A Gate-and-Switch Model for Head Orientation Behaviors in Caenorhabditis elegans

Abstract

The nervous system seamlessly integrates perception and action. This ability is essential for stable representation of and appropriate responses to the external environment. How the sensorimotor integration underlying this ability occurs at the level of individual neurons is of keen interest. In Caenorhabditis elegans, RIA interneurons receive input from sensory pathways and have reciprocal connections with head motor neurons. RIA simultaneously encodes both head orientation and sensory stimuli, which may allow it to integrate these two signals to detect the spatial distribution of stimuli across head sweeps and generate directional head responses. Here, we show that blocking synaptic release from RIA disrupts head orientation behaviors in response to unilaterally presented stimuli. We found that sensory encoding in RIA is gated according to head orientation. This dependence on head orientation is independent of motor encoding in RIA, suggesting a second, posture-dependent pathway upstream of RIA. This gating mechanism may allow RIA to selectively attend to stimuli that are asymmetric across head sweeps. Attractive odor removal during head bends triggers rapid head withdrawal in the opposite direction. Unlike sensory encoding, this directional response is dependent on motor inputs to and synaptic output from RIA. Together, these results suggest that RIA is part of a sensorimotor pathway that is dynamically regulated according to head orientation at two levels: the first is a gate that filters sensory representations in RIA, and the second is a switch that routes RIA synaptic output to dorsal or ventral head motor neurons.

Keywords: C. elegans; behavior; calcium signaling; sensorimotor integration.

Figure 1.

RIA overview and requirement for control of head orientation. A, RIA is a unipolar interneuron whose axon extends into the ventral nerve cord and throughout the nerve ring. There are two RIAs, left and right, one is shown. B, Two axonal compartments within the nerve ring, nrV and nrD, encode head movements via local calcium signals (mCa²⁺) triggered by muscarinic input from SMD head motor neurons. C, Sensory pathways synapse on the loop domain of the RIA axon in the ventral nerve cord. Attractive stimuli lead to whole-axon reduction in calcium, while attractant removal causes whole axon calcium increases (sCa²⁺). D, Model for RIA function in steering behaviors. Where stimuli are symmetric across head bends, inhibitory output from RIA to motor neurons is symmetric. E, If the stimulus changes across head sweeps, RIA output to motor neurons become asymmetric, driving steering behavior. F, Schematic of head orientation measurement in response to a unilaterally presented stimulus. G, Estimation plot of difference between wild-type preference for odor as measured by percentage of the assay time (120 s) with their head within the odor stream and RIA::TeTx animals or animals treated with scopolamine. ANOVA F_(2,50) = 5.7771, p = 0.0055, *p < 0.05, **p < 0.01 post hoc Student’s t test, n = 13, n = 20, n = 20. V, ventral; D, Dorsal. Open circles and lines are mean ± SD (top). Closed circles and lines are mean difference from control ± 95% confidence interval, shown with distribution of estimated means (see Materials and Methods).

Figure 2.

Responses to odor removal depend on posture. A, Schematic of position-velocity plots to illustrate oscillatory head movements. B, Position-velocity trajectories from odor removal (circles) to 2 s after (arrowheads). C, Head displacement in the 2 s immediately following odor removal, with the start position normalized to head orientation at the odor switch. D, Relationship between head position at odor removal and the change in head position 2 s later. Linear regression F_(1,124) = 132.3145, p < 0.0001; n = 126.

Figure 3.

RIA mediates directional head withdrawal. A, To analyze head withdrawal behaviors, which are mirror-symmetric across the dorsoventral axis, head deflection in either the dorsal (D) or ventral (V) direction is defined as positive. Positive velocities correspond to bending away from the body axis and negative velocities indicate head withdrawal. B, Mean plots of peristimulus head orientation at odor off, odor on, or in constant odor (no switch), binned according to whether the head is unbent (top) or bent in either direction (bottom) at the time of stimulus change (t = 0). Dashed “pre” and “post” boxes indicate time windows used for quantitation in D. C, Comparison of head velocity in response to stimulus changes (or no switch) when the head is bent or unbent. Arrow indicates characteristic head withdrawal in response to odor removal when the head is bent. D, Paired mean pre- and post-switch head velocities (upper panels, open circles and lines are mean ± SD) and estimations of the size of the head velocity change (lower panel, closed circles and lines are mean difference ± 95% confidence interval, along with probability distribution of means, see Materials and Methods). Odor switch (on, off, or constant odor) has no effect on head movements when the head is not bent (left panel, repeated measures ANOVA F_(2,263) = 0.9386, p = 0.3925, n = 100, n = 82, n = 84). Sharp decreases in velocity are seen when odor removal occurs when the head is bent, but not for odor presentation or constant odor (middle panel, repeated measures ANOVA, F_(2,109) = 7.7932, p = 0.0007, n = 26, n = 44, n = 42). In gar-3 mutants and animals expressing tetanus toxin in RIA these sharp decreases in head velocity are absent (right panel, repeated measures ANOVA F_(3,85) = 4.4105, p = 0.0062, n = 42, n = 12, n = 18, n = 17); ***p < 0.001, **p < 0.01 post hoc Student’s t test between odor conditions (on, off, no switch) for wild-type animals; ##p < 0.01, #p < 0.05 post hoc Student’s t test between genetic manipulation (wild type, gar-3, gar-3 rescue, RIA::TeTx) for odor off responses during head bending.

Figure 4.

Head orientation gates sensory responses. A, Head position-velocity trajectories color-coded by ∂nr (= nrV – nrD), a measure of normalized mCa²⁺ asymmetry in nrV and nrD. B, Head position-velocity trajectories color-coded by loop calcium signal, n = 126, 40 s each. C, Spontaneous head movements (left) and sCa²⁺ signals in the loop region of the RIA axon (right). Rows are matched and sorted according to head orientation at odor off (t = 10 s). A linear regression of loop response magnitudes and head deflection in either direction was significant (F_(1,124) = 29.98, p < 0.0001, n = 126). D, Mean traces of loop calcium responses 5 s before and after odor removal binned by head orientation at the odor off time point. Shading is SEM. Left to right, n = 14, n = 6, n = 18, n = 22, n = 23, n = 21, n = 13, n = 9. E, Comparison of the magnitude of loop calcium responses to odor removal when the head is bent (head orientation > 0.5 or < –0.5) or unbent (head orientation between –0.5 and 0.5). Pre- and post-measurements are mean loop Ca²⁺ levels in 1s windows just before odor OFF and centered on the peak of the mean response, respectively, shown as paired responses (lines) and mean ± SD (open circles and lines). Estimations of the mean changes are shown below as mean difference ± 95% confidence interval (closed circles and lines) along with probability distributions of the means (see Materials and Methods). In both wild-type (n = 84, n = 42) and gar-3 (n = 20, n = 12) animals, the loop response magnitude is larger when the head is bent (repeated measures ANOVA F_(1,154) = 41.4529, p < 0.0001, ***p < 0.001 post hoc Student’s t test). gar-3 mutants have larger responses under both head orientation conditions, but there is no significant interaction between loop Ca²⁺ responses, head position at odor off, and genotype (repeated measures ANOVA F_(1,154) = 0.2145, p = 0.6439).

Figure 5.

Temporal features of sensory responses in RIA axonal compartments. A, Mean nrV, nrD, and loop responses on odor removal. Shading is SEM, dashed lines indicate position of nrV and nrD mean peaks. B, Head orientation predicts the mean lag between nrV and nrD peaks post-odor removal. ANOVA F_(1,124) = 13.47, p = 0.0004. C, Cross-correlation of nrV, nrD, and mCa²⁺ asymmetry (∂nr) and head orientation, showing mCa²⁺ lag with head movements. gar-3 mutants lack mCa²⁺ and do not show head correlations. Re-expression of gar-3 cDNA in RIA partially rescues this relationship. D, nrV and nrD cross-correlations with head velocity for wild-type, gar-3, and RIA-specific gar-3 rescue in a 6-s time window comprising 2-s pre-odor removal and 4-s post-odor removal in relation to head orientation at odor off. When the head was bent, we observed no lag between head withdrawal and peak calcium responses in the nerve ring compartment ipsilateral to the direction of bending. gar-3 mutants show no distinction between nerve ring compartments. Expression of gar-3 cDNA in RIA in gar-3 mutants does not rescue the temporal features of sensory responses in the nerve ring. Analysis groups are the same as Figure 4.

Figure 6.

Gate and switch model for RIA function in head orientation behaviors. A, RIA circuit diagram, with major upstream sensory pathways and head motor connections. Gating of sensory inputs to the loop domain and the nerve ring connections that constitute the dorsal-ventral switch. B, Ethological model in which head movements simultaneously produce proprioception-gated reafference and efference copies that converge on RIA. RIA output to motor neurons biases the undulatory gait dorsally or ventrally when reafferent stimuli are asymmetric across head sweeps.