Neuroligin genes (Nlgn1-3) are abundantly expressed by neurons, astrocytes, and OPCs in brain as determined by analyses of publicly available RNAseq datasets.

(A-D) Analysis of Nlgn1, Nlgn2, and Nlgn3 mRNA levels in neurons (green), astrocytes (blue), oligodendrocyte lineage cells (orange), microglia (red), and other cell types in the brain (yellow) using the single-cell RNAseq dataset published from the McCaroll lab (Saunders and Macosko et al., 2018, www.dropviz.org) (A), Chan Zuckerberg Initiative (Schaum et al., 2018) (B), Wu lab (Zhang et al., 2014) (C), and Linnarson lab (Zeisel et al., 2018, www.mousebrain.org) (D). Note that although relative expression levels vary greatly between datasets, all datasets support the conclusion that Nlgn1, Nlgn2, and Nlgn3 are broadly expressed in astrocytes and OPCs.

(E) Analysis of Nlgn1, Nlgn2, and Nlgn3 mRNA levels in astrocytes in three different brain regions (cortex, hippocampus, and striatum) using the bulk RNAseq datasets published by the Khakh lab (Chai et al., 2017; Srinivasan et al., 2016, www.astrocyternaseq.org) that examined RiboTag-purified mRNAs.

(F) Analysis of Nlgn1, Nlgn2, and Nlgn3 mRNA levels in astrocytes as a function of age in mice using the bulk RNAseq datasets published by the Barres lab (Clarke et al., 2018; www.brainrnaseq.org). Astrocyte mRNA was purified from three different brain regions (cortex, hippocampus, and striatum) by RiboTag pulldowns in Aldh1l1-eGFP-L10a mice.

Note that Nlgn4 is not measured in the RNAseq experiments shown, probably because its expression levels are low and because its mRNA is very GC rich. For an analysis based on the pioneering original Barres lab data that used a less deep sequencing approach, see Figure S1.

Nlgn1-3 are efficiently and selectively deleted in astrocytes by crossing triple Nlgn1-3 conditional KO mice with Adh1l1-CreERT2 driver mice and inducing Cre-activity with tamoxifen early during postnatal development.

(A) Breeding strategy. Triple conditional KO mice carrying floxed Nlgn1, Nlgn2, and Nlgn3 alleles or mice with a Cre-sensitive tdTomato (tdT) reporter allele (Ai14) were crossed with pan-astrocyte, tamoxifen-inducible Aldh1l1-CreERT2 BAC transgenic mice. Nlgn1-3 cKO mice were crossed for multiple generations until homozygosity was reached (females: Nlgn1f/f 2f/f 3f/f, males: Nlgn1f/f 2f/f 3f/y).

(B & C) Two different tamoxifen-induced Cre-activation schedules were used to delete Nlgn1-3 in astrocytes. Aldh1l1-CreERT2 mice and controls (littermate Nlgn1-3 cKO mice lacking the Aldh1l1-CreERT2 allele) were injected with tamoxifen at P10 and P11 (B) (Trotter et al., 2021) or at P1 (C). Mice were sacrificed at least four (B) or five weeks post Cre induction to ensure complete deletion of neuroligins and decay of any astrocyte- specific neuroligin proteins.

(D & E) To confirm the specificity and efficiency of the deletion of target genes in astrocytes using the Aldh1l1-CreERT2 BAC transgenic mouse line via tamoxifen injection at P1, Cre-recombination was visualized in the hippocampus (D) and primary visual cortex (E) via expression of tdT in reporter mice (magenta). Sections were additionally labeled for NeuN to mark neurons (green) and S100β to mark astrocytes (blue).

(F) The Algh1l1-CreERT2-induced deletion of floxed genes produced by P1 tamoxifen injections is effective and selective for astrocytes as quantified using expression of tdTomato in reporter mice. tdTomato expression was quantified in the CA1 region of the hippocampus (the S. oriens, S. pyramidale, S. radiatum, and S. lacunosum-moleculare), the dentate gyrus (molecular layer [ML], granule cell layer [GCL], and hilus), and layer IV of the primary visual cortex.

Quantification of the Nlgn1-3 conditional KO efficiency in astrocytes by qRT-PCR using RiboTag-mediated isolation of astrocytic mRNAs.

(A) Diagram of the experimental strategy. AAVs (AAV2/5) encoding the RiboTag (Rpl22- HA) and Lck-mVenus (a membrane-targeted fluorescent reporter) driven by the astrocyte- specific GfaABC1D promoter and additionally including a 4x6T cassette of mIR-124 targeting sequences (Gleichman et al., 2023) were injected into the CA1 region of mice with a P0 induction of the astrocyte-specific Nlgn1-3 KO or control mice at P35-40 and analyzed 4 weeks later.

(B) Representative coronal brain section (counterstained with DAPI) showing successful viral targeting of CA1 and adjacent regions indicated by the presence of Lck-mVenus (green) and Rpl22-HA (purple) astrocytes.

(C-E) quantitative RT-PCR validation of the successful pull-down of highly-enriched astrocyte ribosome-bound mRNA from micro-dissected cortical tissue (i.e. Rpl22-HA). Enrichment of astrocyte marker, Aqp4, and de-enrichment of other cell marker genes is expressed as levels in pull-down (i.e. IP) relative to total mRNA (i.e. input).

(D) Confirmation in astro-Nlgn123 cKO mice of reduced levels of Nlgn1, Nlgn2, and Nlgn3

using assays that recognized floxed exons here expressed as levels in IP relative to input.

(E) Same as D, except expressed as a % change relative to wild-type based on the 2- ΔΔCt method.

(F - H) Same as C-E, but for the hippocampal CA1 region.

Data in summary graphs are means ± SEM with each data point representing individual animals. Unpaired two-tailed t-tests were used to test statistical significance of data in D and G, whereas a one sample t-test was used to test statistical significance of data in E and H (*, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p<0.0001).

Conditional deletion of Nlgn1-3 in astrocytes throughout the brain at early postnatal timepoints (P10/11 or P1) has only modest effects on overall neuroligin protein levels and does not significantly alter the synaptic proteome.

(A & B) Representative immunoblots and quantifications of Nlgn1, Nlgn2, and Nlgn3 protein levels from hippocampal (A) and cortical lysates (B) of astrocyte Nlgn1-3 cKO and littermate control mice injected with tamoxifen at P10 and P11 and sacrificed at P38. Proteins were quantified on immunoblots using fluorescent secondary antibodies, with protein levels normalized to β-actin and then to control levels (n = 5, all male).

(C & D) Representative immunoblots and (D) quantification for various synaptic protein levels from hippocampal and cortical lysates of astrocyte Nlgn1-3 cKO and littermate control mice injected with tamoxifen at P10 and P11 and sacrificed at P38. Protein is quantified using fluorescent secondary antibodies, with protein levels normalized to β- actin and then to control levels (n = 5, all male).

(E-H) Same as (A-D) except mice were injected with tamoxifen at P1 and sacrificed at P35 (n = 4, 2 male & 2 female).

Numerical data are means ± SEM with statistical significance determined by unpaired two-tailed t-test (**, p<0.01).

Conditional deletion of Nlgn1-3 in astrocytes starting at P1 does not alter excitatory or inhibitory synapse numbers in the hippocampus as assessed by immunocytochemistry with antibodies to synaptic markers.

(A) Representative images of CA1 and dentate gyrus hippocampal sections from astrocyte Nlgn1-3 cKO and littermate control mice, injected with tamoxifen at P1 and sacrificed at P35, stained for dendritic marker MAP2 (magenta), excitatory presynaptic marker vGluT1 (red), excitatory postsynaptic marker Homer1 (green), and DAPI (blue), taken at 20X (top) and 60X (bottom) magnification.

(B) Quantification of total vGluT1 (top) and Homer1 (bottom) immunofluorescence signal for low magnification imaging (20X) across the layers of the hippocampus (S. oriens, S. pyramidale, S. radiatum, S. lacunosum-moleculare, dentate gyrus molecular layer), first internally normalized to MAP2 and then to average vGluT1 (top) or Homer1 (bottom) immunofluorescence level in S. oriens of control mice.

(C) Quantification of total vGluT1 (top) and Homer1 (bottom) immunofluorescence signal for high magnification imaging (60X) in the CA1 S. pyramidale (left) and S. radiatum (right), first internally normalized to MAP2 and then to average vGluT1 (top) or Homer1 (bottom) immunofluorescence level in control mice.

(D) Representative images of CA1 and dentate gyrus hippocampal sections from astrocyte Nlgn1-3 cKO and littermate control mice stained for dendritic marker MAP2 (magenta), excitatory presynaptic marker vGluT2 (red), excitatory postsynaptic marker Homer1 (green), and DAPI (blue), taken at 20X (top) and 60X (bottom) magnification.

(E) Quantification of total vGluT2 immunofluorescence signal for low magnification imaging (20X) across the layers of the hippocampus, first internally normalized to MAP2 and then to average vGluT2 immunofluorescence level in S. oriens of control mice.

(F) Quantification of total vGluT2 (top) and Homer1 (bottom) immunofluorescence signal for high magnification imaging (60X) in the CA1 S. lacunosum-moleculare (left) and dentate gyrus molecular layer (right), first internally normalized to MAP2 and then to average vGluT2 (top) or Homer1 (bottom) immunofluorescence level in control mice.

(G) Representative images of CA1 and dentate gyrus hippocampal sections from astrocyte Nlgn1-3 cKO and littermate control mice stained for dendritic marker MAP2 (magenta), inhibitory postsynaptic marker Gephyrin (red), inhibitory presynaptic marker GAD67 (green), and DAPI (blue), taken at 20X (top) and 60X (bottom) magnification.

(H) Quantification of total Gephyrin (top) and GAD67 (bottom) immunofluorescence signal for low magnification imaging (20X) across the layers of the hippocampus, first internally normalized to MAP2 and then to average Gephyrin (top) or GAD67 (bottom) immunofluorescence level in S. oriens of control mice.

(I) Quantification of puncta density for Gephyrin (top left), GAD67 (bottom left), Gephyrin having GAD67 (top right), and GAD67 having Gephyrin (bottom right) for high magnification imaging (60X) in the CA1 S. radiatum.

Data are means ± SEM with statistical significance determined by unpaired two-tailed t- test (n=4, 2 male & 2 female).

Conditional deletion of Nlgn1-3 in astrocytes starting at P1 has no major effect on basal excitatory or inhibitory neurotransmission monitored in hippocampal CA1 pyramidal neurons.

(A) Representative traces for miniature EPSCs (mEPSCs) from CA1 pyramidal neurons in acute slices from astrocyte Nlgn1-3 cKO and littermate controls injected with tamoxifen at P1 and recorded at P44 - P50.

(B) Cumulative distribution and summary graph of mEPSC amplitude and (C) frequency.

(D) Summary graph of mEPSC rise (left) and decay (right) times (n = 14-15 cells / 3 mice per genotype).

(E-H) Same as (A-D) except for miniature IPSCs (mIPSCs) (n = 16 cells / 3 mice per genotype).

Data in summary graphs are means ± SEM with each data point representing individual cells. Unpaired two-tailed t-tests were used to test statistical significance of data in bar graphs, and Kolmogorov-Smirnov tests were used for cumulative curves (****, p<0.0001).

Conditional deletion of Nlgn1-3 in astrocytes starting at P1 does not alter excitatory or inhibitory synapse numbers in layer IV of the primary visual cortex as assessed by immunocytochemistry with antibodies to synaptic markers.

(A) Representative images of primary visual cortex (V1) layer IV (L4) astrocyte Nlgn1-3 cKO and littermate control mice, injected with tamoxifen at P1 and sacrificed at P35, stained for dendritic marker MAP2 (magenta), excitatory presynaptic marker vGluT1 (red), excitatory postsynaptic marker Homer1 (green), and DAPI (blue) taken at 60X magnification.

(B) Quantification of total vGluT1 (left) and Homer1 (right) immunofluorescence signal in V1L4 first internally normalized to MAP2 and then to average vGluT1 (left) or Homer1 (right) immunofluorescence level in control mice.

(C) Representative images of V1L4 astrocyte Nlgn1-3 cKO and littermate control mice stained for dendritic marker MAP2 (magenta), excitatory presynaptic marker vGluT2 (red), excitatory postsynaptic marker Homer1 (green), and DAPI (blue) taken at 60x magnification.

(D) Quantification of total vGluT2 immunofluorescence signal in V1L4 first internally normalized to MAP2 and then to average vGluT2 immunofluorescence level in control mice.

(E) Representative images of V1L4 astrocyte Nlgn1-3 cKO and littermate control mice stained for dendritic marker MAP2 (magenta), inhibitory postsynaptic marker Gephyrin (red), inhibitory presynaptic marker GAD67 (green), and DAPI (blue) taken at 60X magnification.

(F) Quantification of total Gephyrin (left) or GAD67 (right) immunofluorescence signal in V1L4 first internally normalized to MAP2 and then to average gephyrin (left) or GAD67 (right) immunofluorescence level in control mice.

(G) Quantification of Gephyrin (left) or GAD67 (right) puncta density.

(H) Quantification of Gephyrin (left) or GAD67 (right) puncta size in V1L4.

(I) Quantification of puncta density for Gephyrin having GAD67 (left) or GAD67 having Gephyrin (right) in V1L4.

Data are means ± SEM with statistical significance determined by unpaired two-tailed t- test (*, p<0.05) (n=4, 2 male & 2 female).

Conditional deletion of Nlgn1-3 in astrocytes starting at P1 does not detectably alter the proteome or the cytoarchitecture of astrocytes in the hippocampus or layer IV of the primary visual cortex.

(A) Representative immunoblots and (B) quantification for various glial protein levels from hippocampal and cortical lysates of astrocyte Nlgn1-3 cKO and littermate control mice injected with tamoxifen at P1 and sacrificed at P35. Protein is quantified using fluorescent secondary antibodies, with protein levels normalized to β-actin and then to control levels (n = 4, 2 male & 2 female).

(C) Representative images of CA1 and dentate gyrus hippocampal sections from astrocyte Nlgn1-3 cKO and littermate control mice, injected with tamoxifen at P1 and sacrificed at P35, stained for astrocytic marker GFAP (white) and DAPI (blue), taken at 20X magnification.

(D) Quantification of total GFAP immunofluorescence signal across the layers of the hippocampus (S. oriens, S. pyramidale, S. radiatum, S. lacunosum-moleculare, dentate gyrus molecular layer), normalized to average GFAP immunofluorescence level in S. oriens of control mice (n = 4, 2 male & 2 female).

(E) To measure astrocyte volume, astrocyte Nlgn1-3 cKO and littermate control mice were injected with tamoxifen at P1, underwent stereotactic injections of AAV expressing membrane-bound mVenus in astrocytes in the hippocampus at P14, and were subsequently sacrificed two weeks later at P35. Representative images of mVenus- expressing astrocytes in CA1 S. radiatum are shown with corresponding 3D volume reconstruction performed in Imaris.

(F) Summary graph of CA1 S. radiatum astrocyte volumes shown averaged per animal with means ± SEM on the bar graph, as well as data points for individual astrocyte volumes. Statistical significance determined by unpaired two-tailed t-test of data averaged per animal (n=3, 1 male & 2 female).

(G & H) Same as E & F, except for primary visual cortex layer IV astrocytes.

Numerical data are means ± SEM. Dots in bar graphs represent independent biological replicates; in F and H, larger dots are independent biological replicates and smaller dots are pseudoreplicates since these are commonly reported in papers to boost statistical significance.

Deletion of Nlgn1-4 in mouse glia co-cultured with wild-type human neurons does not significantly alter spontaneous synaptic events (mEPSCs) mediated by the co-cultured neurons.

(A) Schematic of the experimental strategy

(B-D) Measurements of the mouse Nlgn1, Nlgn2, and Nlgn3 mRNA levels in the mouse glia culture as a function of the expression of ΔCre (Ctrl., negative control) or Cre (cKO) reveals complete deletion of the expression of all neuroligins.

(E-G) mEPSC measurements in human neurons co-cultured with mouse glia expressing (Ctrl.) or lacking neuroligins (cKO) fails to uncover any changes as a function of the neuroligin expression in the glia (E, sample traces; F, cumulative probability plot of the mEPSC amplitudes [inset, summary graph of the mean mEPSC amplitude]; G, cumulative probability plot of the mEPSC interevent intervals [inset, summary graph of the mean mEPSC frequency].

Numerical data are means ± SEM. Dots in bar graphs in F and G represent independent biological replicates.

Conditional deletion of astrocytic Nlgn1-3 at P1 does not alter the number or size of inhibitory synapses in the CA1-region Stratum Radiatum.

(A) Quantification of total Gephyrin (left) or GAD67 (right) immunofluorescence in CA1 Str. Radiatum from astrocyte Nlgn1-3 cKO and littermate control mice. Images of hippocampal sections were taken at 60X magnification. Total immunofluorescent signal was first internally normalized to MAP2 and then to average gephyrin (left) or GAD67 (right) immunofluorescence level in control mice.

(B) Quantification of puncta density for Gephyrin (left) and GAD67 (right) CA1 Str. Radiatum from astrocyte Nlgn1-3 cKO and littermate control mice.

Representative images are shown in Fig. 4G. Data are means ± SEM with statistical significance determined by unpaired two-tailed t-test (n=4, 2 male & 2 female).

Conditional deletion of astrocytic Nlgn1-3 at P1 does not alter CA1 pyramidal neuron membrane properties.

(A) Summary graph of membrane capacitance from CA1 pyramidal neurons in acute slices from astrocyte Nlgn1-3 cKO and littermate controls injected with tamoxifen at P1 and recorded at P44 – P50.

(B) Same as (A) but for membrane resistance.

Data are means ± SEM with statistical significance determined by unpaired two-tailed t- test (n = 14-15 cells / 3 mice per genotype).

Further data characterizing the effect of a deletion of glial Nlgn1-3 on co- cultured human neurons.

(A) Representative images of primary mouse glia cultures from Nlgn1-4 quadruple cKO mice infected with lentiviruses encoding FLP-EGFP (Ctrl.) or Cre-EGFP fusion proteins and stained for GFP, DAPI, and GFAP as glial marker. Note that the levels of EGFP expression differ among glia cells but that based on qRT-PCR measurements the lower levels of Cre-EGFP in some glial cells are sufficient for complete recombination of Nlgn1, Nlgn2, and Nlgn3 genes (Nlgn4 is constitutively deleted).

(B) Summary graph of the ratio of EGFP-expressing to DAPI-stained cells demonstrates that nearly all cells in the culture are infected by the lentiviruses.

(C) Summary graphs of the membrane capacitance (left) and input resistance (right) monitored in human neurons that are co-cultured with mouse glia expressing or lacking all neuroligins (Nlgn1-4).

Data are means ± SEM with statistical significance determined by unpaired two-tailed t- test (n = 14-15 cells / 3 mice per genotype).