- Email: [email protected]
Calorimetric and Fourier transform infrared spectroscopic study of solid proteins immersed in low water organic solvents
Biochimica et Biophysica Acta 1547 (2001) 359^369 www.bba-direct.com
Calorimetric and Fourier transform infrared spectroscopic study of solid proteins immersed in low water organic solvents Vladimir A. Sirotkin a; *, Albert N. Zinatullin a , Boris N. Solomonov a , Djihanguir A. Faizullin b , Vladimir D. Fedotov b a
Department of Chemistry, Kazan State University, Kremlevskaya str. 18, Kazan 420008, Russia b Kazan Institute of Biochemistry and Biophysics, Kazan 420503, Russia Received 24 May 2000; received in revised form 23 March 2001; accepted 4 April 2001
Abstract Calorimetric heat effects and structural rearrangements assessed by means of Fourier transform infrared (FTIR) amide I spectra were followed by immersing dry human serum albumin and bovine pancreatic K-chymotrypsin in low water organic solvents and in pure water at 298 K. Enthalpy changes upon immersion of the proteins in different media are in a good linear correlation with the corresponding IR absorbance changes. Based on calorimetric and FTIR data the solvents were divided into two groups. The first group includes carbon tetrachloride, benzene, nitromethane, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine where no significant heat evolution and structural changes were found during protein immersion. Due to kinetic reasons no significant protein^solvent interactions are expected in such systems. The second group of solvents includes dimethyl sulfoxide, methanol, ethanol, and water. Immersion of proteins in these media results in protein swelling and involves significant exothermic heat evolution and structural changes in the protein. Dividing of different media in the two groups is in a qualitative correlation with the solvent hydrophilicity defined as partial excess molar Gibbs free energy of water at infinite dilution in a given solvent. The first group includes the solvents with hydrophilicity exceeding 2.7 kJ/mol. More hydrophilic second group solvents have this energy values less than 2.3 kJ/mol. The hydrogen bond donating ability of the solvents also assists in protein swelling. Hydrogen bonding between protein and solvent is assumed to be a main factor controlling the swelling of dry solid proteins in the studied solvents. ß 2001 Elsevier Science B.V. All rights reserved. Keywords: Fourier transform infrared; Immersion calorimetry; Organic solvent; Solvent hydrophilicity, swelling; Human serum albumin; Bovine pancreatic K-chymotrypsin
1. Introduction Enzymes suspended in organic solvents with low water contents may catalyze reactions not feasible in aqueous medium [1,2], and demonstrate greatly enAbbreviations: FTIR, Fourier transform infrared; HSA, human serum albumin; CT, bovine pancreatic K-chymotrypsin * Corresponding author. Fax: +7-8432-380-994; E-mail: [email protected]
hanced thermostability [3,4] and `molecular memory' [5,6]. This biotechnological potential of enzymes in organic medium depends strongly on the nature of the organic solvent. For example, solvents may activate or suppress enzymatic activity [7,8], a¡ect the enantioselectivity of the suspended enzymes [9,10], result in exothermic peaks on di¡erential scanning calorimetry (DSC) curves of the protein suspended in organic solvents [11,12], and in£uence ligand binding to the imprinted protein [5].
0167-4838 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 0 1 ) 0 0 2 0 1 - 1
BBAPRO 36434 1-6-01
360
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Therefore, information on solid protein^organic solvent interactions is of importance in explaining various protein activities. Due to the ability to monitor the thermodynamic and structural changes in such heterogeneous systems, the combination of calorimetry and infrared (IR) spectroscopy has a great potential in examining factors governing solid protein^solvent interactions. Earlier we proposed a calorimetric approach for the examination of processes that occur on immersing the solid proteins in water^organic mixtures [13^ 15]. This approach involved the measurement of the enthalpies corresponding to the formation of the `protein+liquid' heterogeneous systems. It was suggested [14,15] that the enthalpy of the formation of human serum albumin (HSA) suspensions in water^ organic mixtures is controlled mainly by two processes. The ¢rst is water desorption/sorption which may follow a Langmuir type trend. The thermodynamics of water sorption by the protein in organic solvents was evaluated using the Langmuir model. The solvent e¡ect on thermodynamic parameters of water sorption by the protein was found to be in a strong correlation with the water solvation thermodynamics in organic medium [14]. The second process attributed to protein swelling in a water^organic mixture [14,16] is associated with signi¢cant heat e¡ects and an increase in the surface area accessible for sorption of water. However, the role of protein conformational changes in calorimetric events observed on suspending the protein in organic medium was not justi¢ed. Fourier transform infrared (FTIR) spectroscopy has been successfully applied to the examination of the secondary structure of proteins in organic solvents [17,18]. It was shown that some organic solvents induce conformational changes in the suspended protein. However, the factors governing the protein structure in low water organic medium are not really clear. In the present study we want to combine isothermal calorimetry and IR spectroscopy in the examination of protein^solvent interactions. The objective of this combined calorimetric and IR study is to elucidate the relations between the e¡ect of solvent nature on protein secondary structure and heat e¡ects of interaction of solid protein with organic liquids.
Bovine pancreatic K-chymotrypsin (CT) and human serum albumin were used as model proteins. As is known, water sorbed by proteins in organic solvents may signi¢cantly in£uence the catalytic properties [9,19], protein conformation [17] and thermostability [3,4,12], and enthalpies of suspension formation [13^15]. Hence, in order to focus on protein^ organic solvent interactions, dry proteins (dehydrated at a water activity less than 0.01) were investigated when immersed in essentially low water solvents. The behavior of dry proteins in nonaqueous solvents was also compared with that in a water environment. 2. Materials and methods 2.1. Materials Human serum albumin (Sigma, product No. A 1887, essentially fatty acid free) and bovine pancreatic K-chymotrypsin (Sigma, C 4129, essentially salt free; EC 3.4.21.1; speci¢c activity 52 units/mg of solid) were used without further puri¢cation. Organic solvents (reagent grade, purity s 99%) were puri¢ed and dried according to the recommendations [20] and then stored over dry 3A molecular sieves for at least 24 h prior to use. Water used was doubly distilled. 2.2. FTIR measurements FTIR spectrometry was carried out on a Vector 22 (Bruker) FTIR spectrophotometer at 4 cm31 resolution. Vibration spectra were obtained with glassy-like protein ¢lms casted from a 2% (w/v) water solution onto a CaF2 window at room humidity. After mounting the windows in the sample cell, the ¢lm was dehydrated by £ushing air dried over P2 O5 powder. The relative vapor pressure over P2 O5 at 298 K does not exceed 0.01 [21]. The protein ¢lm was £ushed until no further spectral changes were detected in the 3450 cm31 water absorbance region and the amide A contour on this side represented a smooth line without any visible shoulders. The cell was then ¢lled with the chosen organic solvent. Spectra were recorded as a function of time until equilib-
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
361
rium was achieved. Then, spectra of pure solvent were recorded and subtracted from the sample spectra. Protein spectra in a water environment were obtained on protein ¢lms wetted by £ushing air at 99% relative humidity. The spectrum of liquid water was then subtracted from the spectra of wet ¢lms in accord with the criteria described in [22]. Changes of the secondary protein structure were analyzed using the known correlation between secondary structure elements in proteins and peak positions in the amide I spectra [23,24].
the proteins, transformed the samples to a transparent dense gel. It is known that some nonaqueous solvents, for example DMSO and formamide, can dissolve proteins [26,27]. However, the mechanism of the protein dissolving potential of anhydrous organic solvents is still not clear [27]. Protein solubility in organic solvents may depend strongly on a variety of factors such as pH, temperature, solvent humidity, and electrolyte additives [26^29]. Probably, deep dehydration of the proteins and absence of salts may strongly decrease the solubility of macromolecules in anhydrous organic solvents.
2.3. Calorimetric measurements
2.5. Solvent hydrophilicity
Commercial lyophilized protein powders were dried under vacuum using a microthermoanalyzer (Setaram, MGDTD-17S) at 298 K and 0.1 Pa until constant sample weight. Water content of the dried protein was estimated as 0.003 þ 0.003 g of water/g of dry protein by the Karl Fischer titration method. Calorimetric heat e¡ects on immersing dry protein powders into organic solvents and water were measured at 298 K with a Setaram BT-215 calorimeter according to the described procedure [13]. Typically, 4^8 mg of protein sample was placed in the calorimetric cell and brought in contact with 4.0 ml of a given solvent. The calorimeter was calibrated using the Joule e¡ect and tested by dissolving sodium chloride in water according to the recommendations [25]. The enthalpy changes on dissolution of the dry HSA and CT powders in water were measured at a protein concentration of 1 g/l.
Partial molar excess Gibbs free energy (G Er ) of water in a solvent at in¢nite dilution and 298 K was used as a measure of solvent hydrophilicity. G Er is determined by Eq. 1 [30]:
2.4. Solubility control
Qr w 1=xw
The solubilities of dry solid HSA and K-chymotrypsin were controlled by optical density measurements of supernatants on a Specord M-40 spectrophotometer at 260^300 nm (sensitivity limit of the assay: 0.01 mg/ml). No protein was observed in the liquid phase. No noticeable variation in the absorbance of the liquid phase was observed after exposing the protein sample for at least 6 h to the studied organic solvents. Insolubility of lyophilized HSA in diverse water^organic mixtures was observed even earlier [13^15]. In our experiments DMSO, while not dissolving
G Er RTUln Q r w
1
where Qr w is a mole fraction basis activity coe¤cient for water at in¢nite dilution. The reference state for the activity coe¤cient is pure liquid water (Qw C1 at xw C1). G Er becomes more negative in more hydrophilic solvents. The GEr values are presented in Table 1. The water activity coe¤cients were calculated by one of the two following methods. (1) It is expected that water activity coe¤cients in water immiscible hydrophobic solvents do not depend on the water concentration over the whole solubility range [31]. Hence, Qr w may be estimated using Eq. 2: 2
where xw is the mole fraction solubility of water in a hydrophobic solvent. This method was used to estimate the water activity coe¤cients in carbon tetrachloride and benzene. The water solubilities in carbon tetrachloride at 297 K and benzene at 299 K were taken from [32]. The temperature e¡ect on water solubility in carbon tetrachloride and benzene was tested using the UNIFAC model [33]. The temperature di¡erence of 1³C involves no more than a 5% change in the activity coe¤cient value. (2) Water activity coe¤cients in hydrophilic solvents were calculated using literature data on va-
BBAPRO 36434 1-6-01
362
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Table 1 Enthalpy changes (vHtot ) on immersing dry HSA and CT into organic solvents and water, normalized areas (vA) corresponding to the positive parts of solvent induced di¡erence spectra and the partial molar excess Gibbs free energies (GEr ) of water at in¢nite dilution in organic solvents at 298 K Solvent
Human serum albumin vHtot (J g31 )
First group 1 Carbon tetrachloride 2 Benzene 3 Nitromethane 4 Acetonitrile 5 1,4-Dioxane 6 n-Butanol 7 n-Propanol 8 Pyridine Second group 9 Ethanol 10 Methanol 11 Water 12 DMSO
GEr (kJ mol31 )
Chymotrypsin vA (cm31 )
vHtot (J g31 )
vA (cm31 )
0.2 þ 3.0
0.2 þ 0.2
1.6 þ 2.5
0.2 þ 0.1
18.5
0.2 þ 3.0 7.9 þ 2.0 7.3 þ 3.1 30.6 þ 1.3 32.1 þ 1.9 30.4 þ 0.9 1.0 þ 3.5
0.2 þ 0.2 ^ 0.7 þ 0.3 0.3 þ 0.3 1.0 þ 0.2 ^ 0.2 þ 0.2
^ 1.8 þ 0.2 9.4 þ 2.8 2.8 þ 2.3 ^ ^ 30.4 þ 1.1
0.4 þ 0.4 ^ 0.7 þ 0.5 0.7 þ 0.4 0.5 þ 0.3 0.8 þ 0.3 0.7 þ 0.3
15.0 9.0 5.0 4.6 4.2 3.4 2.7
358.2 þ 0.7 369.7 þ 0.7 391.8 þ 2.8 389.0 þ 2.0
4.1 þ 0.2 4.4 þ 0.2 7.0 þ 1.0 6.0 þ 0.8
338.1 þ 1.9 363.0 þ 2.5 386.6 þ 2.1 376.3 þ 2.1
4.2 þ 0.7 4.4 þ 0.7 8.5 þ 1.5 6.8 þ 0.8
2.3 1.0 0 33.0
The data are presented as the average þ S.E. from 3^6 independent determinations.
por^liquid equilibrium according to Eq. 3:
Q w yw UPt =xw UPo
3
where yw is the measured mole fraction of water in vapor phase, Pt is the total pressure, Po is the saturated vapor pressure of pure water at the same temperature and xw is the mole fraction of water in the liquid phase. Vapor^liquid equilibrium data for water mixtures with pyridine, 1,4-dioxane, methanol, ethanol, n-butanol, n-propanol, and acetonitrile are from [34]. Vapor^liquid equilibrium data for water mixtures with nitromethane and DMSO are from [35,36] respectively. Water activity coe¤cients at in¢nite dilution Qr w were estimated by extrapolation of activity coef¢cients Qw at higher water concentrations by a polynomial expression. The fourth order polynomial was su¤cient to represent the activity coe¤cients in most mixtures. In water^n-butanol mixtures the second order polynomial was used. 2.6. Water contents of organic solvents The humidities of the solvents, determined by Karl Fischer titration [37], were 0.02 mol/l in 1,4-dioxane, 0.01 mol/l in n-butanol, 0.02 mol/l in n-propanol,
0.005 mol/l in nitromethane, 0.022 mol/l in acetonitrile, 0.021 mol/l in ethanol, 0.07 mol/l in pyridine, 0.23 mol/l in DMSO and 0.1 mol/l in methanol. The water activities, calculated as aw = Qw Uxw , did not exceed 0.01 in all the solvents studied. The water contents of carbon tetrachloride and benzene were 6 0.001 mol/l. The water sorption on HSA suspensions has been measured earlier in a series of organic solvents with low water content [14,15]. From these isotherms, using the Langmuir model, the water amount on the protein could be calculated at the lowest water contents referred above to particular solvents. Thus, total water amount on HSA in DMSO was assessed as 0.002 þ 0.002 g of water/g of dry protein (8 þ 8 mol water/mol protein), 0.007 þ 0.003 g/g (26 þ 11 mol/ mol) in ethanol, 0.011 þ 0.009 g/g (45 þ 36 mol/mol) in 1,4-dioxane and 0.003 þ 0.002 g/g (12 þ 8 mol/mol) in n-butanol. These results agree with those of Halling [38]. It could be deduced from his data that at a water activity of 0.01 in organic liquids the water amounts on bovine serum albumin and chymotrypsinogen (proteins close in properties and structure to human serum albumin and chymotrypsin, respectively) are negligible.
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
363
We could resume that the initial water amount on the dried protein (0.003 þ 0.003 g of water/g of protein) does not change essentially on immersion into the studied organic solvents. 3. Results and discussion 3.1. Enthalpy and integral absorbance changes on the immersion of dry proteins in low water organic solvents Figs. 1A and 2A show CT and HSA spectra, respectively, in carbon tetrachloride, benzene, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. Only subtle amide I absorbance changes (if any) relative to protein spectrum in dry air may be noticed in this case. Figs. 1B and 2B demonstrate much more considerable alterations observed on Fig. 2. Amide I spectra of albumin ¢lms. A, B and C depict the same features as in Fig. 1.
Fig. 1. Amide I spectra of CT ¢lms. (A) In dry air (solid line) and in low water organic solvents (dashed lines) including carbon tetrachloride, benzene, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. (B) In ethanol (1) methanol (2), water (3), DMSO (4) and dry air (5). (C) Di¡erences obtained by subtraction of dry air from solvent immersed protein spectra; numbering corresponds to that in B; curves 1^4 are shifted for clarity. As an example, the area (vA) of the di¡erence spectrum in DMSO is shown shaded.
the immersion of both proteins in methanol, ethanol, DMSO and water. The overall changes in the secondary structure of proteins in solvents as compared with the dry protein state were quantitatively characterized by the positive area of solvent induced spectral di¡erence in the amide I region (vA) of proteins. For example, the spectral changes induced by DMSO are shown as shaded in Figs. 1C and 2C. All vA values were normalized on the amide I peak absorbance of the protein sample recorded in the dry state. Typical heat emission curves recorded on immersing dry protein into DMSO and water are given in Fig. 3. As an example, the area corresponding to the total heat e¡ect (vHtot ) evolved at dissolution of a protein in water is shown as shaded. The vA values, enthalpies vHtot of immersion of proteins in organic solvents and enthalpy changes on dissolution of the dry HSA and CT powders in water are presented in Table 1. The interaction of a protein with its environment depends on a variety of factors. It is evident that the measured calorimetric heat and spectral alterations have to be quite complex quantities. However, a good linear correlation between the vHtot values
BBAPRO 36434 1-6-01
364
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Fig. 3. Calorimetric curves recorded on contacting dry solid proteins with water (1) or DMSO (2). As an example, the area corresponding to the total heat e¡ect (vHtot ) of aqueous dissolution is shown shaded.
and spectral vA parameters was found for both CT and HSA (Eqs. 4 and 5, respectively). These linear correlations are also shown in Fig. 4. 3v H tot 38:9 6:3 12:2 1:3Uv A
4
with correlation coe¤cient r = 0.972, rms deviation so = 10.5, number of points n = 7. 3v H tot 36:6 4:6 15:9 0:9Uv A
5
with correlation coe¤cient r = 0.986, rms deviation so = 7.3, number of points n = 10. Slopes and intercepts of Eqs. 4 and 5, within the con¢dence intervals, were found to be the same, thus indicating that the vHtot versus vA correlation is not sensitive to the structural di¡erences between two proteins. The intercept in Eqs. 4 and 5 is small and close to zero. This shows that calorimetric heat effects re£ect mainly the structural rearrangements in the protein. As such, the similar slopes of Eqs. 4 and 5 indicate that the protein structure changes seen in the vA values contribute proportionally to the calorimetric heat for two di¡erent proteins in a series of di¡erent solvents.
expressed as log P (where P is the partition n-octanol^water coe¤cient) [40], solvent molar volume, solvent donor and acceptor numbers [39], and solvent hydrophilicity based on the partial molar excess Gibbs energy G Er of water in a given solvent at in¢nite dilution (Table 1). No correlation was found for either chymotrypsin or albumin when vH and vA were plotted against the solvent parameters dielectric constant, hydrophobicity, molar volume, donor and acceptor numbers. Signi¢cant discrepancies exceeding the experimental errors can be seen in Fig. 5A^E, where for clarity only CT values are presented. A reasonable trend was obtained only when vH and vA were plotted against solvent hydrophilicity (Fig. 6A for CT and Fig. 6B for HSA). As is seen from the ¢gure, solvents may be divided into two groups according to the measured calorimetric heat and spectral changes and this separation agrees well with the solvent hydrophilicity. The ¢rst group of solvents includes liquids with Er G greater than 2.7 kJ/mol, i.e. carbon tetrachloride, benzene, nitromethane, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. The heat evolved upon interaction of dry proteins with solvents of the ¢rst group is close to zero. The second group consists of more hydrophilic liquids such as water, DMSO, methanol and ethanol. Their G Er values are lower than 2.3 kJ/mol. Immersion of dry proteins into solvents of the second group is followed
3.2. In£uence of solvent nature on enthalpy and integral absorbance changes The correlation between the vHtot and vA values and solvent properties was examined using the solvent dielectric constant [39], solvent hydrophobicity
Fig. 4. vH plotted against vA values for HSA and CT. Linear regression lines and 95% con¢dence bands are shown for CT (solid) and HSA (dashed).
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
365
Fig. 5. Enthalpy changes vH (squares) and spectral di¡erences vA (circles) for CT plotted against the solvent's dielectric constant (A), hydrophobicity (B), molar volume (C), donor (D) and acceptor number (E). Symbol numbering corresponds that of the solvents in Table 1.
BBAPRO 36434 1-6-01
366
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
ple physical adsorption on the surface of the solid protein phase. Such a physical adsorption would not induce signi¢cant changes in sorbent structure. The secondary structure alterations and exothermic heat e¡ects observed for proteins in the solvents of the second group are indicative of swelling process (i.e. dissolution of the organic molecules in solid protein phase). Aqueous swelling ¢nally results in protein dissolution. Dry solid proteins do not demonstrate such a swelling in the solvents of the ¢rst group. The possible reason may be that dehydrated proteins are in a kinetically `frozen' state [41]. The potential barrier for swelling of such dried proteins may be quite high, thus preventing swelling in this group of solvents. This hypothesis is supported by the results of [11] in which the thermostability of HSA suspended in n-hexane^pyridine mixtures has been examined by DSC. It was found that heating the HSA suspension in pyridine (and pyridine^hexane mixtures) results in an exothermic peak. The observed exothermic peak indicates that HSA immersed in pyridine, which is a solvent of the ¢rst group, is in the non-equilibrium state at 298 K. Fig. 6. Enthalpy changes vH (squares) and spectral di¡erences vA (circles) for CT (A) and HSA (B) plotted against partial excess molar Gibbs free energy of water at in¢nite dilution in a series of solvents at 298 K. Symbol numbering corresponds that of the solvents in Table 1.
by signi¢cant exothermic enthalpy changes. The heat e¡ects of immersion into water and DMSO are most exothermic for both proteins. Spectral changes vA observed in di¡erent media ¢t to the same solvent classi¢cation. Solvents of the ¢rst group do not alter signi¢cantly the secondary structure relative to its initial state in dry proteins, thus producing vAW0. Immersion of both proteins into solvents of the second group is followed by signi¢cant changes in their amide I spectra (vA s 4.0 cm31 ). The most signi¢cant di¡erences are again observed in water and in DMSO. It is interesting to note that despite their very di¡erent structures both proteins exhibit very similar trends on solvent properties. The interaction of dry CT and HSA with solvents of the second group can hardly be considered a sim-
3.3. In£uence of hydrogen bond donating ability The hydrogen bond donating ability of a solvent molecule seems to be an important factor in lowering the potential barrier of swelling of solid protein. For example, the hydrophilicities of ethanol and pyridine are close (2.3 and 2.7 kJ/mol, respectively). Nevertheless, the vA and vHtot values are nearly zero for both proteins in the hydrogen bond accepting pyridine but rise to signi¢cant values in hydrogen bond donating ethanol (Table 1). Qualitative evidence of the role of the hydrogen bond donating ability of solvents in protein swelling may be seen when comparing the kinetics of heat evolution on protein immersion. For example, water and DMSO demonstrate the lowest G Er values, 0 and 33 kJ/mol, respectively. Heat evolution on HSA dissolution in water was completed within 40 min. Heat evolution at the interaction between HSA and DMSO is completed within 6 h. Slow heat evolution on immersing the partially hydrated (10% w/w) HSA into water^DMSO mixtures was also observed in [42]. It was found that an increase in the
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
water contents in DMSO reduced signi¢cantly the time of interaction between HSA and DMSO. For example, the heat evolution time on HSA suspending was 5 h in DMSO containing 1 mol/l water, 2.5 h in DMSO+3 mol/l water, and 1.5 h in DMSO+5 mol/l water. The total heat e¡ect of HSA immersion did not depend on water concentration and amounted to 389.0 þ 2.0 J/g protein. The heat evolution kinetics on immersing chymotrypsin in DMSO was also found to be much slower than in water. Similar temporary trends were seen in the spectral data when proteins contacted with DMSO and water. DMSO was the only hydrogen bond accepting solvent resulting in structural changes in dry proteins. However, the kinetics of interaction in DMSO was much slower than in less hydrophilic methanol and ethanol in which the interaction between proteins and solvents was completed after 40 min and 1.5 h, respectively. This supports the hypothesis that the hydrogen bond donating ability of solvents contributes signi¢cantly to decreasing the potential barrier of protein swelling. 3.4. Structural events on swelling in the second group solvents Hydrogen bond donating and accepting solvents a¡ect the protein structure in di¡erent ways. This is evidenced by the IR spectra of albumin in various environments in Fig. 2B,C. Hydrogen bond donating water, methanol and ethanol result in an absorbance increase at 1658 cm31 which indicates an increased helical content [22]. Hydrogen bond accepting DMSO reduces strongly the initial helical content. The increase in the absorbance at 1628 and 1690 cm31 points to the formation of intermolecular Lstructure and aggregation. The structural events on swelling chymotrypsin, unlike those of HSA, reveal a more complex behavior. Contact between CT and alcohols results not only in the generation of some helical structure but also in the subsequent formation of more extensive L-sheets (Fig. 1B,C). Swelling of CT in DMSO shows, as an initial step, relatively fast randomization of the protein structure followed by slow conversion to extensive L-aggregation. Despite the observed diversities the ¢nal vA values of induced spectral changes are close for both proteins.
367
Fig. 7. Qualitative scheme describing the dependence of the potential barrier of protein swelling on the nature of the solvent.
3.5. Di¡erentiation of solvents on their protein swelling ability The behavior of dry solid proteins in organic solvents is conceptualized using the solvent hydrophilicity and hydrogen bond donating ability in Fig. 7. It depicts the potential energy changes along the reaction coordinate for three typical cases considered in this work. Hydrogen bonding between a protein and a solvent results in the decrease of the potential energy barrier of protein swelling. According to this conception, the total energy of formation of the `solid protein+organic solvent' system is in correlation with the decrease of the potential barrier for protein swelling. It is known that GEr measures the deviation of the water state in organic solvents from ideality [30]. This deviation mainly arises from the formation of hydrogen bonds between water and organic molecules. As such, the GEr values may be considered a measure of the intensity of hydrogen bonding between water and solvent molecules. Strongly negative G Er values indicate strong hydrogen bonding between water and solvents. It is hypothesized that the contribution of hydrogen bonding between protein and organic solvent to the total measured energy (immersion heat) correlates with the G Er values. According to this criterion, solvents with low hydrophilicity (GEr positive and exceeding 2.7 kJ/mol) are not e¡ective in hydrogen bonding with a protein. Hence, the potential barrier of protein swelling is high in these solvents. This status is diagrammed by the upper curve in Fig. 7. Raising the temperature may
BBAPRO 36434 1-6-01
368
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
help in passing this potential barrier, as was suggested to explain the exothermic DSC peaks for HSA suspended in pyridine [11]. The solvents of the second group (i.e. ethanol, methanol, water and DMSO) demonstrate high hydrophilicity (G Er less than 2.3 kJ/mol or even negative) and e¡ectiveness in hydrogen bond formation with both water and protein, thus resulting in the swelling of proteins at room temperature. The bottom curve in Fig. 7 corresponds to this case. The hydrogen bond donating ability of solvents strengthens their swelling e¤cacy towards dry proteins. Therefore, at close solvent hydrophilicity the potential barrier of swelling is lower in hydrogen bond donating media compared to hydrogen bond accepting ones. This was exempli¢ed by the slow swelling of albumin in DMSO as compared with that in water and by the absence of swelling in pyridine as compared with ethanol. The e¡ect of hydrophilic hydrogen bond accepting solvents such as DMSO is shown in Fig. 7 by the middle curve. Thus, the potential of a solvent to form hydrogen bonds appears an important factor in controlling the swelling stability of dry solid proteins in di¡erent media. Acknowledgements The authors gratefully acknowledge Mikhail Borisover (Institute of Soil, Water, and Environmental Sciences, the Volcani Center, Bet Dagan, Israel) for helpful discussion. This research was supported by the Russian Foundation for Basic Researches (Grant No. 98-03-32102) and the Center for Fundamental Natural Sciences at Saint-Petersburg State University (Grant No. 97-0-9.3-283). References [1] [2] [3] [4]
A.M. Klibanov, Trends Biochem. Sci. 14 (1989) 141^144. A.M. Klibanov, Chemtech 16 (1986) 354^359. E. Battistel, D. Bianchi, J. Phys. Chem. 98 (1994) 5368^5375. D.B. Volkin, A. Staubli, A. Langer, A.M. Klibanov, Biotechnol. Bioeng. 37 (1991) 843^853. [5] K. Dabulis, A.M. Klibanov, Biotechnol. Bioeng. 39 (1992) 176^185. [6] C.J. Slade, J. Mol. Catal. B Enzym. 9 (2000) 97^105.
[7] O. Almarsson, A.M. Klibanov, Biotechnol. Bioeng. 49 (1996) 87^92. [8] T.C. de Sampaio, R.B. Melo, T.F. Moura, S. Michel, S. Barreiros, Biotechnol. Bioeng. 50 (1996) 257^264. [9] P.A. Fitzpatrick, A.M. Klibanov, J. Am. Chem. Soc. 113 (1991) 3166^3171. [10] A. Ducret, M. Trani, R. Lortie, Enzyme Microb. Technol. 22 (1998) 212^216. [11] M.D. Borisover, D.V. Zakharychev, B.N. Solomonov, J. Therm. Anal. Calorimetry 55 (1999) 85^92. [12] D.V. Zakharychev, M.D. Borisover, B.N. Solomonov, Russ. J. Phys. Chem. (Engl. Transl.) 69 (1995) 162^166. [13] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, J. Phys. Org. Chem. 6 (1993) 251^253. [14] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, Thermochim. Acta 284 (1996) 263^277. [15] V.A. Sirotkin, M.D. Borisover, B.N. Solomonov, Biophys. Chem. 69 (1997) 239^248. [16] M.D. Borisover, V.A. Sirotkin, D.V. Zakharychev, B.N. Solomonov, in: Abstracts, 15th European Chemistry at Interfaces Conference, Jerusalem, 1998, p. 64. [17] K. Griebenow, A.M. Klibanov, Biotechnol. Bioeng. 53 (1997) 351^362. [18] A. Dong, J.D. Meyer, B.S. Kendrick, M.C. Manning, J.F. Carpenter, Arch. Biochem. Biophys. 334 (1996) 406^414. [19] A. Zaks, A.M. Klibanov, J. Biol. Chem. 263 (1988) 8017^ 8021. [20] D.D. Perrin, W.L.F. Armarego, D.R. Perrin, Puri¢cation of Laboratory Chemicals, Pergamon Press, Oxford, 1980. [21] R.C. Weast (Ed.), Handbook of Chemistry and Physics, 58th edn., CRC Press, Cleveland, OH, 1977^1978. [22] A. Dong, P. Huang, W.S. Caughey, Biochemistry 29 (1990) 3303^3308. [23] K. Gribenow, A.M. Klibanov, Proc. Natl. Acad. Sci. USA 92 (1995) 10969^10976. [24] H.R. Constantino, K. Griebenow, P. Mishra, R. Langer, A.M. Klibanov, Biochim. Biophys. Acta 1253 (1995) 69^74. [25] V.A. Medvedev, M.E. E¢mov, Russian J. Phys. Chem. (Engl. Transl.) 49 (1975) 780^783. [26] S.J. Singer, Adv. Protein Chem. 17 (1962) 1^68. [27] J.T. Chin, S.L. Wheeler, A.M. Klibanov, Biotechnol. Bioeng. 44 (1994) 140^145. [28] P. Douzou, Cryobiochemistry, Academic Press, New York, 1977. [29] R.V. Rariy, A.M. Klibanov, Biotechnol. Bioeng. 62 (1999) 704^710. [30] J.M. Prausnitz, Molecular Thermodynamics of Fluid-phase Equilibria, Prentice-Hall, Englewood Cli¡s, NJ, 1969. [31] S. D Christian, A.A. Taha, B.W. Gash, Q. Rev. Chem. Soc. 24 (1970) 20^36. [32] J.A. Riddick, E.E. Toops, Organic Solvents. Physical Properties and Methods of Puri¢cation, 2nd edn., Interscience Publishers, New York, 1955. [33] H.K. Hansen, P. Rasmussen, A. Fredenslund, M. Shiller, J. Gmehling, Ind. Eng. Chem. Res. 30 (1991) 2352^2355. [34] V.B. Kogan, V.M. Fridman, V.V. Kafarov, Ravnovesie
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
[35]
[36] [37] [38]
Mezhdu Zhidkost'u i Parom (Liquid^Vapour Equilibria) (in Russian), V.1, Nauka, Moscow, 1966. G.S. Lyudmirskaya, T.A. Barsukova, A.M. Bogomolnij, Ravnovesia Zhidkost^Par. (Liquid^Vapour Equilibria) (in Russian), Khimia, Moscow, 1987. S.Y. Lam, R.L. Benoit, Can. J. Chem. 52 (1974) 718^722. H.A. Laitinen, W.E. Harris, Chemical Analysis, 2nd edn., McGraw-Hill, New York, 1975. P.J. Halling, Biochim. Biophys. Acta 1040 (1990) 225^228.
369
[39] C. Raichardt, Solvents and Solvent E¡ects in Organic Chemistry, 2nd edn., VCH, Weinheim, 1988. [40] C. Laane, S. Boeren, K. Vos, C. Veeger, Biotechnol. Bioeng. 30 (1987) 81^87. [41] R.B. Gregory, Protein hydration and glass transition behavior, in: R.B. Gregory (Ed.), Protein^solvent Interaction, vol. 264, Marcel Dekker, New York, 1995, pp. 191^264. [42] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, Thermochim. Acta 256 (1995) 175^183.
BBAPRO 36434 1-6-01
Calorimetric and Fourier transform infrared spectroscopic study of solid proteins immersed in low water organic solvents Vladimir A. Sirotkin a; *, Albert N. Zinatullin a , Boris N. Solomonov a , Djihanguir A. Faizullin b , Vladimir D. Fedotov b a
Department of Chemistry, Kazan State University, Kremlevskaya str. 18, Kazan 420008, Russia b Kazan Institute of Biochemistry and Biophysics, Kazan 420503, Russia Received 24 May 2000; received in revised form 23 March 2001; accepted 4 April 2001
Abstract Calorimetric heat effects and structural rearrangements assessed by means of Fourier transform infrared (FTIR) amide I spectra were followed by immersing dry human serum albumin and bovine pancreatic K-chymotrypsin in low water organic solvents and in pure water at 298 K. Enthalpy changes upon immersion of the proteins in different media are in a good linear correlation with the corresponding IR absorbance changes. Based on calorimetric and FTIR data the solvents were divided into two groups. The first group includes carbon tetrachloride, benzene, nitromethane, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine where no significant heat evolution and structural changes were found during protein immersion. Due to kinetic reasons no significant protein^solvent interactions are expected in such systems. The second group of solvents includes dimethyl sulfoxide, methanol, ethanol, and water. Immersion of proteins in these media results in protein swelling and involves significant exothermic heat evolution and structural changes in the protein. Dividing of different media in the two groups is in a qualitative correlation with the solvent hydrophilicity defined as partial excess molar Gibbs free energy of water at infinite dilution in a given solvent. The first group includes the solvents with hydrophilicity exceeding 2.7 kJ/mol. More hydrophilic second group solvents have this energy values less than 2.3 kJ/mol. The hydrogen bond donating ability of the solvents also assists in protein swelling. Hydrogen bonding between protein and solvent is assumed to be a main factor controlling the swelling of dry solid proteins in the studied solvents. ß 2001 Elsevier Science B.V. All rights reserved. Keywords: Fourier transform infrared; Immersion calorimetry; Organic solvent; Solvent hydrophilicity, swelling; Human serum albumin; Bovine pancreatic K-chymotrypsin
1. Introduction Enzymes suspended in organic solvents with low water contents may catalyze reactions not feasible in aqueous medium [1,2], and demonstrate greatly enAbbreviations: FTIR, Fourier transform infrared; HSA, human serum albumin; CT, bovine pancreatic K-chymotrypsin * Corresponding author. Fax: +7-8432-380-994; E-mail: [email protected]
hanced thermostability [3,4] and `molecular memory' [5,6]. This biotechnological potential of enzymes in organic medium depends strongly on the nature of the organic solvent. For example, solvents may activate or suppress enzymatic activity [7,8], a¡ect the enantioselectivity of the suspended enzymes [9,10], result in exothermic peaks on di¡erential scanning calorimetry (DSC) curves of the protein suspended in organic solvents [11,12], and in£uence ligand binding to the imprinted protein [5].
0167-4838 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 0 1 ) 0 0 2 0 1 - 1
BBAPRO 36434 1-6-01
360
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Therefore, information on solid protein^organic solvent interactions is of importance in explaining various protein activities. Due to the ability to monitor the thermodynamic and structural changes in such heterogeneous systems, the combination of calorimetry and infrared (IR) spectroscopy has a great potential in examining factors governing solid protein^solvent interactions. Earlier we proposed a calorimetric approach for the examination of processes that occur on immersing the solid proteins in water^organic mixtures [13^ 15]. This approach involved the measurement of the enthalpies corresponding to the formation of the `protein+liquid' heterogeneous systems. It was suggested [14,15] that the enthalpy of the formation of human serum albumin (HSA) suspensions in water^ organic mixtures is controlled mainly by two processes. The ¢rst is water desorption/sorption which may follow a Langmuir type trend. The thermodynamics of water sorption by the protein in organic solvents was evaluated using the Langmuir model. The solvent e¡ect on thermodynamic parameters of water sorption by the protein was found to be in a strong correlation with the water solvation thermodynamics in organic medium [14]. The second process attributed to protein swelling in a water^organic mixture [14,16] is associated with signi¢cant heat e¡ects and an increase in the surface area accessible for sorption of water. However, the role of protein conformational changes in calorimetric events observed on suspending the protein in organic medium was not justi¢ed. Fourier transform infrared (FTIR) spectroscopy has been successfully applied to the examination of the secondary structure of proteins in organic solvents [17,18]. It was shown that some organic solvents induce conformational changes in the suspended protein. However, the factors governing the protein structure in low water organic medium are not really clear. In the present study we want to combine isothermal calorimetry and IR spectroscopy in the examination of protein^solvent interactions. The objective of this combined calorimetric and IR study is to elucidate the relations between the e¡ect of solvent nature on protein secondary structure and heat e¡ects of interaction of solid protein with organic liquids.
Bovine pancreatic K-chymotrypsin (CT) and human serum albumin were used as model proteins. As is known, water sorbed by proteins in organic solvents may signi¢cantly in£uence the catalytic properties [9,19], protein conformation [17] and thermostability [3,4,12], and enthalpies of suspension formation [13^15]. Hence, in order to focus on protein^ organic solvent interactions, dry proteins (dehydrated at a water activity less than 0.01) were investigated when immersed in essentially low water solvents. The behavior of dry proteins in nonaqueous solvents was also compared with that in a water environment. 2. Materials and methods 2.1. Materials Human serum albumin (Sigma, product No. A 1887, essentially fatty acid free) and bovine pancreatic K-chymotrypsin (Sigma, C 4129, essentially salt free; EC 3.4.21.1; speci¢c activity 52 units/mg of solid) were used without further puri¢cation. Organic solvents (reagent grade, purity s 99%) were puri¢ed and dried according to the recommendations [20] and then stored over dry 3A molecular sieves for at least 24 h prior to use. Water used was doubly distilled. 2.2. FTIR measurements FTIR spectrometry was carried out on a Vector 22 (Bruker) FTIR spectrophotometer at 4 cm31 resolution. Vibration spectra were obtained with glassy-like protein ¢lms casted from a 2% (w/v) water solution onto a CaF2 window at room humidity. After mounting the windows in the sample cell, the ¢lm was dehydrated by £ushing air dried over P2 O5 powder. The relative vapor pressure over P2 O5 at 298 K does not exceed 0.01 [21]. The protein ¢lm was £ushed until no further spectral changes were detected in the 3450 cm31 water absorbance region and the amide A contour on this side represented a smooth line without any visible shoulders. The cell was then ¢lled with the chosen organic solvent. Spectra were recorded as a function of time until equilib-
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
361
rium was achieved. Then, spectra of pure solvent were recorded and subtracted from the sample spectra. Protein spectra in a water environment were obtained on protein ¢lms wetted by £ushing air at 99% relative humidity. The spectrum of liquid water was then subtracted from the spectra of wet ¢lms in accord with the criteria described in [22]. Changes of the secondary protein structure were analyzed using the known correlation between secondary structure elements in proteins and peak positions in the amide I spectra [23,24].
the proteins, transformed the samples to a transparent dense gel. It is known that some nonaqueous solvents, for example DMSO and formamide, can dissolve proteins [26,27]. However, the mechanism of the protein dissolving potential of anhydrous organic solvents is still not clear [27]. Protein solubility in organic solvents may depend strongly on a variety of factors such as pH, temperature, solvent humidity, and electrolyte additives [26^29]. Probably, deep dehydration of the proteins and absence of salts may strongly decrease the solubility of macromolecules in anhydrous organic solvents.
2.3. Calorimetric measurements
2.5. Solvent hydrophilicity
Commercial lyophilized protein powders were dried under vacuum using a microthermoanalyzer (Setaram, MGDTD-17S) at 298 K and 0.1 Pa until constant sample weight. Water content of the dried protein was estimated as 0.003 þ 0.003 g of water/g of dry protein by the Karl Fischer titration method. Calorimetric heat e¡ects on immersing dry protein powders into organic solvents and water were measured at 298 K with a Setaram BT-215 calorimeter according to the described procedure [13]. Typically, 4^8 mg of protein sample was placed in the calorimetric cell and brought in contact with 4.0 ml of a given solvent. The calorimeter was calibrated using the Joule e¡ect and tested by dissolving sodium chloride in water according to the recommendations [25]. The enthalpy changes on dissolution of the dry HSA and CT powders in water were measured at a protein concentration of 1 g/l.
Partial molar excess Gibbs free energy (G Er ) of water in a solvent at in¢nite dilution and 298 K was used as a measure of solvent hydrophilicity. G Er is determined by Eq. 1 [30]:
2.4. Solubility control
Qr w 1=xw
The solubilities of dry solid HSA and K-chymotrypsin were controlled by optical density measurements of supernatants on a Specord M-40 spectrophotometer at 260^300 nm (sensitivity limit of the assay: 0.01 mg/ml). No protein was observed in the liquid phase. No noticeable variation in the absorbance of the liquid phase was observed after exposing the protein sample for at least 6 h to the studied organic solvents. Insolubility of lyophilized HSA in diverse water^organic mixtures was observed even earlier [13^15]. In our experiments DMSO, while not dissolving
G Er RTUln Q r w
1
where Qr w is a mole fraction basis activity coe¤cient for water at in¢nite dilution. The reference state for the activity coe¤cient is pure liquid water (Qw C1 at xw C1). G Er becomes more negative in more hydrophilic solvents. The GEr values are presented in Table 1. The water activity coe¤cients were calculated by one of the two following methods. (1) It is expected that water activity coe¤cients in water immiscible hydrophobic solvents do not depend on the water concentration over the whole solubility range [31]. Hence, Qr w may be estimated using Eq. 2: 2
where xw is the mole fraction solubility of water in a hydrophobic solvent. This method was used to estimate the water activity coe¤cients in carbon tetrachloride and benzene. The water solubilities in carbon tetrachloride at 297 K and benzene at 299 K were taken from [32]. The temperature e¡ect on water solubility in carbon tetrachloride and benzene was tested using the UNIFAC model [33]. The temperature di¡erence of 1³C involves no more than a 5% change in the activity coe¤cient value. (2) Water activity coe¤cients in hydrophilic solvents were calculated using literature data on va-
BBAPRO 36434 1-6-01
362
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Table 1 Enthalpy changes (vHtot ) on immersing dry HSA and CT into organic solvents and water, normalized areas (vA) corresponding to the positive parts of solvent induced di¡erence spectra and the partial molar excess Gibbs free energies (GEr ) of water at in¢nite dilution in organic solvents at 298 K Solvent
Human serum albumin vHtot (J g31 )
First group 1 Carbon tetrachloride 2 Benzene 3 Nitromethane 4 Acetonitrile 5 1,4-Dioxane 6 n-Butanol 7 n-Propanol 8 Pyridine Second group 9 Ethanol 10 Methanol 11 Water 12 DMSO
GEr (kJ mol31 )
Chymotrypsin vA (cm31 )
vHtot (J g31 )
vA (cm31 )
0.2 þ 3.0
0.2 þ 0.2
1.6 þ 2.5
0.2 þ 0.1
18.5
0.2 þ 3.0 7.9 þ 2.0 7.3 þ 3.1 30.6 þ 1.3 32.1 þ 1.9 30.4 þ 0.9 1.0 þ 3.5
0.2 þ 0.2 ^ 0.7 þ 0.3 0.3 þ 0.3 1.0 þ 0.2 ^ 0.2 þ 0.2
^ 1.8 þ 0.2 9.4 þ 2.8 2.8 þ 2.3 ^ ^ 30.4 þ 1.1
0.4 þ 0.4 ^ 0.7 þ 0.5 0.7 þ 0.4 0.5 þ 0.3 0.8 þ 0.3 0.7 þ 0.3
15.0 9.0 5.0 4.6 4.2 3.4 2.7
358.2 þ 0.7 369.7 þ 0.7 391.8 þ 2.8 389.0 þ 2.0
4.1 þ 0.2 4.4 þ 0.2 7.0 þ 1.0 6.0 þ 0.8
338.1 þ 1.9 363.0 þ 2.5 386.6 þ 2.1 376.3 þ 2.1
4.2 þ 0.7 4.4 þ 0.7 8.5 þ 1.5 6.8 þ 0.8
2.3 1.0 0 33.0
The data are presented as the average þ S.E. from 3^6 independent determinations.
por^liquid equilibrium according to Eq. 3:
Q w yw UPt =xw UPo
3
where yw is the measured mole fraction of water in vapor phase, Pt is the total pressure, Po is the saturated vapor pressure of pure water at the same temperature and xw is the mole fraction of water in the liquid phase. Vapor^liquid equilibrium data for water mixtures with pyridine, 1,4-dioxane, methanol, ethanol, n-butanol, n-propanol, and acetonitrile are from [34]. Vapor^liquid equilibrium data for water mixtures with nitromethane and DMSO are from [35,36] respectively. Water activity coe¤cients at in¢nite dilution Qr w were estimated by extrapolation of activity coef¢cients Qw at higher water concentrations by a polynomial expression. The fourth order polynomial was su¤cient to represent the activity coe¤cients in most mixtures. In water^n-butanol mixtures the second order polynomial was used. 2.6. Water contents of organic solvents The humidities of the solvents, determined by Karl Fischer titration [37], were 0.02 mol/l in 1,4-dioxane, 0.01 mol/l in n-butanol, 0.02 mol/l in n-propanol,
0.005 mol/l in nitromethane, 0.022 mol/l in acetonitrile, 0.021 mol/l in ethanol, 0.07 mol/l in pyridine, 0.23 mol/l in DMSO and 0.1 mol/l in methanol. The water activities, calculated as aw = Qw Uxw , did not exceed 0.01 in all the solvents studied. The water contents of carbon tetrachloride and benzene were 6 0.001 mol/l. The water sorption on HSA suspensions has been measured earlier in a series of organic solvents with low water content [14,15]. From these isotherms, using the Langmuir model, the water amount on the protein could be calculated at the lowest water contents referred above to particular solvents. Thus, total water amount on HSA in DMSO was assessed as 0.002 þ 0.002 g of water/g of dry protein (8 þ 8 mol water/mol protein), 0.007 þ 0.003 g/g (26 þ 11 mol/ mol) in ethanol, 0.011 þ 0.009 g/g (45 þ 36 mol/mol) in 1,4-dioxane and 0.003 þ 0.002 g/g (12 þ 8 mol/mol) in n-butanol. These results agree with those of Halling [38]. It could be deduced from his data that at a water activity of 0.01 in organic liquids the water amounts on bovine serum albumin and chymotrypsinogen (proteins close in properties and structure to human serum albumin and chymotrypsin, respectively) are negligible.
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
363
We could resume that the initial water amount on the dried protein (0.003 þ 0.003 g of water/g of protein) does not change essentially on immersion into the studied organic solvents. 3. Results and discussion 3.1. Enthalpy and integral absorbance changes on the immersion of dry proteins in low water organic solvents Figs. 1A and 2A show CT and HSA spectra, respectively, in carbon tetrachloride, benzene, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. Only subtle amide I absorbance changes (if any) relative to protein spectrum in dry air may be noticed in this case. Figs. 1B and 2B demonstrate much more considerable alterations observed on Fig. 2. Amide I spectra of albumin ¢lms. A, B and C depict the same features as in Fig. 1.
Fig. 1. Amide I spectra of CT ¢lms. (A) In dry air (solid line) and in low water organic solvents (dashed lines) including carbon tetrachloride, benzene, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. (B) In ethanol (1) methanol (2), water (3), DMSO (4) and dry air (5). (C) Di¡erences obtained by subtraction of dry air from solvent immersed protein spectra; numbering corresponds to that in B; curves 1^4 are shifted for clarity. As an example, the area (vA) of the di¡erence spectrum in DMSO is shown shaded.
the immersion of both proteins in methanol, ethanol, DMSO and water. The overall changes in the secondary structure of proteins in solvents as compared with the dry protein state were quantitatively characterized by the positive area of solvent induced spectral di¡erence in the amide I region (vA) of proteins. For example, the spectral changes induced by DMSO are shown as shaded in Figs. 1C and 2C. All vA values were normalized on the amide I peak absorbance of the protein sample recorded in the dry state. Typical heat emission curves recorded on immersing dry protein into DMSO and water are given in Fig. 3. As an example, the area corresponding to the total heat e¡ect (vHtot ) evolved at dissolution of a protein in water is shown as shaded. The vA values, enthalpies vHtot of immersion of proteins in organic solvents and enthalpy changes on dissolution of the dry HSA and CT powders in water are presented in Table 1. The interaction of a protein with its environment depends on a variety of factors. It is evident that the measured calorimetric heat and spectral alterations have to be quite complex quantities. However, a good linear correlation between the vHtot values
BBAPRO 36434 1-6-01
364
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
Fig. 3. Calorimetric curves recorded on contacting dry solid proteins with water (1) or DMSO (2). As an example, the area corresponding to the total heat e¡ect (vHtot ) of aqueous dissolution is shown shaded.
and spectral vA parameters was found for both CT and HSA (Eqs. 4 and 5, respectively). These linear correlations are also shown in Fig. 4. 3v H tot 38:9 6:3 12:2 1:3Uv A
4
with correlation coe¤cient r = 0.972, rms deviation so = 10.5, number of points n = 7. 3v H tot 36:6 4:6 15:9 0:9Uv A
5
with correlation coe¤cient r = 0.986, rms deviation so = 7.3, number of points n = 10. Slopes and intercepts of Eqs. 4 and 5, within the con¢dence intervals, were found to be the same, thus indicating that the vHtot versus vA correlation is not sensitive to the structural di¡erences between two proteins. The intercept in Eqs. 4 and 5 is small and close to zero. This shows that calorimetric heat effects re£ect mainly the structural rearrangements in the protein. As such, the similar slopes of Eqs. 4 and 5 indicate that the protein structure changes seen in the vA values contribute proportionally to the calorimetric heat for two di¡erent proteins in a series of di¡erent solvents.
expressed as log P (where P is the partition n-octanol^water coe¤cient) [40], solvent molar volume, solvent donor and acceptor numbers [39], and solvent hydrophilicity based on the partial molar excess Gibbs energy G Er of water in a given solvent at in¢nite dilution (Table 1). No correlation was found for either chymotrypsin or albumin when vH and vA were plotted against the solvent parameters dielectric constant, hydrophobicity, molar volume, donor and acceptor numbers. Signi¢cant discrepancies exceeding the experimental errors can be seen in Fig. 5A^E, where for clarity only CT values are presented. A reasonable trend was obtained only when vH and vA were plotted against solvent hydrophilicity (Fig. 6A for CT and Fig. 6B for HSA). As is seen from the ¢gure, solvents may be divided into two groups according to the measured calorimetric heat and spectral changes and this separation agrees well with the solvent hydrophilicity. The ¢rst group of solvents includes liquids with Er G greater than 2.7 kJ/mol, i.e. carbon tetrachloride, benzene, nitromethane, acetonitrile, 1,4-dioxane, n-butanol, n-propanol and pyridine. The heat evolved upon interaction of dry proteins with solvents of the ¢rst group is close to zero. The second group consists of more hydrophilic liquids such as water, DMSO, methanol and ethanol. Their G Er values are lower than 2.3 kJ/mol. Immersion of dry proteins into solvents of the second group is followed
3.2. In£uence of solvent nature on enthalpy and integral absorbance changes The correlation between the vHtot and vA values and solvent properties was examined using the solvent dielectric constant [39], solvent hydrophobicity
Fig. 4. vH plotted against vA values for HSA and CT. Linear regression lines and 95% con¢dence bands are shown for CT (solid) and HSA (dashed).
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
365
Fig. 5. Enthalpy changes vH (squares) and spectral di¡erences vA (circles) for CT plotted against the solvent's dielectric constant (A), hydrophobicity (B), molar volume (C), donor (D) and acceptor number (E). Symbol numbering corresponds that of the solvents in Table 1.
BBAPRO 36434 1-6-01
366
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
ple physical adsorption on the surface of the solid protein phase. Such a physical adsorption would not induce signi¢cant changes in sorbent structure. The secondary structure alterations and exothermic heat e¡ects observed for proteins in the solvents of the second group are indicative of swelling process (i.e. dissolution of the organic molecules in solid protein phase). Aqueous swelling ¢nally results in protein dissolution. Dry solid proteins do not demonstrate such a swelling in the solvents of the ¢rst group. The possible reason may be that dehydrated proteins are in a kinetically `frozen' state [41]. The potential barrier for swelling of such dried proteins may be quite high, thus preventing swelling in this group of solvents. This hypothesis is supported by the results of [11] in which the thermostability of HSA suspended in n-hexane^pyridine mixtures has been examined by DSC. It was found that heating the HSA suspension in pyridine (and pyridine^hexane mixtures) results in an exothermic peak. The observed exothermic peak indicates that HSA immersed in pyridine, which is a solvent of the ¢rst group, is in the non-equilibrium state at 298 K. Fig. 6. Enthalpy changes vH (squares) and spectral di¡erences vA (circles) for CT (A) and HSA (B) plotted against partial excess molar Gibbs free energy of water at in¢nite dilution in a series of solvents at 298 K. Symbol numbering corresponds that of the solvents in Table 1.
by signi¢cant exothermic enthalpy changes. The heat e¡ects of immersion into water and DMSO are most exothermic for both proteins. Spectral changes vA observed in di¡erent media ¢t to the same solvent classi¢cation. Solvents of the ¢rst group do not alter signi¢cantly the secondary structure relative to its initial state in dry proteins, thus producing vAW0. Immersion of both proteins into solvents of the second group is followed by signi¢cant changes in their amide I spectra (vA s 4.0 cm31 ). The most signi¢cant di¡erences are again observed in water and in DMSO. It is interesting to note that despite their very di¡erent structures both proteins exhibit very similar trends on solvent properties. The interaction of dry CT and HSA with solvents of the second group can hardly be considered a sim-
3.3. In£uence of hydrogen bond donating ability The hydrogen bond donating ability of a solvent molecule seems to be an important factor in lowering the potential barrier of swelling of solid protein. For example, the hydrophilicities of ethanol and pyridine are close (2.3 and 2.7 kJ/mol, respectively). Nevertheless, the vA and vHtot values are nearly zero for both proteins in the hydrogen bond accepting pyridine but rise to signi¢cant values in hydrogen bond donating ethanol (Table 1). Qualitative evidence of the role of the hydrogen bond donating ability of solvents in protein swelling may be seen when comparing the kinetics of heat evolution on protein immersion. For example, water and DMSO demonstrate the lowest G Er values, 0 and 33 kJ/mol, respectively. Heat evolution on HSA dissolution in water was completed within 40 min. Heat evolution at the interaction between HSA and DMSO is completed within 6 h. Slow heat evolution on immersing the partially hydrated (10% w/w) HSA into water^DMSO mixtures was also observed in [42]. It was found that an increase in the
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
water contents in DMSO reduced signi¢cantly the time of interaction between HSA and DMSO. For example, the heat evolution time on HSA suspending was 5 h in DMSO containing 1 mol/l water, 2.5 h in DMSO+3 mol/l water, and 1.5 h in DMSO+5 mol/l water. The total heat e¡ect of HSA immersion did not depend on water concentration and amounted to 389.0 þ 2.0 J/g protein. The heat evolution kinetics on immersing chymotrypsin in DMSO was also found to be much slower than in water. Similar temporary trends were seen in the spectral data when proteins contacted with DMSO and water. DMSO was the only hydrogen bond accepting solvent resulting in structural changes in dry proteins. However, the kinetics of interaction in DMSO was much slower than in less hydrophilic methanol and ethanol in which the interaction between proteins and solvents was completed after 40 min and 1.5 h, respectively. This supports the hypothesis that the hydrogen bond donating ability of solvents contributes signi¢cantly to decreasing the potential barrier of protein swelling. 3.4. Structural events on swelling in the second group solvents Hydrogen bond donating and accepting solvents a¡ect the protein structure in di¡erent ways. This is evidenced by the IR spectra of albumin in various environments in Fig. 2B,C. Hydrogen bond donating water, methanol and ethanol result in an absorbance increase at 1658 cm31 which indicates an increased helical content [22]. Hydrogen bond accepting DMSO reduces strongly the initial helical content. The increase in the absorbance at 1628 and 1690 cm31 points to the formation of intermolecular Lstructure and aggregation. The structural events on swelling chymotrypsin, unlike those of HSA, reveal a more complex behavior. Contact between CT and alcohols results not only in the generation of some helical structure but also in the subsequent formation of more extensive L-sheets (Fig. 1B,C). Swelling of CT in DMSO shows, as an initial step, relatively fast randomization of the protein structure followed by slow conversion to extensive L-aggregation. Despite the observed diversities the ¢nal vA values of induced spectral changes are close for both proteins.
367
Fig. 7. Qualitative scheme describing the dependence of the potential barrier of protein swelling on the nature of the solvent.
3.5. Di¡erentiation of solvents on their protein swelling ability The behavior of dry solid proteins in organic solvents is conceptualized using the solvent hydrophilicity and hydrogen bond donating ability in Fig. 7. It depicts the potential energy changes along the reaction coordinate for three typical cases considered in this work. Hydrogen bonding between a protein and a solvent results in the decrease of the potential energy barrier of protein swelling. According to this conception, the total energy of formation of the `solid protein+organic solvent' system is in correlation with the decrease of the potential barrier for protein swelling. It is known that GEr measures the deviation of the water state in organic solvents from ideality [30]. This deviation mainly arises from the formation of hydrogen bonds between water and organic molecules. As such, the GEr values may be considered a measure of the intensity of hydrogen bonding between water and solvent molecules. Strongly negative G Er values indicate strong hydrogen bonding between water and solvents. It is hypothesized that the contribution of hydrogen bonding between protein and organic solvent to the total measured energy (immersion heat) correlates with the G Er values. According to this criterion, solvents with low hydrophilicity (GEr positive and exceeding 2.7 kJ/mol) are not e¡ective in hydrogen bonding with a protein. Hence, the potential barrier of protein swelling is high in these solvents. This status is diagrammed by the upper curve in Fig. 7. Raising the temperature may
BBAPRO 36434 1-6-01
368
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
help in passing this potential barrier, as was suggested to explain the exothermic DSC peaks for HSA suspended in pyridine [11]. The solvents of the second group (i.e. ethanol, methanol, water and DMSO) demonstrate high hydrophilicity (G Er less than 2.3 kJ/mol or even negative) and e¡ectiveness in hydrogen bond formation with both water and protein, thus resulting in the swelling of proteins at room temperature. The bottom curve in Fig. 7 corresponds to this case. The hydrogen bond donating ability of solvents strengthens their swelling e¤cacy towards dry proteins. Therefore, at close solvent hydrophilicity the potential barrier of swelling is lower in hydrogen bond donating media compared to hydrogen bond accepting ones. This was exempli¢ed by the slow swelling of albumin in DMSO as compared with that in water and by the absence of swelling in pyridine as compared with ethanol. The e¡ect of hydrophilic hydrogen bond accepting solvents such as DMSO is shown in Fig. 7 by the middle curve. Thus, the potential of a solvent to form hydrogen bonds appears an important factor in controlling the swelling stability of dry solid proteins in di¡erent media. Acknowledgements The authors gratefully acknowledge Mikhail Borisover (Institute of Soil, Water, and Environmental Sciences, the Volcani Center, Bet Dagan, Israel) for helpful discussion. This research was supported by the Russian Foundation for Basic Researches (Grant No. 98-03-32102) and the Center for Fundamental Natural Sciences at Saint-Petersburg State University (Grant No. 97-0-9.3-283). References [1] [2] [3] [4]
A.M. Klibanov, Trends Biochem. Sci. 14 (1989) 141^144. A.M. Klibanov, Chemtech 16 (1986) 354^359. E. Battistel, D. Bianchi, J. Phys. Chem. 98 (1994) 5368^5375. D.B. Volkin, A. Staubli, A. Langer, A.M. Klibanov, Biotechnol. Bioeng. 37 (1991) 843^853. [5] K. Dabulis, A.M. Klibanov, Biotechnol. Bioeng. 39 (1992) 176^185. [6] C.J. Slade, J. Mol. Catal. B Enzym. 9 (2000) 97^105.
[7] O. Almarsson, A.M. Klibanov, Biotechnol. Bioeng. 49 (1996) 87^92. [8] T.C. de Sampaio, R.B. Melo, T.F. Moura, S. Michel, S. Barreiros, Biotechnol. Bioeng. 50 (1996) 257^264. [9] P.A. Fitzpatrick, A.M. Klibanov, J. Am. Chem. Soc. 113 (1991) 3166^3171. [10] A. Ducret, M. Trani, R. Lortie, Enzyme Microb. Technol. 22 (1998) 212^216. [11] M.D. Borisover, D.V. Zakharychev, B.N. Solomonov, J. Therm. Anal. Calorimetry 55 (1999) 85^92. [12] D.V. Zakharychev, M.D. Borisover, B.N. Solomonov, Russ. J. Phys. Chem. (Engl. Transl.) 69 (1995) 162^166. [13] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, J. Phys. Org. Chem. 6 (1993) 251^253. [14] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, Thermochim. Acta 284 (1996) 263^277. [15] V.A. Sirotkin, M.D. Borisover, B.N. Solomonov, Biophys. Chem. 69 (1997) 239^248. [16] M.D. Borisover, V.A. Sirotkin, D.V. Zakharychev, B.N. Solomonov, in: Abstracts, 15th European Chemistry at Interfaces Conference, Jerusalem, 1998, p. 64. [17] K. Griebenow, A.M. Klibanov, Biotechnol. Bioeng. 53 (1997) 351^362. [18] A. Dong, J.D. Meyer, B.S. Kendrick, M.C. Manning, J.F. Carpenter, Arch. Biochem. Biophys. 334 (1996) 406^414. [19] A. Zaks, A.M. Klibanov, J. Biol. Chem. 263 (1988) 8017^ 8021. [20] D.D. Perrin, W.L.F. Armarego, D.R. Perrin, Puri¢cation of Laboratory Chemicals, Pergamon Press, Oxford, 1980. [21] R.C. Weast (Ed.), Handbook of Chemistry and Physics, 58th edn., CRC Press, Cleveland, OH, 1977^1978. [22] A. Dong, P. Huang, W.S. Caughey, Biochemistry 29 (1990) 3303^3308. [23] K. Gribenow, A.M. Klibanov, Proc. Natl. Acad. Sci. USA 92 (1995) 10969^10976. [24] H.R. Constantino, K. Griebenow, P. Mishra, R. Langer, A.M. Klibanov, Biochim. Biophys. Acta 1253 (1995) 69^74. [25] V.A. Medvedev, M.E. E¢mov, Russian J. Phys. Chem. (Engl. Transl.) 49 (1975) 780^783. [26] S.J. Singer, Adv. Protein Chem. 17 (1962) 1^68. [27] J.T. Chin, S.L. Wheeler, A.M. Klibanov, Biotechnol. Bioeng. 44 (1994) 140^145. [28] P. Douzou, Cryobiochemistry, Academic Press, New York, 1977. [29] R.V. Rariy, A.M. Klibanov, Biotechnol. Bioeng. 62 (1999) 704^710. [30] J.M. Prausnitz, Molecular Thermodynamics of Fluid-phase Equilibria, Prentice-Hall, Englewood Cli¡s, NJ, 1969. [31] S. D Christian, A.A. Taha, B.W. Gash, Q. Rev. Chem. Soc. 24 (1970) 20^36. [32] J.A. Riddick, E.E. Toops, Organic Solvents. Physical Properties and Methods of Puri¢cation, 2nd edn., Interscience Publishers, New York, 1955. [33] H.K. Hansen, P. Rasmussen, A. Fredenslund, M. Shiller, J. Gmehling, Ind. Eng. Chem. Res. 30 (1991) 2352^2355. [34] V.B. Kogan, V.M. Fridman, V.V. Kafarov, Ravnovesie
BBAPRO 36434 1-6-01
V.A. Sirotkin et al. / Biochimica et Biophysica Acta 1547 (2001) 359^369
[35]
[36] [37] [38]
Mezhdu Zhidkost'u i Parom (Liquid^Vapour Equilibria) (in Russian), V.1, Nauka, Moscow, 1966. G.S. Lyudmirskaya, T.A. Barsukova, A.M. Bogomolnij, Ravnovesia Zhidkost^Par. (Liquid^Vapour Equilibria) (in Russian), Khimia, Moscow, 1987. S.Y. Lam, R.L. Benoit, Can. J. Chem. 52 (1974) 718^722. H.A. Laitinen, W.E. Harris, Chemical Analysis, 2nd edn., McGraw-Hill, New York, 1975. P.J. Halling, Biochim. Biophys. Acta 1040 (1990) 225^228.
369
[39] C. Raichardt, Solvents and Solvent E¡ects in Organic Chemistry, 2nd edn., VCH, Weinheim, 1988. [40] C. Laane, S. Boeren, K. Vos, C. Veeger, Biotechnol. Bioeng. 30 (1987) 81^87. [41] R.B. Gregory, Protein hydration and glass transition behavior, in: R.B. Gregory (Ed.), Protein^solvent Interaction, vol. 264, Marcel Dekker, New York, 1995, pp. 191^264. [42] M.D. Borisover, V.A. Sirotkin, B.N. Solomonov, Thermochim. Acta 256 (1995) 175^183.
BBAPRO 36434 1-6-01