Glial type specific regulation of CNS angiogenesis by HIFα-activated different signaling pathways.

The mechanisms by which oligodendroglia modulate CNS angiogenesis remain elusive. Previous in vitro data suggest that oligodendroglia regulate CNS endothelial cell proliferation and blood vessel formation through hypoxia inducible factor alpha (HIFα)-activated Wnt (but not VEGF) signaling. Using in vivo genetic models, we show that HIFα in oligodendroglia is necessary and sufficient for angiogenesis independent of CNS regions. At the molecular level, HIFα stabilization in oligodendroglia does not perturb Wnt signaling but rather activates VEGF. At the functional level, genetically blocking oligodendroglia-derived VEGF but not Wnt significantly decreases oligodendroglial HIFα-regulated CNS angiogenesis. Blocking astroglia-derived Wnt signaling reduces astroglial HIFα-regulated CNS angiogenesis. Together, our in vivo data demonstrate that oligodendroglial HIFα regulates CNS angiogenesis through Wnt-independent and VEGF-dependent signaling. These findings suggest an alternative mechanistic understanding of CNS angiogenesis by postnatal glial cells and unveil a glial cell type-dependent HIFα-Wnt axis in regulating CNS vessel formation.

T he vasculature of the central nervous system (CNS), which is developed exclusively through angiogenesis, plays a crucial role in providing neural cells with nutrients and oxygen. CNS angiogenesis, the growth of new blood vessels from pre-existing ones, starts during embryonic development and matures during postnatal development in human and rodent brains, for example, by the age of one month in rodents 1 . Dysregulated CNS angiogenesis negatively impacts postnatal brain development and functional recovery from brain injuries [2][3][4] . The current study aimed to dissect the molecular regulation of postnatal CNS angiogenesis using in vivo genetic animal models.
The developing CNS parenchyma is exposed to physiological hypoxia with local oxygen concentration ranging from 0.5 to 7% 5 . Hypoxia-inducible factor α (HIFα) is a critical regulator that adapts neural cells to hypoxic conditions. The transcription factor HIFα, including HIF1α and HIF2α, is subjected to constant degradation. Von Hippel-Lindau (VHL), a negative regulator of HIFα's transcriptional activity, plays an essential role in HIFα degradation. Under low oxygen or upon VHL disruption, HIF1α and HIF2α degradation is impaired and subsequently translocate into the nuclei where they regulate downstream target genes through forming transcriptional active complexes with the constitutive HIF1β 6,7 . Previous data suggest that HIFα function in neural precursor cells is required for embryonic brain vascular development 8 . Recent data including those from our own laboratory show that HIFα function in oligodendroglial lineage cells may play a pivotal role in regulating postnatal angiogenesis in the brain white matter 9 and in the spinal cord 10 . However, the molecular mechanisms underlying oligodendroglial HIFαregulated angiogenesis are still controversial and remain incompletely defined.
The current concept stated that oligodendroglial HIFα promotes CNS angiogenesis through activating signaling pathway of Wnt but not vascular endothelial growth factor (VEGF) 9 . However, this "Wnt-dependent" view was supported only by in vitro studies and pharmacological manipulations 9 , in which "pathological" activation of Wnt signaling, poor cell-type selectivity and/ or off-target effects of small compounds cannot be excluded. In this study, we present in vivo evidence supporting an alternative view in our mechanistic understanding of oligodendroglial HIFαregulated CNS angiogenesis. Our in vivo genetic knockout data reveal that oligodendroglial HIFα regulates endothelial cell proliferation and angiogenesis in a VEGF-dependent but Wntindependent manner and this regulation is independent of CNS regions during postnatal development. This data also demonstrate that postnatal astroglia regulate CNS angiogenesis at least in part through HIFα-activated Wnt signaling, unveiling a glial cell type-specific HIFα-Wnt connection in the CNS.

Results
Oligodendroglial HIFα regulates CNS angiogenesis. Endothelial cell (EC) proliferation is an essential step of angiogenesis and the blood vessel density is an end-point reflection of angiogenesis. Therefore, we used EC proliferation and vessel density as in vivo readouts of angiogenesis 9 . To quantify blood vessel density, we used the basement membrane marker Laminin to label blood vessels and employed a semi-automated approach to calculate the percentage of Laminin-occupying area among total assessed area ( Supplementary Fig. 1). To determine whether oligodendroglial HIFα is required for angiogenesis throughout the postnatal CNS, Cre-LoxP approach was used to genetically ablate or stabilize HIFα and EC proliferation and vessel density were analyzed in the brain and the spinal cord.
We used Cnp-Cre line 11 to generate Cnp-Cre:Hif1α fl/fl (HIF1α conditional knockout, cKO), Cnp-Cre: Hif2α fl/fl (HIF2α cKO), and Cnp-Cre: Hif1α fl/fl :Hif2α fl/fl (HIF1α/HIF2α or HIFα double cKO) mutants ( Supplementary Fig. 2). Mice carrying Cnp-Cre transgene alone did not display any developmental abnormalities compared with non-Cre animals as previously reported 11 and supported by our assessment of CNS angiogenesis at postnatal 10 and motor function at postnatal one month ( Supplementary  Fig. 3). HIF1α cKO or HIF2α cKO did not influence blood vessel density, indicating a compensatory effect of oligodendroglial HIF1α and HIF2α on angiogenesis. In contrast, HIF1α/HIF2α double cKO (refer to as HIFα cKO hereafter) significantly impaired CNS angiogenesis evidenced by reduced blood vessel density (Fig. 1a-c) and diminished EC proliferation not only in the cerebral cortex but also in the spinal cord at postnatal 14 ( Fig. 1d-g), suggesting that oligodendroglial HIFα is necessary for CNS angiogenesis.
The HIFα protein is constantly translated but subjected to rapid turnover via proteasome-mediated degradation, a process in which Von Hippel-Lindau product (VHL) is essential for HIFα degradation. Therefore, we employed Cnp-Cre:Vhl fl/fl transgenic mice to genetically ablate VHL and stabilize HIFα function in oligodendroglial lineage cells ( Supplementary Fig. 4). We found that the density of blood vessels and the proliferation of ECs were significantly increased in the cerebral cortex and spinal cord of Cnp-Cre:Vhl fl/fl mice compared with those of non-Cre control mice at different time points in the early postnatal CNS ( Fig. 1h-j). Stabilizing HIFα in oligodendroglial lineage cells did not have a major effect on the integrity of the blood-brain (spinal cord) barrier in the adult Cnp-Cre:Vhl fl/fl mice ( Supplementary  Fig. 5).
Previous studies have reported that Cnp-Cre primarily targets oligodendroglial lineage cells and also a subpopulation of early neural progenitor cells 12,13 . To corroborate the conclusion derived from Cnp-Cre transgenic mice, we assessed CNS angiogenesis in a different animal strain Sox10-Cre:Vhl fl/fl in which Sox10-Cre mediated HIFα stabilization in the earlier stages of oligodendrocyte development in the CNS. Consistently, CNS angiogenesis was significantly increased in Sox10-Cre:Vhl fl/fl mutants, as assessed by elevated blood vessel density (Fig. 1k), EC-specific Pecam1 mRNA expression (Fig. 1l), and EC densities (Fig. 1m). Taken together, our loss (gain)-of-function results suggest that oligodendroglial HIFα is necessary and sufficient for angiogenesis and that the angiogenic regulation by oligodendroglial HIFα is independent of CNS regions.
Oligodendroglial HIFα does not regulate Wnt signaling. To determine whether HIFα in oligodendroglial lineage cells regulates Wnt/β-catenin signaling, we quantified the Wnt/β-catenin target gene Axin2, Naked1, and Notum, which are reliable readouts for the signaling activation 10 . We found no significant changes in the mRNA levels of those genes in HIFα-stabilized spinal cord and forebrain of Cnp-Cre:Vhl fl/fl mutants at different time points (Fig. 2a) compared with those of non-Cre controls. Consistently, western blot assay showed that the active form of βcatenin (dephosphorylated on Ser37 or Thr41) and Axin2 did not change (Fig. 2b), indicating that stabilizing oligodendroglial HIFα does not perturb the activity of Wnt/β-catenin signaling in the CNS. Furthermore, we found no significant change in the mRNA level of Wnt7a in the CNS of Cnp-Cre:Vhl fl/fl mice compared with non-Cre controls (Fig. 2c). We crossed Wnt reporter transgenic mice BAT-lacZ 14 with Cnp-Cre:Vhl fl/fl mutants and found no difference in lacZ mRNA level in the spinal cord of BAT-lacZ/ Cnp-Cre:Vhl fl/fl mice compared with age-matched BAT-lacZ mice (Fig. 2d).
We further assessed the activity of Wnt/β-catenin signaling in a different animal strain of Sox10-Cre:Vhl fl/fl mice. Consistent with    ( Fig. 2g) and the density of cerebral blood vessels (Fig. 2h). However, the mRNA expression of Wnt target genes Ainx2 and Sp5 (Fig. 2i) and the protein levels of active β-catenin and Naked1 (Supplementary Fig. 8) were indistinguishable between Pdgfrα-CreER T2 :Vhl fl/fl mutants and non-Cre controls, indicating that Wnt/β-catenin signaling activity was not altered by oligodendroglial HIFα stabilization. Previous study reported an autocrine activation of Wnt/βcatenin signaling in OPCs by HIFα stabilization 9 . To assess the autocrine activity of Wnt/β-catenin signaling, we treated purified primary OPCs with HIFα stabilizer DMOG 9 in the presence or absence of HIFα signaling blocker Chetomin 15 ( Supplementary  Fig. 9a). Our results showed that pharmacological stabilizing HIFα activated HIFα signaling target genes (Supplementary Fig. 9b) but did not activate Wnt/β-catenin target genes nor Wnt7a and Wnt7b ( Supplementary Fig. 9c) in primary OPCs isolated from neonatal brain. We also quantified the activity of Wnt/β-catenin signaling in primary OPCs which were isolated from neonatal Sox10-Cre:Vhl fl/fl brain. Consistent with the in vivo data (Fig. 2e), HIFα target genes were significantly increased in primary VHL-deficient OPCs (Fig. 2j). However, neither Wnt/βcatenin target genes Axin2 and Naked1 nor Wnt7a were increased in primary VHL-deficient OPCs (Fig. 2k), suggesting that stabilizing oligodendroglial HIFα does not perturb Wnt/βcatenin signaling in primary OPCs.
To determine whether HIFα deletion affects Wnt/β-catenin signaling, we analyzed Wnt/β-catenin activity in the early postnatal CNS of Cnp-Cre:HIFα cKO and Pdgfrα-CreER T2 :HIFα cKO mutants. Consistent with HIFα-stabilized mutants (Fig. 2), we found no evidence of Wnt/β-catenin signaling perturbation in both strains of HIFα cKO mutants ( Supplementary Fig. 10). Collectively, our in vivo and in vitro data demonstrate that Wnt/ β-catenin signaling is unlikely a downstream target of oligodendroglial HIFα as previously reported 9 and suggest that oligodendroglial HIFα may regulate CNS angiogenesis independent of Wnt/β-catenin signaling.
OPC autocrine Wnt signaling is dispensable for angiogenesis. WLS is an essential factor of Wnt secretion from Wnt-producing cells and its deficiency blocks Wnt ligands from activating the downstream pathways in Wnt-receiving cells [16][17][18][19] . To determine whether WLS deficiency affects Wnt secretion from oligodendroglial lineage cells, we knocked down WLS in primary Wnt7aexpressing OPCs and assessed Wnt secretion and autocrine Wnt/ β-catenin activity (Fig. 3a, b). Because Wnt7a has been shown as one of the major Wnt ligand genes expressed in OPCs at the mRNA level 9,20 , we overexpressed Wnt7a in primary OPCs.
Our enzyme-linked immunosorbent assay (ELISA) of the culture medium showed that WLS knockdown significantly reduced Wnt7a concentration secreted from Wnt7a-expressing OPCs (Fig. 3c). Autocrine Wnt/β-catenin signaling was activated in Wnt7a-expressing OPCs, as evidenced by the increased expression of Wnt target genes Axin2 and Sp5 (Fig. 3d), but this activation was blocked in WLS-deficient OPCs (Fig. 3d). Our data suggest that WLS is required for Wnt secretion from OPCs.
To define the putative in vivo role of Wnt signaling in HIFαregulated CNS angiogenesis, we generated VHL/WLS double mutant hybrids to block Wnt secretion from HIFα-stabilized oligodendroglial lineage cells (Fig. 4a). Because constitutive Sox10-Cre:Vhl fl/fl pups died at very early postnatal ages, we used an inducible Cre line Sox10-CreER T2 to stabilize HIFα and disrupt WLS (Fig. 4b, c) in Sox10 + oligodendroglial lineage cells (OPCs and differentiated oligodendroglia). Our fate-mapping data showed that Sox10-CreER T2 elicited~60% of recombination efficiency and greater than 90% of oligodendroglial specificity in Sox10 + oligodendroglial lineage cells in the early postnatal CNS ( Supplementary Fig. 11). We confirmed that HIFα's function was indeed stabilized in the spinal cord of Sox10-CreER T2 :Vhl fl/fl (HIFα-stabilized mice) and Sox10-CreER T2 :Vhl fl/fl :Wls fl/fl (HIFαstabilized/WLS-disrupted mice), as evidenced by the elevated expression of HIFα target gene Hk2 (Fig. 4d) and Ldha (Fig. 4e) in comparison with non-Cre controls. Our analysis demonstrated that blocking Wnt secretion by disrupting WLS did not alter HIFα stabilization-elicited CNS angiogenesis in HIFα-stabilized/ WLS-disrupted mice compared with HIFα-stabilized mice, which was supported by unchanged levels of endothelial Pecam1 mRNA expression (Fig. 4f) and unchanged densities of blood vessels ( Fig. 4g-j), ERG + ECs (Fig. 4k), and ERG + EdU + dividing ECs (Fig. 4l) in the spinal cord and cerebral cortex of HIFα-stabilized/ WLS-disrupted mice compared with those of HIFα-stabilized mice. These data suggest that oligodendroglial lineage-derived Wnt signaling plays a minor role in HIFα-regulated angiogenesis in the early postnatal CNS.
Astroglial HIFα regulates CNS angiogenesis via Wnt signaling. Astroglial maturation is also temporally and functionally coupled with postnatal CNS angiogenesis. We assess the connection of astroglial HIFα and Wnt/β-catenin activation in the CNS. We first used the mouse Gfap promoter-driven constitutive Cre, i.e. mGfap-Cre to genetically stabilize HIFα in astroglia. The efficiency of mGfap-Cre-mediated recombination among astroglial lineage cells, quantified by Cre-mediated EYFP reporter, was low (~35%) in the CNS in the early postnatal CNS by P10 (Supplementary Fig. 12a-c) and progressively increased during postnatal CNS development ( Supplementary Fig. 12d-f). Our fate-mapping data showed that EYFP reporter, which is an indicator of mGfap-Cre activity, was expressed in GFAP + or S100β + astrocytes, but not in Sox10 + oligodendroglial lineage cells, NeuN + neurons (Fig. 8a), or ERG + ECs (data not shown) in the spinal cord and the cerebral cortex of adult mGfap-Cre:Rosa26-EYFP mice at P60, confirming that mGfap-Cre primarily targets astroglial lineage cells in those CNS regions. We observed a significant increase in the density of Laminin + blood vessels (Fig. 8b, c) and in the mRNA expression of EC-specific PECAM1 (Fig. 8d) throughout the CNS of mGfap-Cre:Vhl fl/fl mutants compared with non-Cre control mice by P30 when Cremediated recombination efficiency was greater than 80% (Fig. 8a,  Supplementary Fig. 12). Double immunohistochemistry showed that blood-borne macromolecule IgG was confined to Laminin + blood vessels in mGfap-Cre:Vhl fl/fl mice, a similar pattern to that in agematched non-Cre controls (Fig. 8e, arrowheads), indicating that the function of the blood-brain (spinal cord) barrier does not appear compromised although the vessel density is elevated.   Unexpectedly, we found that stabilizing HIFα in astroglial lineage cells (Fig. 8f) remarkably activated Wnt/β-catenin signaling in the CNS of mGfap-Cre:Vhl fl/fl mice, as shown by significant elevation in the expression of Wnt/β-catenin signaling target genes Axin2 and Notum in spinal cord and brain (Fig. 8g). Histological (Fig. 8h) and Western blot (cf Fig. 9a, b) assay demonstrated that the active form of β-catenin (dephosphorylated on Ser37 or Thr41) was significantly increased in mGfap-Cre:Vhl fl/fl mice. Double immunohistochemistry confirmed the presence of elevated active β-catenin in PECAM1 + ECs (Fig. 8i,  arrowheads). Collectively, our data suggest that stabilizing HIFα in astroglial lineage cells increases CNS angiogenesis and activates Wnt/β-catenin signaling in ECs.
Wnt/β-catenin signaling activation in ECs by astroglial HIFα stabilization led us to hypothesize that astroglia-derived Wnt signaling may instead play a major role in HIFα-regulated CNS angiogenesis. To test this hypothesis, we generated mGfap-Cre: Vhl fl/fl :Wls fl/fl mice to stabilize HIFα's function and simultaneously disrupting Wnt secretion from HIFα-stabilized astroglia. Our data showed that Wnt signaling activity was significantly reduced in the spinal cord of mGfap-Cre:Vhl fl/fl : Wls fl/fl mice compared with that of mGfap-Cre:Vhl fl/fl mice (Fig. 9a, b), thus verifying the efficacy of blocking astrogliaderived Wnt signaling in vivo by WLS deletion. Intriguingly, disrupting astroglia-derived Wnt signaling significantly reduced the densities of blood vessels (Fig. 9c-e) and ERG + ECs (Fig. 9f-h) in the CNS of mGfap-Cre:Vhl fl/fl :Wls fl/fl double mutant mice compared with mGfap-Cre:Vhl fl/fl mice, indicating that astroglia-derived Wnt signaling is a downstream mediator of astroglial HIFα-regulated CNS angiogenesis.
Our results indicated that the constitutive mGfap-Cre elicited a poor recombination efficiency in early postnatal astrocytes ( Supplementary Fig. 12). To determine whether early postnatal astrocytes regulate CNS angiogenesis through HIFα-activated Wnt signaling, we generated Aldh1l1-CreER T2 :Vhl fl/fl :Wls fl/fl mutants. Our data demonstrated a greater than 90% of recombination efficiency and 95% of astroglial specificity in the spinal cord and cerebral cortex of Aldh1l1-CreER T2 :Rosa26-EYFP at P8 when tamoxifen was injected at P1, P2 and P3 ( Supplementary Fig. 13). Consistent with the data derived from mGfap-Cre:Vhl fl/fl :Wls fl/fl strain, we found that the densities of blood vessels and ECs were significantly increased in Aldh1l1-CreER T2 :Vhl fl/fl mutants compared with those in non-Cre controls and that simultaneous WLS ablation significantly reduced the densities of blood vessels and ECs in the cortex and spinal cord of Aldh1l1-CreER T2 :Vhl fl/fl :Wls fl/fl mutants compared with those of Aldh1l1-CreER T2 :Vhl fl/fl animals at early postnatal age of P8 (Fig. 10). These data provide a strong genetic proof that HIFα-activated Wnt signaling is a major downstream pathway by which astroglia regulate angiogenesis during postnatal CNS development.

Discussion
The maturation of glial cells including oligodendroglia and astroglia in the developing human and murine brain is temporally and functionally coupled with the maturation of the CNS vascular network 21 . The regulation of CNS angiogenesis by glial cells is critical for postnatal CNS development and investigating the molecular underpinnings of CNS angiogenesis has clinical implications in neural repair after CNS damage in which hypoxia is commonly present 4,21,22 . In this study, we employed a battery of genetic mutant mice and presented several significant findings: (1) oligodendroglial HIFα is necessary and sufficient for postnatal CNS angiogenesis and this regulation occurs in a manner independent of CNS regions; (2) in sharp contrast to the previous report 9 , HIFα stabilization in oligodendroglial lineage cells does not perturb Wnt/β-catenin signaling, but remarkably activates VEGF, and genetically blocking oligodendroglia-derived VEGF but not Wnt reduces oligodendroglial HIFα-regulated CNS angiogenesis; (3) Wnt signaling is a downstream pathway by which astroglial HIFα regulates CNS angiogenesis. Our findings represent an alternative view in our mechanistic understanding of oligodendroglial HIFα-regulated angiogenesis from a Wntdependent/VEGF-independent view 9 to a VEGF-dependent/ Wnt-independent one, and also unveil a glial cell type-dependent HIFα-Wnt axis (oligodendroglial vs astroglia) in regulating CNS angiogenesis ( Supplementary Fig. 14).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15656-4 ARTICLE that VEGFA is a direct transcriptional target of HIFα [40][41][42] . By leveraging our unique in vivo genetic models of VHL/VEGFA double cKO, we unequivocally prove that VEGFA is an essential downstream molecule that couples oligodendroglial HIFα function and vascular angiogenesis in the CNS, which is different from Yuen et al. 9 , who reported that VEGFA was unchanged in the CNS of oligodendroglial HIFα-stabilized mutants. The discrepancy may presumably reflect the intrinsic differences of in vitro pharmacological interventions and in vivo genetic manipulations. It has been suggested that Wnt signaling regulates VEGF, or vice versa, to control angiogenesis [43][44][45] . It is possible that Wnt/β-catenin signaling is required for, or synergistically regulates, HIFα-activated VEGFA expression. Our data do not support this possibility. First, stabilizing oligodendroglial HIFα activates VEGFA but not Wnt/β-catenin signaling. Second, VEGF expression is indistinguishable in the CNS of oligodendroglial VHL/WLS double cKO mutants from that of oligodendroglial VHL single cKO mutants (data not shown), indicating that oligodendroglial-derived Wnt signaling plays a minor role in VEGFA expression. Third, Wnt/β-catenin activity is comparable in the CNS of oligodendroglial VHL/VEGFA double cKO mutants and VHL single cKO mutants, implying that oligodendroglia-derived VEGFA has no regulatory role in Wnt/β-catenin activity. Together, our results do not support a major interplay between oligodendroglial HIFα-activated VEGFA and Wnt/β-catenin signaling in modulating CNS angiogenesis. Previous studies including those from our own laboratory 10,12,[46][47][48][49] have shown that dysregulated Wnt/β-catenin activity invariably inhibits oligodendrocyte differentiation and myelination. Given the normal level of Wnt/β-catenin activity in the CNS of oligodendroglial HIFα-stabilized mutants, our study will spark renewed interests in studying Wnt-independent mechanisms underlying the impairment of oligodendroglial differentiation and myelination in HIFα-stabilized mutants as previously reported 9 . Interestingly, HIFα is stabilized and enriched in oligodendroglia in the active demyelinating lesions and normalappearing white matter (NAWM) of multiple sclerosis patient brains [50][51][52][53] . We show that HIFα stabilization in oligodendrocytes remarkably activates the angiogenic and neurotrophic factor VEGF in the CNS. The genetic models generated in our study also provide a powerful tool in determining the role of HIFα stabilization in OPCs and oligodendrocytes in the pathophysiology of demyelination and remyelination in multiple sclerosis and other neurological disorders in which hypoxia-like tissue injury occurs. Animals were housed at 12h light/dark cycle with free access to food and drink, and both males and females were used in this study. The single transgenic mice were crossed to generate double or triple transgenic mice indicated in the study. All Cre transgene was maintained as heterozygous. All transgenic mice were maintained on a C57BL/6 background. Animal protocols were approved by Institutional Animal Care and Use Committee at the University of California, Davis.
HIFα signaling stabilization and inhibition in vitro. Purified brain primary OPCs were pre-incubated with 100 nM Chetomin or DMSO control for 2 h, and then switched to the fresh culture medium with 1 mM Dimethyloxalylglycine (DMOG, D3695, Sigma) in the presence of 100 nM Chetomin (C9623, Sigma) or DMSO (D8418, Sigma) control for 7 h before RNA preparation.
VEGFA ELISA assay. Cell culture medium of primary OPCs from Non-Cre control and Cnp-Cre, Hif1α fl/fl , Hif2α fl/fl were collected for VEGF measurement. Endogenous VEGF concentrations were determined using a mouse-specific VEGF Quantikine ELISA kit (t#MMV00, R&D System) according to the manufacture's instruction.
Transfection in primary OPCs and ELISA assay. Primary OPCs was transfected with Wls-ShRNA (TRCN 0000 234932, Mission ShRNA bacterial Glycerol stock Fig. 8 Astroglial HIFα stabilization promotes CNS angiogenesis and enhances Wnt signaling activity. a Fate-mapping study showing that mGfap-Cremediated EYFP was expressed in GFAP + astrocytes but not in Sox10 + oligodendroglial lineage cells or NeuN + neurons in the spinal cord at P60. WM, white matter, GM, gray matter, Ctx, cortex. EYFP was identified as S100β + astrocytes in the cerebral Ctx. Scale bars = 20 μm. b Representative images of Laminin immunostaining in mGfap-Cre, Vhl fl/fl mutants and non-Cre control mice at P30. Scale bars = 100 μm. c Percentage of Laminin-occupying area among total area at P30. Two-tailed Student's t test, Welch's corrected t (4) = 11.92 spinal crd, t (7) = 10.12 cerebral cortex. n = 4 Ctrl, 5 VHL cKO. d RT-qPCR assay of endothelial Pecam1 at P30. Two-tailed Student's t test, Welch's corrected t (4.046) = 8.564 spinal cord, Welch's corrected t (4.490) = 6.706 forebrain. n = 5 each group. e immunostaining showing that endogenous mouse IgG is restricted to Laminin + blood vessels (arrowheads) in the early adult spinal cord of mGfap-Cre,Vhl fl/fl mutant and control mice at P47. Scale bars = 10 μm. f RT-qPCR assay of the mRNA levels of HIFα target gene Glut1, Hk2 and Ldha in P30 spinal cord. Two-tailed Student's t test with Welch's correction, t (4.099) = 7.947 Glut1, t (4.243) = 9.636 Hk2, t (4.085) = 9.025 Ldha. n = 4 Ctrl, 5 VHL cKO. g RT-qPCR assay of the mRNA levels of Wnt/β-catenin target genes Axin2 and Notum at P30. Two-tailed Student's t test, Welch's corrected t (4.760) = 7.296 spinal cord Axin2, t (8)  Immunohistochemistry and blood vessel quantification. Study mice were perfused with ice-cold phosphate buffered saline (PBS, pH = 7.0, Catalog #BP399-20, Fisher Chemical), and then post-fix in fresh 4% paraformaldehyde (PFA, Catalog #1570-S, Electron Microscopy Science,PA) at room temperature(RT) for 2 h. The CNS tissue was washed in ice-cold PBS for three times, 15 min each time. The samples were cryoprotected with 30% sucrose in PBS (Sucrose, Catalog #S5-3, Fisher Chemical) for 20 h followed by sectioning. Sixteen microns thick sections were serially collected and stored in −80°C. Immunohistochemistry was conducted as below: slices were air dry in RT for at least 1 h, and then were blocked with 10% Donkey (Dky) serum for 1 h at RT. Tissue was incubated with primary antibody overnight at 4°C. Slices were washed for 15 min in PBST (PBS with 0.1% Tween-20) for three times, then incubated with fluorescence conjugated secondary antibody (1:500; Alexa-fluorescence from Jackson ImmunoResearch) for 1.5 h at RT. Slices were washed 15 min in PBST (PBS with 0.1% Tween-20) for three times. The immunostaining was done before incubating with DAPI nuclear staining for 10 min 56,57 . The information of primary antibodies used for immunohistochemistry in the study were listed in Supplementary Table 1. For BrdU immunostaining, sections were pretreated with fresh-made 2N HCl (#320331, Sigma) followed by the above immunostaining procedures.
program. The total area and Laminin-occupying area were derived and from ImageJ and exported to Microsoft Excel for calculating the percent of Lamininoccupying area among assessed total CNS area.
mRNA In Situ hybridization (ISH). We employed the PCR and in vitro transcription to prepare cRNA probes targeting Vegfa and Plp 58 . Targeted sequences of Vegfa and Plp were generated by PCR. The primers used were: Vegfa-Forward: GGATATGTTTGACTGCTGTGGA; Vegfa-Reverse: AGGGAAGATGAGGAA GGGTAAG; Plp-Forward: GGGGATGCCTGAGAAGGT; Plp-Reverse: TGTGA TGCTTTCTGCCCA. We added the T7 (GCGTAATACGACTCACTATAGGG) and SP6 (GCGATTTAGGTGACACTATAG) promoter sequences to the 5' of the forward and reverse primers, respective. The SP6 and T7 promoter sequences are recognized by the SP6 and T7 RNA polymerase, respectively, in the subsequence in vitro transcription. PCR products of Vegfa and Plp amplification were used as DNA templates to transcribe into Vegfa and Plp cRNA probes in vitro using SP6 RNA polymerase. T7 RNA polymerase-mediated transcription of RNA was used as negative control. DIG-UTP or FITC-UTP was used to generate DIG-or FITClabeled cRNA probes. Single or dual mRNA ISH was done using our previous protocols 58 . Frozen sections of 14 μm thickness were used. The concentration of cRNA probe we used was 100 ng/100 μl hybridization buffer. Hybridization was conducted at 65°C for 18-20 h. After hybridization, sections were treated with 10 μg/ml RNase A to eliminate nonspecific cRNA binding. For single mRNA ISH with DIG-labeled cRNA probes, DIG was recognized by alkaline phosphatase (AP)-conjugated anti-DIG (#11093274910, Sigma) antibody and DIG signals were visualized by the NBT/ BCIP (#72091, Sigma) method. For dual fluorescence mRNA ISH (Vegfa and Plp), FITC-labeled Plp cRNA probe and DIG-labeled Vegfa cRNA probe were applied to frozen sections simultaneously during the hybridization step. The FITC signals were visualized by tyramide signal amplification (TSA) fluorescence system (#NEL747A, Perkin Elmer) according to the manufacturer's instructions using horseradish peroxidase (HRP)-IgG Fraction Monoclonal Mouse Anti-Fluorescein (#200-032-037, Jackson ImmunoResearch). DIG signals were visualized by a HNPP fluorescent kit (#11758888001, Sigma) according to the manufacturer's instructions using AP-conjugated anti-DIG Fab 2 antibody (#11093274910, Sigma).