proteins STRUCTURE O FUNCTION O BIOINFORMATICS

pH dependence of conformational fluctuations of the protein backbone Daniel E. Richman,1 Ananya Majumdar,2 and Bertrand Garcıa-Moreno E3* 1 Department of Physics and Astronomy, Johns Hopkins University, Baltimore, Maryland, 21218 2 Biomolecular NMR Center, Johns Hopkins University, Maryland 21218 3 Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218

ABSTRACT Proton binding equilibria (pKa values) of ionizable groups in proteins are exquisitely sensitive to their microenvironments. Apparent pKa values measured for individual ionizable residues with NMR spectroscopy are actually population-weighted averages of the pKa in different conformational microstates. NMR spectroscopy experiments with staphylococcal nuclease were used to test the hypothesis that pKa values of surface Glu and Asp residues are affected by pH-sensitive fluctuations of the backbone between folded and locally unfolded conformations. 15N spin relaxation studies showed that as the pH decreases from the neutral into the acidic range the amplitudes of backbone fluctuations in the ps-ns timescale increase near carboxylic residues. Hydrogen exchange experiments suggested that backbone conformational fluctuations promoted by decreasing pH also reflect slower local or sub-global unfolding near carboxylic groups. This study has implications for structure-based pKa calculations: (1) The timescale of the backbone’s response to ionization events in proteins can range from ps to ms, and even longer; (2) pH-sensitive fluctuations of the backbone can be localized to both the segment the ionizable residue is attached to or the one that occludes the ionizable group; (3) Structural perturbations are not necessarily propagated through Coulomb interactions; instead, local fluctuations appear to be coupled through the co-operativity inherent to elements of secondary structure and to networks of hydrogen bonds. These results are consistent with the idea that local conformational fluctuations and stabilities are important determinants of apparent pKa values of ionizable residues in proteins. Proteins 2014; 82:3132–3143. C 2014 Wiley Periodicals, Inc. V

Key words: staphylococcal nuclease electrostatics; acid and local unfolding; pH sensitivity; NMR relaxation; hydrogen exchange.

INTRODUCTION Ionizable residues determine many important functional and solution properties of proteins, such as stability, solubility, conformation, dynamics, interactions, catalysis, and sensitivity to pH and to the ionic properties of their milieu.1–3 Many properties of ionizable residues depend on their charged state, which is determined by their pKa. Despite the importance of ionizable groups in biology, fundamental aspects of the molecular determinants of their pKa values are not well understood. Structure-based pKa calculations have contributed important insights into the determinants of pKa values, but accurate pKa calculations remain challenging.4–6 Further physical insight is needed for the development of improved methods for structure-based pKa calculations, especially concerning the relationship between changes in protonation states and changes in the conformation of

3132

PROTEINS

side chains and backbone.7–9 Because ionizable groups are exquisitely sensitive to the properties of their microenvironment, measured pKa values reflect populationweighted averages of ionization equilibria experienced in the different conformational microstates sampled during the relatively long timescale of the equilibrium thermodynamic experiments used to measure pKa values.10 An experimental description of the pH sensitivity of fluctuations between conformational microstates and how they

Additional Supporting Information may be found in the online version of this article. Grant sponsor: NSF; Grant number: MCB-0743422; Grant sponsor: NIH; Grant number: GM-073838. *Correspondence to: Bertrand Garcıa-Moreno E; Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218. E-mail: [email protected] Received 14 February 2014; Revised 25 June 2014; Accepted 4 August 2014 Published online 18 August 2014 in Wiley Online Library (wileyonlinelibrary. com). DOI: 10.1002/prot.24673

C 2014 WILEY PERIODICALS, INC. V

pH Dependence of Conformational Fluctuations

influence pKa values is needed for further development of structure-based pKa calculations. This study is focused on description of how backbone fluctuations are affected by pH and the ionization state of carboxylic groups. It is not focused on contributions from side chain dynamics to pKa values for three reasons. (1) Contributions from side chain dynamics have been described experimentally and data are already available for testing structure-based calculations.11,12 In contrast, we are not aware of any previous study on the effects of pH on fluctuations of the backbone. (2) If the backbone reorganizes in response to an ionization event, the effects related to side chain dynamics of surface ionizable residues are moot. (3) Computational methodology already exists to account for side chain dynamics explicitly in pKa calculations with continuum calculations,13,14 whereas explicit treatment of backbone reorganization in a continuum calculation has not been attempted, perhaps for lack of evidence that such reorganization takes place. pKa values are extremely sensitive to the properties of the microenvironments of the ionizable groups. pKa values measured with an ensemble-based experiment, such as NMR spectroscopy, are apparent pKa values that represent the average of the pKa values experienced by the ionizable moieties in different microstates. Here we test experimentally the hypothesis that the pKa values of ionizable residues in proteins, Asp and Glu in particular, reflect the equilibrium between the fully folded state and microstates in which segments of the backbone are unfolded. The fully folded and the locally unfolded microstates represent limiting cases of the range of microstates that can be sampled by an ionizable group. According to linkage theory15,16 decreasing pH near or below the pKa of carboxylic groups is expected to promote locally unfolded states because the pKa values of Asp and Glu residues in the fully folded states of proteins are usually depressed,17 whereas they can be assumed to be almost normal in the locally unfolded conformation (Fig. 1). In other words, as the concentration of H1 increases upon a decrease in pH below the normal pKa of 4.5 and 4.0 for Glu and Asp in water, respectively, conformational states are promoted that favor binding of H1 to the Asp and Glu side chains. These states are those in which the pKa of Glu and Asp are normal. They are normal as a consequence of local unfolding of the backbone because the ionizable groups are fully exposed to water and not under the influence of the rest of the protein. Consequently, the apparent pKa values of carboxylic groups measured with NMR spectroscopy will reflect the local stability of a given region in the backbone, which in turn determines the probability that the backbone will shift between folded and unfolded states. Because the pKa values of Glu and Asp are different in the fully folded and in the locally unfolded microstates, the equilibrium between fully

Figure 1 Hypothesis: Changes in pH can be used to shift the equilibrium between two conformational states when the pKa values of an ionizable group is different in each conformation. The structure on the left is in the fully folded conformation. The Glu side chain has a depressed pKa and it is ionized. In the structure on the right the Glu is in a locally unfolded region of the protein, its pKa is normal and it is in the neutral state. A decrease in pH would shift the equilibrium towards the structure on the right. The locally unfolded conformation, exaggerated for illustrative purposes, is a model created in PyMOL based on the crystal structure of D1PHS SNase (PDB ID: 3BDC).4 [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

folded and locally unfolded microstates can be modulated by changes in pH; decreasing pH will promote states that contain more locally unfolded segments.18 This hypothesis was tested with a highly stable variant of staphylococcal nuclease (SNase) known as D1PHS SNase. SNase contains 8 Asp, 12 Glu, and 2 His residues, with pKa values ranging from 6.5 to 2 [Fig. 2(A)].4 All have depressed pKa values except His-8, Asp-40, Glu-43, Glu-142, which have normal pKa values, and Asp-21, which has an elevated pKa of 6.5. NMR relaxation and hydrogen exchange measurements were used to characterize the backbone motions promoted by acidic pH down to pH 3.3. NMR relaxation experiments performed at pH 7.4, 4.7, and 3.3 examined the fast fluctuations (ps-ns timescale) of the backbone that might indicate changes in rigidity with changes in pH. At pH 7.4, most of the acidic groups are fully ionized [Fig. 2(B)]. At pH 4.7, about half are at least 90% ionized and some are 50% protonated. At pH 3.3 all but five of the ionizable groups are mostly protonated. Hydrogen-deuterium exchange (HDX) properties of the protein were measured at pH* values of 6.7, 4.5, and 3.9 to characterize the extent to which slower local fluctuations are promoted by acidic pH. These local fluctuations presumably are related to local or sub-global unfolding sufficient to increase the solvent accessibility of backbone amides. The conditions of pH used in our experiments were selected to avoid frequent sampling of globally denatured states, to focus our observations on the conformational fluctuations within the native ensemble. The global thermodynamic stability of D1PHS SNase is 11.9 kcal/mol between pH 5 and 8 (at 298 K, with 0.1M KCl), and declines to 11 kcal/mol at pH 4.5, 10 kcal/mol at pH PROTEINS

3133

D.E. Richman et al.

Figure 2 A: Location of a carbons of ionizable groups on SNase and magnitude of the shifts in the side chain pKa values relative to normal pKa values in water. B: The pKa values of the ionizable groups relative to each other and to the pH conditions at which NMR relaxation experiments were conducted in this study. 4 His-8: 6.56 6 0.01, Asp-21: 6.54 6 0.02, His-121: 5.35 6 0.03, Glu-142: 4.49 6 0.04, Glu-43: 4.32 6 0.04, Glu-52: 3.93 6 0.08, Glu-122: 3.89 6 0.09, Asp-40: 3.87 6 0.09, Asp-146: 3.86 6 0.05, Glu-101: 3.81 6 0.10, Asp-143: 3.80 6 0.10, Glu-135: 3.76 6 0.09, Glu-67: 3.76 6 0.07, Glu-129: 3.75 6 0.09, Glu-57: 3.49 6 0.09, Glu-73: 3.31 6 0.01, Glu-75: 3.26 6 0.05, Glu-10: 2.82 6 0.09, Asp-19: 2.21 6 0.07, Asp-77: ms) timescales in the more acidic conditions of pH. The local unfolding events are slow enough to be outside the range in which relaxation measurements are sensitive. However, we infer from a detailed structural analysis of the colocalization of these phenomena (described below) that segments near Asp and Glu residues experience fluctuations between folded and unfolded conformations and higher flexibility in the fast timescale in conjunction with their titration. Propagation of perturbations towards the middle of helix-1

According to NMR relaxation data, the backbone at both ends of helix-1 (residues 55268) is rigid at pH 7.4 and becomes progressively more flexible at pH 4.7 and 3.3 [Fig. 5(A)]. This pH dependence can be attributed to the titration of Glu-57 (pKa 5 3.5) and Glu-67 (pKa 5 3.8), located at the ends of helix-1, where unfolding of the backbone would allow the Glu residues to experience greater contact with bulk water, an environment more conducive to protonation. Fraying events have been proposed to occur at the ends of this helix4,18,34 and have been identified previously based on hydrogen exchange data at low pH.35 Fluctuations in the middle of helix-1, especially the second turn from residue 61 to 64, were notable in light of fluctuations at the ends because the apparent local stabilities revealed by HDX in this segment show strong pH dependence. The cooperativity of hydrogen bonding in regular helical structure would promote breaking of bonds from the ends toward the middle of the helix in a pH-dependent

manner coupled with the titration-driven fluctuations of Glu-57 and Glu-67. As will be discussed later, we propose that propagated destabilization of this nature also occurs in the b barrel. Why are fast fluctuations apparent in the second turn of helix-1 even at pH 7.4, and why does their location and amplitude vary with pH? The flexibility is centered around Lys-64, and while it may be a unique consequence of this protein’s architecture that the backbone fluctuates at all at pH 7.4, the notable feature is the side chain’s Coulomb or H-bond interaction with Glu-67. Increasing protonation of Glu-67 at low pH may shift the distribution of conformational microstates and change concomitantly where the fluctuations are present. Fluctuations at loops and in helix-3

Two loop segments—residues 77286 between b4 and b5 and residues 1132120 leading into helix-3—exhibit fluctuation properties different from each other [Fig. 5(B)]. The 1132120 segment was flexible in the ps-ns timescale even at neutral pH; many of its amides are not detectable in 1H-15N correlation spectra in water. In contrast, residues in the 77286 segment had high rigidity at neutral pH that may be attributable to several intra-loop backbone hydrogen bonds that stabilize the loops. Both loops appear to become more flexible as pH decreases, fully consistent with the hypothesis that ionizable groups increasingly fluctuate between conformational microstates in concert with increased fluctuations of the backbone. It is noteworthy that the Asp and Glu residues with pKa most different from those of model compounds in water are in irregular elements of secondary structure (Fig. 2).4 The 77286 and 1132120 loops make important crossdomain interactions through a complex H-bond network4,36 that may modulate the pH dependence of loop flexibility. Dead-time HDX at the Thr-120 amide is consistent with the H-bond with the Asp-77 side chain breaking as a consequence of titration of Asp-77 (pKa < 2.2) [Fig. 5(B)]. Cross-domain H-bonding between the side chains of Glu-75 and His-121 can be broken by titration of either group at low pH (Glu-75 pKa 5 3.26, His-121 pKa 5 5.35). This may drive local unfolding of the helix as reported by the strongly pHdependent apparent DGlocal of Leu-125 (which makes a backbone H-bond with His-121). The increased flexibility at the backbone of Gln-123 at low pH may result from the fraying of the end of the helix or from the fluctuating field caused by titration of neighbor Glu-122 (pKa 5 3.9). Fast fluctuations in the middle of helix-3 [residues 126–130, Fig. 5(C)] were pronounced at the two lowest pH conditions that were studied and the residues that exhibited the lowest order parameters changed between pH 4.7 and 3.3. This is likely due to a redistribution of PROTEINS

3137

D.E. Richman et al.

Figure 4 A: Location and stability of hydrogen exchange opening events. Apparent DGlocal values derived from hydrogen-deuterium exchange rate are color coded according to where they fall in the range 0–12 kcal/mol. Bright red: 0–3 kcal/mol; orange: 3–6; green: 6–9; blue: 9–12. Dark red indicates sites exhibiting exchange too fast to detect (faster than 25 min). Light gray indicates residues whose resonances did not decay at all in the 70 h exchange time course. Dark gray indicates Pro or sites with no exchange rate extracted due to overlapping residues. Locations of Asp, Glu, and His are marked as spheres, with their pKa shifts with respect to water color coded according to Figure 2(A). B: Location and severity of pH dependence of the apparent DGlocal derived from HDX rates (between pH* 4.5 and 3.3). The DDGlocal/DpH slopes between pH* 4.5 and 3.3 are color coded according to where they fall in the range from around 0 6 0.5 kcal/mol/pH, signifying negligible pH dependence (yellow), to 5 kcal/mol/pH, signifying strong pH dependence (bright red). Dark red and light and dark gray here have the same meanings as in subfigure A. Black marks residues with rate fitting uncertainties precluding the calculation of local DDG. Locations of Asp, Glu, and His are marked as spheres, with their pKa shifts with respect to water color coded according to Figure 2(A).

3138

PROTEINS

pH Dependence of Conformational Fluctuations

Figure 5 H-bonds associated with notable hydrogen exchange phenomena. Color coding as in Figure 4(B). Labels that refer only to CO or HN moieties use one-letter residue codes. (A) Helix-1, (B) Loop-loop interactions, (C) Helix-3, (D) Helix-2, (E) Active site and b strands 1, 2, 3.

PROTEINS

3139

D.E. Richman et al.

conformational states following the titration of Glu-129 (pKa 3.8), similar to the phenomenon seen in helix-1. N and C termini

The N and C termini of the protein are disorganized in crystal structures (residues preceding Leu-7 and following Ser-141 are not shown and therefore not color coded in the present figures) and behave as expected in the NMR relaxation and HDX experiments. They exhibit low rigidity. Even the segments of termini observed in crystal structures and therefore assumed to be wellordered do not experience pH-modulated flexibility as observed in the rest of the protein. This is true in the N terminus of strand-1 despite the presence of His-8, which titrates with a rather normal pKa of 6.6, and Glu-10, which has a depressed pKa of 2.8. It is also true in the C-terminus of helix-3, where Glu-135 titrates with a depressed pKa of 3.8. This is likely due to their inherently low rigidity in solution at neutral pH. Their fluctuations at acidic pH are possibly even larger and on a slower timescale to which the NMR relaxation measurements are not sensitive. Hydrogen exchange in helix-2

In contrast with helices-1 and -3, NMR relaxation showed little evidence of an increase in amplitude of fast fluctuations in helix-2 with decreasing pH [Fig. 5(D)]. However, nearby perturbations indicated by HDX can be interpreted in terms of the titration of Glu-101 (pKa 5 3.8). The apparent DGlocal reported by the amide of Leu-103 indicates pH sensitivity; it can be exposed by local unfolding of the helix coupled to the titration of Glu-101. The apparent DGlocal reported by the amide of Ile-92 in b5 also indicates pH dependence; it may lose its H-bond to the side chain of Asn-100 because the titration of Glu-101 promotes helix fraying. Alternatively, titration of Asp-95 may be responsible for perturbing the b strand directly and modulating the pH dependence of apparent DGlocal of Lys-71. Destabilization of the b1-b2-b3 face of the b barrel

At acidic pH, increases in amplitude of ps-ns timescale fluctuations along strand b1 and at several sites on strands b2 and b3 were detected [Fig. 5(E)]. HDX measurements indicated that H-bonds bridging the b sheets are broken more at lower pH, in a concerted manner. In addition to the strong pH effect, the geometry of strand b1 is kinked and this may make it prone to unfolding, which might lead to destabilization of the rest of the sheet. Kinks occur at Ile-15, which competes with the amide of Lys-16 for H-bonding with the carbonyl of Leu-24, and at Ile-18, which competes with the amide of Asp-19 for H-bonding with the carbonyl of Thr-22. Resi-

3140

PROTEINS

dues along strand-b1 and the b1-b2 turn have low apparent DGlocal or experience dead-time exchange even at neutral pH. Low protection factors for Leu-14 and Ala17—1.4 and 1.0 respectively—reported by Skinner et al. (2012) for the P117G/H124L variant of SNase are consistent with these observations. The pH-dependent destabilization of the b1-b3 sheet may be modulated by disruption of H-bonds through titration of active-site cluster residues Asp-19 (pKa 5 2.2), Asp-40 (pKa 5 3.9), and Glu-43 (pKa 5 4.3). The crystal structure of the D1PHS variant of SNase reveals that the cluster has many possible Coulomb and hydrogen bond interactions in the region defined by the 19221 turn and the 37254 loop. The NMR relaxation and HDX data suggest that fluctuations in this region involve excursions away from the conformation reflected in the crystal structure. In addition to their effects on the pKa values of catalytically essential residues, it is possible that these loop fluctuations are relevant for cycles of substrate binding and release, as observed in proteins such as dihydrofolate reductase.37 An alternative interpretation of these observations is that fluctuation of Glu67 promotes fraying of the end of helix-1 that perturbs the b1 strand via hydrophobic groups packed between the helix and strand.

DISCUSSION NMR relaxation and hydrogen-deuterium exchange experiments monitored with NMR spectroscopy were used to test the hypothesis that the protonation of Asp and Glu residues on the surface of SNase is coupled to local unfolding processes that affect the microenvironments of the ionizable moieties of these residues. Specifically, the hypothesis posits that when the ionizable moiety of Glu or Asp can exist in at least two different microenvironments, one in which its pKa is normal and one in which it is depressed, a decrease in pH will increase the population of the conformation in which the group has a normal pKa once the pH approaches or drops past the normal pKa of the carboxylic groups. Presumably the conformation in which the carboxylic group has the more normal pKa is one in which the ionizable group has extensive contact with bulk water and is not under the influence of other polar or charged groups of the protein. According to this hypothesis, changes in pH can drive transitions between folded and locally unfolded conformations associated with the unprotonated and protonated states, respectively. The local unfolding promoted by decreasing pH near acidic values should increase amide hydrogen-deuterium exchange and lead to loss of rigidity detectable as increased fast fluctuations by NMR relaxation measurements. NMR relaxation experiments performed at several neutral and acidic pH conditions showed that conformational

pH Dependence of Conformational Fluctuations

fluctuations, mostly in the ps-ns timescale, increase in amplitude with decreasing pH especially near the ionizable groups that titrate in that pH range. Hydrogen exchange experiments, also performed at several conditions of pH but in the pH range where the protein still has high global stability, suggested that the locations of these fluctuations are correlated with the disruption of hydrogen bonds concomitant with titration events. Structural perturbations associated with protonation of ionizable groups were detectable at several timescales. The fluctuation between protonated and unprotonated states promoted by decreasing pH seems to be associated with local unfolding, probably related to the breaking of backbone hydrogen bonds. This entails sufficient conformational rearrangement to allow backbone amides to be exposed to water. These events must take place on the ms timescale or longer. Localized fluctuations in a much faster ps-ns timescale, detected through NMR relaxation experiments, were also promoted by decreasing pH. This likely reflects the increased flexibility at locally unfolded residues or segments, concomitant with the breaking of hydrogen bonds or with the loss of Coulomb interactions that might rigidify local structure. These results have important implications for structure-based pKa calculations: since increased backbone flexibility and local unfolding excursions usually accompany each other, structure-based calculations have to account for the titration-induced changes in conformation, and for the fact that these conformational changes represent diverse phenomena that span wide timescales. Even small changes in pH can affect the location of fluctuations and have significant effects on local stabilities; the populations of conformational microstates and extent of fluctuation between them appear to be extremely sensitive to changes in pH. Some residues exhibited greater flexibility at pH 4.7 than at 3.3. This may result from an increase in population of the alternate microstate resulting from a decrease in pH. In lowering the pH to 4.7 the local stabilities decrease and sampling of alternate states increases, but the population is still weighted toward the fully folded conformation in which the carboxylic groups are mostly unprotonated. At pH 3.3 the local stabilities are even lower and the alternate states representing the protonated conformations become dominant. As discussed above, the extent of concerted, local unfolding is still small at pH 3.3 thus the overall population is still dominated by the mostly native state. Upon further decrease in pH, locally unfolded states at many sites around the protein become increasingly stabilized. Taken together these states constitute concerted unfolding driving the overall transition towards the acidunfolded state. As pH decreases, residues whose backbones experience increased fluctuations may be adjacent to the titrating group in sequence, close to the titrating group via tertiary contacts, or both. In the case of Asp-19, Glu-57, Glu-67, and Glu-129, the backbones of these residues or close

sequence neighbors experience increased fluctuations with decreasing pH. Because of broken Coulomb and hydrogen bond interactions in the case of Asp-19, or perturbed packing in the case of Glu-67, titration of these residues may also promote fluctuations in segments with which they make contact through tertiary interactions, such as the loop from 37 to 54 and the b1 strand, respectively. The titration of other Asp or Glu residues located in the 37254 loop over the same range of pH complicates this interpretation. Similarly, there are interactions between residues in the 1132121 and 77286 loops that are interfered with by the titration of nearby Glu and Asp residues. The hypothesis articulated by Whitten et al.18 describes the ensemble of a protein in terms of nativelike states that are described by the crystal structure and states with alternative local conformations that can be treated thermodynamically as unfolded regions in which the ionizable groups achieve normal pKa. To what extent is this model of the protein ensemble in solution consistent with the conformational fluctuations measured in this study? The experimental data suggest that the structural fluctuations can be small—unfolding on the scale of one to four hydrogen bonds being broken in most cases. This suggests that the alternative conformational states that are populated with decreasing pH are less unfolded than in a model in which this local unfolding is represented by a long random coil segment. To represent the thermodynamic coupling between titration events, local fluctuations, and global stability accurately, it may be sufficient to assume that segments carrying ionizable groups can be modeled in an equilibrium between two relatively distinct states—fully native and locally unfolded. On the other hand, this is probably not a sufficiently subtle model to capture the range of conformational differences that determine electrostatic environments that give rise to different pKa values, or to model acid unfolding processes that involve the aggregate action of many locally unfolded microstates. It is unlikely that the observed structural changes originating from changes in the protonation state of carboxylic groups can be interpreted as consequences of changes governed by direct, Coulomb interactions. Rather, the propagation of the perturbations originating from the change in the charge state of a carboxylic group is more easily described in terms of the effect on segmental stability exerted by the promotion of certain conformational states in another segment. Perturbations may propagate through well-structured elements of secondary structure and their attendant networks of hydrogen bonds. In most cases, titration events that perturb local structure promote the breaking of several hydrogen bonds in a segment. This can be explained in terms of the cooperative nature of hydrogen bonding and stability throughout the protein, especially throughout well-structure elements of secondary structure: perturbation at one site propagates to segmental unfolding, and ultimately to global unfolding. Indeed, the PROTEINS

3141

D.E. Richman et al.

computational analysis of SNase in terms of local fluctuations and global unfolding by Whitten et al. identified Glu-10, Asp-19, Asp-21, Glu-75, Glu-77, Asp-83, Asp-95, Glu-101, Glu-129, and Glu-135 as residues whose titrations are coupled, through local fluctuation, with global unfolding. Our analysis supports this classification by identifying all of the above residues as residues that promote an increase in conformational fluctuations with decreasing pH. The exceptions are Glu-10 and Glu-135, which are in terminal segments, and Asp-21, which has an elevated rather than depressed pKa. The fluctuations exhibited by the rest of this set have notable pH dependence. Asp-19 is part of a network of hydrogen bonds shown by HDX to be broken by titration at low pH and that potentially promotes the unfolding of the b barrel. Similarly, increased fluctuations at Glu-75 and Glu-77 were detected at low pH, and those residues are involved in a hydrogen bond network linking the two sub-domains of SNase. Increased mobility was detected at Asp-83 and its surrounding loop segment at low pH. Although increased fast fluctuations were not detected near Asp-95 and Glu-101, HDX revealed pH-modulated conformational fluctuations on slower timescales in the vicinity of those groups. Finally, both NMR relaxation and HDX detected substantial pH-dependent conformational fluctuation at Glu-129 and its surrounding residues in helix-3. The results from these NMR spectroscopy experiments with SNase support the idea that local stabilities and local fluctuations of the backbone are important determinants of pKa values of surface Asp and Glu residues, and probably of all ionizable groups. These results imply that computational algorithms for structure-based calculation of pKa values that treat explicitly the conformational flexibility of side chains without comparable treatment of the backbone13 will not always capture the physical phenomena that determine pKa values. In fact, for the surface ionizable groups of interest to this study, the role of side chain dynamics as a determinant of pKa values become irrelevant if the conformation of the backbone is known to be coupled to ionization events. More accurate calculations must consider alternate backbone conformations explicitly, and account accurately for electrostatic and non-electrostatic contributions to free energy differences between conformational states. Our results also highlight the importance of constant pH molecular dynamics simulations38–41 with techniques such as replica exchange42–46 or accelerated MD47 as a tool for capturing the coupling between changes in ionization state and conformation. Recent constant pH MD studies by Shi et al.48 and Goh et al.49 with variants of SNase with internal ionizable residues attempted to reproduce the highly anomalous pKa values of these residues. In those calculations the conformation of the protein is allowed to change concomitant with the ionization event of interest. Those studies recapitulate the general idea outlined by Whitten et al.18 suggesting

3142

PROTEINS

an approach for pKa calculations in terms of ensembleweighted averaging of different pKa values corresponding to two different microstates. The conformational fluctuations observed in the present NMR spectroscopy study are much more subtle than that ones observed in the constant pH MD simulations by Shi et al. and Goh et al. because the present studies are focused on the properties of surface ionizable residues that are fully hydrated and under relatively weak electrostatic forces, whereas the constant pH MD studies examined ionizable groups in highly destabilizing situations. We are in the process of characterizing conformational reorganization coupled to the ionization of internal residues in attempts to corroborate or falsify the studies by Shi et al. and Goh et al. What is already clear is that the ability of constant pH MD simulations to handle explicitly the coupling between backbone conformation and ionization events is an extremely important advance in computational methodology. Our experimental studies contribute novel physical insight and data that constitute benchmarks essential for guiding the development of more realistic and accurate computational methods for structure-based calculations of proton binding energetics. ACKNOWLEDGMENTS Thanks to Dr. Konstantin Berlin, University of Maryland, for development of and assistance with the ROTDIF software, and to Dr. Aaron C. Robinson, Johns Hopkins University, for thoughtful discussions and proofreading. All NMR data were acquired at the Johns Hopkins Biomolecular NMR Center. REFERENCES 1. Garcıa-Moreno EB. Adaptations of proteins to cellular and subcellular pH. J Biol 2009;8:98. 2. Gutteridge A, Thornton JM. Understanding nature’s catalytic toolkit. Trends Biochem Sci 2005;30:622–629. 3. Pace CN, Grimsley GR, Scholtz JM. Protein ionizable groups: pK values and their contribution to protein stability and solubility. J Biol Chem 2009;284:13285–13289. 4. Casta~ neda CA, Fitch CA, Majumdar A, Khangulov V, Schlessman JL, Garcıa-Moreno EB. Molecular determinants of the pKa values of Asp and Glu residues in staphylococcal nuclease. Proteins Struct Funct Bioinforma 2009;77:570–588. 5. Chimenti MS, Casta~ neda CA, Majumdar A, Garcıa-Moreno EB. Structural origins of high apparent dielectric constants experienced by ionizable groups in the hydrophobic core of a protein. J Mol Biol 2011;405:361–377. 6. Harms MJ, Casta~ neda CA, Schlessman JL, Sue GR, Isom DG, Cannon BR, Garcia-Moreno EB. The pKa values of acidic and basic residues buried at the same internal location in a protein are governed by different factors. J Mol Biol 2009;389:34–47. 7. Karp DA, Gittis AG, Stahley MR, Fitch CA, Stites WE, GarcıaMoreno EB. High apparent dielectric constant inside a protein reflects structural reorganization coupled to the ionization of an internal Asp. Biophys J 2007;92:2041–2053. 8. Song Y. Exploring conformational changes coupled to ionization states using a hybrid Rosetta-MCCE protocol. Proteins Struct Funct Bioinforma 2011;79:3356–3363.

pH Dependence of Conformational Fluctuations

9. Williams SL, Blachly PG, McCammon JA. Measuring the successes and deficiencies of constant pH molecular dynamics: A blind prediction study. Proteins Struct Funct Bioinforma 2011;79:3381–3388. 10. Di Russo NV, Estrin DA, Martı MA, Roitberg AE. pH-dependent conformational changes in proteins and their effect on experimental pKas: the case of nitrophorin 4. PLoS Comput Biol 2012;8: e1002761. 11. Stafford KA, Ferrage F, Cho J-H, Palmer AG. Side chain dynamics of carboxyl and carbonyl groups in the catalytic function of Escherichia coli ribonuclease H. J Am Chem Soc 2013;135:18024–18027. 12. Tomlinson JH, Ullah S, Hansen PE, Williamson MP. Characterization of salt bridges to lysines in the protein G B1 domain. J Am Chem Soc 2009;131:4674–4684. 13. Song Y, Mao J, Gunner MR. MCCE2: improving protein pKa calculations with extensive side chain rotamer sampling. J Comput Chem 2009;30:2231–2247. 14. Aleksandrov A, Polydorides S, Archontis G, Simonson T. Predicting the acid/base behavior of proteins: a constant-pH Monte Carlo approach with generalized born solvent. J Phys Chem B 2010;114: 10634–10648. 15. Wyman J, Gill SJ. Binding and linkage: functional chemistry of biological macromolecules. Mill Valley, CA: University Science Books; 1990. 16. Isom DG, Cannon BR, Casta~ neda CA, Robinson A, Garcıa-Moreno EB. High tolerance for ionizable residues in the hydrophobic interior of proteins. Proc Natl Acad Sci USA 2008;105:17784–17788. 17. Forsyth WR, Antosiewicz JM, Robertson AD. Empirical relationships between protein structure and carboxyl pKa values in proteins. Proteins Struct Funct Bioinforma 2002;48:388–403. 18. Whitten ST, Garcıa-Moreno E B, Hilser VJ. Local conformational fluctuations can modulate the coupling between proton binding and global structural transitions in proteins. Proc Natl Acad Sci USA 2005;102:4282–4287. 19. Karp DA, Stahley MR, Garcıa-Moreno E B. Conformational consequences of ionization of Lys, Asp, and Glu buried at position 66 in staphylococcal nuclease. Biochemistry 2010;49:4138–4146. 20. Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, Bax A. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 1995;6:277–293. 21. Goddard TD, Kneller DG. Sparky 3. University of California, San Francisco. 22. Fushman D, Cahill S, Cowburn D. The main-chain dynamics of the dynamin pleckstrin homology (PH) domain in solution: analysis of 15N relaxation with monomer/dimer equilibration. J Mol Biol 1997; 266:173–194. 23. Walker O, Varadan R, Fushman D. Efficient and accurate determination of the overall rotational diffusion tensor of a molecule from 15N relaxation data using computer program ROTDIF. J Magn Reson 2004;168:336–345. 24. Berlin K, Longhini A, Dayie TK, Fushman D. Deriving quantitative dynamics information for proteins and RNAs using ROTDIF with a graphical user interface. J Biomol NMR 2013;57:333–352. 25. Pond MP, Majumdar A, Lecomte JTJ. Influence of heme posttranslational modification and distal ligation on the backbone dynamics of a monomeric hemoglobin. Biochemistry (Mosc) 2012; 51:5733–5747. 26. Lee D, Hilty C, Wider G, W€ uthrich K. Effective rotational correlation times of proteins from NMR relaxation interference. J Magn Reson 2006;178:72–76. 27. Jones E, Oliphant TE, Peterson P, et al. SciPy: Open Source Scientific Tools for Python. 2001. Available at: http://www.scipy.org. 28. Lipari G, Szabo A. Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. Theory and range of validity. J Am Chem Soc 1982;104:4546–4559. 29. Tjandra N, Feller SE, Pastor RW, Bax A. Rotational diffusion anisotropy of human ubiquitin from 15N NMR relaxation. J Am Chem Soc 1995;117:12562–12566.

30. Jurt S, Zerbe O. A study on the influence of fast amide exchange on the accuracy of 15N relaxation rate constants. J Biomol NMR 2012; 54:389–400. 31. Alexandrescu AT, Ulrich EL, Markley JL. Hydrogen-1 NMR evidence for three interconverting forms of staphylococcal nuclease: effects of mutations and solution conditions on their distribution. Biochemistry 1989;28:204–211. 32. Bai Y, Milne JS, Mayne L, Englander SW. Primary structure effects on peptide group hydrogen exchange. Proteins Struct Funct Bioinforma 1993;17:75–86. 33. Zhang Y-Z. Protein and peptide structure and interactions studied by hydrogen exchange and NMR. 2005. Philadelphia, PA: University of Pennsylvania. 34. Fitch CA, Whitten ST, Hilser VJ, Garcıa-Moreno E. B. Molecular mechanisms of pH-driven conformational transitions of proteins: insights from continuum electrostatics calculations of acid unfolding. Proteins Struct Funct Bioinforma 2006;63:113–126. 35. Skinner JJ, Lim WK, Bedard S, Black BE, Englander SW. Protein dynamics viewed by hydrogen exchange. Protein Sci 2012;21:996– 1005. 36. Baran KL, Chimenti MS, Schlessman JL, Fitch CA, Herbst KJ, Garcıa-Moreno E B. Electrostatic effects in a network of polar and ionizable groups in staphylococcal nuclease. J Mol Biol 2008;379: 1045–1062. 37. Osborne MJ, Schnell J, Benkovic SJ, Dyson HJ, Wright PE. Backbone dynamics in dihydrofolate reductase complexes: role of loop flexibility in the catalytic mechanism. Biochemistry 2001;40:9846– 9859. 38. Mongan J, Case DA, McCammon JA. Constant pH molecular dynamics in generalized Born implicit solvent. J Comput Chem 2004;25:2038–2048. 39. Baptista AM, Martel PJ, Petersen SB. Simulation of protein conformational freedom as a function of pH: constant-pH molecular dynamics using implicit titration. Proteins Struct Funct Bioinforma 1997;27:523–544. 40. Lee MS, Salsbury FR, Brooks CL. Constant-pH molecular dynamics using continuous titration coordinates. Proteins Struct Funct Bioinforma 2004;56:738–752. 41. Donnini S, Tegeler F, Groenhof G, Grubm€ uller H. Constant pH molecular dynamics in explicit solvent with k-dynamics. J Chem Theory Comput 2011;7:1962–1978. 42. Khandogin J, Brooks CL. Toward the accurate first-principles prediction of ionization equilibria in proteins. Biochemistry 2006;45: 9363–9373. 43. Meng Y, Roitberg AE. Constant pH replica exchange molecular dynamics in biomolecules using a discrete protonation model. J Chem Theory Comput 2010;6:1401–1512. 44. Itoh SG, Damjanovic´ A, Brooks BR. pH replica-exchange method based on discrete protonation states. Proteins Struct Funct Bioinforma 2011;79:3420–3436. 45. Wallace JA, Shen JK. Continuous constant pH molecular dynamics in explicit solvent with pH-based replica exchange. J Chem Theory Comput 2011;7:2617–2629. 46. Sabri Dashti D, Meng Y, Roitberg AE. pH-replica exchange molecular dynamics in proteins using a discrete protonation method. J Phys Chem B 2012;116:8805–8811. 47. Williams SL, de Oliveira CAF, McCammon JA. Coupling constant pH molecular dynamics with accelerated molecular dynamics. J Chem Theory Comput 2010;6:560–568. 48. Shi C, Wallace JA, Shen JK. Thermodynamic coupling of protonation and conformational equilibria in proteins: theory and simulation. Biophys J 2012;102:1590–1597. 49. Goh GB, Laricheva EN, Brooks CL. Uncovering pH-dependent transient states of proteins with buried ionizable sesidues. J Am Chem Soc 2014;136:8496–899.

PROTEINS

3143