Biochimica et Biophysica Acta 1804 (2010) 18501856

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / b b a p a p

Stability curves of laboratory evolved thermostable mutants of a Bacillus subtilis lipase

Md. Zahid Kamal, Shoeb Ahmad, Poornima Yedavalli, Nalam Madhusudhana Rao
Centre for Cellular and Molecular Biology, Council for Scientic and Industrial research, Uppal Road, Hyderabad-500007, India

a r t i c l e

i n f o

a b s t r a c t
Shape of the protein stability curves changes to achieve higher melting temperature. Broadly, these changes have been classied as upward shift (increased Gs), rightward shift (increase in Ts) and attening of the stability curves (decrease in Cp). Comparative studies on homologous mesophilicthermophilic protein pairs highlighted the differential contribution of these three strategies amongst proteins. But unambiguous way of identication of the strategies, which will be preferred for a protein, is still not achieved. We have performed comparative thermodynamic studies using differential scanning calorimeter (DSC) on thermostable variants of a lipase from Bacillus subtilis. These variants are products of 1, 2, 3 and 4 rounds of directed evolution and harbor mutations having denite contribution in thermostability unlike natural thermophilic proteins. We have shown that upward and rightward shift in stability curves are prime strategies in this lipase. Our results along with that from the other study on laboratory evolved xylanase A suggest that optimization of suboptimal thermodynamic parameters is having a dominant inuence in selection of thermodynamic strategies for higher thermostability. 2010 Elsevier B.V. All rights reserved.

Article history: Received 10 February 2010 Received in revised form 14 June 2010 Accepted 17 June 2010 Available online 25 June 2010 Keywords: Lipase Thermostability In vitro evolution Thermodynamics Calorimetry Free energy

1. Introduction An unambiguous understanding of protein thermostability demands the insights of sequence structure thermodynamics stability interrelationship. Changes in protein sequence to attain higher thermostability are associated with improvisation of interactions on the native protein structure and modulation of thermodynamic parameters. Although there are many ways to increase protein thermostability, none of them is universal. It has been clearly established that a strategy that is useful in one set of homologous proteins may not be applicable to another set. This has been observed at the level of sequence, structure [1] as well as thermodynamics [2]. Different proteins use different sets of strategies for thermostabilization at all these levels. Success of rational approaches for any protein can be signicantly improved if we would have clues of its requirements. Comparative studies on natural homologous proteins have provided us glimpse of such requirements, but information provided by them are severely compromised because the sequence variation between homologous thermophilicmesophilic proteins can be very high (70%) and many of these variations in sequence are unrelated to thermostability. To overcome limitation imposed by natural proteins without losing advantage of natural evolution (accepting variation in
Corresponding author. Deputy Director, Room: S106, Centre for Cellular and Molecular Biology, Uppal Road, Hyderabad-500007, India. Tel.: + 91 40 27192514; fax: + 91 40 27160591. E-mail address: madhu@ccmb.res.in (N.M. Rao). 1570-9639/$ see front matter 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2010.06.014

sequences which are highly suitable to bear selection pressure), directed evolution is a wonderful technique [37]. It is a laboratory version of natural evolution in which library of protein variants are generated followed by screening/selection of variants showing best adaption against applied selection pressure. Variation in protein sequence is controlled by experimentalist. In the present work, we used multiple thermostable variants of a Bacillus subtilis lipase LipA generated by directed evolution technique to uncover the thermodynamic strategies well suited for this kind of protein structure. Thermodynamic studies of proteins rely on the determination of various thermodynamic quantities such as changes in free energy, enthalpy and entropy during an unfolding transition. Dependence of thermodynamic stability on temperature follows a parabolic curve called stability curve[8] as shown in Fig. 1. It achieves a maximum value (Gs) at a certain temperature Ts and takes zero value at melting temperature (Tm). Other thermodynamic parameters associated with protein unfolding such as enthalpy change (H), entropy change (S) and heat capacity change (Cp) are inherently linked to the stability curve and so they can be derived in temperature dependent fashion from stability curves. A protein can acquire higher melting temperature in one or more of the following three ways i.e., Shifting the entire stability curve upward (by an increase in Gs) or by attening (or decreasing the curvature) of the stability curve (by decreasing Cp) or by shifting the entire stability curve to the right ( by increasing Ts) as shown in Fig. 1. Comparative studies on families of proteins from mesophilic and thermophilic species highlighted the contributions of these strategies towards gaining higher thermostability [2,9,10]. These studies have clearly indicated that different sets

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856

1851

0.1 mg/ml protein in 50 mM buffer of appropriate pH were recorded in 0.1 cm pathlength cuvette using a JASCO J-815 spectropolarimeter between wavelengths from 250 nm to 200 nm. All spectra recorded are an average of 4 accumulations. Wavelength scans were carried out in the ellipticity mode at a scan speed of 50 nm/min, bandpass of 2 nm, at response time of 2 s and wavelength step of 0.5 nm. All spectra were corrected for buffer base line by subtracting the respective blank spectra recorded identically without the protein. 2.4. Differential scanning calorimetry (DSC) DSC measurements were performed on 0.5 mg/ml of proteins in Microcal VPDSC instrument between pH 3.4 to 4.6 at the difference of 0.2 pH unit. 50 mM sodium acetate buffers were used for pH maintenance during thermal denaturation experiments. Before DSC experiments all the samples were degassed extensively. All the DSC scans were performed in-between 10 and 90 C temperature range, at a scan rate 60 C/h, with prescan thermostat of 10 min and a lter time of 16 s unless otherwise mentioned. All the DSC scans were highly reproducible. Prior to analysis of DSC data, buffer/buffer base lines were determined for each experiment and subtracted from protein/buffer scans, followed by data normalization for protein concentration. 2.5. Thermodynamic analysis of DSC data For analysis of DSC scans, thermograms were tted with independent two-state transitions including Cp effects with single transition model present in analysis software provided by manufacturer. In this model calorimetric enthalpy H of each transition is equal to its van't Hoff enthalpy H* and heat capacity change between folded and unfolded state Cp is taken as constant (independent of temperature). Fitting used the equation KTCp KTHT2 + Cp T = B0 + B1 T + 1 + KT 1 + KT2 RT 2 " #

Fig. 1. Stability curves showing various ways to achieve higher melting temperature (Tm). Stability curve of a typical mesophilic protein () changes its shape to achieve higher Tm by increasing the Gs, by increasing Ts and by decreasing Cp resulting in upward shift (), rightward shift () and attening () of curve respectively. Gs is the maximum thermodynamic stability of protein at temperature Ts. Subscript s denotes the point where entropy of unfolding is zero.

of homologous proteins use different mixes of the above mentioned three thermodynamic strategies [8]. However, such studies could neither unambiguously establish the link between sequence changes and strategy nor which of the strategies will be preferred by a protein. Here, we attempted to overcome these limitation using a Bacillus subtilis lipase mutants created by directed evolution techniques to acquire a clearer picture of which thermodynamic strategies is preferred by this protein and why. 2. Materials and methods 2.1. Protein purication and concentration determination All lipase variants were over expressed and puried as described earlier [11,12]. Protein concentrations were estimated by method of Lowry et al. 2.2. Generation of lipase mutants by directed evolution: Thermostable mutants were identied by the methods described earlier [12]. Briey, lipase gene along with the signal sequence cloned in pMK3 Escherichia coliB. subtilis shuttle vector was mutagenized by error-prone PCR using GeneMorph RandomMutagenesis kit from Stratagene according to the manufacturer's instructions. A mutation condition that should generate an error frequency of approximately four to ve mutations per 1000 bases was used. The mutagenic library thus obtained was used to transform E. coli DH5 by electroporation. The thermostability of mutants was assessed by following a three-tier screening protocol. Initially, coarse screening was performed on petri plates followed by a stringent screening in 96-well plate format. Positives were further conrmed by residual activity measurements in tube assays. The base changes were derived from the sequencing the gene insert and mutations were combined by standard molecular biology methods. 2.3. Far UV circular dichorism spectroscopy Secondary structures of proteins were determined in the pH range 3.44.6 and at pH 7.2 using far UV circular dichorism spectroscopy. Sodium acetate buffer was used for pH 3.44.6, while sodium phosphate buffer was used for pH 7.2. pH of all buffers was measured by Orion potentiometer calibrated within error range not more than 0.1 pH unit with standard buffers of pH 4.0 and 7.0. CD spectra of

where Cp(T) is the experimental heat capacity at temperature T. (Bo + B1T) is the linear dependence of heat capacity of the native state while, K(T) and H(T) are equilibrium unfolding constant and enthalpy change. In the equation, equilibrium constant and change in enthalpy can be expressed in terms of their temperature-independent values at melting temperature Tm and the heat capacity change for transition Cp, i.e., HT = Hm + Cp TTm KT = exp      Cp Hm T T 1 TTm T ln Tm Tm RT RT

3. Results Details of the directed evolution methodologies had been reported earlier [1113]. Briey, four rounds of directed evolution were performed on B. subtilis lipase LipA with higher thermostability as the screening criterion. Three best mutations identied after each round of directed evolution have been recombined into single mutant, which parented the next round of directed evolution. We have created lipase variants named 4B1, 2D9, 4D3 and 6B (harboring 3, 6, 9 and 12 thermostabilizing mutations, respectively) after 1, 2, 3 and 4 rounds of directed evolution and recombination. Details of mutations in the variants are given in Table 1. Contribution of each mutation to the protein thermostability has been afrmed by performing detailed characterization of puried lipases with one mutation at time [1113]. Sequence and structural similarity of these variants are very high.

1852

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856

Table 1 Thermodynamic parameters of lipase (LipA) variants. Variant Mutations pH Thermodynamic parameters from tting of DSC scans Tm (C) Wt 3.4 3.6 3.8 4.0 4.2 4.4 4.6 3.4 3.6 3.8 4.0 4.2 4.4 4.6 3.4 3.6 3.8 4.0 4.2 4.4 4.6 3.4 3.6 3.8 4.0 4.2 4.4 4.6 3.4 3.6 3.8 4.0 4.2 4.4 4.6 26.68 0.18 32.63 0.23 35.33 0.14 38.90 0.12 42.09 0.08 44.86 0.05 46.68 0.08 40.85 0.10 44.14 0.06 46.86 0.05 49.45 0.05 52.10 0.03 53.85 0.08 55.40 0.11 49.19 0.09 51.77 0.06 53.96 0.10 56.13 0.07 58.26 0.04 60.24 0.03 61.99 0.04 53.17 0.16 55.68 0.06 57.83 0.05 60.05 0.05 62.23 0.04 64.17 0.06 65.81 0.06 60.50 0.13 62.79 0.02 65.17 0.03 67.47 0.04 69.69 0.02 71.62 0.02 73.26 0.04 Hma (kcal mol1) 43.58 0.83 62.19 0.93 66.54 0.77 74.88 0.67 83.81 0.45 91.70 0.37 94.19 0.54 80.45 0.50 90.33 0.36 95.55 0.38 100.7 0.42 112.1 0.30 119.8 0.96 122.9 1.17 96.47 0.67 105.8 0.52 112.2 0.61 116.3 0.54 125.0 0.42 131.9 0.34 135.0 0.37 103.5 0.64 116.9 0.38 120.1 0.41 123.3 0.43 131.0 0.36 138.2 0.69 140.5 0.73 121.9 0.81 137.7 0.38 143.4 0.47 143.0 0.58 154.2 0.39 158.0 0.37 155.9 0.56 Cpa (kcal mol1 C1) 3.98 0.11 2.33 0.27 3.53 0.18 3.19 0.21 2.69 0.17 2.33 0.14 3.36 0.23 2.29 0.21 2.01 0.15 2.04 0.15 2.22 0.15 0.77 0.11 1.58 0.34 1.13 0.53 1.36 0.26 0.80 020 1.07 0.46 1.45 0.32 1.23 0.26 0.09 0.23 0.01 0.31 0.08 0.60 0.04 0.31 0.45 0.29 0.56 0.28 0.97 0.28 4.91 0.45 5.61 0.52 0.03 0.72 1.04 0.11 0.96 0.27 1.02 0.30 1.09 0.19 0.88 0.20 1.02 0.34 Thermodynamic parameters calculated using DSC data Cpb (kcal mol1 C1) 2.33 Hmc (kcal mol1) 56.96 64.77 71.75 78.84 85.34 91.58 96.20 79.27 88.96 96.96 104.59 112.39 117.55 122.11 97.07 104.90 111.54 118.12 124.57 130.58 135.88 106.19 113.18 119.16 125.34 131.41 136.81 141.37 127.64 133.51 139.62 145.52 151.22 156.17 160.38 Ts (C) 7.0 7.3 7.5 7.8 8.1 8.4 8.6 15.0 15.3 15.5 15.8 16.1 16.3 16.4 18.7 18.9 19.2 19.4 19.6 19.8 20.0 17.2 17.4 17.8 17.9 18.2 18.4 18.6 14.3 14.6 14.9 15.2 15.5 15.8 16.1 Gs (kcal mol1) 2.2 2.8 3.4 4.1 4.7 5.4 5.9 3.3 4.1 4.8 5.5 6.3 6.9 7.4 4.7 5.4 6.0 6.7 7.4 8.1 8.7 6.0 6.7 7.4 8.1 8.8 9.5 10.1 9.1 9.8 10.6 11.5 12.3 13.0 13.6

4B1

L114P, A132D, N166Y

2.94

2D9

4B1 + F17S, N89Y, I157M

3.03

4D3

2D9 + A15S, A20E, G111D

2.78

6B

4D3 + M134E, M137P, S163P

2.56

a b c

Derived from a separate analysis of the individual DSC thermogram. Calculated from Kirchoff's equation, using values of Hm and Tm provided by ttings of DSC scans. Corrected Hm, estimated from linear dependence of Hm on Tm (Fig. 5).

Most thermostable mutant, 6B (having 22 C higher melting temperature than wild type), is N93% similar in sequence to wild type protein, while mutants and wild type crystal structure overlaps very well with an rmsd of b0.391 over the C-atoms. 3.1. Differential scanning calorimetry (DSC) of lipase variants Thermodynamic analysis requires unfolding of proteins that is completely reversible. Thermal denaturation of LipA at physiological pH 7.2 is highly irreversible due to aggregation of protein while heating. Our attempts to prevent aggregation to increase reversibility by adding various chemicals like urea, guanidinium chloride and polyols failed. Earlier report [14] from the lab has suggested that LipA demonstrates a remarkable property of complete recovery of structure and function even after heating at 90 C below pH 5.0. The CD spectrum of the wild type protein at pH 7.2 was identical to the CD spectrum of wild type protein that was heated to 50 C at pH 4.5 and cooled to room temperature. Hence, we used low pH (3.44.6) buffer solutions for thermal unfolding. In the absence of heating, the secondary structures determined by far UV circular dichorism spectroscopy of the variants at these pH values were identical to that at pH 7.2 (Fig. 2). DSC scans for all the ve lipase variants within this pH range show single transition as inferred from the single peak in each scan (Fig. 4). Reversibility of proteins is estimated from the areas under successive DSC scans. Area under the DSC scan gives calorimetric enthalpy of unfolding, which decreases in successive

scans of the sample in case of irreversibility. Good reversibility (N70%) was observed for all the variants at these pH values (Fig. 3). Lowest reversibility (50%) was observed for wild type lipase at pH 4.6. Reversibility was also judged by DSC scans of protein sample taken at different scan speeds (3090 C/h), which overlap very well (data not shown) suggesting nearly complete reversibility. Together these

Fig. 2. Far UV circular dichorism spectra of wild type LipA at pH 7.2 (solid line) and pH 3.4 (broken line) at 10 C.

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856

1853

Fig. 3. Reversibility of DSC scans. Reversibility of wild type and 6B lipases, monitored by successive DSC scans at (a) pH .3.4 and (b) pH 4.6. Solid lines () represent the rst scans while dotted lines () represent the second scans.

observations suggest the existence of thermodynamic equilibrium throughout thermal denaturation of lipases. Fig. 4 shows the buffer subtracted DSC scans of lipase variants obtained at 60 C/h scan speed that were normalized for protein concentration. All the DSC scans were tted with a model describing independent two-state transitions including Cp effects with single transition. In this model calorimetric enthalpy H of each transition is equal to its van't Hoff enthalpy H* and heat capacity change between folded and unfolded state Cp is taken as constant (independent of temperature). The 2 values, which is the measure of goodness of tting, obtained with this model were in the range of 104105. Thermodynamic parameters derived from theses ttings are listed in Table 1. 3.2. Determination of thermodynamic parameter and simulation of stability curve Stability curves of a protein can be simulated using Gibbs Helmholtz equation Gu(T) = Hm(1 T/Tm) Cp [Tm T + T ln (T/Tm)]. Gu(T) is the thermodynamic stability or free energy of unfolding at temperature T, Hm is the enthalpy of unfolding at melting temperature Tm and Cp is the change in heat capacity upon unfolding. As obvious from the equation, simulation of stability curve requires the measurements of Tm, Hm and Cp. Out of the three parameters, rst two can be readily determined from thermal unfolding curves (spectroscopically or calorimetrically) with an uncertainty of about 0.5 C for Tm and 5 kcal/mol for Hm [15]. Determination of a reliable value of Cp is more difcult [1620]. Our tting model of DSC scans estimated Cp associated with each scan.

Fig. 4. DSC scans between pH 3.4 to 4.6 for (a) wild type, (b) 4B1, (c) 2D9, (d) 4D3 and (e) 6B. Solid lines () represent experimental thermograms while dots () represent ttings by model, independent two-state transitions including Cp effects with single transition present in analysis software provided by manufacturer. Experimental pH has been indicated above each scan.

1854

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856

However the values obtained have shown variability between different pHs for a given protein. This inconsistency is possibly the outcome of partial irreversible processes associated with thermal denaturation, causing distortion in post transition baseline of DSC scans and/or tted errors associated with ve different tting variables. We estimated the Cp using modied Kirchoff's equation,   dHm dTm

Cp =

This method has been widely used to obtain a very good estimate of Cp [1820]. For this method, variation in Tm and corresponding Hm was generated using varying concentration of chemical denaturant (e.g., urea and guanidinium chloride) or pH. Slope of the linear plot between Hm and Tm gives the value of Cp. Widespread use of this method is because of the fact that Hm does not depend upon change in these chemical conditions but on Tm and Cp, which is robust to changes in these conditions as well as to temperature range 2080 C [15,2022]. We have used Tm and corresponding Hm obtained calorimetrically at various pH values for all LipA variants to plot Hm vs. Tm (Fig. 5). Slopes of these curves give Cp for all the variants. We have listed these Cp values obtained by this method in Table 1. Hm obtained from tting of any DSC thermogram is associated with larger error than Tm. Hence to make Hm more accurate, its value is estimated from its linear dependence on Tm (Fig. 5) and used for determination of other thermodynamic parameters. We calculated Ts and Gs at all pH values by following equations, T "m # Hm exp Tm Cp !

Fig. 6. Simulated stability curves at pH 4.6. Values given in Table 1 are used for simulation. WT (), 4B1 (), 2D9 (), 4D3 () and 6B ().

4. Discussion We can see that variation in estimated Cp amongst all the lipase variants is very less. Cp of the wild type lipase is 2.33 kcal/mol/C. This increases to 2.94 kcal/mol/C and 3.03 kcal/mol/C for 4B1 and 2D9 respectively. Additional mutations in 4D3 and 6B decrease it to 2.78 kcal/mol/C and 2.56 kcal/mol/C respectively. Considering the linear dependence of Cp on ASA (change in accessible surface area upon unfolding) [23,24], these very small variations in Cp are not unexpected as all the mutations are on protein surface, hence they do not bring any change in ASA. Decrease in Cp of a protein decreases the curvature of stability curve, leading to its attening and increase in melting temperature. This strategy is employed by many thermophilic proteins to achieve higher melting temperature (Tm) over their mesophilic homologs. A thermodynamic comparison of thermophilic ribonuclease H from Thermus thermophilus to that of E. coli [25] showed the decrease in Cp of thermophilic protein. Cp value of T. thermophilus ribonuclese H is only1.8 kcal/mol/C while that of E. coli protein is 2.7 kcal/mol/C. This large difference cannot be attributed to change in size as both the protein is nearly equal in size viz. 166 and 156 amino acid long respectively. In a different study on Sac7d protein of thermophilic origin [26], Cp of this protein when compared with that of similar sized proteins of mesophilic origin, its value is found to be lower . Razvi et al. [2] compared the stability curves for many sets of homologous thermophilic and mesophilic proteins in order to identify the thermodynamic mechanisms employed to achieve higher melting temperature. One of the strong conclusions derived from their study is that 70% of the thermophilic proteins strategically lower their Cp alone or in combination with other thermodynamic strategies to achieve higher Tm. We have theoretically determined the value of Cp for LipA using a program BPPred [27], which can predict the Cp of protein based on predicted ASA, where the number of amino acid as the only input. The theoretical value for this lipase is 2.77 kcal/mol/C. This theoretical value is close to the experimentally determined values of all the lipase variants. In fact, Cp measured by taking all the lipase variants simultaneously in consideration is 2.54 kcal/mol/C, which is also very close to the predicted value. Clearly attening of stability curve by decreasing Cp is not amongst the preferred strategies taken by this lipase for thermostabilization. Comparative thermodynamic characterization of performed on LipA variants clearly shows increase in Ts of the thermostable mutants compared to wild type. Thermostabilization by increasing Ts has been earlier observed in only few of the cases. Thermophilic histone PyA1 is

Ts =

Gs = Cp Ts Tm +

Hm Cp

All these thermodynamic parameters are listed in Table 1. A simulated stability curve for all lipase variants at pH 4.6 is shown in Fig. 6 as representative.

Fig. 5. Calculation of Cp. Slope of Hm at various pH vs. corresponding Tm, from Table1, gives the Cp of all the lipA variants. Straight lines represent the linear ts. WT (), 4B1 (), 2D9 (), 4D3 () and 6B ().

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856

1855

having higher Ts [9] than the same sized mesophilic homolog MfoB. Similarly, Ts of farsenyl diphosphate/geranylgeranyl diphosphate synthase from thermophilic organism Thermococcus kodakiensis is higher than that of its mesophilic homolog [28]. Data base created by Razvi et al. [2] suggested that this thermostabilization strategy is the least used by thermophilic proteins. The low occurrence of this strategy has been attributed to the demand imposed by this strategy i.e., reduction in entropy of unfolding. To meet this requirement, either the entropy of unfolded state of protein should decrease so that it will be closer to that of folded state. Alternatively, entropy of folded state needs to be increased to shorten the gap to the unfolded state. Proteins can achieve the former by mutations to proline, which can constrain the unfolded state. Otherwise, mutations to glycine in the structured region of folded state can make it more exible. In the present study, Ts of LipA mutants increase in the rst and second round of directed evolution at all pH values. For example, at pH4.6 it increases from 8.6 C for wild type lipase to 16.4 C and 20.0 C for 4B1 and 2D9 respectively. Additional two cycles of directed evolution although increases number of mutations in 4D3 and 6B, but does not cause increase in Ts, rather it decreases (to 18.6 C in 4D3 and further to 16.1 C in 6B at pH 4.6). This observation suggest that increasing Ts may be an efcient strategy for acquiring higher thermostability (Tm) for this lipase up to a point after which it may not be as efcient. It should be noticed that over the span of four cycles of directed evolution, LipA has accumulated no glycine but three proline mutations, all are on loop and solvent exposed regions of protein. First proline mutation is L114P, which occurred in rst round of directed evolution. 4B1 mutant, which is the product of rst cycle of directed evolution, harbor this mutation along with other two (A132D and N166Y), shows increase in Ts by 8 C. Two other proline mutations are product of fourth round of directed evolution. These two mutations present in 6B could not increase the Ts, as its value in 6B decreased by 2.5 C compared to parent lipase molecule 4D3. It may also be noticed that although there are no proline mutations in 2nd and 3rd round of directed evolution, yet Ts has increased in 2nd by 3 C and decreased in 3rd by 1.5 C. This trend of Ts change can be better understood in the context of other signicant behavior shown by natural proteins. It has been found that Ts of most of the naturally occurring proteins are around room temperature (20 8 C) irrespective of their habitat temperature (e.g. mesophilic or thermophilic) [29]. It has been argued that this behavior of natural proteins is the outcome of the fact that hydrophobic force plays major role in protein folding and stability and it should be strongest around room temperature because of the least solubility of hydrophobic amino acids around this temperature. Ts (that represents the temperature at which thermodynamic stability is maximum) is low for wild type protein (7.08.6 C at various pH). It shows a jump of 8.0 C in 4B1 and further increase of 2.0 C in 2D9 after rst two rounds of evolution with accumulation of three and six stabilizing mutations respectively. In 2D9 Ts is around room temperature (20 C) which is in accord with aforementioned nding for high protein thermostability. Next two rounds of evolution although decreases Ts by few degrees, Ts of the mutants are still in room temperature zone. It appears that optimization of Ts to be in room temperature zone is probably the primary driving force in early rounds of improvements. It can also be argued that the very low occurrence of natural thermophiles protein using increase in Ts as the thermostabilizing strategy [2] is because of the simple fact that most of the even natural mesophilic proteins have optimized Ts [29]. Except mutation I157M which is product of 3rd round of directed evolution, in our study with lipase, we have not found any other mutation whose role can be directly implicated in increasing hydrophobicity of protein as all the mutations are on

the surface. Mutation I157M also occurred on protein surface, but it shows improved van der Waal's interaction with core residue [12]. However, it should also be noticed that I157M is present in 4D3, whose Ts value is lower than evolutionarily preceding parent molecule 2D9. It seems that mutations accumulated in this lipase, have indirectly improved the hydrophobic interaction in native protein structure, possibly by rigidication of native structure resulting in improve core packing. Gs increment is the most consistent feature during the evolution of this lipase for thermostability. As can be seen in Table 1, this strategy has constantly been used by LipA during directed evolution for higher thermostability. For example, at pH 4.6 Gs increased from 5.9 kcal/mol for wild type lipase to 7.4, 8.7, 10.1 and 13.6 kcal/mol for 4B1, 2D9, 4D3 and 6B respectively. Increasing Gs to achieve higher melting temperature is the most commonly exercised strategy by thermophiles. This strategy has been used in 77% of the thermophilic proteins either alone or in combination with other strategy(s) [2]. The very high occurrence of this particular strategy can readily be attributed to its easy feasibility. It should be noted that at Ts, entropy of unfolding is zero hence Gs (thermodynamic stability at Ts) has contribution only from enthalpy of unfolding. Hence, any approach to increase enthalpic contribution in free energy can increase Gs. Stability contribution of most of the interactions like hydrogen bonds, salt-bridges are of enthalpic origin. Stabilization of native structure in LipA with accumulating mutations has been structurally attributed to increase in number of stabilizing interactions, particularly hydrogen bondings, salt bridges and pi-stacking [11,12]. The role of the three mutations in case of 6B also were in stabilization of the loops either by water bridges or by decreasing the exibility of the region (unpublished data). Evidently, these interactions in native LipA structure add to the enthalpic contribution to the thermodynamic stability resulting in increase in Gs. Our study shows that when B. subtilis lipase was evolved specically for thermostability, both Ts and Gs are enhanced, while decrease in change in heat capacity upon unfolding (Cp) has not been chosen by this lipase as an efcient strategy. A closer look at values of these parameters in wild type protein suggests that Cp is near optimum, while Ts is away from optimum. Also, Gs of wild type protein is lower than achievable for thermophilic proteins (25 30 kcal/mol) [2,30,31]. At this point, it is interesting to argue that a property would be specically subjected to change if that property is suboptimal to the global average. This argument is well supported by the only other study where thermodynamic parameters were compared among xylanase A variants evolved by directed evolution approach [32]. It is interesting to notice that Cp of wild type xylanase A is 6.7 kcal/mol, which is very high than expected from theoretical calculation. Xylanase A is a 184 amino acid long protein and its Cp predicted from BPPred [27] is 2.81 kcal/mol. Hence, thermostabilizing mutations incorporated during three rounds of directed evolution of this protein are ingenious picks to decrease Cp. However, it is noticeable that the most thermostable mutant of xylanase A (product of three rounds of directed evolution) has Cp equals to 0.1 kcal/mol/C, which is far below than its theoretical Cp. The very low Cp of this mutant is not unexpected since the methods of evolution of xylanase A involved rst two rounds of evolution comprised of random mutagenesis by error prone PCR, whereas the third round involves gene shufing of the thermostable mutants created during previous two rounds. A direct consequence of the strategies taken during evolution of xylanase is that during the rst two rounds of evolution, xylanase A was free to select mutations to better suit its requirement to achieve higher thermostability, third round of evolution constrained the protein to select a new combination of mutations from the pool of mutations already been selected during the rst two rounds of evolution. In contrast, in case of LipA evolution, new sequence

1856

Md.Z. Kamal et al. / Biochimica et Biophysica Acta 1804 (2010) 18501856 [10] A. Razvi, J.M. Scholtz, A thermodynamic comparison of HPr proteins from extremophilic organisms, Biochemistry 45 (2006) 4084. [11] P. Acharya, E. Rajakumara, R. Sankaranarayanan, N.M. Rao, Structural basis of selection and thermostability of laboratory evolved Bacillus subtilis lipase, J. Mol. Biol. 341 (2004) 1271. [12] S. Ahmad, M.Z. Kamal, R. Sankaranarayanan, N.M. Rao, Thermostable Bacillus subtilis lipases: in vitro evolution and structural insight, J. Mol. Biol. 381 (2008) 324. [13] S. Ahmad, N.M. Rao, Thermally denatured state determines refolding in lipase: mutational analysis, Protein Sci. 18 (2009) 1183. [14] E. Rajakumara, P. Acharya, S. Ahmad, R. Sankaranaryanan, N.M. Rao, Structural basis for the remarkable stability of Bacillus subtilis lipase (Lip A) at low pH, Biochim. Biophys. Acta 1784 (2008) 302. [15] C.N. Pace, D.V. Laurents, A new method for determining the heat capacity change for protein folding, Biochemistry 28 (1989) 2520. [16] S.S. Alexander, C.N. Pace, A comparison of the denaturation of bovinelactoglobulins A and B and goat-lactoglobulin, Biochemistry 10 (1971) 2738. [17] H. Nojima, A. Ikai, T. Oshima, H. Noda, Reversible thermal unfolding of thermostable phosphoglycerate kinase. Thermostability associated with mean zero enthalpy change, J. Mol. Biol. 116 (1977) 429. [18] D. Shortle, A.K. Meeker, E. Freire, Stability mutants of staphylococcal nuclease: large compensating enthalpyentropy changes for the reversible denaturation reaction, Biochemistry 27 (1988) 4761. [19] W.J. Becktel, W.A. Baase, Thermal denaturation of bacteriophage T4 lysozyme at neutral pH, Biopolymers 26 (1987) 619. [20] P.L. Privalov, Stability of proteins: small globular proteins, Adv. Protein Chem. 33 (1979) 167. [21] L. Swint, A.D. Robertson, Thermodynamics of unfolding for turkey ovomucoid third domain: thermal and chemical denaturation, Protein Sci. 2 (1993) 2037. [22] Y.V. Griko, P.L. Privalov, Calorimetric study of the heat and cold denaturation of beta-lactoglobulin, Biochemistry 31 (1992) 8810. [23] J.R. Livingstone, R.S. Spolar, M.T. Record Jr., Contribution to the thermodynamics of protein folding from the reduction in water-accessible nonpolar surface area, Biochemistry 30 (1991) 4237. [24] R.S. Spolar, J.R. Livingstone, M.T. Record Jr., Use of liquid hydrocarbon and amide transfer data to estimate contributions to thermodynamic functions of protein folding from the removal of nonpolar and polar surface from water, Biochemistry 31 (1992) 3947. [25] J. Hollien, S. Marqusee, A thermodynamic comparison of mesophilic and thermophilic ribonucleases H, Biochemistry 38 (1999) 3831. [26] B.S. McCrary, S.P. Edmondson, J.W. Shriver, Hyperthermophile protein folding thermodynamics: differential scanning calorimetry and chemical denaturation of Sac7d, J. Mol. Biol. 264 (1996) 784. [27] C.D. Geierhaas, A.A. Nickson, K. Lindorff-Larsen, J. Clarke, M. Vendruscolo, BPPred: a Web-based computational tool for predicting biophysical parameters of proteins, Protein Sci. 16 (2007) 125. [28] S. Fujiwara, A. Yamanaka, K. Hirooka, A. Kobayashi, T. Imanaka, E. Fukusaki, Temperature-dependent modulation of farnesyl diphosphate/geranylgeranyl diphosphate synthase from hyperthermophilic archaea, Biochem. Biophys. Res. Commun. 325 (2004) 1066. [29] S. Kumar, C.J. Tsai, R. Nussinov, Maximal stabilities of reversible two-state proteins, Biochemistry 41 (2002) 5359. [30] S. D'Amico, J.C. Marx, C. Gerday, G. Feller, Activitystability relationships in extremophilic enzymes, J. Biol. Chem. 278 (2003) 7891. [31] C. Vieille, G.J. Zeikus, Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for thermostability, Microbiol. Mol. Biol. Rev. 65 (2001) 1. [32] R. Ruller, L. Deliberto, T.L. Ferreira, R.J. Ward, Thermostable variants of the recombinant xylanase A from Bacillus subtilis produced by directed evolution show reduced heat capacity changes, Proteins 70 (2008) 1280.

space was sampled in each round of evolution resulting in pronounced improvement in thermodynamic parameters to achieve higher Tm. Though all the mutations in xylanase and lipase were located on the surface, interestingly they have different effects on the thermodynamic properties of these proteins. Mutations not only affect the native state but also the non-native states of the protein, whose properties determine the thermodynamic properties of the protein. To obtain complete insight into the inuences of a mutation on thermodynamic parameters, inuence of a mutation on non-native states also should be investigated. 5. Conclusion In conclusion, our study on thermostable lipases in combination with studies on xylanase A [32] created by directed evolution techniques, emphasizes the very fact that different proteins show variations in their preferences towards the thermodynamic parameters (and associated strategies) and optimization of suboptimal parameters probably guides the evolution for higher thermostability in proteins. Acknowledgments MZK, SA and PY thank the Council of Scientic and Industrial Research for the senior research fellowships. We also acknowledge the help given by Dr. Rajan Sankaranarayanan in determining the structure of thermostable lipases. This work has been supported by grant zero emission research initiative (NWP0044). References
[1] G.A. Petsko, Structural basis of thermostability in hyperthermophilic proteins, or there's more than one way to skin a cat, Methods Enzymol. 334 (2001) 469. [2] A. Razvi, J.M. Scholtz, Lessons in stability from thermophilic proteins, Protein Sci. 15 (2006) 1569. [3] F.H. Arnold, P.L. Wintrode, K. Miyazaki, A. Gershenson, How enzymes adapt: lessons from directed evolution, Trends Biochem. Sci. 26 (2001) 100. [4] S. Bershtein, D.S. Tawk, Advances in laboratory evolution of enzymes, Curr. Opin. Chem. Biol 12 (2008) 151. [5] C. Jackel, P. Kast, D. Hilvert, Protein design by directed evolution, Annu. Rev Biophys. 37 (2008) 153. [6] T. Matsuura, T. Yomo, In vitro evolution of proteins, J. Biosci. Bioeng. 101 (2006) 449. [7] S.G. Peisajovich, D.S. Tawk, Protein engineers turned evolutionists, Nat. Methods 4 (2007) 991. [8] W.J. Becktel, J.A. Schellman, Protein stability curves, Biopolymers 26 (1987) 1859. [9] W.T. Li, R.A. Grayling, K. Sandman, S. Edmondson, J.W. Shriver, J.N. Reeve, Thermodynamic stability of archaeal histones, Biochemistry 37 (1998) 10563.