INTRODUCTION

Cell surface receptors for ATP can be divided into metabotropic (P2Y/P2U) and ionotropic (P2X) classes. The metabotropic class belong to the superfamily of G protein-coupled receptors with seven transmembrane segments; the ionotropic class are ligand-gated channels, currently thought to be multisubunit proteins with two transmembrane domains per subunit (for review see Ref. 1). P2Z receptors have been distinguished from other P2 receptors in three main ways (2-4). First, activation of this receptor leads not only to an inward ionic current but also to cell permeabilization. Second, 2,3-(4-benzoyl)benzoyl ATP is the most effective agonist, and ATP itself is of rather low potency. Third, responses are strongly inhibited by extracellular magnesium ions, which has been interpreted to indicate that ATP⁴ is the active agonist (for review see Ref. 5).

A seventh member of the P2X receptor family was isolated recently from a rat cDNA library that, when expressed in human embryonic kidney (HEK293) cells, exhibits these three properties (6). This receptor (rP2X₇)¹ is thus considered to represent a P2Z receptor. The protein is structurally related to other members of the P2X family; there is 35-40% amino acid identity in the region of homology, but the C terminus is 239 amino acids long in the rP2X₇ receptor compared with 27-120 amino acids in the others. The rP2X₇ receptor functions both as a channel permeable to small cations and as a cytolytic pore (6). Brief applications of ATP (1-2 s) transiently open the channel, and this is generally similar in properties to other P2X receptors. Repeated or prolonged applications of agonist cause cell permeabilization; reducing the extracellular magnesium concentration much potentiates this effect. The permeabilization involves the cytoplasmic C terminus of the protein because it does not occur with a P2X₇ receptor lacking the last 177 residues, although this truncation does not affect the function as a small cation channel.

The P2Z receptor has been implicated in lysis of antigen-presenting cells by cytotoxic T lymphocytes, in the mitogenic stimulation of human T lymphocytes, as well as in the formation of multinucleated giant cells (7-9). However, the interpretation of the physiological role of P2Z receptor has been complicated by functional differences that seem to exist between rodent and man (10). Therefore, we undertook to clone the human macrophage P2X₇ receptor (hP2X₇). We compared the functional properties of the human receptor expressed in HEK293 cells with those observed in human macrophages, and we attempted to understand some of the differences between the rat and human receptors by exchanging the C-terminal domains in a chimeric receptor.

MATERIALS AND METHODS

A 433-bp fragment of the rat P2X₇ receptor (6) was used as a probe to screen at low stringency a gt10 human monocyte cDNA library (Clontech: 1050a). Phage DNA was prepared from three positive plaques and digested by EcoRI. Fragments were cloned into EcoRI prepared pBluescript (Stratagene) and sequenced by fluorescent sequencing. These three clones encoded partial overlapping cDNAs with high sequence homology to rP2X₇. For functional expression, a clone containing the complete open reading frame of the hP2X₇ receptor was constructed by overlapping PCR using phage DNA as template. The entire coding sequence was subcloned into the NotI site of pcDNA3 (Invitrogene) using NotI sites included in the amplification primers. All PCR amplified material was confirmed by sequencing. The nucleotide sequence was confirmed by reverse transcription PCR on human mRNA from brain, spleen, and the macrophage cell line U937. The sequence was identical except for the finding of either C or T at position 499, which encodes either His or Tyr at amino acid 155 (Tyr in rP2X₇); this probably reflects allelic variation of the human P2X₇ gene because the variation was also found in genomic DNA coming from a single donor.²

Multiple tissue Northern blots (Clontech) were hybridized with a 809-bp fragment (1351-2160) generated by PCR amplification and random-primed with [-³²P]dCTP. Hybridization was at 42 °C in 50% formamide with final washes with 1 × SSC (55 °C for 20 min). Blots were exposed for 4 days at 80 °C.

A silent restriction site (NheI) was introduced at the equivalent positions in the rat and human cDNAs using the Pfu mutagenesis kit (Stratagene) (T to G at 1069 in rP2X₇; G to A at 1072 in hP2X₇). A NheI-XhoI fragment corresponding to 3 extremity of each cDNA were then excised and subcloned in the opposite plasmid (i.e. human 3 end into rat background). The resulting chimeras were h-rP2X₇ (human 1-346 and rat 347-595) and r-hP2X₇ (rat 1-346 and human 347-595) (see Fig. 1). All constructions were sequenced on their entire coding region.

HEK293 cells were transiently transfected with cDNA (1 µg/ml) and Lipofectin dissolved in Optimem (Life Technologies, Inc.) placed into Petri dishes containing four coverslips onto which cells were plated at a density of about 8 × 10⁴/coverslip. Cells were washed 5 h later, and normal Dulbecco's modified Eagle's medium was applied. Electrophysiological studies or dye uptake measurements were carried out 18-49 h later. For each set of transfections, parallel experiments were performed on HEK cells transfected with cDNA encoding both rP2X₇ and hP2X₇ receptors.

Monocyte-derived human macrophage cultures were prepared as described by Blanchard et al. (11, 12). Briefly, monocytes were isolated from leukocyte concentrates obtained from a healthy male volunteer. Leukocytes were resuspended in RPMI 1460 medium (Life Technologies, Inc.) with 20% human serum, 2 mM glutamine, 5 mM HEPES, and 100 µg/ml streptomycin. Cells were allowed to adhere to culture flasks for 1-2 h, after which nonadherent cells were washed away. Adherent cells were cultured for 7-14 days in this medium plus human interferon- (1000 units/ml). Macrophages were recovered from the culture flask by pipetting with cold phosphate-buffered saline and plated onto glass coverslips for electrophysiological and YO-PRO1 uptake experiments that were carried out 12-24 h later.

Whole cell recordings were made using the EPC9 patch-clamp amplifier and Pulse acquisition programs (HEKA, Lambrecht, Germany). Patch pipettes (5-7 M) contained (in mM) CsCl or NaCl 154, EGTA 10, HEPES 5. The normal extracellular solution contained (in mM) NaCl 147, KCl 2, CaCl₂ 2, MgCl₂ 1, HEPES 10, and glucose 12, and the "low divalent" solution had no magnesium and 0.3 mM CaCl₂. Agonists were delivered by U-tube delivery system; antagonists, when applied, were present in both superfusion and U-tube. Experiments were carried out at room temperature.

The Photonics Imaging (IDEA) system for microscopic fluorescence measurements (Photonics, Planegg, Germany) was used. Coverslips were placed on the stage of a Zeiss Axiovert 100 inverted microscope and viewed under oil immersion with a 40× Fluar objective. YO-PRO1 (Molecular Probes, Eugene OR) fluorescence was measured using 491/509 nm excitation/emission wavelengths. Images were obtained at 5-20-s intervals during continuous superfusion (2 ml/min) with YO-PRO1 (2 µM) and varying concentrations of ATP or BzATP. For each experiment, the time course of YO-PRO1 fluorescence was obtained for 10-20 individual cells (e.g. Fig. 3A) and then averaged to obtain the mean fluorescence signal. It usually was not possible to follow YO-PRO1 fluorescence in rP2X₇-expressing cells for more than 3-5 min after application of maximum concentrations of BzATP because the extensive cell lysis caused the cells to detach from the coverslip. Therefore, results were expressed as mean signal at 3 min for rP2X₇, and the signal at 10 min was used for hP2X₇ and human macrophage cells. All experiments were carried out at room temperature.

RESULTS AND DISCUSSION

Three phage with overlapping inserts were isolated from a human monocyte library by low stringency hybridization with a rP2X₇ probe. The clones spanned a region of 3076 bp encoding an open reading frame of 595 amino acids (Fig. 1). This protein is 80% identical with rP2X₇ receptor, with no particular regions of the sequence being more related than others; this identity is less than that found for the rat/human comparisons of P2X₁, P2X₃, and P2X₄ receptors (91, 94, and 88% respectively) (hP2X₁, Ref. 13; hP2X₃³; hP2X₄, Ref. 14).

The cDNA isolated from the monocyte library did not contain a poly(A)⁺ tail in the 3-untranslated region, suggesting a larger size for the mature RNA. Northern blotting detected a single band of about 6 kilobases, with strong signals in pancreas, liver, heart, and thymus, and moderate to low levels in brain, skeletal muscle, lung, placenta, leukocytes, testis, prostate, and spleen. A similar distribution was seen for the rat P2X₇ receptor (data not shown) and indicates that the P2X₇ receptor has a much more widespread distribution than previously considered on the basis of functional responses of the "P2Z" type.

Brief application (1-3 s) of ATP or BzATP on HEK293 cells transiently transfected with hP2X₇ receptors evoked inward currents (at 70 mV) (Fig. 2). The currents were linearly dependent on membrane potential (90 to 30 mV) (Fig. 2C) and were carried by cations; reversal potentials in external sodium (154 mM) or potassium (147 mM) were 1 ± 0.2 mV (n = 4) and 0.5 ± 0.05 mV (n = 3), respectively, and were not significantly altered when internal chloride was replaced with aspartate (n = 6). Removal of extracellular magnesium (and/or calcium) greatly enhanced the responses (Fig. 2A). BzATP was 10-fold more potent than ATP to activate the receptor. The currents evoked by BzATP were blocked by relatively high concentrations of suramin and PPADS; the concentrations causing half-maximal inhibition were similar to those seen in the rat (for hP2X₇, suramin 92 ± 8 µM (n = 4) and PPADS 62 ± 4 µM (n = 4) versus 300 µM BzATP, and for rP2X₇, suramin 78 ± 3 µM (n = 3) and PPADS 51 ± 4 µM (n = 3) versus 30 µM BzATP). In these respects, the properties of the human receptor resemble those of the rat.

There were also marked differences between hP2X₇ and rP2X₇ receptors. First, higher concentrations of agonists were required to activate hP2X₇ receptors. The half-maximal concentration (EC₅₀) of BzATP to activate hP2X₇ receptors was 25-fold greater than for the rP2X₇ receptor (Fig. 2H), and the ATP EC₅₀ value was about 10-fold greater. Second, removal of external magnesium increased the hP2X₇ currents to a greater degree than the rP2X₇ currents (670 ± 80% versus 420 ± 84%, p < 0.05; Fig. 2G); the concentration of magnesium that caused half-maximal inhibition of the current was significantly lower for the hP2X₇ receptor (212 µM versus 780 µM, p < 0.0001). Third, the time courses of the deactivation of the current differed. One of the unusual features of the rP2X₇ receptor is that when the extracellular concentration of divalent cations is reduced, a very prolonged current (up to 10-20 min) is induced by even a very brief agonist application (1-3 s) (6). This prolonged component has a different underlying ionic basis, for the membrane becomes permeable to large cations (such as N-methyl-D-glucamine) in addition to small cations; it becomes progressively more evident when the agonist applications are repeated (6). This behavior was much less marked in the case of the hP2X₇ receptor. Even repeated applications of BzATP (300 µM; 20-30 times for 3-s duration at 1-min intervals) in low extracellular divalent concentrations evoked currents that largely declined to baseline within 10-20 s of discontinuing the application. For the hP2X₇ receptor, the current measured at 30 s was 18 ± 2% of the peak current (n = 16), and the corresponding value for the rP2X₇ receptor was 80 ± 3% (n = 13). On the other hand, the human receptor did exhibit a slow component in the current deactivation if compared with the rP2X₂ receptor; this appeared as a "tail" in the offset of the response, typically lasted for 1-3 min, and accounted for 8-17% of the total current integral (n = 7) (Fig. 2, A-C).

BzATP or ATP evoked currents in macrophages that closely resembled those observed upon activation of heterologously expressed hP2X₇ receptors (Fig. 2, D-H). This was true for inhibition by magnesium (Fig. 2, D and G), block by suramin (Fig. 2E), reversal potential (Fig. 2F), and agonist potency (Fig. 2H).

A striking property of rP2X₇ receptor, when compared with other P2X receptors, is its ability to induce cell lysis; this results from the formation of large pores that are permeant to high molecular mass dyes such as YO-PRO1 (629 daltons). We therefore measured YO-PRO1 uptake into cells expressing hP2X₇ receptors and made comparative measurements in human macrophages, cells expressing rP2X₇ receptors, and cells expressing rP2X₂ receptors. Fig. 3A shows the time course of YO-PRO1 fluorescence from single cells during superfusion with BzATP in normal or low divalent cation concentrations. Much less YO-PRO1 was taken up by cells expressing human P2X₇ receptors and human macrophages than by cells expressing rP2X₇ receptors (Fig. 3A). On the other hand, there was significant uptake by hP2X₇-expressing cells and macrophages when these were compared with cells expressing P2X₂ receptors; this is clearly seen on the expanded scale of Fig. 3B.

YO-PRO1 uptake, measured 5 or 10 min after adding BzATP, was strongly dependent on the agonist concentration (Fig. 3C). Cells expressing rP2X₇ receptors gave a larger "maximal" response and showed YO-PRO1 uptake at much lower concentrations of BzATP (Fig. 3C). It is unlikely that the lower YO-PRO1 uptake into hP2X₇-expressing cells was due to a lower level of receptor expression than in rP2X₇-expressing cells, because the cation current induced by maximal concentrations of BzATP (30 µM for the rat and 300 µM for the human) were not different (1869 ± 286 pA versus 1777 ± 342 pA, n = 12). There is, on the other hand, quite good agreement between the concentrations of BzATP required to induce cation current measured electrophysiologically (Fig. 2H) and YO-PRO1 uptake measured by fluorescence (Fig. 3C). These results are in general agreement with previous work that indicates that higher agonist concentrations are required to induce permeabilization of human macrophages than rodent macrophages (7).

The most obvious difference between the P2X₇ receptor and other P2X receptors is the induction by agonists of an increased permeability to very large ions, including propidium dyes (6). This difference is accounted for, at least in part, by the long C-terminal domain of the P2X₇ receptor because it largely disappears in a P2X₇ receptor in which this domain is greatly truncated (at residue 418). This suggested that the propidium uptake observed for the rP2X₇ receptor, which is greater than that observed for the hP2X₇, might result from differences in this C-terminal domain. Therefore we compared the properties of cells expressing four different receptors: hP2X₇, rP2X₇, h-rP2X₇, and r-hP2X₇, where the h-r and r-h forms were chimeras with exchanged C-terminal domains. These chimeras have completely exchanged cytoplasmic C-terminal domains (248 amino acids) because the point of cross-over (residue 347) was within the second putative transmembrane domains.

In electrophysiological experiments, all four proteins expressed similar peak currents. The deactivation kinetics were largely transferred by exchange of the C-terminal domain in both directions. Thus, cells expressing h-rP2X₇ receptors showed currents in low divalent concentrations that did not decline for minutes after removal of the BzATP (Fig. 4A). The current measured at 30 s was 70 ± 4% of peak, n = 10). Conversely, r-hP2X₇ receptors gave currents that more closely resembled those of wild type hP2X₇ receptors (Fig. 4A); the current measured at 30 s was 31 ± 3% of peak, n = 10). The difference in sensitivity to BzATP between human and rat receptors was also affected, but in this case the exchange was not reciprocal. Thus, the rat C terminus on the human receptor (h-rP2X₇) increased the sensitivity to BzATP, but the human C terminus on the rat receptor (r-hP2X₇) did not reduce the sensitivity to BzATP (Fig. 4B). These experiments suggest that binding of BzATP and subsequent conformational changes leading to channel opening involve concerted conformational changes in both domains of the molecule.

Measurements of YO-PRO1 uptake in cells expressing the chimeric receptors indicated that provision of the rat C-terminal domain to the human receptor (h-rP2X₇) significantly increased permeability to this large cation (Fig. 4C; measured 5 min after adding a maximal concentration on BzATP). On the other hand, substituting the human C terminus onto the rat receptor (r-hP2X₇) did not significantly reduce YO-PRO1 uptake as compared with that seen with wild type rat receptor. The rat receptor truncated at residue 418 (P2X₇C) showed very little YO-PRO1 uptake, even compared with the wild type hP2X₇ receptor, in agreement with our earlier finding (Fig. 4C) (6).

The present experiments have shown three main functional differences between the rat and human P2X₇ receptors. The first is the lower sensitivity to agonists, notably BzATP, of the human receptor. One might have expected that this agonist sensitivity would be determined by the presumed extracellular loop of the receptor, but such a simple interpretation is not tenable in view of the finding that the human receptor with the rat cytoplasmic C-terminal domain is as sensitive as the wild type rat receptor (Fig. 4B). The second difference relates to the time for which the inward current flows following a brief application of agonist; the greatly prolonged currents observed in the rat P2X₇ receptor, which distinguishes it dramatically from other P2X receptors (6), were much less obvious in the case of the human P2X₇ receptor (Figs. 2, A-C, and 4A). The different rate at which the channel closes after removal of the agonist (deactivation) was largely transferred by exchange of the C-terminal domain, suggesting that this cytoplasmic part of the molecule is a determinant of channel closing. We have shown previously that the prolonged component of the current in rP2X₇ receptors is associated with an increased conductance to large cations such as N-methyl-D-glucamine (6); if an initially small channel dilates into a larger pore, then this result implies that the C-terminal domain is a determinant of the rate of dilatation.

The third large difference between species was the uptake of the propidium dye, YO-PRO1. We did not observe complete transfer of this phenotype with the exchange of C-terminal domains, although one might have expected such a result if the slow deactivation kinetics and the larger YO-PRO1 uptake are both related to formation of a large pore. However, the YO-PRO1 uptake reflects the cell permeabilization during several minutes of the continued presence of BzATP, whereas the deactivation kinetics reflect the closure of ion conducting channels that have been opened by a very brief application of BzATP. Clearly much further work will be required to determine the relation between these two properties of the P2X₇ receptor. The experiments have also shown that in all the respects examined, the human P2X₇ receptor cloned from monocytes corresponds in its properties to the ATP receptor of the human macrophage. This would be consistent with the macrophage receptor assembling as a homomultimer of P2X₇ subunits.