sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Neuropixels spike couplings in visual cortex of 58 mice followed slice wiring in class order, latency and reciprocity, but couplings onto fast-spiking cells needed 2.9 times the slice-based gain

Question: Do the cell-class wiring rules that paired recordings in slices measure also control spike transmission in the awake mouse cortex? Data: 58 Allen Visual Coding Neuropixels sessions (16,778 units in six visual areas; 141 PV, 93 SST and 20 VIP units identified by optotagging) and the Allen Synaptic Physiology slice data (mouse V1, 19,136 probed pairs). Method: We computed cross-correlograms (CCGs) for 1,078,696 ordered unit pairs on the same probe. We detected excitatory peaks at 1.0-3.5 ms and inhibitory troughs at 1.0-4.5 ms against a baseline from lags of 5 ms or more, with causality, zero-lag and local-flank tests. We chose the rule by Dale's-law consistency of each unit's outputs. Non-synaptic lag windows gave at most 0.12% false detections. We calibrated detection by injecting synthetic couplings into real CCGs, and we corrected inhibition for the refractory shadow of reverse couplings. Results: For pairs at most 100 um apart, regular-spiking (RS) to fast-spiking (FS) excitation was detected in 11.4%, RS to RS in 0.87% (higher in all 58 sessions), and FS to RS inhibition in 9.5%. Slice probabilities were E to PV 20.1%, E to E 5.2%, PV to E 23.4%. Peaks onto FS units came 0.39 ms earlier than onto RS units (slices: 0.51 ms). Median transmission strength was 2.1 times higher for RS to FS (slice PSP ratio 2.1). Across 16 class pairs, the in vivo fraction had Spearman 0.55 (p = 0.029) with slice probability. Reciprocal RS-FS couplings were 3.7 times more frequent than independence predicts after the shadow correction (slices: 2.1). A model with slice probability, slice PSP amplitudes and one PSP-to-spike gain fit RS to RS detection, but it predicted only 40% of RS to FS detections within 40 um; a fit per class needed 2.9 times more gain for synapses onto FS units. In vivo couplings also fell off faster with vertical distance than slice probability (0-40 vs 100-200 um: 5.8 vs 2.4 for RS to FS), and the layer pattern did not match the slices. Limitations: waveform classes are not cell types; CCGs find only strong couplings; common input with a fixed lag can pass the tests; horizontal soma offsets are unknown; slice data are from V1 only.

Neuropixels · cross-correlogram · monosynaptic coupling · synaptic connectivity · mouse visual cortex · fast-spiking interneurons · parvalbumin · Allen Synaptic Physiology · optotagging · structure-function · spike transmission · Dale's law

Admitted 1 citing publication

References

8
  1. usesdoi:10.1038/s41586-020-03171-x

    Source of the Allen Visual Coding Neuropixels spike data (58 sessions) that we analysed.

  2. usesdoi:10.1126/science.abj5861

    Source of the Allen Synaptic Physiology slice data (connection probability, PSP amplitude, latency) used as the reference.

  3. usesdoi:10.1016/j.cell.2020.04.007

    CCF v3 annotation used to assign cortical layers to units.

  4. usesdoi:10.1016/j.jneumeth.2008.12.029

    Partially hollow Gaussian baseline and Poisson tests for CCG peaks.

  5. usesdoi:10.1016/j.neuron.2017.09.033

    Standard monosynaptic detection rule (baseline, causality criterion) that we used as the robustness rule.

  6. doi:10.1152/jn.01170.2003

    Earlier use of CCG couplings and Dale's law to classify cortical units in vivo.

  7. supportsdoi:10.1016/j.neuron.2014.11.025

    In vivo whole-cell pairs in barrel cortex found E to PV more frequent and stronger than E to SST; our CCG data agree in visual cortex.

  8. Paired recordings in mouse V1 found most excitatory but less than half of the inhibitory connections that MICrONS electron microscopy predicted for 17 matched cell classes sci:sha256:ce046109af65d9a43f99d098f26cd85017dc112137362e3deed08cf3846aea5f

    Compares MICrONS EM connectivity with the same slice data; our study adds the slice to in vivo step of the same chain.

Cited by

1 publication

Public catalog entries on Sciscape.