Insights into the molecular determinants of proton-inhibition in an acid-inactivated DEG/ENaC channel

Abstract

Mammalian ASIC channels of the DEG/ENaC superfamily are gated by extracellular protons and function to mediate touch and pain sensitivity, learning and memory, and fear conditioning. The recently solved crystal structure of chicken ASIC1a and preliminary functional studies suggested that a highly negatively charged pocket in the extracellular domain of the channel might be the primary proton binding domain. However, more recent extensive mutagenesis analysis paints a more complex mechanism of channel gating, involving binding of protons at sites immediately after the first transmembrane domain (TM1) and displacement of inhibitory Ca²⁺ ions from the acidic pocket in the extracellular domain and from another Ca²⁺ binding site at the mouth of the pore. We recently identified and functionally characterized C. elegans ACD-1, the first acid-inactivated DEG/ENaC channel. ACD-1 is expressed in C. elegans amphid glia and functions with neuronal DEG/ENaC channel DEG-1 to mediate acid avoidance and chemotaxis to the amino acid lysine. The post-TM1 residues that were proposed to bind protons in ASIC1a are not conserved in ACD-1, but some of the amino acids constituting the acidic pocket are. However, ACD-1 proton sensitivity is completely independent from extracellular Ca²⁺ and protons appear to bind the channel in a less cooperative manner. We thus wondered if residues in the acidic pocket might contribute to ACD-1 acid sensitivity. We show here that while ACD-1 sensitivity to extracellular protons is influenced by mutations in the acidic pocket, other sites are likely to participate. We also report that one histidine at the base of the thumb and residues in the channel pore influence proton-inhibition in a voltage independent manner suggesting that they affect the coupling of proton-binding with the gating rather than proton-binding itself. We conclude that ACD-1 inhibition by protons is likely mediated by binding of proton ions to multiple sites throughout the extracellular domain of the channel. Our data also support a model in which residues in the acidic pocket contribute to determining the channel state perhaps by changing the strength of the interaction between adjacent thumb and finger domains.

Ion channel subunits of the DEG/ENaC family (named after the C. elegans degenerins and the mammalian epithelial Na⁺ channels) are two transmembrane domain proteins positioned in the membrane such that short N- and C-termini are located intracellularly while a large loop protrudes extracellularly (1, 2). DEG/ENaC subunits come together in trimers (3) to form voltage-independent Na⁺ (1, 4) or Na⁺ and Ca²⁺ (5, 6) channels, are found across species, and are implicated in an extraordinary range of biological functions including mechanosensation (7–9), pain sensation (10, 11), thermo-sensation (12), pheromone perception (13), proprioception (14, 15), learning and memory (16) and Na⁺ reabsorption (1). In mammals, DEG/ENaC subunits expressed primarily in neuronal tissues are called ASICs (Acid-Sensing Ion Channels) and are gated by extracellular protons (17). Mammalian epithelially-expressed DEG/ENaCs are called ENaCs, are inhibited by intracellular (18) but not extracellular protons (19) and mediate transport of Na⁺ across epithelia (20–22). We recently reported the characterization of ACD-1, a novel C. elegans DEG/ENaC channel. ACD-1 is expressed in C. elegans glial amphid sheath cells, is inhibited by both intracellular and extracellular protons, and functions with neuronal DEG/ENaC channel DEG-1 to mediate acid avoidance behavior and chemotaxis to the amino acid lysine (23). ACD-1 shares the highest homology with mammalian INaC and BLINaC, two poorly characterized DEG/ENaC subunits expressed in intestine, liver, and brain, and functionally it resembles ENaCs (24, 25).

Recently the crystal structure of chicken ASIC1a channel has been solved (3). It reveals that ASIC1a subunits resemble a “fist” protruding from the plasma membrane (3). The transmembrane domains, post-transmembrane region, and the extracellular domain resemble a forearm, a wrist and a clenched hand respectively. A close look at the electrostatic potential mapped onto the solvent-accessible surface reveals the existence of a negatively charged pocket between the “thumb” and the “palm” domains containing 3 pairs of acidic amino acids that were suggested to coordinate proton binding. Jasti and colleagues indeed demonstrated that mutations of two of these acidic residues (D346 and D350) to asparagine diminish the channel proton sensitivity affecting either the Ki or the cooperativity between protons, or both (3). A recent more extensive mutagenesis study, however has challenged a direct role of the acidic pocket in proton gating (26). Paukert and colleagues, found instead that two histidines (H72 and H73) and an aspartic acid (D78) right after transmembrane domain 1 (TM1) are major contributors to the allosteric effect of proton binding that results in gating of the channel. A role of post-TM1 residues in ASIC1a proton gating was previously suggested by the Canessa lab (27). Paukert and colleagues also reported that residues in the acidic pocket, along with E425 and E432 in the external mouth of the pore, are likely to be involved in binding Ca²⁺, which exerts an inhibitory effect on ASIC1a stabilizing the closed state (26). The authors suggest that protonation of these residues might displace Ca²⁺ favoring channel opening. Another study by Yang and colleagues suggests that the acidic residues of the acidic pocket contribute to stabilize the interaction between the “thumb” and the “finger” domain and that proton binding at these sites may render this interaction stronger thus triggering channel gating (28). The post-TM1 residues that seem to play such a crucial role in proton binding in ASIC1a are not conserved in ACD-1, however some of the residues in the acidic pocket are. Then again, we show here that ACD-1 pH sensitivity is completely independent from extracellular Ca²⁺. Thus we reasoned that ACD-1 structural and functional features may shed light onto the mechanism by which the acidic pocket contributes to channel acid sensitivity.

We show here that despite the fact that the pH sensitivity of ACD-1 is independent from extracellular Ca²⁺, mutations at the acidic pocket influence proton inhibition. We also show that His356 at the base of the “thumb” domain plays a role in proton inhibition and that when mutated to aspartic acid shifts the dose response curve and reduces the cooperativity between protons in inhibiting the channel. When we simultaneously mutated residues in the acidic pocket and His356, we generated a mutant channel that displayed a pH-insensitive residual current. Since ACD-1 is inhibited rather than activated by protons and its pH sensitivity is Ca²⁺-independent, our data suggest that the acidic pocket, perhaps via interaction with the finger domain, might be involved in determining whether the open or the close state of the channel are more stable. We also report that mutations in the pore affect proton inhibition in a voltage-independent manner suggesting that they affect coupling of proton-binding to the gating rather than proton-binding itself.

Materials and Methods

Oocyte expression, electrophysiology and fluorescence

acd-1 cDNA was subcloned in pGEM-HE vector (23). Mutations were introduced by PCR using QuikChange site-directed mutagenesis kit from Stratagene following standard procedures. Capped RNAs were synthesized using T7 mMESSAGE mMACHINE kit (Ambion), purified (Qiagen RNAeasy columns), and run on denaturating agarose gels to check for size and cRNA integrity. cRNA quantification was then performed spectroscopically. Stage V-VI oocytes were selected among multistaged oocytes dissected by 2 hours collagenase treatment (2mg/ml in Ca²⁺-free OR2 solution) from Xenopus laevis ovaries. Oocytes were incubated in OR2 media, which consists of 82.5 mM NaCl, 2.5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 1 mM Na₂HPO₄, 0.5 g/l polyvinyl pyrolidone and 5 mM HEPES (pH 7.2), supplemented with penicillin and streptomycin (0.1 mg/ml) and 2 mM Na-pyruvate. Oocytes were then injected with 69 nl of cRNA mix for a final amount of 5 ng/oocyte of each cRNA. Oocytes were incubated in OR2 plus 500 μM amiloride to inhibit the ACD-1 current and thus prevent Na⁺ overload (29), at 20°C for 2–4 days before recording.

Currents were measured using a two-electrode voltage clamp amplifier (GeneClamp 500B, Axon Instruments) at room temperature. Electrodes (0.2–0.5 MΩ) were filled with 3M KCl and oocytes were perfused with a NaCl solution containing (in mM): 100 NaCl, 2 KCl, 1 CaCl₂, 2 MgCl₂, 10 HEPES. pH was adjusted at the indicated values using NMDG-Cl or HCl. When we tested solutions at pH lower than 6, we used 10 mM 2-(N-morpholino)ethanesulfonic acid (MES) instead of 10 mM HEPES to buffer the solutions. For experiments using 1 mM NaCl the impermeant ion that substituted for Na⁺ was NMDG⁺ (NMDG-Cl 99 mM). We routinely perfused oocytes with solutions at low pH for only 30 seconds, which we determined was the time required to fully switch the solution in the recording chamber at our perfusion rate. Chemicals were obtained from Sigma and Calbiochem. We used the pCLAMP suite of programs (Axon Instruments) for data acquisition and analysis. Currents were filtered at 200 Hz and sampled at 1 kHz.

For fluorescence measurements, oocytes were preloaded with 30 μM BCECF for 30 min in OR2 buffer at room temperature (30). Fluorescence was measured using a Leica upright microscope equipped with a 20× objective, an Oregon Green Filter cube (exciter, HQ495/30X; dichroic, Q515LP; and emitter, HQ545/50m), and a photodiode (pin-020A, UCT Sensors, Inc.) coupled to an Axopatch 200B patch clamp amplifier (Axon Instruments, Inc.). The objective was focused on the animal pole, and fluorescence was monitored with Clampex 8.2 (Axon Instruments, Inc.).

Results

Protons bind away from the ion permeation pathway to inhibit ACD-1 channel activity

When we perfuse a Xenopus oocyte expressing ACD-1 with a physiological solution at pH 7.2 in which the major carrier ion is Na⁺, we detect a large voltage-independent amiloride-sensitive current (23). This current resembles currents produced by expression of other DEG/ENaC channels including mammalian ENaCs and C. elegans MEC-4(d) and UNC-105(d) (1, 29, 31) (Fig. 1A). If we switch to a solution at pH 5, the current is ~ 90% inhibited (Fig. 1B); return to the solution at pH 7.2 recovers the current to its original level (Fig. 1C). Thus, ACD-1 is reversibly inhibited by extracellular acidification (23). Since ACD-1 is inhibited by intracellular acidification as well, we confirmed that our manipulations of extracellular pH did not affect intracellular pH using the pH sensitive fluorescent dye BCECF (30). Figure 1S shows that perfusion of extracellular solutions at pH 6.5 and 5.5 for more than two minutes does not result in significant decrease in intracellular pH.

Figure 1. ACD-1 currents are reversibly inhibited by application of extracellular acidic solutions. Protons bind away from the ion permeation pathway.

(A) Ionic currents recorded in an oocyte injected with acd-1 cRNA, perfused with a physiological NaCl solution. Voltage steps were from −160 mV to + 100 mV from a holding potential of −30 mV. (B) The same oocyte was perfused with a NaCl solution that was adjusted at pH 5 and was then washed with the solution adjusted at pH 7.2 (C). ACD-1 current is nearly completely inhibited by pH 5, but then recovers to control levels once the oocytes is perfused again with the solution at pH 7.2. Dashed line represents zero current level. (D) Current-voltage relationships of ACD-1 currents measured at pH 7.2 and 5. n was 7 in both cases. (E) We plotted the pH that gives half maximal inhibition (Ki) versus the membrane voltage applied. Ki values were obtained from data reported in (23). The Ki does not change with the voltage indicating that protons do not enter the membrane electric field to inhibit the channel. n=9. Data are expressed as mean ± SE.

Analysis of the current-voltage relationship at pHs 7.2 and 5 indicates that ACD-1 current is inhibited by protons in a voltage-independent manner (Fig. 1D), suggesting that protons do not enter the channel pore. To confirm this, we plotted the pH that gives half-maximal inhibition (Ki) against the voltage (23). We found that the Ki does not change with the voltage supporting the notion that protons do not enter the membrane electric field to inhibit ACD-1 and thus must bind to the extracellular domain of the channel (Fig. 1E).

ACD-1 inhibition by acidic solutions is independent from extracellular Ca²⁺ and Na⁺

Channels formed by human α, β and γ ENaC subunits are potentiated rather than inhibited by extracellular protons (32). This potentiating effect is due to protons releasing Na⁺ self-inhibition (32). Conversely, ASIC channels pH sensitivity is affected by extracellular Ca²⁺ (33). We thus wondered if ACD-1 inhibition by protons had anything to do with protons altering the effect of extracellular ions on channel activity. To test this, we first measured ACD-1 acid sensitivity in the absence of Ca²⁺. When we perfused Xenopus oocytes expressing ACD-1 with a solution without CaCl₂, we detected an increase in current amplitude suggesting that Ca²⁺ exerts some type of blocking effect on ACD-1 channel (Fig. 2A), similarly to what we observed for MEC-4(d) and MEC-4(d,G717E) channels (5). However, when we determined pH dose-response curve for ACD-1 currents in the absence of Ca²⁺, we found that it was identical to the pH dose-response curve determined in extracellular 1 mM CaCl₂ (Ki was 6.4 in both cases) (Fig. 2B). These results indicate that protons do not interfere with Ca²⁺ inhibitory effect on ACD-1 channels.

Figure 2. Extracellular Ca²⁺ and Na⁺ do not interfere with proton inhibition.

(A) Current-voltage relationships of ACD-1 currents recorded in physiological saline containing 1 mM CaCl₂ (squares) and in 0 mM CaCl₂ (triangles) (pH was 7.2 in both cases). ACD-1 current increases upon removal of CaCl₂, suggesting that Ca²⁺ inhibits the channel. n is 9 each. (B) pH dose-response curves for ACD-1 currents recorded in physiological solution containing 100 mM NaCl + 1 mM CaCl₂ (grey curve, from (23)), 100 mM NaCl + 0 mM CaCl₂ (filled triangles) and 1 mM NaCl + 1 mM CaCl₂ (open triangles). All three solutions contained also 2 mM KCl, 2 mM MgCl₂ and 10 mM Hepes or MES (see Methods). Since ACD-1 is impermeable to Ca²⁺ and K⁺ ions (23), the permeant ion is Na⁺ in all cases. Note that the reversal potential of ACD-1 currents was −72 ± 4.2 mV (n=13) in 1mM NaCl (not shown) predicting an intracellular concentration of Na⁺ of ~ 16 mM. This low concentration of intracellular Na⁺ was maintained in oocytes expressing constitutively open ACD-1 channels by our incubation of oocytes in OR2 plus amiloride that prevented Na⁺ overload (see Material and Methods). Data were fitted using the Hill’s equation. Currents were normalized for the maximal current (usually at pH 7.2 or 8) at −160 mV for all the pH dose-response curves presented in all the figures. Ki values were not statistically different and they were 6.37, 6.45 and 6.47 respectively. n is 8 each. Data are expressed as mean ± SE.

We next tested the effect of Na⁺ and measured ACD-1 pH sensitivity in 100 and 1 mM NaCl. Again, dose response curves were identical in these two conditions suggesting that Na⁺ ions do not interfere with protons (Fig. 2B).

The “acidic pocket” and acid sensitivity

To determine if C. elegans proton-inactivated ACD-1 channel coordinates proton binding using residues corresponding to the ones that were proposed to bind protons in ASIC1a (3, 26), we threaded ACD-1 sequence onto chicken ASIC1a crystal structure using CPHmodels server (http://www.cbs.dtu.dk/services/CPHmodels/). We found that none of the relevant residues in the post-TM1 extracellular domain are conserved in ACD-1 and that at 4 of the 6 positions in which acidic amino acids are found in ASIC1a extracellular acidic pocket, ACD-1 encodes neutral amino acids. However, at position 419 (corresponding to position 346 in chicken ASIC1a), and at position 304 (corresponding to position 238 in chicken ASIC1a) ACD-1 encodes aspartic acids (3) (Fig. 3A). When we looked at the location of these two acidic residues mapped on the surface of ACD-1, we noted that residues D304 and D419 are in the “palm” and “thumb” domains respectively but are next to each other and form an acidic pocket/surface resembling the acidic pocket found in ASIC1a (Fig. 3B). To determine if this acidic pocket mediates proton binding in ACD-1, we mutated D419 and D304 to N and assayed channel function. Surprisingly, we found that mutating D419 to N increases rather than decreases ACD-1 proton sensitivity (Ki from 6.4 to 7.3, Fig 3C and (23)). By fitting experimental data points with the Hill’s equation we found that the n coefficient, indicative of cooperativity between protons as they inhibit the channel, decreases from 0.9 to 0.7 in the mutant channel, suggesting increased negative cooperativity between protons. Parenthetically, protons bind ASIC1a in a highly cooperative fashion as the n coefficient in chicken ASIC1a is 8 (3). Our results suggest that mutation D419N may interfere with the interaction between the “thumb” and the “finger” domain perhaps favoring the closed state at higher pHs. However, we cannot exclude the possibility that D419N mutation causes rearrangements of other charges, resulting in increased net negative charges available to protons for binding. We next analyzed ACD-1(D304N) and ACD-1(D304N,D419N). We found that while ACD-1(D304N) channel is inhibited by acidic solutions similarly to wild type, the double mutant channel has a reduced sensitivity to protons (Fig. 3C). Because D304N has unaltered acid sensitivity it is possible that D419 interacts with another residues in the “palm” domain (based on the crystal structure candidates are not immediately apparent). However, our analysis of the double mutant channel suggest a role for both residues in channel acid sensitivity. We conclude the acidic pocket formed by residues D304 and D419 participate to proton inhibition, but that other sites are likely involved since ACD-1 double mutant channel is still sensitive to acidic solutions.

Figure 3. Properties of mutants in the “acidic pocket”.

(A) ACD-1/ASIC1a amino acids alignments highlight the only two conserved acidic residues in ACD-1 “acidic pocket”. D304 corresponds to D238 in chicken ASIC1a, and D419 corresponds to D346 (blue boxes). Chicken ASIC1a has other 4 acidic residues including D350 (green box) that form with D238 and D346 negatively charged pocket on the surface the protein. (B) The surface of one ACD-1 channel subunit threaded onto chicken ASIC1a structure is shown here (3). The other two subunits were removed for clarity and they would be to the right of the one showed. Acidic residues throughout the protein are shown in yellow, histidines are shown in orange, in cyan the two conserved acidic residues D304 and D419. (C) pH dose-response curves for wild type ACD-1, ACD-1(D304N), ACD-1(D419N), and ACD-1(D304N,D419N). Wild type and ACD-1(D419N) dose-response curves were already reported in (23). Data were fitted using the Hill’s equation, Ki are shown on the graph, n was ~ 1 except for ACD-1(D419N) and ACD-1(D304N,D419N), in which it was 0.7. n is 5 each. Average current ratios at pH 6.5 and lower were statistically different by t-Test (p<0.01) between ACD-1(D304N,D419N) and wild type or ACD-1(D304N). (D) Voltage-dependence of the pH that gives half maximal inhibition (Ki) for wild type ACD-1 (grey line), ACD-1(D304N) (filled squares), ACD-1(D419N) (triangles) and ACD-1(D304N,D419N) (open squares). Note that the Ki becomes voltage-dependent at positive voltages in ACD-1(D304N), suggesting that this mutation induces a conformational change in the channel that exposes residues involved in proton binding at more positive potentials. Fit was by linear regression, except for ACD-1(D304N) for which we used the Woodhull equation for voltages between +20 and +100 mV (48). Data are expressed as mean ± SE.

Interestingly, when we analyzed the voltage-dependence of proton block in ACD-1(D304N) we found that while its proton sensitivity is like wild type at negative voltages, it increases as the voltage becomes more positive (Fig. 3D). Since amino acid 304 is not predicted to be in the ion permeation pathway and proton sensitivity increases with depolarization rather than with hyperpolarization, this result suggests that mutations at this site may cause novel residues to move or unmask residues moving out the membrane electric field as the membrane potential becomes more positive. This result thus suggests that at least some of the residues involved in ACD-1 acid inhibition may be located close to the plasma membrane.

Lastly, we wondered if other acidic residues in the “thumb” domain, postulated to play a major role in ASIC1a gating (3, 34), may participate to conferring acid sensitivity to ACD-1. We focused on E428 and E433 because they are the only other two acidic residues that form a clear acidic surface in the “thumb” domain of ACD-1. It is noteworthy however that these two acidic residues are within the “thumb” domain and not between the “thumb” and the “finger” domains. We found that ACD-1(E428N,E433N) double mutant channels are inhibited by protons just like wild type (Suppl. Fig. 2), suggesting that they do not play any role in ACD-1 acid sensitivity.

A histidine at the base of the “thumb” influences ACD-1 acid sensitivity

Our results suggested that residues close to the plasma membrane may be implicated in ACD-1 inhibition by protons. Based on ASIC1a crystal structure, Jasti and colleagues (3) proposed a mechanism of proton-regulated channel gating involving residues at the interface with the plasma membrane. They proposed that Trp 288 in the “thumb” acts like a ball sitting in a socket of the transmembrane domain. They suggested that this residue may couple the movement of the “thumb” caused by protons binding onto the extracellular domain to the movement of the transmembrane domain ultimately resulting in gating. Recent data from the Canessa’s lab confirm this prediction (34). When we looked at ACD-1 sequence we noticed that ACD-1 also encodes a tryptophan at this site. A histidine (His356), a residue that can be protonated, is two residues upstream of this tryptophan and thus it is predicted to be at the interface with the plasma membrane (Fig. 4A). We thus wondered if this histidine may be involved in proton binding or in coupling of proton binding to channel gating.

Figure 4. Role of a histidine at the base of the “thumb” in proton sensitivity.

(A) Secondary and tertiary structure organization of one ACD-1 subunit: alpha helixes are in red, beta sheets in yellow and turns in green (PyMol). The ellipse highlights the thumb domain. At the basis of the thumb histidine 356 sticks out facing away from the ion permeation pathway. A tryptophan in the thumb domain and a tyrosine in the first transmembrane domain are thought to form π–π interaction that stabilize the subunit structure and is crucial for proton-activated gating in ASIC1a (34). These two residues are present in ACD-1 as well (in black) although slightly shifted. (B) pH dose response curves for ACD-1 histidine mutants and D304N/H356D/D419N triple mutant. Data points were fitted using Hill’s equation. Ki values are shown on the graph. The grey line represent fit for wild type ACD-1. Note that at pH 3.5 14% of the current carried by the triple mutant channel is still present. We confirmed this by applying 500 μM amiloride which completely suppressed the current. n is 5 to 10. Data are expressed as mean ± SE. Statistical analysis by t-Test confirms that current ratios of ACD-1(H356D) and ACD-1(D304N,H356D,D419N) are statistically different from wild type (p<0.01), wild type values were obtained from (23).

We mutated His 356 to the neutral amino acid alanine, to the negatively charged amino acid aspartate, and to the positively charged arginine. We found that that introduction of a negative charge at this site had a profound effect on the apparent Hill’s coefficient n and a small effect on the Ki. While elimination of a potential protonation site and introduction of a “permanent” positive charge had no effect on channel function or acid inhibition (Fig. 4B). This result suggests that protonation of His 356 is not needed for channel inhibition by protons. This result does not support our initial prediction, however, our data suggest a role for this residue in channel inhibition since a negatively charged residue at this site interferes with ACD-1 acid sensitivity.

Since we observed the strongest effect on sensitivity to protons in ACD-1(D304N,D419N) and ACD-1(H356D) mutant channels, we decided to introduce all three mutations in ACD-1 at once to analyze the channel sensitivity to acidic solutions. Interestingly, we found that while the Ki of the triple mutant is similar to the Ki of the double mutant (pH 5.7 for the triple mutant and pH 5.6 for the D304/D419N double mutant), the negative cooperativity between protons in exerting their inhibitory effect on the channel is increased in the triple mutant (n is 0.45 in the triple mutant and 0.7 in the D304/D419N double mutant). In fact the Hill coefficient of the triple mutant is similar to the Hill coefficient of H356D (n is 0.5). Another difference we noticed is that while pH 4 is sufficient to fully inhibit all the other mutants we studied, pH 3.5 does not completely inhibit the triple mutant channel: at pH 3.5 14% of ACD-1(D304N,H356D,D419N) current is still present (Fig. 4B). We could not test lower pHs because they affected the oocytes’ health and endogenous current, however the fact that at pH 4 the effect appears saturated suggests that a portion of the current in ACD-1 triple mutant is pH insensitive. Interestingly, this residual pH insensitive current was not observed in the double mutant or H356D, suggesting that the effect of these three combined mutations is not additive, but rather synergistic.

Contribution of the second transmembrane domain to acid inhibition

Residues in the second transmembrane domain (TM2) of DEG/ENaCs have been implicated in controlling channel gating and ion selectivity (2, 29, 35–41). Yang and colleagues have shown that mutations at specific residues within TM2 result in non-functional ASIC1a channels suggesting that they may interfere with proton gating, however other explanations for lack of channel activity may exist (28). ACD-1 is inhibited rather than opened by protons, we thus reasoned that this feature may help shed light on the role of TM2 residues on proton gating, because lack or reduced effect of protons should not result in non-functional ACD-1 channels. We focused on three residues: 1) S513, corresponding to Alanine 713 in MEC-4 (the (d) mutation), that when mutated to bulkier residues including Val or Thr, hyperactivates the MEC-4 channel (2, 31, 37, 39, 42); 2) G516, that forms a kink in ASIC1a TM2 that constricts the ion permeation pathway (3) and that when mutated to glutamic acid in MEC-4 introduces a binding site for extracellular Ca²⁺ (5), and 3) I530 that participates to conferring acid sensitivity to β ENaC in a channel complex formed by α and β only (43) (Fig. 5A and B).

Figure 5. Contribution of pore residues to inhibition of ACD-1 by protons.

(A) The surface of an ACD-1 subunit is shown here. Yellow and orange are for acidic residues and histidines respectively. The subunit is rotated ~ 180° compared to Figures 3B and 4 A. In purple the residues that we mutated in the second transmembrane domain and that line the ion permeation pathway. (B) Amino acids alignment of the second transmembrane domain of mammalian β and γ ENaC, and C. elegans ACD-1. The β ENaC sequence delineated by the box is needed in β/γ chimeras for ENaC channel inhibition by extracellular protons (43). (C) pH dose response curves for ACD-1 pore mutants. Fitting was with the Hill’s equation. Data are expressed as mean ± SE, statistical analysis by t-Test (p<0.01) confirms that current ratios of ACD-1(S513V) and ACD-1(G516E) are statistically different from wild type for pH 6.5 and lower. The grey line represent fit for wild type ACD-1. Ki values are shown on the graph. n is 5 to 9. (D) Voltage dependence of the pH that gives half maximal inhibition (Ki) for ACD-1(S513V) (n=6), ACD-1(G516E) (n=6) and ACD-1(I530V) (n=5). In grey wild type ACD-1 voltage sensitivity of proton block. Data are expressed as mean ± SE and were fitted by linear regression.

We found that introduction of a valine at the (d) site significantly shifts the pH dose-response curve towards more acidic pH (Ki is 5.5 instead of 6.4), indicating that higher protons concentration is needed to close this mutant channel (Fig. 5C). It is noteworthy that ACD-1(d) substitution (S513V) does not cause channel hyperactivation as the channel seems to be fully activated in its wild type form (23). A smaller shift towards more acidic pH was observed for G516E mutant, suggesting that this residue is also part of the proton-gate (Fig. 5B). Interestingly, we did not observe activation of the Xenopus oocytes endogenous Ca²⁺-activated Cl⁻ current, that we used in the past as a measure of MEC-4(d) Ca²⁺ permeability (5), when we perfused oocytes expressing ACD-1(G516E) with a CaCl₂ solution. These results indicate that this substitution alone does not introduce a Ca²⁺ binding site in ACD-1, (not shown, (5)). Finally, I530V substitution does not change the pH dose-response curve, suggesting either that other residues in mammalian β ENaC subunits participate with this site to control the gate (43), or that ACD-1 proton-gate differs substantially from β ENaC’s. Importantly none of the mutations we engineered introduces voltage-dependence to the sensitivity to protons (Fig. 5D), indicating that they do not introduce or expose novel proton binding sites in the ion permeation pathway. Rather, these mutations are likely to affect the coupling of the proton binding with the gate. Of note is that all the mutants we analyzed displayed the same time and voltage-independence of WT channels, as well as similar current amplitudes (except for G516E as we report in the manuscript); they also did not acquired Ca²⁺ permeability (not shown).

Discussion

This study shows that two acidic residues in the extracellular domain and a histidine at the base of the “thumb” domain influence proton gating in the C. elegans acid-inactivated DEG/ENaC channel ACD-1. We also report that residues in the second transmembrane domain possibly mediate coupling of proton binding to channel gating and that, contrarily to ASICs and ENaCs, extracellular sodium and calcium do not interfere with proton inhibition of ACD-1.

The base of the “thumb” and channel gating in the DEG/ENaC channel class

The solution of ASIC1a crystal structure predicted that amino acids at the base of the “thumb” domain would be essential for coupling proton binding to channel gating. Indeed, the base of the thumb is shaped like a tight loop formed by three consecutive prolines, that reaches back towards the plasmamembrane almost touching it. It was thus postulated that this loop would interact with the first transmembrane domain of ASIC1a. Specifically Trp 288 was predicted to form non covalent interaction with sites in TM1 (3, 28). Recent work from the Canessa’s lab has indeed demonstrated that tryptophan 288 forms non-covalent interaction with tyrosine 72 in TM1 and that such interaction is essential for proton-mediated gating: substitution of either amino acids results in shift of proton sensitivity towards extremely acidic pH essentially resulting in the channel not being gated by testable pHs (below pH 4) (34).

ACD-1 encodes a tryptophan at position 358 and a tyrosine at position 127 in TM1 suggesting that the interaction between the base of the thumb domain and TM1 is conserved in C. elegans ACD-1 as well. Two residues upstream to Trp 358, histidine 356 sticks out of the protein structure facing away from the central axis of the channel. Our analysis of substitutions at this position suggests that while a potential protonation site is not essential at this position, a negatively charged residue at this site interferes with coupling of protons binding to channel inhibition. This interference may be a direct effect on the interaction between Trp358 and Tyr72, or an effect mediated by the closely located plasma membrane (for example through repulsion from the negative surface charges of the plasma membrane). Our results confirm that the base of the “thumb” domain plays a key role in gating in the channel class.

Residues lining the pore participate to coupling of protons binding to channel gating

Since the identification of neurotoxic MEC-4 mutant MEC-4(S713V or T), residues in the second transmembrane domain of DEG/ENaC channels have been implicated in gating (2). MEC-4(d) channels are hyperactivated with channel open probability increasing from less than 0.05 in wild type MEC-4 to 0.5 in MEC-4(d) (42), resulting in larger whole cell currents (29). MEC-4, which mediates gentle touch sensation in C. elegans (7), induces neurodegeneration when it becomes hyperactivated (2) probably as a result of Na⁺/Ca²⁺ overload (5). Other members of the DEG/ENaC family including C. elegans UNC-105 and mammalian MDEG, hINaC and BLINaC are hyperactivated by analogous substitutions (24, 25, 31). We previously reported that introducing a Val at the corresponding residue in ACD-1 (ACD-1(S513V)) does not cause channel hyperactivation (23). However, we show here that ACD-1(S513V) channels are less sensitive to extracellular protons. Similarly, G516E substitution shifts the pH dose response curve towards more acidic pHs. Based on threading of ACD-1 sequence onto ASIC1a crystal structure, S513 faces the ion permeation pathway, whereas G516 is facing the inside of the alpha helix. It is possible that substituting S513 with a non-polar residue and G516 with a polar one causes the alpha helix to become enough distorted to affect coupling of the proton binding to the gate. In support of this hypothesis we find that ACD-1(G516E) has reduced whole-cell currents (5% of wild type ACD-1 currents, data not shown). To conclude we have identified two residues in ACD-1 TM2 that are important for coupling proton binding to channel inhibition. Since in previous work, role of residues in TM2 in coupling of proton binding to gating was inferred based on lack of proton gating and thus activity in TM2 mutants (26), we believe that our study provides further and clearer evidence that residues in TM2 couple proton binding to gating.

The extracellular domain and its influence on channel inhibition by protons

Our work shows that the two conserved acidic residues in ACD-1 (D304 and D419) in the extracellular domain of ACD-1 participate in proton gating. However, our results also suggest that other residues are involved, since neutralization of both charges does not abolish ACD-1 sensitivity to protons. Similarly, Li and colleagues recently showed that neutralization of all the negative charges in ASIC1a proton sensor does not eliminate ASIC1a sensitivity to protons (34). The post-TM1 residues that seem to play a big role in proton gating in ASIC1a possibly binding protons themselves (H72, H73 and D78) are not conserved in ACD-1, excluding the involvement of these residues in ACD-1 acid sensitivity. ACD-1 has 40 negatively charged amino acids, and 9 histidines in the extracellular domain respectively, it will be interesting to determine if a discrete number of these charges bind protons to influence channel function, or if the effect of protons is mediated by their interaction with all these sites. Because residues D304, D419 and H356 are located far away from the channel gate, our results and work from other laboratories draw a picture of a highly dynamic extracellular domain of DEG/ENaC channels when they interact with their ligands (3, 28, 34, 44).

Our previous work on homologous C. elegans channel MEC-4, had also suggested that large conformational changes occur in the extracellular/transmembrane domain of the channel during gating (44). We indeed, showed that mutation A149V in MEC-4 palm domain causes hyperactivation of channels when combined with MEC-10(d), while MEC-10(d), MEC-4(A149V) and MEC-10(d)/MEC-4 channels are not hyperactivated. These data suggest that changes of the extracellular domain structure can have profound effects on how residues located in the transmembrane domain affect gating.

ACD-1 structure/function and implications for its role in C. elegans sensory perception

ACD-1 is expressed in C. elegans amphid sheath cells where it functions to coordinate acid avoidance behavior and chemotaxis to lysine-acetate (23). Because ACD-1 is sensitive to both extracellular and intracellular acidification and acid avoidance and chemotaxis to lysine-acetate are expected to cause extracellular and intracellular acidification respectively, we previously proposed a direct link between ACD-1 acid sensitivity and C. elegans behavior. In support of our model, are our data showing that a higher percentage of C. elegans deg-1;acd-1 double mutant animals expressing ACD-1(D419N) escape acidic environments as compared to wild animals or deg-1;acd-1 mutants expressing wild type ACD-1 (23). However, the pH that causes half maximal inhibition of ACD-1 currents in Xenopus oocytes and that induces escape behavior in 50% of the animals do not match (pH 6.4 versus 4 (23)).

So, why ACD-1 channel reconstituted in Xenopus oocytes is more sensitive to acidic solutions than animals are? One possibility is that other DEG/ENaC channels may be expressed in amphid glia and may form with ACD-1 heteromultimeric channels with lower pH sensitivity than homomeric ACD-1 channels. This scenario would not be unprecedented since other DEG/ENaC channels have been shown to have different pH sensitivity depending on whether they are assayed in expression systems or in native tissue (see supplementary table in (45)). Another possibility is that C. elegans endogenous accessory subunits tune ACD-1 acid sensitivity to more acidic pHs in vivo. For example, the stomatin-like protein MEC-2 interacts with the transmembrane domains of MEC-4 affecting the permeability properties and activity of the channel (29), and the paraoxonase-like protein MEC-6 enhances MEC-4 channel activity by direct binding to MEC-4 (46). MEC-2 and MEC-6 do not change ACD-1 properties in Xenopus oocytes (not shown) and are not expected to be expressed in amphid glia (46, 47); however other 7 stomatin-like proteins are encoded by the C. elegans genome and their expression pattern and function are currently unknown. Our structure function analysis suggests that accessory or other DEG/ENaC subunits that interact with residues D304, H356, D419, S513 and G516 may have profound effects on ACD-1 pH sensitivity.

Abbreviations

DEG/ENaC channels

degenerins and the mammalian epithelial Na⁺ channels

ASICs

acid sensitive ion channels

ASIC1a

acid sensitive ion channel 1a

ACD-1

acid sensitive channel degenerin-like

INaC

intestinal Na⁺ channel

BLINaC

brain, liver, intestine Na⁺ channel

TM1

transmembrane domain 1

TM2

transmembrane domain 2

NMDG

N-methyl-D-glucamine

BCECF

2′–7′-bis(carboxyethyl)-5(6)-carboxyfluorescein