Hydrogen-Bonding Partner of the Proton-Conducting Histidine in the Influenza M2 Proton Channel Revealed From ¹H Chemical Shifts

Abstract

The influenza M2 protein conducts protons through a critical histidine (His) residue, His37. Whether His37 only interacts with water to relay protons into the virion, or whether a low-barrier hydrogen bond (LBHB) also exists between the histidines to stabilize charges before proton conduction, is actively debated. To address this question, we have measured the imidazole ¹H^N chemical shifts of His37 at different temperatures and pH using 2D ¹⁵N-¹H correlation solid-state NMR. At low temperature, the H^N chemical shifts are 8–15 ppm at all pH values, indicating that the His37 sidechain forms conventional hydrogen bonds (H bonds) instead of LBHBs. At ambient temperature, the dynamically averaged H^N chemical shifts are 4.8 ppm, indicating that the H-bonding partner of the imidazole is water instead of another histidine in the tetrameric channel. These data show that His37 forms H-bonds only with water, with regular strength, thus supporting the His-water proton exchange model and ruling out the low-barrier H-bonded dimer model.

The influenza M2 protein forms a tetrameric proton channel important for the virus lifecycle ^1-3. Activated by the low pH environment of the endosome, the channel opens to acidify the virion, which causes viral uncoating. The mechanism of proton conduction through M2 has been long debated. Early computational studies and functional data diverged on whether proton conduction occurs by Grotthuss hopping along a water wire ^4,5 or requires conformational changes of the only titratable residue in the transmembrane (TM) domain, His37 ⁶ (Fig. 1a). Recent data has ruled out the water wire model and converged on the active participation of His37 in proton relay. Evidence for proton shuttling by His37 came from magic-angle-spinning (MAS) ¹⁵N NMR spectra showing chemical exchange of the imidazole nitrogens between the protonated (NH) and unprotonated (N) states at the physiological pH of the endosome ⁷. This exchange is accompanied by low-pH specific imidazolium reorientation on the microsecond timescale with an energy barrier comparable to the proton conduction barrier ⁸.

Figure 1.

Two proton conduction models for the influenza M2 channel. (a) Structure of M2TM at pH 6.5 (PDB: 3LBW), showing the location of the key His37 and Trp41. (b) His37-water proton exchange model. (b) His-His low-barrier H-bonded dimer model. The dimer of dimer state is proposed to exist stably in the +2 tetrad to stabilize charge before proton transfer to water.

Despite the general consensus that His37 shuttles protons, the mechanism by which charge is stabilized in the His37 tetrad is still actively debated. The ¹⁵N chemical exchange and imidazolium reorientation led to the proposal that His37-water H-bonding and proton exchange are sufficient for proton conduction (Fig. 1b) ⁷, and that excess protons are stabilized by delocalization over the His37 tetrad and the surrounding water molecules ⁹. In contrast, an alternative model posits a LBHB between a neutral and a cationic histidine in the +2 state of the channel (Fig. 1c) ¹⁰, which stabilizes the charges before channel activation. This model was motivated by the observation of a very high pK_a of 8.2 for the first two protonation steps in DMPC/DMPG bilayer-bound M2 TM peptide (M2TM) ¹⁰, and by computational modeling of the His37 sidechain structure ¹¹. The latter yielded His37 (χ₁,χ₂) torsion angles of (180°, 90°) to establish the putative Nε2-H…Nδ1 H-bond Recently reported chemical shift multiplicity of some of the TM residues ^12,13 although observed at neutral pH, was also interpreted as supporting the LBHB model.

Equilibrium conformation of His37 measured by solid-state NMR 8 and X-ray crystallography ^9,14 at acidic pH have so far shown no direct His-His H-bonding: the His37 χ₁ and χ₂ angles were measured to be ∼180° in both lipid bilayers and detergents, which point the Nε2-H and Nδ1 toward the interior and exterior of the channel rather than toward each other. ¹³C-¹³C 2D correlation spectra of the +2 charged channel displayed no imidazole-imidazolium cross peaks ⁷, also challenging the LBHB model. However, the ¹⁵N NMR spectra showing N<–>NH chemical exchange, can, in principle, be due to either His-water proton transfer or His-His H-bonding. Thus, we sought more definitive evidence for the H-bonding partner of His37 as well as the strength of the His37 H-bond The strength of H-bonds can be discerned through the ¹H chemical shift a proton in a low-barrier or strong H-bond should have a large chemical shift of greater than 16 ppm ^15-18, whereas a proton in a regular unequal-well H-bond should have a smaller chemical shift of 8-15 ppm ^19,20. The identity of the H-bonding partner for membrane proteins in hydrated lipid bilayers can be determined through the temperature dependence of the 1H chemical shift. Between -30°C and +30°C, the diffusion rates of water in the channel change significantly ^21,22, thus a regular N-H…O H-bond should involve only one or few water molecules at low temperature but should undergo rapid exchange with many water molecules at physiological temperature. This should result in a ¹H chemical shift close to the imidazole H^N value at low temperature but a population-weighted value near the water ¹H chemical shift at high temperature. In contrast, for a pK_a-matched N-H…N LBHB ^23,24, the central proton has a much higher activation energy for exchange with water ²⁵; moreover proton transfer dynamics between the two nitrogens is ultrafast. Thus the ¹H chemical shift will be insensitive to temperature of this range and remain large. Thus, the low-temperature ¹H chemical shift reveals the H-bond strength, whereas the high-temperature shift indicates the identity of the H-bonding partner.

We measured the ¹H chemical shift using the 2D ¹⁵N-¹H heteronuclear correlation (HETCOR) experiment. To detect only cross peaks due to direct N-H dipolar coupling without relayed transfer, we suppressed ¹H spin diffusion using ¹H homonuclear decoupling during the evolution period and the ¹H-¹⁵N cross polarization period ²⁶. His37-labeled M2TM bound to a virus-mimetic lipid membrane were measured at pH 6.0, 4.5, and 8.5 ⁸. Since all initial experiments that led to the LBHB model were conducted on M2TM, we used the same construct to avoid potential ambiguities in interpretation. Previous measurement of His37 pK_a's in this virus-mimetic membrane indicated that the channel was 80% in the +2 state at pH 6, in a mixed +3 and +4 state at pH 4.5, and about 90% neutral at pH 8.5 ⁷. Thus, the pH 6.0 sample is the closest to the putative LBHB state. The 2D HETCOR spectra were measured at 245 K to determine the H-bond strength and 296 K to determine the H-bonding partner.

Fig. 2a shows the 2D HETCOR spectra of the pH 6.0 sample. At 245 K, the imidazole Nε2 (τ tautomer) and Nδ1 (π tautomer) peaks at 160-180 ppm exhibit ¹H chemical shifts of 8 - 12 ppm, similar to the backbone amide ¹H chemical shift range. Thus, imidazole H^N lies in a regular H-bond For comparison, histidine hydrochloride (Fig. S1) shows a large Hδ1 chemical shift of 16.8 ppm due to an intermolecular H-bond to a C=O with an N…O distance of 2.63 Å ²⁷. Both ¹⁵N and ¹H shifts reflect the strength of the H-bond: small ¹⁵N and ¹H shifts indicate a stronger covalent N-H bond, while large shifts indicate a more deprotonated nitrogen or a stronger H-bond ^17,18. The correlation gives a slope of ∼3 between the ¹⁵N and ¹H chemical shifts (Fig. 2a). The ¹H shift distribution (Fig. S2), detected for both backbone and imidazole nitrogens, indicate a distribution of H-bond strengths. The backbone distribution is likely due to varying degrees of helix ideality in an ensemble with mixed protonation states, while the imidazole H^N shift distribution can be attributed to the presence of multiple N-H species, including Nε2H(τ), Nδ1H(π), and the Nε2H and Nδ1H of cationic imidazolium (Fig. S3).

Figure 2.

2D ¹⁵N-¹H HETCOR spectra of membrane-bound His37-labeled M2TM at (a) pH 6.0, (b) pH 4.5, and (c) pH 8.5 at 245 K (left) and 296 K (right). The main charged states of the M2TM channel at each pH are indicated. The ¹H dimension of the spectrum was measured with ¹H homonuclear decoupling to eliminate spin diffusion effects. Assignments are shown in red for the neutral τ tautomer, black for the neutral π tautomer, and green for cationic His37.

When the temperature increased to 296 K, the imidazole ¹H chemical shifts decreased uniformly to 4.8 ppm, indicating definitively that the H-bonding partner of His37 is water instead of another His. Since ¹H homonuclear decoupling was applied in the experiment, both rigid and mobile protons were equally detected, thus the observed ¹H chemical shift near the unperturbed water frequency indicates a large number of water molecules in exchange with the imidazole nitrogens. For comparison, the backbone H^N chemical shift is unaffected by temperature, as expected for the persistence of N-H…O=C H-bonds at these temperatures.

The 2D spectra of the pH 4.5 sample (Fig. 2b) further support the His37-water interaction model. Even at low temperature, the 178-ppm ¹⁵N peak already shows a water ¹H cross peak (5.7 ppm) in addition to the Hε2/Hδ1 signal (12-15 ppm), consistent with previous data showing a more hydrated channel at this low pH ²⁸. The Hε2/Hδ1 chemical shift is larger than at pH 6, indicating stronger H-bonds. This is consistent with the previously measured N-H bond elongation at this pH ⁸. The ¹⁵N/¹H chemical shift slope is the same as at pH 6.0 (Fig. S4), as expected for the intrinsic correlation between ¹⁵N protonation and N-H…O H-bond strength. Again, the identity of the H-bonding partner is determined by the high-temperature spectrum, which shows a ¹H chemical shift at the water position of 4.8 ppm, indicating that His37 H-bonds only with water.

At pH 8.5, the high-temperature spectrum retained the dominant water cross peak, but a weak signal at ∼8 ppm was also detected and can be assigned to Hε2. Although the channel does not conduct protons at this pH, some waters are still present, for example between His37 and Trp41 ^8,9,29, thus allowing polarization transfer to ¹⁵N. At low temperature, the unprotonated nitrogen, mainly Nδ1, exhibits a ¹H cross peak at ∼5 ppm due to Hε1, as verified by the spectrum of amino acid histidine (Fig. S1)²⁷.

These low-temperature ¹H chemical shifts are smaller than expected for an LBHB or a strong H bond, while the high-temperature ¹H chemical shifts reveal the H-bonding partner to be water. Thus the data support the direct His37-water interaction model and rule out the His-His LBHB-dimer model. The 1.65 Å crystal structure at pH 6.5 ⁹ detected tightly-clustered water molecules near the His37 tetrad, with N…O distances as short as 2.8 Å, also supporting direct His37-water interactions. On the other hand, all experimental data so far, including the initial ¹⁵N NMR spectra from which the dimer model was proposed ¹⁰, show an absence of imidazole-imidazolium H-bond. An LBHB entails either a single ¹⁵N peak at the averaged chemical shift between N and NH for equal-well potentials or two ¹⁵N peaks centered around the averaged frequency for unequal-well potentials. Instead, the ¹⁵N spectra showed a single peak away from the averaged chemical shift, without the partner peak. Molecular modeling of the HxxxW structure ¹¹ was questionable since it used the putative LBHB as a starting distance restraint to enforce the expected geometry during MD simulations. Finally, the LBHB model implies a hydrophobic environment for the donor and acceptor with a very small pK_a difference ³⁰, which contradict the observed different proton affinities of Nε2 and Nδ1 in His37 and the high hydration of this residue.

In conclusion, temperature-dependent ¹H chemical shifts of His37 sidechain indicate that His37-water H-bonding and proton exchange dominate the equilibrium structure of the His37 tetrad throughout the whole pH range. His-His interactions are indirectly mediated by water. If direct His-His H-bond is too transient to be detectable by NMR, then it cannot be an LBHB and cannot provide stabilization for the dimer state. We propose that charges are stabilized by water-mediated interactions and by cation-π interaction between His37 and Trp41.

This temperature-dependent ¹H chemical shift approach avoids the difficulty of measuring N…N and N…O distances across a H-bond by NMR; moreover it reveals the structure of the most essential player in a proton relay chain. It is applicable to both biological and synthetic proton conductors to understand the nature of the H-bond in proton transport.