Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Oct 21.
Published in final edited form as: J Am Chem Soc. 2026 Apr 27;148(17):18444–18455. doi: 10.1021/jacs.6c04516

High-pH NMR to Identify Macromolecular Hydrogen-Bonds and Foldons

Andrei T Alexandrescu 1, Antonio J Rua 2, Sameena Shah 3, Daniel Fairchild 4, Irina Bezsonova 5
PMCID: PMC13681069  NIHMSID: NIHMS2212415  PMID: 42044207

Abstract

Hydrogen bond restraints are critical for NMR structure determination, yet their experimental identification can be challenging for marginally stable structures affording insufficient exchange protection in D2O. As an alternative, we explored the use of NMR between pH 10 and 11, conditions that promote rapid exchange, for identifying backbone amide protons involved in H-bonds. We analyzed ~750 amide sites distributed across ten proteins with known structures. Survival of amide protons at high pH in standard 2D 1H–15N HSQC spectra for 15N-labeled proteins in H2O, or TOCSY for unlabeled proteins, identifies H-bonds with ~91% accuracy, exceeding the ~80% accuracy of traditional H/D exchange experiments in D2O. For two α-helical coiled coils and three globular proteins, we performed alkaline unfolding experiments taking advantage of amide NMR signal attenuation from unstructured segments. Increasing the pH led to a progressive loss of native amide proton NMR signals, revealing an unfolding hierarchy where “foldons” remaining at the highest pH had the most persistent H-bonds under EX1 exchange conditions. The foldons observed at high pH are consistent with partially folded structures previously characterized near neutral pH by native state hydrogen exchange, equilibrium unfolding, and protein fragment studies. For β-sheet proteins, foldons correspond to regions with high inter-residue contact density, whereas in coiled coils they demarcate regions with high α-helical propensity. High-pH NMR experiments provide a sensitive, fast, and broadly applicable approach to map H-bonding. Additionally, they offer the opportunity to explore uncharted protein dynamics and unfolding pathways under basic pH conditions.

Graphical Abstract

graphic file with name nihms-2212415-f0001.webp

INTRODUCTION

Hydrogen bonds (H-bonds) are cornerstones of protein structures contributing to their stabilities and folding specificities, as well as forming regular interaction patterns such as α-helix and β-sheet secondary structures.1–3 In NMR structure determination, H-bond restraints greatly improve conformational convergence even when supplemented with only sparse experimental data.4,5 By contrast, without explicit H-bond constraints NMR structures rarely approach the accuracy of the gold-standard X-ray structures, or even of Alpha Fold models.6

Historically, H-bonds were included in NMR structure calculations only when validated by hydrogen–deuterium (H/D) isotope exchange in D2O, where amide proton protection from solvent was inferred to indicate H-bonding.7 This protection requirement is overly stringent as exchange depends on both H-bonding and stability,8–10 with weak or transient H-bonds failing to afford exchange protection. Computational approaches such as ANSURR infer and iteratively include H-bonds during cycles of NMR structure refinement but ultimately depend on validation against experimentally determined H-bond data sets.5 H-bonds can be detected directly by NMR using long-range HNCO experiments that establish through-H-bond h3JNC′ couplings.11,12 These experiments identify both H-bond donors and acceptors but depend on the measurement of very small 0.3–1 Hz couplings that require long delays (~133 ms) to detect, and therefore typically need perdeuterated samples for molecular weights above 5–10 kDa.13 These factors combine to make the use of long-range HNCO experiments for determining H-bonds for NMR structure determination rare, with most studies still relying on H/D exchange in D2O to identify H-bonded sites. Because H/D exchange depends on structural stability as well as environmental variables such as temperature and pH, protection can be undetectable even close to optimal conditions: low temperature, and near the ~pH 4.5 acidic minimum for exchange.14 Examples include the unstable SN-OB fragment of staphylococcal nuclease,10 the small protein CspA in the absence of stabilizing osmolytes,15 as well as several small zinc fingers that are a recent interest of our lab.16–18 Particularly problematic are partially folded proteins and folding intermediates where H-bonds rarely persist long enough to afford detectable protection.19

To overcome the limitations of H/D isotope exchange that amide protons need to survive for at least a few minutes to be detected, we examined whether high-pH NMR could serve as an alternative method to identify H-bonds. Since HX is base-catalyzed, rates increase 10-fold with every pH unit above the pH 4.5 minimum for exchange.20 For structured H-bonded sites, the mechanism of exchange shifts from the EX2 limit, where protection depends on the stability of the H-bonded structure, toward the EX1 limit where protection depends on the rates of H-bonds breaking at high pH.20 However, even approaching the EX1 limit at pH 10–11, H-bonding will confer amide exchange protection compared to solvent exposed sites. Using a set of ~750 amide proton sites from ten proteins we established that survival of amide protons identifies H-bonds with high >90% accuracy at the pH 10–11 conditions that promote rapid hydrogen exchange. Moreover, we find that in three globular protein examples and two coiled coils, NMR data at progressively higher pH, reveal a hierarchy of structural stabilities prior to complete alkaline unfolding. This unfolding hierarchy agrees with that established for the proteins by other methods such as equilibrium denaturation, mutagenesis, and native-state HX analyses. Because the high-pH experiments employ sensitive and routine NMR experiments such as 1H–15N HSQC (or TOCSY for unlabeled samples) they are inexpensive, fast, straightforward, and broadly applicable to probe hydrogen bonding. High-pH NMR offers the opportunity to distinguish folded (protected) and unfolded (unprotected) segments in proteins containing a mixture of structured and disorder regions, and to access largely unexplored information on molecular dynamics and unfolding under alkaline conditions.

RESULTS

High pH Causes Loss of Amide Proton NMR Signals in H2O Due to Fast Solvent Exchange

In typical H/D isotope exchange experiments a lyophilized protein sample is dissolved in D2O. Over time, since 1H NMR experiments do not detect 2H, amide proton NMR signals disappear as the exchange-labile hydrogens on the protein are replaced with deuterons from the solvent.21,22 Isotope exchange is effectively irreversible, since 1H isotopes from protein concentrations typically in the submillimolar range are diluted into a pool of 110 molar D atoms from the D2O solvent.

Amide proton NMR signals can also be lost for proteins in H2O through a variety of mechanisms. Figure 1A shows 1H–15N HSQC spectra of the unstructured Alzheimer’s-associated Aβ(40) peptide in H2O at increasing pH values. At pH values below 6.7 (as far down as pH 2.0), all of the 39 expected amide proton crosspeaks from the sequence are observed at a temperature of 20 °C (Figure 1). The first residue has an α-amine nitrogen and is not observed. As the pH is raised by one pH unit from 6.7 to 7.7, most of the amide protons from G, S and T residues, with 15N chemical shifts below 120 ppm in the 1H–15N HSQC spectrum, are lost. A further increase to pH 8.4 causes the loss of all the signals except for that of the C-terminus V40 (pink arrow), which has the slowest intrinsic exchange (kint) rate predicted from the amino acid sequence.23 The results exemplify that the majority of amide protons signals in unstructured proteins and peptides typically disappear over a narrow range of about 1.5 pH units near the physiological pH of 7.4, as seen in numerous other examples such as the intrinsically disordered protein α-synuclein or small unstructured peptides.14,24 In fact, protein NMR experiments are typically done at a slightly acidic pH in the range 6.0–6.8 and/or low temperatures, precisely to avoid loss of amide protons through fast exchange.25,26 Although amide protons at high pH are usually only lost from unstructured regions, these segments can be crucial for obtaining complete sequential NMR assignments, a requirement for comprehensive NMR studies.

Figure 1.

Figure 1.

Protein hydrogen exchange in the EX1 limit. (A) Selected spectra from a pH titration of 15N-labeled Alzheimer’s Aβ(40) peptide at 20 °C. With increasing pH, amide protons from the unstructured Aβ monomers disappear in H2O due to increasingly accelerated kint hydrogen exchange rates. (B) Intrinsic exchange rates (kint) for the Aβ(40) at various pH values in H2O at a temperature of 20 °C, predicted from the amino acid sequence with the program SPHERE.23 Each point at the given pH represents the kint value averaged over the peptide sequence (excluding the N- and C-term residues). Note that the y-axis (kEX) is logarithmic so that kint increase 10-fold with every pH unit. On the left side of the plot are shown kEX ranges for the disappearance of amide proton NMR signals by the various mechanisms described in the main text. The Linderstrom-Lang equation for HX (eq 1) was used to simulate the effects of various closing rates (kcl) on HX using the kint data for Aβ(40). In all these simulations kcl = kop, and therefore Kop = 1. The simulations show that the closing rate kcl determines the pH value at which the rate plateau corresponding to the EX1 limit of HX is reached. For a typical protein folding rate (kcl = 10 μs, green curve) the EX1 plateau is reached at about pH 10.5. We did additional simulations keeping kcl constant and varying kop (Figure S1).

Figure 1B shows the pH dependence of kint rates calculated for the Aβ(40) amino acid sequence with the program SPHERE (https://spin.niddk.nih.gov/bax/nmrserver/sphere/ accessed on Dec 21, 202523). Above the minimum for exchange near pH 4, the intrinsic exchange rate (dependent only on amino acid sequence, pH, and temperature) is expected to increase 10-fold with every unit increase in pH; therefore, a linear relationship is expected between log(kint) and pH. In the left-side inset for Figure 1B, ranges of kEX values are matched with different types of mechanisms for loss of amide proton NMR signals in H2O. On the slowest time scale of about 0.1–10 s, amide proton NMR signals can be attenuated through transfer of solvent suppression. Under these conditions, NMR signals from H2O and labile protein amide protons are in slow exchange on the chemical shift time scale but in fast exchange on the T1 relaxation time scale. During the T1 relaxation time, the necessary suppression of the H2O signal is partially transferred to exchangeable hydrogens on the protein. The effect is most pronounced using PRESAT type pulse sequences,27 but even more selective solvent suppression schemes, such as those using gradient dephasing, can perturb signals on the protein through exchange, albeit to a reduced extent.28 Next, time scales faster than 10 ms will cause loss of NMR signal because the amide proton residence time on the protein becomes shorter than the INEPT period (typically (1/4JHN) = 2.5 ms) used to establish coupling to the nitrogen. Experiments that use cross-polarization (CP) rather than INEPT to transfer magnetization from 1H to 15N such as CP-HISQC can circumvent some of these problems and detect faster exchanging amide protons.29 In another mechanism, amide proton exchange with time scales between about 1 ms and 1 μs will cause loss of signal due to T2-exchange broadening contributions. Finally, on the fastest time scales shorter than 1 μs, assuming a typical chemical shift difference of 1–3 ppm or 500–3000 Hz between protein amide and solvent protons on 500 to 1000 MHz instruments, the protein amide and solvent protons will be in fast exchange on the chemical shift time scale. The population-weighted average chemical shift in this case will be dominated by the large excess of solvent compared to the protein and tend toward the ~4.7 ppm chemical shift of H2O. Taken together, these effects combine to cause a progressive loss of amide proton signals in standard 1H–15N HSQC experiments as the pH is raised beyond about pH 7.24 Most amide protons in the Aβ(40) example disappear by about pH 8.5 (Figure 1A), corresponding to kEX < 103 s−1 (lifetime <1 ms). This suggests that T2ex broadening and lifetimes shorter than the INEPT coupling period are the dominant mechanisms of NMR signal loss under these conditions (Figure 1B).

Solvent exchange can be slowed down considerably by protein structure, particularly H-bonded structure. The Linderstrom-Lang model of HX in structured proteins presupposes a Closed H-bonded state that is exchange-resistant and an Open exchange-susceptible conformation. In this model HX occurs at the kint rate only from the Open state, when it is generated through conformational fluctuations from the Closed state.30,31

Closed→kopkdOpen→kintExchanged (1)

The Closed state itself makes not contribution to exchange. The reaction scheme can be related to the observed exchange rate kEX by20

kEX=kopkintkcl+kint (2)

The equation has two limiting cases. In the EX2 (bimolecular exchange) limit at low pH, kcl≫kint

kEXEX2=kopkintkcl (3)

Since kint can be calculated using programs such as SPHERE,23 the observed exchange rate can be used to calculate8,9,15,20–22 the equilibrium constant for unfolding describing the concentrations of Open and Closed states Kunf=kop/kcl.

In the second limiting case EX1 (monomolecular exchange) at high pH, kcl≪kint and eq 2 reduces to

kEXEX1=kop (4)

Thus, in the EX1 limit at high pH, the observable exchange rate depends on the opening rate describing the breaking of H-bonded structure to allow OH−-catalyzed exchange of exposed amide protons.

Simulations of the effects of different kcl rates calculated from eq 2 on kEX, with the kint rates obtained for Aβ(40) from the SPHERE program,23 are shown in Figure 1B. The simulations show that the pH at which HX enters the EX1 limit, characterized by a pH-independent plateau, depends on the kcl rate. For a protein that folds without intermediates kcl can be equated to the protein folding rate since the entire structure is formed in a single highly cooperative reaction. Numerous kinetic protein folding studies have demonstrated that evolutionary constraints have selected for proteins to fold in a highly cooperative manner without intermediates and on extremely fast time scales.32–34 A survey of folding rates for two-state proteins normalized to a uniform temperature of 25 °C, found values in a range between 0.5 × 103 and 2 × 105 s−1.35 A larger survey of 108 proteins similarly found an average protein folding rate of 0.4 × 103 s−1 for two-state folders.36 Thus, based on these folding rates, most typical proteins should be well in the EX1 exchange limit by about pH 10 to 11, according to the simulations in Figure 1B.

Similar simulations (Figure S1) show that decreasing the unfolding rate kop while keeping kcl constant has the effect of shifting the pH-dependent exchange curves to slower kEX values. At low pH in the EX2 limit, decreasing kop while keeping kcl constant decreases Kop=kop/kcl leading to slower exchange. At high pH in the EX1 limit, lower kop rates lead to slower kEX rates.

Thus, OH−-catalyzed exchange accelerates rates by a factor of 10 with every unit increase in pH. H-bonded structure protects amide protons even at high pH, however, leading to a plateau in the pH-dependence of exchange rates as the process becomes governed by H-bond breaking under EX1 conditions. Exchange protection due to H-bonded structure can be significant. For example, using H/D isotope exchange in D2O we measured kEX rates as slow as 10−4 s−1 in the proteins SN and LysN at pH 11.0,21 a factor 109-slower than the kint rates predicted from the amino acid sequences of the proteins. Even much lower protection allows amide protons participating in H-bonds to persist in NMR spectra obtained under alkaline conditions (pH > 10). Thus, high-pH NMR data should be useful for identifying amide protons participating in H-bonds.

Amide Protons that Survive at High pH Accurately Identify Hydrogen Bonds

As described above, protection from solvent exchange under EX1 conditions at high pH implies H-bonded structure. We reasoned that survival of amide protons in 1H–15N HSQC spectra at high pH might provide an alternative way to identify H-bonds in structures that are too unstable to give measurable solvent exchange protection in D2O, under EX2 conditions favored by neutral or slightly acidic pH. Compared to the more conventional H/D studies, to our knowledge there has not been an attempt to account for HX protection under EX1 conditions in terms of structural properties such as H-bonding. During the course of studies in our lab, we collected pH data on several proteins and obtained additional new data on others (Figures 2 and S2–S9 and Tables S1–S10). This yielded a data set of some 750 amide proton sites from 10 proteins with X-ray or cryoEM structures to investigate the relationship between HX protection at high pH and H-bonded structure. As illustrated in Figure 2 for ubiquitin and the kinesin neck fragment comprised of an unfolded N-terminus and a C-terminal parallel coiled coil dimer, a subset of the 1H–15N HSQC correlations observed near neutral pH, survived in the spectrum under basic conditions at pH > 10.0. Similar results were observed for the three globular proteins CspA, LysN and SN, in an earlier publication.21 Structural analyses showed that most of the protected amide protons were involved in protein H-bonding whereas unprotected amide protons did not participate in H-bonding. Exceptions in the data were amide protons not protected despite participating in H-bonds, which we categorized as “False Negatives” (FN); and amide protons still detectable at high pH despite not being H-bonded, which we designated “False Positives” (FP). We expanded our NMR analysis to 10 proteins (Figures 2 and S2–S9) and analyzed the data on the persistence of amide protons at high pH in terms of H-bonding using X-ray, cryoEM, or NMR data for the proteins (Tables S1–S10).

Figure 2.

Figure 2.

Survival of H-bonded amide protons at high pH. Representative data for (A) ubiquitin at pH 6.0 and 10.0, and (B) the kinesin neck fragment coiled coil at pH 6.4 and 10.3. The ubiquitin NMR spectra were obtained at a temperature of 25 °C while those for the kinesin neck coiled coil were at 37 °C. Peaks that survived at high pH are labeled green, those that did not are labeled red. NMR signals from side chains are labeled orange. Additional data for 8 proteins are shown in Figures S2–S9 and the relationship between amide proton NMR signal survival at high pH and H-bonding is summarized in Tables S1–S10.

The overall results for ~750 amide protons in the 10 proteins are summarized in Table 1. The presence or absence of amide proton NMR signals between pH 10 and 11 can be used to ascertain H-bonds (protected amides) or the absence of H-bonds (unprotected amides) with ~91% accuracy. About 6% of amide protons are FP, surviving at high pH even though they are not H-bonded. Some 3% are FN, not being detected at high-pH despite being H-bonded. Since all of the proteins studied also had data on isotope exchange protection in D2O at slightly acidic pH, we also performed the H-bonding analysis with this data set (Table 1). The accuracy of H-bond identification from the traditional D2O protection data was 80%, slightly worse than that from the high-pH approach. The D2O data set had a higher proportion of both FP (8%) and FN (12%) amide protons. The fraction of FP amides may be due to the threshold to define D2O protection being set too low (typically survival of NMR signals after D2O incubation for 5 min-2 h, near 25 °C, at pH values between 5 and 6). Increasing the D2O incubation time would reduce cases of the FPs but increase the FNs as marginally stable structure would fall under the raised protection threshold. The higher proportion of FN amides in the D2O exchange data sets, is likely to reflect H-bonded structure that is too unstable to afford protection. This is exemplified by strand β5 of the protein CspA where protection cannot be observed in D2O, unless the protein is stabilized with 0.2 M concentrations of the osmolyte TMAO.15 Indeed, the occurrence of such marginally stable structure was a primary factor motivating us to examine the use of amide proton survival in high-pH NMR as an alternative method for the identification of H-bonds.

Table 1.

Accuracy of H-Bond Identificationa

protein crosspeak survival at high pH 1H to 2H isotope exchange in D2O


# HN T pH FP pH FN pH Acc pH (%) # HN T HX FP HX FN HX Acc HX (%)
Ubq 76 70 5 1 92.1 76 57 10 9 75.0
Kin neck 53 50 0 3 94.3 54 46 0 8 85.2
CspA 62 58 3 1 93.6 66 52 2 12 78.8b
LysN 109 103 2 4 94.5 110 94 4 12 85.4
SN 122 112 4 6 91.8 122 93 4 25 76.2
P22-iD 120 97 20 3 80.8 117 86 26 5 73.5
CUS3-iD 104 96 6 2 92.3 107 86 8 13 80.4
GCN4p 32 31 0 1 96.9 32 28 2 2 87.5
BPTI 50 45 5 0 90.0 54 49 5 0 90.7
SecA-Ct 21 19 2 0 90.5 21 18 2 1 85.7
total 749 681 47 21 90.9 759 609 63 87 80.2
a

# HN (number of backbone amide probes), T pH (true values—number of protected amides that are H-bonded + unprotected amides that are not H-bonded), FP pH (false positives—protected amides that are not H-bonded), FN (false negatives—unprotected amides that are H-bonded), remaining columns—same, but for HX isotope exchange protection in D2O. The % accuracy was calculated as [(T)/(T + FP + FN)] × 100.

b

For CspA isotope exchange in D2O containing the stabilizing osmolyte 0.2 M TMAO, the corresponding numbers are T: 54, FP: 8, FN: 3, Acc: 83%.

We used a range between pH 10 and 11 to identify native state H-bonds for the proteins in this study rather than a precise pH value, since we reasoned the predicted 10-fold difference in intrinsic exchange rates over this range would have a small effect on the identification of H-bonds with opening rates several orders of magnitude slower than the intrinsic rate. The lower end of this range (pH 10) might be preferable for future studies since there is less of a chance of unfolding the protein and it avoids deprotonation of lysine side chains (pKa ~ 10.4) that could cause chemical shift changes. For weakly stable proteins, it may be possible to identify H-bonds from persistence of amide protons at even lower pH values than the ones used here. The example Figure 1A, shows that most amide proton signals from the unfolded Aβ(40) peptide are lost by pH 8.5. Thus pH values as low as 8.5 could be used to identify H-bonds though surviving amide proton NMR signals, although lower pH could present a potential trade-off of fewer FN cases (lost NMR signals for amides that are H-bonded) at the expense of more FP cases (surviving signals for amides that are not H-bonded).

Foldon Identification from Amide Proton Survival in High-pH Unfolding Studies

In contrast to D2O exchange experiments under a single set of conditions, the high-pH approach is tunable across a range of exchange rates, from the slowest near neutral pH to the fastest at basic pH. This made us reason that looking at amide exchange protection over a series of increasing pH values would allow us to investigate the stability hierarchy of structural elements within a protein,37 up to the most stable secondary structure at alkaline pH, before base-induced complete unfolding of the protein. As an alternative, we also explored the use of increasing temperature near pH 6 to identify the most persistent H-bonded structure. In addition to the possibility of precipitation due to heating, thermal unfolding experiments had the drawback of giving a combination of 1H–15N HSQC NMR signals from the remaining structured molecules, and unfolded molecules with surviving amide proton NMR signals at slightly acidic pH 5 to 6 conditions near the minimum for intrinsic solvent exchange. In this regard, the high-pH 1H–15N HSQC experiments have the advantage of effectively filtering out signals from unstructured polypeptide segments through fast exchange. Above ~pH 10, exchange rates are some thousand-fold larger than those compatible with detecting signals from unfolded proteins or segments of proteins (Figure 1, at pH below 8), so only amide protons participating in stable H-bonded structure survive in the spectrum at pH 10–11.

Figure 3 shows representative 1H–15N HSQC pH titration data for two proteins: the GCN4p parallel coiled coil dimer,38 and the P22iD I-domain from the bacteriophage P22, that has a globular structure based on a six-stranded β-barrel with a smaller accessory subdomain.39 As the sample pH is raised above 10, there is an increasing loss of amide protons from the spectra suggesting a progressive loss of structure. We infer that the amide protons surviving at the highest pH, correspond to the most persistent elements of H-bonded structure, and refer to these species as “foldons” in analogy to partially unfolded quasi-autonomous units of H-bonded structure detected in native state hydrogen exchange (NSHX) experiments.9 Whereas NSHX experiments detect foldons based on the stability of residual structure to unfolding near neutral pH under EX2 conditions,9 the foldons described in this work at high pH under EX1 conditions, correspond to the portions of the structure with the slowest unfolding rates. The distinction between folding stability and unfolding rates becomes effectively moot if protein folding rates are uniformly fast, as observed for several small proteins,35,36 since then the unfolding rates would then be the principal determinant of folding stability. The interpretation of the disappearance of amide protons in terms of structural stability is supported among other data by comparison of the protein SN and its less stable fragment SN-OB. The SN-OB fragment retains the OB-fold structure of the parent protein SN but is missing the two C-terminal α-helices.40 The structure and most of the sequence of the SN-OB fragment is the same as that of the conserved portion of the parent protein SN,40 but the stability of the SN-OB fragment is 2.4 kcal/mol versus 6.4 kcal/mol for SN.10 Consistent with the differences in stability, at the same temperature of 25 °C, 1H–15N HSQC signals from amide protons in the SN-OB fragment can only be detected up to pH 10.2 while those from the more stable parent protein SN are observed up to pH 11.0.21 In addition to the aforementioned proteins GCN4p and P22iD, we obtained high-pH unfolding data for a second coiled coil dimer—the kinesin neck fragment, as well as the globular proteins CUS3id and ubiquitin. A summary of the pH values at which amide protons are last detected in the five proteins is given in Table S11. All five proteins show a progressive loss of amide protons with increasing pH, suggesting a stability hierarchy of partially unfolded states before complete unfolding at high pH. This is not the case for all of the proteins we looked at. For SN the surviving amides at pH 11.0 correspond to those H-bonded in the X-ray structure (Figure S4 and Table S5). By pH 11.3, all remaining amide protons are lost from the spectrum (not shown), in an apparently highly cooperative global unfolding process.

Figure 3.

Figure 3.

Selected high pH 1H–15N HSQC spectra ued to identify residual H-bonded foldon structure. (A) Spectra of the GCN4p coiled coil from pH 10.2 to 13.0. (B) Spectra of the globular P22iD protein from pH 9.6 to 12.8. Each spectrum is annotated with labels for peaks that disappear above the stated pH. Above pH 13.0 for GCN4p and pH 12.8 for P22iD 1H–15N HSQC crosspeaks are no longer seen. The colors for the peak labels correspond to the pH ranges given in Figure 4.

Foldons Can Consist of Sequential or Nonsequential Structural Motifs

To better understand the mechanisms of amide proton loss with increasing pH, we mapped the data (Table S11) on the structures of the 5 proteins studied (Figure 4). For the GCN4p coiled coil there is a gradient from the N-terminus where amide protons survive only at moderately basic pH to the C-terminus where amide protons survive at the highest pH value (Figure 4A). The 19–27 segment that survives up to pH 13.4 (the highest pH value before all NMR signals disappeared), corresponds to the coiled coil ‘trigger site’ comprised of residues 18–31.41 Trigger sites are segments of coiled coil sequences that have the highest propensity for α-helical structure in the monomers, nucleating assembly and subsequent zippering of the coiled coil structure.41,42 It is interesting to note in this regard, that the segment of the protein that is the first to fold is also the last to unfold—both can be accounted for by the proclivity of the trigger site to form H-bonded α-helical structure.

Figure 4.

Figure 4.

Structural analysis of amide proton persistence at high pH. Mapping of the pH values at which amide protons are last detected onto the structures of: (A) GCN4p coiled coil (PDB 2ZTA), (B) kinesin neck coiled coil fragment (PDB 8TT7), (C) P22iD (PDB 2M5S), (E) CUS3iD (PDB 6MNT), (F) ubiquitin (PDB 1UBQ). We used NMR structures of the isolated domains of P22iD and CUS3iD because the cryoEM structures of the corresponding domains form intact phage capsids have additional H-bondeds involved in capsomer assembly. Note that the pH scales are different in each case since the proteins have distinct stabilities and unfolding cooperativities. Amides persisting at the highest pH values are indicated by brackets for the coiled coils (A,B) and arrows for the globular protein (C–E). Correlations of the pH values at which amide proton NMR signals are last observed (gray bars) with contact densities (blue points, green lines) for the globular proteins P22iD (F), CUS3iD (G) and ubiquitin (H). The contact densities (# of Cα–Cα contacts <5 Å per residue) were smoothed by averaging over a 3-residue window.

For the second kinesin neck fragment coiled coil, there is also a gradient of amide proton NMR signal persistence from the rapidly exchanging N-terminus to the more exchange-resistant C-terminus (Figure 4B). Similar to GCN4p, the 352–365 segment of the kinesin neck fragment with the amide protons that survive at the highest pH value of 11.0, scores the highest for coiled coil preference with the DEEPCOIL1 program (https://toolkit.tuebingen.mpg.de/tools/deepcoil accessed on Jan 26, 202643). This region may correspond to the trigger site of the kinesin neck fragment.

For the globular protein P22iD, the most persistent amide protons map to noncontiguous segments comprising the six β-strands of a conserved antiparallel β-barrel structure (indicated by arrows in Figure 4C). This foldon is similar to one identified in previous NSHX experiments of P22iD at pH 6.0.22 In the previous study, the six-stranded β-barrel had the greatest stability to unfolding while a subdomain consisting of a short α-helix and three short β-strands had a lower stability, consistent with the hierarchy of structure suggested by the high pH unfolding experiments in this work.

To test whether the pH dependence of amide protons is conserved in structurally related proteins we looked at the CUS3iD domain that has the same fold and function as P22iD44 but only 35% sequence identity.45 Like in P22iD, the six-stranded antiparallel β-barrel structure in CUS3iD (indicated by arrows in Figure 4D) has the highest proportion of amide protons surviving at high pH, with the accessory subdomain consisting of three short β-strands and a one-turn α-helix, immediately below the β-barrel structure showing less persistence. The foldon structure consisting of the core of the six-stranded β-barrel, thus appears to be conserved between the structurally related P22iD and CUS3iD domains. Within the six-stranded β-barrel, strands 1 and 6 have a higher proportion of amide protons that exchange at lower pH in both P22iD and Cus3iD, suggesting a less firm connection between these strands. NMR studies support a disruption of the β1–β6 pairing of CUS3iD under acid-denaturing conditions, that eventually leads to amyloid formation from the residual β-sheet structure of the partially open β-barrel (ATA, unpublished observations). We have not yet investigated if the β1–β6 connection that closes the β-barrel is also susceptible to disruption in P22iD, like in CUS3iD.

Finally, we collected high pH unfolding data for the well-studied protein ubiquitin (Figure 4E). A number of partially folded states of ubiquitin have been reported including the A-state stabilized in alcohol under acidic conditions that only has the β1–β2 hairpin and long α-helix components of the structure,46 the related acidic U′ state with only the β1–β2 hairpin,47 the weakly populated acidic U″ state that preserves most of the native structure but experiences an unfolding of residues 65–68 at the beginning of the C-terminal strand β5,47 and a recently described pressure-denatured folding intermediate that involves an H-bond register shift in the β-sheet pairing of residues 65–73 from C-terminal strand β5 with strands β1 and β3.48 The high pH data for ubiquitin in this work does not support loss of any secondary structure elements, since all of those in the native structure have amides that survive at the highest pH: β1(residues 3–6), β2 (15–18), α (26–30), β3 (44–45), β4 (50), β5 (67–68). Together, these persistent amides appear to form a ladder of secondary structure rungs through the portion of the globular structure closest to the N-terminus (arrow in Figure 4E). By contrast, the portion of the globular structure closer to the C-terminus has no amide protons that survive at high pH. This part of the molecule has a high fraction of basic Arg residues (not shown). Neutralization of the Lys residues between pH 10 and 11 may leave an excess of positively charged Arg residues at pH values >12 that could destabilize structure in this region in a manner specific to ubiquitin at high pH, although this proposed mechanism will need further investigation.

Thus, denaturation experiments monitored by amide proton persistence at high pH show a variety of behaviors. For the two coiled coils GCN4p and kinesin neck, we see a gradual breakdown of structure within a single α-helical secondary structure. For the two related globular proteins P22iD and Cus3iD, the most persistent structure corresponds to the six β-strands that are not contiguous in the sequence but come together in the structure to form a conserved antiparallel β-barrel motif. For ubiquitin, the most persistent structure maps to one side of the β-grasp folding motif that has a relatively neutral to negatively charged electrostatic surface. Finally, for some proteins like SN, all H-bonded amides are lost over a narrow pH range, indicative of a cooperative global unfolding transition.

Density of Inter-residue Contacts Is a Determinant of Foldon Stability

In previous papers, we noted a correlation between hydrogen exchange stability in NSHX experiments and the density of inter-residue contacts in protein structures.21,22 This correlation extends even to amyloids probed by quenched HX, where the most protected amides in amyloids correspond to residues that form the largest number of contacts in the amyloid structure.49 We see the same trend for the high-pH persistence of amide protons in the globular proteins studied in this work (Figure 4F–H). The R-values for correlations between the pH value at which amide protons are last observed and the number of Cα–Cα distance contacts in the structure <5 Å, were 0.68 for P22iD, 0.66 for CUS3iD, and somewhat lower at 0.45 for ubiquitin. For all three proteins, the probability that these correlations could arise by chance is p < 0.0001 at a 95% confidence level. Clearly, Cα–Cα contact density cannot be the only determining factor in the pH persistence of amide protons, since in the coiled coil examples the contact density should be nearly uniform through the structure (Figure 4A,B). As previously noted, the propensity of the polypeptide to fold into α-helix structure appears to be the dominant determinant in the high-pH HX behavior of coiled coils.

DISCUSSION

In this work we looked at the pH-dependence of NMR signals of ~750 backbone amide protons from a database of 10 proteins. We found that the survival of amide proton NMR signals under basic conditions between pH 10 to 11 that promote fast HX, can be used to identify amide proton H-bond donors with an accuracy of about 91%. This compares favorably with the traditional method of inferring H-bonding from isotope exchange protection in D2O, which gave an accuracy of about 80% for the same data set. The lower accuracy of the D2O isotope exchange method is due to a higher number of false negatives—amide protons that are H-bonded in marginally stable structure that offers insufficient protection to be detected in D2O. The alternative high-pH HX method has the advantage of being “tunable” over a range of pH values, as long as these do not disturb the structure of the protein. Conversely, the advantage of the D2O exchange method is that the protein is studied under a fixed set of conditions where the only experimental variable is the substitution of D2O solvent for H2O. Temperature coefficients, measured as the change in chemical shift of a nucleus with temperature (e.g., Δδ(HN)/ΔT) provide and alternative way to identify H-bonds both in folded proteins and loosely structured folding intermediates. In particular, Δδ(HN)/ΔT values more positive than −4.6 ppb/K predict H-bonds with an accuracy of 85–90%.22 The accuracy can be further improved to as high as 95% by combining temperature coefficient data for multiple nuclei such as 1HN and 15N.22 While temperature coefficients have a comparable predictive accuracy to the high-pH method, the latter can also be used to probe the stability of structure to alkaline unfolding over a series of pH values. Another potential application of the high-pH method is as a spectral editing tool. High-pH spectra like the ones illustrated in Figures 2 and 3 could be used to preserve signals from the subset of H-bonded amides in structured regions of the protein, while suppressing those from unstructured sites that often give the strongest NMR signals. This could simplify NOESY, or other types of NMR spectra, for large proteins containing a mixture of structured and disordered segments.

In our analysis of the relationship between amide proton protection at high-pH and H-bonding we observed some examples of both false positives and false negatives. The false negatives occur because the stability of H-bonded structure can be insufficient to afford protection. This prompted us to explore the possibility of assessing the stability hierarchy of structure by shifting conditions to increasingly basic conditions where only the most persistent H-bonded structure survives. We used this approach for five proteins, and except for ubiquitin the residual structure observed to survive at the highest pH values agrees with that found in folding intermediates at neutral pH observed in NSHX experiments or protein fragment studies. This argues that the partially folded species observed at extremes of basic pH result from native-like interactions rather than representing an artifact of specific alkaline denaturation conditions. The five proteins show differences among the pH values at which amide protons start to disappear as would be expected from the different stabilities of the structures involved, and in the range of pH values leading to a signal loss due to the different unfolding cooperativities of the proteins.

False positives manifest exchange protection without apparent H-bonding, based on high-resolution structures of the respective proteins. It is generally accepted that participation in H-bonding rather than seclusion from solvent is the predominant requirement for amide proton exchange protection.26,50 Indeed, it is exceedingly rare for polar amide protons to occur in the interior of protein structures without being H-bonded.26 Interestingly, examples of false positives in our high pH experiments often correspond to false positives in the D2O exchange experiments as well. Examples include the amide protons of I36 and D58 for ubiquitin, as well as other amide protons in the nine proteins we studied (Tables S1–S10). Possible explanations for the absence of H-bonding for these protected amides are errors in the X-ray or cryoEM structures, or contributions from H-bonding to solvent that are not discernible in the structures. Alternatively, studies on model compounds and proteins show that in a small number of cases amide protons lacking H-bonds can still be protected by burial in the structure.51 In our data, the fraction of false positive cases accounts for about 6–8% of the total (Table 1).

For H-bonded sites protected in D2O exchange experiments, the degree of protection does not correlate with H-bonding strength as defined by H-bond distances, Δδ(HN)/ΔT temperature coefficients, or the magnitudes of NMR though-H-bond J-couplings.22,52,53 Rather, the degree of protection appears to be correlated with the stability of the structure to unfolding as parametrized by the number of distance contacts with other residues in the structure.15,21,22,54 We found similar correlations in this work between the pH persistence of amide protons and the Cα–Cα contact density in the structure, at least for proteins with predominantly β-sheet structure (Figure 4F–H). For proteins with a coiled coil α-helical structure where the Cα–Cα contact density is uniform throughout the structure, the amide protons that persist at the highest pH values are those from the residues with the highest sequence propensity to adopt α-helical secondary structure.

CONCLUSION

In summary, we described methods to probe H-bonding and the unfolding dynamics of H-bonded structure using high-pH NMR. The approach does not require specialized labeling schemes or specialized NMR equipment and uses sensitive NMR experiments that can be done with unlabeled (2D 1H-TOCSY) or 15N-labeled (2D 1H–15N HSQC) samples. A fingerprint 1H–15N HSQC spectrum for a protein at each pH value can be collected in a matter of minutes. A series of spectra at progressively increasing pH values can provide information on the H-bonded structure most resistant to alkaline unfolding. Since alkaline unfolding of the structure typically occurs at pH values that promote fast exchange of amide protons, unstructured polypeptides do not contribute to the spectrum. For the proteins examined in this work (except ubiquitin), the residual structure at high pH is consistent with that observed in studies of partial unfolding near neutral pH. This indicates that unfolding at high pH occurs through nonspecific structure disruption. The resulting foldon structures reflect the stability hierarchy of substructures, rather than partial unfolding specific to the conditions used to destabilize the protein. The resulting information on partially unfolded species observed in the high-pH unfolding studies could be relevant to understanding folding mechanisms, protein misfolding related to disease, the stability of pharmaceutical proteins, and conformational transitions important in protein function.

MATERIALS AND METHODS

Protein Samples

Unless otherwise noted all the NMR spectra in this study were obtained at a temperature of 25 °C, with protein concentrations ranging from 0.1 to 1 mM. Exceptions were studies of kin-neck at 37 °C, and BPTI at 30 °C. Protein samples for NMR were prepared by Escherichia coli expression and purified according to previously published conditions as follows: Kin-neck,55 CspA,15,56 LysN,57–59 SN,21 GCN4p,38 P22iD,39 Cus3iD.45 Ubiquitin was expressed in E. coli BL21(DE3) cells using a pET28b(+) vector from Novagen (Madison, WI), with the ubiquitin gene inserted via the Nhe1 and NotI restriction sites. Uniform 15N-isotope labeling of ubiquitin and its purification were done as previously described.60 The ubiquitin sample had 150 mM NaCl and 50 mM sodium phosphate buffer. 15N-labeled Aβ(40) was from rPeptide (Watkinsville, GA). BPTI was purchased as bovine lung aprotinin (Cat Number 616370-M, ≥95% pure) from Sigma (St. Louis, MO). SecA-Ct was custom synthesized to ≥98% purity by Biopeptide (San Diego, CA) and studied in the presence of 1:1 ZnCl2 under previously described sample conditions.61 Unless otherwise noted, the samples contained no added buffers or salts.

The pH of samples contained in 1.5 mL Eppendorf centrifuge tubes was adjusted with small (0.3–3 μL) aliquots of 1 N HCl and NaOH stocks. For smaller pH adjustments, the micropipette tip was inserted in the acid or base stocks without withdrawing liquid and blotted against a clean Kimwipe before inserting it into the sample. All pH measurements were made in Eppendorf centrifuge tubes using a Mettler (Columbus OH) MA235 pH meter equipped with a glass InLab micro pH electrode. After the initial pH measurement, the sample was transferred to the NMR tube and then back to the Eppendorf tube again for a final pH measurement, to minimize pH drift, before inserting the sample back into the NMR tube for data collection. The reported sample pH was taken as the average of measurements before and after NMR experiments, which typically differed by <0.1 pH unit.

NMR Data Acquisition and Analysis

NMR assignments at the various pH values employed in this study were extended with 2D titration experiments using 1H–15N HSQC (pulse sequences gNhsqc for Varian and fhsqcf 3gpph for Bruker) or TOCSY (Bruker pulse sequence dipsi2gpph19) spectra, starting from published NMR assignments near neutral pH. Chemical shift assignment files and annotated fingerprint regions of NMR spectra were obtained from the BioMagResBank (BMRB) and publications as follows: ubq (BMRB 645762), kin-neck (BMRB 5207555), CspA (BMRB 429663), LysN,57 SN (BMRB 405264), GCN4p (BMRB 3002738), P22iD (BMRB 1856639), CUS3iD (BMRB 2526345), BPTI (BMRB 4968), SecA-Ct (BMRB 6289). Data on protection from 1H/2H isotope exchange in D2O was obtained from the literature under the following conditions: ubq (1 h, pH 3.5, 22 °C65), kin-neck (0.5 h, pH 6.1, 37 °C55), CspA (3 min, pH 6.0, 30 °C63), LysN (40 min, pH 6.0, 20 °C58), SN (1 h, pH 5.2, 25 °C21), GCN4p (3 min, pH 7.0, 6 °C66), P22iD (2 h, pH 6.0, 37 °C39), CUS3iD (2 h, pH 6.0, 25 °C45), BPTI (100 min, pH 3.5, 36 °C67), SecA-Ct (5 min, pH 6.8, 25 °C61). H-bonds were calculated with the program HBPLUS68 from structures with the following PDB codes: ubq (1UBQ), kin-neck (3KIN), CspA (1MJC), LysN (1BBU_A), SN (1STN), GCN4p (2ZTA_A), P22iD (8I1V_A and ref 52), CUS3iD (8SKG_A and ref 52), BPTI (1G6X), SecA-Ct (1TM6).

NMR spectra were recorded within 15–30 min of adjusting the pH, with the total experiment time for each pH titrations about 6–12 h. Changes in spectra were reversible upon lowering the pH back to neutral, except for the protein SN after all of the peaks in the 1H–15N HSQC spectrum were lost due to complete unfolding at pH 11.3. In the analysis of amide proton NMR signal loss as a function of pH, crosspeaks were deemed to have survived if they were more than one contour above the noise level.

Calculations of distance contacts between residues were done with the MAPIYA server using a contact cutoff of 5 Å and an intracontact filter set to zero (https://mapiya.lcbio.pl accessed on Jan 18, 202669). Intrinsic HX rates were calculated from amino acid sequences with the program SPHERE (https://spin.niddk.nih.gov/bax/nmrserver/sphere/ accessed on Dec 21, 202523).

Supplementary Material

Supplementary Information

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.6c04516.

Simulations of the pH dependence of HX; high pH NMR and H-bonding data for ten proteins; high-pH unfolding data for five proteins (PDF)

Funding

Supported by NSF MB0236316 and NIH GM076661 to A.T.A. and NIH R35GM156397 to I.B.

ABBREVIATIONS

Aβ(40)

40-residue amyloid beta peptide

FN

false negatives (lost NMR signals for amides that are H-bonded)

FP

false positives (surviving NMR signals for amides that are not H-bonded)

H-bond

hydrogen bond

H/D

hydrogen/deuterium

HSQC

heteronuclear single quantum correlation

HX

hydrogen exchange

NMR

nuclear magnetic resonance

NSHX

native state hydrogen exchange

k cl

closing rate

k int

intrinsic exchange rate

k op

opening rate

K unf

equilibrium constant for protein unfolding

Footnotes

The authors declare no competing financial interest.

Contributor Information

Andrei T. Alexandrescu, Department of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut 06269, United States

Antonio J. Rua, Department of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut 06269, United States

Sameena Shah, Department of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut 06269, United States.

Daniel Fairchild, Department of Molecular Biology and Biophysics, UConn Health, Farmington, Connecticut 06030, United States.

Irina Bezsonova, Department of Molecular Biology and Biophysics, UConn Health, Farmington, Connecticut 06030, United States.

REFERENCES

  • (1).Pauling L The hydrogen bond. The Nature of the Chemical Bond, 3rd ed.; Pauling L, Ed.; Cornell University Press: New York, 1960; pp 449–504. [Google Scholar]
  • (2).Branden C; Tooze J Introduction to Protein Structure, 2nd ed.; Garland Science: New York, 1998. [Google Scholar]
  • (3).Pace CN; Fu H; Lee Fryar K; Landua J; Trevino SR; Schell D; et al. Contribution of hydrogen bonds to protein stability. Protein Sci. 2014, 23, 652–661. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (4).Alexandrescu AT Strategy for supplementing structure calculations using limited data with hydrophobic distance restraints. Proteins 2004, 56, 117–129. [DOI] [PubMed] [Google Scholar]
  • (5).Fowler NJ; Albalwi MF; Lee S; Hounslow AM; Williamson MP Improved methodology for protein NMR structure calculation using hydrogen bond restraints and ANSURR validation: The SH2 domain of SH2B1. Structure 2023, 31, 975.e3–986.e3. [DOI] [PubMed] [Google Scholar]
  • (6).Fowler NJ; Williamson MP The accuracy of protein structures in solution determined by AlphaFold and NMR. Structure 2022, 30, 925.e2–933.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (7).Williamson MP; Havel TF; Wuthrich K Solution conformation of proteinase inhibitor IIA from bull seminal plasma by 1H nuclear magnetic resonance and distance geometry. J. Mol. Biol. 1985, 182, 295–315. [DOI] [PubMed] [Google Scholar]
  • (8).Bai Y; Milne JS; Mayne L; Englander SW Protein stability parameters measured by hydrogen exchange. Proteins 1994, 20, 4–14. [DOI] [PubMed] [Google Scholar]
  • (9).Bai Y; Sosnick TR; Mayne L; Englander SW Protein folding intermediates: native-state hydrogen exchange. Science 1995, 269, 192–197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (10).Alexandrescu AT; Dames SA; Wiltscheck R A fragment of staphylococcal nuclease with an OB-fold structure shows hydrogen-exchange protection factors in the range reported for ″molten globules″. Protein Sci. 1996, 5, 1942–1946. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (11).Cordier F; Grzesiek S Direct observation of hydrogen bonds in proteins by interresidue 3HJNC’ scalar couplings. J. Am. Chem. Soc. 1999, 121, 1601–1602. [DOI] [PubMed] [Google Scholar]
  • (12).Jaravine VA; Alexandrescu AT; Grzesiek S Observation of the closing of individual hydrogen bonds during TFE-induced helix formation in a peptide. Protein Sci. 2001, 10, 943–950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (13).Grzesiek S; Cordier F; Jaravine VA; Barfield M Insights into biomolecular hydrogen bonds from hydrogen bond scalar couplings. Prog. Nucl. Magn. Reson. Spectrosc. 2004, 45, 275–300. [Google Scholar]
  • (14).Bai Y; Milne JS; Mayne L; Englander SW Primary structure effects on peptide group hydrogen exchange. Proteins 1993, 17, 75–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (15).Jaravine VA; Rathgeb-Szabo K; Alexandrescu AT Microscopic stability of cold shock protein A examined by NMR native state hydrogen exchange as a function of urea and trimethylamine N-oxide. Protein Sci. 2000, 9, 290–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (16).Harris RE; Rua AJ; Alexandrescu AT Zinc-Induced Folding and Solution Structure of the Eponymous Novel Zinc Finger from the ZC4H2 Protein. Biomolecules 2025, 15, 1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (17).Harris RE; Whitehead RD III; Alexandrescu AT Solution structure of the Z0 domain from transcription repressor BCL11A sheds light on the sequence properties of protein-binding zinc fingers. Protein Sci. 2025, 34, No. e70097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (18).Lee MS; Gippert GP; Soman KV; Case DA; Wright PE Three-dimensional solution structure of a single zinc finger DNA-binding domain. Science 1989, 245, 635–637. [DOI] [PubMed] [Google Scholar]
  • (19).Smith LJ; Alexandrescu AT; Pitkeathly M; Dobson CM Solution structure of a peptide fragment of human alpha-lactalbumin in trifluoroethanol: a model for local structure in the molten globule. Structure 1994, 2, 703–712. [DOI] [PubMed] [Google Scholar]
  • (20).Krishna MM; Hoang L; Lin Y; Englander SW Hydrogen exchange methods to study protein folding. Methods 2004, 34, 51–64. [DOI] [PubMed] [Google Scholar]
  • (21).Watson E; Matousek WM; Irimies EL; Alexandrescu AT Partially folded states of staphylococcal nuclease highlight the conserved structural hierarchy of OB-fold proteins. Biochemistry 2007, 46, 9484–9494. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (22).Newcomer RL; Fraser LCR; Teschke CM; Alexandrescu AT Mechanism of Protein Denaturation: Partial Unfolding of the P22 Coat Protein I-Domain by Urea Binding. Biophys. J. 2015, 109, 2666–2677. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (23).Zhang Y-Z Protein and Peptide Structure and Interactions Studied by Hydrogen Exchange and NMR, Ph.D. Thesis, University of Pennsylvania, 1995. [Google Scholar]
  • (24).Croke RL; Sallum CO; Watson E; Watt ED; Alexandrescu AT Hydrogen exchange of monomeric alpha-synuclein shows unfolded structure persists at physiological temperature and is independent of molecular crowding in Escherichia coli. Protein Sci. 2008, 17, 1434–1445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (25).Muskett FW Sample preparation, data collection and processing. Protein NMR Spectroscopy: Practical Techniques and Applications, 1st ed.; Lian L-Y; Roberts GCK, Eds.; John Wiley & Sons, 2011; pp 1–21. [Google Scholar]
  • (26).Dempsey CE Hydrogen exchange in peptides and proteins using NMR spectroscopy. Prog. Nucl. Magn. Reson. Spectrosc. 2001, 39, 135–170. [Google Scholar]
  • (27).Grzesiek S; Bax A The importance of not saturating water in protein NMR. Application to sensitivity enhancement and NOE measurements. J. Am. Chem. Soc. 1993, 115, 12593–12594. [Google Scholar]
  • (28).Mori S; Abeygunawardana C; Johnson MO; van Zijl PC Improved sensitivity of HSQC spectra of exchanging protons at short interscan delays using a new fast HSQC (FHSQC) detection scheme that avoids water saturation. J. Magn. Reson., Ser. B 1995, 108, 94–98. [DOI] [PubMed] [Google Scholar]
  • (29).Yuwen T; Skrynnikov NR CP-HISQC: a better version of HSQC experiment for intrinsically disordered proteins under physiological conditions. J. Biomol. NMR 2014, 58, 175–192. [DOI] [PubMed] [Google Scholar]
  • (30).Skinner JJ; Lim WK; Bedard S; Black BE; Englander SW Protein dynamics viewed by hydrogen exchange. Protein Sci. 2012, 21, 996–1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (31).Englander SW; Mayne L; Kan ZY; Hu W Protein Folding-How and Why: By Hydrogen Exchange, Fragment Separation, and Mass Spectrometry. Annu. Rev. Biophys. 2016, 45, 135–152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (32).Dobson CM Principles of protein folding, misfolding and aggregation. Semin. Cell Dev. Biol. 2004, 15, 3–16. [DOI] [PubMed] [Google Scholar]
  • (33).Wolynes PG Evolution, energy landscapes and the paradoxes of protein folding. Biochimie 2015, 119, 218–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (34).Jackson SE How do small single-domain proteins fold? Folding Des. 1998, 3, R81–R91. [DOI] [PubMed] [Google Scholar]
  • (35).Manavalan B; Kuwajima K; Lee J PFDB: A standardized protein folding database with temperature correction. Sci. Rep. 2019, 9, 1588. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (36).Glyakina AV; Galzitskaya OV How Quickly Do Proteins Fold and Unfold, and What Structural Parameters Correlate with These Values? Biomolecules 2020, 10, 197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (37).Wang Y; Alexandrescu AT; Shortle D Initial studies of the equilibrium folding pathway of staphylococcal nuclease. Philos. Trans. R. Soc., B 1995, 348, 27–34. [DOI] [PubMed] [Google Scholar]
  • (38).Kaplan AR; Brady MR; Maciejewski MW; Kammerer RA; Alexandrescu AT Nuclear Magnetic Resonance Structures of GCN4p Are Largely Conserved When Ion Pairs Are Disrupted at Acidic pH but Show a Relaxation of the Coiled Coil Superhelix. Biochemistry 2017, 56, 1604–1619. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (39).Rizzo AA; Fraser LC; Sheftic SR; Suhanovsky MM; Teschke CM; Alexandrescu AT NMR assignments for the telokin-like domain of bacteriophage P22 coat protein. Biomol. NMR Assignments 2013, 7, 257–260. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (40).Alexandrescu AT; Gittis AG; Abeygunawardana C; Shortle D NMR structure of a stable ″OB-fold″ sub-domain isolated from staphylococcal nuclease. J. Mol. Biol. 1995, 250, 134–143. [DOI] [PubMed] [Google Scholar]
  • (41).Steinmetz MO; Jelesarov I; Matousek WM; Honnappa S; Jahnke W; Missimer JH; et al. Molecular basis of coiled-coil formation. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 7062–7067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (42).Kammerer RA; Schulthess T; Landwehr R; Lustig A; Engel J; Aebi U; et al. An autonomous folding unit mediates the assembly of two-stranded coiled coils. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 13419–13424. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (43).Ludwiczak J; Winski A; Szczepaniak K; Alva V; Dunin-Horkawicz S DeepCoil-a fast and accurate prediction of coiled-coil domains in protein sequences. Bioinformatics 2019, 35, 2790–2795. [DOI] [PubMed] [Google Scholar]
  • (44).Tripler TN; Kaplan AR; Alexandrescu AT; Teschke CM Conservation and Divergence of the I-Domain Inserted into the Ubiquitous HK97 Coat Protein Fold in P22-Like Bacteriophages. J. Virol. 2019, 93, No. e00007–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (45).Tripler TN; Maciejewski MW; Teschke CM; Alexandrescu AT NMR assignments for the insertion domain of bacteriophage CUS-3 coat protein. Biomol. NMR Assignments 2015, 9, 333–336. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (46).Brutscher B; Bruschweiler R; Ernst RR Backbone dynamics and structural characterization of the partially folded A state of ubiquitin by 1H, 13C, and 15N nuclear magnetic resonance spectroscopy. Biochemistry 1997, 36, 13043–13053. [DOI] [PubMed] [Google Scholar]
  • (47).Adhada ST; Sarma SP Slow Conformational Exchange between Partially Folded and Near-Native States of Ubiquitin: Evidence for a Multistate Folding Model. Biochemistry 2024, 63, 2565–2579. [DOI] [PubMed] [Google Scholar]
  • (48).Masoumzadeh E; Courtney JM; Charlier C; Ying J; Anfinrud P; Bax A Structure of a transient protein-folding intermediate by pressure-jump NMR spectroscopy. Proc. Natl. Acad. Sci. U. S. A. 2025, 122, No. e2519493122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (49).Alexandrescu AT Amide proton solvent protection in amylin fibrils probed by quenched hydrogen exchange NMR. PLoS One 2013, 8, No. e56467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (50).Englander SW Hydrogen exchange and mass spectrometry: A historical perspective. J. Am. Soc. Mass Spectrom. 2006, 17, 1481–1489. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (51).Englander SW; Kallenbach NR Hydrogen exchange and structural dynamics of proteins and nucleic acids. Q. Rev. Biophys. 1983, 16, 521–655. [DOI] [PubMed] [Google Scholar]
  • (52).Alexandrescu AT; Dregni AJ The Temperature Dependence of Hydrogen Bonds Is More Uniform in Stable Proteins: An Analysis of NMR (h3)J(NC’) Couplings in Four Different Protein Structures. Molecules 2024, 29, 2950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (53).Cordier F; Grzesiek S Temperature-dependence of protein hydrogen bond properties as studied by high-resolution NMR. J. Mol. Biol. 2002, 317, 739–752. [DOI] [PubMed] [Google Scholar]
  • (54).Bahar I; Wallqvist A; Covell DG; Jernigan RL Correlation between native-state hydrogen exchange and cooperative residue fluctuations from a simple model. Biochemistry 1998, 37, 1067–1075. [DOI] [PubMed] [Google Scholar]
  • (55).Seo D; Kammerer RA; Alexandrescu AT Solution NMR assignments and structure for the dimeric kinesin neck domain. Biomol. NMR Assignments 2023, 17, 301–307. [DOI] [PubMed] [Google Scholar]
  • (56).Chatterjee S; Jiang W; Emerson SD; Inouye M The backbone structure of the major cold-shock protein CS7.4 of Escherichia coli in solution includes extensive beta-sheet structure. J. Biochem. 1993, 114, 663–669. [DOI] [PubMed] [Google Scholar]
  • (57).Commans S; Plateau P; Blanquet S; Dardel F Solution structure of the anticodon-binding domain of Escherichia coli lysyl-tRNA synthetase and studies of its interaction with tRNA(Lys). J. Mol. Biol. 1995, 253, 100–113. [DOI] [PubMed] [Google Scholar]
  • (58).Alexandrescu AT; Jaravine VA; Dames SA; Lamour FP NMR hydrogen exchange of the OB-fold protein LysN as a function of denaturant: the most conserved elements of structure are the most stable to unfolding. J. Mol. Biol. 1999, 289, 1041–1054. [DOI] [PubMed] [Google Scholar]
  • (59).Alexandrescu AT; Lamour FP; Jaravine VA NMR evidence for progressive stabilization of native-like structure upon aggregation of acid-denatured LysN. J. Mol. Biol. 2000, 295, 239–255. [DOI] [PubMed] [Google Scholar]
  • (60).Korchak EJ; Geddes-Buehre DH; Bezsonova I The C-terminal Tail of Ubiquitin-Specific Protease 7 Facilitates Ubiquitin Release and Ensures Efficient Catalytic Turnover. J. Mol. Biol. 2025, 437, 169298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • (61).Matousek WM; Alexandrescu AT NMR structure of the C-terminal domain of SecA in the free state. Biochim. Biophys. Acta Gen. Subj. 2004, 1702, 163–171. [DOI] [PubMed] [Google Scholar]
  • (62).Ban D; Funk M; Gulich R; Egger D; Sabo TM; Walter KF; et al. Kinetics of conformational sampling in ubiquitin. Angew. Chem. Int. Ed. 2011, 50, 11437–11440. [DOI] [PubMed] [Google Scholar]
  • (63).Feng W; Tejero R; Zimmerman DE; Inouye M; Montelione GT Solution NMR structure and backbone dynamics of the major cold-shock protein (CspA) from Escherichia coli: evidence for conformational dynamics in the single-stranded RNA-binding site. Biochemistry 1998, 37, 10881–10896. [DOI] [PubMed] [Google Scholar]
  • (64).Markley JL; Kainosho M Stable isotope labelling and resonance assignments in larger proteins. NMR of Macromolecules: A Practical Approach; Robertsw GCK, Ed.; Oxford University Press: Oxford, 1993; pp 101–152. [Google Scholar]
  • (65).Pan Y; Briggs MS Hydrogen exchange in native and alcohol forms of ubiquitin. Biochemistry 1992, 31, 11405–11412. [DOI] [PubMed] [Google Scholar]
  • (66).Goodman EM; Kim PS Periodicity of amide proton exchange rates in a coiled-coil leucine zipper peptide. Biochemistry 1991, 30, 11615–11620. [DOI] [PubMed] [Google Scholar]
  • (67).Wagner G; Wuthrich K Amide protein exchange and surface conformation of the basic pancreatic trypsin inhibitor in solution. Studies with two-dimensional nuclear magnetic resonance. J. Mol. Biol. 1982, 160, 343–361. [DOI] [PubMed] [Google Scholar]
  • (68).McDonald IK; Thornton JM Satisfying hydrogen bonding potential in proteins. J. Mol. Biol. 1994, 238, 777–793. [DOI] [PubMed] [Google Scholar]
  • (69).Badaczewska-Dawid AE; Nithin C; Wroblewski K; Kurcinski M; Kmiecik S MAPIYA contact map server for identification and visualization of molecular interactions in proteins and biological complexes. Nucleic Acids Res. 2022, 50, W474–W482. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information

RESOURCES