Human tissue studies were conducted on deidentified specimens obtained from the Southern California Islet Cell Resource Center at City of Hope. The research consents for the use of postmortem human tissues were obtained from the donors' next of kin and ethical approval for this study was granted by the Institutional Review Board of City of Hope (IRB #01046). T2DM was identified based on diagnosis in the donors' medical records as well as the percentage of glycated hemoglobin A1c (HbA1c) of 6.5% or higher. The intimal RNA was isolated from human mesenteric artery by flushing the inner lumen with TRIzol once as previously described (30). Donor information is summarized in Supplementary Table 1.

For most in vitro experiments involving ECs, human umbilical vein ECs, i.e., HUVECs (Cell Applications Inc., San Diego, CA, Catalog # 200p-05n and lot # 3,363 and 2,914) between passages 5–8 pooled from multiple donors were used after testing negative for mycoplasma contamination. The cells were cultured in M199 (Sigma M2520) supplemented with 15% FBS (Hyclone, SH30910.02), β-endothelial cell growth factor (Sigma, E1388), and 100 units/ml penicillin and 100 mg/ml streptomycin (Thermo Fisher Scientific). Human VSMC (HVSMC) were purchased from ATCC (Manassas, VA, USA) (PCS-100-012TM) and cultured in M231 medium with smooth muscle growth supplement (Gibco, Waltham, MA, USA) as previously described (31). All the cells were maintained at 37°C with 5% CO₂ in a hydrated incubator.

For LINC00607-knockdown experiments, an antisense locked nucleic acid (LNA) modified oligonucleotide (GapmeR) specifically targeting LINC00607 (NR_037195.1) was designed by QIAGEN (28). For LINC00607-KD under basal condition, the final concentration of LINC00607-LNA is 50 nM, treated for 48 h. For LINC00607-KD in HUVEC under HT, the final concentration of LINC00607-LNA is 25 nM, treated for 72 h. For Myc-knockdown experiments, a pool of 3 different siRNAs were purchased from Thermo Fisher Scientific. c-Myc-siRNA was used at 20 nM treated for 72 h. LNA GapmeR or siRNAs were transfected into HUVECs with Lipofectamine RNAiMAX following the protocol provided by the manufacturer (Thermo Fisher Scientific). The cells were returned to M199 complete medium 4–6 h after transfection and incubated for 3 days or treated with high glucose (25 mM final concentration) + TNFα treatment (5 ng/ml; Thermo Fisher Scientific Cat # PHC3011) (abbreviated as HT) for 3 days as in our previous study (28). For the reversal experiment, ECs were treated with HT first before transfection. D-mannitol at 25 mM was used as an osmolarity control. HVSMC transfection was done as previously described (31).

Cell lysates were resolved on an SDS-PAGE gel followed by immunoblotting with two different antibodies against c-Myc. Recombinant Anti-c-Myc antibody (ab32072, Abcam, 1:1000 dilution) or c-Myc Rabbit mAb (18,583, Cell Signaling Technology, 1:1000 dilution) were used as primary antibodies, and anti-rabbit (7,074 s, Cell Signaling Technology, 1:5000 dilution) was used as the secondary antibody. β-actin Rabbit mAb (8,457 s, Cell Signaling Technology, 1:1000 dilution) was used for loading control.

Tube formation assay was performed as previously described (32) with slight modifications. Briefly HUVECs were plated on a Matrigel (Corning, Catalog # 356234) coated 24-well plate in M199 complete media, incubated for up to 12 h in 5% CO₂ at 37°C, and examined for capillary tube formation under an inverted microscope and imaged. Three randomly selected views were captured, and the formed tubes were counted.

Total RNA was extracted using TRIzol reagent and cDNAs were synthesized using the PrimeScript™ RT Master Mix containing both Oligo-dT and random hexamer primers. qPCR was performed with Bio-Rad SYBR Green Supermix following the manufacturer's protocol on the Bio-Rad CFX Connect Real Time system with β-actin as internal control. All primer sequences used for qPCR amplification are listed in Supplementary Table 2.

For HUVEC bulk RNA-seq, 500 ng total RNA per sample was subjected to library construction using KAPA mRNA HyperPrep Kit (Roche Diagnostics) following the manufacturer's manual. The libraries were sequenced in HiSeq2500 using the SR50 mode. For HVSMC bulk RNA-seq, 200 ng total RNA per sample was subjected to library construction using KAPA mRNA HyperPrep Kit (Roche Diagnostics) following the manufacturer's manual. The libraries were sequenced in HiSeqX using the PE150 mode.

To analyze abundance of LINC00607, ENCODE RNA-seq data (corresponding to Supplementary Figure 2) were downloaded from GSE78534, GSE78528, GSE78548, GSE78541, GSE88601, GSE78645, GSE78660, GSE88615, GSE78671, GSE78609, GSE78613, GSE78588, GSE78540, GSE78533, GSE78536, GSE88167, GSE90268, GSE90272, GSE78619, GSE78618, GSE78649, and GSE78587.

RNA-seq data were analyzed as previously described (28, 33). The FATSQ files were aligned to human reference genome (Ensembl GRCh38.105) using STAR (v2.4.7a) (34). FeatureCounts from Subread package (v2.3.0) (35) were used to generate counts of the uniquely mapped reads. The raw count matrices were prefiltered to remove genes with very low read counts and used as input for differential gene expression analysis using the R package DESeq2 (36). Differentially expressed genes (DEGs) with P-values < 0.01 were considered significant and genes with Log₂ fold change (FC) > |1| were used for subsequent analysis. Gene ontology (GO) enrichment analysis was performed using the Gene Ontology Consortium platform (37) and Benjamini-Hochberg corrected P-values < 0.05 were considered significantly enriched pathways.

Our published datasets from EC-enriched human mesenteric arterial cells (28) was first analyzed using Cellranger (v6.1.1), the standard pipeline provided by 10X Genomics, which aligned the raw reads to human hg38 reference transcriptome. The aligned data was then processed using the R package Seurat (v3.2.3) following published guidelines (38). Initially, well-established filtering steps were performed to remove genes expressed in <3 cells and cells expressing <200 genes. A threshold was set at the 98th percentile for gene distribution in each sample, and cells above that threshold were removed to account for possible multiplets. Furthermore, cells containing >20% mitochondrial genes were filtered out. Next, the “sctransform” normalization strategy was used where the residuals of negative binomial regression were used to model each gene. Any gene expressing a positive residual indicated that more unique molecular identifiers (UMIs) were observed than predicted considering the sequencing depth and gene average expression. The converse is represented by a negative residual (39). Datasets from the four samples then underwent integration to enable joint dimensionality reduction and clustering without being influenced by donor condition. Cells-pairs in matched biological states between samples were identified as “anchors” and used to integrate the data. The standard pre-processing step of scaling was performed prior to principal component analysis (PCA) on all cells and top 1,000 highly variable genes using the scaled z-score expression values were identified. The top 15 significant principal components (PCs) were used as an input to the uniform manifold approximation and projection (UMAP) algorithm with 0.5 resolution (Seurat default). Cell clusters were generated and annotated using the following markers based on existing literature and PanglaoDB (40–42): for VSMCs, ACTA2, MYH11, WT1, and TAGLN; for ECs, VWF, PECAM1, EGFL7, ID3, GNG11, MCAM, CLDN5, and CDH5; for macrophages, CD68, MARCH1, and CD52; for fibroblasts, LUM and DCN; for T cells, CD3D and CD3E; and for NK cells, KLRD1 and NKG7.

Differential expression analysis was performed on UMI counts that had been normalized using “LogNormalize” and scaled. The Seurat default, non-parametric Wilcoxon test was used to analyze genes expressed in at least 10% of cells. The log fold-change of the average expression was thresholded at 0.25 and a pseudocount of 1 was added to the averaged expression values.

For scRNA-seq of HUVECs, ECs were trypsinized and suspended into a single cell suspension for scRNA-seq library preparation following the 10x Genomics Chromium 3' expression protocol and as described (28). The same steps were utilized to perform scRNA-seq analysis on HUVECs with the following exceptions: (1) Seurat (v.4.0) R package was utilized, (2) the top 30 significant PCAs were used as input for UMAP, (3) high mitochondrial genes were defined as >25% per cell and filtered out, (4) clusters were annotated as either LINC00607 high (>1), mid (0.5–1) or low (<0.5).

Fresh frozen human mesenteric artery cryosections of 10 μm thickness were subjected to smFISH using RNAscope™ Multiplex Fluorescent V2 Assay (ACDBio, 323100) with probes specifically designed to target the LINC00607 transcript NR_037195.1 (RNAscope® Probe—Hs-LINC00607-O1; Catalog # 894351). Briefly, the cryosections were ethanol dehydrated and permeabilized with Protease Plus reagent for 30 min before probe hybridization, signal amplification, and development following the manufacturer's protocol. smFISH in HUVECs was performed similarly with a slight modification. Briefly, cells were grown and treated on glass coverslips coated with Poly-lysine and fibronectin and then fixed with 4% paraformaldehyde. After fixing, the cells were subjected to ethanol dehydration, hydrogen peroxide pre-treatment, permeabilization with Protease III reagent (1:10 dilution) and probe hybridization before developing the signal. Image analysis was performed using FIJI (ImageJ software). For smFISH in HUVECs, LINC00607 signal per nucleus was computed by drawing a mask around the nucleus (with DAPI channel as reference) and determining the average gray value of LINC00607 signal within this mask per nucleus.

For all the high-throughput sequencing experiments, 2–3 biological replicates were used. For other experiments at least three independent experiments were performed. For these experiments, statistical analysis was performed using t-test (unpaired, two-sided) between two groups or ANOVA followed by Bonferroni post-test for multiple-group comparisons. P < 0.05 was considered as statistically significant unless otherwise specified. The quantification and statistical analysis have also been specified and detailed in the results, figure legends, and methods.

All high-throughput sequencing data supporting the current study have been deposited on GEO (Accession# GSE197956). The sample information is provided in Supplementary Table 4. The bulk RNA-seq data for LINC00607-KD in HUVEC under HT is included in Accession# GSE135357.

To map the expression of LINC00607 in human arteries, we leveraged the scRNA-seq data previously obtained from mesenteric arteries of 4 human donors (2 healthy control vs. 2 T2D) (28). Specifically, the arterial tissues were collected using a published method in which the intimal layer was dissociated, followed by enzymatic digestion into single-cell suspension for 10x Genomics scRNA-seq workflow (43). Analysis of >12,000 single cells identified 3,366 ECs and 1,120 VSMC among other cells present in the mesenteric artery (Figure 1A). LINC00607 was detected in ECs and VSMCs by scRNA-seq, which captures transcripts with poly A tails (Figures 1B,C). scRNA-seq analysis also revealed higher levels of LINC00607 in T2D ECs and VSMCs as compared to healthy donors, evident by both average expression levels in single cells and percentage of cells expressing LINC00607 (Figures 1B,C). As cell type controls, CDH5 (encoding vascular endothelial cadherin) and ACTA2 (encoding alpha smooth muscle cell actin) were abundantly expressed in ECs and VSMCs, respectively.

Consistent with scRNA-seq, smFISH revealed the expression of LINC00607 in intima and media of the mesenteric arteries, with remarkably higher signals in the T2D donor as compared to the healthy control (Figure 1D), whereas negative controls yielded no significant signals in either tissue (Supplementary Figure 1). Of note, as a validation of the diabetic state of our select donor-derived tissues, expression levels of many well-known EC dysfunction markers (e.g., CCL2, FN1, SERPINE1, EDN1, etc.) were increased in ECs of T2D vs. control donors (Figure 1E). We also verified the expression pattern of LINC00607 by qPCR using tissues collected from multiple donors. Indeed, LINC00607 levels in the intima-enriched arterial tissues were significantly higher in age-matched T2D donors (n = 8) relative to those in the healthy individuals (n = 6) (Figures 1F–H).

We exploited multiple transcriptomic and epigenomic datasets available at ENCODE and genotype–tissue expression (GTEx) (44, 45) to determine the expression pattern of LINC00607. Based on ChIP-seq data, the genomic region of LINC00607 is indeed highly enriched in H3K27ac and H3K4me1 in HUVECs and VSMCs, indicating an enhancer state in these two cell types (Figure 1I). Furthermore, LINC00607 RNA is transcribed in arterial ECs and VSMCs evident by RNA-seq data from multiple ENCODE datasets (Supplementary Figure 2). Interestingly, FN1, one of the best characterized genes associated with CVD, is located ~175Kb downstream of LINC00607 (Figure 1I). According to the tissue distribution revealed in GTEx, the highest abundance of LINC00607, like that of FN1, was detected in arteries, as compared to other major organs and tissues (Figure 1J). Taken together, these data suggest that LINC00607 is a vascular-enriched SE-derived lncRNA expressed in ECs and VSMCs and is upregulated in diabetes.

Given the active transcription and SE state of LINC00607 in both ECs and VSMC, and that SEs are important to maintain cell lineage and identity, we queried the role of LINC00607 in the regulation of basal function of ECs and VSMCs. To this end, we used a locked nucleic acid (LNA)-GapmeR to knockdown LINC00607 (607-KD) in HUVECs and HVSMC and performed RNA-seq simultaneously in the two cell types. Compared to the scramble control, LINC00607 inhibition led to 1,785 differentially expressed genes (DEGs) with 697 up- and 1,088 down-regulated genes in ECs (Figure 2A). Pathway enrichment analysis of these DEGs revealed that collagen catabolic process, extracellular matrix (ECM) organization, cell adhesion, inflammatory response, and angiogenesis are regulated by LINC00607 in ECs (Figure 2B). Specifically, DEGs involved in ECM organization, cell adhesion, and angiogenesis such as COL4A1, CTGF, FN1, SERPINE1, and TGFB2 were significantly downregulated upon 607-KD in ECs (Figure 2C; Supplementary Figures 3B,E). Consistently, 607-KD decreased the tube forming activity of ECs in a dose-dependent manner (Figures 2D,E), which was particularly significant in the early time points (1–4 h) (Supplementary Figure 4).

With the same cutoff, RNA-seq revealed a total of 1,270 DEGs in VSMCs due to 607-KD, with 427 up- and 843 down-regulated (Figure 2F). Interestingly, while 607-KD affected similar pathways in VSMCs as in ECs, such as ECM organization, cell adhesion, and collagen catabolic process, it also affected different pathways in VSMCs, e.g., cell proliferation and c-AMP signaling (Figure 2G). ECs and VSMCs shared 479 common DEGs, comprising 26.8% in ECs and 37.7% of DEGs in VSMCs (Figure 2H). The top enriched GO terms of these DEGs included cell proliferation, protein transport, angiogenesis, ECM organization, and blood coagulation (Figures 2I,J). Of note, FN1, the key ECM component is downregulated in both ECs and VSMCs (Figure 2J), suggesting LINC00607 RNA participates in the cis regulation of LINC00607 SE. On the other hand, the broad effect in other genes due to 607-KD suggests that LINC00607 RNA may regulate other genes through trans regulation, which may be independent from the SE function. Together, these data support an essential role for LINC00607 in the basal functions of these vascular cells, including angiogenesis and ECM organization.

We previously demonstrated a pro-inflammatory and pro-fibrotic role of LINC00607 in ECs under HT (28). Given the profound regulation of basal EC function by LINC00607 shown in the present study (Figure 2), we hypothesized that LINC00607 is required for EC homeostasis, but that its induction by pathophysiological conditions, through aberrant regulation of the downstream genes, promotes EC dysfunction. Consistent with our previous data from RNA-seq and qPCR (28), we observed induction of LINC00607 by HT in ECs using smRNA FISH, which showed the primary localization of LINC00607 in the nucleus (Figures 3A,B). Moreover, there was a higher enrichment of H3K27ac signal on LINC00607 locus in HT-treated ECs compared to NM control (Supplementary Figure 5), suggesting a more active SE state under HT. Comparison of DEGs identified from basal condition vs. HT-treated ECs revealed a total of 237 common DEGs (Figure 3C), which are involved in a series of intrinsic EC functions, including collagen catabolism and ECM organization, angiogenesis, cell-cell signaling, inflammatory response and type I interferon response (Figure 3D).

TFs play a central role in regulating the transcriptome by binding to promoters and enhancers, with the latter coordinating with the TFs to regulate its transcriptional targets (16, 46, 47). To gain insights into the upstream regulator of LINC00607 and the downstream transcriptional network regulated by LINC00607 RNA, we performed two analyses: (1) TRANSFAC analysis to predict potential TFs that can regulate LINC00607 transcription through binding its promoter; and (2) Ingenuity Pathway Analysis (IPA) to discover TFs that may coordinate with LINC00607 to regulate the DEGs affected by 607-KD (i.e., putative LINC00607 targets) in ECs. Integrating the two analyses revealed c-Myc to be a top candidate TF. Examining the c-Myc ChIP-seq data from HUVEC available in ENCODE and Epigenome Browser, we found specific Myc-binding signals at the promoter regions of LINC00607, which partially match the predicted c-Myc TF binding sites (Figure 3E). In the same dataset, c-Myc also shows clear binding to SERPINE1 and APEX1, two known targets of c-Myc (48, 49) (Figure 3E). IPA analysis independently revealed c-Myc as a top candidate upstream TF of LINC00607-regulated DEGs ranked by the number of target genes and activation scores (Figure 3F; Supplementary Table 3).

To further test whether c-Myc regulates LINC00607, we knocked down c-Myc in HUVECs by small interference RNAs (siRNAs) as confirmed by immunoblotting (Supplementary Figure 6). While c-Myc inhibition did not significantly affect the expression of LINC00607 under basal conditions (Figure 3G), there was a strong reduction in LINC00607 expression upon c-Myc inhibition under HT (Figure 3H). Consistently, the expression of SERPINE1, a proven regulatory target of LINC00607 and c-Myc, was also downregulated upon c-Myc inhibition (Figure 3H). Taken together, c-Myc appears to be a TF controlling the expression of LINC00607, which may act as an integral regulator in the c-Myc transcriptional network especially in the dysfunctional ECs.

It has been well-recognized that metabolic dysfunction associated with diabetes can provoke sustained alterations that contribute to persistent organ damage. In ECs, DNA methylation and histone modifications have been implicated in the sustained effect of hyperglycemia (5, 50, 51). We previously observed that whereas the HT-induced changes of some gene expression, such as eNOS and ICAM1 can be reversed by switching to normoglycemia, LINC00607 and its regulated SERPINE1 cannot be reversed (28). In line with this finding, HT-induced FN1 expression also continued to increase even after the replacement of high glucose with normal glucose (Figures 4A,B). Thus, we asked whether LINC00607 mediates the prolonged effect of high glucose and if so, whether its inhibition can reverse the effect of HT in ECs. To this end, we altered our approach of 607-KD prior to HT treatment (as we did for bulk RNA-seq shown in Figure 3), to instead exposing HUVECs to HT treatment for 4 days and then inhibiting LINC00607 for the next 3 days (Figure 4C). Because the transfection efficiency is unlikely to be 100%, we performed scRNA-seq analysis to resolve the cell heterogeneity of responses. We sequenced 2,806, 1,846 and 3,407 cells in three treatment groups, i.e., ECs kept under NM and transfected with scramble LNA (NM-scr), EC kept under HT and transfected with scramble LNA (HT-scr), and ECs kept under HT first and then transfected with LINC00607 LNA (HT-KD), respectively. The HT induction of LINC00607 and inhibition of LINC00607 was evident by scRNA-seq (Figure 4D). However, UMAP did not reveal the 3 groups of ECs as separated clusters (Supplementary Figure 7A), suggesting that ECs remain largely same population despite the treatments. To identify the specific genes whose expression is altered by HT and attenuated/reversed by 607-KD, we identified DEGs between NM-scr vs. HT-scr and those between HT-scr and HT-KD. Such analysis revealed 60 DEGs (Supplementary Figures 7B,C). This list of reversible genes includes FN1, SERPINE1, and THBS1, all of which were also downregulated by 607-KD in the bulk RNA-seq (Figures 4E,F; Supplementary Figures 7C,E). On the other hand, such analysis also identified 67 genes that were irreversible upon 607-KD such as CCL2, SMAD3, and ICAM1 (Figures 4G,H; Supplementary Figures 7D,F).

Next, we focused on the HT-KD samples to dissect the effect of LINC00607 inhibition in single ECs. Based on LINC00607 levels, the ~3,000 ECs were separated into 7 clusters (Figure 4I), indicating the expected heterogeneity in the transfected ECs. Cluster 6 exhibited the highest abundance of LINC00607, followed by Clusters 0 and 1. In contrast, Cluster 4 exhibited the lowest LINC00607 level (Figures 4I–K). Differential expression analysis of the 3 clusters with highest LINC00607 levels (hereafter called 607-Hi cells) relative to Cluster 4 (called 607-Lo cells) identified 609 DEGs including SERPINE1, THBS1, and COL4A1, which were also 607-KD-reversible genes (Figures 4L–N). In line with the bulk RNA-seq data in which 607-KD was performed prior to HT treatment, 607-Lo cells also expressed lower levels of various pro-inflammatory, pro-ECM remodeling, and pro-fibrotic genes (Figure 4O). These data support that ECs with low LINC00607 levels, most likely due to LINC00607 inhibition, are also ECs with LINC00607-downstream genes suppressed, corroborating a positive role of LINC00607 in the sustained effect of HT.