Downloaded from Pharmgkb (Accessed 11/2018 by Webgestalt 334 Tool)) Databases

bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 Translationally relevant transcriptomic alterations in mouse ischemic cerebral 2 microvessels 3 Keri Callegari 1,*, Sabyasachi Dash1,*, Hiroki Uchida 1,*, Yunkyoung Lee 1, Akira Ito 1, Tuo 4 Zhang2, Jenny Xiang2 and Teresa Sanchez1,3,#

6 1Department of Pathology and Laboratory Medicine, Center for Vascular Biology, Weill Cornell 7 Medicine, New York, NY

8 2Department of Microbiology and Immunology, Weill Cornell Medicine, New York, NY.

9 3Feil Family Brain and Mind Research Institute, Weill Cornell Medicine, New York, NY.

11 * Equal contributions

12 # Correspondence to: Teresa Sanchez, PhD. Department of Pathology and Laboratory 13 Medicine, Vascular Biology Division and Department of Neuroscience, Brain and Mind 14 Research Institute, Weill Cornell Medicine, 1300 York Ave, A607B, New York, NY 10065. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

15 ABSTRACT 16 Increasing evidence implicates cerebral microvascular dysfunction in the pathophysiology of 17 numerous central nervous system pathologies, including stroke. Understanding the molecular 18 alterations in cerebral microvessels in these conditions will provide original opportunities for 19 scientific investigation at the pre-clinical and clinical levels. In this study, we conducted a novel 20 genome-wide transcriptomic analysis of microvessels in a mouse model of transient focal 21 cerebral ischemia. Using a publicly available human ischemic stroke dataset, we identified 22 shared alterations in our microvessel dataset with implications for human pathophysiology. 23 From this unbiased analysis, we report predicted alterations in inter- and intra-cellular signaling, 24 emphasizing perturbations in genes involved in blood brain barrier function, endothelial cell 25 activation and metabolism. Furthermore, our study unveiled previously unreported gene 26 expression changes associated with altered sphingolipid metabolism. Altogether, our results 27 have identified microvessel-specific transcriptomic changes in a number of translationally 28 relevant pathways that support the targeting of these pathways in preclinical studies. The data 29 shared here provide a resource for future investigation of translationally relevant pathways in 30 ischemic stroke. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

31 INTRODUCTION

32 Despite many decades of research, stroke is still a leading cause of mortality and disability 33 worldwide[1, 2]. While stroke research has focused largely on development of neuroprotective 34 agents, none of these drugs unequivocally showed an improvement in clinical outcomes[3, 4]. 35 Thus, there is a need to develop novel therapeutic strategies[5]. There is increasing evidence 36 that cerebral microvascular dysfunction plays a critical role in the exacerbation of neurovascular 37 injury in stroke[6-14]. The cerebrovascular endothelium, in coordination with pericytes [15, 16] 38 and astrocytes[17] plays a critical role in the maintenance of the blood brain barrier (BBB). As 39 the primary barrier between systemic blood supply and the central nervous system, it has great 40 therapeutic potential [18] [19, 20].

41 Stroke induced alterations in BBB integrity and function are important contributors to brain 42 injury related to hypoxia and neuroinflammation [21]. In the acute phase of stroke, pro- 43 inflammatory cytokines (TNF-α, IL-1β) are released and trigger the induction of matrix 44 metalloproteinases 3 and 9 (MMP-3/MMP-9) that degrade the basal lamina and contribute to 45 endothelial activation and BBB permeability[22]. BBB dysfunction exacerbates neurovascular 46 ischemic injury by allowing the entrance of neurotoxic plasma components into the brain 47 parenchyma, increasing intra-cerebral pressure with the risk of brain herniation and vessel 48 compression, further compromising blood flow to the brain [6, 7, 10, 23, 24]. Endothelial 49 activation is marked by dysregulated endothelial function culminating in microvascular 50 thrombosis and tissue damage with progressive cell death. The pathways leading to and 51 resulting from endothelial dysfunction in ischemia require extensive investigation to broaden our 52 understanding of the molecular mechanisms governing disease progression. By expanding our 53 comprehension of the underlying biology at the neurovascular level, we can develop potential 54 therapeutics for clinical intervention. There is an increasingly dire need for novel targeted 55 approaches to prevent BBB disruption in the acute and long-term stroke-related 56 pathophysiology.

57 In order to expand our understanding of how cerebral microvascular dysfunction 58 contributes to ischemic pathology, in the present study, we conducted an unbiased experiment 59 to determine transcriptomic changes in cerebral microvessels after stroke that are relevant to 60 human stroke pathology. We elucidated several signaling pathways relevant to inflammation 61 and metabolic stress in the BBB and identified targetable pathways enriched in microvessel 62 preparations. From this unbiased analysis, we report predicted alterations in genes involved in 63 blood brain barrier dysfunction, endothelial cell activation and metabolism, which are sustained bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

64 in human stroke lesions highlighting the contribution of these pathways to chronic 65 pathophysiology. In addition, given the encapsulating nature of our RNA-sequencing data, we 66 explored transcriptomic changes relevant to sphingolipid metabolism to uncover novel 67 mechanisms for sphingosine-1-phosphate signaling alterations in stroke. Our data provide 68 support for future preclinical studies to explore neurovascular therapies pertinent to human 69 disease.

70 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

71 RESULTS

72 Transcriptomic changes in cerebral microvessels after tMCAO

73 In order to examine gene expression changes at the neurovascular level, microvessels 74 were isolated from the cerebral cortices of mice subjected to transient middle cerebral artery 75 occlusion (tMCAO) or sham surgery (Figure 1A). This protocol was completed as previously 76 described [25]. RNA was isolated from these microvessels and sequenced on an Illumina 77 platform (n=4). Outliers were excluded similarly across all samples and tMCAO samples 78 clustered with similar gene expression patterns (Supplemental Figure 1). Interestingly, a mild 79 subgrouping within sham microvessels was observed (Supplemental Figure 1C-D). General 80 examination of transcript expression alterations revealed 18,491 out of 32,129 genes were 81 altered, of which 6,291 genes were significant (Table 1; p<0.05). From these significant 82 transcripts, 854 genes had a log fold change of greater than 1 while 837 genes had a log fold 83 change of less than -1 (Figure 1B). Other RNA types were also significantly altered and these 84 results are summarized in Table 1.

85 Cerebral microvessels are composed of endothelial cells, pericytes, and astrocytic end 86 foot processes (Figure 1A). In order to examine the extent to which the cellular composition of 87 the microvessels was altered after tMCAO, gene expression markers of these cell types were 88 assessed (Figure 1C). In the RNA-sequencing dataset, the expression levels of these cell- 89 identity markers remained insignificantly changed, whereas markers of neuroinflammation 90 including glial fibrillary acidic protein (Gfap) and e-selectin (Sele) were significantly induced in 91 microvessel preparations after tMCAO (Figure 1C). Similarly, qPCR validation of cell markers 92 for pericytes (platelet derived growth factor receptor-beta (Pdgfr-beta) and CD45), astrocytes 93 (aquaporin 4 (Aqp4)) and endothelial cell markers (zona occludens 1 (ZO-1)/tight junction 94 protein 1 (Tjp1)) resulted in insignificant differences between sham and tMCAO microvessels 95 (Supplemental Figure 2A-D).

96 To identify the functional consequences of gene expression changes after tMCAO, we 97 performed downstream effects (DE) analysis on a subset of the 6,291 significantly altered genes 98 (with a log2 fold change (log2 FC) cut off of < - 0.5 and > 0.5) using Ingenuity Pathway Analysis 99 (IPA). This DE analysis included Disease and Function, Canonical Pathway, and Upstream 100 Regulator analyses that were used to identify predicted alterations in various functional and 101 molecular categories based on gene expression alterations. Disease and Function analysis 102 predicted alterations in various biological processes such as behavior, cell-to-cell signaling and bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

103 interaction, central nervous system development and function, cancer, organismal injury, and 104 neurological diseases (Figure 1D). Canonical Pathway analysis predicted positive pathway 105 activation in eukaryotic translation initiation factor 2 (EIF2) and neuroinflammation signaling 106 (Figure 1E). In contrast, predicted inhibited pathways shared downstream alterations related to 107 G-coupled protein receptor (GPCR) signaling (ex: Gai, G Beta Gamma, and cAMP-mediated 108 signaling) and calcium signaling (ex: CREB, Dopa-DARPP32, and Glutamate Receptor 109 signaling) (Figure 1E). Due to the strength of predicted EIF2 activation indicating a potential halt 110 in protein translation, we also examined the endoplasmic reticulum (ER) stress pathway. ER 111 stress was predicted to be increased with implications for apoptosis (increased transcripts of 112 caspase 3 and 7 (Casp3/7) and tumor necrosis factor (TNF) receptor associated factor 2 113 (Traf2)) along with decreased protein translation (Eif2). Further, we also examined the predicted 114 upstream regulators affected after tMCAO (Figure 1E). The top five predicted upstream 115 regulators were a part of following molecular categories: transcriptional regulators (21.25%), 116 enzymes (13.75%), kinases (8.63%), transmembrane receptors (7.38%), and GPCRs (5.5%).

117 These results highlight the variety of molecular changes in the microvasculature after 118 tMCAO. Neuroinflammation in these microvessel preparations is associated broad changes in 119 ER functioning, calcium, and GPCR signaling that is predictive of both transcriptomic and 120 metabolic changes at the cellular level after tMCAO.

121 Translational relevance of microvessel-associated transcriptome changes

122 As many transcriptomic alterations were observed in microvessel preparations after 123 tMCAO, we next sought to specifically examine how these changes related to human disease. 124 To this end, the WebGestalt tool was used to perform gene set enrichment analysis (GSEA) of a 125 priori defined gene sets related to mouse phenotype and human disease. For mouse phenotypic 126 analysis, GSEA was performed using the Mammalian Phenotype Ontology database from 127 Jackson labs (MP:0000001; Figure 2A). From this analysis, we observed that genes involved in 128 immune response and abnormal lipid levels (ex: TNF super family member 9 (Tnfsf9), C-C motif 129 chemokine ligand 2 (Ccl2), and complement component 2 (C3)) were positively enriched while 130 genes involved in postnatal development and central nervous system transmission were 131 negatively enriched (Supplemental File 1). To further understand the translational relevance of 132 this data, we performed an additional GSEA to identify enriched classes of genes involved in 133 human disease using the GLAD4U database (Figure 2B). As expected, these same genes (ex: 134 Ccl2 and other TNF-family members) were positively enriched with human diseases with a large 135 inflammatory or hypoxic component such as Acute-Phase Response, infections, and aneurysm bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

136 (Supplemental File 1). Of interest, markers of endothelial cell activation including matrix 137 metalloproteinase 9 (Mmp9), Intercellular Adhesion Molecule 1, (Icam1), and TNF family 138 members were also positively enriched in these disease categories. Interestingly, genes 139 involved with more neurological diseases and function were significantly and negatively 140 enriched.

141 Additional analyses were completed using the Disgenet and OMIM databases for human 142 disease ontology and produced comparable results (Supplemental Figure 3). These results 143 highlight the enrichment of inflammatory and neurological disease categories in microvessel- 144 associated gene expression changes after tMCAO. Additionally, these comparisons provide 145 further support for the relevance of our microvessel model in capturing these aspects of stroke 146 pathology.

147 Comparison of mouse microvessels to human lesion samples after ischemia

148 In an effort to understand the specific changes in gene expression relevant to human 149 stroke, we conducted a direct comparison of our microvessel dataset with a publicly-available 150 dataset containing RNA-sequencing data of lesion-site samples from patients post-stroke. 151 Further information on this dataset is summarized in Supplementary Table 1. Human samples in 152 this dataset included tissue from cortical lesion sites of patients with advanced age (67-74 153 years) and a history of non-fatal ischemia within the last 5 years (GSE56267).

154 Initial comparison of these datasets reveal 541 shared genes, accounting for 8.6% of the 155 microvessel dataset and 35.1% of the human dataset (Figure 3A). A heatmap of the Z-scores of 156 the log2FC for these genes is represented in Figure 3B. To further investigate the how these 157 541 genes may have related one another, we performed k-means clustering to divide these 158 genes into 5 distinct clusters (Supplemental Figure 4A and 4B). As could be expected, these 159 distinct clusters were associated with distinct aspects of the shared gene expression patterns 160 between the mouse microvessels and human lesion sites. For example, while clusters 3 and 4 161 both contained genes involved in inflammation, cluster 3 contained general markers of this 162 response (including genes transcribing TNF-family members, NF-kb, and chemokines) and 163 cluster 4 contained 18 vascular disease specific genes involved in inflammation (including heme 164 oxygenase 1 (Hmox1), serpin e family member 1 (Serpine1), pentraxin-related protein 3 (Ptx3), 165 tissue inhibitor of metalloproteinase 1 (Timp1), and CD44) (Figure 3B, Supplemental Figure 4C). 166 Cluster 5 contained negatively enriched genes implicated in neurotransmission and metabolism. 167 Additionally, these clusters corresponded to previously defined diseases with separate clusters bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

168 containing genes enriched for inflammation (cluster 3), central nervous system (cluster 5), and 169 vascular disease (cluster 4) (Supplemental Figure 4D). These results highlight the functional 170 compartmentalization of the gene alterations shared between our mouse microvessel model 171 and human lesion sites.

172 Further, the 541 genes shared between the human lesions and microvessels were 173 expressed in a highly similar manner (Figure 3C; r = 0.76). Functional analysis with of these 174 genes predicted multiple canonical pathways involved in neuroinflammation to be significantly 175 upregulated while g-coupled protein receptor (Protein Kinase A) and calcium dependent 176 signaling processes (Dopa-DARPP) were predicted as significantly suppressed (Figure 3D). As 177 expected, we then see that genes involved in Neuroinflammation (r = 0.86, p < 0.0001, n = 27) 178 and Protein Kinase A (r = 0.87, p < 0.00011, n = 25) were similarly expressed across both 179 datasets (Supplemental File 2). While gene expression in the Dopa-DARPP category as a 180 whole was not correlated (r = 0.30, p = 0.31, n = 13), if we removed the dopamine specific 181 genes to highlight the calcium and g-protein coupled receptor signaling components we saw a 182 dramatic correlation of expression between mouse microvessels and human lesion sites (r = 183 0.98, p = 0.0006, n = 6). These removed genes included voltage-gated potassium channels 184 and DARPP32-specific targets not relevant to microvessel cell types (Supplemental File 2).

185 These findings further highlight the conserved alteration of these inflammation and 186 metabolic pathways in ischemic human lesions and mouse microvessels. Given these results, 187 we interpret that the relevance of these microvessel-associated changes to human 188 pathophysiology may lie largely in the disruption of the BBB.

189 Ischemia-induced endothelial cell activation alters transcripts related BBB function in 190 cortical microvessels

191 In order to more closely examine BBB-related transcriptomic alterations after tMCAO 192 and stroke, we included 34 genes related to endothelial cell activation and BBB function and 193 maintenance (Figure 4A). Expression of these genes was significantly correlated between 194 mouse microvessels and human lesion-sites after stroke (Figure 4B). Molecules involved in 195 maintaining BBB and endothelial cell function are exciting potential targets in neurovascular 196 disease and we observed that several of these genes (including Hmox1, Serpine1, CD44, 197 gardner-rasheed feline sarcoma viral (Fgr), and vitamin D receptor (VDR)) were similarly 198 induced after tMCAO in mice and chronically post-stroke in humans (Figure 4B). Further, 199 endothelial cell activation indicated by robust Sele induction was validated in qPCR (Figure 4C). bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

200 Subsequent validation of some of these translationally-relevant targets revealed novel 201 genes altered in microvessel preparations after tMCAO. Of interest, transcriptional changes in 202 the family of angiopoietins 1/2 (Angpt1/2) and their receptor, tyrosine-protein kinase receptor 203 (TEK), were validated by qPCR. While the transcriptional difference in Angpt1 was insignificant, 204 Angpt2 transcript was significantly induced in the microvessel preparations post-tMCAO (Figure 205 4D). Tek, the gene that transcribes the Tie2 receptor for Angpt1/2, was also down regulated 206 after tMCAO (Figure 4E).

207 In addition, we also analyzed some of critical players of wingless (WNT) signaling in the 208 context of BBB maintenance and function. qPCR-based validation revealed significant down 209 regulation in the mRNA levels of g-protein coupled receptor 124 (Gpr124) and adenomatosis 210 polyposis coli down-regulated 1 (Apcdd1) with an insignificant effect on low density lipoprotein 211 receptor-related protein 1 (Lrp1) mRNA levels in tMCAO microvessels (Figure 4F).

212 Additional analysis of the 541 shared genes between mouse microvessels and human 213 lesion-sites revealed 68 genes that were existing drug targets (Supplemental Figure 5). This list 214 contained several endothelium-associated genes, including Hmox1 and Serpine1, which were 215 robustly shared between the two datasets. The strong induction of these two specific transcripts 216 was validated in microvessels after tMCAO by qPCR (Figure 4G-I). Hmox1 and Serpine1 are of 217 high interest as possible novel and druggable targets for the microvasculature and have had 218 promising experimental and clinical outcomes for stroke (Ref). Thereafter, we sought to validate 219 the enrichment of the selected BBB maintenance and function-associated targets in the 220 microvessel preparations. We performed qPCR quantification of Angpt1/2, Tek, Lrp1, Gpr124, 221 Apcdd1, plasmalemma vesicle-associated protein (Plvap), Serpine 1 and Hmox1 transcripts in 222 the total RNA extracted from whole brain, cortex, and isolated microvessels of naïve mice. 223 These genes were significantly enriched in the microvessels when compared to whole brain and 224 cortex samples suggesting their potential relevance in microvessel-specific functions associated 225 with the maintenance and proper functioning of the BBB (Figure 5A-E).

226 These results support and emphasize the importance and relevance of the 227 neurovasculature to stroke and highlight the potential of targeting microvessel and endothelial 228 cell function in humans. While many of these transcript alterations have been described to play 229 a role in BBB maintenance and function, how these genes are regulated upon injury was still 230 previously unknown. The observed conservation of pathways related to endothelial cell 231 activation and BBB integrity between tMCAO microvessels and human brain samples after bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

232 stroke strongly support the principle that these changes at the neurovasculature level are crucial 233 in sustained pathogenesis after ischemia.

234 Ischemia-associated changes in S1P metabolism and receptor signaling in cerebral 235 microvessels

236 As a novel contributor of endothelial cell activation and BBB function, we decided to 237 examine transcriptomic changes in mediators of sphingolipid metabolism. A succinct summary 238 of this pathway with our qPCR results is represented in Figure 6A and a full heatmap containing 239 125 genes involved in sphingolipid metabolism (GO: 0006665) is available in Supplemental 240 Figure 6A. These 125 genes were expressed in a similar manner between mouse microvessels 241 and human lesion sites (Figure 6B).

242 Specifically related to the synthesis of sphingosine, we conducted qPCR validation of 243 selected targets involved. While there were no significant changes in the serine 244 palmitoyltransferase (SPT) inhibitors, ORMDL sphingolipid biosynthesis regulator 1/2/3 245 (Ormdl1/2/3), after tMCAO there was a modest and significant increase in the SPT genes 246 themselves, serine palmitoyltransferase long chain base subunit 1/2/3 (Sptlc1/2/3) (Figure 6C- 247 D). Further, there were no observed differences in the transcripts for sphingosine kinases 1/2 248 (Sphk1/2) or S1P phosphatase 1 (Sgpp1) after tMCAO (Figure 6E-F). Interestingly, there was a 249 significant elevation of alkaline ceramidase 2 transcript (Acer2) suggesting the potential for an 250 increase in conversion of ceramide to sphingosine (Figure 6G). Additionally, while there was a 251 modest increase in S1P lyase 1 (Sgpl1) transcript and no remarkable differences in 252 sphingomyelin phosphodiesterase (Smpd1/2) transcript expression (Figure 6H-I). To better 253 understand the tissue specificity of S1P metabolism-associated genes we conducted a qPCR- 254 based relative enrichment analysis between the total RNA extracted from whole brain, cortex 255 tissue, and cortical microvessels of naïve mice. The genes Orndl1/3, Sptlc1/2/3, Sphk1/2, 256 Sgpp1, Smpd1/2 and Sgpl1 were significantly enriched in the microvessel preparations when 257 compared to whole brain and cortex samples (Figure 7A-G). Importantly, levels of Acer 2 were 258 significantly and robustly enriched in the microvessel fraction compared to cortex and whole 259 brain (Figure 7E). In addition, the mRNA levels of Ormdl2 seem modestly but insignificantly 260 abundant in the microvessel preparations when compared to whole brain tissue (Figure 7B).

261 Additional qPCR validation of mRNA transcribing S1P’s cognate receptors that mediate 262 sphingolipid signaling was conducted. Significant changes in S1pr1/2/4 were observed in 263 microvessels after tMCAO. Of note, S1pr2 was the most robust and significant receptor bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

264 alteration, in line with previous data from our lab (Supplemental Figure 6D). Supplmental 265 analysis of RNA-sequencing data with differential gene expression data overlaid on the 266 sphingosine receptor signaling pathway from IPA is available in Supplemental Figure 6C. This 267 data is further reflected in a heat map to compare these changes in mouse microvessels and 268 human lesion-site samples (Supplemental Figure 6B).

269 Taken together, we highlight novel transcriptomic alterations in sphingosine metabolism 270 that provide additional support to the investigation of this pathway in stroke. This data suggests 271 the importance of the sphingolipid metabolic axis in the cerebral microvessels that might be 272 essential in the regulation of BBB function. Specifically, Acer2 has emerged as a potential target 273 enriched in the microvessels that warrants further investigation. As a major player in ceramide 274 metabolism, this induction is associated with neuroinflammation-associated endothelial cell 275 activation. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

276 MATERIALS AND METHODS

277 Animals and ischemic model

278 All animal experiments were approved by Weill Cornell Institutional Animal Care and Use 279 Committee. C57/BL6/J (8-10 weeks, male) mice are subjects to transient middle cerebral artery 280 occlusion (tMCAO). During tMCAO surgery, mice were deeply anesthetized with Isoflurane and 281 body temperature was maintained at 37 ˚C with self-regulating heating pad. The flow of middle 282 cerebral artery was temporally occluded with silicon coated intraluminal suture (Doccol, 283 602245PK10Re) for 60 minute and reperfused by withdrawing a suture, as previously described 284 [26]. Animals which unable to walk straight 24 hours after surgery were used for microvessel 285 isolation.

286 Microvessel isolation

287 The cerebral microvessels were isolated as previously described[25]. To minimize cell 288 activation, all procedures were conducted in a cold room. Ipsilateral cortices were homogenized 289 with MCDB131 medium (Thermo Fisher Scientific, 10372019) with 0.5% fatty acid free BSA 290 (Millipore Sigma, 126609). The homogenate was centrifuged at 2000 g for 5 minutes at 4˚C. 291 The pellet was suspended in 15% dextran (molecular weight ~70 kDa, Millipore Sigma, 31390) 292 in PBS and centrifuged at 10000 g for 15 minutes at 4˚C. The pellet was resuspended in 293 MCDB131 with 0.5% fatty acid free BSA and centrifuged at 2000 g for 10 min at 4˚C. The pellet 294 contained the microvessels.

295 RNA extraction from cerebral microvessels

296 Isolated cerebral microvessels were dissolved in appropriate amount of RLT lysis buffer to lyse 297 the multicellular structures. Lysed samples were then loaded onto a column-based shredder 298 (QIAshredder, Qiagen, Germany) and centrifuged (8000g for 2min at room temperature) to 299 eluate the homogenized lysate. Thereafter, the lysate was processed using total RNeasy mini 300 kit (Qiagen, Germany). In brief, the lysate was resuspended in 1:1 volume 70% ethanol and 301 loaded onto extraction column and centrifuged (8000g for 2min) at room temperature. DNase-I 302 digestion was performed on-column for samples with 1U of TURBO-DNase (Invitrogen, 303 ThermoFisher Scientific) and incubated for 15min at room temperature. Then columns were 304 washed with appropriate amount of wash buffer and finally total RNA was eluted using 25-20 µl 305 of nuclease free water.

306 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

307 Quality Check of total RNA extracted from cerebral microvessels

308 To ensure the quality of total RNA preparations from the cerebral microvessels, 200ng of total 309 RNA were electrophoresed in 1% Agarose Gel at 80V for 30min at room temperature. Further, 310 100ng of total RNA preps for each sample were analyzed for A260/280 ratios using UV-visible 311 spectroscopy in a microplate format (Varioskan, ThermoFisher Scientific) and subsequently 312 confirmed with a final analysis using Bioanalyzer quantification (Genomics core facility, Weill 313 Cornell Medicine).

314 RNA-sequencing workflow

315 RNA integrity and other quality control measures were performed, as stated above, along with 316 ribosomal RNA depletion before cDNA library preparation. cDNA libraries were made using the 317 Illumina TruSeq Stranded mRNA Library Prep kit and were sequenced in a single lane with pair- 318 end 101 bps on Illumina HiSeq4000 instrument. Cutadapt was used to trim low quality bases 319 and adapters [27]. STAR was used to align raw sequencing reads to the mouse GRCm38 - 320 mm10 reference genome[28] Raw read counts were calculated using HTseq-count[29]. 321 Differential expression analysis was performed using DEseq2 package [30]. Volcano and bar 322 plots along with pie chart were generated in GraphPad Prism.

323 Gene Set Enrichment and Functional Prediction Analysis

324 The downstream analysis (including Disease and Function, Canonical Pathways, and Upstream 325 Regulator analyses) reported in this study were generated through the use of IPA (QIAGEN 326 Inc., https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis)[31] Significant 327 genes for downstream analysis in IPA were included with a cut-off of -0.5 and 0.5 log2 fold 328 change (p<0.05). GSEA analysis was completed using the WebgestaltR package in R Studio 329 with all significantly altered[32]. Mouse phenotype GSEA was completed using the Mammalian 330 Phenotype Ontology database from Jackson Labs (MP:0000001; Accessed on 11/14/2018 by 331 WebGestalt tool). Human Disease GSEA was completed using DisgeNET (Version 5.0, 332 05/28/2017), Online Mendelian Inheritance in Man (OMIM; https://www.omim.org/), and 333 GLAD4U (Disease terms were downloaded from PharmGKB (Accessed 11/2018 by WebGestalt 334 tool)) databases. Genes associated with individual disease term were inferred using GLAD4U.

335

336 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

337 Dataset comparison

338 Data processing was completed using the R Studio statistical computing environment 339 (http://cran.us.rproject.org). The Vennerable and limma packages in R Studio were used to 340 compare datasets and generate a venn diagram [33]. The pHeatmap package was used to 341 generate the gene expression heatmaps. Linear regression and Pearson’s correlation tests 342 were used to measure associated gene expression between mouse microvessels and human 343 lesion sites. Gene expression heatmaps with named genes were clustered in R with the hclust 344 and kmeans packages and further generated in excel. All clustering was conducted with a seed 345 set at 25.

346 cDNA synthesis and quantitative real time PCR (qPCR)

347 100-200 ng of total RNA was used as template to prepare cDNA for each sample in the 348 presence of Verso-Reverse Transcriptase and Random Hexamer Primers (Thermo Scientific, 349 USA) at 42 °C for 30 min in a final reaction volume of 20µl. Then, the cDNA prep was diluted 5- 350 fold with nuclease free water to obtain a working stock at a final volume of 100 µl. SYBR Green 351 chemistry with Rox dye signal normalization (Quanta bio, MA, USA)-based quantitative PCR 352 was performed to determine the relative expression levels of target gene. qPCR reactions were 353 run either in duplicates or, triplicates on a 96 well plate with 2µl of cDNA as template in a final 354 reaction volume of 10µl per well for each sample. Finally, the plate was loaded on to the ABI- 355 7500 Sequence Detection System PCR machine (Applied Biosystems, USA). Amplification of 356 target genes were performed using the following cycling conditions with initial denaturation at 357 95oC for 10min followed by 30cycles of 95 oC for 10s; 56 oC for 30s (annealing), extension at 72 358 oC for 30s with melt curve analysis at 60 oC for 30s at a linear ramp rate of 0.5 oC/s followed by 359 acquisition at 0.5 oC intervals. First, the expression levels (Ct values) of target gene were 360 normalized to expression levels of Hprt rRNA as ΔCt values. For Sham and Stroke samples, the 361 relative expression levels of target gene were then expressed as ΔΔ Ct values by comparing the 362 ΔCt values of each sham to the average ΔCt of the sham group. Fold change in target gene 363 expression was calculated by comparing the 2-Δ Ct values of the stroke samples with that of 364 sham samples. A detailed list of primers sequences for the target genes used in this study is 365 provided in table 2.

366

367 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

368 Data Availability

369 The human lesion site data used in these analysis is previously published and available in the 370 NIH GEO Omnibus under accession number GSE56267[34].

371 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

372 ACKNOWLEDGEMENTS

373 This work was supported by internal funds provided by the Department of Pathology and 374 Laboratory Medicine, Weill Cornell Medicine and American Heart Association Grant-in-Aid 375 12GRNT12050110, NIH HL094465 and Leducq Foundation grants to TS. AI was partially 376 supported by LeRoche foundation and Tri I TDI.

377 The authors declare no competing financial interests. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

378 FIGURE LEGENDS:

379 Figure 1: Transcriptomic changes in cerebral microvessels after tMCAO. A) Graphical 380 summary of RNA-sequencing workflow and microvessel cell composition. B) Volcano plot of 381 differential gene expression in microvessels after tMCAO. Red dots represent significantly 382 altered genes (p<0.05). C) Table of microvessel component gene changes and 383 neuroinflammation/endothelial activation gene changes. D) Bar plot of Disease and Function 384 analysis generated by IPA. E) Bar plot of Canonical Signaling Pathway analysis generated by 385 IPA. Prediction Z-score is overlaid on the bars with orange representing predicted pathway 386 activation and blue representing predicted inhibition. F) Pie chart of predicted upstream 387 regulators altered after tMCAO generated by IPA.

388 Figure 2: Translational relevance of microvessel-associated transcriptome changes. A) 389 GSEA analysis of significant differentially-expressed genes in microvessels after tMCAO using a 390 mouse phenotype database through WebGestalt. Ratios reflect number of genes in our dataset 391 over the size of the a priori defined dataset. Blue bars represent a positive enrichment with an 392 false discovery rate (FDR) of <0.05 and yellow bars represent a negative enrichment with an 393 FDR of >0.05 but a p-value of <0.05. B) GSEA analysis of significant differentially expressed 394 genes in microvessels after tMCAO using a human disease database (GLAD4U) through 395 WebGestalt. Ratios reflect number of genes in our dataset over the size of the a priori defined 396 dataset. Blue bars represent a positive enrichment with an FDR of <0.05 and orange bars 397 represent a negative enrichment with an FDR of <0.05.

398 Figure 3: Comparison of mouse microvessels to human lesion samples after ischemia. A) 399 Venn diagram of gene expression comparisons between mouse microvessels and human 400 lesion-sites. B) Heatmap of the differential gene expression Z-scores of 541 shared genes 401 ranked by k-means clustering (seed = 25). C) Pearson correlation of transcript expression Z- 402 scores between mouse microvessels and human lesion-sites (r = 0.76; p < 2.2e-16). D) Bar plot 403 of Canonical Signaling Pathway analysis of 541 shared genes generated in IPA. Predicted 404 activation and inhibition is overlaid on the bars with the Z-score printed adjacent.

405 Figure 4: Ischemia-induced endothelial cell activation alters BBB integrity in cortical 406 microvessels. A) Heatmap of genes relevant to endothelial cell activation and BBB integrity 407 and function ordered by hierarchical clustering. B) Pearson correlation of transcripts 408 represented in 4A between RNA-sequencing data of mouse microvessels after tMCAO and 409 human stroke lesion-sites (r = 0.52; p < 0.01). QPCR validation of transcripts involved in BBB bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

410 function including: C) endothelial cell activation, D-E) angiopoietin ligands and receptor, F) WNT 411 signaling, G-I) vascular permeability and endothelial function. (*p<0.05; **p<0.01; ***p<0.001; ns 412 = not significant).

413 Figure 5: Target genes related to BBB function are enriched in cortical microvessels. A- 414 E) qPCR validation of the enrichment of selected BBB target genes in isolated mouse 415 microvessels compared to whole brain and cortex. (*p<0.05; **p<0.01; ***p<0.001); 416 ****p<0.0001; ns = not significant).

417 Figure 6: Ischemia-associated changes in S1P metabolism genes in cerebral 418 microvessels. A) Simplified summary of sphingolipid metabolism with qPCR results included 419 (green: upregulated; blue: no change; red: downregulated). B) Pearson correlation of transcript 420 expression from mouse microvessels and human lesion-sites including genes from sphingolipid 421 metabolic processes (GO: 0006665). qPCR validation of transcripts involved in sphingosine 422 metabolism separated into functional groups of genes for C) sphingolipid biosynthesis 423 regulators (Ormdl1/2/3), D) serine palmitoyltransferases (Sptlc1/2/3), E) sphingosine kinases 424 (Sphk1/2), F) S1P phosphatase (Sgpp1), G) alkaline ceramidase 2 (Acer2), H) 425 sphingomyelinases (Smpd1/2), and I) S1P lyase (Sgpl1). (*p<0.05; ****p<0.0001; ns = not 426 significant).

427 Figure 7: Specific target genes related to S1P metabolism are enriched in cortical 428 microvessels. A-G) qPCR validation of the enrichment of sphingolipid metabolism target genes 429 in isolated mouse microvessels compared to whole brain and cortex. (*p<0.05; **p<0.01; 430 ***p<0.001; ****p<0.0001; ns = not significant).

431 Table 1: Table of general overview of RNA-sequencing dataset from mouse microvessels after 432 tMCAO.

433 Table 2: List of primer sequences used in qPCR experiments. 434

435 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

436 Supplemental Figure 1: A) Cook’s distances of all samples used in differential expression 437 analysis. B) MA plot depicting global view of the relationship between the expression change 438 between sham and tMCAO samples (log2 FC, M) and the average expression strength of the 439 genes (mean of normalized counts, A) with significance threshold overlaid (red = p < 0.05). C) 440 Heatmap of Euclidean sample-to-sample distance matrix across all samples. D) Principal 441 Component Analysis (PCA) of sham and tMCAO (MCO) samples.

442 Supplemental Figure 2: qPCR validation of cell component transcripts in microvessels isolated 443 from sham and tMCAO mice. (**p<0.01; ***p<0.001); ns = not significant).

444 Supplemental Figure 3: A) GSEA analysis of significant differentially expressed genes in 445 microvessels after tMCAO using a human disease database (OMIM) through WebGestalt. B) 446 GSEA analysis of significant differentially expressed genes in microvessels after tMCAO using a 447 human disease database (Disgenet) through WebGestalt. Dark blue bars represent a positive 448 enrichment with an FDR of <0.05, light blue bars represent a positive enrichment with an FDR 449 of >0.05 but a p-value of <0.05, yellow bars represent a negative enrichment with an FDR of 450 >0.05 but a p-value of <0.05, and orange bars represent a negative enrichment with an FDR of 451 <0.05.

452 Supplemental Figure 4: A) Table of clusters generated by k-means clustering with a seed set 453 to 25. B) Heatmap of k-means clusters (k=5, seed = 25). C) GSEA analysis of genes in each 454 cluster using a pathway database (David Kegg). Clusters 2 and 4 resulted in no significant 455 pathway predictions. D) GSEA analysis of genes in each cluster using a human disease 456 database (GLAD4U). Clusters 1 and 2 resulted in no significant disease predictions.

457 Supplemental Figure 5: A) Heatmap of differential gene expression Z-scores of druggable 458 gene targets in mouse microvessels and human lesion-sites identified through IPA.

459 Supplemental Figure 6: A) Heatmap of differential gene expression Z-scores for sphingolipid 460 metabolic processes (GO: 0006665) including mouse microvessels and human lesion-sites. B) 461 Heatmap of differential gene expression Z-scores of sphingosine receptor signaling 462 components. C) Schematic representation of sphingosine receptor signaling generated by IPA 463 and overlaid with differential gene expression in microvessels after tMCAO. D) qPCR validation 464 of sphingosine receptor transcripts. (*p<0.05;***p<0.001; ns = not significant).

465 Supplemental Table 1: General characteristics of mouse microvessel and human lesion site 466 datasets. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

467 Supplemental File 1: Excel file containing clustering data from 541 shared genes between 468 mouse microvessels after tMCAO and human lesion sites after stroke.

469 Supplemental File 2: Excel file containing correlations between mouse microvessels after 470 tMCAO and human lesion sites after stroke and gene lists from IPA canonical pathway 471 categories: Neuroinflammation, Protein Kinase A, and Dopa-DARPP32 signaling. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

472 References

473 1. Lloyd-Jones, D., et al., Heart disease and stroke statistics--2009 update: a report from 474 the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. 475 Circulation, 2009. 119(3): p. 480-6.

476 2. Benjamin, E.J., et al., Heart Disease and Stroke Statistics-2018 Update: A Report From 477 the American Heart Association. Circulation, 2018. 137(12): p. e67-e492.

478 3. Savitz, S.I. and M. Fisher, Future of neuroprotection for acute stroke: in the aftermath of 479 the SAINT trials. Ann Neurol, 2007. 61(5): p. 396-402.

480 4. Fisher, M. and W. Schaebitz, An overview of acute stroke therapy: past, present, and 481 future. Arch Intern Med, 2000. 160(21): p. 3196-206.

482 5. Saver, J., Lutsep, H, Lyden, P, Barsan, W, Becker, K, DeVeber, G, Frank, J, Guluma, K, 483 Haley, C, Hill, M, Jahan, R, Khatri, P, Kwiatkowski, T, Meyer, B, Nogueira, R, Wechler, L, Final 484 Report of the Stroke Progress Review Group. NINDS report, 2011.

485 6. Ames, A., 3rd, et al., Cerebral ischemia. II. The no-reflow phenomenon. Am J Pathol, 486 1968. 52(2): p. 437-53.

487 7. del Zoppo, G.J. and T. Mabuchi, Cerebral microvessel responses to focal ischemia. J 488 Cereb Blood Flow Metab, 2003. 23(8): p. 879-94.

489 8. Saver, J.L., Improving reperfusion therapy for acute ischaemic stroke. J Thromb 490 Haemost, 2011. 9 Suppl 1: p. 333-43.

491 9. Molina, C.A. and J. Alvarez-Sabin, Recanalization and reperfusion therapies for acute 492 ischemic stroke. Cerebrovasc Dis, 2009. 27 Suppl 1: p. 162-7.

493 10. Hamann, G.F., Y. Okada, and G.J. del Zoppo, Hemorrhagic transformation and 494 microvascular integrity during focal cerebral ischemia/reperfusion. J Cereb Blood Flow Metab, 495 1996. 16(6): p. 1373-8. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

496 11. Murthy, S.B., et al., Perihematomal Edema and Functional Outcomes in Intracerebral 497 Hemorrhage. Influence of Hematoma Volume and Location, 2015. 46(11): p. 3088-3092.

498 12. Macdonald, R.L., Delayed neurological deterioration after subarachnoid haemorrhage. 499 Nat Rev Neurol, 2014. 10(1): p. 44-58.

500 13. Pluta, R.M., et al., Cerebral vasospasm following subarachnoid hemorrhage: time for a 501 new world of thought. Neurol Res, 2009. 31(2): p. 151-8.

502 14. Macdonald, R.L., R.M. Pluta, and J.H. Zhang, Cerebral vasospasm after subarachnoid 503 hemorrhage: the emerging revolution. Nat Clin Pract Neurol, 2007. 3(5): p. 256-63.

504 15. Bell, R.D., et al., Apolipoprotein E controls cerebrovascular integrity via cyclophilin A. 505 Nature, 2012. 485(7399): p. 512-6.

506 16. Armulik, A., et al., Pericytes regulate the blood-brain barrier. Nature, 2010. 468(7323): p. 507 557-561.

508 17. Abbott, N.J., L. Ronnback, and E. Hansson, Astrocyte-endothelial interactions at the 509 blood-brain barrier. Nat Rev Neurosci, 2006. 7(1): p. 41-53.

510 18. Faraci, F.M., Vascular protection. Stroke, 2003. 34(2): p. 327-9.

511 19. Aird, W.C., Spatial and temporal dynamics of the endothelium. J Thromb Haemost, 512 2005. 3(7): p. 1392-406.

513 20. Sanchez, T., Sphingosine-1-Phosphate Signaling in Endothelial Disorders. Curr 514 Atheroscler Rep, 2016. 18(6): p. 31.

515 21. Yang, Y. and G.A. Rosenberg, Blood-brain barrier breakdown in acute and chronic 516 cerebrovascular disease. Stroke, 2011. 42(11): p. 3323-8.

517 22. Lakhan, S.E., et al., Matrix metalloproteinases and blood-brain barrier disruption in acute 518 ischemic stroke. Front Neurol, 2013. 4: p. 32. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

519 23. Hu, X., et al., Cerebral Vascular Disease and Neurovascular Injury in Ischemic Stroke. 520 Circ Res, 2017. 120(3): p. 449-471.

521 24. Keep, R.F., et al., Brain endothelial cell junctions after cerebral hemorrhage: Changes, 522 mechanisms and therapeutic targets. J Cereb Blood Flow Metab, 2018: p. 271678X18774666.

523 25. Lee, Y., et al., The isolation and molecular characterization of cerebral microvessels. . 524 Nature Protocols, 2019.

525 26. Kim, G.S., et al., Critical role of sphingosine-1-phosphate receptor-2 in the disruption of 526 cerebrovascular integrity in experimental stroke. Nat Commun, 2015. 6: p. 7893.

527 27. Martin, M., Cutadapt Removes Adapter Sequences From High-Throughput Sequencing 528 Reads. EMB.Net Journal, DOI: https://doi.org/10.14806/ej.17.1.200.

529 28. Dobin, A., et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 2013. 29(1): 530 p. 15-21.

531 29. Anders, S., P.T. Pyl, and W. Huber, HTSeq--a Python framework to work with high- 532 throughput sequencing data. Bioinformatics, 2015. 31(2): p. 166-9.

533 30. Love, M.I., W. Huber, and S. Anders, Moderated estimation of fold change and 534 dispersion for RNA-seq data with DESeq2. Genome Biol, 2014. 15(12): p. 550.

535 31. Kramer, A., et al., Causal analysis approaches in Ingenuity Pathway Analysis. 536 Bioinformatics, 2014. 30(4): p. 523-30.

537 32. Liao, Y., et al., WebGestalt 2019: gene set analysis toolkit with revamped UIs and APIs. 538 Nucleic Acids Res, 2019. 47(W1): p. W199-W205.

539 33. Ritchie, M.E., et al., limma powers differential expression analyses for RNA-sequencing 540 and microarray studies. Nucleic Acids Res, 2015. 43(7): p. e47.

541 34. Huttner, H.B., et al., The age and genomic integrity of neurons after cortical stroke in 542 humans. Nat Neurosci, 2014. 17(6): p. 801-3. bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

543 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was A. not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Cerebral Microvessels Blood Brain Barrier PDGFR-B B.

Basal Lamina Pericyte

TJPs ZO-1 GLUT1 Endothelial SELE Cell GFAP

Mouse Brain Astrocyte End-Feet RNA-sequencing workflow Microvessel RNA rRNA cDNA Data Read Differential RNA extraction quality control depletion library processing Alignment Expression Analysis

C. D. Gene Fold Change p-value TJP1 (ZO-1) 1.186 n.s SLC2A1 (GLUT-1) 1.129 n.s PDGFRB 1.55 n.s AQP4 1.208 n.s ACTA2 (SMA) 1.123 n.s

GFAP 3.651 <0.00001 SELE 3.104 <0.01

E. F.

Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A.

22/27 18/21 4/8 4/8 7/7 12/14 12/14 21/27 11/13 11/13 6/6 1/7 1/7 1/5 1/5 1/5 1/5 6/6 10/12 8/11

20/25 31/42 25/36 22/32 21/32 12/21 17/26 18/26 14/22 13/22 25/32 14/15 31/38 73/130 39/57 52/65 74/108 38/49 75/89 68/84

Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A. B. Mouse Microvessel Human Lesion Site

Mouse Human Microvessel Lesion Site Vascular Disease-Related Vascular

C. D.

Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was A. not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. HMOX1 IL1B MMP9 SERPINE1 ICAM1 EDN1 PTGS2 PLVAP SELE ANGPT2 VCAM1 PTAFR TBXA2R EDN3 TF TBXAS1 PTGS1 CAV1 MFSD2A ANGPT1 VWF EDN2 TJP1 TJP2 TJP3 OCLN APCDD1 AGER PPIA SLC2A1 TEK LRP1 GPR124 MMP2 Microvessels Human Lesion Site

B. C. D. Sele Angpt1 Angpt2 150 *** 4 ns 4 ** 100 3 3 50 10 2 2 8 Fold Change Fold Fold Change Fold 6 Change Fold 1 1 4 E-selectin / Hprt RNA / E-selectin mu Angpt2 / Hprt RNA / Angpt2 mu 2 Hprt RNA / Angpt1 mu 0 0 0 Sham tMCAO Sham tMCAO Sham tMCAO

E. F. Tek Lrp1 Gpr124 Apcdd1 2.0 2.0 2.0 ** 2.0 ** ns ** 1.5 1.5 1.5 1.5

1.0 1.0 1.0 1.0 Fold Change Fold Fold Change Fold Fold Change Fold 0.5 Change Fold 0.5 0.5 0.5 mu Tek / Hprt RNA / Tek mu mu Lrp1 / Hprt RNA Lrp1 mu mu Apcdd1 / Hprt RNA / Apcdd1 mu mu Gpr124 / Hprt RNA / Gpr124 mu 0.0 0.0 0.0 0.0 Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO

G. H. I. Plvap Serpine1 Hmox1 50 ** 50 * 20 *** 40 40 15 30 30 10 20 20 Fold Change Fold Fold Change Fold Fold Change Fold

Plvap / Hprt RNA / Plvap 5 10 10 Hprt RNA / Hmox1 mu Serpine1 / Hprt RNA / Serpine1 mu 0 0 0 Sham tMCAO Sham tMCAO Sham tMCAO Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A. Angpt1 Angpt2 Tek 5 * 4 25 **** 4 * 20 3 3 15 2 2 10

1 Hprt RNA / Tek Angpt2 / Hprt RNA / Angpt2

Angpt1 / Hprt RNA 1 5 Relative Enrichment Relative Relative Enrichment Relative Relative Enrichment Relative 0 0 0

Cortex Cortex Cortex

Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels

B. Lrp1 Gpr124 Apcdd1 4 */** 25 30 **** *** 20 3 20 15 2 10

1 Hprt RNA / 10 Lrp Gpr124 / Hprt RNA / Gpr124 1 5 Hprt RNA Apcdd1/ Relative Enrichment Relative Relative Enrichment Relative Relative Enrichment Relative 0 0 0

Cortex Cortex Cortex Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels

C. D. E. Plvap Serpine1 Hmox1 20 **** 1.5 ***/* 8 **/*** 15 6 1.0

10 Hprt RNA/ 4

0.5 5 2 Plvap / Hprt RNA / Plvap Hmox1 / Hprt RNA / Hmox1 Relative Enrichment Relative Relative Enrichment Relative Serpine1 Relative Enrichment Relative 0 0.0 0

Cortex Cortex Cortex

Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels

Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was A. not certifiedSphingolipid by peer review) Metabolism is the author/funder. All rights reserved. No reuse allowedB. without permission.

C. D. Ormdl1 Ormdl2 Ormdl3 Sptlc1 Sptlc2 Sptlc3 1.5 ns 2.0 2.5 2.5 3 1.5 ns * * * ns 2.0 2.0 1.5 1.0 .0 2 1.5 1.5 1.0 1.0 1.0

0.5 .5 1

Fold Change Fold

Fold Change Fold Fold Change Fold 0.5 Change Fold

Ormdl1/Hprt RNA Ormdl1/Hprt 0.5 0.5

mu Sptlc1/Hprt RNA Sptlc1/Hprt mu

mu Sptlc2/Hprt RNA Sptlc2/Hprt mu

mu Sptlc3/Hprt RNA Sptlc3/Hprt mu

mu Ormdl2/Hprt RNA Ormdl2/Hprt mu mu Ormdl3/Hprt RNA Ormdl3/Hprt mu 0.0 0.0 0.0 0.0 0.0 0 Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO E. F. G. H. Sphk1 Sphk2 Sgpp1 Acer2 Smpd1 Smpd2 3 ns 2.0 2.0 4 1.5 ns 1.5 ns ns **** 1.5 1.5 ns 3 2 1.0 1.0 1.0 1.0 2

1 0.5 0.5

Fold Change Fold Fold Change Fold

Fold Change Fold

Fold Change Fold Fold Change Fold

0.5 0.5 1 Change Fold

mu Acer2/Hprt Acer2/Hprt mu RNA

mu Sphk2/Hprt RNA Sphk2/Hprt mu

mu Sphk1/Hprt RNA Sphk1/Hprt mu

mu Sgpp1/Hprt RNA Sgpp1/Hprt mu

mu Smpd2/Hprt RNA Smpd2/Hprt mu mu Smpd1/Hprt RNA Smpd1/Hprt mu 0 0.0 0.0 0 0.0 0.0 Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO Sham tMCAO

I. Sgpl1 2.0 * 1.5

1.0

Fold Change Fold 0.5 mu Sgpl1/Hprt RNA Sgpl1/Hprt mu 0.0 Sham tMCAO

Figure 6 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was A. Ormdl1 not certified by peer review) isOrmdl the author/funder. 2 All rights reserved. No reuseOrmdl3 allowed without permission. 4 3 4 ***/** ns ** 3 3 2 2 2 1 1 Foldchange 1 Ormdl3/ Hprt RNA Ormdl3/ Ormdl2 / Hprt RNA / Ormdl2 Ormdl1 / Hprt RNA / Ormdl1 Relative Enrichment Relative Relative Enrichment Relative 0 0 0

Cortex Cortex Cortex

Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels

B. Sptlc1 Sptlc2 Spllc3 4 4 6 **** **** * 3 3 4 2 2 2 1 1 Sptlc3 / Hprt RNA Sptlc3 Sptlc2 / Hprt RNA / Sptlc2 Sptlc1 / Hprt RNA / Sptlc1 Relative Enrichment Relative Relative Enrichment Relative Relative Enrichment Relative 0 0 0

Cortex Cortex Cortex

Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels

Sphk1 Sphk2 Sgpp1 Acer2 C. ns / * D. E. 4 2.0 8 25 **** ** ** 20 3 1.5 6 15 2 1.0 4 / Hprt RNA/ 10 Foldchange Foldchange Foldchange 1 0.5 2 Acer2 Sgpp1/ Hprt RNA Sgpp1/ Sphk2/ Hprt RNA Sphk2/ Sphk1/ Hprt RNA Sphk1/ 5 Relative Enrichment Relative 0 0.0 0 0

Cortex Cortex Cortex Cortex

Whole brain Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels Microvessels

F. Smpd1 Smpd2 G. Sgpl1 5 **** 8 5 *** ****/*** 4 4 6 3 3 4 2 2 Foldchange Foldchange

2 Hprt RNA / Sgpl1 Smpd1/ Hprt RNA Smpd1/ 1 Hprt RNA Smpd2/ 1 Relative Enrichment Relative 0 0 0

Cortex Cortex Cortex

Whole brain Whole brain Whole brain Microvessels Microvessels Microvessels Figure 7 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A. B.

C. D.

Supplemental Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ZO-1 mRNA CD45 mRNA A. B. ns (Zona occludens 1) 1.5 1.5 ns

1.0 Hprt RNA

Hprt RNA 1.0

Fold Change 0.5 Fold Change 0.5 mu CD45 / CD45 mu mu ZO-1 / ZO-1 mu

0.0 0.0 Sham Stroke (tMCAO) Sham Stroke (tMCAO)

mu PDGFR-β Aqp4 mRNA (Platelet derived growth factor receptor-β) (Aquaporin 4) C. D. ns 1.4 ns 1.5

1.2 Hprt RNA 1.0 Hprt RNA β / 1.0 Fold Change Fold Change 0.5

mu Aqp4 / Aqp4 mu 0.8 mu PDGFR-

0.6 0.0 Sham Stroke (tMCAO) Sham Stroke (tMCAO)

Supplemental Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A.

Supplemental Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A.

# of Genes Cluster avg MV DE avg human DE KEGG Pathways Disease (GLAD4U) 131 1 0.903362 1.953676 Phagosome, T-cell leukemia virus 1 infect, PI3K-Akt , Cancer None 39 2 -0.8696133 1.913478 None None 100 3 1.6906824 2.423374 NF-KB/Cytokine Receptor Interactions Cancer, Inflammation, Pathology 18 4 3.6557395 2.269004 None Sepsis, inflammation, vascular disease 253 5 -0.7941709 -2.415199 Axon Guidance, Neurotransmission, Metabolism CNS diseases

B. C.

Kegg Pathway 5 Retrograde endocannabinoid signaling Dopaminergic synapse GABAergic synapse Glutamatergic synapse 4 Axon guidance Metabolic pathways Cytokine-cytokine receptor interaction NF-kappa B signaling pathway 2 PI3K-Akt signaling pathway Phagosome Cluster 5 Pathways in cancer Cluster 3 1 Human T-cell leukemia virus 1 infection Cluster 1 0 2 4 Enrichment Ratio 3

D. Disease GLAD4U Autism Spectrum Disorder Autistic Disorder Epilepsy Mental Disorders Schizophrenia Inflammation Necrosis Respiratory Tract Diseases Lung Diseases Chronic Disease Head and Neck Neoplasms Metaplasia Cluster 5 Neovascularization, Pathologic Cluster 4 Pathologic Processes Inflammation Cluster 3 0 2 4 6 8 10 Enrichment Ratio Supplemental Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. A. Mouse Microvessel Mouse Human Lesion Site Mouse Microvessel Mouse Human Lesion Site ANXA1 ATP1A1 BIRC3 CDC7 C1R CDK5 C1S EPHA3 C5AR1 PPP3CB CASP1 PRKCZ CCR5 SMS CDK2 SRD5A1 COL4A2 CACNA1A CSF1 CACNB4 FAS ESRRG FCGR1A GABRB2 FGF2 GABRD FLT4 GAD2 IL6R GLRA2 INHBA GRIA4 IRAK4 KCND2 ITGAL KCNJ11 LTB4R MMP24 LY75 PDP1 MMP14 PPP3CA NOS3 SCN1B PNP SCN2B PTGS2 SLC1A6 PTPN1 SNAP25 PTPN6 SV2A PTPRC TUBA4A RIPK1 XK TNFRSF1A TP53 CCL2 CD44 COL4A1 FGR HMOX1 P2RY2 SERPINE1 TNFRSF12A TNFRSF9 VDR

Supplemental Figure 5 tgla6A3galt2 St6galnac6 St6galnac3 Smpdl3a St6galnac5 Col4a3bp B4galnt1 A. Pla2g15 Alox12b B4galt5 Serinc5 B3galt4 B4galt3 B4galt6 Serinc1 B3galt1 B3galt2 Gal3st1 Ormdl3 Fam57b St8sia4 St8sia1 St8sia3 Symbol Sgms1 Aloxe3 Psapl1 Smpd4 Samd8 Smpd1 Smpd3 Asah2 Degs2 Cers6 Abca2 Cers1 Sgpp2 P2rx7 Cers4 Hacd2 Sgpp1 Hacd3 Elovl6 Elovl2 Elovl4 Cerkl Ugt8a Gm2a Map7 Sirt3 Serinc3 Neu1 Neu2 Neu4 Gba2 Ugcg l8Serinc2 Hexa Cln8 Kdsr Smpdl3b Hexb Fa2h Psap Tecr Galc Cel Kit bioRxiv preprint

9130409I23Rik Microvessels Human Cortex St6galnac4 St6galnac1 Ormdl1 Ormdl2 St8sia2 St8sia6 B3gnt5 Symbol Pla2g6 Pdxdc1 Sgms2 Smpd2 Smpd5 A4galt Sphk2 Sphk1 Sptssb Sptssa Degs1 Sptlc1 Sptlc2 Asah1 Sptlc3 Cers3 Prkcd Cers2 P2rx1 Spns2 Acer1 Acer2 Hacd4 Hacd1 Acer3 Cers5 Elovl5 Elovl7 Elovl3 Enpp7 Elovl1 Sgpl1 Plpp1 Crem Neu3 Ccn1 Cerk Pemt Naga Itgb8 Sftpb Cln6 Arv1 Atg7 doi: Fut7 Agk not certifiedbypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. Bax Gba Gla https://doi.org/10.1101/829820

Microvessels Human Cortex B. ADCY4 ADCY9 ADCY5 ADCY3 ADCY1 ADCY2 ADCY8 SMPD3 PIK3R5 ACER2 CASP8 CASP3 CASP7 CASP1 PTK2B CASP4 GNAI3 GNAI2 PLCH2 GNAI1 RHOA FGFR3 PDIA3 PLCB1 PLCE1 PLCL2 RHOC S1PR2 S1PR3 S1PR1 RHOG S1PR5 AKT1 RHOF RAC1 RND2 RHOJ ;

Microvessels this versionpostedNovember4,2019. Human Corte x D. C. Fold Change Fold Change Fold Change S1PR5 / Hprt RNA muS1PR3 / Hprt RNA muS1PR1 / Hprt RNA 0.0 0.5 1.0 1.5 0 1 2 3 4 0.0 0.5 1.0 1.5 2.0 2.5 hmtMCAO Sham hmtMCAO Sham hmtMCAO Sham S1PR3 S1PR5 S1PR1 ns The copyrightholderforthispreprint(whichwas ns * Supplemental Figure 6

Fold Change Fold Change mu S1PR4 / Hprt RNA mu S1PR2 / Hprt RNA 0.0 0.5 1.0 1.5 10 15 0 5 hmtMCAO Sham hmtMCAO Sham S1PR4 S1PR2 *** * bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Transcriptome Classification Number

Total 18491 (32129) Significant Changes (p <0.05) 6291 mRNA > 1 Log2FC increase (p<0.05) 854 < -1 Log2FC decrease (p<0.05) 837 Total 13638 (32129) Significant Changes (p <0.05) 554 lncRNA 151 Noncoding RNA Antisense RNA 76 miRNA 14

Others (Pseudogenes, SnoRNA, ScaRNA, etc)

Table 1 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Table 2. List of primer sequences used in qPCR experiments.

Target genes qPCR primer (5’ 3’ orientation) Ref Seq ID (Mus musculus) E-selectin FP-CCGTCCCTTGGTAGTTGCA NM_011345.2 RP-CAAGTAGAGCAATGAGGACGATGT Angpt1 FP- TGCTCTGAAGAGTTGACACAGG XM_006520323.2 RP- ATGGTGGCCGTGTGGTTTTG Angpt2 FP- AGGAAGACTGCTGTTAATGACG NM_007426.4 RP- GGCTTTGTAGATGTCCCAGGT Tek FP- AGGAGAACTGGAGGTTCTTTGTA NM_013690.3 RP- GGAGGACAGTGTGGAAGCTGTA Lrp1 FP- ACACGAGTTCCTGTCACACC NM_008512.2 RP- TCTGTACCACGGTGACGTTG Gpr124 FP- GCGGTCAACATCCACAACTAC NM_054044.2 RP- GATCCTGTTACCCTGCTTTGG Apcdd1 FP- GCTGTGAAGTAAGGTCGGGT NM_133237.4 RP- GAGGGTGTAGGTGGGGTTTG Plvap FP- AGCCAGGTGGTTGGACTATC NM_032398.2 RP- TAGCGGCGATGAAGCGATTA Serpine1 FP- GGCACAGTGGCGTCTTCCT NM_008871.2 RP- TGCCGAACCACAAAGAGAAAG Hmox1 FP- AAACCAGCAGCCCCAAATCC NM_010442.2 RP- TACGAGACAGAAATGTCTGGAAAC Ormdl1 FP- CTGGGAACAGCTGGACTACG NM_145517.4 RP- ATGCAGCTGTGGCATTTTGG Ormdl2 FP- CCTGAAGCTGGCAGGAGACA NM_024180.6 RP- CAGATGCCCCGACTGTTCAT Ormdl3 FP- GCATCTGGCTCTCCTACGTG NM_025661.4 RP- ACCTTGCTTTGCCTTGGTCT Sptlc1 FP- GAATGCACTCGCTTCTGTCG XR_382233.1 RP- AACCCGAAACACTGGATTCTCT Sptlc2 FP- ACTGGACAGGCCCTTTGATG NM_011479.4 RP- ATTGGCTCAGAAAGGCCACA Sptlc3 FP- CTGGAAATTCCTGACCGGTTG NM_001356507.1 RP- CATTGGTCGGATGTGCTGGA Sphk1 FP- GACTGGGAGACTGCCATCCA NM_011451.3 RP- CGTACCCAGCATAGTGGTTCAC Sphk2 FP- CATTCACAGTGAGCGCTTCAG NM_203280.3 RP- TCCACGGTAGGTATGCAACGA Sgpp1 FP- GCAGTTGTGCCCAGGTTTTT NM_030750.3 RP- AAGTCAAGTGTGCGGTAGCA Acer2 FP- CCATGCAACGCTGAGTTTCC NM_139306.3 RP- TGAACCTGCCCCTGTCATTC Smpd1 FP - ACTACCCCGGAAGCTCTCAT NM_011421.2 RP- ATGCGGTAGACCAGGTTGTG Smpd2 FP- GCTCACCTATCCAGATGCACA XM_017313862.2 RP- GCCCTTGCAAGATGAGGGT Sgpl1 FP - GCAGAGGACACTTGCTCCAT NM_009163.4 bioRxiv preprint doi: https://doi.org/10.1101/829820; this version posted November 4, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

RP- TCTGCTTCCTTTTAGCGGGG Zo-1 FP - CCAAGGTCACACTGGTGAAGTC NM_009386.2 RP- CTTGAATGTTACCATCTCTTGCTG Aqp4 FP - TGGTTGGAGGATTGGGAGTCAC NM_009700.3 RP- CAG TTCGTT GGAATCACAGCTG CD45 FP- ATGCCAGCTACATTGATGGCTTC NM_001111316.2 RP- ATCGTGTGACCATGACAATAACTG Pdgfr-β FP- CCACGTGGATCAGCCACTGTC NM_001146268.1 RP- TAA GGCCAGGATGGCTGAGATC S1pr1 FP- CCGGATCGTATCTTGTTGCA NM_007901.5 RP- AAATTCCATGCCTGGGATGA S1pr2 FP- ATGGGCGGCTTATACTCAGAG NM_010333.4 RP- GCGCAGCACAAGATGATGAT S1pr3 FP- GCCTAGCGGGAGAGAAACCT NM_010101.4 RP- CCGACTGCGGGAAGAGTGT S1pr4 FP- TTAGAGTGGTCCGAGCCAATG NM_010102.2 RP- GATCATCAGCACGGTGTTGAGT S1pr5 FP- ATTGCTTTAGAGCGCCACCT NM_053190.2 RP- TTTCCAGGCGTCCTAAGCAG Hprt FP- ACCTCTCGAAGTGTTGGATACAG NM_013556.2 RP- TTCACTAATGACACAAACGTGATTC Syt11 FP - CAAGACCCCGCCATACAAGT NM_018804.3 RP- CCTCTGGGATCCTTGTCGTG Slc2a1 FP- GCCTAAGGTCACATGAAGAAGG XM_006502908.2 RP- AGAGACCAAAGCGTGGTGAG CD144 (Cdh5) FP- CAACTTCACCCTCATAAACAACCAT NM_009868.4 RP- ACTTGGCATGCTCCCGATT Spns2 FP- AGGAGGCCTTCAGTAGCGAT NM_001276383.1 RP- GTCCAGAAGAACACCCTGCT Cav1 FP- GACCCCAAGCATCTCAACGA NM_007616.4 RP- AAATGCCCCAGATGAGTGCC Ager (Rage) FP- AACTCGGTGGGTTGAAGGAA XM_006523501.3 RP- CCCCCATCTCCCATCTCGTT ApoM FP- TACAATCGGTCACCACACCC NM_018816.2 RP- GTCACTGGTCACTTGCTGGA Mfsd2a FP- ACGGCTTCCTTTGGTACCTG NM_029662.2 RP- TTCTGTATGCCGTGGCTGAG Ocln FP- CCCTCTTTCCTTAGGCGACA NM_008756.2 RP- TTCAAAAGGCCTCACGGACA Cldn5 FP- CCCAGTTAAGGCACGGGTAG NM_013805.4 RP- GGCACCGTCGGATCATAGAA Ascl3 FP- AGGCAACGAGTCAAGTGTGT NM_020051.1 RP- GGCTGTTCGAGGGTTCTTCT

NIH Accession Dataset Cell/Tissue Type Disease Model Profiling Method Lab Details Number

Teresa Sanchez - Mouse Cortical microvessels Transient MCAO RNA-seq Weill Cornell GSEXXX Microvessels Medicine

Human Stroke Patients (post- Jonas Frisen - Cortex (lesion-site) RNA-seq GSE56267 Lesion-site mortem) Karolinska Institutet

Supplementary Table 1