Paper #16 - Dissecting a central flip-flop circuit that integrates contradictory sensory cues in C. elegans feeding regulation

TitleDissecting a central flip-flop circuit that integrates contradictory sensory cues in C. elegans feeding regulation

Year: 2012


In this paper, they sought to understand how C. elegans controls its pharyngeal pumping ('feeding') behavior in response to attractive and repulsive chemical stimuli. They use diacetyl as an attractive stimulus, and quinine or high concentrations of isoamyl alcohol as repulsive stimuli.

To present isoamyl alcohol or diacetyl stimuli to the worm, they put a 2uL drop of the volatile organic on the top of the lid, sealed the plate, and then started watching worms 5 minutes later. For the quinine, they instead dissolve it in M9 buffer and then spread it over the plate.

First, they that in general diacetyl increases pumping rates, and isoamyl alcohol (at >10% in the volatile drop they put on the lid) and quinine reduce pumping rates. They show dose-response curves, but the x-axis is either 'concentration in the drop' or 'concentration of the quinine solution applied to the plate', so it's not a very replicatable or quantitative response. However, it's quite clear that they do modulate their pumping rates in response to these chemicals.

They next show that sensing these chemicals is essential to modulating pharyngeal pumping (hopefully not surprisingly). Knocking out odr-7  (a gene necessary for AWA development [AWA is 'the' diacetyl sensing neuron) or the diacetyl receptor odr-10 itself eliminate the pharyngeal accelaration due to diacetyl. Similarly, knocking out osm-9 - a TRPV1 homolog that has been implicated in detecting noxious chemicals in the neuron ASH - eliminates the quinine-induced pumping deceleration. They also note that eat-4 (a glutamate transmission mutant) also does not display quinine-induced pumping deceleration.

The authors knew in advance that exogenous serotonin could stimulate feeding behavior, but they confirmed it anyway. In contrast to previous work, they found that adding exogeneous serotonin while the worms were on food could still upregulate pumping (the effect is apparently much stronger when worms are off food).  If pumping rates somehow saturate though, perhaps this isn't surprising (as an analogy, small changes in the input do nothing to the output of an amplifier if the amplifier is already saturated!).

In support of serotonin modulating pumping acceleration, the tph-1 mutant (discussed here recently) does not display diacetyl-induced pumping acceleration, and further, expressing tph-1 specifically in NSM only rescues diacetyl-induced pumping acceleration.

(This paper is thus a nice backstory to Steve Flavell's work that came out a year and a half later.)

The authors also identified glr-7 as a gene that disrupts diacetyl-induced pumping acceleration. glr-7 is a non-NMDA glutamate receptor that is expressed in NSM and a few other pharyngeal neurons (I1, I2, I3, I6, MI). They thus hypothesize that glr-7 is the receptor by which NSM receives the cue to secrete serotonin.

What receives the serotonin? They made mutants of each serotonin receptor, and found that only MOD-1 was necessary for the pumping response to diacetyl. Using cell-specific rescue, they found that MOD-1 was necessary only in RIM and RIC (the tyraminergic and octopaminergic neurons discussed here a few days back).

Tyramine and octopamine were thought to oppose the behavior of serotonin, so they wondered if serotonin inhibited  tyramine/octopamine signaling in RIM/RIC. To test this, they asked if a tdc-1 mutation could rescue a mod-1 mutant's lack of pumping acceleration. A tdc-1; mod-1 double mutant does display diacetyl-induced pumping acceleration. I think this suggests that serotonin acts directly on other targets as well to induce pumping acceleration, but they don't really address what the other targets are.

While tdc-1 mutants upregulate pumping just fine, they do not downregulate in the presence of quinine! Similarly, they found ser-2 (a tyramine receptor) mutants do not downregulate  pumping either. ser-2 is expressed in many neurons including AIY, AIZ, RIA, as well as NSM and pharyngeal muscle cells.  Cell-specific ser-2 expression in ser-2 mutants could rescue quinine-induced pumping deceleration when ser-2 was expressed in NSM.

At this point, the data is suggesting reciprocal inhibition between NSM and RIM/RIC.

They go on to look at calcium imaging of NSM and RIM in moving animals, in the presence of diacetyl and of quinine. They find that generally diacetyl drives NSM activity and induces RIM, and vice versa for quinine.

They go on to argue that the mutual inhibition between these two neurons makes their behavior somewhat bistable. They attempt to show this by comparing a diacetyl dose-response curve in the presence of some quinine for the wild-type vs a mod-1; ser-2 double mutant (which should eliminate mututal inhibition). They also do an analogous experiment, with some background quinine and then varying the diacetyl concentration. In both cases, the response transition point between downregulation and upregulation of pumping is much sharper for the wild-type than for the mutant without reciprocal inhibition.

Questions I still have:

  • In the absence of tyramine and octopamine and serotonin, animals still pump. How do they do this? What drives 'basal' pumping?
  • Not particularly related to the paper, but in one of the figures they plot dF/F - change in fluorescence normalized to 'baseline' fluorescence.  This normalization seems reasonable and appropriate (and enables averaging many trials!) in the case of sparsely spiking neurons, but somehow doesn't seem super appropriate for neurons which may contain analog-like information. What if you're sometimes normalizing to a high state and sometimes normalizing to a low state?
  • I think their results are probably still all correct, but this paper commits this same basic statistical sin discussed here a few days back: to see if the response to X in condition A is different from the response to X in condition Y, you cannot compare the p-value for a non-zero response in condition A to the p-value for a non-zero response in condition B!