|Home | About | Journals | Submit | Contact Us | Français|
Voltage-gated proton (Hv1) channels are dimers, where each subunit has a separate permeation pathway. However, opening of the two pathways is highly cooperative. It is unclear how Hv1 channels open their permeation pathways, because Hv1 channels lack a classic pore domain. Using voltage clamp fluorometry, we here detect two conformational changes reported by a fluorophore attached to the voltage sensor S4 in Hv1 channels. The first is voltage dependent and precedes channel opening, with properties consistent with reporting on independent S4 charge movements in the two subunits. The second is less voltage dependent and closely correlates with channel opening. Mutations that reduce dimerization or alter the intersubunit interface affect both the second conformational change and channel opening. These observations suggest that, following an initial S4 charge movement in the two subunits, there is a second, cooperative conformational change, involving interactions between subunits, that opens both pathways in Hv1 channels.
Voltage-gated proton (Hv1) channels were first discovered in snail neurons (Thomas and Meech, 1982). Subsequently, Hv1 have been found in a variety of cell types and have been implicated in a wide variety of biological functions, such as the respiratory burst in phagocytic cells and the capacitation of the sperm (DeCoursey, 2010; El Chemaly et al., 2010; Iovannisci et al., 2010; Lishko et al., 2010; Morgan et al., 2009). In neurons, Hv1 channels are thought to play a role in the control of intracellular pH by extruding protons in response to excess intracellular acidification (Decoursey, 2003) and have been recently implicated in enhancement of brain damage in ischemic stroke (Wu et al., 2012). Hv1 channels belong to the superfamily of voltage-gated cation channels (Ramsey et al., 2006; Sasaki et al., 2006). However, in contrast to other members of this superfamily, such as Kv and Nav channels, Hv1 channel subunits have just four transmembrane segments (S1–S4; Fig. 1A), which correspond to the S1–S4 voltage sensor domain in Kv channels (Ramsey et al., 2006; Sasaki et al., 2006). Hv1 channels lack the characteristic pore domain that is made up of transmembrane segments S5 and S6 in other voltage-gated cation channels (Jiang et al., 2003; Long et al., 2005). In other voltage-gated cation channels, the intracellular end of S6 constitutes the activation gate that opens and closes the pore (Liu et al., 1997). The lack of S6 in Hv1 channels makes it unclear what constitutes the gate in Hv1 channels. We here characterize conformational changes that are coupled to opening and closing (i.e. gating) of Hv1 channels, in order to elucidate how Hv1 channels are gated.
Hv1 channel subunits oligomerize as dimers (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008), not tetramers as other members of the superfamily of voltage-gated cation channels (Jiang et al., 2003; Long et al., 2005). This dimerization is most likely through the formation of a coiled-coil structure of the two C-termini from the two subunits in the dimer (Fujiwara et al., 2012; Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008). Each subunit of the dimer has its own permeation pathway (Koch et al., 2008; Tombola et al., 2008), but there is a strong cooperativity between the two subunits during channel activation (Fujiwara et al., 2012; Gonzalez et al., 2010; Tombola et al., 2010). Using voltage clamp fluorometry and cysteine accessibility to thiol reagents on Hv1 channels, we previously showed that the fourth transmembrane segment S4 with its three positively charged residues moves outward in response to depolarizations, as if S4 functions as the voltage sensor (Gonzalez et al., 2010). Cysteines introduced in the external end of S4 are exposed to the extracellular solution at depolarized membrane potentials, but not at hyperpolarized potentials. Conversely, cysteines introduced in the internal end of S4 are exposed to the intracellular solution at hyperpolarized membrane potentials, but not at depolarized potentials (Gonzalez et al., 2010). In addition, in response to depolarizations, a fluorophore attached close to S4 displays a fluorescence decrease that precedes the proton current (Gonzalez et al., 2010), suggesting that the fluorescence decrease reports on S4 charge movement preceding channel opening. The time course of the fluorescence decrease raised to a power of two overlaps well the time course of the proton currents for moderate depolarizations (Gonzalez et al., 2010). Using a Hodgkin-Huxley type analysis, we previously interpreted the time courses of the fluorescence and currents to denote that S4s in both subunits need to activate before either of the two proton permeation pathways in the two subunits is activated. This suggests that there is a high degree of cooperativity between the two subunits in the Hv1 dimer during channel opening. Using linked Hv1 subunits with different voltage dependences, Tombola and coworkers showed also that there is strong cooperativity between the two Hv1 subunits during Hv1 channel opening (Tombola et al., 2010).
Recently, Musset and coworkers (Musset et al., 2011) identified a conserved aspartate (D112 in hHv1) in the first transmembrane segment (S1) as the selectivity filter for the permeation of protons in Hv1 channels (D160 in Ciona-Hv1: Fig. 1A). Berger and Isacoff proposed a possible interaction between D112 and the third positively charged residue of S4 (Fig. 1A) at the selectivity filter in the open conformation of the channel (Berger and Isacoff, 2011), suggesting a close interaction between S4 and the selectivity filter in open Hv1 channels. However, it is still unknown how the Hv1 channel opens and closes its permeation pathways. In addition, the mechanism underlying the strong cooperativity in opening between the two subunits in the dimer during voltage activation is also unclear. A cysteine cross-linking study showed two dimeric interfaces between the two subunits of the Hv1 channel: one at the extracellular side of S1, including the adjacent loop, and a second one at the intracellular C terminus (Lee et al., 2008). We here test whether intersubunit interactions in these regions are involved in the cooperative opening in dimeric Hv1 channels.
Using voltage clamp fluorometry, we here detect two distinct conformational changes during the opening and closing of Hv1 channels. The first conformational change is highly voltage dependent and is consistent with the previously proposed, independent charge movements of the two S4s in the two subunits preceding channel opening (Gonzalez et al., 2010). The second conformational change is less voltage dependent and correlates with channel opening. Mutations in the two dimeric interfaces right shift the voltage dependence of both the second conformational change and channel opening, suggesting that subunit interactions contribute to the second conformational change and the cooperative gating of the two permeation pathways in the Hv1 channel.
Using voltage clamp fluorometry (VCF) and cysteine accessibility on cysteines introduced in and around S4 of Hv1 channels, we have previously shown that S4 moves as if it functions as the voltage sensor in Hv1 channels (Gonzalez et al., 2010). To perform VCF, we introduce a cysteine at position S242 at the external end of S4 in Ciona intestinalis Hv1 (Ci-Hv1) channels (Fig. 1A–B). We express S242C channels in Xenopus oocytes and subsequently label these oocytes with the thiol-reactive fluorophore Alexa488-maleimide. In VCF, changes in fluorescence from the introduced fluorophore are assumed to report on conformational changes of the protein region in which the fluorophore is introduced (Gonzalez et al., 2010). As expected for a voltage sensor, we measure voltage-dependent fluorescence changes from Alexa488-maleimide attached to S242C in S4 of Hv1 channels (Fig. 1D). This fluorescence change is due to fluorophores attached to S242C, because no fluorescence change was observed from Alexa488-maleimide treated oocytes expressing wt Hv1 channels (Supplementary Fig. S1A).
In response to small depolarizing voltage steps that did not open a significant number of Hv1 channels, the fluorescence slowly decreased during the voltage step and then slowly returned to the fluorescence baseline when the voltage returned to the hyperpolarizing holding potential (Fig. 1C–E). In response to depolarizations large enough to open most Hv1 channels, the fluorescence developed in a biphasic manner (Figs. 1D and 1F). Two phases with opposite polarity are most clearly seen in the ‘tail’ fluorescence during the hyperpolarizing pulse following the depolarizing pulse: the fluorescence rapidly decreases to a minimum, followed by a slow return of the fluorescence to the resting level (Fig. 1F). Similar biphasic fluorescence changes are observed from Alexa488-maleimide labeled 241C Hv1 channels and tetra-methyl-rhodamine-maleimide labeled S242C Hv1 channels (Supplementary Fig. S1B–C), suggesting that the biphasic fluorescence changes from fluorophores attached to S4 report on conformational changes of S4.
The biphasic nature of the fluorescence is not due to changes in local pH, which in Hv1 channels has a strong effect on gating (Cherny et al., 1995). We performed our recordings in the presence of high concentrations of proton buffers (100 mM HEPES) on both sides of the membrane to minimize changes in pH due to the proton currents. Furthermore, the biphasic fluorescence is also observed in oocytes expressing low amounts of 242C channels and in oocytes expressing 242C channels with the N264R mutation, shown to reduce the proton currents by over 95% (Sakata et al., 2010) (Suppl. Fig. S1D).
What gives rise to the biphasic fluorescence changes during the depolarizing voltage pulse and the following hyperpolarizing voltage pulse? One possibility is the following hypothesis: 1) Following the depolarizing voltage step, the outward movement of the two independent S4 voltage sensors changes the environment around the fluorophores attached to the S4s, thereby decreasing the fluorescence (Fig. 2A). 2) As both S4s become activated in a channel, the channel undergoes a second conformational change that opens the permeation pathways. This second conformational change leads to further changes in the environment around the fluorophores, which partially restore (increase) the fluorescence (Fig. 2A). 3) The net result during the depolarizing voltage step is that the fluorescence first decreases as the two S4s activate and then the fluorescence increases as the channel opens (Fig. 2A). 4) Following the return step to the hyperpolarizing holding potential, the open channel closes quickly, resulting in a rapid reduction in fluorescence, producing a hook in the fluorescence signal, and then 5) more slowly the two S4s move back inward, leading to an increase in fluorescence back to the starting level (Fig. 2A; see Supplementary Fig S2A–B for more details).
If the above hypothesis about the origin of the two phases of fluorescence change is correct, then the appearance of the initial decrease (from here on called Fhook) in the tail fluorescence during the hyperpolarizing voltage pulse is an indication that some Hv1 channels have undergone the second conformational change and channel opening during the preceding depolarizing voltage pulse. Then the appearance and the amplitude of Fhook should correlate with the time course of Hv1 channel opening. In contrast, the fluorescence minimum in the tail fluorescence (from here on called Ftail) is a measure of the fluorescence after the second fluorescence change has quickly reversed, but before the first fluorescence change has more slowly reversed. Therefore, Ftail should correlate more with the initial S4 charge movement, and not with channel opening (see Figure 2B for definition of Fhook and Ftail).
The amplitude of the Fhook increases with both the amplitude of the depolarization (Fig. 1D) and the duration of the depolarization (Fig. 2B). We observe that there is a close correlation between the development of Fhook in response to increasing lengths of the depolarized voltage pulses and the development of the proton current during the depolarized pulse, whereas the development of Ftail clearly precedes both the proton currents and Fhook (Fig. 2C). Thus, the time development of Fhook shows that Fhook is correlated to the opening of the proton permeation pathways in Hv1 channels.
We also compare the kinetics of the two fluorescence phases in the tail fluorescence with the closing kinetics of the channel during the hyperpolarizing voltage pulse (Fig. 2D). The kinetics of the two fluorescence phases are both dependent on the hyperpolarizing voltage: the more hyperpolarized the voltage, the faster the kinetics are for both fluorescence phases (Fig. 2E). The tail currents during the hyperpolarizing voltage pulse also display two kinetic components, indicated by the fact that the tail currents cannot be fitted by a single exponential function (Suppl. Fig. S2C–D). A channel with one open state and several closed states (as in Fig. 2A) is expected to have a multiexponential closing kinetics (Hoshi et al., 1994; Zagotta et al., 1994), especially if the channel has a significant probability to reopen again from the first closed state (Cact) during the hyperpolarizing voltage step. The kinetics of the two fluorescence phases are similar to the kinetics of the two current components during channel closing (Fig. 2E). Thus, the time dependence of Fhook show that Fhook is correlated to both the opening and closing of the proton permeation pathways in Hv1 channels.
In order to assess the voltage dependence of the amplitude of the two kinetic phases during the tail fluorescence, we plot the amplitude of Fhook and the amplitude of the minimal fluorescence during the tail (called Ftail) versus voltage (see Fig. 2B for definition of Fhook and Ftail). The normalized proton conductance versus voltage curve, G(V), superimposes closely the Fhook versus voltage curve, Fhook(V), whereas the Ftail versus voltage curve, Ftail(V), is shifted in the hyperpolarized direction compared to the G(V) (Fig. 2F). That Ftail(V) has a V1/2 that is more hyperpolarized than the V1/2 of the G(V) is expected for a multi subunit channel, such as Hv1, in which activation of more than one voltage sensor is necessary to turn on the conductance. This is similar to the left- shifted F(V) relative to the G(V) in the homologous Kv and Nav channels (Cha and Bezanilla, 1997; Mannuzzu et al., 1996). That Fhook is correlated to the activation of the current suggests that Fhook reports on a conformational change that opens Hv1 channels. Thus, both the time and voltage dependences of Fhook show that Fhook is correlated to the opening and closing of the proton permeation pathways in Hv1 channels.
In contrast, no Fhook was detected in the VCF signal from the Ciona voltage sensitive phosphatase (Ci-VSP), which is homologous to the Ciona voltage sensitive proton channel (Ci-Hv1) but lacks proton channel activity (Murata et al., 2005). The voltage dependent fluorescence change of Ci-VSP showed only one component (Supplementary Fig. S2E), most likely reporting on the outward movement of the S4 segment upon membrane depolarization (Murata et al., 2005), suggesting that the additional fluorescence change (Fhook) in the Hv1 channel might be due to the opening and closing of the proton permeation pathway.
All these findings are consistent with our hypothesis that the fluorescence component Ftail reports on the outward movement of the positively charged S4 segment that precedes (in both time and voltage) the opening of the proton permeation pathways, whereas the fluorescence component Fhook reports on a second conformational change that both reverses part of the change in fluorescence and opens the gates of the Hv1 channel (Fig. 2A). Below, we explore this working hypothesis further using mutant Hv1 subunits.
It is known that changing the oligomeric state of Hv channels changes the voltage dependence and kinetics of Hv channels (Koch et al., 2008; Tombola et al., 2008). To investigate the origin of the biphasic fluorescence in Hv1 channels, we therefore measured the fluorescence from Hv1 channels that form only monomers (Fig. 3A), due to the deletion of the N- and C-terminal regions of the protein (ΔNΔC Hv1 channels) (Koch et al., 2008; Tombola et al., 2008). The midpoint of the fluorescence change Ftail is similar in the monomer channel and in the wild type dimeric channel (F1/2 = +16.2 ± 1.1 mV, n = 6, versus F1/2 = +19.1 ± 1.9 mV, n = 4). In contrast, the midpoint of the conductance versus voltage curve, V1/2, for the monomeric ΔNΔC Hv1 channels is shifted more than 40 mV towards more positive potentials compared to wild type dimeric Hv1 channels (Fig. 3B). In addition, the fluorescence hook is notably smaller in ΔNΔC Hv1 channels than in the dimeric Hv1 channels (Fig. 3A inset). The second component of the fluorescence change, as indicated by the fluorescence hook, and the conductance are similarly shifted to more depolarized potentials, as if the second fluorescence component is coupled to channel opening. Because both channel opening and the second conformational change are dramatically shifted to a more depolarized voltage range in the monomeric Hv1 channel, this suggests that changes in inter-subunit interactions contribute to the second conformational change and the cooperative opening of the dimeric channels.
Inter-subunit interactions are most likely localized at the interfaces of two subunits. One such interface is the coiled-coil structure of the two C-termini from the two subunits in the Hv1 dimer (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008). Another dimeric interface in Hv1 channels is localized at the extracellular end of S1 (Lee et al., 2008). We looked for residues in this region that could underlie the cooperative interactions and the second conformational change during Hv1 channel opening. At the extracellular end of S1, we identified D171, which is conserved in all Hv1 channels (Musset et al., 2011) (Fig. 4A), as a residue that could possibly be involved in intersubunit interactions.
Substituting the aspartate D171 by a neutral alanine (D171A) shifts the Ftail(V) for D171A around 14 mV in the hyperpolarized direction compared to the F(V) from the wild type channels (F1/2 = +5.0 ± 2.5 mV, n = 3, Fig. 4E). However, as for ΔNΔC/S242C channels, the G(V) curve in D171A channels is shifted to more depolarized potentials relative the Ftail(V) (Fig. 4E) and the D171A/S242C channels have a much smaller fluorescence hook than S242C channels (Fig. 4B and 4D). The second component of the fluorescence change, as indicated by the fluorescence hook, and the conductance are similarly shifted to more depolarized potentials (Fig. 4E), suggesting that D171 contribute toboth the second conformational change and channel opening.
The currents of the D171A and dimeric wild-type Ci-Hv1 channels both have a sigmoidal time course (Fig. 4C), in contrast to the mono-exponential current time course of the monomeric ΔNΔC channels. The sigmoidal time course in wt Hv1 channels has been proposed to be due to cooperative channel opening of the two subunits (Gonzalez et al., 2010). The sigmoidal time course of D171A channels suggests that the D171A channels still function as cooperative dimers. However, the effects of the D171A mutation on the voltage dependence of the currents and the second conformational change are similar to those of monomeric ΔNΔC channels. Therefore, we conducted further experiments to rule out that the D171A mutation prevents dimerization of D171A subunits. To further test whether the D171A mutant is expressed a dimer, we conducted FRET experiments on D171A/S242C channels. We have previously demonstrated that 242C channels display a large FRET efficiency, E, (E = 65 % ± 07 %, n = 12), showing that 242C channels are expressed as dimers in the plasma membrane (Koch et al., 2008). D171A/S242C channels also display a large FRET efficiency (58 % ± 13 %, n = 8), confirming that D171A/S242C channels are expressed as dimers in the plasma membrane. This shows that the effects of the D171A mutation are not due to a loss of dimerization of D171A Hv1 subunits.
To test whether the effects of the D171A mutation and the ΔNΔC deletion were additive, we made the D171A mutation in the ΔNΔC construct (ΔNΔC/D171A). The voltage dependence of channel opening is very similar in ΔNΔC and ΔNΔC/D171A channels, showing that the effects of the ΔNΔC deletion and the D171A mutation on channel opening are not additive (Suppl. Fig. S3). One possibility for this lack of additivity is that, in dimeric channels, D171 interacts with a residue in the other subunit, whereas, in monomeric ΔNΔC channels, this interaction is lost due to the loss of dimerization. Thus, the D171A mutation would have no functional effect in the monomeric ΔNΔC channels.
D171 is located at the external end of S1, a region where introduced cysteines have been shown to crosslink the subunits (Lee et al., 2008). For example, introducing a cysteine in the human Hv1 channel at a residue homologous to residues 168 or 170 in Ci-Hv1 allows a disulfide bond to be formed between the two subunits (Lee et al., 2008). This suggests that D171 of one Ci-Hv1 subunit is located close to the D171 of the other Ci-Hv1 subunit. Then, D171 residues from adjacent subunit may interact electrostatically across the intersubunit interface and thereby drive the second conformational change. We tested this possibility by other mutations at position 171. The proton current and fluorescence for D171E are similar to wt Hv1 channels (Suppl. Fig. S4). The voltage dependence of the proton current from the D171R mutant is shifted to more depolarized potentials (Suppl. Fig. S4). However, both D171E and D171R restored the biphasic feature of the fluorescence signal (Fig. 5). The fact that both negative and positive charged residues at position 171 restored the fluorescence hook supports our hypothesis that an electrostatic repulsion between 171 residues across subunits contributes to the second conformational change and channel opening.
The ΔNΔC deletion, which prevents dimerization, and the D171A mutation, which is localized at the S1 dimeric interface, impaired the second conformational change and channel opening, suggesting that subunit interfaces are important for cooperative channel gating. These mutations shifted the second conformational change to very depolarized voltages, so that we could measure the initial outward movement of S4 separately (seen in the VCF signal as the decreasing fluorescence signal in response to a depolarization). We also identify another mutation, R255A, that allowed us to measure the second conformational change separately. We have previously shown that neutralization mutations of the most external S4 charge, R255, restricted the inward S4 movement, so that S4 is prevented from reaching the normal resting position of S4 at negative voltages (Gonzalez et al., In Press). However, channel opening and closing is retained in R255 charge-neutralization mutations (Gonzalez et al., In Press). Figure 6A shows the current and fluorescence from the R255A/S242C mutant. The fluorescence change from R255A/S242C shows only one component and the polarity of the fluorescence change is the same as the second fluorescence component in wt Hv1 channels (Fig. 6A). Moreover, the voltage dependence of the fluorescence change is similar to the voltage dependence of channel opening (Fig. 6B), and the kinetics of the fluorescence increase during the depolarizing pulses superimposes the kinetics of channel opening (Fig. 6C) and the kinetics of the fluorescence decrease during the hyperpolarizing pulses superimposes the kinetics of channel closing (Fig. 6D). In contrast to the sigmoidal time course of the currents from wt Hv1 channels, the time course of the current from R255A channels are not sigmoidal, as if there is only one conformational change during R255A channel activation. This conformational change in R255A is similar to the second conformational change in wild type Hv1 channels. The fact that the fluorescence from R255A channels superimposes on the currents, further supports that the second conformational change in wild type Hv1 channels is closely related to channel opening.
Our data suggests that there are two types of transitions during opening and closing of Hv1 channels. In our gating model, we call the first type of transition S4 activation (α) and the second type of transition S4 stabilization (δ) in response to a depolarizing voltage pulse (cartoon in Fig. 7). In response to a subsequent hyperpolarizing pulse (following the depolarizing voltage pulse), the two conformational changes occur in the reverse order. We call these transitions S4 destabilization (γ) followed by S4 deactivation (β) (cartoon in Fig. 7).
In order to determine the rates and voltage dependence of each transition in our model, we measured the S4 activation time constant (τact), S4 stabilization time constant (τstb), S4 destabilization time constant (τdestb) and S4 deactivation time constant (τdeact) individually at different voltages by using three different stimulation protocols (Fig. 7A, C, E) to isolate the different transitions as much as possible from each other.
As seen in Figure 1, the S4 activation and S4 stabilization fluorescence changes are convolved in wild type Ci-Hv1 channels during depolarizations, making it difficult to isolate the S4 activation conformational change (especially at voltages with significant channel opening). However, during repolarization S4 stabilization is quickly reversed, as shown by the fast rise of the fluorescence hook, while S4 activation is reversed only slowly (Fig. 2A). This allows us to estimate the amount of S4 activation that has occurred at the end of the depolarizing pulse by measuring the tail fluorescence Ftail after S4 destabilization has occurred (Ftail measured as in Fig. 7A inset). We therefore determined the S4 activation rate constant by measuring the amplitude Ftail in response to different lengths of activation voltage steps (Fig. 7A and 7B inset). We measure the S4 activation rate at different voltages to obtain the voltage dependence of the S4 activation rate. The effective gating charge calculated from the S4 activation rate is 1.61 e0 ± 0.32 e0, (n = 4) (Fig. 7B).
To isolate S4 stabilization from S4 activation, we first open the channel by stepping to a highly depolarized voltage. Subsequently, a 20-ms hyperpolarizing pulse is applied, allowing S4 destabilization to occur without much S4 deactivation. We then step back to different depolarizing voltages to measure the rate of S4 stabilization (Fig. 7C). During this second depolarizing pulse, the channel only goes through S4 stabilization, because the channels have already undergone S4 activation during the first depolarizing pulse (cartoon in Fig. 7). The effective gating charge calculated for the S4 stabilization rate is 0.97 e0± 0.09 e 0, (n = 3) (Fig. 7D).
By stepping to different hyperpolarizing voltages after channel opening, we measure the voltage dependence of the kinetics of S4 destabilization and S4 deactivation (Fig. 7E). In the fluorescence tail in response to the hyperpolarizing pulses, we interpret the fast fluorescence decrease as S4 destabilization and the subsequent slower fluorescence recovery to the fluorescence baseline as S4 deactivation (Fig. 7E inset). The S4 destabilization and S4 deactivation fluorescence changes can be measured separately, because, in response to a hyperpolarization, the channel undergoes S4 destabilization well before S4 deactivation takes place. The calculated effective gating charges for the S4 destabilization rate and S4 deactivation rate are 0.43 e0 ± 0.04 e0, (n =5) (Fig. 7F) and 0.81 e0± 0.03 e 0, (n = 3) (Fig. 7G), respectively.
Because the second conformational change (S4 stabilization and S4 destabilization) is significantly affected in the monomeric ΔNΔC Hv1 channels and by a mutation at the extracellular intersubunit interface in dimeric Hv1 channels, it is likely that this conformational change involves some inter-subunit interactions in the dimeric Hv1 channel. In contrast, the first conformational change is not significantly shifted in the monomeric ΔNΔC Hv1 (ΔNΔC: F1/2 = +16.2 ± 1.1 mV, n = 6, versus wt: F1/2 = +19.1 ± 1.9 mV, n = 4), as if this conformational change involves independent S4 movements in the two subunits in the dimeric Hv1 channel (Gonzalez et al., 2010). We therefore model the first conformational change as an S4 movement that occurs independently in the two subunits, whereas the second conformational change occurs as a concerted conformational change simultaneously in the two subunits (Fig. 8A). With the assumption that the initial S4 movement is independent in the two subunits (Gonzalez et al., 2010), the gating charges associated with this conformational change should be counted twice (Fig. 8A). With this assumption, the sum of the translocated charges in these two conformational changes, (1.61*2+0.81*2 +0.43+0.97) e0 = 6.24 e0, is very close to the total gating charges movement in the dimeric Hv1 channel (~6 e0) (Gonzalez et al., 2010), suggesting that all of the major charge movements in Hv1 channels have been identified by our study.
Using all the measured rates and voltage dependences of each transition, we simulated currents and fluorescence by our model (Fig. 8A) in response to voltage protocols identical to those used in our experiments. One prediction of this model is that channels should open with very different kinetics (more than 10-fold different), whether the channels are opened from the fully rested state (C0) or from the state just preceding the open state (C2) (Fig. 8B). This can best be seen in a simulation of the model in response to a triple-pulse protocol, in which we first activate the channels by a step to +60 mV, then close the channels briefly by a voltage pulse to −120 mV with different durations (Δt = 4 ms), and then reopen the channels again by a voltage pulse to +60 mV (Fig. 8B inset). In response to this triple pulse protocol, the model channel opens with very different time course in response to the two +60 mV voltage pulses: 1) in response to the first voltage pulse to +60 mV, the channels open slowly because the channels are initially in the C0 state and need to undergo the slow S4 movement in each subunit before they can open. 2) after the shorter hyperpolarized voltage pulses at −120 mV, most channels are in the C2 state and can quickly reopen by undergoing the fast cooperative opening step. 3) after longer hyperpolarized voltage pulses at −120 mV, most channels have had time to undergo transitions to the C1 and C0 states from which the channels open slowly, because the channels need to undergo the slow S4 movement back to the C2 state before they can open.
To test this prediction, we conducted experiment on wt Ci-Hv channels in excised patches to be able to well resolve the kinetics for short voltage pulses (Fig. 8C). The channels open with the normal slow activation kinetics in response to the first pulse to +60 mV (τ = 1551 ± 432 ms, n = 3). In contrast, most channels open with a much faster kinetics in response to the second pulse to +60 mV (τ = 51.3 ± 12.6 ms, n = 3) after the shorter hyperpolarizing voltage pulses (Fig. 8C and Supplementary Fig. S5A inset). The proportion of channels that open with the faster time course decreases with the length of the hyperpolarized voltage pulse, until most channels open with the same slow activation kinetics during the second +60 mV voltage step as during the first +60 mV voltage step (Fig. 8C). The Hv1 blocker 2-GBI (Hong et al., In press) applied at 200 μM blocked the majority of the currents during both +60 mV voltage pulses (Supplementary Fig. S5A–B), showing that the current during both voltage pulses is due to Hv1 channels and not some endogenous channels. Also the change in time course is not due to changes in local pH, which previously has been shown to alter the kinetics and voltage dependence of Hv1 channels, because the reversal potential of the proton currents was still close to the expected 0 mV in symmetrical solutions as measured in between the two depolarizing voltage pulses (Supplementary Fig. S5C). This drastic change in the time course of opening in response to the triple pulse protocol is exactly what is predicted by our model (Fig. 8B–C). The fluorescence curves generated by our model were also similar to our experimental data (cf. Fig. 7A, 7C, and 7E with Fig. 8E–G, respectively), including the biphasic tail fluorescence (cf. Fig. 1D and Fig. 8D).
This paper describes that Hv1 channel opening involves at least two conformational changes in the voltage sensor domain in response to membrane depolarization. Both conformational changes are voltage dependent. The first conformational change precedes channel opening and is consistent with the main outward charge movement of the voltage sensor S4. The second conformational change correlates with channel opening and might represent the elusive gate movement of Hv1 channels. The second conformational change was significantly decreased by the truncation of the cytosolic domains in ΔNΔC Hv1 and by the D171A mutant, suggesting that an inter-subunit rearrangement contributes to this second conformational change in which D171 plays an important role. Interestingly, the second conformational change required a charged residue at position 171, because both positive and negative charges at position 171 restored the second conformational change. Therefore, we propose that the second conformational change is driven by mutual electrostatic repulsion of the two D171 residues in the two subunits. The movement of the extracellular ends of the two S1 segments relative to each other would relieve this electrostatic repulsion. This movement of the two S1 segments would allow for a rearrangement of the voltage sensing domain, thereby generating the second conformational change of S4. In addition, this second conformational change is correlated with channel opening. Interestingly, the homologous residues to D160 in S1 and R261 in S4 have been shown to be part of the selectivity filter of human Hv1 channels (Musset et al., 2011) (Berger and Isacoff, 2011). Therefore, we speculate that an S1 movement, driven by D171 electrostatic interactions, positions D160 in the appropriate position in the selectivity filter for high proton flux through the channel, and/or that the second conformational change of S4 rearranges the side chain of R261 in the selectivity filter to allow high proton flux through the channel (Fig. 8A open state).
The S4 stabilization conformational changes are >10 times faster than the S4 activation conformational changes. This suggests that the main charge movement of S4 (S4 activation) is the rate limiting step for Hv1 channel opening. We could directly demonstrate the presence of these different steps in the activation process by our triple-pulse protocol, from which the time course of channel opening is either rate limited by the slow rate S4 activation step or by the much faster S4 stabilization step depending on the prepulse protocol (Fig. 8C). The model we propose suggests that the faster and slower components in H+ current deactivation are due to the faster S4 destabilization and slower S4 deactivation conformational changes, respectively. This model, in which the initial outward S4 movement occur independently in the two subunits and in which the outward S4 movement is rate limiting for channel opening, also explain why the current time course is similar to the fluorescence time course to the second power in dimeric Hv1 channels (Gonzalez et al., 2010). This simple model recapitulates the major features of the fluorescence and currents of Hv1 channels (Fig. 8). However, a more complex model with additional transitions is most likely needed to explain the details of the kinetics of Hv1 channels.
The monomeric ΔNΔC channels and the mutated D171A channels still open and display a second fluorescence component, albeit significantly smaller and shifted to more depolarized voltages. This suggests that the electrostatic interaction of D171 across subunits is not the only molecular interactions that drive the second conformational change and channel opening, but that other parts of the protein are also involved in the opening conformational change. There is no evidence of a second component (or hook) in the fluorescence change from the mutant monomer D171A/ΔNΔC (Suppl. Fig. S4), suggesting that in D171A/ΔNΔC channels the fluorophore experiences a different environment during the opening conformational change than in wt channels. This could be due to either that the conformational change is slightly different in this mutation or that the interactions of the fluorophore with the protein are different in this mutation.
In the R255A mutant, the resting position of S4 is different from that in wt channels, as if S4 has already undergone the first charge-moving conformational change (Gonzalez et al., In Press). Indeed, the polarity, voltage dependence, and kinetics of the fluorescence change in the R255A mutant (Fig. 6) are similar to those of the second conformational change in wt Hv1 channels, suggesting that the R255A mutant only undergoes the second conformational change. This conformational change has the same voltage dependence and kinetics as channel opening, consistent with our hypothesis that the second conformational change is closely related to opening and closing of the channel gate. The fact that the current and fluorescence superimpose, and that there is no sigmoidicity in the current from the R255A mutant, supports our model that the second conformational change is a concerted conformational change in the two subunits that opens the two permeation pathways simultaneously.
In conclusion, we have shown here that two distinct conformational changes lead to Hv1 channel opening. The first conformational change is consistent with the previously reported outward movement of the positively charged S4 upon membrane depolarization. The second one is consistent with a concerted conformational change that opens the two permeation pathways in the two subunits in the dimeric Hv1 channel. Our data suggest that this second conformational change is a cooperative conformational change involving interactions between the two subunits of Hv1 channels that are, at least partly, mediated by the S1–S1 interface.
We performed site-directed mutagenesis, in vitro transcription of cRNA, and injection of cRNA encoding the Ciona Ci-VSOP (here called Ci-Hv1) into Xenopus Laevis oocytes as described previously (Gonzalez et al., 2010). The ΔNΔC Ci-Hv1 was constructed with a stop codon at Val270 and initiator methionine replacing Glu129(Gonzalez et al., 2010).
We performed VCF experiments as described previously (Gonzalez et al., 2010). Briefly, we labeled oocytes for 30 min with 100 μM Alexa-488 maleimide (Molecular Probes) in Na+ Ringer’s solution. Fluorescence was monitored through a FITC filter cube: exciter, HQ480/40; dichroic, Q505LP; and emitter, HQ535/50. Fluorescence intensities were low-pass filtered at 200–500 Hz and digitized at 1 kHz. We performed two-electrode voltage clamp (TEVC) recordings as described earlier. Solutions for TEVC contained 88 mM NaCl, 1 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 100 mM HEPES (pH = 7.4). We injected oocytes with 50 nl of 1M HEPES (pH = 7.0) to minimize pH changes due to the proton currents. This results in approximately 100 mM HEPES in the cytosol. We also added 100 mM HEPES (pH = 7.4) to the external solutions for these recordings. Currents were leak subtracted off-line, assuming ohmic leak and using currents from potentials between −80 and −40 mV. To compare the kinetics of the fluorescence (F) and the currents (I), we fit the later half of the current and fluorescence traces to single exponentials and normalized them to their steady-state amplitudes.
We performed excised inside-out patch clamp recordings as described earlier (Gonzalez et al., 2010). Both pipette and bath solutions for excised patch recordings contained in mM: 100 HEPES, 2 MgCl2, and 1 EGTA (pH = 7 with NMDG). 2-Guanidinobenzymidazole (2-GBI: SIGMA) was dissolved in water (10 mM) and then added to the cytosolic solution at 200 μM.
We performed FRET measurements as described previously (Koch et al., 2007; Koch and Larsson, 2005). Briefly, we labelled approximately 20% of the ΔNΔC S242C subunits in an oocyte with the donor fluorophore Alexa488-Maleimide. The donor-only fluorescence was measured on a Leica fluorescence microscope with a FITC filter cube. The oocyte was subsequently labelled to saturation with TMR-MTS (2-((5(6)-tetramethylrhodamine) carboxylamino) ethyl-methanethiosulfonate)) acceptor fluorophore and the donor fluorescence was measured in the presence of the acceptor fluorophore with a FITC filter cube. We determined the FRET efficiency E by the donor quenching method: E = 1- Fdonor(in the presence of acceptor)/Fdonor(in the absence of acceptor) (Koch et al., 2007; Koch and Larsson, 2005).
Fluorescence and currents from the Hv1 models were simulated using Berkeley Madonna (Berkeley, CA). Rate constants for each transitions were of the form ki(V) = ki(0 mV)*exp(−ziFV/RT), where ki (0 mV) and zi were determined by the fits from Figure 7. R, T, and F have their usual thermodynamic meaning. The only free parameter in the simulation was the relative fluorescence intensity of the open state. Note that the rates determined from the fits of the time constants are only approximations of the true rates, because all rates in the model will contribute to the time constants in any macroscopic experiments. However, the experimental protocols used to determine the time constants (Fig. 7) were developed to bias the rates so that mostly one rate would be dominating the time constant (α for τact, β for τdeact, γ for τstb, and γ for τdestb).
We thank Drs. Rene Barro-Soria, Fredrik Elinder, Karl Magleby, Wolfgang Nonner, and Ramon Latorre for comments on the manuscript. This work was partially funded by a grant from NHLBI (R01-HL095920) to HPL, and fondecyt 1120802 and ACT 10224 to CG.
The authors declare no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.