|Perk of attending chemosensory conferences: the candy.|
In the second section, he presented recordings from the taste ganglia. They stuck a 2p-endoscope into the taste ganglia to see what combination of tastes the neurons responded to. Assuming only five taste modalities (sweet, sour, salty, bitter, and umami, and ignoring carbonation or fat), there are 31 possible combinations of taste receptive fields, including pairs, triplets, quadruplets, and the full taste house. Using a "magic" dye, they found only seven of these combinations, the five basic tastes, and two pairs (the slide went by before I could note both). Zuke breezed through this quickly, as the researcher will present later in the conference.
Zuke seems quite enthusiastic to pursue taste into "higher" brain regions, which he showed in the last two sections. Having described taste hotspots in gustatory cortex, he wanted to show they were necessary and sufficient. To show they are necessary, the lab trained mice in a go/no-go task where the go cue was a sweet taste, and the no-go was a bitter taste, and the task was to lick (or not) following the cue . Thirsty mice were rewarded with water. Then the lab bilaterally cannulated the mice in either the sweet or bitter hotspot, through which they injected a "neuronal silencer." When they injected in the sweet area, the mice failed to stop licking during the no-go trials, showing they may not have sensed sweet. Conversely, when they silenced the bitter area, the mice did not lick for the go cue. Thus they showed that these areas are required for sensing their respective modalities.
To show the areas are sufficient for tasting these modalities, they employed Channelrhodopsin. First, the bitter area. They infected the bitter hotspot with AAV-ChR2-YFP. They then water-deprived mice, and trained them to lick a water "sipper." The mice would lick freely, until they turned on the light in the bitter hotspot, which caused the mouse to wince and stop licking. This could be turned on and off for a few seconds at a time, showing the effect was nearly instantaneous. And if they turned the "power" of the light up (unclear whether intensity or frequency), the mice would wince and gape from the awful sensation.
For the sweet hotpost, they trained mice to occasionally lick a water spout. Since they didn't want the mice licking for liquid reward, they used slaked mice. For some of the licks, the licks would trigger the light in the sweet hotspot; for other licks, it would trigger nothing. They then measured the number of licks when the light was functional versus not, and found the mice licked more when the licking triggered the light.
In the last section of the talk, Zuker focused on the downstream areas from gustatory cortex. They have injected AAV-GFP and -tdTomato in both the sweet and bitter hotspot, and identified four downstream areas: amygdala, hypothalamus, entorhinal cortex, and the nucleus accumbens. The labeling in the amygdala did not label the entire amygdala, but one sub-field.
I'm not an expert in amygdala function, but it is involved in valence. For unclear motivation, they infected the sweet hotspot with AAV-ChR2, and stuck a light fibre in the amygdala. They then stimulated the sweet axon terminals in the amygdala while the mice were tasting a bitter compound. When the light was on, the mice licked frequently; however, with the light off, the mice did not lick. The effect of light could also be blocked by NBQX.
For future directions, Zuker basically highlighted most interesting questions in central taste. How are mixtures encoded? How does internal state (i.e. satiety) alter circuits? How are olfaction and taste integrated as flavor? And they now have chronic 2p endoscopes working in behaving mice.
All in all, I nice overview of many exciting experiments. One large outstanding question remains from before: given that all e-phys recordings have shown that neurons are broadly tuned, why hasn't Zuker's lab been able to replicate that (Zuker notably presented almost all imaging, and one e-phys slide regarding a ChR2 control). More generally, it will be fun to look at these results in a complete paper. Right now it's not clear to me that the cannula and light fibre experiments are specific enough to make conclusions about bitter versus sweet taste modalities, or whether it is simply aversive versus pleasant. And the amygdala experiment seems more like a fun pilot experiment rather than a complete story. In any case, it's nice to see a big lab presenting unpublished data to a wide audience.
Taste coding in mammals
Aside from the main talks, ISOT runs three parallel sessions at a time. Since I want to work in taste, I will focus on the taste sessions, including this afternoon's session on taste coding from the periphery to central areas.
First up was G. Hellekant, who presented some simple experiments recording from taste nerve fibres. He started by showing that mammalian taste receptors can vary 50-70% in homology between species. Specifically, ruminant mammals are quite different from primates. Then he showed taste nerve recording from chimpanzees and marmosets, that showed the responded to a wide variety of taste modalities, from sweet to bitter. In contrast, cows and pigs responded to a smaller subset, notably lacking many sweet responses.
In the second half, he mentioned two small facts. First, there is a chemical, gymnemic acid, which blocks sweet receptors in humans (lactisole also does the same). Second, he showed that a calcium channel, Calhm, is expressed in T1R2 cells, and that Calhm KO mice do not have a sweet response.
The second speaker was Susan Travers, who focused on taste responses in the NST and PBN of anesthetized rats. Previously, she had reported that there are four clusters of taste responses - Sweet, NaCl, Acid, and Bitter - based on their average firing rate during 20s of taste presentation. However, recognizing that taste responses are dynamic, she focused on the temporal aspect of the responses. She showed that during the first second of the response, many cells were broadly tuned, responding to multiple tastes. Notably, sweet cells responded quickly, while bitter cells took some time to ramp their response.
To look more closely at the temporal information, she performed a metric space analysis, which I did not quite follow. From this, she concluded that including temporal information increases total information, which seemed obvious from simply looking at how dynamic the responses were. Also of note is how slow these responses are developing over seconds, when a lick is 150ms. In any case, it appeared that a subset of PBN neurons were still broadly tuned.
Third up was Sid Simon, whose main point was that taste neurons can have highly diverse responses, both narrowly and broadly tuned. He showed recordings from rats with electrodes implanted four places: OFC, insular cortex (IC), nucleus accumbens (Nacc), and amygdala. These rats were then trained in a go/no-go task. Each of the recorded places had special kinetics. IC neurons responded to licking; OFC neurons anticipated the cue; amygdala neurons responded to the reward. For IC neurons, some were inhibited by feeding, while other were excited or unchanged; and these same neurons would become inhibited, excited or unchanged during sleep.
In the last section, he showed data from a rat trained to discriminate between different concentrations of NaCl. 60% of IC cells didn't respond to the taste of NaCl. Of those that did respond, the responses were dynamic. Some responded during the first lick before fading, while others increased their response over time. Some just responded to water, but not NaCl; and of those, some could differentiate between reward water, and cue water. All in all, if you can imagine a taste receptive field, some neuron probably has it.
The fourth speaker, Claire Murphy, presented fMRI data on satiety. I'm not a big fan of fMRI, but her main point was that hunger increases the magnitude of taste responses in IC, OFC, amygdala, caudate, etc.; and that some areas do not respond in sated humans, but only when they're hungry.
That's it for day one.