Contribution of Calcium Ions to P2X Channel Responses

Abstract

Ca²⁺ entry through transmitter-gated cation channels, including ATP-gated P2X channels, contributes to an array of physiological processes in excitable and non-excitable cells, but the absolute amount of Ca²⁺ flowing through P2X channels is unknown. Here we address the issue of precisely how much Ca²⁺ flows through P2X channels and report the finding that the ATP-gated P2X channel family has remarkably high Ca²⁺ flux compared with other channels gated by the transmitters ACh, serotonin, protons, and glutamate. Several homomeric and heteromeric P2X channels display fractional Ca²⁺ currents equivalent to NMDA channels, which hitherto have been thought of as the largest source of transmitter-activated Ca²⁺ flux. We further suggest that NMDA and P2X channels may use different mechanisms to promote Ca²⁺ flux across membranes. We find that mutating three critical polar amino acids decreases the Ca²⁺ flux of P2X₂ receptors, suggesting that these residues cluster to form a novel type of Ca²⁺ selectivity region within the pore. Overall, our data identify P2X channels as a large source of transmitter-activated Ca²⁺ influx at resting membrane potentials and support the hypothesis that polar amino acids contribute to Ca²⁺ selection in an ATP-gated ion channel.

Introduction

Transmitter-gated cation channels (TGCCs) are key transmembrane proteins of excitable and non-excitable cells. Mammalian TGCCs can be divided into three main families on the basis of gene and protein sequences and known and predicted channel structures (Green et al., 1998). Cys-loop receptors for ACh, serotonin, GABA, and glycine form one family, and glutamate-gated channels form the second (Green et al., 1998). The third major family of mammalian TGCCs are the ATP-gated P2X channels that have a relatively simple structure compared with the other two families. To date, the quaternary structure of the channel is thought to be composed of specific combinations of three of the seven (P2X₁–P2X₇) known subunits (Nicke et al., 1998; Stoop et al., 1999; Jiang et al., 2003). Each subunit possesses two transmembrane segments (North, 2002), the second of which is thought to line the ion channel pore (Rassendren et al., 1997; Egan et al., 1998). P2X channels are widely expressed throughout the nervous system and underlie fast ATP neurotransmission at some neuro-neuronal synapses (Norenberg and Illes, 2000; North, 2002). For instance, P2X₂ channels mediate an EPSP between myenteric neurons (Galligan and Bertrand, 1994; Khakh et al., 2000; Ren et al., 2003), and P2X₁ channels underlie a fast excitatory junction potential at neuro-effector junctions (Brain et al., 2002; Lamont and Wier, 2002; Lamont et al., 2003). In addition, P2X channels are found presynaptically where they modulate neurotransmitter release (Gu and MacDermott, 1997; Khakh and Henderson, 1998; MacDermott et al., 1999; Hugel and Schlichter, 2000; Kato and Shigetomi, 2001; Nakatsuka and Gu, 2001; Khakh et al., 2003). In many instances, the physiological response is triggered by the influx of Ca²⁺ through P2X channels; prime examples include presynaptic responses at neuro-neuronal synapses and postsynaptic responses at neuroeffector junctions.

Most TGCCs, including P2X channels, have Ca²⁺ permeabilities equal to or greater than those of the 100-fold more abundant Na⁺ (Burnashev, 1998). When these channels open, Ca²⁺ moves down its electrochemical gradient and into the cell. The resulting influx of Ca²⁺ exerts wide-ranging physiological effects that can last for seconds, days, or weeks (Berridge et al., 2003), adding a spatial and temporal dimension to transmitter signaling that may outlast the initial millisecond time scale surge of neurotransmitter and the accompanying depolarization by factors of >10⁹. Previous studies revealed that ATP gates a transmembrane Ca²⁺ flux pathway (Benham and Tsien, 1987; Rogers and Dani, 1995) that subsequently was shown to contribute to the panoply of physiological responses in cells throughout the body, including neurons, muscle, glia, immune cells, and epithelia (Khakh, 2001; Inoue, 2002; North, 2002; Schwiebert and Zsembery, 2003). However, a comprehensive examination of Ca²⁺ flux through the ATP-gated P2X channel family is not reported, and there is little quantitative information about how much Ca²⁺ flows through P2X channels in relation to other TGCCs.

In the present study, we directly measured Ca²⁺ flow through 11 functional recombinant P2X channels and compared these values with other TGCCs. We used patch-clamp photometry to measure fractional calcium currents (Pf%) carried by TGCCs (Schneggenburger et al., 1993). Our data suggest that, on average, P2X channels are the most Ca²⁺-permeable TGCCs and suggest that polar amino acids may provide the counter charges needed to partially dehydrate Ca²⁺ ions in a narrow part of the pore.

Materials and Methods

Molecular biology. The stable cell lines used were HEK293 cells expressing the rat 5-HT_3A channel (Sarah Lummis, Medical Research Council, Laboratory of Molecular Biology, Cambridge, UK) or the human α4β2 nicotinic channel (Alison Rush, Research Labs, San Diego, CA). Wild-type and mutant rat P2X₂ cDNAs were available from previous work; these mutants were generated using routine methods. Drs. Richard Evans (University of Leicester, Leicester, UK), R. Alan North (Institute of Molecular Physiology, Sheffield, UK), David Julius (University of California San Francisco, San Francisco, CA), Thomas Küner (MPI, Hiedelberg, Germany), and Henry A. Lester (California Institute of Technology, Pasadena, CA) supplied cDNAs encoding human P2X₁, human P2X₄, rat vanilloid receptor 1 (VR1), rat NR1/NR2A, and chick α4/β2 nicotinic channels, respectively. Plasmid cDNAs were transfected into HEK293 cells plated on 35 mm plastic culture dishes using Effectene (Qiagen, West Sussex, UK) and following the protocol of the manufacturer.

Patch-clamp photometry. The experimental setup is depicted in Figure 1 A. Cells expressing the protein of interest were replated at low density onto 13 mm borosilicate glass coverslips (BDH Chemicals, Poole, UK) 12–18 hr before the start of the experiment. Coverslips were transferred to a recording chamber positioned on the stage of a Nikon (Tokyo, Japan) Diaphot 200 inverted microscope equipped with a Fluor 40× objective, 100 W xenon lamp, Uniblitz (Rochester, NY) shutter, liquid light guide, and custom-made coupling. In most cases, whole-cell current was recorded at a holding potential of –55 mV using an Axopatch 1-D amplifier (Axon Instruments, Foster City, CA) and a low resistance (1.5–5 MΩ) glass microelectrode filled with an intracellular solution of the following composition (in mm): 140 CsCl, 10 tetraethylammonium Cl, 10 HEPES, 2 fura-2 K⁵ (lot 3491-11; Molecular Probes, Eugene, OR), and 4.8 CsOH, pH 7.3. Previous findings demonstrate that fura-2 reaches a steady-state intracellular concentration in HEK293 cells after 5–6 min of passive diffusion from the recording electrode (Schneggenburger et al., 1993; Burnashev et al., 1995; Schneggenburger, 1998; Frings et al., 2000). Consistent with this, we saw no increase in the steady level of resting fura-2 fluorescence 10 min after going whole cell (data not shown), and we always waited 10 min after rupturing the patch before continuing with the experiment. Agonists were applied for 0.2–4.0 sec once every 2–3 min using triple-barreled theta glass and a rapid solution changer system (Perfusion Fast-Step System SF-77; Warner Instruments, Hamden, CT). The extracellular bath solution contained the followinig (in mm): 140 NaCl, 1 MgCl₂, 2 CaCl₂, 10 HEPES, 10 glucose, and 5 NaOH to adjust the pH to 7.4.

Figure 1.

Determination of fractional Ca²⁺ currents. A, Representation of the experimental setup used for measuring Pf%. HEK293 cells expressing molecularly defined channels were plated onto glass coverslips and patch clamped with electrodes filled with intracellular solution containing 2 mm fura-2. ATP was applied rapidly using an automated fast solution switcher. Photons were captured through a 40× objective lens. The emitted light was filtered and directed toward a photo multiplier attached to the microscope side port, and photon counts were measured in volts. The bottom photomicrograph (B) shows images of Fluoresbrite beads that were used to calibrate the voltage signal produced by the photomultiplier tube (PMT): for illustration purposes, the images shown were captured on a confocal microscope with the iris fully open. The beads have a mean diameter of 4.6 μm. C, ATP-evoked currents of increasing duration in pure Ca²⁺ extracellular solutions at P2X₂ channels. The holding potential was –60 mV. D shows the integral of these currents, and E shows the corresponding changes in fura-2 fluorescence at 380 nm: the time course of the change in F₃₈₀ matches the time course of Q_T for all of the traces. F, A graph of ΔF₃₈₀ and Q_T for each of the traces shown in C–E. They all superimpose and fall on a straight line. The slope of this line represents the proportionality constant between Q_T and ΔF₃₈₀ bead units per picocoulomb.

We determined fractional calcium current by simultaneously measuring the total membrane current and the change in emission of fura-2 excited at 380 nm (F₃₈₀). Previous work shows that Pf% values can be determined accurately with a single excitation wavelength (Schneggenburger et al., 1993). An advantage of this technique is that it does not require that the absolute change in [Ca²⁺]_i be quantified directly and thus avoids the time-consuming step needed to switch between excitation wavelengths. Rather, Ca²⁺ flux is measured by monitoring the change in fluorescence of fura-2 at a single wavelength, thus allowing for high time resolution measurements with a single photomultiplier tube. Whole-cell fluorescence at 510 nm was collected using a model 714 Photomultiplier Detection System (Photon Technology International, South Brunswick, NJ). The excitation filter was 380AF10 (XF1094), the dichroic was 415DCLP (XF2002), and the emitter was 510WB40 (XF3043; all from Omega Optical, Brattleboro, VT). Background fluorescence was minimized by limiting the field of view of the photomultiplier tube to the immediate vicinity of the cell under investigation using an adjustable pinhole. We controlled for day-to-day variations in the efficiency of our system by normalizing the biological signal to that of the average fluorescence of five carboxy Bright Blue 4.6 μm microspheres (Polysciences, Warrington, PA) that had settled on the bottom of the bath chamber filled with the normal extracellular solution (Fig. 1 B). Thus, in keeping with previous work (Schneggenburger et al., 1993), we present changes in fura-2 fluorescence in bead units (BU) rather than volts. All data were sampled at 5 kHz using Digitdata 1200 hardware and pClamp 8.0 software by Axon Instruments; fluorescence was low-pass filtered offline at 300 Hz.

Data analysis. The Pf% was calculated as follows:

where Q_T (the integral of the ligand-gated ionic current) and Q_Ca are given by the following:

F_max is the calibration constant used to relate ΔQ_Ca to ΔF₃₈₀. It was calculated in a separate series of experiments under conditions in which Q_T is expected to equal Q_Ca (Fig. 1C–E). This was achieved by the following: (1) measuring ATP-gated membrane current and fluorescence from cells expressing either P2X₂ or P2X₄ channels; (2) holding these cells at a membrane potential (–60 mV) at which the outward flow of ions is negligible; and (3) replacing extracellular Na⁺ with 112 mm Ca²⁺. Under these conditions, ATP elicits membrane currents carried exclusively by Ca²⁺, and F_max can be determined from the slope of the plot of Q_T versus F₃₈₀ (Fig. 1 F). F_max, measured in this way, was 0.012 ± 0.002 BU/nC (n = 37). Data were analyzed in Clampfit 8.1 (Axon Instruments), and calculations were performed using macros written in Igor Pro (WaveMetrics, Lake Oswego, OR). Results are reported as mean ± SEM for the number of cells (n) included in the study. Pf% was often measured many times in a single cell; these measures were then averaged to give the Pf% for the individual cell. Significant differences among groups were determined by one-way ANOVA with Tukey's post hoc or the Student's t test. A p value of <0.01 was considered significant.

Chemicals. All chemicals used were from Tocris Cookson (Bristol, UK), Molecular Probes, or Sigma.

Results

Pf% for channels gated by glutamate, serotonin, acetylcholine, and protons

Patch-clamp photometry is the only method providing an absolute measure of Ca²⁺ flux for ion channels that is independent of cell type, endogenous buffer capacity, channel density, and current amplitude (Schneggenburger et al., 1993). It is also superior to reversal potential-based methods because the measurements are made in physiological solutions at resting membrane potentials and because the values for flux do not make Goldman-Hodgkin-Katz assumptions (Frings et al., 2000). Using patch-clamp photometry, we began by measuring the fractional Ca²⁺ current of nicotinic α4β2, serotonin 5-HT_3A, proton-gated VR1, and glutamate NR1/NR2A channels (Fig. 2) because these are well characterized channels that are known to transport Ca²⁺ across membranes in neurons (Burnashev, 1998; MacDermott et al., 1999; Montell, 2001; Reeves and Lummis, 2002). Our Pf% values for the human α4β2, chick α4β2, and rat NR1/NR2A were 3.1 ± 0.6 (n = 5), 3.1 ± 0.8 (n = 4), and 14.1 ± 0.9% (n = 15) and are in excellent agreement with published reports (Ragozzino et al., 1998; Jatzke et al., 2002; Lax et al., 2002; Watanabe et al., 2002). The Pf% values of the rat 5-HT_3A and rat VR1 channels are not reported previously, and we find them to be 4.7 ± 0.3% (n = 8) and 3.5 ± 0.3% (n = 30), respectively. Together, these data validate our methods and provide guideposts by which to compare the Ca²⁺ flux of different TGCC families.

Figure 2.

Representative traces for Q_T and ΔF₃₈₀ at transmitter-gated channels. For all panels in this figure, the black traces are transmitter-evoked currents (in amperes) and the integral of the current Q_T (in nanocoulombs), whereas the gray traces are the ΔF₃₈₀ (in bead units). Appropriate agonists were applied for the times indicated by the solid bars above the traces. These agonists were as follows: 100 μm glutamate in 0 mm extracellular Mg²⁺ with 100 μm glycine for NR1/NR2A, 100 μm (–)-nicotine for α4β2, 10 μm serotonin for 5-HT_3A, and pH 5.5 for VR1.

Pf% for homomeric P2X channels gated by ATP

Previous detailed experiments have shown that intact HEK293 cells loaded with cell-permeable Ca²⁺ indicator dyes express endogenous P2Y receptors that mobilize intracellular Ca²⁺ in a G-protein-dependent manner (Fischer et al., 2003; He et al., 2003). We measured the effect of 30 μm ATP on untransfected and mock-transfected HEK293 cells to determine whether P2X channel-independent changes in [Ca²⁺]_i occur under the conditions used in the experiments described in this study. In so doing, we used exactly the same experimental conditions used for the measurement of Pf% values, which is whole-cell dialysis of the cell with intracellular solution containing fura for at least 10 min. Figure 3A shows that an application of 30 μm ATP had no measurable affect on either the holding current (bottom trace) or F₃₈₀ (top trace) in a mock-transfected cell. In a sample of eight cells, whole-cell fluorescence measured at 4 sec after the start of an application of 30 μm ATP was 99.9 ± 0.2% of that measured immediately before the start; this translates to a negligible decrease of 0.007 ± 0.015 BU in fura fluorescence, which implies negligible release of Ca²⁺ from intracellular stores and no presence of endogenous P2X channels (Fig. 3A, bottom trace). We suggest that differences between our experiments showing negligible contribution of P2Y receptors and previous studies (Fischer et al., 2003; He et al., 2003) are attributable to the unavoidable experimental requirements of our approach, namely complete intracellular dialysis of the cell constituents (Schneggenburger et al., 1993). We expect that, under these conditions, metabotropic ATP receptor effects are impaired.

Figure 3.

Controls and calibrations for measurements of Pf% at P2X channels. A, Lack of effect of 30 μm ATP on membrane current (bottom trace) and F₃₈₀ (top trace) in a mock-transfected HEK293 cell. B–D, Representative traces for an HEK293 cell expressing P2X₂ channels and activated with 10 μm ATP for increasing durations, as indicated by the length of the bars in B. B shows the current traces (for clarity, we show only 5 of the 9 traces), and C shows the corresponding integrated currents (Q_T, left traces) and changes in fluorescence (ΔF₃₈₀, right traces). D shows a plot of ΔF₃₈₀ versus Q_T. The data fall on a straight line, as expected if P2X channels are the only source of Ca²⁺. The traces and data are colored coded in B–D so that any individual response can be compared across all graphs. In D, there are four additional data points that, for the sake of simplicity, are not illustrated in B and C.

We saw robust inward currents and decreases in F₃₈₀ in transfected cells expressing P2X channels. Current and fluorescence were measured while applying 10 μm ATP for varying lengths of time (0.2–3.5 sec, as indicated by the lengths of the solid bars in Fig. 3B) to a HEK293 cell expressing homomeric P2X₂ channels (Fig. 3B). Of critical importance are data showing that the time course of the decrease in F₃₈₀ mirrored that of the integral of the ATP-gated current (Fig. 3C) as expected if Ca²⁺ enters through P2X₂ channels (Schneggenburger et al., 1993). Furthermore, the plot of ΔF₃₈₀ versus Q_T was linear (Fig. 3D). Both results suggest that the P2X channels are the sole source of the rise in [Ca²⁺]_i. This conclusion is supported by data showing that calcium-induced calcium release is negligible in HEK293 cells studied with fura-2 (Alonzo et al., 2003). Our finding that ATP activation of endogenous P2Y receptors does not affect [Ca²⁺]_i. under our experimental conditions probably reflects disruption of a signaling cascade mechanism caused by the obligatory 10 min dialysis of the cell interior with the contents of the recording electrode. All in all, the data provide strong evidence against a detectable contribution of endogenous metabotropic ATP receptors to the responses described in this study.

We next studied Ca²⁺ flux at homomeric rat P2X channels. ATP evoked inward currents and measurable decreases in fluorescence of fura-2 at 380 nm (ΔF₃₈₀) in cells expressing functional homomeric P2X channels (Fig. 4). In all cases, the time course of the change in fluorescence paralleled the accumulation of charge (Q_T), as expected if the underlying cause of the rise in Ca²⁺ is the ATP-gated current. Pf% values determined for all six functional rat homomeric channels were as follows: P2X₁, 12.4 ± 1.6% (n = 5); P2X₂, 5.7 ± 0.3% (n = 18); P2X₃, 2.7 ± 0.9% (n = 5); P2X₄, 11.0 ± 0.7% (n = 14); P2X₅, 4.5 ± 0.5% (n = 5); and P2X₇, 4.6 ± 0.5% (n = 12). We found that Pf% was not a function of agonist concentration for P2X₂ channels (5.8 ± 0.3, 5.5 ± 0.5, and 6.3 ± 0.9% for 10, 30, and 100 μm ATP, respectively, n = 13, 4, and 3, respectively; p > 0.05), although we did not study this relationship in detail for all channels. With the exception of the P2X₃ channel, all ATP-gated channels show fractional Ca²⁺ currents that are equal to or greater than α4β2, 5-HT_3A, and VR1 channels (Fig. 5A). Furthermore, both P2X₁ and P2X₄ channels display fractional Ca²⁺ currents that are equivalent (p > 0.05) to the NR1/NR2A NMDA channel that was thought previously to be an unparalleled source of transmitter-activated Ca²⁺ flux. The human homologues, hP2X₁ and hP2X₄, also gate large fractional Ca²⁺ currents of 10.8 ± 1.1% (n = 7) and 15.0 ± 1.5% (n = 7), respectively, demonstrating that high Ca²⁺ flux is conserved across species.

Figure 4.

Representative traces for Q_T and ΔF₃₈₀ at homomeric P2X channels. For all panels in this figure, the black traces are transmitter-evoked currents (in amperes) and the integral of the current Q_T (in nanocoulombs), whereas the gray traces are the ΔF₃₈₀ (in bead units). Appropriate agonists were applied for the times indicated by the solid bars above the traces. These agonists were as follows: 3 μm ATP for P2X₁ and P2X₃; 30 μm ATP for P2X₂, P2X₄, and P2X₅; and 100 μm benzoylbenzoyl ATP for P2X₇.

Figure 5.

Fractional Ca²⁺ currents for transmitter-gated channels. A, Mean data for Pf% of TGCCs compared with homomeric rat and human P2X channels. Note there is no data point for the P2X₆ channel because no ATP-evoked responses could be measured. B, Representative raw data of three heteromeric rat P2X channels. These traces show ATP-evoked currents (black), Q_T (black), and ΔF₃₈₀ (gray) for P2X_1/5, P2X_2/3, and P2X_2/6 channels. An appropriate concentration of ATP (either 3 or 30 μm) for each channel was used to evoke current. C, Mean data for the measured Pf% values for ATP-evoked currents recorded from cells expressing combinations of P2X subunits. The stars indicate significant differences between the heteromeric assemblies and their homomeric counterparts.

Pf% for heteromeric P2X channels gated by ATP

We next cotransfected HEK293 cells with pairs of cDNAs to record the Pf% values of heteromeric rat P2X channels (Fig. 5B). Four pairs of subunits (P2X_1/5, P2X_2/3, P2X_2/6, and P2X_4/6) are known to form functional complexes (North, 2002). The current through two of these pairs, P2X_1/5 and P2X_2/3, can be reliably measured without significant contamination by the unpaired, homomeric channels that may also be expressed. Cotransfection of P2X₁ and P2X₅ produces a population of channels composed almost exclusively of heteromeric P2X_1/5 channels (Torres et al., 1998). The Pf% of heteromeric P2X_1/5 channels was 3.3 ± 0.2% (n = 9), a value similar to homomeric P2X₅ channels (Fig. 5C). Cells cotransfected with cDNAs for P2X₂ and P2X₃ were studied using αβ-methylene ATP to isolate heteromeric responses (North, 2002). The Pf% of heteromeric P2X_2/3 channels was 3.5 ± 0.5% (n = 9). In both cases, the heteromeric channels show the Ca²⁺ phenotype of the less permeable channel. Functional isolation of heteromeric P2X_2/6 and P2X_4/6 responses from possible homomeric channels is less straightforward because there are no agonists that separate homomeric P2X₂ channels from the heteromeric pairs (North, 2002). However, we found that co-transfection of cDNAs encoding P2X₂ and P2X₆ subunits produced a population of ATP-gated channels that had a Pf% that was significantly greater (7.7 ± 0.7%; n = 14) than that of the P2X₂ channel alone (p = 0.0074) (Fig. 5C), and these data suggest that a considerable percentage of this channel population is heteromeric (King et al., 2000). Seemingly, the P2X₆ subunit imparts a sizable increase in Ca²⁺ flux through the pore. In contrast, cotransfection of rat P2X₄ and P2X₆ cDNAs resulted in an ATP-evoked current with a Pf% (11.3 ± 1.0%; n = 6) not significantly different from transfection of rat P2X₄ cDNA alone, although it is possible that our measurement of Pf% at P2X_4/6 channels may be dominated by homomeric P2X₄ channels that are likely also expressed.

On the role of pore lining polar residues in determining Ca²⁺ flux at P2X₂ channels

The high Ca²⁺ flux at nicotinic and NMDA channels occurs because of fixed charge in the pore region that may concentrate or select Ca²⁺ ions for permeation (Premkumar and Auerbach, 1996; Sharma and Stevens, 1996; Burnashev, 1998; Unwin, 2000; Watanabe et al., 2002). How do the structurally distinct P2X channels select for Ca²⁺ ions over the equally sized but ∼100-fold more abundant Na⁺? We focused on the involvement of the second transmembrane domain (TM2) (Fig. 6A) because it lines the channel pore (North, 2002) and ion permeation is altered in TM2 mutants (Migita et al., 2001). We focused on homomeric P2X₂ channels because most of our understanding of P2X channel structure and function is derived from studies of these channels, and this wealth of data provides an appropriate framework for hypothesis-driven mutagenesis (North, 2002). P2X channels contain conserved aspartates that may occupy sites near the pore vestibules. However, we believe that these aspartates do not make an obvious contribution to Ca²⁺ flux because neutralizing the negative charge at D315 and D349 with asparagines had no effect on Pf% values of P2X₂ channels (Fig. 6B). Next, we systematically mutated the polar amino acids that may provide a favorable environment for ion flow through the channel. Changing the character of polar residues that are primarily external (T330 and N333) or immediately internal (S345) to the putative channel gate near G342-V343-G344 (North, 2002) produced little effect on Ca²⁺ flux. However, increasing the hydrophobicity of three critical amino acids just extracellular to the gate (T336, T339, and S340) led to significant decreases in Pf%, suggesting that permeating ions normally interact with these polar residues in this domain. This would likely only occur in a narrow region of the pore. In keeping with this hypothesis, we found that increasing the size of these same amino acids led to either near complete absence of Ca²⁺ flux (T339Y and S340Y) or less informatively loss of channel function (T336Y). These data are in accord with the findings of Migita et al. (2001) (Fig. 6C), but, because our measurements report Ca²⁺ flow directly, they provide unequivocal evidence for a domain that regulates Ca²⁺ flux, just external to the gate, in the pore of P2X₂ channels.

Figure 6.

Toward a molecular basis for high Ca²⁺ flux at P2X₂ channels. A, The bar graph shows Pf% values for P2X₂ mutants, as indicated with amino acids substituted for changes in side chain size, charge, and hydrophobicity. WT, Wild type. The stars indicate statistical significance. B, The diagram illustrates the presently understood membrane topology of a P2X subunit and a helical net model of the second transmembrane segment. The secondary structure of TM2 is unknown, although it is not unreasonable to assume that this segment conforms to a general pattern of helical, pore-forming domains of other ion channels (Spencer and Rees, 2002). If so, then the residues influencing Pf% at P2X₂ channels cluster on one face of a predicted α helix, perhaps representing the molecular determinants of a Ca²⁺ selectivity filter. The graph shows the relative P_Ca/P_Cs data by Migita et al. (2001) plotted against the Pf% data shown in B. For the sake of clarity, mutants are designated by the last two digits of their position in the sequence of P2X₂ and by a letter code designating the substituted amino acid. For example, T339Y is designated as 39Y. Also included are permeability and Pf% data for the mutant T339E. This mutant showed elevated Ca²⁺ permeability and flux, as expected after addition of fixed negative charge to a critical position within the pore (Heinemann et al., 1992).

Discussion

Because P2X channels are widely expressed in excitable and non-excitable cells throughout the body (Norenberg and Illes, 2000; North, 2002; Schwiebert and Zsembery, 2003), it is important to understand the magnitude and mechanisms of ion flow through their pores. Several laboratories have quantified Ca²⁺ movement through the pores of some, but not all, P2X channels by measuring the relative permeability of Ca²⁺ to a reference monovalent cation (called P_Ca/P_M), with the reported values ranging from ∼1 to 4 for different members of the family (for review, see North, 2002). Many of these studies used different extracellular and intracellular concentrations of ions, different values for ion activities, and different algorithms to calculate relative permeability. Thus, it is not possible to directly compare values between studies, and consequently a precise understanding of Ca²⁺ flux for the P2X family in relation to other channels has been lacking. Here, we used a single, direct and model-independent method and a uniform set of physiological ionic conditions to quantify Ca²⁺ flux through all known homomeric and heteromeric P2X channels in relation to other transmitter-gated channels. We find that Ca²⁺ flux ranges from ∼3 to 15% of total current through the P2X pore in a manner that depends on the subunit composition of protein. This corresponds to P_Ca/P_M values between ∼1 and 3 if current follows constant field assumptions (Burnashev et al., 1995).

The main finding of the present study is that ATP-gated channels conduct an unexpectedly large flux of Ca²⁺ across cell surface membranes at resting membrane potentials in physiological solutions. As a family, P2X channels conduct more Ca²⁺ on average than do ACh, proton, serotonin, or glutamate-gated channels. Furthermore, the fractional Ca²⁺ current at the brain forms of P2X channels (P2X₂, P2X₄, P2X_2/6, and perhaps P2X_4/6) at ∼6–14% is significantly larger than that of AMPA (∼0.5–3.9%) and kainate channels (∼0.2–2%) (Burnashev et al., 1995), and, in the case of P2X₂, similar to that of P2X channels in sympathetic neurons (Rogers and Dani, 1995). Several P2X channels (P2X₁, P2X₄, P2X_2/6 and perhaps P2X_4/6) display fractional calcium currents between ∼8 and 15% that are larger than those of α7 nicotinic channels (Fucile et al., 2003) and equal to or greater than distinct NMDA channels at 8–14% (Burnashev et al., 1995). Furthermore, Ca²⁺ inflow through NMDA channels decreases at potentials more negative than –30 mV because of Mg²⁺ block of the pore (Burnashev et al., 1995), whereas this does not occur for P2X channels. Thus, it is likely that Ca²⁺ flux through P2X channels may dominate over NMDA channels at resting membrane potentials when Mg²⁺ blocks the NMDA channel. Furthermore, our data indicate that, in contrast to nicotinic and glutamate channels, P2X₂ channels may not use rings of fixed charge to select Ca²⁺. Rather, a critical domain (Migita et al., 2001) in the center of the pore influences Ca²⁺ flux (Fig. 6). Although the precise mechanism of ion selection is unknown, polar amino acids may provide the countercharge needed to partially dehydrate Ca²⁺ ions in a narrow part of the pore. At present, we do not know if this countercharge is supplied by backbone carbonyl oxygens, as is the case for K⁺ channels (Doyle et al., 1998), or directly by the side chains themselves. Indeed, the lack of sequence conservation in this domain among the family favors the latter hypothesis (Fig. 7). However, unlike the side chains of the amino acids that constitute the selectivity filter of K⁺ channels, the side chains of T336, T339, and S340 of the P2X₂ channel face into the aqueous environment of the pore (Egan et al., 1998) and therefore are well positioned to influence the flow of ions across the membrane. Different members of the family display a variable Ca²⁺ flux, and it will be interesting to see whether this range reflects the sequence variability in TM2. Future experiments designed to study the effects of site-directed mutagenesis at homologous sites of other family members are needed before a complete hypothesis about the mechanics of calcium permeability and flux can be presented. Furthermore, it is not apparent in the sequences presented in Figure 7 why the P2X₁ and P2X₄ channels should have such extraordinarily high Ca²⁺ fluxes; the explanation may reside at distant sites An obvious place to look would be TM1 because it, like TM2, is thought to line the pore (North, 2002). Alternatively, Ca²⁺ may be selected by parts of the protein besides the transmembrane domains. The identity of these distant sites may remain hidden until the structure of the channel is solved.

Figure 7.

Sequence alignment of the putative second transmembrane segments of P2X subunits. A stretch of 28 adjoining amino acids thought to span the membrane is shown for each of the six functional homomeric channels. The boxed amino acids are identical in all family members. Mutations of the three shaded amino acids of the P2X₂ channel result in changes in Pf%.

The present experiments on molecularly defined channels now call for similar experiments on endogenously expressed P2X channels in brain neurons. From this perspective, we note that previous estimates of Pf% for NMDA channels endogenously expressed in brain neurons (Schneggenburger et al., 1993), and heterologously expressed in HEK293 cells (Burnashev et al., 1995; Watanabe et al., 2002) differ by ∼7%, perhaps suggesting that presently unknown factors in neurons may regulate Pf%. It will be interesting to determine whether a similar situation exists for natively expressed P2X channels. The experiments with native P2X channels in brain neurons are challenging because not all neurons express P2X channels. Moreover, our recent systematic analysis of P2X channel expression in different fields of the hippocampus suggests that P2X channels may be trafficked to nerve terminals and dendrites rather than being expressed in the soma (Khakh et al., 2003). An additional consideration is that it is difficult to discriminate distinct P2X channels from each other because of the paucity of selective agonists and antagonists (Khakh et al., 2001). This problem is heightened by the fact that most neurons contain mRNA for multiple, and in some cases all, P2X subunits (Collo et al., 1996; North, 2002), suggesting that neurons may express mixtures of distinct P2X channels on their surfaces. However, our estimate of Pf% at ∼6% for P2X₂ is consistent with that measured previously for natively expressed P2X₂-like channels in SCG neurons (Rogers and Dani, 1995). The highest Pf% values for P2X₄ and P2X₁ at 10–15% channels are also broadly consistent with previous estimates at 15% from medial habenula neurons on the basis of reversal potentials (Edwards et al., 1997). Our demonstration of high Ca²⁺ flux for the whole P2X channel family offers a molecular interpretation of several physiological responses. Recent elegant studies show profound P2X₁ channel-mediated changes in intracellular Ca²⁺ in smooth muscle cells during fast ATP synaptic transmission (Brain et al., 2002; Lamont and Wier, 2002; Lamont et al., 2003). P2X₁ channels are well suited to this task because they are a large source of transmitter-activated Ca²⁺ flux (Fig. 4). The diverse presynaptic and postsynaptic effects of ATP on synaptic transmission and long-term potentiation (MacDermott et al., 1999; North, 2002) may be explained by variable expression of different P2X subunits and the associated differences in Ca²⁺ flux. P2X channels are also abundantly expressed in non-excitable cells, including epithelia, astrocytes, and microglia (Inoue, 2002; Schwiebert and Zsembery, 2003), in which the physiological response may be triggered by Ca²⁺ entry rather than a depolarization. Thus, P2X₄ channel activation in epithelial cells results in sustained Ca²⁺ entry that may trigger Cl^– secretion and prove to be of benefit in cystic fibrosis (Zsembery et al., 2003). Interestingly, recent studies show that upregulated P2X channels in microglia may trigger the release of factors such as cytokines and trigger allodynia (Tsuda et al., 2003). It is possible that the trigger for this is likely to be the substantial Ca²⁺ entry through P2X₄ subunit-containing channels. Mutant P2X channels with calibrated Ca²⁺ fluxes like those reported here could be used to further explore this possibility in vitro and in vivo and possibly in genetic approaches to treat P2X channel-associated pathologies (Tsuda et al., 1999, 2000, 2003). P2X channel subunits have been localized to brain nerve terminals (Vulchanova et al., 1996; Vulchanova et al., 1997; Le et al., 1998; MacDermott et al., 1999) and the periphery of dendritic spines (Rubio and Soto, 2001) by light and electron microscopy. Synaptically released ATP acting on P2X₁ channels evokes postsynaptic Ca²⁺ changes (Brain et al., 2002; Lamont and Wier, 2002; Lamont et al., 2003), and exogenous and endogenously released ATP causes a form of Ca²⁺-dependent presynaptic facilitation at some interneuron synapses (Khakh et al., 2003). Our data indicating that distinct P2X channels support significant, but variable, Ca²⁺ flux provides a molecular interpretation of these physiological studies and a rational to determine whether ATP gates a Ca²⁺ pathway in single dendritic spines that are known to express P2X channels (Rubio and Soto, 2001).