Article Cite This: J. Phys. Chem. B XXXX, XXX, XXX−XXX
pubs.acs.org/JPCB
Imidazole Nitrogens of Two Histidine Residues Participating in N−H···N Hydrogen Bonds in Protein Structures: Structural Bioinformatics Approach Combined with Quantum Chemical Calculations Abhishek Hariharan Iyer,‡ R. N. V. Krishna Deepak,†,‡ and Ramasubbu Sankararamakrishnan* Department of Biological Sciences and Bioengineering, Indian Institute of Technology Kanpur, Kanpur 208016, India S Supporting Information *
ABSTRACT: Protein structures are stabilized by different types of hydrogen bonds. However, unlike the DNA double helical structure, the N−H···N type of hydrogen bonds is relatively rare in proteins. N−H···N hydrogen bonds formed by imidazole groups of two histidine residues have not been investigated. We have systematically analyzed 5333 highresolution protein structures with resolution 1.8 Å or better and identified 285 histidine pairs in which the nitrogen atoms of the imidazole side chains can potentially participate in N−H···N hydrogen bonds. The histidine pairs were further divided into two groups, neutral− neutral and protonated−neutral, depending on the protonation state of the donor histidine. Quantum chemical calculations were performed on imidazole groups adopting the same geometry observed in the protein structures. Average interaction energies between the interacting imidazole groups are −6.45 and −22.5 kcal/mol for neutral−neutral and protonated−neutral, respectively. Hydrogen bond interaction between the imidazole moieties is further confirmed by natural bond orbital analyses of the model compounds. Histidine residues involved in N−H···N hydrogen bonds are relatively more buried and have low B-factor values in the protein structures. N−H···N hydrogen bond formed by a pair of buried histidine residues can significantly contribute to the structural stability of proteins.
■
INTRODUCTION Aromatic residues have aided in the structural and folding studies of proteins using spectroscopic techniques. Their involvement in different types of noncovalent interactions has been shown to be structurally and functionally significant. The π-electron cloud of aromatic residues has been shown to participate in cation···π, anion···π, π···π, O−H···π, and C−H···π interactions. An exhaustive search of X−H···π hydrogen bonds in high-resolution structures identified N−H···π, O−H···π, S− H···π, and C−H···π interactions in different secondary structural elements of proteins, and Trp residue has been identified as the most efficient π-acceptor.1 More recently, aromatic residues including histidine have been shown to participate in lone pair···π interactions with the carbonyl oxygen atoms of backbone functional groups and water oxygen atoms.2,3 Atoms in the periphery of the aromatic ring have been shown to take part in different types of hydrogen bonds. In mutation experiments and structural studies, it has been demonstrated that hydrogen bonds formed by the tyrosine −OH groups provide stability to enzymes such as RNAse Sa and Sa3 and lysozyme.4 Petrella and Karplus analyzed and found out that the unusual side-chain rotamers of tryptophan residues in high-resolution protein crystal structures are due to the three center C−H···X···H−C hydrogen bonds.5 Their © XXXX American Chemical Society
subsequent free-energy calculations exhibited that the hydrogen bonds formed by the C−H groups of tryptophan side chains can stabilize the off-rotamer conformation. Crystal structure analysis and ab initio calculations have also led to the identification of interactions between phenylalanine and cysteine residues.6 Spectroscopy studies and density functional calculations demonstrated the existence of T-shaped N−H···π interactions between histidine and phenylalanine side chains.7 Preference of aromatic residues for the membrane−water interface in integral membrane proteins has been well established by experimental studies and structural analyses.8,9 Conformational rearrangements of large aromatic side chains are often found in protein−protein binding sites.10 Important factors in protein-carbohydrate recognition include stacking interactions between the aromatic residues and sugar pyranose rings.11 Clusters of aromatic residues have been observed in the interior region of several protein structures, and the stabilization of the hydrophobic core often occurs through aromatic−aromatic interactions.12 Tryptophan and histidine are the only two aromatic residues having at least one nitrogen Received: November 29, 2017 Revised: December 25, 2017 Published: December 27, 2017 A
DOI: 10.1021/acs.jpcb.7b11737 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B atom as part of the aromatic ring systems. The nitrogen atoms of indole and imidazole groups of these two residues participate in unusual interactions with water molecules in which water molecules approach the heterocyclic nitrogen perpendicular to the aromatic plane.13 Chakrabarti and his colleagues examined high-resolution protein structures to understand the environment of tryptophan side chains and found that the nitrogen atom in the indole rings participated in the highest number of interactions and a significant number of them can be described as hydrogen bonds.14 The nitrogen atom of the imidazole ring can participate as a hydrogen bond acceptor or donor depending on the protonation state of histidine residues. To our knowledge, a systematic study of two histidine residues participating in hydrogen bonding interactions through their imidazole rings has not been carried out. We have recently examined the ultrahigh-resolution protein structures and very high-resolution protein structures and analyzed the occurrence of six different types of hydrogen bonds, three with oxygen as acceptor (N−H···O, O−H···O, and C−H···O) and three with nitrogen as acceptor (N−H···N, O−H···N, and C−H···N).15 Hydrogen bonds with nitrogen as acceptor constitute less than 2% of all of the hydrogen bonds analyzed. Proline and histidine are the two residues that predominantly participate in hydrogen bonds with nitrogen as acceptor. Histidine imidazole ring can participate in N−H···N type of hydrogen bond either with the main-chain nitrogen or with the side-chain nitrogen. N−H···N hydrogen bonds formed between histidine imidazole nitrogen as acceptor and mainchain nitrogen as donor have been characterized and their possible structural significance has been discussed.16 However, N−H···N hydrogen bonds formed between the imidazole rings of two histidine residues have not been investigated. In this study, using structural bioinformatics approach, we first identified the N−H···N contacts between two histidine residues using specific geometric criteria. We subsequently used quantum chemical calculations on model compounds to evaluate the strength of such hydrogen bonds and found that they are most likely to be present in the interior of proteins and can provide structural stability to proteins.
Quantum Chemical Calculations Involving Model Compounds. To find out the strength of N−H···N hydrogen bonds formed by the imidazole rings of two histidine residues, we used quantum chemical calculations. The histidine residue participating in N−H···N hydrogen bond as a donor can be either neutral or protonated. Hence, histidine residues involved in N−H···N hydrogen bonds can be classified into two groups. In the first group, both the donor and acceptor histidine residues are neutral. In the second category, the donor histidine is protonated and the acceptor histidine is neutral. A representative example, one from each category, is shown in Figure 1. In our quantum chemical calculations, both histidine
METHODS We have previously identified N−H···N contacts in the dataset containing high-resolution protein structures from Protein Data Bank (PDB)17 determined as of March 31, 2015.15,16 These structures were determined using X-ray diffraction with resolution 1.8 Å or better, R-value ≤0.20, and R-free ≤0.25. This dataset contained a nonredundant set of 5542 polypeptide chains from 5333 structures, and the hydrogen positions were generated using the program REDUCE.18 With the possibility of modeling wrong orientation of His side chain in ambiguous density in X-ray structure determination, REDUCE has the option of flipping the orientation of His side chain if it promotes hydrogen bonds and avoids steric clashes. Our aim is to find the N−H···N hydrogen bonds involving two histidine residues. After building hydrogen positions using REDUCE, we used the following distance [d(ND···NA) ≤ 3.5 Å and d(H···NA) ≤ 2.5 Å] and angle [θ(ND−H···NA) ≥ 90° and θ(H···NA−AA) ≥ 90°] criteria to identify N−H···N type of hydrogen bonds.19 ND and NA are, respectively, the donor and acceptor nitrogen atoms belonging to the imidazole rings of two different histidine residues. AA represents the acceptor antecedent atom. The same geometric criteria were used in our earlier studies.15,16
residues were replaced by imidazole groups. The imidazole structure was obtained from PubChem database.20 The imidazole groups were first individually geometry optimized using the electronic structure program package ORCA version 3.0.221 with BP86 density functional theory22,23 and Ahlrichs’ triple-ζ def2-TZVP basis set.24,25 Two such imidazole groups were superposed on the histidine imidazole side chains of PDB structures that are involved in N−H···N hydrogen bonds. This is to ensure that the geometry observed in the PDB structure is retained by the model compounds. In our calculations, the imidazole group that donates hydrogen for the N−H···N hydrogen bond is either neutral or protonated. Single point energy calculations were carried out on all of the identified examples. Interaction energies between the imidazole moieties participating in N−H···N hydrogen bonds are calculated using the following equation
Figure 1. Representative examples of two histidine residues involved in N−H···N type of contacts: (A) both the participating histidine residues are neutral; (B) histidine that donates hydrogen for the N− H···N hydrogen bond interaction is protonated and the acceptor histidine is neutral. Standard colors are used to represent the different atoms. The four-letter unique PDB ID is mentioned at the bottom of each example. Residue numbers of participating histidines are shown. The distance between the hydrogen and the acceptor nitrogen is displayed. UCSF Chimera software30 is used to generate this figure and subsequent molecular plots.
■
E int = E DH ‐ AH − E DH − EAH
(1)
In the above equation, Eint is the interaction energy and EDH‑AH is the single point energy of the whole system consisting of both interacting imidazole groups. EDH and EAH are the single point energies of the donor and acceptor imidazole groups, respectively. The program Gaussian 0926 along with M06-2X B
DOI: 10.1021/acs.jpcb.7b11737 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B density functional theory27 and AUG-cc-pVQZ basis set28 was used in single point energy calculations. Boys and Bernardi’s standard counterpoint correction method29 was employed to account for the basis set superposition error (BSSE). Natural Bond Orbital (NBO) Calculations. NBO analysis31,32 was performed on systems consisting of both imidazole groups participating in N−H···N interactions. The systems with the most and least favorable interaction energies were considered for this purpose. Conformations found in the PDB structures were used with M06-2X model chemistry27 and the AUG-cc-pVQZ basis set.28 Second-order perturbation theory, as implemented in NBO version 3.133 (available as part of Gaussian 0926), was used to calculate the stabilization energies. B-Factor Analysis. The inherent thermal mobility of protein atoms around their equilibrium positions is measured in X-ray crystallography as B-factors (also referred to as temperature factors or atomic displacement parameters). Bfactors have been used to design mutagenesis experiments,34 to predict protein binding interfaces and affinities,35,36 and to predict intrinsically disordered regions in proteins.37,38 Bfactors have been used to confirm the lone-pair−π interactions between aromatic residues and water molecules.2 In this study, we analyzed the B-factors of histidine residues participating in N−H···N hydrogen bonds to find whether the interacting His pairs can be discriminated on the basis of B-factor analysis. We first normalized the B-factors of every protein in the dataset using the approach adopted in an earlier study,39 and the following equation is used to normalize the B-factors B N = (B − ⟨B⟩)/σ(B)
Figure 2. Scatter plot showing the d(ND···NA) distance (in angstrom) vs θ(ND−H···NA) angle (in degrees), where ND and NA, respectively, denote the donor and acceptor nitrogen atoms of imidazole rings of two histidine residues. Contacts satisfying all of the criteria for N−H··· N hydrogen bonds are shown in green. Cutoff distances ≤3.5 Å but not satisfying either the hydrogen−acceptor distance or at least one of the two angle criteria are shown in blue. All contacts >3.5 Å are shown in red.
All of the 285 histidine−histidine pairs that are involved in N−H···N type of hydrogen bonds are given in Table S1. This was further divided into two groups, namely, neutral−neutral and protonated−neutral. In the first case, both the histidine residues participating in N−H···N hydrogen bonds are neutral, and in the second case, the histidine that donates hydrogen for the N−H···N hydrogen bond is protonated. Out of 285, we have found 223 examples of histidine pairs where both are neutral. In the remaining 62 pairs, the donor histidine is in the protonated form. We have performed quantum chemical calculations on both sets of histidine pairs, as described in the Methods section. Histidine residues are represented by the imidazole model compounds. Distributions of interaction energies of neutral−neutral and protonated−neutral imidazole pairs participating in N−H···N hydrogen bonds are shown in Figure 3. Interaction energy for each individual example is also provided in Table S2. It is clear that the interaction energies of both sets of imidazole pairs are favorable. Interaction energies of most of the imidazole pairs in which both imidazole groups are neutral are between −6 and −8 kcal/mol (Figure 3A). Only three examples resulted in a positive interaction energy, indicating that the close approach of N−H···N contacts is not favorable. In two of them (PDB IDs: 3SFV and 1L3K), the hydrogen of one imidazole group approaches the N−H group of another neutral imidazole group. In the third example (PDB ID: 4JEV), the distance between the hydrogen and acceptor nitrogen is too close (1.51 Å). All of the three examples are shown in the Supporting Information (Figure S1). We have earlier shown that histidine residues participate in N−H···N hydrogen bond with the main-chain amino hydrogen.16 This was represented by the model compounds imidazole and Nmethylacetamide, and the calculated interaction energy was very similar in magnitude to the current interaction energies between the two imidazole groups that are neutral and are involved in the same type of interaction. It is interesting to note that when the donor imidazole is protonated, the interaction energies become much more favorable and fall in the range of −20 to −25 kcal/mol (Figure
(2)
where BN is the normalized B-factor and B is the B-factor of the heavy atoms of the individual histidine residue. ⟨B⟩ and σ(B) are, respectively, the average and standard deviation of B-factors of all Cα atoms of the structure under consideration.
■
RESULTS AND DISCUSSION Using the geometric criteria defined in the Methods section, we identified 285 examples of N−H···N contacts formed between two histidine residues from the dataset of 5542 polypeptide chains belonging to 5333 high-resolution X-ray structures. To further confirm that the 285 His−His pairs form true N−H···N hydrogen bonds, we identified all of the His−His pairs whose imidazole nitrogen atoms are within 7.5 Å. We have presented the distance d(ND···NA) versus angle θ(ND−H···NA) scatter plot in Figure 2, where ND and NA represent donor and acceptor imidazole nitrogens, respectively. The scatter plot very clearly shows the ND−H···NA hydrogen bond contacts (shown in green in Figure 2), which satisfy all of the distance and angle criteria, and they are very clearly separated from all other contacts. Contacts that satisfy the donor−acceptor distance criteria that mostly lie between 3.3 and 3.5 Å (blue in Figure 2) do not satisfy either the hydrogen−acceptor distance cutoff or at least one of the two angle criteria. When the donor−acceptor distance is >3.5 Å, then their ND−H···NA angles spread over the entire region from 0 to 180°. This analysis clearly demonstrates that the potential hydrogen bonded contacts between the imidazole nitrogen atoms of two histidine residues are clustered separately in the ND···NA distance and the ND− H···NA angle landscape, reiterating that these contacts, shown in green in Figure 2, are indeed due to true N−H···N hydrogen bond interactions. C
DOI: 10.1021/acs.jpcb.7b11737 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 3. Histograms showing the distribution of BSSE-corrected interaction energies for imidazole pairs participating in N−H···N hydrogen bonds: (A) both imidazole rings are neutral and (B) donor imidazole is protonated and the acceptor imidazole is neutral.
Table 1. Natural Bond Orbital (NBO) Analysis of N−H···N Hydrogen Bonds Formed by the Imidazole Nitrogens of Two Histidine Residues NPA chargesa
orbital occupancy
PDB ID/chain IDc donor His−acceptor Hisd
acceptor nitrogen
donor nitrogen
hydrogen bonded to donor nitrogen
lp(N)e
σ*(N−H)f
n → σ* stabilization energy (kcal/mol)b
4RT5/A His18− His17g 3PIU/A His35−His32g 2EHZ/A His52− His53h 3UGF/A His45− His64h
−0.50784 −0.46430 −0.55688 −0.54985
−0.55296 −0.53348 −0.47900 −0.47772
0.45695 0.43118 0.47796 0.48760
1.91338 1.92421 1.86880 1.88610
0.03297 0.01678 0.08357 0.06368
10.90 0.10 40.10 25.16
a
Natural population analysis (NPA) charges obtained from the NBO analyses are furnished for the atoms involved in N−H···N hydrogen bonds from the imidazole groups of two histidine residues. bSecond-order perturbation energies corresponding to the N−H···N hydrogen bond formed by the imidazole groups of two histidine residues. cThe four-letter unique PDB ID and the chain ID of the polypeptide chain are provided. dResidue numbers of donor and acceptor histidine residues are given. eOrbital occupancy for the lone-pair electrons for the acceptor nitrogen. fOrbital occupancy of non-Lewis orbitals of the N−H group involved in N−H···N hydrogen bond. gExamples with the most and least favorable interaction energies in which both imidazole groups were considered neutral. The BSSE-corrected interaction energies are −8.43 and −2.17 kcal/mol, respectively, for the systems corresponding to the PDB IDs 4RT5 and 3PIU. hExamples with the most and least favorable interaction energies in which the donor imidazole group is protonated and the acceptor imidazole group is neutral. The BSSE-corrected interaction energies are −25.1 and −4.1 kcal/mol, respectively, for the systems corresponding to the PDB IDs 2EHZ and 3UGF.
in protonated−neutral pair is much stronger because of the protonated state of donor imidazole. This is reflected in both the calculated NPA charges and the orbital occupancies. When N−H···N hydrogen bond is present between two histidine residues involving a protonated histidine as a donor, it is likely to be a major stabilizing factor in proteins. Hence, we wanted to analyze two important factors, namely, accessible surface area and B-factors, to further understand the nature and the environment of the participating histidine residues in the dataset of protein structures considered for this study. Solvent Accessible Surface Area (SASA) Analysis. We wanted to find out whether the histidine residues that form the N−H···N hydrogen bonds show any preference to be present on the surface or in the interior of proteins. We calculated the solvent accessible surface areas (SASAs) of histidines involved in N−H···N hydrogen bonds and compared them with all other histidine residues in our dataset of protein structures. SASAs of histidine residues were calculated using the gmx_sasa module of the GROMACS software package.43 Figure 4 shows the average and standard errors of SASA values for those histidine residues that participate in N−H···N hydrogen bond with another histidine residue through their imidazole rings. In the same figure, we have also plotted the SASA average with standard error for the 28 970 other histidines that serve as a control data. When the histidine imidazole side chain interacts with another histidine imidazole side chain through N−H···N hydrogen bond, its average SASA value is 253 Å2. The average
3B). This is almost 4- to 5-fold stronger than in the imidazole pairs in which both are neutral, and it could be considered as one of the strongest noncovalent interactions along with saltbridge and cation−π interactions.40−42 To further understand the nature of the N−H···N contact in both sets of histidine pairs, we performed NBO calculations on the systems with the least and most favorable interaction energies from both sets and the results of the NBO analyses are presented below. NBO Analyses of Imidazole Pairs with N−H···N Hydrogen Bonds. The N−H···N contacts formed by the two histidine residues were further characterized by performing NBO analyses on the imidazole pairs. Systems with the least and most favorable interaction energies from neutral−neutral and protonated−neutral sets were considered for this purpose (Figure S2 in Supporting Information). The orbital occupancy and natural population analysis (NPA) charges of the atoms involved in N−H···N contacts along with the second-order perturbation energies due to n → σ* orbital overlap are provided in Table 1. In the case of neutral−neutral imidazole pairs, the second-order perturbation energies for the system with the most and least favorable interaction energies are 10.9 and 0.1 kcal/mol, respectively. For the protonated−neutral pairs, the stabilization energies are 40.1 and 25.16 kcal/mol, respectively, for the examples that resulted in the most and least favorable interaction energies. This indicates that the charge transfer due to N−H···N contacts results in favorable interactions in all of the examples. However, the interaction D
DOI: 10.1021/acs.jpcb.7b11737 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
involved in strong interactions, then it is expected that the Bfactors of these residues will be lower than those of the other residues.2 In this study, we analyzed the B-factors of histidine residues that take part in N−H···N hydrogen bond interactions between the imidazole side chains and compared them with all other histidine residues. Averages of normalized B-factors of the histidine residues for both the categories are plotted in Figure 6. We can easily conclude that when the imidazole group of a
Figure 4. Plots showing the average solvent accessible surface areas with standard error for histidine residues involved in N−H···N hydrogen bonds through their imidazole rings (light color bar) and all other histidine residues (dark color bar).
SASA for all other histidine residues is 287 Å2, indicating that the SASA of histidine residues that are involved in N−H···N hydrogen bonds is significantly lower compared to that of all other histidine residues (p-value