The Existence of an Isolated Hydronium Ion in the Interior of Proteins.

Neutron diffraction analysis studies reported an isolated hydronium ion (H3 O+ ) in the interior of d-xylose isomerase (XI) and phycocyanobilin-ferredoxin oxidoreductase (PcyA). H3 O+ forms hydrogen bonds (H-bonds) with two histidine side-chains and a backbone carbonyl group in PcyA, whereas H3 O+ forms H-bonds with three acidic residues in XI. Using a quantum mechanical/molecular mechanical (QM/MM) approach, we analyzed stabilization of H3 O+ by the protein environment. QM/MM calculations indicated that H3 O+ was unstable in the PcyA crystal structure, releasing a proton to an H-bond partner His88, producing H2 O and protonated His88. On the other hand, H3 O+ was stable in the XI crystal structure. H-bond partners of isolated H3 O+ would be practically limited to acidic residues such as aspartic and glutamic acids in the protein environment.

Water molecules can serve as the proton donor and acceptor in the hydrogen bond (H‐bond) network of the protein interior, forming a proton transfer pathway with titratable residues. In particular, when water molecules are strongly H‐bonded, the activation energy for proton transfer is the lowest, without involving formation of an isolated hydronium ion, H3O+.1 On the other hand, H3O+ was proposed to be present in H, K‐ATPase2 or at the end of the proton transfer pathway in bacteriorhodopsin3 (but see also Ref. 4). Neutron diffraction analysis of rubredoxin suggested the presence of H3O+ on the protein surface.5 In all these examples, it is assumed that H3O+ is stabilized by the donation of OH groups to the acceptor water molecules (for example, [H2O⋅⋅⋅H⋅⋅⋅OH2]+). In contrast, “isolated H3O+”, which exists in the absence of other water molecules, was not identified in protein crystal structures until neutron diffraction analysis of metal‐removed d‐xylose isomerase (XI)6 and the more recent neutron diffraction analysis of phycocyanobilin‐ferredoxin oxidoreductase (PcyA)7 were reported. It has been proposed that in the PcyA neutron structure, H3O+ donates H‐bonds to Nδ of His74, Nϵ of His88, and the backbone carbonyl O of Leu243 (Figure 1 a).7 The H‐bond network of the H3O+‐binding moiety is also proposed to be involved in the proton transfer pathway, which would be necessary for the endovinyl reduction of biliverdin IXα.8 In the XI neutron structure, H3O+ was proposed to donate H‐bonds to Glu181, Glu217, and Asp245, which was originally the metal‐binding moiety of the enzyme6 (Figure 1 b). The existence of isolated H3O+ is possible only when pK a of H3O+/H2O [pK a(H3O+)] is higher than that of all of the H‐bond acceptor groups. However, in water, pK a(H3O+) is −1.7, which is significantly lower than that for Asp (4.0), Glu (4.4), Nδ of His (6.6), and Nϵ of His (7.0) in water.

Figure 1

H3O+‐binding sites proposed by neutron diffraction analysis in a) PcyA (PDB code: 4QCD) and b) XI (PDB code: 3KCJ). Yellow dotted lines indicate H‐bond interactions.

The pK a of H‐bond donor and acceptor moieties in H‐bonds can be analyzed from the potential energy profiles of the H‐bonds (Supporting Information, Figure S1).9 In H‐bonds, a proton is more likely to populate the moiety with the higher pK a value between the two moieties (Supporting Information, Figure S2).9c The energy difference between the H‐bond donor and acceptor moieties corresponds to the pK a difference (Supporting Information, Figure S3). This feature also holds true for H‐bonds in protein environments,9b,9c, 10 which are typically analyzed at the density functional theory (DFT) level. Calculations performed at the DFT level are likely to stray away from correct description of geometry towards better description of energy.11 Therefore, H‐bonds should be evaluated based on not only the distances but also the potential‐energy profiles, as suggested by Schutz and Warshel.9b In particular, low‐barrier H‐bonds (LBHB), which also exist in [H2O⋅⋅⋅H⋅⋅⋅OH2]+, can be unambiguously defined by the potential‐energy profile at the DFT level (Supporting Information, Figure S4), because identical pK a values of the donor and acceptor moieties is the requirement for LBHB formation (that is, asymmetric single minimum H‐bonds are not LBHB, as suggested by Schutz and Warshel9b). Further description of the H‐bonds may be obtained with the solution of the nuclear Schrödinger equation.12

Herein, to understand how the protein environment can stabilize an isolated H3O+, we analyzed the potential energy profiles of H‐bonds in the proposed H3O+ binding moieties by adopting a quantum mechanical/molecular mechanical (QM/MM) approach based on the neutron structures of PcyA and XI.

The stability of H3O+ in PcyA was investigated. The neutron structure of PcyA has two conformers: conformer I, corresponding to the case with protonated biliverdin and ionized Asp105; and conformer II, corresponding to deprotonated biliverdin and protonated Asp105.7 QM/MM calculations indicated that H3O+ at the His74, His88, and Leu243 moiety was unstable, and that in both conformers I (Figure 2; Supporting Information, Table S1) and II (Supporting Information, Table S2), it released a proton to Nϵ of His88 to form H2O and doubly protonated His88; the resulting geometries were practically the same in the two conformers. Below, we focus on conformer I.

Figure 2

H3O+ binding sites proposed by neutron diffraction analysis of PcyA a) in the neutron structure and b) the QM/MM‐optimized geometry. The red dotted line indicates the newly formed H‐bond in the QM/MM‐optimized geometry.

Notably, the H‐bond lengths Owater−NϵHis88 and Owater−NδHis74 in the original neutron structure (2.7 and 2.8 Å7) were reproduced even in the QM/MM‐optimized geometry, where H3O+ was absent but H2O and doubly protonated His88 were present (2.7 and 2.7 Å, respectively; Supporting Information, Table S1). Thus, the geometry of the heavy atom position in the neutron structure can be explained without assuming the presence of H3O+. Although the present study suggests that H3O+ is absent in the PcyA neutron structure, one H+ shows high probability of existence at the NϵHis88 moiety. When assuming H2O, deprotonated NϵHis88, and deprotonated NδHis74, the resulting root‐mean‐square deviations (RMSD) of the QM/MM‐optimized geometry from the neutron structure were significantly large (Supporting Information, Table S1).

The potential energy profile of H‐bonds also indicated that in the O ‐OLeu243 H‐bond, the proton is located at the H3O+ moiety, that is, pK a(O=CLeu243)H3O+) (Figure 3). On the other hand, in the O −NHis74 and O −NHis88 H‐bonds, the protons are located at the histidine moieties (Figure 3). The energy difference between the H3O+ and His88 moieties is larger than that between the H3O+ and His74 moieties, which suggests pK a(O=CLeu243)H3O+)His74)His88) in the PcyA protein environment (Figure 3). In water, pK a(H3O+) is −1.7 and pK a(NϵHis) is about 7. The proposed H3O+ binding site in PcyA does not have acidic residues; hence, pK a(H3O+) cannot be increased to overcome the original pK a difference of over 8 pK a units (see similar discussions in Ref. 4). His88 can accept an H‐bond from H3O+ but cannot decrease pK a(H3O+) to pK a(NϵHis88) because of the absence of a negative charge. The nearest acidic residue Glu76 is about 7 Å away from the proposed H3O+ binding site.

Figure 3

Energy profiles along the H‐bonds in the H3O+‐binding moiety in PcyA. a) Detailed energy profiles. b) Energy profiles along the O⋅⋅⋅H⋅⋅⋅O bond axes.

The reasons for the higher pK a(NϵHis88) as compared to pK a(NδHis74) are the presence of 1) Asp105 near His88 (3.7–5.5 Å7), which stabilizes the doubly protonated His88; and 2) Lys72 near His74 (5.1 Å7), which destabilizes the doubly protonated His74 (Table 1).

Table 1

Residues that shift pK a(NϵHis88) by more than 2 pK a units in PcyA.[a]

		ΔpK a(NϵHis88)	ΔpK a(NδHis74)
	Asp105	4.6	1.9
	Asp245	2.1	2.5
	Lys72	−3.2	−6.0

[a] For comparison, influences of the same residues on pK a(NδHis74) are also shown. See the Supporting Information for experimental procedures.

For the presence of stable H3O+ at the PcyA binding moiety, pK a(NδHis74) and pK a(NϵHis88) must be lower than pK a(H3O+), that is, Nδ of His74 and Nϵ of His88 must be deprotonated. The absence of protonation at Nδ of His74 in the H‐bond potential‐energy profile (Figure 3) is consistent with the interpretation based on the neutron diffraction analysis reported by Unno et al.7 On the other hand, the protonated Nϵ of His88 in the H‐bond potential energy profile (Figure 3) is inconsistent with the interpretation based on neutron diffraction analysis. Since the neutron diffraction analysis was carried out in crystals, unstable H3O+ might have been be trapped in them. If this is the case, the crystallographers are urged to make efforts to confirm their “interpretation” by chemical data (for example, pK a, chemical shift). In support of the presence of H3O+, Unno et al. stated7 that “in nuclear magnetic resonance (NMR) studies by Kohler et al.,13 His88 was reported to be singly protonated in native PcyA”. However, according to the original report by Kohler et al.,13 it was the D105N mutant that had singly protonated His88, and not the native PcyA.13 His88 was doubly protonated in biliverdin‐free native PcyA; the protonation state of His88 could not be determined for biliverdin‐bound native PcyA.13 Even in biliverdin‐bound D105N mutant, where His88 was confirmed to be singly protonated at Nϵ,13 the experimentally measured NMR chemical shift of 7.86 ppm for the 1Hϵ of His88 was far from those observed for LBHB (typically 17–22 ppm;14 Supporting Information, Figure S4). It is likely that the NMR results13 do not directly support the interpretation by Unno et al.7 for the presence of H3O+ and deprotonated NϵHis88.

The stability of H3O+ in XI was then investigated. QM/MM calculations reproduced the presence of H3O+, observed as D3O+ in the XI neutron structure (Supporting Information, Table S3). The neutron structure could be interpreted to indicate that Glu181, Glu217, and Asp245 are the H‐bond acceptors of H3O+.6 On the other hand, the RMSD of the QM/MM‐optimized geometry from the neutron structure was the lowest, at 0.27 Å, when Glu217, Asp245, and Asp287 were the H‐bond acceptors of H3O+ (Supporting Information, Table S3). All the other H‐bond patterns resulted in RMSD of 0.35–0.39 Å, which are even higher than the value (0.33 Å) obtained assuming the presence of NH4 + (Supporting Information, Table S4). Below, we focus on this QM/MM‐optimized structure, where Glu217, Asp245, and Asp287 are the H‐bond acceptors of H3O+.

In contrast to the neutron structure of PcyA, the neutron structure of XI shows two remarkably short H‐bonds, O −OAsp245=2.3 Å and O −OGlu217=2.4 Å.6 Intriguingly, QM/MM calculations reproduced this result: O −OAsp245=2.5 Å and O −OGlu217=2.5 Å (Figure 4; Supporting Information, Table S3). In sharp contrast to PcyA, the potential energy profile of the H‐bonds indicates that the energy minimum is localized at the H3O+ moiety in all the three H‐bonds with an acidic residue (Figure 5), confirming that H3O+ exists in the protein interior of XI. The energy difference between the H3O+ and proton acceptor moieties suggests that pK a(Asp287)Glu217)≈pK a(Asp245)≤pK a(H3O+) (Figure 5). In water, pK a(Asp, Glu)≈4 is lower than pK a(NϵHis)≈7 but still higher than pK a(H3O+)=−1.7. In XI, four acidic residues are present at the H3O+‐binding moiety. These four acidic residues can stabilize the protonated state of H3O+ and cause a significant increase in pK a(H3O+), leading to pK a(H3O+)≥pK a(Asp, Glu) in XI.

Figure 4

H3O+‐binding sites proposed by neutron diffraction analysis of XI in a) the neutron structure and b) the QM/MM‐optimized geometry with the lowest RMSD. The red dotted line indicates the H‐bond that differs between the neutron structure and the QM/MM‐optimized geometry.

Figure 5

Energy profiles along the H‐bonds in the H3O+ binding moiety in XI. a) Detailed energy profiles. b) Energy profiles along the O⋅⋅⋅H⋅⋅⋅O bond axes.

Based on the analysis of the two independent neutron structures, the present study helps in understanding how the two protein environments of the proposed H3O+ binding moieties are markedly different.

In XI, the potential energy surface of the H3O+ binding moiety, shaped as a symmetric funnel (Figure 6), would be a prerequisite for the existence of isolated H3O+. Isolated H3O+ is stable in the protein interior only when the energy minimum of H+ is localized in the H3O+ moiety, that is, pK a of H3O+ must be equal to or larger than pK a of all the three H‐bond acceptors (Figure 6). These H‐bond partners would be practically limited to acidic residues such as aspartic acid and glutamic acid in the protein environment.

Figure 6

Difference in the energy profiles between proton transfer pathway (PcyA; left) and isolated H3O+ binding site (XI; right). Blue arrows indicate migration of H+ in H‐bonds.

The potential energy profile of the isolated H3O+‐binding moiety in XI is in sharp contrast to that of PcyA (Figure 6). In PcyA, deprotonated histidine may accept a H‐bond from H3O+ but cannot decrease pK a(H3O+) because of the absence of negative charge, thus allowing formation of H2O and protonated histidine. The backbone carbonyl O at Leu243 is non‐titratable; thus, the proton needs to be delocalized over the other two titratable groups His74 and His88 in PcyA, facilitating proton transfer between the two residues. This is consistent with a common view that His74 and the water molecule form a proton transfer pathway (proton shuttle) to His88 in PcyA.8b,8c Isolated H3O+ is unlikely to exist unless the protonated carbonyl of Leu243 is stable in PcyA. Notably, backbone carbonyl groups also exist as H‐bond acceptors for water molecules in the proton‐conducting water chain of photosystem II,10b, 15 where formation of H3O+ is inhibited for efficient proton transfer.1a

The proton stabilized in the form of H3O+ (for example, XI6) may not be readily available as a transferable H+, that is, catalytically important H+. Isolated H3O+ binding sites may be suitable metal‐binding sites, as is originally the case with XI;6 isolated H3O+ plays a role in binding the negatively charged residues and stabilize the protein structure prior to metal binding.

These results may also provide a key to understanding the requirement for the protein environment of efficient proton transfer pathways, for example, photosystem II and bacteriorhodopsin.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.