PEDS Advance Access originally published online on August 10, 2005
Protein Engineering Design and Selection 2005 18(9):425-433; doi:10.1093/protein/gzi051
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
How to improve nature: study of the electrostatic properties of the surface of 
-lactalbumin
1Institute for Biological Instrumentation, Russian Academy of Sciences, Pushchino, Moscow Region, Russia, 2Department of Biochemistry and Molecular Biology, Penn State University, College of Medicine, Hershey, PA 17033, 3Department of Chemistry and Biochemistry, University of Denver, Denver, CO 80208, 4Center for Computational Biology and Bioinformatics, Department of Biochemistry and Molecular Biology, School of Medicine, Indiana University – Purdue University, Indianapolis, IN 46202 and 6Department of Veterinary Biosciences, The Ohio State University, Columbus, OH 43210, USA
5 To whom correspondence should be addressed. E-mail: vuversky{at}iupui.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
It was recently shown that
-lactalbumin interacts with histones and simple models of histone proteins such as positively charged polyamino acids, suggesting that some fundamental aspects of the protein surface electrostatics may come into play. In the present work, the energies of charge–charge interaction in apo- and Ca2+-loaded -lactalbumin were calculated using a Tanford–Kirkwood algorithm with either solvent accessibility correction or using a finite difference Poisson–Boltzmann method. The analysis revealed two major regions of -lactalbumin that possessed highly unfavorable electrostatic potentials: (a) the Ca2+-binding loop and its neighboring residues and (b) the N-terminal region of the protein. Several individual mutants were prepared to neutralize specific individual surface acidic amino acids at both the N-terminus and Ca2+-binding loop of bovine -lactalbumin. These mutants were characterized by intrinsic fluorescence, differential scanning microcalorimetry and circular dichroism. The structural and thermodynamic data agree in every case with the theoretical predictions, confirming that the N-terminal region is very sensitive to changes in charge. For example, desMet D14N mutation destabilizes protein and decreases its calcium affinity. On the other hand, desMet E1M and desMet D37N substitutions increase the thermal stability and calcium affinity. The Met E1Q is characterized by a marked increase in protein stability, whereas desMet E7Q and desMet E11L display a slight increase in calcium affinity and thermal stability. Examination of the unfavorable energy contributed by Glu1 and the energetically favorable consequences of neutralizing this residue suggests that nature may have made an error with bovine -lactalbumin from the viewpoint of stabilizing structure and conformation.
Keywords: -lactalbumin/calcium binding/electrostatic interactions/site-directed mutagenesis/thermal stability
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
-Lactalbumin (-LA) is a small (14 kDa), highly abundant calcium-binding protein from milk, which has, after decades of intensive study, become one of the best characterized proteins in protein science [for a review, see Permyakov and Berliner (2000)
]. One of the reasons for the high interest in -LA is its ability to convert under mild denaturing conditions into the equilibrium molten globule state, representing what is now considered a general intermediate in protein folding (Dolgikh et al., 1981). Recently, several reports have uncovered bactericidal and antitumor activities of -LA and its fragments. For example, Pellegrini et al. and colleagues have found that proteolytic digestion of -LA yields three peptides with bactericidal properties (Pellegrini et al., 1999). Hakansson et al. have established that a multimeric form of human -LA derivative (MAL) isolated from the casein fraction of milk is a potent apoptosis-inducing agent with broad, yet selective, cytotoxic activity (Hakansson et al., 1995). MAL was shown to kill all transformed, embryonic and lymphoid cells tested, but spared mature cells. It was shown that the apoptosis-inducing fraction contained -LA oligomers that underwent a conformational change toward a molten globule-like state (Svensson et al., 1999). MAL crosses the plasma membrane and enters the nucleus where it induces DNA fragmentation through some direct effect at the nuclear level (Hakansson et al., 1999).
Similar results were obtained with HAMLET (human -LA made lethal to tumor cells) or BAMLET (bovine -LA made lethal to tumor cells), which is native -LA converted in vitro to an apoptosis-inducing form of the protein when in a stoichiometric complex with oleic acid (Svensson et al., 1999, 2000). HAMLET was shown to trigger apoptosis in tumor and immature cells, but healthy cells were resistant. HAMLET passed through the cytoplasm to the nucleus and accumulated in the cell nucleus. In tumor cells in vivo, HAMLET co-localized with histones and perturbed the chromatin structure (Duringer et al., 2003). HAMLET was found to bind histone HIII strongly and to lesser extent histones HIV and HIIB. The binding of histones by HAMLET impaired their interaction with DNA. Based on these observations, it was concluded that HAMLET interacts with histones and chromatin in tumor cell nuclei, locking the cells into the apoptotic pathway via an irreversible disruption of chromatin organization (Duringer et al., 2003).
Recently, it was found that non-fatty acid-bound monomeric bovine and human -LAs interacted electrostatically with basic proteins, histones and positively charged polyamino acids as simple models of histone proteins (Permyakov et al., 2004). Thus, complexation of -LA with oleic acid is not required for the interaction with histone proteins. The intrinsic ability of -LA to interact strongly with charged polymers suggests that the protein surface possesses some electrostatic properties which complement this interaction.
Mutations of surface charged residues in multi-subunit proteins can increase their stability. For example, mutation of Glu165 to Gln or Lys in tetrameric malate dehydrogenase caused a dramatic increase in thermal stability at pH 7.5 (increase by about 24°C) (Bjork et al., 2004). Remarkably, the crystal structures of the two mutants showed only minor structural changes localized in close proximity to the mutated residues, indicating that the observed stability changes were caused by subtle changes in the complex network of electrostatic interactions at the dimer–dimer interface.
Minagawa et al. constructed a thermostable mutant of lactate oxidase (Minagawa et al., 2003). Their molecular modeling suggested that the substitution of Gly for Glu at position 160 reduced the electrostatic repulsion between the negative charges of Glu160 and Glu130 in the (ß/)8 barrel structure, but thermal inactivation experiments on the five different single-mutant lactate oxidases at position 160 (E160A, E160Q, E160H, E160R and E160K) showed that it was the side-chain molecular volume of the residue at position 160 that was the primary contribution to the thermostability.
Protein engineering experiments on glycosyl hydrolase showed that the thermostabilization resulted as a consequence of numerous favorable ionic interactions in the 83–124 sequence with the other parts of protein matrix that became more rigid and less accessible to thermally activated solvent molecules (Bismuto et al., 2003).
Seven hyper-stable multiple mutants of staphylococcal nuclease have been constructed by various combinations of eight different stabilizing single mutants (including mutations of negatively charged residues—D21N and D21K) (Chen et al., 2000). Their thermal denaturation midpoint temperatures were 12.6–22.9°C higher than that of the wild-type. The crystal structures of these mutants were solved at high resolution, yet no major structural changes were found, with most changes localized around the site of mutation. Rearrangements were observed in the packing of side chains in the major hydrophobic core, although none of the mutations were in the core. Surprisingly, detailed structural analysis showed that packing had improved, with the volume of the mutant hydrophobic core decreasing as protein stability increased. The authors believed that these results indicate that optimization of packing follows as a natural consequence of increased protein thermal stability and that good packing is not necessarily the proximate cause of high stability (Chen et al., 2000). The mutants showed that increased numbers of electrostatic and hydrogen bonding interactions are not obligatory for large increases in protein stability. Based on the electrostatic energy calculations, it has been suggested that at least two of mutants, D21N and D21K, increase stability by removing unfavorable electrostatic interactions (Chen et al., 2000).
The development of reliable methods for the prediction of result of mutations in protein requires the knowledge of the force field. This is a complex task that should take into account the delicate balance between the different energy terms that contribute to protein stability. The force fields usually use an effective physical energy function or they are based on statistical potentials where energies are derived from the frequency of residue or atom contacts in the protein database or they use empirical data obtained from experiments. For example, a computer algorithm, FOLDEF (for FOLD-X energy function), has been developed to provide a fast and quantitative estimation of the importance of the interactions contributing to the stability of proteins and protein complexes (Guerois et al., 2002). At the same time, the free energy of protein unfolding includes several contributions, some of which are difficult to estimate. For this reason, the predictive power of methods developed so far is still not as high as desired and researchers continue to suggest new approaches. Mozo-Villarias et al. have developed a thermostability criterion for a protein in terms of a quasi-electric dipole moment (contributed by its charged residues) defined for an electric charge distribution whose total charge is not zero (Mozo-Villarias et al., 2003). It was found that minimization of the modulus of this dipole moment increased its thermal stability. In spite of these efforts to create methods for prediction of effects of mutations on stability of proteins, we still do not have reliable and simple approaches to solve this problem.
In the work presented here, the energies of charge–charge interactions in apo- and Ca2+-loaded -LA were calculated using a Tanford–Kirkwood algorithm with solvent accessibility correction or using the finite difference Poisson–Boltzmann method. These continuum electrostatic models have been shown to capture successfully the interactions between ionizable residues on the protein surfaces (for reviews, see Klapper et al., 1986; Gilson, 1995; Elcock and McCammon, 1998; Schaefer et al., 1998; Schutz and Warshel, 2001; Dong and Zhou, 2002; Bashford, 2004; Dominy et al., 2004; Feig et al., 2004; Garcia-Moreno and Fitch, 2004). The analysis revealed that several residues in the -LA sequence have unfavorable charge–charge interactions. The theoretical predictions were confirmed by site-directed mutagenesis of those charged surface residues and detailed physico-chemical characterization of each mutant protein. The experimental data obtained were in good qualitative agreement with the theoretical predictions.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Materials
Bovine -LA (lot 60K7002) was purchased from Sigma Chemical (St. Louis, MO) and used without further purification. Protein concentrations were determined spectrophotometrically using an extinction coefficient E1%, 280 nm = 20.1 (Kronman et al., 1964). EDTA standard solutions (Fisher Scientific) were used for calcium titration. All other chemicals were of reagent grade or higher. All solutions were prepared with distilled, deionized water.
Recombinant proteins were prepared as described previously (Anderson et al., 1997). The mutant plasmids were prepared by the method of Kunkel (Kunkel, 1985). All proteins were characterized by absorption, fluorescence and CD spectroscopy.
All of the recombinant -LAs retained the additional N-terminal methionine residue which was previously found to contribute to both decreased thermal stability and lower calcium affinity (Ishikawa et al., 1998; Chaudhuri et al., 1999; Veprintsev et al., 1999). Selective removal of the N-terminal Met was achieved using Aeromonas proteolitica aminopeptidase (Wilkes et al., 1973; Prescott and Wilkes, 1976) as described by Veprintsev et al. (1999). Only the amino terminal Met is removed as the activity of this aminopeptidase is stopped by N-terminal amino acids with large negatively charged side chain (i.e. the first residue, Glu1, in the native bovine -LA sequence). Briefly, the digestion reaction was performed at a substrate to enzyme ratio of about 100:1 (-LA 
2 mg/ml, 2 h, 37°C, 10 mM HEPES, pH 8.0). Excess EDTA (final concentration, 10 mM) was added to quench the reaction. Proteins were separated by gel filtration on a Sephadex G-100 column and fractions containing -LA were collected, dialyzed against 10 mM ammonium bicarbonate and lyophilized.
Saturation of intact and mutant -LA by calcium ions was achieved via addition of 1 mM CaCl2. The use of higher concentrations of calcium could result in saturation of the weaker secondary calcium-binding site of -LA (Aramini et al., 1992). Removal of Ca2+ was achieved via addition of 1 mM EDTA. Higher concentrations of EDTA could promote EDTA binding to -LA (Permyakov et al., 1987).
Instrumentation and methods
Fluorescence measurements
Fluorescence measurements were performed on a Perkin-Elmer LS50B or a laboratory-made instrument with a precision titrator device and Peltier temperature-controlled cell holder described previously (Permyakov et al., 1977). The excitation wavelength was 280.4 nm. All spectra were corrected and fit to log-normal curves using non-linear regression analysis (Marquardt, 1963) to obtain the emission maximum for each spectrum (Burstein and Emelyanenko, 1996). In all fluorescence experiments, illumination time and UV irradiation power levels were minimized in order to avoid UV-induced structural rearrangements (Vanhooren et al., 2002; Permyakov et al., 2003).
Temperature scans were performed stepwise, allowing the sample to equilibrate at each temperature for at least 5 min. Temperature was monitored directly inside the cell. The fraction of conversion from the native to the thermally unfolded state was calculated as described previously (Permyakov and Burstein, 1984; Permyakov, 1993).
The calcium binding affinity of each -LA mutant was measured by spectrofluorimetric back-titration of the calcium-loaded protein with a strong calcium chelator (e.g. EDTA) at fixed pH. Calculations of the calcium association constant from the experimental data were based on a competitive binding scheme between protein (P) and chelator (H) for calcium ions (Permyakov et al., 1985; Permyakov, 1993):
 | (1) |
) based upon K2 values from Schwarzenbach and Flaschka (1965).
Calcium binding and protein stability
The role of calcium on stability is linked to the difference in binding affinity for the folded protein and the unfolded protein (Permyakov et al., 1985):
 | (2) |
G0 and –G1 and the two vertical equilibria by binding constants K0 and K1, respectively, then the thermodynamic cycle results in the rigorous relation
 | (3) |
Because calcium has a much higher affinity for the folded state (i.e. K1 >> K0), the folding equilibrium of the calcium-loaded state is shifted towards higher temperatures (i.e. G1 >> G0).
In principle, in the presence of calcium, the total concentration of the folded protein is the sum of the two species, [F] and [FCa] and similarly for the unfolded state, [U] and [UCa]. Thus, the overall equilibrium constant can be described as
 | (4) |
-LA (Permyakov et al., 1987). Likewise, the use of higher concentrations of calcium may result in saturation of the weaker secondary calcium-binding site of -LA (Aramini et al., 1992).
Scanning calorimetry
Scanning calorimetric measurements were carried out on a VP-DSC differential scanning microcalorimeter (Microcal, Northhampton, MA) at a 0.5 K/min or 1 K/min heating rate in 10 mM HEPES–KOH buffer, pH 7.7. A pressure of 30 psi was maintained in order to prevent degassing of the solutions during heating. Protein concentrations were 0.3–1 mg/ml. The heat sorption curves were baseline corrected. Protein specific heat capacity (Cp) was calculated as described by Privalov and Potekhin (1986). The partial molar volume was calculated according to Hakel et al. (1999). The temperature dependence of Cp was fitted to a simple two-state model, assuming that the difference between heat capacities of the denatured and native proteins (Cp) was independent of temperature (all values were normalized by molecular weight):
 | (5) |
 |
Here Cp,D is the specific heat capacity of the denatured protein, linearly extrapolated to the transition region. The fitting parameters, Cp, HVH (van't Hoff's enthalpy of protein denaturation) and T0 (mid-transition temperature) were estimated with the Origin 5.0 software provided with the MicroCal VP-DSC. The free energy change of thermal denaturation, G, was calculated as follows:
 | (6) |
Circular dichroism Circular dichroism measurements were performed on either an AVIV 62DS, Applied Photophysics PiStar or Jasco J-500A spectropolarimeter. Typical instrument conditions were scan rate 5 nm/min and time constant 8 s. The pathlength was 0.19 mm for the far-UV and 10 mm for the near-UV region. All data were baseline corrected.
Electrostatics calculations
The energies of charge–charge interaction were calculated using a TK-SA procedure, implementation of which is described in detail elsewhere (Ibarra-Molero et al., 1999; Loladze et al., 1999; Makhatadze et al., 2003). Briefly, the energy of pairwise interactions between unit charges was calculated according to a Tanford–Kirkwood algorithm (Tanford and Kirkwood, 1957) with the solvent accessibility correction as proposed by Matthew and Gurd (1986). The mean field approximation was used for calculating the effect of charge–charge interactions on the pKa of ionizable groups from their model compound values (Asp, 4.0; Glu, 4.5; Lys, 10.5; Arg, 12.0; His, 6.3; Tyr, 10.5; N-terminus, 7.7; C-terminus, 3.6). Calculations were performed on the PDB entries 1F6R (apo-form) and 1F6S (Ca2+-bound form) for bovine -LA (Chrysina et al., 2000). Missing atoms were reconstructed using the default option of the SwissPDB Viewer. Calculations of the charge–charge interactions for each of the six structural subunits in each PDB file were performed and average values were reported. For comparison of native -LA with the various protein mutants, it was assumed that the charge–charge interactions in the unfolded states of these proteins were similar. The results of TK-SA calculations were also compared with the calculation done using the finite difference Poisson–Boltzmann (FDPB) method as implemented in the UHBD software package (Antosiewicz et al., 1994) as described (Fitch et al., 2002; Makhatadze et al., 2004). Both methods gave qualitatively identical results (see below). In addition, charge–charge interactions in the unfolded state were calculated using a Gaussian chain model as suggested by Zhou (2002a). Because the residues of interest are located on the protein surface and have high (>50%) solvent accessibility, the effects of solvation were not explicitly taken into account.
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Energetics of charge–charge interactions in
-LA
Calculations of the energies of the charge–charge interactions were carried out for three -LA forms: apo-protein (1F6R), holo-protein (1F6S
[PDB]
) and an apo-form based on the 1F6S structure with Ca2+ ion excluded from the calculations. Comparison of the results from these calculations (Figure 1) reveals several interesting features:
- In the apo-form, many acidic side chains have unfavorable charge–charge interactions, with 11 residues (Glu1, Glu7, Glu11, Asp63, Asp64, Asp78, Asp82, Asp83, Asp84, Asp87 and Asp88) possessing significantly unpropitious charge–charge interactions.
- Minor structural rearrangements accompanying Ca2+ binding to -LA (Chrysina et al., 2000) do not lead to any significant changes in energies of charge–charge interactions. This can be seen from comparison of charge–charge interactions calculated for apo--LA with the results obtained on Ca2+-bound structure without including Ca2+ in the calculations.
- It is clear that Ca2+-binding has the most pronounced effect on residues directly involved in cation coordination. For each of these three residues, Asp82, Asp87 and Asp88, the charge–charge interaction energy, being large and unfavorable in the apo-form, becomes comparably favorable after Ca2+ ion binding. Charge–charge interactions of the other two residues in the Ca2+-binding loop, Asp83 and Asp84, are also significantly affected by Ca2+ binding. On the other hand, Ca2+ binding only slightly affects residues more distant from the Ca2+-binding site and the energy contributions from Glu1, Glu7, Glu11, Asp63 and Asp64, which have large, unfavorable charge–charge interactions in the apo-form, remained essentially unchanged in the Ca2+-bound form.
|
It is also remarkable that both TK-SA and FDPB/UHBD calculations give qualitatively similar results (Figure 1) despite the differences in the assumptions each model incorporates. In both treatments of energetics of charge–charge interactions it is assumed that the contribution of the residual charge–charge interactions in the unfolded state is small and should not affect qualitatively the results of the calculations. This can be seen from the comparison of the results of FDPB/UHBD calculations for the native state with the calculations done using a Gaussian model of the unfolded chain proposed by Zhou (2002a)
. It is clear that in the case of -LA charge–charge interactions in the unfolded state are relatively small and hence justify the assumption that in a qualitative approximation the mutations have an effect only on the charge–charge interactions in the native state. Based on these observations, one can predict that mutations, which neutralize side chains that are involved in unfavorable charge–charge interactions (e.g. Glu1, Glu7, Glu11), should result in an increase in protein stability. Conversely, substitutions leading to charge neutralization on side chains that provide favorable charge–charge interactions (i.e. Asp14, Glu25) should result in a decrease in protein stability relative to the native protein. Furthermore, from the analysis presented above, the most dramatic changes would be expected from mutations in the calcium-binding loop (including those non-coordinating residues) and secondly the N-terminal region of the protein.
The energies calculated for the calcium loop were not unexpected and the pronounced stabilization of the apo-form by Ca2+ binding originated from the negative charge ‘compensation’ in the coordination site was demonstrated earlier (Permyakov et al., 2001). Until recently, however, the N-terminal region of -LA had not been examined with respect to charged amino acid residues and protein stability. Consequently, we constructed a set of mutants covering this region in the 3-D structure of -LA: E1Q, E7Q, E11L, D14N, D37N, including two additional mutants previously reported: E1 (E1M) and E25A (Veprintsev et al., 1999; Permyakov et al., 2000). Since recombinant -LA contains an extra methionine residue at the N-terminus, which was found to destabilize protein structure, thermal stability and calcium affinity (Ishikawa et al., 1998; Chaudhuri et al., 1999; Veprintsev et al., 1999), selective removal of the additional N-terminal Met from recombinant proteins was performed using A.proteolitica aminopeptidase enzyme [with the exception of the mutants E1 (E1M) and E1Q].
Physico-chemical properties of -lactalbumin mutants and comparisons with electrostatic calculations
The correctness of folding of -LA mutants is assessed, in part, from the maximum wavelength positions of the intrinsic fluorescence spectra (
max) of Ca2+-loaded (1 mM CaCl2) -LA forms (see Table I). This spectral parameter reflects the mobility and polarity of the environment of emitting residues in proteins and generally reflects the degree of accessibility to solvent molecules (Permyakov, 1993). All of the mutants studied here, except for desMet E25A, possess similar max values, within 4–5 nm, which indicates only slight structural perturbations of the environment of their tryptophan residues. The increase of 8–9 nm in max for desMetE25A -LA reflects a significantly increased Trp exposure to solvent compared with native -LA.
|
The far-UV circular dichroism analyses suggest a high
-helical content in all of the mutants studied (data not shown), confirming that all retain a native-like -LA fold.
Typical spectrofluorimetric thermal denaturation curves for the apo- (1 mM EDTA, pH 7.7) and calcium-loaded (1 mM CaCl2) states of desMet-D37N -LA are shown in Figure 2. A red shift in the max value with temperature corresponds to a progressive exposure of Trp residues to solvent water accompanying protein denaturation. It is remarkable that the D37N substitution shifts the thermal unfolding transition toward higher temperature, both in the absence and presence of calcium. Similar observations were found from scanning microcalorimetry (Figure 3).
|
|
Although very similar thermal transition mid-temperature (tm) values were extracted from the DSC data, some differences have been found earlier between fluorescence and calorimetric data with apo-
-LA (Veprintsev et al., 1997). We report tm values from the fluorescence experiments here since they were run at significantly lower protein concentrations than required for DSC, which minimizes the possibilities for aggregation and also allows the use of high EDTA:protein ratios in the calcium back-titrations. This was particularly important with mutants exhibiting increased calcium affinity.
Table I summarizes all of the thermal transition mid-temperatures (tm) for the -LA species in this work. Every mutant studied possessed a thermal transition, even in the absence of calcium. Hence none of these mutations resulted in a total loss of protein tertiary structure. Yet desMet D14N -LA displayed the most significant decrease in thermal stability, in both the absence and presence of calcium, resembling recombinant wt -LA. A similar, but less pronounced, destabilization was observed for desMet E25A -LA. On the other hand, E1(E1M) and desMet D37N -LA exhibited significantly increased thermal stabilities (see Table I). Overall, the charge-neutralizing mutations in the N-terminal region of -LA resulted in protein forms differing in thermal stability by more than 17°C, suggesting very unusual electrostatic interactions in this region. Yet other substitutions that were predicted to increase thermal stability, namely desMet E7Q and desMet E11L, resulted in relatively minor effects (<2°C).
The results obtained with E1Q -LA are especially interesting. Although this mutant exhibited slightly lower thermal stability (1–3°C) with respect to native -LA, this particular mutant contains an additional N-terminal Met residue, thus E1Q -LA should be compared with recombinant wild-type -LA. Consequently, removal of the negative charge at Glu1 in wild-type -LA results in a substantial recovery of protein thermal stability. The same effect was observed for the native protein, i.e. substitution of Glu1 by Met1 in the E1(E1M) mutant significantly increased the thermal stability of -LA (Table I). The stabilizing effects of both mutations were easily rationalized from the electrostatic calculations in Figure 1, which showed that the carboxylate side chain of Glu1 provides distinctly unfavorable contributions; hence charge neutralization of the Glu side chain should add protein stabilization (as confirmed from the experimental data). Furthermore, electrostatic calculations for wild-type protein showed that the extra N-terminal Met residue resulted in a decrease in the favorable contributions from the positively charged -amino group (Figure 1). Hence substantial destabilization is caused by moving the N-terminus away from a region of highly negative potential, again in accordance with experimental data (see Table I). It therefore appears that the pronounced destabilization of recombinant wild-type -LA noted earlier is due to unfavorable charge–charge interactions.
In order to compare quantitatively the experimental data with the electrostatic calculations the mutation-induced changes in the Gibbs free energy of thermal denaturation (Gn
m) of apo- and Ca2+-loaded -LA species were estimated from the mid-transition temperatures (tm) in Table I. The calculations were based on DSC data for native -LA (see Figure 3), which were analyzed according to the simple two-state model followed by calculation of the free energy change upon thermal denaturation (G) according to Equation (6), given in Materials and methods:
 |
The resulting Gnm values are presented in Table I. A comparative plot of Gnm versus the change in the energy of charge–charge interactions (Gq-q) is shown in Figure 4. Overall, the correlation qualitatively predicts the effects of every substitution with the possible exception of the D37N mutant. The TK-SA algorithm for this latter substitution predicted little change in stability due to changes in charge–charge interactions, whereas the experimental results showed a significant increase in stability. Since the TK-SA calculations reflect changes only in charge–charge interactions, other potential effects such as changes in hydrophobicity, hydrogen bonding, configurational entropy or secondary structure propensities may account for the additional stabilization.
|
Another distinguishing property of the studied
-LA mutants is their calcium affinity. Figure 5 depicts a typical spectrofluorimetric titration for desMet D37N -LA. The Ca2+-loaded protein was back-titrated with EDTA at 45°C (pH 7.6), which is well above the thermal transition for the apo- protein and well below the thermal transition for the Ca2+-bound form (see also Figures 2 and 3). The calcium equilibrium association constants were calculated as described in Materials and methods according to Scheme (1) (Permyakov et al., 1985; Permyakov, 1993). Apparent calcium binding association constants for all -LA species analyzed in this study are listed in Table I. The results show that the substitution-induced changes in calcium affinity of a given protein correlate well with concomitant changes in its thermal stability, i.e. the more thermostable mutants possess a higher affinity for Ca2+. Accordingly, recombinant wild-type and desMet D14N -LA bind Ca2+ about one order of magnitude weaker, whereas desMet D37N and E1 -LA bind Ca2+ 3–6 times more strongly than native -LA. At 45°C, where the binding of Ca2+ by intact and mutant -LA converts denatured apo-protein into native calcium-bound form (see Table I), pair-wise reduction of free energy terms corresponding to denatured apo- and calcium-saturated forms of intact and mutant proteins easily results in equality of the mutation-induced change in the free energy of denaturation of Ca2+-loaded -LA Gnm and of the mutation-induced change in free energy of calcium binding (taken with the opposite sign):
 |
-LA thermal stability described here is related to an enhancement of calcium binding affinity.
|
|
Role of charge–charge interactions in structural stabilization
Computational analysis of charge–charge interactions in bovine -LA conducted in this work was qualitatively validated by experiment. The N-terminal sequence (residues 1–11) is characterized by a high proportion of negatively charged residues that cluster on the surface of the native protein. Neutralization of unfavorable charge–charge interactions in the N-terminus results in stabilization of both the apo- and Ca2+-bound protein. As also demonstrated above, an increase in thermal stability is related to an increase in calcium binding affinity. If one considers the binding of basic proteins, histones or positively charged polyamino acids, these interactions are accompanied by different effects (Permyakov et al., 2004). Although the interaction with basic polymers increased the thermal stability of apo--LA, the calcium-saturated protein was destabilized (the calcium affinity was diminished also). An example that illustrates the comparable mechanism is the effect of Asp87 replacement, which directly coordinates the calcium ion in -LA, with the neutral side chain Asn (Permyakov et al., 2001). The resulting decrease in calcium binding affinity was about two orders of magnitude with an accompanying decrease in thermal stability of the Ca2+-loaded protein. It is likely that the same destabilization of the Ca2+-saturated form occurs in the case of -LA interactions with the basic proteins, i.e. involvement of the highly negatively charged calcium-binding loop in the binding of basic proteins results in a decrease in calcium affinity, consequently decreasing the thermal stability of the Ca2+-loaded protein. The interaction of apo-protein with a basic polymer stabilizes -LA via the neutralization of the highly unfavorable negative charge distribution of the calcium binding-loop (see Figure 1). Hence the N-terminus may be involved in similar compensating interactions with histones, positively charged polyamino acids and other polycations in vivo.
It should be noted that, despite many years of protein engineering studies with -LA, rarely have mutations been found that improve protein physico-chemical properties. We are aware of only two cases where a significant increase in -LA thermal stability was found and most of these mutations were not planned via rational computational approaches (Greene et al., 1999; Veprintsev et al., 1999). This strategy was first proposed by Loladze et al. in 1999 (Loladze et al., 1999) based on experimental data from several papers, which showed that mutations in surface charges can significantly alter protein stability (Grimsley et al., 1999; Paoli et al., 1999; Perl et al., 2000; Spector et al., 2000). It was further reinforced when it was shown that computational approaches are capable of qualitatively predicting the consequences of surface charge mutations on protein stability (Perl and Schmid, 2001; Sanchez-Ruiz and Makhatadze, 2001; Forsyth et al., 2002; Zhou, 2002b; Luisi et al., 2003; Makhatadze et al., 2003; Schwehm et al., 2003).
The results presented here provide experimental validation for rational optimization of charge–charge interactions on a protein surface as a tool to modulate protein stability. Examination of the energy contributed by Glu1 and the energetically favorable consequences of neutralizing this residue clearly suggests that nature may have made an error with bovine -LA (as opposed to the human and most other species, which contain an N-terminal Lys!). Or was this destabilizing contribution intentional for other physiological reasons? Further studies on the differences between bovine and other -LA species in physiological function may shed more light on this question.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
We acknowledge Takuji Kobayashi and Richard Kristinsson for their technical assistance. This work was supported in part by grants from the USPHS (NIH GM 56970 to L.J.B.), the National Science Foundation (MCB-0110396 to G.I.M.), the Programs of the Russian Academy of Sciences (‘Fundamental science for medicine’ and ‘Molecular and cellular biology’), a grant of the Russian Fund for Basic Research (04-04-97322 to E.A.P.) and INTAS (01-2347 CA to V.N.U., E.A.P. and S.E.P.).
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionAcknowledgementsReferences |
|---|
Anderson,P.J., Brooks,C.L. and Berliner,L.J. (1997) Biochemistry, 36, 11648–11654.[CrossRef][Medline]
Antosiewicz,J., McCammon,J.A. and Gilson,M.K. (1994) J. Mol. Biol., 238, 415–436.[CrossRef][Web of Science][Medline]
Aramini,J.M., Drakenberg,T., Hiraoki,T., Ke,Y., Nitta,K. and Vogel,H.J. (1992) Biochemistry, 31, 6761–6768.[CrossRef][Medline]
Bashford,D. (2004) Front. Biosci., 9, 1082–1099.[Web of Science][Medline]
Bismuto,E., Febbraio,F., Limongelli,S., Briante,R. and Nucci,R. (2003) Proteins, 51, 10–20.[CrossRef][Web of Science][Medline]
Bjork,A., Dalhus,B., Mantzilas,D., Sirevag,R. and Eijsink,V.G. (2004) J. Mol. Biol., 341, 1215–1226.[CrossRef][Web of Science][Medline]
Burstein,E.A. and Emelyanenko,V.I. (1996) Photochem. Photobiol., 64, 316–320.
Chaudhuri,T.K. et al. (1999) J. Mol. Biol., 285, 1179–1194.[CrossRef][Web of Science][Medline]
Chen,J., Lu,Z., Sakon,J. and Stites,W.E. (2000) J. Mol. Biol., 303, 125–130.[CrossRef][Web of Science][Medline]
Chrysina,E.D., Brew,K. and Acharya,K.R. (2000) J. Biol. Chem., 275, 37021–37029.
Dolgikh,D.A., Gilmanshin,R.I., Brazhnikov,E.V., Bychkova,V.E., Semisotnov,G.V., Venyaminov,S. and Ptitsyn,O.B. (1981) FEBS Lett., 136, 311–315.[CrossRef][Web of Science][Medline]
Dominy,B.N., Minoux,H. and Brooks,C.L.,III (2004) Proteins, 57, 128–141.[CrossRef][Web of Science][Medline]
Dong,F. and Zhou,H.X. (2002) Biophys. J., 83, 1341–1347.[Web of Science][Medline]
Duringer,C., Hamiche,A., Gustafsson,L., Kimura,H. and Svanborg,C. (2003) J. Biol. Chem., 278, 42131–42135.
Elcock,A.H. and McCammon,J.A. (1998) J. Mol. Biol., 280, 731–748.[CrossRef][Web of Science][Medline]
Feig,M., Onufriev,A., Lee,M.S., Im,W., Case,D.A. and Brooks,C.L.,III (2004) J. Comput. Chem., 25, 265–284.[CrossRef][Web of Science][Medline]
Fitch,C.A., Karp,D.A., Lee,K.K., Stites,W.E., Lattman,E.E. and Garcia-Moreno,E.B. (2002) Biophys, J., 82, 3289–3304.[Web of Science][Medline]
Forsyth,W.R., Antosiewicz,J.M. and Robertson,A.D. (2002) Proteins, 48, 388–403.[CrossRef][Web of Science][Medline]
Garcia-Moreno,E.B. and Fitch,C.A. (2004) Methods Enzymol., 380, 20–51.[Web of Science][Medline]
Gilson,M.K. (1995) Curr. Opin. Struct. Biol., 5, 216–223.[CrossRef][Web of Science][Medline]
Greene,L.H., Grobler,J.A., Malinovskii,V.A., Tian,J., Acharya,K.R. and Brew,K. (1999) Protein Eng., 12, 581–587.
Grimsley,G.R., Shaw,K.L., Fee,L.R., Alston,R.W., Huyghues-Despointes,B.M., Thurlkill,R.L., Scholtz,J.M. and Pace,C.N. (1999) Protein Sci., 8, 1843–1849.[Web of Science][Medline]
Guerois,R., Nielsen,J.E. and Serrano,L. (2002) J. Mol. Biol., 320, 369–387.[CrossRef][Web of Science][Medline]
Hackel,M., Hinz,H.J. and Hedwig,G.R. (1999) J. Mol. Biol., 291, 197–213.[CrossRef][Web of Science][Medline]
Hakansson,A., Zhivotovsky,B., Orrenius,S., Sabharwal,H. and Svanborg,C. (1995) Proc. Natl Acad. Sci. USA, 92, 8064–8068.
Hakansson,A., Andreasson,J., Zhivotovsky,B., Karpman,D., Orrenius,S. and Svanborg,C. (1999) Exp. Cell Res., 246, 451–460.[CrossRef][Web of Science][Medline]
Ibarra-Molero,B., Loladze,V.V., Makhatadze,G.I. and Sanchez-Ruiz,J.M. (1999) Biochemistry, 38, 8138–8149.[CrossRef][Medline]
Ishikawa,N., Chiba,T., Chen,L.T., Shimizu,A., Ikeguchi,M. and Sugai,S. (1998) Protein Eng., 11, 333–335.
Klapper,I., Hagstrom,R., Fine,R., Sharp,K. and Honig,B. (1986) Proteins, 1, 47–59.[CrossRef][Medline]
Kronman,M.J. Aandreotti,R. and Vitols,R. (1964) Biochemistry, 28, 1152–1160.[CrossRef]
Kunkel,T.A. (1985) Proc. Natl Acad. Sci. USA, 82, 488–492.
Loladze,V.V., Ibarra-Molero,B., Sanchez-Ruiz,J.M. and Makhatadze,G.I. (1999) Biochemistry, 38, 16419–16423.[CrossRef][Medline]
Luisi,D.L., Snow,C.D., Lin,J.J., Hendsch,Z.S., Tidor,B. and Raleigh,D.P. (2003) Biochemistry, 42, 7050–7060.[CrossRef][Medline]
Makhatadze,G.I., Loladze,V.V., Ermolenko,D.N., Chen,X. and Thomas,S.T. (2003) J. Mol. Biol., 327, 1135–1148.[CrossRef][Web of Science][Medline]
Makhatadze,G.I., Loladze,V.V., Gribenko,A.V. and Lopez,M.M. (2004) J. Mol. Biol., 336, 929–942.[CrossRef][Web of Science][Medline]
Marquardt,D.W. (1963) J. Soc. Ind. Appl. Math., 11, 431–441.[CrossRef]
Matthew,J.B. and Gurd,F.R. (1986) Methods Enzymol., 130, 413–436.[Web of Science][Medline]
Minagawa,H., Shimada,J. and Kaneko,H. (2003) Eur. J. Biochem., 270, 3628–3633.[Web of Science][Medline]
Mozo-Villarias,A., Cedano,J. and Querol,E. (2003) Protein Eng., 16, 279–286.
Paoli,P., Taddei,N., Fiaschi,T., Veggi,D., Camici,G., Manao,G., Raugei,G., Chiti,F. and Ramponi,G. (1999) Arch. Biochem. Biophys., 363, 349–355.[CrossRef][Medline]
Pellegrini,A., Thomas,U., Bramaz,N., Hunziker,P. and von Fellenberg,R. (1999) Biochim. Biophys. Acta, 1426, 439–448.[Medline]
Perl,D. and Schmid,F.X. (2001) J. Mol. Biol., 313, 343–357.[CrossRef][Web of Science][Medline]
Perl,D., Mueller,U., Heinemann,U. and Schmid,F.X. (2000) Nat. Struct. Biol., 7, 380–383.[CrossRef][Web of Science][Medline]
Permyakov,E.A. (1993) Luminescent Spectroscopy of Proteins. CRC Press, Boca Raton, FL.
Permyakov,E.A. and Berliner,L.J. (2000) FEBS Lett., 473, 269–274.[CrossRef][Web of Science][Medline]
Permyakov,E.A. and Burstein,E.A. (1984) Biophys. Chem., 19, 265–271.[CrossRef][Web of Science][Medline]
Permyakov,E.A., Burstein,E.A., Sawada,Y. and Yamazaki,I. (1977) Biochim. Biophys. Acta, 491, 149–154.[Medline]
Permyakov,E.A., Morozova,L.A. and Burstein,E.A. (1985) Biophys. Chem., 21, 21–31.[CrossRef][Web of Science][Medline]
Permyakov,E.A., Murakami,K. and Berliner,L.J. (1987) J. Biol. Chem., 262, 3196–3198.
Permyakov,E.A., Permyakov,S.E., Deikus,G.Y., Morozova-Roche,L.A., Grishchenko,V.M., Kalinichenko,L.P. and Uversky,V.N. (2003) Proteins, 51, 498–503.[CrossRef][Web of Science][Medline]
Permyakov,S.E., Veprintsev,D.B., Brooks,C.L., Permyakov,E.A. and Berliner,L.J. (2000) Proteins, 40, 106–111.[CrossRef][Web of Science][Medline]
Permyakov,S.E., Uversky,V.N., Veprintsev,D.B., Cherskaya,A.M., Brooks,C.L., Permyakov,E.A. and Berliner,L.J. (2001) Protein Eng., 14, 785–789.
Permyakov,S.E., Pershikova,I.V., Khokhlova,T.I., Uversky,V.N. and Permyakov,E.A. (2004) Biochemistry, 43, 5575–5582.[CrossRef][Medline]
Prescott,J.M. and Wilkes,S.H. (1976) Methods Enzymol., 45, 530–543.[Medline]
Privalov,P.L. and Potekhin,S.A. (1986) Methods Enzymol., 131, 4–51.[Medline]
Sanchez-Ruiz,J.M. and Makhatadze,G.I. (2001) Trends Biotechnol., 19, 132–135.[CrossRef][Web of Science][Medline]
Schaefer,M., van Vlijmen,H.W. and Karplus,M. (1998) Adv. Protein Chem., 51, 1–57.[Web of Science][Medline]
Schutz,C.N. and Warshel,A. (2001) Proteins, 44, 400–417.[CrossRef][Web of Science][Medline]
Schwarzenbach,G. and Flaschka,H. (1965) Die Komplexometrische Titration. Ferdinand Enke Verlag, Stuttgart.
Schwehm,J.M., Fitch,C.A., Dang,B.N., Garcia-Moreno,E.B. and Stites,W.E. (2003) Biochemistry, 42, 1118–1128.[CrossRef][Medline]
Spector,S., Wang,M., Carp,S.A., Robblee,J., Hendsch,Z.S., Fairman,R., Tidor,B. and Raleigh,D.P. (2000) Biochemistry, 39, 872–879.[CrossRef][Medline]
Svensson,M., Sabharwal,H., Hakansson,A., Mossberg,A.K., Lipniunas,P., Leffler,H., Svanborg,C. and Linse,S. (1999) J. Biol. Chem., 274, 6388–6396.
Svensson,M., Hakansson,A., Mossberg,A.K., Linse,S. and Svanborg,C. (2000) Proc. Natl Acad. Sci. USA, 97, 4221–4226.
Tanford,C. and Kirkwood,J.G. (1957) J. Am. Chem. Soc., 79, 5333–5339.[CrossRef]
Vanhooren,A., Devreese,B., Vanhee,K., Van Beeumen,J. and Hanssens,I. (2002) Biochemistry, 41, 11035–11043.[CrossRef][Medline]
Veprintsev,D.B., Permyakov,S.E., Permyakov,E.A., Rogov,V.V., Cawthern,K.M. and Berliner,L.J. (1997) FEBS Lett., 412, 625–628.[CrossRef][Medline]
Veprintsev,D.B., Narayan,M., Permyakov,S.E., Uversky,V.N., Brooks,C.L., Cherskaya,A.M., Permyakov,E.A. and Berliner,L.J. (1999) Proteins, 37, 65–72.[CrossRef][Web of Science][Medline]
Wilkes,S.H., Bayliss,M.E. and Prescott,J.M. (1973) Eur. J. Biochem., 34, 459–466.[Web of Science][Medline]
Zhou,H.X. (2002a) Proc. Natl Acad. Sci. USA, 99, 3569–3574.
Zhou,H.X. (2002b) Biochemistry, 41, 6533–6538.[CrossRef][Medline]
Received April 5, 2005; revised June 28, 2005; accepted June 30, 2005.
Edited by Gideon Schreiber
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. V. Gribenko, M. M. Patel, J. Liu, S. A. McCallum, C. Wang, and G. I. Makhatadze Rational stabilization of enzymes by computational redesign of surface charge-charge interactions PNAS, February 24, 2009; 106(8): 2601 - 2606. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Halskau Jr, R. Perez-Jimenez, B. Ibarra-Molero, J. Underhaug, V. Munoz, A. Martinez, and J. M. Sanchez-Ruiz Large-scale modulation of thermodynamic protein folding barriers linked to electrostatics PNAS, June 24, 2008; 105(25): 8625 - 8630. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||