Contributions of the C-terminal Domain to the Control of P2X Receptor Desensitization*

The P2X purinergic receptor channels (P2XRs) differ among themselves with respect to the rates of desensitization during prolonged agonist stimulation. Here we studied the desensitization of recombinant channels by monitoring the changes in intracellular free Ca2+ concentration in cells stimulated with ATP, the native and common agonist for all P2XRs. The focus in our investigations was on the relevance of the P2XR C terminus in controlling receptor desensitization. When expressed in GT1 cells, the P2XRs desensitized with rates characteristic to each receptor subtype: P2X1R = P2X3R > P2X2bR > P2X4R > P2X2aR > P2X7R. A slow desensitizing pattern of P2X2aR was mimicked partially by P2X3R and fully by P2X4R when the six-amino acid sequences of these channels located in the cytoplasmic C terminus were substituted with the corresponding arginine 371 to proline 376 sequence of P2X2aR. Changing the total net charge in the six amino acids of P2X4R to a more positive direction also slowed the receptor desensitization. On the other hand, substitution of arginine 371-proline 376 sequence of P2X2aR with the corresponding sequences of P2X1R, P2X3R, and P2X4R increased the rate of receptor desensitization. Furthermore, heterologous polymerization of wild-type P2X2aR and mutant P2X3R having the C-terminal six amino acids of P2X2aR at its analogous position resulted in a functional channel whose desensitization was significantly delayed. These results suggest that composition of the C-terminal six-amino acid sequence and its electrostatic force influence the rate of receptor desensitization.

The P2X purinergic receptor channels (P2XRs) differ among themselves with respect to the rates of desensitization during prolonged agonist stimulation. Here we studied the desensitization of recombinant channels by monitoring the changes in intracellular free Ca 2؉ concentration in cells stimulated with ATP, the native and common agonist for all P2XRs. The focus in our investigations was on the relevance of the P2XR C terminus in controlling receptor desensitization. When expressed in GT1 cells, the P2XRs desensitized with rates characteristic to each receptor subtype: P2X 1 R ‫؍‬ P2X 3 R > P2X 2b R > P2X 4 R > P2X 2a R > P2X 7 R. A slow desensitizing pattern of P2X 2a R was mimicked partially by P2X 3 R and fully by P2X 4 R when the six-amino acid sequences of these channels located in the cytoplasmic C terminus were substituted with the corresponding arginine 371 to proline 376 sequence of P2X 2a R. Changing the total net charge in the six amino acids of P2X 4 R to a more positive direction also slowed the receptor desensitization. On the other hand, substitution of arginine 371-proline 376 sequence of P2X 2a R with the corresponding sequences of P2X 1 R, P2X 3 R, and P2X 4 R increased the rate of receptor desensitization. Furthermore, heterologous polymerization of wild-type P2X 2a R and mutant P2X 3 R having the C-terminal six amino acids of P2X 2a R at its analogous position resulted in a functional channel whose desensitization was significantly delayed. These results suggest that composition of the C-terminal six-amino acid sequence and its electrostatic force influence the rate of receptor desensitization.
ATP-gated receptor channels (P2XRs) 1 are expressed in a number of tissues where they are involved in regulation of a wide variety of cellular functions, including central and peripheral neurotransmission, smooth muscle contraction, platelet activation, and hormone secretion (1)(2)(3). The regulation of these functions by the activated P2XRs requires or is susceptible to local and/or global changes in intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) (3). Binding of extracellular ATP to P2XRs is associated with a rise in [Ca 2ϩ ] i , which is mediated by Ca 2ϩ influx through these channels as well as by depolarization of cells and activation of voltage-gated Ca 2ϩ channels. The pattern of [Ca 2ϩ ] i signaling by native P2XRs is highly variable, depending on the cell type examined (3). In contrast to other Ca 2ϩ -conducting channels, however, the relevant structural base of the P2XRs contributing to the control of [Ca 2ϩ ] i signaling function has not been completely elucidated.
Using molecular cloning techniques, seven subunits of P2XRs have been obtained so far and are named P2X 1 R to P2X 7 R (4 -10). They can form Ca 2ϩ -permeable pores through homo-and heteropolymerization (11,12). Each subunit is proposed to have two putative transmembrane helices connected with a large extracellular loop, and both the N and C termini are located in the cytoplasm. From their N termini to the second transmembrane domain, the cloned subunits exhibit a relatively high level of amino acid (aa) sequence homology compared with their C termini, which are variable in lengths and show no apparent sequence homology except for the proximal region near the second transmembrane domain (1). During continuous exposure to agonist, current signals generated by recombinant P2XRs are desensitized gradually, an action that should effectively attenuate or terminate direct and indirect Ca 2ϩ influx.
Based on the observed differences in their current desensitization kinetics, recombinant P2XRs are generally divided into two groups: the rapidly desensitizing (P2X 1 R and P2X 3 R) and the slowly desensitizing (P2X 2 R, P2X 4 R, P2X 5 R, P2X 6 R, and P2X 7 R) (3). Experiments with chimera subunits composed of P2X 2 R and P2X 1 R or P2X 3 R subunits suggested that the rapid desensitization requires interactions between two transmembrane domains of receptor subunits (13). A large C terminus of P2X 7 R has also been suggested to account for the nondesensitizing pattern of these channels during repetitive stimulation (10). Recently, a new view of P2XR desensitization has emerged; the C-terminal splice variant of P2X 2 R, termed P2X 2b R or P2X 2-2 R, was found to lack a stretch of 69 aa and to desensitize faster than the full-length P2X 2 R, termed P2X 2a R (14 -16). Amino acids responsible for such a functional difference between the spliced and full-length channels are localized to the initial six residues (Arg 371 -Pro 376 ) within the spliced segment (17).
Here, we examined the importance of the C-terminal 6-aa sequences of P2XRs, which correspond to Arg 371 -Pro 376 of P2X 2a R (Fig. 1), to the desensitization pattern of these channels. For this purpose, P2XRs and their mutants were expressed in GT1 immortalized neurons, and the impact of receptor desensitization on the pattern of [Ca 2ϩ ] i signaling was analyzed by monitoring single cell [Ca 2ϩ ] i . These cells are excitable and express two types of voltage-gated calcium channels, T-and L-type (18). On the other hand, neither P2XRs nor Ca 2ϩ -mobilizing P2Y receptors are native for these cells (16).
Our results indicate that the pattern of receptor desensitization is unique for each homo-and heteropolymeric channel and that the structural differences in the C-terminal small region in part account for such variety of responses.
All C-terminal mutants were made by a PCR-based overlap extension method (21) using the wild-type receptor cDNA as a template. Entire PCR fragments carrying mutations were subcloned into pBluescript II, sequenced, and transferred to C termini of the wild-type expression constructs using the following restriction sites: ClaI/Hin-dIII, BspEI/HindIII, PstI/HindIII, and SmaI for P2X 2a R, P2X 3 R, P2X 4 R, and P2X 7 R mutants, respectively. All restriction enzymes were obtained from New England Biolabs (Beverly, MA). The full-length P2X 3 R and its chimera mutant with P2X 2a R were transferred from pBluescript to a bicistronic enhanced fluorescent protein vector, pIRES2-EGFP (CLONTECH, Palo Alto, CA), at its XhoI/EcoRI site, generating pIRES/ P2XR 3 and pIRES/P2X 3 /X 2a R, respectively.
Cell Cultures and Transfection-GT1 cells were cultured in Dulbecco's modified Eagle's medium/Ham's F-12 (1:1) containing 10% (v/v) fetal bovine serum, 100 g/ml streptomycin, and 100 units/ml penicillin. Procedures for transient transfection in GT1 cells were performed as described (17) with minor modifications. Briefly, cells were plated on coverslips coated with poly-L-lysine at a density of 0.5-1.0 ϫ 10 5 cell/ 35-mm dish and allowed to grow for 24 h. On the day of transfection, the total amount of 1.2 g of expression constructs encoding wild-type or mutant P2XRs was mixed with 8 l of cationic lipid, LipofectAMINE TM , in 1.2 ml of reduced serum medium, Opti-MEM (Life Technologies, Inc.) for 15 min at ambient temperature. The DNA mixture was then applied to cells for 3-5 h and replaced by normal culture medium. The cells were subjected to experiments 24 -48 h after the transfection. For co-transfection experiments, a ratio of 1:2 DNA for P2X 2a R and pIRES constructs was used, keeping the total DNA amount same as above.
[Ca 2ϩ ] i Measurements-For single cell [Ca 2ϩ ] i measurements, cells were incubated at 37°C for 60 min with 1 M fura-2 AM in phenol redand ATP-free Dulbecco's modified Eagle's medium. The cells were subsequently washed with assay buffer containing 137 mM NaCl, 5 mM KCl, 1.2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES (pH 7.4), and 10 mM glucose and kept for at least 30 min in this medium before measurements. Apyrase (grade I) was purchased from Sigma and used at 20 g/ml throughout the incubation process indicated. Coverslips with cells were mounted on the stage of an Axiovert 135 microscope (Carl Zeiss, Oberkochen, Germany) attached to the Attofluor digital fluorescence microscopy system (Atto Instruments, Rockville, MD). [Ca 2ϩ ] i responses were examined under a 40ϫ oil immersion objective during exposure to alternating 340-and 380-nm light beams, and the intensity of light emission at 520 nm was measured. The ratio of light intensities, F 340 /F 380 , which reflects changes in [Ca 2ϩ ] i , was simultaneously followed in several single cells. Cells expressing fluorescence protein were optically detected by an emission signal at 520 nm when excited by 488-nm ultraviolet light and were not detectable by 340-or 380-nm excitations. We thus considered the emission signal from fluorescence protein by 340-or 380-nm excitations within our background level. In co-transfection experiments of P2X 2a R cDNA and pIRES vectors, about 75% of fluorescent protein-positive cells responded to 100 M ATP stimulation and were considered to be co-transfected. Lower co-transfection efficiency below this level tended to show small response, and such experiments were excluded from further analysis.
Calculations-Since transfection efficiencies and expression levels of recombinant receptors were variable among the cells, only cells showing the peak [Ca 2ϩ ] i amplitude of more than 1.0 in the F 340 /F 380 ratio scale were further used for evaluating the desensitization rates. The time course of [Ca 2ϩ ] i response to ATP was fitted to one or two exponential functions using SigmaPlot 5.0 (Jandel Scientific software, San Rafael, CA) with the tolerance value set at 10 Ϫ5 . Significant differences were determined by either Student's t test or one-way analysis variance (ANOVA) followed by Scheffe's test, if applicable; p Ͻ 0.05 was considered as significantly different.  1. Schematic representation of the P2XR with two putative transmembrane domain model and amino acid sequences of the proximal C termini. Shading indicates conserved residues. In the terminal six-amino acid segments shown, negatively charged residues are indicated by squares, and positively charged residues are indicated by circles. The arrow indicates the beginning of the spliced site in P2X 2a R C terminus. M1 and M2 indicate the transmembrane domains 1 and 2, and N and C indicate the N and C termini, respectively. activation of P2X 2b R and P2X 4 R showed a relatively slow decrease compared with those generated by the P2X 1 R and P2X 3 R. In contrast, P2X 2a R-and P2X 7 R-expressing cells induced long-lasting [Ca 2ϩ ] i signals. The P2X 2a R-induced [Ca 2ϩ ] i signals usually decreased to one-third of maximum response, and this steady-state plateau level was reached within 10 min of stimulation. The P2X 7 R-expressing cells showed no obvious decline in [Ca 2ϩ ] i response for more than 10 min. In all cases, except for P2X 7 R, single exponential functions were sufficient to describe the decline rates of [Ca 2ϩ ] i responses. The mean values of calculated rate constants are shown in Fig. 2B. The rank order of desensitization rates derived from these data

Desensitization Pattern of Homopolymeric
was highly comparable with that observed in current measurements (22), confirming the validity of single cell [Ca 2ϩ ] i measurements as a method for comparative evaluation of receptor desensitization.
Role of a C-terminal 6-aa Sequence in Receptor Desensitization-Recently, we found that the Arg 371 -Pro 376 sequence, located in the cytoplasmic region of P2X 2a R, is necessary for the slow desensitizing pattern of these receptors (17). As shown in Fig. 1, this 6-aa region is located near the second putative transmembrane domain, and the C-terminal difference in amino acid sequences among the members of P2XRs starts from this region. To study the possible role of structural diversity of this region in the control of receptor desensitization, the Arg 371 -Pro 376 sequence of P2X 2a R was introduced to P2X 3 R and P2X 4 R instead the native Thr 362 -Lys 367 and Glu 376 -Gly 381 sequences, respectively. Such C-terminal chimeric subunits were termed P2X 3 /X 2a R and P2X 4 /X 2a R (Figs. 1 and 3). Cells expressing the P2X 3 /X 2a R mutant responded to 100 M ATP with an apparent delay in desensitization of [Ca 2ϩ ] i signals compared with the wild-type P2X 3 R (Fig. 3A). Furthermore, the averaged spike [Ca 2ϩ ] i amplitude was significantly higher in cells expressing P2X 3 /X 2a R than in cells expressing P2X 3 R (F 340 /F 380 : 1.6 Ϯ 0.1 (n ϭ 31) versus 2 Ϯ 0.05 (n ϭ 34), respectively). As seen in wild-type P2X 3 R, the addition of apyrase to the bath solution was necessary to detect the [Ca 2ϩ ] i change in P2X 3 /X 2a R-expressing cells, indicating that sensitivity of P2X 3 R to ATP was not largely affected by introducing Arg 371 -Pro 376 sequence to its C terminus.
Cells expressing P2X 4 /X 2a R also showed a slower desensitization rate compared with the cells transfected with the wildtype P2X 4 R (Fig. 3B). The calculated time constants for the desensitization rates were 12 Ϯ 2 ms Ϫ1 (n ϭ 12) for P2X 4 R versus 4 Ϯ 0.7 ms Ϫ1 (n ϭ 12) for P2X 4 /X 2a R. The averaged peak [Ca 2ϩ ] i amplitudes induced by 100 M ATP were elevated in the mutant P2X 4 /X 2a R compared with wild-type P2X 4 R (F 340 /F 380 : 2.1 Ϯ 0.1 (n ϭ 72) and 1.6 Ϯ 0.1 (n ϭ 55), respectively). On the other hand, the half maximal doses of ATP required to induce maximum peak [Ca 2ϩ ] i response were comparable for wild-type and chimeric channels (0.8 and 1.6 M, respectively), indicating that ATP is equipotent for both wild-type P2X 4 R and mutant P2X 4 /X 2a R.
In further studies, the reverse mutation was performed at P2X 2a R C terminus by substituting the Arg 371 -Pro 376 sequence for the corresponding 6-aa sequence of P2X 3 R, P2X 4 R, and P2X 1 R, and the mutant P2X 2a R subunits were termed P2X 2a / X 3 R, P2X 2a /X 4 R, and P2X 2a /X 1 R, respectively. As shown in Fig.  4, traces b, c, and d, all chimeric subunits formed functional channels in GT1 cells with enhanced desensitization rates when compared with the wild-type P2X 2a R (trace a). However, none of the P2X 2a R mutants was able to mimic the rate of P2X 2b R desensitization (trace e). In terms of averaged peak [Ca 2ϩ ] i amplitudes, no significant difference in [Ca 2ϩ ] i re-sponses was observed between wild-type and mutant P2X 2a R.
The relevance of the 6-aa sequence in the control of desensitization in wild-type and chimeric channels are summarized in Fig. 3C and Table I. The rate constant for the desensitization of chimeric P2X 4 /X 2a R was indistinguishable from that of P2X 2a R (Fig. 3B). Also, no significant differences were observed between P2X 4 R and P2X 2a /X 4 R (Table I), indicating that the structural element responsible for the distinct pattern of P2X 2a R and P2X 4 R desensitization is localized to the region of the 6-aa C-terminal sequence. In contrast, the desensitization rates of P2X 2a /X 1 R and P2X 2a /X 3 R were significantly enhanced by mutations but not enough to reproduce responses seen in cells expressing wild-type P2X 1 R and P2X 3 R (Table I). Similarly, the desensitization rate of P2X 3 /X 2a R was considerably reduced, but not to the level seen in cells expressing wild-type P2X 2a R. Therefore, a part of the subunit molecule other than the C terminus also participates in the control of P2X 1 R and P2X 3 R desensitization. This agrees with literature data (13) suggesting that the transmembrane domains are critical for the fast desensitization process of P2X 1 R and P2X 3 R signaling.
We have also analyzed the relevance of P2X 7 R C-terminal 6-aa sequence on receptor desensitization. The optimized results of the amino acid sequence alignments for P2XRs revealed that Val 392 -Pro 397 sequence of P2X 7 R corresponds to the Arg 371 -Pro 376 sequence of P2X 2a R. However, a 18-aa stretch from Cys 362 to Val 379 of P2X 7 R is not related to any other subunit C termini. Thus, an insertion was needed in the alignment (1). Substitution of Val 392 -Pro 397 sequence with the corresponding Glu 376 -Gly 381 sequence of P2X 4 R generated a functional P2X 7 /X 4 R mutant receptor, the Ca 2ϩ -signaling function of which was indistinguishable from that of wild-type P2X 7 R. This suggests that the C-terminal 6-aa sequence of P2X 7 R subunit is not functionally equivalent to those found in other subunits in terms of controlling receptor desensitization.
Identification of Residues Contributing to the Receptor Desensitization-The in-frame splicing of the P2X 2a R C terminus starts at Val 370 (indicated by an arrow in Fig. 1) and effectively removes three positively charged residues from the Arg 371 -Pro 376 sequence. On the other hand, negatively charged residue(s) were found in the corresponding segment of P2X 1 R, P2X 3 R, P2X 2b R, and P2X 4 R (Fig. 1), and all these receptors apparently showed faster desensitization than P2X 2a R. Such a structural difference raised the possibility that charged residues in this region may play an important role in P2XR desensitization. This is especially relevant for P2X 2a /X 4 R and P2X 4 / X 2a R mutants, whose desensitization patterns were indistinguishable from the wild type P2X 4 R and P2X 2a R, respectively.
To test this, we next evaluated the effects of single amino acid substitution in the P2X 4 R C terminus on the pattern of ATP-induced [Ca 2ϩ ] i response. The negatively charged residues in the 6-aa sequence of P2X 4 R were replaced with neutral ones, or the basic residues of the Arg 371 -Pro 376 sequence were introduced. As shown on Fig. 5, both elimination of negatively  Table I. charged residues from and introduction of positively charged residue to the position of Glu 376 in P2X 4 R decreased the rate of desensitization. Mutation of Glu 379 to a basic lysine residue also resulted in a significant decrease in the rate of desensitization. However, the point mutations at Glu 376 or Glu 379 could not produce the same extent of desensitization as seen in cells expressing wild-type P2X 2a R and mutant P2X 4 /X 2a R. In addition, we constructed two mutant P2X 4 R subunits having 3-aa substitutions, with native acidic moieties of P2X 4 R changed to neutral (P2X4E376Q/D377Q/E379Q) or three basic amino acids of P2X 2a R introduced at the corresponding sites of P2X 4 R (P2X4RE376R/E379K/Q380H). Both triple mutants desensitized slower than the wild-type P2X 4 R (Fig. 5). Thus, changing the amino acid composition and the net charge of this particular C-terminal segment altered the duration of Ca 2ϩ influx induced by P2X 4 R. Within the Arg 371 -Pro 376 of P2X 2a R, Thr 372 represents an optimum phosphorylation site, (R/K)XX(T/S), for the type II calcium/calmodulin-dependent protein kinase (23). At the analogous position of Thr 372 , either glutamate or aspartate was found in P2X 1 R, P2X 3 R, P2X 2b R, and P2X 4 R (Fig. 1). We speculated that phosphorylation of the single Thr 372 residue or introduction of a negatively charged moiety at this position may change the desensitization rate of P2X 2a R. However, substitution of this residue with glutamate or with glutamine did not affect the desensitization rate of mutant channels compared with that of the wild type. The desensitization constants for glutamate-and glutamine-containing mutants were 4 Ϯ 0.3 ms Ϫ1 (n ϭ 35) and 5 Ϯ 0.3 ms Ϫ1 (n ϭ 21), respectively, and 4 Ϯ 0.2 ms Ϫ1 (n ϭ 26) for wild-type channels. Also, no obvious difference was observed in the maximum amplitude of [Ca 2ϩ ] i response and the EC 50 values for ATP between P2X 2a R and each C-terminal mutant receptors examined (not shown).
Desensitization Pattern of Heteropolymeric P2XRs-The coexpression of P2X 2a R and P2X 3 R has been shown to form a heteropolymeric P2XR channel, which exhibits a distinct pharmacological profile and desensitization pattern from those seen in homopolymeric channels (11,12). We used this particular subunit combination to analyze the impact of the C-terminal structure on the [Ca 2ϩ ] i -signaling pattern of heteropolymeric channels. GT1 neurons expressing only P2X 3 R responded to 20 M AMP-CPP with a rise in [Ca 2ϩ ] i , the amplitude and desen-sitization rate of which were highly comparable with that observed in 100 M ATP-stimulated cells (Fig. 2). When co-expressed, these two subunits generated a channel that also responded to 20 M AMP-CPP but with the amplitude of response comparable with that observed in homopolymeric P2X 2a R stimulated with 100 M ATP (Fig. 6A). In contrast, cells co-transfected with an empty vector and cDNA for P2X 2a R did not respond to this agonist (Fig. 6A).
Under the continuous presence of AMP-CPP, the subsequent application of 100 M ATP further increased [Ca 2ϩ ] i , suggesting the existence of two channels with distinct pharmacological features in a co-transfected cell: AMP-CPP-sensitive and -insensitive (Fig. 6A). When the co-transfected cells were initially stimulated with 100 M ATP, the subsequent addition of 100 M AMP-CPP failed to induce any change in [Ca 2ϩ ] i (not shown). Therefore, all P2XRs-expressed cells were sensitive to ATP and desensitized completely during the initial stimulation. This suggests that the AMP-CPP-mediated rise in [Ca 2ϩ ] i was initiated by the activation of heteropolymeric P2X 2a R ϩ P2X 3 R, which desensitized more rapidly than the homopolymeric P2X 2a R but slowly when compared with P2X 3 R (Figs. 2  and 6).
The co-expression of the P2X 2a R and C-terminal mutant P2X 3 /X 2a R also resulted in the formation of AMP-CPP-sensitive heteropolymeric channels (Fig. 6). However, the desensiti- zation of [Ca 2ϩ ] i signals generated by these channels was significantly slower than in P2X 2a R ϩ P2X 3 R-expressing cells (9.5 Ϯ 0.3 (n ϭ 63) versus 13.5 Ϯ 0.9 (n ϭ 34), respectively). These results indicate that the C-terminal structure of the participating subunits influences the desensitization of heteropolymeric P2X 2a R ϩ P2X 3 R channels in a manner comparable with that observed in homopolymeric channels. It is also important to stress that the exact subunit stoichometry of the heteropolymeric P2X 2a R ϩ P2X 3 R and P2X 2a R ϩ P2X 3 /X 2a R in our experiments is not known and might vary within the cells. To overcome the possible impact of such heterogeneity on conclusions driven from these experiments, the number of recordings from which means were derived was elevated compared with other experiments (see above). DISCUSSSION In this study, we employed GT1 neurons as an expression system to analyze the impact of receptor desensitization on the Ca 2ϩ -signaling function of recombinant P2XRs and to identify the role of C terminus in receptor desensitization. The time scale for [Ca 2ϩ ] i measurements in our experiments was set to be sufficiently long (5-10 min) to incorporate the consequence of a pore dilation effect during ATP stimulation reported by others (24,25) on Ca 2ϩ signaling. Under such recording conditions, activation of all homopolymeric P2XR by ATP caused a rise in [Ca 2ϩ ] i , the pattern of which was highly specific for each particular channel. Receptors differed among themselves with respect to the amplitude of Ca 2ϩ response. [Ca 2ϩ ] i signals initiated by the homopolymeric P2XRs also desensitized with rates characteristic to each receptor subtype: The kinetics of receptor desensitization estimated in single cell [Ca 2ϩ ] i measurements should be interpreted with two reservations. First, voltage-gated calcium channels, which are expressed by GT1 cells (18), were not silenced during [Ca 2ϩ ] i recording. Second, although the rank orders for receptor desensitization estimated in current and [Ca 2ϩ ] i measurements were highly comparable, the time needed to reach the steady desensitized states for P2XRs was significantly longer when estimated in [Ca 2ϩ ] i measurements (22). This probably reflects the slow kinetics of Ca 2ϩ elimination from the cytoplasm, which was additionally enhanced by the integration of voltage-gated Ca 2ϩ influx during receptor stimulation. Thus, [Ca 2ϩ ] i recordings are of limited use for studies on the dynamics of channel behavior, and the calculated rates should be interpreted only as a relative indicator of the status of receptor desensitization, especially when compared with results obtained in electrophysiological measurements.
However, we previously found a marked and comparable difference in the desensitization rates of P2X 2a R and spliced P2X 2b R in both current recordings and single cell [Ca 2ϩ ] i measurements, the later done under conditions where voltage-gated calcium channels were silent (16). In addition, the amplitudes of [Ca 2ϩ ] i response induced by these channels were significantly reduced when Ca 2ϩ influx through L-type Ca 2ϩ channels was blocked by nifedipine, but the estimated rate of receptor desensitization was not affected (17). Finally, the rank order of receptor desensitization estimated in [Ca 2ϩ ] i measurements was highly comparable to those observed by others in current measurements for P2X 1 R, P2X 3 R, P2X 2a R, P2X 2b R, and P2X 7 R (22).
In our experimental conditions, the rapidly desensitizing P2X 1 R and P2X 3 R were unable to induce a measurable increase in [Ca 2ϩ ] i when GT1 cells were bathed in physiological solution, but the inclusion of apyrase in extracellular space recovered receptors from the desensitized state. This suggests that the spontaneous release or pathological leakage of cellular ATP may prevent the fast-desensitizing receptors from responding to ATP stimulation with an increase in [Ca 2ϩ ] i . P2X 4 R exhibited fast desensitization when expressed in oocytes and slow desensitization when expressed in HEK293 cells (19,26,27), indicating that experimental conditions, including a difference in amphibian and mammalian expression systems, may change the recombinant channel behavior (24). In our expression system, P2X 4 R completely desensitized during continuous agonist stimulation. With respect to the rate of desensitization, this channel should be considered as a relatively slow desensitizing one.
These observations support the validity of [Ca 2ϩ ] i measurements as an indicator of P2X receptor desensitization. Since the pattern of [Ca 2ϩ ] i response initiated by P2XRs represents signals that encode the activity of these receptors for controlling cellular functions, the time course for desensitization estimated by single cell [Ca 2ϩ ] i measurements is highly relevant. Also, activation of voltage-gated Ca 2ϩ influx is a physiological mechanism by which P2XRs amplify [Ca 2ϩ ] i signals in excitable cells in addition to conducting Ca 2ϩ through their pores. All together, these results indicates that selective expression of P2XR subunits in an excitable cell can serve as an effective mechanism for generating specific Ca 2ϩ signals.
Our results further indicate that the variable subunit C termini in part accounts for the observed differences in the desensitization of channels. The structural element controlling the desensitization pattern of P2X 2a R and P2X 4 R appears to be exclusively localized to the 6-aa C-terminal sequence. Within the Arg 371 -Pro 376 of P2X 2a R, Thr 372 represents an optimum phosphorylation site, (R/K)XX(T/S), for type II calcium/calmodulin-dependent protein kinase and is not present in P2X4R. Nonetheless, this residue is not responsible for the difference in the desensitization of these two channels, as documented in experiments with single amino acid mutations.
It is likely that total net charge in the Arg 371 -Pro 376 sequence of P2X 2a R and in the equivalent sequences of P2X 3 R, and P2X 4 R is an important determinant of the extent of Ca 2ϩ influx during sustained ATP stimulation. In a case of mutant P2X 2a R/ X 4 R, the replacement of three positive net charges with three negative ones in the equivalent segment increased the rate of desensitization compared with P2X 2a /X 1 R or P2X 2a /X 3 R mutant receptors with two and one negative residue, respectively. The relevance of single mutations at the charged residues to P2X 4 R desensitization was also confirmed. Both removal of negatively charged residues and introduction of positive ones resulted in a significant delay of desensitization. Furthermore, the individual residues in the 6-aa sequence of the P2X 4 R C terminus were not exactly equivalent in terms of their effects on receptor desensitization. Substitution of amino acids closer to the second transmembrane domain had a more profound impact on the pattern of [Ca 2ϩ ] i response. This domain is suggested to form a pore-lining region of P2XR (3). Since the 6-aa region of P2XR C termini is located relatively close to the cytoplasmic mouth of the pore, we may speculate that the electrical charge in the 6-aa region has an influence on the overall efficiency of conformational transition to the desensitized state.
In parallel to electrophysiological measurements (11,12), [Ca 2ϩ ] i measurements showed that heteropolymerization of P2X 2a R and P2X 3 R results in a channel sensitive to AMP-CPP, which desensitizes with a pattern different from those seen in cells expressing homopolymeric channels. The contribution of each subunit C terminus to [Ca 2ϩ ] i -signaling profile was additive in this combination of subunits. Furthermore, desensitization of the heteropolymeric channels composed of P2X 2a R and P2X 3 R subunit was delayed by the C-terminal mutation intro-duced only in the P2X 3 R subunit. Based on the putative pseudosymmetrical orientation of the channel subunits, heteropolymeric P2X 2a R ϩ P2X 3 R could have the 6-aa C-terminal region of each subunit at an equivalent position around the pore axis. Such a ring-like structure made by the C-terminalcharged residues could determine the pattern of P2XR signaling.
In contrast to the other receptor subunits, mutation at the C-terminal 6-aa sequence of P2X 7 R did not result in modulation of receptor desensitization. This is not a surprise, since the structure of P2X 7 R C terminus is distantly related to others. It is the longest C terminus among P2XRs and contains the structural basis underlying the unique feature of this channel, the cell-lytic pore formation. It also shows the least amino acid sequence similarity compared with the other P2XR C termini (10). The amino acid sequence alignment of P2XRs revealed that an 18-aa stretch of P2X 7R , with no similarity to others, is positioned proximal to the Val 392 -Pro 397 sequence, which corresponds to the Arg 371 -Pro 376 sequence of P2X 2a R (1). This insertion should effectively move the charged 6-aa residues away from the internal mouth of the pore.
A role of the C terminus in receptor signaling is not unique for P2XRs. The contribution of the C terminus to channel gating was reported for other channels, including acetylcholine-gated channels, inward rectifier potassium channels, and mechano-sensitive channels (28 -31). In the case of inwardly rectifying potassium channels, which have the same topological architecture as P2XRs, the important determinants of the pore block by Mg 2ϩ and polyamines are mediated by negatively charged residues located in the C terminus and the second transmembrane region (29). The mechanism for P2XR desensitization by the charged C-terminal small segment needs to be further examined by means of structural and biophysical studies.
In conclusion, we show here that the activation of P2XRs by ATP leads to an increase in [Ca 2ϩ ] i , the pattern of which is highly specific for each channel expressed and determined by the rate of receptor desensitization. The variable C termini of receptor subunits in part account for the observed difference in desensitization rates of homo-and heteropolymeric receptors. The charged residues in the 6-aa C-terminal sequence appear to serve as a common factor influencing the desensitization rates of P2X 3 R, P2X 2a R, P2X 2b R, and P2X 4 R. The structural element responsible for the difference in desensitization rates among P2X 2a R, P2X 2b R, and P2X 4 R is exclusively localized to the 6-aa C-terminal sequence. In the case of P2X 1 R and P2X 3 R subunits, the structure other than the C termini also participates in the control of desensitization. Finally, P2X 7 R has a distinct C terminus, as well as a distinct pattern of [Ca 2ϩ ] i signaling, and further experiments are required to establish the possible relationship between the structure of this terminus and the sustained Ca 2ϩ influx.