International Journal of Pharmaceutics 509 (2016) 41–49

Suppression of protein inactivation during freezing by minimizing pHchanges using ionic cryoprotectants�Lubica Krauskováa,c, Jitka Procházkováa, Martina Klaškováa, Lenka Filipováa,Radka Chaloupkováb,c, Stanislav Malýd, Ji�rí Damborskýb,c, Dominik Hegera,c,*aDepartment of Chemistry, Faculty of Science, Masaryk University, Kamenice 5/A8, 625 00 Brno, Czech Republicb Loschmidt Laboratories, Department of Experimental Biology, Faculty of Science, Masaryk University, Kamenice 5/A13, 625 00 Brno, Czech RepubliccResearch Centre for Toxic Compounds in the Environment RECETOX, Faculty of Science, Masaryk University, Kamenice 5/A29, 625 00 Brno, Czech RepublicdCentral Institute for Supervising and Testing in Agriculture, Hroznová 2, CZ-656 06, Czech Republic

A R T I C L E I N F O

Article history:Received 4 January 2016Received in revised form 9 May 2016Accepted 16 May 2016Available online 17 May 2016

Keywords:Protein cryopreservationFreezing/thawingProtein aggregationProtein stabilityQuality by designFreezing stressFreezing-induced pH shift

A B S T R A C T

Freezing and lyophilization are often used for stabilization of biomolecules; however, this sometimesresults in partial degradation and loss of biological function in these molecules. In this study weexamined the effect of freezing-induced acidity changes on denaturation of the model enzymehaloalkane dehalogenase under various experimental conditions. The effective local pH of frozensolutions is shown to be the key causal factor in protein stability. To preserve the activity of frozen-thawed enzymes, acidity changes were prevented by the addition of an ionic cryoprotectant, a compoundwhich counteracts pH changes during freezing due to selective incorporation of its ions into the ice. Thisapproach resulted in complete recovery of enzyme activity after multiple freeze-thaw cycles. We proposethe utilization of ionic cryoprotectants as a new and effective cryopreservation method in researchlaboratories as well as in industrial processes.

ã 2016 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

International Journal of Pharmaceutics

journal homepage: www.elsev ier .com/locate / i jpharm

1. Introduction

Although lyophilization is a key fundamental process in proteinproduction, it still faces serious challenges in the freezing anddrying process (Cicerone et al., 2015; Kasper et al., 2013; Kolheet al., 2010b); some of the stresses that proteins undergo duringlyophilization are unavoidable (low temperature, crystal forma-tion, increased concentration, water removal), but changes in pHduring freezing can be minimized to mitigate unnecessarydegradation (Bhatnagar et al., 2007; Cicerone et al., 2015; Kasperand Friess, 2011). Up to now, no general approach has existed forregulating the pH of frozen solutions. If uncontrolled, the pH in thefreeze-concentrated solution (FCS) in the veins surrounding icecrystals can be greatly altered compared to that of the originalaqueous solution and this can detrimentally influence the othersolutes present (Bronshteyn and Chernov, 1991; Heger et al., 2006;Reinsch et al., 2015; Takenaka et al., 2006; Vajda, 1999; Williams-Smith et al., 1977). Such pH changes can be of the utmost

* Corresponding author at: Department of Chemistry and RECETOX, Faculty ofScience, Masaryk University, Kamenice 5/A8, 625 00 Brno, Czech Republic.

E-mail address: hegerd@chemi.muni.cz (D. Heger).

http://dx.doi.org/10.1016/j.ijpharm.2016.05.0310378-5173/ã 2016 Elsevier B.V. All rights reserved.

importance, not only in protein storage and production, but also inthe fields of food conservation (Soltanizadeh et al., 2014),transfusion medicine (Asghar et al., 2014), and studies on theorigin of self-replicating life (Attwater et al., 2013; Mutschler et al.,2015) as well as for bromine activation and ozone depletion eventsin natural ice (Abbatt et al., 2012; Bartels-Rausch et al., 2014; Prattet al., 2013).

A constant pH is essential for the stability of many biomole-cules. Buffers are typically used to maintain pH values in solutions;however, many buffers cease working in frozen solutions (Muraseand Franks, 1989; Sundaramurthi et al., 2010b). The reason is thatduring freezing, ice forms essentially pure crystals, resulting in aconcentration increased by several orders of magnitude in the FCS(Bartels-Rausch et al., 2014; Bogdan et al., 2014; Heger et al., 2005;Kania et al., 2014; Kasper and Friess, 2011; Krausko et al., 2015a,b,2014). This FCS will eventually also solidify, depending on thetemperature. Often, the solubility limits of the buffer are reached,causing crystallization of the less soluble component and thuspronounced acidity changes.

Another mechanism leading to changes in acidity duringfreezing of salt solutions is the difference in the distribution ofions between the FCS and the ice (Bronshteyn and Chernov, 1991).

42 L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49

One type of ion (either the cation or the anion) is preferentiallyincorporated into the ice lattice, whereas the excess of counter-ions accumulates in the FCS. The process is known as selective ionincorporation. As a result, an electrical potential called freezingpotential (FP) develops between the FCS and the ice when asolution of an ionic compound is frozen (Bronshteyn and Chernov,1991; Lefebre, 1967; Lodge et al., 1956; Murphy, 1970; Wilson andHaymet, 2008; Workman and Reynolds, 1950). This chargeimbalance is then neutralized by the diffusion of H+ or OH� fromthe liquid into the ice. If the anions of the salt are incorporated intothe ice more than the cations, the excess negative charge in the iceis neutralized by a flow of protons from the solution into the iceand the FCS becomes more basic. An example of such salt is NaCl(Bronshteyn and Chernov, 1991; Heger et al., 2006; Sola and Corti,1993; Takenaka et al., 2006; Workman and Reynolds, 1950). Incontrast, greater incorporation of the cations into the ice leads topositively charged ice, and the consequent flow of OH� from thesolution to the ice results in acidification of the FCS if salts likeNH4Cl (Heger et al., 2006), Na2SO4 (Robinson et al., 2006), or(NH4)2SO4 (Robinson et al., 2006) are used.

The desire to protect the compounds being lyophilized fromdegradation is achieved by combination of solution compositionand freeze-drying procedure. Numerous factors contribute to thesuccess of the lyophilization e.g., choice of the bulking agents,buffers and stabilizers. Previously, the change of pH duringfreezing and lyophilization was monitored and then minimizedby the proper choice of buffer, or its dilution by other compoundse.g. sugars. Despite invested effort, the problem of pH shift duringthe freezing has not been solved systematically till now andremains an empirical endeavor (Badawy and Hussain, 2007;Reinsch et al., 2015).

For the formulation processes the strongest most recom-mended method to avoid the pH jump is the right choice of buffer.It was well documented that partial crystallization of Na2H-PO4.12H2O results in sharp decrease by 3 units of pH duringfreezing of sodium phosphate (Na-P) buffer from initially neutralsolution, whereas potassium phosphate (K-P) buffer showed muchsmaller pH shift (Murase and Franks, 1989; Sundaramurthi et al.,2010a; Van den Berg, 1959). TRIS buffer (2-Amino-2-hydroxy-methyl-propane-1,3-diol) was observed to increase the pH by asmuch as 3 units at freezing (Kolhe et al., 2010a; Williams-Smithet al., 1977). Succinate buffer of initial pH 4 exhibited consequentpH increase to 8 followed be decrease to 2 at freezing(Sundaramurthi et al., 2010b). Sodium citrate and acetate, histidinehydrochloride and acetate accounts among the buffers not alteringthe pH considerably at freezing and are therefore most recom-mendable for these purposes (Izutsu et al., 2009; Jameel, 2010;Kolhe et al., 2010a). The temperature independent pH buffer wasformed as a mixture of HEPES and K-P (Sieracki et al., 2008). Thisblend retains the pH close to 7 also after the freezing and down to77 K as measured by absorption spectra changes of two pH-indicator dyes. The protection ability of the buffer was shown onoxacillin and methemoglobin. Both of these degrade at 253 K inindividual HEPES and K-P but nearly do not degrade in the blendedbuffer. For most of these buffers, the pH shift was observed also todepend on the initial solution pH and concentration of the buffer(Gomez et al., 2001).

The methods determining the apparent change of pH duringfreezing and lyophilization were reviewed (Badawy and Hussain,2007); these are the slurry low-temperature electrode method,and the techniques employing molecular probes. The strengthsand weaknesses of the methods were compared and possibleproblems of reciprocal validation were discussed. Less commonlyapplied approaches to assess the microenvironmental pH utilizesbare eye perception (Orii and Morita, 1977) or measurement offluorescence (Fu et al., 2000; Wren and Donaldson, 2012), nuclear

magnetic resonance (Robinson et al., 2006) or electron paramag-netic resonance (Williams-Smith et al., 1977) characteristics infrozen or lyophilized state.

The pH shift caused by buffer partial crystallization was oftenobserved to be decreased by the presence of sugars added ascryoprotectant or bulking agents (Lu et al., 2009; Orii and Morita,1977). The sugars are often added as bulking agents into theformulation anyhow, so they may have served as freezing pH shifteliminators tacitly. In this respect the role of bovine serumalbumin, polyvinylpyrrolidone (Anchordoquy and Carpenter, 1996)trehalose and inulin were also studied (Eriksson et al., 2003). Therole of glycerol as a dilution medium of the salts was recognizedalready by Lovelock who proposed the use of it as a “salt buffer” toprotect the red blood cells at freezing (Lovelock, 1953). Theaddition of polymers was also shown to influence the pH jumpsubstantially; in one study of citrate buffer the polyvinylpyrroli-done was found to cause pH increase whereas the dextran pHdecrease (Lu et al., 2009). The successful use (glycol)protein wasemployed on red blood cells cryopreservation (Deller et al., 2013).

Here we present a practical empirical approach to proteincryopreservation based on experimental estimation of the effec-tive pH in the frozen state and consequent minimization offreezing-induced pH changes by addition of an appropriatelychosen ionic compound to play the role of ionic cryoprotectant thatminimizes the change in effective pH via the selective ionincorporation mechanism in such a way that it opposes the aciditychange connected to the proteins and the excipients. This originalapproach leads to complete recovery of the activity of the modelenzyme, haloalkane dehalogenase DbjA, after multiple freeze-thaw cycles. The close relationship observed between the freezing-induced pH change and the activity of the frozen-thawed enzymeprovides a persuasive picture of the power and usefulness of thisnewly proposed cryopreservation strategy.

2. Methods

2.1. Model enzyme

Hydrolytic haloalkane dehalogenase DbjA (EC 3.8.1.5) fromBradyrhizobium japonicum USDA110 was used as the modelenzyme in our experiments. It catalyzes hydrolytic cleavage ofcarbon-halogen bond in halogenated aliphatic hydrocarbons. DbjAhas a broad pH optimum with maximum activity at pH 9.7, andretains more than 90% of the maximum activity in the pH rangefrom 7.7 to 10.4. DbjA preserves both secondary and tertiarystructure in the pH range 6.2–10.1. In acidic conditions, DbjA losesits tertiary structure at pH below 6.2 and it aggregates when pH isdecreased below 5.3. Under alkaline conditions, the enzyme losesmost of its tertiary structure at pH above 10.7. At the pH 10.3–11.5DbjA occurs in disordered conformation and remains soluble(Chaloupkova et al., 2011).

2.2. Enzyme preparation

His-tagged DbjA was overexpressed in Escherichia coli BL21using a previously described method (Sato et al., 2005) andpurified using the Ni-NTA Superflow Cartridge (Qiagen, Hilden,Germany). The enzyme was bound to the resin in equilibratingbuffer (20 mM potassium phosphate buffer, pH 7.5, containing0.5 M sodium chloride, and 10 mM imidazole). Unbound andweakly bound proteins were washed out with the buffercontaining 10 mM imidazole. The target enzyme was eluted by abuffer containing 500 mM imidazole. The active fractions werepooled and dialyzed against 50 mM potassium phosphate buffer(pH 7.5). The enzyme was kept at 277 K during the purificationprocedure. Enzyme concentration was determined by the Bradford

L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49 43

assay (Sigma-Aldrich, St. Louis, USA) (Bradford, 1976). Purity waschecked by SDS-PAGE. The enzyme in 50 mM sodium phosphatebuffer (pH 7.5) was prepared by dialysis from the purified enzymesolution. The enzyme was stored in 50 mM potassium or sodiumphosphate buffer at 277 K until use.

2.3. Activity assay

Dehalogenation activity of DbjA was assayed as a rate ofdehalogenation of 1,2-dibromoethane to 1-bromoethanol either byIwasaki method (Iwasaki et al., 1952) or by GC analysis. In theIwasaki method, bromides released from the substrate duringenzymatic reaction were analyzed spectrophotometrically at460 nm after the reaction with mercuric thiocyanate and ferricammonium sulfate. Sunrise microplate reader (Tecan, Grödig/Salzburg, Austria) was used for the measurements. In case ofaddition of tetramethylammonium chloride (TMACl) to thesolution of enzyme, Iwasaki method could not be used due tothe interference of chlorides. Then a loss of the substrate in timewas quantified by GC analysis (HP 6890, Hewlett-Packard). Toperform a reaction, 10 ml of substrate 1,2-dibromoethane wasinjected into 10 ml of glycine buffer (100 mM, pH = 8.6) in 25 mlReacti flask closed by Mininert valve. The reaction mixture wasincubated in a water bath at 310 K. The dehalogenation reactionwas initiated by addition of 0.2 ml of enzyme solution in theconcentration of 0.1–0.2 mg/ml. The reaction was monitored bywithdrawing 1 ml samples at 4, 8, 12, 16, and 20 min from thereaction mixture. The aliquots were immediately mixed with 35%nitric acid (for Iwasaki analysis) or with methanol (for GC analysis)to terminate the reaction. Each activity was measured in at leastthree independent replicates for both non-frozen and frozen-thawed samples. The ratio of specific activity of frozen-thawedsample to specific activity of non-frozen sample is denoted asrecovery activity.

2.4. Freeze-thaw cycle

1 ml of DbjA enzyme solution in a concentration of 0.1–0.2 mg/ml and initial pH of 7.5 was frozen in a 15 ml polypropylene tubeimmersed for 10 min into liquid nitrogen (77 K) or ethanol coolingbath with a desired temperature (typically 233 K). To achieve veryslow cooling rates, 1 ml of the enzyme solution in a vial was frozenin a cryostat (Optistat DN2, Oxford Instruments). The sample wasprecooled to 273 K and then the temperature of the sample waslowered by 0.5 K/min to 223 K. The temperature was monitored bya sensor located above the sample. In this case, ice nucleationtemperature was also recorded; it ranged between 258 and 263 K.Afterwards, the frozen solutions were thawed in the water bath(295 K) for 15 min. The freeze-thaw cycles were repeated if a morepronounce effect of enzyme degradation was desired. If not statedotherwise, data from 3 freeze-thaw cycles are presented.

2.5. UV–vis measurements

The acidities of frozen solutions were estimated by measuringUV–vis spectra of the molecular probe cresol red (CR). CR belongsto the group of sulfonephthalein indicators that are known to workat frozen conditions and in a dry state (Govindarajan et al., 2006;Heger et al., 2006). In solution it exists in three different formsdepending on pH: diprotonated (A, lmax = 518 nm), monoproto-nated (B, lmax = 434 nm), and deprotonated (C, lmax = 573 nm).pKa

1 and pKa2 of CR were reported to be 1.10 (Dean, 1992) and 8.15

(Perrin, 1963), respectively. The three forms of CR are distinguish-able also in frozen samples. At 77 K, the absorption maxima of allthree forms are red-shifted to some extent and the absorptionbands belonging to species A and C are split in two, probably due to

restricted rotation along the partial double bond between thecentral carbon and the benzene rings (Heger et al., 2006). CR in afinal concentration of 2.5 �10�5M was added to the solutions andwas used as an internal probe (Heger et al., 2006). The sampleswere frozen directly in Plastibrand cuvettes with a path length of1 cm. The UV–vis spectra were recorded immediately afterremoving the cuvettes from a cooling bath using Cary 5000 UV–vis-NIR spectrophotometer with integrating sphere DRA 2500(both Agilent, Santa Clara, USA). Although the sample temperaturewas not controlled during the recording of absorption spectra, nochanges in the spectra were observed within the time period whichwas necessary for the measurement. Each sample was measured inat least three independent replicates. Pure ice was used as areference.

2.6. Hammet acidity functions H0

Since the pH scale is used as a measure of acidity only for diluteaqueous solutions, Hammett acidity functions, H0, were used forquantification of the acidity of frozen solutions. The extent ofindicator protonation in a frozen solution with a given H0 value isthe same as that in a dilute liquid solution of the same pH. The UV–vis spectra of frozen CR solutions with initial pH of 1.1, 8.2 and 12.1,respectively, were measured to get the spectra of pure forms A, B,and C in ice. Those were fitted into the spectra of frozen buffersolutions and the relative abundances (cA, cB and cC) of forms A, B,and C were obtained. From these relative abundances the acidityfunctions H0 were calculated using equations

H0 ¼ pK1a þ log

cBcA

for the dissociation A II B

þ Hþ pK1a ¼ 1:10

� �; or

H0 ¼ pK2a þ log

cCcB

for the dissociation BAII C

þ Hþ pK2a ¼ 8:15

� �

Since the pKa values of CR frozen in ice are not known (and maydiffer from the values in liquid solution), pKa values of CR insolution were used for calculation of H0 of frozen samples.Therefore, absolute H0 values may differ from the pH values ofpartially frozen solutions measured with a low temperatureelectrode. However, H0 values can be used for comparing theacidity of frozen solutions in relative terms (Williams-Smith et al.,1977). Similar approach was used to evaluate H0 of lyophiles(Govindarajan et al., 2006).

2.7. Circular dichroism spectroscopy

Circular dichroism (CD) spectra were recorded at roomtemperature (293 K) using a Chirascan spectropolarimeter (Ap-plied Photophysics, Leatherhead, United Kingdom) equipped witha Peltier thermostat. Data were collected from 185 to 260 nm (inpure 50 mM phosphate buffer, pH 7.5) at a scan rate of 100 nm/min,1 s response time and 2 nm bandwidth using a 0.1 cm quartzcuvette containing the enzyme. Each spectrum shown is theaverage of five individual scans and was corrected for absorbancecaused by the buffer. CD data were expressed in terms of the meanresidue ellipticity (QMRE) using the Eq. (1):

QMRE ¼ Qobs:Mw:100n:c:l

ð1Þ

where Qobs is the observed ellipticity in degrees, Mw is thepolypeptide molecular weight in g/mol, n is number of residues, cis the polypeptide concentration, l is the cell path length and the

44 L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49

factor 100 originates from the conversion of the molecular weightto mg/dmol. Secondary structure content was calculated from thespectra by using of CDSSTR (Compton and Johnson, 1986;Manavalan and Johnson, 1987; Sreerama et al., 2000), Selcon3(Sreerama et al., 1999, 2000), K2D (Andrade et al., 1993) andCONTIN (Provencher and Glockner, 1981; Vanstokkum et al., 1990)methods implemented in DichroWeb server (Lobley et al., 2002;Whitmore and Wallace, 2004, 2008).

2.8. Freezing potential measurements

Freezing potential was measured with an apparatus describedin a published paper (Roubal et al., 2011). A sample was put into aglass tube (height 20 cm, internal diameter 3 mm) with twoplatinum wire electrodes placed inside the tube. One electrode wassealed through the bottom of the tube, the other was placed in anupper part of the tube. The tube was cooled from the bottom andthe voltage between the upper electrode (in contact with a liquidsample) and bottom electrode (in contact with ice) was measured.Solutions of NaCl, NH4Cl, MgCl2, TMACl in a concentration rangefrom 10�5M to 10�3M were used as the samples.

Fig. 1. Effect of freezing rate and buffer type on activity of DbjA. a, Recovered activities a233 K. Preserved a-helical content is the ratio of a-helical content in frozen-thawed sampand d. Error bars represent standard deviations of recovered activities from 6, 6, 9, an(p < 0.001) and buffer (p < 0.001) on the recovered activity. b, UV–vis spectra of cresol redliquid solutions of buffers before freezing; red and orange spectra represent frozen solutare indicated. c and d, Comparison of CD spectra of DbjA in 50 mM K-P (c) or 50 mM Na-Pthe magnitude of negative peaks at 208 and 222 nm indicates loss of a-helical conformatreferred to the web version of this article.)

3. Results

3.1. Effect of freezing rate and buffer type on DbjA activity

Samples of DbjA were exposed to freeze-thaw stress undervarious conditions to study the impact of freezing-induced aciditychanges on denaturation of this model enzyme. Three freezingmethods were used: fast freezing in liquid nitrogen (77 K) andslower freezing in an ethanol cooling bath (233 K) and very slowcontrolled freezing in cryostat (0.5 K/min). Under all conditionsexamined, the loss of activity from DbjA fast frozen at 77 K wassignificantly higher than that from samples frozen at 233 K orhigher temperatures (Fig. 1a and S1, S4). Additional cooling of thesamples frozen at 233–77 K and subsequent holding in the frozenstate had no effect on the recovery of DbjA activity (Figs. S1 and S2).There is thus compelling evidence that the enzyme degradationoccurs during the freezing process. DbjA denaturation wasirreversible � the enzyme did not regain its activity within 24 hafter thawing when incubated at 277 K (Fig. S3). Throughout theexperiments, several batches of DbjA enzyme were used. Althoughthe specific and recovered activities of DbjA were consistent within

nd preserved a-helical contents of DbjA frozen in 50 mM Na-P or K-P at 77 K and atle to that in non-frozen sample, calculated from the CD spectra presented in panels cd 6 replicates, respectively. Two-way ANOVA revealed the effects of temperature

(CR) in 50 mM Na-P and K-P, with an initial pH of 7.5. The grey spectrum representsions of Na-P and K-P, respectively, at 77 K. Peaks belonging to forms A, B, and C of CR

(d) before freezing and after 5 freeze-thaw cycles at 77 K and at 233 K. A decrease inion. (For interpretation of the references to colour in this figure legend, the reader is

L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49 45

a batch, they differed in absolute values among various batches ofthe enzyme. The reproducibility of the results reported throughoutthe paper is expressed by the standard deviations being indicatedin the Figures by error bars.

In the experiments, two types of phosphate buffer were used �sodium phosphate (Na-P) and potassium phosphate (K-P). They aresimilar in composition in the liquid state, but differ substantiallywhen frozen. Na-P acidifies extensively during freezing due toprecipitation of Na2HPO4, while there is only a minor change inacidity during freezing of K-P (Murase and Franks, 1989). Thus thispair of buffers makes it possible to evaluate only the effect of pHchange, with other factors being very similar. We observed a clearrelationship between acidity function H0 of the frozen samples andrecovery of DbjA activity (Fig. 2). DbjA frozen in 50 mM K-P buffer(H0 = 6.6; recovery (66 � 6)%) exhibits greater recovery of activitythan DbjA frozen in 50 mM Na-P (H0 = 1.4; recovery (16 � 9)%). AsDbjA in solutions loses its tertiary structure at pH values below 6.2and aggregates when the pH is decreased to below 5.3(Chaloupkova et al., 2011), an increase in the acidity of the FCSis therefore expected to denature the enzyme. Lowering the Na-Pconcentration to 1 mM resulted in both smaller acidity increasesduring freezing and much better recovery of activity after thawing(H0 = 6.1; recovery (36 � 5)%). This effect was even more pro-nounced when the enzyme was dialyzed against pure water with

Fig. 2. Effect of acidity of FCS on recovery of DbjA activity. a, Recovery of DbjA activities inand corresponding H0 values for frozen buffer solutions at 77 K. Tukey test revealed signError bars represent standard deviations of recovery activities from 9, 12, 6, and 6 replicatpanel a. The grey spectrum represents liquid solutions of buffers before freezing. Peaks beof DbjA activity after 3 freeze-thaw cycles at 77 K. Recovered activities of DbjA in 50 mM pphosphate buffers with the addition of 0.1 M TMACl, frozen-thawed under otherwise idenTMACl, 50 mM Na-P with 0.1 M TMACl, and 50 mM K-P with 0.1 M TMACl. The initial pH oflegend, the reader is referred to the web version of this article.)

an initial pH of 7.5 adjusted using NaOH (H0 = 6.8; recovery(61 �8)%) as shown in Fig. 2. All the data obtained in all ourexperiments show a relationship between the loss in enzymaticactivity after freezing and thawing and the change in acidity duringfreezing.

To investigate the relationship between inactivation andstructural changes in DbjA upon freezing, CD spectra of theenzyme in 50 mM Na-P and K-P buffers were measured before andafter multiple freeze-thaw cycles at 233 and 77 K (Fig. 2). Far-UVCD spectra of the native enzyme, measured in both buffers beforefreezing, showed one positive peak at 195 nm and two negativefeatures at 208 and 222 nm, characteristic of an a-helix (Fasman,1996). The a-helical content of the native enzyme estimated frommeasured spectra was about 25%. Inspection of CD spectrarecorded after subjecting DbjA in both buffers to multiplefreeze-thaw cycles revealed changes in intensity in comparisonwith the CD spectra recorded under native conditions. Structuralchanges in DbjA caused by multiple freeze-thaw cycles at 77 Kwere significantly more pronounced than those at 233 K. Theestimated content of a-helical structure preserved in DbjA after 5freeze-thaw cycles is displayed in Fig. 1a. The observed conforma-tional changes in the enzyme are in good agreement with theinactivation of DbjA upon multiple freeze-thaw cycles at 233 and77 K (Fig. 1a).

various buffer solutions with an initial pH of 7.5 after 3 freeze-thawing cycles at 77 K,ificant differences (p < 0.001) between all the means except 50 mM K-P and water.es, respectively. b, UV–vis spectra of CR in frozen solutions corresponding to those inlonging to forms A, B, and C of CR are indicated. c, Effect of adding TMACl on recoveryhosphate buffers are significantly lower than recovered activities of DbjA in 50 mMtical conditions (p < 0.001). d, UV–vis spectra of CR in frozen solutions (77 K) of 0.1 M

all the solutions was 7.5. (For interpretation of the references to colour in this figure

46 L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49

3.2. Acidity changes caused by freezing of salts

A series of common salts were examined for freezing-inducedacidity changes associated with selective ion incorporation(Figs. 3b). These experiments were undertaken to probe the effectof the salt alone and therefore the solutions used in thisexperiments contained only the studied salts and CR as the pHindicator. Initial pH of the solutions was 5.0 or 6.0 as indicated inFig. 3b. Since the pKa values of CR in ice are not known, calculatedH0 values may be offset from absolute pH values measured withlow temperature electrode. Therefore the resulting UV–vis spectraand H0 values of the frozen samples are compared to the spectraand H0 of the frozen CR solution at the same conditions. In general,salts with bulky anions, e.g. sulfates, increased the acidity of thefrozen solution, while halides increased the basicity of the frozensolution (Fig. 3b). This correlates with the results of FP measure-ments, where relatively small anions (e.g. halides) were usuallyincorporated better into the ice lattice than almost any cation,while relatively large anions (nitrate, acetate) were not usuallyincorporated into the ice (Workman and Reynolds, 1950).Ammonium salts do not follow this trend, since NH3 isincorporated into the ice to a large extent, more so than, forexample, Cl�, thus making the FCS acidic (Workman and Reynolds,1950). A very high level of basification was observed when TMAClwas frozen. This is not surprising, since the bulky tetramethy-lammonium cation would be expelled from the ice more than thechloride anion. The extent of this basification was dependent onthe concentration of TMACl, with a higher concentration of TMAClresulting in a more basic frozen solution: solutions of 0.001 M,0.01 M and 0.1 M TMACl, all with an initial pH of 5.0, resulted in H0

Fig. 3. Selective ion incorporation. a, Scheme showing selective ion incorporation into icsolution, middle—onset of freezing, bottom—neutralization of charge imbalance. Left-hanincorporated into the ice leaving an excess of negatively charged chloride anions in the freSubsequent neutralization of the charge imbalance results in acidification of the FCS. Rigwhere chloride anions are incorporated into the ice to a greater extent than bulky tetrafrozen solutions. Signs (+ or �) of the FP were taken from the literature (Workman and Reice. A negative FP is associated with acidification of the FCS, while a positive FP is associatefrozen solutions (77 K) of 0.01 M salts with an initial pH of 5.0 or 6.0.

values of 7.5, 7.9 and 8.1, respectively, when the solution was frozen(Fig. S5).

3.3. Neutralization of freezing-induced acidity changes by addition ofionic cryoprotectant

We have proposed and investigated a new method ofneutralizing buffer-induced acidity changes by the addition of acarefully chosen salt in the role of ionic cryoprotectant; this doesnot change pH of a non-frozen solution, but during freezing itcauses an acidity change opposite to that resulting from the buffercrystallization, so that the overall change in acidity is negligible. Tocounterbalance buffer-induced acidification by the mechanism ofcrystallization of individual buffer component, an ionic cryopro-tectant that causes the basification of the FCS must be applied. Forthis purpose, TMACl at various concentrations was added to the50 mM Na-P solution and the acidity of the frozen mixed solutionwas monitored. The addition of TMACl even at low concentrationsresulted in reduced acidification during freezing in comparisonwith that undergone by the buffer solution alone. Almost completeinhibition of freezing-induced acidification was observed at aTMACl concentration of 0.1 M (Fig. S6). This concentration ofTMACl had no negative effect on the specific activity of non-frozenenzyme (Fig. S7).

The effect of adding TMACl on the recovery of DbjA activity afterfreezing and thawing was positive over the whole range of testedconcentrations (from 0.01 M to 1 M). Recovered activities of DbjAafter freezing and thawing in the presence of TMACl were higherthan those in the buffer alone. When slow freezing (0.5 K/min) wasapplied, the recovered activity of DbjA enzyme in 50 mM Na-P

e and consequent charge neutralization leading to acidity change. Top—non-frozend column: During freezing of NH4Cl solution, ammonium cations are preferentiallyeze-concentrated solution (FCS). A negative freezing potential (FP) can be measured.ht-hand column: The opposite situation applies during freezing of TMACl solution,methylammonium cations. b, Effects of various salts on the FP and acidities (H0) ofynolds,1950) or measured. FPs are expressed as potentials of the FCS with respect tod with basification of the FCS. H0 values were calculated from UV–vis spectra of CR in

L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49 47

buffer without addition of TMACl was (43 � 16)%. If 0.1 M TMAClwas added to the solution to counterbalance the pH shift onfreezing, the recovered activity of DbjA was increased to(112 � 27)%. To demonstrate the protective effect of the ioniccryoprotectant more clearly, fast freezing at 77 K was used, as thismethod of freezing caused more damage to the unprotectedenzyme than slower rates of freezing. The effect of adding 0.1 MTMACl to 50 mM phosphate buffers on the recovered activity ofDbjA after 3 freeze-thaw cycles is shown in Fig. 2c. The recoveredactivity of DbjA in 50 mM Na-P buffer alone was (16 � 9)%. Theaddition of 0.1 M TMACl prevented the loss of activity of DbjA inphosphate buffers during freezing and thawing; DbjA activity wascompletely recovered to (105 � 24)%. Similarly, the recoveredactivity of DbjA in 50 mM K-P increased from (66 � 6)% to(106 � 14)% when 0.1 M TMACl was added to the solution priorto freezing.

4. Discussion

We demonstrate a strong link between losses in the activity ofthe model enzyme DbjA and an increase in the acidity of FCS undervarious experimental conditions. The enzymatic activity loss,which was irreversible, was accompanied by changes in thestructure of the enzyme, as confirmed by CD spectroscopy. Whenthe enzyme was frozen in Na-P buffer, which acidifies more than K-P during freezing, there was a greater loss of the activity incomparison with the enzyme frozen in K-P buffer under otherwiseidentical conditions. Lowering of the Na-P concentration resultedin less change in acidity during freezing and also led to greaterrecovery of enzyme activity. These findings support those ofprevious research in which minimization of protein losses duringfreezing was found to be linked to minimizing acidity changes inthe FCS by making an optimal choice of buffer salts and using themat a low concentration (Bhatnagar et al., 2007; Jiang and Nail,1998;Kasper and Friess, 2011; Sieracki et al., 2008). Lower freezing ratesresulted in lower losses of DbjA activity in comparison with higherfreezing rates when the enzyme was frozen in K-P, in Na-P and alsoin water without any buffer. This is in agreement with previousstudies in which lower losses of enzyme activities were observedduring slow freezing (Bhatnagar et al., 2008; Cao et al., 2003;Chang et al., 1996; Chen and Cui, 2006; Jiang and Nail, 1998;Strambini and Gabellieri, 1996). However, there are other studiesthat show the opposite trend (Pikal-Cleland et al., 2000; Wei et al.,2012). We believe that this contradiction is caused by two mutuallycounteracting phenomena that are expected to occur during slowfreezing: a) formation of larger ice crystals, hence a smaller icesurface area, and b) more extensive precipitation of the less solublebuffer component, hence a greater acidity change (Williams-Smithet al., 1977). Smaller ice surface area results in an increase enzymerecovery after freezing and thawing since the contact of proteinwith the ice surface is thought to be responsible for surface-induced denaturation (Jameel, 2010), while acidity changes causeprotein degradation.

Here we validated a new approach to minimize the aciditychange in the FCS by addition of the ionic cryoprotectant thataffects the acidity of the frozen sample in a manner opposite to thatof the protein and excipients. The increase in acidity related tosequential crystallization of Na-P was canceled out by the additionof TMACl, which turns the FCS more basic during freezing due toselective ion incorporation. Addition of 0.1 M TMACl to 50 mM Na-P led to complete recovery of DbjA activity after 3 freeze-thawcycles when both very slow freezing (0.5 K/min) and very fastfreezing (77 K) were applied. Since all the other conditions (the rateof freezing, concentration of the enzyme and the buffer etc.) arepreserved for freezing of the samples with and without addition ofTMACl to the solutions, the excessive degradation of the enzyme

frozen in the buffer alone can be assigned mostly to the pH shift. Allthe other effects (ice crystal formation, self-buffering capacity ofprotein etc.) are not supposed to be greatly affected by thepresence of TMACl and therefore their impact on the extent ofdegradation of the enzyme during freezing should be approxi-mately the same in the samples with and without the addition ofTMACl.

TMACl was proved to be a useful ionic cryoprotectant for DbjAenzyme during freezing in phosphate buffers due to increase of H0

of the frozen solution and eliminator of buffer-induced acidifica-tion. Similar effect can be expected when other salts will be used.To prevent unwanted acidification, salts with the anion incorpo-rating more into the ice than the cation (generating positive FP)should be used as the ionic cryprotectants. On the other hand, saltswith the cation incorporating more into the ice than the anion(generating negative FP) should be used as the ionic cryprotectantsto prevent unwanted basification. Examples of such salts are givenin Fig. 3. It should be emphasized, that any salts present in thesolution to be frozen, for example those used to adjust ionicstrength, may substantially change the resulting acidity of frozensamples. Since pH change related to buffer crystallization, and alsoto selective ion incorporation, is dependent on the rate of freezing,the initial pH, type and concentration of the buffer/salt used, theamount of the ionic cryoprotectant which is necessary forneutralization of the buffer-induced pH change is dependent onthe experimental conditions during the process of freezing. Theresulting acidity of the frozen solution should be monitored andcompared to the pH optimum of the protein, because the additionof the ionic cryoprotectant to the buffer solution may affect thesolubility of the individual buffer substances during freezing(Larsen, 1973) and cause undesirable effects.

Previously, the freezing-induced pH change, if recognized as aproblematic factor, was empirically minimized either by the choiceof different buffer or by its dilution by the bulking agent (Badawyand Hussain, 2007; Lu et al., 2009). On the other hand, our newlyproposed method of applying ionic cryoprotectant is based on theunderstanding of physicochemical phenomenon of selective ionincorporation. Many studies of microenvironmental pH conclude,that the freezing or lyophilization optimization deserve a priori pHtest since there are little rules available to predict the pH behaviorin the microenvironment of frozen active pharmaceutical com-pound (Badawy and Hussain, 2007). The method of applying ioniccryoprotectants may give to the researches one of the tools that canhelp in the design of successful freezing protocols.

Conjunction of already established methods and our newlyproposed ionic cryoprotection method may bring some practicalrecommendations and new possibilities for the formulationprotocols. Among the common features of all cryoprotectionmethods is the choice of the buffers known to exhibit a small pHshift upon freezing and applying the buffer in the smallest possibleconcentration that suffice the solution buffering capacity sincehigh buffer concentrations at freezing can become problematic.The distinct element of proposed ionic cryoprotection methodfrom previous methods is the possibility to neutralize theremaining pH shifts by justified choice of ionic compounds. Thechoice can be made in a way that the FCS would have propertiesdesirable for future workup of the frozen or lyophilized formula-tion (e.g., keeping high glass transition temperature). The use ofionic cryoprotectants seems to be practical for most of the freezingcycles as it proved operational for extremely fast (immersion intothe liquid nitrogen) as well as very slow (0.5 K/min) freezingmethods. It should be stressed out, that our proposed approachneeds to be tested and verified for each particular case of cryo(lyo)-protection and its possible industrial application. The cooperativeeffect of the ionic cryoprotectant and the bulking agent (e.g.saccharides) is a matter of our further research; we expect further

48 L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49

protein stabilization in comparison to saccharides alone (Lu et al.,2009). Also the effect of added salts to glass transition temperatureshould be considered.

In this work, we intentionally applied the harshest freezingconditions (fast freezing and extensively crystallizing buffer) todemonstrate that even these conditions can be rendered benign byan appropriate choice of the ionic cryoprotectant. Thus theavailability of fast freezing methods, which previously causedextensive protein degradation, may now become advantageous,because under these conditions protein aggregation can beminimized and dissolution of a protein in its active form facilitated.We believe that a properly designed approach to freezing usingionic cryprotectants will pave the way to more efficient storage ofproteins both in research laboratories and in industry.

Conflict of interest

Authors declare they have no conflict of interest.

Author’s contributions

D.H. devised the idea for the project and directed the research. J.D. and R.C. provided the enzyme, methodology and facilities foractivity measurements. �L.K. conducted and coordinated theexperimental work, analyzed the data and wrote the initial draftof the manuscript. J.P., L. F. and M.K. conducted parts of theexperimental work and analyzed the data. R.C. and M.K. measuredand R.C. analyzed the CD spectra. S.M. evaluated the datastatistically. All authors contributed to discussions and writingof the final paper.

Acknowledgements

The research was supported by Czech Science Foundation(GA15-12386S) and the RECETOX research infrastructure sup-ported by the projects of the Czech Ministry of Education (LO1214)and (LM2011028).

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.ijpharm.2016.05.031.

References

Abbatt, J.P.D., Thomas, J.L., Abrahamsson, K., Boxe, C., Granfors, A., Jones, A.E., King,M.D., Saiz-Lopez, A., Shepson, P.B., Sodeau, J., Toohey, D.W., Toubin, C., vonGlasow, R., Wren, S.N., Yang, X., 2012. Halogen activation via interactions withenvironmental ice and snow in the polar lower troposphere and other regions.Atmos. Chem. Phys. 12, 6237–6271.

Anchordoquy, T.J., Carpenter, J.F., 1996. Polymers protect lactate dehydrogenaseduring freeze-drying by inhibiting dissociation in the frozen state. Arch.Biochem. Biophys. 332, 231–238.

Andrade, M.A., Chacon, P., Merelo, J.J., Moran, F., 1993. Evaluation of secondarystructure of proteins from UV circular-dichroism spectra using an unsupervisedlearning neural-network. Protein Eng. 6, 383–390.

Asghar, W., El Assal, R., Shafiee, H., Anchan, R.M., Demirci, U., 2014. Preservinghuman cells for regenerative, reproductive, and transfusion medicine.Biotechnol. J. 9, 895–903.

Attwater, J., Wochner, A., Holliger, P., 2013. In-ice evolution of RNA polymeraseribozyme activity. Nat. Chem. 5, 1011–1018.

Badawy, S.I.F., Hussain, M.A., 2007. Microenvironmental pH modulation in soliddosage forms. J. Pharm. Sci. 96, 948–959.

Bartels-Rausch, T., Jacobi, H.W., Kahan, T.F., Thomas, J.L., Thomson, E.S., Abbatt, J.P.D.,Ammann, M., Blackford, J.R., Bluhm, H., Boxe, C., Domine, F., Frey, M.M., Gladich,I., Guzmán, M.I., Heger, D., Huthwelker, T., Klán, P., Kuhs, W.F., Kuo, M.H., Maus,S., Moussa, S.G., McNeill, V.F., Newberg, J.T., Pettersson, J.B.C., Roeselová, M.,Sodeau, J.R., 2014. A review of air–ice chemical and physical interactions (AICI):liquids quasi-liquids, and solids in snow. Atmos. Chem. Phys. 14, 1587–1633.

Bhatnagar, B.S., Bogner, R.H., Pikal, M.J., 2007. Protein stability during freezing:separation of stresses and mechanisms of protein stabilization. Pharm. Dev.Technol. 12, 505–523.

Bhatnagar, B.S., Pikal, M.J., Bogner, R.H., 2008. Study of the individual contributionsof ice formation and freeze-concentration on isothermal stability of lactatedehydrogenase during freezing. J. Pharm. Sci. 97, 798–814.

Bogdan, A., Molina, M.J., Tenhu, H., Bertel, E., Bogdan, N., Loerting, T., 2014.Visualization of freezing process in situ upon cooling and warming of aqueoussolutions. Sci. Rep. 4, 7414.

Bradford, M.M., 1976. Rapid and sensitive method for quantitation of microgramquantities of protein utilizing principle of protein-dye binding. Anal. Biochem.72, 248–254.

Bronshteyn, V.L., Chernov, A.A., 1991. Freezing potentials arising on solidification ofdilute aqueous-solutions of electrolytes. J. Cryst. Growth 112, 129–145.

Cao, E.H., Chen, Y.H., Cui, Z.F., Foster, P.R., 2003. Effect of freezing and thawing rateson denaturation of proteins in aqueous solutions. Biotechnol. Bioeng. 82, 684–690.

Chaloupkova, R., Prokop, Z., Sato, Y., Nagata, Y., Damborsky, J., 2011. Stereoselectivityand conformational stability of haloalkane dehalogenase DbjA fromBradyrhizobium japonicum USDA110: the effect of pH and temperature. FEBS J.278, 2728–2738.

Chang, B.S., Kendrick, B.S., Carpenter, J.F., 1996. Surface-induced denaturation ofproteins during freezing and its inhibition by surfactants. J. Pharm. Sci. 85,1325–1330.

Chen, Y.H., Cui, Z., 2006. Effect of salts on the freezing denaturation of lactatedehydrogenase. Food Bioprod. Process. 84, 44–50.

Cicerone, M.T., Pikal, M.J., Qian, K.K., 2015. Stabilization of proteins in solid form.Adv. Drug Deliv. Rev..

Compton, L.A., Johnson, W.C., 1986. Analysis of protein circular-dichroism spectrafor secondary structure using a simple matrix multiplication. Anal. Biochem.155, 155–167.

Dean, J.A., 1992. Lange’s Handbook of Chemistry, 14th ed. McGraw Hill, New York.Deller, R.C., Congdon, T., Sahid, M.A., Morgan, M., Vatish, M., Mitchell, D.A., Notman,

R., Gibson, M.I., 2013. Ice recrystallisation inhibition by polyols: comparison ofmolecular and macromolecular inhibitors and role of hydrophobic units.Biomater. Sci. 1, 478–485.

Eriksson, J.H.C., Hinrichs, W.L.J., de Jong, G.J., Somsen, G.W., Frijlink, H.W., 2003.Investigations into the stabilization of drugs by sugar glasses: III. The influenceof various high-pH buffers. Pharm. Res. 20, 1437–1443.

Fasman, G.D., 1996. Circular Dichroism and the Conformational Analysis ofBiomolecules. Plenum Press, New York, NY.

Fu, K., Pack, D.W., Klibanov, A.M., Langer, R., 2000. Visual evidence of acidicenvironment within degrading poly(lactic-co-glycolic acid) (PLGA)microspheres. Pharm. Res. 17, 100–106.

Gomez, G., Pikal, M.J., Rodriguez-Hornedo, N., 2001. Effect of initial buffercomposition on pH changes during far-from-equilibrium freezing of sodiumphosphate buffer solutions. Pharm. Res. 18, 90–97.

Govindarajan, R., Chatterjee, K., Gatlin, L., Suryanarayanan, R., Shalaev, E.Y., 2006.Impact of freeze-drying on ionization of sulfonephthalein probe molecules intrehalose-citrate systems. J. Pharm. Sci. 95, 1498–1510.

Heger, D., Jirkovsky, J., Klan, P., 2005. Aggregation of methylene blue in frozenaqueous solutions studied by absorption spectroscopy. J. Phys. Chem. A 109,6702–6709.

Heger, D., Klanova, J., Klan, P., 2006. Enhanced protonation of cresol red in acidicaqueous solutions caused by freezing. J. Phys. Chem. B 110, 1277–1287.

Iwasaki, I., Utsumi, S., Ozawa, T., 1952. New colorimetric determination of chlorideusing mercuric thiocyanate and ferric ion. Bull. Chem. Soci. Jpn. 25 226–226.

Izutsu, K.-i., Kadoya, S., Yomota, C., Kawanishi, T., Yonemachi, E., Terada, K., 2009.Stabilization of protein structure in freeze-dried amorphous organic acid buffersalts. Chem. Pharm. Bull. 57, 1231–1236.

Jameel, F., 2010. Formulation and Process Development Strategies forManufacturing Biopharmaceuticals. In: Hershenson, S. (Ed.), John Wiley & Sons.

Jiang, S., Nail, S.L., 1998. Effect of process conditions on recovery of protein activityafter freezing and freeze-drying. Eur. J. Pharm. Biopharm. 45, 249–257.

Kania, R., Malongwe, J.K.E., Nachtigallová, D., Krausko, J., Gladich, I., Roeselová, M.,Heger, D., Klán, P., 2014. Spectroscopic properties of benzene at the air–iceinterface: a combined experimental–computational approach. J. Phys. Chem. A118, 7535–7547.

Kasper, J.C., Friess, W., 2011. The freezing step in lyophilization: physico-chemicalfundamentals, freezing methods and consequences on process performanceand quality attributes of biopharmaceuticals. Eur. J. Pharm. Biopharm. 78, 248–263.

Kasper, J.C., Winter, G., Friess, W., 2013. Recent advances and further challenges inlyophilization. Eur. J. Pharm. Biopharm. 85, 162–169.

Kolhe, P., Amend, E., Singh, S.K., 2010a. Impact of freezing on ph of buffered solutionsand consequences for monoclonal antibody aggregation. Biotechnol. Prog. 26,727–733.

Kolhe, P., Holding, E., Lary, A., Chico, S., Singh, S.K., 2010b. Large-scale freezing ofbiologics: understanding protein and solute concentration changes in acryovessel-part 2. Biopharm. Int. 23, 40–49.

Krausko, J., Runštuk, J., Ned�ela, V., Klán, P., Heger, D., 2014. Observation of a brinelayer on an ice surface with an environmental scanning electron microscope athigher pressures and temperatures. Langmuir 30, 5441–5447.

Krausko, J., Malongwe, J.K.E., Bi9canová, G., Klán, P., Nachtigallová, D., Heger, D.,2015a. Spectroscopic properties of naphthalene on the surface of ice grains

L. Krausková et al. / International Journal of Pharmaceutics 509 (2016) 41–49 49

revisited: a combined experimental–computational approach. J. Phys. Chem. A119, 8565–8578.

Krausko, J., Ondrušková, G., Heger, D., 2015b. Comment on photolysis of polycyclicaromatic hydrocarbons on water and ice surfaces and on nonchromophoricorganic matter suppresses polycyclic aromatic hydrocarbon photolysis in iceand at ice surfaces. J. Phys. Chem. A 119, 10761–10763.

Larsen, S.S., 1973. Studies on stability of drugs in frozen systems. Arch. Pharm.Chemi. Sci. Ed. 1, 41–53.

Lefebre, V., 1967. Freezing potential effect. J. Colloid Interface Sci. 25 (263-&).Lobley, A., Whitmore, L., Wallace, B.A., 2002. DICHROWEB: an interactive website

for the analysis of protein secondary structure from circular dichroism spectra.Bioinformatics 18, 211–212.

Lodge, J.P., Baker, M.L., Pierrard, J.M., 1956. Observations on ion separation in dilutesolutions by freezing. J. Chem. Phys. 24, 716–719.

Lovelock, J.E., 1953. Het mechanism of the protective action of glycerol againsthaemolysis by freezing and thawing. Biochim. Biophys. Acta 11, 28–36.

Lu, E.X., Ewing, S., Gatlin, L., Suryanarayanan, R., Shalaev, E., 2009. The effect ofbulking agents on the chemical stability of acid-sensitive compounds in freeze-dried formulations: sucrose inversion study. J. Pharm. Sci. 98, 3387–3396.

Manavalan, P., Johnson, W.C., 1987. Variable selection method improves theprediction of protein secondary structure from circular-dichroism spectra. Anal.Biochem. 167, 76–85.

Murase, N., Franks, F., 1989. Salt precipitation during the freeze-concentration ofphosphate buffer solutions. Biophys. Chem. 34, 293–300.

Murphy, E.J., 1970. Generation of electromotive forces during freezing of water. J.Colloid Interface Sci. 32, 1–11.

Mutschler, H., Wochner, A., Holliger, P., 2015. Freeze–thaw cycles as drivers ofcomplex ribozyme assembly. Nat. Chem. 7, 502–508.

Orii, Y., Morita, M., 1977. Measurement of pH of frozen buffer solutions by using pHindicators. J. Biochem. 81, 163–168.

Perrin, D.D., 1963. Buffers of low ionic strength for spectrophotometric pKdeterminations. Aust. J. Chem. 16, 572.

Pikal-Cleland, K.A., Rodriguez-Hornedo, N., Amidon, G.L., Carpenter, J.F., 2000.Protein denaturation during freezing and thawing in phosphate buffer systems:monomeric and tetrameric beta-galactosidase. Arch. Biochem. Biophys. 384,398–406.

Pratt, K.A., Custard, K.D., Shepson, P.B., Douglas, T.A., Pohler, D., General, S., Zielcke, J.,Simpson, W.R., Platt, U., Tanner, D.J., Gregory Huey, L., Carlsen, M., Stirm, B.H.,2013. Photochemical production of molecular bromine in Arctic surfacesnowpacks. Nat. Geosci. 6, 351–356.

Provencher, S.W., Glockner, J., 1981. Estimation of globular protein secondarystructure from circular-dichroism. Biochemistry 20, 33–37.

Reinsch, H., Spadiut, O., Heidingsfelder, J., Herwig, C., 2015. Examining the freezingprocess of an intermediate bulk containing an industrially relevant protein.Enzyme Microb. Technol. 71, 13–19.

Robinson, C., Boxe, C.S., Guzman, M.I., Colussi, A.J., Hoffmann, M.R., 2006. Acidity offrozen electrolyte solutions. J. Phys. Chem. B 110, 7613–7616.

Roubal, Z., Szabó, Z., Steinbauer, M., Heger, D., Kubásek, R., 2011. The design of high-impedance and high-voltage input amplifier for measurement ofelectropotentials on solid-liquid phase boundary. Progress in ElectromagneticsResearch Symposium 1162–1166.

Sato, Y., Monincova, M., Chaloupkova, R., Prokop, Z., Ohtsubo, Y., Minamisawa, K.,Tsuda, M., Damborsky, J., Nagata, Y., 2005. Two rhizobial strains, Mesorhizobiumloti MAFF303099 and Bradyrhizobium japonicum, USDA110, encode haloalkanedehalogenases with novel structures and substrate specificities. Appl. Environ.Microbiol. 71, 4372–4379.

Sieracki, N.A., Hwang, H.J., Lee, M.K., Garner, D.K., Lu, Y., 2008. A temperatureindependent pH (TIP) buffer for biomedical biophysical applications at lowtemperatures. Chem. Commun. 823–825.

Sola, M.I., Corti, H.R., 1993. Freezing induced electric potentials and ph changes inaqueous-solutions of electrolytes. An. Asoc. Quim. Argent. 81, 483–498.

Soltanizadeh, N., Mirmoghtadaie, L., Nejati, F., Najafabadi, L.I., Heshmati, M.K., Jafari,M., 2014. Solid-state protein-carbohydrate interactions and their application inthe food industry. Compr. Rev. Food Sci. Food Saf. 13, 860–870.

Sreerama, N., Venyaminov, S.Y., Woody, R.W., 1999. Estimation of the number ofalpha-helical and beta-strand segments in proteins using circular dichroismspectroscopy. Protein Sci. 8, 370–380.

Sreerama, N., Venyaminov, S.Y., Woody, R.W., 2000. Estimation of protein secondarystructure from circular dichroism spectra: inclusion of denatured proteins withnative proteins in the analysis. Anal. Biochem. 287, 243–251.

Strambini, G.B., Gabellieri, E., 1996. Proteins in frozen solutions: evidence of ice-induced partial unfolding. Biophys. J. 70, 971–976.

Sundaramurthi, P., Shalaev, E., Suryanarayanan, R., 2010a. Calorimetric anddiffractometric evidence for the sequential crystallization of buffer componentsand the consequential ph swing in frozen solutions. J. Phys. Chem. B 114, 4915–4923.

Sundaramurthi, P., Shalaev, E., Suryanarayanan, R., 2010b. pH swing in frozensolutions-consequence of sequential crystallization of buffer components. J.Phys. Chem. Lett. 1, 265–268.

Takenaka, N., Tanaka, M., Okitsu, K., Bandow, H., 2006. Rise in the pH of an unfrozensolution in ice due to the presence of NaCl and promotion of decomposition ofgallic acids owing to a change in the pH. J. Phys. Chem. A 110, 10628–10632.

Vajda, T., 1999. Cryo-bioorganic chemistry: molecular interactions at lowtemperature. Cell. Mol. Life Sci. 56, 398–414.

Van den Berg, L., 1959. The effect of addition of sodium and potassium chloride tothe reciprocal system � KH2PO4-Na2HPO4-H2O on pH and composition duringfreezing. Archi. Biochem. Biophys. 84, 305–315.

Vanstokkum, I.H.M., Spoelder, H.J.W., Bloemendal, M., Vangrondelle, R., Groen, F.C.A., 1990. Estimation of protein secondary structure and error analysis fromcircular-dichroism spectra. Anal. Biochem. 191, 110–118.

Wei, Y.Y., Huang, C.W., Chou, W.Y., Lee, H.J., 2012. alpha-Crystallin protects humanarginosuccinate lyase activity under freeze-thaw conditions. Biochimie 94,566–573.

Whitmore, L., Wallace, B.A., 2004. DICHROWEB, an online server for proteinsecondary structure analyses from circular dichroism spectroscopic data.Nucleic Acids Res. 32, W668–W673.

Whitmore, L., Wallace, B.A., 2008. Protein secondary structure analyses fromcircular dichroism spectroscopy: methods and reference databases.Biopolymers 89, 392–400.

Williams-Smith, D.L., Bray, R.C., Barber, M.J., Tsopanakis, A.D., Vincent, S.P., 1977.Changes in apparent pH on freezing aqueous buffer solutions and theirrelevance to biochemical electron-paramagnetic-resonance spectroscopy.Biochem. J. 167, 593–600.

Wilson, P.W., Haymet, A.D.J., 2008. Workman-Reynolds freezing potentialmeasurements between ice and dilute salt solutions for single ice crystal faces. J.Phys. Chem. B 112, 11750–11755.

Workman, E.J., Reynolds, S.E., 1950. Electrical phenomena occurring during thefreezing of dilute aqueous solutions and their possible relationship tothunderstorm electricity. Phys. Rev. 78, 254.

Wren, S.N., Donaldson, D.J., 2012. Laboratory study of pH at the air–ice interface. J.Phys. Chem. C 116, 10171–10180.