RNA-seq analysis reveals altered gene expression profiles in HMEC-1 cells overexpressing KRAS gene associated with brain arteriovenous malformation
Kexin Yuan, Yahui Zhao, Haibin Zhang, Ke Wang, Yunfan Zhou, Yu Chen, Xiaolin Chen, Yuanli Zhao, Qiang Hao
TL;DR
This study finds that overexpressing the KRAS gene in endothelial cells changes gene activity in ways that may contribute to brain arteriovenous malformation.
Contribution
The study identifies novel gene expression patterns linked to KRAS overexpression in a model of brain arteriovenous malformation.
Findings
KRAS overexpression in HMEC-1 cells alters pathways related to cell adhesion and signaling.
Upregulated genes are associated with cell-substrate junctions, while downregulated genes relate to ribosomes and mitochondria.
These findings suggest potential molecular mechanisms underlying brain arteriovenous malformation development.
Abstract
Brain arteriovenous malformation (bAVM) is a rare vascular disorder that can lead to severe neurological symptoms. The molecular mechanisms driving bAVM development and progression of bAVM remain poorly understood. This study aimed to investigate the molecular changes potentially associated with bAVM pathogenesis by performing RNA-seq on human microvascular endothelial cells (HMEC-1) overexpressing KRAS, a key driver of BAVM. HMEC-1 cells overexpressing KRAS were established as an in vitro model of bAVM. RNA-seq were conducted and transcriptomic analysis revealed that differentially expressed genes in HMEC-1 cells overexpressing KRAS were predominantly enriched in pathways related to cell adhesion, signaling, and transport, which may contribute to bAVM pathogenesis. Specifically, upregulated genes were mainly located in the cell–substrate junctions and focal adhesion, whereas…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5- —National Key Research and Development Program of China
- —Natural Science Foundation of China
- —Natural Science Foundation of China
- —Beijing Municipal Administration of Hospitals Incubating Program
- —Natural Science Foundation of Beijing
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsVascular Malformations Diagnosis and Treatment · Vascular Malformations and Hemangiomas · Vascular Anomalies and Treatments
Background
Brain arteriovenous malformation (bAVM) is a type of cerebrovascular disease characterized by the abnormal connection between intracranial arteries and veins through a malformation nidus. The tangled vasculature forms a high-flow, low-resistance shunt between the arterial and venous circulation due to the absence of a normal capillary bed [1, 2]. One of the most common manifestations of bAVM is intracranial hemorrhage, which can lead to high mortality rate and severe neurological complications, especially in children and young adults [3, 4]. Despite advancements in diagnosis and therapeutic strategies, the molecular mechanisms underlying bAVM development and progression remain poorly understood, highlighting the need for research into its pathogenesis.
Recent studies have implicated the proto-oncogene KRAS in bAVM pathogenesis [5–7]. KRAS encodes a GTPase involved in signal transduction pathways that regulate cell growth and differentiation. The pathogenic role of KRAS mutation is well-established in multiple human cancers, including lung, pancreatic, and colorectal carcinomas [8–13]. Moreover, overexpression of KRAS in endothelial cells has been shown to recapitulate the features of human bAVM in vitro [14], strongly suggesting that the dysregulation of KRAS signaling may contribute to the etiology of bAVM.
In this study, we sought to delineate the transcriptomic changes mediated by KRAS in an endothelial context. We established a cellular model by introducing overexpression of KRAS in the immortalized human microvascular endothelial cell line (HMEC-1), a standard model for vascular research [15, 16]. Subsequent RNA-seq analysis enabled identification of key differentially expressed genes and pathways that may underlie BAVM development. Our study provides a preliminary delineation of molecular alterations induced by KRAS in endothelial cells, which might shed light on the pathogenic mechanisms of bAVM.
Methods
Data availability statement
All data are available from the corresponding author upon reasonable request.
Cell culture
HMEC-1 cells were purchased from Wuhan prosai Life Technology Co. Ltd (Wuhan, China). After resuscitation, the cells were seeded into 25-cm^2^ culture flasks pre-coated with 1% gelatin. The cells were cultured in MCDB131 medium (Genom, Hangzhou, China) supplemented with 10% (v/v) fetal bovine serum (FBS; Ausbian, Cat. No. VS500T), 10 ng/mL EGF (Corning, Cat. No. 354052), 1 mM L-glutamine, and 1 µg/mL hydrocortisone. The cultures were maintained in a cell incubator (Thermo science, USA) at 37 °C with 5% CO_2_. Cells were passaged when they reached 80–90% confluence, as observed under an inverted microscope.
Construction of KRAS vector virus
The lentiviral construct encoding KRAS (NM_033360, G12V) was generated and packaged by GeneChem (Shanghai, China). The construction workflow included vector linearization, PCR amplification of the KRAS insert, homologous recombination into the GV492 backbone, bacterial transformation, colony screening via PCR, sequence verification, and large-scale lentiviral packaging. Viral preparations were quality-controlled and supplied at high titer for subsequent HMEC-1 transduction.
Lentiviral transduction and RNA preparation
Third-generation HMEC-1 cells were used for lentiviral transduction. When cell density reached approximately 20%, cells were inoculated into six-well plates containing 2 mL infection medium per well. Lentiviral particles (2 × 10^5^ TU/well) were added, corresponding to a multiplicity of infection (MOI) of 20 based on an estimated 1 × 10^4^ cells per well. After 8–16 h of incubation, the viral medium was replaced with fresh complete medium.
For RNA extraction, cells were harvested, and total RNA quality was assessed. RNA degradation and contamination were examined using 1% agarose gel electrophoresis. RNA purity (OD260/280 and OD260/230) was measured using a NanoPhotometer® spectrophotometer (IMPLEN, CA, USA). RNA integrity was evaluated using the RNA Nano 6000 Assay Kit on the Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA).
RNA-seq library preparation, sequencing, and data processing
For each sample, 1 µg of total RNA was used for library construction. mRNA was purified with poly-T oligo-attached magnetic beads, fragmented in First-Strand Synthesis Reaction Buffer (5X). mRNA was enriched using poly-T oligo-attached magnetic beads and fragmented using divalent cations in First-Strand Synthesis Reaction Buffer. First-strand cDNA synthesis was performed using random hexamer primers and M-MuLV Reverse Transcriptase (RNase H–), followed by second-strand cDNA synthesis with DNA Polymerase I and RNase H. Following end repair, 3′ adenylation, and adapter ligation, cDNA fragments of approximately 370–420 bp were selected using AMPure XP beads. Libraries were subsequently amplified with Phusion High-Fidelity DNA Polymerase, purified, and evaluated for quality on the Agilent 2100 Bioanalyzer. Cluster generation was performed on an Illumina cBot system using the TruSeq PE Cluster Kit v3-cBot-HS, and libraries were sequenced on an Illumina NovaSeq platform to generate 150-bp paired-end reads (Fig. 1A, B).Fig. 1. Clustering and sequencing of index-coded samples using Illumina technology. A Clustering of index-coded samples using TruSeq PE Cluster Kit v3-cBot-HS (Illumina). B Sequencing of library preparations on Illumina Novaseq platform. C Volcano plot showing all differentially expressed genes (DEGs) between KRAS overexpression HMEC-1 and control. D Hierarchical clustering heat map showing expression levels of all DEGs
Raw reads containing > 2 ambiguous nucleotides were removed, and adapter sequences and low-quality bases were trimmed using the FASTX-Toolkit. Reads shorter than 20 nt or containing > 50% low-quality positions were discarded. Clean reads were aligned to the reference genome using HISAT2 v2.0.5 with up to four mismatches allowed. Transcript assembly was performed using StringTie v1.3.3b, and gene-level read counts were obtained with featureCounts v1.5.0-p3. Gene expression levels were quantified as FPKM based on read counts and transcript length.
Quantitative real-time PCR (qPCR)
cDNA was synthesized from RNA using Superscript III first strand synthesis system (Invitrogen, Cat# 18080051). qPCR was performed using GoTaq qPCR Master Mix (Promega, Cat# A6102) on CFX96 thermal cycler (Bio-Rad), which was conducted with 3 replicates per group. The upstream primer of sequence reference gene is GCGTGACATTAAGGAGAAGC, downstream primer sequence is CCACGTCACACTTCATGATGG, and amplified fragment length is 236. The upstream primer of KRAS gene is AGTTGGAGCTGGTGGCGTAG, downstream primer sequence is CCTCATGTACTGGTCCCTCATT, and amplified fragment length is 196.
Differential gene expression and enrichment analysis
Differential expression analysis between control and KRAS-overexpressing groups (two biological replicates each) was performed using the DESeq2 R package (v1.20.0). DESeq2 applies a negative binomial model, and P values were adjusted using the Benjamini–Hochberg method. Genes with adjusted P < 0.05 were considered differentially expressed. For genes lacking biological replicates, edgeR (v3.22.5) was applied using a single scaling normalization factor. Corrected P value of 0.05 and absolute fold change of 2 were set as the threshold for significantly differential expression.
Gene Ontology (GO) enrichment analysis of DEGs was performed using the clusterProfiler R package, with corrections for gene length bias. GO terms with adjusted P < 0.05 were regarded as significantly enriched. KEGG pathway enrichment was also conducted using clusterProfiler based on KEGG database annotations (http://www.genome.jp/kegg/). Gene set enrichment analysis (GSEA) was performed using a local installation of the Broad Institute GSEA tool (http://www.broadinstitute.org/gsea/index.jsp), with GO and KEGG datasets analyzed independently.
Statistical analysis
The experimental data are expressed by mean ± SD, and statistical analysis is carried out by SPSS v22.0 (IBM SPSS statistics for Windows) and R (3.22.5) software. Group comparisons were performed using one-way ANOVA followed by Tukey’s post hoc test. Statistical significance was defined as P < 0.01.
Result
Morphological observation of HMEC-1 cells
HMEC-1 cells exhibited an epithelioid morphology with irregular sizes, abundant cytoplasm, and round to oval nuclei. The cell–cell boundaries appeared indistinct, consistent with previously reported characteristics of this line. At 72 h after KRAS overexpression with lentiviral transduction, no major morphological differences were observed compared with control cells, and the proliferation rate remained similar between the groups. Only a small number of floating cells and cell debris were noted in both cultures (Fig. 2).Fig. 2. Morphological characteristics of HMEC-1 cells before (A) and after (B) KRAS transfection
qPCR validation of KRAS overexpression
Quantitative PCR confirmed robust overexpression of KRAS mRNA in the overexpression (OE) group compared with the negative control (NC) group (Fig. 3). KRAS expression increased by approximately 107-fold relative to control, indicating highly efficient lentiviral transduction and stable expression of the KRAS construct.Fig. 3RT-qPCR confirmation of successful KRAS overexpression in HMEC-1 cells. KRAS mRNA levels were markedly elevated in the overexpression (OE) group relative to the negative control (NC) group. Expression values were normalized to ACTB. Data represent mean ± SD from three biological replicates. ***P < 0.001
Quality assessment of RNA sequencing
RNA-seq produced high-quality reads across all samples, with Q20 and Q30 values exceeding 96% and 91%, respectively, and GC content ranging from 48.64 to 50.95% (Table 1). Clean reads were effectively mapped to the reference genome, and the distribution of reads with a unique location on the genome in each region was calculated (Table 2), confirming that the sequencing depth and data quality were appropriate for downstream analysis. Table 1. Valid sequence tableSampleLibraryraw_readsraw_basesclean_readsclean_baseserror_rateQ20Q30GC_pctMeanSDGV341_120h_1FRAS220254031-1r44,442,7286.67G43,700,4226.56G0.0396.3290.9350.45267.37359.3929GV341_120h_2FRAS220254032-1r41,142,4086.17G40,491,2606.07G0.0396.4991.4950.95268.74360.6332GV341_120h_3FRAS220254033-1r46,211,1366.93G45,182,9246.78G0.0396.5891.4650.55265.43362.6148KRAS_120h_1FRAS220254034-1r45,618,7866.84G44,675,0706.7G0.0396.5991.4749.89281.89063.4755KRAS_120h_2FRAS220254035-1r41,695,8106.25G41,175,4466.18G0.0396.5491.2148.64285.99156.8738KRAS_120h_3FRAS220254036-1r46,981,5407.05G46,289,3066.94G0.0396.9692.149.65282.44759.8469Table 2Mapping of clean reads on the reference genomeSampleGV341_120h_1GV341_120h_2GV341_120h_3KRAS_120h_1KRAS_120h_2KRAS_120h_3total_reads43,700,42240,491,26045,182,92444,675,07041,175,44646,289,306total_map40,472,167 (92.61%)37,593,657 (92.84%)42,062,920 (93.09%)41,529,433 (92.96%)38,309,605 (93.04%)43,267,123 (93.47%)unique_map39,576,743 (90.56%)36,744,540 (90.75%)41,112,470 (90.99%)40,643,077 (90.97%)37,528,628 (91.14%)42,339,400 (91.47%)multi_map895,424 (2.05%)849,117 (2.1%)950,450 (2.1%)886,356 (1.98%)780,977 (1.9%)927,723 (2.0%)read1_map20,080,223 (45.95%)18,657,707 (46.08%)20,745,339 (45.91%)20,520,938 (45.93%)18,967,352 (46.06%)21,271,606 (45.95%)read2_map19,496,520 (44.61%)18,086,833 (44.67%)20,367,131 (45.08%)20,122,139 (45.04%)18,561,276 (45.08%)21,067,794 (45.51%)positive_map19,740,468 (45.17%)18,334,021 (45.28%)20,507,433 (45.39%)20,275,586 (45.38%)18,727,965 (45.48%)21,133,225 (45.65%)negative_map19,836,275 (45.39%)18,410,519 (45.47%)20,605,037 (45.6%)20,367,491 (45.59%)18,800,663 (45.66%)21,206,175 (45.81%)splice_map16,747,775 (38.32%)15,745,028 (38.89%)17,428,081 (38.57%)16,978,801 (38.01%)15,144,553 (36.78%)17,563,085 (37.94%)unsplice_map22,828,968 (52.24%)20,999,512 (51.86%)23,684,389 (52.42%)23,664,276 (52.97%)22,384,075 (54.36%)24,776,315 (53.52%)proper_map37,512,368 (85.84%)35,017,752 (86.48%)39,180,238 (86.71%)38,766,518 (86.77%)35,831,848 (87.02%)40,580,668 (87.67%)
Transcriptomic alterations induced by KRAS overexpression
RNA-seq results showed that 24,809 genes were co-expressed across the OE and NC groups. Among them, 4737 genes were differentially expressed, including 2619 upregulated and 2118 downregulated genes (fold change ≥ 2 or ≤ 0.5, FDR < 0.05) (Fig. 1C). The heatmap revealed clear separation between the two groups and high correlation within biological replicates (Fig. 1D).
GO analysis of upregulated genes demonstrated enrichment in cellular components associated with adhesion, including the cell–substrate junction, focal adhesion, and adherens junction. Enriched molecular functions included cadherin binding and cell adhesion molecule binding, while enriched biological processes involved Golgi vesicle transport, endomembrane system organization, and nuclear transport (Fig. 4A). Downregulated genes were enriched in ribosomal subunits, ribosomes, mitochondrial protein complexes, and mitochondrial inner membranes. Moreover, the molecular function of the downregulated genes included structural constituent of ribosome. Generally, the downregulated genes were mainly involved in protein targeting to the endoplasmic reticulum (Fig. 4B).Fig. 4. Gene ontology annotations of overexpression of KRAS gene on the transcriptome of HMEC-1 cells. A All upregulated genes DEGs between OE and NC group. B All downregulated genes DEGs between OE and NC group
KEGG analysis showed that the upregulated pathways in the KRAS-overexpressed HMEC-1 cells included the protein processing in endoplasmic reticulum, proteoglycans in cancer and adherens junction, and so on (Fig. 5A). The downregulated pathways included the ribosome, oxidative phosphorylation, and Parkinson disease pathways (Fig. 5B).Fig. 5KEGG pathway enrichment analysis of overexpression of KRAS gene on the transcriptome of HMEC-1 cells. A All upregulated genes DEGs between OE and NC group. B All downregulated genes DEGs between OE and NC group
Discussion
KRAS encodes a small GTPase that functions as a central node in the RAS–MAPK/ERK signaling cascade and regulates cell proliferation, differentiation, and survival. Recent genomic studies have identified somatic activating KRAS mutations in the endothelial compartment of most sporadic brain arteriovenous malformations (bAVM), implicating its critical role in bAVM pathogenesis [17, 18]. In this study, we investigated transcriptomic alterations induced by KRAS overexpression in HMEC-1 endothelial cells to explore molecular mechanisms potentially contributing to bAVM, hoping to provide insights into the molecular mechanisms underlying the pathogenesis of bAVM. Although KRAS overexpression did not lead to obvious morphological changes within 72 h, substantial transcriptional shifts were observed, indicating early molecular remodeling preceding overt phenotypic alteration.
KEGG analysis revealed the upregulated pathways following KRAS overexpression in HMEC-1 were predominantly associated with protein processing in the endoplasmic reticulum, proteoglycans in cancer, and adherens junction signaling—all processes implicated in vascular remodeling and pathological angiogenesis. Consistently, GO analysis also showed downregulated genes were mainly involved in protein targeting to the endoplasmic reticulum, suggesting that the ER stress response may be activated upon KRAS overexpression. ER stress has been implicated in various diseases and is increasingly recognized as an important contributor to vascular disorders. In endothelial cells, chronic activation of the unfolded protein response promotes oxidative stress, apoptosis, senescence-associated inflammation, and impaired nitric oxide bioavailability, thereby driving endothelial dysfunction and atherosclerotic lesion development [19, 20]. In the setting of ischemic stroke, cerebral ischemia/reperfusion triggers ER stress in brain microvascular endothelial cells, which exacerbates blood–brain barrier disruption and secondary injury, whereas pharmacological attenuation of ER stress has been shown to preserve barrier integrity and reduce infarct size in experimental stroke models [21, 22]. Furthermore, ER stress has been linked to endothelial senescence and pro-inflammatory phenotypes in response to vasoactive stimuli such as angiotensin II, highlighting its role in maladaptive vascular remodeling [23, 24]. Together with our transcriptomic data, these observations implicate that KRAS-induced ER stress may contribute to endothelial dysfunction and microvascular instability in bAVM, although this hypothesis requires direct functional validation in future studies.
Similarly, alterations in proteoglycan expression and adherens junction integrity may influence endothelial barrier stability and cell–cell interactions, both critical factors in abnormal vessel formation and might be related with pathological vascular remodeling in bAVMs. Proteoglycans such as syndecan-4 have been shown to facilitate VEGFA-induced VE-cadherin internalization, thereby promoting pathological angiogenesis [25]. Karthika et al. reported elevated expression of adhesion molecules, including VE-cadherin and N-cadherin, in bAVM nidus compared with normal brain tissue, suggesting altered junctional dynamics [26]. In a related vascular model, Tang et al. showed that inhibiting TLF4 signaling reduced phosphorylation of key adherens junction proteins—VE-cadherin, γ-catenin, and p120-catenin—thereby stabilizing junctional structure and mitigating cavernous malformation formation [27, 28]. Our results indicated KRAS overexpression may contribute to dysregulated proteoglycan signaling and adherens junction remodeling, highlighting the potential involvement of these processes in bAVM pathogenesis.
Downregulated pathways in KRAS-overexpressed HMEC-1 cells included ribosomal function and oxidative phosphorylation, suggesting altered mitochondrial activity and reduced biosynthetic capacity [29]. Mitochondrial metabolism is increasingly recognized as a critical regulator of endothelial cell homeostasis, vascular stability, and angiogenic potential [30]. Dysfunctional mitochondria in endothelial cells lead to elevated mitochondrial ROS, diminished nitric oxide bioavailability, altered calcium handling, and increased endothelial permeability—hallmarks of vascular destabilization [31, 32]. A study by Zhang et al. demonstrated that mitochondrial dysfunction was present in mouse bAVM tissues, while inhibition of mitochondrial oxidative phosphorylation could suppress bAVM angiogenesis and improve neurological deficits in a mouse model [33]. Although the Parkinson’s disease pathway emerged among downregulated terms, the relationship is likely indirect and reflects shared mitochondrial or proteostatic stress mechanisms rather than disease-specific processes [34–36]. Taken with our transcriptomic results, KRAS overexpression might contribute to the development of bAVM via regulating mitochondrial activity and oxidative phosphorylation, representing an important direction for future mechanistic investigation.
The upregulated genes in GO analysis were primarily located in the cell–substrate junction, focal adhesion, and cell–substrate adherens junction, indicating that alterations in cell–substrate adhesion and signaling may play a crucial role in bAVM pathogenesis. Previous studies have reported that changes in the adhesiveness and the deformability may facilitate migration of cells [37]. Cell–substrate interactions are implicated multiple pathways relevant to vascular remodeling and angiogenesis [38]. Given the association between adhesion remodeling and endothelial plasticity, these transcriptional changes may reflect early features of endothelial-to-mesenchymal transition following KRAS overexpression, which is a process increasingly implicated in vascular malformations. EndoMT has been reported in cerebral cavernous malformations (CCM) and other vascular disorders, where it promotes endothelial dysfunction and abnormal vessel formation [39, 40]. However, whether KRAS overexpression drives EndoMT in bAVMs remains speculative and requires dedicated experimental validation.
In general, our findings provide valuable insights into the molecular mechanisms following KRAS overexpression and suggest potential targets for therapeutic intervention. KRAS activation triggers extensive transcriptional remodeling in endothelial cells affecting adhesion, ER stress, and mitochondrial function—processes that may contribute to bAVM development. These results provide a foundation for future mechanistic and functional studies.
This study has several limitations that should be acknowledged. First, although GO and KEGG analyses identified multiple dysregulated pathways, functional analyses to determine whether these transcriptional changes result in measurable endothelial phenotypes were lacking. Second, KRAS overexpression was confirmed only at the mRNA level, and protein-level validation (e.g., western blot or immunostaining) was not conducted. Third, all experiments were performed in a single immortalized endothelial cell line cultured in a two-dimensional system, which does not fully recapitulate the cellular heterogeneity, extracellular matrix complexity, or hemodynamic forces present in bAVMs. Finally, as a preliminary transcriptomic study, the findings are descriptive and hypothesis-generating, and further mechanistic and in vivo validation will be essential to clarify the biological relevance of the pathways identified.
Conclusion
In conclusion, this RNA-seq study provides a preliminary delineation of transcriptional changes induced by KRAS overexpression in HMEC-1 cells, highlighting dysregulated ER stress, proteoglycan signaling, cell–substrate adhesion, and mitochondrial oxidative phosphorylation. Overall, our study provides a foundation for further investigations into the molecular mechanisms of bAVM pathogenesis, as well as potential targets for therapeutic intervention.