Patch–clamp coordinated spectroscopy shows P2X₂ receptor permeability dynamics require cytosolic domain rearrangements but not Panx-1 channels

Panx-1 Channels Do Not Contribute to the P2X₂ Receptor I₂ State.

We determined whether Panx-1 channels contribute to P2X₂ ion permeability increases and YOPRO1 uptake, which reflect the I₂ state (2, 3, 5, 9, 20). We chose to work on P2X₂ because of the wealth of available structure-function data (7, 8) and because P2X₂ receptor activation does not lead to cell death and downstream effects like those reported for P2X₇ (7). P2X₂ thus represents the more direct way to probe the mechanisms associated with the I₂ state. We used Panx-1 overexpression (21) and selective blockade to directly test whether these channels contribute to P2X₂ receptor permeability dynamics. We measured currents indicative of native Panx-1 in HEK cells (127 ± 24 pA for 100 mV jumps from −40 mV; n = 6) [supporting information (SI) Fig. S1A]. These were increased in cells heterologously expressing Panx-1 (to 493 ± 51 pA; n = 6) (Fig. S1B), and carbenoxolone (CBX) (10 μM) abolished the depolarization evoked currents (to 82 ± 18 pA; n = 5; Fig. S1C). Thus, Panx-1 currents are blocked >80% by CBX, as shown in ref. 21.

We next used imaging to measure ATP-evoked YOPRO1 uptake in cells expressing Panx-1, P2X₂ receptors (3) alone, in combination with Panx-1, or with CBX present in the bath (Fig. 1 B–E). At saturating concentrations of ATP (100 μM; Fig. 2A), if Panx-1 channels are a YOPRO1 conduit, then one expects elevated YOPRO1 uptake in cells overexpressing Panx-1, and reduced YOPRO1 uptake in cells where Panx-1 channels are blocked (16, 18). On the contrary, we found that ATP-evoked YOPRO1 uptake for P2X₂ receptors was unaffected by Panx-1 or CBX (Fig. 1 B–E). ATP did not trigger YOPRO1 uptake in cells expressing Panx-1 alone (Fig. 1C), and the P2X₂ receptor-mediated signals were abolished by PPADS (10 μM; Fig. 1 B and D), which is an antagonist for P2X receptors at this dose. We next examined ATP-evoked time-dependent NMDG⁺ permeability increases for P2X₂ receptors. In these experiments, the cells were bathed in NMDG⁺ containing extracellular solutions, and the reversal potential (E_rev) of the ATP-activated current was measured over time. Consistent with past work with WT P2X₂ (2, 4–6), immediately after applying ATP, the current had a negative E_rev (approximately −65 mV), indicating a low permeability to NMDG⁺ relative to Na⁺ (the intracellular cation). However, ≈6 s later, the E_rev had shifted to more positive voltages by ≈20 mV, indicating increased NMDG⁺ permeability as a function of ATP activation history. In contrast, we did not measure ATP-evoked currents in cells expressing Panx-1 alone and thus could not measure any E_rev (n = 6) (Fig. 1G). Our data for WT P2X₂ reproduce past work with P2X₂ (2, 4, 6, 19) and allowed us to determine whether Panx-1 and CBX affected NMDG⁺ permeability. We found that the shift in NMDG⁺ E_rev was identical from cells expressing WT P2X₂, P2X₂ with Panx-1, or P2X₂ with CBX (+19 ± 2, +20 ± 2, and +19 ± 2 mV, respectively; n = 10, 9, and 8) (Fig. 1 F–I). Because Panx-1 channels are activated at voltages positive to −40 mV (Fig. S1D reproduces work in refs. 21 and 22), one expects minimal contributions of the Panx-1 conductance to the P2X₂ current-voltage relation near the I₁ and I₂ state reversal potentials. This was observed in our experiments (Fig. 1 F–I). P2X₂ receptors also show NMDG⁺ permeability increases and YOPRO1 uptake in Xenopus oocytes that are devoid of native Panx-1 channels (Fig. S2 A–C) (18, 21). Taken together, our data argue against the Panx-1 model. However, we cannot formally exclude roles for an unknown channel, for which, to our knowledge, there is no evidence in the literature.

The Need for an Approach to Site-Specifically Label P2X₂ Receptors.

In its simplest form, the gating model posits that the I₂ state arises because of an intrinsic conformational change in the P2X₂ protein. If this is so, then state-specific conformational changes (i) must exist, (ii) need to be triggered by ATP, (iii) would develop in seconds with a time constant τ of ≈6 s, and (iv) decay over tens of seconds like the I₂ state (3, 4). The cytosolic domain is appropriate to look for such conformational changes, because this domain is needed for the I₂ state (3, 4, 6, 23). Fluorescent protein (XFP) tagging of the N termini of P2X₂ severely impaired function (Fig. S3 and Table S1), and the use of cysteine reactive fluorophores in a “cysteineless” channel background (24) is not possible for P2X receptors because replacement of native cysteines produces large functional deficits in receptor properties (25). Because of this, we explored the use of tetracysteine (4C) tags (26) (CCxxCC). Because of their small size, they can be inserted where XFPs are not tolerated (26). 4C tags can be specifically labeled with fluorescein arsenical hairpin (FlAsH) binder fluorophores (26). In the past, 4C tags and FlAsH-based FRET have been used to measure conformational changes in metabotropic receptors (27). We used tetracysteine (4C) tagging to the N and C termini of the P2X₂ receptors to generate P2X₂-NT-4C and P2X₂-CT-4C receptors (Fig. 2A). These functioned in a manner indiscernible from WT P2X₂ (Fig. 2A and Table S1). We also found that P2X₂ receptors carrying cytosolic domain 4C tags could be labeled with FlAsH (Fig. 2 B–D). For subsequent patch–clamp coordinated imaging experiments, we used dispersed and rounded up cells (3) (representative images in Fig. S4). We also found that FlAsH labeling of P2X₂-NT-4C and P2X₂-CT-4C receptors did not affect receptor function (Fig. 2A and Table S1). Moreover, in a specific set of experiments they also displayed NMDG⁺ permeability increases similar those of WT P2X₂ (WT, P2X₂-NT-4C, and P2X₂-CT-4C channels displayed E_rev shifts of 38 ± 6, 37 ± 7 and 25 ± 10 mV, respectively; P > 0.05 n = 5–6).

Site-Specific Optical Signals in the Cytosolic Domain of P2X₂ Receptors.

We simultaneously measured ATP-evoked currents (I_ATP) and changes in FlAsH fluorescence intensity (ΔF) from HEK293 cells expressing FlAsH-labeled P2X₂-NT-4C or P2X₂-CT-4C receptors (Fig. 3). If the N and C termini move, one expects the movement to alter the environment of the fluorophore, resulting in a change in its emission (28, 29). We used rapid puff applications of 100 μM ATP (for 0.1 s) to record changes in I_ATP and ΔF. This is because such brief ATP applications trigger entry into a permissive I₂ state (2, 3). This is a technical requirement, because brief ATP applications allowed us to measure ΔF during minimal ion flow. We measured robust and equally sized ATP-evoked currents for both constructs. However, we measured distinct optical signals for the N and C termini (Fig. 3 A and B).

The rise time of the optical signal for P2X₂-NT-4C constructs was described by a single exponential with a time constant τ of 2.0 ± 0.2 s (rate ≈0.5 s⁻¹; n = 8) (Fig. 3C), whereas the signal for P2X₂-CT-4C constructs was described by a double exponential with τ₁ at 2.0 ± 0.1 s (rate ≈0.5 s⁻¹) and τ₂ of 5.6 ± 1.4 s (rate ≈0.17 s⁻¹) (Fig. 3D) with amplitudes of each component at 59 ± 4 and 41 ± 4%, respectively (n = 11). The conformational changes at the N and C termini thus proceed over ≈2 s, and the C tail also moves over an additional ≈6 s. Thus, the total time for the optical signal at the C tail at ≈8 s is almost identical to opening to the I₂ state (2–5). We cannot formally exclude the possibility that the optical signals may also reflect desensitization. However, this is unlikely, because 4C-tagged P2X₂ receptors labeled with FlAsH do not desensitize during ≈0.1 s ATP applications (<1% desensitization) (Fig. 3 A and B) and desensitize incompletely even for ≈20-s applications of 100 μM ATP (35 ± 5 and 48 ± 8%, for P2X₂-NT-4C and P2X₂-CT-4C, respectively; n = 5 and 13). We are unaware of another demonstration that 4C tags and FlAsH can be used to report site-specific optical changes in membrane proteins.

We considered it important to determine whether the site-specific optical signals were impacted by ion flow through P2X channels. To this end, we repeated the experiments while holding the cells at +60 mV, i.e., when net ion flow is outward (7). Representative traces for ATP-evoked currents and ΔF are shown in Fig. 3 E and F; optical signals at +60 mV were preserved even though net ion flow was outward [at the N terminus, the τ was 1.9 ± 0.4%, whereas, at the C terminus, the decay was biphasic with τ (and amplitude) values of 1.9 ± 0.4 (56 ± 4%) and 5.9 ± 1.0 (44 ± 4%)]. We also used a variety of other approaches to rule out the possibility that the optical signals were due to ion flow (Fig. S5). We did not measure kinetically similar, small, reversible, and voltage-independent signals for P2X₂ tagged with YFP on the C tail (Fig. S5E).

Strategies That Disrupt the P2X₂ I₂ State Abolish Optical Signals for the N and C Termini.

We tested the hypothesis that the optical signals for P2X₂-NT-4C and P2X₂-CT-4C were due to conformational changes associated with entry to the I₂ state. This is because the ΔF kinetics for FlAsH-labeled P2X₂-NT-4C and P2X₂-CT-4C receptors recalled the slow kinetics for entry into the NMDG⁺ permeable I₂ state (2–5). We reasoned that, if the optical signals are due to conformational changes associated with the I₂ state, impairing the I₂ state should also abolish the optical signals. Previous work suggests that the I₂ state occurs only at concentrations of ATP >10 μM (6, 7). Consistent with this, 3 μM ATP evoked inward currents but led to negligible optical signals at the N or C termini (Fig. 3 I and J) and minor increases in NMDG⁺ permeability (the shift in the NMDG⁺ E_rev value was 4 ± 0.5 mV; n = 8). We also studied P2X₂-NT-4C and P2X₂-CT-4C receptors carrying T18A mutations (30), which results in channels that do not enter the I₂ state (3, 30). We detected ATP-evoked currents but no changes in ΔF (Fig. 3 K and L). We found that T18A mutations did not express as well as WT P2X₂ receptors (Fig. 3 I and J). To circumvent this limitation, we also used F31A mutants in TM1 that hinder the I₂ state (19); these express at levels equal to WT P2X₂. We found that peak current densities for F31A mutants in the P2X₂-NT-4C and P2X₂-CT-4C backgrounds were not significantly different to the cognate WT 4C-tagged receptors (Fig. 3 A, B, M, and N). We also found that the ATP-evoked optical signals for FlAsH bound to the N and C tails were abolished (Fig. 3 M and N) and, conversely, that F31A mutants displayed much reduced NMDG permeability increases, indicating a much reduced I₂ state (shift in NMDG E_rev was 7 ± 2 mV; n = 8). Thus, two mutations (T18A and F31A) and low concentrations of ATP that do not lead to the I₂ state also reduce or abolish the optical signals recorded at the N and C termini.

Spectral Studies.

We used a model peptide (26) carrying a 4C tag (WEAAAREACCPGCCARA) to characterize the optical properties of FlAsH for conditions relevant to the electrophysiological studies. The FlAsH was nonfluorescent alone, but it became brightly fluorescent upon binding to the peptide (Fig. 4A). We then measured the fluorescence intensity of peptide-bound FlAsH in solutions expected to increase hydrophobicity of the fluorophores environment by increasing the amount of glycerol (0, 30, 60, and 90% glycerol) (Fig. 4B). We also probed the optical signals for P2X₂ receptors carrying FlAsH on the cytosolic domain by measuring emission spectra from cells before and during ATP applications (29). In direct comparisons for P2X₂-NT-4C receptors, the ΔF value was −2 ± 0.4 and −2.5 ± 0.4% as measured by imaging and spectroscopy, respectively (n = 8 and 10). For P2X₂-CT-4C receptors, the ΔF value was −2.6 ± 0.5 and −3.3 ± 0.4% for imaging and spectroscopy, respectively (n = 11 and 10). This indicates that spectral imaging reports optical signals with signal-to-noise similar to that with conventional imaging (Fig. 3). We examined the peak emission wavelength of peptide-bound FlAsH under a variety of conditions and compared it with the peak emission wavelength of FlAsH bound to P2X₂-NT-4C and P2X₂-CT-4C receptors in the absence of ATP (Fig. 4E). We found that the peak was shifted to longer wavelengths in hydrophobic solutions containing glycerol (Fig. 4E). Moreover, the peak emission wavelength of FlAsH bound to P2X receptors (in the absence of ATP) was similar to that measured for peptide-bound FlAsH in glycerol, but significantly longer than that recorded for peptide-bound FlAsH in water, NaCl, or KCl solutions. In principle, an ATP-evoked decrease in FlAsH fluorescence intensity for P2X₂ receptor-bound FlAsH could be due to movement of the fluorophore, leading to increased quenching or a shift in the emission spectrum. We found that, for 4C-tagged P2X₂ receptors, ATP caused mainly a decrease in fluorescence intensity (Fig. 3A and Table S1). The subtle shift in emission peak of ≈2 nm for P2X₂-NT-4C receptors was minor when compared with the broad emission peak (Fig. 4F). Overall, for 4C-carrying constructs, the major effect of ATP activation of P2X receptors is to decrease the fluorescence emission intensity with very subtle or no effects on the emission peaks. This result suggests that the fluorophores move to sample an environment, where they may be more susceptible to quenching by nearby residues (28, 29).

Summary.

An unresolved issue in the P2X field has been how the I₂ state occurs for P2X receptors (7, 12). For instance, P2X₂, P2X₄, and P2X₇ receptors display NMDG⁺ permeability increases and YOPRO1 uptake, indicative of the I₂ state. This implies that the mechanisms underlying the I₂ state may be conserved between these receptor types from the same super family of membrane proteins. However, recent reports on P2X₇ receptors suggested that YOPRO1 uptake for P2X₇ receptors was mediated by accessory Panx-1 channels (11, 16–18). This raised the possibility that Panx-1 channels may underlie the I₂ state for other P2X receptors, such as P2X₂. We tested this possibility and found no evidence to support a role for Panx-1 channels in mediating the I₂ state of P2X₂ receptors. For several reasons, we did not examine P2X₇ receptors in the present study. First, P2X₇ receptor activation leads to a number of downstream effects in cells that complicate data interpretation (7, 11). For example, it is problematic to unequivocally conclude whether a given ATP-evoked response, such as YOPRO1 uptake, is due to activation of P2X₇ itself or some other downstream mechanism. These complications are avoided with P2X₂, because there is no evidence for comparable downstream signaling (7). Second, P2X₂ receptors are mechanistically well studied (7, 8), and these data provide both the tools and framework to interpret the results of our experiments, as discussed below.

What other evidence is there that may argue in favor or against Panx-1 channels as mediators of the I₂ state of P2X receptors? First, P2X₇ receptor mediated NMDG⁺ permeability increases and YOPRO1 uptake have been observed in oocytes that lack native Panx-1 (2, 31, 32). In contrast, a recent study concluded that these properties could only be measured when Panx-1 was coexpressed (18). Second, mutant P2X₂ and P2X₄ (2, 5) and human P2X₅ receptors (33) open to the I₂ state with little or no time delay and, in many cases, in oocytes that lack native Panx-1 (21). Third, in one study, Panx-1 was not part of the P2X₇ receptor signaling complex (34), whereas in another it was (16, 17). Fourth, a variety of mutations map throughout the cytosolic and transmembrane domains of P2X receptors and impair the I₂ state (3–5, 19, 35). We suggest the most straightforward interpretation is that these mutations impair conformational changes in P2X receptors themselves, rather than in Panx-1. Fifth, recent experiments suggest P2X₇ properties are not altered when Panx-1 channels are blocked (36). Our data do not detract from previous work on P2X₇, because the reported role of Panx-1 in P2X₇ receptor-mediated YOPRO1 uptake may be due to cell death/lysis (16–18). However, our data argue against Panx-1 channels as fundamental mediators of the I₂ state.

If the I₂ state is not due to Panx-1, then how does it arise? If the gating model has value, then conformational changes must exist that accompany opening to the I₂ state. We attempted to measure these directly and therefore provide a quantitative basis to probe the mechanistic basis of the I₂ state. Our initial approach was to use XFP tags, but we found that N-terminal fusions led to channels with altered properties. We overcame this limitation by using small 4C tags, which could be inserted at the N- and C-terminal tails without detriment. Moreover, the tags could be labeled with FlAsH. Therefore, we used these tagged channels to probe conformational changes occurring when ATP is applied and investigated their relation to the I₂ state. The ATP-evoked optical signals were not due to ion flow but were correlated with entry to the I₂ state. This correlation was based on (i) the kinetic similarity between the optical signals and entry into the I₂ state, (ii) mutants (T18A and F31A) that disrupt the optical signals and the I₂ state, and (iii) the ATP concentration dependency of the I₂ state. Similar approaches could now be used to study conformational dynamics of other ion channels, receptors, and transporters in single living cells.

We used FRET to measure conformational changes in P2X₂ receptors (3). However, the present study extends and goes beyond that work. First, we have now measured a conformational change at the N terminus. Second, we found that the conformational change at the C tail has two kinetic phases. Third, our analyses imply that the N terminus moves before the C terminus during permeability changes. Fourth, our data provide direct support for the suggestion that the N and C termini undergo functional and spatial interactions (30, 37). Thus, our data show site-specific conformational changes at the N and C termini that are needed for the channels to enter the I₂ state. In the simplest interpretation, cytosolic domain motions are intrinsic to P2X₂ receptors, but the optical signals may be impacted with interactions with other accessory proteins, a possibility that we are currently exploring in depth. There is no crystal structure for P2X receptors, and so a structural view of gating is not yet possible. P2X receptors share organizational and topological similarity to ASIC1b channels for which a structure is known (38), although they share no sequence similarity. The P2X pore is lined by TM2, where the selectivity filter resides (8). Thus, our experiments provide evidence of conformational changes that accompany gating to the I₂ state in a domain distinct from the selectivity filter itself. Extension of this approach to other sites will lead to a more complete understanding of how P2X receptors change during gating. Our data also demonstrate the utility of tetracysteine tags and biarsenical fluorophores to study conformational changes in membrane proteins.

Molecular Biology.

We used rat WT P2X₂ in pcDNA3.1 as a template; this was available from the work in ref. 3. The tetracysteine sequence (CCPGCC) was added by PCR and point mutations were made by Quick Change Mutagenesis (Stratagene). All constructs were sequenced.

HEK293 Cell Culture, Transfection, Labeling, and Recording.

HEK293 cells were maintained and transfected as described in detail by us (39). FlAsH labeling was carried out on the day of recording, using the recommendations in ref. 26, and the detailed protocol is described in SI Methods. Electrophysiology was performed as described in ref. 3 (see SI Methods). ATP was applied by using a Picospritzer III (Intracell). Data were collected by using pCLAMP9.2 and analyzed in Clampfit 9.0. Cells were imaged by using an Olympus BX50WI microscope (3) equipped with an Imago CCD camera, lamp, and condenser (Till Photonics). FlAsH fluorescence was excited at 488 nm, using a monochromator, and we used a 535/25 filter for emission. YOPRO1 (10 μM) uptake experiments were performed by using methods described in ref. 3. In some experiments (Fig. 2), confocal imaging was also used. In these cases, the cells were plated on glass-bottomed dishes (Matek), and imaged on Nikon Eclipse TE300 microscope equipped with a Bio-Rad Radiance 2000 confocal optics. To obtain fluorescent spectra of FlAsH bound to a tetracysteine motif, we used a model peptide of the following sequence: WEAAAREACCPGCCARA (40). A stock solution of 1 mM peptide was labeled with 5 μM FlAsH in the presence of 1 mM DTT. This stock solution was then diluted 200 times in different solvent so that the fluorescence of 250 nM of FlAsH bound to 5 μM peptide is measured. All readings were done in triplicate, using the same settings. Readings were performed by using a LS50B luminescence spectrometer (Perkin–Elmer). Data acquisition was performed by using WinLab software, Version 4.00.02 (Perkin–Elmer). The maximum excitation was set at 507 nm, and the maximum emission was set at 527 nm. For spectrometry on HEK293 cells expressing tagged channels, we used a microscope mounted Ocean Optics USB 2000 Spectrometer and the OOIbase32 software (all from Ocean Optics). This spectrometer allowed us to capture the emission spectrum from a single cell. In short, we positioned the cell on a pin hole in the emission path connected to the spectrometer.

Data Analysis.

All analysis was performed with Clampfit software, Version 9.0 and 10.1 (Molecular Devices Axon Instruments); Origin software, Versions 6.1 or 7.5 (OriginLab), GraphPad Instat software, Version 3.0; or ImageJ (National Institutes of Health). Data are mean ± SEM from at least five experiments.