In one MICrONS mouse, shared trial-to-trial variability did not predict synapses between excitatory neurons after signal similarity and per-neuron fixed effects (728 connected pairs)

Question: do synaptically connected excitatory neurons in mouse visual cortex share more trial-to-trial (noise) variability than nearby unconnected neurons? Approach: we used the MICrONS pair table of Ding et al. (2025) and kept pairs with both neurons in the same imaging scan: 728 connected and 23,602 axon-dendrite proximate unconnected pairs, 148 presynaptic and 8,123 postsynaptic neurons, 13 scans. We extracted fluorescence traces from the public NWB files (DANDI:000402) by partial HTTP reads, deconvolved them, and used 6 movie clips repeated 10 times. For each pair we calculated signal correlation (SC) and noise correlation (NC), also after removal of running, pupil, eye position and 1 or 5 population modes. Poisson models of synapse count had presynaptic and postsynaptic fixed effects, a co-travel offset, soma distance and field covariates. Result: the rate ratio per SD of NC was 1.01 (95% CI 0.90-1.14), 1.00 (0.90-1.12) after behaviour removal and 1.01 (0.90-1.13) after population-mode removal, while SC gave 1.21 (1.09-1.36). Results were the same in high-pupil and low-pupil bins and for pairs in different imaging fields. Connected pairs had a small raw NC excess (0.034 vs 0.026), which SC, proximity and per-neuron effects explained. Limitations: one mouse, slow calcium signals, 600 s of repeats per scan, NC split-half reliability 0.59; reliability-corrected effects above a rate ratio of about 1.2 per SD are excluded, smaller effects are not. The mice almost never ran, so running was not tested.

MICrONS · connectomics · noise correlation · signal correlation · like-to-like wiring · mouse visual cortex · two-photon imaging · fixed effects · structure-function · arousal · DANDI 000402 · null result