NompC TRP Channel Required for Vertebrate Sensory Hair Cell Mechanotransduction

See allHide authors and affiliations

Science  04 Jul 2003:
Vol. 301, Issue 5629, pp. 96-99
DOI: 10.1126/science.1084370


The senses of hearing and balance in vertebrates rely on the sensory hair cells (HCs) of the inner ear. The central element of the HC's transduction apparatus is a mechanically gated ion channel of unknown identity. Here we report that the zebrafish ortholog of Drosophila no mechanoreceptor potential C (nompC), which encodes a transient receptor potential (TRP) channel, is critical for HC mechanotransduction. In zebrafish larvae, nompC is selectively expressed in sensory HCs. Morpholino-mediated removal of nompC function eliminated transduction-dependent endocytosis and electrical responses in HCs, resulting in larval deafness and imbalance. These observations indicate that nompC encodes a vertebrate HC mechanotransduction channel.

Mechanical stimuli such as sound and movement excite sensory hair cells (HCs) by deflecting their apical stereocilia. The process by which HCs convert stereociliary deflections into electrical activity is termed mechanotransduction (1). A “gating spring” model for mechanotransduction (24) proposes that the deflections of the stereocilia exert tension on extracellular filaments, the tip links (5, 6), directly opening mechanosensitive cation channels. The resulting inward transduction current depolarizes the HC membrane and subsequently triggers neurotransmitter release at the HC afferent synapse (1, 3). Despite detailed insights into HC physiology, the molecular identity of the transduction channel remains unknown (3, 4).

Ion channels of the transient receptor potential (TRP) superfamily (4, 7, 8) (Fig. 1) have recently been implicated in a number of sensory processes, including vision (9), chemosensation (10), thermosensation (11), and osmosensation (12, 13). In addition, TRP channels exhibit biophysical and pharmacological properties consistent with those of an HC transduction channel, and have therefore been put forward as potential candidates for that channel (3, 4). The only TRP family member implicated thus far in neurosensory mechanotransduction is Drosophila NOMPC (14), an unconventional TRP channel exhibiting an unusually high number (29) of ankyrin (ANK) repeats (a protein-protein interaction motif) in its N terminus (fig. S1). Mechanoreceptor currents in touch-sensitive bristle neurons of nompC homozygous mutant flies are reduced (14), resulting in touch insensitivity and uncoordinated motility (15). The Caenorhabditis elegans nompC ortholog is also expressed in mechanosensory neurons (14). A vertebrate nompC homolog would therefore be an attractive candidate for an HC transduction channel. Extensive searches of genome databases and cDNA libraries, however, have until now failed to identify a vertebrate copy of the nompC gene (3, 4).

Fig. 1.

Phylogenetic tree of the TRP superfamily [see (8) for classification nomenclature], based on alignments done with the Clustal program (Megalign, DNASTAR), of one vertebrate representative member [zebrafish (Dr) or human (Hs)] and two invertebrate representative members [Drosophila melanogaster (Dm) and C. elegans (Ce)] of each class. The time scale at the bottom is in millions of years.

Combining computational and molecular biology techniques enabled us to identify a nompC homolog in zebrafish (Danio rerio) (GenBank accession number AY313897). The zebrafish nompC cDNA encodes a predicted 1614–amino acid protein, the overall domain structure of which is identical to that of invertebrate NOMPC (fig. S1 and Fig. 2B). Zebrafish NompC shares 45% identity and 62% similarity with Drosophila NOMPC, and 42% identity and 58% similarity with C. elegans NOMPC. In addition to their high degree of conservation at the protein level, all three genes share similar genomic organizations (Fig. 2A). Surprisingly, we were unable to find an obvious homolog of zebrafish nompC in either the human or mouse databases. However, it should be noted that the sequencing of either genome is not yet complete. In terms of TRP channels, zebrafish NompC belongs to the same subgroup (TRPN) as invertebrate NOMPCs within the phylogeny of the TRP superfamily (Fig. 1). Phylogenetic analyses indicate that TRPNs are most closely related to TRPVs, members of which are involved in osmosensation and have been implicated in mechanosensation (12, 13). This further supports the notion that mechanosensation and osmosensation are functionally related (3, 13) and suggests that TRPNs and TRPVs may have evolved from a common TRP mechanosensory ancestor. Finally, nompC appears to be present in a single copy in the fly, worm, and fish genomes. Altogether, these results suggest that the zebrafish gene we identified is an ortholog of invertebrate nompC.

Fig. 2.

MO-mediated reduction of nompC expression. (A) Zebrafish nompC genomic structure. The 31 exons are indicated by black boxes. Introns of known and unknown (not to scale) sequences are designated by black and dark gray upper lines, respectively. gt MO and atg MO target sites (arrows) and exon/intron boundaries conserved between fly and fish (asterisks) and worm and fish (arrowheads) are indicated. The nompC 3′ end is generated by transplicing of exons 30 and 31. Individual ANK repeats, the linker, and the TRP channel region are nearly always encoded by single exons (an example is indicated by the lines extending between exon 28 and the linker domain). (B) Schematic diagrams of the full-length NompC protein and of the truncated NompCgtMO variant. The gt MO–induced aberrant splice junction between exons 27 and 29 introduces a frameshift resulting in a nonrelated 59–amino acid C-terminus (striped box) followed by a premature stop codon. Parentheses indicate the last amino acid in wild-type NompC and the start of the frameshift in the truncated form. (C) “Morphotype” of nompC gt MO–injected zebrafish larvae. The upper RT-PCR product (790 bp) is derived from full-length transcripts, whereas the lower product (350 bp) is derived from exon 28–deleted transcripts. Each lane contains RT-PCR products derived from a pool of mRNA isolated from 5 to 10 individual larvae of one given phenotype (either WILD-TYPE or DEAF) at indicated stages (72, 96, 108, or 120 hpf). Larvae were selected for pooling only if they displayed an obvious behavioral phenotype.

nompC is expressed in zebrafish embryonic and larval HCs. nompC mRNA was first detectable in the embryonic ear by in situ hybridization at 48 hours post fertilization (hpf) (Fig. 3A). In larvae 5 days post fertilization (dpf), low levels of nompC transcript were found in all five innerear sensory patches, and nompC transcript was restricted to the HC bodies found in the upper two-thirds of the neuroepithelia (Fig. 3, B to D). nompC mRNA was not detected by this method in lateral-line HCs, but reverse transcription polymerase chain reaction (RT-PCR) with adult skin RNA preparations containing lateral-line HCs yielded a nompC-specific band (Fig. 3E). Transient expression of nompC was also detected at earlier stages (24 to 60 hpf) in discrete regions of the brain (the pineal organ and isthmus), in the dorsal tip of the caudal neural tube, and in migrating neural crest cells in head and anterior trunk regions (16). In adult tissues, we detected nompC message in the ear and eye, and to a lesser extent in the brain, gills, and testes (Fig. 3E). In C. elegans, nompC is expressed in nonmechanosensory neurons as well (14), which together with our data suggest that nompC may have other functions in addition to its function in HCs.

Fig. 3.

nompC expression in sensory HCs in the zebrafish inner ear. In situ hybridizations of whole embryos or larvae are shown. (A) Lateral view of the embryonic ear at 48 hpf. Labeling of the sensory patches of the anterior macula (am) and posterior macula (pm) is shown. (B) Lateral view of the larval ear at 120 hpf. The focal plane includes the medial crista (mc) and posterior crista (pc) in the developing semicircular canals. Some labeling that is apparent in epithelial columns of the semicircular canals and the inner cavity may represent background staining. (C and D) Dissected inner ear sensory patches from larvae 120 hpf. (C) Isolated anterior macula. (D) Isolated medial crista from a semicircular canal. Scale bar in (D) indicates 50 μm in (A) and (B); 10 μm in (C); and 12 μm in (D). ov, otic vesicle; hc, hair cells; sc, supporting cells. (E) Expression of nompC in adult zebrafish tissues. Homogenates of the indicated tissues were subjected to semiquantitative RT-PCR. A PCR product obtained with primers for elongation factor 1 α (ef1α) is shown as a loading control. The left lane shows a 1-kb ladder.

To investigate the function of nompC in vivo, we used a morpholino (MO) antisense oligonucleotide loss-of-function approach. MOs directed to splice donor sites (gt MOs) disrupt gene function by interfering with the targeted splicing event (17). We designed a gt MO targeted to the splice donor site of intron 28 (Fig. 2A). Binding of this gt MO to the endogenous pre-mRNA should lead to a deletion of exon 28, thereby introducing a frameshift resulting in a truncated NompC protein devoid of the entire TRP ion channel region (Fig. 2B). Comparison of RT-PCR products of uninjected control and morphant mRNA extracts (morphotyping) allowed us to monitor the MO efficiency directly (Fig. 2C). Subsequent sequencing of the RT-PCR products confirmed that in the morphant transcript (nompCgtMO mRNA), exon 28 was indeed deleted (16).

nompC morphant larvae were scored for the acoustic startle reflex starting at 72 hpf and were subsequently morphotyped. In contrast to uninjected control larvae that exhibited a normal startle reflex, nompC morphants did not respond to acoustic stimuli, indicating that they were deaf (Fig. 2C, DEAF; movie S1). The number of HCs and their morphology were unaffected in nompC morphants, ruling out the possibility that their deafness phenotype resulted from MO interference with HC development or from abnormal HC degeneration. In addition, nompC morphants could be induced to swim by touch stimuli. In contrast to uninjected control larvae, nompC morphants often swam erratically in a circular path or at a tilt, indicating a defect in balance or vestibular function (16). As expected from the inherent mosaicism and gradual dilution of the MO at late stages, the penetrance and expressivity of these phenotypes were variable and also depended on the genetic background. However, at 72 hpf, all deaf morphants were found to almost exclusively express nompCgtMO transcripts, whereas sibling morphants of wild-type phenotype always expressed high levels of full-length transcripts in addition to nompCgtMO mRNA (Fig. 2C). Furthermore, all 72 hpf deaf nompC morphants recovered by 96 hpf, and their morphotype at this later stage had also recovered to the wild-type condition, with the amount of full-length transcripts exceeding the amount of nompCgtMO transcripts (Fig. 2C). Hence, loss of nompC function and deafness and imbalance phenotypes were correlated, indicating an important role for nompC in auditory and vestibular function.

To confirm this notion and to further control for phenotypic specificity, we performed additional MO injections. A MO targeted to the nompC 5′ end (atg MO), which is expected to block NompC protein synthesis, phenocopied the gt MO deafness and imbalance phenotypes (fig. S2) (16). In contrast, neither a 4–base pair (bp) mismatch variant of the nompC gt MO (4bpmm MO) nor a MO targeted to an independent gene expressed in HCs (the 2P domain K+ channel encoding gene twik2) had any effect on auditory and vestibular function (16). Altogether, these results pointed to a specific requirement for nompC in larval hearing and balance.

To test whether nompC is required for mechanotransduction, we examined HC function using a vital dye. In intact zebrafish, superficial HCs of the lateral-line system take up the membrane marker FM1-43 in a mechanotransduction-dependent fashion via apical endocytosis (18). In gt MO and atg MO nompC morphants lacking an acoustic startle reflex, FM1-43 uptake was abolished (Fig. 4B and fig. S3) although HC morphology was normal (Fig. 4E and fig. S3). In contrast, FM1-43 uptake was not compromised in 4bpmm MO– and twik2 MO–injected larvae (fig. S3). Within the mosaic population of nompCgtMO larvae, the reduction of FM1-43 uptake correlated with the severity of the behavioral phenotype (Fig. 4C). After recovery at 96 hpf, when all gt MO–injected larvae mainly expressed the full-length nompC transcript (Fig. 2C), FM1-43 uptake was restored (Fig. 4D). Together, these results show that FM1-43 uptake by HCs requires nompC function and are thus consistent with a critical role for NompC in mechanotransduction.

Fig. 4.

Absence of mechanotransduction-dependent apical endocytosis and extracellular potentials in nompC gt–morphant HCs. [(A) to (F)] Uptake of FM1-43, a marker of endocytosis, in lateral-line HCs of intact zebrafish larvae. Clusters of HCs (neuromasts) in the head region are shown. (A) Un-injected control larva at 80 hpf (n = 10 larvae). (B) Lack of apical endocytosis in a nompC morphant with balance defects and without acoustic startle reflex at 84 hpf (n = 11). (C) Apical endocytosis in a nompC MO–injected larva with normal balance and acoustic startle reflex (n = 5). (D) Normal endocytosis in a nompC morphant that recovered after 24 hours (restored balance and hearing) (n = 7). [(E) and (F)] Higher-magnification view of a neuromast in a nompC morphant [the same as shown in (B)]. (E) Bright-field view of morphant HCs. Arrowheads indicate two HC bodies, which are in focus. Arrow indicates the apical HC bundles. (F) Absence of apical endocytosis. Scale bar in (E) indicates 130 μm in (A) to (D) and 10 μm in (E) and (F). [(G) to (I)] Extracellular potentials are reduced or absent in nompC morphant HCs at 80 hpf. (G) Upper trace, microphonic potentials of a wild-type neuromast in response to a saturating sinusoidal bundle deflection (at 20 Hz for 200 ms; see bottom). Middle traces, recordings from a nompC morphant neuromast (no microphonics are detectable) and from a recovering nompC morphant neuromast (some microphonics are detectable). Lower trace, control recording under the same conditions but remote from a neuromast, to establish the recording noise level. Traces are averages of 262 presentations. (H) Power spectra of responses during stimulus presentation. The spectrum of the wild-type response shows clear peaks at integer multiples of the stimulus frequency, especially at 20 and 40 Hz. The power spectrum from nompC morphants is not distinguishable from the recording noise level. (I) Summary of microphonic responses, quantified by the total power between 17 and 43 Hz (± SD). The power in nompC morphant responses (n = 10) is significantly lower than in wild-type responses (n = 8; P < 0.0001, Wilcoxon rank-sum test) and is indistinguishable from noise controls (n = 8).

To more directly assess mechanotransduction in wild-type versus nompC morphant HCs, we measured stimulus-evoked microphonic potentials in lateral-line HCs of intact zebrafish larvae. Microphonic potentials are generated by the transepithelial return currents that result from the inward transduction current in the stereocilia and the voltage-dependent currents in the basal membrane of HCs (19). Because all membrane currents of HC responses are triggered by the transduction current, microphonic potentials should be completely abolished by inactivation of the transduction channel but only partially affected by inactivation of voltage-gated channels (19). Microphonic responses of control larvae to saturating (>10°) sinusoidal hair bundle deflections showed the characteristic oscillatory pattern with strong frequency components at once and twice the stimulus frequency (20 and 40 Hz, respectively; Fig. 4, G and H) (20). Response intensities, quantified by the total power between 17 and 43 Hz, exceeded the recording noise level by a factor of >250 (Fig. 4, H and I). Because the nompC gt MO had a reversible effect on hearing behavior and FM1-43 uptake, we measured microphonic potentials in both deaf and recovering injected larvae. In all deaf nompC morphants examined, no microphonic potentials were detected and recordings were indistinguishable from the recording noise level (Fig. 4, G to I). In contrast, substantial responses were always observed in recovering larvae (Fig. 4G). Hence, microphonic potentials correlated with the severity of the behavioral phenotype. Together, these results show that nompC is required for electrical responses of HCs to mechanical stimuli.

Our study demonstrates that nompC encodes an essential component of vertebrate HC mechanotransduction. Because removal of nompC function completely abolishes HC microphonic responses, we propose that nompC encodes a mechanotransduction channel. To confirm this notion, further studies will be required to localize the channel in HCs and to characterize its physiological properties. The finding that nompC is required for HC mechanotransduction also provides further evidence that the Drosophila sensory bristle and vertebrate HC transduction machinery may have evolved before arthropods and chordates diverged. Indeed, the physiological properties of both mechanosensory systems are strongly analogous, because both invertebrate and vertebrate transduction apparatus bathe in a K+-rich endolymph, show directional sensitivity and microsecond latencies, and adapt to sustained mechanical stimuli (3, 4, 14).

Supporting Online Material

Materials and Methods

Figs. S1 to S3


Movie S1

References and Notes

Stay Connected to Science

Navigate This Article