Chemical Rescue of Histidine Selectivity Filter Mutants of the M2 Ion Channel of Influenza A Virus*

The influenza virus M2 proton-selective ion channel activity facilitates virus uncoating, a process that occurs in the acidic environment of the endosome. The M2 channel causes acidification of the interior of the virus particle, which results in viral protein-protein dissociation. The M2 protein is a homotetramer that contains in its aqueous pore a histidine residue (His-37) that acts as a selectivity filter and a tryptophan residue (Trp-41) that acts as a channel gate. Substitution of His-37 modifies M2 ion channel properties drastically. However, the results of such experiments are difficult to interpret because substitution of His-37 could cause gross structural changes to the channel pore. We described here experiments in which partial or, in some cases, full rescue of specific M2 ion channel properties of His-37 substitution mutants was achieved by addition of imidazole to the bathing medium. Chemical rescue was demonstrated for three histidine substitution mutant ion channels (M2-H37G, M2-H37S, and M2-H37T) and for two double mutants in which the Trp-41 channel gate was also mutated (H37G/W41Y and H37G/W41A). Currents of the M2-H37G mutant ion channel were inhibited by Cu(II), which has been shown to coordinate with His-37 in the wild-type channel. Chemical rescue was very specific for imidazole. Buffer molecules that were neutral when protonated (4-morpholineethanesulfonic acid and 3-morpholino-2-hydroxypropanesulfonic acid) did not rescue ion channel activity of the M2-H37G mutant ion channel, but 1-methylimidazole did provide partial rescue of function. These results were consistent with a model for proton transport through the pore of the wild-type channel in which the imidazole side chain of His-37 acted as an intermediate proton acceptor/donor group.

The M2 protein of influenza A virus permits protons to enter virus particles during virion uncoating in endosomes, and the M2 channel also causes the equilibration of pH between the acidic lumen of the trans-Golgi network and the cytoplasm (1)(2)(3). The activity of the M2 ion channel is inhibited by the antiviral drug amantadine (4 -6). The mature M2 protein con-sists of a 23-residue N-terminal extracellular domain, a single internal hydrophobic domain of 19 residues that acts as a transmembrane (TM) 1 domain and forms the pore of the channel, and a 54-residue cytoplasmic tail (7). Chemical cross-linking studies (8 -10) and statistical analysis of the ion channel activity of mixed oligomers (11) showed the active state of the M2 ion channel protein to exist minimally as a homotetramer.
Despite the small size of the active M2 oligomer, several lines of evidence indicate that ion channel activity is intrinsic to the M2 protein. First, ion channel activity has been observed in three different expression systems, Xenopus oocytes (4, 6), mammalian cells (6,12), and yeast (13,14). Second, M2 channel activity has also been recorded in artificial lipid bilayers from a reconstituted peptide corresponding to the TM domain of the M2 protein (15) and from purified M2 protein (16,17). Thus, due to its structural simplicity, the M2 ion channel is a potentially useful model for studying ion channels in general.
The ion selectivity of the M2 ion channel has been studied with voltage clamp (18,19) and ion flux studies (17,19) indicating that the channel is nearly perfectly selective for protons. The channel is inactive for extracellular pH values higher than pH 7.5 but becomes active when extracellular pH is lowered. This greater activity is due to two factors, greater abundance of protons and increased channel opening due to an interaction between protonated His-37 and the indole side chain of Trp-41, the putative "gate" of the ion channel (20,21). Experiments in which His-37 of the TM domain was replaced by site-directed mutagenesis with Gly, Ala, Glu, Lys, and Arg (22) indicate that His-37 is essential for the proton selectivity of the channel and its activation by low pH. However, it is possible that substitution of even a single residue in the closely packed pore of this protein might bring about large structural changes in the protein, making a comparison of the mutant and wild-type (wt) channels difficult. For some enzymes that transport protons as part of their catalytic cycle, substitution of histidine in the active site with alanine or glycine resulted in loss of enzymatic activity, e.g. carbonic anhydrase II (23)(24)(25), copper amine oxidase (26), aldolase (27), (S)-mandelate dehydrogenase (28), bacterial luciferase (29), the reaction center of photosynthetic bacteria Rhodobacter spheroides (30), and protein kinases (31). However, one way to overcome the argument that the amino acid substitution caused gross structural changes in the protein was achieved by partial or complete chemical rescue of the mutant protein on introduction of imidazole or an imidazole analog into the buffer solution. This chemical rescue demonstrated that the loss of catalytic function resulted from the lack of the histidine residue and not from large scale structural changes. Chemical rescue experiments also provided insight into the function of the histidine residue in the wt enzyme, providing direct evidence that the imidazole side chain of histidine acts as an intermediate proton acceptor/donor, in which protons were accepted by the histidine molecule from one chemical moiety and subsequently donated to a second chemical moiety.
Chemical rescue of mutant ion channels has been reported in only a few instances. However, an important property of the M2 ion channel suggested that histidine substitution mutants might be susceptible to chemical rescue by imidazole. This property is the ability of the channel to be inhibited by amantadine, a molecule of approximately the same size as imidazole. We reasoned that it was likely that the diameter of the channel pore might be large enough to accommodate the imidazole molecule, and we applied imidazole buffer to three histidine substitution mutants. It was found that partial rescue was possible for each mutant M2 ion channel and that the imidazole-enhanced currents were able to be inhibited by Cu(II). These results confirm the role of the imidazole side chain of histidine in proton transport across the M2 channel pore and support the hypothesis that the imidazole side chain of histidine acts as an intermediate proton acceptor/donor to relay protons through the aqueous pore of the M2 ion channel while obstructing the flow of larger cations.
Recording  Tables I  and II, Na ϩ ions were substituted with equimolar concentrations of N-methyl-D-glucamine (NMDG) or K ϩ . Cl Ϫ ions were substituted with equimolar concentrations of methanesulfonic acid. Titration of 50 mM imidazole buffer used in chemical rescue and ion substitution experiments required the addition of significant amounts of HCl, resulting in a concomitant increase in osmolality. In this case, control solutions were made isoosmotic by addition of mannitol. In experiments with 50 mM buffer (either MES or imidazole), the solutions contained (in mM) 88 Na ϩ , 1 KCl, 2.4 NaHCO 3 , 50 buffer (MES or imidazole or HEPES), 0.3 NaNO 3 , 0.71 CaCl 2 , and 0.82 MgSO 4 . The osmolality of solution with imidazole buffer was ϳ260 mosM/kg. The osmolality of the control solutions buffered with MES and HEPES was adjusted with mannitol to ϳ260 mosM/kg. The composition of solutions used in Na ϩ ion substitution experiments (Table III and Table IV 8.5) or 15 MES/MOPSO/imidazole/lutidine/ 1,2,4-triazole/1-methylimidazole (for pH 5.5), osmolality ϳ202 mosM/kg. The order of presentation of test solutions, the time of recovery in normal solution, and the time during which oocytes were exposed to low pH solutions were important for obtaining recordings in which reversible changes in current flow and acidification occurred and are described in Supplemental Information 1.
Measurement of Membrane Current and Intracellular pH of Oocytes-Whole-cell currents were measured using a two-electrode voltage clamp amplifier (Dagan TEV 200A) at ϳ25°C using electrodes filled with 3 M KCl. Oocyte holding potential was Ϫ20 mV unless stated otherwise. Recording of intracellular pH was done with silanized microelectrodes filled with protonophore as described previously (19).
Immunofluorescence of Living Oocytes-The relative expression levels of the mutant M2 proteins at the surface membrane of the oocytes were measured with respect to that of the wild-type protein. This was done by incubating the oocytes after recording with a solution containing a monoclonal antibody directed against the N-terminal ectodomain of the influenza A M2 protein (monoclonal antibody 14C2) (33). Individual oocytes were washed with ND96 lacking pyruvate and gentamycin (4°C, three times), incubated in ND96 containing 2% bovine serum albumin (4°C, 1 h), incubated in primary antibody (1:500 in 2% bovine serum albumin, 4°C, 1 h), washed with ND96 (4°C, three times for 10 min), incubated in secondary antibody (goat anti-mouse IgG 1 (␥1) labeled with Alexa Fluor® 546 (catalog number A21123, Molecular Probes (Medford, OR), 10 or 20 g/ml in 2% bovine serum albumin in ND96, 4°C), and washed with ND96 (4°C, five times for 10 min). Fluorescence was quantified using a PTI image master microfluorometer (London, Ontario, Canada) with a 20ϫ 0.5 NA objective. Approxi-  Ion selectivity of the wt and the M2-H37G mutant ion channels. V rev of the amantadine-sensitive component of current was measured. V rev measured in oocytes expressing the wt ion channel is always more negative than predicted from extracellular pH and resting intracellular pH because oocytes acidify rapidly when bathed in low pH medium (18). Note that V rev for the mutant ion channel is significantly decreased by lowering ͓Na ϩ ͔, but the difference is less at pH 5.5 than at pH 7.5 (see Table I). Reversal voltage for the wt channel can only be measured at low pH when the channel is open and is unaffected by ͓Na ϩ ͔. Reversal voltage for the M2-W41A channel is not affected by ͓Na ϩ ͔; the V rev shown was for oocytes with an average current of approximately the same amplitude at pH 5.5 as for oocytes expressing the wt channel; V rev for this very active channel quickly changes due to oocyte acidification.  with 50 mM imidazole buffer Effect of imidazole buffer on the ion selectivity of the M2-H37G mutant ion channel. V rev was measured using the amantadine-sensitive component of current. Note that V rev for the mutant channel was less affected by lowering [Na ϩ ] than when HEPES buffer was used (Table I) and that oocyte acidification (induced by lowering pH of the bathing medium from pH 8.5 (HEPES) to imidazole-buffered solution at pH 7.5 for 30 sec) was not affected by reduced [Na ϩ ] (see Supplemental Information 2). mately 20% of the lower oocyte surface was imaged upon the photocathode of the CCD camera. The emission spectrum of the fluorescence was confirmed to peak at 546 nm, consistent with acceptably low autofluorescence of the oocyte. For quantification of the fluorescence of each oocyte, the excitation wavelength was 556 nm, and emission was measured for wavelengths longer than 574 nm. Calculation of Free [Cu 2ϩ ]-The HYSS program (www.chem. leeds.ac.uk/People/Peter_Gans/HySS.htm) was used to calculate the amount of total Cu(II) salt needed to achieve the desired free concentration. This program uses dissociation constants from the NIST data base (version 46).

RESULTS
When oocytes expressing the wt M2 protein are bathed in solutions of low pH, but not solutions of high pH, an inward proton current flows across their membrane, and the cytoplasm becomes acidified (Fig. 1A). Previous work had shown that oocytes expressing histidine substitution mutant proteins have membrane currents, but these currents flowed at both high pH and low pH (22). We tested whether proton flow was included in the currents of oocytes expressing the histidine substitution mutant proteins by recording intracellular pH of the oocytes and found that acidification occurred when the oocytes were bathed in solutions of low pH (Fig. 1B), showing that protons can pass through the pore of the mutant ion channel in the absence of histidine (see Supplemental Information 2). However, results with mutant enzymes that normally contain histidine as part of a proton transport pathway indicated that proton transport rates are increased when imidazole buffer is used (23)(24)(25). We thus applied imidazole buffer to oocytes expressing histidine substitution mutant proteins, and we found that current amplitude and acidification were increased significantly (p Ͻ 0.05, n ϭ 4) (Fig. 1B). Increases were not observed for oocytes expressing the wt protein (Fig. 1A). These observations suggested that the presence of imidazole buffer might cause the properties of the mutant ion channels to resemble more closely those of the wild-type ion channel (i.e."rescue" the channel properties). We thus analyzed the key properties of the mutant ion channels with and without imidazole buffer. Our studies focused on the M2-H37G mutant ion channel because the currents of oocytes expressing this genotype were more reproducible from experiment-to-experiment than those of oocytes expressing the other mutant ion channels.

Ion Selectivity of Histidine Substitution Mutants in MES Buffer
The ion selectivity for the M2-H37G, M2-H37S, and M2-H37T mutant ion channels was studied by measuring their current-voltage relationships in media of various pH, buffer, and ionic composition ( Fig. 2 and Tables I and II). These mutant ion channels displayed much less proton selectivity than the wt ion channel when studied at high pH (pH 7.5 or pH 8.5) and low pH (pH 5.5) in MES buffer.
Measurements at High pH-Oocytes expressing these mutant ion channels had standing currents when bathed at pH 7.5 or pH 8.5 ( Fig. 2A). The standing current for the M2-H37G mutant ion channel became outward when Na ϩ in the bathing medium was replaced with NMDG (holding voltage Ϫ20 mV). This suggested that the mutant ion channel conducted Na ϩ , unlike the wt channel that does not conduct Na ϩ at all. To test if Na ϩ carried the inward current of the M2-H37G mutant ion channel at pH 7.5, reversal voltage (V rev , Table I) was measured and found to be less negative in media containing Na ϩ (Ϫ7.6 Ϯ 0.2 mV, n ϭ 3 mean Ϯ S.E.) than in media containing NMDG (Ϫ42 Ϯ 1.7 mV, n ϭ 3), consistent with the conclusion that Na ϩ is conducted through the mutant channel.
Measurements at Low pH-Oocytes expressing the wt M2 ion channel and the M2-H37G mutant ion channel had steady inward currents at pH 5.5 (Ϫ20 mV holding voltage, Fig. 2A). The inward current amplitude of oocytes expressing the M2-H37G mutant ion channel decreased when Na ϩ in the medium . Current is fully amantadinesensitive (arrow labeled amantadine). B, M2-H37G mutant ion channel. Current and acidification rates are much larger upon lowering pH of the bathing medium for imidazole buffer than for MES buffer (p Ͻ 0.05, n ϭ 4). Note that imidazole buffer results in larger amplitude of current and acidification for the mutant ion channel and that upon return to pH 8.5 buffer (unlabeled arrow, upper) an outward current flowed, indicating that the channel did not close at this high pH. Amantadine was applied at the time shown by the labeled arrow. Buffer concentrations, 50 mM; amantadine concentration, 100 M.

TABLE IV
Reversal voltages measured at pH 5.5 with 50 mM buffer Effect of imidazole buffer on the ion selectivity of the wt M2 ion channel and the M2-H37G mutant ion channel. V rev was measured using the amantadine-sensitive component of current. Note that V rev for the mutant ion channel changes much less when Na ϩ is replaced in imidazole buffer than in MES buffer.
was replaced by NMDG, suggesting that the mutant ion channel has Na ϩ conductance at low pH. However, the amplitude of the decrease achieved by this replacement was not as large when studied at pH 5.5 (0.55 Ϯ 0.05 A, n ϭ 6) as it was at pH 7.5 (0.99 ϩ 0.2 A, n ϭ 3; p Ͻ 0.05). Consistent with the interpretation that the channel has Na ϩ conductance at pH 5.5, Na ϩ replacement made the V rev considerably more negative for oocytes expressing the M2-H37G mutant ion channel (Table  II). In contrast, neither inward current amplitude nor the V rev (Table II) of oocytes expressing the highly proton-selective wt channel was affected by replacement of Na ϩ in the bathing medium with NMDG (6,18,19). To test for the possibility that the mutant ion channels also conduct K ϩ at pH 5.5, [K ϩ ] in the bathing medium was increased to 88 mM by replacing Na ϩ . The value of V rev found under this condition was considerably more positive than for bathing medium containing NMDG and approximately equal to that for medium containing Na ϩ . Moreover, the amplitude of the inward current was larger in the medium with 88 mM K ϩ than in medium with 88 mM Na ϩ . The results of these alterations of V rev with the composition of the bathing medium and the larger amplitude of inward current in high K ϩ medium indicate that the mutant channel also has significant conductance for K ϩ (Table II). To test for possible conduction of Cl Ϫ by the M2-H37G mutant ion channel, [Cl Ϫ ] in the bathing medium was replaced with methane sulfonate, but the V rev did not change significantly (data not shown). Thus, the M2-H37G mutant protein forms a poorly selective cation channel.

Ion Selectivity of Histidine Substitution Mutant Ion Channels in Imidazole Buffer
To test the dependence of ion selectivity of the M2-H37G mutant ion channel on the concentration of imidazole buffer in the bathing medium, we measured the current-voltage relationship (Fig. 2B) and V rev for oocytes that were bathed in solutions with each of several concentrations of imidazole buffer between 1.5 and 50 mM at pH 5.5 (Fig. 3). For these measurements we studied the amantadine-sensitive component of current. Neither the amplitude of the inward current (Fig. 2B) nor the V rev of oocytes expressing the wt channel varied significantly with imidazole concentration (Fig. 3A), nor was V rev affected by replacement of Na ϩ in the bathing medium with NMDG (6,18). However, the V rev of oocytes expressing the M2-H37G mutant ion channel became more positive with increased imidazole concentration ( Fig. 2B and Fig. 3B), consistent with increased proton conductance resulting from higher imidazole concentration. The current-voltage relationship for the M2-H37G mutant ion channel was increased by replacing MES buffer with imidazole buffer (Fig. 2B); this increase in slope occurred for both inward and outward current. In contrast, the current-voltage relationship for the wt ion channel was changed very little by imidazole buffer (Fig. 2B). To test ion selectivity further, we measured the effect on V rev of replacement of Na ϩ in the bathing medium with NMDG (50 mM buffer, pH 5.5; Table IV). This was done while the cells were bathed in medium buffered with either MES or imidazole (pH FIG. 2. Comparison of the properties of the wt M2 ion channel with those of histidine substitution mutant ion channels. A, inward current amplitude (total current) is plotted versus pH for oocytes expressing the wt (n ϭ 5), the M2-H37G (n ϭ 5), M2-H37S (n ϭ 6), and M2-H37T (n ϭ 5) mutant proteins. Note that the currents of the mutant ion channels are less dependent upon pH than those of the wt ion channels. B, current-voltage relationships of the wt M2 channel (upper) and the M2-H37G mutant channel (lower). Note that substantial currents remain in the presence of amantadine for the mutant channel (4) and that the slope conductance of the channel is not significantly affected by the pH of the bathing medium in HEPES versus MES buffers (1 versus 2). Imidazole buffer (3) significantly increases the slope at low pH over that in MES buffer (2).

FIG. 3. Dependence of reversal voltage (V rev ) on imidazole concentration for wt M2 ion channel (A) and the M2-H37G mutant ion channel (B).
Note that V rev for the wt ion channel (n ϭ 5) is independent of imidazole concentration within this range but that V rev becomes more positive for the mutant ion channel (n ϭ 5), approaching that for wt, for increasing imidazole concentration (p Ͻ 0.0001). 5.5). The V rev for the M2-H37G mutant was much less affected by Na ϩ replacement when the oocyte was bathed in solutions buffered with imidazole than when it was bathed in solutions buffered with MES (Table IV, paired t test, p Ͻ 0.05). When measured at pH 7.5 in imidazole buffer (Table III), the V rev of oocytes expressing the M2-H37G mutant ion channel differed slightly for bathing media containing Na ϩ and NMDG (p Ͻ 0.0001, n ϭ 7), in contrast to the results obtained at pH 5.5 in imidazole buffer (Table IV) for which the V rev was unchanged. Examination of the reversal voltages for oocytes expressing the M2-H37S and M2-H37T mutant channels revealed that their ion selectivity was improved by imidazole (see Supplemental Information 3). Thus, the presence of imidazole buffer changed the ion selectivity of the histidine substitution ion channels to make it similar to that of the wt ion channel.

Specific Activity of Histidine Substitution Mutant Ion Channels in Imidazole Buffer
A notable characteristic of the wt M2 ion channel is its low single channel conductance or specific activity (17,34). We therefore compared the relative activity of the histidine substitution mutant ion channels expressed in individual oocytes with that of the wt ion channel by first measuring the oocyte current in imidazole buffer (50 mM) and then quantifying the relative surface expression of M2 protein expressed in each oocyte with immunofluorescence using an antibody specific for the N-terminal ectodomain of the protein (33). This epitope would not be expected to be altered by mutation of the transmembrane domain histidine residue (monoclonal antibody 14C2) (see "Experimental Procedures" and Fig. 4). The currents of the oocytes expressing the three mutant ion channels were much higher than those expressing the wt ion channel, but the relative levels of surface expression were lower for the mutant ion channel. Relative to the wt ion channel, the ratio of current to immunofluorescence for the mutant ion channels was (mean Ϯ S.E.) M2-H37G, 6.8 Ϯ 1.9, n ϭ 3; M2-H37S, 12 Ϯ 2.3, n ϭ 3; and M2-H37T, 4.1 Ϯ 1.0, n ϭ 3. This finding is consistent with the mutant ion channels having a higher specific activity than the wt ion channel.

Gating of Histidine Substitution Mutant Ion Channels in MES Buffer
Substitution of histidine adversely affects gating by pH out , an important functional property of the M2 ion channel. The wt M2 ion channel is closed when the pH of the solution bathing the ectodomain is high and opens at low pH in order to permit influx of protons into the virion, and thus the plot of activity versus pH of the medium bathing the ectodomain can be thought of as an "activation curve." The wt channel closes to proton efflux when extracellular pH is high, even if cytoplasmic pH is lowered by bathing in acidic medium or by injection of acid into the cytoplasm (21). Thus, current does not flow outward from acidified oocytes expressing the wt ion channel into solutions of high pH (21). Both of these properties are altered in the histidine substitution ion channels. First, the relationship between amplitude of the inward current and pH of the bathing medium is quite steep for the wt ion channel but is not steep for the mutant ion channels (Fig. 2A). Second, oocytes expressing the mutant ion channels have significant currents when the pH of the bathing medium is pH 7.5 or higher. Third, outward currents flow from acidified oocytes after return to neutral pH bathing solution (Fig. 1B, unlabeled arrow). Thus, the histidine substitution mutant ion channels show signs of altered gating function.

Gating of Histidine Substitution Ion Channels in Imidazole Buffer
Modest changes in gating occurred in imidazole buffer for the M2-H37G mutant ion channel. The slope of the current versus pH relationship was greater for the M2-H37G mutant ion channel with imidazole buffer (Fig. 5A) than with MES buffer (see also Fig. 2A), although the range of pH for which the slope of the relationship was maximal differed for the M2-H37G mutant ion channel and the wt ion channel. The maximal slope FIG. 4. Relative specific activity of mutant ion channels measured with imidazole buffer. Surface expression of the mutant proteins was measured with respect to that of the wild-type protein with immunofluorescence of living oocytes after determination of ion channel activity recorded from the same oocytes. The ratio of the current to the surface expression is higher for the mutant ion channels than for the wild-type ion channel, showing that these ion channels have high specific activity. Number of oocytes measured for each genotype are as follows: wt, 7; M2-H37G, 3; M2-H37S, 3; M2-H37T, 3; M2-H37G/W41A, 3; and M2-H37G/W41Y, 4.

FIG. 5. Dependence of amplitude of inward current (A) and percentage amantadine-sensitive current (B) of oocytes expressing the wt M2 (n ‫؍‬ 3) and M2-H37G mutant (n ‫؍‬ 4) proteins upon pH of the bathing medium for two concentrations of imidazole buffer.
Note that, unlike the wt ion channel, the mutant ion channel is not completely inhibited by amantadine at low pH. occurred within the range pH 6.5 to pH 8.5 for the mutant ion channel and the maximal slope occurred within the range pH 4.5 to pH 6.5 for the wt ion channel. Thus, imidazole buffer partially modified gating of the mutant ion channel.

Inhibition of Histidine Substitution Ion Channels by Amantadine and Cu(II)
The wt M2 channel is completely inhibited by amantadine, and this inhibition is essentially irreversible (4,5). In contrast, the activity of the histidine substitution mutant ion channels was only partially inhibited by amantadine. For example, the currents of oocytes expressing the M2-H37G mutant ion channel, studied in MES buffer, were inhibited only 62 Ϯ 2.8% (mean Ϯ S.E., n ϭ 4) when the inhibitor (100 M) was applied in bathing media of either high or low pH (see Fig. 2B and Table VI). The amantadine inhibition of the mutant ion channels was reversible within the time scale of these experiments (ϳ85-100% reversibility in 5 min). Thus, amantadine inhibition of the histidine substitution mutant ion channels was incomplete and reversible, in contrast to the complete and irreversible inhibition found for the wt ion channel. Cu(II) causes a rapid, incomplete inhibition of the M2-H37G mutant ion channel, in contrast to the slow, complete inhibition observed for the wt ion channel (35).

Inhibition by Amantadine and Cu(II) in the Presence of Imidazole Buffer
These two inhibitors have been shown to inhibit the wt ion channel, and we thus tested the influence of imidazole buffer on their ability to inhibit the M2-H37G mutant ion channel. Imidazole buffer increased inhibition of the M2-H37G mutant ion channel by amantadine (100 M). Inhibition measured at pH 5.5 was greater in imidazole buffer (82 ϩ 0.5% mean Ϯ S.E., n ϭ 4) than in MES buffer (62 Ϯ 2.8%), and the inhibition of the mutant ion channel became greater for higher values of pH of the bathing medium with imidazole buffer (Fig. 5B). In the presence of either MES or imidazole buffer, the currents and acidification caused by the wt ion channel were fully inhibited by amantadine (100 M). Possible explanations for these results are as follows: 1) imidazole buffer acted on the M2-H37G mutant ion channel to increase proton current and decrease the current for other cations; and 2) amantadine inhibited only the proton current. The specificity for the proton current could explain the increased inhibition of the mutant ion channel by amantadine.
Inhibition of the wt ion channel by Cu(II) is important because it is known that inhibition occurs by coordination with His-37 (35), and if imidazole buffer is able to occupy a site(s) in the mutant channel pore, Cu(II) would be expected to inhibit the mutant ion channel. We therefore tested the ability of Cu(II) to cause inhibition of inward current and acidification of oocytes expressing the M2-H37G mutant ion channel when bathed in imidazole buffer. The inward current and acidification of the M2-H37G mutant channel were partially inhibited by Cu(II) (220 M) in 15 mM imidazole-buffered solution ( Fig.  6B; inhibition of the inward current was 38 ϩ 0.2% mean Ϯ S.E., n ϭ 4; p Ͻ 0.05). At this value of pH, less than 1% of the total imidazole in the bulk solution was complexed with Cu(II), and thus the inhibition observed could not be attributed to a decrease in the concentration of free imidazole buffer. Although the percentage inhibition of the M2-H37G mutant by Cu(II) was lower in imidazole buffer than in MES buffer, the absolute amount of inhibition was larger in imidazole buffer (0.6 Ϯ 0.06 A) than in MES buffer (0.4 Ϯ 0.1 A, p Ͻ 0.05, n ϭ 3), demonstrating that the imidazole-sensitive portion of the current is inhibited by Cu(II). Most surprisingly, the wt ion chan-

Intracellular Injection of Imidazole Was without Significant Effect Upon the M2-H37G Mutant Ion Channel
If bulk flow of protonated imidazole is responsible for the increased current in imidazole buffer, then injection of imidazole into the cytoplasm would be expected to increase outward currents. Thus, imidazole was injected intracellularly into oocytes expressing the M2-H37G mutant ion channel. The injection of imidazole to a final concentration of 4 -14 mM (confirmed with fluorescent tracer) resulted in an insignificant (Ͻ5%) increase of current amplitude for cells measured in MES buffer (pH 5.5, 50 mM). This is in contrast to the 5-fold increase (500 Ϯ 89%, n ϭ 5) measured when 15 mM imidazole buffer was introduced into the bathing medium.

The Chemical Nature of Rescuing Buffers
To determine whether only the naturally occurring imidazole side chain of histidine was capable of increasing current and conductance of the M2-H37G mutant channel, we applied a number of other buffer molecules while measuring inward current and conductance, and we compared these variables with their values measured in MES buffer (Fig. 7, 15 mM buffer concentration, pH 5.5). MOPSO buffer (pK a 6.9, identical to that of imidazole) did not increase either current or conductance of the mutant ion channel. The currents and conductance with 1-methylimidazole were significantly higher than those measured with MES buffer, but the currents measured with 1,2,4-triazole were not. However, it should be noted that limited proton currents are to be expected for triazole, as the pK a of triazole (2.3) is significantly lower than the pH of the bathing solution (pH 5.5) (34). In contrast to these findings, current and conductance for the mutant ion channel decreased in solutions buffered with 2,6-lutidine. Imidazole buffer produced the greatest increase in inward current and conductance among compounds tested. Thus, the common feature of buffers that are capable of increasing the current or conductance of the M2-H37G mutant channel is that they resemble the naturally occurring side chain of histidine and carry a positive charge in the protonated state.

Independence of Chemical Rescue from Trp-41
Chemical rescue of the proton transfer function of the inactive carbonic anhydrase II H64A mutant protein by 4-methylimidazole achieved about 40% of the wt activity (36,37). The crystal structure of the enzymatically rescued protein showed that the 4-methylimidazole moiety was held in place through -stacking interactions with Trp-5 of the protein; if Trp-5 was replaced with Ala then enzymatic rescue was not achieved (25). An essential, conserved feature of the TM domain of the M2 ion channel protein is the HXXXW motif. We tested whether the increase in current of the M2-H37G mutant ion channel that occurred when MES buffer was replaced with imidazole buffer depended upon interactions with Trp-41. This was done by introducing a second mutation in which Trp-41 was replaced with either tyrosine or alanine. Before doing these experiments, however, we performed tests of the ion selectivity of the M2-W41A mutant ion channel by measuring its V rev in various solutions at pH 5.5 (Table II). We found that V rev was very similar to that of the wt ion channel and that V rev was not altered by ion substitutions, indicating that this mutant ion channel has proton selectivity similar to that of the wt ion channel. The double mutant ion channels were expressed in oocytes, and their current, conductance, and V rev were studied in MES and imidazole buffers (Table V). When imidazole buffer replaced MES buffer in the bathing medium (pH 5.5), inward current amplitude increased, and V rev became more positive for both of these double mutant ion channels (Table V). In addition, conductance increased for both double mutants (ratio of conductance in imidazole buffer to MES buffer was 2.4 Ϯ 0.1, n ϭ 4, for M2-H37G/W41A and 4.4 Ϯ 0.4, n ϭ 3, for M2-H37G/ W41Y). The specific activity of these double mutant proteins was determined by measuring currents and surface expression relative to those for the wt ion channel protein (Fig. 4). Relative to the wt ion channel, the ratio of current to immunofluorescence for the double mutant ion channels was (mean Ϯ S.E.): M2-H37G/W41A, 4.0 Ϯ 0.43, n ϭ 3, and M2-H37G/W41Y, 9.6 Ϯ 5.1, n ϭ 4. Neither double mutant ion channel was significantly inhibited by amantadine (as low as 1% inhibition for the M2-H37G/W41Y mutant ion channel; Table VI). Thus, partial rescue of ion selectivity and proton transport can be achieved independent of Trp-41. DISCUSSION This study confirms that the TM domain histidine residue of the M2 ion channel protein of influenza A virus is essential for its ability to conduct protons with high selectivity and suggests that histidine acts as an intermediate proton acceptor/donor group. When the His-37 was replaced by Gly, Ser, or Thr, the mutant ion channels were capable of transporting protons, but the proton selectivity of each mutant ion channel, measured in MES buffer, was diminished. The use of imidazole buffer improved proton selectivity, increased proton transport, and improved gating of the mutant ion channels. The finding that partial rescue of three histidine substitution mutant ion channel proteins can be achieved with imidazole added to the bathing medium suggests that the mutation of four closely positioned residues in the TM pore did not cause gross structural alterations that would invalidate the comparison of mutant and wt proteins. Rescue is specific to imidazole, the side chain of histidine, supporting the conclusion that histidine plays a key role in proton transport and selectivity. The ability of three different histidine substitution mutant ion channels to be rescued by added imidazole, together with the ability of Cu(II) to inhibit the currents in the presence of imidazole, suggests that rescue was probably achieved by imidazole acting as an intermediate proton acceptor/donor group in which the imidazole molecule accepted protons from the buffered solution on the outside of the ion channel pore and donated them to the solution on the other side of the pore. These results are consistent with histidine acting in the wt ion channel as an intermediate proton acceptor/donor group.

Properties of Histidine Substitution Mutant Ion Channels Studied in MES Buffer-
The histidine substitution mutant ion channels M2-H37G, M2-H37S, and M2-H37T were found to be deficient in ion selectivity, were not activated by low pH, and were not fully inhibited by amantadine. In addition, Cu(II) inhibition of the M2-H37G mutant channel was incomplete and rapidly reversible. However, the present study demonstrated that the M2-H37G mutant ion channel conducts protons and that the proton flux can be partially inhibited by amantadine. The ion selectivity of the M2-H37G mutant ion channel measured in MES buffer, inferred from its V rev , showed that it possesses substantial Na ϩ conductance at pH 8.5, that its relative Na ϩ conductance decreased with pH, that it possesses substantial Na ϩ and K ϩ conductance at pH 5.5, and had undetectably low Cl Ϫ conductance. A possible explanation for the observation that oocytes expressing this mutant ion channel undergo nearly equal amounts of acidification as oocytes expressing the wt ion channel, but with smaller amplitude of inward membrane current than oocytes expressing the wt ion channel (Fig. 1), is that under these conditions considerable outward K ϩ current flows through the mutant channel, balancing the inward proton and Na ϩ current. In summary, although a histidine substitution mutant ion channel is capable of proton transport, the ion selectivity of the channel is greatly compromised by substitution of histidine.
Evaluation of Chemical Rescue-According to four criteria for ion channel function, imidazole buffer achieves substantial chemical rescue of histidine substitution mutant channels. First, ion selectivity is enhanced, in part by preventing Na ϩ ions from being transported (Fig. 3 and Tables III and IV). Second, the transport of protons is increased (Fig. 1). Third, the gating of the channel is partially restored (Fig. 5A). Fourth, sensitivity to amantadine (Fig. 5B) and Cu(II) is increased. However, even in the presence of a high concentration of imidazole buffer, the properties of the mutant channels are not identical to those of the wt channel, and thus the rescue of the histidine substitution mutant ion channels by imidazole buffer is incomplete.
Chemical rescue of ion selectivity of the M2-H37G mutant ion channel by imidazole buffer was probably brought about by a combination of increased proton conductance and decreased conductance for Na ϩ and K ϩ . In the presence of imidazole buffer the rate of acidification was greater than in MES buffer (Fig. 1), and the change in V rev due to replacing Na ϩ in the bathing medium by NMDG was lower than in MES buffer (Tables III and IV). It is unlikely that the latter difference can be explained solely by an increase in proton conductance, and the most likely explanation is that decreased Na ϩ conductance also resulted from bathing in imidazole buffer.
Gating of the M2-H37G mutant ion channel was improved modestly by imidazole buffer (Fig. 5), but this mutant channel did not achieve the same extent of closing at high pH as the wt channel. This might have been due to the inability of the imidazole buffer molecules to interact with the indole side chain of Trp-41 of the channel. In the wt M2 ion channel, gating depends on the interactions between His-37 and Trp-41 (20,21). Current does not flow through the wt M2 ion channel when bathed in solutions of high pH (21,34). The present study found that partial rescue of ion selectivity (Table V) and proton transport are possible for the double mutants M2-H37G/W41A and M2-H37G/W41Y in the presence of imidazole buffer. As partial rescue could be obtained in the absence of Trp-41, it is unlikely that the rescuing imidazole molecules interacted with the Trp-41 gate in the M2-H37G mutant ion channel in order to effect rescue. Thus, rescue of gating was incomplete. This independence from Trp is in contrast to observations of the catalytic function of mutant human carbonic anhydrase II (HCA II-H64A), for which partial enzymatic rescue by imidazole is only possible in the presence of the native Trp-5 residue (36,37). The present findings suggest that the positioning of the rescuing imidazole molecule(s) in the mutant ion channels does not depend upon interaction with Trp-41.
Specificity of Chemical Rescue-Chemical rescue of the M2 ion channel is very specific to imidazole buffer. Buffers that are neutral when protonated (MES and MOPSO) are not able to achieve rescue of proton transport of the M2-H37G mutant ion channel. Of the several buffers tested, the only other one that achieved rescue was structurally related to imidazole (1-methylimidazole), and rescue by this compound was less than that by imidazole itself (Fig. 7).
In a study of proton transfer rates by chemical rescue of the bacterial reaction center, the second order rate constant was found to depend upon the pK a of the rescuing buffer over a 9 log unit range (38). This dependence had a slope of about 10-fold per pH unit, a finding that was interpreted to indicate the presence of an intermediate proton acceptor/donor group located between the source of protons at the surface of the reaction center and their destination at Glu-212 of the reaction center. Unfortunately, an analysis of this type cannot be applied to the M2 ion channel because chemical rescue is specific to imidazole.
Mechanism for Partial Chemical Rescue of Histidine Substitution Mutant Ion Channels-The finding that it is possible for three histidine substitution ion channels to be rescued by imidazole supports a mechanism in which His-37 acts as an intermediate proton acceptor/donor group in these rescued mutant ion channels. This conclusion is further supported by several additional findings. First, imidazole buffer causes an increase in proton transport, as determined by rate of acidification (Fig. 1), and proton selectivity, as determined by relative independence of V rev from [Na ϩ ] in the bathing medium (Tables III and IV). Second, the extent of rescue increases with the concentration of imidazole (Fig. 3). Third, low concentrations of Cu(II) insufficient to reduce bulk free imidazole concentration (see below) inhibit proton transport in the presence of imidaz- ole buffer (Fig. 6). Fourth, intracellular injection of imidazole does not achieve chemical rescue. Thus, the present results point to imidazole functioning as an intermediate proton acceptor/donor in the rescued mutant ion channels. We do not know the location of the rescuing imidazole buffer molecules in the mutant ion channel pore, but indirect evidence suggests that the buffer might interact with the amide carbonyl groups of the pore-lining residues. The pore-lining residues of the wt channel, identified by cysteine-scanning mutagenesis and interaction with transition elements, are Ala-30, Gly-34, His-37, and Trp-41 (4,35,39,40). As the side chains of none of these pore-lining residues are likely candidates for binding to a positively charged buffer molecule, it is suggested that the buffer molecule binds with low affinity to amide carbonyl groups. The need for a high concentration of imidazole (Ͼ15 mM; see Fig. 3) to achieve rescue of function and the observed rapid rate of loss of rescue after removal of imidazole (within 1 s; see Figs. 1 and 6) are both consistent with a low affinity interaction expected for amide carbonyl groups.
Inhibition of currents in the presence of imidazole buffer by Cu(II) is inconsistent with a mechanism for rescue in which bulk imidazole is conducted through the pore of the channel. Inhibition can be observed with 220 M free Cu(II) in the presence of 15 mM imidazole, and under these conditions only about 1% of the imidazole is coordinated with Cu(II). This makes it unlikely that a decrease in the free imidazole concentration in the bulk solution causes the inhibition observed. A much more likely explanation is that Cu(II) coordinates with imidazole molecules that are located in the mutant ion channel pore, inhibiting their ability to bind or conduct protons. This explanation is strengthened by two observations made with the M2-H37G mutant ion channel. First, in the presence of imid-azole buffer the slope of the current-voltage relationship for both outward and inward currents was equally enhanced. As the free imidazole concentration in the cytoplasm is negligible, it is unlikely that protonated imidazole molecules carry an outward proton current. Second, intracellular injection of imidazole is without effect. Thus, it is likely that imidazole molecules located in the pore are responsible for the observed chemical rescue.
We postulate that imidazole molecules act as intermediate acceptor/donors of protons conducted through the pore of the mutant ion channels (Fig. 8). Each of the three mutant ion channel proteins, when studied in MES buffer, is capable of conducting H ϩ , K ϩ , and Na ϩ through their pore (Fig. 8B). These ion channels lack specificity because they lack poreoccluding imidazole side chains of His-37. When studied in imidazole buffer, imidazole occupies the channel pore (Fig. 8C) and alters channel function in two ways. First, imidazole partly occludes the pore, thereby impeding flow of K ϩ and Na ϩ . Second, imidazole molecules act as intermediate proton acceptor/ donor molecules, enhancing H ϩ transport. Cu(II) is postulated to coordinate with pore-occupying imidazole molecules in this model, inhibiting their ability to bind protons, disabling them from acting as intermediate proton acceptor/donor molecules (Fig. 8D).
Implications for Proton Transport by the Wild-type M2 Ion Channel-The mechanism by which protons are transported through the pore of the wt M2 ion channel is not known. However, measurement of the kinetic isotope effect shows that it is unlikely that passage of hydronium ions is the principal mechanism (18). The remaining two possibilities are as follows: 1) His-37 acts as an intermediate proton acceptor/donor group, i.e. each proton binds to His-37 and is subsequently released; FIG. 8. Postulated mechanism for chemical rescue of histidine substitution mutants of the M2 ion channel protein by imidazole. Extracellular solution of low pH is shown in purple (upper portion of each schematic), and intracellular solution of neutral pH is shown in blue (lower), and lipid bilayer is shown in white. Two of the TM domain helices of the tetrameric ion channel pore are shown crossing the bilayer. The conserved TM domain tryptophan residue is omitted from this schematic. A, wt M2 ion channel. Two of the four imidazole side chains are shown occluding the pore for all ions. These imidazole moieties accept protons from the acidic extracellular side of the pore and donate them to the neutral side of the pore. B, histidine substitution mutant ion channel. In the absence of the imidazole side chains the pore is large enough to allow passage of K ϩ , Na ϩ , and H ϩ . C, mutant ion channel with imidazole buffer. The imidazole molecules occupy unidentified critical locations within the pore, inhibiting the flow of K ϩ and Na ϩ but facilitating the flow of H ϩ by serving as proton acceptor/donor that relays protons. D, mutant ion channel with imidazole buffer and Cu(II) inhibitor. Cu(II) coordinates with the imidazole molecules that occupy the mutant ion channel pore to prevent the proton relay mechanism from functioning, thereby causing inhibition of the ion channel. and 2) a "proton wire" forms through which protons are transported (41,42). The finding that the single channel conductance of the channel is very low (17,18) favors the former interpretation, although it is still possible that a very short lived proton-wire could provide for a low average conductance, albeit with a very high conductance during the time of opening (41,42). Our results point to imidazole functioning as an intermediate proton acceptor/donor in the rescued mutant ion channels, and it is consistent that the naturally occurring imidazole side chain of His-37 functions in a similar fashion in the wt ion channel. However, although the present results support the intermediate proton acceptor/donor mechanism, they do not exclude the short lived proton wire mechanism, and further studies will be needed to resolve this question.
In summary, these experiments confirm the essential role of His-37 in the selective transport of protons through the pore of the M2 ion channel protein and support the conclusion that mutation of His-37 in all four subunits of the tetrameric channel does not necessarily cause a gross distortion of the channel architecture. The results suggest that His-37 acts as an intermediate proton acceptor/donor group in the function of the wt ion channel. Three histidine substitution mutant ion channel proteins (M2-H37G, M2-H37S, and M2-H37T), studied in MES buffer, have very high specific activities and function as large, ungated cation pores. It is remarkable that the addition of imidazole buffer to the bathing medium is capable of partial rescue of the selectivity of proton transport and gating of these mutant ion channels.