Permeation and Activation of the M2 Ion Channel of Influenza A Virus*

The M2 ion channel protein of influenza A virus is essential for mediating protein-protein dissociation during the virus uncoating process that occurs when the virus is in the acidic environment of the lumen of the secondary endosome. The difficulty of determining the ion selectivity of this minimalistic ion channel is due in part to the fact that the channel activity is so great that it causes local acidification in the expressing cells and a consequent alteration of reversal voltage, Vrev. We have confirmed the high proton selectivity of the channel (1.5–2.0 × 106) in both oocytes and mammalian cells by using four methods as follows: 1) comparison of Vrev with proton equilibrium potential; 2) measurement of pHin and Vrev while Na+ out was replaced; 3) measurements with limiting external buffer concentration to limit proton currents specifically; and 4) comparison of measurements of M2-expressing cells with cells exposed to a protonophore. Increased currents at low pHout are due to true activation and not merely increased [H+]out because increased pHout stops the outward current of acidified cells. Although the proton conductance is the biologically relevant conductance in an influenza virus-infected cell, experiments employing methods 1–3 show that the channel is also capable of conducting NH4 +, probably by a different mechanism from H+.

The M 2 protein of influenza A virus is thought to function as an ion channel that permits protons to enter virus particles during virion uncoating in endosomes. In addition, in influenza virus-infected cells, the M 2 protein causes the equilibration of pH between the acidic lumen of the trans-Golgi network and the cytoplasm (reviewed in Refs. 1 and 2). The activity of the M 2 ion channel is inhibited by the antiviral drug amantadine (3)(4)(5). The mature M 2 protein consists of a 23-residue N-terminal extracellular domain, a single internal hydrophobic domain of 19 residues that acts as a transmembrane domain and forms the pore of the channel, and a 54-residue cytoplasmic tail (6). Chemical cross-linking studies showed the M 2 protein to exist minimally as a homotetramer (7)(8)(9). Statistical analysis of the ion channel activity of mixed oligomers also indicated that a homotetramer is the minimal active oligomeric form of the protein (10).
Despite the small size of the active M 2 oligomer, several lines of evidence indicate that ion channel activity is intrinsic to the M 2 protein. First, ion channel activity has been observed in three different expression systems, Xenopus oocytes (3,11,12), mammalian cells (5,13), and yeast (14). Second, M 2 channel activity has also been recorded in artificial lipid bilayers from a reconstituted peptide corresponding to the transmembrane domain of the M 2 protein (15) and from purified M 2 protein (16). Thus, due to its structural simplicity, the M 2 ion channel is a potentially useful model for the study of ion channels in general.
Although a great deal of evidence indicates H ϩ is the biologically relevant ion for the role of M 2 protein in the life cycle of the influenza virus (1,3,(17)(18)(19)(20)(21)(22), other ions have been shown to be capable of flowing through the channel (12). In addition, the ion selectivity measured for the M 2 channel has been found to differ depending on whether the activity was measured in Xenopus oocytes or mammalian cells. When M 2 protein was expressed in oocytes, V rev was found to differ from the proton equilibrium potential, E Hϩ as [H ϩ ] out was varied (12). On the other hand, when M 2 protein was expressed in MEL cells, V rev was found to agree with E Hϩ (5). In a recent study (23), we found I Hϩ of the M 2 ion channel to be so large that it was capable of decreasing [H ϩ ] out in the locale of the extracellular pore of the channel if the expressing cells were bathed in medium of low buffer concentration. One possible explanation for the different results may be that the channel is also capable of acidifying the interior of some expressing cells, thereby altering reversal voltage, V rev . Shimbo and co-workers (12) found that replacement of Na ϩ with Li ϩ decreased currents, and replacement of Na ϩ with NH 4 ϩ increased currents. In principle, these effects could have resulted from one of two mechanisms. Either these ions affected the proton current, I Hϩ , or the replacing ions permeated the M 2 ion channel. In this study, we were able to study the effects of these ion replacements on proton currents specifically by taking advantage of the finding that inward H ϩ currents are limited when the concentration of buffer in the bathing medium is reduced (23) to distinguish between these possibilities.
The M 2 ion channel current is increased in amplitude when the pH of the extracellular domain is lowered (3,5,24). This increase in current occurs within the range of pH values expected for titration of histidine (24). The only amino acid in the transmembrane domain of the M 2 protein with a titratable group in this pH range is His 37 , and when His 37 is replaced by Ala, Gly, or Glu, the proton selectivity of the channel is greatly reduced, and the channel is conductive over a wider range of pH (3,24). It has been proposed that His 37 forms a selectivity filter for protons and that H ϩ conduction may occur by tautomerization of the imidazole side chain of His 37 (25). Although the H ϩ current of the M 2 ion channel protein is increased by elevated [H ϩ ] in the extracellular medium, this increased current may be due to either the increased abundance of the conducting species or activation of the channel at low pH, or both factors operating together. One way to distinguish pH-dependent changes in activity from the effects of increased abundance of H ϩ at low pH is to compare the efflux of H ϩ from acidified cells that express the M 2 protein to the efflux from acidified cells treated with the electrogenic protonophore FCCP. 1 Cell acidification can be achieved by lowering the pH of the medium bathing M 2 -expressing or FCCP-treated cells. If the M 2 ion channel is indeed activated by low pH out and conversely deactivated by neutral or alkaline pH out , then the efflux of H ϩ should be greater for FCCP-treated cells than for M 2expressing cells upon return to a bathing solution of neutral or alkaline pH.
In this study we measured ionic currents and pH in in two M 2 expression systems to ensure that the results obtained were not specific to the cell type. The results demonstrate that under normal physiological conditions the M 2 ion channel specifically conducts H ϩ . We also demonstrate that NH 4 ϩ can permeate the channel, by a mechanism that differs from that for H ϩ permeation. Furthermore, by comparison of the outward currents of acidified, M 2 -expressing cells and FCCP-treated cells, we confirm that M 2 ion channel activity is modulated by the pH of the solution bathing the extracellular N-terminal domain of the channel.

EXPERIMENTAL PROCEDURES
mRNA Synthesis-The cDNA to the A/Udorn/72 mRNA was cloned into the BamHI site of pGEM3 such that mRNA sense transcripts could be generated by using the bacteriophage T7 RNA polymerase promoter and T7 RNA polymerase. For in vitro transcription, plasmid DNAs were linearized downstream of the T7 promoter and the M 2 cDNA with XbaI. In vitro synthesis and quantification of 7m G(5Ј)ppp(5Ј)G-capped mRNA was carried out as described previously (3).
Culture and Infection of CV-1 Cells-CV-1 cells were cultured and infected with recombinant simian virus 40 expressing the M 2 protein from influenza A/Udorn/72 (rSV40-M 2 ), as described previously (13). Briefly, CV-1 cells grown to confluency at 37°C, 5% CO 2 in culture media (Dulbecco's modified Eagle's medium ϩ 10% fetal calf serum ϩ penicillin ϩ streptomycin) were trypsinized, pelleted, and resuspended in culture medium. Resuspended cells were incubated in the presence of high titer SV40-M 2 (100 l of resuspended CV-1 cells ϩ 1 ml of virus stock) for 4 h. Infected cells were then diluted 1:1 in culture medium and seeded onto 5-mm square glass coverslips arranged in 3.5-cm Petri dishes (2 ml total volume/dish). Infected cells were then incubated for 48 h before recording to ensure adequate M 2 protein expression.
Measurement of Membrane Current of CV-1 Cells-M 2 currents were recorded from CV-1 cells using the whole cell patch clamp technique as described previously (13). Briefly, patch pipettes having tip diameters of ϳ2-3 m were pulled from borosilicate capillary glass, fire-polished, and then partially filled with pipette solution which contained, in mM, 145 KCl, 5 EGTA, 1 MgCl 2 , 5 NaCl, 15 HEPES, pH 7.4, adjusted with KOH, osmolality 300 -310 mosmol/Kg. Pipettes filled with this solution typically had resistances of ϳ3-4 M⍀. CV-1 cells attached to glass coverslips were transferred to a recording chamber filled with a solution that contained, in mM, 140 NaCl, 5.3 KCl, 0.55 MgSO 4 , 1.8 mM CaCl 2 , 5.5 glucose, 15 HEPES, pH 7.4, or 15 mM MES, pH 6.2, osmolality ϳ300 mosmol/kg. Seals (in excess of 10 G⍀) were made by gently pressing the patch pipette against a CV-1 cell and then applying ϳ12 mm Hg suction without delay. The whole cell configuration was achieved by using a brief pulse of high voltage combined with gentle pipette suction. In the whole cell configuration, cells had access resistances of Ͻ10 M⍀. Cells were generally bathed in pH 7.4 solution and held at Ϫ20 mV. Whole cell currents were recorded after the bathing solution was changed from pH 7.4 to pH 6.2 using a Fast Step Perfusion System (model SF77B, Warner Instruments Corp., Hamden, CT). By using this system, solution changes could be made in less than 100 ms. M 2 -specific currents were identified by sensitivity to block by 100 M amantadine.
Measurement of Membrane Current of Oocytes-Whole cell currents were measured using a two-electrode voltage clamp. Electrodes were filled with 3 M KCl, and the oocytes were bathed in either Barth's solution, which contained, in mM, 88 NaCl, 1 KCl, 2.4 NaHCO 3 , 0.3 NaNO 3 , 0.71 CaCl 2 , 0.82 MgSO 4 , 15 HEPES, pH 7.5, osmolality ϳ210 mosmol/kg or a modified solution during the recording. Continuous current-voltage (I-V) relationships were measured with ramps of membrane voltage since the M 2 channel shows no rapid voltage-or time-dependent gating. These ramps typically spanned a range of 120 mV in 2 s. Oocyte holding potential was Ϫ20 mV unless stated otherwise.
Measurement of pH in of Oocytes-Microelectrodes were silanized and filled with protonophore as described previously (12). The electrodes were calibrated before each experiment with four pH values spanning the range encountered during the experiment. The response time of these electrodes, determined with a stepping motor device that changed solution pH within 100 ms, was less than 10 ms.
Measurement of pH in of CV-1 Cells-We used the fluorometric indicator 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM) t measure pH in . Control cells or cells infected with rSV40-M 2 were incubated (37°C for 1 h) in a solution containing BCECF-AM in 0.25% Me 2 SO carrier with a final dye concentration of 0.25 g/ml. These cells were placed on the stage of an epifluorescence microscope equipped with a ϫ 20, 0.75 NA (Nikon) objective that allowed up to seven CV-1 cells to be imaged in its field at one time, an intensified CCD camera and MagiCal image analysis software (Applied Imaging, Sunderland, UK). The dye was excited with 490 nm illumination to observe pH-induced changes in fluorescence. To measure the intracellular concentration of dye and thus allow calibration of pH in of the CV-1 cells, illumination was applied at 435 nm (the isosbestic wavelength) at the beginning and end of the measurements from one field of cells. Emission was recorded at 520 nm. Calibration of the pH in from fluorescence measurements was done using the FCCP equilibration method (26). Briefly, the cells were treated with the protonophore FCCP to allow equilibration of the [H ϩ ] across the plasma membrane. The emission at 520 nm as a result of excitation at 490 and 435 nm was measured while the cell was bathed in solutions with pH spanning the range of pH values expected to be encountered during the measurements (pH 4.0, pH 6.7, and pH 9.0), and the resulting ratios (F 490 /F 435 ) were used to construct a calibration curve (see Equation 1 of Ref. 26).

Reversal Voltage Changes within a Few Seconds after Lowering the pH of the Bathing Solution
The M 2 protein from influenza A/Udorn/72 virus was used for this study. If the M 2 ion channel is highly proton-selective, then the reversal voltage of the current (V rev ) should change according to the equilibrium potential for H ϩ (E Hϩ ) when the difference between pH in and pH out is altered. The reversal voltage of the currents of cells expressing the M 2 protein can be measured from continuous current-voltage relationships measured with ramps of membrane voltage because the M 2 ion channel is not voltage-activated on the time scale of the ramps of voltage that are practical to use.
Oocytes-We measured the V rev of amantadine-sensitive currents in M 2 -expressing Xenopus oocytes at 20-s intervals using two-microelectrode voltage clamp. Oocytes whose membrane voltage was clamped to Ϫ20 mV produced a large amantadine-sensitive inward current (ϳ1 A) when the pH of the bathing solution was lowered from pH 8.5 to pH 5.8 (Fig. 1A). During the time when the oocyte was bathed in low pH medium, V rev , measured using voltage ramps, became more positive within a few seconds and reached a maximum prior to the time when the inward current reached its maximum amplitude (Fig. 1B). The average peak V rev , calculated from the current-voltage relationship of the amantadine-sensitive current, was 51.4 Ϯ 1.22 mV S.E. (n ϭ 34), a value more negative than that for E Hϩ (85.2 mV Ϯ 1.86 mV S.E., n ϭ 34) which was calculated from the known pH out and the value of pH in measured with a pH microelectrode at the beginning of the experiment. We calculated the permeability of H ϩ relative to that for Na ϩ assuming 20 mM intracellular Na ϩ concentration for these 34 cells, using the Goldman-Hodgkin-Katz (GHK) equation, and we found that the relative permeability was about 1.5 ϫ 10 6 . However, in two cells that had small amantadine-sensitive currents, the peak V rev nearly reached the value calculated for E Hϩ (ϩ89 and ϩ91 mV). At the time when the peak value of V rev occurred, the current-voltage relationship was steeper than that obtained at pH 8.5, indicating that the conductance had increased. Shortly after reaching its peak value, the inward current decreased in amplitude over a period of a few minutes (Fig. 1A). The current decrease was accompanied by a return of V rev back toward more negative potentials.
CV-1 Cells-The membrane currents and V rev were measured using the whole cell patch clamp technique in CV-1 cells that expressed the M 2 protein (Fig. 2), whereas pH out was reduced in less than 100 ms from pH 7.4 to pH 6.2 with a stepping-motor device (see "Experimental Procedures"). The pH in of CV-1 cells, measured fluorimetrically in a separate experiment described below, was used to calculate E Hϩ . The peak V rev measured in a solution of pH 6.2 was 41.8 Ϯ 3.5 mV S.E. (n ϭ 12), and E Hϩ was 38.9 Ϯ 4.7 mV S.E. (n ϭ 13). In less than 1 min after lowering pH of the bathing solution, V rev reached a peak near E Hϩ and then returned to more negative values (Fig. 2). We calculated the permeability of H ϩ relative to that for Na ϩ assuming 2 mM residual intracellular Na ϩ concentration for these CV-1 cells, using the GHK equation, and found that the relative permeability was about 1.8 ϫ 10 6 .
Since V rev changed after reaching a peak in the low pH solutions in a rapid and systematic manner in both M 2 -expressing CV-1 cells and oocytes, we studied the reason for the systematic changes in V rev , and we employed other means to study the ion selectivity of the channel. There are a number of possible explanations for the decrease of inward current and the return of V rev to more negative voltages after reaching a peak value that was observed in M 2 -expressing cells bathed in low pH solutions. 1) The influx of ions through the M 2 channel could activate an endogenous outward current that opposes the current flowing through the M 2 channel. 2) The M 2 ion channel might undergo an activity dependent change in ion selectivity.
3) The M 2 ion channel might inactivate after long periods in low pH solutions. 4) A constant influx of protons through the M 2 channel might cause acidification of the cell cytoplasm and thus decrease the driving force on protons.
For explanations 1 and 2, the possibility was tested that the shift in V rev observed in oocytes and CV-1 cells might be due to activation of an amantadine-insensitive, endogenous current or an activity-dependent change in ion selectivity by measuring currents under ionic conditions chosen to minimize all but H ϩ currents. The experiments were conducted in CV-1 cells, for which it was possible to control the composition of both the intracellular and extracellular solutions. We found that it was not possible to perform these experiments by changing the internal composition in oocytes using the cut-open technique because small leaks that developed were indistinguishable from M 2 currents, except by application of amantadine, the effects of which are not reversible on the time scale of these experiments. The principal endogenous currents of CV-1 cells were found to be similar to those of HEK293 cells, i.e. inward Cl Ϫ and outward K ϩ currents (27). Whole cell currents were measured in CV-1 cells when Cl Ϫ in the bathing solution was replaced with methane sulfonate and KCl in the pipette solution with tetraethylammonium chloride in order to reduce the endogenous currents of the cells. As before, V rev was measured at frequent intervals as the external pH was lowered. Upon lowering the extracellular pH from pH 7.4 to pH 6.2, a large inward current developed in M 2 -expressing CV-1 cells held at Ϫ20 mV, similar to the results obtained with control solutions (Fig. 2). This inward current was accompanied by a shift of V rev to positive values. The peak V rev (38.8 Ϯ 2.01 mV S.E., n ϭ 5) was close to the value of E Hϩ (calculated as above, 38.9 Ϯ 4.7 mV S.E., n ϭ 13). As observed in the control solution, the inward current began to decrease a few seconds after reaching its maximum amplitude. The decrease of inward current was A, time course of the membrane current as the pH of the bathing medium was lowered from pH 8.5 to pH 5.8 while membrane voltage was held at Ϫ20 mV. Note that the current decreased spontaneously while the bathing solution was held at pH 5.8 and that the current was fully inhibited by 100 M amantadine (Amant). Upper interrupted line represents zero current, and lower interrupted line represents maximal inward current amplitude. Vertical deflections are the result of the voltage ramps used to measure the current-voltage relationships such as those shown in B at the times indicated after pH was reduced. B, current-voltage relationship measured in the same cell shown in A. Note the return of V rev to negative values and the increase in conductance that occurred after 2 min in low pH bathing medium. E Hϩ is shown with the vertical interrupted line and was calculated from the known pH out and pH in measured using an intracellular pH electrode. again accompanied by a return of V rev to more negative voltages and an increase in the slope conductance of the I-V relationship. The fact that the return of V rev to negative values still occurred in the absence of other conducting ions suggests that the change of V rev was not the result of either activating an endogenous current or an activity-dependent change in ion selectivity. We calculated the permeability of H ϩ relative to that for Na ϩ using the mean values of 38.8 and 38.9 mV for the V rev and E Hϩ, respectively, and assuming 2 mM residual intracellular Na ϩ concentration for these CV-1 cells, using the GHK equation, and we found that the relative permeability was about 2 ϫ 10 7 . However, if the actual values differed by as little as 3 mM, the relative permeability would have been about 2 ϫ 10 6 .
For explanation 3, it was also unlikely that the decrease of current and negative shift of V rev observed in low pH was the result of inactivation of the M 2 ion channel. This is because the slope conductance of the I-V relationship actually increased in both oocytes and CV-1 cells during the shift (Figs. 1B and 2). This observation is the opposite of what would have been expected if the channel had been inactivating. The remaining and most likely explanation for the return of V rev to more negative values for cells bathed in low pH solutions was that intracellular acidification occurred.
For explanation 4, we tested whether the return of V rev to negative potentials and the decrease in amplitude of the inward current observed after their peak values occurred were the result of cell acidification by measuring the shift under conditions designed to minimize acidification. These conditions were achieved as discussed in the following two paragraphs.
If the return of V rev to more negative values is due to acidification, then a larger negative shift should be observed when there is a larger inward driving force on H ϩ . This can be obtained either by lowering the pH of the bathing solution further or by making the holding voltage still more negative than E Hϩ . Since most cells become unstable in very low pH solutions (ϽpH 5.5), we decided to modulate the size of the inward current of oocytes by varying the cell holding voltage while the cells were bathed in Barth's solution of pH 5.8 (Fig.  3). Holding oocytes at voltages more negative than Ϫ20 mV produced larger inward currents, accelerated the onset and rate of current decrease, and resulted in a faster rate of negative shift of V rev over time (Fig. 3A). It was also found that V rev could be directed to more positive values following the shift in V rev in low external pH by making the holding voltage more FIG. 3. Alteration of V rev as a result of manipulation of holding potential. A, holding potential of an oocyte expressing the M 2 protein was adjusted to each of several values (top record) and V rev was measured using voltage ramps as shown in Fig. 1B. B, holding the membrane at ϩ50 mV after reducing pH out from pH 8.5 to pH 5.8 prevented V rev from returning to more negative values. Note that V rev was closer to E Hϩ for holding potentials closer to E Hϩ and that V rev was able to be manipulated toward both more negative and more positive voltages. Holding at a membrane voltage of Ϫ20 mV caused a rapid decrease in V rev . The interrupted line in A and B shows E Hϩ calculated from the known pH out and the pH in measured with an intracellular pH electrode.

FIG. 2. Current-voltage relationships recorded from a CV-1 cell ex-
pressing the M 2 protein at pH 7.4 at various times (numerals above curves) after the pH of the bathing medium was lowered from pH 7.4 to pH 6.2. Note that the return of V rev to more negative values and the increase in conductance that occurred with time in low pH bathing medium were faster than those measured in oocytes (Fig. 1B). E Hϩ (vertical interrupted line) was calculated from the known pH out and the pH in was measured fluorometrically in separate experiments. positive (Fig. 3A). Furthermore, the return of V rev to negative values could be prevented by holding membrane voltage at a large positive value that was close to E Hϩ (Fig. 3B). These results demonstrate that V rev can be manipulated by holding voltage, consistent with the current that flows at the holding voltage being capable of altering pH in of the cell.
To minimize the acidification of M 2 -expressing cells in low pH solutions, the intracellular buffer concentration was increased. Again, as the ionic composition of the ooplasm could not be controlled, this experiment was performed in M 2 -expressing CV-1 cells using the whole cell patch clamp technique. We studied the effect of elevated concentrations of the buffer of the pipette solution. The results obtained from pipettes containing 15 and 120 mM HEPES buffer on V rev and the amplitude of the inward current as pH was lowered from pH 7.4 to pH 6.2 were compared. It was observed that for both 15 and 120 mM buffer in the pipette, lowering the external pH led to an increase of inward current in cells held at Ϫ20 mV within a few seconds. It was also observed that for both 15 and 120 mM buffer in the pipette, after lowering the external pH, the amplitude of the inward current reached a peak and then decreased within a few seconds, and during the decrease of inward current V rev returned to more negative voltages. Thus, changes of V rev could not be prevented by increasing the intracellular buffer concentration. Reports in the literature show that in order to control intracellular pH adequately, even with high concentrations of buffer in the pipette solution, the pipette diameter must be at least 1 ⁄3 of the diameter of the cell (28). The pipette diameters used in our experiments were on average 3-4 m diameter, and the CV-1 cells from which we recorded were ϳ100 -150 m diameter. The small ratio of pipette diameter to cell diameter used in these experiments probably explains why even a high buffer concentration did not stabilize V rev , consistent with poor stabilization of pH in . Taken together, these results indicate that the consistent change of V rev to more negative values is the result of cell acidification while bathed in solutions of low pH.

Measurements of pH in during Exposure to Low pH out
It was found with direct measurements of pH in that the return of V rev to negative values observed in M 2 -expressing cells bathed in solutions of low pH was indeed accompanied by cell acidification.
For Oocytes-Intracellular pH was measured using an electrode, and the cells were voltage-clamped to measure the membrane current, V rev , and membrane conductance (Fig. 4A). Lowering the pH out from pH 8.5 to pH 5.8 produced a large inward current and a rapid shift of V rev to more positive values, as noted earlier (Fig. 1). Immediately after reaching a peak, the amplitude of the inward current began to decrease. The V rev reached a peak value prior to the amplitude of the inward current and returned to more negative voltages as the amplitude of the inward current decreased. Measurement of pH in during this time (Fig. 4A) revealed that pH in did not change immediately after introduction of low pH bathing solution, despite the presence of a detectable inward current. However, after ϳ120 s bathing in low pH solution, just after the amplitude of the inward current reached a maximum, pH in began to decrease steadily. The onset of this acidification lagged the decrease of inward current and negative shift of V rev by ϳ100 s. Plots of V rev versus pH in (Fig. 4B) revealed that the initial shift of V rev to more positive values and subsequent return to more negative values occurred independently of changes in pH in that were recorded with a pH microelectrode. After ϳ150 s in the low pH solution, the observed changes in V rev and pH in occurred together. Recovery of pH in to control values was ob-served both when the pH of the bathing solution was returned from pH 5.8 to pH 8.5 and also when 100 M amantadine was added to the solution (pH 5.8). The recovery of pH under both of these conditions followed an attenuation of the inward current. These results demonstrate that oocytes expressing the M 2 protein acidify when there is a large inward H ϩ current. This H ϩ influx causes an acidification of the cytoplasmic solution accessible to the pH microelectrode after a delay with respect to the time when V rev reaches its peak value. These results are consistent with a diffusional delay for H ϩ in the cytoplasm of the expressing cells, between the membrane and the location of the tip of the pH electrode. This would result in a delay between a decrease of pH at the cytoplasmic opening of the pore of the M 2 channel, which determines V rev , and the decrease of pH at the tip of the pH microelectrode. These changes were reversible upon return to bathing medium of pH 7.4 (Fig. 5A). Cells treated with amantadine did not undergo this change in pH in when bathed in medium of pH 6.2. The membrane voltage of the CV-1 cells expressing the M 2 protein was not clamped in these measurements, and thus it was possible that alterations of driving force might have influenced the membrane proton currents. To control membrane voltage, we took advantage of the potassium ionophore valinomycin to help maintain the membrane voltage at a value determined by the ratio of [K ϩ ] across the membrane. This was done by introducing valinomycin (20 M final concentration in 0.02% Me 2 SO carrier; exposure to carrier alone produced no changes in fluorescence) into the bathing medium. It was found that the rate of acidification increased for lower [K ϩ ] of the bathing medium, consistent with increased driving force for protons caused by a more negative membrane potential (Fig.  5B).
The systematic variation of V rev observed with both M 2expressing oocytes and CV-1 cells has important implications for determining the ion selectivity of this channel. When measured in low pH solutions, the value obtained for V rev will depend upon the time when it is measured. This time dependence of V rev is probably the result of acidification of the bulk solution accessible to the cytoplasmic mouth of the pore of the M 2 ion channel. Thus, even the most appropriate measurement, that of the peak value of V rev , is likely to be distorted by acidification.

Ion Substitution Studies
The possibility that the inward current of the M 2 ion channel might in part be carried by ions other than the proton was tested by replacing other extracellular ions with large, presumably impermeant, ions. We tested for Na ϩ permeability by replacing Na ϩ , the major extracellular cation, with other ions. These experiments were performed in oocytes. We also replaced NaCl with mannitol. Changes in the peak V rev , conductance, and pH were measured after reducing pH of the bathing medium. The principle of this experiment is that if Na ϩ normally flows through the channel, replacing Na ϩ with an impermeant cation should decrease the amplitude of the inward FIG. 4. Measurement of pH in , V rev , and membrane conductance of an oocyte expressing the M 2 channel as it was exposed to a bathing solution of low pH. A, time course of the changes. Note that pH in (top graph) decreased after V rev (middle graph) reached its peak value while conductance (lower graph) increased steadily during exposure to low pH bathing medium. B, plot of V rev against pH in for each of the times shown (in seconds) after lowering pH out from pH 8.5 to pH 5.8. The relationship displayed three phases as follows: (i) from 0 to 20 s, immediately after pH out was reduced, V rev reached a peak value, whereas pH in changed very little; (ii) from 20 to 120 s when V rev returned to more negative values with a small (ϳ0.1 pH unit) change in pH in ; and (iii) for t Ͼ120 s when pH in changed slowly as V rev reached its plateau value. current but not change the acidification rate. NaCl in the extracellular medium was replaced with equimolar concentrations of N-methyl-D-glucamine Cl, LiCl, NH 4 Cl, or iso-osmotically with mannitol. To determine the effect of these Na ϩ substitutions, we first measured pH in , conductance, and V rev in the control solution (containing NaCl), at both pH 8.5 and pH 5.8. This was followed by measurements in a pH 5.8 solution in which NaCl was replaced. Finally, the measurements were repeated in the control solution at pH 5.8 to check for reversibility before applying 100 M amantadine. Oocytes were bathed in control solution of pH 8.5 between exposures to low pH solutions to allow recovery from intracellular acidification. A full recovery of pH in to control values typically took 15-20 min in the pH 8.5 solution. As the M 2 ion channel is closed at this pH and there was no residual current, it is thought that the restoration of pH in to control values was the result of a non-electrogenic endogenous H ϩ exchanger.
It was found that substitution of Na ϩ with large, presumably impermeant, cations such as NMDG ϩ or replacing NaCl isoosmotically with mannitol had no detectable effect on peak V rev , conductance, or oocyte acidification rate in low pH solutions (Table I). This result demonstrates that the M 2 ion channel does not conduct detectable amounts of Na ϩ ions. However, when NaCl was replaced with LiCl or NH 4 Cl the results differed oppositely from those in control solutions. Replacement of NaCl with LiCl decreased conductance and acidification rate (Table I) but had no detectable effect on the peak V rev (Fig. 6A), and the peak of V rev occurred at about the same time as it did in Na ϩ -containing solutions, about 20 s after changing solutions. Replacement of NaCl with NH 4 Cl, on the other hand, increased conductance, increased acidification rate (Table I), and shifted V rev to potentials more positive than those observed  in NaCl (Fig. 6B). The peak of V rev also occurred in NH 4 ϩcontaining solutions at about the same time as it did in Na ϩcontaining solutions, about 20 s after changing solutions. This increase in conductance was fully sensitive to amantadine (100 M) and control uninjected oocytes exposed to NH 4 ϩ -containing solutions at pH 6.2 or lower did not display inward currents (in contrast, we have found that oocytes bathed in NH 4 ϩ containing solutions at pH 7.5 or above exhibit large endogenous currents (29)). These results can be interpreted by either Li ϩ and NH 4 ϩ replacing Na ϩ in permeating the pore, flowing independently through the pore, or Li ϩ actually interfering with conduction through the pore. To distinguish among these possibilities, low external buffer concentrations were used to limit specifically the component of current carried by protons, I Hϩ.

Currents Measured with Low External Buffer Concentration to Limit I Hϩ
Advantage was taken of the limitation of H ϩ currents that can be achieved for M 2 -expressing cells by reducing the buffer concentration of the bathing medium (23). This means was used to limit H ϩ currents to determine if the alterations in amplitude of the M 2 current we observed with Li ϩ and NH 4 ϩ were due to an effect on H ϩ currents or due to an effect on other ionic currents. The limitation of H ϩ currents from low external buffer concentration results from a decrease in the [H ϩ ] near the extracellular mouth of the pore of the M 2 ion channel (23). This decrease in [H ϩ ] is reflected in a decrease of current seen during a 2-s-long voltage clamp pulse (Fig. 7). If ion substitution inhibits H ϩ conduction through the channel, then the decrement in amplitude of the inward current during a voltage clamp pulse applied while bathed in a solution of low buffer concentration should be proportional to the decrease of current due to the ion replacement. If, on the other hand, the replacing ion permeates the channel by a mechanism independent of H ϩ conduction, then the decrement in amplitude of the inward current while bathed in a solution of low buffer concentration should be unaffected by the replacement. The decrease in amplitude of the inward M 2 current was recorded during a 2-s hyperpolarizing voltage clamp pulse to Ϫ120 mV from a holding voltage of Ϫ20 mV at pH 5.8 in the presence of low (0.15 mM) buffer concentration in the bathing medium. We measured the diminution in current amplitude during the pulse and the final current amplitude at the end of the pulse in low buffer, Na ϩ -containing medium. We then measured these variables in low buffer media in which Na ϩ was replaced by Li ϩ or NH 4 ϩ , and we compared the values. Replacement of Na ϩ by Li ϩ in a Note the decrease in conductance for the Li ϩ substitution and the increase in both conductance and V rev for the NH 4 ϩ substitution. Measurements were made 20 s after changing solutions, at the time when V rev reached its peak value (see Fig. 1). solution containing 0.15 mM buffer caused a decrease in the steady-state current amplitude, measured at the end of the hyperpolarizing pulse (to 64 Ϯ 2.7% S.E., n ϭ 10). We found that the diminution of current amplitude during the 2-s voltage pulse was reduced proportionally (to 63 Ϯ 4.5% S.E., n ϭ 10) in a medium of low buffer concentration (Fig. 7). This result is consistent with Li ϩ interfering with proton conduction through the M 2 ion channel.
Replacing Na ϩ with NH 4 ϩ , on the other hand, caused an increase in the steady-state current amplitude in low buffer, measured at the end of the hyperpolarizing pulse (to 315 Ϯ 20% S.E., n ϭ 8). However, the buffering capacity of this solution, which contained 0.15 mM MES at pH 6.2, was unavoidably increased by the presence of 88 mM ammonium buffer (to 0.66 mM total buffering capacity at pH 5.8). The decrement of current amplitude during the voltage pulse was reduced to 61 Ϯ 5.5% S.E. (n ϭ 8) of that observed in Na ϩ -containing low buffer (0.15 mM) solutions (Fig. 7), an effect that could be attributed to the modest increase of buffering capacity imparted by the 88 mM NH 4 ϩ Cl solution (Mould et al. (23)). However, the large increase in final current amplitude in the NH 4 ϩ -containing solution cannot be explained by this modest increase in buffer capacity and demonstrates that an additional current flows which is not affected by external buffer concentration. Similar results were obtained when Na ϩ was replaced with NH 3 OH ϩ (data not shown). Finally, replacement of Na ϩ with NMDG ϩ and NaCl with mannitol had no detectable effect on either the final current amplitude at the end of the pulse or the diminution of current during voltage pulses in low external buffer concentration (data not shown).

Comparison of M 2 -specific Oocyte Acidification with That
Obtained from "Pure" Proton Currents Using an Electrogenic Protonophore A final way that we tested the ion selectivity of the M 2 ion channel was by comparing the acidification rate, normalized to the amplitude of the inward current, of M 2 -expressing cells bathed in solutions of low pH with that obtained in cells exposed to the electrogenic protonophore FCCP (30). If the M 2 channel conducts only protons, then for a given inward current amplitude the acidification rate of an oocyte expressing the M 2 protein at low pH should be equal to that observed in the presence of FCCP.
Oocytes Treated with FCCP-Membrane current, V rev , and pH in over time were measured in uninjected oocytes bathed in pH 5.8 Barth's solution in the presence or absence of 20 M FCCP (in 0.02% Me 2 SO carrier; exposure to carrier alone produced no membrane currents). Oocytes clamped at a holding voltage of Ϫ20 mV developed an inward current at low pH after FCCP was added to the bathing medium (Fig. 8A, upper record). The amplitude of the inward current of FCCP-treated cells was generally less than that of M 2 -expressing cells studied at the same pH (compare upper and lower records of Fig. 8A), but it was found that applying higher concentrations of FCCP resulted in deterioration of the condition of the cells. The inward current normally appeared within 1 min of exposing cells to FCCP, a delay that was probably due to the time required for incorporation of FCCP into the oocyte plasma membrane. V rev measured using voltage ramps shifted within a few seconds to positive potentials after the pH of the bathing medium was reduced (Fig. 8B, lower record). As observed for M 2 -expressing oocytes held at Ϫ20 mV, the V rev of oocytes exposed to FCCP (70.4 Ϯ 1.6 mV S.E., n ϭ 10) was close to E Hϩ predicted from the known pH out and pH in measured with an intracellular pH electrode (81.8 Ϯ 5.2 mV S.E., n ϭ 10). After reaching a maximum, V rev began to return to more negative potentials, and the current began to decrease in amplitude a few minutes after the pH of the bathing medium was decreased (Fig. 8). The pH in of oocytes treated with FCCP did not begin to decrease until about 100 s after the changes in current and V rev occurred (Fig.  8B, upper record). One important difference was noted between the behavior of oocytes expressing the M 2 protein and those into which FCCP had been incorporated. When the pH of the bathing solution of FCCP-treated oocytes was returned to pH 8.5 following oocyte acidification, a large, transient outward current appeared (Fig. 8A, upper record). The appearance of this transient outward current was accompanied by an overshoot of V rev to potentials more negative than those observed before lowering pH of the bathing medium (Fig. 8B, lower  record). This was in contrast to the measurements of M 2 -expressing oocytes for which no outward current flowed upon return to bathing medium of pH 8.5 after prolonged bathing in low pH medium (Fig. 8A, lower record). The incorporation of FCCP into the plasma membrane appeared to be reversible, as re-exposure of oocytes to medium of pH 5.8 following washout of FCCP in medium of pH 8.5 failed to produce an inward current (data not shown).
The maximal rate of acidification, normalized to the maximum inward current amplitude, of M 2 -expressing oocytes was compared with the same value obtained from uninjected oocytes treated with the FCCP ionophore. It was found that the ratio of maximal rate of acidification to maximal inward current to be similar in both cases (0.085 Ϯ 0.006 pH unit/min/A for FCCP-treated cells, n ϭ 8, versus 0.081 Ϯ 0.11 pH unit/ min/A for M 2 -expressing cells, n ϭ 20).
CV-1 Cells Treated with FCCP-We performed two types of experiments. In the first type of experiment CV-1 cells that did not express the M 2 protein were employed, and membrane currents were measured at low pH out in the presence of FCCP (Fig. 9). In the second type of experiment, we studied the effect of FCCP treatment on the pH in of M 2 -expressing cells after inhibiting M 2 currents with amantadine (Fig. 5A). The results of both types of experiments were similar to those obtained with oocytes. Lowering the pH of FCCP-treated cells that did not express the M 2 protein from pH 7.4 to pH 6.2 resulted in an inward current flow accompanied by an initial increase in V rev to positive voltages and an increase in conductance (Fig. 9). If the holding voltage was adjusted to give very little inward current while the cells were bathed in low pH medium, the measured V rev was 65.8 Ϯ 3.5 mV S.E. (n ϭ 6). This value was close to the 69.6 mV value of E Hϩ calculated from the known pH out and an assumed pH in of pH 7.4. However, when the holding voltage was made more negative, the V rev quickly returned to more negative voltages (Fig. 9).
The effect of FCCP on pH in was also measured in CV-1 cells (Fig. 5A). In this experiment, M 2 -expressing cells were exposed first to pH 6.2 for ϳ200 s and then allowed to recover from acidification in a bathing medium of pH 7.4. The cells were then reexposed to pH 6.2 in the presence of 100 M amantadine, and in this solution acidification did not occur. Finally, the same cells were exposed to pH 6.2 medium in the presence of FCCP for 200 s. This treatment resulted in acidification once again. Finally, the cells were allowed to recover from acidification in medium of pH 7.4 ϩ FCCP. Bathing FCCP-treated cells in solution of pH 6.2 resulted in an acidification that occurred with a slightly more rapid time course than that observed for M 2 -expressing cells (Fig. 5A). The pH in of the M 2 -expressing cells was pH 6.96 Ϯ 0.15 pH units S.E. (n ϭ 5) at pH 7.4 out before application of FCCP, and pH in decreased by 0.88 Ϯ 0.095 pH units (n ϭ 5) after reducing pH out from pH 7.4 to pH 6.2 in the presence of FCCP. As also observed with oocytes, these changes of pH in were reversible upon return to bathing medium of pH 7.4. It was not possible to study CV-1 cells in solutions of pH 8.5 because irreversible changes occurred at this alkaline pH value. These two experiments demonstrate that acidification of CV-1 cells can also be achieved by treatment with FCCP and that acidification results from a mechanism that is not affected by the presence of the M 2 protein. DISCUSSION This study confirms the very high proton selectivity of the M 2 ion channel under physiological conditions, demonstrates that Li ϩ inhibits the channel, and provides additional evidence that the channel allows the permeation of quaternary ammonium ions, probably by a different mechanism than that for H ϩ . These results also provide evidence for the restricted diffusion negative values during exposure to FCCP at low pH out , and the overshoot of V rev to very negative potentials after return to pH out to pH 8.5.

FIG. 8. Effects of FCCP on oocytes.
A, comparison of effects of lowering pH of the bathing medium on the membrane currents of uninjected oocytes exposed to FCCP (upper record) and oocytes expressing the M 2 protein (lower record). Note that when pH out was decreased from pH 8.5 to pH 5.8 the uninjected cell did not display an inward current until FCCP was introduced. Upon return to pH out of pH 8.5, the FCCP-exposed cell developed a large outward current (upper record), whereas the M 2 -expressing cell did not (lower record). The interrupted line shows the zero current. B, pH in and V rev of oocytes treated with FCCP. Time course of pH in and V rev measured in a single uninjected oocyte when pH out was lowered from pH 8.5 to pH 5.8. Note the decrease of pH in after lowering pH out to pH 5.8 only occurred in the presence of FCCP. Note the recovery of pH in when pH out was returned to pH 8.5. Full recovery occurred but is not shown. Note also that V rev increased to a peak positive value after FCCP was applied and returned to more Note that V rev was near E Hϩ when cells were held at ϩ30 mV and that V rev became negative when the membrane voltage was held at Ϫ30 mV for the times indicated.
of H ϩ in the cytoplasm of oocytes and support the notion that the M 2 channel is gated by changes in pH out .
These results are consistent with the proposed roles for M 2 ion channel in the life cycle of the influenza A virus. However, the only known ion channel encoded by influenza A virion is the M 2 ion channel, and a pure proton conductance for the M 2 channel would depolarize the virion membrane. This depolarization would elevate virion membrane potentials to high enough values to risk dielectric breakdown of the membrane and decrease the driving force on protons. Thus, this depolarization would limit the extent of virion acidification that is possible. A simple calculation can estimate the number of protons that could flow into the virion without causing an excessive virion membrane potential that would lead to dielectric breakdown of its lipid bilayer. If we assume that the virion starts with a membrane potential of about Ϫ70 mV (the resting potential of many epithelial cells) and assume that it can withstand a membrane voltage of ϩ100 mV, then about ϳ2.5 ϫ 10 Ϫ22 Eq of protons will be able to enter the virion during acidification, changing its membrane voltage by altering the charge across its membrane capacitance, before placing the virion membrane in danger of dielectric breakdown. Assuming that the volume of the virion is about 1.5 ϫ 10 Ϫ19 liters and that the interior of the virion is well buffered, the pH inside the virion will decrease by only a few tenths of a pH unit if the pH decrease is distributed uniformly throughout the interior of the virion. The reason that the calculated pH decrease is so small is that every proton that enters the virion carries a charge that increases the membrane voltage of the virion, but only a small fraction of the entering protons contribute to the free proton concentration (lower pH) because the buffering capacity of the virion proteins can sequester the majority of the protons. Thus, the calculated decrease of pH of the virion, if it occurred uniformly throughout the virion, would only lower pH of the virion from an assumed initial value of approximately pH 7.4 to a value slightly higher than pH 7.0. This decrease of pH is inadequate to cause release of ribonucleoprotein complexes from the M1 matrix protein (31) contained within the entire virion. Perhaps the acidification of only a zone of the virion immediately below the virion surface membrane might be adequate (32,33), as this is the zone where the M1 protein is concentrated. It is intriguing that we found evidence for a localized pH decrease in the zone near the cytoplasmic entrance to the pore of the M 2 channel in oocytes.
The Difficulty of Determining M 2 Ion Selectivity by Comparing V rev with E Hϩ Calculated from Measured pH in -In a previous study Shimbo et al. (12) found that when M 2 -expressing oocytes were bathed in a solution of low pH, V rev was 20 -30 mV more negative than the value predicted from E Hϩ calculated from the pH in measured with a micro pH electrode and the pH of the bathing solution. In that study, V rev measurements were made ϳ2 min after the oocytes were exposed to bathing solutions of low pH, at approximately the time when the amplitude of the inward membrane current reached a maximum value, to ensure adequate equilibration of the low pH solution. The relative permeability of H ϩ to that of Na ϩ was found to be about 10 5 . In the present study it was found that upon lowering extracellular pH, V rev measured in M 2 -expressing oocytes shifted within a few seconds toward positive potentials near E Hϩ and then fell to more negative values before inward current reached its maximum amplitude (Fig. 1B, oocytes). Similar results were obtained in CV-1 cells that expressed the M 2 protein (Fig. 2, CV-1 cells). The return of V rev to more negative values was not the result of activating an endogenous current, as the shift still occurred in CV-1 cells expressing the M 2 protein in the absence of other conducting ions. The return of V rev to negative values was also not due to inactivation of the M 2 ion channel, as the slope conductance increased as V rev became more negative. This finding is consistent with results obtained for HEK293 cells expressing the Kv2.1 delayed rectifier channel (34). In HEK293 cells the activity of the channel produces changes in [K ϩ ] in that mimic inactivation and results in apparent changes in ion selectivity. By using the peak value of V rev and the pH in measured at the time of peak V rev , the permeability of H ϩ relative to that of Na ϩ was found to be about 1.5-2.0 ϫ 10 6 , consistent with the values found by Chizhmakov and co-workers (5) and consistent with results from reconstituted M 2 protein in vesicles (35,36). For CV-1 cells studied in the absence of Cl Ϫ and K ϩ ions and two oocytes studied in control solution, the measured values of V rev and the calculated values of E Hϩ were within a few mV, consistent with a high proton permeability.
Two lines of evidence indicate that acidification of the cytoplasm of cells expressing the M 2 ion channel is the explanation for the return of V rev to more negative values, after reaching a peak positive value, when cells are bathed in medium of low pH. First, the rate of the return of V rev was smaller when cells were held at voltages closer to E Hϩ (Fig. 3). Second, for times longer than 120 s after introduction of the low pH bathing solution, the change in V rev and the change in pH in had a similar time course (Fig. 4).
Several observations indicate that H ϩ diffusion in oocytes is restricted by the presence of immobile buffers (37)(38)(39). First, the decrease in pH in occurred with a delay after the onset of inward current (Fig. 4). Second, the return of V rev to negative values following the occurrence of its peak value near E Hϩ occurred prior to the change in pH in detected using an intracellular pH electrode (Fig. 4). The data presented here are also consistent with previous conclusions that it is very difficult to control the pH in of a cell when the diameter of the patch pipette is much smaller than that of the cell. Even with very high concentrations of buffers, the diameter of the patch pipette needs to be no less than one-third of the cell diameter in order to control pH in adequately (28). As the CV-1 cells used in our patch clamp experiments were 100 -150 m diameter and the largest pipettes we were able to use were 3-4 m diameter, it would be expected that pH in was not well controlled in our experiments, even when high concentrations of buffer were used. Chizhmakov and co-workers (5) found that the V rev of M 2 -expressing MEL cells was close to the value predicted for E Hϩ without taking special precautions to minimize I Hϩ at the holding voltage, and a change in V rev with time was not reported. The most likely reasons for these data are that the currents of these cells are much smaller than those of CV-1 cells, that MEL cells are rather small (ϳ10 m diameter) and thus are small in relation to the patch pipettes usually employed, and that the patch pipette contained high concentrations of buffer.
Ion Substitution and Low Buffer Studies-Two other ways that we tested the ion selectivity of the M 2 channel were to replace Na ϩ in the bathing medium with other impermeant cations or NaCl iso-osmotically with mannitol and to reduce the buffer concentration in the bathing medium in order to limit the H ϩ currents. The effects of these alterations were studied in M 2 -expressing oocytes while the V rev , acidification rate, and conductance were measured. Replacement of NaCl with Nmethyl-D-glucamine or iso-omotically with mannitol had no detectable effect on V rev , acidification rate, or conductance. In addition the decrease of current amplitude due to low external buffer concentration was unchanged by these substitutions. These results are consistent with the interpretation that the M 2 ion channel does not conduct detectable amounts of Na ϩ