Intersubunit physical couplings fostered by the left flipper domain facilitate channel opening of P2X4 receptors

P2X receptors are ATP-gated trimeric channels with important roles in diverse pathophysiological functions. A detailed understanding of the mechanism underlying the gating process of these receptors is thus fundamentally important and may open new therapeutic avenues. The left flipper (LF) domain of the P2X receptors is a flexible loop structure, and its coordinated motions together with the dorsal fin (DF) domain are crucial for the channel gating of the P2X receptors. However, the mechanism underlying the crucial role of the LF domain in the channel gating remains obscure. Here, we propose that the ATP-induced allosteric changes of the LF domain enable it to foster intersubunit physical couplings among the DF and two lower body domains, which are pivotal for the channel gating of P2X4 receptors. Metadynamics analysis indicated that these newly established intersubunit couplings correlate well with the ATP-bound open state of the receptors. Moreover, weakening or strengthening these physical interactions with engineered intersubunit metal bridges remarkably decreased or increased the open probability of the receptors, respectively. Further disulfide cross-linking and covalent modification confirmed that the intersubunit physical couplings among the DF and two lower body domains fostered by the LF domain at the open state act as an integrated structural element that is stringently required for the channel gating of P2X4 receptors. Our observations provide new mechanistic insights into P2X receptor activation and will stimulate development of new allosteric modulators of P2X receptors.

P2X receptors are trimeric membrane ion channels (1,2) activated by extracellular ATP (3). So far, seven P2X subtypes (P2X1-P2X7) have been identified, which are expressed virtu-ally in almost all mammalian tissues, including nervous, immune, and cardiovascular systems (3)(4)(5). P2X receptors possess a molecular architecture distinct from other ion channel protein families (4,6,7) and are implicated in a wide range of physiological and pathological processes (2,4,5,8), such as neuroinflammation, synaptic transmission, primary afferent signaling, chronic pain, central control of respiration, vascular remodeling, and the regulation of blood pressure. Accordingly, P2X receptors hold great interest as new therapeutic targets for inflammation and v cardiovascular and neurological diseases (8 -14). For this purpose, it is essential to fully understand the detailed gating mechanism of P2X receptors at the atomic level (9,15,16).
High resolution X-ray structures at apo/closed and ATPbound open states are available for zebrafish P2X4 (zfP2X4) (7,9), human P2X3 (hP2X3) (17), Amblyomma maculatum P2X (AmP2X) (18), and panda P2X7 (19) receptors, which greatly aid our understanding of the working principles of those unique receptors, such as the ligand recognition, pore architecture, and conformational changes associated with the channel activation. Based on these structures, optical control of P2X receptors independent of natural stimulus has been achieved via powerful optogating approaches (20 -22). The structural comparison of the closed and open structures suggests a possible gating mechanism of P2X receptors (Fig. 1, A and B) (9,16,23,24). First, at the ATP-binding site, ATP promotes the jaw closure between the head and dorsal fin (DF) 3 domains, making the DF domain move upward to the head domain to accommodate ATP. Meanwhile, the bound ATP pushes the left flipper (LF) domain out of the ATP-binding site. Second, because both the LF and DF domains are structurally coupled with the lower body domain, the movements of those two domains lead to a concomitant outward flexing of lower body domains in the open state, which markedly expands the central vestibule (Fig. 1B). Finally, lower body domains are directly coupled with two transmembrane (TM) domains TM1 and TM2, and therefore their outward flexing can directly promote the opening of ion channel pore by causing the TM helices to expand in an iris-like motion (Fig. 1B). Recent studies have partially confirmed this hypothesis (16,23). Multiple approaches, including voltageclamp fluorometry (25,26), fast-scanning atomic force microscopy (27), electron microscopy (28), engineering metal bridge (29 -32), substituted cysteine accessibility (33), normal mode analysis, and molecular dynamics (MD) simulations (30, 32, 34 -37), have demonstrated that the dynamics of the head domain (25,27,28,30,35), tightening the ATP-binding site jaw (30,35), and the expansion of upward (27) and central (33,38) vestibules are essential for the ATP recognition, ion permeation, channel activation, desensitization, and sustained activation of P2X receptors. However, more pronounced allosteric changes of the extracellular domain (26 -28) were observed in biochemical, biophysical, and computational analysis than those observed in the open structure of the P2X4 receptor. Recent studies (22,31) also suggested that the TM region was distorted due to the unexpected absence of intersubunit interactions in the X-ray open structure of zfP2X4 receptor. Thus, although the apo/closed and ATP-bound open X-ray structures of the zfP2X4 receptor have provided a blueprint for the mechanism of P2X activation, the detailed conformational transition during channel gating requires further exploration (18,22,24,31,32,37,39,40).
The proposed gating mechanism based on the open and closed X-ray structures highlights that the repelling action of ATP on the LF domain would favor the outward flexing of lower body domains and concomitant pore dilation (9). The reason why the motion of the LF domain favors this outward flexing of the lower body domains is yet unknown. Recently, it has been demonstrated that ATP binding-induced alteration in interdomain hydrophobic interactions and the concomitant relative motions between the LF and DF domains are indispensable allosteric events for channel activation of P2X4 receptors (32), although the underlying mechanism of this process remains undetermined. One possibility is that the hydrophobic interactions between the LF and DF domains at the resting state build up an energy barrier that prevents the activation of P2X receptors (32). The expelling of the LF domain from the ATP-binding pocket might help to overcome this energy barrier and favor the channel activation of P2X4. If so, bound ATP would finally soothe this loop structure to reduce the energy barrier during the channel opening. However, more inter-and intrasubunit contacts among the LF, DF, and lower domains were established after bound ATP pushed the LF domain out of the ATPbinding pocket (details see below and Ref. 9), implying a less flexible LF domain at the open state was developed after the repulsion. Thus, previous studies and structural models (9,32) only partially elucidate the function of the LF domain during channel gating. What exactly is the reason behind the bound ATP-induced repulsion of this flexible loop out of the ATPbinding pocket? Is it a passive allosteric change for only adapting the LF domain to accommodate the allostery of other domains or a more important allostery making a major contribution to couple the ATP binding to the final pore opening? Getting a clear understating of this allostery will provide new mechanical insights into the gating process of P2X4 receptors.
Using multidisciplinary approaches, we proposed that ATPbound induced conformational changes of the LF domain enable it to establish intersubunit physical couplings among the DF and two lower body domains, which are essential for the channel opening of P2X4 receptors. This observation will enrich our understandings of the role of the LF domain in channel gating and provide new mechanistic insights into the channel activation of P2X receptors.

Intersubunit physical contacts among the lower body and DF domains established by the LF domain are essential for the channel function of P2X4 receptors
Revealed by the homology models of rat P2X4 (rP2X4) receptors (32,37) built from zfP2X4 X-ray structures of the apo and open states (9), the LF domain is a loop structure surrounded by the head, DF, right flipper, and lower body domains ( Fig. 2A). Its N and C termini are covalently coupled with the ␤12 and ␤13 sheets of the lower body domain (Fig. 2, B and C), respectively. Alanine-scanning mutagenesis of all residues of the LF domain (Fig. 2, D and E), ranging from Arg-277 to Tyr-292 of rP2X4, indicated that the residues in the N and C termini rather than  (Fig. 3, B and C), indicating a crucial role of these residues in both rP2X4 and zfP2X4.
A comparison of the LF domain at the open and closed states based on the homology models of rP2X4 revealed that all mutants impairing the maximal current amplitude virtually contribute to the newly established intersubunit physical couplings among the two lower body and DF domains after the LF domain was repelled out of the ATP-binding site (Fig. 2, B and C). Arg-278 and Asp-280 foster an intrasubunit salt bridge in both the closed and open structures of P2X4, whereas an additional hydrogen bond (H-bond) was formed between the side chain of Arg-282 and the oxygen atom of the main chain of Arg-278 at the open state (Fig. 2B, lower panel). ATP binding also contributes to the formation of the H-bond between the side chain of Asn-192 (located in the lower body domain of another subunit) and the main chain atom of Arg-282 (Fig. 2B, lower panel). N192A mutant partially reduced the maximal current amplitude of P2X4 (Fig. 2, D and E). Additionally, the atoms of the main chain of Val-288, Ser-289, and Pro-290 developed new contacts with the side chain of Arg-203 (in the lower body domain of another subunit) (Fig. 2C, lower panel).
Alanine substitution of Arg-203 significantly impaired channel activation of rP2X4 receptors (Fig. 2, D and E). Mutations on the identical residue (Arg-206) in zfP2X4 significantly reduced ATP (1 mM)-induced currents (Fig. 3, B and C), suggesting this newly established contact is also crucial for the channel function of zfP2X4 receptors. The intersubunit hydrophobic contacts among Val-288, Leu-214, and Ile-205 were significantly changed after ATP binding ( Fig. 2C) (32). Mutations on Val-288, Leu-214, Ile-205, and the identical residues of zfP2X4 significantly reduced maximal current densities of both rP2X4 and zfP2X4 receptors (32). In contrast, although Glu-245 forms intersubunit contact with Arg-282 via H-bonds to tighten the LF domain at the resting state (Fig. 2B, upper panel), E245A and E245R had no effect on the maximal current amplitude and the EC 50 of ATP (the concentration of ATP yielding current that is half of the maximum) of rP2X4 (see below), indicating that this contact is redundant for the channel function. Thus, ATP binding-induced conformational changes of the LF domain and the following intersubunit physical couplings among the DF and two lower body domains fostered by the deformed LF domain at the open state are pivotal for the channel function of both rP2X4 and zfP2X4 receptors.
Because of the low potency of ATP on the zfP2X4 receptor expressing in mammalian cells (7,32,37), the following mutagenesis, protein-expression measurements, and electrophysiological recordings were carried out on rP2X4 receptors.

Physical couplings fostered by left flipper domain
Additionally, although the essential role of the residues involved in forming intersubunit physical couplings after ATP binding was conserved between rP2X4 and zfP2X4 receptors (Fig. 3, A and D) (32), some residues of the middle region of the LF domain are not exactly the same (Fig. 3D). Thus, to get a more appropriate prediction about the functional studies of the rP2X4 receptors, the following free-energy profile reconstructions and MD simulations were also based on the homology models of rP2X4 rather than on the crystal structures of the zfP2X4 receptor.

Impaired intersubunit physical couplings significantly influence channel gating rather than channel assembly, protein stability, and ATP-EC 50 of P2X4 receptors
Multiple factors can influence the maximal current densities of P2X receptors. The fact that the surface expression levels of these loss-of-function mutants exhibited no pronounced changes when compared with wild-type (WT) rP2X4 (Fig. 4, A and B) suggests that the decreased ATP currents were not acting on channel expression and trafficking, except for R278A and D280A. The complete abolishment of the ATP-induced currents in rP2X4 R278A and rP2X4 D280A (Fig. 2, D and E) may partially attribute to a decreased surface expression of those two mutants. Another possibility is that the LF domain and the established intersubunit physical couplings may structurally support the trimeric assembly or prevent the neighboring subunit from clashing with each other upon ATP binding. We introduced additional mutations, R203A and R282A, into mutant V288C/T211C, which could form trimeric receptors in non-reduced SDS-PAGE, due to the formation of intersubunit disulfide bonds as we previously demonstrated (32). Trimeric bands of mutants, V288C/T211C/R203A and V288C/T211C/ R282A, were observed in the non-reducing Western blotting (Fig. 4C), indicating that the impaired intersubunit physical couplings did not render a significant deficiency in the channel assembly of P2X4 receptors. Additionally, we incubated the transfected HEK-293 cells with cycloheximide (CHX, 20 g/ml), an inhibitor of protein biosynthesis (41), in time-scale experiments. In contrast to the profound reduction in maximum currents, after incubation with CHX for 10 h, the protein level of the mutants made no significant changes (Fig. 4, D and E), indicating that alterations in those positions did not render channel instability of P2X4 receptors. Moreover, in contrast to the profound reduction in maximum currents (Fig. 4, F and G), little to no change was observed in the surface expression (  (Fig. 4J). Thus, the decreased maximal current densities in those mutants were not related to surface protein expression, channel assembly, protein stability, or EC 50 of ATP.
To gain more structural information about the role of those newly established intersubunit physical couplings in the channel function of P2X4 receptors, the free-energy profiles for those interactions were reconstructed by metadynamics (42)(43)(44), "a powerful algorithm that can be used for both reconstructing the free energy and accelerating rare events in systems described by complex Hamiltonians" (45)

Physical couplings fostered by left flipper domain
ble CV similar to that of the resting state (CV R ) than that of open state (CV O ), indicating that the LF domain-mediated establishment of intersubunit physical couplings is an allosteric change correlated well with the ATP-bound open state. Because these mutations had little to no changes in the EC 50 of ATP, the protein expression and channel assembly, despite profound reduction in maximum currents (Fig. 4), they may impair the channel gating of P2X4 receptors.
To test this hypothesis, we performed single channel recordings in the outside-out configuration. The control experiment showed that the main conductance state of channels opened by saturated ATP (100 M) in patches excised from HEK-293 cells expressing WT rP2X4 (Fig. 5D) was similar to that shown in previous reports (46 -48). The channel with a unitary conductance of ϳ10 pS at Ϫ120 mV was observed in the most of the patches. Channels opened and closed frequently and were highly flickery, and thus the precise determination of the open and shut time could not be made. Additionally, unitary currents could only be observed at the first 5 s and disappeared after ϳ10 s (

Physical couplings fostered by left flipper domain
currents could be observed both in rP2X4 R203A and rP2X4 R282A (Fig. 5G). Similar to a previous finding (48), IVM exhibited a small effect on the unitary current amplitude (Fig. 5, G and H) of rP2X4 R203A , rP2X4 R282A , and WT and significantly prolonged the open time of channels (Fig. 5G). Perhaps this is because IVM directly acts on the interface between transmembrane domains TM1 and TM2 (50), the only entrance of various ions. The generation of unitary currents of rP2X4 R203A and rP2X4 R282A in the patches exposed to IVM and ATP further supported the idea that the mutated channels with impaired intersubunit physical couplings fostered by the LF domain behave with normal surface expression, channel assembly, and the EC 50 of ATP. It has been well established that IVM is able to significantly increase the channel opening probability (P o ) of P2X4 receptors (48). Thus, the absence of unitary P2X4 currents in mutated channels exposed to saturated ATP and the significant channel activity observed in the presence of both ATP and IVM suggested that the impairment in intersubunit physical couplings significantly decreased the open probability of P2X4 receptors.

Engineered intersubunit metal bridges that change the intersubunit physical couplings remarkably influence the channel gating of P2X4 receptors
To further examine this idea, we applied different metal bridges to weaken or strengthen the physical couplings among the DF and two lower body domains fostered by the LF domain at the open state. We have recently shown that an intradomain Zn 2ϩ bridge in the LF domain produces an unexpected inhibition (32) on the current of mutant channel rP2X4 His-286/V288H without identifying the underlying mechanism. Here, the structural model of rP2X4  in the open state suggested that Zn 2ϩ might form an intersubunit rather than an intrasubunit Zn 2ϩ bridge with the introduced histidine residue V288H and natural histidine residue His-286 of one subunit, as well as with the main chain oxygen atoms of Pro-207 and Ile-209 of another subunit (Fig. 6A). Indeed, post-application of Zn 2ϩ after ATP markedly reduced the remaining currents of rP2X4 His-286/V288H (ratio ϭ 0.26 Ϯ 0.06, n ϭ 4, green arrow, Fig.  6, B and C). This blockage was specific to the presence of histidine at both 288 and 286 positions because Zn 2ϩ application only caused mild inhibition on the single histidine mutant H286A/V288H as well as WT P2X4 (His-286/Val-288) receptor (Fig. 6C). Introduction of the additional mutation into His-286/V288H (His-286/V288H/P207A and His-286/ V288H/I209A) affected both the inhibition (Fig. 6C) and the dose-response curve of Zn 2ϩ (Fig. 6D) (Fig. 6A, right panel). This idea is further supported by the measurement of the angle between three C␣ atoms of residues His-288, Arg-278, and Ile-205 of P2X4 V288H/His-286 (Fig. 6, A and E). The C␣ atoms of residues Arg-278 and Ile-205 align horizontally, whereas the C␣ atoms of His-288 and Arg-278 form an angle to Arg-278 and Ile-205, suggesting a tilted shape of the LF domain (Fig. 6, A and E). The tilting angle of the LF domain along the horizontal axis in P2X4 His-286/V288H

Physical couplings fostered by left flipper domain
ranged from 23-25° (Fig. 6E) during MD simulations, a value smaller than that of WT rP2X4 (25-28°) (Fig. 6E), indicating that the C terminus moved downwards and became closer to the flexible loop ␤9-␣3 than WT rP2X4 (Fig. 6A). Because the N terminus of the LF domain was located in the rigid lower body domain (␤12) of another subunit (Fig. 6A), the C-terminal movement toward the flexible loop ␤9-␣3 significantly weakened intersubunit physical couplings between the DF and two lower body domains fostered by the deformed LF domain after ATP binding. To prevent this movement, we introduced an additional substitution in loop ␤9-␣3 (His-286/V288H/I209C) to push the C terminus away from the flexible loop ␤9-␣3 through an additional coordination bond between Zn 2ϩ and the free thiol group of I209C (Fig. 6F), and to regain contacts between the C terminus and rigid the ␣3 helix of the DF domain (Fig. 6F). This additional replacement on Ile-209 by cysteine significantly increased the angle formed by C␣ atoms of residues His-288, Arg-278, and Ile-205 of rP2X4 His-286/V288H/I209C during MD simulations (Fig. 6, E and F), indicating that the LF domain had escaped from the flexible loop ␤9-␣3 and produced more contacts with the rigid ␣3 helix of the DF domain (Fig. 6F,  right lower panel).
Indeed, the additional application of Zn 2ϩ after ATP led to a remarkable potentiation (ratio ϭ 3.55 Ϯ 0.36, n ϭ 12) rather than an inhibition on the remaining current of His-286/ V288H/I209C (right trace of Fig. 6, B, green arrow, and C, green column), revealing the pivotal role of intersubunit physical couplings between the DF and two lower body domains established by the LF domain in the open state. This potentiation requires the presence of both histidine residues His-288 and His-286 in the LF domain of one subunit, and the Cys-209 located in the interface between the DF and lower body domains of another subunit, because Zn 2ϩ application only slightly inhibited rather than potentiated the currents of H286A/V288H/I209C, His-286/Val-288/I209C, and H286A/Val-288/I209C (Fig. 6C).
The effect of Zn 2ϩ on the unitary rP2X4 His-286/V288H/I209C currents was also measured in the outside-out configuration of patch clamp. There were no channel openings resembling unitary rP2X4 currents in the presence of saturated ATP (Fig. 7A), although currents with a unitary conductance of ϳ10 pS at Ϫ120 mV were evoked when ATP and Zn 2ϩ were co-applied (Fig. 7A). In additional macroscopic recordings, a few unitary P2X4 currents (ϳ10 pS) were observed when only 100 M ATP was applied (Fig. 7B); however, a following co-application of ATP and Zn 2ϩ evoked a large current (ϳ30 channels simultaneously opening) that rapidly declined to a steady-state level, where individual openings and closings can be measured, indicating that the engineered metal bridge rendered a significant increase in the open probability of rP2X4 receptors but had no remarkable effects on the rP2X4 desensitization. The unitary current conductance was similar before (ϳ10 pS) and after Zn 2ϩ treatment (ϳ10 pS, Fig. 7, A and the lower panel of B), suggesting these physical couplings had no effect on the unitary current conductance of rP2X4 receptors. Thus, physical couplings among the DF and two lower body domains fostered by the LF domain increase the open probability rather than change the current unitary conductance and channel desensitization of rP2X4 receptors.

Restraining the LF domain from fostering physical couplings via intersubunit disulfide cross-linking impairs channel activation of P2X4 receptors
To further examine the contribution of the intersubunit physical couplings fostered by the deformed LF domain during the channel gating of rP2X4 receptors, we immobilized the LF domain at the resting state by introducing cysteine residues that form the intersubunit disulfide bridge (Fig. 8A). The  (Fig. 8, B and C) were longer than that of a disulfide pair (Ͻ5 Å) (51). Therefore, the interdomain disulfide bond at those positions could immobilize the LF domain at the resting state and restrain the LF domain from fostering intersubunit physical coupling at the open state (Fig.   Figure 7. Effect of Zn 2؉ on the unitary rP2X4 His-286/V288H/I209C currents. A and B, representative current recordings from excised outside-out membrane expressing single channel (A) or multiple channels (B) exposed to ATP and the following ATP-Zn 2ϩ co-application for the mutant His-286/V288H/I209C. Full opening (O) and closing (C) are indicated by black and yellow lines, respectively. y axis denotes the ratio of the number of events to the number of bins (the bin number is set to 320). Similar results were obtained in at least three other independent recordings. 8A). The WT rP2X4 and single mutants S201C, D283C, and L284C migrated on SDS-polyacrylamide gels predominantly at a position expected for monomeric form (ϳ57 kDa; Fig. 8D). In contrast, no obvious monomeric form was observed for the subunits containing cysteine at both positions (S201C/D283C and S201C/L284C). The observed higher molecular weights presumably represent disulfide bond trimer, because ␤-mercaptoethanol (␤-ME, 1%) reduced those to a monomeric size (Fig. 8D). Although ␤-ME caused a modest shift of WT and single cysteine replacement mutants as we previously reported because of its effects on the large number of native cysteine residues (32), it should still be reasonable to conclude that interdomain/intersubunit disulfide bonds are actually formed between the LF and lower body domains in S201C/D283C and S201C/L284C.

Physical couplings fostered by left flipper domain
ϳ4 -5-and ϳ2-3-fold (ratio ϭ 4.49 Ϯ 0.67, n ϭ 3, p ϭ 0.003 and 2.19 Ϯ 0.39, n ϭ 4, p ϭ 0.006), respectively (Fig. 8, E, G, and  H), which were reversed by applications of H 2 O 2 (Fig. 8E), suggesting that immobilization of the LF domain led to impaired channel activation of P2X4. In contrast, DTT slightly reduced ATP-evoked currents of the WT rP2X4 receptors, which was reversed by H 2 O 2 (Fig. 8F). All the results suggested that the breaking of disulfide bonds is responsible for DTT-induced increases in current amplitudes of S201C/D283C and S201C/ L284C. Additionally, increasing ATP concentration (1 mM) had no effect on DTT-induced potentiation efficacy on rP2X4 S201C/D283C and rP2X4 S201C/L284C currents (Fig. 8 Fig. 8C) at the open state. Therefore, a rebuilt Zn 2ϩ bridge (Cys-201…Zn 2ϩ …Cys-283, Fig. 9A; C ␤ -C ␤ distance of Cys-201 and Cys-283 ϭ 5.2 Å) after DTT application on P2X4 S201C/D283C could perturb the conformation of the middle region of the LF domain. Indeed, post-administration of Zn 2ϩ after ATP inhibited 51.7 Ϯ 2.3% of the remaining ATP currents of P2X4 S201C/D283C but not WT P2X4 (n ϭ 9, p ϭ 0.0006, Fig. 9, B and C) under the condition that the disulfide bond has been previously interrupted by DTT at the resting state. This point was further tested by a measurement of state-dependent cross-linking of rP2X4 S201C/D283C . After DTT breaking, the cross-linking between S201C and D283C rebuilt more quickly after rP2X4 S201C/D283C was treated by ATP, when it is compared with channels without ATP treatment (Fig. 9D). Alanine-scanning mutations on the LF domain have demonstrated that the residues of the middle region (Figs.  2, D and E, and 10A) were not as crucial as the residues in the N-and C-terminal regions in the channel activation of P2X4 receptors. However, an immobilization of the LF domain using disulfide cross-linking or slightly shortening the pair-residue distance (from 7.1 to 5.2 Å) between the middle region of the LF and lower body domains using a metal bridge rendered a significantly impaired channel activation of rP2X4 receptors. Therefore, intersubunit physical couplings among the DF and two lower body domains fostered by the LF domain at the open state act as a whole structural element that is stringently required by the channel opening of P2X4 receptors, and any impairment in its integrity will lead to an impaired channel activation.
Moreover, deletions of a single amino acid (⌬283, ⌬284, ⌬285, ⌬286, and ⌬287), two amino acids (⌬283-284, ⌬284 -285, ⌬285-286, and ⌬286 -287), and even three amino acids (⌬283-285, ⌬284 -286, and ⌬285-287) in the middle region produced little change on the maximal current of P2X4 receptors (Fig. 10, B and C). However, truncating four amino acids (⌬283-286) fully abolished the channel activation of P2X4 (n ϭ 8, Fig. 10, B and C), without changing the channel expression and trafficking of P2X4 receptors (Fig. 10, D and E). Thus, a proper length of the middle region (at least two residues remained among those of five amino acids) is prerequisite for the LF domain being functional, further confirming that the middle region is a structural element required by the P2X4 receptors at the open state (see "Discussion").

Weakening intersubunit physical couplings via covalent modifications impairs the channel activation of P2X4 receptors
Finally, we interrupted the conformation of the N terminus of the LF domain through covalent modifications to partially weaken these intersubunit physical couplings at the open state (Fig. 11A). Upon channel activation, the intrasubunit salt bridge between Arg-278 and Asp-280 as well as the two H-bonds (Arg-282…Arg-278 and Arg-282…Asn-192) formed an "interaction network" that was crucial for the deformation of the LF domain at the open state (Fig. 2B, lower panel). Trp-194 is located adjacent to those four amino acids (Fig. 11A and the lower panel of Fig. 2B). We substituted Trp-194 with cysteine with the intention to prevent the interactions between Arg-278, Asp-280, and Arg-282 via covalent modifications (Fig. 11A). Using a cysteine modification technique such as methanethiosulfonate (33), Ellman's (53), and alkylating reagents (54) (Fig. 11B), we introduced charged groups (the aminoethyl group of 2-aminoethyl methanethiosulfonate (MTSEA) and sulfonatoethyl group of 2-sulfonatoethyl methanethiosulfonate (MTSES)), a bulky group (1-phenylpyrrolidine-2,5-dione group of N-phenylmaleimide (NPM)), and both charged and bulky groups  (Fig. 11, C and D), which were partially or fully reversed by DTT treatments (Fig. 11C), except NPM treatments, because the alkylating reaction is irreversible. In contrast, DTNB, MTSEA, MTSES, and NPM had no effects on the current amplitude of P2X4 S201C (Fig. 11, C and D), a mutant with a cysteine at 201 that stays slightly away from Trp-194, Arg-278, Asp-280, and Arg-282 (Figs. 11A and the lower panel of 2B). In addition, MTSEA, MTSES, DTNB, and NPM did not change the current amplitude of WT P2X4 (Fig. 11, C and D). Thus, specific covalent modifications at W194C are responsible for their inhibitory effects on the P2X4 W194C .

Physical couplings fostered by left flipper domain
Similarly, covalent modification may also interrupt both conformations of the LF domain at resting and open states. To provide direct evidence that covalent modification could prevent the LF domain from fostering intersubunit physical couplings at the open state, MTSEA (1 mM) was applied after ATP to perturb the conformation of the N terminus of the LF domain at the open state. Post-administration of MTSEA inhibited 55 Ϯ 7.2% of the remaining ATP currents of rP2X4 W194C but only slightly or had no effects on rP2X4 S201C and WT rP2X4 (Fig. 11, E and F, n ϭ 3). Thus, the specific covalent modification-induced inhibitory effects may be due to perturbing the "interaction networks" among Arg-278, Asp-280, and Arg-282 at the open state, at least partially, which is essential for the formation of intersubunit physical couplings at the open state. Therefore, any conformational changes of the LF domain that interrupt the formation and stability of intersubunit physical couplings at the open state will affect the channel gating of P2X4 receptors.

Discussion
Here, we propose that the LF domain, a flexible loop structure, underwent allosteric changes after ATP binding, which promotes the formation of intersubunit physical couplings among the two lower body domains and the DF domain that facilitates opening of P2X4 receptors (Fig. 12). At the open state, bound ATP repels the LF domain out of the ATP-binding site, which coordinates the LF, DF, and two lower body domains into an integrated structural element. The integration of the LF domain in the cleft among those domains was achieved through the salt-bridge Arg-278…Asp-280, the newly formed hydrogen-bonding contacts Arg-282…Arg-278, Asn-192…Arg-282, Val-288…Arg-203, and Ser-289…Arg-203, and new hydrophobic interactions among Val-288, Ile-205, and Leu-214 after ATP binding (Fig. 12). The following evidence demonstrated that this new integrated structural element is stringently required by

Physical couplings fostered by left flipper domain
the channel gating of the P2X4 receptors. First, mutations of the residues involved in the establishment of physical couplings significantly reduced ATP currents of P2X4 receptors. The gel analysis has excluded the possibility that this newly integrated structural element made contributions to the channel stability and channel assembly. The absence of unitary current in R203A and R282A in response to saturated ATP and the "gain-of-function" of channel activity in these two mutants when IVM was co-applied with ATP confirmed the crucial role of intersubunit physical couplings fostered by the LF domain in the channel gating of P2X4 receptors. Covalent modifications on W194C-induced weakened channel activations of rP2X4 further supported this idea. Second, a slight alteration in the interdomain interactions among the LF, DF, and lower body domains, for example, introducing different engineered metal bridges between the LF and DF domains, switched the mutant rP2X4 receptors from "loss-of-function" (His-286/ V288H) to "gain-of-function" (His-286/V288H/I209C) channels. Third, although alanine-screening mutagenesis revealed that the middle region of the LF domain exerts less influence in the channel activation, introducing intersubunit/interdomain disulfide cross-linking or metal bridge between the LF and lower body domains significantly affected the channel activation of rP2X4, suggesting that proper intersubunit/interdomain contacts, even in a region regarded as not so important, is still stringently required by the integrated structural element at the open state. Thus, bound ATP-induced repulsion of this flexible loop from the ATP-binding pocket is not a "passive"/ inessential allosteric change that just accommodates itself to the allostery of the DF domain, but is an essential process that renders the establishment of new intersubunit physical couplings among the DF and two lower body domains, which integrates all those domains into a structural element that is stringently required by channel gating of rP2X4 receptors.
Still, how the conformational changes of the LF domain together with the DF and two lower body domains affect the gating transition of P2X4 receptors needs to be explained. The deformed LF domain after ATP binding is sloped in a direction along the horizontal axis, making it possible to contact with both the rigid lower body and DF domains rather than the flexible loop ␤9-␣3 (Fig. 6A), and thus possesses an ability in strengthening the physical couplings between the DF and two lower body domains (Fig. 12). Additionally, because mutations of W194A, W194E, W194R, E245A, and E245R (Figs. 4G and 11, G and H) and covalent modifications on S201C (Fig. 11, C-F) only produced small or no changes in the activation of the rP2X4, there should be a certain distance between the middle region of the LF domain of one subunit and the lower body domain of another subunit. This distance can avoid additional contacts between the middle region of the LF domain and the lower body domain during the outward moving of two lower body domains at the open state (Fig. 2, A and B). All of those features made the LF domain fit for coordinating both outward flexing of two domains and upward motion of the DF domain (Fig. 1, A and B). Because the DF domain is structurally coupled to the lower body domain through loop ␤9-␣3 (Figs. 1, A and B,  and 2, A-C), its upward motion evoked by ATP binding will cause the outward flexing of the lower body domains. Three lower body domains form the big central vestibule (9) of P2X receptors (Fig. 1, A and B), and the outward flexing of lower body domains leads to the expansion (9) of this big central vestibule (Fig. 1, A and B). It is worth noting that the expansion of the central vestibule is crucial for the channel gating of trimeric ion channels (1,2,38), such as P2X receptors and ASIC channels. Small molecules (53,55), toxins (56), and covalent modifications (53) acting on the residues of this region can directly affect the channel activation of ASIC channels. The outward flexing of the lower body domains and the expansion of the central vestibule might facilitate the channel activation of P2X receptors by the following reason. The central rigid lower body domains are structurally coupled with TM domains and the pore region (Fig. 1B). Therefore, the deflection of the LF domain evoked by bound ATP directly causes the motions of TM region through those rigid lower body domains, which may facilitate the gating transition from the resting state to the open state (Fig. 1A). Thus, the established physical couplings between the DF and two lower body domains by the deformed LF domain are pivotal to the outward flexing of lower body domains and the concomitant pore dilation of P2X4 receptors.
Our data also showed that the flexible middle region of the LF domain, consisting of seemingly negligible residues, affected channel gating of P2X4 receptors. We have recently suggested that hydrophobic interactions between Val-288, Ile-205, Leu-214, and the aliphatic chain of Lys-190 in P2X4 receptors may develop an energy barrier for the channel gating (32). As revealed by changes in both apparent affinity and maximal cur-

Physical couplings fostered by left flipper domain
rent before and after DTT application on disulfide cross-linking in mutant P2X4 V288C/T211C , bound ATP-induced repelling action on Val-288 from the ATP-binding site may behave with two main functions. One is to reduce the energy barrier for channel gating, and the other is to accommodate ATP molecules. The N terminus of the LF domain is relative rigid because of its structural coupling with the lower body domain and the existence of a salt bridge between Arg-278 and Asp-280 in this region; thus, only a flexible middle region can buffer the repelling action of ATP on Val-288. Following the expulsion of Val-288 from the ATP-binding site and structural rearrangements of Val-288, Ile-205, Leu-214, and Lys-190, a lot of new intraand intersubunit contacts were established, including Arg-282. At this stage, the middle region located on the interface between two "interaction clusters," Asp-280…Arg-278…Arg-282 and Val-288…Arg-203…Ser-289, may act as a linker to stabilize those two "clusters" and maintains the intersubunit physical couplings between lower body domains and the DF domain at the open state. Thus, the middle region may contribute to the flexibility at the resting state to buffer the repelling action of ATP on Val-288 and provide the proper length between two termini ( Fig. 10) that facilitate the LF domain to foster proper intersubunit physical couplings among the two lower body and DF domains at the open state.
Finally, despite the indispensable role of the LF domain in channel activation of P2X4, we cannot neglect the fact that the sequence of the LF domain is not conserved among various subtypes of P2X receptors (Fig. 3D). At the open state, Arg-203 contacts with the main chain atoms of Val-288 and Ser-289, contributing to the establishment of intersubunit physical couplings. However, the arginine is replaced by glycine at the identical position of P2X2 and P2X3 subtypes (Fig. 3D). Similarly, the salt bridge formed by Arg-278 and Asp-280 in P2X4 is also absent in the P2X1 subtype (Fig. 3D). The identical residue of Arg-282 in P2X4 is absent in the P2X1 subtype (Fig. 3D). The sequence variation is much greater in the middle region of the LF domain throughout the P2X receptor family, which is completely absent in P2X6. However, this absence is not the only reason why no one could record ATP current in cells expressing P2X6 (57). Because of those non-conserved sequences, the three-dimensional (3-D) architecture of the LF domain varies in different subtypes. Further studies are required to determine how those distinct sequences make up the functional LF domain in various subtypes. Nevertheless, it provides a foundation for developing subtype-specific blockers of P2X4 receptors by targeting on this non-conserved region of the LF domain throughout the P2X receptor family.
In summary, ATP binding-induced repelling action on the LF domain promotes the formation of physical couplings among two lower body domains and the DF domain and facilitates the outward flexing of lower body domains (Fig. 12), leading to the expansion of the central vestibule and the concomitant pore dilation. This study provides new mechanistic insights into the channel gating of P2X4 receptors and may contribute to develop new strategies for subtype-specific blockers of P2X receptors.

Drugs, cell culture, mutagenesis, and receptor expression
ATP, ZnCl 2 , and most of the other drugs were purchased from Sigma. The plasmid rP2X4 and zfP2X4.1 are the gifts from Drs. Lin-Hua Jiang, Alan North, and Eric Gouaux. Each mutant was constructed by the QuikChange mutagenesis kit and was verified by DNA sequencing. All constructs were expressed in cultured HEK-293 cells in DMEM at 37°C in a humidified atmosphere of 5% CO 2 and 95% air. Transfections of plasmids were performed using Hilymax (Dojindo Laboratories, Kumamoto, Japan). Electrophysiological measurements were performed on HEK-293 cells 24 -48 h after transfection.

Electrophysiology
As in our previous descriptions (32,53), conventional wholecell configuration under the voltage clamp at room temperature (23 Ϯ 2°C) was used for electrophysiological recordings. Patch pipettes were pulled from glass capillaries using the twostage puller PP-830 (Narishige Co., Ltd.), and the resistance between the recording electrode filled with pipette solution and the reference electrode in bath solution ranged from 3 to 5 megohms. Membrane currents were filtered at 2 kHz using a low pass Bessel filter and measured with an Axon 200B patchclamp amplifier (Molecular Devices). All currents were sampled and analyzed in Digidata 1440 interface using Clampex and Clampfit 10.0 software (Molecular Devices). Cells were incubated in bath solution containing 150 mM NaCl, 5 mM KCl, 10 mM glucose, 10 mM HEPES, 2 mM CaCl 2 , and 1 mM MgCl 2 at the conditional neutral pH 7.35-7.40. Patch electrodes were filled with standard internal solution containing 30 mM NaCl, 120 mM KCl, 1 mM MgCl 2 , 0.5 mM CaCl 2 , and 5 mM EGTA at the conditional neutral pH 7.35-7.40. During electrophysiological recordings, 80 -90% of the series resistance was compensated, and the membrane potential was held at Ϫ60 mV throughout the experiment. As we described previously (32), ATP solutions were prepared for 2 h in the batch buffer and applied using a fast pressure-driven computer-controlled microperfusion system OctaFlow08P (ALA Scientific Instrument). ATP currents were normalized to cell membrane capacitance. Dose-response curves data were collected from the recording of a range of ATP concentrations; the corresponding currents were normalized to the maximal current amplitude; ATP-gated currents were recorded after regular 3-5-s ATP application every 2-8 min. Pulses were spaced up to 8 -20 min to avoid receptor desensitization at higher ATP concentration (10 -100 M) applications. The amphotericin-perforated patch-clamp technology (58) was also used for recordings of dose-dependent responses and covalent modifications of WT P2X4, P2X4 W194C , and P2X4 S201C . During this procedure, ATP-gated currents were recorded after regular 15-20-s ATP applications every 8 -10 min to avoid receptor desensitization. Single-channel recordings using outside-out configuration were carried out in HEK-293 cells at room temperature (23 Ϯ 2°C) 24 -48 h after transfection. Recording pipettes were pulled from borosilicate glass (World Precision Instruments, Inc.) and fire-polished to yield resistance of 5-10 megohms. The holding potential was Ϫ120 mV. The external solution and internal solutions are the same as those of whole-cell recordings. Single channel recordings were sampled at 50 kHz with a 2-kHz filter and a low-pass filtered at 200 Hz, using an AxoPatch 200B amplifier in conjunction with pClamp 10 software (Axon Instruments). Occasional large brief noise spikes were visually identified and removed from current traces.

Cell-surface biotinylation and Western blotting analysis
Cell-surface biotinylation and Western blotting were performed according to our previous descriptions (32,59). Briefly, HEK-293 cells expressing rP2X4 or its mutants were washed in chilled PBS ϩ/ϩ and then were incubated with sulfo-NHS-LCbiotin. The reaction was then terminated by treating the cells with glycine in PBS. Then the cells were collected and lysed with RIPA buffer. Using agarose resin linked to NeutrAvidin, the biotinylated proteins were then separated from the intracellular protein fraction. The resins were washed, and bound proteins were eluted with the boiling SDS sample buffer, whereas 10% of the volume of the supernatant was diluted and used as the total protein fraction. The samples were separated by SDS-PAGE and transferred to the polyvinylidene difluoride (PVDF) membrane and then were incubated overnight at 4°C with anti-EE tag (1:1000, Abcam, catalogue number ab40767) or anti-GAPDH (1:1000, Sungene Biotech, catalogue number KM9002) antibodies. Appropriate HRP-conjugated secondary antibodies for EE tag (25°C, 1 h, 1:1000, goat-rabbit IgG(HL)-HRP; Sungene Biotech, catalogue number: LK2001) or GAPDH (25°C, 1 h, 1:3000, goat-mouse IgG(HL)-HRP; Sungene Biotech, catalogue number: LK2003) were further incubated and finally visualized by exposure with the ImageQuant RT ECL system (GE Healthcare) for 1-3 min in ECL solution (Thermo Fisher Scientific). All Western blottings and gels are accompanied by the location of molecular weight markers (Thermo PageRuler Prestained Protein Ladder 10 -170 kDa, catalogue number 26617). Protein expression analysis of each mutant and WT receptors was repeated at least by three independent experiments.

Homology modeling
For homology modeling of rP2X4 and its mutants using program MODELLER (60), the structures of the closed (PDB code 4DW0) and open (PDB code 4DW1) zfP2X4 receptors were taken as the templates. Zinc bridge models were also constructed according to our previous procedure (32) using MODELLER. Briefly, for zinc-binding site reconstructions, a distance constraint (2.0 -2.6 Å) was added between zinc and the coordination atoms of histidine or free cysteine residues. Then, these models applied by OPLS_2005 force field (61) were further minimized by DESMOND (62). The resulting models were further optimized by 1.2-ns MD simulations using program DESMOND with OPLS_2005 force field. After such a time scale of MD simulations, various parameters of zinc bridges, including distances between atoms, bond angles, and dihedral angles, were very close to those obtained by analysis of crystals (52).

MD simulations
As we described previously (32,63), all MD simulations were performed using the program DESMOND (62) with a constant number of particles, pressure, and temperature and periodic boundary conditions, which use a particular "neutral territory" method called the midpoint method (62) to efficiently exploit a high degree of computational parallelism. A default OPLS_ 2005 force field (61), following the functional form of the OPLS-AA family of force fields with additional stretch, bend, and torsional parameters for better coverage of ligand functional groups, was employed for the protein, ions, and ligand molecules. The energy-minimized homology models of rP2X4 and its mutants at the resting or open states were used as the starting structures for MD simulations. The large dimyristoylphosphatidylcholine bilayers in various simulation systems were constructed to generate a suitable membrane system where the TM region of the WT P2X4 and its mutants could be embedded. The protein/dimyristoylphosphatidylcholine system was then solvated in a bath of simple point charge water molecules. Counter ions were subsequently added to compensate for the net negative charge of the system. NaCl (150 mM) was added into the simulation box that represents background salt at physiological conditions. To maintain the system at a constant temperature of 300 K and constant pressure, Berendsen thermostat and barostat algorithms were applied to couple protein and other molecules. All of the bond lengths, including hydrogen atoms, were constrained by the Linear Constraint Solver algorithm. Electrostatic interactions between charged groups at a distance of less than 12 Å were calculated explicitly; long range electrostatic interactions were calculated using the smoothed particle mesh Ewald method. All of the MD simulations were run on the DAWNING TC2600 (AMD Opteron TM 8374HE CPUs). Preparation, analysis, and visualization were performed on a 12-CPU CORE DELL T7500 graphic working station. The MD trajectory analysis were performed using Simulation Even Analysis and Simulation Interactions Diagram tools of DESMOND.

Metadynamics
Metadynamics (43)(44)(45) is a technique where the potential for one or more chosen variables ("collective variables") is modified by periodically adding a repulsive potential of Gaussian shape at the location given by particular values of the variables. All metadynamics analysis were conducted by the program DESMOND (62) under NPT and periodic boundary conditions using the default parameters at constant temperature (320 K) and pressure (1 bar) by using the Berendsen method. All simulations used the all-atom OPLS_2005 force field for proteins, ions, lipids, and the simple point charge waters. The parameters for height, width of the Gaussian, and the interval were set to 0.12 kcal/mol, 0.05 Å, and 0.09 ps, respectively. The sum of the Gaussians and the free-energy surface were generated by Metadynamics Analysis Tools of DESMOND.

Data analysis
The results are expressed as the means Ϯ S.E. Statistical comparisons were made using one-way ANOVA and Student's t test, where p Ͻ 0.05 (*) or p Ͻ 0.01 (**) was considered significant. Concentration-response relationships for ATP activation of WT or mutated channels were obtained by measuring currents in response to different concentrations of ATP, and all of

Physical couplings fostered by left flipper domain
the results used to generate a concentration-response relationship were from the same group. The data were fit to Hill Equation 1, I/I max ϭ 1/(1 ϩ (EC 50 /[ATP]) n ) (Eq. 1) where I is the normalized current at a given concentration of ATP; I max is the maximum normalized current; EC 50 is the concentration of ATP yielding a current that is half of the maximum, and n is the Hill coefficient.