Linkage of thioredoxin stability to titration of ionizable groups with perturbed pKa

The highly conserved, buried, Asp 26 in Escherichia coli thioredoxin has a pKa = 7.5, and its titration is associated with a sizable destabilization of the protein [Langsetmo, K., Fuchs, J., & Woodward, C. (1991) Biochemistry (preceding paper in this issue)]. A fit of the experimental pH dependence of thioredoxin stability to a theoretical expression for the pH/stability relation in proteins agrees closely with a pKa value of 7.5 for Asp 26. The agreement between the experimental and theoretical changes in protein stability due to substitution of Asp 26 by alanine is also good. The local structure in the vicinity of Asp 26 in the low-pH crystal structure (with uncharged Asp 26) is hydrophobic, indicating that the aspartate would be highly destabilized. In theoretical calculations, the desolvation penalty for deprotonating Asp 26 in this environment is similar to the total protein folding energy. As a consequence, the Asp 26 pKa would be much greater than 7.5, and/or the protein might not fold. This suggests that a compensating process partially stabilizes the Asp 26 carboxyl group when it is charged. A simple model for this proposed, whereby the Lys 57 side chain rotates to form a salt bridge with Asp 26 when it is deprotonated.     

Conformation, stability, and active-site cysteine titrations of Escherichia coli D26A thioredoxin probed by Raman spectroscopy.

The active-site cysteines (Cys 32 and Cys 35) of Escherichia coli thioredoxin are oxidized to a disulfide bridge when the protein mediates substrate reduction. In reduced thioredoxin, Cys 32 and Cys 35 are characterized by abnormally low pKa values. A conserved side chain, Asp 26, which is sterically accessible to the active site, is also essential to oxidoreductase activity. pKa values governing cysteine thiol-thiolate equilibria in the mutant thioredoxin, D26A, have been determined by direct Raman spectrophotometric measurement of sulfhydryl ionizations. The results indicate that, in D26A thioredoxin, both sulfhydryls titrate with apparent pKa values of 7.5+/-0.2, close to values measured previously for wild-type thioredoxin. Sulfhydryl Raman markers of D26A and wild-type thioredoxin also exhibit similar band shapes, consistent with minimal differences in respective cysteine side-chain conformations and sulfhydryl interactions. The results imply that neither the Cys 32 nor Cys 35 SH donor is hydrogen bonded directly to Asp 26 in the wild-type protein. Additionally, the thioredoxin main-chain conformation is largely conserved with D26A mutation. Conversely, the mutation perturbs Raman bands diagnostic of tryptophan (Trp 28 and Trp 31) orientations and leads to differences in their pH dependencies, implying local conformational differences near the active site. We conclude that, although the carboxyl side chain of Asp 26 neither interacts directly with active-site cysteines nor is responsible for their abnormally low pKa values, the aspartate side chain may play a role in determining the conformation of the enzyme active site.     

Proton-transfer effects in the active-site region of Escherichia coli thioredoxin using two-dimensional 1H NMR.

A series of two-dimensional (2D) correlated 1H NMR spectra of reduced and oxidized Escherichia coli thioredoxin have been used to probe the effects of pH in the vicinity of the active site, -Cys32-Gly-Pro-Cys35-, using the complete proton resonance assignments available for thioredoxin. In either oxidation state, the majority of residues of the thioredoxin molecule remain unchanged between pH 5.7 and pH 10, as indicated by the identical chemical shifts of the C alpha H, C beta H, and other protons. In reduced thioredoxin, a fairly widespread region around the active-site dithiol is affected by the titration of a group or groups with pKa approximately 7.1-7.4 in 2H2O. Another titration, with pKa approximately 8.4, affects a smaller region of the protein. Oxidized thioredoxin contains a disulfide and no free thiol groups; nevertheless, the proton resonances of many groups in the active-site region were observed to titrate with a pKa of 7.5, probably as a result of an abnormally high pKa value for the carboxyl group of the buried Asp-26 residue. For reduced thioredoxin, the results indicate that Asp-26 is titrating in this pH range, as well as both thiol groups. The new results are strongly suggestive that the mechanism of thioredoxin-catalyzed protein disulfide reduction may be critically dependent on proton transfer as well as electron transfer within the active site.     

Buried, charged, non-ion-paired aspartic acid 76 contributes favorably to the conformational stability of ribonuclease T1.

The side-chain carboxyl of Asp 76 in ribonuclease T1 (RNase T1) is buried, charged, non-ion-paired, and forms three good intramolecular hydrogen bonds (2.63, 2.69, and 2.89 A) and a 2.66 A hydrogen bond to a buried, conserved water molecule. When Asp 76 was replaced by Asn, Ser, and Ala, the conformational stability of the protein decreased by 3.1, 3.2, and 3.7 kcal/mol, respectively. The stability was measured as a function of pH for wild-type RNase T1 and the D76N mutant and showed that the pH dependence below pH 3 was almost entirely due to Asp 76. The pK of Asp 76 is 0.5 in the native state and 3.7 in the denatured state. Thus, the hydrogen bonding of the carboxyl group of Asp 76 contributes more than half of the net stability of RNase T1 at pH 7. In addition, the charged carboxyl of Asp 76 stabilizes structure in the denatured states of RNase T1 that is not present in D76N, D76S, and D76A.     

Escherichia coli thioredoxin inhibition by cadmium: two mutually exclusive binding sites involving Cys32 and Asp26.

Observations of thioredoxin inhibition by cadmium and of a positive role for thioredoxin in protection from Cd(2+) led us to investigate the thioredoxin-cadmium interaction properties. We used calorimetric and spectroscopic methods at different pH values to explore the relative contribution of putative binding residues (Cys32, Cys35, Trp28, Trp31 and Asp26) within or near the active site. At pH 8 or 7.5 two binding sites were identified by isothermal titration calorimetry with affinity constants of 10 x 10(6) m(-1) and 1 x 10(6) m(-1). For both sites, a proton was released upon Cd(2+) binding. One mole of Cd(2+) per mole of reduced thioredoxin was measured by mass spectrometry at these pH values, demonstrating that the two binding sites were partially occupied and mutually exclusive. Cd(2+) binding at either site totally inhibited the thiol-disulfide transferase activity of Trx. The absence of Cd(2+) interaction detected for oxidized or alkylated Trx and the inhibition of the enzymatic activity of thioredoxin by Cd(2+) supported the role of Cys32 at the first site. The fluorescence profile of Cd(2+)-bound thioredoxin differed, however, from that of oxidized thioredoxin, indicating that Cd(2+) was not coordinated with Cys32 and Cys35. From FTIR spectroscopy, we inferred that the second site might involve Asp26, a buried residue that deprotonates at a rather high and unusual pK(a) for a carboxylate (7.5/9.2). The pK(a) of the two residues Cys32 and Asp26 have been shown to be interdependent [Chivers, T. P. (1997) Biochemistry36, 14985-14991]. A mechanism is proposed in which Cd(2+) binding at the solvent-accessible thiolate group of Cys32 induces a decrease of the pK(a) of Asp26 and its deprotonation. Conversely, interaction between the carboxylate group of Asp26 and Cd(2+) at a second binding site induces Cys32 deprotonation and thioredoxin inhibition, so that Cd(2+) inhibits thioredoxin activity not only by binding at the Cys32 but also by interacting with Asp26.     

Contribution of proton release to the B2 photocurrent of bacteriorhodopsin.

The contribution of proton release from the so-called proton release group to the microsecond B2 photocurrent from bacteriorhodopsin (bR) oriented in polyacrylamide gels was determined. The fraction of the B2 current due to proton release was resolved by titration of the proton release group in M. At pH values below the pKa of the proton release group in M, the proton release group cannot release its proton during the first half of the bacteriorhodopsin photocycle. At these pH values, the B2 photocurrent is due primarily to translocation of the Schiff base proton to Asp85. The B2 photocurrent was measured in wild-type bR gels at pH 4.5-7.5, in 100 mM KCl/50 mM phosphate. The B2 photocurrent area (proportional to the amount of charge moved) exhibits a pH dependence with a pKa of 6.1. This is suggested to be the pKa of the proton release group in M; the value obtained is in good agreement with previous results obtained by examining photocycle kinetics and pH-sensitive dye signals. In the mutant Glu204Gln, the B2 photocurrent of the mutant membranes was pH independent between pH 4 and 7. Because the proton release group is incapacitated, and early proton release is eliminated in the Glu204Gln mutant, this supports the idea that the pH dependence of the B2 photocurrent in the wild type reflects the titration of the proton release group. In wild-type bacteriorhodopsin, proton release contributes approximately half of the B2 area at pH 7.5. The B2 area in the Glu204Gln mutant is similar to that in the wild type at pH 4.5; in both cases, the B2 current is likely due only to movement of the Schiff base proton to Asp85.     

Asp79 makes a large, unfavorable contribution to the stability of RNase Sa

The two most buried carboxyl groups in ribonuclease Sa (RNase Sa) are Asp33 (99% buried; pK 2.4) and Asp79 (85% buried; pK 7.4). Above these pK values, the stability of the D33A variant is 6kcal/mol less than wild-type RNase Sa, and the stability of the D79A variant is 3.3kcal/mol greater than wild-type RNase Sa. The key structural difference between the carboxyl groups is that Asp33 forms three intramolecular hydrogen bonds, and Asp79 forms no intramolecular hydrogen bond. Here, we focus on Asp79 and describe studies of 11 Asp79 variants. Most of the variants were at least 2kcal/mol more stable than wild-type RNase Sa, and the most interesting was D79F. At pH 3, below the pK of Asp79, RNase Sa is 0.3kcal/mol more stable than the D79F variant. At pH 8.5, above the pK of Asp79, RNase Sa is 3.7kcal/mol less stable than the D79F variant. The unfavorable contribution of Asp79 to the stability appears to result from the Born self-energy of burying the charge and, more importantly, from unfavorable charge-charge interactions. To counteract the effect of the negative charge on Asp79, we prepared the Q94K variant and the crystal structure showed that the amino group of the Lys formed a hydrogen-bonded ion pair (distance, 2.71A; angle, 100 degrees ) with the carboxyl group of Asp79. The stability of the Q94K variant was about the same as the wild-type at pH 3, where Asp79 is uncharged, but 1kcal/mol greater than that of wild-type RNase Sa at pH 8.5, where Asp79 is charged. Differences in hydrophobicity, steric strain, Born self-energy, and electrostatic interactions all appear to contribute to the range of stabilities observed in the variants. When it is possible, replacing buried, non-hydrogen bonded, ionizable side-chains with non-polar side-chains is an excellent means of increasing protein stability.     

Characterization of Escherichia coli thioredoxin variants mimicking the active-sites of other thiol/disulfide oxidoreductases.

Thiol/disulfide oxidoreductases like thioredoxin, glutaredoxin, DsbA, or protein disulfide isomerase (PDI) share the thioredoxin fold and a catalytic disulfide bond with the sequence Cys-Xaa-Xaa-Cys (Xaa corresponds to any amino acid). Despite their structural similarities, the enzymes have very different redox properties, which is reflected by a 100,000-fold difference in the equilibrium constant (K(eq)) with glutathione between the most oxidizing member, DsbA, and the most reducing member, thioredoxin. Here we present a systematic study on a series of variants of thioredoxin from Escherichia coli, in which the Xaa-Xaa dipeptide was exchanged by that of glutaredoxin, PDI, and DsbA. Like the corresponding natural enzymes, all thioredoxin variants proved to be stronger oxidants than the wild-type, with the order wild-type < PDI-type < DsbA-type < glutaredoxin-type. The most oxidizing, glutaredoxin-like variant has a 420-fold decreased value of K(eq), corresponding to an increase in redox potential by 75 mV. While oxidized wild-type thioredoxin is more stable than the reduced form (delta deltaG(ox/red) = 16.9 kJ/mol), both redox forms have almost the same stability in the variants. The pH-dependence of the reactivity with the alkylating agent iodoacetamide proved to be the best method to determine the pKa value of thioredoxin's nucleophilic active-site thiol (Cys32). A pKa of 7.1 was measured for Cys32 in the reduced wild-type. All variants showed a lowered pKa of Cys32, with the lowest value of 5.9 for the glutaredoxin-like variant. A correlation of redox potential and the Cys32 pKa value could be established on a quantitative level. However, the predicted correlation between the measured delta deltaG(ox/red) values and Cys32 pKa values was only qualitative.     

Multiple role of hydrophobicity of tryptophan-108 in chicken lysozyme: structural stability, saccharide binding ability, and abnormal pKa of glutamic acid-35.

Trp108 of chicken lysozyme is in van der Waals contact with Glu35, one of two catalytic carboxyl groups. The role of Trp108 in lysozyme function and stability was investigated by using mutant lysozymes secreted from yeast. By the replacement of Trp108 with less hydrophobic residues, Tyr (W108Y lysozyme) and Gln (W108Q lysozyme), the activity, saccharide binding ability, stability, and pKa of Glu35 were all decreased with a decrease in the hydrophobicity of residue 108. Namely, at pH 5.5 and 40 degrees C, the activities of W108Y and W108Q lysozymes against glycol chitin were 17.3 and 1.6% of that of wild-type lysozyme, and their dissociation constants for the binding of a trimer of N-acetyl-D-glucosamine were 7.4 and 309 times larger than that of wild-type lysozyme, respectively. For the reversible unfolding at pH 3.5 and 30 degrees C, W108Y and W108Q lysozymes were less stable than wild-type lysozyme by 1.4 and 3.6 kcal/mol, respectively. As for the pKa of Glu35, the values for W108Y and W108Q lysozymes were found to be lower than that for wild-type lysozyme by 0.2 and by 0.6 pKa unit, respectively. The pKa of Glu35 in lysozyme was also decreased from 6.1 to 5.4 by the presence of 1-3 M guanidine hydrochloride, or to 5.5 by the substitution of Asn for Asp52, another catalytic carboxyl group. Thus, both the hydrophobicity of Trp108 and the electrostatic interaction with Asp52 are equally responsible for the abnormally high pKa (6.1) of Glu35, compared with that (4.4) of a normal glutamic acid residue.(ABSTRACT TRUNCATED AT 250 WORDS)     

pH dependence of stability of the 10th human fibronectin type III domain: a computational study

We present detailed computational studies based on electrostatic calculations to evaluate the origins of pKa values and the pH dependence of stability for the 10th type III domain of human fibronectin (FNfn10). One of our goals is to validate the calculation protocols by comparison to experimental data (Koide, A.; Jordan, M. R.; Horner, S.; Batori, V.; Koide, S. Biochemistry 2001, 40, 10326-10333). Another goal is to evaluate the sensitivity of the calculated ionization free energies and apparent pKa values on local structural fluctuations, which do not alter the structural convergence to a particular architecture, by using a complete ensemble of solution NMR structures and the NMR average minimized structure of FNfn10 (Main, A. L.; Harvey, T. S.; Baron, M.; Boyd, J.; Campbell, I. D. Cell 1992, 71, 671-678). Our calculations demonstrate that, at high ionic strength, FNfn10 is more stable at low pH compared to neutral pH, in overall agreement with experimental data. This behavior is attributed to contributions from unfavorable Coulombic interactions in a surface patch for the pairs Asp7-Glu9 and Asp7-Asp23. The unfavorable interactions are decreased at low pH, where the acidic residues become neutral, and are further decreased at high ionic strength because of increased screening by salt ions. Elimination of the unfavorable interactions in the theoretical mutants Asp7Asn (D7N) and Asp7Lys (D7K) produce higher calculated stabilities at neutral pH and any ionic strength compared to the wild-type, in agreement with the experimental data. We also discuss subtleties in the calculated apparent pKa values and ionization free energies, which are not in agreement with the experimental data. This work demonstrates that comparative electrostatic calculations can provide rapid predictions of pH-dependent properties of proteins and can be significant aids in guiding the design of proteins with tailored properties.     

Hydrogen bonding and catalysis: a novel explanation for how a single amino acid substitution can change the pH optimum of a glycosidase

The pH optima of family 11 xylanases are well correlated with the nature of the residue adjacent to the acid/base catalyst. In xylanases that function optimally under acidic conditions, this residue is aspartic acid, whereas it is asparagine in those that function under more alkaline conditions. Previous studies of wild-type (WT) Bacillus circulans xylanase (BCX), with an asparagine residue at position 35, demonstrated that its pH-dependent activity follows the ionization states of the nucleophile Glu78 (pKa 4.6) and the acid/base catalyst Glu172 (pKa 6.7). As predicted from sequence comparisons, substitution of this asparagine residue with an aspartic acid residue (N35D BCX) shifts its pH optimum from 5.7 to 4.6, with an approximately 20% increase in activity. The bell-shaped pH-activity profile of this mutant enzyme follows apparent pKa values of 3.5 and 5.8. Based on 13C-NMR titrations, the predominant pKa values of its active-site carboxyl groups are 3.7 (Asp35), 5.7 (Glu78) and 8.4 (Glu172). Thus, in contrast to the WT enzyme, the pH-activity profile of N35D BCX appears to be set by Asp35 and Glu78. Mutational, kinetic, and structural studies of N35D BCX, both in its native and covalently modified 2-fluoro-xylobiosyl glycosyl-enzyme intermediate states, reveal that the xylanase still follows a double-displacement mechanism with Glu78 serving as the nucleophile. We therefore propose that Asp35 and Glu172 function together as the general acid/base catalyst, and that N35D BCX exhibits a "reverse protonation" mechanism in which it is catalytically active when Asp35, with the lower pKa, is protonated, while Glu78, with the higher pKa, is deprotonated. This implies that the mutant enzyme must have an inherent catalytic efficiency at least 100-fold higher than that of the parental WT, because only approximately 1% of its population is in the correct ionization state for catalysis at its pH optimum. The increased efficiency of N35D BCX, and by inference all "acidic" family 11 xylanases, is attributed to the formation of a short (2.7 A) hydrogen bond between Asp35 and Glu172, observed in the crystal structure of the glycosyl-enzyme intermediate of this enzyme, that will substantially stabilize the transition state for glycosyl transfer. Such a mechanism may be much more commonly employed than is generally realized, necessitating careful analysis of the pH-dependence of enzymatic catalysis.     

Binding of calcium ions to bacteriorhodopsin.

Adding Ca2+ or other cations to deionized bacteriorhodopsin causes a blue to purple color shift, a result of deprotonation of Asp85. It has been proposed by different groups that the protonation state of Asp85 responds to the binding of Ca2+ either 1) directly at a specific site in the protein or 2) indirectly through the rise of the surface pH. We tested the idea of specific binding of Ca2+ and found that the surface pH, as determined from the ionization state of eosin covalently linked to engineered cysteine residues, rises about equally at both extracellular and cytoplasmic surfaces when only one Ca2+ is added. This precludes binding to a specific site and suggests that rather than decreasing the pKa of Asp85 by direct interaction, Ca2+ increases the surface pH by binding to anionic lipid groups. As Ca2+ is added the surface pH rises, but deprotonation of Asp85 occurs only when the surface pH approaches its pKa. The nonlinear relationship between Ca2+ binding and deprotonation of Asp85 from this effect is different in the wild-type protein and in various mutants and explains the observed complex and varied spectral titration curves.     

Relationship between electrostatics and redox function in human thioredoxin: characterization of pH titration shifts using two-dimensional homo- and heteronuclear NMR.

The electrostatic behavior of potentially titrating groups in reduced human thioredoxin was investigated using two-dimensional (2D) 1H and 15N nuclear magnetic resonance (NMR) spectroscopy. A total of 241 chemical shift titration curves were measured over the pH range of 2.1-10.6 from homonuclear 1H-1H Hartmann-Hahn (HOHAHA) and heteronuclear 1H-15N Overbodenhausen correlation spectra. Nonlinear least-squares fits of the data to simple relationships derived from the Henderson-Hasselbalch equation led to the determination of pKas for certain isolated ionizable groups, including the single histidine residue at position 43 (pKa = 5.5 +/- 0.1) and a number of aspartic and glutamic acid carboxylate groups. Many of the titration curves demonstrate complex behavior due to the effects of interacting titrating groups, the long range of electrostatic interactions through the protein interior, and, perhaps, pH-induced conformational changes on the chemical shifts. Unambiguous assignment of the pKas for most of the 38 potentially ionizing groups of human thioredoxin could therefore not be made. In addition, there was no clear evidence that Asp-26 titrates in a manner corresponding to that observed in the Escherichia coli protein [Dyson, H. J., Tennant, L. L., & Holmgren, A. (1991) Biochemistry 30, 4262-4268]. The pKas of the active site cysteines were measured, however, with Cys-32 having an anomalously low value of 6.3 +/- 0.1 and that of Cys-35 between 7.5 and 8.6. These pKas are in agreement with proposed mechanisms for redox catalysis of thioredoxin and previously measured pKas within the active site of E. coli thioredoxin [Kallis, G. B., & Holmgren, A. (1980) J. Biol. Chem. 255, 10261-10265]. The stabilization of a thiolate anion at physiological pH can be explained by the interaction of the S gamma of Cys-32 with the amide of Cys-35 observed in the previously determined high-resolution solution structure of reduced human thioredoxin [Forman-Kay, J. D., Clore, G. M., Wingfield, P. T., & Gronenborn, A. M. (1991) Biochemistry 30, 2685-2698].     

In a staphylococcal nuclease mutant the side-chain of a lysine replacing valine 66 is fully buried in the hydrophobic core

The crystal structure of the staphylococcal nuclease mutant V66K, in which valine 66 is replaced by lysine, has been solved at 1.97 A resolution. Unlike lysine residues in previously reported protein structures, this residue appears to bury its side-chain in the hydrophobic core without salt bridging, hydrogen bonding or other forms of electrostatic stabilization. Solution studies of the free energy of denaturation, delta GH2O, show marked pH dependence and clearly indicate that the lysine residue must be deprotonated in the folded state. V66K is highly unstable at neutral pH but only modestly less stable than the wild-type protein at high pH. The pH dependence of stability for V66K, in combination with similar measurements for the wild-type protein, allowed determination of the pKa values of the lysine in both the denatured and native forms. The epsilon-amine of this residue has a pKa value in the denatured state of 10.2, but in the native state it must be 6.4 or lower. The epsilon-amine is thus deprotonated in the folded molecule. These values enabled an estimation of the epsilon-amine's relative change in free energy of solvation between solvent and the protein interior at 5.1 kcal/mol or greater. This implies that the value of the dielectric constant of the protein interior must be less than 12.8. Lysine is usually found with the methylene groups of its side-chain partly buried but is nevertheless considered a hydrophilic surface residue. It would appear that the high pKa value of lysine, which gives it a positive charge at physiological pH, is the primary reason for its almost exclusive confinement to the surface proteins. When deprotonated, this amino acid type can be fully incorporated into the hydrophobic core.     

Titration study of acetylated lysozyme.

To study the interaction between carboxyl groups and amino groups in native lysozyme [EC 3.2.1.17], and to identify the positions and the pK values of the abnormal carboxyl groups, N-acetylated lysozyme was prepared. The acetylation did not affect the molecular shape of the enzyme, but changed six amino groups to a non-ionizable form, leaving one amino group free; this was determined to be Lys 33. In addition, pH titration of the acetylated lysozyme in 0.2 or 0.02 M KCl aqueous solution indicated fewer titratable groups with pK(int) of 7.8 or 10.4 compared with the native protein, though the number of titratable carboxyl groups was not affected by the acetylation. From the pH titration results and structural considerations, the unititratable carboxyl groups were suggested to be Asp 48, Asp 66, and Asp 87. On the other hand, spectrophotometric titration in 0.2 M KCl showed that all three tyrosine residues are titratable in the acetylated protein, although an abnormal tyrosine residue exists in the native state. Tyr 20 was suggested to be untitratable in the pH range of 8-12.6.     

G33D mutant thioredoxin primarily affects the kinetics of reaction with thioredoxin reductase. Probing the structure of the mutant protein

Escherichia coli thioredoxin is a redox-active protein. A mutant protein with an aspartic acid substitution for the largely conserved glycine at position 33 (G33D) in the active site of thioredoxin has been generated to study the effects of a negatively charged residue in the active site of the protein. Despite the close proximity of the negative-charged Asp to the redox active cysteines, the effective concentration of the cysteines does not deviate significantly from that of the wild-type protein. The redox potential (E(o)') measured by the equilibrium between NADPH and the mutant thioredoxin is also close to that of the wild-type. Kinetic measurements of the reaction between thioredoxin and thioredoxin reductase show that G33D mutant and the wild-type proteins have identical kcat values. However, the Km for G33D mutant is approximately 10-fold higher than that for the wild-type protein. In vivo assay of the growth of E. coli strain carrying wild-type or G33D mutant thioredoxin on methionine sulfoxide indicates that the G33D mutant protein is a slower electron donor for methionine sulfoxide reductase. Structural stability of the oxidized protein is not altered by the G33D substitution, as illustrated by the same unfolding free energies studied by urea. The substitution does not show significant change of the near UV and far-UV circular dichroic (CD) and the fluorescence spectra for either the reduced or the oxidized protein. Therefore, the global structure of the G33D protein is not changed. However, the surface of the active site has been altered locally by G33D substitution, which accounts for the above kinetically poor behaviors. A model of G33D structure is constructed based on these studies.     

pH-induced structural changes in bacteriorhodopsin studied by Fourier transform infrared spectroscopy.

Previous C13-NMR studies showed that two of the four internal aspartic acid residues (Asp-96 and Asp-115) of bacteriorhodopsin (bR) are protonated up to pH = 10, but no accurate pKa of these residues has been determined. In this work, infrared spectroscopy with the attenuated total reflection technique was used to characterize pH-dependent structural changes of ground-state, dark-adapted wild-type bacteriorhodopsin and its mutant (D96N) with aspartic acid-96 replaced by asparagine. Data indicated deprotonation of Asp-96 at high pH (pKa = 11.4 +/- 0.1), but no Asp-115 titration was observed. The analysis of the whole spectral region characteristic to complex conformational changes in the protein showed a more complicated titration with an additional pKa value (pKa1 = 9.3 +/- 0.3 and pKa2 = 11.5 +/- 0.2). Comparison of results obtained for bR and the D96N mutant of bR shows that the pKa approximately 11.5 characterizes not a direct titration of Asp-96 but a protein conformational change that makes Asp-96 accessible to the external medium.     

Increasing protein stability by altering long-range coulombic interactions.

It is difficult to increase protein stability by adding hydrogen bonds or burying nonpolar surface. The results described here show that reversing the charge on a side chain on the surface of a protein is a useful way of increasing stability. Ribonuclease T1 is an acidic protein with a pI approximately 3.5 and a net charge of approximately -6 at pH 7. The side chain of Asp49 is hyperexposed, not hydrogen bonded, and 8 A from the nearest charged group. The stability of Asp49Ala is 0.5 kcal/mol greater than wild-type at pH 7 and 0.4 kcal/mol less at pH 2.5. The stability of Asp49His is 1.1 kcal/mol greater than wild-type at pH 6, where the histidine 49 side chain (pKa = 7.2) is positively charged. Similar results were obtained with ribonuclease Sa where Asp25Lys is 0.9 kcal/mol and Glu74Lys is 1.1 kcal/mol more stable than the wild-type enzyme. These results suggest that protein stability can be increased by improving the coulombic interactions among charged groups on the protein surface. In addition, the stability of RNase T1 decreases as more hydrophobic aromatic residues are substituted for Ala49, indicating a reverse hydrophobic effect.     

Stabilization of a fibronectin type III domain by the removal of unfavorable electrostatic interactions on the protein surface

It is generally considered that electrostatic interactions on the protein surface, such as ion pairs, contribute little to protein stability, although they may play important roles in conformational specificity. We found that the tenth fibronectin type III domain of human fibronectin (FNfn10) is more stable at acidic pH than neutral pH, with an apparent midpoint of transition near pH 4. Determination of pK(a)'s for all the side chain carboxyl groups of Asp and Glu residues revealed that Asp 23 and Glu 9 have an upshifted pK(a). These residues and Asp 7 form a negatively charged patch on the surface of FNfn10, with Asp 7 centrally located between Asp 23 and Glu 9, suggesting repulsive electrostatic interactions among these residues at neutral pH. Mutant proteins, D7N and D7K, in which Asp 7 was replaced with Asn and Lys, respectively, exhibited a modest but significant increase in stability at neutral pH, compared to the wild type, and they no longer showed pH dependence of stability. The pK(a)'s of Asp 23 and Glu 9 in these mutant proteins shifted closer to their respective unperturbed values, indicating that the unfavorable electrostatic interactions have been reduced in the mutant proteins. Interestingly, the wild-type and mutant proteins were all stabilized to a similar degree by the addition of 1 M sodium chloride at both neutral and acidic pH, suggesting that the repulsive interactions between the carboxyl groups cannot be effectively shielded by 1 M sodium chloride. These results indicate that repulsive interactions between like charges on the protein surface can destabilize a protein, and protein stability can be significantly improved by relieving these interactions.     

Hydrogen bonding markedly reduces the pK of buried carboxyl groups in proteins

The ionizable groups in proteins with the lowest pKs are the carboxyl groups of aspartic acid side-chains. One of the lowest, pK=0.6, is observed for Asp76 in ribonuclease T1. This low pK appeared to result from hydrogen bonds to a water molecule and to the side-chains of Asn9, Tyr11, and Thr91. The results here confirm this by showing that the pK of Asp76 increases to 1.7 in N9A, to 4.0 in Y11F, to 4.2 in T91V, to 4.4 in N9A+Y11F, to 4.9 in N9A+T91V, to 5.9 in Y11F+T91V, and to 6.4 in the triple mutant: N9A+Y11F+T91V. In ribonuclease Sa, the lowest pK=2.4 for Asp33. This pK increases to 3.9 in T56A, which removes the hydrogen bond to Asp33, and to 4.4 in T56V, which removes the hydrogen bond and replaces the -OH group with a -CH(3) group. It is clear that hydrogen bonds are able to markedly lower the pK values of carboxyl groups in proteins. These same hydrogen bonds make large contributions to the conformational stability of the proteins. At pH 7, the stability of D76A ribonuclease T1 is 3.8 kcal mol(-1) less than wild-type, and the stability of D33A ribonuclease Sa is 4.1 kcal mol(-1) less than wild-type. There is a good correlation between the changes in the pK values and the changes in stability. The results suggest that the pK values for these buried carboxyl groups would be greater than 8 in the absence of hydrogen bonds, and that the hydrogen bonds and other interactions of the carboxyl groups contribute over 8 kcal mol(-1) to the stability.