HTRA1 deficiency in COL4A1 mutant hiPSC-derived astrocytes, a convergent mechanism of cerebral small vessel disease

Cerebral small vessel disease ( cSVD) is a major contributor to stroke and cognitive decline, ultimately leading to vascular dementia (VaD). Genetic factors play a key role in the disease susceptibility and progression, and variants in COL4A1 cause one of the most common genetic cSVD. COL4A1 encodes the 1 subunit of type IV collagen, the principle extracellular matrix (ECM) protein in the basement membrane of vasculature. In the central nervous system (CNS), the neurovascular unit (NVU) has the unique astrocyte-derived parenchymal basement membrane (pBM), in addition to the vascular basement membrane (vBM), which together contributing to the regulation of the blood- brain barrier (BBB) function. However, the role of pBM in c SVD remains under investigated and poorly understood. The lack of relevant human models has limited our ability to dissect specific cell-cell and cell-matrix interactions, hindering the identification of effective therapeutic targets. In this study, we hypothesi sed that astrocyte-mediated ECM remodelling contributes to BBB dysfunction in COL4A1-associated c SVD. To investigate this, human induced pluripotent stem cells (hiPSCs) derived from a patient carrying the COL4A1G755R variant and its isogenic control line were differentiated into astrocytes and brain microvascular endothelial cells (BMECs). Comparing to isogenic controls, the COL4A1G755R astrocytes significantly reduced the expression of ECM- related genes and abnormally increased glutamate uptake. ECM preparations from COL4A1G755R astrocytes significantly damaged the tight junction (TJ) structure formed by control iPSC-derived BMECs and failed to rescue the compromised TJ integrity in COL4A1G755R BMECs. The secretome from COL4A1G755R astrocytes exaggerated the ECM abnormality in COL4A1G755R BMECs. Most importantly, reduced expression of HTRA1, a crucial serine protease known to regulate both ECM turnover and homeostasis, and increased TGF-β signalling was observed in COL4A1G755R astrocytes. Functional rescue by recombinant human HTRA1 protein restored the disrupted TJ continuity in COL4A1G755R BMECs and normalized TGF-β signalling and glutamate uptake in astrocytes. Together, these findings defined a previously unrecognised astrocyte-driven pBM mechanism in COL4A1-associated cSVD and highlight HTRA1 in ECM remodelling as a therapeutic target.

COL4A1, Cerebral small vessel disease, HTRA1, extracellular matrix, blood- brain barrier, human induced pluripotent stem cells. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 3

Cerebral small vessel disease ( cSVD) is the leading cause of stroke and vascular cognitive impairment, contributing to approximately 25% of all strokes and nearly 50% of dementia cases globally1-3. It accounts for up to 20% of ischemic strokes and a significant proportion of intracerebral haemorrhages (ICH) worldwide 4. With a trend of global aging, the burden of cSVD is expected to increase substantially in coming decades 5. The aetiology of cSVD is heterogeneous with the two major subtypes being arteriosclerosis resulting in “sporadic SVD ”, for which the major risk factors include hypertension, diabetes and smoking, and cerebral amyloid angiopathy (CAA) related to amyloid deposition in the more superficial cerebral vessels. Monogenic causes of SVD are also being increasingly recognised, the most common of which are due to mutations in the NOTCH3, HTRA1, and COL4A1/2 genes. Increasing evidence suggests these mutations, and many of the common genetic variants also associated with sporadic cSVD,6-8 share disease mechanisms converging on disruption of the extracellular matrix (ECM) and associated proteins comprising the matrisome .2, 9 This raises the possibility that therapeutically targeting matrisome disruption could offer novel treatment approaches. To achieve this more detailed understanding of the underlying molecular pathways is required. COL4A1 and COL4A2 encode the 1 and 2 subunits of type IV collagen , which assemble into α1α1α2 heterotrimer via the Gly-X-Y motif.10, 11 Collagen IV is the key ECM component of the basement membrane (BM) in blood vessels. They form a covalently cross-linked network that connects to the laminin scaffold via bridging proteins, including nidogens and heparan sulphate proteoglycans (perlecan).12 Interestingly, in the central nervous system (CNS), two distinct BMs exist: the vascular BM (vBM) and parenchymal (or glial) BM (pBM) 13, 14. The vBM, produced mainly by brain microvascular endothelial cells (BMECs) and mural cells (MCs), is enriched in collagen IV 1/2, laminin α4, perlecan, and nidogen-1 13, 15. In contrast, The pBM, is primarily secreted by astrocyte end-feet and mainly contains laminin α2 and α1, nidogen-2, collagen IV 1/2, and astrocyte-derived ECM proteins such as tenascin-C and SPARC-like 1 13. While the vBM has been extensively studies as a determinant of the blood-brain barrier (BBB) integrity,13, 16 the pBM remains comparatively overlooked. As the most abundant glial cells in the CNS, astrocytes are essential for neurovascular coupling and BBB function by direct interaction with BMECs in cerebral microvessels , but the contribution of astrocyte- secreted pBM to cSVD has not attracted wide attention. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 4 This gap is particularly relevant in the context of pathogenic variants in COL4A1/A2 (Gould syndrome ), which cause a spectrum of cSVD phenotypes , including stroke, intracranial haemorrhage, and white matter damage.17-19 While both vBMs and pBM are central in supporting the function of neurovascular unit (NVU) that is composed of multiple neurovascular cell types including BMECs, MCs, astrocytes and neurons , essential for perivascular homeostasis,15, 20, 21 however, how the mutant collagen leads to these brain pathologies remains unclear. Using induced pluripotent stem cells (iPSCs) from COL4A1/COL4A2 cSVD patients, our previous study reported detrimental effects of the mutant MC-secreted vBM in supporting BBB function 22, 23. Using the same iPSC model, this study has focused on the role of astrocytes and astrocyte -secreted vBM components in the disease pathology. We identified global ECM disorganisation in astrocytes derived from iPSCs (iPSC-astrocytes) of the COL4A1G755R cSVD patient. ECM preparations from the mutant astrocytes significantly damaged tight junction (TJs) of iPSC-derived BMECs (iPSC-BMECs), and secretome from the mutant astrocytes reduced its ability to rescue the integrity of TJ in mutant BMECs , comparing to that of the isogenic control . Importantly, we identified HTRA1, a serine protease involved in multiple genetic cSVDs, that was significantly downregulated in COL4A1G755R iPSC-astrocytes. Treatment of COL4A1G755R iPSC- BMECs with recombinant human HTRA1 significantly reversed the BBB integrity. Our findings provide new insights into the role of astrocytes and the pBM in the regulation of BBB function and driving cSVD pathology, which may inform the development of targeted therapies.

Transcriptomics of COL4A1G755R iPSC-astrocytes highlight ECM pathology The COL4A1G755R and isogenic control iPSC lines were reported in a previous publication 23. The COL4A1G755R variant and pluripotency of iPSCs were also confirmed in this study (Fig. S1 ). We first differentiated iPSCs into astrocytes using a two-stage protocol involving neural progenitor cell (NPC) induction followed by astrocyte differentiation as illustrated in Figure 1A. Successful differentiation of NPCs from b oth the mutant and isogenic control iPSCs were confirmed by their rosette-like structure and the expression of NPC-specific markers, PAX6 and SOX1 (Fig. 1B and C). The subsequent astrocyte differentiation from iPSC-NPCs was carried out and characterised on differentiation Day 60 (Fig. 1). The “astro” shape d astrocytes (Fig. 1B) gradually emerged during differentiation. Immunostaining and RT-qPCR demonstrated robust expression of glial .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 5 fibrillary acidic protein (GFAP) and S100 calcium-binding protein B (S100B), confirming the identity of iPSC-astrocytes (Fig. 1B and C). To evaluate the functional maturity of iPSC-astrocytes, we performed glutamate uptake assay (Fig. 1D). Cells were exposed to a defined amount of glutamate for two hours followed by colorimetric quantification glutamate in both cell lysates and medium, where increased cellular glutamate and concomitant depletion from the medium indicated active glutamate uptake. Result showed that astrocytes derived from both the mutant and isogenic control iPSCs had the ability of up-taking glutamate from the medium (Fig. 1D). Interestingly, the COL4A1G755R mutant astrocytes had significantly enhanced glutamate uptake compared to the isogenic control (Fig. 1D), suggesting a potential metabolic stress or a reactive phenotype. Calcium imaging of the iPSC derived astrocytes revealed spontaneous baseline calcium oscillation and responses to glutamate stimulation (Fig. 1E and S2). These assessments on cell morphology, marker gene expression and functional assay validated the iPSC-astrocytes to be used for modelling astrocyte-related phenotypes. To identify molecular mechanisms of COL4A1-related cSVD, global transcriptomic approach was conducted using RNA sequencing (RNAseq). PCA plot revealed a clear separation between samples of COL4A1G755R-astrocytes and isogenic control s on PC1 that explains 67% of the total variance, indicating a distinct transcriptomic signature of the mutant astrocytes (Fig. 2A). Data analysis identified 1,284 significantly differentially expressed genes (DEGs) between the two groups, of which 833 were upregulated and 451 downregulated (Fig. 2B). Gene otology ( GO) analysis revealed significant ECM- related alterations across all three major GO categories (Fig. 2C). The most significantly enriched GO terms were “extracellular matrix organization” and “extracellular structure organization” under the biological process (BP) category, underscoring a central defect in ECM pathways in mutant astrocytes. We then conducted GSEA analysis to uncover the coordinated changes of genes involved in ECM function. Results show that gene sets involved in “collagen fibril organization” and “regulation of extracellular matrix organization” were negatively enriched (Fig. 2D). Defects of pBM in COL4A1G755R iPSC-astrocytes and associated HTRA1 downregulation and TGF-β signalling dysregulation To further dissect ECM alterations in the COL4A1G755R astrocytes, we analysed DEGs under the GO term “extracellular matrix organization ” as shown on t he heatmap ( Fig. 2E), and these DEGs were also annotated on the volcano plot (Fig. 2B). The DEG list contains key genes encoding pBM proteins secreted by astrocytes including LAMC1, NID2, FBLN2, FBLN5, FBN1, and ECM remodelling genes such as MMP9 and HTRA1, .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 6 most of which were downregulated. We then verified the RNAseq results for some of these genes using RT-qPCR (Fig. 2F). Western blotting (WB) confirmed that Nidogen 2 (encoded by NID2), the pBM -specific nidogen type, was significantly reduced in COL4A1G755R astrocytes (Fig. S3A), and immunofluorescent staining of ECM secreted from the COL4A1G755R astrocytes revealed spar se nidogen signal, comparing to the isogenic control (Fig. S3B). COL4A1 expression was not significantly altered (Fig. S3C), since the COL4A1G755R variant is predicted to affect posttranscriptional assembly of the collagen triple -helix; however, its secretion to extracellular space tends to be compromised (Fig. S3B), which is in line with the previous finding23. Among the differentially expressed pBM genes, HTRA1 attracted our specific interest as it is a causative gene for monogenic cSVDs, with homozygous mutation s resulting in CARASIL (cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy), while heterozygous mutations result in a milder clinical phenotype in CADASIL-2 (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy-2).24, 25 HTRA1 is a serine protease that plays a key role in maintaining the homeostasis of the BM by degrading misfolded ECM proteins and regulating TGF-β activity26. Using RT-qPCR and WB we confirmed a significant down regulation of HTRA1 in COL4A1G755R astrocytes (Fig. 3A and B). Immunofluorescence staining of both iPSC-astrocytes and ECM preparations also revealed diminishing of HTRA1 around the COL4A1G755R astrocytes and in the ECM (Fig. 3C). Concomitantly, a marked increase of ECM-associated TGF-β1 deposition was observed in the ECM preparation from COL4A1G755R astrocytes compared to the isogenic contro l (Fig. 3D), suggesting increased extracellular availability of TGFβ1 and potential activation of the TGF-β signalling. This was confirmed by phosphorylated Smad2/3 (P-Smad2/3) staining which demonstrated a distinct shift of P -Smad2/3 from cytoplasmic to predominant nuclear localisation in COL4A1G755R astrocytes (Fig. 3E and F), a hallmark of canonical TGF-β signalling activation. RNAseq analysis also highlighted increased expression of TGFB2 (Fig. 2E), and RT-qPCR confirmed the elevation of both TGFB1 and TGFB2, and decreased LTBP1 (Fig. 3G). ECM perpetrations from COL4A1G755R iPSC-astrocytes damage s TJs of iPSC- BMECs As the astrocyte deposited pBM is a key regulator of BMEC function and BBB integrity, we evaluated the direct effect of the COL4A1G755R astrocyte-secreted ECM on BMECs. We differentiated iPSCs into BMECs using a protocol modified from our previous publication 27 and the publication by Pediaditakis et al 28 (Fig. S4A-C). We first confirmed previous findings on the same iPSC model where TJs in COL4A1G755R BMECs were .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 7 significantly damaged as stained for claudin 5, occluding and ZO-1, and trans-endothelial electrical resistance (TEER) was also significantly reduced compared to the isogenic control ( Fig. S4D and E) 23. We then grew iPSC-BMECs on the ECM substrates synthesised by iPSC-astrocytes. Results showed that when the isogenic control BMECs grown on the ECM from COL4A1G755R mutant astrocytes, they developed significantly disrupted TJ architecture characterised by irregular, discontinuous, and punctate staining patterns, compared to control BMECs growing on ECM preparations from the control astrocytes (Fig. 4A and B), suggesting a detrimental effect of the mutant astrocyte ECM on BBB integrity. We then grew the COL4A1G755R mutant BMECs on ECM preparations of the isogenic control or mutant iPSC -astrocytes and found that the control astrocyte ECM significantly restored the TJ integrity of the mutant BMECs (Fig. 4C and D), suggesting the importance of a healthy pBM in supporting BBB function. iPSC-astrocyte secretome enhances BMEC barrier function with reduced efficacy from COL4A1G755R iPSC-astrocytes To better understand the contribution of astrocyte-derived soluble factors, in addition to ECM, to the disease mechanism in COL4A1-associated cSVD, we investigated the effects of the astrocyte secretome on BBB integrity. We cultured isogenic control iPSC- BMECs in conditioned medium from either the mutant COL4A1G755R iPSC-astrocytes or the isogenic control iPSC -astrocytes in a Trans-well insert (Corning) for four days .

showed that while the TJ continuity of the control BMECs were not further enhanced by either the control or mutant astrocyte secretome (Fig. 5A and B), both the control and mutant astrocyte secretomes i nduced significantly increased protein levels of ZO-1, occludin and claudin-5 in control BMECs, but this effect was significantly weaker by the mutant astrocyte secretome (Fig. 5C). These results suggest that the secretome of astrocytes carrying the COL4A1G755R mutation did not completely lose their ability to support BMEC TJ integrity, but their supportive capacity is compromised . Notably, the mRNA expression of ZO-1, CLND5 and OCDN in control BMECs were unchanged when cultured in conditioned medium from either the control or mutant astrocytes (Fig. 5E), suggesting a posttranslational mechanism. We next examined the effects of secretomes from the isogenic control and mutant iPSC- astrocyte on the TJ integrity of the COL4A1G755R mutant BMECs using the same Trans- well co-culture setting. Both secretomes could significantly restore the disorganised TJs of the mutant BMECs, but again, the extent of the restoration by the mutant astrocyte secretome was significantly compromised compared to that of the control (Fig. 5A and D). However, both the healthy and mutant astrocyte secretomes were not able to increase TJ protein levels in COL4A1G755R mutant BMECs, suggesting an intrinsic or .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 8 autonomous defect on the regulation of TJ proteins in mutant BMECs (Fig. 5F). This finding, taking the TJ-rescue capability by the mutant astrocyte secretome (Fig. 5A and C) into account, also indicate that the observed improvement in TJ organisation is likely mediated by redistribution or reorganisation of existing TJ proteins rather than changes in their overall levels. Accordingly, the mRNA expression of ZO-1, CLND5 and OCDN remained unchanged in mutant BMECs when cultured in conditioned medium from either the mutant or control astrocytes (Fig. 5G). COL4A1G755R iPSC-astrocyte secretome exaggerates the ECM dysregulation in COL4A1G755R BMECs To model the disease -relevant microenvironment of COL4A1-related CSVD, we performed bulk RNA sequencing on COL4A1G755R iPSC-BMECs cultured in conditioned medium from COL4A1G755R iPSC-astrocytes, alongside isogenic control iPSC-BMECs exposed to isogenic astrocyte-conditioned medium . Differential expression analysis identified 656 DEGs, including 395 upregulated and 261 downregulated transcripts in COL4A1G755R BMECs (Fig. 6A). In contrast, the DEG numbers were much larger on the RNA-seq results comparing the monoculture of COL4A1G755R ipSC-BMECs and control iPSC-BMECs where the cells were not exposed to their respective astrocyte conditioned medium (300 upregulated and 663 downregulated DEGs) (Fig. 6B). This suggests that both control and mutant secretomes have reduced the global transcriptomic variability between the mutant and control iPSC-BMECs, which is beneficial to BMECs, in line with our finding in ( Fig. 5A and C ) above. GO analysis of the astrocyte-medium-exposed mutant and control iPSC-BMECs enriched changes on GO terms of “extracellular matrix organisation”, “extracellular structure organisation ”, and “collagen-containing extracellular matrix” (Fig. 6C), which are similar to the RNA -seq results found in the mutant astrocytes mentioned earlier (Fig. 2C). However, these matrix-related GO terms were not significantly enriched in RNA-seq results comparing transcriptomics between the monocultured COL4A1G755R iPSC-BMECs and control iPSC-BMECs that were not exposed to their respective astrocyte conditioned medium (Fig. S5A). Notably, HTAR1 appeared among the significantly downregulated gene set under the GO term “extracellular matrix organisation” (Fig. 6A and D), which was confirmed by RT-qPCR and western blotting (Fig. S6A-C). In the RNA-seq data on monocultured COL4A1G755R BMECs, HTRA1 was not statistically significant ly downregulated (adjusted P = 0.22), although there was a trend of decrease ( Fig. 6B ), but s ubsequent independent experiments using RT -qPCR did show statistically significant decrease of HTRA1 expression in monocultured COL4A1G755R BMECs (Fig. S 6B). Interestingly, the COL4A1G755R iPSC-astrocyte conditioned medium has driven the HTRA1 downregulation .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 9 from 29-fold (Fig. S6A) to 48-fold (Fig. S6B), suggesting a detrimental role of the mutant astrocyte secretome in regulating HTRA1 in BMECs. Additionally, among the significantly altered ECM-related transcripts, MMP9 was upregulated in COL4A1G755R BMECs exposed to conditioned medium from the mutant astrocytes (Fig. 6A and S6D) comparing to that of the control, implicating the likelihood of matrix degradation and subsequent BBB disruption. However, similar to the HTRA1 result, MMP9 were not among the significantly altered DEGs in RNA -seq data on monocultured COL4A1G755R and control iPSC-BMECs (Fig. 6B ), but it was found statistically increased in separate RT-qPCR experiments (Fig. S 6E). H owever, t he mutant astrocyte conditioned medium has driven the MMP9 expression from ~72-fold (Fig. S6E ) to an average of 327 -fold in the mutant iPSC-BMECs (Fig. S 6D). This phenomenon suggests that both cell autonomous effects caused by the COL4A1 mutation and the mutant astrocyte -derived signals synergistically amplify pathological changes in BMECs. Taken together, these data support a model in which astrocytes dysfunction significantly drives ECM related ECM pathology in COL4A1-associated SVD. The results also emphasise the importance of employing multicellular co-culture system in cSVD disease mechanism studies. HTRA1 treatment rescue s impaired TJs of COL4A1G755R iPSC-BMECs and improves astrocyte function Given the observed downregulation of HTRA1 in COL4A1G755R iPSC-derived astrocytes and BMECs and its established role in ECM remodelling and signalling regulation, as well as its involvement in other genetic cSVD, we hypothesised that correction of HTRA1 deficiency could rescue the disruption of TJ integrity, a hallmark feature of BBB dysfunction in COL4A1-related cSVD. To test this hypothesis, recombinant human HTRA1 (20 ng/ml) was administered to COL4A1G755R iPSC-derived BMECs for 48 hours. Immunofluorescence analysis revealed that HTRA1 treatment significantly rescued the TJ architecture (Fig. 7). The untreated COL4A1G755R BMECs displayed fragmented and discontinuous TJ patterns of TJ proteins including ZO -1, occludin, and claudin -5, and often interrupted at fusion points . I n contrast, HTRA1 treated cells exhibited a more uniform and linear staining pattern (Fig. 7A). To further investigate the mechanism underlying the protective role of BBB by HTRA1, we investigated the involvement of TGFβ signalling as hypothesised earlier. RT-qPCR analysis showed that HTRA1 supplementation significantly suppressed the elevated mRNA expression of TGFB1, TGFB2, and TGFBR2 in COL4A1G755R iPSC-BMECs (Fig. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 10 8A). We then tested if TGF-β receptor inhibitor could replicate the rescue effect of HTRA1 by applying SB431542 (10 μM) to the COL4A1G755R iPSC-BMEC culture. To our expectation, immunofluorescence analysis demonstrated that SB431542 treatment has significantly improved TJ organization, as evidenced by the improved continuity of ZO-1 and occludin staining at cell boundaries and fusion points (Fig. 8B-D). Next, we determined if HTRA1 could rescue astrocyte function in COL4A1 cSVD. Interestingly, supplementation of recombinant human HTRA1 protein to astrocyte culture significantly reversed the abnormally enhanced glutamate uptake by the COL4A1G755R iPSC-astrocytes (Fig. 8E and F ). Remarkably, the reversal reached to the level comparable to that of the isogenic control iPSC-astrocytes. HTRA1 treatment did not significant change glutamate uptake by the isogenic control iPSC-astrocytes (Fig. 8F), suggesting that supplementation specifically corrects the deficit in the mutant astrocyte without altering glutamate transport in healthy astrocytes. In contrast, inhibition of TGF- β signalling by SB431542 resulted in a significant enhancement of glutamate uptake in both COL4A1G755R and the isogenic control iPSC -astrocytes, rather than rescue the abnormally enhanced glutamate by the mutant astrocytes (Fig. S 7), suggesting an alternative mechanism underlying the damaged astrocyte function , other than through TGF-β signalling. Together, these findings suggest that reduced HTRA1 expression in COL4A1G755R BMECs contributes to BBB dysfunction at least in part via upregulation of TGF -β signalling, and in turn disrupts TJ protein expression. Restoration of HTRA1 activity or pharmacological inhibition of the TGF -β pathway may therefore represent promising therapeutic strategies for COL4A1-related small vessel disease.

Using iPSCs derived from COL4A1 cSVD patient, this study investigated how the COL4A1G755R variant altered astrocyte function and subsequently affected BBB integrity. Transcriptomic profiling highlighted ECM dysregulation in the COL4A1G755R iPSC derived astrocytes as compared to its isogenic control, which emphasised the importance of the pBM in COL4A1 cSVD pathology. Importantly, we identified downregulation of HTRA1 in COL4A1G755R iPSC derived astrocytes and BMECs, and supplement of human recombinant HTRA1 rescued the BBB integrity and astrocyte function, suggesting a potentially novel therapeutic target for this group of disease. Linking HTRA1 to COL4A1 cSVD highlights a key pathomechanism of clinical importance. HTRA1 is a secreted serine protease that directly cleaves ECM proteins .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 11 including fibronectin, aggrecan, collagens, and perlecan 26. Under physiological conditions, HTRA1 contributes to ECM remodelling during development, degrades aged and damaged ECM components, and helps releasing ECM -embedded bioactive molecules, acting as a master regulator of ECM homeostasis 26. Prior studies have already demonstrated contributions of HTRA1 to other genetic cSVD . Biallelic HTRA1 variants cause CARASIL characterized by early-onset ischemic strokes and progressive cognitive deterioration with spondylosis and alopecia 29. Heterozygous variants in HTRA1 cause CADASIL-2,30 the second most common hereditary cSVD. Interestingly, the rare pathogenic HTRA1 variants are also found in the general populations, e.g., the UK Biobank, which increase the risk of stroke and dementia 31, and common variants (SNPs) in HRTA1 and COL4A1 increase the risk of white matter hyperintensity (WMH, the MRI biomarker of cSVD) 32-34. Most cSVD-associated HTRA1 variants are loss-of- function. Consistent with this, a recent proteomics study of more than 40,000 individuals with MRI markers of cSVD found that lower plasma HTRA1 levels were associated with dementia and with extensive hippocampal perivascular spaces, a marker of cSVD 35, and loss-of-protease function of HTRA1 variants was associated with ischemic stroke and coronary artery disease 36. Interestingly, a HTRA1 loss -of-function profile was also reported in autopsy brain vessel samples from patients of CADASIL 24, the most common genetic cSVD caused by NOTCH3 variants, whereas HTRA1 was found to be sequestered into the aggregated NOTCH3 extracellular domain ( N3ECD) in CADASIL microvessels. Additionally, HTRA1 was found in aggregates of Aβ in microvessels of patients with CAA 25, 37, a common sporadic cSVD and age related dementia, suggesting an alternative mechanism underpinning HTRA1 loss-of-function in cSVD. Researchers have long been seeking a convergent mechanism that explains the diverse forms of cSVD which helps to reveal shared therapeutic targets. Although matrisome has been proposed to be a promising candidate 9, a more specific gene or pathway remains elusive. Based on literature findings described above, HTRA1 seems to be a specific target, however, among the most common genetic cSVDs caused by gene variants such as NOTCH3, HTRA1 and COL4A1/A2, there has not been direct evidence demonstrating the contribution of HTRA1 to pathologies specific to COL4A1/A2 disease. In this study, we provided experimental evidance where HTRA1 was significantly downregulated in astrocytes of the patient -specific iPSC model of COL4A1 cSVD, suggesting HTRA1 inactivation is likely a promising convergent mechanism contributing to the development of the common types of genetic , and possibly sporadic , cSVD. Most importantly, treatment of the iPSC COL4A1 cSVD model with human recombinant HTRA1 significantly rescued the impaired TJs of COL4A1G755R iPSC-BMECs, reversed the BBB integrity, and improved astrocyte function. To our knowledge, this is the first HTRA1 - .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 12 targeted therapy on cSVD models. However, it is not clear how the COL4A1 mutation leads to downregulation of astrocytes HTRA1 in our study. It was reported that HtrA1 is a specific marker in adult mouse forebrain astrocytes and plays an important role in astrocyte development 38. The expression of HtrA1 is regulated by BMP4 signal ling in postnatal astrocytes and responses to brain injury 38. Elucidating the molecular mechanisms underlying the HTRA1 downregulation in COL4A1 cSVD is critical in our future study. One of the important functions of HTRA1 is its regulation of the TGF- signalling activity. It has been established that HTRA1 negatively regulates TGF - signalling through proteolytic degradation of the latency-associated peptide (LAP) 39-41. Consistent with this, we found upregulated TGF- signalling components in the COL4A1 mutant astrocytes, including increased TGF-1 deposits in ECM and increased P-Smad2/3 in nuclei , alongside significantly reduced HTRA1. Our findings are in line with previous reports, where TGF- signalling was activated, TGF-1 protein was accumulated in tunica media, and levels of brain mRNA and protein of TGF-β, Smad2, and Smad3 in were elevated in human samples or a mouse model of CARASIL 40-42. In contrary, a study on a different CARASIL mouse model reported a downregulation of TGF- signalling due to lack of HtrA1 cleavage of latent TGF-1 binding protein-1 (LTBP-1) 43. The reason behind the controversial findings is unclear, but it is more likely that the TGF- signalling is context or cell type dependent. However, a more recent report described an elevated TGF -β signalling in a Col4a1 mutant mouse model of Gould syndrome, where suppressing of TGF-β signalling prevented vascular smooth muscle cell (VSMC) loss and reduce intracerebral haemorrhage 44, which support our finding. The improved BBB integrity by TGF-β inhibitor treatment in our iPSC COL4A1 cSVD model further demonstrated a gain- of-function mechanism on TGF-β signalling. Of note, although HTRA1 treatment reduced the expression of TGFB1, TGFB2 and TGFBR2 in our study, we believe this is not the only mechanism underlying the beneficial role of HTRA1 on BBB integrity. It is likely that supplement of the deficient HTRA1 protease also restored its protein quality control function in eliminating the fragmented and misfolded ECM proteins like the mutant collagen IV α1, therefore rescue the cell function 26. The exact mechanism warrants further study. Previous studies, including our own, have established the importance of the ECM in regulating BBB function and its contributing to cSVD when disrupted 9, 13, 21, 45-47. However, the specific role of the pBM in cSVD pathogenesis remains largely unexplored. Elucidating the function of astrocyte-derived pBM using in vivo models is challenging. In this study, the iPSC-based model enabled us to generate pBM matrices from astrocytes and assess its specific contributions to BBB function in cSVD. Our findings reveal cell .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 13 type specific neurovascular interactions and provided a foundation for the development of targeted therapeutic strategies. Astrocyte secreted s oluble factors play a critical role in the establishment and maintenance of the BBB 48. Indeed, we found that conditioned medium from the control iPSC-derived astrocytes could effectively rescue the damaged TJ in mutant BMECs, and this function of the mutant astrocyte secretome was compromised. Interestingly, the increased TJ protein level and improved TJ organisation was not associated with increased mRNA expression of TJ genes . As key soluble factors secreted from astrocytes, such as Sonic Hedgehog (Shh) and glia-derived neurotropic factor (GDNF), regulate TJs at the transcription level, our results may suggest an unsignificant involvement of these soluble factors from astrocyte in COL4A1 cSVD pathology. Additionally, the exaggerated transcriptomic changes in the mutant BMECs after exposing to the mutant astrocyte secr etome also implies that both cell autonomous effects by the COL4A1 variant and the mutant astrocyte-derived signals synergistically amplify disease pathologies, potentially accelerating BM degradation and BBB damage in COL4A1-associated cSVD. Therefore, multicellular models are important in future studies to mimic the full spectrum disease phenotype.

Our work has limitations. We have used the conventional 2D setting to culture the iPSC derived astrocytes and BMECs. While this helps understanding the functionalities of individual cell types, considering the special organi sation and cell-cell communications in human intact tissues, it is ideal to use 3D models like brain and vascular organoids or organ-on-chip models, which are our future works. Additionally, iPSC-derived astrocytes and BMECs may not fully capture the maturity and heterogeneity seen in adult humans. This is the current limitation for iPSC-based models in general. Long term culture of brain organoids, especially vascularized brain organoids, could improv e model maturity. Furthermore, the study focused on a single COL4A1 variant, COL4A1G755R. Although this variant represents one of the 85–95% cSVD-associated COL4A1 missense variants that involve a glycine within the Gly–X–Y motif essential for collagen triple helix formation, it is ideal to include multiple patient lines to avoid interpersonal variations , which is particularly important for validating therapeutic candidates and understanding genotype– phenotype variability. Nevertheless, the use of an isogenic control ensured that the phenotypes observed in this study resulted from the COL4A1G755R variant.

.CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 14 iPSC culture The COL4A1G755R iPSC line derived from a cSVD patient and its isogenic control line used in this study were reported in previous23. IPSCs were maintained in 6-well plates in Essential 8 medium (A15170001, Thermo Fisher) at 37°C with 5% CO2. The plates were pre-coated with Corning® Matrigel® Basement Membrane Matrix (Matrigel) (Corning, CLS356234). When the iPSCs were approximately 70% confluence, the cells were washed with phosphate-buffered saline ( PBS) and dissociated with 0.5 mM UltraPure EDTA solution (Invitrogen, 15575020) at 37°C for 3-5 minutes. Small cell clusters were seeded in 1.5 mL fresh E ssential 8 medium containing 10 μM Rho kinase (ROCK) Inhibitor (Y-27632) (Sigma, SCM075) in 6-well plates. IPSCs used in this study were between passages 20 to 40. Differentiation of iPSCs into brain microvascular endothelial cells (BMECs) The method used for BMEC differentiation from iPSCs was adapted from a reported paper28. IPSCs were seeded on Matrigel-coated 6-well plates with Essential 8 medium and 10 μM ROCK inhibitor at a density approximately 10,000 cells / cm2, with medium refreshed daily. Three days after, medium was changed to DeSR1 with 6 μM CHIR99021 (72052, StemCell Technologies), termed Day 0. The DeSR1 medium is composed of DMEM/Ham’s F12 (31331, Life Technologies) with 1× MEM -NEAA (11140, Life Technologies), 0.5× GlutaMAX (25030024, Life Technologies), and 0.1 mM β- mercaptoethanol (31350, Life Technologies). Twenty-four hours later, the medium was replaced by DeSR2 medium, that was DeSR1 medium supplemented with 1× B27. The DeSR2 medium was changed daily until day 6 , when it was changed to hECSR1 and further cultured for 48 hours. The hECSR1 is composed of human endothelial serum- free medium (hESFM; Thermo Fisher Scientific), 20 ng/mL bFGF (R&D Systems), 10 μM all-trans retinoic acid (Sigma), and 1 × B27. On day 8, the medium was switched to hECSR2 that is hESFM supplemented by 1× B27. On day 9, cells were dissociated with TrypLE (Thermo Fisher Scientific) and seeded onto Matrigel-coated plates or 12-well trans-well inserts with 0.4 μm pore size membranes (Corning) . After 20 mins, the plate was rinsed using a medium composed of hESFM supplemented with 2% platelet -poor plasma-derived serum and 10 µM Y27632 , as a selection step to remove any undifferentiated cells. BMECs were then left in the same medium overnight to allow cell attachment and growth. Astrocyte differentiation from iPSCs Neuronal progenitor cell (NPC) induction: The iPSCs were first differentiated into NPCs based on methods from previous literature49. IPSCs were passaged onto Matrigel-coated 6-well plates and cultured in Essential 8 medium supplemented with 10 μM ROCK .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 15 inhibitor at 100% confluency. After 24 hours , the medium was replaced with neural induction medium (NIM) 2 ml per well, termed Day 0. The NIM is DMEM/Ham’s F12 mixed with equal volume of Neurobasal (12348, Life Technologies) and supplemented with 1× B-27 supplement, 1× N-2 supplement (17502048, Life Technologies), 5 μg/ml insulin (I9278, Sigma), 100 μM β-mercaptoethanol, 100 μM nonessential amino acids, 1 mM L-glutamine, 500 μM Sodium Pyruvate (S8636, Sigma), 50 μg/ml Pens/Strep (15140, Life Technologies), 10 μM SB431542 (1614, Tocris), and 1 μM Dorsomorphin (3093, Tocris). The NIM was refreshed daily. On Day 12, cells were dissociated by 200 ul Dispase (07923, Stem Cell Technologies) , mixed with 2 ml NIM medium, and seeded onto Laminin-coated plates. After an overnight culture on Day 13, NIM was switched to neural maintenance medium (NMM) supplemented with 20 ng/ml FGF2. NMM is NIM without SB431542 and Dorsomorphin. On Day 17 when rosettes became visible, cells were dissociated using Dispase and reseeded onto Laminin-coated plates at a 1:2 ratio. The NMM medium was changed daily. On Day 25, cells were dissociated into single cells using Accutase and plated onto Laminin-coated plates, the plating density was set to a 1:1 ratio. The medium was refreshed after 24 hours and then every 48hrs subsequently. Cells were passaged in a 1:2 ratio when reaching 90%-100% confluency. Astrocyte differentiation: NPCs were dissociated with Accutase and seeded onto Matrigel pre-coated 6-well plates in NMM and cultured at 37°C. This timepoint was termed as Day 0 of iPSC -astrocytes differentiation. On day 1, medium was replaced with 2 ml STEMdiffTM Astrocyte Differentiation Medium (ADM; 100-0013, Stem cell Technologies) that was refreshed daily. On Day 7, cells were passaged with Accutase and seeded onto Matrigel-coated plates with ADM in a density approximately 1.5 -2×105 cells/cm2. The ADM was changed every 2-3 days. On Day 14, cells were passaged . On Day 21, the cells were passaged again , and medium was replaced with STEMdiffTM Astrocyte Maturation Medium (AMM; 100-0016, Stem cell Technologies), which was every 2-3 days. On both Day 28 and Day 35, cells were passaged using Accutase and reseeded on Matrigel-coated glass coverslips in 24 -well plates or 6 well plates in a density of 1.5 - 2×105 cells/cm2 and cultured at 37°C with 5% CO2. After two passages in AMM, astrocyte identity was assessed by immunochemistry and RT -qPCR using astrocyte -specific marker S100β and GFAP . Immunofluorescence staining Cells were washed twice with PBS and fixed with 4% paraformaldehyde (PFA) 200μl per well in 24 -well plate for 15 minutes at room temperature (RT). Paraformaldehyde was removed, and cells were washed 3 times with 2ml PBS permeabilized with 0.1% triton X-100 in PBS for 10 minutes at RT. After further PBS wash, cells were incubated in 1 ml blocking buffer (10% donkey serum in PBS) for 30 min at RT to block the non -specific .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 16 binding. Primary antibodies were diluted in blocking buffer and incubated with cells for 1 hour at RT. After further PBS washes, cells were incubated with appropriate Alexa Fluorescent conjugated secondary antibodies for 1 hour at RT in dark. Antibody details and dilution s are listed in Table S1 and S2 . Cell nuclei were counterstained using ProLong™ Gold Antifade Mountant with DAPI (Thermo Fisher Scientific). Fluorescent images were acquired using the EVOS™ FLoid™ Cell Imaging Station . Images were processed and analyzed using Fiji ImageJ (http://imagej.net/Fiji/Downloads). Quantification of TJ Integrity: To assess TJ integrity, two complimentary image analysis approaches were applied. The first approach measures discontinued membrane staining of TJ markers (ZO-1, Claudin-5, and Occludin), which is presented as numbers of TJ breaks per cell. The second approach assessed the discontinuous fusion points by measuring the loss of fluorescence intensity of TJ staining at cell junctions, which was presented as discontinued fusion points per cell. total 9 images, with 3 randomly selected images taken from each experiment, repeated across 3 biological replicates. Trans-endothelial electrical resistance (TEER) assay On day 9 of the i PSC BMEC differentiation, cells were seeded onto 12-well Trans-well inserts inserts with a 0.4 μm pore size. Twenty-four hours after, TEER measurements were recorded daily using an EVOM2TM Epithelial Volt/Ohm Meter with STX2 electrodes (World Precision Instrument) . The cell culture hECSR2 medium was refreshed 30 minutes before measurement. TEER values were normalized by subtracting the

TEER from a blank well and then multiplied by the surface area (1.12 cm 2 of 12-well plate) of the Trans-well filter. Results are presented as ohms x cm2. All TEER experiments were performed with at least 3 triplicate wells. Quantitative real time polymerase chain reaction (RT-qPCR) Total RNA was prepared using the RNeasy Mini kit according to the manufacturer’s instructions (74104, Qiagen) followed by cDNA synthesis 250 ng total RNA and High- Capacity RNA -to-cDNATM kit (4387406, Thermofisher) according to manufacturer’s instruction. RT-qPCR was then performed using PowerUpTM SYBRTM Green Master Mix (A25742, Thermofisher). QuantStudio 3 (Applied Biosystems) Real -Time PCR System. Cycling conditions were as follows: 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. The melt curve analysis was performed to confirm amplification specificity (95°C for 15 seconds, 60°C for 60 seconds, then gradual increase to 95 °C). Data was analysed using the 2^-ΔΔCt method. GAPDH was used as endogenous control Primers were listed in Table S3. Calcium imaging .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 17 Four-compartment glass-bottom chamber slides (Cellvis, D35C4 -20-1.5-N) were pre - coated with Matrigel at a concentration of 8.5 μg/cm2 per compartment and incubated at 37°C for 2 hours. After removing the excess Matrigel, iPSC -derived astrocytes were seeded into each compartment at a density of 1 x 105 cells in 0.5 mL astrocyte medium. Cells were cultured for 2–3 days at 37°C in a humidified incubator with 5% CO2 to allow for attachment and network formation. Five μL of 50 μM Fluo 4 AM (Invitrogen, ThermoFischer Scientific) working solution was added directly to each chamber (final concentration: ~0.5 μM), followed by incubation at 37°C and 5% CO 2 for 30 min in the dark. After incubation, cells were gently washed with PBS to remove excess dye and replaced with fresh astrocyte medium. Calcium transients were stimulated by applying 25 μ M L -glutamate (ab83389, Abcam) directly into the imaging medium during live imaging. Fluorescence signals were recorded using a Zeiss spinning -disk confocal microscope (Axio Observer Z1) with a 20×/0.8 NA objective and a 488 nm excitation laser. Time-lapse images were captured every 0.39 seconds for a total duration of 30 minutes. Regions of in terest (ROI) were selected for imaging for individual astrocytes. Fluorescence intensity over time was analysed using ZEN (Zeiss), Fiji (ImageJ), and GraphPad Prism software to quantify transient and oscillatory calcium responses. All experiments were conducted in three independent replicates. Glutamate uptake assay iPSC-derived astrocytes were seeded in Matrigel-coated 48-well plates at a density of 1 x 10 6 cells per well and cultured at 37°C in a 5% CO 2 incubator overnight. On the following day, the cell culture medium was replaced with 200 μL of pre-warmed, serum- free Hank’s Balanced Salt Solution (HBSS; 14025092, Gibco) containing 100 nM glutamate. Cells were incubated at 37°C in a 5% CO ₂ incubator for 2 hours to allow glutamate uptake, the cell culture medium was then collected. Cells from each well were washed once with cold PBS and then lysed with 100 μL in ice cold PBS. The cell lysates were homogenised by pipetting and then centrifuged at 12,000 × g for 15 minutes at 4°C, and the supernatant was collected. To quantify glutamate levels, 50 μL of medium or cell lysate was respectively mixed with 50 μL of the Glutamate Detection Reagent, prepared according to the manufacturer’s instructions (Glutamate Assay Kit; Abcam, ab83389). Samples were shaken for 30 –60 seconds and incubated at room temperature for 60 minutes, protected from light. Luminescence was measured using a multimode plate reader. ECM synthesis and cell seeding Twenty-four well plates were pre-coated with Laminin (Sigma, L2020) at a concentration of 10 μg/mL (equivalent to approximately 1 –2 μg/cm²) in sterile PBS and incubated at 37 °C for 4 hours. IPSC-derived astrocytes were seeded on the Laminin-coated 24-well .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 18 plate at a density of 1-2 x 105 cells / well. After 24 hours, the medium was replaced with astrocyte medium (AM; SC-1801, Sciencell) supplemented with 50 μg/mL of L-ascorbic acid. The media was refreshed every 24-48 hours for 7 days. The cell monolayer was then decellularised by applying 0.1% Triton X -100 and 20 mM ammonium hydroxide (NH4OH) in PBS for 5 minutes at room temperature . Each well was then gently rinsed three times (30 seconds each time) with 500ul PBS to remove cell debris. To ensure complete decellularization, additional 500 μL PBS was added and the plate was incubated on a gentle orbital shaker overnight at 4°C. On t he following day, any remaining cell debris was removed with additional PBS washes. Synthesised ECM was fixed with 4% paraformaldehyde (PFA) or used for cell seeding experiments. Prior to cell seeding, synthesised ECM -coated wells were pre -equilibrated by adding 1 mL of pre - warmed culture medium and incubated at 37°C for 1 hour to promote ECM hydration and temperature equilibration. The medium was then aspirated immediately before cell plating. iPSC-BMECs were dissociated into single -cell suspensions using TrypLE and seeded onto the pre-warmed ECM at a density of 1 × 10⁴ cells per well in 24-well plates with 500 μL of hECSR2 medium and 10 μM Y-27632. Cultures were maintained at 37°C in a 5% CO₂ humidified incubator, and the hECSR2 medium was refreshed every 24–48 hours. After 4 days of culture, cells were fixed with 4% PFA for downstream immunocytochemistry. Western Blotting: Cells were washed with cold PBS and lysed using RIPA buffer (150 mM NaCl, 1% NP - 40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris HCl, pH 7.4) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). Lysates were incubated on ice for 30 minutes with periodic vortexing, then centrifuged at 12,000 × g for 15 minutes at 4 °C. The supernatant containing total protein was collected, and concentration was determined using the BCA assay (Thermo Fisher). Protein samples (20–30 μg) were mixed with 4x Laemmli sample buffer, denatured at 100 °C for 5 minutes, and loaded onto homemade SDS-PAGE gel (4% stacking gel and 12% resolving gel). Gels ran in SDS running buffer (25 mM Tris base, 192 mM glycine, 0.1% SDS ) at 50 V for 15 minutes, followed b y 120 V for ~40 minutes until the dye front approached the bottom. Proteins in the gel were transferred onto Nitrocellulose membrane (0.45 87 μm, GE Healthcare) at 15V for 1 hour in blotting buffer (25 mM Tris, 192 mM glycine, 20% methanol). The membrane was blocked in 5% Marvel in TBS (20 mM Tris-HCl, 150 mM NaCl, pH 7.6) for 1 hour and then incubated with primary antibodies in 5% Marvel in TBS-T (TBS with 0.1% Tween 20) at 4°C overnight. Following three 5-minute washes in TBS-T, membranes were incubated with HRP-conjugated secondary antibodies (1:2000 1:5000 dilution) for 1.5 hours at room temperature in 5% Marvel in TBS-T. After final .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 19 washes, membranes were developed with enhanced chemiluminescence (ECL) substrate (Bio-Rad) for 1 minute, and signals were visualized using a digital imager. Quantification of band intensity was performed using Fiji (ImageJ). Antibodies were listed in Table S4. HTRA1 and small molecule treatments Recombinant human HTRA1 protein (20 ng/mL; ab134441, Abcam) was administered to iPSC-derived BMECs carrying the COL4A1G755R mutation. Treatments were performed in hECSR2 medium supplemented withHTRA1 recombinant protein, and maintained for 4 days, with media refreshed every 48 hours. To inhibit TGF-β signalling, SB431542 (10 μM; Cayman Chemical, 1614) was added to the hECSR2 culture medium of COL4A1G755R iPSC-derived BMECs for 4 days with media changes every 48 hours. RNA Sequencing Bioinformatic Analysis Total RNA was extracted from iPSC-BMECs, iPSC-astrocytes and COL4A1G755R mutant and isogenic control lines using the RNeasy Mini Kit (Qiagen, 74104), according to the manufacturer’s protocol as described in Section 2.8. Cells from each condition were harvested in biological triplicates. RNA concentration and purity were assessed using a SimpliNano spectrophotometer (Biochrom, 29061711), and all samples were submitted to Azenta Life Sciences for sequencing. RNA quality control and sequencing were conducted by Azenta Life Sciences. RNA integrity was confirmed using the Agilent Bioanalyzer, and only samples with high quality RNA (RIN > 8) were used for downstream analysis. Paired -end sequencing was performed on an Illumina platform, achieving high sequencing depth for robust transcriptome coverage. Bioinformatic processing and diff erential expression analysis were performed by Azenta using a standardized pipeline. Raw reads were trimmed and aligned to the human reference genome (GRCh38/hg38), and gene -level counts were generated. Differential gene expression analysis was conducted u sing DESeq2 (v1.36.0) in R Studio, with genes defined as differentially expressed if they had an adjusted p value 1. Principal component analysis (PCA) and hierarchical clustering were used to assess sample variability and global expression profiles. Visualization and enrichment analyses —including volcano plots, gene ontology (GO) enrichment, heatmaps, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, gene set enrichment analysis (GSEA), and Venn diag rams of differentially expressed genes (DEGs)—were performed using the WeiShengXin online platform (https://www.bioinformatics.com). Protein –protein interaction (PPI) networks were constructed using the STRING database, highlighting key interaction cluster s among significantly downregulated genes. Statistics .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 20 All statistical analyses were performed using GraphPad Prism software (version 9). Data distribution was assessed for normality using the Shapiro –Wilk test, and variance homogeneity was verified with the Brown–Forsythe test. For comparisons between two groups, two-tailed unpaired Student’s t-tests was used. For comparisons involving more than two groups, one -way analysis of variance (ANOVA) was conducted, followed by Tukey’s post hoc test. In cases involving two or more independent variables, two -way ANOVA with Šídák's post hoc analysis were used . Data are presented as mean ± standard error of the mean (SEM). Statistical significance was defined as p < 0.05. The number of biological replicates (n) is specified in each figure legend. For experiments involving iPSC-derived cells, each “n” typically refers to an independ ent differentiation unless otherwise noted. Acknowledgments The Bioimaging Facility microscopes used in this study were purchased with grants from BBSRC, Wellcome and the University of Manchester Strategic Fund. Special thanks go to Steven Marsden and David Spiller for their help with the microscopy. Funding The work was supported by a Stroke Association priority program award in Advancing Care and Treatment of Vascular Dementia (grant 16VAD_04) in partnership with the British Heart Foundation and Alzheimer’s Society to SA, ZC, AG, KH, HM, SS, TW, TVA. Competing interests The authors declare no competing interests.

1. Wardlaw, J.M., Smith, C. & Dichgans, M. Small vessel disease: mechanisms and clinical implications. Lancet Neurol 18, 684-696 (2019). 2. Markus, H.S. & Joutel, A. The pathogenesis of cerebral small vessel disease and vascular cognitive impairment. Physiol Rev 105, 1075-1171 (2025). 3. Cannistraro, R.J. et al. CNS small vessel disease: A clinical review. Neurology 92, 1146-1156 (2019). .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 21 4. Best, J.G., Jesuthasan, A. & Werring, D.J. Cerebral small vessel disease and intracranial bleeding risk: Prognostic and practical significance. Int J Stroke 18, 44-52 (2023). 5. De Silva, T.M. & Faraci, F.M. Contributions of Aging to Cerebral Small Vessel Disease. Annu Rev Physiol 82, 275-295 (2020). 6. Haffner, C., Malik, R. & Dichgans, M. Genetic factors in cerebral small vessel disease and their impact on stroke and dementia. J Cereb Blood Flow Metab 36, 158-171 (2016). 7. Mishra, A. et al. Gene-mapping study of extremes of cerebral small vessel disease reveals TRIM47 as a strong candidate. Brain 145, 1992-2007 (2022). 8. Traylor, M. et al. Genetic basis of lacunar stroke: a pooled analysis of individual patient data and genome -wide association studies. Lancet Neurol 20, 351-361 (2021). 9. Joutel, A., Haddad, I., Ratelade, J. & Nelson, M.T. Perturbations of the cerebrovascular matrisome: A convergent mechanism in small vessel disease of the brain? J Cereb Blood Flow Metab 36, 143-157 (2016). 10. Casino, P. et al. Structures of collagen IV globular domains: insight into associated pathologies, folding and network assembly. IUCrJ 5, 765-779 (2018). 11. Khoshnoodi, J., Cartailler, J.P., Alvares, K., Veis, A. & Hudson, B.G. Molecular recognition in the assembly of collagens: terminal noncollagenous domains are key recognition modules in the formation of triple helical protomers. J Biol Chem 281, 38117-38121 (2006). 12. Hohenester, E. & Yurchenco, P.D. Laminins in basement membrane assembly. Cell Adh Migr 7, 56-63 (2013). 13. Thomsen, M.S., Routhe, L.J. & Moos, T. The vascular basement membrane in the healthy and pathological brain. J Cereb Blood Flow Metab 37, 3300-3317 (2017). 14. Hallmann, R. et al. The role of basement membrane laminins in vascular function. Int J Biochem Cell Biol 127, 105823 (2020). 15. Wright, S.A., Lennon, R. & Greenhalgh, A.D. Basement membranes' role in immune cell recruitment to the central nervous system. J Inflamm (Lond) 21, 53 (2024). 16. Zhang, X. et al. The endothelial basement membrane acts as a checkpoint for entry of pathogenic T cells into the brain. J Exp Med 217 (2020). 17. Gould, D.B., Marchant, J.K., Savinova, O.V., Smith, R.S. & John, S.W. Col4a1 mutation causes endoplasmic reticulum stress and genetically modifiable ocular dysgenesis. Hum Mol Genet 16, 798-807 (2007). .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 22 18. Gould, D.B. et al. Role of COL4A1 in small -vessel disease and hemorrhagic stroke. N Engl J Med 354, 1489-1496 (2006). 19. Wardlaw, J.M. et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822-838 (2013). 20. Agrawal, S. et al. Dystroglycan is selectively cleaved at the parenchymal basement membrane at sites of leukocyte extravasation in experimental autoimmune encephalomyelitis. J Exp Med 203, 1007-1019 (2006). 21. Xu, L., Nirwane, A. & Yao, Y. Basement membrane and blood-brain barrier. Stroke Vasc Neurol 4, 78-82 (2019). 22. McNeilly, S. et al. Collagen IV deficiency causes hypertrophic remodeling and endothelium-dependent hyperpolarization in small vessel disease with intracerebral hemorrhage. EBioMedicine 107, 105315 (2024). 23. Al-Thani, M. et al. A novel human iPSC model of COL4A1/A2 small vessel disease unveils a key pathogenic role of matrix metalloproteinases. Stem Cell Reports 18, 2386-2399 (2023). 24. Zellner, A. et al. CADASIL brain vessels show a HTRA1 loss -of-function profile. Acta Neuropathol 136, 111-125 (2018). 25. Haffner, C. The emerging role of the HTRA1 protease in brain microvascular disease. Front Dement 2, 1146055 (2023). 26. Clausen, T., Kaiser, M., Huber, R. & Ehrmann, M. HTRA proteases: regulated proteolysis in protein quality control. Nat Rev Mol Cell Biol 12, 152-162 (2011). 27. Zhang, W. et al. Induced pluripotent stem cell model revealed impaired neurovascular interaction in genetic small vessel disease Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy. Front Cell Neurosci 17, 1195470 (2023). 28. Pediaditakis, I. et al. Modeling alpha-synuclein pathology in a human brain-chip to assess blood-brain barrier disruption. Nat Commun 12, 5907 (2021). 29. Fukutake, T. Cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy (CARASIL): from discovery to gene identification. J Stroke Cerebrovasc Dis 20, 85-93 (2011). 30. Nozaki, H. et al. Distinct molecular mechanisms of HTRA1 mutants in manifesting heterozygotes with CARASIL. Neurology 86, 1964-1974 (2016). 31. Cho, B.P .H. et al. Association of Vascular Risk Factors and Genetic Factors With Penetrance of Variants Causing Monogenic Stroke. JAMA Neurol 79, 1303-1311 (2022). .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 23 32. Mishra, A. et al. Association of variants in HTRA1 and NOTCH3 with MRI-defined extremes of cerebral small vessel disease in older subjects. Brain 142, 1009- 1023 (2019). 33. Malik, R. et al. Whole-exome sequencing reveals a role of HTRA1 and EGFL8 in brain white matter hyperintensities. Brain 144, 2670-2682 (2021). 34. Sargurupremraj, M. et al. Cerebral small vessel disease genomics and its implications across the lifespan. Nat Commun 11, 6285 (2020). 35. Caro, I. et al. Proteogenomics in cerebrospinal fluid and plasma reveals new biological fingerprint of cerebral small vessel disease. Nat Aging 5, 2514-2531 (2025). 36. Malik, R. et al. Genetically proxied HTRA1 protease activity and circulating levels independently predict risk of ischemic stroke and coronary artery disease. Nat Cardiovasc Res 3, 701-713 (2024). 37. Zellner, A. et al. Proteomic profiling in cerebral amyloid angiopathy reveals an overlap with CADASIL highlighting accumulation of HTRA1 and its substrates. Acta Neuropathol Commun 10, 6 (2022). 38. Chen, J. et al. BMP-Responsive Protease HtrA1 Is Differentially Expressed in Astrocytes and Regulates Astrocytic Development and Injury Response. J Neurosci 38, 3840-3857 (2018). 39. Oka, C. et al. HtrA1 serine protease inhibits signaling mediated by Tgfbeta family proteins. Development 131, 1041-1053 (2004). 40. Shiga, A. et al. Cerebral small-vessel disease protein HTRA1 controls the amount of TGF-beta1 via cleavage of proTGF -beta1. Hum Mol Genet 20, 1800 -1810 (2011). 41. Hara, K. et al. Association of HTRA1 mutations and familial ischemic cerebral small-vessel disease. N Engl J Med 360, 1729-1739 (2009). 42. Chuanfen, L., Xiaoling, W., Wen, J., Bingzhen, C. & Min, W. HtrA1L364P leads to cognitive dysfunction and vascular destruction through TGF -beta/Smad signaling pathway in CARASIL model mice. Brain Behav 12, e2691 (2022). 43. Beaufort, N. et al. Cerebral small vessel disease -related protease HtrA1 processes latent TGF-beta binding protein 1 and facilitates TGF -beta signaling. Proc Natl Acad Sci U S A 111, 16496-16501 (2014). 44. Branyan, K. et al. Elevated TGFbeta signaling contributes to cerebral small vessel disease in mouse models of Gould syndrome. Matrix Biol 115, 48 -70 (2023). 45. Thomsen, M.S., Birkelund, S., Burkhart, A., Stensballe, A. & Moos, T. Synthesis and deposition of basement membrane proteins by primary brain capillary .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 24 endothelial cells in a murine model of the blood-brain barrier. J Neurochem 140, 741-754 (2017). 46. Trempel, M.A. et al. Pericytes repair engineered defects in the basement membrane to restore barrier integrity in an in vitro model of the blood-brain barrier. Mater Today Bio 35, 102361 (2025). 47. Zhao, X. et al. Selective vulnerability of cerebral vasculature to NOTCH3 variants in small vessel disease and rescue by phosphodiesterase-5 inhibitor. Sci Adv 12, eaeb1134 (2026). 48. Alvarez, J.I., Katayama, T. & Prat, A. Glial influence on the blood brain barrier. Glia 61, 1939-1958 (2013). 49. Shi, Y., Kirwan, P . & Livesey, F.J. Directed differentiation of human pluripotent stem cells to cerebral cortex neurons and neural networks. Nat Protoc 7, 1836- 1846 (2012). Figure legends Figure 1. Astrocyte differentiation and characterization. (A). Schematic protocol of astrocyte differentiation from iPSCs. (B). Phase contrast microscopy and immunofluorescent staining of cells during astrocyte differentiation. NPC markers PAX6 (red) and SOX1 (green) were stained on Day 20, and astrocyte markers GFAP (green), S100β (green) and Glut-1(red) stained on Day 60. Nuclei were stained with DAPI (blue). Scale bars, 100 μm. (C). qRT-PCR quantification of NPC and astrocyte specific marker genes PAX6, SOX1, S100β, and GFAP in isogenic contro l (IsoCtrl) and COL4A1G755R cells on differentiation on Day 0, 18, and 64, respectively. GAPDH was used as the endogenous control. (D). Glutamate uptake assay. IPSC-derived astrocytes were incubated with 100 nM glutamate at 37°C for two hours followed by measuring the remaining glutamate in the cell culture medium and cell lysates. Data in (C and D) are mean ± SEM from triplicate reactions of 3 biological replicates, n=3. Unpaired Student’s t-test in (C) and one -way ANOVA followed by Tukey’s post hoc test in (D) , **p<0.01, ***p<0.001, ****p<0.0001. (E). Calcium signals measured by Fluo-4 AM indicator in COL4A1G755R and IsoCtrl iPSC-derived astrocytes . Scale bars, 100 μm. NIM, neural induction medium; NMM, neural maintenance medium ; ADM, astrocyte differentiation medium; AMM, astrocyte maturation medium. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 25 Figure 2. Transcriptome Profiling of iPSC-derived astrocytes. Astrocytes derived from COL4A1G755R and isogenic control iPSCs from 3 independent iPSC differentiations were subjected to RNA sequencing (RNA-seq) followed by data analysis. (A). Principal component analysis (PCA). (B). Volcano plot displaying differentially expressed genes (DEGs) in COL4A1G755R astrocytes compared to the isogenic control with an adjusted p- value cutoff p<0.05. Log2 fold change greater than 1 are indicated by red dots, representing significantly up-regulated DEGs; and l og2 fold change less than -1 are indicated by blue dots, represent ing significantly down-regulated DEGs. (C). Gene ontology (GO) analysis of DEGs demonstrates enriched GO terms of biological process (BP), cellular components (CC), and molecular functions (MF). (D). GSEA plot of DEGs revealing enriched gene sets relating to the indicated GO terms. (E). Heatmap showing DEGs of COL4A1G755R and isogenic control astrocytes under the GO term “extracellular matrix organization”. (F). RT-qPCR conformation of RNA-seq results on basement membrane related genes. Data are means ± SEM from triplicate reactions of 3 biological replicates, n=3. Unpaired Student’s t-test, **p<0.01, **p<0.01, ***p<0.001, ****p<0.0001. Figure 3. Changes of HTRA1 in COL4A1G755R iPSC-astrocytes and dysregulation of TGF-β signalling. Astrocytes were differentiated from COL4A1G755R and isogenic control (IsoCtrl) iPSCs and subjected to RT-qPCR and immunofluorescent staining. (A). HTRA1 expression determined by RT-qPCR. (B). HTRA1 protein levels determined by western blotting (WB). WB of iPSC samples of COL4A1G755R and IsoCtrl were also included as controls. Right panel shows quantification of WB images. (C). Immunofluorescent staining of HTRA1 in iPSC-derived astrocytes (a and b) and ECM preparations from astrocytes (c and d). (D). Immunofluorescent staining of TGF-β in ECM preparations of iPSC-derived astrocytes. (E). Immunofluorescent staining of p-SMAD2/3 in iPSC-derived astrocytes. (F) Quantification of nuclear signals of the p -SMAD2/3 staining in (E) . (G). TGFB1, TGFB2, and LTBP1 expression in iPSC-derived astrocytes were determined by RT-qPCR. Data in (A, B, F and G) are means ± SEM from triplicate reactions of 3 biological replicates, n=3. Unpaired Student’s t-test, **p<0.01, **p<0.01, ***p<0.001, ****p<0.0001. Scale bar in (C-E), 100 μm. Figure 4. ECM secreted from COL4A1G755R iPSC-astrocytes damages tight junctions formed by iPSC -BMECs. IPSC-derived BMECs were seeded onto extracellular matrix (ECM) synthesised from iPSC-derived astrocytes for 4 days, followed by immunofluorescent staining of tight junction (TJ) markers ZO -1, Claudin -5, and Occludin. ( A). Isogenic BMECs were grown on ECM secreted from isogenic and COL4A1G755R astrocytes. (B). Qualifications of (A) on the integrity of TJs. Figure shows the number of TJ breaks per cell (left panel) and discontinued fusion points per cell (right .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 26 panel). (C). COL4A1G755R BMECs were grown on ECM secreted from isogenic and COL4A1G755R astrocytes. (D). Qualifications of (B) on the integrity of TJs. Figure shows the number of TJ breaks per cell (left panel) and discontinued fusion points per cell (right panel). Data in (B and D) are presented as means ± SEM from 9 images, with 3 randomly selected images taken from each experiment, repeated across 3 biological replicates, n=9. Unpaired Student’s t-test, **** p<0.0001. Figure 5. Effect of iPSC-astrocyte secretome on tight junction integrity in iPSC- BMECs. BMECs and astrocytes were differentiated from iPSCs of COL4A1G755R and its isogenic control (IsoCtrl). Total RNA and cells were collected from COL4A1G755R iPSC- derived BMECs after 4 days of treatment with astrocyte-conditioned medium (ACM) from patient or control iPSC -astrocytes, mixed at a ratio of ACM: hECSR2 = 7:3. Protein of COL4A1G755R iPSC-derived BMECs were collected from a Trans-well® co-culture system after co-cultured with iPSC-astrocytes for 4 days. (A). Immunofluorescent staining of tight junction (TJ) markers ZO -1, Claudin-5, and Occludin in COL4A1G755R or IsoCtrl iPSC- BMECs cultured without astrocyte medium (left column), with IsoCtrl astrocyte medium (middle column), or with COL4A1G755R astrocyte medium (right column), respectively. Scale bar, 100 μm. (B-C). Qualifications of (A) on the integrity of TJs and presented as the number of TJ breaks and discontinued fusion points per cell. The TJ integrity in IsoCtrl BMECs exposed to either IsoCtrl or COL4A1G755R astrocyte medium is shown in (B); and the TJ integrity in COL4A1G755R BMECs exposed to either IsoCtrl or COL4A1G755R astrocyte medium is shown in (C). (D) WB of TJ proteins ZO-1, Claudin-5, and Occludin in IsoCtrl iPSC-BMECs co-culture with either IsoCtrl or COL4A1G755R astrocyte. Quantifications of band densities are shown on the right panel. (E). RT-qPCR

of COL4A1G755R BMECs exposed to either IsoCtrl or COL4A1G755R astrocyte medium. (F). WB of TJ proteins ZO -1, Claudin-5, and Occludin in COL4A1G755R iPSC- BMECs co-culture with either IsoCtrl or COL4A1G755R astrocyte. Quantifications of band densities are shown on the right panel. Data are means ± SEM from triplicate reactions of 3 biological replicates, n=3. Two-way ANOVA followed by Šídák's post hoc test, **p<0.01, **p<0.01, ***p<0.001, ****p<0.0001. Figure 6. Transcriptomics of iPSC-derived BMECs exposed to secretome of iPSC- derived astrocytes. BMECs and astrocytes were differentiated from iPSCs of COL4A1G755R and its isogenic control (IsoCtrl) from 3 independent experiments. The iPSC-derived COL4A1G755R and IsoCtrl BMECs were then cultured in their corresponding iPSC-astrocyte conditioned medium mixed with hECSR2 medium in a 7:3 ratio for 4 days, followed by RNA sequencing (RNA-seq). RNA-seq were also conducted on COL4A1G755R and IsoCtrl iPSC -derived BMECs that were not exposed to astrocyte conditioned .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 27 medium for comparison. (A-B). Volcano plots showing differentially expressed genes (DEGs) between COL4A1G755R BMECs exposed to COL4A1G755R astrocyte medium and IsoCtrol BMECs exposed to IsoCtrl astrocyte medium (A), and between COL4A1G755R BMECs and IsoCtrol BMECs (B). (C). Gene ontology (GO) analysis of DEGs between COL4A1G755R BMECs exposed to COL4A1G755R astrocyte medium and IsoCtrol BMECs exposed to IsoCtrl astrocyte medium. (D). Heatmap showing DEGs under the GO term “extracellular matrix organisation”. Results comparing DEGs between COL4A1G755R BMECs exposed to COL4A1G755R astrocyte medium and IsoCtrol BMECs exposed to IsoCtrl astrocyte medium. Figure 7. HTRA1 treatment rescue impaired tight junctions of COL4A1G755R iPSC- BMECs. BMECs derived from COL4A1G755R iPSCs were treated with or without human HTRA1 recombinant protein (20 ng/mL) for 4 days, with medi um refreshed every 48 hours. The cells were then subjected to immunofluorescent staining for tight junction (TJ) markers ZO-1, Claudin-5 and Occludin. (A). Immunofluorescent images of TJs. Scale bar, 100 μm. (B). Quantification of TJ continuity in (A) which are presented as the number of TJ breaks per cell (left panel) and discontinued fusion points per cell (right panel). Data are means ± SEM derived from 9 images, with 3 randomly selected images taken from each experiment, repeated across 3 biological replicates, n=9. Unpaired Student’s t-test, ****p<0.0001. Figure 8. Involvement of TGF-β signaling in HTRA1 rescue of impaired tight junctions in COL4A1G755R iPSC-BMECs. (A). BMECs derived from COL4A1G755R iPSCs were treated with or without human HTRA1 recombinant protein (20 ng/mL) for 4 days, with medium refreshed every 48 hours. RT-qPCR showing reduced expression of TGFB1, TGFB2 and TGFR2 after HTRA1 treatment. Data are means ± SEM , n=3. Unpaired Student’s t-test, **p<0.01, ***p<0.001, ****p<0.0001. ( B). BMECs derived from COL4A1G755R iPSCs were treated with or without TGF-β inhibitor SB431542 (10 μM) for 4 days, followed by immunofluorescent staining of tight junction (TJ) proteins ZO-1 and Occludin. Scale bar, 100 μm. ( C-D). Quantification of TJ continuity in (A) which are presented as the number of TJ breaks per cell ( C) and discontinued fusion points per cell (D). Data are means ± SEM derived from 9 images, with 3 randomly selected images taken from each experiment, repeated across 3 biological replicates, n= 9. Unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ( D). Astrocytes derived from COL4A1G755R and isogenic control (IsoCtrl) iPSCs were treated with or without human HTRA1 protein (20 ng/mL), respectively, for 4 days, with medium refreshed every 48 hours. Glutamate uptake was then measured. Data are means ± SEM from triplicate reactions of 3 biological experiments, n=3. Unpaired Student’s t-test, *p<0.05. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 28 Figure 1. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 29 Figure 2. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 30 Figure 3. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 31 Figure 4. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 32 Figure 5. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 33 Figure 6. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 34 Figure 7. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint 35 Figure 8. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 13, 2026. ; https://doi.org/10.64898/2026.05.12.724691doi: bioRxiv preprint