ATP-gated P2X channels are the simplest of the three families of transmitter-gated ion channels. Some P2X channels display a time- and activation-dependent change in permeability as they undergo the transition from the relatively Na+-selective I1 state to the I2 state, which is also permeable to organic cations. We report that the previously reported permeability change of rat P2X2 (rP2X2) channels does not occur at mouse P2X2 (mP2X2) channels expressed in oocytes. Domain swaps, species chimeras, and point mutations were employed to determine that two specific amino acid residues in the cytosolic tail domain govern this difference in behavior between the two orthologous channels. The change in pore diameter was characterized using reversal potential measurements and excluded field theory for several organic ions; both rP2X2 and mP2X2 channels have a pore diameter of ∼11 Å in the I1 state, but the transition to the I2 state increases the rP2X2 diameter by at least 3 Å. The I1 to I2 transition occurs with a rate constant of ∼0.5 s−1. The data focus attention on specific residues of P2X2 channel cytoplasmic domains as determinants of permeation in a state-specific manner.

## Introduction

ATP-gated cationic P2X channels define one of the three major families of transmitter-gated ion channels (Khakh, 2001). Although there are seven distinct P2X subunits, much of our understanding is based on studies of the P2X2 subunit which forms homomeric channels. Available evidence indicates that each P2X2 subunit possesses two transmembrane domains, cytosolic NH2 and COOH termini and a large cysteine-rich extracellular loop. There is strong evidence that transmembrane domain 2 (TM2) lines the pore, because mutations here affect organic cation permeability (Khakh et al., 1999; Virginio et al., 1999b) and Ca2+ permeability (Migita et al., 2001), and because substituted cysteines in TM2 are modified by extracellular hydrophilic methanethiosulphonate reagents when the channel opens (Rassendren et al., 1997; Egan et al., 1998). The gate appears to be close to a conserved glycine (G342), whereas a conserved aspartate (D349) is internal to it. P2X channel pores are cation selective with almost equal permeability to Na+/K+ and significant permeability to Ca2+.

Some P2X channels also display a time and activation-dependent increase in permeability to organic cations and fluorescent dye molecules before they desensitize. This phenomenon was first reported for, and considered unique to, the P2X7 channel (Surprenant et al., 1996). But subsequent studies demonstrated permeability changes for various recombinant and natively expressed P2X channels in neurons (Khakh et al., 1999; Virginio et al., 1999b). The permeability change of the P2X2 channel is just one example of the recent observations that some channels change their selectivity on the time scale of milliseconds in response to diverse stimuli, including membrane voltage, neurotransmitter binding, and second messengers (Khakh and Lester, 1999). The structural bases and biophysical properties of these changes are poorly understood.

The goal of this study was to provide a more complete understanding of the history-dependent changes in cation permeability at P2X2 channels. We found that the previously reported permeability change of rat P2X2 (rP2X2) channels does not occur at mouse P2X2 (mP2X2) channels expressed in oocytes. We used domain swaps, chimeras, point mutations, reversal potential measurements, kinetics, and excluded field theory to track and quantify permeability changes. Surprisingly, we found that permeation of one P2X2 state (I2) is affected by two specific residues in the cytosolic tail domain, whereas the permeation of a preceding state (I1) is not. We conclude that opening to the I2 state requires conformational changes in the C tail domain, whereas opening to I1 does not.

## Materials And Methods

### Cloning, Expression, and Recording

The mP2X2 cDNA was cloned from a mouse testis library using primers designed to the rat P2X2 cDNA (antisense primer GAA TTC TCA AAG TTG GGC CAA ACC T, sense primer GGA TCC ATG GTC CGG CGC TTG G). Mouse P2X2 cDNA were amplified using PCR Master (Roche) according to manufacturer's instructions. The PCR product was gel purified and cloned into TOPO TA pCR2.1 (Invitrogen) according to manufacturer's instructions. The five individual clones were sequenced (Roche Bioscience sequencing laboratory) on both strands. mP2X2 and rP2X2 cDNA were subcloned into pcDNA3.

The C tail swap chimeras were made by digesting rP2X2 and mP2X2 cDNA with HpaI (native site in both cDNAs) and XhoI (in the 3′ polylinker) to produce two fragments: (a) the vector and coding region for the entire P2X2 cDNA, minus part of TM2 and the C tail, and (b) DNA coding for the remainder of TM2 and the COOH-terminal domain. The DNA fragments were separated by agarose gel electrophoresis and purified (QIAGEN gel extraction kit). Thus, the mP2X2 C tail domain was ligated into rP2X2, and vice versa, and restriction analysis was used to confirm the construction.

The C tail domain chimeras were constructed in two PCR steps (High Fidelity PCR Master; Roche) using synthetic oligonucleotides (Caltech Polymer Synthesis Facility). In the first step, two PCR reactions produced overlapping pieces, one using rP2X2 as the template and the other using mP2X2 as the template. In the second step, the PCR products from the first steps were used as the template and the final product was a connected chimeric fragment spanning the DNA coding for TM2 through the cytoplasmic tale and into the polylinker. The chimeric fragment and wild-type DNA were digested with HpaI and XhoI. The pieces of the digestion were separated on an agarose gel and extracted. The chimeric fragment and the wild type vector were ligated. Sequencing confirmed the correct chimeras.

Single site mutants were made using Quick Change Mutagenesis (Stratagene). DNA sequencing was used to confirm the mutation. All cDNAs were transcribed in vitro using the mMESSAGE mMACHINE kit (Ambion).

Xenopus laevis oocytes were prepared and used for electrophysiological recordings described using described methods. Two-electrode voltage-clamp recording of oocytes was performed using the Geneclamp 500 amplifier (Axon Instruments, Inc.). Electrodes were pulled from borosilicate glass (Sutter Instrument Co.) and back filled with 3 M KCl to yield resistances of 1–2 MΩ. Recordings were made in solution consisting of 98 mM NaCl, 5 mM HEPES, and 1 mM MgCl2 at pH 7.35–7.4, which was superfused over the oocytes by gravity flow at a rate of ∼3 ml min−1 (chamber volume was ∼300 μl). In some experiments, equimolar substitutions of organic cations were made for Na+. The organic cations tested were dimethylammonium, 2-(methyl-amino)-ethanol, Tris+, and N-methyl-d-glucamine. Solutions containing ATP were applied to the oocyte using a solenoid-operated solution switcher (General Valve Company); complete solution exchange around the oocyte occurred within 0.5–1.0 s. Voltage-clamp experiments were controlled by a Digidata 1200 interface and a personal computer running pCLAMP 7 or pCLAMP 8 software (Axon Instruments, Inc.). In some experiments, the voltage was ramped at a rate of 0.36–0.6 mV/ms. Data were filtered at 200–500 Hz and digitized at 3–5 times this rate. Current-voltage relation data were filtered at 1 kHz and digitized at 3 kHz. All experiments were performed at 18–20°C.

### Data Analysis

Data were analyzed using Clampfit (Axon Instruments, Inc.) or Origin 5.0 (Microcal Software, Inc.), and appear in the text and graphs as mean ± SEM from n determinations as indicated (>4). We employed the principles developed by Hille (Dwyer et al., 1980; Hille, 1992). A transform of the GHK voltage equation under bionic conditions from

$Erev=\frac{RT}{zF}\mathrm{ln}\frac{P_{X^{\mathrm{+}}}\left[X^{\mathrm{+}}\right]_{o}}{P_{Y^{\mathrm{+}}}\left[Y^{\mathrm{+}}\right]_{i}}$

to

$\frac{P_{X^{\mathrm{+}}}}{P_{Y^{\mathrm{+}}}}=\left({\left[Y^{\mathrm{+}}\right]_{i}}/{\left[X^{\mathrm{+}}\right]_{o}}\right)\mathrm{exp}\left({zErevF}/{RT}\right)$

provides a way to measure the permeability, P, of ion X+ with reference to ion Y+. However, under the assumption that the intracellular ion concentration is constant, the above equation can be used to measure the permeability of ion X+ relative to Y+, as

$\frac{P_{X^{\mathrm{+}}}}{P_{Y^{\mathrm{+}}}}=\left({\left[X^{\mathrm{+}}\right]_{o}}/{\left[Y^{\mathrm{+}}\right]_{o}}\right)\mathrm{exp}\left(zF{\left(ErevX^{\mathrm{+}}{-}ErevY^{\mathrm{+}}\right)}/{RT}\right)$

where X+ is the organic cation and Y+ is Na+, with a reversal potential close to 0 mV for P2X2 channels (see results). Under fixed ion concentrations this equation simplifies to

$\frac{P_{X^{\mathrm{+}}}}{P_{Na^{\mathrm{+}}}}=\mathrm{exp}\left(\frac{{\Delta}ErevF}{RT}\right)\mathrm{,}$

where ΔErev is the shift in reversal potential and F, R and T have their usual meaning (Khakh et al., 1999). The size of the narrowest region of the pore was determined using excluded field theory (Dwyer et al., 1980; Cohen et al., 1992), which relates ion permeability ratios, and sizes of ions (treated as spheres) to the narrowest region of the pore, which is approximated as a cylinder with a narrow constriction. The theory does not provide information about the shape of the constriction, or the pore, but can be used to measure the dimensions of the narrowest region, as explained below. For a channel pore of radius Rc, and ion X+ with a radius Rx (RNa+ is the hydrated radius of Na+ at 2.15 Å) the PX+/PNa+ is given by

$\frac{P_{X^{\mathrm{+}}}}{P_{Na^{\mathrm{+}}}}=\left(\frac{Rc{-}Rx}{Rc{-}RNa^{\mathrm{+}}}\right)^{2}\mathrm{.}$

A square root transformation gives

$\sqrt{\frac{P_{X^{\mathrm{+}}}}{P_{Na^{\mathrm{+}}}}}=a{-}bRx\mathrm{,}$

where

$a=\frac{Rc}{Rc{-}RNa^{\mathrm{+}}}\mathrm{and}b=\frac{1}{Rc{-}RNa^{\mathrm{+}}}$

and the diameter of the narrowest region of the pore is 2a/b. In the above equations a and b are obtained from linear regressions of plots of (PX+/PNa+)1/2 versus Rx (as in Fig. 4). In such formalisms a convenient check is to determine the diameter of Na+, which is given by RNa+ = (a1)/b, and using a = 1.57 Å−1 and b = 0.28 (average from regressions of rP2X2 I1, mP2X2 I1, and I2 shown in Fig. 4). This gives RNa+ as 2.04 Å, which is close to the hydrated radius (Kielland, 1937) of Na+ in water at 2.15 Å. These numbers indicate that there is ∼5% error in the calculations reported in this study. The radius of ion X+ (Rx) was determined as

$Rx=\frac{\sqrt[3]{D1{\times}D2{\times}D3}}{2}\mathrm{,}$

where D1, D2, and D3 are the three dimensions of the smallest box that will house the ion. The fold change (Δ) in PX+/PNa+ is the ratio of PX+/PNa+ for I1 and I2. Slope conductance values for I1, I2, and INa+ were measured from current-voltage relationships over a range of 40 mV around the reversal potential. The values were calculated separately for each cell, before averaging. For I2 there was a small difference between these values and those calculated from the average values of I and Erev(NMDG) (as shown in Table I). These differences are expected with small numbers for the numerators, as in the NMDG+ data. For Na+ currents, inward rectification was quantified as G+60 mV/G−60 mV. Rate constants (1/τ; s−1) were determined from single exponential fits to the data. Concentration-effect curves were fitted where appropriate, as indicated, to the Hill equation. Statistical tests were performed using the paired or unpaired Student's t test, as appropriate, and a P < 0.05 was taken to indicate significance.

## Results

### Initial Observations with rP2X2 Channels

Fig. 1 illustrates permeability changes for rP2X2 channels expressed in oocytes, and introduces the measurements that are used throughout this paper. Table I contains data for various measurements from nine cells, from experiments of the type illustrated in the representative waveform shown Fig. 1 A. At a holding potential of −60 mV and in extracellular solutions that contain NMDG+, application of ATP evokes an initial transient outward current (termed I1) at a holding potential of −60 mV. Over a time course of seconds the currents become inward and reach steady-state values (I2), but on switching to Na+ solutions the currents are larger (INa+; Fig. 1 A). The INa+/I2 ratio calculated from the individual cells was 7.3 ± 1.9 (n = 9; Table I legend).

We also measured the reversal potential of the ATP-evoked currents between I1, I2, and INa+ with voltage ramps (Fig. 1 B), and expressed these values as permeability ratios relative to Na+ (see materials and methods). The inward I2 current develops with a rate constant (k+1 = 0.3 ± 0.01 s−1) similar to that for the shift in reversal potential (k+1 = 0.4 ± 0.1 s−1) from −66.7 mV for I1 to −35.5 mV for I2, thus showing a change in PNMDG+/PNa+ from 0.07 for I1 to 0.34 for I2 (Table I). The rate constant for the increase in NMDG+ permeability is ∼0.4 s−1, similar to that reported previously in mammalian cells (Virginio et al., 1999b). Thus, with time NMDG+ becomes more permeable through rP2X2 pores, but it does not become so permeable as Na+ (Erev 2.7 mV; n = 9, mean values are shown in Table I).

In the present study rP2X2 channels underwent permeability changes in all batches of oocytes, but the extent of the change varied from batch to batch (for example our results show values for PNMDG+/PNa+ between 0.17 and 0.34). These observations suggest the presence of uncontrolled variable(s) that influence the permeability change. To control for this variability, all comparisons between mutant and wt channels were made from the same batches. The remainder of this paper analyzes the nature of the changes that occur during the growth of the current from I1 to I2, and the accompanying shifts in reversal potentials (Fig. 1, A and B).

### Species Differences Reveal a Rationale to Study the Basis of Permeability Changes

We cloned and expressed mP2X2 channels in Xenopus oocytes. Rat and mouse P2X2 channels were similar with respect to peak INa+, but the ATP EC50s differed, being 2 ± 1 and 12 ± 4 μM with Hill slopes of 1.9 ± 0.2 and 2.0 ± 0.2, respectively, under conditions where we expect the pore to have dilated to the I2 state (n = 5; Fig. 2 A). Because of the difference in EC50, and to minimize any errors on the linear part of the curve, all subsequent experiments were performed at saturating concentrations of ATP (≥100 μM). rP2X2 and mP2X2 channels displayed similar current-voltage relations in Na+ solutions: the INa+ reversal potentials were close to 0 mV (−4.1 ± 0.1 and −6.5 ± 1.5 mV; P > 0.05) and the rectification indices (G+60 mV/G−60 mV) were 0.6 ± 0.2 and 0.5 ± 0.2 (P > 0.05), for rP2X2 and mP2X2 (Fig. 2 B). In short, there were no major differences between rP2X2 and mP2X2 channels in Na+ solutions.

When ATP was applied in extracellular solutions that contained NMDG+ in place of Na+ at −60 mV, a clear difference between rP2X2 and mP2X2 channels was apparent. Fig. 3 illustrates this distinction. In NMDG+ solutions application of ATP to mP2X2 evoked an initial outward current (I1), that decreases over 50 s to reach a steady-state value (I2). In contrast, at rP2X2 channels the current actually switches direction: I1 is outward and I2 is inward. Current-voltage relations based on voltage ramps (Fig. 3, B and D) show that the development of I2 for both rP2X2 and mP2X2 channels is associated with shifts in reversal potential over time, but the shift is much greater for rP2X2 than for mP2X2. Thus, for rP2X2 channels PNMDG+/PNa+ is 0.07 ± 0.002 for I1 (500 ± 77 nA), and at the steady-state I2 (−1241 ± 209 nA) the PNMDG+/PNa+ is 0.15 ± 0.007 (P < 0.01). In contrast for mP2X2 channels at the initial I1 peak (615 ± 92 nA) the PNMDG+/PNa+ is 0.07 ± 0.003, and at the steady-state I2 current (−12 ± 140 nA) the PNMDG+/PNa+ is 0.09 ± 0.006 (P < 0.05). In summary, with time NMDG+ becomes much more permeable through rP2X2 pores, but only slightly more permeable through mP2X2 pores.

### Sizing the P2X2 Pore States Using Excluded Field Theory

For the above experiments we studied the permeability to NMDG+ (radius ∼4.5 Å) because this was the standard protocol in previous work on rP2X2 channels. We next tested the hypothesis that smaller cations also change their permeability during ATP activation of the rP2X2 channel. To address this we used four organic cations (dimethylammonium, 2-(methyl-amino)-ethanol, Tris+ and N-methyl-D-glucamine) with radii between 2.7 and 4.5 Å and determined their permeability relative to Na+ (PX+/PNa+). While both mP2X2 and rP2X2 channels showed increased permeability in the I2 state to all four test ions, the increases for rP2X2 were much more dramatic. Fig. 4 shows the data plotted on graphs to determine the size of the narrowest part of the channel pore, using excluded field theory (EFT).* EFT was used previously to size the muscle nicotinic channel pore at a diameter of 6.5–8.4 Å (Dwyer et al., 1980; Cohen et al., 1992), which agrees well with the pore diameter revealed by imaging the open channel (Unwin, 1995). Using EFT we determined the diameter of the narrowest part of the rP2X2 pore in the I1 state as 11.4 Å, which is similar to the mP2X2 pore I1 states at 11.0 Å, and 11.6 Å diameter in the I2 states (Table II).

Thus, both I1 and I2 states of the mP2X2 channel, as well as the I1 state of the rP2X2 channel, have diameters equal to within 10%, but markedly smaller than the rP2X2 I2 state (Fig. 4); what is the size of the latter state? The data at hand do not give an accurate estimate because all tested organic cations acquired significant permeability. For example, the permeabilities (relative to Na+) of Tris+ and 2-(methyl-amino)-ethanol+ were 0.5 and 0.7, are roughly proportional to their mobilities in free solution (0.6 and 0.7 for Tris+ and 2-(methyl-amino)-ethanol+, relative to Na+) for the rP2X2 I2 state, whereas for the I1 state the permeability values decline more steeply than mobility with Rx, as expected for ions that are slowed by interactions with the channel wall as they traverse the I1 state pore (the free solution mobility values are presented in the legend for Fig. 4). On the basis of the plots shown in Fig. 4 we can state that the rP2X2 I2 state diameter is at least 3 Å larger than the ∼11 Å that characterizes the rP2X2 I1 and mP2X2 I1 and I2 states. This change increases the area across the narrowest region in the rP2X2 pore by at least ∼60 Å2, for a circular filter, or ∼76 Å2 for a square filter.

### A Role for the C Tail in Permeability Changes

What is the structural basis for the differences between rP2X2 and mP2X2 with respect to I2? As expected, the mouse cDNA has all the hallmarks of a bona fide P2X channel (North, 1996; Khakh, 2001), including sequence identity in TM2 with rP2X2. An alignment for the rP2X2 and mP2X2 protein sequence is shown in Fig. 5. Of 472 residues, 14 differ between rP2X2 and mP2X2, but none are in the pore segment (Rassendren et al., 1997; Egan et al., 1998), near the presumed ATP binding sites (Ennion et al., 2000; Jiang et al., 2000) or in TM1 a region known to influence the link between ATP binding to gating of the pore (Haines et al., 2001; Jiang et al., 2001). Seven residue differences between rP2X2 and mP2X2 map to the extracellular loop and the other seven are in the 119-residue COOH-terminal cytosolic domain (Fig. 5). Although intact C tail domains are not critical for P2X function, rP2X2 splice variants that have shorter C tail domains display faster desensitization (Simon et al., 1997; Koshimizu et al., 1998, 1999; Smith et al., 1999), and truncation of C tail domains impairs permeability in some P2X channels (Virginio et al., 1999b). We reasoned that a comparative study between rP2X2 and mP2X2 channels would represent a direct way to identify residues in the C tail domain that affect permeation.

### Focusing on Residues that Are Permissive for Permeability Changes in the C Tail Domain

Fig. 5 highlights the differences in the C tail domains between rP2X2 and mP2X2. In our initial experiments, we asked which of the 14 amino acid differences between rP2X2 and mP2X2 determine the presence of I2 and the associated increases in PNMDG+/PNa+ in rP2X2, but not in mP2X2? Moreover, can rP2X2 I2-like phenotypes be imparted on to mP2X2 channels by swapping an appropriate domain? To this end we made chimeric channels. For example, transplanting the entire rP2X2 C tail domain onto mP2X2 (mEL/rCT) produced functional responses that were indistinguishable from those of rP2X2 with respect to I1 PNMDG+/PNa+ and I2 PNMDG+/PNa+ (Fig. 6). Conversely, transplanting the mP2X2 C tail domain onto rP2X2 (rEL/mCT) gave rise to responses that were identical to those of mP2X2 channels. Seemingly, the differences between rP2X2 and mP2X2 reside entirely in the seven residue differences in the C tail domain, illustrated in Fig. 5. But which residues are involved?

We next asked whether any single amino acid is sufficient to make mP2X2 resemble rP2X2, and vice versa. Analysis of the seven appropriate single-site mutants in mP2X2 and the complementary seven mutants in rP2X2 indicate that this is not the case (Fig. 6): for simplicity these mutants have been pooled in Figs. 7 and 8. Thus, any one rP2X2 residue is not enough to endow the mP2X2 with I2, and any one mP2X2 residue is not enough to impair I2 in rP2X2 (Fig. 6). Between two and seven residues are needed to switch the I2 phenotype between mP2X2 and rP2X2.

We next made a series of chimeras in the C tail domain with progressively increasing numbers of rP2X2 C tail domain residues in the mP2X2 C tail domain. This approach tests for effects of substituting sequential residues, beginning at the end of the tail. Results are shown in Fig. 7. Remarkably, from the data it is clear that the last (most COOH-terminal) two amino acids, of the seven differences between mP2X2 and rP2X2, are sufficient to confer rP2X2-like behavior on the mP2X2 channel with respect to the I2 and PNMDG+/PNa+ changes. We call this chimera m430r, because until residue 430 the sequence is mP2X2, and after residue 430 the sequence is rP2X2. Increasing the number of rP2X2 residues from two to four produces progressively smaller increases in the I2 phenotype, but clearly the largest jump occurs for transplantation of the last two residues. In rP2X2 they are I432 and G444 and for mP2X2 these residues are S432 and D444 (Fig. 5).

We also made the reverse series of chimeras, with progressively increasing numbers of mP2X2 C tail domain residues in the rP2X2 C tail domain. Here again a clear picture is apparent: the last two mP2X2 residues are sufficient to eliminate I2 and PNMDG+/PNa+ changes. Increasing the number of mP2X2 residues in the rP2X2 tail from two to four produces no greater effect than simply substituting two (Fig. 7).

There are two outliers to the trend (shown by gray arrows). Transplantation of five rP2X2 residues into mP2X2 (m407r) produces a channel that desensitizes rapidly within 1–2 s, whereas the other mutants resemble wt rP2X2 and desensitize negligibly over a time period of up to 60s. Similarly, transplanting four mP2X2 residues into rP2X2 produces a channel that expresses poorly and desensitizes completely. In this chimera, reversal potentials could not be measured at a time point when I2 is expected to occur; therefore, these constructs are not informative with respect to changes in permeability. For these chimeras we expect that the channels close before spending appreciable time in the I2 state. This behavior is similar to our previous data with rP2X1 and rP2X3, which also desensitize rapidly, and previous work with P2X2 channels that indicate mutants in the C tail domain affect desensitization (Smith et al., 1999).

### Permeability Changes Are Unrelated to I1, but Correlate with the Growth of I2

The dataset presented in this study allows us to address a number of questions about the relationships between I1 and I2 states. For instance, is the extent of pore dilation in the I2 states determined by the permeability of the preceding I1 state? In other words, do channels with a lower PNMDG+/PNa+ in the I1 state result in a lower PNMDG+/PNa+ for the I2 state, and vice versa? Moreover, are PNMDG+/PNa+ changes related to I1 amplitude, and therefore ion flow? There was a clear correlation for all point mutants, chimeras, tail swaps, and wt channels between I2 amplitude and I2 PNMDG+/PNa+ changes, but not between I1 amplitude and I2 PNMDG+/PNa+ (Fig. 8), arguing that the extent of pore dilation in the I2 state is not related to ion flow in the I1 state. There was also no correlation between I1 PNMDG+/PNa+ and I2 PNMDG+/PNa+. We next asked, do permeability changes merely happen secondarily to the growth of I2 or do they mirror it? Table III shows the rates for increases in PNMDG+/PNa+, as well as for the growth from I1 to I2 for rP2X2, mEL/rCT, and m430r channels. The rate constants for rP2X2, mEL/rCT, and m430r channels are the same for PNMDG+/PNa+ changes and growth of I2 from I1 at −60 mV. Overall, these data show that (a) I2 amplitude and PNMDG+/PNa+ provide measures of the same underlying phenomenon because they are well correlated and occur with the same rates across a range of mutants, and (b) PNMDG+/PNa+ changes are not a consequence of ion flow during the outward I1. Moreover, the data clearly show that whereas the tail swaps and chimeras produce profound effects on I2, they have no effect on I1 amplitude or I1 PNMDG+/PNa+ across experiments from 26 distinct channels (2 wild type channels, 14 point mutations, 10 chimeras; Figs. 68). A lack of effect on I1 serves as an internal control against any effects on overall protein conformation. But more so the data sharply focus attention on amino acids that specifically affect permeation of a channel state: I2.

## Discussion

There is convincing evidence that the P2X2 channel COOH-terminal domain is cytosolic, including studies of functional concatenated subunits and access of antibodies raised to the cytosolic domain in permeabilized, but not non permeabilized cells (Torres et al., 1998; Stoop et al., 1999). The main finding of the present study is a significant extension and sharpening of the view that the COOH-terminal cytosolic domain affects permeation of the P2X channels in a dilated state. We exploited the observation that rat P2X2 (rP2X2) channels display a much more robust history-dependent permeability change than do mouse P2X2 (mP2X2) channels. The present study shows that replacement of just two amino acid residues abolishes the change in permeability in rP2X2, and just two residues allows mP2X2 channels to undergo permeability changes. The “negative controls” of the mP2X2 channel and of the various rP2X2/mP2X2 chimeras, which show identical physiology in the I1 state but completely different physiology in the I2 state, allow us to conclude even more confidently that the I2 state is not an artifact of voltage-clamp fidelity, Ca2+-activated conductances, or any other vaguely imaginable event. Instead, the I2 state results from a change in the pore, in a transition that is discussed below.

Our finding that the pore dilates by at least 3 Å is consistent with measurements using dye uptake studies (Khakh et al., 1999; Virginio et al., 1999a,b). Because cationic dye flux studies are interpreted in a binary fashion (the dye either permeates or not), they provide a cutoff for the pore size. For instance, YOPRO1 (16.8 × 12.8 × 8.2 Å) does not permeate P2X channels that lack the I2 state because two of its dimensions are larger than the I1 pore, which we estimated to be 11 Å in the present study for P2X2 channels. However, YOPRO1 permeability through P2X2 channels increases in a time-dependent manner during ATP application, reaching steady-state in 30–50 s. This time course is slower than opening to the I1 state, which occurs in <300 ms (Fig. 1 A), but resembles the time for opening to the dilated state (Fig. 1 A), as measured either by the growth of I1 to I2, or by increases in PNMDG+/PNa+. These data imply that YOPRO1 does not permeate the 11 Å I1 states because of steric hindrance, but can permeate the dilated I2 state. These considerations imply that the pore of P2X channels must dilate by at least 2 Å, consistent with the electrophysiological data presented in this study indicating that the pore dilates by at least 3 Å. A more accurate estimate of I2 pore size is required, but the present paucity of approximately spherical monovalent cations larger than NMDG+ vitiates further EFT experiments. Moreover, an attempt to use polymers of variable length has proved unfruitful (Virginio et al., 1999a). Perhaps novel approaches are required to accurately size the I2 state.

We compare our data to observations on P2X7 channels. Early studies of P2X7 channels carrying a truncated C tail suggested a role for the C tail domain in permeability changes, but did not pinpoint individual or stretches of important residues (Surprenant et al., 1996). A single nucleotide polymorphism of the P2X7 channel gene in a population of humans with deficits in lymphocyte and monocyte function results in functionally impaired, but appropriately membrane localized, P2X7 channels carrying a E496A mutation in the COOH-terminal tail domain (Gu et al., 2001). These observations support our present findings that distinct COOH-terminal domain residues affect P2X channel function, but our data are enlightening because they suggest a molecular explanation for the functional deficits in humans with the P2X7 channel E469A polymorphism (Gu et al., 2001). In the simplest case, polymorphic E469A channels display impaired dye uptake through the I2 states (Gu et al., 2001) because the C tail domains are locked in a nonpermissive state, as revealed by the appropriate mutants and chimeras reported here. The two nonpermissive residues identified in this study (S432 and D444 in mP2X2) and the E469A mutation in human polymorphic P2X7 channels all locate to a region of the C tail domain that is ∼100 residues from the pore-lining segment.

### Permeability Measurements and Conductances

The present study adds quantitative details on the extent of the change from a relatively selective I1 state to a less selective I2 state, which permeates organic cations with ease. Our most complete dataset, for NMDG+, allows us to compare three related parameters for NMDG+ permeation in the rP2X2 I2 state: relative free solution mobility, relative conductance, and relative permeability. For the experiment specifically designed to compare permeability and conductance, these parameters for NMDG+ are, respectively, 0.49 (Barry and Lynch, 1991; Ng and Barry, 1995), 0.68 (Table I), and 0.34 (Table I), relative to Na+. As noted in results, the absolute magnitude of the permeability change varied somewhat between batches of oocytes, and the measured PNMDG+/PNa+ range was 0.17–0.34 across all our experiments. The slope conductance for NMDG+ in the I2 state is therefore surprisingly high relative to that for Na+.

Conductance is governed by several mechanisms not directly related to permeability, which primarily reflects a single rate-limiting filter. The interestingly high ratio of macroscopic NMDG+ to Na+ conductance calls for single-channel measurements of the I2 state. The pioneering single-channel studies of P2X2 channels (Ding and Sachs, 1999a,b) have not systematically explored the optimal conditions for production of the I2 state (Khakh et al., 1999; Virginio et al., 1999b): very low extracellular Ca2+ and concentrations of ATP >10 μM (Virginio et al., 1999b). Moreover, rP2X2 channels in excised patches display inactivation kinetics that are quite different to those in whole-cell mode, implying the loss of necessary cytosolic components (Ding and Sachs, 2000).

In the absence of single-channel I2 state recordings, we review several previously reported single-channel characteristics of the I1 state in monovalent metal cation solutions (NMDG+ single-channel currents are below the resolution of this state). At least two mechanisms decrease the macroscopic conductance (Ding and Sachs, 1999b). There is an Na+ ion binding site ∼20% of the distance through the electric field (from the outside), with an affinity of ∼90 mM at −60 mV (Ding and Sachs, 1999a). Under our conditions, this site would be 50% saturated. Single-channel recordings also show pronounced flickering; the maximal open probability is 0.6. One of these mechanisms might operate more strongly for Na+ than for NMDG+ in the I2 state, accounting for the interestingly high ratio of NMDG+ to Na+ slope conductance.

### Toward a View of P2X Channel Gating

We observed profound effects for point mutants, chimeras, and tail swaps on I2 amplitude and I2 PNMDG+/PNa+ changes, but no effect on I1 amplitude or I1 PNMDG+/PNa+. A lack of effect on I1 across experiments from 26 distinct channels serves as a good internal control against effects on overall protein conformation, and strongly argues that the effect of the mutants and chimeras is specific to the permeation of the I2 state. With this control dataset in hand we can conclude that residues in the C tail affect the I2 state. But what are the structural details of this transition? P2X channels are the newest members of the transmitter-gated ion channels to be identified, they have little sequence homology with any other ion channels and relatively little is known about their structure-function relationships. Nonetheless, we present some plausible views.

Present evidence indicates that ATP binds to regions just extracellular to the transmembrane domains in P2X channels (Ennion et al., 2000; Jiang et al., 2000). In the present view, ATP binding causes motions in TM1 (Haines et al., 2001; Jiang et al., 2001), perhaps as the outer part of TM1 moves with respect to the outer part of TM2, that allow the pore to open at or near G342 in TM2 (Rassendren et al., 1997; Egan et al., 1998; Migita et al., 2001). Our data suggest (Fig. 9) that the transition from I1 to I2 increases the area across the narrowest region in the rP2X2 pore by at least ∼60 Å2, for a circular filter, or ∼76 Å2 for a square filter. This implies an increase in the pore diameter by at least 3 Å, but an accurate estimate of the I2 state pore size requires further work. The change is regulated by the disposition of side chains in a particular region of the cytoplasmic tail, but as previously demonstrated the change likely occurs at the pore (Khakh et al., 1999; Virginio et al., 1999b). Our data favor the model in which channels go from closed to open 1 (I1; pore diameter 11 Å) and then to open 2 (I2; pore diameter >14 Å) during a conformational change(s) in the C tail domain. This conformational change(s) proceeds at a rate of 0.5 s−1.

In the conventional view, the shape change in the permeation pathway during permeability changes is occurring on the axis of the multimeric channel, in a region where the cytoplasmic tails of several subunits approach the axis. Recent biochemical and electrophysiological evidence indicates that the NH2 and COOH termini are close to one another in functional P2X2 channels (Boue-Grabot et al., 2000), suggesting at the very least that the cytoplasmic regions are ordered and structured, and lending credence to the idea that these regions could contribute to the inner aspect of a conduction pathway. Consistent with this hypothesis are recent data on nicotinic channels (Miyazawa et al., 1999), Shaker (Sokolova et al., 2001), and Na+ channels (Sato et al., 2001), revealing the existence of cytoplasmic hanging baskets that split the conduction pathway. If such fenestrations also occur for P2X channels, then there might be multiple parallel and radially symmetric dilations of the conduction path.

The COOH-terminal tails of P2X2 channels contain several Pro-rich regions, which often bind other proteins (Fig. 5). Work published since submission of this paper has identified protein partners for the P2X7 channel (Kim et al., 2001), and it is possible that a similar situation exists for P2X2 channels. Thus, permeability changes might involve interactions with other cytoplasmic partners, which may differ between mP2X2 and rP2X2 at the crucial two residues we have identified in this study. In particular, it has not escaped our attention that phosphorylation events on a time scale of milliseconds to seconds might participate in the transition to the I2 state. Indeed, P2X2 channel kinetics appear to be regulated by phosphorylation (Boue-Grabot et al., 2000). A further possibility is that the C tail domain itself may not line the pore directly, but may be involved in the conformational change that allows the pore to dilate. In this scenario, the residues that lead to absence of I2 would effectively ‘lock’ the C tail in a nonpermissive state (Fig. 9). In such a model the transition from the I1 to the I2 state might involve extensive changes in the C tail domain. A large shape change was recently suggested for the gating of the MscL channel (Sukharev et al., 2001): the location of the C tail domain would change during opening as it moves radially from the inner aspect of the pore. Because MscL and P2X channels have similar membrane topologies and both permeate large ions (Khakh, 2001), it is possible to consider that a similar gating mechanism may operate. The above hypotheses reveal several avenues of future experiments to better understand the contribution of the C tail domain to permeability changes at P2X2 channels, but solving the channel structure at atomic resolution in the closed and open states, and deciphering the nature of the transitions, may provide the only completely satisfying view.

## Acknowledgments

We thank K. Kostenko and H. Li for help with preparation of Xenopus oocytes, G. Shapovalov for measurements of 2-(methyl-amino)-ethanol mobility, other members of the group for comments, and Nigel Unwin for discussion and comments on the paper.

Work in our labs was supported by a Wellcome Trust (UK) Prize Fellowship, Roche Bioscience, National Institutes of Health (NS-11756), and the Medical Research Council.

*

Abbreviation used in this paper: EFT, excluded field theory.

## References

References
Barry, P.H., and J.W. Lynch.
1991
. Liquid junction potentials and small cell effects in ptach-clamp analysis.
J. Membrane Biol.
121
:
101
–117.
Boue-Grabot, E., V. Archambault, and P. Seguela.
2000
. A protein kinase C site highly conserved in P2X subunits controls the desensitization kinetics of P2X(2) ATP-gated channels.
J. Biol. Chem.
275
:
10190
–10195.
Cohen, B.N., C. Labarca, N. Davidson, and H.A. Lester.
1992
. Mutations in M2 alter the selectivity of the mouse nicotinic acetylcholine receptor for organic and alkali metal cations.
J. Gen. Physiol.
100
:
373
–400.
Ding, S., and F. Sachs.
1999
a. Single channel properties of P2X2 purinoceptors.
J. Gen. Physiol.
113
:
695
–720.
Ding, S., and F. Sachs.
1999
b. Ion permeation and block of P2X2 purinoceptors: single channel recordings.
J. Membr. Biol.
172
:
215
–223.
Ding, S., and F. Sachs.
2000
. Inactivation of P2X2 purinoceptors by divalent cations.
J. Physiol.
522
:
199
–214.
Dwyer, T.M., D.J. Adams, and B. Hille.
1980
. The permeability of the endplate channel to organic cations in frog muscle.
J. Gen. Physiol.
75
:
469
–492.
Egan, T.M., W.R. Haines, and M.M. Voigt.
1998
. A domain contributing to the ion channel of ATP-gated P2X2 receptors identified by the substituted cysteine accessibility method.
J. Neurosci.
18
:
2350
–2359.
Ennion, S., S. Hagan, and R.J. Evans.
2000
. The role of positively charged amino acids in ATP recognition by human P2X1 receptors.
J. Biol. Chem.
275
:
29361
–29367.
Evans, R.J., C. Lewis, C. Virginio, K. Lundstrom, G. Buell, A. Surprenant, and R.A. North.
1996
. Ionic permeability of, and divalent cation effects on, two ATP-gated cation channels (P2X receptors) expressed in mammalian cells.
J. Physiol.
497
:
413
–422.
Gu, B.J., W. Zhang, R.A. Worthington, R. Sluyter, P. Dao-Ung, S. Petrou, J.A. Barden, and J. Wiley.
2001
. A Glu-496 to Ala polymorphism leads to loss of function of the human P2X7 receptor.
J. Biol. Chem.
276
:
11135
–11142.
Haines, W.R., K. Migita, J.A. Cox, T.M. Egan, and M.M. Voigt.
2001
. The first transmembrane domain of the P2X receptor subunit participates in the agonist-induced gating of the channel.
J. Biol. Chem.
276
:
32793
–32798.
Hille, B. 1992. Ionic channels of excitable membranes. 2nd Edition.
Jiang, L.H., F. Rassendren, V. Spelta, A. Surprenant, and R.A. North.
2001
. Amino acid residues involved in gating identified in the first membrane-spanning domain of the rat P2X(2) receptor.
J. Biol. Chem.
276
:
14902
–14908.
Jiang, L.-H., F. Rassendren, A. Surprenant, and R.A. North.
2000
. Identification of amino acid residues contributing to the ATP binding site of a P2X receptor.
J. Biol. Chem.
275
:
34190
–34196.
Khakh, B.S., P.P. Humphrey, and A. Surprenant.
1995
. Electrophysiological properties of P2X-purinoceptors in rat superior cervical, nodose and guinea-pig coeliac neurones.
J. Physiol.
484
:
385
–395.
Khakh, B.S., X. Bao, C. Labarca, and H.A. Lester.
1999
. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.
Nat. Neurosci.
2
:
322
–330.
Khakh, B.S., and H.A. Lester.
1999
. Dynamic selectivity filters in ion channels.
Neuron.
23
:
653
–658.
Khakh, B.S.
2001
. Molecular physiology of P2X receptors and ATP signalling at synapses.
Nat. Rev Neurosci.
2
:
165
–174.
Kielland, J.
1937
. Individual activity coefficients of ions in aqueous solutions.
J. Am. Chem. Soc.
59
:
1675
–1678.
Kim, M., L.H. Jiang, H.L. Wilson, R.A. North, and A. Surprenant.
2001
. Proteomic and functional evidence for a P2X7 receptor signalling complex.
EMBO J.
20
:
6347
–6358.
Koshimizu, T., M. Koshimizu, and S.S. Stojilkovic.
1999
. Contributions of the C-terminal domain to the control of P2X receptor desensitization.
J. Biol. Chem.
274
:
37651
–37657.
Koshimizu, T.-A., M. Tomi, M. Koshimizu, and S.S. Stojilkovic.
1998
. Identification of amino acid residues contributing to desensitization of the P2X2 receptor channel.
J. Biol. Chem.
273
:
12853
–12857.
Migita, K., W.R. Haines, M.M. Voigt, and T.M. Egan.
2001
. Polar residues of the second transmembrane domain influence cation permeability of the ATP-gated P2X2 receptor.
J. Biol. Chem.
276
:
30934
–30941.
Miyazawa, A., Y. Fujiyoshi, M. Stowell, and N. Unwin.
1999
. Nicotinic acetylcholine receptor at 4.6 A resolution: transverse tunnels in the channel wall.
J. Mol. Biol.
288
:
765
–786.
Ng, B., and P.H. Barry.
1995
. The measurment of ionic conductivities and mobilities of certain less common organic ions needed for junction potential corrections in electrophysiology.
J. Neurosci. Methods.
56
:
37
–41.
North, R.A.
1996
. Families of ion channels with two hydrophobic segments.
Curr. Opin. Cell Biol.
8
:
474
–483.
Rassendren, F., G. Buell, A. Newbolt, R.A. North, and A. Surprenant.
1997
. Identification of amino acid residues contributing to the pore of a P2X receptor.
EMBO J.
16
:
3446
–3454.
Sato, C., Y. Ueno, K. Asai, K. Takahashi, M. Sato, A. Engel, and Y. Fujiyoshi.
2001
. The voltage-sensitive sodium channel is a bell-shaped molecule with several cavities.
Nature.
409
:
1047
–1051.
Simon, J., E.J. Kidd, F.M. Smit, I.P. Chessell, R. Murrell-Lagnado, P.P. Humphrey, and E.A. Barnard.
1997
. Localization and functional expression of splice variants of the P2X2 receptor.
Mol. Pharmacol.
52
:
237
–248.
Smith, F.M., P.P.A. Humphrey, and R.D. Murrell-Lagnado.
1999
. Identification of amino acids within the P2X2 receptor C-terminus that regulate desensitization.
J. Physiol.
520
:
91
–99.
Sokolova, O., L. Kolmakova-Partensky, and N. Grigorieff.
2001
. Three-dimensional structure of a voltage-gated potassium channel at 2.5 nm resolution.
Structure.
9
:
215
–220.
Stoop, R., S. Thomas, F. Rassendren, E. Kawashima, G. Buell, A. Surprenant, and R. North.
1999
. Contribution of individual subunits to the multimeric P2X2 receptor: estimates based on methanethiosulfonate block at T336C.
Mol. Pharmacol.
56
:
973
–981.
Sukharev, S., M. Betanzos, C.S. Chiang, and H.R. Guy.
2001
. The gating mechanism of the large mechanosensitive channel MscL.
Nature.
409
:
720
–724.
Surprenant, A., F. Rassendren, E. Kawashima, R.A. North, and G. Buell.
1996
. The cytolytic P2Z receptor for extracellular ATP identified as a P2X receptor (P2X7).
Science.
272
:
735
–738.
Torres, G.E., T.M. Egan, and M.M. Voigt.
1998
. Topological analysis of the ATP-gated ionotropic P2X2 receptor subunit.
FEBS Lett.
425
:
19
–23.
Unwin, N.
1995
. Acetylcholine receptor channel imaged in the open state.
Nature.
373
:
37
–43.
Virginio, C., A. MacKenzie, R.A. North, and A. Surprenant.
1999
a. Kinetics of cell lysis, dye uptake and permeability changes in cells expressing the rat P2X7 receptor.
J. Physiol.
519
:
335
–346.
Virginio, C., A. MacKenzie, F.A. Rassendren, R.A. North, and A. Surprenant.
1999
b. Pore dilation of neuronal P2X receptor channels.
Nat. Neurosci.
2
:
315
–321.