Stereotaxic injection of C1ql2-expressing AAV into Bcl11b cKO DGN restores C1ql2 levels.

a Experimental design to analyze the functions of C1ql2 in the MFS as a downstream target of Bcl11b. b AAV constructs injected in the DG of Bcl11b cKO and control littermates. c Western blot and d relative C1ql2 protein levels in mouse hippocampal homogenates. N=3. All data are presented as means; error bars indicate SEM. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP-2A-C1ql2: ns, p=0.11; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP: *p=0.015; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP-2A-C1ql3: *p=0.019; ns, not significant. e Immunohistochemistry of C1ql2 (cyan), GFP (green), and Bcl11b (red) on hippocampal sections. Scale bar: 200 μm. f Immunohistochemistry of C1ql2 (cyan), vGlut1 (magenta) and Homer1 (yellow) in the SL of CA3. White arrowheads indicate co-localizing puncta of all three proteins. Scale bar: 15 μm

C1ql2 reintroduction in Bcl11b cKO DGN rescues SV recruitment in MFS.

a Electron microscope images of MFS and proximal SV. White bars mark synapse length from the postsynaptic side. Scale bar: 100 nm. b Average synapse score. Control+EGFP, N=3; Bcl11b cKO+EGFP, Bcl11b cKO+EGFP-2A-C1ql2, Bcl11b cKO+EGFP-2A-C1ql3, N=4. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: ***p=0.0002, and vs. Bcl11b cKO+EGFP-2A-C1ql3: ***p=0.0003; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP and vs. Bcl11b cKO+EGFP-2A-C1ql3: ***p<0.0001; ns, not significant. c Relative frequency of synapse scores. d Number of docked vesicles per 100 nm AZ profile length. Control+EGFP, Bcl11b cKO+EGFP-2A-C1ql3, N=3; Bcl11b cKO+EGFP, Bcl11b cKO+EGFP-2A-C1ql2, N=4. All data are presented as means; error bars indicate SEM. Points represent the individual examined AZ and SV, respectively. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: *p=0.024, and vs. Bcl11b cKO+EGFP-2A-C1ql3: *p=0.045; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP: *p=0.026, and vs. Bcl11b cKO+EGFP-2A-C1ql3: *p=0.049; ns, not significant. e Diameter of docked vesicles. Control+EGFP, Bcl11b cKO+EGFP-2A-C1ql3, N=3; Bcl11b cKO+EGFP, Bcl11b cKO+EGFP-2A-C1ql2, N=4; Two-way ANOVA. ns, not significant. f Cumulative and g relative frequency of the number of docked vesicles per 100 nm AZ profile length. h Western blot of mouse hippocampal homogenates. i Average synapse score. Control+EGFP, N=3; Control+EGFP-2A-C1ql2, N=6; Bcl11b cKO+EGFP-2A-C1ql2, N=4. Two-way ANOVA. ns, not significant. j Relative frequency of synapse scores. Data for Control+ EGFP-2A-C1ql2 from i-j in this figure are compared with Control+EGFP and Bcl11b cKO+EGFP-2A-C1ql2 data from Figure 2.

C1ql2 reintroduction in Bcl11b cKO DGN rescues mossy fiber LTP.

a Representative fEPSP traces showing baselines before HFS (black), fEPSP changes 30-40 min after HFS (cyan) and following the application of 3 µM DCG-IV (red). b Time course of fEPSP slopes. The black arrow indicates HFS and the dashed line is the baseline level. c Quantification of fEPSP facilitation at four different time intervals after HFS. Changes in the fEPSP slope are shown as the percentage of the mean baseline fEPSP. Control+EGFP, 7 slices from 6 mice; Bcl11b cKO+EGFP, 8 slices from 5 mice, Bcl11b cKO+EGFP-2A-C1ql3, 8 slices from 6 mice; Bcl11b cKO+EGFP-2A-C1ql2, 6 slices from 4 mice; All data are presented as means; error bars indicate SEM. One-way ANOVA followed by Bonferroni’s PHC for each time interval. 20-30 min: Control+EGFP vs. Bcl11b cKO+EGFP: **p=0.002, and vs. Bcl11b cKO+EGFP-2A-C1ql3: *p=0.011; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP: *p=0.023; 30-40 min: Control+EGFP vs. Bcl11b cKO+EGFP: ***p<0.001, and vs. Bcl11b cKO+EGFP-2A-C1ql3: **p=0.002; Bcl11b cKO+EGFP-2A-C1ql2 vs. Bcl11b cKO+EGFP: *p=0.01 and vs. Bcl11b cKO+EGFP-2A-C1ql3: *p=0.0523; ns, not significant. d Representative fEPSP traces showing baselines before forskolin application (black), fEPSP changes 105-120 min after the start of application (cyan) and following the addition of 3 µM DCG-IV (red). e Time course of fEPSP slopes. The black solid line indicates forskolin perfusion and the dashed line is the baseline level. f Quantification of fEPSP facilitation at two different time intervals after the start of the forskolin application. Changes in fEPSP slope are shown as percentage of the mean baseline fEPSP. 8 slices from 5 mice. All data are presented as means; error bars indicate SEM. Unpaired t-test for both time intervals. 15-30 min: **p=0.005; 105-120 min: **p=0.0025.

KD of C1ql2 in DGN of WT mice impairs SV recruitment and LTP.

a Experimental design to analyze the MFS after AAV-mediated KD of C1ql2 in WT DGN. b Relative C1ql2 mRNA levels in DGN. N=4. All data are presented as means; error bars indicate SEM. Unpaired t-test: ***p=0.0002. c Western blot of mouse hippocampal homogenates. d Immunohistochemistry of C1ql2 (red) and GFP (green) on hippocampal sections. Scale bar: 200 μm. e Average synapse score. N=4. Unpaired t-test. *p=0.025. f Number of docked vesicles per 100 nm AZ profile length. N=3. All data are presented as means; error bars indicate SEM. Points represent the individual examined AZ. Unpaired t-test. *p=0.018. g Representative fEPSP traces showing baselines before HFS (black), fEPSP changes 30-40 min after HFS (cyan) and following the application of 3 µM DCG-IV (red). h Time course of fEPSP slopes. The black arrow indicates HFS and the dashed line the baseline level. i Quantification of fEPSP facilitation at four different time intervals after HFS. Changes in fEPSP slope are shown as percentage of the mean baseline fEPSP. +shNS-EGFP, 6 slices from 6 mice; +shC1ql2-EGFP, 7 slices from 7 mice. All data are presented as means; error bars indicate SEM. Mann-Whitney U-test for each time interval. 10-20 min: **p=0.0012; 20-30 min: **p=0.0023; 30-40 min: **p=0.0023; ns, not significant.

C1ql2-Nrxn3 interaction recruits vGlut1 in vitro.

a Immunocytochemistry of myc-tagged C1ql2, C1ql2.K262E or myc-tag (magenta) expressing HEK293 cells and GFP-Nrxn3α(25b+) (cyan) from contacting hippocampal neurons. Scale bar: 5 μm. b Nrxn3α(25b+) recruitment by differentially transfected HEK293 cells. N=3. All data are presented as means; error bars indicate SEM. One-way ANOVA and Tuckey’s PHC. myc-C1ql2 vs. myc-tag: *p=0.016, and vs. myc-K262E: *p=0.022; ns, not significant. c Immunocytochemistry of myc-tagged C1ql2, C1ql2.K262E or myc-tag (magenta) expressing HEK293 cells and vGlut1 (cyan) from contacting hippocampal neurons. Scale bar: 5 μm. d vGlut1 recruitment by differentially transfected HEK293 cells. N = 3. All data are presented as means; error bars indicate SEM. One-way ANOVA and Tuckey’s PHC. myc-C1ql2 vs. myc-tag: *p=0.04, and vs. myc-K262E: **p=0.007; ns, not significant. e Immunocytochemistry of myc-tagged C1ql2 (magenta) expressing HEK293 cells and vGlut1 (cyan) from contacting control, Nrxn123 KO or Nrxn123 KO with Nrxn3α(25+) rescued hippocampal neurons. Scale bar: 5 μm. f vGlut1 recruitment by HEK293 cells in presence or absence of neuronal Nrxns. N = 3. All data are presented as means; error bars indicate SEM. One-way ANOVA and Tuckey’s PHC. inactive Cre vs. Cre: *p=0.023, and vs. Cre+ Nrxn3α(25+): p=0.21; ns, not significant. g Trimeric structures of C1ql2 (PDB_ID: 4QPY, upper panels) and the variant C1ql2.K262E (lower panels). Residue 262 is the central residue (red, left & middle panels) of a larger area underneath the C1ql2-specific calcium and receptor binding loops (magenta, middle panel). The mutation K262E alters the charge of that surface area negative (yellow-circled area, right panels) and makes it potentially repulsive to bind Nrxn3(25b+).

C1ql2-Nrxn3(25b+) interaction is important for C1ql2 localization at the MFS and SV recruitment.

a Experimental design to analyze the MFS after AAV-mediated expression of C1ql2.K262E in Bcl11b cKO DGN. b Western blot and c relative C1ql2.K262E protein levels in mouse hippocampal homogenates. N=3. All data are presented as means; error bars indicate SEM. Mann-Whitney U-test. ns, not significant. d Immunohistochemistry of C1ql2 (red) and GFP (green) in hippocampal sections. Scale bar: 200 μm. Upper panels depict close-ups of C1ql2 staining from the SL of CA3. Scale bar: 15 μm. e Integrated density of C1ql2 fluorescence in the SL of CA3. N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. *p=0.008. f Electron microscope images of MFS and proximal SV. White bars mark synapse length from postsynaptic side. Scale bar: 100 nm. g Average synapse score. Control+EGFP, Bcl11b cKO+EGFP-2A-K262E: N=3; Bcl11b cKO+EGFP: N=4. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: ***p=0.0004, and vs. Bcl11b cKO+EGFP-2A-K262E: ***p=0.0002; ns, not significant. h Number of docked vesicles per 100 nm AZ profile length. Control+EGFP, Bcl11b cKO+EGFP-2A-K262E: N=3; Bcl11b cKO+EGFP: N=4. All data are presented as means; error bars indicate SEM. Points represent the individual examined AZ. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: *p=0.0434, and vs. Bcl11b cKO+EGFP-2A-K262E: *p=0.0196; ns, not significant. Data for Control+EGFP and Bcl11b cKO+EGFP-2A-K262E from f-h in this figure are compared with Bcl11b cKO+EGFP data from Figure 2.

Nrxn KO perturbs C1ql2 localization at the MFS and SV recruitment.

a Experimental design to analyze the MFS after AAV-mediated Nrxn KO. b Relative Nrxn3 mRNA levels. N=4. All data are presented as means; error bars indicate SEM. Unpaired t-test. **p=0.007. c Immunohistochemistry of C1ql2 (red) and GFP (green) in hippocampal sections. Scale bar: 200 μm. Upper panels depict close-ups of C1ql2 staining from the SL of CA3. Scale bar: 15 μm. d Integrated density of C1ql2 fluorescence in the SL of CA3. N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. *p=0.02. e Immunohistochemistry of ZnT3 (red) and GFP (green) in hippocampal sections. Scale bar: 200 μm. Upper right corner of ZnT3 panels depicts close-ups from the SL of CA3. Scale bar: 15 μm. f Electron microscope images of MFS and proximal SVs. White bars mark synapse length from postsynaptic side. Scale bar: 100 nm. g Average synapse score. N=3. Unpaired t-test. **p=0.009. h Number of docked vesicles per 100 nm AZ profile length. N=3. All data are presented as means; error bars indicate SEM. Points represent the individual examined AZ. Unpaired t-test. **p=0.007.

C1ql2 mRNA and protein are lost upon Bcl11b cKO in DGN.

a mRNA in situ hybridization of C1ql2 on hippocampal sections. Scale bar: 200 μm. b Immunohistochemistry of C1ql2 (cyan) and Bcl11b (magenta) on hippocampal sections. Scale bar: 200 μm.

C1ql2 reintroduction in Bcl11b cKO DGN does not rescue MFS number and MFB complexity.

a Immunohistochemistry of C1ql3 (red) and GFP (green) on hippocampal sections. Scale bar: 200 μm. b Relative C1ql3 mRNA levels in DGN. Control+EGFP, Bcl11b cKO+EGFP, Bcl11b cKO+EGFP-2A-C1ql2, N=4; Bcl11b cKO+EGFP-2A-C1ql3, N=3. All data are presented as means; error bars indicate SEM. Two-way ANOVA. ns, not significant. c Active zone length. Control+EGFP, Bcl11b cKO+EGFP-2A-C1ql3, N=3; Bcl11b cKO+EGFP, Bcl11b cKO+EGFP-2A-C1ql2, N=4. All data are presented as means; error bars indicate SEM. Two-way ANOVA. ns, not significant. d vGlut1 and Homer1 double positive puncta in selected CA3 SL ROIs. N=3. All data are presented as means; error bars indicate SEM. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: *p=0.047, and vs. Bcl11b cKO+EGFP-2A-C1ql2: *p=0.029; ns, not significant. e Electron microscopy images of MFBs (purple) and contacting postsynaptic spines (yellow). Scale bar: 500 nm. f MFB perimeter-to-area ratio. N=5. All data are presented as means; error bars indicate SEM. Two-way ANOVA and Tuckey’s PHC. Control+EGFP vs. Bcl11b cKO+EGFP: *p=0.035, and vs. Bcl11b cKO+EGFP-2A-C1ql2: **p=0.0014; ns, not significant.

Bcl11b cKO and C1ql2 KD in DGN do not affect basal synaptic transmission.

a Representative fEPSP traces recorded in slices from a control and Bcl11b cKO animal showing responses to a 20 V electrical stimulation (black, black arrowheads indicate stimulation). The signal is almost entirely blocked by the application of 3 µM DCG-IV (red). b Input-output curves generated by plotting fEPSP slope against fiber volley amplitude at increasing stimulation intensities. Control+EGFP, 35 slices from 16 mice; Bcl11b cKO+EGFP, 32 slices from 14 mice. The data are presented as means, error bars represent SEM. c Representative fEPSP traces recorded in slices from a control and C1ql2 KD animal, showing responses to a 20 V electrical stimulation (black, black arrowheads indicate stimulation). The signal is almost entirely blocked by the application of 3 µM DCG-IV (red). d Input-output curves generated by plotting fEPSP slope against fiber volley amplitude at increasing stimulation intensities. +shNS-EGFP, 16 slices from 9 mice; +shC1ql2-EGFP, 12 slices from 8 mice. The data are presented as means, error bars represent SEM.

C1ql2 KD in DGN of WT mice impairs SV recruitment.

a Immunohistochemistry of C1ql3 (red) and GFP (green) on hippocampal sections. Scale bar: 200μm. b Relative C1ql3 mRNA levels in DGN. N=4. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant. c Relative frequency of synapse scores (refer to Figure 4e). d Cumulative and e relative frequency of the number of docked vesicles per 100 nm AZ profile length (refer to Figure 4f). f Active zone length. N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant. g Diameter of the docked vesicles. N=3. All data are presented as means; error bars indicate SEM. Points represent the individual examined SV. Unpaired t-test. ns, not significant. h Immunohistochemistry of ZnT3 (red) and GFP (green) on hippocampal sections. Scale bar: 200μm. i Number of MFB from selected ROIs in SL of CA3. +shNS-EGFP: N=4; +shC1ql2-EGFP: N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant.

C1ql2-Nrxn3(25b+) interaction is important for SV recruitment at the MFS.

a Relative Nrxn3 mRNA levels in DGN. N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant. b Immunohistochemistry for C1ql3 (red) and GFP (green) on hippocampal sections. Scale bar: 200 μm. c Relative frequency of synapse scores (refer to Figure 6g). d Active zone length. Control+EGFP, Bcl11b cKO+EGFP-2A-K262E: N=3; Bcl11b cKO+EGFP: N=4. All data are presented as means; error bars indicate SEM. Two-way ANOVA. ns, not significant. e Diameter of the docked vesicles. Control+EGFP, Bcl11b cKO+EGFP-2A-K262E: N=3; Bcl11b cKO+EGFP: N=4. All data are presented as means; error bars indicate SEM. Points represent the individual examined SV. Two-way ANOVA. ns, not significant. Data for Control+EGFP and Bcl11b cKO+EGFP-2A-K262E from b-f in this figure are compared with Bcl11b cKO+EGFP data from Figure 2. f Cumulative and g relative frequency of the number of docked vesicles per 100 nm AZ profile length (refer to Figure 6h). h Representative fEPSP traces showing baselines before HFS (black), fEPSP changes 30-40 min after HFS (cyan) and following the application of 3 µM DCG-IV (red). i Time course of fEPSP slopes. The black arrow indicates HFS and the dashed line is the baseline level. j Quantification of fEPSP facilitation at four different time intervals after HFS. Changes in the fEPSP slope are shown as the percentage of the mean baseline fEPSP. Data for C1ql2.K262E from f-h in this figure are compared with control and Bcl11b cKO data from Figure 3a-c. Control+EGFP, 7 slices from 6 mice; Bcl11b cKO+EGFP, 8 slices from 5 mice; Bcl11b cKO+EGFP-2A-K262E, 5 slices from 4 mice. All data are presented as means; error bars indicate SEM. One-way ANOVA followed by Bonferroni’s PHC for each time interval. 0-10 min: Control+EGFP vs. Bcl11b cKO+EGFP-2A-K262E: *p=0.037; 20-30 min: Control+EGFP vs. Bcl11b cKO+EGFP: *p=0.047; Bcl11b cKO+EGFP vs. Bcl11b cKO+EGFP-2A-K262E: *p=0.042; 30-40 min: Control+EGFP vs. Bcl11b cKO+EGFP: **p=0.009; Bcl11b cKO+EGFP vs. Bcl11b cKO+EGFP-2A-K262E: *p=0.031; ns, not significant. k Western blot analysis with flag-tag antibody upon co-immunoprecipitation of HEK293 cells protein extract expressing GluK2-myc-flag and GFP-C1ql2 (lanes 1,2,3 and 4) or GluK2-myc-flag and GFP-K262E (lanes 5,6,7 and 8). Protein extract was precipitated with magnetic beads coupled with anti-flag antibody (lanes 3 and 7) or anti-C1ql2 antibody (lanes 4 and 8). Anti-IgG antibody was used as negative. Expression of GluK2 was verified using protein lysate as input for the co-immunoprecipitation (lanes 1 and 5).

Nrxn KO perturbs SV recruitment at the MFS.

a Relative Nrxn1 and Nrxn2 mRNA levels in DGN. N=4. All data are presented as means; error bars indicate SEM. Unpaired t-test. *p=0.02; ns, not significant. b Relative C1ql2 mRNA levels in DGN. N=4. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant. c Immunohistochemistry for C1ql3 (red) and GFP (green) on hippocampal sections. Scale bar: 200 μm. d Relative frequency of synapse scores (refer to Figure 7g). e Cumulative and f relative frequency of the number of docked vesicles per 100 nm AZ profile length (refer to Figure 7h). g Active zone length. N=3. All data are presented as means; error bars indicate SEM. Unpaired t-test. ns, not significant. h Diameter of the docked vesicles. N=3. All data are presented as means; error bars indicate SEM. Points represent the individual examined SV. Unpaired t-test. ns, not significant.