Temperature Sensing by an Olfactory Neuron in a Circuit Controlling Behavior of C. elegans

See allHide authors and affiliations

Science  09 May 2008:
Vol. 320, Issue 5877, pp. 803-807
DOI: 10.1126/science.1148922


Temperature is an unavoidable environmental cue that affects the metabolism and behavior of any creature on Earth, yet how animals perceive temperature is poorly understood. The nematode Caenorhabditis elegans “memorizes” temperatures, and this stored information modifies its subsequent migration along a temperature gradient. We show that the olfactory neuron designated AWC senses temperature. Calcium imaging revealed that AWC responds to temperature changes and that response thresholds differ depending on the temperature to which the animal was previously exposed. In the mutant with impaired heterotrimeric guanine nucleotide–binding protein (G protein)–mediated signaling, AWC was hyperresponsive to temperature, whereas the AIY interneuron (which is postsynaptic to AWC) was hyporesponsive to temperature. Thus, temperature sensation exhibits a robust influence on a neural circuit controlling a memory-regulated behavior.

If wild-type C. elegans individuals are cultivated at a certain temperature, ranging from 15° to 25°C, for 3 hours with bacteria as food source and are then placed on a temperature gradient from 15° to 25°C, most of the animals migrate to the previous cultivation temperature (Fig. 1) (1). This behavior is called thermotaxis, and its plasticity provides an opportunity to understand molecular and neural circuit mechanisms of thermosensation, learning, and memory (13). By ablation of particular cells with a laser microbeam and evaluation of the consequent behavioral effects, a simple neural circuit essential for thermotaxis has been identified (Fig. 2A) (3).

Fig. 1.

Thermotaxis behavior of C. elegans. (A) Procedure for the population thermotaxis assay (21). A linear temperature gradient was established along the 14-cm agar surface in the thermotaxis (TTX) plate (10 cm by 14 cm). The steepness of temperature gradient was kept at 0.44°C/cm. Animals cultivated at 17°, 20°, or 23°C were placed at the surface of agar at 20°C and left for 60 min. The TTX plate was divided into eight regions with scores from –4 to +4, and the TTX index was calculated as shown. (B) Distributions (left) and TTX indices (right) of the animals cultivated at 17°, 20°, or 23°C. Error bars indicate SEM. n = 545 to 1063 animals for each genotype. Single and double asterisks indicate P < 0.05 and P < 0.01, respectively, in an analysis of variance (ANOVA) for multiple comparisons (see SOM). (C) Single-animal thermotaxis assay with a radial temperature gradient (22). The assay plate (a 9-cm-diameter petri dish) was divided into three areas: 0 to 1.5 cm (at about 17°C), 1.5 to 3 cm (at about 20°C), and 3 to 4.5 cm (at about 25°C)—indicated as blue, yellow, and orange areas, respectively—as measured from the center of the plate (left panel). The steepness of temperature gradient was not linear. A single animal was placed on the “x” indicated on the assay plate. Animals' tracks were categorized into four groups. Animals that moved only within the blue, yellow, or orange areas were classified as “17,” “20,” or “25,” respectively. Animals that moved randomly, or within both 17° and 25°C regions, were classified as “17/25” (further classified in detail as shown). (D) The tracks of wild-type or eat-16 mutant animals in which the indicated neuron was ablated. All the animals were cultivated at 20°C before the radial gradient assay. (E) Cell-specific rescue experiments for defective thermotaxis of eat-16 mutants on a radial temperature gradient. Each animal was cultivated at 20°C. n = 60 to 514 animals for each genotype. Error bars indicate SEM. Double asterisk indicates P < 0.01 in ANOVA for multiple comparisons.

Fig. 2.

A neural circuit model controlling thermotaxis and olfactory signaling. (A) The simplified original neural circuit model for thermotaxis (3). Temperature is sensed by AFD and an unidentified “X” sensory neuron. The activated AIY-RIA circuit accelerates the movement to a temperature higher than the cultivation temperature (thermophilic drive or “T”), the activated AIZ-RIA circuit accelerates the movement to a temperature lower than the cultivation temperature (cryophilic drive or “C”), and the RIA interneuron integrates thermal signaling of “T” and “C.” (B) A model for G protein/cGMP–mediated olfactory signaling in AWC.

Although wild-type animals migrate up or down the temperature gradient until reaching their cultivation temperature, nj8 mutants that were isolated in a genetic screen for thermotaxis-defective mutants migrated toward colder temperatures than those to which they were previously exposed (Fig. 1, B, D, and E). The nj8 mutation corresponded to a nearly loss-of-function mutation in the eat-16 gene encoding a homolog of the mammalian regulator of G protein signaling (RGS) proteins, which are negative regulators for the α subunit of the G proteins (4).

A fusion gene encoding wild-type EAT-16 fused to green fluorescent protein (eat-16::gfp), which is under the control of the eat-16 promoter, was broadly expressed in neurons (fig. S2A). To identify neurons in which EAT-16 is required for thermotaxis, we expressed eat-16 cDNA in eat-16(nj8) mutants with neural promoters to drive EAT-16 expression in various sets of neurons (Fig. 1E and fig. S2B). Expressing EAT-16 in nearly all or in many neurons restored the normal migration of eat-16 mutants toward their cultivation temperature (Fig. 1E). Although expression of EAT-16 in several subsets of neurons, including neurons essential for the neural circuit controlling thermotaxis (Fig. 2A), did not restore the normal migration of eat-16 mutants, expression of EAT-16 from odr-3, gpa-13, and odr-1 promoters (57) did (Fig. 1E). These three promoters allow the expression in a single type of sensory neuron: AWC, which is characterized as an olfactory neuron (Fig. 1E and fig. S2B) (8). These results suggest that eat-16 acts cell-autonomously in AWC sensory neurons and that EAT-16 RGS is required in AWC for thermotaxis.

Using Ca2+ imaging, we tested whether AWC responds to temperature. We monitored temperature-evoked changes in the intracellular concentration of free Ca2+ in intact AWC neurons with a genetically encoded Ca2+ indicator: cameleon (Fig. 3, A to D) (9). The Ca2+ concentration in the AWC neuron of wild-type animals increased upon warming and decreased upon cooling (Fig. 3, A and B, and fig. S3). We also tested whether two other sensory neurons respond to temperature. Although the responses of the ASE neuron were minimal, the small responses of the ASH neuron were detected (10, 11) (Fig. 3B). This weak responsiveness may implicate the possibility that temperature information is conveyed by multiple sensory neurons. AWC neurons in eat-16 mutants were hyperresponsive to temperature changes (Fig. 3C). Thus, the EAT-16 RGS protein appears to negatively regulate temperature detection in AWC neurons of wild-type animals. AWC neurons responded to increments of warming above a threshold temperature that is set by the cultivation temperature (Fig. 3D). Consequently, thermal responses of AWC are modulated by memory of the past cultivation temperature, similar to those of the major thermosensory neuron AFD (12, 13).

Fig. 3.

Ca2+ imaging of the neurons in the thermotaxis circuit in wild-type animals and mutants defective in G protein–coupled signaling of AWC. In vivo calcium ratio imaging of AWC (A to D), AIY (E), and AIZ (F) in the individual genotypes. Relative increases or decreases in the intracellular Ca2+ concentration were measured as increases or decreases in the yellow fluorescent protein/cyan fluorescent protein (YFP/CFP) fluorescence of the cameleon protein ratio (“ratio change”). Temperature change per unit time is shown at the bottom of each graph. (B to F) Each graph represents the average response to temperature stimuli. n = 10 to 17 animals. (A) Representative FRET signal in AWC of wild-type animals cultivated at 20°C and subjected to temperature change. (B) Ca2+ imaging of AWC, ASE, and ASH sensory neurons in wild-type animals subjected to warming and cooling. The average of maximum ratio change from baseline ± SEM was 16.95 ± 2.17 for AWC, 3.18 ± 2.17 for ASE, and 5.35 ± 2.12 for ASH. n = 14 to 17 animals. (C) Changes in Ca2+ concentration. The average of maximum ratio change from baseline ± SEM was 17.529 ± 2.07 for wild type, 25.81 ± 2.14 for eat-16, 6.72 ± 1.33 for odr-3, 4.01 ± 2.19 for tax-4, and 27.74 ± 7.42 for wild type expressing excess ODR-3 in AWC. n = 13 to 17 animals. (D) Response of AWC neurons to steplike temperature changes above a threshold temperature that nearly corresponds to the cultivation temperature. The average of maximum ratio change from baseline ± SEM was 26.61 ± 2.45 for a 15°-cultivated animal, 17.78 ± 2.69 for a 20°-cultivated animal, and 3.70 ± 2.34 for a 25°-cultivated animal. n = 10 to 12 animals. (E and F) Ca2+ imaging in AIY and AIZ neurons. The average of maximum ratio change from baseline ± SEM was 10.87 ± 1.10 for wild type (AIY), 2.67 ± 0.83 for eat-16 (AIY), 10.41 ± 1.52 for wild type (AIZ), and 9.35 ± 1.13 for eat-16 (AIZ). n = 15 to 16 animals.

In the neural circuit model for thermotaxis, the presence of a second thermosensory neuron with a less critical role than that of the AFD neuron has been proposed (Fig. 2A) (3). To test whether this unidentified thermosensory neuron might be the AWC neuron, we conducted laser ablation experiments (3). Although ablating AFD neurons of wild-type animals resulted in aberrant migration (Figs. 1D and 4A) (3), ablating AWC neurons of wild-type animals did not cause detectable defects in a radial gradient assay (Figs. 1D and 4A). Ablating both AFD and AWC neurons in wild-type animals resulted in no temperature preference (Figs. 1D and 4A): a phenotype similar to that caused by mutations in the tax-4 gene encoding a guanosine 3′, 5′-monophosphate (cGMP)–gated channel that is expressed in both AFD and AWC neurons (Fig. 4B) (14). Moreover, ablation of AWC neurons in eat-16 mutants restored normal migration to the mutant animals (Figs. 1D and 4A). Thus, our ablation results are consistent with the notion that AWC is a secondary thermosensing neuron.

Fig. 4.

Thermotaxis of animals with ablated neurons or cGMP signaling mutants. (A) Summary for thermotaxis of animals with ablated neurons. All the animals were cultivated at 20°C before the radial gradient assay. Error bars indicate a twofold increase in SD (see SOM). n = 6 to 78 assays. (B to E) Thermotaxis of animals cultivated at 20°C on a radial temperature gradient. n = 60 to 236 assays (B), n = 48 to 183 assays (C), n = 99 to 495 assays (D), and n = 44 to 119 assays (E) for each genotype. Error bars indicate SEM. Double asterisk indicates P < 0.01 in ANOVA for multiple comparisons.

G protein signaling (which increases the intracellular concentration of cGMP) is required for olfactory signaling in AWC neurons (Fig. 2B): ODR-1 guanylyl cyclase (GCY) and TAX-4 cGMP-gated channels are localized at the AWC sensory endings where odorants are initially received, and mutations in genes encoding either of these proteins lead to defective behavioral responses to AWC-sensed odorants (6, 7, 14). odr-1 mutants exhibited almost normal thermotaxis after cultivation at 20°C (Fig. 4, B and C, and fig. S4, B and C), and tax-4 mutants were not responsive to temperature (Fig. 4, B and C) (14). GCY-23, GCY-8, and GCY-18 all localize at sensory endings of AFD neurons and are required for thermosensory signaling in AFD (15). The gcy-23 gcy-8 gcy-18 triple mutant showed abnormal migration, quite similar to that of wild-type animals in which the AFD neuron was ablated (Fig. 4, A and B) (15). Mutation in the odr-1 gene enhanced the severity of the thermotaxis defect of gcy-23 gcy-8 gcy-18 triple mutants, because gcy-23 gcy-8 gcy-18; odr-1 quadruple mutants were unresponsive to temperature, similar to that observed in wild-type animals in which the AFD and AWC neurons had been ablated or in tax-4 mutants (Fig. 4, A and B). odr-1 mutation suppressed abnormal migration of eat-16 mutants, and tax-4 is epistatic to eat-16 (mutation in tax-4 masks the effects of mutation in eat-16) (Fig. 4C and fig. S4B). Thermal responses of AWC in tax-4 mutants revealed by Ca2+ imaging were diminished (Fig. 3C and fig. S3, B and C). Thus, mutations in odr-1 and tax-4 likely interrupt AWC thermosensory function, further suggesting that temperature signaling may occur through a cGMP signaling similar to that mediating odor sensing (Fig. 2B).

We explored Gα proteins mediating AWC thermosensation by examining Gα genes—gpa-7, odr-3, gpa-2, and gpa-13—that are expressed in AWC neurons (5, 7, 16). After cultivation at 20°C, gpa-7 and odr-3 mutants showed nearly normal thermotaxis, whereas gpa-2 and gpa-13 mutants exhibited a partially abnormal phenotype in radial temperature gradients (Fig. 4D and fig. S4A). Only odr-3 mutation suppressed abnormal migration of eat-16 mutants (Fig. 4D and fig. S4, B and C). Expression of a dominant-negative form of ODR-3 also suppressed the defect of eat-16 animals (Fig. 4D). Small responses of AWC neurons in odr-3 mutants were detected by Ca2+ imaging (Fig. 3C and fig. S3C). Excess expression of wild-type ODR-3 in AWC neurons of wild-type animals induced abnormal migration, which was reminiscent of that of eat-16 mutants (Fig. 4D and fig. S4, A and B), whereas excess expression of GPA-13 and another Gα protein (EGL-30) in AWC neurons of wild-type animals did not (Fig. 4D and fig. S4B). We used Ca2+ imaging to reveal that AWC neurons expressing excess wild-type ODR-3 in wild-type animals were hyperresponsive to temperature changes, as were the AWC neurons in eat-16 mutants (Fig. 3C). Thus, G protein–cGMP signaling appears to mediate thermosensation in AWC neurons. Notably, in contrast to eat-16 mutants, odr-3 mutants previously cultivated at 17°, 20°, or 23°C migrated toward a higher temperature than their cultivation temperature in a linear temperature gradient (Fig. 1B), which implicates RGS-mediated Gα thermosensory signaling of AWC neurons.

We evaluated AWC temperature signaling in a neural circuit for thermotaxis by Ca2+ imaging. The AIY neuron receives synaptic inputs from both AFD and AWC neurons, AIZ receives synaptic inputs from AIY, and RIA receives synaptic inputs from both AIY and AIZ (17). The AIY neuron is thought to mediate movement toward warmer temperatures, AIZ is thought to mediate movement toward colder temperatures, and the regulation between AIY and AIZ activities through RIA is thought to be important for proper migration to a memorized temperature on a temperature gradient (Fig. 2A) (3). Upon warming, fluorescence resonance energy transfer (FRET) ratios in AIY neurons changed less in eat-16 mutants than in wild-type animals (Fig. 3E), whereas thermal responses of AIZ were nearly normal (Fig. 3F). Given that wild-type animals in which AIY is ablated and AIY-defective mutants migrate toward temperatures colder than the cultivation temperature (Fig. 4E) (3, 18), the decrease of AIY activity in eat-16 mutants revealed by Ca2+ imaging (Fig. 3E) may cause abnormal migration to colder temperatures (fig. S5). Thus, regulation of AWC-AIY connectivity may be critical for the circuit controlling thermotaxis. The importance of AWC-AIY wiring was also reported recently for olfactory behavioral responses (19). A thermotactic defect in myo-inositol monophosphatase–defective ttx-7 mutants is caused by aberrant synaptic localization on the RIA neuron (Figs. 2A and 4E) (20). Abnormal migration of eat-16 mutants was suppressed by ttx-7 mutation (Fig. 4E), which is consistent with the model that the transmission of neural signals from AWC and AIY neurons to RIA is also indispensable for thermotaxis (fig. S5). Our current work helps explain a complex mechanism for control of behavior from stimulus sensing through the neuronal circuit that produces behavioral output.

Supporting Online Material

Materials and Methods

Figs. S1 to S7


Movie S1

References and Notes

View Abstract

Navigate This Article