RiboTag RNA Sequencing Identifies Local Translation of HSP70 in Astrocyte Endfeet After Cerebral Ischemia

Abstract

Brain ischemia causes disruption in cerebral blood flow and blood–brain barrier integrity, which are normally maintained by astrocyte endfeet. Emerging evidence points to dysregulation of the astrocyte translatome during ischemia, but its effects on the endfoot translatome are unknown. In this study, we aimed to investigate the early effects of ischemia on the astrocyte endfoot translatome in a rodent cerebral ischemia and reperfusion model of stroke. To do so, we immunoprecipitated astrocyte-specific tagged ribosomes (RiboTag IP) from mechanically isolated brain microvessels. In mice subjected to middle cerebral artery occlusion and reperfusion and contralateral controls, we sequenced ribosome-bound RNAs from perivascular astrocyte endfeet and identified 205 genes that were differentially expressed in the endfoot translatome after ischemia. The main biological processes associated with these differentially expressed genes included proteostasis, inflammation, cell cycle/death, and metabolism. Transcription factors whose targets were enriched amongst upregulated translating genes included HSF1, the master regulator of the heat shock response. The most highly upregulated genes in the translatome were HSF1-dependent Hspa1a and Hspa1b, which encode the inducible HSP70. Using qPCR, Western blot, and immunohistochemistry, we confirmed that HSP70 is upregulated in astrocyte endfeet after ischemia. This coincided with an increase in ubiquitination across the proteome that suggests that ischemia induces a disruption in proteostasis in astrocyte endfeet. These findings suggest a robust proteostasis response to proteotoxic stress in the endfoot translatome after ischemia. Modulating proteostasis in endfeet may be a strategy to preserve endfoot function and BBB integrity after ischemic stroke.

1. Introduction

Ischemic stroke is a leading cause of death in the United States and worldwide [1,2,3]. The vast majority of strokes are caused by cerebral ischemia, which results in cerebral edema [4], impaired cerebral blood flow (CBF) [5], neuroinflammation [6], blood–brain barrier (BBB) disruption [7], and widespread cell death [8]. Astrocytes play a central role in maintaining an intact central nervous system (CNS) and responding to brain injury. In particular, astrocytes are responsible for maintaining BBB integrity [9], regulating cerebral blood flow [10], and locally transporting ions and metabolites [11,12]. Many of these functions occur at perivascular endfeet, which directly contact and ensheathe cerebral blood vessels as part of the neurogliovascular unit [13]. As such, astrocytes and their perivascular endfeet are important potential therapeutic targets for improving outcomes in ischemic stroke.

As we have recently reported, astrocyte endfeet exhibit highly dynamic responses to cerebral ischemia, including cationic fluxes that lead to cerebral edema [14]. Yet classically, the cellular response to ischemia and other molecular stresses also involves the activation of highly-conserved transcriptional programs, such as the heat shock response [15] and the hypoxia response [16]. It is poorly understood how such responses might arise in cellular compartments, like astrocyte perivascular endfeet, that are far away from the nucleus. This is a question of general importance to stress responses in the nervous system, in which numerous cell types have complex morphology with essential compartments at great distances from the nucleus. The susceptibility of the BBB and the essential role of astrocyte perivascular endfeet in maintaining it make the mechanisms of local stress response in distal compartments particularly relevant to ischemic stroke.

It is increasingly evident that local protein translation is essential to the function of such cellular compartments [17]. In astrocytes, perivascular endfeet possess the necessary machinery, including the endoplasmic reticulum and Golgi apparatus [18], to sustain a specialized proteome [19,20]. Recently, protein translation of a large repertoire of endfoot-localized mRNA has been reported [18,21]. Similarly, local translatomes have been identified in perisynaptic processes [22,23]. These seminal findings were made following the conception and optimization of the RiboTag approach to isolate ribosome-bound RNA from specific brain cell types from RiboTag transgenic mice [24,25]. This approach has been used to show that major shifts in the astrocyte translatome occur after ischemia, including the translation of stress response and transcription factor genes [26,27]. Yet it is not known whether a distinct translational response occurs specifically in astrocyte endfeet in response to ischemia. In the present study, we aimed to characterize the translatome of perivascular endfeet at an early reperfusion time point in mice following severe ischemia. We observe a wide range of translatomic responses in astrocyte endfeet after ischemia, most notably a marked increase in the translation of the canonical heat shock protein HSP70, suggesting a locally encoded stress response occurring far away from the soma.

2. Results

2.1. Validation of Endfoot RiboTag Isolations After Middle Cerebral Artery Occlusion (MCAo)

To characterize the astrocyte endfoot-specific translatome from the post-ischemic brain, we immunoprecipitated (IP’d) RiboTag and ribosome-bound RNA from mechanically isolated microvessels [28], as previously reported (Figure 1A) [18]. The RiboTag IP relies on the transgenic expression of a GFP-tagged RPL10a ribosomal protein driven by an Aldh1l1 promoter, which largely restricts expression to astrocytes [18,25]. Astrocyte specificity has been shown to persist following brain ischemia [27].

We first sought to validate the approach for use in a model of stroke in which mice are subjected to 2 h of ischemia via MCAo and 6 h reperfusion. Ischemia has previously been shown to induce morphological changes in perivascular endfeet such as swelling [29] and degeneration [30]. We immunolabelled for GFP in brain section from the transgenic mice following MCAo and reperfusion, and found that GFP was detected in S100B+ astrocytes, including in perivascular domains with minimal localization over CD31+ vessels (Figure 1B). This suggested that GFP-tagged ribosomes were expressed specifically in astrocytes. The GFP signals were detected in isolated microvessels from the post-ischemic brain including co-labelling over AQP4⁺ astrocyte endfeet, suggesting that RiboTag expression survives the mechanical isolation of microvessels (Figure 1C). STED imaging to visualize nanometer-sized GFP puncta in isolated post-ischemic microvessels revealed that GFP remains primarily perivascular (Figure 1D). Using a magnetic bead-conjugated anti-GFP antibody to IP eGFP-tagged RPL10a (Supplementary Figure S2), we confirmed detectable levels of RiboTag in GFP-IP’d post-ischemic microvessel lysates (Figure 1E). qPCR of RiboTag-IP’dmRNA from isolated microvessels after stroke revealed enrichment of astrocyte markers Aqp4 and Gfap [31,32], as well as depletion of vascular markers including Pecam1, Cldn5 (endothelial cells), Pdgfrb (pericytes), and Acta2 (vascular smooth muscle cells) (Figure 1F). To confirm that the preparation excludes astrocyte soma, we verified that the astrocytic transcription factor SOX9 was absent in mechanically isolated post-ischemic microvessels (Supplemental Figure S1A) [33]. Furthermore, the nuclear pore complex (NPC) protein NUP153 was depleted in enzymatically digested and magnetically sorted ACSA2+ astrocyte endfeet (Supplemental Figure S1B) [19]. These results collectively verified that perivascular astrocytic RiboTag and ribosome-bound mRNA were detectable in post-ischemic isolated microvessels, and further established a molecular signature in perivascular astrocyte endfeet distinct from that in the soma.

2.2. Quality Assessment of RiboTag RNA Sequencing

To identify changes in the translatome induced by stroke, we extracted RNA from endfoot ribosomes and performed bulk RNA sequencing. As it has been reported that poly(A) selection can introduce bias into RNA sequencing data sets [34], we sought to determine whether adequate sequencing depth could be achieved without selection. We sequenced 67 ± 18 million reads per sample, of which 63 ± 1% were mapped uniquely. Of those reads not uniquely mapped, we aligned 1 million per sample against the genomic sequence of Mus musculus 45S pre-ribosomal RNA (NR_046233.2) and determined that 82 ± 0.02% were ribosomal in origin (Figure 2A). This suggests that the relatively low unique mapping rate in our samples is a result of the absence of complete rRNA depletion. As the proportion of rRNA is expected to be much higher in whole transcriptome samples, our results suggest that the RiboTag approach substantially enriches for mRNA, which may serve as an internal control for the quality of ribosome isolation. As another quality control measure, we found read counts to be >90% correlated with each other across all samples (Figure 2B).

2.3. Features of the Post-Ischemic Endfoot Translatome

Principal component analysis of the transcriptome data set revealed two major components contributing to the variation in the data (Figure 3, Table S3). Contralateral (CTR) and ischemia/reperfusion (Stroke) samples were clearly distinguished by PC2. From our data set, we found that 86 genes were upregulated and 119 genes were downregulated in post-ischemic endfeet compared to contralateral endfeet. Of the six genes that contribute most to the variation in PC2, five are significantly differentially expressed: Hspa1a (log₂Fold Change: 3.51, s-value: 1.37 × 10⁻⁸), Hspa1b (log₂Fold Change: 3.06, s-value: 3.21 × 10⁻⁷), Ptx3 (log₂Fold Change: 2.93, s-value: 2.72 × 10⁻⁸), Sfrp5 (log₂Fold Change: −2.58, s-value: 4.51 × 10⁻⁵), and Heyl (log₂Fold Change: −2.32, s-value: 1.65 × 10⁻⁴) (Figure 3C and Figure 4, Table S4). We also detected the upregulation of three mitochondrial genes (mt-Cytb, mt-Nd2, and mt-Nd1), which cytosolic ribosomes would not typically translate. This may reflect the release of mitochondrial transcripts from fragmented mitochondria in the setting of ischemia [35].

We next asked whether the translatome response to stroke in astrocyte endfeet resembles that previously reported in whole astrocytes in a model similar, but not identical, to our own [27]. To maintain consistency, we re-aligned and analyzed raw data files from RiboTag sequencing reads of astrocytes at four hours after ischemia. We found a highly significant overlap between differentially expressed genes (DEGs) in endfeet and whole astrocytes, especially among upregulated genes (Figure 2D). This included a prominent activation of heat shock response genes like Hspa1a and Hspa1b, as well as hypoxia response genes like Hmox1 and acute phase response genes like Ptx3. Nevertheless, 45% of genes detected as upregulated in endfeet were not detected as upregulated in whole astrocytes. Only two downregulated genes overlapped between the two data sets, the Hes5 transcription factor gene and the poorly characterized gene RP23-391P21.4. As we detect transcription factors and somatic marker genes in our data set, we cannot completely exclude the presence of RNA from cell bodies.

GSEA revealed a range of enriched pathways associated with cell cycle/death, inflammation, metabolism, and proteostasis (Figure 5A, Table S5). Amongst downregulated genes, all enriched pathways were related to metabolism, including oxidative phosphorylation and two pathways related to lipid metabolism (bile acid metabolism and adipogenesis). Notably, the most enriched pathway is TNFα signaling via NF-κB. Pathways for the unfolded protein response and the mTOR pathway were also significantly enriched among upregulated genes.

As we recently showed marked calcium influx in perivascular astrocyte endfeet induced by ischemia [14], we wondered whether calcium-responsive genes were differentially expressed in stroke (Figure 5B, Table S6). The Gene Ontology Biological Process term ‘Cellular Response to Calcium Ion’ was found to be enriched in upregulated genes (Benjamini–Hochberg-adjusted p = 3.3 × 10⁻², NES = 1.7) and is the only enriched ion-response term (Table S5). The calcium-regulated transcriptome was recently defined in glomerular podocytes based on calcium ionophore challenge [36]. Like astrocytes, glomerular podocytes participate in regulating vascular permeability [37]. We found no overlap in downregulated genes, but a significant overlap in upregulated genes (odds ratio = 9.7, Holm–Bonferroni-adjusted Fisher exact test p = 2.7 × 10⁻⁴).

Given the prominent role that the heat shock response plays in the post-ischemic astrocyte endfoot translatome, we sought to further validate these changes using qPCR (Figure 4B). In addition to the upregulation of Hspa1a and Hspa1b, we find that Sfrp5, the most downregulated gene in our translatome data set, is also decreased in expression based on qPCR. Similarly, the most upregulated and downregulated calcium-responsive genes in Thbs1 and Itpkb, respectively, are also significantly changed in expression in our data set.

2.4. Regulatory Features of Differentially Expressed Genes

We next investigated potential mechanisms of regulation of the endfoot translatome. Given recent evidence that RNA structural features are associated with expression in the endfoot translatome, we tested whether sequence length and GC content differed in control and stroke tissue (Figure 6A,B). We found that upregulated genes tend to have longer 5′-UTRs and protein-coding sequences, while downregulated genes tend to have shorter 3′-UTRs and protein-coding sequences. On average, both upregulated and downregulated genes have higher GC content in the protein-coding region than do genes in the translatome as a whole, but only downregulated genes tend to have a higher GC content in the 3′-UTR. Reasoning that these features may affect translation in part through selection by specific RNA-binding proteins, we tested the association between differential expression in stroke and dependence on the Quaking RNA-binding protein (QKI). It was recently shown that deletion of QKI alters the translatome of maturing astrocytes [38]. We found that 9.2% of downregulated genes (odds ratio = 10.7, Holm–Bonferroni-adjusted Fisher exact test p = 3.2 × 10⁻⁸), but no upregulated genes, were QKI-dependent (Figure 6C). Conversely, an unbiased search of transcription factors using the TFLink database [39] revealed only transcription factors enriched in upregulated genes (Figure 6D, Tables S7 and S8). These included HSF1, the master regulator of the heat shock response, as well as a number of transcription factors associated with inflammation (e.g., NFκB1) and cell cycle/death (e.g., FOSL1, RB1). Interestingly, DAND5 has been reported to be a negative regulator of angiogenesis [40]. These transcription factors form a network of shared regulation, with many genes under the influence of multiple transcription factors (Figure 6E). While Hspa1a and Hspa1b, canonical genes in the heat shock response, are transcriptionally activated by HSF1, they differ in terms of additional genetic interactions with other transcription factors. FOSL1, for example, regulates Hsp1a but not Hspa1b, while DAND5 and SP3 regulate Hspa1b but not Hspa1a.

2.5. Upregulation of HSP70 and Protein Ubiquitination in Post-Ischemic Endfeet

Given that Hspa1a and Hspa1b were the most upregulated genes in the post-ischemic endfoot translatome, we asked whether their protein product, inducible HSP70, was more abundant in endfeet after stroke. We isolated astrocyte endfeet from post-ischemic or control microvessels following proteolytic dissociation to single cells and positively selection for ACSA2+ endfeet by magnetic separation (Figure 7A). Immunoblotting for HSP70 revealed its upregulation in astrocyte endfeet after ischemia (Figure 7A,B). The constitutively expressed heat shock cognate HSC70, whose corresponding gene was more modestly upregulated in the stroke endfoot translatome, shows a small but significant upregulation in post-ischemic endfeet (Figure 7A,B). This confirmed our expectation that inducible HSP70 would be upregulated more than its constitutively expressed paralog. Immunolabelling of HSP70 corroborated the upregulation of HSP70 found in immunoblotting, with increased HSP70 immunoreactivity in AQP4-positive endfeet after stroke (Figure 7C). We hypothesized that the robust upregulation of HSP70 and other proteostasis components may have been in response to proteotoxic stress induced by ischemia. Consistent with this, we note a global increase in ubiquitination in stroke (Figure 7D). These results suggest that active translation of Hspa1a and Hspa1b in the astrocyte endfeet after stroke are likely in response to global disruptions in endfoot proteostasis.

3. Discussion

In this study, we report the translatomic response to ischemia in perivascular astrocyte endfeet. In so doing, we address a fundamental question regarding translational responses to acute molecular stress in cellular compartments located far away from the nucleus. We have identified a set of 205 differentially expressed genes in the astrocyte endfoot translatome after a 2 h ischemia/6 h reperfusion insult to the brain territory supplied by the middle cerebral artery. While dysfunction of local translation in astrocytes has been implicated in other pathologies such as leukoencephalopathy [21] and ALS [41], it has not been demonstrated previously in stroke.

The changes we have observed in the endfoot translatome overlap with, but are distinct from, those previously reported in the whole astrocyte translatome [26,27]. One common feature is the robust activation of the heat shock response. Consistent with our results, genes encoding HSP70 have been shown to be upregulated in the whole astrocyte translatome at 4 h [27], but not at 72 h, after stroke [26]. Overexpression of HSP70 rescues cells exposed to ischemia from apoptosis [42] and decreases the size of the ischemic lesion in vivo [43]. The upregulation of HSP70 that we observe in astrocyte endfeet after ischemia may be explained by a rise in proteotoxic stress and disruption in proteostasis, as indicated by increased protein ubiquitination. We have shown using various assays, including RiboTag RNA-sequencing, immunohistochemistry, qPCR, and immunoblotting, that HSP70 is induced in astrocyte endfeet after cerebral ischemia and reperfusion.

To our knowledge, stress-induced expression of HSP70 has not been reported in astrocyte endfeet previously. We have also shown that the presence of HSP70 protein in endfeet after ischemia is a result, at least in part, of local translation. Our findings add to the growing body of evidence for local proteostasis responses in cellular processes and compartments distant from the nucleus, such as axons and synapses [44]. For example, transcripts encoding HSP70 have been observed in regenerating axons of cultured dorsal root ganglion neurons [45]. After acute molecular stress—classically, from heat, but also ischemia and hypoxia—the transcription factor HSF1 trimerizes and is translocated to the nucleus, where it activates a broad transcriptional program known collectively as the heat shock response [46,47,48,49]. The molecular chaperone HSP70 is a central part of this response and rapidly becomes one of the most highly expressed proteins in the cell after acute stress [50]. Our results suggest that the localization of this response to distal cellular processes may be dependent not on the trafficking of HSP70 protein synthesized in the soma, but rather on translation of a local pool of mRNA. It is unknown whether these mRNA molecules are transported to the endfoot acutely after stress or are stored under normal conditions and selectively engaged by the ribosome in response to stress. In either case, the mechanisms that regulate HSP70 in endfeet appear to deviate in important ways from the canonical heat shock response. As the heat shock response is a general, highly conserved response to a wide variety of proteotoxic stresses, this may have broad implications for proteostasis in cells with complex morphology.

HSP70 is upregulated in neurons, astrocytes, and microglia after cerebral ischemia [51,52]. Cerebral ischemia induces a range of other changes in proteostasis, including the unfolded protein response in the endoplasmic reticulum and activation of autophagy [53]. An open question is how these pathways may be altered in astrocyte endfeet, and what the implications are of such responses to injury and repair after ischemia.

We observed that the rise in HSP70 coincides with a substantial increase in ubiquitination. A widespread increase in ubiquitination in astrocyte endfeet after ischemia has not been reported previously, but our results are reminiscent of earlier findings of ubiquitination in post-synaptic densities after ischemia [54]. Our results suggest that ischemia induces local disruptions in proteostasis that subsequently lead to the activation of proteostasis pathways. These effects of ischemia are not unexpected, as many components of the proteostasis network, including the proteasome and a range of molecular chaperones, are ATP-dependent. Therefore, in the ATP-starved environment of ischemia, the maintenance of proteostasis is impaired. As we have assayed astrocyte endfeet 6 h after reperfusion, we may be observing the response to damage accumulated during the 2 h ischemic period. Whether this is salutary or a contributor to reperfusion injury remains to be determined.

While our study has focused on the early events after ischemia and reperfusion, a recent study of the whole-astrocyte translatome revealed important differences between 4 h and 3 days after ischemia [27]. Future studies on the time course of translatomic and proteomic changes may yield mechanistic insights by defining the sequence of proteostasis changes that occur after ischemia. Similarly, understanding changes in ubiquitination across the proteome at various time points after ischemia will be important to better characterize disruptions of proteostasis that occur in astrocyte endfeet in this disease.

In addition to the prominent role of the heat shock response, our findings suggest other intriguing local response mechanisms. Pathway analysis suggests that these center around four major functions: cell cycle/death, inflammation, metabolism, and proteostasis. The transcription factors whose targets are enriched among our differentially expressed genes are consistent with this pattern. We recently reported a significant calcium influx in astrocyte endfeet after ischemia [14]. Consequently, we note that a significant number of calcium-regulated genes are differentially expressed in the endfoot translatome after stroke. The most upregulated calcium-responsive gene is Thbs1, which encodes the cell adhesion protein thrombospondin 1 (TSP1) that plays a complex role in angiogenesis [55,56]. Itpkb, the most downregulated calcium-response gene, encodes inositol trisphosphate-3 (IP₃) kinase-B, which regulates the IP₃ second messenger system with wide-ranging effects on cellular function. It has a putative role in neurodegenerative diseases [57,58], but its potential role in stroke has not been described. Other differentially expressed genes have also been implicated in angiogenesis. Ptx3 encodes the acute-phase protein pentraxin-3, important in the classical complement pathway that has been shown to promote neurogenesis and angiogenesis in rodent models of stroke [59,60] and to regulate the blood–brain barrier [61]. Sfrp5 encodes the secreted frizzled-related protein 5, a Wnt-signaling pathway antagonist and anti-inflammatory adipokine that has been shown to regulate angiogenesis [62,63].

Beyond highlighting specific genes, our findings also suggest potential mechanisms by which translatomic changes in astrocyte endfeet might be regulated. The enrichment of specific transcription factor targets could reflect either transport of de novo synthesized mRNAs from the nucleus or the maintenance of selective pools of mRNA in endfeet that ribosomes translate after a stimulus. Our analysis points towards possible mechanisms for such regulation, including basic features of the transcript and association with RNA-binding proteins. Actively translating transcripts in perisynaptic astrocyte processes have been shown to have longer 5′-UTRs, coding regions, and 3′-UTRs than their somatic counterparts [22]. They also have decreased RNA secondary structure in their 3′-UTR, but similar GC content throughout. We find that some of these features are also associated with the post-ischemic endfoot translatome. In particular, genes upregulated and downregulated in response to ischemia and reperfusion differ in length. They also differ from the unchanged translatome in terms of GC content. This suggests that some of the same features that help to regulate localization of transcripts to peripheral processes may play a role in guiding their response-specific translation in these compartments. Similarly, a significant number of the transcripts known to be regulated by the RNA-binding protein QKI in whole astrocytes overlap with genes downregulated in response to ischemia–reperfusion in astrocyte endfeet. QKI and other RNA-binding proteins may be important in governing mRNA transport to and ribosome binding in astrocyte endfeet.

One limitation of our study is the specificity of current isolation methods. It has been reported that perisynaptic astrocyte process preparations do not completely remove neurons [22]. We have shown that our microvessel preparations are depleted in SOX9 protein, an astrocyte-specific transcription factor. In addition, affinity purification of astrocyte endfeet from microvessels using ACSA2⁺ selection reveals an absence of the NUP153 nuclear pore protein, as would be expected for a non-somatic compartment. However, transcriptomic approaches are likely highly sensitive to contamination from cell bodies, which have a larger absolute abundance of mRNA. As we do observe the mRNA expression of transcription factors and somatic markers (Table S4) in our astrocyte endfoot RiboTag preparations, we cannot fully exclude some degree of somatic contamination. That said, the establishment of putative somatic markers is dependent on the sensitivity of the assay and the detection cutoff used. As sequencing methods advance, it may become apparent that some markers are more widely expressed than previously recognized. For example, we detect mRNA expression of the transcription factor gene Hes5 in our endfoot preparations. This may result from somatic contamination or from wider expression of Hes5 than previously recognized. Protein expression of HES5 has been observed outside the nucleus in the extended processes of cochlear pillar and Deiters’ cells [64]. Whether this reflects additional functions beyond transcription factor activity is unknown. In some cases, such as that of SOX9, there is relatively low protein expression despite high mRNA expression, which may suggest additional post-transcriptional or post-translational regulation.

That we see a translatome in endfoot preparations that is distinct from whole astrocyte responses to ischemia suggests that we are able to identify endfoot-specific signals. In addition, we find overall low expression of endothelial markers in our RiboTag data set. Finally, we have confirmed our observations about HSP70 using more targeted approaches. In particular, immunohistochemistry of coronal brain sections is independent of isolation techniques and provides strong validation for the localization of HSP70 in astrocyte endfeet after stroke.

4. Materials and Methods

4.1. Mice and Surgical Procedure

Astrocyte-specific RiboTag mice (FVB-Tg(Aldh1l1-EGFP/Rpl10a)JD133Htz/J, #030248, Jackson Laboratory, Bar Harbor, ME, USA) were bred and maintained in-house (University of Maryland School of Medicine, Baltimore, MD, USA). Transgene expression was confirmed by genotyping using collected tail samples (Transnetyx, Cordova, TN, USA). Surgery to induce middle cerebral artery occlusion (MCAo) in male (10–12 week-old) mice was performed as previously described [14], that is, common carotid artery ligation and intraluminal occlusion using a 6-0 silicon filament occluder (Doccol Corp, Sharon, MA, USA). Following placement of the occluder, MCAo animals recorded > 70% reduction in relative cerebral blood flow (rCBF) as measured using laser Doppler flowmetry (DRT2, Moor Instruments, Axminster, Devon, UK). Following 2 h occlusion, the filament occluder was removed and the mice were allowed to recover.

4.2. Brain Microvessel and Astrocyte Endfeet Isolation

Following 6 h of reperfusion, mice were terminally anesthetized by pentobarbital overdose, then transcardially perfused with ice-cold normal saline (0.9%). MCA territories were dissected, then homogenized by hemisphere in a 7 mL glass Dounce homogenizer (12–15 strokes) in 1× HBSS containing Ca²⁺/Mg²⁺ (Gibco, Waltham, MA, USA) and supplemented with HEPES. Homogenates were centrifuged at 2000× g for 10 min at 4 °C. Pellets were resuspended in 18% dextran (from Leuconostoc spp., M_r~70,000, MilliporeSigma, Burlington, MA, USA), then centrifuged at 10,000× g for 15 min at 4 °C. Myelin-cleared pellets were resuspended in 1% bovine serum albumin, then filtered through a 20 μm nylon mesh. Microvessels caught in the mesh were retrieved and centrifuged at 2000× g for 10 min. For immunostaining, microvessels were cytospun onto glass microscope slides and fixed in 4% paraformaldehyde for 10 min at 4 °C prior to storage at −20 °C. For RiboTag immunoprecipitation, solutions were supplemented with 100 µg/mL cycloheximide and 200 U/mL RNAsin. Astrocyte endfeet isolation was performed as previously described [19]. In brief, isolated microvessels were dissociated into single cells following incubation in 100 µg/mL Liberase DL (MilliporeSigma) at 37 °C for 30 min. Larger cells and debris were removed by centrifugation at 300× g at 4 °C for 30 min. The supernatant was centrifuged at 25,000× g at 4 °C for 30 min, and the pellet was resuspended in AstroMACS separation buffer (Miltenyi, Bergisch Gladbach, Germany). Following labelling with magnetic-bead-conjugated anti-ACSA2 (Miltenyi), ACSA2+ astrocyte endfeet were positively selected and immediately processed for protein analysis.

4.3. RiboTag Immunoprecipitation (IP) and RNA-Seq

For RiboTag IP, all steps were performed in 4 °C and RNase-free (RNaseZAP, ThermoFisher, Waltham, MA, USA) conditions. Microvessels were lysed in freshly prepared lysis buffer (150 mM KCl, 20 mM HEPES [pH 7.4]), 5 mM MgCl₂, 1 mM DTT, 1% NP-40, 100 µg/mL cycloheximide, and 30 mM DHPC [1,2-dicaproyl-sn-glycero-3-phosphocholine, 07:0]) supplemented with 200 U/mL RNAsin (ProMega, Madison, WI, USA) and cOmplete protease inhibitor (MilliporeSigma). Following homogenization using a handheld pestle motor mixer, lysates were centrifuged at 18,000× g at 4 °C for 10 min. IP was performed by incubating the soluble fraction with magnetic bead-conjugated anti-GFP antibody (#67090, Cell Signaling Technology, Danvers, MA, USA) or magnetic bead-conjugated control IgG (#5873, Cell Signaling Technology) overnight (~16 h) with end-over-end rotation. Following magnetic separation from the supernatant (DynaMag-2, ThermoFisher), beads were washed three times in a high-salt (300 mM KCl) lysis buffer supplemented with 200 U/mL RNAsin. Unpurified RNA was collected from the magnetic beads using the TriZOL reagent (Invitrogen, Waltham, MA, USA) according to the manufacturer’s instructions. Total mRNA (in ultrapure water) following RiboTag IP was quantified using the Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA). Strand-specific RNA sequencing was performed by Maryland Genomics in NovaSeq S4 flow cells (NovaSeq6000, Illumina, San Diego, CA, USA) with 100 bp paired-end reads, with 67 ± 18 million (mean ± SD) reads per sample.

4.4. Sequence Alignment

Raw R1 and R2 fastq files were trimmed to remove adapter sequences and low-quality reads using the Trimmomatic software v0.33 under the following parameters: simple clip threshold (7), seed mismatches (2), palindrome threshold (40), minimum sequence length (3), and trailing quality (20). Trimmed sequences were followed by quality control evaluation using FastQC v0.12.0. Trimmed sequences were aligned using STAR v2.7.5a [65] to the Mus musculus mm39 primary assembly. These were quantified at the gene level using the STAR geneCounts function to obtain reverse/forward strand-specific counts. Analysis of unmapped, multi-mapped, and ambiguously mapped reads was performed using blastn [66] with default settings.

4.5. Differential Gene Expression Analysis

Analysis was performed using DESeq2 v1.44.0 [67] in R (v4.4.1). There were 10,862 genes that had ≥10 counts per sample in ≥3 samples. We assessed the pairwise correlation of gene counts in these samples after rlog transformation. We generated a model for differential expression analysis using two variables: (1) individual animal (ID 1, 2, or 3) and (2) disease state (Control or Stroke). We performed fold change shrinkage using the adaptive shrinkage method ashr. Significance testing was performed against the null hypothesis that |log₂(Fold Change)| < 0.322, known as the s-value, corresponding to gene expression in the ischemia/reperfusion condition of <4/5 or >5/4 that of controls [68].

4.6. Principal Component Analysis (PCA)

PCA of the gene counts was performed using the R package PCAtools v2.16.0. The correlation between the first 6 principal components and the sample variables was performed using the eigencorplot function. We generated the biplot as described in PCAtools, plotting each sample using the centered input data multiplied by the variable loadings (i.e., “rotated data”) for principal components 1 and 2. Loadings are rescaled, plotted in the manner used by PCAtools, with a scaling factor r generated from the variable loadings (var_x) and rotated loadings (rot_x), where each plotted principal component x:

$r=1.5\times \left({\displaystyle \frac{\mathrm{range}\left(ro{t}_{PC1}\right)}{\mathrm{range}\left(va{r}_{PC1}\right)}},{\displaystyle \frac{\mathrm{range}\left(ro{t}_{PC2}\right)}{\mathrm{range}\left(va{r}_{PC1}\right)}}\right)$

where the range is the difference between the maximum and minimum value of the given variable.

4.7. Gene Set Enrichment Analysis (GSEA)

GSEA was performed using the R package fgsea v1.30.0 [69,70,71]. Three sets of pathway databases were obtained from the mouse Molecular Signatures Database [72]: hallmark gene set (MH, v2024.1), canonical pathway curated gene set (M2.CP, v2024.1), and Gene Ontology (GO) biological processes gene set (M5.GO.BP, v2024.1). To compare pathways to our gene set, a mapping table was generated of Ensembl gene IDs to gene symbols by automated conversion supplemented with manual curation. A pathway was included in enrichment analysis if our gene expression data set included at least 10 genes from that pathway and 50% of total genes in the pathway. This resulted in 44/50 MH, 965/1730 M2.CP, and 3719/7713 pathways included in downstream analysis. Enrichment was performed using the fgsea function fgseaMultilevel, which uses the adaptive multilevel splitting Monte Carlo approach, with default settings. Ranking of genes for enrichment analysis was done in one of two ways, based on (1) fold change or (2) the Wald statistic, and a pathway significance threshold was set at Benjamini–Hochberg adjusted p < 0.00833 (≈0.05/6 to account for multiple tests across 3 pathway gene sets with 2 ranking approaches). Consensus enriched pathways were identified that were significant by both the fold change and Wald statistic metrics. Pathways were then manually annotated into categories as shown in Figure 5.

4.8. Gene Structural Analysis

From the GENCODE vM35 protein-coding transcript sequences file, sequences and annotations for 5′-UTR and 3′-UTR sequences were extracted.

4.9. Transcription Factor Target Enrichment Analysis

Transcription factor and target mapping were obtained from the TFLink database (TFLink_Mus_musculus_interactions_All_GMT_ncbiGeneID_v1.0.gmt) [39]. To compare transcription factor target sets to our gene set, a mapping table was generated of Ensembl gene IDs to NCBI gene IDs by automated conversion supplemented with manual curation. Enrichment analysis was performed using Fisher’s exact test with multiple hypothesis correction performed using the Benjamini–Hochberg method.

4.10. Immunohistochemistry (IHC), Fluorescence Imaging, and Stimulated Emission Depletion (STED) Imaging

After terminal anesthesia using excess pentobarbital, mice were transcardially perfused with ice-cold normal saline (0.9%), then fixed with 4% paraformaldehyde. Whole brains were dissected, post-fixed overnight, and cryopreserved in 30% sucrose (in 1× PBS) for at least 48 h. Then, 12 μm thick coronal brain sections were prepared on glass microscope slides by cryosectioning. Tissue sections or isolated microvessels on glass microscope slides were immunostained as previously described [19]. Primary and fluorophore-conjugated secondary antibodies are listed in Table S1. Slides were coverslipped using either a ProLong Gold Antifade mount (Invitrogen) for fluorescence imaging or the liquid antifade mountant (Abberior, Heidelberg, Germany) for STED imaging. Omission of primary antibody was included to verify specificity of the immunolabelling. Fluorescent images were acquired using an upright NiE fluorescence microscope system (Nikon, Tokyo, Japan) accompanied with the 20× (Nikon Plan Apo λ 20×/0.75 NA, WD 1.0 mm) or 100× (Nikon Plan Apo λD 100×/1.45oil NA) objectives. Excitation and emission wavelengths were as follows (in nm): 488/525, 561/620, and 640/710. Images were processed and analyzed using NIS Elements v5.30 (Nikon). STED imaging was performed using the STED Facility Line microscopy system (Abberior). Raw STED images were acquired using the 60× (Olympus UPlanSApo 60×/1.30sil NA. Olympus Microscopy Technologies, Tokyo, Japan) objective. The signal for STED imaging was excited at 640 nm with pulsed depletion at 775 nm. Images were acquired using the Imspector Image Acquisition software and further processed on ImageJ v1.54f (National Institutes of Health, Bethesda, MD, USA). Both fluorescent and STED images were acquired at equal laser strength and exposure times between non-ischemic control and ischemic conditions.

4.11. Quantitative Reverse-Transcription PCR (qPCR) of Translating Endfeet RNA

Following RiboTag IP and RNA extraction, RNA amount was quantified using the Qubit 4 Fluorometer (ThermoFisher). First-strand cDNA was generated from 40 ng RNA using the SuperScript IV reverse transcription kit (ThermoFisher) according to the manufacturer’s recommendations. qPCR reactions were prepared using the PowerUp SYBR Green master mix (Invitrogen) with (per reaction) 1 µL cDNA and 200 nM each of the forward and reverse primers. qPCR was performed using the ABI7300 (Applied Biosystems, Waltham, MA, USA) with the following thermal cycling conditions: 50 °C for 2 min, 95 °C for 2 min, then 40 cycles of 95 °C for 15 s and 60 °C for 30 s. Melting curve analysis was included in all qPCR reactions to confirm that a single product was amplified per reaction. Primers (sequences listed in Table S2) were designed to span exon–exon junctions. Raw C_T values were normalized to that of Gapdh for each biological sample, then quantified relative to the mean of the fold changes of microvessels or non-stroke control mRNA.

4.12. Immunoblotting

Protein was extracted from brain homogenates and isolates using 1× RIPA buffer supplemented with cOmplete protease inhibitor (MilliporeSigma). Following homogenization using a handheld pestle motor mixer, samples were centrifuged at 18,000× g at 4 °C for 10 min. Supernatant containing soluble proteins was separated from the pelleted insoluble fraction. Protein amount was quantified using the Qubit 4 fluorometer and Qubit protein assay kit (ThermoFisher). Lysates were denatured at 70 °C for 5 min in NuPAGE LDS sample buffer and sample reducing agent (ThemoFisher). Conventional protein gel electrophoresis, blot transfer, and immunoblotting (in 5% nonfat milk in 1× TBST) were performed. Primary and secondary antibodies used for immunoblotting are listed in Table S1. Blots were visualized using the SuperSignal West Atto Ultimate Sensitivity substrate (ThermoFisher). Densitometry analysis was performed using ImageJ v1.54f.

4.13. Statistical Analysis and Figure Preparation

Unless specified otherwise, data are presented as mean ± SEM, and the statistical analyses employed are described in the figure legends. Figures were prepared using NIS Elements (Nikon), ImageJ (Fiji), and the R package ggplot2 v3.5.1 [73]. Schematics were created using BioRender.com.

5. Conclusions

Taken together, our findings demonstrate that early phases of cerebral ischemia and reperfusion induce a rapid change in the translatome of perivascular astrocyte endfeet. Given the critical role of astrocyte endfeet in maintaining the BBB and the significance of endfeet swelling after ischemic stroke, the regulation of the endfoot proteome is highly relevant to the pathophysiology of stroke. In particular, we find a local upregulation of HSP70 in this cellular compartment distant from the astrocyte soma, suggesting a non-canonical activation of the heat shock response. Future investigations of the endfoot translatome would help to resolve several remaining questions, including the time course of translatomic changes, the differences between the translatome and stored pools of mRNA in endfeet, and the mechanisms of mRNA transport to endfeet. We anticipate that such efforts to determine the regulation of the astrocyte endfoot translatome will yield new biomarkers and therapeutic targets for ischemic stroke.

Author Contributions

Conceptualization, B.S., P.C., V.G. and J.M.S.; methodology, B.S. and P.C.; validation, B.S., P.C., C.T., R.S. and K.K.; formal analysis, B.S., P.C., V.G. and J.M.S.; investigation, B.S., P.C., C.T., R.S. and N.T.; data curation, B.S., P.C., C.T. and V.G.; writing—original draft preparation, B.S. and P.C.; writing—review and editing, B.S., P.C., V.G. and J.M.S.; visualization, B.S., P.C. and V.G. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was approved by the Institutional Animal Care and Use Committee (protocol #0522007, approved 23 August 2022) of the University of Maryland School of Medicine (Baltimore, MD, USA).

Informed Consent Statement

Not applicable.

Data Availability Statement

All data presented in this study are available upon request from the corresponding author.

Acknowledgments

We acknowledge Shilpa D. Kumar and the Confocal Microscopy Core, Center for Innovative Biomedical Resources (CIBR), University of Maryland School of Medicine (Baltimore, MD, USA), for their technical expertise with STED imaging. We thank Luke J. Tallon of Maryland Genomics, Institute for Genome Sciences (IGS), University of Maryland School of Medicine (Baltimore, MD, USA), for the RNA sequencing.

Conflicts of Interest

J.M.S. and V.G. have filed a U.S. Provisional Patent Application (Number 63/324,492, filed 28 March 2022) titled “Methods and Compositions for the Treatment of Stroke”. All other authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures

Figure 1. Detection of RiboTag and ribosome-bound mRNA from post-ischemic astrocyte endfeet. (A) Schematic illustrating the induction of stroke via transient MCAo/R, followed by isolation of microvessels from the post-ischemic brain tissue and RiboTag IP. (B) Representative high-powered image of the post-ischemic cortex with immunolabelling for GFP (cyan), S100B (yellow), and CD31 (red) (scale bar = 20 µm). (C) Representative image of microvessels mechanically isolated from the post-ischemic tissue with immunolabelling for GFP (cyan), AQP4 (yellow), and CD31 (red) (scale bar = 10 µm). (D) High-powered images without or with STED of isolated post-ischemic microvessels immunolabelled for GFP (white) and CD31 (red) (scale bar = 10 µm). Yellow arrowheads indicate STED-resolved GFP signals. (E) Representative immunoblot following RiboTag IP from post-ischemic brain lysate or microvessels. (F) qPCR of post-ischemic microvessels before RiboTag IP (MVs), after RiboTag IP (GFP+), or non-IP’d flowthrough (GFP−). Data expressed as mean ± SEM (n = 3 biological replicates).

Figure 2. RNA sequencing analysis of ribosome–bound RNA from post-ischemic astrocyte endfeet. (A) Percentage of reads not uniquely mapped to the mouse genome that were subsequently mapped to mouse ribosomal DNA. (B) Correlation matrix of the entire sample set (3 contralaterals C1–C3, 3 MCAo S1–S3) used in this study. Pearson r was used for the heatmap color scheme. (C) MA scatterplot showing the relationship of fold change to mean normalized counts after log fold change shrinkage. Red and blue points are differentially expressed. (D) Venn diagrams of the 205 differentially expressed endfoot genes (86 upregulated and 119 downregulated, as indicated by the up and down arrows, respectively) after ischemia plotted against DEGs identified in an existing astrocyte translatome during the hyperacute phases of ischemia [27]. (* p [less than] 0.05, **** p [less than] 0.0001, Holm-Bonferroni-adjusted Fisher exact test).

Figure 3. Principal component analysis of RiboTag RNA sequencing from post–ischemic astrocyte endfeet. (A) Scree plot of principal components (PCs) with threshold at PC2 indicated by the red dashed line. (B) Correlation matrix plotted for the tested variables (Disease State or Animal) against the six PCs, with Pearson r2 displayed. (C) Principal component biplot of the first two PCs showing individual endfoot samples C1–C3 and S1–S3 overlayed with the most negative and positive gene loadings contributing to variation explained by PC2. (* p [less than] 0.05, Holm-Bonferroni-adjusted p-value for Pearson correlation).

Figure 4. Differentially expressed genes and validation of top hits. (A) Volcano plot depicting changes in post-ischemic endfoot gene expression. Vertical gray lines show threshold of log2 fold change significance (±0.322). Gray line shows threshold of significance (s-value [less than] 0.05). (B) qPCR validation of the most upregulated (Hspa1a, Ptx3, Thbs1. CTR vs. Stroke in grey and red bars, respectively) and downregulated (Sfrp5, Heyl, Nrarp, Itpkb. CTR vs Stroke in grey and blue bars, respectively) genes. Relative fold changes per gene normalized to Gapdh are expressed as mean ± SEM (* p [less than] 0.05, ** p [less than] 0.01, **** p [less than] 0.0001, Holm-Bonferroni-adjusted paired two-tailed t-test).

Figure 5. Pathway analysis of differentially expressed genes. (A) Significance (Benjamini–Hochberg-adjusted p-values) and normalized enrichment score from GSEA analysis for hallmark pathways found to be enriched in upregulated or downregulated genes based on both fold change and Wald statistic metrics. (B) Volcano plot as in Figure 4A, highlighting calcium-regulated genes in blue.

Figure 6. Mechanisms of regulation of the astrocyte endfoot translatome. (A) Length and (B) GC content of the translatome (black), upregulated genes (red), and downregulated genes. Statistical significance was determined using the Wilcoxon–Mann–Whitney test with Holm–Bonferroni adjustment. (C) Percentage of the translatome (black), upregulated (red, 0%, not visualized), and downregulated (blue) genes that are known to be dependent on Quaking-RNA-binding protein [38]. Statistical significance was determined using the Holm-Bonferroni-adjusted Fisher Exact Test. (D) Transcription factors enriched in upregulated genes relative to whole translatome. (E) Network diagram of transcription factor and gene target interactions generated with R package igraph using the Fruchterman–Reingold layout algorithm. Grey arrows point from transcription factors towards genes they are known to regulate. Only those transcription factor gene targets that are differentially expressed in astrocyte endfeet after MCAo and reperfusion are displayed. (* p [less than] 0.05, ** p [less than] 0.01, *** p [less than] 0.001, **** p [less than] 0.0001).

Figure 7. Upregulation of HSP70 and protein ubiquitination in post-ischemic astrocyte endfeet. (A) Schematic illustrating the enrichment of perivascular endfeet, which involves proteolytic dissociation of isolated microvessels followed by magnetic separation of ACSA2+ endfeet. (B) Representative immunoblots of HSP70, HSC70, GFAP, and the loading control α-tubulin from isolated ACSA2+ endfoot lysates. Densitometric quantification of HSP70 and HSC70 normalized to α-tubulin (CTR vs. Stroke, in grey and red bars respectively) are plotted as mean ± SEM (n = 6 biological replicates). (C) Representative high-powered images of the post-ischemic cortex with immunolabelling for HSP70 (green) and AQP4 (red) (scale bar = 10 µm). (D) Representative immunoblots of ubiquitin and the loading control α-tubulin from isolated ACSA2+ endfoot lysates. (* p [less than] 0.05, ** p [less than] 0.01, Holm-Bonferroni-adjusted paired two-tailed t-test).

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms26010309/s1.

References

1. Virani, S.S.; Alonso, A.; Aparicio, H.J.; Benjamin, E.J.; Bittencourt, M.S.; Callaway, C.W.; Carson, A.P.; Chamberlain, A.M.; Cheng, S.; Delling, F.N. et al. Heart Disease and Stroke Statistics-2021 Update: A Report From the American Heart Association. Circulation; 2021; 143, pp. e254-e743. [DOI: https://dx.doi.org/10.1161/CIR.0000000000000950]

2. Tsao, C.W.; Aday, A.W.; Almarzooq, Z.I.; Anderson, C.A.M.; Arora, P.; Avery, C.L.; Baker-Smith, C.M.; Beaton, A.Z.; Boehme, A.K.; Buxton, A.E. et al. Heart Disease and Stroke Statistics-2023 Update: A Report From the American Heart Association. Circulation; 2023; 147, pp. e93-e621. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36695182]

3. Ahmad, F.B.; Cisewski, J.A.; Anderson, R.N. Leading Causes of Death in the US, 2019–2023. JAMA; 2024; 332, pp. 957-958. [DOI: https://dx.doi.org/10.1001/jama.2024.15563] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39116093]

4. Simard, J.M.; Kent, T.A.; Chen, M.; Tarasov, K.V.; Gerzanich, V. Brain oedema in focal ischaemia: Molecular pathophysiology and theoretical implications. Lancet Neurol.; 2007; 6, pp. 258-268. [DOI: https://dx.doi.org/10.1016/S1474-4422(07)70055-8]

5. Premilovac, D.; Sutherland, B.A. Acute and long-term changes in blood flow after ischemic stroke: Challenges and opportunities. Neural Regen. Res.; 2023; 18, pp. 799-800. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36204841]

6. Lee, H.G.; Lee, J.H.; Flausino, L.E.; Quintana, F.J. Neuroinflammation: An astrocyte perspective. Sci. Transl. Med.; 2023; 15, eadi7828. [DOI: https://dx.doi.org/10.1126/scitranslmed.adi7828]

7. Candelario-Jalil, E.; Dijkhuizen, R.M.; Magnus, T. Neuroinflammation, Stroke, Blood-Brain Barrier Dysfunction, and Imaging Modalities. Stroke; 2022; 53, pp. 1473-1486. [DOI: https://dx.doi.org/10.1161/STROKEAHA.122.036946] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35387495]

8. Iadecola, C.; Buckwalter, M.S.; Anrather, J. Immune responses to stroke: Mechanisms, modulation, and therapeutic potential. J. Clin. Investig.; 2020; 130, pp. 2777-2788. [DOI: https://dx.doi.org/10.1172/JCI135530]

9. Alvarez, J.I.; Katayama, T.; Prat, A. Glial influence on the blood brain barrier. Glia; 2013; 61, pp. 1939-1958. [DOI: https://dx.doi.org/10.1002/glia.22575]

10. MacVicar, B.A.; Newman, E.A. Astrocyte regulation of blood flow in the brain. Cold Spring Harb. Perspect. Biol.; 2015; 7, a020388. [DOI: https://dx.doi.org/10.1101/cshperspect.a020388]

11. Diaz-Castro, B.; Robel, S.; Mishra, A. Astrocyte Endfeet in Brain Function and Pathology: Open Questions. Annu. Rev. Neurosci.; 2023; 46, pp. 101-121. [DOI: https://dx.doi.org/10.1146/annurev-neuro-091922-031205] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36854317]

12. Wang, Y.F.; Parpura, V. Astroglial Modulation of Hydromineral Balance and Cerebral Edema. Front. Mol. Neurosci.; 2018; 11, 204. [DOI: https://dx.doi.org/10.3389/fnmol.2018.00204] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29946238]

13. Iadecola, C. The Neurovascular Unit Coming of Age: A Journey through Neurovascular Coupling in Health and Disease. Neuron; 2017; 96, pp. 17-42. [DOI: https://dx.doi.org/10.1016/j.neuron.2017.07.030] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28957666]

14. Stokum, J.A.; Shim, B.; Negoita, S.; Tsymbalyuk, N.; Tsymbalyuk, O.; Ivanova, S.; Keledjian, K.; Bryan, J.; Blaustein, M.P.; Jha, R.M. et al. Cation flux through SUR1-TRPM4 and NCX1 in astrocyte endfeet induces water influx through AQP4 and brain swelling after ischemic stroke. Sci. Signal; 2023; 16, eadd6364. [DOI: https://dx.doi.org/10.1126/scisignal.add6364]

15. Higashi, T.; Nakai, A.; Uemura, Y.; Kikuchi, H.; Nagata, K. Activation of heat shock factor 1 in rat brain during cerebral ischemia or after heat shock. Brain Res. Mol. Brain Res.; 1995; 34, pp. 262-270. [DOI: https://dx.doi.org/10.1016/0169-328X(95)00163-M]

16. Bergeron, M.; Yu, A.Y.; Solway, K.E.; Semenza, G.L.; Sharp, F.R. Induction of hypoxia-inducible factor-1 (HIF-1) and its target genes following focal ischaemia in rat brain. Eur. J. Neurosci.; 1999; 11, pp. 4159-4170. [DOI: https://dx.doi.org/10.1046/j.1460-9568.1999.00845.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10594641]

17. Ament, S.A.; Poulopoulos, A. The brain’s dark transcriptome: Sequencing RNA in distal compartments of neurons and glia. Curr. Opin. Neurobiol.; 2023; 81, 102725. [DOI: https://dx.doi.org/10.1016/j.conb.2023.102725] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37196598]

18. Boulay, A.C.; Saubamea, B.; Adam, N.; Chasseigneaux, S.; Mazare, N.; Gilbert, A.; Bahin, M.; Bastianelli, L.; Blugeon, C.; Perrin, S. et al. Translation in astrocyte distal processes sets molecular heterogeneity at the gliovascular interface. Cell Discov.; 2017; 3, 17005. [DOI: https://dx.doi.org/10.1038/celldisc.2017.5] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28377822]

19. Stokum, J.A.; Shim, B.; Huang, W.; Kane, M.; Smith, J.A.; Gerzanich, V.; Simard, J.M. A large portion of the astrocyte proteome is dedicated to perivascular endfeet, including critical components of the electron transport chain. J. Cereb. Blood Flow. Metab.; 2021; 41, pp. 2546-2560. [DOI: https://dx.doi.org/10.1177/0271678X211004182]

20. Kameyama, T.; Miyata, M.; Shiotani, H.; Adachi, J.; Kakuta, S.; Uchiyama, Y.; Mizutani, K.; Takai, Y. Heterogeneity of perivascular astrocyte endfeet depending on vascular regions in the mouse brain. iScience; 2023; 26, 108010. [DOI: https://dx.doi.org/10.1016/j.isci.2023.108010] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37829206]

21. Avila-Gutierrez, K.; Slaoui, L.; Alvear-Perez, R.; Kozlowski, E.; Oudart, M.; Augustin, E.; Claveau, C.; Mailly, P.; Monnet, H.; Mignon, V. et al. Dynamic local mRNA localization and translation occurs during the postnatal molecular maturation of perivascular astrocytic processes. Glia; 2024; 72, pp. 777-793. [DOI: https://dx.doi.org/10.1002/glia.24503] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38189217]

22. Sakers, K.; Lake, A.M.; Khazanchi, R.; Ouwenga, R.; Vasek, M.J.; Dani, A.; Dougherty, J.D. Astrocytes locally translate transcripts in their peripheral processes. Proc. Natl. Acad. Sci. USA; 2017; 114, pp. E3830-E3838. [DOI: https://dx.doi.org/10.1073/pnas.1617782114]

23. Mazare, N.; Oudart, M.; Moulard, J.; Cheung, G.; Tortuyaux, R.; Mailly, P.; Mazaud, D.; Bemelmans, A.P.; Boulay, A.C.; Blugeon, C. et al. Local Translation in Perisynaptic Astrocytic Processes Is Specific and Changes after Fear Conditioning. Cell Rep.; 2020; 32, 108076. [DOI: https://dx.doi.org/10.1016/j.celrep.2020.108076] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32846133]

24. Sanz, E.; Yang, L.; Su, T.; Morris, D.R.; McKnight, G.S.; Amieux, P.S. Cell-type-specific isolation of ribosome-associated mRNA from complex tissues. Proc. Natl. Acad. Sci. USA; 2009; 106, pp. 13939-13944. [DOI: https://dx.doi.org/10.1073/pnas.0907143106] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19666516]

25. Doyle, J.P.; Dougherty, J.D.; Heiman, M.; Schmidt, E.F.; Stevens, T.R.; Ma, G.; Bupp, S.; Shrestha, P.; Shah, R.D.; Doughty, M.L. et al. Application of a translational profiling approach for the comparative analysis of CNS cell types. Cell; 2008; 135, pp. 749-762. [DOI: https://dx.doi.org/10.1016/j.cell.2008.10.029]

26. Rakers, C.; Schleif, M.; Blank, N.; Matuskova, H.; Ulas, T.; Handler, K.; Torres, S.V.; Schumacher, T.; Tai, K.; Schultze, J.L. et al. Stroke target identification guided by astrocyte transcriptome analysis. Glia; 2019; 67, pp. 619-633. [DOI: https://dx.doi.org/10.1002/glia.23544]

27. Hernandez, V.G.; Lechtenberg, K.J.; Peterson, T.C.; Zhu, L.; Lucas, T.A.; Bradshaw, K.P.; Owah, J.O.; Dorsey, A.I.; Gentles, A.J.; Buckwalter, M.S. Translatome analysis reveals microglia and astrocytes to be distinct regulators of inflammation in the hyperacute and acute phases after stroke. Glia; 2023; 71, pp. 1960-1984. [DOI: https://dx.doi.org/10.1002/glia.24377]

28. Boulay, A.C.; Saubamea, B.; Decleves, X.; Cohen-Salmon, M. Purification of Mouse Brain Vessels. J. Vis. Exp.; 2015; 10, 53208. [DOI: https://dx.doi.org/10.3791/53208-v]

29. Ito, U.; Hakamata, Y.; Kawakami, E.; Oyanagi, K. Temporary [corrected] cerebral ischemia results in swollen astrocytic end-feet that compress microvessels and lead to delayed [corrected] focal cortical infarction. J. Cereb. Blood Flow. Metab.; 2011; 31, pp. 328-338. [DOI: https://dx.doi.org/10.1038/jcbfm.2010.97] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20588315]

30. Ito, U.; Hakamata, Y.; Kawakami, E.; Oyanagi, K. Degeneration of astrocytic processes and their mitochondria in cerebral cortical regions peripheral to the cortical infarction: Heterogeneity of their disintegration is closely associated with disseminated selective neuronal necrosis and maturation of injury. Stroke; 2009; 40, pp. 2173-2181.

31. Cahoy, J.D.; Emery, B.; Kaushal, A.; Foo, L.C.; Zamanian, J.L.; Christopherson, K.S.; Xing, Y.; Lubischer, J.L.; Krieg, P.A.; Krupenko, S.A. et al. A transcriptome database for astrocytes, neurons, and oligodendrocytes: A new resource for understanding brain development and function. J. Neurosci.; 2008; 28, pp. 264-278. [DOI: https://dx.doi.org/10.1523/JNEUROSCI.4178-07.2008]

32. Zamanian, J.L.; Xu, L.; Foo, L.C.; Nouri, N.; Zhou, L.; Giffard, R.G.; Barres, B.A. Genomic analysis of reactive astrogliosis. J. Neurosci.; 2012; 32, pp. 6391-6410. [DOI: https://dx.doi.org/10.1523/JNEUROSCI.6221-11.2012]

33. Sun, W.; Cornwell, A.; Li, J.; Peng, S.; Osorio, M.J.; Aalling, N.; Wang, S.; Benraiss, A.; Lou, N.; Goldman, S.A. et al. SOX9 Is an Astrocyte-Specific Nuclear Marker in the Adult Brain Outside the Neurogenic Regions. J. Neurosci.; 2017; 37, pp. 4493-4507. [DOI: https://dx.doi.org/10.1523/JNEUROSCI.3199-16.2017]

34. Viscardi, M.J.; Arribere, J.A. Poly(a) selection introduces bias and undue noise in direct RNA-sequencing. BMC Genom.; 2022; 23, 530. [DOI: https://dx.doi.org/10.1186/s12864-022-08762-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35869428]

35. Kislin, M.; Sword, J.; Fomitcheva, I.V.; Croom, D.; Pryazhnikov, E.; Lihavainen, E.; Toptunov, D.; Rauvala, H.; Ribeiro, A.S.; Khiroug, L. et al. Reversible Disruption of Neuronal Mitochondria by Ischemic and Traumatic Injury Revealed by Quantitative Two-Photon Imaging in the Neocortex of Anesthetized Mice. J. Neurosci.; 2017; 37, pp. 333-348. [DOI: https://dx.doi.org/10.1523/JNEUROSCI.1510-16.2016] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28077713]

36. Struk, T.; Nair, V.; Eichinger, F.; Kretzler, M.; Wedlich-Soldner, R.; Bayraktar, S.; Pavenstadt, H. Transcriptome analysis of primary podocytes reveals novel calcium regulated regulatory networks. FASEB J.; 2020; 34, pp. 14490-14506. [DOI: https://dx.doi.org/10.1096/fj.201902493RR]

37. Buniatian, G.H.; Hartmann, H.J.; Traub, P.; Wiesinger, H.; Albinus, M.; Nagel, W.; Shoeman, R.; Mecke, D.; Weser, U. Glial fibrillary acidic protein-positive cells of the kidney are capable of raising a protective biochemical barrier similar to astrocytes: Expression of metallothionein in podocytes. Anat. Rec.; 2002; 267, pp. 296-306. [DOI: https://dx.doi.org/10.1002/ar.10115]

38. Sakers, K.; Liu, Y.; Llaci, L.; Lee, S.M.; Vasek, M.J.; Rieger, M.A.; Brophy, S.; Tycksen, E.; Lewis, R.; Maloney, S.E. et al. Loss of Quaking RNA binding protein disrupts the expression of genes associated with astrocyte maturation in mouse brain. Nat. Commun.; 2021; 12, 1537. [DOI: https://dx.doi.org/10.1038/s41467-021-21703-5]

39. Liska, O.; Bohar, B.; Hidas, A.; Korcsmaros, T.; Papp, B.; Fazekas, D.; Ari, E. TFLink: An integrated gateway to access transcription factor-target gene interactions for multiple species. Database; 2022; 2022, baac083. [DOI: https://dx.doi.org/10.1093/database/baac083]

40. Popovic, N.; Hooker, E.; Barabino, A.; Flamier, A.; Provost, F.; Buscarlet, M.; Bernier, G.; Larrivee, B. COCO/DAND5 inhibits developmental and pathological ocular angiogenesis. EMBO Mol. Med.; 2021; 13, e12005. [DOI: https://dx.doi.org/10.15252/emmm.202012005] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33587337]

41. Barton, S.K.; Gregory, J.M.; Chandran, S.; Turner, B.J. Could an Impairment in Local Translation of mRNAs in Glia be Contributing to Pathogenesis in ALS?. Front. Mol. Neurosci.; 2019; 12, 124. [DOI: https://dx.doi.org/10.3389/fnmol.2019.00124] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31164803]

42. Mosser, D.D.; Caron, A.W.; Bourget, L.; Meriin, A.B.; Sherman, M.Y.; Morimoto, R.I.; Massie, B. The chaperone function of hsp70 is required for protection against stress-induced apoptosis. Mol. Cell Biol.; 2000; 20, pp. 7146-7159. [DOI: https://dx.doi.org/10.1128/MCB.20.19.7146-7159.2000]

43. van der Weerd, L.; Lythgoe, M.F.; Badin, R.A.; Valentim, L.M.; Akbar, M.T.; de Belleroche, J.S.; Latchman, D.S.; Gadian, D.G. Neuroprotective effects of HSP70 overexpression after cerebral ischaemia--an MRI study. Exp. Neurol.; 2005; 195, pp. 257-266. [DOI: https://dx.doi.org/10.1016/j.expneurol.2005.05.002] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15936758]

44. Bechtold, D.A.; Rush, S.J.; Brown, I.R. Localization of the heat-shock protein Hsp70 to the synapse following hyperthermic stress in the brain. J. Neurochem.; 2000; 74, pp. 641-646. [DOI: https://dx.doi.org/10.1046/j.1471-4159.2000.740641.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10646515]

45. Willis, D.; Li, K.W.; Zheng, J.Q.; Chang, J.H.; Smit, A.B.; Kelly, T.; Merianda, T.T.; Sylvester, J.; van Minnen, J.; Twiss, J.L. Differential transport and local translation of cytoskeletal, injury-response, and neurodegeneration protein mRNAs in axons. J. Neurosci.; 2005; 25, pp. 778-791. [DOI: https://dx.doi.org/10.1523/JNEUROSCI.4235-04.2005] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15673657]

46. Lindquist, S. The heat-shock response. Annu. Rev. Biochem.; 1986; 55, pp. 1151-1191. [DOI: https://dx.doi.org/10.1146/annurev.bi.55.070186.005443] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/2427013]

47. Morimoto, R.I. Regulation of the heat shock transcriptional response: Cross talk between a family of heat shock factors, molecular chaperones, and negative regulators. Genes. Dev.; 1998; 12, pp. 3788-3796. [DOI: https://dx.doi.org/10.1101/gad.12.24.3788] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9869631]

48. Bjork, J.K.; Sistonen, L. Regulation of the members of the mammalian heat shock factor family. FEBS J.; 2010; 277, pp. 4126-4139. [DOI: https://dx.doi.org/10.1111/j.1742-4658.2010.07828.x]

49. Fujimoto, M.; Nakai, A. The heat shock factor family and adaptation to proteotoxic stress. FEBS J.; 2010; 277, pp. 4112-4125. [DOI: https://dx.doi.org/10.1111/j.1742-4658.2010.07827.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20945528]

50. Sharp, F.R.; Zhan, X.; Liu, D.Z. Heat shock proteins in the brain: Role of Hsp70, Hsp 27, and HO-1 (Hsp32) and their therapeutic potential. Transl. Stroke Res.; 2013; 4, pp. 685-692. [DOI: https://dx.doi.org/10.1007/s12975-013-0271-4]

51. Sharp, F.R.; Lu, A.; Tang, Y.; Millhorn, D.E. Multiple molecular penumbras after focal cerebral ischemia. J. Cereb. Blood Flow. Metab.; 2000; 20, pp. 1011-1032. [DOI: https://dx.doi.org/10.1097/00004647-200007000-00001] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10908035]

52. Kim, J.Y.; Kim, J.W.; Yenari, M.A. Heat shock protein signaling in brain ischemia and injury. Neurosci. Lett.; 2020; 715, 134642. [DOI: https://dx.doi.org/10.1016/j.neulet.2019.134642] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31759081]

53. Thiebaut, A.M.; Hedou, E.; Marciniak, S.J.; Vivien, D.; Roussel, B.D. Proteostasis During Cerebral Ischemia. Front. Neurosci.; 2019; 13, 637. [DOI: https://dx.doi.org/10.3389/fnins.2019.00637] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31275110]

54. Liu, C.L.; Martone, M.E.; Hu, B.R. Protein ubiquitination in postsynaptic densities after transient cerebral ischemia. J. Cereb. Blood Flow. Metab.; 2004; 24, pp. 1219-1225. [DOI: https://dx.doi.org/10.1097/01.WCB.0000136706.77918.21] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15545915]

55. Shih, H.J.; Chen, C.L.; Torng, P.L. IGFBP3 inhibits angiogenesis through intracellular regulation of THBS1 expression. Am. J. Cancer Res.; 2020; 10, pp. 1728-1744.

56. Bender, H.R.; Campbell, G.E.; Aytoda, P.; Mathiesen, A.H.; Duffy, D.M. Thrombospondin 1 (THBS1) Promotes Follicular Angiogenesis, Luteinization, and Ovulation in Primates. Front. Endocrinol.; 2019; 10, 727. [DOI: https://dx.doi.org/10.3389/fendo.2019.00727] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31787928]

57. Yu, E.; Lariviere, R.; Thomas, R.A.; Liu, L.; Senkevich, K.; Rahayel, S.; Trempe, J.F.; Fon, E.A.; Gan-Or, Z. Machine learning nominates the inositol pathway and novel genes in Parkinson’s disease. Brain; 2024; 147, pp. 887-899. [DOI: https://dx.doi.org/10.1093/brain/awad345]

58. Zhang, Y.; Xu, C.; Nan, Y.; Nan, S. Microglia-Derived Extracellular Vesicles Carrying miR-711 Alleviate Neurodegeneration in a Murine Alzheimer’s Disease Model by Binding to Itpkb. Front. Cell Dev. Biol.; 2020; 8, 566530. [DOI: https://dx.doi.org/10.3389/fcell.2020.566530]

59. Rajkovic, I.; Wong, R.; Lemarchand, E.; Rivers-Auty, J.; Rajkovic, O.; Garlanda, C.; Allan, S.M.; Pinteaux, E. Pentraxin 3 promotes long-term cerebral blood flow recovery, angiogenesis, and neuronal survival after stroke. J. Mol. Med.; 2018; 96, pp. 1319-1332. [DOI: https://dx.doi.org/10.1007/s00109-018-1698-6] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30315331]

60. Rodriguez-Grande, B.; Varghese, L.; Molina-Holgado, F.; Rajkovic, O.; Garlanda, C.; Denes, A.; Pinteaux, E. Pentraxin 3 mediates neurogenesis and angiogenesis after cerebral ischaemia. J. Neuroinflamm.; 2015; 12, 15. [DOI: https://dx.doi.org/10.1186/s12974-014-0227-y]

61. Shindo, A.; Maki, T.; Mandeville, E.T.; Liang, A.C.; Egawa, N.; Itoh, K.; Itoh, N.; Borlongan, M.; Holder, J.C.; Chuang, T.T. et al. Astrocyte-Derived Pentraxin 3 Supports Blood-Brain Barrier Integrity Under Acute Phase of Stroke. Stroke; 2016; 47, pp. 1094-1100. [DOI: https://dx.doi.org/10.1161/STROKEAHA.115.012133]

62. Ouchi, N.; Higuchi, A.; Ohashi, K.; Oshima, Y.; Gokce, N.; Shibata, R.; Akasaki, Y.; Shimono, A.; Walsh, K. Sfrp5 is an anti-inflammatory adipokine that modulates metabolic dysfunction in obesity. Science; 2010; 329, pp. 454-457. [DOI: https://dx.doi.org/10.1126/science.1188280] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20558665]

63. Ding, N.; Zheng, C. Secreted frizzled-related protein 5 promotes angiogenesis of human umbilical vein endothelial cells and alleviates myocardial injury in diabetic mice with myocardial infarction by inhibiting Wnt5a/JNK signaling. Bioengineered; 2022; 13, pp. 11656-11667. [DOI: https://dx.doi.org/10.1080/21655979.2022.2070964]

64. Hartman, B.H.; Nelson, B.R.; Reh, T.A.; Bermingham-McDonogh, O. Delta/notch-like EGF-related receptor (DNER) is expressed in hair cells and neurons in the developing and adult mouse inner ear. J. Assoc. Res. Otolaryngol.; 2010; 11, pp. 187-201. [DOI: https://dx.doi.org/10.1007/s10162-009-0203-x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20058045]

65. Dobin, A.; Davis, C.A.; Schlesinger, F.; Drenkow, J.; Zaleski, C.; Jha, S.; Batut, P.; Chaisson, M.; Gingeras, T.R. STAR: Ultrafast universal RNA-seq aligner. Bioinform.; 2013; 29, pp. 15-21. [DOI: https://dx.doi.org/10.1093/bioinformatics/bts635]

66. Altschul, S.F.; Gish, W.; Miller, W.; Myers, E.W.; Lipman, D.J. Basic local alignment search tool. J. Mol. Biol.; 1990; 215, pp. 403-410. [DOI: https://dx.doi.org/10.1016/S0022-2836(05)80360-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/2231712]

67. Love, M.I.; Huber, W.; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol.; 2014; 15, 550. [DOI: https://dx.doi.org/10.1186/s13059-014-0550-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25516281]

68. Stephens, M. False discovery rates: A new deal. Biostatistics; 2017; 18, pp. 275-294. [DOI: https://dx.doi.org/10.1093/biostatistics/kxw041]

69. Serguishichev, A.A. An algorithm for fast preranked gene set enrichment analysis using cumulative statistic calculation. bioRxiv; 2016; pp. 1-9.

70. Subramanian, A.; Tamayo, P.; Mootha, V.K.; Mukherjee, S.; Ebert, B.L.; Gillette, M.A.; Paulovich, A.; Pomeroy, S.L.; Golub, T.R.; Lander, E.S. et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA; 2005; 102, pp. 15545-15550. [DOI: https://dx.doi.org/10.1073/pnas.0506580102]

71. Mootha, V.K.; Lindgren, C.M.; Eriksson, K.F.; Subramanian, A.; Sihag, S.; Lehar, J.; Puigserver, P.; Carlsson, E.; Ridderstrale, M.; Laurila, E. et al. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet.; 2003; 34, pp. 267-273. [DOI: https://dx.doi.org/10.1038/ng1180] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12808457]

72. Castanza, A.S.; Recla, J.M.; Eby, D.; Thorvaldsdottir, H.; Bult, C.J.; Mesirov, J.P. Extending support for mouse data in the Molecular Signatures Database (MSigDB). Nat. Methods; 2023; 20, pp. 1619-1620. [DOI: https://dx.doi.org/10.1038/s41592-023-02014-7]

73. Wickham, H. ggplot2: Elegant Graphics for Data Analysis; 2nd ed Springer-Verlag: New York, NY, USA, 2016.