HHMI-NIH Research Scholars
Learn about the HHMI-NIH Research Scholars Program, also known as the Cloister Program. More

Janelia Farm Research Campus
Learn about the new HHMI research campus located in Virginia. More

Summary: David Corey studies mechanically activated ion channels and sensory transduction in the inner ear.

Work in my laboratory is focused on understanding the gating and regulation of mechanically sensitive ion channels. Much of this work involves the hair cells of the inner ear, which convert the mechanical stimulus of a sound wave into an electrical signal that is sent to the brain. The mechanosensitive organelle of the hair cell is a bundle of stereocilia that protrudes from the top surface of the cell. Stereocilia are connected at their tips by fine filaments called tip links, which are pulled each time the bundle is deflected in one direction by a sound vibration. Tip links are thought to pull directly on ion channels in the tips of the stereocilia, which open in response to the tension, allowing electric current in the form of potassium ions to flow into the hair cell to change its internal voltage.

Mechanics of Adaptation
We wish to understand the action of Ca²⁺ in mediating a fast adaptation in hair cells. Because fast adaptation apparently produces a fast movement of the hair bundles, we used a gradient-force optical trap ("laser tweezers") to apply minute forces to hair bundles within microseconds and to measure the resulting movement. We simultaneously measured the channel opening and closing with electrical recording.

	Mechanics of adaptation depend on membrane potential... Mouse cochlea infected with an adenovirus... Mouse utricle infected with an adenovirus...

Ca²⁺ enters through transduction channels and acts at an intracellular site to close them. We asked where in the transduction apparatus Ca²⁺ acts, by measuring the movement that occurs when Ca²⁺ is allowed into stereocilia. Ca²⁺ could bind to an internal elastic protein to change its stiffness, it could bind to a protein that then relaxes by a fixed distance, or it could bind to the transduction channel itself to change the relationship between force and open probability. Each model predicts a different dependence of movement on the static bias applied to the bundle. Only one model fits the data: Ca²⁺ binding to the channel (or a closely associated subunit) tends to close the channel, so that it takes 2–3 piconewtons more force to open when Ca²⁺ is bound. We also confirmed earlier measurements that indicate the channel moves by about 2 nanometers when it opens.

Molecules of the Transduction Apparatus
A hair cell only needs about 100 mechanosensitive ion channels to sense sound stimuli. Consequently, there are very few copies of each protein in the transduction complex, and identifying them has been difficult. To find a candidate for this transduction channel, we guessed that it might be a member of the TRP (transient receptor potential) ion channel family, many members of which are involved in sensory transduction, including mechanosensation.

We first asked whether any of the 33 mouse TRPs are made by hair cells, by testing in situ hybridization probes for each one. Probes for one, called TRPA1, labeled several hair cell regions in the inner ear. To determine whether TRPA1 is made by hair cells themselves, and where it is located in the cell, we made antibodies to the protein. Antibodies labeled the tips of the stereocilia in both frog and mouse hair cells, where we expect transduction channels to be located. Surprisingly, the TRPA1 labeling leaves the stereocilia in several minutes if the tip links are cut, as happens for the tip link protein itself, consistent with the idea that TRPA1 is a component of the transduction apparatus.

We then asked if disruption of TRPA1 expression causes a functional deficit in hair cells. We cloned the zebrafish TRPA1 and injected morpholino oligonucleotides against this TRPA1 into zebrafish embryos, to inhibit translation or splicing of the messenger RNA for the zebrafish channel. As expected, hair cells generated a much smaller electrical response in morpholino-treated fish, and a fluorescent dye that normally passes through transduction channels to label the cells could no longer enter. In mice, we used another strategy to disrupt the mRNA for TRPA1: Embryonic hair cells, which had not yet made TRPA1, were removed from mice and maintained in culture. They were infected with adenoviruses encoding small interfering RNAs (siRNAs) to cause degradation of the TRPA1 mRNA as soon as it was made, and then they were tested for function a few days later. Again, the electric current evoked by mechanical stimulation was nearly abolished in these cells, and they no longer permitted dye entry. Together, these tests indicate that TRPA1 is likely to be part of the mechanically activated ion channel of vertebrate hair cells.

TRPA1 and a relative called TRPN1, which may also be mechanosensitive, are unusual among the TRP family in that they have a long chain of ankyrin domains leading up to the pore-forming part of the channel. These domains, when several dozen are stacked in series, form a curved structure that looks like one turn of a spring. Since biophysical measurements had indicated that hair cells have some sort of springy element leading to the channel, we wondered if this part of the TRPA1 and TRPN1 proteins could form a biological spring. In collaboration with Marcos Sotomayor and Klaus Schulten (University of Illinois), we used molecular dynamics calculations to explore the elasticity of the domain. Calculations indicate that such structures are elastic, that they relax back in a few nanoseconds after being stretched, and that the elasticity is close to that measured in hair cells. Moreover, forces that are excessive for hair cells are calculated to pull apart the stack without breaking the protein, perhaps as a safety release to protect against breakage from loud noises.

Genomics of Hearing and Deafness
A hair cell probably expresses between 10 and 15 thousand genes, a significant fraction of the entire genome. Many of these are "housekeeping" genes, necessary for such basic functions as protein synthesis, energy production, and cytoplasmic transport. Hundreds or thousands of other genes play roles in the unique mechanosensory function of the hair cell. To understand hair cell function fully, we must start by identifying these genes and the timing and determinants of their expression. Conventional molecular biology methods that focus on one gene at a time are too slow for a task of this size, so my lab is mapping gene expression in hair cells with oligonucleotide arrays.

We first looked at changes in gene expression during development. Although the hair cell sensory epithelium is a complex structure with half a dozen cell types, we can identify the hair cell genes by chemically dissecting away the nerve layer to assay just hair cells and supporting cells, and then killing hair cells with aminoglycoside antibiotics. To group genes into functional pathways, we use a self-organizing-map algorithm to cluster genes with similar temporal patterns of expression. For instance, this analysis clustered a large number of genes that are consistently up- or down-regulated from ages E14 through P12.

In collaboration with Zheng-Yi Chen (Massachusetts General Hospital), we looked at the function of one such gene—that encoding the retinoblastoma (Rb) protein. In mice with a conditional deletion of the Rb gene, hair cells of the inner ear do not stop cell division at E14 as in wild-type mice, but continue to divide and produce new hair cells. Remarkably, the hair cells differentiate and show mechanotransduction while continuing to divide. Further exploration of this pathway may reveal ways to promote hair cell regeneration as a clinical treatment for hearing loss.

Some of these studies are supported by the National Institute on Deafness and Other Communication Disorders.

Last updated: March 1, 2007