pH value at the end of

reaction had decreased to 7.0. At 1.0 M HEPES the opti-

mum reaction yield was 36 mM (72 % mol mol-1) and the

final pH value was found to be 7.2. Finally, for the reaction

with 1.5 M HEPES the product yield increased to 43 mM

ABT (86 % mol mol-1) and the pH only decreased to 7.4.

In this case, the time needed to achieve this maximum yield

was also quicker by 8 h in comparison with the reactions at

lower HEPES concentrations.

Since substrate inhibition with the amino acceptor (Ery)

used is expected to be negligible [16], the effect of using

higher initial concentrations of Ery to increase conversion

of the amino donor used (MBA in this case) as well as

increasing the final product concentration (i.e. ABT) was

also examined. When the reaction in 1.5 M HEPES was

repeated at an initial Ery concentration of 70 mM there was

almost complete conversion of 50 mM MBA and a final

product concentration of 49 mM ABT was obtained in 20 h

as also shown in Fig. 6. The 40 % (mol mol-1) excess of

Ery added at the start of the reaction helped further

increase the initial reaction rate to nearly twofold over that

when the substrates were initially present at equal molar

ratio. This finding was repeatable in larger scale pH-stat

experiments suggesting that the use of higher buffer con-

centrations in microwell experiments is an acceptable way

to examine potential process benefits arising from imple-

mentation of second enzyme systems.

ISPR using reduced pressure

The third bioprocess option investigated was operation at

reduced pressure to facilitate evaporation of unfavourable or

Fig. 5 Influence of second enzyme system on CV2025 x-TAm

bioconversion kinetics: (R)-1-phenylethanol (filled triangles) produc-

tion from the preliminary study where the pH dropped to 4.5 after 5 h

reaction and AP (filled circles) production from standard CV2025 x-

TAm reaction (‘‘Standard CV2025 x-TAm bioconversion kinetics

and yield’’) where the pH remained constant throughout the reaction.

Reaction conditions: 0.3 g L-1 CV2025 x-TAm in lysate form;

0.2 mM PLP; [MBA] and [Ery] were 50 mM each; 1 mM NADPH;

70 mM glucose; ADH (1.2 g L-1); GDH (0.02 g L-1); 30 �C; pH

7.5; shaken at 300 rpm in 50 mM HEPES buffer; total volume of

300 lL in a 96-glass microwell plate. Bioconversions performed as

described in ‘‘Enzyme reactions and microscale methods’’

Fig. 6 Influence of buffer concentration on measured CV2025 x-

TAm bioconversion kinetics using the second enzyme system: a Ery

consumption and b ABT production. The reaction mixture contained

50 mM MBA, 50 mM Ery in different HEPES (pH 7.5) concentra-

tions of 0.5 M (filled squares) 1.0 M (filled circles) or 1.5 M (filled

triangles) and 70 mM Ery in 1.5 M HEPES (filled diamonds).

Reaction conditions: 0.3 g L-1 CV2025 x-TAm; 0.2 mM PLP;

1 mM NADPH; 70 mM glucose; ADH (1.2 g L-1); GDH

(0.02 g L-1); 30 �C; pH 7.5; shaken at 300 rpm; total volume of

300 lL in a 96-glass microwell plate. Bioconversions performed as

described in ‘‘Enzyme reactions and microscale methods’’


123

inhibitory co-product from the system. This option, how-

ever, is only applicable when the co-product is more volatile

than the other substrates and the target product. This is the

case here for the inhibitory AP by-product [32]. 300 lL

scale reactions were performed in un-covered 96-glass mi-

crowell plate integrated with an automation compatible

96-well vacuum manifold as depicted in Fig. 1, Option 3.

Considering the limited bioconversion volume, a ‘mild’

vacuum was initially used of 5 torr (*6.67 mbar) to

minimise water evaporation from the reaction medium.

Control experiments showed that the average rates of water

evaporation under standard and reduced pressure condi-

tions were 0.9 and 6.8 mg hr-1, respectively [7]. Ery,

MBA and ABT did not show any measurable evaporation

over 16 h while the evaporation rate of AP was quantified

to be 1.22 mM hr-1 suggesting the possibility of selective

removal. All substrate and product concentrations reported

here are corrected for water evaporation.

Kinetic profiles of the CV2025 x-TAm reaction carried

out at 5 torr are shown in Fig. 7. To minimise any artefacts

due to water evaporation a high CV2025 x-TAm concen-

tration (i.e. 1.0 g L-1) was used to ensure the reaction

could reach completion in 24 h or less. When equivalent

concentrations (i.e. 50 mM) of both substrates were used,

the ABT yield was 28 mM in 12 h. This was equivalent to

a 30 % (mol mol-1) increase over the yield of the standard

CV2025 x-TAm bioconversion (Fig. 2). When the initial

Ery concentration was increased to 70 mM, as previously

observed with the second enzyme reaction system, the

product concentration increased slightly further to 33 mM

ABT, some 40 % (mol mol-1) higher than the standard

CV2025 x-TAm bioconversion.

By comparison with the results obtained using the second

enzyme system (Fig. 6) and the study on AP inhibition

(Fig. 3), it is likely that the relatively small increases in

bioconversion yield at reduced pressure are a consequence

of only limited removal of the inhibitory AP by evaporation.

While more AP could be removed by operating at lower

pressures, this would lead to further problems with water

evaporation, especially at the microwell scale. Neverthe-

less, the method established here still enables the potential

benefits from operation at reduced pressure to be quantita-

tively evaluated in a high throughput, automated manner.

ISPR using adsorbent resins

The final approach investigated to alleviate low biocon-

version reaction yields is ISPR using adsorbent resins

(Fig. 1, Option 4). This relies on selective binding of the

unwanted by-product onto a solid-phase resin [13, 29]

added into each microwell and kept in suspension by

shaking. Three commercially available hydrophobic poly-

meric resins were investigated: AmberliteTM XAD 7, XAD

1180 and XAD 16. These were chosen particularly because

they have been previously shown to adsorb AP from the

reaction solution [29] and because bound AP could sub-

sequently be desorbed using isopropyl alcohol.

Adsorption isotherm studies were initially conducted to

measure the capacity and selectivity of each resin towards

the various solutes present in the bioconversions [4, 30].

Based on these studies, the equilibrium binding capacity

for AP on all the selected resins was much greater than

MBA, Ery and ABT [7]. From the resins examined, XAD

16 demonstrated the highest binding capacity for AP

(0.31 g g-1) with XAD 7 and XAD 1180 showing signif-

icantly less binding at around 0.04 and 0.07 g g-1,

respectively. The measured binding capacities for Ery and

MBA on XAD-16 were 0.0009 and 0.0002 g g-1, respec-

tively. Considering that solute binding to the beads should

be reversible, any bound substrates should partition off the

resin and back into solution as the reaction proceeds.

Fig. 7 Influence of ISPR by operation at reduced pressure on

CV2025 x-TAm bioconversion kinetics: a MBA (filled circles)

consumption and AP (open circles) production in a reaction using

50 mM Ery and MBA (filled triangles) consumption and AP (cross

symbols) production in a reaction using 70 mM Ery; b ABT

production (values plotted are corrected for evaporation) in reactions

using 50 mM Ery (filled circles) and 70 mM Ery (filled triangles).

Experiments performed with 50 mM MBA; Ery at either 50 or

70 mM; 1 g L-1 CV2025 x-TAm in lysate form; 0.2 mM PLP and

50 mM HEPES buffer. Reaction conditions: 30 �C; pH 7.5; 5 torr

vacuum; total volume of 300 lL in a 96-glass microwell plate using

the vacuum manifold shown in Fig. 1


123

Likewise, any AP formed should be adsorbed by the resin,

at least until the binding capacity of the beads is reached,

thus driving the reaction towards completion [29].

In order to obtain a complete kinetic profile regarding

the amount of solute adsorbed onto the resins, a ‘sacrificial

well’ approach was used. At each sampling interval, all the

liquid in the well was carefully aspirated off and isopropyl

alcohol immediately added. The microwells were then left

to shake for 5 h to ensure all solute was fully desorbed.

Besides isopropyl alcohol, nearly any water-miscible

organic solvent can be used to desorb the adsorbate from

the AmberliteTM XAD resins [4, 27].

Preliminary bioconversions with CV2025 x-TAm clar-

ified lysate in the presence of 20 % (w w-1) adsorbent

resins were found to only proceed up to 5–10 % (mol -

mol-1) conversion (data not shown). These were lower

than the level achieved in the standard batch bioconversion

which reached 26 % (mol mol-1) conversion (Fig. 2). The

reason for this low yield was unclear since the concentra-

tion of total protein in the supernatant remained constant,

i.e. the enzyme itself did not bind to the resin. One pos-

sibility could be a selective binding of the PLP co-factor

although this was not explored further.

Consequently, experiments were also performed using a

whole cell form of the CV2025 x-TAm biocatalyst. With a

mean diameter of 0.2–0.5 lm the individual cells should

not enter the pores of the resin which are approximately

0.01 lm [14, 27]. Bioconversions in the presence of 20 %

(w/w) of either XAD 7 or XAD 16 reached a 50 % (mol -

mol-1) conversion whilst a marginally lower conversion of

46 % (mol mol-1) was attained with XAD 1180 Fig. 8a.

These yields represent 12 and 8 % (mol mol-1) improve-

ments, respectively, over the standard CV2025 x-TAm

whole cell bioconversion without the resin adsorption [7].

As shown in Fig. 8b AP was totally removed from the

aqueous bioconversion medium owing to the presence of

the resins. MBA was also adsorbed onto each resin at

equivalent aqueous phase concentrations of 13 mM (XAD

7), 11 mM (XAD 16) and 17 mM (XAD 1180) as shown in

Fig. 8c. Thus concentrations of MBA in excess of at least

30 mM remained in the aqueous solution to act as amino

donors. Likewise, approximately the same amount of Ery

was adsorbed onto each resin leaving approximately

40 mM of substrate free in the aqueous solution.

Further bioconversions were performed at increased

mass loadings of resin. These studies indicated that the

reactions with 20 % (w w-1) resin (as described above)

were the best overall. Reactions attempted with 33.3 and

66.7 % (w w-1) loaded resins failed to increase the reac-

tion conversion higher than the standard bioconversion

(data not shown). Even so, a 20 % (w w-1) resin mass

loading is very high for commercial application [8]. Should

implementation of ISPR have led to significant increases in

Fig. 8 Influence of ISPR using adsorbent resins on CV2025 x-TAm

whole cell bioconversion kinetics: XAD 7 (filled circles); XAD 1180

(filled triangles); XAD 16 (filled squares) addition compared with

standard whole cell bioconversion (filled diamonds). a ABT produc-

tion (total concentration based on both aqueous and resins phases);

b AP production (in the aqueous phase only); c MBA concentrations

adsorbed in resin phase. Reaction conditions: 0.3 g L-1 CV2025 x-

TAm in whole cell form; 0.2 mM PLP; [MBA] and [Ery] were

50 mM each, 60 mg (wetted) adsorption resin; 30 �C; pH 7.5; shaken

at 500 rpm in 50 mM HEPES buffer; total volume of 300 lL in

96-glass microwell plate. Bioconversions performed as described in

‘‘Enzyme reactions and microscale methods’’. Error bars represent

one standard deviation about the mean (n = 3)


123

reaction yield further experiments would be needed to

screen a wider range of resins and to optimise (minimise)

the bead loading while still maximising the reaction yield.

Dual substrate and product bioconversions as studied here

represent particularly challenging ISPR applications unless

highly selective resins are available [13]. Literature sug-

gests ISPR is most successful for bioconversions involving

single substrate and to overcome issues of substrate/prod-

uct toxicity to whole cell biocatalysts [6, 17, 29].

Conclusions

This work has established a series of microscale experi-

mental methods (Fig. 1) to rapidly evaluate bioprocess

options to increase the yield of bioconversions that exhibit

either unfavourable reaction equilibria or strong substrate

and/or product inhibition. It has also demonstrated how the

methods can be used to optimise bioconversion conditions

quickly and cost effectively using only minimal quantities

of material. The utility of the methods were demonstrated

for chiral amine synthesis using the CV2025 x-transami-

nase. The increases in yield obtained with an exemplar

reaction for each of the bioprocess options investigated are

summarised in Table 1. In this particular case, implemen-

tation of the second enzyme reaction (‘‘Second enzyme

system for in-situ co-product removal’’) resulted in the

highest increase in product yield followed closely by the

use of an alternative amino donor which is probably the

fastest and most economically viable bioprocess option to

implement industrially. Most important, however, is the

fact that the microscale methods described here are generic

and can in principle be applied to a wide range of bio-

conversions. Current work is addressing the automation of

these methods so that each of the bioprocess options can be

evaluated in parallel for libraries of different enzymes.

Acknowledgments The Ministry of Higher Education of Malaysia

(MOHE) and The Mexican National Council for Science and

Technology (CONACYT) are acknowledged for the studentships

provided to support MH and LRS, respectively. The UK Engi-

neering and Physical Sciences Research Council (EPSRC) is

thanked for the support of the multidisciplinary Biocatalysis Inte-

grated with Chemistry and Engineering (BiCE) programme (GR/

S62505/01) at University College London (London, UK). Aspects of

the work were also supported by the European Union’s Seventh

Framework Programme FP7/2007-2013 under Grant Agreement No.

266025 (BIONEXGEN).

References

1. Bea HS, Seo Y-M, Kim B-G, Yun H (2010) Kinetic resolution of

a-methylbenzylamine by recombinant Pichia pastoris expressing

x-transaminase. Biotechnol Bioprocess Eng 15(3):429–434

2. Cohen SA, Michaud DP (1993) Synthesis of a fluorescent de-

rivatizing reagent, 6-aminoquinolyl-N-hydroxysuccinimidyl car-

bamate, and its application for the analysis of hydrolysate amino

acids via high performance liquid chromatography. Anal Bio-

chem 211:279–287

3. Dominik K, Ivan L, Dorina C, David R, Wolfgang K (2008)

Asymmetric synthesis of optically pure pharmacologically rele-

vant amines employing x-transaminases. Adv Synth Catal

350:2761–2766

4. Du X, Yuan Q, Li Y, Zhou H (2008) Preparative purification of

solanesol from tobacco leaf extracts by macroporous resins.

Chem Eng Technol 31(1):87–94

5. Faber K (2004) Biotransformations in organic chemistry.

Springer, Berlin

6. Guo JL, Mu X-Q, Xu Y (2009) Integration of newly isolated

biocatalyst and resin-based in situ product removal technique for

the asymmetric synthesis of (R)-methyl mandelate. Bioprocess

Biosyst Eng 33(7):797–804

7. Halim M (2012) Microscale approaches to the design and opti-

mization of equilibrium controlled bioconversions. PhD Thesis,

University College London, London

8. Hilker I, Wohlgemuth R, Alphand V, Furstoss R (2005) Micro-

bial transformations 59: first kilogram scale asymmetric micro-

bial Baeyer–Villiger oxidation with optimized productivity using

a resin-based in situ SFPR strategy. Biotechnol Bioeng 92(6):

702–710

9. Hohne M, Kuhl S, Robins K, Bornscheuer UT (2008) Efficient

asymmetric synthesis of chiral amines by combining transami-

nase and pyruvate decarboxylase. ChemBioChem 9:363

10. Kaulmann U, Smithies K, Smith MEB, Hailes HC, Ward JM

(2007) Substrate spectrum of x-transaminase from Chromobac-

terium violaceum DSM30191 and its potential for biocatalysis.

Enzyme Microb Technol 41:628–637

11. Koszelewski D, Tauber K, Faber K, Kroutil W (2010) x-Trans-

aminases for the synthesis of non-racemic a-chiral primary

amines. Trends Biotechnol 28:324–332

12. Lye GJ, Shamlou PA, Baganz F, Dalby PA, Woodley JM (2003)

Accelerated design of bioconversion processes using automated

microscale technique. Trends Biotechnol 21:1

13. Lye GJ, Woodley JM (1999) Application of in situ product-

removal techniques to biocatalytic processes. Trends Biotechnol

17(10):395–402

14. Parsons SA, Jarvis PJ, Dixon DW, Sharp E (2007) Treatment of

waters with elevated organic content. J Am Water Works Assoc

15. Rios SL, Halim M, Cazare A, Morris P, Ward JM, Hailes H,

Dalby PA, Baganz F, Lye GJ (2011) A microscale toolbox for

rapid evaluation of multi-step enzymatic syntheses comprising a

Table 1 Summary of increases in product concentration and bio-

conversion yield obtained using the various microscale methods

Bioprocess option

investigated

Maximum [ABT]

achieved (mM)

Fold increase

in yield

Use of alternative amino

donor (IPA)

36 2.8

Use of second enzyme

system

43 3.3

ISPR under reduced

pressure

28 2.2

ISPR using adsorbent

resins

25 1.3

Yield enhancements calculated relative to the product concentration

(13 mM) obtained in the standard CV2025 TAm lysate bioconversion

using initial [Ery] and [MBA] concentrations of 50 mM each


123

mix and match E. coli expression system. Biocatal Biotransform

29(5):192–203

16. Rios SL, Bayir N, Halim M, Du C, Ward JM, Baganz F, Lye GJ

(2013) Non-linear kinetic modelling of reversible bioconversions:

application to the transaminase catalysed synthesis of chiral

amino-alcohols. Biochem Eng J 73:38–48

17. Rudroff F, Alphand V, Furstoss R, Mihovilovic MD (2006)

Optimizing fermentation conditions of recombinant Escherichia

coli expressing cyclopentanone monooxygenase. Org Process Res

Dev 10:599–604

18. Shin JS, Kim BG (2009) Transaminase-catalyzed asymmetric

synthesis of L-2-aminobutyric acid from achiral reactants. Bio-

technol Lett 31(10):1595–1599

19. Shin JS, Kim BG (2002) Substrate inhibition mode of w-trans-

aminase from Vibrio fluvialis JS17 is dependent on the chirality

of substrate. Biotechnol Bioeng 77(7):832–837

20. Shin JS, Kim BG (2001) Comparison of the x-transaminases

from different microorganisms and application to production of

chiral amines. Biosci Biotechnol Biochem 65:1782–1788

21. Shin JS, Kim BG, Liese A, Wandrey C (2001) Kinetic resolution

of chiral amines with x-transaminase using an enzyme-mem-

brane reactor. Biotechnol Bioeng 73(3):179–187

22. Smith MEB, Chen BH, Hibbert EG, Kaulmann U, Smithies K,

Galman JL, Baganz F, Dalby PA, Hailes HC, Lye GJ, Ward JM,

Woodley JM, Micheletti M (2010) A multidisciplinary approach

toward the rapid and preparative-scale biocatalytic synthesis of

chiral amino alcohols: a concise transketolase-/x-transaminase-

mediated synthesis of (2S,3S)-2-aminopentane-1,3-diol. Org

Process Res Dev 14:99–107

23. Sutin L, Anderson S, Bergquist L, Castro VM, Danielsson E,

James S, Henriksson M, Johansson L, Kaiser C, Flyren K, Wil-

liams M (2007) Oxazolones as potent inhibitors of 11b-

hydroxysteroid dehydrogenase type 1. Bioorg Med Chem Lett

17:4837–4840

24. Tufvesson P, Ramos JL, Jensen JS, Al-Haque N, Neto W,

Woodley JM (2011) Process considerations for the asymmetric

synthesis of chiral amines using transaminases. Biotechnol Bio-

eng 108(7):1479–1493

25. Truppo MD, Turner NJ (2010) Micro-scale process development

of transaminase catalysed reactions. Org Biomol Chem 8:

1280–1283

26. Truppo MD, Rozzell DJ, Moore J, Turner NJ (2009) Rapid

screening and scale-up of transaminase catalysed reactions. Org

Biomol Chem 7(2):395–398

27. Vicenzi JT, Zmijewski MJ, Reinhard MR, Landen BE, Muth WL,

Marler PG (1997) Large-scale stereoselective enzymatic ketone

reduction with in situ product removal via polymeric adsorbent

resins. Enzyme Microb Technol 20:495–499

28. Weckbecker A, Hummel W (2006) Cloning, expression, and

characterisation of an (R)-specific alcohol dehydrogenase from

Lactobacillus kefir. Biocatal Biotransform 24(5):380–389

29. Yang ZH, Zeng R, Chang X, Li X-K, Wang G-H (2008) Toxicity

of aromatic ketone to yeast cell and improvement of the asym-

metric reduction of aromatic ketone catalysed by yeast cell with

the introduction of resin adsorption. Food Technol Biotechnol

46(3):322–327

30. Yang WC, Shim WG, Lee JW, Moon H (2003) Adsorption and

desorption dynamics of amino acids in a non-ionic polymeric

sorbent XAD-16 column. Korean J Chem Eng 20(5):922–929

31. Yun H, Kim J, Kinnera K, Kim BG (2005) Synthesis of enatio-

merically pure trans-(1R,2R)-and cis-(1S, 2R)-1-amino-2-indanol

by lipase and x-transaminase. Biotechnol Bioeng 93(2):391–395

32. Yun H, Cho B-K, Kim B-G (2004) Kinetic resolution of (R, S)-

sec-butylamine using omega-transaminase from Vibrio fluvialis

JS17 under reduced pressure. Biotechnol Bioeng 87(6):772–778

33. Yun H, Hun Y, Cho B-K, Hwang B-Y, Kim B-G (2003)

Simultaneous synthesis of enantiomerically pure (R)-1-phenyl-

ethanol and (R)-a-methylbenzylamine from racemic a-methyl-

benzylamine using x-transaminase/alcohol dehydrogenase/

glucose dehydrogenase coupling reaction. Biotechnol Lett 25:

809–814


123

microscale methods to rapidly evaluate bioprocess options for increasing bioconversion yields:...

Documents