Extended Data Fig. 10: Bergmann glial ABCA1 mediate phagocytosis and HOKR learning.

a-d, Data from Tabula Muris. a, t-Distributed Stochastic Neighbor Embedding (t-SNE) plot of individual non-myeloid cell in the brain. Colors correspond to each cell type. b, c, Abca1 was enriched mainly in oligodendrocyte precursor cell, BG, and astrocytes. d, BG expressed multiple type of PS-dependent phagocytic pathway molecules including Abca1, Megf10, Lrp1, Mertk, Bai1, Itgav, Itgb5, as well as astrocytes markers. These features were similar to the general astrocyte population. e, Immunohistochemical detection of ABCA1 in the cerebellum. ABCA1 staining showed a similar morphological pattern of BG soma and processes (arrowheads). Scale bar: 100 µm. We have done 1 independent experiment. f, Schema for conditional targeting Abca1^flox/flox with human GFAP-Cre. g, Protein quantification of ABCA1 by western blotting from the cerebellum. Conditional knockout (cKO) of Abca1 gene decreased ABCA1 protein in the cerebellar tissue (n = 2 mice). We have done 2 independent experiments. h, Schema of the experimental design (top left). AAV2/9-L7-4mCMV-tTA injected into tetO-pHRed; Abca1^flox/flox with or without hGFAP-Cre mice to create PCs-specific pHRed expression in ABCA1 cKO mice. Frequency of finding of pHRed puncta within BG soma was apparently less in ABCA1 cKO mice (bottom left). pHRed intensity in BG was significantly less in ABCA1 cKO (middle graph) and no such difference between the control was observed for pHRed intensity in PC (right graph), (n = 3 mice). Two-sided unpaired t-test (p = 0.0051 for BG; p = 0.75 for PC). i, Schema for conditional targeting Abca1^flox/flox with GLAST-Cre^ERT2 mice (above) and the experimental schedule (below). Tamoxifen was injected at 8 weeks for consecutive 5 days. HOKR experiments were done after three weeks of wait time to achieve efficient KO of ABCA1. j, HOKR learning curves of three types of control groups and the cKO group are shown (For control: n = 9 (Abca1^fl/fl: n = 5; Abca1^fl/fl + Tam: n = 2; Abca1 ^fl/fl, Glast-Cre^ERT2 n = 2) For ABCA1 cKO: n = 3 mice (Abca1^fl/fl, Glast-Cre^ERT2 + Tam). k, Significant improvement in the HOKR gain was observed from the 4th training session on Day 1 to the test session on Day 2 for the control group. However, such improvement was absent in the ABCA1 cKO group (control: n = 9; ABCA1 cKO n = 3 mice). Two-sided paired t-test (4^th session vs 2^nd day: p = 0.0081 for control; p = 0.072 for ABCA1 cKO). Unpaired t-test (p = 0.011 control vs ABCA1 cKO). All data are shown as mean ± s.e.m.

Source data