Knockout of the ASIC2 channel in mice does not impair cutaneous mechanosensation, visceral mechanonociception and hearing

Methods

Generation of ASIC2 knockout mice

The generation of the ASIC2−/− mice used here has been reported by us elsewhere (Ettaiche et al. 2004). Briefly, a loxP sequence and a neomycin resistance cassette flanked by two loxP sequences were inserted into the HindIII and XhoI sites flanking exon 8, which codes for the second transmembrane domain. The targeted homozygotes were crossed with mice that express CRE recombinase, leading to deletion of exon 8 and thus to a truncation of the ASIC2 protein before TM2. The TM2 deletion affects all known splice variants of ASIC2: ASIC2a and ASIC2b. The ASIC2 null mice were viable, fertile and did not have any apparent morphological or behavioural abnormalities. Animal experiments were carried out in accordance with European Communities Council Directive 86/6609/EEC. Experiments were also in accordance with the German Animal Protection Law (1987) and approved by the District Government or in accordance with French legislation.

Expression of ASIC2a from wild-type and knockout mice in Xenopus oocytes

Both ASIC2 wild-type and ASIC2 knockout cDNA was amplified from ASIC2+/− mouse brain by RT-PCR with the primers AGAATTCGCCGCCACCATGGACCTCAAGGAGAG (sense) and ATCTCGAGTCAGCAGGCAATCTCCT (anti-sense) and subcloned into a Xenopus oocyte expression vector (Waldmann et al. 1996). Capped cRNA was prepared using standard techniques and 1 ng of cRNA was injected into Xenopus laevis oocytes, obtained from frogs which were hummanely killed after the final collection. Currents activated by extracellular acid were recorded 1 day later essentially as described by Champigny et al. (1998). The bath medium contained (mm): NaCl, 140; MgCl₂, 2; CaCl₂, 1.8; and Hepes, 10; adjusted to pH 7.4 with NaOH. In the acidic extracellular solution, Hepes was replaced by 10 mm sodium acetate, adjusted to pH 4.5 with acetic acid.

Primary cultured mouse hippocampal neurones

Six 2-day-old C57BL/6 mice were killed by decapitation. Hippocampi were dissected and sliced (7–8 slices per hippocampus). Slices were collected at 4°C in DMEM (Sigma; pH 7.4) and incubated in 0.2% trypsine (Seromed), 1875 U ml⁻¹ of DNAse IV (Sigma) in the digestion buffer (mm: NaCl, 137; KCl, 5; Na₂HPO₄, 7; NaHCO₃, 4; Hepes, 25) at room temperature for 5 min. The enzymatic digestion was stopped by an 8 min incubation in 1 mg ml⁻¹ Type I Trypsin inhibitor (Sigma) at room temperature. Slices were washed with DMEM before being mechanically dissociated in the presence of 1875 U ml⁻¹ of DNAse IV, followed by two further mechanical dissociations. Supernatant was collected and centrifuged for 15 min, 300 g at 4°C. The pellet was resuspended in DMEM, and cells were preplated at a density of 100 000 per 35 mm tissue culture dish (central drop) coated with polyornithine (Sigma) and matrigel (Becton Dickinson) in culture medium (MEM {Sigma} with 6 g l⁻¹ glucose, 2.2 g l⁻¹ NaHCO₃, 2 mm glutamine, 10% decomplemented fetal calf serum, 3.6 g l⁻¹ Hepes, 100 mg l⁻¹ transferrine {Sigma}, 30 mg l⁻¹ insulin {Sigma}, 100 mg l⁻¹l-ascorbic acid {Sigma}, 0.1 mg l⁻¹ biotin {Sigma}, 1.5 mg ml⁻¹ vitamin B12 {Sigma} and 2 μg ml⁻¹ gentamycin {Sigma}) for 2–3 h before filling the dishes (2 ml culture medium per dish). After 1 day of culture, the culture medium was complemented with B27 medium (1:50 dilution; Gibco InVitrogen), 2 μm uridine (Sigma), 2 μm 5-fluoro-2′ -deoxyuridine (Fluka), and this medium was renewed every 4 days. Neurones were kept in an incubator with 95% air–5% CO₂ at 37°C and used for electrophysiological recordings 7–20 days after plating. Neurones with triangular cell bodies, a typical feature of pyramidal neurones, were selected for recording.

Patch-clamp recordings of hippocampal ASIC-like currents

Ion currents were recorded using the whole-cell patch-clamp technique (Hamill et al. 1981). Data were sampled at 500 Hz and low-pass filtered at 3 kHz using pClamp8 software (Axon Instruments). The statistical significance of differences between sets of data was estimated by Student's unpaired t test. The pipette solution contained (mm): 140 KCl, 5 NaCl, 2 MgCl₂, 5 EGTA, 2 K₂ATP and 10 Hepes (pH 7.35), while the bath solution contained (mm): 150 NaCl, 5 KCl, 2 MgCl₂, 2 CaCl₂, 10 glucose and 10 Hepes (pH 7.45). Twenty micromolar CNQX, 10 μm kynurenic acid, 7 mm MgCl₂ and 10 μm bicuculline (all from Sigma) were added in order to inhibit glutamate- and GABA-induced currents. Mes or acetate was used instead of Hepes to buffer bath solution pH ranging from 6 to 5, and from 4.5 to 3, respectively. ZnCl₂ was coapplied with acidic solutions at 300 μm (Baron et al. 2001), whereas the spider toxin psalmotoxin 1 (PcTx1) was applied at 20 nm before the pH drop (Escoubas et al. 2000). ASIC-like currents were activated every 2 min by an extracellular pH drop induced by shifting one out of six outlets of a microperfusion system in front of the cell. Experiments were carried out at room temperature (20–24°C). Bovine serum albumin (0.1%) was added in extracellular solutions containing the toxin PcTX1 to prevent its adsorption to tubing and containers.

Western blots

The Western blot with an antibody directed against the ASIC2a NH₂ terminus (MDLKESPSEGSLQPSSC) was carried out as described (Ettaiche et al. 2004). Briefly, total protein was prepared from mouse brain and from COS cells (monkey kidney cell line) transfected with an ASIC2a or ASIC3 expression vector (PCI, Promega, France). Proteins were separated on 8% Laemmli gels and transferred to HybondC (Amersham). The first antibody was a 1:200 dilution of affinity purified anti-ASIC2a antibody. The second antibody was a 1:10 000 diluted peroxidase labelled anti-rabbit IgG antibody (Pharmacia).

In vivo measurements of auditory threshold

The technique for recording gross cochlear potentials in mice has been described elsewhere (Guitton et al. 2003). Briefly, mice were anaesthetized intraperitoneally with 0.3 ml kg⁻¹ of 6% sodium pentobarbitone (Sanofi, Montpellier, France). The right bulla was opened through a posterior auricular surgical procedure (dorsal approach) to expose the cochlea. A recording electrode was placed on the bony edge of the round window. A reference electrode was placed in the neck muscles. Cochlear potentials were elicited with tone burst with a 1 ms rise–fall time and a 9 ms total duration generated by an arbitrary function generator (type 9100R, LeCroy Corporation, Chestnut Ridge, NY, USA). The signals were passed through a programmable attenuator and presented to the ear in free field via a JBL 075 earphone (JBL, Northridge, CA, USA). Ten frequencies were tested (2, 4, 6, 8, 10, 12, 16, 20, 26 and 32 kHz), increasing in 5 dB increments from 0 to 100 dB sound pressure level (SPL). The rate of presentation was 10 bursts s⁻¹. Cochlear responses were amplified (gain 2000) by a differential amplifier (Grass P511K, Astro-Medical Inc., West Warwick, RI, USA), averaged (256 samples) and saved on the hard disk of a PC. The stored potentials were digitally low-pass filtered at 2.5 kHz to measure the compound action potential (CAP) of the auditory nerve, and the summating potential (SP) reflecting the summed intracellular DC receptor potential, mainly generated by the inner hair cells (Cody & Russell, 1988). To extract the cochlear microphonic (CM), the potential reflecting summed intracellular ac receptor potentials, mainly generated by outer hair cells (Patuzzi et al. 1989), cochlear potentials were filtered with a band-pass filter centred on the frequency of tone burst stimulation.

In vitro skin–nerve preparation

Extracellular recordings from single nerve fibres were performed as previously published (Kress & Guenther, 1999; Bernardini et al. 2002). Briefly, KO mice and WT litter mates were killed by exposure to 100% CO₂ and the hairy skin with the saphenous nerve attached was dissected from either hind paw. The preparation was transferred to a chamber filled with synthetic interstitial fluid (SIF) consisting of (mm): 107.8 NaCl, 26.2 NaHCO₃, 9.64 sodium gluconate, 7.6 sucrose, 5.05 glucose, 3.48 KCl, 1.67 NaH₃PO₄, 1.53 CaCl₂ and 0.69 MgSO₄ at 32°C, which was adjusted to pH 7.4 by continuously gassing with 95% oxygen–5% CO₂. After the nerve had been placed into a separate chamber filled with liquid paraffin for electrical insulation, fine filaments were dissected from the main nerve trunk with sharpened watchmaker's forceps and monopolar action potential activity was recorded from the single filaments with a gold wire electrode with the reference electrode positioned nearby. Signals were amplified 10 000-fold and recorded online on the hard disk of a PC. For recording and offline analysis, the SPIKE/SPIDI software package was used (Forster & Handwerker, 1990). Receptive fields were identified by probing the skin with a blunt glass rod as a search stimulus. For mechanical threshold determination, a calibrated set of von Frey hairs with uniform tip diameter was used. Mechanical response characteristics of myelinated RA and SA fibres were determined with a custom-made automated constant force stimulator. Heat stimuli were applied from a feedback-controlled radiant heat source (linear temperature rise from 32 to 46°C in 21 s with passive cooling) and for cold stimulation SIF of +4°C was applied to the receptive field from a syringe.

CGRP release measurements

The descending colon of the mice was dissected in one piece, cleaned, mounted with the peritoneal side exposed on an inflatable tube with a pressure transducer and washed for 30 min with SIF at 37°C as previously described (Roza & Reeh, 2001). Five glass tubes were each filled with 1.2 ml of SIF at 32°C and the experiment started when the mounted preparation was transferred to the first tube for an incubation period of 5 min. After two 5 min baseline incubations the tube was inflated to a pressure of 60 mmHg during the third incubation and this stimulation was followed by two 5 min washout periods. For CGRP enzyme immunoassay (EIA) analysis, 100 μl aliquots were taken from the total volume and mixed with commercial EIA buffer (SPIbio, Montigny le Bretonneux France). CGRP content was determined as provided by the producer and all EIA plates were analysed photometrically with a microplate reader (Dynatech, Chantilly VA, USA). Total concentrations of CGRP were calculated with reference to 1 g fresh weight and normalized values (to the control incubation preceding the mechanical stimulation) are given in the figures as means ±s.e.m.

Results

Deletion of the second transmembrane domain (TM2) inactivates ASIC2

TM2 has previously been shown to participitate in the ionic pore of this ion channel family (Waldmann et al. 1995). Therefore the TM2 deletion in our knockout mice (Ettaiche et al. 2004) should abolish channel activity. To verify this, we expressed both wild-type ASIC2a and the targeted ASIC2a transcript in Xenopus laevis oocytes (Fig. 1). While a transient acid-activated inward current of at least 100 nA was recorded in six out of ten oocytes injected with wild-type ASIC2a cRNA, no H⁺-gated current could be detected in any of the ten oocytes injected with the ASIC2a transcript from ASIC2 knock-out mice. To further confirm the absence of functional ASIC2 channels, ASIC-like currents were examined in hippocampal neurones from ASIC2−/− mice and compared with currents recorded from ASIC2+/+ mice. We tested the effect of the toxin PcTx1, a specific inhibitor of homomeric ASIC1a channels (Escoubas et al. 2000), and the effect of zinc, a coactivator of ASIC2a-containing channels (Fig. 2; Baron et al. 2001). The ASIC-like current of ASIC2 +/+ mouse hippocampal neurones was half inhibited (I/I_control= 0.5 ± 0.1, n= 15) by 20 nm PcTx1 (Fig. 2C) and increased by about 1.4-fold (I/I_control= 1.36 ± 0.09, n= 28) by 300 μm zinc (Fig. 2D). Thus, the ASIC-like current of ASIC2+/+ mouse hippocampal neurones flows through a mixture of PcTx1-inhibited, zinc-insensitive homomeric ASIC1a channels and of PcTx1-resistant, zinc-sensitive ASIC2a-containing heteromeric channels. These results are similar to those previously reported for rat hippocampal neurones (Baron et al. 2002). In hippocampal neurones from ASIC2−/− mice, PcTx1 (20 nm) induced a massive inhibition (I/I_control= 0.10 ± 0.02, n= 17, Fig. 2C) of the ASIC-like current, whereas zinc (300 μm) did not increase but rather inhibited the current amplitude. (I/I_control= 0.62 ± 0.08, n= 13, Fig. 2D). These properties show that homomeric ASIC1a channels are highly predominant in neurones from ASIC2−/− mice and that ASIC2a subunits do not participate in the formation of functional ASIC channels in ASIC2−/− mice.

Effect of ASIC2 knockout on hearing

Hearing is the sense for which mechanoperception meets its biggest challenge, since extremely small mechanical forces are transformed within microseconds into electrical signals. This short delay makes a direct coupling of mechanosensitive channels to the mechanical trigger highly likely (reviewed by Strassmaier & Gillespie, 2002). The hair cell transduction channel is blocked by the diuretic amiloride and by the lanthanide Gd³⁺, both rather unspecific channel blockers at the high concentrations required. ASIC channels are also inhibited by amiloride (Waldmann et al. 1998) and certain subunit combinations are blocked by Gd³⁺ (Babinski et al. 2000). Thus ASICs are potential candidates for the hair cell transduction channel. To evaluate whether ASIC2 is required for hearing, we recorded compound action potentials (CAPs) of the auditory nerve (Fig. 3). CAPs reflect the electrical activity of the auditory nerve and thus the output of the cochlea towards the central nervous system. The sound intensity thresholds that provoked CAPs were virtually identical in wild-type and ASIC2 knockout mice over the entire frequency range tested (2–64 kHz; Fig. 3).

Analysis of cochlear potentials reflecting the activity of hair cells also showed no significant difference between wild-type and ASIC2 knockout mice, as follows. (i) Cochlear microphonics (CMs) are AC responses with a frequency identical to that of the stimulus, which are considered to be primarily generated by the basal turn outer hair cells (Patuzzi et al. 1989). CM amplitudes measured at saturation (85 dB SPL) were 152 ± 35.13 μV for ASIC2+/+ and 128 ± 31.25 μV (n= 6, ±s.e.m.) for ASIC2−/− mice. (ii) Summating potentials (SPs) are DC shifts of cochlear potentials that persist for the duration of the tone burst and reflect the intracellular DC receptor potentials, mainly generated by the inner hair cells (Dallos, 1986; Cody et al. 1988). The maximum SP amplitudes at 100 dB SPL were 125.6 ± 25.6 and 116.3 ± 32.7 μV (n= 6, ±s.e.m.) in wild-type and ASIC2 KO mice, respectively.

The almost identical CAP audiograms, cochlear microphonics and summating potentials recorded from ASIC2+/+ and ASIC2−/− mice suggest that ASIC2 is not required for auditory perception and is not part of the hair cell transduction channel.

It also seems unlikely that other ASICs play a major role in hair cell transduction since the biophysical properties of the known members of the ENaC/Deg ion channel family (Waldmann et al. 1998) do not match the known properties of the hair cell transduction channel (Strassmaier et al. 2002), as follows. (i) The hair cell channel conductance was estimated by several groups to be around 100 pS. Conversely, all ENaC/Deg channels have rather small conductances, below 15 pS. (ii) The mechanosensitive ion channel in hair cells discriminates poorly between cations and Ca²⁺ is favoured over monovalent cations. Conversely, ASIC channels have a rather strong preference for monovalent cations. The only ASIC subunit with a significant Ca²⁺ permeability is ASIC1 (Na⁺:Ca²⁺permeability ratio of 16 (Sutherland et al. 2001) or 2.5 (Waldmann et al. 1997b).

Effect of ASIC2 knockout on visceral mechanonociception

Mechanosensitive afferents in hollow visceral organs, such as the GI tract and urinary bladder, are thought to signal the degree of distension and thus filling, but also signal discomfort and pain when excessive pressure builds up. The colon is one of the best characterized models for visceral mechanosensation. Both mechanosensitive Aδ and C fibres innervate the colon (Su & Gebhart, 1998). Neurosecretory nociceptive fibres release neuropeptides such as CGRP and tachykinins upon stimulation (Roza et al. 2001) and the amount of peptide released is an indicator of the activity of those nociceptive nerve endings. To evaluate whether knockout of the ASIC2 gene affects visceral mechanonociception, we examined the pressure-stimulated CGRP release from colon in vitro. An increase in pressure and thus strain increased CGRP release from the colon of ASIC2+/+ mice and their ASIC2−/− littermates. However, there was clearly no effect of the ASIC2 knockout (Fig. 4). Thus ASIC2 is not part of the mechanosensor of primary CGRP-releasing nociceptive afferents innervating the colon.

Little is known about the pathways that lead to activation of visceral afferents after mechanical stimulation. To our knowledge, no electrophysiological recordings of mechanosensitive ion channels from identified sensory neurones that innervate the viscera have yet been reported. One would expect that very high speed (≪ 1 s) is not that crucial for visceral mechanosensation, since rather slowly changing parameters are signalled, such as the degree of filling. Indirect pathways that do not involve a direct coupling of a mechanosensor to an ion channel might play a prominent role in visceral nociception. Epithelial mechanostimulated ATP release and activation of ionotropic purinergic receptors on sensory nerve endings were proposed to mediate visceral mechanosensation in the gut and other hollow organs (Burnstock, 1999).

Effect of ASIC2 knockout on cutaneous nociception

Cutaneous mechanosensation is one of the best characterized but maybe also the most complex form of mechanoperception. Several types of mechanosensitive fibres innervate the skin (for review see Koltzenburg et al. 1997). Myelinated Aβ fibres comprise RA (rapidly adapting) fibres and SA (slowly adapting) fibres, which innervate hair follicles and Merkel cell complexes in touch domes, respectively. Two types of thinly myelinated Aδ fibres, rapidly adapting D hair (DH) and slowly adapting SA fibres, low- and high-threshold C mechanoreceptors and polymodal Aδ and C nociceptors add further complexity to mammalian cutaneous mechanosensation.

We analysed the responses of different classes of sensory fibres to mechanical, chemical and thermal stimulation of their receptive fields by extracellular recording of action potential discharge activity in a skin–nerve in vitro preparation. We detected no significant effect of ASIC2 knockout on conduction velocity, mechanical and heat sensitivity of cutaneous afferents. The median von Frey threshold of C fibres was 22.6 mN (n= 12–13) for both wild-type and ASIC2 knockout mice and that of RA fibres was 4 mN (n= 12) for ASIC2+/+ and 1 mN (n= 13) for ASIC2−/− mice. The average heat threshold of mechano-heat-sensitive C fibres was 39.43 ± 0.81°C (n= 13, ±s.e.m.) for wild-type and 38.55 ± 0.63°C (n= 12, ±s.e.m.) for ASIC2−/− mice. So far our data are in good agreement with a previously published report by Price et al. (2000), who found that mechanostimulated responses of DH and AM fibres, as well as mechanical, acid and heat responses of C fibres, were unaltered by the ASIC2 knockout. However, Price et al. (2000) reported a decreased stimulus–response coding of RA fibres and, to a rather minor extent, of SA fibres in ASIC2−/− mice when compared to control mice.

RA fibres typically respond with a brief discharge at the onset of the mechanical stimulus and again when the applied force is relieved (Koltzenburg et al. 1997). Thus, RA fibres sense changes in stimulus intensities (Grigg & Del Prete, 2002) and represent typical sensors with differential transduction characteristics. However, in our experiments, the stimulus–response coding during both stimulus onset and offset were not decreased in ASIC2−/− mice when compared to age-matched wild-type littermates (Fig. 5). The stimulation protocols differ somewhat between our study and that of Price et al. (2000). We stimulated the skin with a defined force (controlled stress), whereas Price et al. (2000) used a protocol in which the skin was stretched by a defined distance (controlled strain). Both techniques are equivalent, since stress causes strain and vice versa (Grigg et al. 2002). Thus, loss of function of an important component of the RA fibre mechanosensor would affect the responses in both experimental configurations. Furthermore, in both our study and that of Price et al. (2000), the activation (von Frey) threshold of RA fibres was unaffected by the ASIC2 knockout, a fact that in our opinion strongly argues against ASIC2 as a crucial component of the RA fibre mechanosensor.

Discussion

Eukaryotic Na⁺-permeable mechanosensitive cation channels attract great interest because of their involvement in nociception, proprioception and hearing. In C. elegans, mutation of degenerin channel subunits affects mechanosensation of the nematode. In consequence, the degenerins were proposed to form mechanosensitive cation channels. By analogy, ASICs, the neuronal mammalian degenerin homologues, appeared to be very attractive candidates for mechanosensitive cation channels. However, our data do not favour this hypothesis, since auditory perception, visceral mechanonociception and cutaneous mechanosensation were unaltered after disruption of the ASIC2 gene.

Are C. elegans degenerins or certain ASIC channels mechanosensitive ion channels? The polls on this question are still open. In C elegans, functional degenerins (MEC-4 and MEC-10) and associated proteins (e.g. MEC-2 and MEC-6) are required for body touch sensation of the nematode (Huang et al. 1994). Gain-of-function mutants of the degenerins form ion channels and coexpression of such constitutively active degenerin mutants with ion channel properties and activity (Chelur et al. 2002; Goodman et al. 2002). Thus, degenerins can form ion channel complexes together with associated proteins that are required for touch sensation. Furthermore, the calcium increase provoked by light touch in C. elegans ALM neurones in vivo is abolished in mec-4 loss-of-function mutants, suggesting that mec-4 activity is required for the mechanosensitivity of the touch receptors. The loss-of mechanosensitivity of mec-4 null worms is probably not due to a general alteration of neuronal excitability, since the calcium signals after potassium depolarization in primary culture were similar for wild-type and mec-4 null worms (Suzuki et al. 2003). These data make a very clear case for degenerin ion channel activity being required for certain forms of mechanoperception of the nematode.

However, electrophysiological evidence that degenerins form mechanosensitive cation channels in C. elegans or after heterologous expression is still lacking. Furthermore, there is no evidence that loss of degenerin function causes loss of mechanosensitive currents in acutely dissected worms or in cultured C. elegans cells. In a very recent study, Jospin et al. (2004) analysed the gating properties of the UNC-105 degenerin, a proposed mechanosensitive cation channel (Liu et al. 1996) in acutely dissected worms. While a constitutively active amiloride-sensitive UNC-105 gain-of-function mutant has been recorded in muscle cells of a C. elegans unc-105(n506) strain, no mechanically gated amiloride-sensitive current has been detected in muscle cells from wild-type worms.

Mechanosensation in C. elegans is complex and requires the correct interplay of various transduction pathways. For example, loss of function of EGL-3, a proprotein convertase thought to be involved in neuropeptide processing (Kass et al. 2001), causes insensitivity to body touch. Surprisingly, the same mutation rescues the insensitivity to nose touch caused by genetic inactivation of the ionotropic glutamate receptor GLR-1 in interneurones that are postsynaptic to touch receptors. Therefore, it cannot currently be ruled out that the role of degenerin channels in mechanosensation of the nematode is indirect. For example, a transmitter molecule could be released after mechanical stimulation by the touch receptor or surrounding cells and activate or modulate degenerin channel activity. A similar mechanism, mechanostimulated ATP release and activation of ionotropic purinergic receptors, was proposed for the mechanosensation in mammalian hollow organs (Burnstock, 1999).

As for the ASICs, no alteration of ASIC channel activity by mechanical stimulation could be detected despite efforts by G. Champigny (personal communication) and others (Garcia-Anoveros et al. 2001). Furthermore, the biophysical properties (single-channel conductance, ion selectivity) and pharmacology of mechanosensitive cation channels recorded in primary cultures of mammalian sensory neurones (Cho et al. 2002; Drew et al. 2002) do not match those of heterologously expressed ASIC channels (Waldmann et al. 1997b; Champigny et al. 1998; Sutherland et al. 2001).

While the benzamil sensitivity of the mechanosensitive whole-cell current in the first study of mechanosensitive currents in sensory neurones suggested that ASIC channels might be mechanosensitive cation channels (McCarter et al. 1999), recent studies rather propose that other non-ASIC channels are the mechanosensors. Cho et al. (2002) identified two types of mechanosensitive cation channels in some 1400 excised patches from dorsal root ganglion sensory neurones: ‘low threshold’ (LT) and ‘high threshold’ (HT) cation channels with activation pressure thresholds of 10–20 and 60 mmHg, respectively. Both ion channels discriminate only poorly between Na⁺ and Ca²⁺, whereas ASIC2 exhibits a strong preference for monovalent cations. The LT channel has a single channel conductance of 51 pS, which is much higher than that of ASIC2a (13.4 pS) or the heteromultimeric ASIC2a/ASIC2b channel (13.7 pS; (Lingueglia et al. 1997). Conversely, the conductance of the HT channel, 13.7 pS, is very similar to that of ASICs. However, both HT and LT channels were resistant to 200 μm amiloride, a concentration that blocks ASIC2 (K_d= 28 μm; Champigny et al. 1998) and most other ASICs almost entirely. The amiloride insensitivity and a block of mechanically activated currents in rat sensory neurones by Ruthenium Red (IC₅₀= 503 μm), a drug that does not affect ASIC2 (G. Champigny, unpublished observations), was also demonstrated in a different study (Drew et al. 2002). The fact that an ASIC2 or ASIC3 knockout does not affect the amplitude and kinetics of mechanically activated currents recorded from cultured sensory neurones (Drew et al. 2004) also suggests that other non-ASIC channels are the mechanosensors in mammalian sensory neurones.

Degenerin ion channels in C. elegans are associated with other proteins (e.g. MEC-6) that are required for touch perception (Chelur et al. 2002) and, of course, yet unidentified associated proteins might alter channel properties and could be required for the mechanosensitivity of ASICs. To date, two families of stretch-activated ion channels have been identified and reconstituted in heterologous expression systems: the bacterial non-selective mechanosensitive cation channels (MscL and MscS; Booth & Louis, 1999) and the mammalian mechanosensitive K⁺ channels (TREK and TRAAK; Patel et al. 2001). Both families of mechanosensitive channels do not require coexpression of associated proteins for mechanosensitivity, indicating that tethering to other non-ubiquitous proteins is not a general requirement for mechanosensitivity of ion channels.

In knockout animals, altered expression of related or unrelated transcripts can mask or generate a phenotype. However, the unaltered mechanosensation in the ASIC2 knockout mice is probably not due to compensation by other ASIC subunits since: (i) expression of ASIC1 and ASIC3 is unaltered in ASIC2 knockout mice (Ettaiche et al. 2004); and (ii) ASIC2 is extremely conserved between species (human vs. mouse, 99% identical protein), suggesting that the principal physiological function of ASIC2 cannot tolerate mutations. Therefore, it seems rather unlikely that another ASIC subunit (<65% protein sequence identity) could, at unaltered expression levels, simply replace the function of ASIC2.

The fact that C. elegans degenerins and mammalian ASIC channels both belong to the same protein family does not necessarily imply a similar function, since the protein homology is rather low (25% homology, 13% identity between mec-4 and ASIC2) and both phyla separated more than 1 billion years ago (Hedges, 2002).

In conclusion, the lack of impact (present study) or the rather limited impact (Price et al. 2000) of an ASIC2 knockout on mammalian mechanosensation suggests that ASIC2 is not part of a mechanically gated cation channel important for cutaneous mechanosensation, hearing and visceral mechanonociception.