Thank you for visiting Nature.com.The browser version you are using has limited support for CSS.For the best experience, we recommend that you use an updated browser (or turn off compatibility mode in Internet Explorer).In the meantime, to ensure continued support, we will display the site without styles and JavaScript.
The Hv1 voltage-gated proton channel is a dimeric complex consisting of two voltage-sensing domains (VSDs), each of which contains a gated proton permeation pathway.Dimerization is controlled by the cytoplasmic coiled-coil domain.The transition from the closed state to the open state in both VSDs is known to occur cooperatively; however, little is known about the underlying mechanisms.Intersubunit interfaces play a key role in allosteric processes; however, such interfaces have not been identified in open Hv1 channels.Here we demonstrate that 2-guanidinothiazole derivatives block two Hv1 VSDs in a cooperative manner and use one of these compounds as a probe for allosteric coupling between open subunits.We found that the extracellular end of the first transmembrane fragment of VSD forms an intersubunit interface that mediates coupling between binding sites, whereas the coiled-coil domain is not directly involved in this process.We also found strong evidence that the channel’s proton-selective filter controls blocker-binding cooperativity.
Voltage-gated proton channels play important roles in a variety of organisms, from phytoplankton to humans1.In most cells, these channels mediate proton efflux from the proton membrane and regulate the activity of NADPH oxidase.The only known voltage-gated proton channel in humans is Hv1, which is the product of the HVCN1 gene2,3.Hv1 (aka VSOP) has been shown to play a role in B cell proliferation4, production of reactive oxygen species by the innate immune system5,6,7,8, sperm cell motility9 and pH regulation of airway surface fluid10. This channel is involved in several Overexpressed in cancer types such as B-cell malignancies4,11 and breast and colorectal cancers12,13.Excessive Hv1 activity was found to increase the metastatic potential of cancer cells 11, 12 .In the brain, Hv1 is expressed by microglia, and its activity has been shown to exacerbate brain damage in models of ischemic stroke.
The Hv1 protein contains a voltage-sensing domain (VSD), which consists of four transmembrane segments called S1 to S414.The VSD resembles the corresponding domains of voltage-gated Na+, K+ and Ca2+ channels and voltage-sensitive phosphatases, such as CiVSP from Ciona gutis15.In these other proteins, the C-terminus of S4 is attached to an effector module, the pore domain or enzyme.In Hv1, S4 is linked to the coiled-coil domain (CCD) located on the cytoplasmic side of the membrane.The channel is a dimeric complex consisting of two VSDs, each containing a gated proton permeation pathway16,17,18.These two Hv1 subunits were found to open cooperatively19,20,21,22, suggesting that allosteric coupling and inter-subunit interactions play an important role in the gating process.The interface between the subunits within the coiled coil domain is well-defined because the crystal structures of the two isolated domains are available22,23.On the other hand, the interface between VSDs within the membrane is not well understood.The crystal structure of the Hv1-CiVSP chimeric protein does not provide information about this interface, as the trimeric organization of the crystallized channel complex may result from the replacement of the native Hv1 CCD by the yeast leucine zipper GCN424.
A recent study of the subunit organization of the Hv1 channel concluded that two S4 helices transition into the CCD without severe disruption of secondary structure, resulting in long helices that start from the membrane and project into the cytoplasm.Based on cysteine crosslinking analysis, this study proposes that Hv1 VSDs contact each other along the S4 segment.However, other studies have proposed alternative interfaces between VSDs.These interfaces include S1 segments 17, 21, 26 and the outer ends of S2 segment 21.A possible reason for the conflicting results of these studies is that allosteric coupling between VSDs was examined in relation to the gating process, which depends on the closed and open states, and the interface between VSDs may vary in different state changes in conformation.
Here, we found that 2-guanidinothiazole inhibits Hv1 channels by synergistically binding to two open VSDs, and used one of the compounds, 2-guanidinobenzothiazole (GBTA), to probe the interaction between subunits in the open state. interface.We found that the GBTA binding curve can be well described by a quantitative model in which binding of an inhibitor to one subunit results in an increase in the binding affinity of the adjacent subunit.We also found that residues D112, selectivity filters for the channel 27, 28 and part of the guanidine derivative binding site 29 control GBTA binding cooperativity.We show that cooperative binding is maintained in the Hv1 dimer, where the CCD separates from the S4 segment, suggesting that the intersubunit interface in the CCD does not directly mediate allosteric coupling between GBTA-binding sites.In contrast, we find that the S1 fragment is part of the interface between the subunits and propose an arrangement of adjacent VSDs with the extracellular end of the S4 helix away from the center of the dimer to allow the S1 fragment to be in the open state.
Small molecule inhibitors of Hv1 are useful as anticancer drugs and neuroprotective agents.However, to date, few compounds have been able to inhibit the channel30,31,32,33.Among them, 2-guanidinobenzimidazole (2GBI, compound [1] in Fig. 1a) and its derivatives were found to block VSD29,32 permeation of protons through the channel.Binding of such compounds is thought to occur independently in the two open subunits.2-guanidinobenzothiazole (GBTA, compound [2] in Figure 1a) was previously shown to inhibit Hv1 almost as effectively as 2GBI when tested at a concentration of 200 μM (Figure 1b).We examined other thiazole derivatives and found that some of them inhibited the channel with similar or greater potency than GBTA (Fig. 1 and Supplementary text).We determined the concentration response curves of four thiazole derivatives (GBTA and compounds [3], [6] and [11], Fig. 1c) and found that they were steeper than those of 2GBI.The Hill coefficients (h) of thiazole derivatives ranged from 1.109 ± 0.040 to 1.306 ± 0.033 (Fig. 1c and Supplementary Fig. 1).In contrast, the Hill coefficient for 2GBI was 0.975 ± 0.024 29, Fig. 2a and Supplementary Fig. 1).A Hill coefficient above 1 indicates binding cooperativity.Because each Hv1 subunit has its own inhibitor-binding site,29,32 we reasoned that the binding of a thiazole derivative to one subunit could enhance the binding of a second inhibitor molecule to the adjacent subunit.GBTA was the test compound with the highest Hill coefficient.Therefore, we chose this compound to further study the mechanism of binding synergy and used 2GBI as a reference negative control.
(a) Test compounds: [1] Reference Hv1 inhibitor 2-guanidino-benzimidazole (2GBI).[2] 2-guanidino-benzothiazole (GBTA), [3] (5-trifluoromethyl-1,3-benzothiazol-2-yl)guanidine, [4] naphtho[1,2-d][1, 3] Thiazol-2-yl-guanidine, [5](4-methyl-1,3-thiazol-2-yl)guanidine, [6](5-bromo-4-methyl-1,3-thiazol- 2-yl)guanidine, [7] famotidine, [8] 2-guanidino-5-methyl-1,3-thiazole-4-carboxylic acid ethyl ester, [9] 2-guanidino-4-methyl Ethyl-1,3-thiazole-5-carboxylate, [10](2-guanidino-4-methyl-1,3-thiazol-5-yl)ethyl acetate, [11]1-[4-(4 -Chlorophenyl)-1,3-thiazol-2-yl]guanidine, [12]1-[4-(3,4-dimethoxyphenyl)-1,3-thiazol-2-yl]guanidine .(b) Inhibition of human Hv1 activity by the indicated guanidinothiazoles and the reference compound 2GBI (blue-green bars).Hv1 proton currents were measured in inside-out plaques of Xenopus oocytes in response to depolarization from a holding potential of -80 mV to +120 mV.Each inhibitor was added to the bath at a concentration of 200 μM.pHi = pHo = 6.0.Data are mean±SEM (n≥4).(c) Concentration-dependent inhibition of human Hv1 by compounds [2], [3], [6] and [11].Each point represents the mean inhibition ± SD of 3 to 15 measurements. The line is a Hill fit used to obtain the apparent Kd values reported in Supplementary Table 1.Hill coefficients were determined from the fits reported in Supplementary Fig. 1: h(1) = 0.975 ± 0.024 h(2) = 1.306 ± 0.033, h(3) = 1.25 ± 0.07, h(6) = 1.109 ± 0.040, h (11) = 1.179 ± 0.036 (see Methods).
(a,b) Compounds 2GBI and GBTA inhibited dimeric and monomeric Hv1 in a concentration-dependent manner.Each point represents the mean inhibition ± SD of 3 to 8 measurements and the curve is a Hill fit.The Hill coefficients (h) shown in the inset histograms were determined from the fits reported in Supplementary Figs 3 and 4.The concentration response of GBTA shown in (a) is the same as that in Fig. 1c.See Supplementary Table 1 for apparent Kd values.(c) Modeling of the cooperative binding of GBTA to dimeric Hv1.The solid black line represents the fit to the experimental data by equation (6), which describes the binding model shown in (d).The dashed lines labeled Sub 1 and Sub 2 represent the bimolecular association-dissociation equilibrium curves of the first and second binding events, respectively (Sub 1: OO + B ⇄ BO*, Kd1 = 290 ± 70 μM; Sub 2: BO* + B ⇄ B*O*, Kd2 = 29.3 ± 2.5 μM).(d) Schematic diagram of the proposed mechanism of the Hv1 block.In the case of GBTA, binding to one open subunit increases the affinity of the adjacent open subunit (Kd2 < Kd1).Subunits are grey in the low affinity state and blue in the high affinity state.For GBTA binding to monomeric Hv1, the dissociation constant for the reaction: O° + B ⇄ B° is between Kd1 and Kd2, but closer to the latter (grey-blue stripes indicate moderate affinity).
Hv1 channels can be monomerized by replacing their N- and C-terminal cytoplasmic domains with corresponding portions of the Ciona Intestinalis voltage-sensitive phosphatase CiVSP (Hv1NCCiVSP)18,34.We measured the concentration dependence of inhibition of monomeric Hv1 by GBTA and 2GBI and compared them to that of dimeric channel (wild-type) inhibition (Fig. 2a,b).We found that the difference in binding cooperativity between the two compounds was eliminated in monomeric Hv1, supporting the explanation that GBTA binding to one subunit increases the affinity of the other subunit for the inhibitor.We reasoned that GBTA binding to one Hv1 subunit might lead to rearrangement of the binding site (inducing fit35), which would trigger the rearrangement of the vacant binding site of the adjacent subunit, leading to increased binding affinity.
To quantitatively describe the cooperative binding of GBTA to the channel, we used a model in which either subunit can bind the first inhibitor molecule following a bimolecular reaction with a dissociation constant Kd1 (Fig. 2c, Sub 1, OO+ B ⇆ BO* ).Binding causes the channel to adopt a state in which the remaining empty subunits bind to the inhibitor following a unique bimolecular reaction with a dissociation constant Kd2, where Kd2 < Kd1 (Fig. 2c, Sub 2, BO* + B ⇆ B*B *). Once the second inhibitor molecule is bound, both subunits have a dissociation constant Kd2 (Fig. 2d).Channel inhibition was measured under depolarizing conditions (+120 mV).Channel species with one open and one closed subunit (see transition CC ⇄ OC ⇄ OO in Fig. 2d) were previously found to contribute negligibly to the Hv1 current under these conditions19,20 and thus they were not included in the binding model .
The solid black line in Figure 2c is the fit of the model equation to the experimental concentration-response curve, yielding a Kd1 of ~290 μM and a Kd2 of ~29 μM (Methods section, equation (6)).The model also describes the binding of 2GBI, where Kd2 ≈ Kd1 = Kd (Fig. 2d).In monomeric Hv1, there is only one binding site and one Kdm (O° + B ⇆ B°).In the case of GBTA, Kdm is around 54 μM, which is more similar to Kd1 than Kd2 (Fig. 2d).In other words, the binding site of the monomer is in a configuration (O°) more similar to the high-affinity state (BO*) than the low-affinity state (OO) of the dimer, most likely due to the elimination of the interface between subunits.
As previously shown for 2GBI32, we found that GBTA suppressed Hv1 currents by blocking the channel when it was open rather than by making it harder to open (Supplementary Fig. 2a,b,d).2GBI is also known to induce slowly decaying tail currents in response to membrane repolarization, but this phenomenon was not observed in GBTA (Supplementary Fig. 2c).In the presence of Hv1 inactivation in the presence of 2GBI, the gates in each subunit cannot close until the blocker leaves the binding site (the “foot gate” mechanism), and 2GBI unbinds slower than the gates close.If one Hv1 subunit is unblocked and closed, while the adjacent subunit remains blocked, unbinding of the remaining 2GBI molecules becomes slower (blocker capture).Long-lived channel species with only one blocked subunit conduct protons transiently before closing and make a significant contribution to the slowly decaying tail current.
The finding that the decay of Hv1 tail currents was not significantly slowed in the presence of GBTA (Supplementary Fig. 2c) is consistent with a synergistic binding mechanism for this blocker (Fig. 2d).Once a GBTA molecule is unbound from the Hv1 subunit, the affinity of the blocker to the adjacent subunit drops about 10-fold, favoring the unbinding process.This means that the second molecule of GBTA has a much higher chance of unraveling before it gets trapped in the channel.Long-lived channel species that produce only one blocking subunit are expected to be much more abundant in the presence of GBTA than in the presence of 2GBI, resulting in faster tail current decay.
In order for GBTA to bind cooperatively with both Hv1 subunits, the binding sites must be allosterically coupled.Each binding site must be capable of: 1) triggering a chain of events that communicate with adjacent subunits bound to the inhibitor, and 2) mediating the transition from low-affinity to high-affinity binding.We reasoned that if specific residues within the binding site contributed to the cooperative process, their mutation should alter the Hill coefficient of the concentration-response curve of GBTA inhibition of Hv1.Residues D112, F150, S181 and R211 were previously shown to be part of the 2GBI29 binding environment and we hypothesized that they would similarly be involved in GBTA binding (Fig. 3A).We measured inhibitory concentration-response curves of GBTA for mutant channels D112E, F150A, S181A, and R211S (Figure 3) and compared their Hill coefficients to those of Hv1 wild-type, using 2GBI as a reference.Residue V109 is on the same face of the S1 segment as D112 and has one extra helical turn inside the cell.Since V109 was not involved in the binding of 2GBI29, we used the V109A mutant as a control (Fig. 3b).
(a) Suggested Hv1 residues involved in guanidine derivative binding.Dashed blue curves surround previously proposed side chains that interact with different parts of compound 2GBI29.(bf) Concentration-dependent inhibition of the indicated Hv1 mutants by compounds 2GBI (cyan) and GBTA (dark red).Each point represents the mean inhibition of 3 to 12 measurements ± SD V109 as a negative control.Curves are Hill fits used to obtain apparent Kd values (see Supplementary Table 1).The Hill coefficients (h) shown in the inset histograms were determined from the fits reported in Supplementary Figs 3 and 4.Reference h values for Hv1 WT are shown as dashed lines.Asterisks indicate statistically significant differences between mutant and WT passages (p < 0.05/14, see Methods).
We found that the D112E mutation significantly reduced the Hill coefficient (p < 0.05/14 see Methods), but not for 2GBI, compared with the wild type of GBTA (Fig. 3c and Supplementary Figs 3 and 4).The S181A mutation caused significant changes in the Hill coefficients of GBTA and 2GBI (p < 0.05/14, Fig. 3e and Supplementary Figs 3 and 4), whereas the mutations V109A, F150A and R211S did not cause significant changes in the Hill coefficients (Fig. 3 and 4). Supplementary Figures 3 and 4).We conclude that GBTA binding cooperativity is abolished in the D112E channel and partially reduced in the S181A channel.However, the S181A mutation was found to reduce the Hill coefficient for 2GBI binding to values significantly below 1, with little effect on potency (Fig. 3e), suggesting that the mutation created some negative cooperativity between binding sites.In support of negative cooperativity, we found less (and no longer statistically significant) changes in the Hill coefficient for 2GBI inhibition induced by the S181A mutation in the monomeric background (Supplementary Fig. 5).
The finding that the Hill coefficient for GBTA binding is higher than 1 in the Hv1 dimer and becomes ~1 in the monomeric channel (Fig. 2b) is consistent with the existence of allosteric interactions between the binding sites in the two subunits.If this is the case, the Hill coefficient of GBTA binding to the dimer to which the blocker has bound to one subunit should be ~1.We tested this prediction in the Hv1 F150A-WT linked dimer (Fig. 4a), where the affinity of the F150A subunit for GBTA was more than 2 orders of magnitude higher than that of the WT subunit (Figs. 1c and 3d).We then measured concentration-response curves for inhibition of the WT subunit at a basal concentration of 2 μM GBTA.At this concentration, the blocker is expected to bind approximately 99% of the F150A subunit and <2% of the WT subunit, resulting in a {F150A}b-WT (or BAO*) blocked hemichannel (Figure 4a) .The reduction in proton current measured after the addition of 2 μM GBTA to the F150A-WT channel (Fig. 4b) is consistent with the inhibition of the F150A subunit.Concentration-response curves for blocked hemichannels yielded Hill coefficients very close to 1 (Fig. 4c), confirming that GBTA-binding cooperativity arises from allosteric coupling between binding sites on adjacent subunits, rather than intra-subunit binding coupling between sites.
(a) Schematic diagram of the F150A-WT junction dimer used to generate the blocking hemichannel ({F150A}b-WT/BAO*).White diamonds show the location of the mutation.In the BAO* and BA*B* states, the affinity of the F150A subunit is expected to be higher than that of the WT subunit.(b) Proton currents from F150A-WT channels measured in response to changes in membrane potential from −80 mV to +120 mV.pHi = pHo = 6.0.Grey traces represent currents measured in the absence of inhibitor.The black trace (control) represents the current measured after addition of 2 μM GBTA.At this concentration, the WT subunit was not significantly inhibited, whereas the F150A subunit bound almost completely to GBTA ({F150A}b-WT).A further increase in GBTA concentration resulted in blockage of the WT subunit (orange and red traces).(c) Concentration-dependent inhibition of {F150A}b-WT dimer (inhibitor pre-bound to F150A subunit).Each point represents the mean inhibition ± SD of 3 to 7 measurements. The curves were Hill fits used to obtain the apparent Kd values reported in Supplementary Table 1.The Hill coefficients shown in the inset histograms were determined from the fits reported in Supplementary Figs 3 and 4.The h values of Hv1 WT are shown as dashed lines.Asterisks indicate statistically significant differences compared to WT dimer Hv1 (p < 0.05/14, see Methods).
Since GBTA inhibition of Hv1 was measured in the open channel, we reasoned that GBTA binding cooperativity could be used to study the mechanism of intersubunit coupling in the open state.These two Hv1 subunits were previously found to be cooperatively gated 19,20,21,22 and the intersubunit coupling involved in gating was proposed to be mediated by the cytoplasmic coiled-coil domain (CCD)22,25.Therefore, we asked whether CCD is also involved in the coupling between GBTA-binding sites in the open state.Triglycine mutations at the interface between the S4 transmembrane fragment and the coiled-coil domain were shown in earlier studies to decouple this domain from the rest of the channel while maintaining dimerization integrity.Therefore, we tested the effect of mutations V220G, K221G and T222G (Fig. 5a, GGG mutant) on GBTA binding cooperativity.We found a statistically insignificant (p > 0.05) decrease in the Hill coefficient of GBTA binding to GGG channels compared with wild type (Fig. 5c), suggesting that despite its role in keeping the two subunits together important, but CCD does not directly mediate intersubunit allosteric coupling between GBTA binding sites.
(a) Schematic diagram of the Hv1 dimer with a triglycine mutation at the inner end of S4 designed to disrupt the intersubunit coupling mediated by the cytoplasmic coiled-coil domain (blue arrows).(b) Schematic representation of Hv1 dimers and ligation dimers with indicated mutations, designed to test S1 fragments involved in coupling between subunits (blue arrows).(c–h) 2GBI (cyan) and GBTA (dark red) inhibit the indicated constructs in a concentration-dependent manner.Each point represents the mean inhibition ± SD of 3 to 10 measurements. The curve was a Hill fit used to obtain apparent Kd values (see Supplementary Table 1).Hill coefficients in the interpolated histograms were determined as described in the Methods section (see Supplementary Figs 3 and 4).Reference h values for Hv1 WT are shown as dashed lines.Asterisks indicate statistically significant differences between mutant and WT passages (p < 0.05/14, see Methods).
Since we ruled out CCD as a direct mediator of GBTA cooperative binding, we asked whether interactions between VSDs are involved in allosteric coupling between binding sites.We found that GBTA-binding cooperativity was abolished by conservative mutation of residue D112 (Fig. 3c) located in the S1 transmembrane segment.S1 contains two other negatively charged residues, E119 and D123, located on the same side of the helix as D112, but closer to the outer end 24 of the fragment.Early molecular dynamics simulations indicated that these residues were linked to D112 via waterlines36,37.Therefore, we tested whether E119 and D123 are involved in allosteric coupling between GBTA-binding sites (Fig. 5b).As a negative control, we tested the conserved positively charged position K125, which is located near D123 but on the other side of the S1 helix (Fig. 5b).
We measured the concentration-dependent inhibition of E119A, D123A and K125A channels by 2GBI and GBTA and found that E119A and K125A mutations did not significantly alter the Hill coefficient of either inhibitor compared to wild type (p > 0.05, Fig. 5d,i ) ). On the other hand, mutation D123A significantly reduced the Hill coefficient of GBTA (p < 0.05/14, Fig. 5e and Supplementary Fig. 4), suggesting that electrostatic interactions between charged residues involve the interaction between GBTA-binding sites Allosteric coupling.We further investigated this possibility by examining the effect of positive charge at position 123 on binding cooperativity.The D123R mutation increased the Hill coefficient of GBTA channel inhibition compared with wild type (Fig. 5f).However, the magnitude of the increase was small and did not meet our criteria for statistical significance (p > 0.05/14).
Since neutralizing the charge at position 123 on both subunits resulted in a strong change in GBTA binding cooperativity, while reversing the charge on both subunits had only a small effect, we extended the analysis to include only one subunit with the inversion channel of charge.We generated Hv1-linked dimers with D123R substitution in the C-terminal subunit (Fig. 5b) and measured the concentration-response inhibition by GBTA and 2GBI.We found that the Hill coefficient of GBTA binding to WT-D123R channels was significantly higher than that of wild-type Hv1 (p < 0.05/14, Fig. 5g and Supplementary Fig. 4), suggesting that the allosteric coupling between binding enhances the point (increased positive synergy for GBTA binding).In contrast, the Hill coefficient of 2GBI binding to WT-D123R channels was significantly lower than that of wild-type Hv1 (p < 0.05/14, Fig. 5g and Supplementary Fig. 3).The reduction of the coefficient to a value below 1 suggests that manipulation of allosteric coupling between binding sites can lead to an increase in negative cooperativity for 2GBI binding.
We also found that the D112E mutation abolished the increase in the GBTA-binding Hill coefficient produced by the WT-D123R background (Fig. 5b,h and Supplementary Fig. 4).The effect on the Hill coefficient of 2GBI binding was also eliminated (Fig. 5h and Supplementary Fig. 3).These findings suggest that the enhanced allosteric coupling between subunits caused by opposite charges at position 123 does not translate into stronger binding cooperativity when the binding site at position D112 is disturbed.
We reasoned that if the D123 residues on adjacent open subunits were close enough to electrostatically interact with each other, the repulsive interaction between the negative charges would be converted to a combination of negative and positive charges by substitution of aspartic acid to arginine. attractive interaction between them.One subunit (WT-D123R dimer).Attractive interactions can strengthen the interface between the outer ends of the S1 fragment, leading to stronger allosteric coupling between the subunits and increased GBTA-binding cooperativity.
To support this hypothesis, we looked for a way to strengthen the interface between the outer ends of the S1 segment, which is distinct from the charge switching at position 123.Lee et al.Mutation of Hv1 residue I127 to cysteine was previously found to result in spontaneous intersubunit cross-linking17.Furthermore, the crystal structure of the Hv1-CiVSP chimeric VSD indicated that I127 is separated from the outer end of the S1 helix by only one residue24.We therefore asked whether the formation of a covalent bond between cysteines at position 127, which is expected to result in a stronger S1-S1 interaction, has an effect on GBTA binding cooperativity similar to that observed by changing the charge at position 123 .
We measured the concentration-dependent inhibition of Hv1 I127C by GBTA and 2GBI in the presence and absence of 10 mM β-mercaptoethanol (βME) (Fig. 6a).At the same time as the inhibitor is added to the intracellular solution, the reducing agent is added to the extracellular solution.A linked dimer with cysteine substitution in only one subunit (WT-I127C) was used as a negative control (Fig. 6a).We were unable to measure any effect of spontaneous intersubunit cross-linking between the I127C subunits on 2GBI inhibition, because the Hill coefficient for I127C binding to Hv1, with or without βME, was significantly different from that for binding to monocysteine substituted dimers. The Hill coefficient could not differentiate WT-I127C (Fig. 6b and Supplementary Fig. 3).In stark contrast, we measured the dramatic effect of spontaneous intersubunit crosslinking on GBTA inhibition.The Hill coefficient for binding to Hv1 I127C in the absence of βME (crosslink on) was significantly higher than measured in the presence of βME (crosslink off) or using WT-127C to link the dimer (in the absence of crosslink) (Fig. 6c and Supplementary Fig. 4), indicating that the allosteric coupling between binding sites is greatly enhanced by the formation of disulfide bonds between the outer ends of the S1 fragments.These results support our interpretation that the attractive electrostatic interaction of aspartate and arginine at position 123 in the WT-D123R dimer results in increased allosteric coupling between the subunits.
Open-state coupling is mediated by direct physical interactions between the outer ends of the S1 segment in the two subunits.
(a) Schematic representation of mutant Hv1 dimers and linker dimers, designed to investigate how intersubunit cysteine crosslinks affect GBTA binding cooperativity.(bd) Concentration-dependent inhibition of the indicated constructs by 2GBI (b,d) and GBTA (c,e) in the presence or absence of 10 mM βME in extracellular solution.Each point represents the mean inhibition ± SD of 3 to 10 measurements. The curve was a Hill fit used to obtain Kd values (see Supplementary Table 1).Hill coefficients in the inset histograms were determined from the fits reported in Supplementary Figs 3 and 4.Asterisks in (c,e) indicate statistically significant differences between different GBTA inhibition conditions (p < 0.0001, see Methods).The corresponding difference in 2GBI inhibition [(b,d)] was not statistically significant (p > 0.05).
We also found that in the absence of βME, the Hill coefficient of GBTA binding to the D112E I127C Hv1 dimer was significantly higher than that measured in the presence of reducing agents (Fig. 6e and Supplementary Fig. 4), implying that The D112E mutation did not abolish the increase in GBTA binding cooperativity resulting from the cross-linking of cysteine 127.The Hill coefficient of 2GBI binding to D112E, I127C Hv1 dimers was also not significantly affected by the D112E mutation (Figure 1).6d and Supplementary Fig. 3).
Taken together, these findings suggest that GBTA binding is enhanced by the interaction between the outer ends of S1 fragments in adjacent Hv1 subunits through attractive electrostatic interactions or through the formation of covalent bonds between substituted cysteines Allosteric coupling between sites increases and results in binding cooperativity.While the effect of attractive electrostatic interactions on cooperativity can be abolished by mutation D112E, the effect of covalent bonds cannot.
In our exploration of the chemical space available for binding guanidine derivatives to Hv1 channels, we found that 2-guanidinothiazoles like GBTA have a steeper concentration dependence than 2-guanidinobenzimidazoles (Fig. 1c) .The Hill coefficient analysis of GBTA binding to both the dimeric and monomeric channels (Fig. 2a,b) and to the dimeric channel in which one subunit was pre-bound to the inhibitor (Fig. 4) led us to conclude that the inhibition of Hv1 by GBTA The action is a synergistic process, and the binding sites of the compounds in the two subunits are allosterically coupled.The discovery that GBTA binds to the open channel, as previously shown for the related compound 2GBI32, suggests that allosteric coupling can be specifically assessed in the open state.Our cooperative binding model was able to quantitatively describe the inhibition of Hv1 by GBTA (Fig. 2c) and explain the different effects of 2GBI and GBTA on the decay of channel tail currents after membrane repolarization, which supports our interpretation of the binding process.
The maximum Hill coefficient that can be achieved in an allosteric protein with two binding sites (such as Hv1) is 2.We measured GBTA to bind Hv1 wild-type with a coefficient of 1.31, which increased to 1.88 in Hv1 I127C.The synergistic free energy, the difference between the binding free energies of the lowest and highest affinity sites (see Methods), was 1.3 kcal/mole in the case of Hv1 wild-type and 2.7 kcal/mole in the case of Hv1 I127C.The binding of oxygen to hemoglobin is the most famous and well-studied example of the collaborative process38.For human hemoglobin (a tetramer with four allosterically coupled binding sites), the Hill coefficient ranges from 2.5-3.0, with values ranging from 1.26 to 3.64 kcal/mol, depending on experimental conditions38.Thus, in terms of global energetics, the cooperativity of GBTA binding to Hv1 is not substantially different from that of O2 binding to hemoglobin when the different numbers of protein subunits in the two systems are considered.
In our synergy model, the binding of GBTA molecules to one subunit results in an increase in the binding affinity of the adjacent subunit.We envision a process in which a rearrangement of the binding environment resulting from the first binding event (induced fit) leads to changes in the interactions between subunits.In response to these changes, adjacent subunits change their binding sites, resulting in tighter GBTA binding.During this process, the S1 aspartate D112 is responsible for the rearrangement of the binding site associated with increased binding affinity.D112 was previously shown to be part of the Hv1 proton permeation pathway and to act as selectivity filter27,28. Our results show that selectivity filters in the two Hv1 subunits are allosterically coupled in the open state.Figure 7a shows the approximate location of the GBTA binding site and the location of residues D112, D123, K125 and I127 on a schematic of the Hv1 VSD based on the crystal structure of the Hv1-CiVSP chimera.Suggested directions for allosteric coupling involving the extracellular end of S1 are shown with black arrows.
(a) Schematic of the Hv1 VSD.Based on the crystal structure of the Hv1-CiVSP chimera, the helical segments are shown to be cylindrical.In this configuration, the inner end of the S4 segment merges with the CCD (not shown).The predicted locations of bound GBTA are shown as grey ovals.Black arrows indicate pathways involved in allosteric coupling between binding sites in two adjacent subunits.The positions of the investigated S1 residues are marked with colored spheres.(b) Schematic illustration of Hv1 subunits arranged in two different dimer configurations as seen from the extracellular side of the membrane plane.On the left panel, the dimer interface is formed by the S4 helix.Rotation of the two VSDs 20 degrees clockwise around an axis perpendicular to the membrane plane, accompanied by separation of the outer ends of the two S4 helices (dashed arrows), produces the arrangement shown on the right.In this configuration, residues D123 and I127 from adjacent subunits are allowed to come close together.(c) Schematic representation of the CiVSP VSD in the dimer configuration found in the 4G80 crystal structure.Position P140 in CiVSP corresponds to position D123 in Hv1.
Our results show that the repulsive electrostatic interaction between residue 123 (123D/123D or 123R/123R) is associated with a “normal” level of GBTA-binding cooperativity in the open state and shifts the interaction from repulsion to Attracted (123D/123R), increased cooperation (Fig. 5g).It is expected that removal of the repulsive interaction with alanine substitution would also lead to increased cooperativity.However, a decrease in cooperativity of the 123A/123A dimer was observed (Fig. 5e).One possible explanation is that the destabilizing effect of placing hydrophobic residues in a hydrophilic environment may outweigh the stabilizing effect due to the elimination of repulsive interactions between D123 residues, resulting in an overall reduction in binding cooperativity.Mony et al.39 recently reported that solvent accessibility at position D171 of Ciona gutis Ci-Hv1 (corresponding to D123 in human Hv1) is increased upon activation, supporting the notion that D123 is located in a hydrophilic environment in the open state.
Gating of Hv1 channels is known to occur through multiple transitions19,20,26,39,40,41.Qiu et al26 found that upon membrane depolarization, the voltage sensor of Ci-Hv1 undergoes a conformational change that leaves the channel activated but still closed, followed by a distinct transition that causes protons to open conduction in both subunits path.The conformational change of the voltage sensor was monitored by voltage-clamp fluorescence, and it was found that the second transition was selectively perturbed by the mutation at position D171.The perturbation of the fluorescent signal is consistent with the presence of electrostatic interactions between the D171 residues of adjacent subunits at some point along the reaction coordinates of the conformational change.This interpretation is consistent with our finding that the D123 residue electrostatically interacts in the open state and mediates allosteric coupling between GBTA binding sites.
Fujiwara et al25 proposed that the dimer interface of the cytoplasmic coiled-coil domain extends into the membrane to contain two S4 helices (Fig. 7b, left panel).A model of this intersubunit interaction is based on cysteine cross-linking analysis covering the entire VSD and functional analysis of the region connecting the S4 helix and the CCD.In the absence of a transmembrane pH gradient, Hv1 channels require considerable membrane depolarization to open, and cysteine crosslinking occurs under conditions where the channel is predominantly closed.Therefore, the detected S4-S4 interface likely reflects the off-state subunit configuration.Other studies have found evidence for the involvement of S1 and S2 in intersubunit interactions during gating17,21,26 suggesting that channels may adopt different subunit configurations in the open and closed states, an idea consistent with the findings of Mony et al. S1 moves 39 during gating.
Here, we show that, in the open state, the extracellular ends of the S1 helix are close enough to support direct electrostatic interactions that mediate allosteric coupling between subunits.In the dimer configuration with extended S4-S4 interactions, the S1 helices are too far apart from each other to interact directly.However, a 20° clockwise rotation of the two VSD subunits around an axis perpendicular to the membrane plane, combined with the separation of the outer ends of the two S4 helices, yielded an S1-S1 configuration consistent with our findings (Fig. 7b, right panel).We recommend this configuration for the channel in the open state.
Although the CiVSP enzyme is thought to function as a monomer, the crystal structure of its isolated VSD is captured in a dimer state.The outer ends of the S1 helices in this dimer are spatially close together, and the overall configuration is similar to that proposed for Hv1 (Fig. 7c).In the CiVSP dimer, the closest residue from the adjacent subunit was proline at position 140 (Fig. 7c).Interestingly, P140 in CiVSP corresponds to position D123 in Hv1.The similarity in subunit configuration between Hv1 and CiVSP dimers suggests that the VSDs of these proteins have an intrinsic tendency to form an interface where the extracellular ends of S1 interact.
The essential role of Hv1 in sperm cell activation makes this channel an attractive drug target for the control of male fertility.Furthermore, activation of Hv1 in microglia has been shown to worsen recovery from ischemic stroke.Enhanced Hv1 activity was found to be associated with lower survival in patients with breast 12 or colorectal cancer 13 and is thought to contribute to B-cell malignancies 11 .Therefore, small molecule drugs targeting Hv1 can be used as neuroprotective agents or anticancer therapeutics.The discovery that guanidinothiazole derivatives can induce conformational rearrangements in the open Hv1 subunit, leading to increased binding affinity, may lead to the development of more potent drugs targeting Hv1 channels.
Site-directed mutagenesis of human Hv1 was performed using standard PCR techniques.In the Hv1NCCiVSP construct, residues 1-96 and 228-273 of Hv1 were replaced by residues 1-113 and 240-576 of CiVSP18.In the Hv1-linked dimer, the C-terminus of one subunit is linked to the N-terminus of the second subunit through the GGSGGSGGSGSGGSGG linker.The pGEMHE plasmids containing the various constructs were linearized with Nhe1 or Sph1 restriction enzymes (New England Biolabs) and RNA synthesis was performed with the T7 mMessage mMachine transcription kit (Ambion).1-3 days before electrophysiological measurements, cRNA was injected into Xenopus oocytes (50 nl per cell, 0.3-1.5 μg/μl).Stage V and VI oocytes from Xenopus laevis (NASCO) were obtained from Ecocyte Bioscience.Following RNA injection, cells were maintained in ND96 medium containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl, 1 mM MgCl, 10 mM HEPES, 5 mM pyruvate, 100 μg/ml gentamicin, pH 7.2 18°C.
2-guanidino-benzimidazole[1], 2-guanidino-benzothiazole[2], (4-methyl-1,3-thiazol-2-yl)guanidine [5], (5-bromo-4 -methyl-1,3-thiazol-2-yl)guanidine) [6], ethyl 2-guanidino-5-methyl-1,3-thiazole-4-carboxylate [8], ethyl 2-guanidino-4- methyl-1,3-thiazole-5-carboxylate [9] and (2-guanidino-4-methyl-1,3-thiazol-5-yl)ethyl acetate [10] were from Sigma-Aldrich.Famotidine [7] was from MP Biomedicals.1-[4-(4-Chlorophenyl)-1,3-thiazol-2-yl]guanidine[11] and 1-[4-(3,4-dimethoxyphenyl)-1,3- Thiazol-2-yl]guanidine [12] was from Matrix Scientific.These compounds are of the highest purity commercially available.They were dissolved in dry DMSO to generate a 100 mM stock solution, which was then diluted in the recording solution at the desired final concentration.Compounds 3 and 4 were synthesized below with at least 99% purity.
To a suspension of 2-amino-4-(trifluoromethyl)benzenethiol hydrochloride (1.02 g, 4.5 mmol) in 25 mL of aqueous hydrochloric acid (2.5 N) was added solid dicyandiamide (380 mg, 4.5 mmol), and the resulting heterogeneous mixture was refluxed with vigorous stirring for 4 hours.The reaction mixture was cooled to room temperature and neutralized by gradual addition of 10N potassium hydroxide.The white precipitate formed was filtered, washed with cold water (3 × 50 ml), dried in an oven (65 °C) for several hours, and then recrystallized from ethyl acetate/petroleum ether to give a white solid (500 mg, 48 %); mp 221–222 °C (light 225–226 °C) 45; 1H NMR (500 MHz, DMSO-d6): δ [ppm] = 7.25 (very broad s, 4 H), 7.40 (d, 1 H, J = 8.1 Hz), 7.73 (s, 1 H), 7.92 (d , 1 H, J = 8.1 Hz).13C NMR (200 MHz, DMSO-d6): δ = 114.2 (d, J = 3.5 Hz), 117.5 (d, J = 3.5 Hz), 121.7, 124.6 (q, J = 272 Hz), 126.1 (q, J = 272 Hz) = 31.6 Hz), 134.8, 152.1, 158.4, 175.5.HRMS (ESI): m/z calculated value.For C9H8F3N4S (M + H)+: 261.0416, found: 261.0419.
Naphtho[1,2-d]thiazol-2-amine (300 mg, 1.5 mmol), synthesized as previously described, was heated to 200 °C in an oil bath in a small test tube.300 mg (large excess) cyanamide and 1.0 ml conc.To the hot compound was added hydrochloric acid rapidly and the mixture was kept in the oil bath for about 2 minutes, during which time most of the water evaporated.The reaction mixture was then cooled to room temperature and the resulting solidified material was broken into small pieces and washed with water to provide a pale yellow amorphous solid.(38 mg, 10%) mp 246-250 °C; 1H NMR (500 MHz, DMSO-d6, D2O): δ [ppm] = 7.59 (t, 1 H, J = 8.2 Hz), 7.66 (t, 1 H, J = 8.3 Hz), 7.77 (d, 1 H, J = 8.6 Hz), 7.89 (d, 1 H, J = 8.6 Hz), 8.02 (d, 1 H, J = 8.2 Hz), 8.35 (d , 1 H, J = 8.3 Hz).13C NMR (150 MHz, DMSO-d6): δ = 119.9, 122.7, 123.4, 123.6, 126.5, 127.1, 128.7, 132.1, 140.7, 169.1.HRMS (ESI): m/z calculated value.For C12H11N4S (M + H)+: 243.0699, found: 243.0704.
Proton currents were measured in internal and external patches of oocytes expressing different constructs using an Axopatch 200B amplifier controlled by an Axon Digidata 1440A (Molecular Devices) with pClamp10 software.Intracellular and extracellular solutions have the same composition: 100 mM 2-(N-morpholino)ethanesulfonic acid (MES), 30 mM tetraethylammonium (TEA) mesylate, 5 mM TEA chloride, 5 mM ethylene glycol-bis(2-aminoethyl)-N,N,N’,N’-tetraacetic acid (EGTA), adjusted to pH 6.0 with TEA hydroxide.All measurements were performed at 22 ± 2 °C.The pipette has an access resistance of 1.5–4 MΩ.Current traces were filtered at 1 kHz, sampled at 5 kHz and analyzed with Clampfit10.2 (Molecular Devices) and Origin8.1 (OriginLab).
Solutions containing various concentrations of Hv1 inhibitor and in some cases 10 mM βME were introduced into the bath by gravity through a manifold connected to a VC-6 perfusion valve system (Warner Instr.), which was passed through the pClamp software TTL (Transistor-Transistor Logic) signals.Rapid perfusion experiments were performed using a multi-tube perfusion pencil (AutoMate Sci.) with a 360 μm diameter delivery tip mounted in front of a patch pipette.Channel inhibition was determined by isochronal amperometric measurements at the end of the +120 mV depolarizing pulse.GV measurements were performed as previously described18,20.Tail currents were recorded at -40 mV after depolarization steps at different voltages from -20 mV to +120 mV unless otherwise stated.The reference pulse preceding the test pulse is used to correct for current decay 18 .The GV graph fits the Boltzmann equation:
The apparent dissociation constants (Kd) for different combinations of channels and inhibitors (Supplementary Table 1) were determined by fitting the concentration dependence of inhibition (mean %inhib values) with the Hill equation:
where [I] is the concentration of inhibitor I and h is the Hill coefficient.To calculate the Hill coefficient, equation (2) is rearranged as:
Post time: Jun-07-2022
