GEARBOCS: An Adeno Associated Virus Tool for In Vivo Gene Editing in Astrocytes

Motivation Astrocytes are indispensable for brain development, function, and health. However, molecular tools to study astrocyte biology and function in vivo have been largely limited to genetically modified mice. Here, we developed a CRISPR/Cas9-based gene editing strategy within a single AAV vector that enables efficient genome manipulations in astrocytes. We designed and optimized this easy-to-use viral tool to understand gene expression, protein localization and function in astrocytes in vivo.

• GEARBOCS is a single AAV-based CRISPR tool to target mouse astrocytes in vivo.
• GEARBOCS can be used to knockout, tag, or gene-trap genes of interest in astrocytes in vivo.
• Using GEARBOCS, we confirmed astrocytes express Vamp2 and found that astrocytic Vamp2 is required for maintenance of excitatory and inhibitory synapse numbers in vivo.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 this version posted January 19, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023

GEARBOCS mediated Gene Editing in Mouse
Graphical Abstract (Main Points): Introduction CRISPR-Cas9 (i.e., Clustered regularly interspaced short palindromic repeatsassociated endonuclease 9) mediated gene-editing is widely used to engineer the mammalian genome (Doudna and Charpentier, 2014;Guan et al., 2022;Hsu et al., 2014;Knott and Doudna, 2018). Cas9 cleaves the genome at specific guide RNA (gRNA) target sites generating double-strand breaks. These breaks are repaired by two cell-intrinsic repair mechanisms: non-homologous end-joining (NHEJ) or homology-directed repair (HDR) (Heyer et al., 2010;Mao et al., 2008). Cas9-mediated cleavage events continue until repair induces insertions or deletions in the genome (i.e., indels) precluding gRNA recognition and further cleavage by Cas9. Indels within coding sequences often result in frameshifts and premature stop codons generating gene knockouts. HDR and NHEJ following CRISPR-Cas9-mediated genome cleavage can also be used to introduce engineered sequences into the genome even in postmitotic cells such as neurons (Suzuki and Izpisua Belmonte, 2018;Suzuki et al., 2019). These knock-in strategies utilize inframe donor sequences to insert reporters or tags within the open reading frames of target genes.
For example, HDR-based SLENDR and vSLENDR use in utero electroporation or Adeno Associated Virus (AAV)-based methods to deliver gene-editing machinery into the mitotic radial glial stem cells or postmitotic neurons (Mikuni et al., 2016;Nishiyama et al., 2017).
However, the efficiency of HDR-mediated donor insertion in postmitotic cells is limited (Mikuni et al., 2016). The homology-independent targeted integration (HITI) method offers more effective genome editing and endogenous tagging of proteins of interest in postmitotic cells such as neurons (Suzuki et al., 2016). Homology-independent Universal Genome Engineering (HiUGE) utilizes universal guide/donor pairs to enable rapid alterations of endogenous proteins without the need for designing target-specific donors (Gao et al., 2019). Even though these gene-editing tools have versatile in vitro and in vivo applications, they depend on multiple vectors harboring the essential components required for genome engineering. Moreover, these strategies are developed primarily for neuronal gene manipulation.
In the CNS, non-neuronal glial cells, called astrocytes, play critical roles in neuronal connectivity and health. Astrocytes are characterized by their complex, ramified structures which are arborized into several thin filamentous branches (Halassa et al., 2007;Oberheim et al., 2006). Astrocytes are indispensable for the functioning of the brain. They provide trophic support to neurons, maintain the blood-brain barrier, regulate ion/water homeostasis as well as control synaptic transmission (Heithoff et al., 2021;Lyon and Allen, 2021;Mederos et al., 2018;Olsen et al., 2015). These essential roles of astrocytes in synapse formation, maturation, elimination, and maintenance depend on a complex bidirectional interaction with neurons (Chung et al., 2015;Shan et al., 2021) mediated by secreted and adhesion molecules (Saint-Martin and Goda, 2022;Tan and Eroglu, 2021).
Despite the undeniable importance of astrocytes in brain development and function, our molecular understanding of this glial cell type is still in its infancy. To address this knowledge gap, many recent studies revealed astrocyte-specific gene expression profiles and proteomes in the mammalian brain both in normal and disease conditions Bayraktar et al., 2020;Endo et al., 2022;Farhy-Tselnicker et al., 2021;Rayaprolu et al., 2022;Takano et al., 2020;Zhang et al., 2014;Zhang et al., 2016;Chai et al., 2017;Endo et al., 2022). These -omics approaches provide a rich resource for future discoveries; however, efficient tools for rapid in vivo genome editing of mouse astrocytes are needed to speed up discovery.
Here, we developed an astrocyte-specific CRISPR-based gene-editing tool, Vamp2 (also known as synaptobrevin II) is a component of the soluble NSF attachment protein receptor (SNARE) complex and its functions at neuronal synapses for regulation of neurotransmitter release are well-characterized (Salpietro et al., 2019;Wang et al., 2020;Zhang et al., 2014). However, the expression and functional relevance of Vamp2 and the role of the SNARE complex in astrocytes has been controversial (Slezak et al., 2012;Yamamoto et al., 2012). Using GEARBOCS, we verified the expression of Vamp2 in astrocytes in vivo and tested its roles in controlling synapse numbers in the mouse cortex.

GEARBOCS is designed as a single AAV tool for in vivo genome editing in astrocytes.
To facilitate rapid, astrocyte-specific, in vivo genome editing, we designed a CRISPR tool within a single AAV vector, which we named GEARBOCS. GEARBOCS vector is designed to be used in combination with an AAV capsid such as PHP.eB (Chan et al., 2017) which is delivered into the CNS of mice via retro-orbital injection. For GEARBOCS strategy to work, transgenic mice that express spCas9 in a Cre recombinase-dependent manner are required ( Figure 1A). For all the experiments, we used GEARBOCS AAVs produced with AAV-PHP.eB capsids and retro-orbitally injected them into the Cas9 (loxP-STOP-loxP) or Cas9-EGFP (loxP-STOP-loxP) mice at postnatal day (P) 21 ( Figure 1A). After 3 weeks at P42, brains were harvested and analyzed. For consistency, we focused our analyses to the V1 visual cortex.
The GEARBOCS vector (5743bp) harbors four essential components for in vivo genome editing ( Figure 1B)-1) A U6 expression cassette, including a human U6 promoter, a gRNA cloning site, and a 76bp gRNA scaffold (Ran et al., 2013). U6 promoter drives the expression of a small guide RNA (gRNA), which is cloned into the Sap1 restriction site within the gRNA cloning site ( Figure 1B). The gRNA is designed to match a 20bp genomic target site, which is next to a Protospacer Adjacent Motif (PAM) ( Figure   1C). 2) A Donor Insertion Site (DIS) wherein the donor DNA can be cloned between the BamH1 and Sal1 restriction sites ( Figure 1B). The donor DNA is omitted for generating gene KOs. Alternatively, donor sequences, ranging from a small epitope tag, like HA or V5, or a larger reporter tag, such as EGFP or mCherry (up to 2kb), can be cloned into the DIS. The donor sequence is flanked with the guide RNA target site sequences on both ends ( Figure 1D). 3) A synthetic human glial fibrillary acidic protein (GFAP) promoter (Lee et al., 2008), gfaABC1D, to drive the expression of Cre in astrocytes. Cre removes the STOP codon and turns on Cas9 or Cas9-EGFP expression in astrocytes. 4) Two inverted terminal repeats (ITRs) for AAV packaging. The ITR sequences in GEARBOCS are based on the AAV2 serotype (Rabinowitz et al., 2002).
The GEARBOCS approach uses the intrinsic DNA repair mechanism, NHEJ. Thus, when combined with HITI of donor sequences (Suzuki et al., 2016) GEARBOCS can also be used for knocking in a donor sequence in frame with the gene of interest ( Figure 1F). To do so, a donor sequence, flanked with the reverseoriented genomic gRNA target sites on both ends, is cloned into the DIS of the GEARBOCS vector ( Figure 1B and D). The gRNAs guide spCas9s to the genomic loci and the donor sequence within the viral genome to generate double-stranded breaks in both sequences. These cleavage events result in two blunt ends in the genome and two at the donor fragment ( Figure 1F). The donor DNA is then integrated into the genomic target site via HITI (Suzuki et al., 2016).
GEARBOCS with donor can be used for endogenous tagging of proteins of interest (TagIN) in astrocytes. To do so, the donor is designed to insert a protein coding sequence in frame with the endogenous coding sequence so that the endogenous start and stop codons and the polyA tails are kept intact ( Figure 1F).
When the donor sequence is designed to insert in frame following the endogenous START site but has its own STOP codon and a synthetic polyA tail, this will lead to a gene trap with a subsequent knockout ( Figure 1F). This GeneTRAP approach is particularly useful to label and visualize astrocytes in which a gene of interest is knocked out. Reverse orientation of gRNA target sequences flanking the donor ( Figure 1D) is critical for avoiding further cleavage, thereby facilitating the insertion of donor in correct orientation . The GEARBOCS vector design enables us to interchange the donors quickly with a single cloning step. Hence, it provides an "all-in-one" simple and powerful CRISPR tool to target astrocytes in vivo.

GEARBOCS enables efficient astrocyte-specific genome editing.
GEARBOCS uses the astrocyte-specific minimal promoter, gfaABC1D (Lee et al., 2008) to drive the Cre expression and turn on the production of Cas9 or Cas9-EGFP in astrocytes of the commercially available Cas9 mice (Chiou et al., 2015;Platt et al., 2014).
This approach overcomes the difficulty of expressing large-sized Cas9 orthologs through the AAV system.
Previous studies have raised the caveat that human gfaABC1D promoter may permit Cre-mediated genome recombination in some neuronal cells, when AAVs are directly injected at high titers into the mouse brain (O'Carroll et al., 2020). Thus, we first tested the astrocyte-specificity of GEARBOCS in the mouse cortex for its in vivo applications. To do so, we used the AAV capsid PHP.eB to package the GEARBOCS vector. This capsid efficiently transduces CNS cells (Chan et al., 2017) when retroorbitally injected, eliminating the need for invasive surgery and direct injection into the CNS. We found that retro-orbitally injected AAV-GEARBOCS indeed transduces the mouse brain with high efficiency which is evidenced by the abundance of Cas9-EGFP positive cells across many brain regions (Figure 2A-B) including V1 cortex ( Figure 2C) To investigate the astrocyte specificity of Cre-mediated Cas9 expression, we performed the co-immunostaining of Cas9-EGFP with different CNS cell type-specific markers. We used NeuN to label neurons (Gusel'nikova and Korzhevskiy, 2015), Olig2 for oligodendrocyte lineage cells (Yokoo et al., 2004) and Sox9 that labels the majority of the astrocytes (Sun et al., 2017) ( Figure 2D-F). The quantitation of marker colocalization with the EGFP/Cas9-expressing (i.e., EGFP/Cas9+) cells in the cortex revealed that they were overwhelmingly Sox9+ (70.14±2.59%) compared to NeuN+ (9.83±1.60%) or Olig2+ cells (3.58±1.88%) ( Figure 2G). Of all the Sox9+ astrocytes, 89.80±3.29% were EGFP/Cas9+, whereas only 3.47±0.98% and 5.86±3.31% percent of NeuN+ or Olig2+ cells were EGFP/Cas9+, respectively ( Figure 2H). These results show that GEARBOCS targets Cas9 expression to cortical mouse astrocytes with considerable specificity and high efficiency.

GEARBOCS-mediated gene editing of Gfap in mouse cortical astrocytes
To evaluate the effectiveness of GEARBOCS as a molecular CRISPR tool for manipulation of astrocyte genome in multiple ways, we first targeted Gfap. Gfap is a class-III cytoskeletal intermediate filament protein that is highly abundant in astrocytes . To knockout Gfap in astrocytes, we first applied the GEARBOCS KO strategy ( Figure 1C). To do so, we selected a unique gRNA targeting the first exon, located 49bp after the start codon ( Figure 3A). The selected gRNA1 was cloned into GEARBOCS to generate the AAV-GEARBOCS-Gfap-KO and it was retro-orbitally injected into P21 Cas9-EGFP (loxP-STOP-loxP) mice. We used AAV-GEARBOCS without a gRNA as our control. Immunostaining showed that the Gfap expression is With gRNA1, we targeted the N-terminal of Gfap protein for TagIN purposes; however, this approach can lead to the KO of the gene in some cells, if the donor integration does not happen at the locus prior to NHEJ. This caveat can be avoided by tagging proteins at their C-terminal. To demonstrate this possibility, we used another gRNA (gRNA2) (Gao et al., 2019) to introduce a hemagglutinin (HA) epitope-tag at the Cterminal of the Gfap protein ( Figure 3L). The donor sequence containing HA-tag and inverted gRNA target sequences were cloned into the GEARBOCS containing gRNA2 to generate AAV-GEARBOCS-Gfap-TagIN-HA. Upon AAV injection, we observed the colocalization of HA immunostaining with endogenous Gfap ( Figure 3M-P). Altogether, these results demonstrate the usefulness of GEARBOCS to tag an astrocytic protein of interest, broadening its potential applications to study the astrocyte proteome.
Finally, we tested the GEARBOCS tool's utility to achieve a GeneTRAP-i.e., the simultaneous KO of an astrocytic gene of interest and insertion of a reporter gene into its locus ( Figure 1D). To do so, we used the gRNA1 targeting the Gfap gene in exon 1 ( Figure 3E-G). We designed and generated the mCherry-CAAX reporter donor with its own STOP codon and polyA tail. This GeneTRAP strategy caused loss of endogenous Gfap expression, while the mCherry-CAAX was produced in these KO cells facilitating their visualization ( Figure 3Q-T). Collectively, these results demonstrate that GEARBOCS is an efficient tool for gene KO, endogenous tagging, or gene trap purposes, greatly facilitating study of astrocyte cell biology in the mouse CNS. Hence, this "all-inone" single AAV CRISPR tool could help address the limitations and challenges in understanding the development and function of astrocytes in vivo.

Mouse cortical astrocytes express Vamp2 in vivo
Astrocytes secrete synapse-modulating proteins, peptides, and neuroactive small molecules (Verkhratsky et al., 2016;Allen and Eroglu, 2017;Baldwin and Eroglu, 2017); however, the mechanisms underlying their release are unknown. Vamp2 is an integral component of the SNARE complex which mediates calcium-dependent vesicular exocytosis (Salpietro et al., 2019;Wang et al., 2020). In response to an increase in cytosolic calcium levels, vesicular Vamp2 forms ternary SNARE complexes with the plasma membrane proteins such as syntaxin and SNAP23 to cause membrane fusion and release of vesicular cargo ( Figure 4A).
While the presence of calcium transients in astrocytes have been well documented (Lia et al., 2021), the expression of Vamp2 in astrocytes and its functions are still controversial. This is primarily due to the overwhelming abundance of neuronal Vamp2 compared to astrocytes and the limited specificity of both chemical (McMahon et al., 1993;Yamamoto et al., 2012) and genetic approaches to target astrocytic Vamp2 (Slezak et al., 2012). Therefore, it has been difficult to discriminate the astrocytic expression of Vamp2 and its functions from its neuronal counterpart.
To capture astrocytic Vamp2 expression by conventional immunohistochemical methods and imaging, we immunostained Vamp2 in brain sections from mice in which cortical astrocytes were transduced with AAV-gfaABC1D-mCherry-CAAX. Co-localization of mCherry+ astrocytes with Vamp2 was detected; however, due to the intense Vamp2 staining in brain tissue and the resolution limit of light microscopy, it was difficult to confirm Vamp2 expression within astrocytes ( Figure 4B-D). Using the Imaris software, we reconstructed the fluorescence signal from an mCherry-filled astrocyte to map the spatial Vamp2 expression in astrocytes ( Figure 4D).
To verify the specificity of observed astrocytic Vamp2 staining with GEARBOCS, we used a Vamp2-targeting gRNA which was previously described (Horvath et al., 2017).
Using the GeneTRAP method, we knocked-in an mCherry-CAAX donor with a STOP codon and polyA tail at its second exon, 22bp after the start codon ( Figure 4E). This strategy allowed us to confirm the endogenous Vamp2 promoter activity driving mCherry-CAAX expression, while simultaneously knocking out Vamp2 expression.
Immunohistochemical analysis of mCherry and Vamp2 in cortical astrocytes transduced with AAV-GEARBOCS-Vamp2-GeneTRAP revealed greatly diminished Vamp2 staining within the mCherry+ astrocytes ( Figure 4F-H). These results show that Vamp2 locus is transcriptionally active in astrocytes further suggesting that astrocytes express Vamp2.
To further corroborate this finding, we applied GEARBOCS TagIN  Astrocytes control the formation and maintenance of synapses through secreted and cell adhesion molecules (Allen and Eroglu, 2017;Saint-Martin and Goda, 2022). The synaptic functions of astrocytes are mediated through their intricate morphologies. In fact, astrocyte morphogenesis coincides with the period of accelerated synaptogenesis in the mouse visual cortex. Importantly, disruption of astrocyte-neuron cell adhesions not only impacts astrocyte morphology but also alters the fine balance between inhibition and excitation in the CNS (Freeman, 2010;Stogsdill et al., 2017). However, whether astrocytic vesicular exocytosis is involved in astrocyte morphology or synaptogenic function remains enigmatic. Because we found Vamp2 to be expressed in mouse cortical astrocytes ( Figure 4), we investigated its role in cortical astrocyte morphology and synaptogenic ability.
To do so, we deployed the GEARBOCS-GeneTRAP method to KO astrocytic Vamp2 with concurrent mCherry-labeling of KO astrocytes. We retro-orbitally injected AAV-GEARBOCS-Vamp2-GeneTRAP ( Figure 5A). To determine if astrocytic Vamp2 has any effects on astrocyte morphology, we analyzed how this genetic manipulation impacted neuropil infiltration by the fine perisynaptic astrocytic processes. These processes interact with synapses and their abundance per unit brain volume is a measure of astrocyte complexity (Baldwin et al., 2021;Stogsdill et al., 2017;Takano et al., 2020).
We did not see any differences in the neuropil infiltration volume (NIV) from Vamp2 KO astrocytes compared to control in layer II/III ( Figure 5B-F) and layer IV ( Figure 5G-K).
This result shows that astrocytic Vamp2 is not required to maintain astrocyte morphology.
Next, we investigated if astrocytic Vamp2 is required to maintain proper synaptic connectivity. To do so, we quantified the numbers of excitatory and inhibitory synapses within the territories of Vamp2-GeneTRAP and control astrocytes. For synapse quantification, we used an immunohistochemical method that takes advantage of the close proximity of pre-and post-synaptic proteins at synaptic junctions (Baldwin et al., 2021;Stogsdill et al., 2017;Takano et al., 2020). Even though these proteins are at different cellular compartments (i.e., axons and dendrites, respectively), they appear to co-localize at synapses due to the resolution limit of light microscopy.
To quantify excitatory cortical synapses, we labeled the brain sections with excitatory presynaptic markers VGluT1 (intracortical) or VGluT2 (thalamocortical) (Kaneko and Fujiyama, 2002) and postsynaptic marker PSD95. The inhibitory synapses were identified as the co-localization of presynaptic VGAT and postsynaptic Gephyrin. Taken together, our results elucidate a previously unknown role for Vamp2 in mouse cortical astrocytes in maintaining the balance between the excitatory and inhibitory synapses. These findings suggest that Vamp2 controls the release of synapse-modifying factors. Our findings also demonstrate that GEARBOCS is a useful genome-editing tool for investigation of astrocyte-mediated complex cellular and molecular mechanisms in CNS development and function.

Discussion
To further our understanding of astrocyte biology in CNS homeostasis and function an efficient molecular tool for rapid in vivo genome-editing in mouse astrocytes is needed.
In this study, we devised a HITI-based, single virus, CRISPR tool, GEARBOCS, to target mouse cortical astrocytes in vivo and demonstrated its multiple applications in manipulating the astrocyte genome. Off-target activity of gRNAs and imprecise insertion of donors pose significant challenges in CRISPR-mediated genome editing (Zhang et al., 2015). Therefore, careful validation of the gRNAs for the genes of interest are critical for experimental success. In this study, we used guides which were previously validated (Gao et al., 2019;Horvath et al., 2017).
Future studies using GEARBOCS with newly developed guides should screen for offtarget activity or precise insertion of tags at endogenous loci (Naeem et al., 2020).
In this study, we also investigated the expression, localization, and function of Vamp2 in astrocytes in vivo using GEARBOCS. Calcium-regulated vesicular exocytosis is a key feature of neuronal synapses and is mediated by the molecular machinery of the SNARE complex including Vamp2 (Salpietro et al., 2019;Wang et al., 2020). Several studies suggested that Vamp2, and other SNARE complex proteins, are involved in secretion of small neuroactive molecules and proteins from astrocytes (Verkhratsky et al., 2016;Araque et al., 2014;Harada et al., 2015). However, it has been difficult and controversial to provide evidence for Vamp2 expression and function in astrocytes (McMahon et al., 1993;Slezak et al., 2012;Yamamoto et al., 2012). In this study, we used GEARBOCS to delineate the expression of Vamp2 in mouse cortical astrocytes in vivo. We were able to detect reporter expression under the control of Vamp2 endogenous promoter via GeneTRAP method and we found vesicle-like HA-tag staining within Vamp2-TagIN astrocytes. These data provide further evidence that Vamp2 is expressed by astrocytes in vivo.

GEARBOCS-mediated endogenous tagging and visualization of proteins such as
Vamp2 circumvents the challenges and limitations of antibody-based immunolabeling of proteins in astrocytes in vivo. Thus, GEARBOCS facilitates the study of essentially any gene of interest in astrocytes, even in the absence of specific antibodies or conditional alleles. Moreover, this CRISPR tool allows for the determination of astrocyte-specific localization and distribution of proteins, even when the protein of interest, such as Vamp2, is a lot more abundant in another CNS cell type, such as neurons. In addition to these advantages, GEARBOCS can be used to replace the strategy of ectopic expression of tagged proteins which may alter their localization and function in astrocytes.
Astrocytes form an integral part of the synapse and control excitatory and inhibitory synaptogenesis through the secretion of synapse-modifying proteins (Chung et al., 2015;Fossati et al., 2020;Lyon and Allen, 2021). Moreover, astrocytes have been proposed to participate in the regulation of neural circuits by secreting neuroactive small molecules such as D-serine and ATP that modify synaptic activity (Harada et al., 2015;Haydon and Nedergaard, 2014). Because Vamp2 is well established to control synaptic neurotransmitter release in neurons (Salpietro et al., 2019), as well as protein secretion in non-neuronal cell types (Mielnicka and Michaluk, 2021), we postulated that loss of Vamp2 will alter the regulation of excitatory and inhibitory synapse numbers locally. In agreement with this possibility, we found that loss of Vamp2 in astrocytes after the end of synaptogenic period (P21) significantly altered synapse numbers. Loss of astrocytic Vamp2 increased intracortical and thalamocortical excitatory synapse numbers within the territories of KO astrocytes, whereas the number of inhibitory synapses were reduced.
These observations indicate that Vamp2-mediated secretion is required for maintaining the excitation/inhibition balance.
Manipulation of astrocytic calcium through opto-and chemo-genetic tools result in strong changes in synapse activity and animal behavior (Hirrlinger and Nimmerjahn, 2022).
Vamp2 controls Ca 2+ -dependent exocytosis in neurons (Schoch et al., 2001;Takamori et al., 2006). Our results suggest that Vamp2 can also mediate astrocytic exocytosis of synapse-modifying molecules. Thus, our finding reveals a rich and unexplored aspect of astrocytic-secretion and provides the rationale for future studies to study Vamp2 function in astrocytes in vivo.
In summary, our results show that GEARBOCS is an effective and multifunctional gene-editing tool that offers a quick and efficient way to investigate astrocyte biology at the cellular and molecular levels in mice. The simple design of GEARBOCS make it a powerful and versatile 'all-in-one' CRISPR tool for both basic and translational research.
Thus, future studies using GEARBOCS will likely provide new insights into the roles of astrocytes in the pathophysiology of neurodevelopmental and neurodegenerative diseases. were randomly assigned to experimental groups for all experiments.

Plasmids and CRISPR guides
To generate GEARBOCS, Cre expression cassette was cloned first into the pZac2.1-GfaABC1D-Lck-GCaMP6f (A gift from Dr. Baljit Khakh; Addgene plasmid #52924) by replacing Lck-GCaMP6f. U6 expression cassette along with the donor insertion sites (DIS) were synthesized as gBlocks (IDT) and cloned upstream of gfaABC1D promoter to generate GEARBOCS. pAAV-gfaABC1D-mCherry-CAAX was generated by cloning mCherry-CAAX into pZac2.1-GfaABC1D-Lck-GCaMP6f by replacing Lck-GCaMP6f. All the gRNAs used in this study were cloned into the Sap1 site of GEARBOCS and the donors were cloned between the Sal1 and BamH1 site in the DIS.

AAV production and purification
Purified AAVs were produced as previously described (Uezu et al., 2016). Briefly, HEK293T cells grown on 150mm dishes were transfected with GEARBOCS plasmid, helper plasmid pAd-DeltaF6 and the serotype plasmid AAV PHP.eB. After three days, cell lysates were prepared with 15mM NaCl, 5mM Tris-HCl, pH 8.5 followed by three repeats of freeze-thaw cycles. The cell lysates were centrifuged for 30min at 4000rpm to collect the supernatant. Benzonase-treated supernatant (50U/ml, 30 min at 37°C) was added to an Optiprep density gradient (15%, 25%, 40% and 60%) for ultracentrifugation at 60,000 rpm for 1.5hr using a Beckman Ti-70 rotor. AAV fraction was collected from the gradient and concentrated along with the multiple washes with DPBS in a 100 kDa filtration unit. AAV titers were quantified by qPCR based on SYBR green technology using primer pair targeting AAV2 ITR (Aurnhammer et al., 2012).
Harvested brains were post-fixed overnight in 4% PFA, cryoprotected in 30% sucrose and the brain blocks were prepared with O.C.T. (TissueTek) to store at −80°C. 25µm thick brain sections were obtained through cryo-sectioning using a Leica CM3050S (Leica, Germany) vibratome and stored in a mixture of TBS and glycerol at −20°C for further freefloat antibody staining procedures.

Immunohistochemistry
For immunohistochemistry, frozen tissue sections were washed three times in 0.2% TBST (0.2% Triton X-100 in 1x TBS). The sections were blocked and permeabilized with a blocking buffer (5% normal goat serum in 0.2% TBST) for 1hr at RT followed by the overnight incubation with the primary antibodies at 4°C. Primary antibodies were

Synaptic Puncta Colocalization Analysis
Confocal images of pre-and postsynaptic puncta were prepared for analysis using ImageJ (https://imagej.nih.gov/ij/) to convert the raw images into RGB type images with the presynaptic marker (VGlut1/VGlut2 or VGAT) in the red channel, the postsynaptic marker in the green channel (PSD95 or Gephyrin), and the astrocyte marker (mCherry) in the blue channel. Images were then Z-projected to make one projection for every 3 images, representing 1 um of depth in our imaging setup. Several of these Z-projections were then used to train a random forest model using the ilastik software for thresholding images of each of the synaptic markers. To restrict the analysis to the territory of a transduced astrocyte, each image was manually cropped. An ImageJ macro for counting colocalized synaptic puncta, SynBot [https://github.com/Eroglu-Lab/Syn_Bot], was then used to count the number of colocalized synaptic puncta in each of the cropped images using the corresponding ilastik models to threshold each image channel. The same ilastik models and parameters were used to analyze the different groups within each experiment. Colocalized synaptic puncta counts were then normalized to the area of the corresponding cell territory. Colocalized synaptic puncta density was then averaged for the images from each animal used in the experiment.

Neuropil infiltration volume analysis
Astrocyte morphology was analyzed by calculating the volume of peri-synaptic astrocyte processes present in the surrounding neuropil. Immunostained astrocytes over-expressing mCherry-CAAX were imaged by confocal microscopy. High magnification (63x plus 2x optical zoom) 15um Z-stack micrographs of individual astrocytes were acquired using the Olympus Fluoview FV3000 confocal microscope by imaging the middle 1/3 of the astrocyte containing the cell soma and surrounding arbor. The images were reconstructed on Imaris Bitplane 9.9.0 software for 3D reconstructions. For every astrocyte analyzed, three 9.45µm x 9.45µm x 5.25µm regions of interest (ROIs, 75 pixels x 75 pixels x 15 pixels) devoid of the soma and large branches were reconstructed using the surface tool in Imaris, similar to (Stogsdill et al., 2017). Astrocyte NIVs were statistically analyzed using Nested t-test.

Statistical Analysis
For data collection, brains from the healthy mice in each experimental group were collected and processed. All data are represented as mean ± standard error of the mean and P < 0.05 were considered to indicate statistical significance.