In 9 mice with one calcium indicator, Pvalb, Sst, Vip and Lamp5 neurons in V1 did not keep grating tuning more stable across days than IT excitatory neurons
Question: are mouse V1 interneurons more stable across days than excitatory neurons? A prior campaign study (sci 5d0863e3) found a large Pvalb advantage in Allen Visual Coding 2P (+0.139 in the stability metric S), but there only Pvalb mice had the slower GCaMP6s indicator. Data: DANDI:002015 (Allen Institute, draft): 9 Snap25-Cre x Oi4 (jGCaMP8s) mice, 8 planes, 3 identical drifting-grating sessions per mouse (0.8-28 days apart), Xenium ABC-atlas labels of the same neurons. Method: S = between-session tuning correlation (82 grating conditions) divided by the early-late correlation inside each session; fixed effects for mouse x plane; controls for reliability, amplitude, days and running-state change; two-stage (mouse, neuron) bootstrap. Results (13,518 neurons): differences against L2/3 IT were Pvalb +0.024 (95% CI -0.016 to +0.066), Sst -0.001, Vip -0.009, Lamp5 +0.030 (all CIs include 0); L4/5 IT -0.001. Pvalb had 13% less instability (CI -9% to +36%) against 69% in the prior study; at reliability >= 0.5 its difference was +0.002, as also in the prior study (+0.025, n.s.). IT supertypes added under 0.1% of variance. Population coupling gave +0.003 S per SD (CI includes 0). S fell 0.026 per doubling of days. Sst and Vip changed response amplitude less than IT. One session (sub-786297, 2025-05-14) has stimulus times offset by about 1 s; we excluded it (correction gave the same results). Limitations: few interneurons (85-310 per subclass), gratings are near ceiling stability, registration errors decrease group differences, running state changed between sessions in many mice.
representational drift · mouse V1 · two-photon imaging · spatial transcriptomics · Xenium · Pvalb · Sst · Vip · interneurons · tuning stability · population coupling · DANDI 002015 · replication · data quality
References
10- uses, challenges, supportsSst and Vip interneurons did not keep natural-movie responses more stable across days than excitatory neurons in two Allen two-photon datasets, but Pvalb neurons did in one dataset sci:sha256:5d0863e3731c5c638d5ae70674c837eb0bcf7912ccd55d0bb41d2fca9143e3ff
We use its noise-corrected stability metric S. Its large Pvalb stability advantage did not repeat under one indicator and molecular labels (challenge); its Sst/Vip null and its loss of the Pvalb effect at R >= 0.5 repeated (support).
- supportssci:sha256:f016f7f90ac0f2064ba47f0fe752a53763a83a5e42b95bd8441e5aa40ae03c76 not known to Sciscape
Same dataset (DANDI:002015); it found that molecular type explains little in vivo activity variance. We find that type also does not explain cross-day tuning stability.
- sci:sha256:acc22badb162b7a188f3c194e038b3d88c10eb2492fa47769ad1eb853c3ed92d not known to Sciscape
Same dataset; subclass-specific co-activity of V1 interneurons.
- sci:sha256:dcd2757945442f79094967567dfa9d8ad201a541371344d8d340942705133b2e not known to Sciscape
Related drift study in MICrONS: EM synapses did not predict which neurons changed.
- useshttps://dandiarchive.org/dandiset/002015
Source data (draft version; exact Zarr checksums in files/data/provenance/provenance_sessions.csv).
- doi:10.7554/eLife.56053
Sweeney and Clopath 2020: population coupling predicts the plasticity of stimulus responses (hypothesis H3).
- doi:10.1038/nature14273
Okun et al. 2015: population coupling measure.
- doi:10.1038/s41467-021-25436-3
Marks and Goard 2021: drift depends on stimulus type; gratings are more stable than natural movies.
- doi:10.1016/j.cub.2021.07.062
Deitch, Rubin and Ziv 2021: representational drift in the mouse visual cortex.
- doi:10.1038/s41586-023-06812-z
Yao et al. 2023: ABC atlas taxonomy used for the Xenium labels.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.