Regionally mapped astrocytic responses to cortical and white matter stroke show differential roles in astrocyte-induced vascular remodeling

Mice

Animal procedures were performed in accordance with the US National Institutes of Health Animal Protection Guidelines and the University of California Los Angeles Chancellor’s Animal Research Committee, protocol 2000-159. All mice were housed in a facility with 12hr-12hr light-dark cycle, controlled temperature and humidity, and had ad libitum access to food and water. All studies used male mice. Mice underwent surgery at 3–4 months of age. Ribotag experiments: The mGFAP-Cre 77.6 mouse strain was provided by Dr. Michael Sofroniew⁴⁶ and is available from Jackson Laboratory (strain 024098); this line shows robust and widespread astrocytic recombination^111,112. The Ribotag mouse strain⁴⁵ was obtained through Jackson Laboratory (strain 011029) and were crossed with mGFAP-Cre 77.6 mice to Ribotag homozygosity. All other experiments: C57BL/6J mice were obtained through Jackson Laboratory (strain 000664). Animals were randomized between groups for each experiment.

Cell lines

HEK-293 cells (ATCC #CRL-3216) were used for lentivirus and AAV packaging. SVEC4-10 cells (ATCC #CRL-2181) were used to assess CRISPR Lamc1 DNA editing efficiency in vitro.

Stroke induction

Mice were anesthetized with isoflurane, placed in a stereotaxic frame (Kopf Instruments), and maintained a body temperature of 37.0°C ± 0.5°C throughout surgery via a homeothermic monitoring system. Middle cerebral artery occlusion (MCAO) was performed by electrocoagulation of a distal branch of the MCA, as described³¹. The temporal muscle on the left side of the skull was bisected and a craniotomy was performed over the distal MCA branch. The branch point was then electrocoagulated to permanently occlude the MCA, followed by 15min bilateral clamp of the jugular vein. White matter stroke was induced via local injection of vasoconstrictor N5-(1-iminoethyl)-L-ornithine (L-Nio) as described previously³². L-Nio was resuspended in sterile saline at a final concentration of 27mg/ml; 200nl of L-Nio was injected into subcortical white matter at 3 sites along the anterior-posterior axis. Sites: 1) A/P −0.75, M/L 0.96, D/V −2.25; 2) A/P −1.00, M/L 0.96, D/V −2.20; 3) A/P −1.25, M/L 0.96, D/V −2.15. White matter stroke sections were evaluated immediately after sectioning; animals with clear involvement of non-white matter tissue in the lesion were excluded.

Cortical L-Nio-induced stroke was performed via pulled glass capillary pipette injection of two injections of 400nl of L-Nio (27mg/ml in sterile saline) at two depths in the cortex. The pipette was lowered to the first depth and left in place for 2min before the first injection. After the first injection, the pipette was left in situ for 10min, then lowered to the second depth and the injection was repeated. After the second injection, the pipette was left in situ for 5min before withdrawal. Sites: 1) A/P 0, M/L 2.75, D/V −0.3; 2) A/P 0, M/L 2.75, D/V −0.75.

Tissue processing and immunohistochemistry

Animals were perfused with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) (most experiments) or periodate-lysine-PFA (PLP; Col1a1 experiments: 2% PFA, 0.01M NaIO4, 0.1M lysine, 0.175M sodium phosphate buffer, pH 7.4). Brains were postfixed for 3–4 hrs (single astrocyte morphological reconstructions) or overnight (all other experiments) in 4% PFA (most experiments) or PLP (Col1a1 experiments), cryoprotected 2–3d in 30% sucrose, and frozen in powdered dry ice. Frozen brains were sectioned on a microtome (American Optical Company, model 860) at 100μm thickness (single astrocyte morphological reconstructions) or on a cryostat (Leica CM3050) at 40μm thickness (all other experiments) and stored in a glycerol-based antifreeze solution at −20°C until immunostaining. Sections were immunostained as previously reported³¹; briefly, sections were washed in PBS, underwent antigen retrieval in 10mM sodium citrate pH 6 at 80°C for 30min, blocked in 5% normal donkey serum (NDS) and 0.3% TritonX-100 in PBS for 1 hr; in experiments with mouse primary antibodies, block was supplemented with Mouse-On-Mouse block (Vector Labs, #MKB-2213). Sections were incubated with primary antibody (PBS, 2% NDS, 0.3% TritonX-100) at 4°C for 3–5 days (single astrocyte morphological reconstructions) or 1–2 days (all other experiments). In experiments that included EdU detection, sulfo-azide EdU detection was performed between antigen retrieval and block, as previously described³². Sections were then washed in PBS with 0.3% TritonX-100 and incubated with secondary antibody for 3–4 hrs (single astrocyte morphological reconstructions) or 1–2hrs (all other experiments) (1:1000 dilution; PBS, 2% NDS, 0.3% TritonX-100). Sections were washed, mounted on gelatinized slides, cleared with increasing concentrations of ethanol and xylene, and coverslipped with DPX (Electron Microscopy Services, #13512). Primary antibodies used for immunostaining: rat anti-GFAP (Thermo Fisher Scientific, #13-0300, 1:1000), rabbit anti-Iba1 (Wako, #019-19741, 1:500), goat anti-AldoC (Santa Cruz Biotechnology, #sc-12066, 1:100), rabbit anti-phosphoERK (Cell Signaling Technology, #4370, 1:500), rabbit anti-Aldh1l1 (Abcam, #ab87117, 1:400), goat anti-Lcn2 (R&D Systems, #AF1857, 1:500), mouse anti-Ki67 (BD Biosciences, #556003, 1:100), rat biotinylated anti-Ki67 (Thermo Fisher Scientific, #13-5698-80, 1:1000), rabbit anti-Kir4.1 (Alomone Labs, #APC-035, 1:400), rabbit anti-AQP4 (Millipore, #AB3594, 1:200), guinea pig anti-Glt1 (Frontier Institute, #GLT1-GP, 1:200), rabbit anti-FLAG (Thermo Fisher Scientific, #PA1-984B, 1:400), chicken anti-V5 (Abcam, #ab9113, 1:500), human anti-V5 (Absolute Antibody, #Ab00136-10.0, 1:400), goat anti-Myc (Abcam, #ab19234, 1:500), rat anti-HA (Roche, #11867423, 1:100), sheep anti-Trem2 (R&D Systems, #AF1729, 1:100), rat anti-CD31 (BD Biosciences, #550274, 1:100), rabbit anti-Myc 488 labeled (Biotium, #20436, 1:400), rabbit anti-Myc 543 labeled (Biotium, #20437, 1:400), rabbit anti-FLAG 543 labeled (Biotium, #20433, 1:400), rabbit anti-Col1a1 (Abcam, #ab21286, 1:200), mouse anti-NF160 (Abcam, #ab7794, 1:500), chicken anti-GFAP (Rockland, #200-901-D60, 1:500), rabbit anti-CD3 (Cell Signaling, #78588S, 1:500), rat anti-Foxp3 (Thermo Fisher Scientific, 14-5773-82, 1:100), goat anti-MBP (Neuromics, #GT22104, 1:2000), mouse anti-SMA (Millipore Sigma, #A5228, conjugated to biotin; conjugated, 1:200). Anti-SMA biotin conjugation via biotinylation kit (Biotium, #92266) per the manufacturer’s instructions.

Plasmid construction

Lentivirus and adeno-associated virus (AAV) transfer plasmids were generated using Gibson cloning¹¹³ as previously described³¹. Briefly, target inserts were generated by PCR; backbones were generated by restriction digest. Short sequences, including gDNA sequences and FLAG tag to identify exogenous Lamc1, were generated by synthesis or integrated into primers (Integrated DNA Technologies).

Lentiviral vectors (astrocyte morphology, Lamc1 overexpression): We used the minimal GFAP promoter, GfaABC1D, to drive expression in astrocytes, and included a miRNA-124 targeting cassette to minimize off-target expression in neurons and progenitor cells¹¹⁴. To fully elaborate astrocytic morphology, we expressed membrane-targeted spaghetti monster reporters using an N-terminal Lck tag.

Dual AAV vectors for Lamc1 knockout: one virus drove expression of BhCas12b in astrocytes, taking advantage of the GfaABC1D promoter and the miR124T targeting sequence to achieve astrocytic high specificity^31,114. The second virus contained either a) a U6 promoter driving expression of a negative control guide DNA (gDNA), or b) two U6 promoters, each driving expression of Lamc1-specific gDNA; these cassettes were followed by an astrocytic spaghetti monster-Myc (smMyc)⁴⁴ reporter to identify transduced astrocytes.

Backbone plasmids: AAV, AAV-pmSyn1-EBFP-Cre¹⁰⁴; AAV pmSyn1-EBFP-Cre was a gift from Hongkui Zeng (Addgene plasmid # 51507; http://n2t.net/addgene:51507; RRID:Addgene_51507). Lentivirus, pLenti-MP2¹⁰³; pLenti-MP2 was a gift from Pantelis Tsoulfas (Addgene plasmid # 36097; http://n2t.net/addgene:36097; RRID:Addgene_36097). Inserts: pCAG_smFP V5 (Addgene plasmid # 59758; http://n2t.net/addgene:59758; RRID:Addgene_59758), pCAG_smFP FLAG (Addgene plasmid # 59756; http://n2t.net/addgene:59756; RRID:Addgene_59756), and pCAG_smFP Myc (Addgene plasmid # 59757; http://n2t.net/addgene:59757; RRID:Addgene_59757) were gifts from Loren Looger⁴⁴. BhCas12b_v4¹⁰⁶ and U6¹⁰⁵: pZ149-pcDNA3-BhCas12b_v4 (Addgene plasmid # 122446; http://n2t.net/addgene:122446; RRID:Addgene_122446) and AAV:ITR-U6-sgRNA(Kras)-U6-sgRNA(p53)-U6-sgRNA(Lkb1)-pEFS-Rluc-2A-Cre-shortPA-KrasG12D_HDRdonor-ITR (AAV-KPL) (Addgene plasmid # 60224; http://n2t.net/addgene:60224; RRID:Addgene_60224) were gifts from Feng Zhang. The open reading frame of Mus musculus Lamc1 was purchased (Transomic Technologies, #BC165945). The negative control gDNA was designed based on a sequence not found in the Mus musculus genome¹¹⁵, with very limited predicted off target sequences (CasOFFinder¹¹⁶). The lck tag was included on reporters to ensure membrane insertion and full astrocytic process labeling¹¹⁷. DNA fragments were enzymatically assembled using HiFi DNA Assembly (New England Biolabs). Sequences were confirmed by Sanger sequencing (Laragen Sequencing and Genotyping) and whole-plasmid sequencing (Primordium Labs). Plasmids were transformed in NEB Stable high efficiency E. coli (New England Biolabs).

Virus preparation and injection

Lentivirus was packaged in HEK293 cells using 3^rd generation packaging plasmids. Briefly, HEK293 cells were split at ~50% confluency into 15cm dishes coated with poly-D-lysine (5μg/ml in PBS; coated 1hr RT). Cells were maintained in DMEM media + GlutaMAX, glucose, and pyruvate, and supplemented with heat-inactivated FBS. ~24 hrs after splitting, cells were transiently transfected with PEI¹¹⁸; per 15cm plate, cells received: 27.6μg transfer plasmid, 13.8μg pMDLg/pRRE plasmid, 6.9μg pRSV-REV plasmid, and 6.9μg pMD2.G plasmid. The next day, the media was changed and supplemented with 2mM caffeine. Lentivirus-containing media was collected at 48hr and 72hr post-transfection. Media was centrifuged at 1500 × g for 10min at 4°C, filtered through a 0.45μm SFCA membrane, and stored at 4°C until centrifugation. Lentivirus was concentrated first by centrifugation in a SW 32 Ti rotor (Beckman Coulter) at 20,000 RPM for 2hr 20min at 4°C. The lentiviral pellet was resuspended DPBS and loaded onto a 30% sucrose gradient, then centrifuged in a SW 55 Ti rotor at 32,000 RPM for 2hr 20min at 4°C. The final pellet was resuspended in DPBS, centrifuged at 7K × g for 5min at room temperature, and the supernatant was aliquoted and stored at −80°C. Lentivirus titer was measured with an ELISA-based p24 assay (Lenti-X p24 Rapid Titer Kit) per the manufacturer’s instructions. On the day of injection, aliquots were thawed, centrifuged at 7K × g for 5min at room temperature, and diluted as needed in sterile saline. Virus (single astrocyte morphology and white matter astrocytic endfeet: 1–3 × 10^9 vg/ml; Lamc1 or control overexpression: 1×10^11 vg/ml) was loaded into pulled glass capillary pipettes for injection and stored on ice until use.

Lentiviruses trigger rapid transgene expression after injection and their larger capsid size leads to a smaller transduced area; therefore, the use of lentiviral vectors allowed us to inject virus immediately after stroke, and observe robust expression within several days. Single astrocyte morphology: viruses were injected at different distances from the injury and morphology assessed seven days post-stroke as a function of distance from the infarct border. Lamc1 overexpression: lentiviral constructs expressing FLAG-tagged Lamc1 or a reporter protein (smMyc) were locally delivered to white matter; the relatively small transduction area of lentivirus allowed expression patterns largely restricted to white matter. Single astrocyte morphology, white matter: A/P −0.85, M/L +0.81, D/V −2.3; A/P −1.10, M/L +1.11, D/V −2.17; A/P −1.15, M/L: −0.78, D/V −2.26. Single astrocyte morphology, cortex: 3 ipsilesional and one contralesional sites per mouse between A/P +0.5 to −0.8, M/L ±1.5 to 2.3; all, D/V −0.75. Lentivirus was injected in a range of locations due to the large and variable infarct location of MCAO, to label astrocytes at a range of distances. Lamc1 or control overexpression, white matter: A/P −0.85, M/L +0.81, D/V −2.3; A/P −1.10, M/L +1.11, D/V −2.17. All: 200nl/site.

AAV was packaged in HEK293 cells in the PHP.eB serotype¹¹⁹. Briefly, cells were transiently transfected with PEI, media was collected at 3d post-transfection, and media and cells were collected at 5d post-transfection¹¹⁸. Cells were lysed with salt-activated nuclease incubation. AAV particles were separated on an iodixanol gradient via ultracentrifugation. Iodixanol was removed and viral particles were concentrated via buffer exchange columns using DPBS with 0.001% Pluronic F-68³¹. AAV titer was measured via RT-PCR using primers to amplify a region of Cas12b or hGH.

Given the longer expression course of AAV, two weeks before stroke, Lamc1 or control knockout viruses were injected intracortically into the area that would encompass peri-infarct cortex to allow time for transgene expression. Diluted virus (2×10^12 vg/ml per each virus in sterile saline) was injected intracortically in 2 sites, two weeks prior to stroke/sham surgery (effects of Lamc1 knockout in stroke) or three weeks prior to euthanasia (effects of Lamc1 knockout in uninjured tissue). To visualize astrocytic endfeet in Lamc1 vs control knockout astrocytes, the injected viral cocktail was supplemented with 2×10^9 vg/ml PHP.eB::GfaABC1D-lck-smV5-miR124T. Injection sites: A/P 0, M/L +2.75, D/V −0.75; A/P −0.8, M/L +2.50, D/V −0.75; 400nl/site. To visualize astrocytic endfeet in cortical stroke vs sham astrocytes, mice were retroorbitally injected with PHP.eB::GfaABC1D-lck-smV5-miR124T (1×10^11 vg/mouse) two weeks prior to stroke/sham surgery.

Laser capture microdissection and ribosomal immunoprecipitation

To prevent mRNA degradation during subsequent laser capture microdissection while minimizing crosslinking that would prevent ribosomal immunoprecipitation, we used brief, minimal fixation followed by immediate sucrose perfusion¹²⁰. Mice were euthanized via exposure to terminal levels of isofluorane and perfused with PBS for 1min, 0.5% paraformaldehyde for 5min, and 20% sucrose for 5min; rate of 5ml/min. Brains were immediately extracted, frozen in powdered dry ice, and stored at −80C until sectioning. Brains were sectioned on a cryostat, relevant sections were mounted onto PET-membrane slides, and slides were briefly stored in 100% ethanol until laser capture.

Relevant brain areas were sequentially excised in both stroke models, using a combination of light and fluorescent microscopy to identify the autofluorescent infarct core; matched regions in sham-operated animals were also microdissected. Samples immediately collected in supplemented homogenization buffer¹²¹ and stored on ice until homogenization. Brain regions in which we observed substantial non-astrocytic HA expression (cingulate cortex; secondary somatosensory cortex) were avoided during the laser capture process; therefore, while morphological analysis of single astrocytes included zone 4 cortical astrocytes to the midline, laser capture ended prior to cingulate cortex. Collected zone samples were further diluted to ~1.5% w/v in additional supplemented homogenization buffer, estimated from the volume of tissue collected.

Samples were homogenized for 30sec using a motorized tissue grinder (Fisher, #12-1413-61). After homogenization, ribosomal immunoprecipitation via HA antibody binding and RNA extraction proceeded as described¹²¹.

RNA-sequencing and analysis

RNA-sequencing was carried out using TrueSeq with RiboZero kit (Illumina). Libraries were indexed and sequenced by HiSeq4000 sequencer (Illumina) with 50-bp paired end reads. Quality control (QC) was performed on base qualities and nucleotide composition of sequences, to identify problems in library preparation or sequencing. Reads were aligned to the Mouse reference genome (mm10) using the STAR spliced read aligner (version 2.4.0)¹⁰⁸. Read counts were obtained using HT-seq program¹⁰⁹ using mouse Ensembl gene model (GRCm38) as reference. Average input read counts were 57.5M per sample and average percentage of uniquely aligned reads was 74.2%. Limma-voom, which is not dependent on negative binomial model, was used for the differential expression analysis¹²². Four outlier samples were removed based on normalized z-scores greater than 2.0 standard deviations; one sample was removed due to low uniquely aligned read rate (22.6%). Lowly expressed genes were first filtered based on counts per million reads (CPM), Trimmed Mean of M-values (TMM) method was used for normalization, and Empirical Bayes smoothing of standard errors was applied to the linear model. Moderate T-tests were performed for each individual and F-test for multiple comparisons. FDR<0.1 was used to define differentially regulated genes.

Relative enrichment/de-enrichment of samples for genes specific to different cell types was assessed using the log(2) fold change of IP samples versus input samples in uninjured cortex and white matter, as well as pooled zones 1–3 from cortical and white matter stroke to ensure that the level of astrocyte enrichment in IP samples did not change with injury. Zone-specific IP samples were clustered within each brain region (cortex vs white matter) using the top 250 most differentially expressed genes using pheatmap. Significantly enriched gene sets in zones 1 and 2 of each stroke model were assessed using pre-ranked weighted Gene Set Enrichment Analysis^48,123. Genes in zones 1 and 2 per model were sorted by (log2FC*-log10 pVal) to incorporate both the magnitude of the change and the statistical significance, and compared to all GO gene sets (v2023.2); overlapping sets were visualized using DeepVenn¹²⁴. The top gene sets common to all 4 zones were sorted by highest NES in any zone. To aid in visualization of -logFDR, any gene set with an FDR of 0 was increased to 0.00001.

Weighted Gene Correlation Network Analysis was performed on all zone-specific IP samples (zones 1–5 and sham from both models using filtered gene lists within models to exclude lowly expressed genes). Reactivity level was coded 5 (zone 1) to 0 (sham); this was applied to samples from both models (Reactivity) or region-specific models (Reactivity, ctx and Reactivity, wm), in which samples from the non-coded model were designated N/A. Modules were assigned using a signed network with a minimum module size of 30. Module relationships with reactivity were quantified by Pearson correlation of module eigengene and coded reactivity level; significant correlations = p<0.05 with Bonferroni correction. GO enrichment analysis using PANTHER⁵⁰ (v. 18) was used to assess the potential functional relevance of gene modules (cut-off of p<0.05; Fisher’s exact test.) To assess whether the patterns of magenta module gene changes observed in ischemic stroke astrocytes were common in reactive astrocytes in other conditions, log2 fold change in other datasets was compared to the gene significance of reactivity of each magenta module gene (GS.reactivity, correlation between magenta module gene expression and coded reactivity for both models; p<0.05; 375 genes total). These datapoints for each gene were fit with a linear regression; goodness of fit and whether the slope of the linear regression was significantly non-zero (GraphPad Prism) were used to assess the strength of the relationship between magenta module gene expression in other datasets and ischemic astrocytes. Log2 fold change values for other datasets: spinal cord inury⁴⁷: downloaded from https://astrocyte.rnaseq.sofroniewlab.neurobio.ucla.edu/. LPS-induced inflammation⁵⁴: available on GEO (GSE148610). Stab wound injury⁵³ microarray dataset: robust multi-array averages (log2 RMA, stab wound) were downloaded from GEO (GSE66370) and used to generate log2FC. Ipsilesional MCAO³⁵ microarray dataset: counts were downloaded from GEO (GSE35338). Limma-voom via Galaxy (Galaxy tool ID: toolshed.g2.bx.psu.edu/repos/iuc/limma_voom/limma_voom/3.50.1+galaxy0) was used to convert probe IDs to gene name for both microarray studies and calculate log2FC (ipsilesional MCAO). The probe ID with the highest log2FC per gene was used in downstream comparisons. Aged motor cortex⁵⁵: available in Boisvert et al. supplemental table S2. Alzheimer’s disease, tau and APP/PS1 models⁵⁶: available in Jiwaji et al. supplementary data 2 and 4. Huntington’s disease, Q175 and R6/2 models⁵⁷: logCPM values were downloaded from GEO (GSE124846) and used to calculate log2FC.

Ingenuity Pathway Analysis was used to assess differences in predicted diseases and functions associated with zone 1 in both stroke models, using log2 fold change of zone 1 vs sham in each model (cutoff FDR 0.1). Predicted associated diseases/functions were sorted by the largest difference in activation z-score in cortex vs white matter. Genes within the “development of vasculature” function were sorted by associated cellular compartment, using GONet¹²⁵ to identify genes in the extracellular region. Genes were further sorted by relative association with cortical reactivity vs white matter reactivity by WGCNA reactivity gene significance (GS) (GS.reactivity.ctx – GS.reactivity.wm).

In situ hybridization

In situ hybridization (ISH) was performed using hybridization chain reaction¹²⁶ (Molecular Instruments). Perfused tissue sections were washed 2x in SSC (0.75M NaCl, 0.075M Na citrate), then pre-hybridized in probe hybridization buffer, 30min 37°C. Sections were then incubated in probe hybridization buffer + probes (0.4pmol mixed probes/100μl hybridization buffer) overnight at 37°C. Sections were washed 4×15min in probe wash buffer, 37°C, followed by 3×5min washes in SSC at room temperature. Hairpin amplifiers were prepared by 90sec heat pulse at 95°C, then allowed to cool to room temperature for 30min, protected from light. Sections were incubated in amplification buffer for 30min at room temperature, followed by overnight incubation in amplification buffer + hairpin amplifiers (6pmol/100μl amplification buffer). Sections were washed in SSC: 1×5min, 2×30min; one 30min wash was supplemented with Dapi. Sections were washed with 1×5min PBS, incubated in TrueView (Vector Labs) to minimize background autofluorescence, washed 1×5min in PBS, mounted on gelatinized slides, and coverslipped with Fluoro-Gel mounting medium (Electron Microscopy Services).

Post-ISH immunostaining

Combined ISH/immunostaining for Bcas1⁺ cells (ISH) and MBP (immunostaining) was performed sequentially. Sections were processed for ISH as above, mounted, coverslipped, and imaged. Coverslips were then removed and slides were washed 3×5min in PBS before antigen retrieval (10mM sodium citrate, pH 6, 80°C 30min). Slides were washed 3×5min in PBS and blocked in normal donkey serum (5% NDS, 0.3% TritonX-100 in PBS) for 1hr at room temperature. Slides were incubated in primary antibody overnight (anti-MBP, Neuromics goat 1:2000; 0.3% TritonX-100, 2% NDS, in PBS). The next day, slides were washed 3×5min in PBS, incubated in secondary antibody 1–2 hrs, washed 3×5min PBS, and coverslipped.

Vessel leak

Animals were retroorbitally injected 7d post-stroke with a combined solution of biocytin-TMR (Thermo Fisher Scientific, #T12921; 50μl of 1% solution in saline) and albumin-647 (Thermo Fisher Scientific, #A34785, 20μl of 1% solution in saline) and euthanized 30min later. Brains were postfixed for 24hr in 4% PFA, cryoprotected in 30% sucrose for 2 days, frozen in powdered dry ice, and stored at −80°C until sectioning; samples were protected from light throughout processing. Sections were stained for blood vessels and virus (CD31 and Myc/Flag, as above).

Lamc1 CRISPR knockout efficiency analysis

SVEC4-10 cells, which naturally express relatively high levels of Lamc1, were transiently transfected with expression plasmids containing U6-gDNA (negative control, Lamc1 gDNA1, Lamc1 gDNA2, or Lamc1 gDNA 1 & 2), BhCas12b_v4, and CMV-GFP-Neomycin using TransIT-X2 (Mirus Bio). pLenti CMV GFP Neo (657-2) was a gift from Eric Campeau & Paul Kaufman (Addgene plasmid # 17447; http://n2t.net/addgene:17447; RRID:Addgene_17447)¹⁰⁷. The day after transfection, media was replaced and supplemented with G418 (500μg/ml, Invivogen); this was repeated every 2–3 days to kill untransfected cells. Seven days post-transfection, cells were lysed and DNA was collected for use in the T7 endonuclease I mismatch cleavage assay, per the manufacturer’s instructions (Alt-R Genome Editing Detection Kit, Integrated DNA Technologies). DNA fragments were separated by agarose gel, imaged, and fragment intensity was quantified using ImageJ.

In vivo editing efficiency: sections from animals that received negative control or Lamc1 gDNA + Cas12b AAVs before cortical stroke underwent ISH for Lamc1, Slc1a3 (astrocytes), and Cas12b. Lamc1 puncta were assessed with Imaris (Bitplane), spots function. Lamc1 spots were then classified by location: within Slc1a3⁺ surface (astrocytic) vs external to Slc1a3, within the first 300μm from infarct border in the area of Cas12b expression. The proportion of Lamc1 puncta in Slc1a3 surfaces vs total Lamc1 puncta was calculated. gDNA groups were compared using a two-tailed unpaired Student’s t test; n = 6 mice per condition, 2 sections per animal.

Peanut butter trough task

Peanut butter troughs were designed using TinkerCad and printed using resin, 100% infill. Troughs were designed to be 3mm × 3mm × 8mm, for easy directional pick-up by adult mice. Alternating sides were painted with two colors of nail polish to aid identification of turns on video. Five troughs were filled with peanut butter and put in a clear plexiglass arena (15cm × 15cm × 25cm) with the animal for 10 minutes. Animals received 4 training sessions, then were tested before stroke (baseline) and 1, 2, and 3 weeks post-stroke. Animals were food deprived for 8–10 hours prior to training and testing. Animals were recorded from below via mirror to assess trough handling behavior.

Phenotypic and morphologic analysis

To analyze markers with very low or no expression in astrocytes pre-stroke (EdU, Ki67, Lcn2, pERK), we identified positive astrocytes post-stroke and measured their location in relation to the infarct border. The infarct border was identified as the edge of the astrocytic border using one of three astrocytic markers (Aldh1l1/AldoC/GFAP); this border closely mimics the onset of the area of high autofluorescence delineating the infarct core. Markers were colocalized with pan-astrocytic markers Aldh1l1/Aldoc, or GFAP; while GFAP is absent in many healthy gray matter astrocytes, it is robustly and globally upregulated in reactive astrocytes, allowing its use as a pan-astrocytic marker in the peri-infarct cortex. Positive astrocytes from 3–5 sections per animal were combined, binned in 100μm intervals from the infarct border, and the percentage of cells per bin was plotted as a histogram.

To analyze markers with relative change (GFAP, Kir4.1, Glt1, AQP4), a ratiometric analysis was conducted on ipsilesional fields every 100μm from the infarct border. These analyses spanned the dorsal-ventral thickness of white matter or cortex, respectively, including all cortical layers in the latter. We included tissues sections spaced throughout the anterior-posterior axis of the stroke; no overt differences were found along the anterior-posterior axis. We quantified both changes in intensity and distribution of signal. To quantify the latter, we measured the skewness of the field to assess how asymmetrically the signal was distributed throughout the image¹²⁷. This assessment of signal distribution is complementary to measurements of signal intensity. Mean image intensity and skewness of each field was analyzed using ImageJ. Each field was normalized to five averaged contralesional fields in the same image to account for variation between tissue sections. Contralateral tissue in cortical stroke was designated as the equivalent area of cortex. In the white matter stroke model, the infarct was localized above the ipsilesional ventricle; therefore, we designated the white matter above the contralesional ventricle as the contralesional region. However, given that the white matter is a continuous structure, the relatively short distance between the infarct border and the white matter above the contralesional ventricle, and the proximity of the subgranular zone, we reasoned stroke-induced changes were possible in this area; therefore, we also evaluated changes in the white matter lateral to the contralesional ventricle. Sham ipsilateral tissue was matched to the stroke area. Three sections were averaged per phenotypic marker. All analyses passed normality testing (Kolmogorov-Smirnov, Shapiro-Wilk tests). Stroke vs sham ipsilesional/averaged contralesional fields as a measure of distance from infarct were compared with two-way ANOVAs with Sidak’s multiple comparisons tests.

Single transduced astrocytes were imaged on a confocal microscope (Nikon Instruments C2) at 100x with 0.2μm Z step sizes. Astrocytes were chosen based on the following criteria: no adjacent astrocytes transduced with the same reporter virus; full morphology contained within the section; strong signal-to-noise. The distance from the astrocytic soma to the infarct border was measured. Single astrocytes were optically isolated using Imaris (Bitplane) surfaces module and morphology was reconstructed using filaments module. Mean process diameter and number of branch points in each zone per stroke model passed normality (Shapiro-Wilk testing) and were compared with one-way ANOVAs with Dunnett’s multiple comparisons tests; for Sholl analysis, the area under the curve was computed, then compared with one-way ANOVAs with Dunnett’s multiple comparisons test.

Quantification of newborn blood vessels

Colocalized CD31⁺ EdU⁺ cells were analyzed using Imaris image analysis software (Bitplane) with a combination of automated analysis followed by manual curation by a blinded investigator to ensure accurate identification. Positive cells were further quantified based on vessel diameter as large vessels (>10μm) or capillaries (<10μm)¹²⁸. Cell counts were normalized to the tissue area: region-specific zone, Figure 5B; virally transduced cortex within 300μm of the infarct core (Figure 5G) or white matter (Figure 6D). All experiments: pooled data from 3 sections along the anterior-posterior axis of the stroke per animal. Groups were compared using one-way (viral expression experiments, Figures 5G and 6C) or two-way (stroke model by zone, Figures 5B and S6A) ANOVAs with Sidak’s multiple comparisons tests; all passed normality testing (Shapiro-Wilk test). Vessel identity was analyzed by co-staining for SMA; CD31⁺EdU⁺ cells enveloped by SMA⁺ mural cells were identified as arterioles, while SMA⁻ vessels were identified as venules. The percentage of SMA+ vessels in each viral group passed normality testing and were compared by unpaired two-tailed t-test.

Quantification of vessel diameter per CD31⁺EdU⁺ cell

CD31⁺ EdU⁺ cell spots on Imaris were split into individual objects and the diameter of the vessel at that spot was analyzed by placing small spots on opposing sides of the vessel. Vessel diameter distribution was analyzed by both cumulative histograms and histograms with 2μm bin sizes. Cumulative histograms of viral groups were compared using nonlinear regression of each dataset via comparison of fits extra sum-of-squares F test (GraphPad Prism).

Quantification of fibrotic scar and denervated region

The fibrotic scar (Col1a1) and axons (NF160) in animals with chronic white matter stroke (4 weeks post-stroke), injected with control or Lamc1 overexpression virus at the time of stroke, were analyzed using ImageJ. The Col1a1⁺ and NF160⁻ areas were outlined and measured. n = 5 mice per group; averaged 3 sections per animal chosen along anterior-posterior axis. Lamc1 vs control virus groups passed normality testing (Shapiro-Wilk) and were compared using two-tailed unpaired t-tests.

Quantification of T cells

The number and distribution of T cells and regulatory T cells in the white matter 4 weeks post-stroke was analyzed with Imaris. White matter and the infarct core were manually delineated using autofluorescence. Spots on T cells were automatically assigned and manually curated, and further manually subdivided into CD3⁺Foxp3⁺ Treg cells. n=12 mice per group; pooled data from 3 sections along the anterior-posterior axis of the stroke per animal. Groups were compared using two-way (Lamc1 vs control virus by distance from the infarct) ANOVAs with Sidak’s multiple comparisons tests.

Quantification of Bcas1⁺ cells

ISH and immunostained images were aligned in Imaris using Dapi signal from both imaging sessions. Bcas1⁺ cells in the white matter were manually labeled with spots. The area of low MBP was manually traced and the resulting surface was used for distance transformation to quantify the location of Bcas1⁺ cells in relation to MBP signal. n = 7–8 mice/virus, combined 3 sections/mouse. Virus groups passed normality testing (Shapiro-Wilk test) and were compared using unpaired two-tailed t test.

Quantification of ISH analysis of Lamc1 in astrocytes in L-Nio-induced cortical stroke

Astrocytic Lamc1⁺ puncta were quantified using Imaris. Zone 1 (50μm from infarct border) or matched contralateral cortex was manually delineated with a surface. Slc1a3⁺ territory was automatically identified using surfaces and manually curated. Lamc1⁺ spots within Slc1a3⁺ surfaces were automatically identified. Lamc1⁺ puncta were normalized to Slc1a3 volume in matched contralateral tissue within the same section; density was averaged across two sections per mouse, n=5 mice. Ipsilesional and contralesional Lamc1 density in astrocytic territory passed normality testing (Shapiro-Wilk test) and were compared using paired two-tailed t-test.

Quantification of infarct size, 7d post-stroke

Infarct border was determined by autofluorescence in combination with CD31 and EdU signal and quantified using Imaris surfaces. n=6 animals per virus group; average infarct area per section from 3 sections per animal. Virus groups passed normality testing (Shapiro-Wilk test) and were compared using two-tailed unpaired t-test.

Quantification of vessel leak

Albumin-647 and biocytin-TMR, along with the viral tag (Myc/FLAG) and CD31, were imaged in both ipsilesional and contralesional tissue. Using Imaris, the intravascular signal (CD31) was masked, and the periinfarct tissue was delineated and used to create a surface (cortical stroke, 300μm from infarct border; white matter stroke, area of viral transduction). Remaining mean extravascular signal intensity was normalized to contralateral extravascular signal intensity of the same tissue section in the equivalent region (white matter or cortex). Virus groups within each stroke model were compared using two-way ANOVA, Sidak’s multiple comparisons test. The masked vascular surface was further used to quantify intravascular albumin signal intensity, normalized to contralateral intravascular signal, to assess vessel perfusion. High biocytin levels were observed in the infarct core; the area of this biocytin-heavy core signal was also measured. Virus groups were compared using unpaired t-tests. 5–6 animals per virus, averaged 2 sections per mouse; all groups passed normality testing.

Peanut butter trough behavioral task quantification and analysis

Animals were scored on the first pick-up of each of 5 troughs. Paw preference in picking up troughs was assessed by which paw was first used to grasp the trough: left, right, or both (both paws contacting trough within 2 frames). Degree of preference for the stroke-affected limb (right paw) was scored as two points for use of only the right paw, one point for use of both, and zero points for use of the left paw, normalized to the number of troughs picked up over the testing period (0 to 5). Fine motor movement was assessed by quantifying the number of quarter-turns of the trough the animal completed per pickup on the first pickup per trough divided by the average number of turns completed per trough during baseline testing. n = 12 Lamc1 virus + stroke; 10 control virus + sham; 11 control virus + stroke. Animals were excluded if they picked up fewer than 4 of the 5 troughs over the 10 min session (2 animals in sham condition; 1 in Lamc1 stroke; 2 in control virus stroke) or if the stroke was later found to be off-target (Lamc1, 2 animals; control virus, 3 animals). Stroke groups were compared to sham animals within each time point using two-way ANOVA with Dunnett’s multiple comparisons test.