Leu85 in the β1-β2 Linker of ASIC1 Slows Activation and Decreases the Apparent Proton Affinity by Stabilizing a Closed Conformation

Acid-sensing ion channels (ASICs) are proton-activated channels expressed in neurons of the central and peripheral nervous systems where they modulate neuronal activity in response to external increases in proton concentration. The size of ASIC1 currents evoked by a given local acidification is determined by the number of channels in the plasma membrane and by the apparent proton affinities for activation and steady-state desensitization of the channel. Thus, the magnitude of the pH drop and the value of the baseline pH both are functionally important. Recent characterization of ASIC1s from an increasing number of species has made evident that proton affinities of these channels vary across vertebrates. We found that in species with high baseline plasma pH, e.g. frog, shark, and fish, ASIC1 has high proton affinity compared with the mammalian channel. The β1-β2 linker in the extracellular domain, specifically by the substitution M85L, determines the interspecies differences in proton affinities and also the time course of ASIC1 macroscopic currents. The mechanism underlying these observations is a delay in channel opening after application of protons, most likely by stabilizing a closed conformation that decreases the apparent affinity to protons and also slows the rise and decay phases of the current. Together, the results suggest evolutionary adaptation of ASIC1 to match the value of the species-specific plasma pH. At the molecular level, adaptation is achieved by substitutions of nonionizable residues rather than by modification of the channel proton sensor.

Acid-sensing ion channels (ASICs) 2 are proton-activated sodium channels (1) expressed in the nervous system of species from the vertebrate lineage. ASIC1 modulates synaptic transmission (2) and other functions by depolarizing the plasma membrane and subsequently allowing influx of calcium into neurons (3).
Three identical or homologous subunits associate to form functional channels (4). Each subunit has a large extracellular domain that contains the proton sensor. The identity and location of the proton sensor remain elusive, but it is generally accepted that ionizable residues, most likely negatively charged, constitute the proton binding sites that control opening of the channel.
Two properties define the magnitude of the ASIC1 response to changes in the concentration of external protons. These are the proton affinity for half-maximal activation or pH 50A , and the proton affinity for half-maximal steady-state desensitization or pH 50D . The value of the former determines the size of the peak current evoked by a rapid increase in the concentration of external protons, whereas the latter determines the fraction of channels available for activation at any given resting pH. In mammals, the pH 50A of ASIC1 is in the range of 6.4 to 6.6, and the pH 50D is ϳ7.2 (5). The steady-state desensitization of the mammalian ASIC1 exhibits a high degree of cooperativity to protons with a Hill coefficient Ն6; thus, a small decrease in external pH such as 0.1 units can lead to inactivation of as much as 80% of channels. These observations underscore the importance of baseline pH in determining the magnitude of the ASIC1 response to local acidification. However, physiological values of baseline external pH are not the same in all vertebrates but spread over a wide range from 7.38 to 8.0, raising the question of whether species with high plasma pH control ASIC1 responses in the same way as mammals. Although, in general, species with high plasma pH also have lower temperature, lower partial pressure of CO 2 , and lower sodium concentration than mammals (6), all of which change the degree of ionization of water and of charged groups in proteins, it remains to be shown whether the compound effect of those factors is sufficient to render the ASIC1 response to equivalent levels across species.
In this work, we examined the proton response of ASIC1 from two species with high arterial pH: frog and elephant shark. We found indeed that ASIC1s from these species have higher proton affinities than the mammalian channel and that residues underlying the differences in proton affinities are in the ␤1-␤2 linker of the extracellular domain rather than in the proton sensor. The mechanism consists in stabilization of a closed conformation by the ␤1-␤2 linker specifically; the substitution M85L decreases the apparent proton affinity and slows the rise and decay phases of the currents by delaying channel openings.

EXPERIMENTAL PROCEDURES
Cloning of ASIC cDNAs-Cloning of ASICs were conducted by a combination of RT-PCR and 5Ј and 3Ј extension as described previously (7). Total RNA was extracted from frog brain with TRIzol reagent and used for synthesis of single strand cDNA with SuperReverse transcriptase III (Invitrogen) and oligo(dT) priming. The full-length cDNAs were ligated into pCDNA3.1 vector (Invitrogen), which added a V5 epitope tag at the C terminus. High fidelity Taq polymerase (Invitrogen) was used in all experiments. DNA of eight independent clones from two different RT-PCR reactions were sequenced in both directions by the Keck facility at Yale University. Elephant shark ASIC was cloned by RT-PCR from a brain cDNA library provided by Dr. Byrappa Venkatesh (Institute of Molecular and Cell Biology, Biopolis). We used specific N and C termini DNA primers designed to match sequences from the DNA genomic data bank of elephant shark (8). The use of frogs was in accordance to guidelines approved by Institutional Animal Care and Use Committee of Yale University.
Mutagenesis-Point mutations were introduced into cDNAs using QuikChange (Agilent). All products were sequenced to confirm the presence of mutations.
Biotinylation of Proteins and Western Blotting-Fifteen oocytes per conditions were biotinylated with sulfo-NHS-SSbiotin (Pierce) on ice. After lysis of cells with 1% Triton X-100, the biotinylated proteins were recovered with streptavidin beads (Pierce). The beads were washed with 1% Triton X-100 and 300 mM NaCl. Biotinylated proteins were eluted from beads, resolved in 10% SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Immobilon). ASIC was detected with anti-V5 monoclonal antibody (Invitrogen). Sig-nals were developed with ECLϩ (Amersham Biosciences) and exposed to BioMax MR film (Eastman Kodak).
Two-electrode Voltage Clamp-Xenopus laevis oocytes were injected with 5 ng cRNA in a volume of 50 nl and were incubated at 16°C for 2-3 days. For two-electrode voltage clamp experiments, oocytes were placed in a recording chamber (400 l) perfused by gravity at a rate of 4 ml/min. Oocytes were impaled with two glass microelectrodes filled with 3 M KCl having resistance Ͻ1 megohm. Membrane potential was held at Ϫ60 mV, and whole-cell currents were recorded with a clamp OC-725B (Warner Instrument Corp., Hamden, CT) and digitized at a sampling rate of 2 kHz (PowerLab 4/30, ADInstruments). Composition of the standard bath solution was as follows: 120 mM NaCl, 2 mM KCl, 1.5 mM CaCl 2 , and 15 mM HEPES-MES. The pH was adjusted to obtain a series of solutions differing by 0.1 pH units from pH 8.0 to 5.0. The same set of solutions was used for all experiments to minimize variations owing to calibration of pH. All experiments were conducted at room temperature. Apparent proton affinities were calculated with Equation 1 as shown below.
Time constants for the rise and decay phases were obtained by fitting the time course of currents to the function (Equation 2), where I o is a scaling factor.
Patch Clamp Recordings in the Outside-out Configuration-Patch pipettes were pulled from PG150T glass (Warner Instruments) to a tip diameter of 2-5 m after heat polishing. Pipette solution was as follows: 120 mM KCl, 5 mM EDTA, and 10 mM HEPES titrated with KOH to pH 7.4. Bath and activating solutions were as follows: 120 mM NaCl, 2 mM KCl, 1.5 mM CaCl 2 , and 15 mM HEPES-MES titrated from pH 8.0 to 5.0. Activating solutions were applied using a modified Perfusion Fast Step device SF-77B (Warner Instruments) onto which eight perfusion pipettes were mounted. The data acquisition program Pulse in turn controlled the SF-77B. Time resolution of the solution exchange device is 0.5 ms as described previously (9). Recordings were made using a EPC-9 amplifier and the Pulse acquisition program (HEKA Electronic). Membrane potential was held at Ϫ60 mV. Experiments were conducted at room temperature.

RESULTS
Cloning of X. laevis ASIC1-We cloned two cDNAs from brain X. laevis that are 94% identical and encode proteins that differ only in 17 out of 525 amino acids, including one deletion. The relation of the frog proteins to other ASICs is shown in the tree of Fig. 1A, and the amino acid sequences are presented in supplemental Fig. 1A. We refer to the frog proteins as xASIC1.1 and xASIC1.2. They are products of different genes and not the result of differential splicing because the amino acid differences are scattered in the whole cDNA rather than being restricted to one exon. The presence of two copies of the same gene and expression of the corresponding proteins in brain of X. laevis is the consequence of a previous tetraploidization event dated ϳ30 million years ago (10).
The protein cloned from elephant shark, Callorhinchus milii, (8) is 79% identical to frog and 77% identical to rat ASIC1a and corresponds also to ASIC1 as indicated in Fig. 1A. The amino acid sequence is shown in supplemental Fig. 1A.
Xenopus ASIC1 Has High Proton Affinity for Steady-state Desensitization-Expression of xASIC1.1 and xASIC1.2 in oocytes produced inward currents evoked by pH 6.0 of mean peak amplitude 20 Ϯ 7 A/cell (n ϭ 44). The magnitude and time course of currents from the two channels were indistinguishable. Fig. 1B shows representative examples of whole-cell currents of xASIC1.1 and xASIC1.2 evoked by increasing concentrations of protons from pH 7.2 to 6.0. The apparent proton affinity for activation (pH 50A ) was measured by preconditioning cells with pH 8.0 followed by stimuli of pH 7.2 to 5.5. The apparent proton affinity for activation was calculated with the peak currents giving values for pH 50A of 7.0 and a Hill coefficient of 3. (Fig. 1C). The apparent affinity of protons for steady-state desensitization pH 50D , i.e. desensitization from the closed state, was measured by lowering the preconditioning pH by 0.1 units starting from pH 8.0 to 6.8. The calculated pH 50D had values of 7.8 for both xASIC1.1 and xASIC1.2. To contrast the differences with rat ASIC1a, the graph in Fig. 1C also shows the corresponding proton affinities of rat ASIC1a measured under identical conditions: pH 50D 7.2 and pH 50A 6.3. Thus, frog ASIC1 has higher affinity for protons than the mammalian ASIC1 channel.
Amino Acid Composition of the ␤1-␤2 Linker Changes ASIC1 Apparent Affinity to Protons-The determinants of proton sensitivity of ASIC1 currently are not known, but it usually is assumed that differences in charged residues in the extracellular domain underlie the variations in proton affinities. Comparison of negatively and positively charged amino acids of frog and rat ASIC1 proteins showed a high degree of conservation, suggesting that ionizable residues do not account for the differences in proton sensitivity. We examined instead the ␤1-␤2 linker that is a short segment in the extracellular domain that connects the ␤1 with the ␤2 strands. We have shown previously that the ␤1-␤2 linker plays an essential role in conferring proton sensitivity to lamprey ASIC1 (11) and also modifies the rate of decay of fish ASIC1 currents (12). The amino acid composition of the ␤1-␤2 linker of xASIC1.1 is Pro 83 -Lys 84 -Met 85 , whereas that of rat ASIC1a is Ser 83 -Gln 84 -Leu 85 (Fig. 1D). These three residues were substituted in frog to generate the triple mutant xASIC1.1-P83S-K84Q-M85L (SQL). The mutations decreased proton affinity for steady-state desensitization; the apparent pH 50D was shifted from 7.8 to 7.35, whereas the pH 50A changed from 7.0 to 7.1 (Fig. 1E). These results were duplicated by the single substitution M85L, indicating that position 85 is the main contributor of the three residues in the ␤1-␤2 linker.
Residues in the ␤1-␤2 Linker Slow the Time Course of xASIC1 Currents-Macroscopic currents of xASIC1.1 rapidly increase and decay upon proton stimulation. Fig. 2A contrasts the time course of normalized frog and rat ASIC1a (rASIC1a) currents evoked by pH 6.8 recorded under identical conditions. Frog and rat channels completely desensitize in the continuous presence of external protons, but both the rise and the decay phases are faster in xASIC1.1 than in rASIC1.
Analysis of xASIC1.1 mutants showed that the substitutions in the ␤1-␤2 linker also slow channel kinetics. Traces in   presented in an expanded time scale (Fig. 2C). Of the three single substitutions, M85L has the largest effect, but together, they produce a further small additive effect. To examine whether other residues have a similar effect in position 85, we tested the amino acids indicated in Table 1. Currents of single mutants were examined at pH 7.1. The time constants for the rise ( a ) and the decay phases ( d ) were derived from the fit of whole-cell currents to Equation 2.
Only leucine slowed the currents; all other residues either did not change significantly the time constants or decreased the values, threonine and phenylalanine. Currents produced by M85R were too small to be examined.

Delay of the First Opening Underlies the Slow Time Course xASIC1-SQL Channels-
To investigate the mechanism underlying the functional changes produced by substitutions in the ␤1-␤2 linker, we examined xASIC1.1 wild type and the mutant SQL in outside-out patches exposed sequentially to a series of activating solutions with increasing concentrations of protons (pH 7.4, 7.3, 7.2, 7.1, 7.0, 6.8, and 6.0). From wild type xASIC1.1, we recorded 10 to 20 sweeps of 4-s exposure to the activating pH and 8-s recovery at pH 8.0. For mutant channels, owing to the slow time course of the currents, the duration of the sweeps was lengthened to a 20-s exposure to activating pH, followed by a 10-s recovery. Representative examples of xASIC1.1 wild type (Fig. 3A) and xASIC1.1-SQL (Fig. 3B) currents evoked by increasing concentration of protons show the slow time course of the mutant currents compared with wild type particularly in the pH range from 7.3 to 7.1. Time constants of activation a and desensitization d calculated at several proton concentrations are presented in Table 2. The constants exhibit proton dependence in the pH range from 7.3 to 6.0. At pH Յ 6.0, the a reached a maximal value without further decrease, suggesting that at very high concentrations of protons, when diffusion rate is not a factor, the activation rate saturates in both channels.
Analysis of patches indicates that the fraction of channels opening late after application of the stimulus and the length of the delay decreased as the concentration of protons increased (Fig. 3, C and D). At very high proton concentrations, pH Յ6.5, almost all channels open immediately after the stimulus producing a sharp increase in the inward current and reducing the duration of the sweep, black traces in Fig. 3, A and  B. The main difference observed in the mutant ASIC1.1-SQL is a larger delay in first openings (Fig. 3D) that markedly broadens the peak current and slows the time course of the current produced by the ensemble of channels. Consequently, it increases the values of both a and d constants.
The Substitution M85L Stabilizes a Closed Protonated Conformation-To understand the effect of the ␤1-␤2 linker in the reaction cycle of ASIC1, we considered the simplest scheme  (Fig. 4A). Because desensitization occurs without channel opening, the reaction is written in a cyclical form. The arrows represent the rate constants that lead to or leave the state. Frog and rat ASIC1 completely desensitize in the continuous presence of protons; thus, the rate constant ␦ is assumed to be negligible as channels move irreversibly from the open to an absorbing desensitized state (9).
Mean open times o of wild type and mutant channels were calculated by fitting histograms of the duration of O H to single exponentials (Fig. 4, B and C). Values of o were not significantly different, 6.8 versus 4.5 ms for wild type and mutant; thus, we conclude that stability of the open state and therefore the rate constants that leave O H (␤ and ␥) are not significantly modified by mutations in the ␤1-␤2 linker.
We next examined whether the mutation M85L decreases the value of the transition rate ␣. The opening rate of the ensemble of channels in the patch (1/ a ) is a function of both ␣ and of the fraction of channels in the C H state, which, in turn, increase with the concentration of protons. Table 2 shows steep proton dependence of a in the range of pH Ն6.0 and larger values of a for the mutant than for wild type. However, at high concentration of protons, a reached the similar minimum value in both channels ϳ2 ms. Fig. 4D shows the rising phase of normalized currents (average of 10 sweeps) of patches containing mutant channels (continuous lines) activated by pH 6.5, 6.0, and 5.0 and wild type (dashed line) activated by pH 5.0. Activation of mutant channels by pH 6.0 significantly decreases the onset time compared with pH 6.5, whereas pH 5.0 does not produce a further decrease. Although pH 6.0 nearly saturates the channel binding sites (same peak current), binding of protons is still the rate-limiting process. In contrast, there is only a small decrease in onset time as the concentration is raised to pH 5.0, suggesting that at this high proton concentration, the onset approximates the value of ␣ in the scheme of Fig. 4A. Because the onset time at pH 5.0 does not differ in wild type and mutant SQL, the mutation does not significantly change the opening rate ␣. This conclusion holds even if the values obtained for the opening time are underestimated by our perfusion system because such a small value cannot explain the more than one order of magnitude difference in time course of wild type and mutant currents. Together, these observations suggest that the ␤1-␤2 linker most likely modifies a step before channel opening. This is represented in the scheme by the transition C to C H , which describes the sum of several Currents are scaled to the same peak current value. The region of the onset curve Ͼ20% of maximum is well described by a single exponential function. This part of the curve is determined primarily by the channel opening rate rather than agonist binding so that the time constant of the exponential function approximates the reciprocal of the opening rate. Onset times were as follows: 2 ms at pH 6.5, 1.5 ms at pH 6.0, and 1.4 ms at pH 5.0. events including the binding of protons and subsequent conformational changes of closed protonated channels. Thus, the substitution M85L in the ␤1-␤2 linker by decreasing the number of channels in the C H state delays channel openings and slows the time course of the macroscopic currents. The same mechanism decreases the number of channels entering the desensitized state, leading to a decrease in the apparent affinity for steadystate desensitization (transition C H to D H in the scheme of Fig. 4A).
The Effect of M85L Applies to ASIC1 of Other Species-To ascertain the role of residue 85 in modulating the function of ASIC1 and whether it works also in other species, we examined channels with Met 85 in the ␤1-␤2 linker. The previously cloned fish ASIC1 and shark ASIC1b both have Met 85 and have been shown to have fast kinetics and high sensitivity to protons (12,13). ASIC1 cloned here from EFS also has Met 85 and produces fast and small proton-activated currents with an average magnitude of 0.3 Ϯ 0.15 A/cell. Further analysis of the properties of wild type EFS was not possible owing to the small currents. However, after substitution of the three residues in the ␤1-␤2 linker Ala 83 -Asn 84 -Met 85 to Ser 83 -Gln 84 -Leu 85 or M85L alone the currents became ϳ100-fold larger, with an average 35 Ϯ 7 A/cell at pH 6.5 (n ϭ 20) (Fig. 5, A and B). Representative examples of wild type and mutant EFS ASIC1 shown in the same time scale but a 10-fold expanded current scale make evident the fast time course of currents from the wild type channel (Fig. 5C). Calculated time constants of the rise and decay phases of the currents were 3 and 17 ms, whereas time constants for the triple mutant were 26 and 140 ms, respectively. The fast kinetics, above the limits of the two-electrode voltage clamp perfusion system, result in small values for the peak currents of wild type EFS channels. Despite this limitation, the important result is the mark slowing of channel kinetics induced by the mutations. Addi-tional evidence that the small current of wild type EFS results from fast kinetics and not from low expression level is provided by the equal amount of protein found at the cell surface of cells expressing wild type or mutant EFS-SQL channels, whereas the difference in current in the same cells was 100-fold (Fig. 5E). The apparent affinities for protons of EFS-SQL channels were pH 50D 7.1 and pH 50A 6.7 (Fig. 5D) similar to those of rat.

DISCUSSION
In summary, the results of this work show that inter species differences in the response to protons of ASIC1 stem, at least in part, from variations in amino acid composition of the ␤1-␤2 linker. Subtle substitutions of neutral residues but not of negatively charged residues in the proton-sensor are a strategy used by nature to achieve diversity in pH sensitivity of ASIC1. From an evolutionary point of view, this approach must be more cost effective than changing the structure of the proton-sensor. Differences in proton affinities of ASIC1 may represent an adaptation to maintain similar levels of ASIC1 response to a given change in pH across vertebrates thereby compensate for the high baseline plasma pH of amphibian and aquatic species.
We determined that delay of channel openings upon proton stimulation is the mechanism whereby the apparent proton affinity is reduced. A decrease in affinity of any of the proton binding sites could slow the opening rate not because the ␤1-␤2 linker directly binds protons but because the M85L substitution could perturb the structure of the proton sensor indirectly decreasing the affinity of the proton bind site(s). Alternatively, the ␤1-␤2 linker stabilizes a closed conformation through the interaction of residue 85 with other yet to be identified amino acids. Among the residues tested in position 85, leucine was the only one that delayed the progression of C to C H , whereas other amino acids increased the rate of the transition ( Table 1), suggesting that Leu 85 forms the strongest contact. The crystal structure of chicken ASIC1 (4,14), which is most likely in the desensitized state, shows Leu 85 exposed to water, but the side chain points toward the interior of the protein consistent with our proposal of interacting with other residue. Taken together, the results from this work strengthen the functional importance of the segment following transmembrane domain 1 (␤1 strand and ␤1-␤2 linker) in channel gating and provide the first evidence that it participates in channel opening.