Research ResourceVASCULAR BIOLOGY

Gene expression profiles of brain endothelial cells during embryonic development at bulk and single-cell levels

See allHide authors and affiliations

Sci. Signal.  11 Jul 2017:
Vol. 10, Issue 487, eaag2476
DOI: 10.1126/scisignal.aag2476

Building the blood-brain barrier

The blood-brain barrier (BBB) is a highly selective physical barrier that protects the central nervous system from pathogens and toxins, but it also prevents many therapeutic drugs from reaching the brain. Hupe et al. used translational profiling to identify factors that are involved in BBB development and distinguish this tissue from other vascular beds. Various brain vasculature–specific genes encoding transport, adhesion, and extracellular matrix factors were differentially expressed in brain endothelial cells compared to endothelial cells from other organs during embryonic development. Several transcription factors that act downstream of Wnt signaling, which is required for BBB formation, were associated with maturation of the BBB, and two of these (Foxf2 and Zic3) were sufficient to stimulate the expression of BBB markers when overexpressed in human umbilical vein endothelial cells. These data offer a rich resource for understanding the unique developmental and functional properties of this highly specialized tissue.

Abstract

The blood-brain barrier is a dynamic interface that separates the brain from the circulatory system, and it is formed by highly specialized endothelial cells. To explore the molecular mechanisms defining the unique nature of vascular development and differentiation in the brain, we generated high-resolution gene expression profiles of mouse embryonic brain endothelial cells using translating ribosome affinity purification and single-cell RNA sequencing. We compared the brain vascular translatome with the vascular translatomes of other organs and analyzed the vascular translatomes of the brain at different time points during embryonic development. Because canonical Wnt signaling is implicated in the formation of the blood-brain barrier, we also compared the brain endothelial translatome of wild-type mice with that of mice lacking the transcriptional cofactor β-catenin (Ctnnb1). Our analysis revealed extensive molecular changes during the embryonic development of the brain endothelium. We identified genes encoding brain endothelium–specific transcription factors (Foxf2, Foxl2, Foxq1, Lef1, Ppard, Zfp551, and Zic3) that are associated with maturation of the blood-brain barrier and act downstream of the Wnt–β-catenin signaling pathway. Profiling of individual brain endothelial cells revealed substantial heterogeneity in the population. Nevertheless, the high abundance of Foxf2, Foxq1, Ppard, or Zic3 transcripts correlated with the increased expression of genes encoding markers of brain endothelial cell differentiation. Expression of Foxf2 and Zic3 in human umbilical vein endothelial cells induced the production of blood-brain barrier differentiation markers. This comprehensive data set may help to improve the engineering of in vitro blood-brain barrier models.

INTRODUCTION

Although all vascular beds share certain features, their specification in distinct organ systems occurs by different modes (1). This results in blood vessels that exhibit properties tailored to each organ’s specific needs (2, 3). The central nervous system (CNS) is rather unique regarding its vascular system. The CNS endothelium forms a tightly sealed physical blood-brain barrier (BBB) that is quite distinct from the more permeable vasculature present in other organs. The BBB is critical for brain homeostasis and neural activity and is perturbed in various CNS disorders like edema, stroke, or brain tumors (4). Despite its critical role, the barrier prevents potential therapeutic agents from reaching the CNS.

Vascularization of the murine CNS begins at embryonic day (E) 9.5 with angiogenic vascular sprouting from the perineural vascular plexus that surrounds the neurectoderm. BBB-defining features are induced in the brain endothelium and regulated by the neuroectodermal microenvironment at the earliest stage of endothelial cell (EC) invasion into the developing CNS (5). One of the notable properties of the BBB is the high abundance of tight junction proteins and various transporters that control the exchange of molecules between the brain and the circulation. Despite the undeniable importance of the BBB, cellular and molecular mechanisms that regulate vascular development and differentiation in the CNS are still poorly understood.

Important insights were provided by studies revealing the requirement of Wnt signaling to establish the BBB endothelial phenotype (68). Members of the Wnt family are involved in cell proliferation and various differentiation events during embryonic development. Wnt signaling causes the cytosolic stabilization of the transcriptional cofactor β-catenin (Ctnnb1), which translocates into the nucleus to control transcription of a specific set of target genes (9). It has been proposed that Wnt7a and Wnt7b act directly on the CNS endothelium through the canonical Wnt signaling pathway to promote CNS-specific angiogenesis and early BBB differentiation in vivo (6, 8). Furthermore, Liebner et al. (7) have reported that canonical Wnt signaling is also required for postnatal maturation and maintenance of the BBB. However, the molecular mechanism by which Wnt signaling regulates the behavior and organ-specific proliferation and differentiation of ECs is still elusive.

Gene expression profiling of the CNS endothelium has been used to obtain further insight into brain vascular development and differentiation (1012). Although RNA sequencing (RNA-seq) permits profiling at high resolution, the relatively small number of ECs in the developing brain and the fact that these cells are dispersed among other cell types make it challenging to perform comprehensive profiling studies of the CNS endothelium. We previously developed an inducible translating ribosome affinity purification (TRAP) mouse model, mCherryTRAP (13), that facilitates large-scale CNS endothelium profiling. Upon Cre-dependent activation of the mCherryTRAP transgene in ECs, these mice express an mCherry-tagged ribosomal protein L10a (Rpl10a) that is incorporated into polysomes and enables the identification of unique translational profiles of different EC populations. Although TRAP provides valuable insights on a global scale, it does not identify heterogeneity at the cellular level. The endothelium within any particular organ consists of cells that contribute to both arterial and venous macrovascular and microvascular vessels. Furthermore, during sprouting angiogenesis, endothelial tip and stalk cells exhibit different morphological, migratory, and adhesion properties (4). Single-cell gene expression analysis has the potential to facilitate characterization of this cellular heterogeneity (14) and to provide further insight into the formation and maturation of the BBB.

Here, we used TRAP followed by RNA-seq (TRAP-seq) to generate high-resolution translational profiles of the developing vasculature in the embryonic mouse forebrain. We show that the translatome of the brain endothelium is distinct from that of endothelia in other organs and highly dynamic during embryonic development. Deletion of Ctnnb1 revealed a requirement of Ctnnb1 for differentiation and maintenance of the BBB but not induction of CNS-specific angiogenesis. We identified transcription factors specifically enriched in brain ECs that act downstream of Wnt-Ctnnb1 signaling and are linked to temporal BBB maturation. Despite a remarkable heterogeneity among individual ECs, high abundance of these transcription factors correlated with the expression of genes encoding BBB differentiation markers. Finally, we show that Foxf2 and ZIC3 are able to induce expression of distinct BBB-associated transcripts in vitro.

RESULTS

The translational profile of the brain endothelium differs from that of other organs

To better understand the unique ability of the CNS endothelium to form the BBB, we compared molecular signatures of different organ-specific vascular systems (Fig. 1A). We crossed mCherryTRAP mice with mice carrying the endothelial Cdh5CreERT2 driver line to achieve tamoxifen-inducible activation of the TRAP construct in all ECs of the embryo. To generate both organ-specifc EC translatomes and whole-organ transcriptomes, we purified both translated and transcribed RNA from E14.5 head (without brain), heart, limb, liver, and lung, and performed RNA-seq on each set of transcripts. The EC translatomes and whole-organ transcriptomes for brain and kidney used in this analysis had been published previously (13).

Fig. 1 TRAP-seq to study brain-specific EC differentiation.

(A) Schematic of the experimental procedure to compare the molecular profiles of seven different vascular beds in E14.5 mouse embryos expressing the mCherry-Rpl10a fusion protein in ECs (red). mCherry+ cells were isolated from the indicated tissues for analysis. (B) PCA of the seven EC populations profiled showing individual replicates. PC1, first principal component explaining 22% of variance; PC2, second principal component explaining 17% of variance. n = 3 biological replicates. (C) Number of transcripts either specifically expressed, enriched, or depleted in ECs derived from different organs. (D) Venn diagram showing transcripts considered as translated in the brain endothelium at E14.5. (E and F) Expression of transporters (E) and transcription factors (F) that were highly enriched in the brain vasculature compared to vascular beds from other organs and to whole-tissue samples.

Principal components analysis (PCA) of the different organ-specific vascular translatomes showed that brain and liver ECs had molecular signatures that are much different from ECs in other organs (Fig. 1B). The strict clustering of independent triplicates further supports these conclusions. For further analysis, we first removed organ-specific background transcripts using DESeq (15). Transcripts exhibiting a significant [adjusted P value (padj) < 0.001] fourfold or greater expression in the EC translatome compared to the corresponding organ transcriptome, with a minimum expression of five reads per kilobase per million mapped reads (RPKM), were considered to be translated in the vasculature of an organ. Translational profiles of different organ-specific EC populations were compared to one another using a threshold of fourfold difference in the expression and a padj < 0.001 (Fig. 1C). Using these criteria, we identified 1025 transcripts as being translated in brain ECs at E14.5, with 98 of these genes also expressed in all vascular beds, representing general EC markers. We identified 63 transcripts that were specifically enriched in CNS vasculature compared to all other endothelial beds. Fourteen transcripts that were translated in other vascular beds were depleted in brain ECs (Fig. 1D; fig. S1, A to D; and tables S1 to S4).

Gene ontology (GO) analysis revealed that terms such as “vasculature development,” “blood vessel development,” and “angiogenesis” were strongly enriched among transcripts expressed in the CNS vasculature and those shared among all vascular beds. In contrast, only genes enriched in brain-derived ECs produced an abundance of transport-related terms (fig. S1, E to H). Manual inspection revealed that about 30% of these transcripts encode membrane transport proteins (Fig. 1E and fig. S2). The appearance of physical BBB properties was further highlighted by the high abundance of transcripts encoding tight junction molecules and the low expression of Plvap, a marker for fenestration, in brain ECs, respectively (fig. S3).

Cell differentiation and specification are critically shaped by transcription factors that govern gene expression programs. Our TRAP data revealed six highly enriched transcripts encoding transcription factors specific for CNS-derived ECs: Foxf2, Foxl2, Foxq1, Lef1, Ppard, and Zfp551 (Fig. 1F and fig. S4). Zfp551 has not been previously reported to be enriched in the brain endothelium. In addition, the data implied that Myt1l, reported to be expressed only in differentiating neurons or oligodendroglia (16, 17), was expressed in brain ECs. This inconsistency might be due to very rare nonspecific expression of the Cdh5CreERT2 driver line in some polydendrocytes (fig. S5). Therefore, Myt1l was excluded from further analysis. To further validate our TRAP-seq data for brain EC–enriched transcripts, we performed in situ hybridization for transcripts that had not previously been reported to be specifically expressed in CNS vasculature. Three markers that we tested (Nkd1, Ppard, and Ttyh2) were expressed in vessel-like structures in embryonic brain at E14.5 and E16.5 (fig. S6, A and B) and colocalized with erythrocytes, which are autofluorescent (18).

Gene expression profiles undergo extensive changes during CNS endothelium development

To shed light on brain vasculature specialization over time, we compared the translatomes of ECs derived from embryonic brains at 1-day intervals between E11.5 and E17.5 (Fig. 2A). PCA separated the endothelial samples according to their temporal origin (Fig. 2B), indicating that distinct expression patterns are characteristic of different developmental stages. After filtering of background transcripts and comparison between EC profiles, we found 1461 transcripts to be translated in the CNS vasculature during embryonic development. Among these, 331 transcripts showed rather stable expression within the time frame examined, 295 transcripts increased over time, and 437 transcripts decreased over time. Only 98 genes showed a significant peak in translation within the examined time frame (Fig. 2C and tables S5 to S10).

Fig. 2 CNS vascular maturation during embryonic development.

(A) Schematic of the experimental procedure to compare molecular profiles of the brain vasculature at different developmental stages in embryos expressing the mCherry-Rpl10a fusion protein in the vasculature (red). The tissues shaded in blue were isolated for analysis. (B) PCA of temporally distinct EC populations profiled showing three biological replicates. PC1, first principal component explaining 31% of variance; PC2, second principal component explaining 11% of variance. (C) Diagram depicting significant (DESeq padj < 0.001) changes of different groups of transcripts translated in the CNS vasculature over time during embryonic development. (D and E) Box plots of calculated z scores for expression during developmental stages E11.5 to E17.5 for (D) 63 genes identified as enriched in E14.5 brain ECs or (E) 20 genes encoding markers of angiogenesis. n = 3 biological replicates. Statistical significance was calculated using nonparametric Friedman with post hoc Nemenyi test (***P < 0.001, **P < 0.01, *P < 0.05).

To obtain a more detailed picture of gene expression dynamics during vascular development in the CNS, we investigated the temporal behavior of different classes of transcripts (Fig. 2, D and E; fig. S7; and tables S11 to S15). The 63 transcripts that were specifically enriched in CNS vasculature compared to all other endothelial beds showed increasing expression during development. The high expression of Slc2a1, an early marker for BBB differentiation that encodes a glucose transporter, observed at E11.5 supports the hypothesis that the differentiation of the CNS endothelium is initiated very early (19). However, most of the other genes encoding transporters showed increasing expression over time (fig. S2A), suggesting a continuous maturation of the BBB and its physical properties during embryonic development (fig. S3). On the contrary, the vast majority of transcripts classified as shared EC genes at E14.5, and especially markers for angiogenesis or proliferation, became depleted in brain ECs, either starting at E15.5 or continuously over time (Fig. 2E and fig. S7, G and H).

The CNS-enriched transcription factors defined above fell into two subgroups based on their temporal expression pattern. Foxf2, Foxl2, Foxq1, and Lef1 were already highly expressed at E11.5, suggesting that they play a role in the differentiation program of the CNS endothelium at early stages of development. Expression of Ppard and Zfp551 was low at E11.5 and increased subsequently, indicating a potential involvement at later developmental or maturation stages. Notably, another gene encoding a transcription factor, Zic3, was also confirmed as enriched in the CNS endothelium by our temporal data set, showing a rapid fourfold increase in expression (fig. S4A).

Ctnnb1 promotes differentiation and maintenance of the BBB but not induction of CNS-specific angiogenesis

Maturation of the BBB requires Wnt signaling (68). To further our understanding of the impact of canonical Wnt signaling on CNS vascular development, we analyzed the effect of conditional deletion of Ctnnb1 on translational profiles of brain ECs at E14.5 and E17.5 (Fig. 3A). A floxed Ctnnb1 allele was crossed into the Cdh5CreERT2;mCherryTRAP mouse line, and then, translated and transcribed RNAs from Ctnnb1 knockout (KO) embryo forebrains were isolated and subjected to RNA-seq. RNA-seq data from Ctnnb1+/+;Cdh5CreERT2;mCherryTRAP embryos were used as control.

Fig. 3 Loss of Ctnnb1 in brain ECs reverses their BBB maturation progress.

(A) Schematic of the experimental procedure to compare molecular profiles of brain ECs from wild-type (WT) and Ctnnb1-deleted (KO) E14.5 and E17.5 embryos expressing the mCherry-Rpl10a fusion protein in the vasculature (red). Ctnnb1 deletion was induced by tamoxifen (Tm) treatment 2 days before isolating the tissue for analysis. (B) PCA for WT and KO EC populations at E14.5 and E17.5 showing individual replicates. n = 3 to 4. PC1, first principal component explaining 25% of variance; PC2, second principal component explaining 11% of variance. (C) Venn diagram showing depleted transcripts in KO compared to control ECs at E14.5 and E17.5. (D and E) GO analysis of depleted transcripts in KO compared to control ECs at E14.5 (D) and E17.5 (E). (F) Venn diagram showing enriched transcripts in KO compared to control ECs at E14.5 and E17.5. (G and H) GO analysis of enriched transcripts in KO compared to control ECs at E14.5 (G) and E17.5 (H). (I and J) Box plots of calculated z scores for expression during developmental stages E11.5 to E17.5 in WT or KO ECs at E14.5 and E17.5 for (I) 63 genes identified as enriched in brain ECs or (J) 20 genes encoding angiogenesis markers. n = 3 to 4 biological replicates. Statistical significance was calculated using nonparametric Friedman with post hoc Nemenyi test (***P < 0.001, *P < 0.05).

PCA revealed marked differences in translational profiles between Ctnnb1-deficient endothelia and the corresponding control samples (Fig. 3B). After background filtering, we found expression of 102 and 69 transcripts being depleted in Ctnnb1 KO ECs at E14.5 and E17.5, respectively (>2-fold difference in expression, padj < 0.001) (Fig. 3C and tables S16 and S17). Most of these depleted genes were related to transmembrane transport (Fig. 3, D and E). We also identified 67 (E14.5) and 132 (E17.5) transcripts with increased expression in Ctnnb1 KOs compared to control ECs (Fig. 3, F to H, and tables S18 and S19). Most of these enriched genes were related to cellular adhesion and extracellular matrix–receptor interactions.

Deletion of Ctnnb1 led to significantly decreased translation for 76% (E14.5) or 52% (E17.5) of the brain-enriched transcripts (fig. S8 and table S20). The strong decrease in expression for nearly all brain EC–specific genes encoding transporters (fig. S2A) mirrored the impaired differentiation of CNS endothelium in embryos lacking Ctnnb1 (Fig. 3I and fig. S3), supporting the suggested role for Ctnnb1 in the maintenance of the BBB (20, 21). Similarly, our results revealed that genes encoding brain EC–enriched transcription factors were expressed to a lower extent in the absence of Ctnnb1 (fig. S4A), pointing toward a key role for Ctnnb1 in activating BBB-specific gene expression.

The increased expression of various collagens, integrins, and laminins in Ctnnb1 KO ECs could be caused by cellular stress or might represent a compensatory mechanism of the KO endothelium for establishing a minimal barrier function (2224). Deletion of Ctnnb1 resulted in significantly increased expression of transcripts encoding angiogenic (55%) and general EC markers (46%) at E17.5, whereas virtually none of the angiogenesis markers and only about 15% of general EC markers were significantly suppressed (fig. S8 and tables S21 and 22). Notably, the increased expression of genes encoding angiogenesis markers was only seen at E17.5 and not detectable at E14.5 (Fig. 3J). Expression of E14.5 brain EC–depleted genes was mainly increased for both time points in Ctnnb1 KOs (table S23). In contrast, general proliferation was seemingly unaltered upon Ctnnb1 deletion (table S24). Together, our data indicate that canonical Wnt signaling through Ctnnb1 promotes maturation and maintenance of brain-derived ECs but does not induce CNS-specific angiogenesis.

Single-cell profiling of brain ECs reveals a high level of heterogeneity

To achieve a better resolution of cellular differences within the ECs comprising the BBB, we analyzed transcriptomes of 80 brain-derived individual cells from forebrains of Cdh5CreERT2;mCherryTRAP embryos after sorting for mCherry expression (Fig. 4A). We chose E14.5 because of the observed expression peak of various BBB-associated genes in the TRAP-seq data. We obtained a minimum count of 125,000 reads per cell with about 40% of these reads mapped uniquely to a single genomic location (fig. S9). Hierarchical clustering of the single-cell transcriptomes suggested the presence of four major groups of cells (Fig. 4B). Sixty-one cells were identified as ECs, 4 as microglia, and 12 as neuronal cells because of their distinct gene expression profile (Fig. 4C). The poor enrichment for ECs was caused by sorting cells using a 488-nm laser that provides only about an 8% excitation of mCherry. The identity of three cells remained undefined. One endothelial subset showed a high expression of transcripts related to proliferation and cell cycle. We found a remarkable heterogeneity in the expression of genes encoding well-known general EC markers (fig. S10A) and BBB-specific markers (fig. S11) among individual ECs. The amount of heterogeneity for genes encoding housekeeping markers was higher in our ECs compared to our neuronal cell population or previously published HEK293T cells (fig. S12 and tables S25 to S28) (25). However, our single-cell data further verified that Nkd1, Ppard, and Ttyh2 were expressed by distinct brain ECs (fig. S6, C and D). Notably, we found two microglia expressing mCherry but no general endothelial or BBB markers (fig. S10B), suggesting rare nonspecific expression of the Cdh5CreERT2 driver. Most of the ECs expressed the previously characterized transcription factors Foxf2, Foxq1, Lef1, Ppard, and Zic3, albeit at different abundances and in varying combinations between individual cells (fig. S13). In contrast, Foxl2 and Zfp551 were expressed by only a few ECs. Overall, these data indicate a strong heterogeneity among brain-derived ECs and suggest an increasing complexity of the CNS endothelium.

Fig. 4 Single-cell RNA-seq analysis distinguishes ECs from other cell lineages.

(A) Schematic of the experimental procedure to compare molecular profiles of individual single brain ECs from two E14.5 embryos expressing the mCherry-Rpl10a fusion protein in the vasculature (red). (B) Unsupervised hierarchical clustering of expression in 80 individual cells (columns) across all genes (rows) with ≥5 RPKM in at least one sample. The following groups of cells were identified: 35 ECs (EC 1, red), 26 ECs with a proliferation signature (EC 2, yellow), 4 microglia (blue) and 12 neurons (green). The identity of three cells remained unclear. (C) PCA for 80 profiled individual cells from E14.5 embryos.

Foxf2, Foxq1, Ppard, and Zic3 associate with ECs differentiation

Despite the observed high degree of heterogeneity between individual ECs, we attempted to pinpoint specific pattern characteristic of the maturation of brain ECs. Unsupervised hierarchical clustering of the 61 individual ECs using E14.5 brain-specific transcripts identified two subgroups of ECs characterized by either a low or high degree of BBB maturation (Fig. 5, A to C). These groups did not differ in the expression of genes encoding general EC markers (Fig. 5D), but cells that showed high expression of BBB differentiation markers also showed increased expression of angiogenesis-associated transcripts (Fig. 5E) and reduced expression of genes related to proliferation and cell cycle (Fig. 5F). The increased expression of genes encoding differentiation markers observed in a subpopulation of ECs coincided with the increased expression of Foxf2, Foxq1, Ppard, and Zic3 (Fig. 5G), suggesting that these transcription factors potentially contribute to the BBB maturation program.

Fig. 5 EC differentiation correlates with expression of Foxf2, Foxq1, Ppard, or Zic3.

(A) Unsupervised hierarchical clustering of 61 E14.5 individual ECs across brain vasculature–enriched transcripts. Colors indicate the two groups of single cells that exhibit low (blue) or high (yellow) expression of genes encoding differentiation markers. (B) PCA for profiled ECs using brain-only transcripts enriched in the endothelium. Colors indicate low (blue) or high (yellow) expression of genes encoding differentiation markers. (C to G) Box plots of calculated z scores of genes encoding differentiation (C), general EC (D), angiogenesis (E), proliferation and cell cycle (F) markers, or individual brain EC–enriched transcription factors (G) showing low (blue) and high (yellow) abundance of brain EC–enriched transcripts. Statistical significance was calculated using Welch’s two-sample t test for normally or Wilcoxon signed-rank test for not normally distributed data (***P < 0.001, **P < 0.01, *P < 0.05). Normal distribution was tested using the Shapiro-Wilk test.

Foxf2 and Zic3 induce expression of genes encoding BBB marker in vitro

The complexity of the BBB has made developing in vitro systems that reproduce the many features and unique functions of the BBB challenging. We tested the three transcripts encoding transcription factors—Foxf2, Foxq1, and Zic3—that were the most statistically significantly enriched in brain ECs for their ability to induce the expression of BBB-associated transcripts in two EC lines, human brain microvascular endothelial cells (HBMECs) and human umbilical vein endothelial cells (HUVECs), as well as in human embryonic kidney 293 (HEK293FT) cells. Cells were infected with lentiviral vectors carrying Foxf2, Foxq1, ZIC3, or all three combined and cultured for 7 days, after which we quantified the expression of BBB-associated transcripts by quantitative real-time polymerase chain reaction (qRT-PCR) analysis. Overexpression of Foxf2 and ZIC3 individually as well as the combination of all three transcription factors induced the expression of the BBB markers ABCB1, SLCO2B1, and TNFRSF19 in HBMEC and HUVEC but not in HEK293FT cells (Fig. 6A).

Fig. 6 Foxf2 and ZIC3 induce expression of BBB-related transcripts in vitro.

(A) qRT-PCR analysis of purified RNA from HBMEC, HUVEC, and HEK293FT cells expressing Foxf2, Foxq1, ZIC3, or all three of these transcription factors (TFs) revealed transcription factor–mediated induction of BBB markers. Values are means ± SD of n = 3 independent experiments. (B) PCA for HUVECs transfected with control (empty) vector or vector containing Foxf2, Foxq1, ZIC3, or all three transcription factors combined, showing three biological replicates. PC1, first principal component explaining 30% of variance; PC2, second principal component explaining 11% of variance. (C) Venn diagram showing the number of transcripts induced (fold change > 4, RPKM > 5) by Foxf2, ZIC3, or the three transcription factors combined in HUVECs. (D) Expression of increased BBB transcripts in HUVECs transduced with control (empty vector), Foxf2, Foxq1, ZIC3, or all three transcription factors combined. n = 3 independent experiments.

To further explore the capacity of these transcription factors to induce transcripts associated with BBB differentiation in ECs, we performed RNA-seq on HUVECs infected with Foxf2, Foxq1, or ZIC3 individually; a combination of all three transcription factors; or the control vector. PCA revealed that cells expressing Foxf2 or the combination of all three transcription factors differed considerably from other groups (Fig. 6B). We identified transcripts specifically induced by each transcription factor using DESeq (>4-fold higher expression and RPKM > 5). Foxq1 did not induce the expression of BBB transcripts, possibly due its low expression in the infected HUVECs. On the other hand, we found 12 and 4 BBB-associated transcripts induced by Foxf2 and ZIC3, respectively (Fig. 6, C and D). These genes overlapped highly with the 18 genes induced by the combination of transcription factors. This suggests that Foxf2 and Zic3 are directly involved in the maturation of the BBB, but they may promote the expression of distinct sets of genes.

DISCUSSION

Although gene expression profiles of the CNS endothelium have been investigated previously (1012), cellular and molecular mechanisms regulating brain-specific vascular development and differentiation are still poorly understood. This study provides a comprehensive data set for obtaining further insights into the maturation of the brain vasculature during development. We defined translational profiles of murine brain ECs at different embryonic stages that reveal the distinct nature of the brain endothelium compared to other organ-specific vascular systems. It has been shown that the formation of the BBB is already initiated at E9.5 (19); however, without contradiction, our data indicate that the vast majority of BBB-associated transcripts begin to be translated at E11.5 or E12.5. Further assessment of the expression of these transcripts during embryonic development revealed notable diversity, with different sets of transcripts increasing at different developmental stages. This observation supports the idea of a continuous maturation process of the BBB during development to the point of full maturation in the adult (5). In addition, we identified genes enriched in the CNS vasculature (Nkd1, Ppard, and Ttyh2) that have not been previously reported to be specifically enriched in brain ECs. The enrichment of Myt1l in the brain ECs from our TRAP-seq data remains an unsolved issue. The identification of microglia expressing mCherry but no general endothelial or BBB marker indicates a rare nonspecific expression of the Cdh5CreERT2 driver line. However, Cdh5CreERT2 is a well-established and commonly used driver line, and no nonspecificity has been reported. The absence of high Myt1l–expressing cells in our single-cell data leaves the possibility of Myt1l being a strong background transcript in our TRAP-seq data. Although our TRAP-seq data provide detailed insights into the expression of known brain EC transcripts during embryonic development, novel CNS vasculature–enriched transcripts must be verified in further studies.

The observation of angiogenesis increasing during early postnatal development (26) led to the assumption that the rate of angiogenesis increases from E9.5 until the early postnatal period (5). Surprisingly, we observed a declining expression of transcripts encoding angiogenic markers in the brain endothelium at later embryonic stages. Therefore, we hypothesize that angiogenesis declines in late embryonic stages, followed by rapid burst right after birth. Further studies are required to verify this hypothesis.

The deletion of Ctnnb1 at E12.5 disturbed the normal dynamics of the brain endothelium–specific translatome and strongly reduced the expression of genes encoding BBB markers at E14.5, supporting the importance of Wnt signaling during early BBB development. A similar effect was observed at E17.5 after 2 days of tamoxifen treatment to delete Ctnnb1, suggesting a continuing role for this factor in later BBB development. It has been shown that canonical Wnt signaling can be detected during formation and maintenance of the CNS blood vessel network in embryonic and adult zebrafish (20). Considering the observed magnitude of decline in expression of genes encoding differentiation-associated marker in Ctnnb1-depleted ECs, our data are in agreement with the findings of Zhou et al. (21), who showed an essential role for the Wnt signaling pathway in the maintenance of the BBB.

Notably, the deletion of Ctnnb1 led to increased expression of genes encoding various extracellular matrix components. Increased abundance of extracellular matrix molecules is linked to the cellular stress response in various cancer models (22). Given that collagens, integrins, and laminins are involved in regulating cell adhesion, migration, and differentiation, the observed increase in expression of genes encoding extracellular matrix proteins might represent a compensatory mechanism for stimulating Ctnnb1 signaling. For example, it has been shown that laminin α5 can positively regulate the Wnt signaling pathway in the intestine or skin (24). Alternatively, the increased expression of genes encoding extracellular matrix components might represent a compensatory mechanism of the KO endothelium for establishing an endothelial barrier function by increasing cell-matrix adhesion properties (23). Future experiments are needed to better characterize the nature of the increased expression of extracellular matrix proteins.

The published literature is consistent in noting a requirement for the canonical Wnt pathway for vascularization of the CNS and induction of BBB differentiation, but there are contradictory observations regarding the impact of Ctnnb1 signaling on angiogenesis (5). The assumption that canonical Wnt signaling positively regulates CNS-specific angiogenesis originated from the observed vascularization defect in mice with Ctnnb1 deletion in the brain endothelium (6, 8). Xu et al. (27) reported that activation of the canonical Wnt-Ctnnb1 pathway in HBMECs induced angiogenesis in vitro. On the other hand, Ctnnb1 signaling inhibits angiogenesis in a mouse glioma model (28). Further complexity is added by the observation that Dickkopf1 suppresses angiogenesis, whereas Dickkopf2 promotes angiogenesis, although both are known Wnt antagonists that inhibit Ctnnb1 signaling (29). In our hands, deletion of Ctnnb1 in the brain endothelium did not suppress the expression of angiogenesis-associated transcripts. This agrees with the observation that a Ctnnb1 KO impairs vascular formation in the CNS but does not completely abolish it, because angiogenic sprouts penetrating the neuroepithelium are still detectable (6, 21). Therefore, we suggest that Ctnnb1 signaling does not positively regulate the induction of angiogenesis during embryonic development. Because our data indicate that the canonical Wnt pathway does not influence the proliferation of ECs in general, we speculate that Ctnnb1 signaling might regulate CNS vascularization by either promoting vessel anastomosis (30) or stabilizing newly formed vessels (31). The very observation of increased expression of genes encoding angiogenic markers in Ctnnb1-deficient compared to control ECs at E17.5 suggests that the degree of CNS vasculature maturation negatively correlates with the induction of angiogenesis.

The observed heterogeneity within the brain EC population further challenges our understanding of CNS vascularization during development. Although certain variation in the expression of BBB-associated genes was expected on the basis of previously published data (32), we were surprised not only by the difference in the expression of genes encoding factors required for BBB formation but also by the difference in the expression of general EC and housekeeping markers between individual ECs at E14.5. It has been suggested that the cell cycle phase makes an important contribution to transcriptional bursting activity (33). E14.5 brain ECs are highly proliferative and plastic (34). However, we were not able to find a statistically significant correlation between the transcriptional heterogeneity and the expression of genes encoding proliferation and cell cycle markers. The heterogeneity might also represent specific adaptations and specializations of ECs to very selective requirements of their immediate neighborhood, supporting the idea of functional education of ECs by the microenvironment (1). Further studies analyzing higher numbers of single ECs and additional time points might provide more insights into the reasons for heterogeneity of brain ECs during embryonic development.

Our analysis revealed a set of genes encoding transcription factors (Foxf2, Foxl2, Foxq1, Lef1, Ppard, and Zic3) that may direct maturation of the BBB. Although these transcription factors have been reported as being specifically expressed in the CNS vasculature previously (8, 10, 11), our data indicate that the expression of Foxf2, Foxq1, Ppard, and Zic3 is actually associated with the progress of BBB differentiation in vivo. Furthermore, we provide evidence that Foxf2 and ZIC3 can induce features of the BBB in vitro. Foxf2 induced TNFRSF19 expression in HUVECs, which suggests a role for Foxf2 downstream of Ctnnb1 but upstream of Tnfrsf19 (10). Further investigations will be necessary to verify the potential role of the identified transcription factors in orchestrating distinct signaling pathways that underlie BBB differentiation. It remains to be seen whether these transcription factors act in parallel or rather serially, and the synergy between Foxf2 and ZIC3 already indicates that combinatorial actions of several factors may be needed to firmly establish BBB characteristics.

This study provides insights into the temporal and cellular complexities of the developing embryonic brain endothelium. Our data raise many questions and provide a basis for further studies aiming to unravel cellular and molecular mechanisms that underlie vascular development and differentiation in the CNS. We anticipate that further investigation of the identified transcription factors will be a fruitful approach for understanding and eventually manipulating the BBB development and to engineer more advanced BBB in vitro models.

MATERIALS AND METHODS

Animals and genotyping

Animal care and research protocols were in accordance with institutional guidelines and approved by the Etiska Nämnden on animal use. For induction of Cre drivers, tamoxifen (Sigma) was resuspended in corn oil (20 mg/ml) and administered to pregnant females by oral gavage 2 days before harvesting and analysis of the embryos. For staging of embryos, the morning a vaginal plug was observed was designated as E0.5. Gt(ROSA)26Sor-mCherry-Rpl10a (mCherryTRAP) (13), Tg(Cdh5-cre/ERT2)1Rha (Cdh5CreERT2) (35), and β-cateninflox/flox (floxed Ctnnb1) (36) mice and embryos were genotyped by PCR using the following primers: R26-mCherry-Rpl10a-forward (TACACCATCGTGGAACAGTAC), R26-mCherry-Rpl10a-reverse (GTAGTTCTTCAGGCTGATCTG); R26-wt-forward (GCGGATCACAAGCAATAATA), R26-wt-reverse (TTTCTGGGAGTTCTCTGCTG); Generic-Cre-forward (CACGACCAAGTGACAGCAAT), Generic-Cre-reverse (AGAGACGGAAATCCATCGCT); and Ctnnb1-forward-1 (AAGGTAGAGTGATGAAAGTTGTT), Ctnnb1-forward-2 (TACACTATTGAATCACAGGGA), and Ctnnb1-reverse (CACCATGTCCTCTGTCTATTC) (6, 13). To activate the mCherry transgene in the endothelium, we crossed male Cdh5CreERT2+/− with female mCherryTRAP+/+ mice. To generate Ctnnb1 KO animals, we first crossed male Cdh5CreERT2+/−;mCherryTRAP+/− with female Ctnnb1flox/flox mice and subsequently obtained male Cdh5CreERT2+/−;mCherryTRAP+/− Ctnnb1flox/WT with female Ctnnb1flox/flox mice.

Histological analysis

Cortical slices from adult mCherryTRAP mice were obtained and stained as described (13). In brief, adult brains were removed fresh, immersion-fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight at 4°C, and then placed in PBS with 30% sucrose for at least 72 hours at 4°C before sectioning on a cryostat. Brains were sectioned at 30 to 40 μm and kept free-floating in PBS. For immunohistochemistry, slide-mounted sections were incubated in a blocking and permeabilization solution containing 10% donkey serum and 0.25% Triton X-100 for 1 hour at room temperature, followed by incubation in primary antibody solution overnight at 4°C. The sections were incubated with appropriate secondary antibodies conjugated to Alexa fluorophores for 2 hours at room temperature before mounting in Immu-Mount (Thermo Scientific). The brain sections were stained free-floating and subsequently mounted onto slides. Confocal microscopy was performed on a Zeiss LSM510 confocal microscope. The following primary antibodies were used for immunostaining: a rabbit antibody recognizing dsRed (Clontech, 632496, 1:500) and a rat antibody recognizing Pecam1 (CD31, BD Biosciences, 553370, 1:500). Primary antibodies were detected using the following secondary antibodies: a donkey antibody recognizing rat immunoglobulin G (IgG) coupled with Alexa Fluor 488 (Thermo Fisher Scientific, A-21208, 1:1000) and a goat antibody recognizing rabbit IgG coupled with Alexa Fluor 647 (Thermo Fisher Scientific, A-21245, 1:1000).

Bulk RNA purification and RNA-seq library preparation

Embryos were harvested and scored for mCherry expression using a fluorescence microscope. The brain (forebrain and part of the diencephalon), head (without brain), heart, kidney, limb, liver, or lung was dissected in ice-cold PBS. Respective tissue from one to six embryos (age-dependent) was immediately homogenized in ice-cold polysome extraction buffer [20 mM Hepes (pH 7.4), 150 mM KCl, 5 mM MgCl2, 0.5% NP-40, 0.5 mM dithiothreitol, cycloheximide (100 mg/ml) (Sigma), protease inhibitors (Roche), and recombinant ribonuclease (RNase) inhibitor (40 U/ml) (Promega)] with a hand glass homogenizer. Homogenates were incubated at 4°C with end-over-end rotation for 10 min. Subsequently, crude extracts were cleared by three centrifugation steps (2600, 8600, and 16,100g each for 5 min at 4°C). A small fraction of the clear extract was taken for total RNA purification. For TRAP, anti–red fluorescent protein magnetic beads (M165-9, MBL), washed two times with polysome extraction buffer, were added to the supernatant, and the mixture was incubated at 4°C with end-over-end rotation for 30 min. Beads were subsequently collected on a magnetic rack and washed four times with a high-salt polysome wash buffer [20 mM Hepes (pH 7.4), 350 mM KCl, 5 mM MgCl2, 0.5% NP-40, 0.5 mM dithiothreitol, cycloheximide (100 mg/ml), protease inhibitors, and recombinant RNase inhibitor (40 U/ml)]. RNA was eluted from the beads by incubating beads in RLT buffer (Rneasy Micro Kit, Qiagen) + β-mercaptoethanol (10 μl/ml) for 5 min at room temperature. Eluted TRAP, as well as total RNA, was purified using RNeasy Micro Kit (Qiagen) following the manufacturer’s instructions including in-column deoxyribonuclease (DNase) digestion. The RNA quality was validated using the Agilent RNA 6000 Pico or Nano Kit and an Agilent 2100 bioanalyzer (Agilent Technologies Inc.). RNA (100 ng to 1 mg) was used to generate RNA-seq libraries with the TruSeq RNA Sample prep kit v2 kit (Illumina) following the manufacturer’s protocol. The sequencing libraries were validated using the Agilent DNA 1000 kit and Agilent bioanalyzer. All samples had an average size of ~300 base pairs. Equal amounts of three to four indexed sequencing libraries were pooled and diluted to a final concentration of 2 nM and sequenced on the Illumina HiSeq 2000 platform. The RNA-seq data for E14.5 forebrain and kidney had been published previously (13).

Purification of single cells and RNA-seq library preparation

Two forebrains each of positive and negative Cdh5CreERT2;mCherryTRAP E14.5 embryos were dissected in ice-cold PBS. The tissue was diced with a scalpel and subsequently enzymatically dissociated in PBS containing collagenase type II (1 mg/ml) (Worthington) and DNase I (1 μl/ml) (Roche) at 37°C for 30 min with gentle sequential trituration. The cell suspension was filtered through a 40-μm cell strainer, and 20 ml of Dulbecco’s modified Eagle’s medium (DMEM; Thermo Fisher Scientific) containing 10% fetal bovine serum (FBS; PAA) was added. After centrifugation at 800g for 5 min, the cell pellet was resuspended in 2 ml of PBS containing 2% FBS. Cells were sorted on a BD FACSAria III Cell Sorting system (BD Bioscience) based on mCherry fluorescence using the 488-nm laser. Note that the 488-nm laser provides only a roughly 8% excitation of mCherry. mCherryTRAP-negative embryos were used to define background fluorescence. Forward and side scatter analyses were used as gates to avoid cell debris and doublets. Single cells were collected in a 96-well plate containing 4 μl of a mild hypotonic lysis buffer [0.1% Triton X-100 (Sigma), recombinant RNase inhibitor (1 U/μl) (Promega), 2.5 μM anchored oligo-dT primer (5′-AAGCAGTGGTATCAACGCAGAGTACT30VN-3′, Biomers), and 2.5 mM dNTP mix (Fermentas)] in each well. Libraries were generated following Smart-seq2 (25) and Tn5 transposase tagmentation (37) protocols. Single-cell lysates were denatured at 72°C for 3 min and immediately placed on ice afterward. Seven microliters of the first-strand reaction mix [0.50 μl of SuperScript II reverse transcriptase (200 U/μl, Invitrogen), 0.25 μl of recombinant RNase inhibitor (20 U/μl, Promega), 2 μl of Superscript II First-Strand Buffer (5×, Invitrogen), 0.25 μl of dithiothreitol (100 mM, Invitrogen), 2 μl of betaine (5 M, Sigma), 0.9 μl of MgCl2 (100 mM, Sigma), 1 μl of template-switching oligonucleotides (10 μM, 5′-AAGCAGTGGTATCAACGCAGAGTACATrGrG+G-3′, Exiqon), and 0.1 μl of nuclease-free water (Gibco)] was added to each sample. Reverse transcription reaction was carried out by incubating at 42°C for 90 min, followed by 10 cycles of 50°C for 2 min, 42°C for 2 min. The reverse transcriptase was inactivated by incubation at 70°C for 15 min. Fifteen microliters of PCR preamplification mix [25 μl of KAPA HiFi HotStart ReadyMix (2×, KAPA Biosystems), 1 μl of ISPCR primers (10 μM, 5′-AAGCAGTGGTATCAACGCAGAGT-3′, Biomers), and 14 μl of nuclease-free water (Gibco)] was added to each sample. Preamplification reaction was carried out by incubating at 98°C for 3 min, then 18 cycles of 98°C for 20 s, 67°C for 15 s, 72°C for 6 min, with a final extension at 72°C for 5 min. PCR was purified using a 1:1 ratio of AMPure XP beads (Beckman Coulter), with the final elution performed in 15 μl of elution buffer (Qiagen). Library size distribution was checked on a High-Sensitivity DNA chip (Agilent Bioanalyzer). Tagmentation was carried out using 500 pg of DNA, 4 μl of 40% (w/v) polyethylene glycol 8000, 2 μl of 10× TAPS buffer (100 mM TAPS-NaOH, 50 mM MgCl2, pH 8.5, at 25°C), and 1 μl of Tn5 (provided by G. Winberg) in a 20-μl reaction and incubated for 10 min at 55°C. Tn5 was stripped off from tagmented DNA by 10-min incubation at room temperature with 5 μl of 0.2% SDS. The final indexing PCR amplification [25 μl of tagmented DNA, 10 μl of Fidelity Buffer (5×, KAPA Biosystems), 1.5 μl of dNTPs (10 mM, KAPA Biosystems), 1 μl of KAPA HiFiDNApolymerase (1 U/μl, KAPA Biosystems), 9.5 μl of water (Gibco), and 1 μl each of Index primer 1 (0.5 μM, N7xx, Biomers), Index primer 2 (0.5 μM, N5xx, Biomers), and Anchoring primer mix (10 μM, Biomers)] was carried out by incubation for 3 min at 72°C, 30 s at 95°C, 14 cycles of 10 s at 98°C, 30 s at 63°C, 30 s at 72°C, followed by 5 min at 72°C. Purification of the final libraries was performed with 50 μl of Agencourt AMPure XP beads, and the final elution was done with 15 μl of EB solution. Samples quality was checked using Agilent DNA 1000 kit, and quantity was analyzed with Qubit high-sensitivity DNA kit (Invitrogen). Fifty-seven libraries were pooled for sequencing on the Illumina HiSeq 2000 instrument at a final concentration of 2 nM each.

RNA-seq analysis

Sequence reads were mapped against the Mus musculus genome assembly (Genome Reference Consortium GRCm38, UCSC version mm10) using the RUM v. 2.04 pipeline (38). Gene expression (RPKM) calculations were performed using rpkmforgenes.py (39). The mm10 refGene.txt file was downloaded from UCSC 20130114. Differential expression analysis was performed with the Bioconductor package DESeq (15). GO was carried out using DAVID 6.7 (40, 41). Cluster 3.0 was used for PCA and hierarchical clustering (centered correlation centroid linkage) (42, 43). Heatmaps were generated using TreeView 1.1.6r2 (44). For statistical analysis, R has been used. For comparison of two normally distributed data, Welsh two-sample t test has been used. For non-normally distributed data sets, Wilcoxon rank sum test was used. Normality was tested by Shapiro-Wilk test. Analysis of variance (ANOVA) by ranks for more than two dependent samples was calculated according to Friedman with Nemenyi as post hoc test. ANOVA on the absolute deviations from the median for groups was calculated as robust Brown-Forsythe Levene–type test.

Cell culture

HBMEC (Cell Systems) and HUVEC (PromoCell) were cultured in EBM2 bullet kit media (Lonza) on gelatin-coated flasks according to the instructions provided by the supplier. HEK293FT cells (Thermo Fisher Scientific) were cultured in high-glucose DMEM (Thermo Fisher Scientific) containing 10% FBS (PAA) at 37°C in 5% CO2.

Viral transduction

The lentiviral vector was constructed from pRRLSIN.cPPT.PGK-GFP.WPRE (Addgene plasmid #12252) by replacing the PGK/GFP cassette with the CMV promotor from pCDNA3.1. Mouse Foxf2 (MGC clone BC137947), mouse Foxq1 (MGC clone BC047155), or human ZIC3 (MGC clone BC113393) was cloned downstream of the CMV promotor. Lentiviral plasmids were transiently transfected into HEK293FT using Lipofectamine 2000 (Thermo Fisher Scientific) following the manufacturer’s instructions. Viral particles were produced in RPMI 1640 media (Thermo Fisher Scientific) containing 10% FBS, 1% MEM Non-Essential Amino Acids, 1% Hepes, and 1% l-glutamine; harvested after 24 hours; and filtered through a 0.45-μm pore-size filter. Transduction of the target cell lines was carried out in gelatin-coated 12-well plates containing 5 × 104 cells per well in 600 μl of RPMI medium with 200 μl of viral supernatant for 4 hours, and cells were cultivated for 7 days in growth medium.

Quantitative real-time polymerase chain reaction

qRT-PCR was performed as described (13). Data were normalized to GAPDH and presented as means ± SD. The following primers were used: GAPDH, CACCGTCAAGGCTGAGAACG (forward) and GCCCCACTTGATTTTGGAGG (reverse); ABCB1, TGAATCTGGAGGAAGACATGAC (forward) and CCAGGCACCAAAATGAAACC (reverse); APCDD1, GGAGTCACAGTGCCATCACAT (forward) and CCTGACCTTACTTCACAGCCT (reverse); SLCO2B1, GGCGAAAGGTCTTAGCAGTCA (forward) and GGCCATCCTGCTTCTTCGT (reverse); and TNFRSF19, ACATGGAGTGTGTGCCTTGTG (forward) and GACGCGATCTTCACGAGGTT (reverse).

In situ hybridization

In situ hybridization of paraffin-embedded E14.5 C57BL/6 head sections was performed as described (45). Partial cDNA sequences were amplified from mouse brain cDNA library, cloned into pcDNA3 vector (Invitrogen) using Bam HI and Xba I, and used to prepare digoxigenin-labeled RNA probes. The following primers were used: Nkd1, GGGAAAGGATCCCCTCGAATCTCCAACCCCAC (forward) and GGGAAATCTAGAGGTGTTCGTGTCTCTGGACG (reverse); Ppard, GGGAAAGGATCCTATGCGCATGGGACTCACTC (forward) and GGGAAATCTAGAGATCTTGCAGATCCGATCGC (reverse); and Ttyh2, GGGAAAGGATCCGACCTAGAACAGCACCTGGC (forward) and GGGAAATCTAGAGTCAATGAGCGCTGGAAGG (reverse). Probe for Flt1 was published previously (46).

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/487/eaag2476/DC1

Fig. S1. Gene expression in the embryonic CNS vasculature at E14.5.

Fig. S2. Translated transcripts encoding transporters in the embryonic CNS vasculature.

Fig. S3. Translated transcripts encoding molecular markers related to physical barrier properties in the embryonic CNS vasculature.

Fig. S4. Translated transcripts encoding transcription factors in embryonic vasculature.

Fig. S5. Cdh5CreERT2 driver line activity is not strictly limited to ECs only.

Fig. S6. Validation of TRAP-seq results by in situ hybridization.

Fig. S7. Brain EC maturation during embryonic development.

Fig. S8. Ctnnb1 deletion–dependent molecular changes in brain ECs.

Fig. S9. Single-cell sequence reads and mapping.

Fig. S10. High EC heterogeneity at the single-cell level for genes encoding general EC markers or mCherry.

Fig. S11. High EC heterogeneity at the single-cell level for genes encoding specific BBB markers.

Fig. S12. High level of EC heterogeneity.

Fig. S13. High EC heterogeneity at the single-cell level for genes encoding transcription factors identified as enriched in CNS vasculature by TRAP-seq.

Table S1. Specifically translated genes in brain ECs.

Table S2. Brain EC–enriched transcripts.

Table S3. Transcripts shared between ECs from all other organs.

Table S4. Brain EC–depleted transcripts.

Table S5. Specifically translated genes in brain ECs during embryonic development.

Table S6. Regulated transcripts in brain ECs during embryonic development.

Table S7. Transcripts with increasing translation in brain ECs during embryonic development.

Table S8. Transcripts with decreasing translation in brain ECs during embryonic development.

Table S9. Genes with increasing and subsequently decreasing translation in brain ECs during embryonic development.

Table S10. Genes with decreasing and subsequently increasing translation in brain ECs during embryonic development.

Table S11. E14.5 brain EC–enriched transcripts during embryonic development.

Table S12. E14.5 shared EC transcripts during embryonic development.

Table S13. E14.5 brain EC–depleted genes during embryonic development.

Table S14. Genes encoding angiogenesis marker in brain ECs during embryonic development.

Table S15. Genes encoding proliferation and cell cycle marker in brain ECs during embryonic development.

Table S16. Ctnnb1 KO–deleted transcripts in brain ECs at E14.5.

Table S17. Ctnnb1 KO–deleted transcripts in brain ECs at E17.5.

Table S18. E14.5 brain EC–enriched transcripts in Ctnnb1 KO brain ECs.

Table S19. Ctnnb1 KO–enriched transcripts in brain ECs at E14.5.

Table S20. Ctnnb1 KO–enriched transcripts in brain ECs at E17.5.

Table S21. E14.5 shared EC transcripts in Ctnnb1 KO brain ECs.

Table S22. Genes encoding angiogenesis marker in Ctnnb1 KO brain ECs.

Table S23. E14.5 brain EC–depleted genes in Ctnnb1 KO brain ECs.

Table S24. Genes encoding proliferation and cell cycle marker in Ctnnb1 KO brain ECs.

Table S25. Genes encoding housekeeping markers in single cells.

Table S26. Genes encoding EC markers in single cells.

Table S27. Genes encoding neuronal markers in single cells.

Table S28. Genes encoding housekeeping markers in single HEK293T cells.

This is an article distributed under the terms of the Science Journals Default License.

REFERENCES AND NOTES

Acknowledgments: We are grateful to G. Winberg for providing Tn5 transposase; H. Storvall, D. Ramsköld, and R. Sandberg for invaluable support and discussions. Mapping of RNA-seq data was performed on resources provided by the Swedish National Infrastructure for Computing (SNIC) through Uppsala Multidisciplinary Center for Advanced Computational Science (UPPMAX) under Project b2012223. Funding: This work was supported by Ludwig Institute for Cancer Reasearch Ltd., Cancerfonden (CAN 2010/679; CAN 2013/747 to J.M.S.), the Deutsche Forschungsgemeinschaft [SFB 688/TP A16], Wenner-Gren Stiftelserna (postdoctoral fellowship to M.H.), and Wenner-Gren Fellows (to J.M.S.). Author contributions: The study was designed by M.H. and J.M.S. Experiments were performed by M.H., M.X.L., J.M.S., C.Y., J.K.-O., and B.H. Bioinformatics analysis and interpretation of data were done by M.H., J.M.S., S.K., D.D., and M.G. The manuscript was written by M.H., D.D., and M.G. Competing interests: The authors declare that they have no competing interests. Data and materials availability: RNA-seq data have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE79306 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE79306). The RNA-seq data for E14.5 forebrain and kidney are published (13) and are accessible through GEO Series accession number GSE51619 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE51619).
View Abstract

Navigate This Article