A Channel at Large

Since that 1983 paper by Corey & Hudspeth (cited 250 times), reams of data have supported the channel's existence. But the narrative of the hair cell channel story diverges from that of other channels. Numerous channels have been characterized, purified, sequenced, and cloned in the years since the hair cell mechanotransduction channel was discovered. Yet the hearing field has been left spinning its wheels in a molecular biology ditch. Scientists are still working to culture the cells, and the molecular identity of the transduction channel remains a mystery.

There are a number of reasons. For one, the amount of channel protein is in such low abundance that no one has been able to isolate it. "The first night that I was visiting Jim Hudspeth's lab, some people showed me the frog sacculus preparation. I had nightmares, literally, of how little material there was," Gillespie says.

"Even though I said it's frustrating to do sIRNA, in a way we've got to get that to work. We're not going to make [knock-out] mice for 25 proteins." 

-Peter Gillespie

The frog is used as a model because of its large hair cells, but it has only a few thousand of these cells with 100 stereocilia per cell. Other animals (fish, mice, humans) have comparably small numbers of hair cells. On each stereocilium likely exists only one mechanotransduction channel. Compare that to the retina, which has six orders of magnitude more cone photoreceptors than hair cells, and each outer segment of the photoreceptor possesses millions of copies of rhodopsin. To purify the mechanotransduction channel from hair cells, scientists "would need hundreds of thousands of animals," says Corey, now at Harvard Medical School.

They'd also need a high-affinity ligand. Tetrodotoxin binds specifically to sodium channels, and the purification technique Numa and colleagues used pulls out proteins bound to the toxin.³ For the mechanotransduction channel, "people were finding it was a nonselective cation channel," says Corne Kros at the University of Sussex, but no one yet has identified a specific channel blocker. In the 1990s Kros and his colleagues began examining a number of different molecules, in particular, amiloride and its derivatives. They found that amiloride could plug up the open transduction channel.⁴ "We found [that the transduction channels] were very different from almost everything, apart from the stretch sensitive channel in frog oocytes," Kros says. Amiloride seemed like a promising channel blocker, but scientists in the field of hearing would come to realize that hair cells seem to have more of an affinity for false leads than for available ligands.

Several years later, using the fluorescent dye FM1-43, Kros found that amiloride and antibiotics can slip through the transduction channel.⁵ Though Kros says he was probably wrong in his initial interpretation that amiloride was not passing through the channel, it was an exciting finding because of evidence that antibiotic use can lead to deafness in patients. "On the basis of that [finding], we think that the transduction channel has a much bigger pore than was initially presumed," Kros says.

Kros also determined that FM1-43 permanently blocks the transduction channel. But, the block was not enough to do what Numa and his colleagues had done with the sodium channel. "If you want to use a blocker to isolate a channel, you need high affinity in the nanomolar range to use it as a handle to pull the channel out. Even FM1-43 is not quite in that class."

Even as one way to identify the channel proved fruitless, other opportunities soon became apparent. In 1987, researchers found a new way to clone channels. They had found a mutant Drosophila whose legs shook when the fly was anesthetized with ether. These now well-known "shaker" flies had alterations in their potassium channel currents, and a team from the University of California, San Francisco, was able to use the mutations to identify the potassium channel gene.⁶

That finding afforded a new way to find the mechanotransduction channel. "The mutant strategy, in principle, should work," says Corey. Dozens of hearing mutants or balance mutants have been found in flies, zebrafish, mice, and humans. In 2000, one such mutant in Drosophila, nompC, put the hearing field at the edge of its seat.

On a July afternoon, the lunchtime crowd empties out of a courtyard patched with sun and shade at Oregon Health and Science University (OHSU). Richard Walker finishes a quesadilla and recounts how he pulled nompC from the crowd of a genetic screen. Charles Zuker's laboratory at the University of California, San Diego, had screened flies for defects in mechanotransduction. The collection was relatively uncharacterized when Walker joined the lab as a postdoc, and as he began to scrutinize the flies electrophysiologically, a particularly interesting mutant with four alleles caught his attention. "These flies are completely uncoordinated," Walker says. "They flop around like a drunken sailor, and [the mutation] is actually lethal in adults."

In physiological examinations Walker found that one of the alleles had a transduction current that would adapt much faster than wild-type flies. "The other three alleles all had almost all the transduction current abolished. We put two and two together and said, "'that's probably an important gene.'"

It was not just an important gene, but also perhaps the gene for the mechanotransduction channel. Walker did positional cloning to identify the nompC gene, which is homologous to members of the transient receptor potential (TRP) channel family.⁷ He and his colleagues titled a paper on their findings (cited 198 times): "A Drosophila mechanosensory transduction channel." Gillespie says the feeling was, "I think we found it." Walker says, "it was a pretty big deal. It was heralded as a definite step forward in understanding mechanotransduction."

But the excitement was short-lived: nompC isn't present in any mammals. Teresa Nicolson, also at OHSU, found the gene in zebrafish, but humans, mice, even other species of fish lack the gene. "It's a mystery why it disappeared in higher vertebrates," Nicolson says.

It's clear the nompC channel - which is also called the TRPN1 channel - is not the transduction channel in mammals, and it appears not to be the only channel in animals that do express it, since nompC null mutants still have a small transduction current. "NompC might be partly responsible for the transduction channel, but it doesn't explain the whole story because of the residual current," Nicolson says. Whatever is responsible for that current is still unknown, but it's possible that therein might be the answer to the mammalian transduction channel, says Walker.

"It's as likely to be a TRP channel as it is not. Personally, I don't think it is."

-Robert Fettiplace

While hair cell research has frustrated some researchers, it hasn't been all failures. Identities of the other components of the transduction apparatus have emerged over the years. Most recently, Gillespie, Ulrich Müller at the Scripps Research Institute, and their colleagues - in perhaps an uncommon instance of serendipity and relative ease in the hearing field - nailed down the identity of the tip link, the molecule that stretches between any two stereocilia and is thought to open the transduction channel gate.

Müller's lab had been working on cell-to-cell interactions at the time when other researchers, including Nicolson, began cloning genes from deafness mutants. One gene that emerged from the screens, cadherin 23, is an adhesion molecule that caught Müller's attention. "The properties were similar to what we knew about the biochemical features of the tip link," Müller says. Mechanotransduction was nearly gone in the mutants, and stereocilia were splayed rather than bundled together. While Müller's lab was not involved in hair cells at the time, he decided to pursue the research. He recruited a postdoc to pioneer hair cell research in his laboratory, and sure enough, cadherin 23 turned out to be a main component of the tip link.⁸

While molecular biologists are teasing out the tip link and other proteins in the transduction apparatus, biophysicists are whittling away the channel's mysteries by uncovering some of its physical properties. "We're trying to create a fingerprint of what this native channel is like, so as we try to compare to candidate channels, we know whether we're right or wrong," says Tony Ricci at Stanford University.

Ricci has found that the channel has unique properties that don't commit it to any known family. The biophysical properties and structure are similar to a TRP channel and an amiloride-sensitive sodium channel (ENac), "but not specifically any one of them." The transduction channel is 60% permeable to calcium, similar to a cyclic nucleotide channel, but the conductance is different between them.

The pore is about 1.2 nm in diameter, with a vestibule on the outside that allows large molecules to enter. The structure might explain how large molecules such as amiloride and antibiotics can enter the stereocilium, but once inside, can't get back out. "Like a Trojan horse, they don't go out anymore and wreak havoc in [the] cell," says Kros.

Ricci says determining the properties of the channel has been "tricky, because a lot of properties associated with the channel might not be intrinsic to the channel ... but might be part of other machinery" associated with the channel. For this same reason, many hair cell researchers doubt the ability to take advantage of another technique widely used in channel research: expression of the channel in frog oocytes. Oocyte expression is used to confirm properties of a channel in the absence of confounding cellular components, but in the transduction channel, the tip link and other machinery might be necessary for function. Gillespie says such challenges beg the question: "How do you really prove what you've got is right?"

"I had nightmares, literally, of how little material there was."

-Peter Gillespie

After Walker and his colleagues suggested that the transduction channel might be a TRP channel based on their work on nompC, Corey took the approach of a biased genetic screen, scanning through all the roughly 30 mouse TRP channels to see if any matched the transduction channel. One candidate, TRPA1, possessed qualities essential for the transduction channel: It was expressed in hair cells right around the time, developmentally, when the cells become mechanically sensitive, and an antibody to the channel became bound to the tips of the stereocilia, where the channel is thought to be located.⁹ Furthermore, inhibiting the function of the channel resulted in far less transduction current. "That was the strongest evidence; we were really convinced by this," says Corey. (His paper describing this candidate channel was cited 47 times within a year of being published, and a total of 151 times.)

To get conclusive evidence, however, Corey and his team created a mouse with a large chunk of the channel excised. Corey and others thought that 2005 might have been the year of cornering the channel, but the knockout mouse data were disappointing: The animals could hear just fine. As he wrote in a 2006 review article, "In the past 10 years, a variety of candidates have appeared, only to disappear, wraith-like, in the clear light of further experiments."¹⁰

Robert Fettiplace at the University of Cambridge says that of the TRP channels, only one, TRPP, has nearly identical size and calcium permeability to the transduction channel,¹¹ yet there's no evidence of its existence in hair cells. While TRP channels still remain a possibility, "it's as likely to be a TRP channel as it is not," says Ricci. "Personally, I don't think it is."

The possibility of identifying the transduction channel thrills researchers, not only because the hunt for this elusive molecule could finally come to a close, but also because so many research possibilities would open up with having the transduction channel in hand. "There are wonderful things we could do - find out how it's connected to the tip link, to the actin cytoskeleton," Corey says. Understanding how the channel fits in to the transduction apparatus and the stereocilia could help illuminate repair mechanisms, such as the restringing of tip links after hearing loss from a rock concert, and could help in determining why those repair mechanisms break down.

Moreover, knowing the sequence of the channel and its "friends" could lead to understanding perhaps some of the hundred forms of nonsyndromal deafness, says Hudspeth, now at Rockefeller University. "In the long run one of our goals is at least to understand that, and for those who wanted to be cured, to try to alleviate that," he adds. One of Hudspeth's major lines of research, how the ear amplifies sound, might also have something to do with the channel, and having the channel in hand could go a long way in resolving how the ear accomplishes amplification (see 12 If those cells turn out to be physiologically and structurally similar to hair cells in vivo, the culture could lend itself to RNAi experiments and provide material in greater abundance than ears.

Ulrich isn't the first to think of it. "A really obvious experiment is to do siRNA suppression of everything you can think of," Gillespie says. One of his postdocs spent years trying to design a culture for siRNA experiments, but the cells didn't stay healthy in culture and measuring mechanotransduction was not reliable. Still, Gillespie says it needs to be done, and he's willing to invest more of his laboratory in developing an assay. "Even though I said it's frustrating to do siRNA, in a way we've got to get that to work. We're not going to make [knockout] mice for 25 proteins."

Gillespie's other approach is biochemical: Take all the proteins in the hair cell and examine them with mass spectrometry, in what he calls shotgun proteomics - "sequence everything." The difficulty here is that the transduction channel will be one of the least abundant proteins in the mix, and there aren't many criteria for whittling down the candidates, except at least a transmembrane domain and some sort of intracellular domain.

But then what? In terms of convincing the field that a candidate is the real deal, "the standard will be extremely high, particularly because TRPA1 was such a good target ... but the genetics didn't work out," says Ulrich. "Even a knockout is not good enough in my mind," Gillespie says, "because all it says is that [the] channel is essential for hair cells, not that it's the transduction channel." Gillespie is creating mice with mutations in various TRP channel candidates, which would allow it to be inhibited from outside the hair cell. Then, Gillespie can stimulate the hair cell and observe whether the inhibitor affects the transduction current.

Corey's laboratory at Harvard Medical School is filled with cutting-edge equipment: a new two-photon microscope to track calcium's movements into and out of cells, floating floors to cut down on vibrations, and a machine shop so lab members can make their own equipment. In one of the floating rooms, Corey points out a laser-equipped rig that can be used to manipulate tiny movements in stereocilia using optical tweezers. Corey wants the transduction channel to be identified so he can use this equipment to do the really interesting experiments, such as determining how the whole transduction apparatus is constructed and how the system has developed such remarkable sensitivity. "We hope to find the transduction channel, and move on."