The Hair Cell's Transduction Channel

A hair cell fails to hear when a gene is missing, and the gene’s protein sits in the right place. Is that enough to call the protein the channel? It is not, and the history of TMC1 shows how much more is asked.

Hearing and balance covers tip links, the gating spring and the driving force, and gives TMC1 one sentence as a pore component. Here we start where that sentence stops: what each experiment proved, what it left open, and why the answer is a complex of proteins rather than one. The gating logic of channels is in ion channels as receptors, and earlier in this course mechanotransduction channels compare candidates across senses.

The idea: what would count as proof of a channel

A protein can be required for a current without forming the pore. A trafficking factor, an anchor or a lipid-modifying enzyme could all be needed, and deleting any of them would silence the cell. The field therefore asks for a ladder of evidence, and each rung rules out a different alternative.

• Localization. The protein sits at the lower end of the tip link, where the channel is thought to be.
• Requirement. The conventional current disappears when the gene is removed.
• Rescue. Putting the gene back restores the current, in cells of the mutant genotype.
• Permeation. A change in the protein’s sequence changes which ions pass, not only how much current flows.
• Acute accessibility. A reagent applied to a cell that already carries current alters permeation through an engineered residue. This excludes a story in which the mutation merely changed how the protein was made or delivered.
• Purified activity. The isolated protein, in a membrane with no hair cell around it, responds to force.

No single paper climbs every rung, and the rungs differ in species and preparation. Keep that in view through the rest of the lesson. Also note what “the channel” means. Even a complete ladder says that the protein forms or lines the pore. It does not say how force from the tip link reaches it.

Requirement and rescue, then permeation

Kurima and colleagues (2002) linked human TMC1 mutations to dominant deafness (DFNA36) and to recessive deafness (DFNB7/B11). That was a human pedigree result: the gene matters for hearing. It said nothing about function.

The first functional rung came from mice. Kawashima and colleagues (2011) recorded from neonatal mouse hair cells, utricular cells at P1 to P8 and cochlear outer hair cells at P2 to P4 (base) and P5 to P7 (apex), in excised tissue. Cells lacking both Tmc1 and Tmc2 had no conventional transduction current. In double-null cultures, expressing either TMC1 or TMC2 restored it. Tmc2 is expressed only transiently in the early postnatal cochlea, which is why a Tmc1 single mutant can still show early currents. This is requirement and replaceability in neonatal mouse hair cells. It does not separate a pore from a factor the pore needs.

The next step changed the protein instead of removing it. The Beethoven allele of Tmc1 substitutes lysine for methionine at residue 412 in the mouse protein (Vreugde and colleagues, 2002, described the line as a model of dominant progressive hearing loss). Pan and colleagues (2013) recorded from excised neonatal mouse inner hair cells with Tmc2 removed, so that only mutant TMC1 was left. Calcium permeability was reduced, and single-channel currents were smaller than in cells with one wild-type Tmc1 allele and no Tmc2. Hair cells carrying Tmc2 without functional Tmc1 showed large single-channel currents. The authors proposed that mixed assemblies of TMC subunits could explain some of the unitary-current differences; that was an interpretation, not a subunit count.

Two cautions. Beethoven is a mouse allele and the residue number is the mouse protein’s, so nothing here is a diagnosis of the composer. And permeation can be altered by changing a partner, so this rung is strong but not conclusive.

Accessibility and reconstitution

Pan and colleagues (2018) built the acute-accessibility test. They delivered Tmc1 constructs carrying single cysteine substitutions into Tmc1/Tmc2 double-null mouse hair cells with a viral vector at P1, excised the cochleae at P5 to P7 and cultured them for a further 3 to 13 days. They then added a membrane-impermeant cysteine-reactive reagent to cells that were already carrying current. Such reagents react with cysteine side chains in water. The reagent changed the permeation properties of the TMC1-mediated current, and for a D569C construct, the fastest-reacting of the cysteine constructs tested, the current was inhibited with a time constant of 191 +/- 62 ms (n = 7, 2 mM reagent); other constructs reacted far more slowly, up to about 17 s. Because the current existed before the reagent was added, a change cannot be explained by altered synthesis or delivery. The authors called this compelling evidence that TMC1 is a pore-forming component. The cells were still cultured hair cells with their native partners, and no purified mammalian TMC1 was tested alone.

Jia and colleagues (2020) asked what happens without the hair cell. They produced truncated turtle TMC1 and budgerigar TMC2 constructs in insect cells, purified them, put them in liposomes, and found that applied pressure activated channel activity. This supports intrinsic mechanical responsiveness in a reduced system. The proteins were not mammalian, and it was not a native complex.

In 2025 two groups addressed a question left over from the older literature: can TMC1 and TMC2 make a mechanically gated current in a non-hair-cell host? Fu and colleagues reported mechanically activated currents from wild-type human TMC1 and TMC2 in cultured cells after a genome-wide CRISPR interference screen identified a way to raise surface expression. Chen and colleagues obtained mechanosensitive activity from full-length mouse TMC1 and TMC2 expressed in PIEZO1-knockout HEK293T cells after adding a lipidation tag, and TMIE raised the channel activity in that assay. These are recent, heterologous results with assisted targeting. They strengthen the case for the pore, and they leave the native force coupling open. (The CRISPR lesson covers screens of this kind.)

Other mechanically activated currents do not automatically count. A PIEZO2-associated reverse-polarity current in neonatal mouse outer hair cells (Wu and colleagues) is distinct from conventional tip-link transduction, and mouse TMEM63A and TMEM63B, but not TMEM63C, gave stretch-activated currents in HEK cells (Murthy and colleagues, 2018) without any test of native hair-cell necessity or rescue.

A complex, not a protein: TMIE, LHFPL5, CIB2 and a worm structure

Jeong and Clark and colleagues (2022) purified a tagged native TMC-1 complex from bulk-grown C. elegans and solved it by cryo-EM. The main assembly held two copies each of TMC-1, CALM-1 and TMIE, and each TMC-1 protomer had ten transmembrane helices. CALM-1 is a worm calcium-binding protein; it is not a resolved mammalian CIB2. The human complex in the paper was a homology model built on the worm structure. See the structure listed with this lesson (7USW) and treat it as a worm structure.

Partners change what you measure:

• TMIE. In mouse cochlear outer hair cells, TMC1 and TMC2 did not support transduction without TMIE (Cunningham and colleagues, 2020, P3 explants). Deafness-linked Tmie mutations lowered unitary current and changed calcium permeability in P3 outer hair cells, an age at which TMC2 is the TMC protein prominent in the bundle, and charged residues in the cytoplasmic end of TMIE bound phospholipids including PIP2.
• LHFPL5 (TMHS). It binds the tip-link protein PCDH15, and the neonatal Tmhs-null cochlea showed reduced transducer conductance (Xiong and colleagues, 2012).
• CIB2. Mouse hair cells without functional CIB2 had no detectable normal-polarity transduction at P4 to P7, and CIB2 interacted with TMC1 and TMC2 in biochemical assays (Giese and colleagues, 2017). Later, CIB3 expression rescued Cib2-null cochlear cells, and a co-crystal showed CIB3 bound to a TMC1 intracellular fragment (Liang and colleagues, 2021).

The localization evidence conflicts. Giese and colleagues used antibodies and did not detect mislocalization of TMC1 or TMC2 in Cib2 mutants. Liang and colleagues used endogenous epitope-tagged TMC1 and TMC2 and could not detect them in stereocilia of Cib2-null cells. Which method to trust is open.

Worked problem: a conductance from a real recording. Beurg and colleagues (2006) recorded from rat inner hair cells (P6 to P11, excised organ of Corti, low-frequency position, 1.5 mM external calcium). The mean unitary current was 15.0 +/- 1.6 pA (SD, n = 7) at -84 mV, at room temperature (19 to 22 C). Take the reversal potential as +3 mV and assume a linear current-voltage relation.

• (a) The driving force is -84 - (+3) = -87 mV. Conductance g = i / (V - E_rev) = (-15.0 pA) / (-87 mV) = 0.172 nS, or about 172 pS.
• (b) For a made-up ensemble of 100 identical, independent channels at open probability 0.20: I = N x i x P_open = 100 x (-15.0 pA) x 0.20 = -300 pA, an inward current of 300 pA. The count and probability are teaching assumptions, not measurements.
• (c) The same paper estimated about 260 pS for inner hair cells, which differs because that estimate was made at 0.02 mM calcium, not 1.5 mM. Calcium is thought to impede passage of monovalent cations, so a conductance without its ionic conditions is not a complete number.

g = i / (V - E_rev), and the whole ensemble current is I = N x i x P_open

There is no single hair-cell conductance. In the low-calcium estimates of Beurg and colleagues (2006), outer hair cells rose from 145 to 210 pS between sampled low- and mid-frequency positions in the rat cochlea, whereas inner hair cells were about 260 pS at both. Beurg and colleagues (2018) found a conductance gradient in outer hair cells of Tmc2-null mice (P3 to P7), so TMC2 is not needed to produce a gradient, and they proposed that varying numbers of cooperating channels per complex contribute. Neonatal gradients do not give the composition of the adult complex.

Adaptation: two time scales, one unsettled calcium story

During a maintained deflection, the transduction current declines and the activation curve shifts. This is adaptation, and it has a fast and a slow phase. Fast and slow name kinetics, not molecules. In the 2020 work the fast phase ran at about a millisecond or faster and the slow phase at about ten milliseconds.

In mouse inner hair cells (P6 to P9 in the cells of that figure, fluid jet, -81 mV, room temperature 22 to 24 C, n = 6, mean +/- SEM), Corns and colleagues (2014) reported time constants of 0.82 +/- 0.31 ms (fast) and 11.4 +/- 4.3 ms (slow) and concluded that calcium entry drives adaptation. Peng, Effertz and Ricci (2013) stimulated rat hair cells (P6 to P10), mainly with a stiff probe and with fluid-jet controls, and found fast adaptation that persisted without calcium entry. The two studies differ in species, stimulus and handling of bundle motion. Caprara and colleagues (2019) showed that step-like force and step-like displacement stimuli give different manifestations of adaptation in rat outer hair cells, so stimulus mechanics could contribute to the disagreement. The authors presented this as reconciling the data in favour of calcium-independent fast adaptation; the lesson leaves the question open.

The slow phase has its own dispute. Selective inhibition of an engineered Myo1c blocked adaptation in mouse utricular hair cells (Holt and colleagues). In 2020, Caprara, Mecca and Peng compared hair-bundle creep with slow adaptation in several preparations and found them uncorrelated; cochlear slow adaptation did not require Myo1c activity, and the Myo1c result differed between cochlear and vestibular cells. That challenges the classic motor model of slow adaptation. It does not rule out every function of myosin.

Design problem

A laboratory reports mechanically gated current from a new candidate protein in a cultured cell line. Design the experiments that would let the field call it the hair-cell transduction channel, and say what each cannot show.