The surgical effect of snipping the dendrites of the AFD neurons is essentially the same as the surgical effect of killing the AFD neurons. This result suggests that thermosensory measurement by AFD is localized to the distal end of the dendrite. It is not obvious that thermosensory measurement needs to be localized at the sensory endings. Unlike chemosensory measurements, which must be localized to pores in the worm's cuticle skin, thermal conduction allows thermosensory measurements to be carried out, in principle, anywhere inside the worm's body. Genetic evidence also suggests that AFD thermosensation is localized to the dendritic tips: mutations which disrupt the morphology of the ciliated tips of AFD also disrupt thermotactic behavior [14
]. However, it is unclear whether genetic disruption of the ciliated tips is necessarily to blame for the thermosensory phenotype, or whether such mutations have pleiotropic effects. Severing the AFD dendrite midway between the ciliary tips and the AFD cell body, as depicted in Fig. , isolates any thermosensory measurement at ciliated tips of the AFD dendrite while leaving the AFD cell body, axon, and synaptic architecture intact. This observation also argues against the possibility that the role of AFD in thermotaxis is only to transmit thermosensory signals received from its presynaptic partners; severing the AFD dendrite should only disrupt information transmission from the AFD sensory endings to the cell body, not alter synaptic transmission into or out of AFD.
In our experiments, AFD surgeries consistently reduce cryophilic bias at T
without increasing cryophilic bias at T
. If AFD were part of a thermophilic pathway, we would have expected AFD surgeries to increase cryophilic bias, especially at T
. On the other hand, our observation that AFD surgery weakens without abolishing the mechanism for generating cryophilic bias might explain the variable outcomes of atactic and cryophilic phenotypes after AFD laser killing in the experiments of Mori and Ohshima [3
]. Those experiments also reported that killing the AIZ interneuron leads to thermophilic aggregation, and argued that AFD contributes to driving this thermophilic aggregation. Our investigation of AIZ laser killing showed that although killing AIZ abolishes cryophilic bias, killing AIZ does not produce thermophilic bias. In short, this study does not support a model in which C. elegans
thermotaxis is explained as the output of cryophilic and thermophilic pathways, nor a role for AFD and AIY within a putative thermophilic pathway. This study supports the existence of a mechanism for generating cryophilic bias, and it is possible that AIZ is part of this mechanism.
Our study agrees with Mori and Ohshima [3
] in that the AFD, AIY, and AIZ neurons contribute to cryophilic movement, but disagrees in the manner of their contribution. We find no evidence that AFD and AIY oppose the mechanism for generating cryophilic bias by generating thermophilic bias. We favor a regulatory model in which AFD and AIY gate the mechanism for generating cryophilic bias. One possibility is that AFD is required to stimulate a mechanism for generating cryophilic bias at T
and AIY is required to suppress this mechanism at T
. It is also possible that at T
, AFD synaptic output to AIY prevents AIY from suppressing the mechanism for generating cryophilic bias via AIZ (a possibility schematized in Fig. ). Because surgery to AFD weakens cryophilic bias at T
even in a ttx-3
background, it appears that AFD might also able to stimulate the mechanism for generating cryophilic bias through another channel of synaptic output, either a synapse that has not yet been identified or a gap junction to AIB, the only other output of AFD identified by electron microscopy [8
]. In order to evaluate these alternatives, it will be necessary to further map the thermotactic circuit by laser ablation and isolate new thermotactic mutants.
Figure 4 A gating model for AFD and AIY in generating cryophilic bias. In this schematic, the synaptic connections between AFD and AIY and between AIY and AIZ are inhibitory, and AIZ contributes directly to generating cryophilic bias. We suggest that patterns (more ...)
Direct measurements of AFD neuronal activity also illuminate the functional role of AFD in thermotaxis. Kimura et al
] showed that intracellular calcium dynamics of the AFD neuron are evoked by positive temperature ramps only at absolute temperatures above Tcult
. The pattern of AFD calcium activity reflects the condition that T
, a condition that applies to the normal display of cryophilic bias. A physiological study of spontaneous synaptic release by the AFD neuron showed strong synaptic activity when T
or when T
, but less at T ~ Tcult
]. The study of synaptic release is more difficult to reconcile with a model in which AFD either stimulates the mechanism for generating cryophilic movement when T
or suppresses this mechanism when T
, but not both. One possibility is that a more complicated model is required. Another possibility is that the synaptic release observed when T
might indicate the artifact of gene-expression in AFD, also encountered in this study, which leads to slight cryophilic bias at T
Recently, several studies have investigated the roles of neurons in navigational behavior, by systematically disrupting neurons and quantifying the effects on forward movement and reorientation maneuvers [19
]. In particular, Tsalik and Hobert showed that disruptions of AFD and AIY lower and raise the rate of spontaneous reorientation maneuvers, respectively [18
]. One possibility is that AIY acts to inhibit reorientation maneuvers, and that AFD acts to inhibit AIY. Tsalik and Hobert's model of AFD and AIY with respect to reorientation maneuvers bears comparison to our gating model of AFD and AIY with respect to generating cryophilic bias that is schematized in Fig. . It is likely that neural circuits operate in similar ways even as they contribute to different aspects of behavior.
It is important to stress the difference between conventional assays of C. elegans
thermotactic behavior that quantify thermophilic and cryophilic aggregation and our microdroplet assay that quantifies the worm's ability to alter movement patterns in response to positive and negative temperature gradients [1
]. We interpret the ability of the worm to bias its movements towards lower temperatures, as measured in the microdroplet assay, underlies the worm's ability to crawl down thermal gradients and exhibit cryophilic aggregation in conventional assays. However, it is a current controversy whether C. elegans
has a capacity to actively crawl up thermal gradients and exhibit thermophilic aggregation [2
The study of behavior in C. elegans often begins when genes or neurons are associated with behavioral endpoints like cryophilic aggregation. But in order to understand the functional organization of genes and neurons, the mechanisms that comprise complex behavior must be identified. This effort requires behavioral assays designed to disentangle and analyze each mechanism of complex behavior, and physiological assays to manipulate and monitor neuronal structure and function. In this study, we used femtosecond laser ablation in combination with a quantitative assay of cryophilic bias to clarify the role of the AFD sensory neuron in thermotaxis. Our observations suggest a model in which the AFD neuron regulates cryophilic performance, not by generating thermophilic bias, but by gating a separate mechanism for generating cryophilic bias.