CRISPR/Cas9-mediated gene knockout reveals a guardian role of NF-κB/RelA in maintaining the homeostasis of human vascular cells

Vascular cell functionality is critical to blood vessel homeostasis. Constitutive NF-κB activation in vascular cells results in chronic vascular inflammation, leading to various cardiovascular diseases. However, how NF-κB regulates human blood vessel homeostasis remains largely elusive. Here, using CRISPR/Cas9-mediated gene editing, we generated RelA knockout human embryonic stem cells (hESCs) and differentiated them into various vascular cell derivatives to study how NF-κB modulates human vascular cells under basal and inflammatory conditions. Multi-dimensional phenotypic assessments and transcriptomic analyses revealed that RelA deficiency affected vascular cells via modulating inflammation, survival, vasculogenesis, cell differentiation and extracellular matrix organization in a cell type-specific manner under basal condition, and that RelA protected vascular cells against apoptosis and modulated vascular inflammatory response upon tumor necrosis factor α (TNFα) stimulation. Lastly, further evaluation of gene expression patterns in IκBα knockout vascular cells demonstrated that IκBα acted largely independent of RelA signaling. Taken together, our data reveal a protective role of NF-κB/RelA in modulating human blood vessel homeostasis and map the human vascular transcriptomic landscapes for the discovery of novel therapeutic targets. Electronic supplementary material The online version of this article (10.1007/s13238-018-0560-5) contains supplementary material, which is available to authorized users.

ABSTRACT Vascular cell functionality is critical to blood vessel homeostasis. Constitutive NF-κB activation in vascular cells results in chronic vascular inflammation, leading to various cardiovascular diseases. However, how NF-κB regulates human blood vessel homeostasis remains largely elusive. Here, using CRISPR/Cas9-mediated gene editing, we generated RelA knockout human embryonic stem cells (hESCs) and differentiated them into various vascular cell derivatives to study how NF-κB modulates human vascular cells under basal and inflammatory conditions. Multi-dimensional phenotypic assessments and transcriptomic analyses revealed that RelA deficiency affected vascular cells via modulating inflammation, survival, vasculogenesis, cell differentiation and extracellular matrix organization in a cell typespecific manner under basal condition, and that RelA protected vascular cells against apoptosis and modulated vascular inflammatory response upon tumor necrosis factor α (TNFα) stimulation. Lastly, further evaluation of gene expression patterns in IκBα knockout vascular cells demonstrated that IκBα acted largely independent of RelA signaling. Taken together, our data reveal a protective role of NF-κB/RelA in modulating human blood vessel homeostasis and map the human vascular transcriptomic landscapes for the discovery of novel therapeutic targets.

INTRODUCTION
Blood vessels carry nutrients and oxygen via blood stream throughout the body. Blood vessel homeostasis is critical to health. Disorders of blood vessels impair vascular cell function, resulting in the pathogenesis of severe cardiovascular diseases like myocardial infarction, atherosclerosis, and stroke.
Blood vessels consist of three layers, namely the tunica intima, tunica media, and tunica adventitia. The major cell types of these three layers are vascular endothelial cells (VECs), vascular smooth muscle cells (VSMCs), and mesenchymal stem cells (MSCs), respectively. VECs compose a monolayer of cells called endothelium that modulates nutrients transport, vascular tone, host defense, homeostasis and angiogenesis (Galley and Webster, 2004). VSMCs constitute the majority of blood vessels and regulate local blood pressure via vasoconstriction and vasodilation. MSCs are part of the adventitia that modulates vascular inflammatory response, trophic supply, and vessel damage repair (Breitbach et al., 2018;Caplan and Correa, 2011;da Silva Meirelles et al., 2008). Vascular dysfunction manifests various cellular abnormalities like neointimal growth, thrombosis, inflammation, and vascular apoptosis and further leads to ischemia and infarction, thus providing prognostic parameters of cardiovascular diseases (Nedeljkovic et al., 2003;Zhang et al., 2018).
NF-κB is a transcription factor that modulates inflammatory response apart from regulating proliferation and survival (Perkins and Gilmore, 2006). In most cell types, the predominant form of NF-κB is a heterodimer composed of RelA/ p65 and p50, which can be activated by various stimuli such as TNFα and IL1β (Tilstra et al., 2011). NF-κB is sequestered in the cytoplasm in an inactive complex bound to the inhibitory proteins, inhibitors of NF-κB (IκBs). Upon stimulations, the degradation of IκB releases NF-κB, which subsequently translocates to the nucleus to modulate genes for the maintenance of cellular homeostasis (Liu et al., 2008;Perkins, 2007). Previous studies have shown that NF-κB activation is present in atherosclerotic lesion, resulting in chemokine secretion, adhesion of circulating monocytes to endothelium and persistent inflammatory response in VECs (Baker et al., 2011;Brand et al., 1996;Hajra et al., 2000). In VSMCs, NF-κB activation promotes extracellular matrix expression, neointimal proliferation, aberrant inflammatory response, and cell apoptosis, leading to vascular restenosis and plaque rupture (Rudijanto, 2007). Conversely, inhibition of NF-κB in vascular cells often exhibits therapeutic effects on cardiovascular diseases, such as the amelioration of vascular atherosclerotic lesion in apolipoprotein E (ApoE)deficient mice (Chiba et al., 2006;Gareus et al., 2008;Mallavia et al., 2013). However, little has been illustrated how NF-κB regulates the physiological functions of different human vascular cells.
Here, using CRISPR/Cas9-mediated genome editing, we generated RelA-deficient human embryonic stem cells (ESCs) to abolish NF-κB activity. We obtained RelA-deficient VECs, VSMCs and MSCs via directed differentiation and uncovered a protective role of NF-κB/RelA in different human vascular cells under basal and inflammatory conditions. RelA deficiency in VECs resulted in impaired vasculogenesis and monocyte-endothelium adhesion, as well as excessive proliferation. In MSCs, loss of RelA resulted in defective proliferation and dysregulated differentiation potential. Additionally, RelA deficiency promoted TNFα-induced apoptosis in different human vascular cells. Further transcriptomic analysis of RelA-deficient vascular cells revealed the cellular changes in vascular matrix organization, angiogenesis, inflammatory response, cell proliferation and survival. Lastly, analysis of gene expression patterns in various IκBα knockout vascular cells showed that IκBα acted largely independent of RelA signaling under basal condition and upon TNFα stimulation. Taken together, our data illustrate a novel protective role of NF-κB/RelA in human vascular cells and provide a platform for the study of NF-κB function in human adult stem cells and somatic cells.

Generation of RelA-deficient human ESCs
We generated RelA-deficient human ESCs (hESCs) targeting the exon 1 of RelA by CRISPR/Cas9-mediated genome editing (Fig. 1A). Successful removal of the targeted exon was verified by PCR (Fig. 1B) and the resulting loss of RelA protein was verified by Western blot (Fig. 1C). The RelA −/− ESCs exhibited common pluripotent stem cell features including typical colony morphology, expression of pluripotency markers OCT4, SOX2 and NANOG ( Fig. 1D and 1E). The in vivo differentiation ability of RelA −/− ESCs was validated by teratoma formation assay (Fig. 1F). Furthermore, karyotype and cell proliferation were each normal in RelA −/− ESCs when compared to wildtype (WT) controls ( Fig. 1G and 1H). These data suggest that the RelA −/− ESCs maintained typical hESC features.  To study how RelA is involved in human vasculature homeostasis, we generated human VECs, VSMCs and MSCs via directed differentiation of RelA −/− and WT ESCs. Cells were purified by fluorescent-activated cell sorting (FACS) using proper cell surface markers ( Fig. 2A-C). Cell purity was confirmed by immunofluorescent staining of additional VEC-specific markers, vWF and CD31 (Fig. 2D) and VSMC-specific markers, SM22 and Calponin (Fig. 2E). While RelA was predominantly retained in the cytoplasm of wildtype vascular cells, loss of RelA protein was verified in different types of RelA-deficient vascular cells by western blotting and immunofluorescent staining ( Fig. 2F and 2G).

RelA deficiency impaired vasculogenesis in VECs and perturbed differentiation potential in MSCs
We next investigated the functional consequences of RelA deficiency in different vascular cells. Although RelA −/− VECs had comparable ability to uptake acetylated low-density lipoprotein (Ac-LDL) compared to that of WT VECs (Fig. 3A), RelA deficiency severely interrupted tube formation of VECs in vitro (Fig. 3B), indicative of dysregulated VEC function.
Functional MSCs undergo adipogenesis, osteogenesis and chondrogenesis for regeneration in vivo (Uccelli et al., 2008). Here we tested whether RelA deficiency interferes with the differentiation potential of MSCs into adipocytes, osteoblasts and chondrocytes. Adipogenesis was slightly enhanced from RelA −/− MSCs, evidenced by an increase in oil red O staining (Fig. 3C) and upregulation of adipocytespecific genes like FABP4, PPARG and LPL, which were dramatically enriched in WT adipocytes relative to WT MSCs (Fig. 3D). Despite of increased osteoblast-specific gene expression, there were less calcium deposits stained by Von Kossa in derived RelA −/− osteoblasts ( Fig. 3E and 3F), indicative of aberrant osteogenesis from RelA −/− MSCs. Moreover, RelA deficiency resulted in defective chondrogenesis with less condensate structure stained by toluidine blue dye (Fig. 3G). These data indicate that RelA was required for maintaining the homeostasis of VECs and MSCs.

RelA deficiency impeded inflammatory response in human vascular cells
RelA is a well-known component of NF-κB heterodimer, which is a major mediator to regulate inflammation (Barnes and Karin, 1997;Salminen et al., 2008). Accordingly, we measured the mRNA levels of several NF-κB target genes involved in inflammation, including vascular cell adhesion molecule 1 (VCAM1), monocyte chemoattractant protein 1 (MCP1), IL6, and IL8 in human VECs, VSMCs, and MSCs. Loss of RelA led to reduced mRNA levels of these NF-κB target genes in rested VECs, VSMCs and MSCs (Fig. 4A). As expected, while RelA was translocated from cytosol to nucleus in WT VECs, VSMCs and MSCs upon TNFα treatment, no RelA immunofluorescence signal was observed in RelA −/− cells (Fig. 4B).
Given that monocyte recruitment to sites of vessel injury and subsequent adhesion to endothelium is critical for inflammation to repair damage and maintain blood vessel homeostasis (Kirton and Xu, 2010;Schober and Weber, 2005), we tested how RelA deficiency affected the adherence of monocytes to endothelium. RelA −/− VECs exhibited compromised ability in recruiting monocytes compared to WT VECs under basal condition, and this defect was even evident upon TNFα stimulation (Fig. 4C). Furthermore, RelA deficiency resulted in reduced expression of genes implicated in monocyte adhesion to vascular endothelium including Selectin E, VCAM1 and intercellular adhesion molecule 1 (ICAM1) upon TNFα treatment ( Fig. 4D and 4E). Thus, our data demonstrate that RelA deficiency impeded inflammatory response in human vascular cells.

RelA deficiency altered proliferative ability in human vascular cells
We next checked how RelA affected the proliferative ability of various human vascular cells. RelA deficiency enhanced proliferation in VECs, but not in VSMCs ( Fig. 5A-D). In MSCs, however, RelA deficiency led to decreased proliferative ability at early passages (Passage 3) ( Fig. 5E and 5F). Given that MSCs as adult stem cells have the self-renewal ability, we further studied how RelA affected MSC proliferation after serial passaging and found even worse proliferative ability in late-passage RelA −/− MSCs (Passage 7) ( Fig. 5G and 5H). Therefore, RelA demonstrated distinct regulatory activities towards the proliferation of VECs, VSMCs and MSCs. RelA safeguards human vascular cell homeostasis

RESEARCH ARTICLE
RelA deficiency promoted TNFα-induced apoptosis in human vascular cells NF-κB modulates the expression of anti-apoptotic genes, thus promoting cell survival (Kucharczak et al., 2003). Here we tested how RelA affected the survival of human vascular cells under basal and inflammatory conditions. Consistently, the protective effects of NF-κB on inhibiting TNFα-induced apoptosis were impaired by RelA deficiency in various vascular cells. Interestingly, at baseline RelA deficiency exhibited an anti-apoptotic effect in VSMCs, but not in VECs or MSCs ( Fig. 6A and 6B). These data indicate that RelA protected human vascular cells against TNFα-induced cell apoptosis.

RESEARCH ARTICLE
Ping Wang et al.

MSCs exhibited similar transcriptional signatures upon TNFα or IL1β stimulation
Inflammatory cytokines such as TNFα and IL1β are strong activators of NF-κB signaling pathway (Kida et al., 2005;Osborn et al., 1989). To explore how these cytokines modulate human vascular cells via NF-κB signaling pathway, we measured the mRNA levels of two NF-κB target genes, ICAM1 and VCAM1, upon TNFα or IL1β stimulation in a dose-and time-dependent manner. Upon TNFα induction, the mRNA levels of ICAM1 and VCAM1 were both increased with peak expression at 20 ng/mL and by 8 h of treatment ( Fig. 7A). By comparison, ICAM1 and VCAM1 were upregulated even at low IL1β concentrations with peak expression by 4 h of treatment (Fig. 7B). We next mapped the transcriptomic landscapes induced by TNFα or IL1β in WT and RelA-deficient MSCs by genome-wide RNA-seq (

RESEARCH ARTICLE
Ping Wang et al. stimulation in WT MSCs, and were abolished by RelA deficiency (Fig. 7D). A total of 618 upregulated genes induced by TNFα and 589 upregulated genes induced by IL1β were found in WT MSCs, while only 20 upregulated genes induced by TNFα and 55 upregulated genes induced by IL1β were found in RelA −/− MSCs (Fig. 7E). Most were shared in common (Fig. 7F) and contributed to cell proliferation, survival, defense response, immune response, and inflammatory response (Fig. 7G).
Due to the aforementioned similarity in gene expression patterns upon TNFα or IL1β induction, we used TNFα as the vascular inflammatory stimulus in the subsequent studies.

Transcriptomic analysis revealed a guardian role of RelA in maintaining blood vessel homeostasis
Genome-wide RNA-seq was performed in WT and RelA −/− VECs, VSMCs and MSCs under basal and TNFα-induced conditions ( Fig. S1A and 8A). Under basal condition, RelA deficiency largely retained the transcriptional profile in vascular cells (Fig. 8B and 8C). Venn diagram analysis revealed only two commonly upregulated genes and three downregulated by RelA deficiency among VECs, VSMCs and MSCs (Fig. 8D), despite that RelA expressed substantially in all these cell types (Fig. S1B). These results suggest that RelA modulated vascular cells in a cell type-specific manner. The two commonly upregulated genes were transcription elongation factor A like 1 (TCEAL1) and peroxiredoxin 4 (PRDX4), and three downregulated genes were C-type lectin domain containing 11A (CLEC11A), DnaJ heat shock protein family member C15 (DNAJC15), and synaptotagmin 11 (SYT11) (Fig. 8E). The expression of CLEC11A, one of the downregulated genes and a growth factor that modulates stem cell proliferation and osteogenic development (Hiraoka et al., 2001;Yue et al., 2016), was re-confirmed by qPCR (Fig. 8F). Similarly to our observation with RelA deficiency, CLEC11A knockdown also impaired MSC proliferation (Fig. 8G, 8H and S1C), suggesting that RelA deficiency blocked MSC self-renewal at least in part through transcriptional silence of CLEC11A.
Further analysis revealed the specific aspects of biological process affected by RelA deficiency. In VECs, RelA deficiency repressed inflammatory response, cytokine production, and NF-κB signaling pathway, supporting a role of RelA in mediating vascular inflammatory response (Fig. 8I). In both VSMCs and MSCs, RelA deficiency led to aberrant extracellular matrix organization and skeletal development that were important to the maintenance of vascular structure and function. In addition, RelA deficiency affected blood vessel formation in VSMCs and promoted the ossification and inhibited the migration of MSCs ( Fig. 8J and 8K).

TNFα-NF-κB axis modulated vascular gene expression in a cell type-specific manner
We next investigated how RelA regulated gene expression in TNFα-treated vascular cells. RelA deficiency dramatically abrogated TNFα-induced transcriptional changes in all types of vascular cells, especially in MSCs (Fig. 9A). Venn diagram analysis revealed that a total of 606 genes in WT VECs, 580 genes in WT VSMCs, and 604 genes in WT MSCs were upregulated by TNFα in a RelA-dependent manner (Fig. 9B). In VECs, TNFα induced a set of genes that modulate inflammatory response and cytokine signaling pathway (Fig. S1D). In VSMCs the upregulated genes contributed to defense response and cell proliferation (Fig. S1E). In MSCs the induced genes were associated with immune response and apoptotic process (Fig. S1F). All these effects were abolished in RelA −/− vascular cells.
Next, we compared the three sets of TNFα-induced RelAtarget genes in WT VECs, VSMCs and MSCs. A total of 134 genes were commonly upregulated among the three types of vascular cells (Fig. 9C). These genes contributed to biological processes including immune response, cytokine signaling, cellular adhesion and apoptotic process, indicating that they may be critical to protect vascular cells from inflammatory damage in a NF-κB-dependent manner ( Fig. 9D and S1G). Furthermore, PCA analysis revealed that RelA deficiency increased the transcriptomic differences in vascular cells upon TNFα stimulation (Fig. 9E).
Taken together, our data demonstrate that RelA deficiency interfered with inflammatory response and cytokine signaling in vascular cells, indicating a protective role of RelA in maintaining vascular homeostasis during vascular inflammation.

Effect of IκBα deficiency on NF-κB pathway
Previous studies have shown that IκBs are essential for the inhibition of RelA and that RelA can be released and activated under inflammatory stimulus. To investigate whether NF-κB is negatively modulated by IκBs in vascular cells, we generated IκBα knockout (IκBα −/− ) hESCs by targeting the first exon of IκBα using CRISPR/Cas9-mediated genome editing (Fig. 10A). Removal of the targeted exon was verified in IκBα −/− ESCs (Fig. 10B) and the resulting protein loss was confirmed by western blotting (Fig. 10C). To further investigate the role of IκBα in vascular cells, we differentiated WT and IκBα −/− ESCs to VECs, VSMCs and MSCs, respectively. Loss of IκBα expression was verified in diverse IκBα −/− cells (Fig. 10D).
To understand how IκBα regulates the homeostasis of different vascular cells, we performed genome-wide RNAseq in WT and IκBα −/− VECs, VSMCs, and MSCs under basal and TNFα-induced conditions (Fig. S2A). IκBα deficiency resulted in 79, 78 and 99 genes upregulated, and 104, 97, 90 genes downregulated in untreated VECs, VSMCs, and MSCs, respectively (Fig. S2B). Venn diagram and GO enrichment analysis revealed that IκBα regulated gene expression in a cell type-specific manner (Fig. S2B-E). To further determine whether RelA modulates gene expression dependently of IκBα, we compared the upregulated genes in IκBα −/− vascular cells with the downregulated genes in RelA −/− vascular cells. Only a small fraction of genes were overlapped in each of the three vascular cell types (Fig. 10E-H), regardless of the absence or presence of TNFα. Consistently, we did not observe marked differences in the subcellular localization of RelA in IκBα-depleted cells (Fig. 10I). These data suggest that IκBα acted largely independent of RelA in various human vascular cells.

DISCUSSION
In this study, we generated RelA-deficient human ESCs using CRISPR/Cas9-mediated genome editing and differentiated them into different types of vascular cells to investigate how NF-κB/RelA regulates human vascular cells under basal and inflammatory conditions. Via genome-wide RNA sequencing analysis, we mapped the NF-κB-regulated transcriptomic landscapes to systematically understand the physiological roles of RelA in various vascular cells. Our findings suggest that RelA modulated vascular cells in a cell type-specific manner by affecting multiple aspects of cellular events including extracellular matrix organization, ossification, vasculogenesis, inflammatory response, proliferation and survival. In addition, our data indicate that IκBα regulated gene expression in vascular cells primarily in a RelAindependent manner (Fig. 11).
The correlation between NF-κB activation and vascular dysfunction have attracted great attention during the last decades. For instance, the nuclear translocation of RelA has frequently been observed in human atherosclerotic lesion vascular cells (Brand et al., 1996;Mallavia et al., 2013). In ApoE −/− mice the activation of NF-κB signaling pathway in endothelial cells and smooth muscle cells is accompanied with enhanced secretion of inflammatory factors and increased atherosclerotic lesions (Gareus et al., 2008;Mallavia et al., 2013). In addition, TNFα-induced NF-κB activation impairs cell proliferation and causes premature senescence in HUVECs (Khan et al., 2017); knockdown of RelA promotes cell proliferation and reduces apoptosis in high glucose-treated HUVECs (Chen et al., 2011a). Yet, the role of NF-κB in different human vascular cells has not been well studied via a side-by-side comparison.
Here, we took advantage of the various types of differentiated RelA-deficient vascular cells to reveal the

RESEARCH ARTICLE
Ping Wang et al. physiological functions of NF-κB in these cells, providing direct, unbiased characterization of the cell type-specific transcriptional regulation by NF-κB in vascular cells. Our platform is of superior advantages over traditional tools such as gene-engineered mice (Ijaz et al., 2016), RNAi-mediated knockdown in human vascular cells (Chen et al., 2011a;Chen et al., 2011b), as well as small-molecule agonists and antagonists of NF-κB (Jakkampudi et al., 2016;Janssen-Heininger et al., 2000). Our data provided a comprehensive understanding on how NF-κB regulates the homeostasis of various human vascular cells under basal and inflammatory conditions. For instance, we observed enhanced adipogenesis and osteogenesis in RelA −/− MSCs, which suggests that RelA deficiency may cause atherosclerosis-associated abnormal lipid accumulation and vascular ossification. Our data also showed that the self-renewal and differentiation abilities of MSCs were directly regulated by RelA, implying that RelA inactivation might cause vessel dysfunction via compromising the regenerative ability of the adventitia-localized stem cells. Furthermore, we showed that RelA deficiency impaired vasculogenesis in VECs and VSMCs. Thus, our study offers detailed illustrations how RelA regulates gene expression in a vascular cell type-specific manner and supports the notion that RelA is implicated in the multi-perspective maintenance of human vascular homeostasis.
IκBs sequester NF-κB in the cytoplasm, thus inhibiting NF-κB activity. Degradation of IκBs releases NF-κB, leading to its translocation to the nucleus and the subsequent activation of NF-κB target genes (Simeonidis et al., 1999). There are at least eight dedicated IκB proteins: IκBα, IκBβ, IκBγ, IκBε, IκBz, IκBNS, Bcl-3 and Drosophila Cactus (Morris et al., 2016), with IκBα generally regarded as the predominant isoform (Fagerlund et al., 2015). In patients, heterozygous nonsense mutation in IκBα leads to NF-κB haploinsufficiency, resulting in ectodermal dysplasia and immune deficiency (Courtois et al., 2003). In this study, we genetically ablated IκBα in hESCs and obtained various vascular cells via directed differentiation to study the impact of IκBα deficiency on the functions of NF-κB signaling. Our results showed that IκBα acted largely independent of RelA signaling under basal condition as well as upon TNFα stimulation, suggesting that NF-κB is not predominantly regulated by IκBα, but perhaps by other IκB isoforms in vascular cells.
To date, inhibition of NF-κB during vascular inflammation has provided a promising therapeutic target for the treatment of various cardiovascular disorders. Various inhibitors that target NF-κB pathway have been developed to extend the choice for clinical applications (Tas et al., 2009). In our study, we have provided a platform that facilitates the in-depth mechanistic interpretation of how RelA regulates human vascular diseases and aging. The differential responses in various types of vascular cells to RelA deficiency sound a cautionary note for diseases being considered for anti-NF-κB treatment. Moderate suppression is likely to be beneficial to the maintenance of vascular homeostasis by alleviating the injury from vascular inflammation, whereas excessive suppression may lead to the abnormal behaviors for different layers of vascular cells. Lastly, given that the regulation of NF-κB is of cell type specificity, our study has demonstrated the importance of highly targeted therapy in the treatment of vascular inflammatory diseases via the inhibition of NF-κB activity.

Flow cytometry analysis
Cell apoptosis analysis was performed with Annexin V-EGFP Apoptosis Detection Kit (Vigrous Biotechnology). Cells were pretreated with 10 ng/mL TNFα (Peprotech) for 24 h before performing apoptosis assay. For ICAM1 expression analysis, cells were suspended in 10% FBS/PBS with ICAM1-PE (BD Biosciences, 560971, 1:100) for 30 min, and IgG-PE was used as an isotype control. The results were determined by FlowJo7.6.1 software. For Dil-Ac-LDL uptake analysis, VECs were cultured with Dil-Ac-LDL (Molecular probes, 1:400) for 6-8 h at 37°C. VECs were harvested and washed twice by PBS and analyzed by FACS detected by FITC channel.

RNA extraction and RT-qPCR
Cells were harvested in TRIzol reagent (Invitrogen) and RNA was extracted according to the manufacturers' instruction. 2 μg RNA was used for cDNA synthesis using reverse transcription master Mix (Promega). RT-qPCR was performed in SYBR Green supermix (Thunderbird) on a CFX-384 RT-qPCR system. The relative expression of genes was normalized by 18S rRNA transcript. All qPCR primers used were listed as follows: 18S forward primer, GTA ACC CGT TGA ACC CCA TT, reverse primer, CCA TCC AAT CGG TAG TAG CG. RelA forward primer, GTG GGG ACT ACG ACC TGA ATG, reverse primer, GGG GCA CGA TTG TCA AAG ATG.

RESEARCH ARTICLE
Ping Wang et al.
mL for 4 h before harvest. RNA-seq libraries were constructed following a previous protocol . RNA integrity was verified by the Bioanalyzer 2100 system (Agilent Technologies). Sequencing libraries were obtained using NEBNext ® Ultra TM RNA Library Prep Kit for Illumina ® (NEB) and index codes were tagged to each samples for sequencing. The libraries were sequenced on an Illumina Hiseq platform.

RNA-seq data processing
RNA-seq data processing was performed as previously reported . Briefly, low quality reads were trimmed and then paired clean reads were mapped to hg19 human genome using hisat2 (v2.0.4) (Kim et al., 2015). Reads were then counted by HTSeq (v0.6.1) (Anders et al., 2015). Differentially expressed genes (DEGs) were calculated using DESeq2 with the cutoff Benjamini-Hochberg adjust P value 0.05 and absolute fold change more than 1.5 (Love et al., 2014). To evaluate the correlation between replicates of each sample, the Pearson correlation coefficient (R) was computed based on log 2 (FPKM + 1). Gene Ontology (GO) and KEGG pathway enrichment analysis was carried out by ClusterProfiler R package (Yu et al., 2012). The RNA-seq data have been deposited to the NCBI Gene Expression Omnibus (GEO) database with accession number GSE115311.
Matrigel tube formation assay 2 × 10 4 VECs were seeded in each well of a 24-well plate coated with matrigel (BD Biosciences) in triplicate and incubated for 6-8 h at 37°C, and then stained by Calcein-AM (Invitrogen) . The tube formation was assessed by taking photomicrographs and measured by ImageJ2x 2.1.4.7 software.
Monocyte adhesion assay 2 × 10 5 VECs were seeded in each well of 12-well plate coated with collagen. The next day, VECs were treated with or without 10 ng/mL TNFα for 4 h. 2 × 10 6 monocytes were co-cultured with VECs for 1 h and then rinsed by PBS for 3 times carefully (Lee et al., 2015). The monocytes adhered on endothelium were analyzed by ImageJ2x 2.1.4.7 software.

Colony formation assay
VECs, VSMCs and MSCs were each seeded by 2,000 cells per well in 12-well plates (Pan et al., 2016), and cultured until clear cell colonies formed. Cells were fixed in 4% paraformaldehyde for 30 min and stained with crystal violet (Biohao Biotechnology). The images were analyzed by ImageJ software.

Teratoma experiment
3 × 10 6 human ESCs were injected into NOD-SCID (Non-obese diabetic severe combined immunodeficiency) mice under subcutaneous regions. After three months, mice were sacrificed and teratomas were excised . The teratomas were fixed in 4% paraformaldehyde, dehydrated in 30% sucrose solution, embedded in O.C.T compound and frozen in liquid nitrogen. Samples were sectioned and analyzed by immunofluorescent staining.
The animal experiments were approved by the Institute of Biophysics, Chinese Academy of Sciences.

Statistical analysis
All the results were presented as mean ± SEM. Two-tailed Student`s t-test was performed to compare the differences between two groups. All quantitative experiments were repeated at least 3 times independently.
Song and Jing Qu declare that they have no conflict of interest. All institutional and national guidelines for the care and use of laboratory animals were followed.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.