In truth it was established quite early on that these synapses are rather odd. To begin with, inner hair cells are not technically neurons, they are epithelial cells that transduce mechanical energy in the cochlea to electrochemical energy in the nervous system. They also don’t fire action potentials, but are electrotonically depolarized by mechanical stimulation of the organ of Corti in the cochlea as sound waves enter the inner ear. The mechanical stimulation causes shearing forces on the hairs atop the IHCs, resulting in mechanical gating of cation channels. The ensuing depolarization opens voltage gated calcium channels, resulting in synaptic vesicule fusion. To be clear, this is synaptic transmission by something technically defined as an epithelial cell. Furthermore, these cells have ribbon synapses, meaning they have a long electron dense synapse, that is very crowded with vesicles, presumably to increase auditory sensitivity. Finally, they are strange on the molecular level as well, since on the one hand they lack synaptotagmins and complexins that are generally important for normal synaptic release while on the other, to function properly, they require molecules that aren’t very common (otoferlin and RIBEYE).
Now though, Nouvian and friends have added another peculiarity to the list. As I mentioned in my description of the meat and potatoes of the paper, when the group applied individual botulinum toxins (BoNTs) to specifically cleave each type of SNARE (synaptobrevins, SNAP-25 and syntaxin) vesicular fusion continued as usual. It is important to note that Nouvian and friends couldn’t study vesicular fusion by measuring postsynaptic responses; to do their recordings they had to remove the IHCs from the cochlea but unfortunately, the only postsynaptic targets of the IHCs have their cell bodies in the superior olivary complex in the brain stem, and were thus left behind when cochlea was removed. Instead, Nouvian and friends measured exocytic membrane capacitance increases that occur when a vesicle fuses and adds its membrane to the presynaptic terminal. To do this they did whole cell patch clamp recordings, which allowed them to simultaneously depolarize the cell enough to cause vesicular fusion and measure current changes. They could also apply the toxins directly into the cell through the patch pipette. Thus they were able to assess whether vesicles were fusing with the membrane without recording postsynaptic responses, although they would presumably occur.
Likely realizing that claiming synaptic transmission without the canonical vesicular fusion machinery is not an easy sell (although not terribly outlandish either) Nouvian and friends tackled this question very thoroughly and from numerous angles.
The Many Angles
To start, they verified that their toxins worked with two different methods. First they applied each toxin to their complementary SNAREs (BoNT/D cleaves synaptobrevin; BoNT/E cleaves SNAP25; and BoNT/C cleaves syntaxin) and verified by western blot that the toxins actually cleaved them. Second, the group examined the effect of their toxins on vesicular fusion in chromaffin cells. Chromaffin cells are neurotransmitter-releasing cells in the autonomic system. They are predominantly found in the adrenal cortex where they release epinenphrin and norepinephrine from dense core vesicles into the blood stream, instead of across a synapse. I believe the reason Nouvian and friends used these cells to examine vesicular fusion is because the dense core vesicle that they use are quite large, and thus it is easier to measure the incremental capacitance increase caused by addition of that large chunk of vesicular membrane to the release site. When Nouvian and friends “poisoned” these cells by applying the BoNTs through a patch pipette, they stopped seeing any change in membrane capacitance following applied current, indicating that vesicular fusion had been shut down by the toxins, as expected. However, as I said before, when they applied the same test to the IHCs, vesicular fusion carried on as usual, with the expected depolarization-induced incremental increases in membrane capacitance, indicating ongoing vesicular fusion.
Since the tetanus toxin insensitivity in IHCs was perhaps unexpected, Nouvian and friends, whether of their own accord or after prompting from referees, undertook a lot of control experiments and alternative tests to verify the validity of their initial finding. To verify that a homeostatic increase in calcium influx did not account for the maintenance of fusion, they verified that capacitance increase with respect calcium influx was the same in poisoned vs control and found that and found that this was the case. They then verified that the the toxins were actually making it into the cells by loading the cells with a fluorescently labelled BoNT/E (conjugated with Alexa 488) and imaging. They found that loading was reliable, and that even though this conjugated toxin still cleaved SNAP-25, it had no effect on capacitance increases. Then, taking things up a notch, Nouvian and friends opted to allow the BoNT/E more time to complete cleavage, and then use stronger current injection and photolytic calcium uncaging to induce vesicular fusion. Once again – the toxin had not effect. They repeated barrage of tests with the other two toxins as well, so up to this points they have their bases covered.
Thankfully for Nouvian and friends there are knock out mice for synaptobrevin-1, -2/3 and SNAP-25 that allowed them to assess the total loss of function of these SNAREs. Synaptobrevin-1 KO mice are the only ones that survive after birth, and they showed robust excocytic incremental capacitance increases in their IHCs. On the other hand, synaptobrevin-2/3 and SNAP-25 KO’s do not survive past birth. So to get around this, Nouvian and friends made organotypic cultures of the organ of Corti from embryonic mice. Once again, in these organotypics, capacitance increases occurred in response to depolarization in the IHCs, indicating in tact vesicular fusion.
Convinced yet? Too bad. At this point Nouvian and friends actually went looking for the physical presence of SNAREs and their transcripts. They did find transcripts for each of the SNAREs with real time PCR, although they amplified at lower levels than genes known to be expressed in IHCs (otoferlin and parvalbumin). Then they knocked pHluorin-tagged synaptobrevin-1 (aka Synapto-pHluorin) into the endogenous synaptobrevin-1 locus. pHluorin is a pH sensitive variant of GFP that fluoresces when exposed to neutral pH. When on a vesicle, the pH sensing domain of synapto-pHluorin is oriented inside the vesicle, where pH is low. Then, when the vesicle fuses to the membrane, the pHluorin domain orients into the neutral extracellular space. Thus, synapto-pHluorin is a reporter for active vesicle fusion. In these mice, neither depolarization of hair cells nor application of an acidic solution revealed any fluorescence. They couldn’t detect GFP in the knock-in haircells by immunohistochemistry either, but they could see it in neuronal terminals that provide feedback onto the hair cells from the superior olivary complex, providing a good postive control.
Finally they went looking for the actual proteins immunohistochemically. Using a total of 13 different antibodies, Nouvian and friends failed to identify any of the 4 neuronal SNAREs that may have been in hair cells, effectively poopooing a previous claim that SNAREs are present. Meanwhile, the terminals of that feedback pathway I mentioned above provided positive controls for all these antibodies.
In their concluding remarks, Nouvian and friends offer the unlikely explanation that perhaps all those 13 epitopes and the botulinum toxin cleavage sites are uniquely shielded in IHCs by interacting proteins, and that perhaps secondary SNAREs come to the rescue and compensate for the knocked-out SNAREs in their transgenic mice. But I think they have presented some pretty convincing work. They end the article by saying, “The most likley explanation of our data is that IHCs, being epithelial cells, make use of other SNARE proteins for synaptic vesicle exocytosis than neurons, which remain to be discovered.” In other words, we should have been expecting this outcome for years. I felt like I’d been duped when I read that, but after pondering it a bit I realized that it isn’t really so cut and dry. Yes, hair cells are technically classified as epithelials, but they are epithelial cells that make chemical synapses, making them pretty unique. So don’t feel too bad if you didn’t know the outcome of this paper without even reading it.
While it is a bit exhausting to be pummelled with so many controls and back up experiments, and while I feel for the experimenters who pushed through what were probably a lot of suggestions from reviewers, its nice to see such a lucid, straightforward paper that covers all its bases. As always, the Brief Communication format didn’t do justice to the amount of work that went into their pretty strong conclusion, but I will assume they got it into the journal the wanted. That being said, it would have been nice to see one more thing: a behavioural test for hearing in the synaptobrevin-1 KO mouse. This probably would have been quite easy and would have been the one piece of evidence proving that there is indeed transmission occurring, not just something that looks vesicular fusion.
Finally, I’d like to point out that this study may have a translational role to play. While a lot is known about the genes involved in congenital hearing loss, it looks like a good number syndromes have yet to be sorted out genetically. Perhaps comparing uncharacterized mutant loci associated with congenital hearing loss will help sort out what genes are implicated in IHC vesicular fusion, improving our understanding of congenital hearing loss and perhaps informing future gene replacement strategies.