A C-terminal Lysine That Controls Human P2X4 Receptor Desensitization*

Receptor desensitization can determine the time course of transmitter action and profoundly alter sensitivity to drugs. Among P2X receptors, ion currents through homomeric P2X4 receptors exhibit intermediate desensitization when compared with P2X1 and P2X3 (much faster) and P2X2 and P2X7 (slower). We recorded membrane currents in HEK293 cells transfected to express the human P2X4 receptor. The decline in current during a 4-s application of ATP (100 μm) was about 30%; this was not different during whole-cell or perforated patch recording. Alanine-scanning mutagenesis of the intracellular C terminus identified two positions with much accelerated desensitization kinetics (Lys373: 92% and Tyr374: 74%). At position 373, substitution of Arg or Cys also strongly accelerated desensitization: however, in the case of K373C the wild-type phenotype was fully restored by adding ethylammonium methanethiosulfonate. At position 374, phenylalanine could replace tyrosine. These results indicate that wild-type desensitization properties requires an aromatic moiety at position 374 and an amino rather than a guanidino group at position 373. These residues lie between previously identified motifs involved in membrane trafficking (YXXXK and YXXGL) and implicates the C-terminal also in rearrangements leading to channel closing during the presence of agonist.

Receptor desensitization can determine the time course of transmitter action and profoundly alter sensitivity to drugs. Among P2X receptors, ion currents through homomeric P2X 4 receptors exhibit intermediate desensitization when compared with P2X 1 and P2X 3 (much faster) and P2X 2 and P2X 7 (slower). We recorded membrane currents in HEK293 cells transfected to express the human P2X 4 receptor. The decline in current during a 4-s application of ATP (100 M) was about 30%; this was not different during whole-cell or perforated patch recording. Alanine-scanning mutagenesis of the intracellular C terminus identified two positions with much accelerated desensitization kinetics (Lys 373 : 92% and Tyr 374 : 74%). At position 373, substitution of Arg or Cys also strongly accelerated desensitization: however, in the case of K373C the wild-type phenotype was fully restored by adding ethylammonium methanethiosulfonate. At position 374, phenylalanine could replace tyrosine. These results indicate that wild-type desensitization properties requires an aromatic moiety at position 374 and an amino rather than a guanidino group at position 373. These residues lie between previously identified motifs involved in membrane trafficking (YXXXK and YXXGL) and implicates the C-terminal also in rearrangements leading to channel closing during the presence of agonist.
P2X receptors belong to a family of ATP-gated ion channels, seven subunits of which have been currently identified and cloned (P2X 1-7 ) (1). Functional P2X receptors probably form as trimers, incorporating the same or different subunits (2,3). The P2X subunits are considered to have two hydrophobic transmembrane domains with intracellular N and C termini and a large extracellular ectodomain containing conserved cysteines (1); the ectodomain also contains residues that appear to contribute to the binding site for ATP (4,5).
In the central nervous system, P2X receptors have been implicated in mediating ATP-dependent fast excitatory neurotransmission (6 -9). The P2X 4 subunit is highly abundant in the central nervous system, with expression in both spinal cord and brain, including dentate gyrus, CA1/ CA3 pyramidal, and cerebellar Purkinje cells (10 -12); they are particularly localized to glutamatergic synapses (12). Upon activation, P2X 4 can regulate cellular Ca 2ϩ levels via direct permeation and by the activation of voltage-dependent Ca 2ϩ channels. P2X receptor activity has therefore been proposed to be important in synaptic plasticity (13,14).
The time course of the effect of ATP at P2X receptors is strongly influenced by receptor desensitization, a feature common to most other ligand-gated ion channels. Desensitization manifests itself as a decline in current amplitude during agonist occupancy of the receptor, and in the case of P2X 4 , it can be modified pharmacologically (15). Recombinant P2X receptors display varying degrees of desensitization; membrane current recordings show that P2X 1 and P2X 3 undergo fast desensitization (tens of milliseconds), P2X 4 exhibits moderate desensitization (several seconds), and P2X 2 , P2X 5 , and P2X 7 show less desensitization (1,16,17). For various P2X receptors, desensitization can be regulated by cellular signaling events, which further suggests its importance in P2X receptor physiology. Several studies have demonstrated regulation of P2X desensitization by protein kinase activity, either via direct N-terminal phosphorylation (at a highly conserved protein kinase C site) or via phosphorylation of an associated protein (18,19). P2X 3 desensitization can be decreased by the calcineurin-mediated dephosphorylation of the N terminus (20).
Experiments using chimeric constructs of P2X 1 and P2X 2 subunits identified desensitization critical domains as the transmembrane segments and the intracellular juxtamembrane regions (up to 15 amino acids in length). Other domains within the C terminus of P2X 2 have been identified by functional analysis of splice variants (21,22). P2X receptors have very different C-terminals, and the regions important for P2X 4 subunits are not known. Given the likely importance of the P2X 4 receptor in central synaptic physiology, it seemed useful to understand further the molecular mechanisms that may contribute to P2X 4 desensitization.

EXPERIMENTAL PROCEDURES
Human P2X 4 subunit with a C-terminal EMYPME epitope tag subcloned into pcDNA3.1(ϩ) was used as the template for all mutagenesis reactions. Mutagenesis was achieved using Pfu turbo polymerase and QuikChange methodology (Stratagene). Truncated receptors were generated by the introduction of a premature stop codon by mutagenesis. HEK293 cells were grown in 10% fetal calf serum in Dulbecco's minimal essential medium at 37°C with 5% CO 2 in a humidified incubator. Cells were transiently transfected with plasmids encoding wild-type or mutant P2X receptor (1 g) and enhanced green fluorescent protein (0.1 g) using Lipofectamine 2000 (Invitrogen).
Whole-cell recordings were made from HEK293 cells at room temperature, 24 -48 h after transfection. The extracellular solution contained (mM): 145 NaCl, 2 KCl, 2 CaCl 2 , 1 MgCl 2 , 13 D-glucose, and 10 HEPES. The intracellular pipette solution contained (mM): 145 NaCl, 10 HEPES, and 10 EGTA. Solutions had pH 7.3 after titration with 5 M NaOH. Chemicals were purchased from Sigma (Poole, UK). Pipettes had resistances of 3-5 M⍀. For perforated patch-clamp, perforation was achieved by the addition of 120 g/ml amphotericin in the patch pipette solution. Perforation typically occurred 5-8 min after gigaseal formation. ATP was applied using an RSC 200 system (Biological Science Instruments, Grenoble, France). Ethyl ammonium methane thiosulfonate (MTSEA 2 : Toronto Research Chemicals) was prepared as a concentrated stock, stored at Ϫ20°C, and diluted in extracellular solution immediately prior to application. MTSEA was applied by superfusion at a rate of ϳ2 ml/min. For immunohistochemistry, transfected cells were fixed with 4% paraformaldehyde for 5 min followed by permeabilization with 0.1% Triton X-100. Cells were blocked for 30 min (phosphate-buffered saline, 0.5% bovine serum albumin) and incubated with primary antibody (rabbit anti-EE, 1:500; Bethyl Laboratories) in blocking solution for 1-2 h at room temperature. After washing, cells were incubated with fluorescein isothiocyanate-conjugated secondary antibody (goat anti-rabbit, 1:200; Jackson ImmunoResearch) for 1-2 h at room temperature.
Numerical data are expressed as mean Ϯ S.E., and differences were tested by Student's t test and analysis of variance.

RESULTS
Decline in Response with Repeated ATP Application-We distinguish two forms of desensitization at the human P2X 4 receptor. The first is the decline in the peak current with repeated brief applications of ATP. Thus, in whole-cell recording, the peak current amplitude progressively declined when ATP (100 M, 2 s) was applied at 1-min intervals (Fig.  1a). This decline was well fit with a single exponential ( ϭ 93 Ϯ 6.2 s, n ϭ 25 cells), reaching a plateau level at ϳ20% of the initial peak amplitude. We observed no recovery from desensitization with repeated application when the application interval was increased to 5, 10, or 20 min (data not shown). P2X 4 currents recorded using the perforated patch configuration (amphotericin 120 g/ml) did not show decline with repeated applications ( Fig. 1, a and b). However, we were unable to influence the decline in response by a range of other perturbations. Ivermectin (3 M, 3 min) did not alter the kinetics of current decline ( ϭ 98 Ϯ 5.4 s; n ϭ 8, p Ͼ 0.05) (Fig. 1a), although it increased peak current amplitudes ϳ3-fold. The rate of current decline was unchanged in external solutions containing 0 mM calcium and also not different when the recording pipette contained ATP (1 mM), GTP (1 mM), staurosporine (100 nM), genistein (100 M), phosphatidylinositol 4,5-bisphosphate (50 M), or U73122 (10 M). P2X 4 ⌬Y378, which is truncated so as to lack the endocytic motif YEQGL (Fig. 2), showed properties not different from wild type receptors ( ϭ 91 Ϯ 5.2 s, n ϭ 12 cells, p Ͼ 0.05).
Decline in Response during ATP Application-The second form of desensitization is the decline in the membrane current during the continued application of ATP, and this is the subject of the remainder of this paper. Currents evoked by ATP (100 M) rose to their peak amplitude in 410 Ϯ 8.2 ms, and the average peak amplitude was 151 Ϯ 30 pA (n ϭ 25 cells). During a 5-s application of ATP (100 M), wild-type currents declined to a value that was 32 Ϯ 7.9% (n ϭ 25 cells) of their initial peak amplitude. These values were obtained in the whole-cell configuration, but the decline in current during the application was not different when recorded using the perforated patch configuration (27 Ϯ 5.2%, n ϭ 12 cells, p Ͼ 0.05) (Fig. 1c). However, the decline during the application was  it was also independent of holding potential (Ϫ60 mV versus ϩ60 mV).
Truncations and Alanine Scanning in the C Terminus-The C terminus of the human P2X 4 receptor extends from residue Leu 358 , at the end of the second transmembrane domain, to Gln 388 . A truncation that deleted the final eleven amino acids (⌬Y378) had properties not different from wild type P2X 4 ; it showed a similar desensitization during the application (Fig. 2) and a similar decline with repeated applications. Removal of the residues from and including Lys 373 or Lys 362 resulted in receptors that were non-functional (Fig. 2b).
These results indicate that maintained channel opening in the presence of ATP requires Lys 373 (or the equivalent side chain -S-S-(CH 2 ) 2 -NH 3 ϩ ). We asked therefore in how many subunits of a homotrimeric receptor was the lysine required? We co-expressed wild-type and P2X 4 [K373A] receptors in nominally equal amounts. Whole-cell recordings showed typical ATP-activated currents, but the decline during the application was intermediate between that seen for cells expressing either wild-type or P2X 4 [K373A] receptors alone (Fig. 4c). In fact, the current declined by 58 Ϯ 8.4% over 4 s (n ϭ 12 cells) during a 4-s application. This decline in the current was not different from that which would have been observed from an equal mixture of wild type and [K373A] substituted receptors. If one assumes that heterotrimer formation occurs readily between subunits containing alanine and those containing lysine and that only two (rather than four) phenotypes can occur, then this result indicates that a single alanine-containing subunit does not exert a dominant effect with respect to the desensitization kinetics (Fig. 4c).

DISCUSSION
Our results distinguish two modes of P2X 4 receptor desensitization. In our whole-cell experimental conditions, P2X 4 responses declined both during ATP applications lasting a few seconds and with applications repeated at 1-min intervals (Figs. 1a and 2a). With perforated patch recording to minimize dialysis of intracellular contents, the decline with repeated application ("run-downP") was completely prevented, whereas the decline during ATP applications was the same as in whole-cell recordings (Fig. 1). The prevention of P2X 4 rundown by per- forated patch recording is comparable with the work by Lewis and Evans (25), who showed that the run-down of a P2X 1 -like current in smooth muscle cells did not occur with amphotericin-permeabilized recordings. The phenomena of run-down has been observed for P2X 1 , P2X 3 , P2X 4 , and P2X 7 ; other P2X receptors give robust steady-state currents with repeated ATP applications. A likely explanation for P2X 4 run-down in whole-cell recording is that a diffusible cytosolic factor that tonically regulates the receptor is lost during intracellular dialysis. The identity of such a factor is not obvious, but from our experiments it seems not to be ATP, GTP, or phosphatidylinositol 4,5-bisphosphate. The mechanism underlying the decline in P2X 4 response with repeated ATP application is also not likely to involve AP2-mediated receptor internalization. This is because the truncation P2X 4 [⌬ Y378], and the point mutations Y378A and L382A, each of which lack the AP2-dependent endocytic motif YXXG⌽ responsible for receptor internalization in neurons (27), exhibited a decline in response with repeated ATP application that was not different from wild type receptors. These truncated receptors would presumably use the YXXV motif (beginning at Tyr 372 ) for their normal trafficking (24).
Ivermectin increases the currents evoked by ATP at P2X 4 receptors (26). A recent careful study by Priel and Silberberg (15) distinguished two effects: the first is an increase in the maximum current that is seen at concentration around 0.3 M, and the second is a slowing of deacti-vation that occurs at 10-fold higher concentrations. Both these effects were observed in our experiments, using 3 M ivermectin. However, in ivermectin there was no change in the decline of the peak currents observed with repeated application of ATP. Parenthetically, we also noted that the increase in current by ivermectin was observed in the truncated P2X 4 receptor P2X 4 [⌬Y378].
The C terminus of P2X 4 is short (30 residues) relative to that of other P2X receptors: 62% of these residues are either charged or aromatic (tyrosine) (Fig. 2a). By alanine scanning we identified two residues (Lys 373 and Tyr 374 ) as being important for the time course of desensitization during the ATP application. Both K373A and Y374A mutations resulted in receptors with much accelerated decline in response during ATP application as compared with wild-type receptor currents (Fig. 3a). These residues lie between two other functionally important regions for P2X receptors. These are the YXXXK motif that is found in all P2X family members (Tyr 367 to Lys 371 , human P2X 4 ) and the YEQGL sequence (Tyr 378 to Leu 382 ) found uniquely in P2X 4 subunits. The first of these contributes to receptor stabilization at the plasma membrane (23). The second has been shown to bind directly to the 2 subunit of the adaptor protein AP-2, where it occupies a site comparable to that seen for the more common endocytic motif YXX⌽ (24). This binding is key to clathrin-mediated receptor internalization undergone by P2X 4 receptors (27). These two residues that we have identified form the heart of the YKYV motif that is not normally used for clathrin-mediated endocytosis of P2X4 receptors (24,27).
We asked how critical was the requirement for tyrosine at position 374 and found that P2X 4 [Y374F] functioned essentially as the wild-type receptor with respect to the decline in current during the ATP application (Fig. 3a). This indicates that the requirement for aromaticity at this position for the channel to remain open during agonist occupancy. It further indicates that Tyr 374 is unlikely to be phosphorylated: for example, previous work (28) on P2X 7 receptors has indicated that repeated activation leads to receptor dephosphorylation and that this underlies the decline in the responses in that case.
On the other hand, arginine could not substitute for lysine at position 373 (Fig. 3a). This suggests that the critical moiety is not simply the positive charge but other features of the lysine side chain (such as the hydrocarbon chain). We tested this by expressing the P2X 4 [K373C] receptor and using MTSEA to add a side chain of similar length (-S-S-CH 2 -CH 2 -NH 3 ϩ ) to that of lysine (-CH 2 -CH 2 -CH 2 -CH 2 -NH 3 ϩ ).
This completely restored the wild-type phenotype. Control experiments showing that MTSEA did not modify the wild-type channel or the fast desensitizing mutant P2X 4 [K373A], and that the effect of MTSEA was prevented by intracellular cysteine (29 -31), strongly support the interpretation that MTSEA is acting by modification of the cysteine side chain at position 373. This indicates that lysine rather than arginine is required for structural reasons (e.g. the arginine side chain -CH 2 -CH 2 -CH 2 -NH-C(NH 2 ) ϭ NH 2 ϩ is too long to be accommodated) or perhaps that the lysine is subject to posttranslational modification. For example, post-translational modification of lysine has been shown to be important for the function of other ion channels (32). However, there are no consensus motifs for lysine modification apparent at position 373 and the relatively fast time course (5-10 min: Fig. 4b) with which MTSEA restored the wild type phenotype effect also argues against such an interpretation. The simplest explanation of why K373A mutation accelerates the decline in response during ATP application is that the critical positioning of this positive charge is required for channel opening to be maintained when ATP is bound. This implies that gating transitions involve not only conformational changes in the ectodomain and transmembrane hydrophobic regions but also the first part of the C terminus (33)(34)(35)(36). It is difficult to make further mechanistic interpretation, given that the residue at this position is lysine in one other P2X subunit (P2X 1 , which shows rapid desensitization) and negatively charged (Glu or Asp) in all others. In a previous study by Koshimizu et al. (22), the region 376 -381 (EDYEQG) was identified as being important in controlling P2X 4 desensitization during sustained applications of ATP. They measured the decline in intracellular calcium concentration over periods of several hundred seconds, so it is difficult to interpret those results with respect to the present studies in which ionic current was measured over a period of 5 s. Under our experimental conditions, these residues do not participate in receptor desensitization, as determined by alanine substitution or truncation. Previous work by Royle et al. (27) has shown that this region contains the first four amino acids of the motif (YEQGL) that interacts with the 2 subunit of the adaptor protein AP2 and is thus involved in receptor internalization. In general terms, this would be consistent with our interpretation that the decline in the current during the time frame of a few seconds that we have studied is unrelated to receptor internalization.
P2X receptors are generally considered to operate as trimers (1). Our final question was whether the rapid desensitization was observed when only one of the three subunits carried alanine at position 373 (i.e. P2X 4 [K373A]⅐P2X 4 ⅐P2X 4 ) (Fig. 4); in other words, did this mutation confer a dominant phenotype. The result of the co-expression of wild-type and mutant subunits in equal amounts (Fig. 4c) strongly indicates that accelerated desensitization is seen only in channels with two or more subunits carrying the alanine (i.e. P2X 4 [K373A]⅐P2X 4 [K373A]⅐P2X 4 ). This would imply that normal channel function can occur even though one of the three subunits is mutated to alanine at this position. We cannot exclude the alternative explanation for the results of the co-expression, which is that the lysine to alanine mutation prevents the formation of heteromeric channels, and that the currents observed flow through approximately equal parts of the homomeric wild type P2X 4 and mutant P2X 4 [K373A] channels. This seems unlikely given that the P2X 4 [K373A] has properties similar in many respects to wild type channels, which shows that alanines in this position (at least with three of them) do not themselves prevent formation of functional homomers.
In summary, we have identified Lys 373 and Tyr 374 in the C terminus of  the P2X 4 receptor as being key determinants of P2X 4 receptor desensitization on the time scale of seconds. This implicates the juxtamembrane C terminus in playing an important role in determining the duration of the physiological action of ATP at the P2X 4 receptors.