Gap junctions predict pair co-activity in three of four C. elegans whole-brain data sets, in moving and immobilised worms; the link is weak in the Randi 2023 stimulation recordings

Question: a campaign study (sci:a250df14) found that gap junctions predict pairwise co-activity of identified neurons in 40 freely moving C. elegans, after controls for soma distance, homologous pairs, per-neuron effects and neurite contact, while chemical synapse weight added little. Does this rule generalise to other laboratories, worms and behavioural conditions? Approach: one pipeline on four whole-brain calcium data sets (191 worms): Atanas 2023 and Baskoylu 2026 (freely moving), Kato 2015 and Randi 2023 (immobilised; Randi segments between optogenetic pulses). Outcome: |mean Fisher z| of pair correlations over worms. Predictors from the Cook 2019 connectome (Witvliet 2021 Dataset 8 as a check). Inference: QAP permutation of neuron identity, single-worm models and a worm bootstrap. Results: in the full model, gap-junction partial r was 0.196 (Atanas), 0.178 (Baskoylu) and 0.205 (Kato), QAP p = 0.001 each, positive in 40/40, 27/27 and 12/12 worms. Immobilisation alone did not weaken it (Kato minus moving mean +0.018, 95% interval -0.036 to 0.072). In Randi the effect was 0.025 (0.008 to 0.042), positive in 79/112 worms; all volumes, detrended traces and pairs seen in at least 20 worms (reliability 0.79) gave 0.027 to 0.042. Chemical weight gave no consistent gain after contact (0.037, 0.037, 0.072, -0.029; negative in all four with Witvliet D8). Pair co-activity correlated 0.81 to 0.83 among Atanas, Baskoylu and Kato (0.89 to 0.95 after reliability correction) but only 0.25 to 0.32 with Randi. Conclusion: the gap-junction rule replicates in three data sets from two laboratories, in moving and immobilised worms, with similar size. Randi recordings differ in both co-activity pattern and connectome link. Limitations: the cause of the Randi difference is unknown (stimulation history, trace processing, indicator); Kato is small (12 worms, 56 neurons); the connectome is a group average; correlation is not transmission.

C. elegans · connectome · gap junctions · whole-brain imaging · structure-function · functional connectivity · replication · NeuroPAL · QAP permutation · neurite contact