Astrocytic Neuroligins Are Not Required for Synapse Formation or a Normal Astrocyte Cytoarchitecture

Astrocytes exert multifarious roles in the formation, regulation, and function of synapses in the brain, but the mechanisms involved remain unclear. Interestingly, astrocytes abundantly express neuroligins, postsynaptic adhesion molecules that bind to presynaptic neurexins. A pioneering recent study reported that loss-of-function of neuroligins in astrocytes impairs excitatory synapse formation and astrocyte morphogenesis. This study suggested a crucial synaptic function for astrocytic neuroligins but was puzzling given that constitutive neuroligin deletions do not decrease excitatory synapse numbers. Thus, we here examined the function of astrocytic neuroligins using a rigorous conditional genetic approach with deletion of all major neuroligins (Nlgn1–3) in astrocytes. Our results show that early postnatal deletion of neuroligins from astrocytes has no effect on cortical or hippocampal synapses and does not alter the cytoarchitecture of astrocytes. Thus, astrocytic neuroligins are unlikely to shape synapse formation or astrocyte development but may perform other important functions in astrocytes.


INTRODUCTION
detections of change of <20% difficult. The finding of a ~25% decrease in Nlgn3 levels 208 in cortex after the P1 induction ( Figure 3F) is consistent with this assessment since 209 Nlgn3 is the most abundantly expressed neuroligin isoform in astrocytes (Figure 1). 210 We next quantified the levels of selected synaptic proteins as a function of the astrocytic 211 Nlgn1-3 deletion (Figure 3C, D, G, and H). We analyzed 12 synaptic proteins as well as 212 calbindin as a marker of subsets of inhibitory neurons. No significant changes in any 213 protein analyzed were detected in hippocampus or cortex.  Figure 4A). Because excitatory synapses in the 222 hippocampus are too dense to be individually resolved by confocal imaging, we used 223 the overall staining intensity as a proxy for synapse density. Low magnification (20x) 224 imaging across the layers of the CA1 region of the hippocampus and the molecular 225 layer of the dentate gyrus revealed no effect of the astro-Nlgn123 cKO on vGluT1 or 226 Homer1 staining intensity ( Figure 4B). In order to increase the signal-to-noise ratio and 227 improve our ability to detect small changes in staining intensity, we additionally 228 performed high-magnification (60x) confocal imaging of the S. pyramidale and S. 229 radiatum in the CA1 region ( Figure 4A). Again, vGluT1 and Homer1 staining intensities  Figure 4D). We also detected no changes in the vGluT2 staining intensity at 234 either low or high magnifications ( Figure 4E, F). 235 To assess whether astrocytic neuroligins are required for inhibitory synapse formation, 236 we labeled hippocampal sections with antibodies to the inhibitory presynaptic marker 237 . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint GAD67, the inhibitory postsynaptic marker gephyrin and MAP2 and counterstained the 238 sections with DAPI ( Figure 4G). Again, no changes in gephyrin or GAD67 staining 239 intensity were observed after P1 deletion of astrocytic Nlgn1-3 using low or high    CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  we quantified the overall staining intensity as a proxy for synapse number. High 280 magnification imaging revealed that loss of astrocytic neuroligins had no effect on the 281 staining intensity of vGluT1 or vGluT2, but revealed a small decrease (~15%) in the 282 staining intensity for Homer1 ( Figure 6B & 6D).

283
To assess the effect of deleting astrocytic neuroligins on inhibitory synaptogenesis in 284 L4, sections were co-stained with antibodies to GAD67, gephyrin, and MAP2 and again 285 counterstained with DAPI ( Figure 6E). The staining intensity of GAD67 and gephyrin 286 were unaffected by the astrocytic neuroligin deletion ( Figure 6F). Since inhibitory puncta 287 can be resolved with confocal imaging, we also quantified the density and size of 288 GAD67 and gephyrin puncta ( Figure 6G, H), as well as the density of colocalized 289 GAD67-gephyrin puncta ( Figure 6I). Deletion of astrocytic Nlgn1-3 caused no change in 290 any of these measures. These data suggest that, in our hands, astrocytic neuroligins  Astrocytic neuroligins are not essential for astrocyte morphogenesis 295 It is possible that astrocytic neuroligins could be involved in astrocyte morphogenesis 296 even if they are not contributing to synapse formation. Thus, we asked whether 297 . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint astrocytic neuroligins contribute to the morphogenesis of astrocytes and their complex 298 cytoarchitecture. We first measured the levels of a series of glial proteins in lysates of 299 the hippocampus and cortex of astro-Nlgn123 cKO and control mice that had been 300 injected with tamoxifen at P1 and were analyzed at P35 ( Figure 7A). However, we failed 301 to uncover major changes ( Figure 7B). Next, we immunostained astrocytes in CA1 302 hippocampal sections for glial fibrillary acidic protein (GFAP) that is constitutively 303 expressed in mouse hippocampal astrocytes but did not detect any alterations in GFAP 304 expression in astro-Nlgn123 cKO mice at P35 ( Figure 7C, D).

305
Finally, to directly test the claim that astrocytic neuroligins control astrocyte size 306 (Stogsdill et al. 2017), we applied tamoxifen to astro-Nlgn123 cKO and control mice at 307 P1 and stereotactically injected AAVs expressing membrane-targeted mVenus under 308 control of the GFAP promoter into their hippocampus or primary visual cortex at P21. 309 We then imaged relatively thick sections (100 μm) from the hippocampus or primary  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint cortex (Halassa et al. 2007), hippocampus (CA1 S. radiatum) (Bushong et al. 2001), and 320 cerebellar cortex (Spacek 1985), astrocytes (or their cousins, the Bergmann glia in the 321 cerebellum) occupy non-overlapping territories, effectively "tiling" the neuropil. Within was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint different neuroligin isoforms in astrocytes varied among studies, but all studies 349 concurred in the conclusion that the mRNA levels of the three major neuroligins (Nlgn1- RNAseq studies owing to its low abundance and/or the high GC content of its mRNA).

352
Thus, neuroligins are clearly not neuron-specific, which is consistent with the possibility 353 that neuroligins in astrocytes function to embed astrocytes in tripartite synapses and to 354 enable astrocytes to contribute to the formation and performance of synapses, a 355 hypothesis that was proposed in a pioneering paper by Stogsdill et al. (2017).

356
Motivated by this paper that has since guided all reviews on the subject, we here 357 investigated the function of neuroligins expressed in astrocytes. We aimed to achieve    was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint First, deletion of Nlgn1-3 from astrocytes has no detectable effect on the density of 379 excitatory or inhibitory synapses in the hippocampus or cortex, or on the properties of 380 spontaneous excitatory or inhibitory synaptic transmission in the hippocampus. We

457
In summary, it seems likely that expression of Nlgn1-3 in astrocytes has no direct 458 function in synapse formation or in shaping the cytoarchitecture of astrocytes, but that 459 Nlgn1-3 perform other important roles in astrocytes that remain to be identified. 460 Continual advances in the tools available to access astrocytes, in concert with genetic 461 models that allow temporally-defined manipulation of genes, will be key to discovering 462 these roles that may provide new insights into astrocyte biology. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Mice. Aldh1l1-Cre/ER T2 BAC transgenic mice were purchased from The Jackson  decay kinetics were calculated automatically as 10%-90% and 90%-10% peak 565 amplitude of currents. Cumulative plots were generated based on the first 50 events 566 from each recordings session that was either 5 minutes long or had at least 300 events.

567
Plasmids. To assess astrocyte morphology, the following plasmids were used: pAAV-       was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made     was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint magnification imaging (60X) in the CA1 S. radiatum. 919 Data are means ± SEM with statistical significance determined by unpaired two-tailed t-920 test (n=4, 2 male & 2 female).

939
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint mice injected with tamoxifen at P1 and sacrificed at P35. Protein is quantified using 979 fluorescent secondary antibodies, with protein levels normalized to β-actin and then to 980 control levels (n = 4, 2 male & 2 female).    was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted April 12, 2023. ; https://doi.org/10.1101/2023.04.10.536254 doi: bioRxiv preprint Figure S2: Conditional deletion of astrocytic Nlgn1-3 at P1 does not alter the