Behavioral control by depolarized and hyperpolarized states of an integrating neuron

Coordinated transitions between mutually exclusive motor states are central to behavioral decisions. During locomotion, the nematode Caenorhabditis elegans spontaneously cycles between forward runs, reversals, and turns with complex but predictable dynamics. Here we provide insight into these dynamics by demonstrating how RIM interneurons, which are active during reversals, act in two modes to stabilize both forward runs and reversals. By systematically quantifying the roles of RIM outputs during spontaneous behavior, we show that RIM lengthens reversals when depolarized through glutamate and tyramine neurotransmitters and lengthens forward runs when hyperpolarized through its gap junctions. RIM is not merely silent upon hyperpolarization: RIM gap junctions actively reinforce a hyperpolarized state of the reversal circuit. Additionally, the combined outputs of chemical synapses and gap junctions from RIM regulate forward-to-reversal transitions. Our results indicate that multiple classes of RIM synapses create behavioral inertia during spontaneous locomotion.

Methods
Behavioral states were extracted from the State array generated by BargmannWormTracker. The tracker can be accessed at: https://github.com/navinpokala/BargmannWormTracker

Local search event frequencies per minute were calculated 4-8 min after removal from food. Global search frequencies per minute were calculated 36-40 min after removal from food. Only tracks that were continuous for the entire four-minute time interval were included in frequency analysis. When calculating frequencies, tracks taken on a single day from a single assay plate were averaged to give a single data point, e.g. in Figure 2B and 2D.

Distributions of reversal parameters and forward run durations were calculated using events observed during local search, 4-8 minutes after removal from food. All reversals were included; only forward runs over 2 s in length were included. Reversal length is the path length calculated using the X-Y coordinates, worm length, and pixel size extracted from the tracker. Reversal and forward run speed are the average of mean and median speed extracted from the tracker.

Tracks that were less than 5 minutes long, tracks approaching the copper barrier, and tracks that did not include a complete reversal or forward run were not included in reversal and forward run parameter analyses.

Custom MATLAB functions used to generate these data sets are included here (SordilloCode1.m, SordilloCode2.m).

Imaging data was tracked using an ImageJ Macro originally used in in Larsch et al., 2013 (https://doi.org/10.1073/pnas.1318325110). Tracks were cleaned and processed using custom Python and  MATLAB functions included here (SordilloCode3.m, SordilloCode4).