Acid-sensing ion channel 2 (ASIC2) modulates ASIC1 H+-activated currents in hippocampal neurons.

Hippocampal neurons express subunits of the acid-sensing ion channel (ASIC1 and ASIC2) and exhibit large cation currents that are transiently activated by acidic extracellular solutions. Earlier work indicated that ASIC1 contributed to the current in these neurons and suggested its importance for normal behavior. However, the specific contribution of ASIC1 and ASIC2 subunits to acid-evoked currents in hippocampal neurons remained uncertain. To decipher the individual role of the ASIC subunits, we studied H(+)-gated currents in neurons from both ASIC1 and ASIC2 null mice. We found that much of the current was produced by ASIC1a/2a heteromultimeric channels, and individual subunits made distinct contributions. The ASIC1a subunit was key in establishing current amplitude. The ASIC2a subunit had little effect on amplitude but influenced desensitization, recovery from desensitization, pH sensitivity, and the response to modulatory agents. We also found heterogeneity in the contribution of ASIC2 throughout the neuronal population, with individual neurons expressing both ASIC1a homomultimeric and ASIC1a/2a heteromultimeric channels. Studies of neurons heterozygous for disrupted ASIC alleles indicated that the properties of H(+)-gated currents are dependent on the proportion of the individual subunits. These findings indicate that the absolute and relative amounts of ASIC subunits determine the amplitude and properties of hippocampal H(+)-gated currents and therefore may contribute to normal physiology and pathophysiology.

Although the extracellular pH of the central nervous system is tightly controlled, both localized and global reductions in pH have important physiologic and pathophysiologic consequences (1). For example, repetitive nerve activity transiently reduces the local pH, perhaps by releasing the acidic contents of synaptic vesicles (2)(3)(4). Ischemia and seizures also cause extracellular acidosis that may contribute to the pathophysiology of those diseases (1,5). In situations where the extracellular pH falls, protons may act as ligands, activating H ϩ -gated channels (4,6). Similarities between the biophysical properties of transient H ϩ -gated cation currents in hippocampal neurons (7,8) and currents generated by recombinant ASICs 1 suggested they might be responsible (6,8).
The contribution of individual ASIC subunits to H ϩ -gated currents in central neurons remains uncertain. In an earlier study, we found that the loss of ASIC1 abolished pH 5-evoked currents in hippocampal neurons (41). This result seemed surprising, because ASIC1a is not the only ASIC subunit expressed in the central nervous system; hippocampal neurons also express ASIC2a and -2b (12,28,38,39). Although recombinant ASIC2b generates no current, heterologous expression of ASIC2a does elicit acid-evoked cation currents (12,42,43). In addition, Zn 2ϩ , which modulates ASIC2a-containing channels but not those produced by ASIC1a, potentiated hippocampal acid-evoked current (8,30). Therefore, the goal of this work was to learn which ASIC subunits contribute to central H ϩ -gated currents and to determine whether they function as homomultimers or heteromultimers. To accomplish this, we examined the characteristics of whole-cell H ϩ -gated currents in hippocampal neurons from mice with disrupted ASIC1 and ASIC2 genes. We focused on channel properties that differentiate the individual ASIC subunits including pH sensitivity, desensitization, recovery from desensitization, and sensitivity to FMRFamide and zinc. In light of earlier reports, we were particularly interested in testing the hypothesis that ASIC2 contributes to these currents in hippocampal neurons.

EXPERIMENTAL PROCEDURES
Cells and Culture-Mouse hippocampal neuron cultures were prepared from postnatal day 1-2 mice as described previously (44) with the addition of insulin-transferrin-sodium selenite (41). Hippocampi from 4 -10 pups were pooled and plated in 24-well dishes on 35-mm collagencoated cover slips. Cytosine ␤-D-arabinofuranoside was added at day 3-4 to inhibit glial growth. Whole-cell patch clamp was performed on large pyramidal cells maintained in culture for 1 to 3 weeks. Neurons from ASIC1 Ϫ/Ϫ (41), ASIC2 Ϫ/Ϫ (34), and heterozygous mice were compared with neurons from the corresponding wild-type animals plated and cultured at the same times. We observed no differences between the groups of wild-type neurons, and they were combined for comparison to heterozygous neurons (ASIC1 ϩ/Ϫ and ASIC2 ϩ/Ϫ).
Mouse ASIC1a, -2a, and -2b were cloned into pMT3 for heterologous expression as described (24). Chinese hamster ovary (CHO) cells were transfected with 1-3 g of DNA using the Transfast lipid reagent (Promega, Madison, WI). CHO cells were used because of the ease of studying by patch clamp, the efficiency of transfection, and their lack of native H ϩ -gated currents. The properties of ASICs expressed in CHO cells, COS-7 cells, Xenopus oocytes, and lipid bilayers can vary (4, 11-16, 22, 24, 26, 28, 30, 42, 45). However, mouse ASICs studied in CHO cells produced properties very similar to those we reported previously in COS cells (26); compare data in Ref. 24 to that in this manuscript. The pC1EGFP vector encoding enhanced green fluorescent protein (Clontech, Palo Alto, CA) was added at 20% the DNA concentration to identify transfected cells using epifluorescence microscopy. For experiments assessing the function of ASIC2b with ASIC2a, pMT3 was added to maintain a constant final DNA concentration. Cells were studied 2-3 days following transfection.
Electrodes had a resistance of 3-5 megohms, and series resistance was compensated by 70% after establishing the whole-cell configuration. Neurons were held at Ϫ70 mV, and solutions were changed by gravity-fed perfusion pipes. Data were acquired using an AXOPATCH 200B amplifier with pCLAMPex 8.1 software (Axon Instruments, Foster City, CA). Data were analyzed using Clampfit (Axon Instruments) and IGOR Pro (WaveMetrics, Lake Oswego, OR). Amplitude was determined by subtracting the baseline current at pH 7.4 from the peak current amplitude determined in Clampfit (Axon Instruments). Capacitance was measured for each neuron in Clampex (Axon Instruments). The tau of desensitization ( d ) was calculated by fitting the data to a single exponential using IGOR Pro (WaveMetrics). Because of the complex nature of current desensitization with FRRFamide application, the Td1 ⁄2 of desensitization was used as a quantitative measurement of desensitization. The Td1 ⁄2 of desensitization was recorded as the time from the peak current amplitude to half-maximal peak current (as described above).
Because of current run-down, data for pH dose-response, zinc potentiation, and recovery from desensitization were normalized to the average of the preceding and following control pH applications. All data were obtained from at least two different culture preparations representing 8 -20 mice. Because current amplitude can vary depending on the time in culture, neurons of different genotypes were prepared within 24 h of one another, cultured identically, and analyzed on the same day after culture. Statistical significance was evaluated using the paired or unpaired Student's t test as appropriate.
The presence of ASIC2a in the hippocampus made it a candidate to generate the H ϩ -gated currents in ASIC1 Ϫ/Ϫ cells. One feature that distinguishes heterologously expressed ASIC2a and ASIC2a/2b channels from channels containing ASIC1 subunits is their slower desensitization (4,11,12,42,43). We found that transient currents from ASIC1 Ϫ/Ϫ neurons desensitized more slowly than those in wild-type neurons (Fig.  1, A and C). Thus, the expression of ASIC2a and -2b in the hippocampus, the failure of ASIC2b expression to generate current, the requirement for very low pH solutions to activate, and the slow desensitization ( d ) suggest that ASIC2a (either alone or as a heteromultimeric channel with ASIC2b) generated currents in ASIC1 Ϫ/Ϫ neurons.
Loss of ASIC2 Enhances pH Sensitivity and Slows Desensitization of Acid-evoked Currents in Hippocampal Neurons-To further test the hypothesis that ASIC2 subunits contribute to H ϩ -gated currents, we compared wild-type and ASIC2 Ϫ/Ϫ neurons; disruption of the ASIC2 gene eliminates both ASIC2a and ASIC2b subunits (34). A pH 5 application generated currents with comparable peak amplitudes in wild-type and ASIC2 Ϫ/Ϫ neurons (Fig. 2, A and B). This result suggests that ASIC2 made little contribution to current amplitude. Therefore, we studied additional properties. Earlier work showed that currents from heterologously expressed ASIC2a channels required more acidic solutions for activation than ASIC1a channels, and ASIC2a/1a heteromultimers had intermediate pH sensitivity (24,28). In wild-type hippocampal neurons, reducing extracellular pH increased current amplitude (Fig. 2C), consistent with earlier work (8). Eliminating ASIC2 increased the pH sensitivity; for example, pH 6.3 solutions stimulated 59% of the pH 5-induced current in ASIC2 Ϫ/Ϫ neurons versus only 26% in wild-type neurons.
Loss of ASIC2 also prolonged desensitization of H ϩ -gated currents as measured by the desensitization rate ( d ) (Fig. 2, A  and D). Because homomultimeric ASIC2a desensitizes more slowly than other ASIC subunits, removing its contribution might have been expected to shorten, rather than lengthen d . To investigate the basis of this change, we measured the d of recombinant ASIC1a and ASIC2 subunits expressed in heterologous cells (Fig. 3, A and B). For this and all other studies, we used mouse ASIC subunits to facilitate comparison with the properties of ASICs in the mouse neurons. Consistent with earlier reports, ASIC1a currents desensitized more rapidly than ASIC2a (24, 28), and ASIC1a/2a currents had a d shorter than either subunit alone (24). Because the ASIC1a/2a combination generated currents that desensitized faster than the sum of the components, these data indicated that heteromultimeric channels generated the current. The fact that ASIC2 Ϫ/Ϫ neurons desensitized more slowly than wild-type further suggests that the acid-evoked current in wild-type hippocampal neurons arose, at least in part, from heteromultimeric channels composed of ASIC1a and ASIC2a. Although ASIC2b is expressed in hippocampus, earlier reports indicated that it neither generated current nor altered ASIC1a currents (12). Consistent with that observation, adding ASIC2b failed to alter d compared with ASIC1a alone (Fig. 3B). These results suggest that the faster desensitization of wild-type neurons is because of the ASIC2a subunit.

ASIC2 Contributes to the Modulatory Effects of FRRFamide
and Zinc-To further test the hypothesis that ASIC2 contributes to acid-evoked currents in hippocampal neurons, we tested the effect of agents that selectively modulate channels containing ASIC1a or ASIC2a. In heterologous expression systems, FRRFamide, FMRFamide, and related neuropeptides potentiate ASIC1a-mediated currents, whereas ASIC2a is unaffected (20,29). In wild-type hippocampal neurons, FRRFamide slowed desensitization and in some cases generated sustained acid-evoked currents, consistent with the presence of ASIC1a (Fig. 4, A and B). Surprisingly, eliminating ASIC2 attenuated this response. These results suggest that the presence of ASIC2 subunits enhanced peptide modulation of ASIC1a-containing channels. To investigate this possibility, we expressed the subunits in CHO cells and found that adding ASIC2a enhanced the response of ASIC1a (Fig. 4B). These data suggest that heteromultimeric channels are responsible for H ϩ -gated currents in hippocampal neurons and that the normal response to FRRFamide requires both ASIC1a and ASIC2a subunits. These results have a parallel in a previous study (29) showing that heteromultimeric channels composed of ASIC3 and ASIC2a responded more robustly to FMRFamide-related peptides than ASIC3 alone.
Earlier work showed that zinc selectively increased H ϩgated currents from ASIC2a and ASIC2a-containing heteromultimeric channels (8,30). We found that zinc potentiated H ϩ -gated currents in wild-type neurons, whereas it slightly inhibited current in ASIC2 Ϫ/Ϫ neurons (Fig. 4, C and D). These results support the conclusion that ASIC2 contributes to hippocampal H ϩ -gated currents and is required for zinc potentiation.
ASIC2a Determines the Fast Recovery from Desensitization in Wild-type Neurons-Because ASIC2a influences recovery from desensitization, we also tested this property. ASIC2a and ASIC1a/2a channels recover rapidly following a pH 5 application, whereas ASIC1a homomultimers recover more slowly (24). We found that after 1 s at pH 7.4, wild-type hippocampal neurons recovered 61% of the pH 5-evoked current (Fig. 5, A and B). Deleting ASIC2 extended the recovery time, so that in 1 s, only 4% of the current had recovered, and even by 10 s, recovery was only 58%. Because ASIC2b is reported to not affect ASIC1a current (12), we assumed that ASIC2a was the subunit responsible for speeding recovery. To test this assumption, we expressed recombinant ASIC1a with ASIC2a or -2b and measured recovery rate. Fig. 5C shows that ASIC2a but not -2b accelerated the recovery of ASIC1a currents. These striking differences indicate that ASIC2a determined the fast recovery from desensitization in hippocampal neurons.
Channels with Heterogeneous Subunit Composition Contribute to H ϩ -gated Hippocampal Currents-The results suggested that hippocampal H ϩ -gated current reflect predominantly ASIC1a/2a heteromultimers. However, they did not exclude the presence of ASIC1a homomultimeric channels. Several observations are consistent with this possibility. First, the peptide toxin PCTX1, which was reported to selectively inhibit ASIC1a homomultimeric channels, reduced H ϩ -gated current in central nervous system neurons (8,19). Second, recovery from desensitization by wild-type neurons (Fig. 5B) was slower than recombinant ASIC1a/2a heteromultimers (Fig. 5C). Third, wild-type neurons (Fig. 2C) were slightly more pH-sensitive than ASIC1a/2a heteromultimers (24,28). Therefore, we tested for heterogeneity of channel types within individual neurons. We reasoned that, because ASIC1a homomultimers recover much more slowly from desensitization than heteromultimers containing ASIC2a (see Fig. 5C and Ref. 24), we could largely eliminate the contribution of ASIC1a homomultimers by limiting the time of recovery from desensitization. This would uncover currents from other homo-and heteromultimeric channels, which we might recognize by a change in d . To test this, we applied a pH 5 solution, measured the d ( d 0) and then continued the perfusion for 10 s to desensitize acid-evoked currents. We then stopped the pH 5 solution, allowed 2 s at pH 7.4 for recovery, and reapplied pH 5 solution to measure the d a second time ( d 2). Fig. 6A shows the ratio of d after 2 s of recovery relative to that prior to desensitization ( d 2/ d 0). We expected that if all the channels expressed on a cell were of identical subunit composition, then the d would be unchanged after desensitization and the ratio would be 1. To test this notion, we expressed recombinant ASIC1a, -2a, and -1a/2a channels and found a ratio of ϳ1 in all cases. Thus, with a homogeneous population of channels, d was the same irrespective of whether the channels showed fast or slow recovery. In contrast, wild-type neurons had a d 2/ d 0 ratio less than 1, suggesting that these neurons contain a population of rapidly recovering channels with a short d (consistent with ASIC1a/2a channels) and a population of slowly recovering channels with a longer d (consistent with ASIC1a homomultimers). Supporting this conclusion, ASIC2 Ϫ/Ϫ neurons had a d 2/ d 0 ratio of 1. Thus, individual wild-type neurons appeared to express ASIC1a/2a heteromultimeric channels plus channels with properties consistent with ASIC1a homomultimers.
We also tested for heterogeneity in the contribution of ASIC2a throughout the population of neurons. The rate of recovery from desensitization is one of the characteristics most strikingly influenced by ASIC2a; ASIC1a/2a heteromultimeric channels recover quickly, whereas ASIC1a homomultimeric channels recover more slowly. Therefore, we desensitized current with a 10-s pH 5 perfusion and then measured the percentage of current recovery 2 s after returning to a pH 7.4 solution. Wild-type neurons showed substantial variability (Fig. 6B). For example, within 2 s, some neurons had recovered ϳ20% of the initial current, whereas others had recovered all of the initial current. In contrast, when ASIC2 was eliminated, none of the neurons showed more than 25% recovery at 2 s. These results suggest that ASIC2a subunits contributed to the properties of H ϩ -gated currents in the majority of neurons.
ASIC2b Is Expressed in Excess of ASIC2a in Neonatal Hippocampal Neurons-The results described above suggested a paradox. On one hand, only 43% of ASIC1 Ϫ/Ϫ neurons possessed transient currents evoked by pH 4. Yet, on the other hand, ASIC2a influenced the properties of acid-evoked currents in the vast majority of wild-type hippocampal neurons. If ASIC2a were present in most neurons, why were ASIC2a currents not more common in ASIC1 Ϫ/Ϫ neurons? A potential explanation for the discrepancy would be that ASIC1 disruption reduces ASIC2 expression. Although an earlier study showed that ASIC2 expression was the same in wild-type and ASIC1 Ϫ/Ϫ brains, hippocampal expression was not specifically measured (41). Using RT-PCR, we found ASIC2a and -2b transcripts in hippocampi of both wild-type and ASIC1 Ϫ/Ϫ animals (Fig. 7A). Northern analysis indicated that ASIC2 transcripts were present equally in wild-type and ASIC1 Ϫ/Ϫ hippocampi (Fig. 7B). Interestingly, the Northern blot also revealed that ASIC2b transcripts were much more abundant than ASIC2a transcripts. A greater abundance of ASIC2b mRNA compared with ASIC2a mRNA has also been reported in several other brain regions (34,38,39,50). This observation raised the possibility that ASIC2b might affect ASIC2a current in ASIC1 Ϫ/Ϫ neurons.
ASIC2b subunits do not generate H ϩ -gated currents when FIG. 6. Heterogeneity of H ؉ -gated currents in individual neurons and in the population of hippocampal neurons. A, data are ratio of the d of desensitization 2 s after recovery at pH 7.4 ( d 2) versus before desensitization ( d 0). Measurements were made from pH 5-activated currents in CHO cells expressing ASIC1a (n ϭ 12), ASIC2a (n ϭ 5), and ASIC1 ϩ ASIC2a (n ϭ 13) and from ASIC2 ϩ/ϩ (n ϭ 25) and ASIC2 Ϫ/Ϫ (n ϭ 17) hippocampal neurons. The asterisk indicates value different from 1.0, p Ͻ 0.02. B, recovery from desensitization was measured as shown in Fig. 5A by applying pH 5 for 10 s, pH 7.4 for 2 s, and then a second pH 5 application. We measured the percentage recovery with the second acid stimulus for 32 ASIC2 ϩ/ϩ and 17 ASIC2 Ϫ/Ϫ neurons. The graph shows percentage current recovery (y axis) for a series of individual neurons (x axis). Each data point indicates an individual neuron; squares indicate ASIC2 ϩ/ϩ, and circles indicate ASIC2 Ϫ/Ϫ neurons.

FIG. 7. Effect of ASIC2b on ASIC2ainduced currents.
A, RT-PCR from ASIC1 ϩ/ϩ and ASIC1 Ϫ/Ϫ hippocampus using primers that detected both ASIC2a and -2b or either alone. B, Northern analysis of RNA from ASIC1 ϩ/ϩ and ASIC1 Ϫ/Ϫ hippocampus using probes that recognize the indicated transcript. C, relationship between pH and H ϩ -gated current in CHO cells expressing ASIC2a alone (n ϭ 10 -17) or ASIC2a and -2b at ratios of 1:1 (n ϭ 9 -14), 1:2 (n ϭ 6 -13), and 1:3 (n ϭ 6 -8). In all cases the amount of ASIC2a cDNA was constant, and pMT3 DNA was varied to maintain a constant amount of total DNA. Data were normalized to response to pH 3 stimulus. D, pH 3-induced current amplitude in cells expressing ASIC2a with different amounts of ASIC2b (n ϭ 17-19). expressed alone, but they were reported to form heteromultimeric channels with ASIC2a (12,42,43). We reasoned that if ASIC2b were present in excess of ASIC2a, then in ASIC1 Ϫ/Ϫ neurons most channels would be ASIC2a/2b heteromultimers. To test the consequences of such a situation, we expressed ASIC2a with varying amounts of ASIC2b in CHO cells. Coexpressing ASIC2b with ASIC2a tended to reduce the pH sensitivity, although the effects were small (Fig. 7C); this result is consistent with earlier studies (12,43). However, ASIC2b reduced current amplitude produced by pH 3 application (Fig.  7D). Moreover, as the relative amount of ASIC2b increased, current amplitude fell. These results may explain the low percentage of ASIC1 Ϫ/Ϫ neurons with H ϩ -gated currents and their relatively small current amplitude. In those neurons, ASIC2b likely dampens current from channels mediated by ASIC2a, and perhaps currents were only observed in neurons expressing a higher proportion of ASIC2a.
ASIC1 and ASIC2 Heterozygote Neurons Show Altered H ϩgated Currents-The contribution of ASIC1 and -2 subunits to H ϩ -gated currents suggested that hippocampal neurons heterozygous for the ASIC1 and ASIC2 alleles might have altered acid-evoked currents. We found that compared with wild-type controls, the amplitude of pH 5-evoked currents fell by approximately half in ASIC1 ϩ/Ϫ neurons but remained unchanged in ASIC2 ϩ/Ϫ neurons (Fig. 8A). ASIC1 ϩ/Ϫ neurons also showed an accelerated d and ASIC2 ϩ/Ϫ neurons a slowed d (Fig. 8B). H ϩ -gated currents from ASIC2 ϩ/Ϫ neurons showed little shift in pH sensitivity, but recovery from desensitization was delayed (Fig. 8, C and D). Despite the dramatic reduction in current amplitude in ASIC1 ϩ/Ϫ neurons (Fig. 8A), the pH dose-response and recovery from desensitization were not altered significantly (Fig. 8, C and D). Together, these results further support the conclusion that both ASIC1a and ASIC2 subunits influence acid-evoked currents in hippocampal neurons. ASIC1 seems to be important in determining current amplitude, whereas the predominant role of ASIC2 appears to be in influencing the properties of the current.

DISCUSSION
It has been known for many years that acidic extracellular solutions activate transient cation currents in hippocampal neurons (7). The biophysical properties of those currents (8) and the expression of ASIC1a and ASIC2 subunits in the hippocampus (4,12,26,28,38,39) suggested that ASIC subunits were responsible. ASIC subunits might generate H ϩ -gated currents in two general ways. ASIC1a and ASIC2a could each form homomultimeric channels, with net current representing the sum of the individual currents. Alternatively, the subunits could combine to form heteromultimeric channels. Our data indicate that at least two ASIC subunits, ASIC1a and -2a, contribute to H ϩ -activated currents in mouse hippocampal neurons. They also suggest that both heteromultimeric and homomultimeric channels generate current, and individual ASIC subunits make distinct contributions.
Hippocampal Neurons Contain Both Hetero-and Homo-multimeric ASIC Channels-Heteromultimeric ASIC1a/2a channels produced H ϩ -gated current in cultured hippocampal neurons. Support for this conclusion came from the finding that d in wild-type neurons exceeded that in both ASIC1 Ϫ/Ϫ and ASIC2 Ϫ/Ϫ neurons and in heterologous cells expressing either ASIC1a or -2a alone. Thus, the d could not be explained by the sum of currents from homomultimeric channels. FRRFamide addition also revealed responses that could not be attributed to an aggregate of homomultimeric channels. Although FRRFamide does not alter ASIC2a currents, it produced a greater response in wild-type neurons and in heterologous cells expressing ASIC1a/2a heteromultimeric channels than in ASIC2 Ϫ/Ϫ neurons or cells expressing ASIC1a homomultimers alone. These results indicate that ASIC1a/2a channels have unique properties compared with homomultimeric channels, and those channels generate H ϩ -gated currents in hippocampal neurons. An advantage for making this conclusion is that our studies used neurons with specific gene disruptions. Thus, they do not depend on comparing absolute values of channel properties in neurons with those of ASICs expressed in heterologous cells, where values could vary depending on the cell type and associated proteins. Nevertheless, there was complete qualitative agreement between data with neurons and recombinant channels, and the quantitative responses were very similar.
By forming heteromultimeric ASIC channels, hippocampal neurons follow a pattern found in other cells expressing DEG/ ENaC subunits. Epithelial cells form heteromultimeric channels from ␣, ␤, and ␥ ENaC subunits (54 -56). Caenorhabditis elegans neurons form channels from the combination of Mec4 and Mec10 (57), and dorsal root ganglion neurons combine ASIC1, -2, and -3 subunits to generate H ϩ -gated currents (24,33). However, differences between subunit desensitization and recovery from desensitization suggested the presence of ASIC1a homomultimeric channels in hippocampal neurons. This conclusion is supported by the observation that the peptide toxin PCTX1 reduced H ϩ -gated current in central nervous system neurons (8). Expression of both heteromultimeric and homomultimeric channels has not been reported in other cell types expressing more than one DEG/ENaC subunit. This finding suggests the opportunity for hippocampal neurons to construct channels with a complex variety of subunit compositions and characteristics.
ASIC Subunits Confer Distinct Properties on Hippocampal Acid-evoked Currents-ASIC1a and -2a subunits made distinct contributions to H ϩ -activated hippocampal currents. ASIC1a seemed to be important in establishing current amplitude. Eliminating ASIC1 from neurons abolished pH 5-induced current, and its absence reduced average pH 4-induced current by ϳ95%. Most convincingly, removing one ASIC1 allele approximately halved H ϩ -gated current.
In contrast, ASIC2 Ϫ/Ϫ and ϩ/Ϫ neurons generated the same current amplitude as wild-type neurons, indicating that ASIC2 plays a minor role in determining this quantitative aspect of H ϩ -gated current. Instead, ASIC2 influenced the biophysical properties, defining, in part, acid sensitivity, desensitization, the rate of recovery from desensitization, and the response to neuropeptide modulators and zinc. A predominantly modulatory role for ASIC2a subunits may be analogous to the contributions of some ␥-aminobutyric acid-A receptor subunits to that channel complex (58).
The contribution of ASIC2b is less certain. Earlier studies reported that co-expressing ASIC2b with ASIC1a had no effect on H ϩ -gated currents (12). When expressed with ASIC2a, ASIC2b reduced pH sensitivity to a small extent (12,43), consistent with our results. In addition, our data indicate that ASIC2b reduced ASIC2a current amplitude in response to a maximal stimulus. This finding, together with a greater relative abundance of ASIC2b versus -2a transcripts in the hippocampus, provides a potential explanation for the minimal H ϩ -gated current in ASIC1 Ϫ/Ϫ neurons. An alternative factor that we cannot exclude, but that could influence current properties, is the possibility that other channels or proteins might differentially regulate ASICs in null and wild-type animals (59).
Contribution of ASIC to Function of Hippocampal Neurons-Specific ASIC subunits make distinct contributions to hippocampal acid-evoked currents, and there is heterogeneity in currents within and between individual neurons. This varia-bility could influence the contribution of ASICs to normal physiology and to pathophysiology. For example, in the peripheral nervous system relatively small changes in H ϩ -gated currents alter sensory transduction (34 -36). Moreover, in the mouse brain, disrupting ASIC1 impaired synaptic plasticity and performance in several behavioral tests (40,41). Pathologic conditions may also skew ASIC levels. For example, ischemia is reported to increase ASIC2a levels (60); such an increase could reduce the pH sensitivity of acid-evoked currents, making them less responsive to an acid insult. Following seizures, the relative contribution of various subunits to the complex may also change (50), perhaps altering the responsiveness of hippocampal neurons to seizure-associated acidosis. In addition, if mutations in ASIC genes are discovered in humans, our data suggest that haplo-insufficiency could have a substantial impact on H ϩ -gated currents and perhaps behavior. Finally, these results suggest that molecules targeting specific ASIC subunits might have value as therapeutics.