ABSTRACT
The ATP activated P2X4 receptor plays a prominent role in pain perception and modulation and thus may constitute an alternative therapeutic target for controlling pain. Given the biomedical relevance of P2X4 receptors, and poor understanding of molecular mechanisms that describe its gating by ATP, fundamental understanding of functional mechanism of these channels is warranted. Through classical all-atom molecular dynamics (MD) simulations we investigated the number of ATP molecules required to open (activate) the receptor for it to conduct ions. Since crystal structures of human P2X4 are not yet available, the crystal structures of highly-homologous zebrafish P2X4 (zfP2X4) structures were utilized for this study. It has been identified that at least two ATP molecules are required to prevent the open state receptor collapsing back to a closed state. Additionally, we have discovered two metal binding sites, one at the intersection of the three monomers in the ectodomain (MBS1) and the second one near the ATP binding site (MBS2), both of which are occupied by the potassium ions. This observation draws its comparison to the gulf coast P2X receptor that it possess the same two metal binding sites, however, MBS1 and MBS2 in this receptor are occupied by zinc and magnesium, respectively.
1 INTRODUCTION
Purinergic (P2X) receptors are ligand-gated cation channels that reside in the plasma membrane (1). P2X4 receptors are expressed in almost all mammalian tissues (2). P2X4 receptor is activated by adenosine triphosphate (ATP) (3–6), during which the channel opens and allows rapid flow of ions such as calcium (Ca2+), magnesium (Mg2+) and potassium (K+) (6). P2X4 receptor exhibits high Ca2+ permeability and contributes to depolarization of the cell membrane, whereby various Ca2+ sensitive intracellular processes are triggered (6–8). P2X4 receptors are implicated in the regulation of cardiac function, ATP-mediated cell death, synaptic strengthening, activating of the inflammasome in response to injury and multiple sclerosis (5, 9–12). The P2X4 receptors are linked to neuropathic pain mediated by microglia (13, 14). Studies have found that P2X4 receptors are upregulated following injury (15). More importantly, P2X4 activation is both necessary and sufficient for neuropathic pain (16).
Despite their biomedical relevance, several structural (role of intracellular fragments and pore dilation) and functional aspects (activation mechanism, allosteric modulation by agonists such as ivermectin (IVM), the number of ATP needed to activate the channel, and desensitization) of P2X4 channels are still debated (17, 18). This makes it difficult to understand the extent to which P2X4 activation drives cellular responses and its potential druggability. A pressing limitation is the lack of human P2X4 crystal structures. However, zfP2X4 structures have been crystallized in multiple relevant conformations and in different functional states (19, 20). zfP2X4 shares ∼59% sequence identity with human P2X4 (17) and has been shown to form functional homomeric channels with properties comparable to mammalian orthologs (21). Hence, zfP2X4 structures are reasonable surrogates for modeling of the human variants of this channel.
P2X4 channels are trimeric (1) structures. zfP2X4 is a homotrimer (19, 22), with each subunit consisting of a transmembrane domain (TMD) and an extracellular ectodomain (Fig. 1). Each TMD contains two TM helices (TM1 and TM2). The TM2 of each sub unit together forms the TM pore that controls the gating of the channel (19, 22). ATP binds in an orthosteric binding site (Fig. 1) between the two neighboring subunits and facilitates a conformational transition i.e., from closed to open state, which was confirmed by the zfP2X4 crystal structures (19, 22). However, many details are still elusive including the number of ATP required to activate the channel and the role of ions in modulating the channel.
Ions play a critical role in normal functioning of many proteins (23–26). Divalent cations including zinc, magnesium, copper, cadmium, silver and mercury are reported to modulate the P2X receptors. For instance, it was demonstrated that zinc and copper modulate rat P2X4 receptors differently, i.e., zinc potentiates whereas copper inhibits ATP current (27, 28). Further, it was demonstrated via site directed mutagenesis the role of residue C132 in zinc potentiation (29), and residues D138 and H140 in copper inhibition in rat P2X4 receptors (29, 30). Similarly, it was reported that cadmium facilitates where as mercury inhibits ATP mediated currents in rat P2X4 receptors (31). Also, it was hypothesized that there are atleast three metal binding sites in P2X channels (31). Kasuya et al. reported an X-ray structure of Gulf Coast P2X receptor complexed with ATP and zinc ion (32). They have also reported two different metal binding sites, each for zinc and magnesium, respectively. Le et al. identified magnesium binding site near the ATP binding domain in the human P2X3 and demonstrated that magnesium enables ATP mediated activation of the channel. They have also proposed two different binding modes for magnesium in the presence and absence of ATP (33). The impact of divalent ions is not uniform among all P2X receptors. For example, it was proposed that magnesium does not have a role in P2X2 and P2X4, where as in P2X1 and P2X3 it does coordinate ATP binding (34). As explained above monovalent ions including sodium and potassium are permeable through all seven P2X receptors. Studies have reported that monovalent ions potassium and sodium modulate the human P2X7 receptor-operated single channel currents (35). This study also reported a binding site for sodium on the extracellular side. One study demonstrated a modulatory role of protons in rat P2X7 (36).
MD simulations enable studying the structural and functional aspects of membrane proteins at an atomic level (24). The P2X4 computational studies conducted thus far are either based on short nanosecond level unbiased MD simulations or enhanced MD simulations and were only successful in implicating either a part of a channel or a particular residue or interaction in activation (37–44), thus its gating dynamics remain unresolved. A complete rigorous atomic level description of the channel activation is warranted. Hence, using microsecond level equilibrium MD simulations we studied (a) the number of ATP required to activate the channel and (b) the binding sites of monovalent/divalent ions and their potential role in modulating the zfP2X4 channel. Based on ≈13.5 µs of MD simulation data we show that minimum two ATP are required to activate zfP2X4. Additionally, we identified two metal binding sites in zfP2X4, similar to gulf coast P2X.
2 METHODS
2.1 Molecular dynamics simulations
The open state crystal structure of zfP2X4 bound with three ATP molecules (PDB: 4DW1 (20)) was utilized for this study. Initially, the open state zfP2X4 crystal structure was downloaded from the crystal data bank and all waters were removed. The two and one ATP bound open state zfP2X4 systems were generated by deleting ATP from the three ATP bound crystal structure. All three systems were further processed and built for the MD simulation using the CHARMM-GUI web server (45, 46). Each system was placed in the POPC bilayer and solvated using the TIP3P water (47), and K+ and chloride (Cl−) ions were added to balance the charges as well as to attain a concentration of 0.15 M. Each system contains one P2X4 protein, ≈42,425 TIP3P water molecules, ≈388 POPC lipid molecules, 0.15 M KCl and 3/2/1 ATP molecules. The size of each system is ≈155 × 160 × 170 Å3 and the total number of atoms in each system is ≈195,116. Amber force-field (ff12SB) (48) was used to treat the entire system. Each system was energy minimized for 100,000 steps using conjugate gradient algorithm (49), and further relaxed in a multistep procedure (which spans for ≈1.5 ns), wherein the lipid tails, protein side chains, and backbone were restrained and then released in a step wise manner as explained elsewhere (45) in the NVT ensemble. Further, production simulations were conducted under periodic boundary conditions in NPT ensemble. Three replicas of each system were carried, each replica for ∼ 1 µs (1µs × 3 trials × 3 systems = 9µs). AMBER16 (50) was used for conducting the MD simulations. A 2 fs time step was used for the initial relaxation as well for the follow up production simulations. Temperature was maintained at 310 K using langevin thermostat and Nose-Hoover Langevin piston method was used to maintain a 1 atm pressure (51, 52). The non-bonded interactions were cut-off at 12 Å and the particle mesh Ewald (PME) method (53) was used to treat the long range electrostatics. Trajectories were saved every 20 ps. The SHAKE algorithm was used to constrain the hydrogen bonds (54).
In addition, all three systems described above are re-simulated (three replicas for each system) again but with the Mg2+ ions docked near the site of ATP binding as described in the references (32, 33). Please note that in the 2-ATP and 1-ATP systems Mg2+ was only docked to the sites that contains ATP. An additional 2-ATP and 1-ATP system were generated but with Mg2+ docked in all three domains irrespective of presence of ATP molecules. Overall, five Mg2+ docked open state zfP2X4 systems were modeled for MD simulations; a 3-ATP system with Mg2+ found at each docking site, two 2-ATP systems, and two 1-ATP systems. The 2-ATP and 1-ATP had one system each with Mg2+ docked in all three domains and one system each where Mg2+ only docked in domains where ATP is present. All simulations were conducted on a local GPU cluster.
Data analysis was conducted using VMD and its various plugins (55) and five data points per each ns was used for analysis. Simulation input files and generated data are available upon request.
3 RESULTS AND DISCUSSION
To study the ATP activation of P2X4, three open state P2X4 systems, each docked with three (3-ATP), two (2-ATP) and one ATP (1-ATP) molecules were simulated via unbiased MD simulations. An open state zfP2X4 crystal structure bound with three ATP (56) was utilized as a starting structure for this study. All three systems reached an apparent steady state at around 4 Å after 700 ns as evidenced by the backbone RMSDs (Fig. S1J). The backbone RMSD of the TMDs that form the TM pore were stabilized after 700 ns (Fig. S1 D-I) as well in all three systems (≈5-6 Å).
3.1 Two ATP molecules are atleast required to activate the zfP2X4
P2X receptors are activated by ATP facilitating the channel opening and transporting of the cations including K+, sodium (Na+) and Ca2+ across the biological membrane. ATP binds in the ectodomain and facilitate opening of the narrowest part of the channel (TM pore) formed by the TMD. P2X receptors are trimeric (homo/hetero) and are capable of binding three ATP molecules in the ectodomain (56). However, the minimum number of ATP required to activate the P2X receptors is still debated, particularly the P2X4s. Therefore, to study this aspect in detail we simulated three zfP2X4 open state receptors each docked with one, two and three ATP molecules, respectively. Since the activation of the channel opens the TM pore, we measured the pore radius as a function of the simulation time Fig. S2. Pore radius was estimated as the distance of Cα of L316 to the center of the TM pore. L316 was reported as the narrowest part of the TM pore (Fig. 3) (56). The TM pore of the 1-ATP system collapsed immediately (within the first 5 ns), where as the 2-& 3-ATP systems resisted the pore collapse hinting that at-least 2-ATP are required to activate the zfP2X4 channel. The average pore radius in the case of 1-ATP system was ≈3.8 Å (Fig. 3), where as in the case 2- and 3-ATP systems it was always ≥5.8 Å (Fig. 3), except for one 3-ATP trail (Fig. S2). Note that the open zfP2X4 channel was used a starting system for all these simulations.
3.2 Two metal binding sites were identified in zfP2X4
Ions are key for normal functioning of various proteins/receptors in humans. Particularly in P2X receptors, a range of cations bind and modulate their structure, dynamics and function. Studies have reported that divalent cations such as magnesium, zinc, copper and cadmium modulate P2X receptors (27, 28, 31–34). Studies have also identified the role of external monovalent cations in modulating P2X7 receptors, particularly potassium and sodium (35). A sodium binding site has also been identified in P2X7 in the same study. Given their importance, the role of cations have been investigated in this study in zfP2X4. We have identified two metal binding sites in the zfP2X4, similar to what was observed in the gulf coast P2X (32). The first metal binding site (MBS1) was identified at the intersection of three isomers in the upper part of the ectodomain (Fig. 4C). K+ ions were found binding in the MBS1 and coordinating the three isomers via interacting with the polar (such as S101 and Y99) and negatively charged (such as E98, D99 and D323) residues of the protein (Fig. 4C,D). Further, the number of K+ binding in the MBS1 (i.e., the number of K+ ions that are within 8 Å of all three isomers in the vestibule) as a function of simulation time were estimated. On average two to three K+ ions were consistently present in MBS1 in the 3-ATP and 2-ATP systems, whereas in the 1-ATP system they reduced from three to one with time (Fig. 5A). We hypothesize that this is due to the collapse of the channel in the 1-ATP system, thus collapsing MBS1. This supports our claim that atleast two ATP are required to activate the channel (or to keep the TM pore open). Earlier studies reported that zinc (Zn2+) binding in the MBS1 potentiates ATP mediated currents in the gulf coast P2X channel through allostery (32). In the MBS1 residues that interact with the potassium ions are E98, D99, Y99, S101 and D323. This is a highly conserved site across the P2X4 receptors and residues that are interacting with the zfP2X4 are similar to other P2X receptors (32).
The second metal binding site (MBS2) was identified near the ATP binding sites, which were present at the intersection of two neighboring isomers in the ectodomain (Fig. 4A). Several reports suggested that MBS2 was occupied by the Mg2+ ions (32, 33), coordinating the binding of ATP to the P2X2 and p2X4 receptors. In this study we have identified that K+ ions were occupying the MBS2 and interacting with the ATP (Fig. 4A). Further, we estimated that atleast one K+ ion was binding in the MBS2 (Fig. 5B-D). Additionally, we identified correlation between ATP binding to the channel and K+ ions binding to the MBS2, i.e., when no ATP binds in a ATP binding site no K+ ions binds to the respective MBS2, and vice versa. For instance, in the 3-ATP system, three ATP molecules binds to the three ATP binding sites and subsequently we observed K+ ions binding to all three MBS2s. However, in the case of 2-ATP system K+ ions were binding near the two ATP binding sites that were occupied by ATP but not near the ATP binding site that was not occupied with ATP (i.e., site between domains P2 and P3) (Fig. 5 B-D). Similarly in the 1-ATP system, K+ was binding in the MBS2 adjacent to the site occupied with ATP (i.e., between the domains P1 and P3) and absent in the other two MBS2 sites. This is in contrast to the hP2X3, in which Mg2+ ions binds to all three MBS2s irrespective of ATP binding to the ATP binding sites (33). In the MBS2, residues primarily interacting with the K+ ions were D145 and E171 (Fig. 4B). This is similar to others including gulf coast P2X (32) and hP2X3 (33).
It has been reported that ions binding to the MBS2 in different P2X receptors have diversified roles. For instance, magnesium binding to the MBS2 modulates the ATP mediated gating of P2X2 and P2X4 but not in P2X1 and P2X3 (32). Similarly, it was demonstrated that zinc binding to this site do not show potentiation of ATP mediated currents in gulf coast P2X (32). Further studies are required to establish the functional role of various ions binding to these two metal binding sites in zfP2X4, including the significance of various residues.
4 CONCLUSION
zfP2X4 is a homologue of human P2X4 and thus could provide critical insights into the P2X4-dependent progression of neuropathic pain. In this report we studied via MD simulations the activation of zfP2X4 by ATP and show that atleast two ATP molecules are required to activate the zfP2X4. We also identified two metal binding sites in the zfP2X4 similar to other P2X receptors such as gulf coast P2X. The first metal binding site is located at the intersection of three isomers in the upper part of the vestibule and the second metal binding site is located near the ATP binding site. We have identified K+ ions binding to both the metal binding sites. The first metal binding site is dominantly occupied with negatively charged and polar residues thereby facilitating the binding of metal cations. K+ ions binding in the second binding site is coordinating the binding of ATP with the receptor. Further studies are required to understand the role of ions binding to these two binding sites given the divergent role of different ions at different P2X receptors. The information gained from this study will help design strategies aimed at manipulating P2X receptors via small molecule modulators to treat diseases such as multiple sclerosis, neuropathic pain, thrombus, and rheumatoid arthritis (18).
5 AUTHOR CONTRIBUTIONS
KI conducted simulations, analyzed the data and wrote the manuscript. PKH designed the project and wrote the manuscript.
7 SUPPLEMENTARY MATERIAL
6 ACKNOWLEDGMENTS
Research reported in this publication was supported by the Maximizing Investigators’ Research Award (MIRA) (R35) from the National Institute of General Medical Sciences (NIGMS) of the National Institutes of Health (NIH) under grant number R35GM124977. This work used the Extreme Science and Engineering Discovery Environment (XSEDE) (57), which is supported by National Science Foundation grant number ACI-1548562.
ACRONYMS
- Ca2+
- calcium. 1, 4;
- Cl−
- chloride. 3;
- K+
- potassium. 1, 3–6, 8;
- MD
- molecular dynamics. 1–4, 8, 14, 16;
- Mg2+
- magnesium. 1, 4, 6;
- Na+
- sodium. 4;
- Rg
- radius of gyration. 5, 15;
- RMSF
- Root mean squared fluctuations. 16;
- TMD
- transmembrane domain. 2, 4;
- zfP2X4
- zebrafish P2X4. 1–6, 8, 14;
- Zn2+
- zinc. 5;