Proteomic and Functional Analysis on Endothelial Cell Heterogeneity Identifies Key Regulators in Hyperglycemia-Induced Dysfunction

Abstract

Endothelial cells (ECs) play a critical role in managing vascular homeostasis and neovascularization. EC functions vary significantly depending on their anatomic locations, especially for ECs forming macrovascular versus microvascular vessels. ECs possess heterogeneous signaling pathways, energy metabolism, and cellular behaviors that enable them to handle both physiological and hyperglycemic conditions. These variations can impact the efficacy of pharmacotherapy and influence the likelihood of unexpected side effects. In this study, we compared human aortic ECs (HAECs) and human dermal microvascular ECs (HDMVECs) to observe the functional and proteomic differences potentially contributing to EC heterogeneity. Compared to HAECs, HDMVECs exhibited faster proliferation, but lower migration and permeability. Under high glucose, migration was worsened for both cell types, whereas proliferation was unaffected, and permeability increased for HDMVECs. Using proteomic analysis, we identified 126 proteins whose abundance was significantly different between HAECs and HDMVECs. DAVID analysis revealed their biological processes, cellular compartments, molecular functions, and pathways. under high glucose, WARS1 increased while SOD2 decreased. Reversing WARS1 or SOD2 expression levels improved HDMVEC migration and permeability functions. The combined treatment of WARS1 knockdown and SOD2 overexpression fully restored EC migration and reduced permeability to levels comparable to those of their counterparts under normal glucose conditions. Furthermore, in the cutaneous wound in type 2 diabetic mice, the combination therapy of Wars1 knockdown and SOD2 overexpression accelerated the wound closure and augmented wound angiogenesis. Our studies provide novel molecular insights into EC heterogeneity and identified WARS1 and SOD2 as potential targets for dermal angiogenesis during tissue repair.

NEW & NOTEWORTHY

Our research highlights the functional and proteomic differences between HAECs and HDMVECs, Importantly, we identified WARS1 as a novel target in HDMVECs. Together with SOD2, correcting the abnormalities of these two molecules, HDMVECs’ migration and permeability can be fine-tuned under high glucose conditions. Furthermore, WARS1 knockdown and SOD2 overexpression accelerated wound healing in type 2 diabetic mice, highlighting the therapeutic potential of targeting WARS1 and SOD2 to address delayed wound healing in diabetes.

Graphical Abstract

INTRODUCTION

Endothelial cells (ECs) form the inner lining of arteries, veins, and capillaries and play a critical role in managing vascular homeostasis (2, 4). Not only do they actively control vascular tone, but they also regulate the immune cell attachments and the exchange of a large number of substances, including nutrients, metabolites, and hormones, between the blood vessels and the surrounding tissue (13, 18). ECs may function differently according to their anatomic locations, especially in macrovascular and microvascular beds, and thus their dysfunction induces distinct clinical manifestations under pathological conditions. Organ/tissue-specificity may also impact EC function and gene expression because the physical characteristics, blood flow rates, and the peri-vascular environment of that particular organ could impact how the ECs respond to abnormalities (10, 29). For example, human placental microvascular ECs and human umbilical vein ECs have different expression levels of genes in G-protein coupled receptor signaling pathways, angiogenesis, L-lysine transmembrane transport, and blood vessel remodeling (16). Therefore, characterizing the molecular and functional phenotypes of ECs from micro- and macro-vascular beds can aid in more targeted interventions with fewer side effects.

Skin wound healing is a dynamic process of replacing damaged tissues and restoring epidermal and dermal layers. It has four overlapping phases, including hemostasis, inflammation, proliferation, and remodeling (8). During the healing course, the proliferation phase requires the increase of blood flow to the wound site to support re-epithelialization and granulation tissue formation, hence the most active phase for dermal microvascular EC-mediated neovascularization (1, 17, 31, 32). Acute wounds typically heal completely within a few weeks, depending on the wound size and depth. However, a small wound can develop into chronic or non-healing ulcers in patients with diabetes mellitus (6, 11, 30). Reduced blood flow due to insufficient neovascularization slows down wound closure and consequently increases the risk of infection in wounds in patients with diabetes (5, 9, 14). In many in vitro studies that attempted to investigate the molecular mechanisms underlying wound angiogenesis, researchers often used ECs from large vessels, such as umbilical veins, aortas, or EC cell lines (12, 36). However, there is a lack of information on how tissue-specific heterogeneity affects EC behaviors in hyperglycemia and how this might interfere with therapeutic interventions in treating wounds in patients with diabetes.

In this study, we compared the cellular behavioral differences between human aortic ECs (HAECs) and human dermal microvascular ECs (HDMVECs) in the absence and presence of high glucose. Through proteomic analysis, we identified remarkable heterogeneity in protein abundance profiles between these two types of ECs. We hypothesized that the alteration of key proteins in HDMVCs was responsible for the vulnerability of HDMVECs under high glucose conditions and observed that reversing the abnormalities of two key proteins, WARS1 and SOD2, rescued the impaired functions of HDMVECs. These molecular insights into EC heterogeneity may facilitate therapy development for wound healing in the near future.

MATERIALS AND METHODS

Cell Culture and Treatment

Human aortic ECs and human dermal microvascular ECs were purchased from Lonza Walkersville, Inc. and cultured in either endothelial growth media-2 (EGM-2; Lonza, CC-3162) or microvascular endothelial growth media-2 (EGM-2 MV; Lonza CC-4147) at 37 °C and 5% CO₂. These ECs were used at passages 4 to 10 (donor information: HDMVECs n = 5, age 53.4 ± 5.84 yrs, male/female = 0:5, for all studies; HAECs n = 5 for proteomic profiling, age 51.00 ± 8.89 years, male/female = 3:2; HAECs n = 4 for Western blot and functional studies, age 51.75 ± 3.49 yrs, male/female = 3:1). In some experiments, HAECs and HDMVECs were treated with a control (5 mM glucose media with 20 mM Mannitol) or high glucose (25 mM glucose) and incubated for 72 hours before experiments were performed.

Cell Function Assays

EC functions were assessed by examining different aspects of the cellular behaviors such as proliferation, migration and EC layer permeability. EC proliferation was evaluated using the BrdU Cell Proliferation Assay Kit (Cell Signaling, #6813). Optical density at 450 nm was recorded using a BioTek Epoch spectrophotometer. Cell migration was evaluated with the Boyden chamber migration assay (Transwell Costar, 6.5 mm diameter and 8 μm pore size) and migrated cells were counted using an EVOS fluorescence microscope under ×20 magnification. For EC permeability, cells were seeded into a Transwell Costar chamber (12 mm diameter and 0.4 μm pore size) and grown to 100% confluency before Fluorescein isothiocyanate (FITC) –dextran (Sigma Aldrich, FD10S-100MG) was added. Fluorescence was measured every 30 minutes for 2.5 hours using the GloMax Explorer Multimode Microplate Reader.

Proteomic Profiling

Proteomics has emerged as a powerful tool to assay endpoints of the proteome to reveal information on the underlying biochemistry and physiologic state of tissue or cells. We have adopted the tandem mass tags (TMT) approach, which offers the capability to quantify multiple samples (each sample labeled with a unique TMT tag) under different conditions in a single proteomics experiment. This enables us to quantify thousands of proteins simultaneously and understand their regulatory network under normal physical and pathological conditions with or without treatments. The proteomics data can then be used to select relevant markers and pathways and further analyzed by functional assays.

The HAECs and HDMVECs were cultured at passage 4 and lysed using CellLytic buffer (Sigma) mixed with Halt Protease and Phosphatase Inhibitor Cocktail (Thermofisher Scientific, 1:100 dilution). BCA Protein assay was performed to quantify the resulting cell lysate, and 250 μg of proteins from each sample were subjected to proteomic analysis as we previously described (23) with recent modifications. In brief, the proteins were digested in-solution by trypsin. The concentration of resulting peptides was measured using a peptide fluorometric assay and 100ug peptide from each sample was labeled with 10 unique Tandem Mass Tags (TMT) out of the TMT11plex (Thermo Fisher Scientific). After checking labeling efficiency using LC-MS/MS (all >95%), TMT-tagged samples were combined and fractioned using high pH reverse phase fractionation kit. Each fraction was further purified by solid-phase extraction (C18 ZipTip, Millipore, Billerica, MA) and subjected to high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (HPLC-ESI-MS/MS) analysis. HPLC-ESI-MS/MS was performed on a Thermo Scientific^™ Orbitrap Fusion^™ Lumos^™ Tribrid^™ mass spectrometer, fitted with a nanospray flex Ion source (Thermo Fisher, San Jose, CA). Online HPLC was performed using a Thermo Scientific^™ Dionex^™ UltiMate 3000 RSLC nano-System: NCS-3500RS, with a C18-reversed phase column (75 μm ID, 45cm length) packed in-house with ReproSil-Pur C18–3μm resin (Dr. Maisch GmbH, Germany), maintained at 60°C. The peptides were separated with a binary buffer system of 0.1% FA in water (buffer A) and 0.1% FA in acetonitrile (buffer B), at a constant flow rate of 300 nL/min, with a linear gradient chromatography of 2–45% buffer B (100% ACN and 0.1% FA in H2O) over a 220 minute method; starting with 2% B and increase to 8% B at 10min, to 30% B in 165 min, 45% B in 180 min, to 90% B in 3 mins, hold at 90% B for 10 min, and then 2% B at a flow rate of 300 nL/min for the remaining of 220 minutes.

A “top 3 second” data-dependent tandem mass spectrometry approach was utilized to identify peptides in the samples. MS1 scan, the survey scan, was acquired using orbitrap mass analyzer with a scan range of 400–1600 m/z, with a resolution of 120,000, followed by higher-energy collisional dissociation (HCD) mass spectra MS2, obtained using the orbitrap. Top intensity ions were selected for fragmentation during the 3s time window, using a 1.3 m/z isolation window and a HCD collision energy of 35, 40 and 45%. Fragment ions were detected in orbitrap at resolution of 30,000 with maximum injection time of 200 ms.

The RAW mass spectrometry files were analyzed using MaxQuant proteomics software package (version 2.0.3.0), and searched against a database with forward and reversed Uniprot homo-sapiens protein sequences downloaded from www.uniprot.org, using the default MS2 settings of MaxQuant for TMT11 plex. The parent mass tolerance was set to 5 parts per million (p.p.m.), and the fragment mass tolerance was 0.5 Daltons (Da). Up to two missed trypsin cleavage sites were permitted. Variable modifications included methionine oxidation (M), phosphorylation (STY), and N-terminal acetylation (Protein N-term). Carbamidomethylation (C) was specified as a fixed modification. The false discovery rate (FDR) for both proteins and peptides (with a minimum length of 7 amino acids) was maintained at 0.01. Only proteins identified with a minimum of two unique peptides were considered.

A total of 4004 proteins were identified. The normalized TMT reporter ion intensities were used to calculate the relative abundance for each protein in the samples using the method described in our previous publication (23). The resulting normalized protein ratios were log 2 transformed and compared between HAECs and HDMVECs using HAECs as the control group. Independent t-test was used to assess the statistical significance of the findings. A total of 126 differentially expressed proteins were identified based on fold change (1.5-fold increase or 0.67-fold decrease) and p-values (p<0.05). These 126 proteins were then analyzed through Database for Annotation, Visualization, and Integrated Discovery (DAVID) to obtain gene ontology biological processes, cellular compartments, and molecular functions as well as KEGG pathway analysis for proteins of interest.

Western Blot Analysis

Western blot analysis was performed on cell lysate for HDMVECs using primary antibodies directed against each target molecule: BASP1 (1:500, 703692), WARS1 (1:1000, PA5–29102), and RCN3 (1:1000, PA5–98793) from Thermo Fisher; LXN (1:1000, ab154744), MYO1B (1:1000, ab194356), and MnSOD (1:1000, ab13533) from Abcam; WARS2 (1:1000, NBP1–54653) from NovusBio; and β-actin (1:10,000, no. 12262) from Cell Signaling. The blots were incubated with secondary antibodies, including IRDye 680 RD goat anti-mouse (926–68070) and IRDye 800 CW anti-rabbit (926–32211) from Li-Cor at a 1:20,000 dilution. After washing, the blots were then imaged with an Odyssey CLx Imaging System (Li-Cor) and quantified using Image Studio Lite Ver 5.2 (Li-Cor).

Silencing RNA Transfection

HDMVECs were plated in six-well plates at 80% confluence before transfection with small interfering RNA (siRNA). DharmaFECT transfection reagent was used according to the manufacturer’s instructions from Horizon Discovery. WARS1 siRNA (5nM) and a scramble siRNA control (5nM) were incubated for 72 hours in either 25mM glucose media or the control (5 mM glucose media with 20 mM mannitol). Cells were then used for function assays or real-time PCR to confirm transfection success.

Plasmid Overexpression

The SOD2 overexpression plasmid (#RC202330) and the according control plasmid pCMV6-Entry (#PS100001) were purchased from ORIGENE. TurboFectin reagent was used as per the manufacturer’s protocol. HDMVECs were treated in 6-well plates for 72 hours in combination with the high glucose or mannitol treatment before being used in function assays.

Real-time PCR

Total RNA from cells was isolated by the RNeasy Mini Kit from Qiagen and then reverse transcribed to cDNA using High Capacity cDNA Reverse Transcription kit (Applied Biosystems, 4368814). For mRNA expression analysis, quantitative real-time PCR was performed using primers synthesized by Integrated DNA Technologies. Amplification and detection of specific gene products were performed with the LightCycler 480 (Roche) System using Fast SYBR^™ Green Master Mix (Thermo Fisher Scientific). The PCR cycling conditions: 94°C for 5 min, and then 35 cycles of 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s, followed by extension at 72°C for 5 min. The Ct value was normalized by subtracting the Ct value of the housekeeping gene GAPDH, which gave the ΔCt value for each sample. The mean ΔCt value of the control group was calculated (Mean ΔCt_control). Finally, the relative expression of the target gene for each sample was calculated using the following equation: 2-(ΔCt_individual - Mean ΔCt_control). The data were presented as the fold-change vs. the control group.

Combined Wars1 Knockdown and SOD2 overexpression for Skin Wound Healing In Vivo

Adult male type 2 diabetic mice (BKS.Cg-m^+/+•Lepr^db/J, db/db, age 16 weeks, plasma glucose 527.7 ± 50.6 mg/dL, n = 10) and their age- and sex-matched nondiabetic healthy littermates (BKS.Cg-m^−/−•Lepr^db/− lean, db/+, plasma glucose 201.8 ± 16.82 mg/dL, n = 5) were purchased from The Jackson Laboratory (Bar Harbor, ME). A full-thickness excisional wound was created on the dorsal skin of the mouse using a 7-mm punch biopsy without hurting the muscle, as we previously described. For Wars1 siRNA and SOD2 plasmid co-transfection in the wound bed, immediately after wounding, an aliquot of 1 μM Wars1 siRNA and 2μg/mL SOD2 plasmid in 60μL of Opti-MEM mixed with Lipofectamine^™ 2000 reagent (ThermoFisher) was dispensed onto the wound bed. We have used this protocol in previous studies (33, 34), which has been proven to yield desirable tissue gene knockdown/overexpression. Wounds were sealed with transparent oxygen-permeable wound dressing (3M). The dressings were changed every other day. The transfection of siWars1 was repeated every two days, while the transfection of SOD2 was repeated every 4 days when changing dressings. Wound closure rates were measured by tracing the wound area onto acetate paper. The tracings were digitized, and the areas were calculated with a computerized algorithm and converted to percent wound closure (ImageJ). Wound closure rates were calculated as percentage closed (y%) = [(area on day₀ − open area on day_x)/area on day₀] × 100. The wound and the adjacent skin were collected for CD31 immunochemistry staining. All animal procedures and care were performed according to Wayne State University Institutional Animal Care and Use Committee guidelines (protocol No. IACUC-23–05-5829).

Statistical Analysis

All values are expressed as mean ± SD. The normality of the data is assessed by the Shapiro-Wilk normality test. For data that follows the normal distribution, the statistical significance of differences between the 2 groups was determined by the Student t test. When more than two groups were compared, one-way ANOVA was used across all the groups, and if significant, pairs of focused groups were tested with 2-sample t test, and P values were adjusted with Benjamini Krieger, and Yekutieli’s adjustment (3). This adjustment controlled the false discovery rate at less than 0.05 due to multiple hypothesis testing. For data that failed the Shapiro-Wilk tests, the statistical significance of the difference between the two groups was determined with the Mann-Whitney U test. When more than two groups were compared, the Kruskal-Wallis test was used across all the groups, and if significant, we then tested the pairs of our primary interest based on scientific rationale using the Mann-Whitney U test with Benjamini, Krieger, and Yekutieli’s adjustment for multiple comparisons. For Western blot, paired t-test was used to compare two groups. For in vivo wound closure, two-way repeated-measures ANOVA, followed by Bonferroni post hoc testing and Benjamini Krieger, and Yekutieli’s adjustment for p values to compare both differences between treatments at different time points. p < 0.05 was considered statistically significant. All the statistical analyses were performed using GraphPad Prism 10 (GraphPad Software).

RESULTS

HAECs and HDMVECs have phenotypic functional differences under basal conditions

To observe the phenotypic differences between HAECs and HDMVECs under basal conditions, we performed functional assays for proliferation, migration, and permeability. Compared with HAECs, HDMVECs proliferated faster (Figure 1A). The migration was comparable between HAECs and HDMVECs (Figure 1B, p = 0.0653). Meanwhile, HDMVEC layers were less permeable than HAEC layers (Figure 1C). These findings suggest robustly divergent behaviors of ECs located in microvasculature and microvasculature.

Figure 1: Proliferation, migration, and permeability function differ between HAECs and HDMVECs.

A. Cell proliferation as measured by BrdU incorporation in HAECs and HDMVECs. B. Cell migration as measured by Transwell Assay in HAECs and HDMVECs. Representative images showing migrated cells were shown in the right panel. Scale bar = 200 μm. C. EC layer permeability as measured by FITC-Dextran Diffusion Assay in HAECs and HDMVECs. In all the experiments, HAECs were from 4 donors, and HDMVECs were from 5 donors. Assays were conducted a minimum of 3 times. Data were expressed as mean ± SD. Mann-Whitney tests were used. *p < 0.05, NS, not significant.

Distinct Proteomic Profiles between HAECs and HDMVECs

To determine whether these functional differences between HAECs and HDMVECs were due to the different protein profiles between the cell types, we utilized comparative proteomic analysis and identified 4004 total proteins presented in both cell types, where 126 proteins met our criteria, that are, substantial differences (>1.5 fold higher or lower) in HDMVECs vs. HAECs and had p-values <0.05 (Table S2). Figure 2A shows the overview of all the proteins identified, with non-significant proteins in gray and 126 proteins highlighted either in red (high abundance in HDMVECs) or in blue (low abundance in HDMVECs). Of these 126 proteins, 70 were higher, and 56 were lower in HDMVECs compared with HAECs. Through the DAVID analysis, these 126 proteins were significantly enriched in 42 biological processes, 38 cellular compartments, and 23 molecular functions. In biological processes, the highest number of proteins were involved in cell adhesion, signal transduction, and proteolysis as shown in Figure 2B. Important processes related to HDMVEC function in wound healing, angiogenesis, extracellular matrix organization, and regulation of cell proliferation are also highly enriched categories. Processes that could affect EC migration include anatomical structure morphogenesis and cell morphogenesis. Actin filament bundle assembly, actin filament organization, and processes involved with cell adhesion could impact EC permeability. The analysis revealed that proteins from most cellular compartments were differently presented, but the top 20 categories, shown in Figure 2C, suggested that proteins located in extracellular exosomes, plasma membranes, and cytosol are the most enriched locations. In the meantime, 23 total molecular functions were identified and the top 20 have been listed in Figure 2D. The highest numbers of proteins were involved in receptor binding, actin binding, and calcium ion binding. Interestingly, proteins involved in EC migration or angiogenesis, including actin and actin filament binding, collagen binding, growth factor binding, structural constituent of the cytoskeleton, and extracellular matrix structural constituent, were also distinct between HDMVECs and HAECs.

Figure 2: Differentially expressed proteins and their distribution over biological processes, cellular compartments, molecular functions, and KEGG pathways in HAECs versus HDMVECs.

HAECs and HDMVECs were cultured, and protein lysates were subjected to proteomic analysis. Both HAECs and HDMVECS were from 5 donors. A. Volcano plot shows the overview of all the proteins identified, with non-significant proteins in gray and 126 proteins that met our selection criteria highlighted either in red (high abundance in HDMVECs) or in blue (low abundance in HDMVECs). The best targets are above the horizontal lines (set at p <0.05) and the outside of vertical lines (fold change > 1.5 or < 0.67). The proteins with gene names were the candidates that we further studied. B-D. Proteins were analyzed for their distribution in biological processes (B), cellular compartments (C), and molecular functions (D) revealed by DAVID analysis. These enriched categories are shown on the Y-axis. The smaller the p-value (listed top right), the more likely a pathway is significantly enriched. The size of the dots indicates protein numbers (listed top right). E. KEGG pathways of the most upregulated vs downregulated proteins. F. Selected KEGG pathways closely associated with wound healing and diabetes. Proteins highlighted in red are higher in HDMVECs and proteins highlighted in blue were low in HDMVECs, compared with HAECs.

For KEGG pathway analysis, upregulated and downregulated proteins from the list of 126 proteins were analyzed either separately (Figure 2E) or together (Figure 2F). We found that proteins of higher abundance in HDMVECs are primarily involved in inflammation and infection processes, whereas proteins of higher abundance in HAECs (hence lower abundance in HDMVECs) are primarily involved in arterial wall modeling and shear stress (Figure 2E). When all 126 proteins were analyzed together, multiple pathways involved in EC function, wound healing, or diabetes were enriched. Figure 2F gives a combined view of KEGG pathways (left) and the proteins associated with them (right). Out of the 5 pathways shown in our selection based on the overall high-fold enrichment, p values, and the relevance to core EC activities, the ECM-receptor interaction is most associated with wound healing.

While the low p-value of a category represents that the proteins from our list of 126 proteins are specifically associated (enriched) in that category by more than random chance, a high-fold enrichment of a category indicates the enrichment of the proteins from the proteins from our list of 126 proteins in that category as compared to the background (the total human proteins), showing how prominent these proteins are in a particular pathway. Note that even if a pathway has the most proteins, it does not necessarily have the best p-value and fold enrichment scores. Therefore, we organized the DAVID analysis for biological processes (Table 1), cellular compartments (Table 2), and molecular functions (Table 3) by p-values below 0.05 and high fold enrichment. Finally, we organized the total 126 proteins by the most marked differences (either higher or lower) and by the best p-values in HDMVECs as compared to HAECs. Of the top 10 proteins, 5 of them were most increased, and 5 of them most decreased in HDMVECs, as shown in Table 4.

Table 1:

Top 20 biological processes and their corresponding proteins.

Biological Processes	Genes	P-Value	Fold Enrichment
cell adhesion	VWF, SRC, ITGA1, FN1, LYVE1, ICAM1, DPP4, CLDN11, RELN, CDH2, COL5A1, ITGA10, SIRPA, PGM5	0.00003	4.184
cell-cell adhesion	PDLIM1, CDH2, SRC, ITGA10, ITGA1, ICAM1, MPZL2	0.001	6.072
cell-matrix adhesion	CD63, ITGA10, ITGA1, FN1, LYVE1	0.004	7.759
cell adhesion mediated by integrin	ITGA10, ITGA1, ICAM1, FBN1	0.002	16.503
signal transduction	ANGPT2, SRC, HHIP, GMFG, RASAL3, NDRG1, NRGN, NFKB2, AKAP12, BST1, ANPEP, TNFSF4, PDE3A, FBN1	0.042	1.812
stimulatory C-type lectin receptor signaling pathway	SRC, PLCG2, PSMB10, PSMB9	0.002	14.709
integrin-mediated signaling pathway	SRC, ITGA10, ITGA1, FN1	0.028	6.096
viral entry into host cell	DPP4, SCARB1, CTSL, ANPEP, CD55, ICAM1	0.0004	9.854
positive regulation of T cell cytokine production	TNFSF4, HLA-A, CD55	0.0018	46.13
angiogenesis	JAG1, ANGPT2, COL4A2, ANPEP, FLT4, TNFAIP2, FN1	0.004	4.644
extracellular matrix organization	COL4A2, COL5A1, ITGA10, ITGA1, PAPLN, NFKB2	0.005	5.342
actin filament bundle assembly	EPS8, MARCKS, MYO1B, DPYSL3, AIF1L	0.0001	22.859
cell morphogenesis	RELN, CDH2, CDH13, CLU, PSMB10	0.003	8.211
actin filament organization	MARCKS, MYO1B, TPM2, SORBS2, CGNL1	0.013	5.457
regulation of cell proliferation	JAG1, FLT4, CLU, NDRG1, PTGS1	0.020	4.833
actin crosslink formation	EPS8, MARCKS, DPYSL3	0.00296	36.24782609
proteolysis	DPP4, THBD, RELN, CTSL, ANPEP, HTRA1, PRSS23, CTSS	0.018	2.955
lipid metabolic process	LRP1, FABP5, PDE3A, PLCG2, CLU	0.041	3.827
heart development	PDLIM1, FN1, SOD2, WDR11, FBN1	0.041	3.827
positive regulation of peptidase activity	RCN3, CTSL, CTSS	0.0015	50.74695652

Bold: higher in HDMVECs, regular: lower in HDMVECs. Fold Enrichment refers to the enrichment of a term for the gene list we identified as compared to the background population of genes. The higher the Fold enrichment, the more proteins involved in a term in the gene list we identified compared to the whole human genome background. The P-values refers to the probability of a term in which the gene list we identified is specifically enriched as compared to the whole human genome background. The smaller the P-Value, the less likely a term is significantly enriched just by chance.

Table 2:

Top 20 cellular compartments and their corresponding proteins.

Cellular Compartments	Genes	P-Value	Fold Enrichment
extracellular exosome	SCARB1, PYGL, HNMT, CLU, PRSS23, SYNE2, ICAM1, EPS8, TUBB8, BASP1, CTSL, ANPEP, LAMP2, COTL1, SIRPA, VWF, ANXA13, ITGA1, HLA-A, SERPINB8, TUBA4A, PROCR, BDH2, COL4A2, PSME2, CDH13, FTL, PFN2, CD63, TTYH3, SRC, HTRA1, PTTG1IP, RASAL3, NDRG1, PTGS1, DPP4, SELENBP1, SEC14L2, PLCG2, TSPAN3, A2M, CD55, FN1, GGH, SOD2, LYVE1, PSMB9, AIF1L, BST1, MYO1B, MARCKS, FABP4, FABP5, CHMP2B	2.36E-21	4.155
extracellular space	CD63, LXN, HTRA1, TNFAIP2, CLU, CTSS, ICAM1, SELENBP1, THBD, RELN, CTSL, ANPEP, LAMP2, DPYSL3, A2M, ANGPT2, VWF, ANXA13, FN1, GGH, HLA-A, SERPINB8, DKK3, PROCR, COL4A2, COL5A1, FABP5, TNFSF4, CDH13, FBN1	3.06E-06	2.551
extracellular region	CD63, HHIP, FLT4, GMFG, PAPLN, HTRA1, PYGL, CLU, CTSS, DPP4, RELN, CTSL, COTL1, A2M, CD55, ANGPT2, JAG1, VWF, FN1, GGH, TUBA4A, DKK3, BST1, PROCR, COL4A2, COL5A1, FABP5, CDH13, FTL, FBN1	2.28E-05	2.299
extracellular matrix	RELN, COL4A2, COL5A1, VWF, PAPLN, FN1, FBN1	0.004	4.574
endoplasmic reticulum lumen	RCN3, COL4A2, CDH2, COL5A1, FN1, PRSS23, FBN1	0.010	3.809
cytoplasm	SCARB1, TRIO, HHIP, FHL1, PYGL, HNMT, CLU, SYNE2, SMG6, EPS8, TUBB8, AKAP12, UBASH3B, CDH2, BASP1, ANPEP, DPYSL3, SH3BGRL2, ANKS1A, ANXA13, SERPINB8, PALD1, TUBA4A, BDH2, DPYD, PSME2, CDH13, TRIM16, KANK3, FTL, PFN2, LXN, SRC, PTTG1IP, RASAL3, NDRG1, PTGS1, PDLIM1, SEC14L2, RELN, SULT1E1, PLCG2, ZC3H14, PPP1R14A, NEK7, SORBS2, WDR11, NFKB2, AIF1L, SULT1B1, MYO1B, MARCKS, FABP4, FABP5, CHMP2B	0.0001	1.613
plasma membrane	SCARB1, HHIP, FLT4, FHL1, ICAM1, EPS8, AKAP12, THBD, CDH2, BASP1, CTSL, ANPEP, LAMP2, COTL1, SIRPA, PLA2G4C, ANXA13, ITGA1, HLA-A, CLDN11, PROCR, PLSCR4, CDH13, TRIM16, CD63, TTYH3, GYPC, LRP1, SRC, HTRA1, MGST1, NDRG1, DPP4, FAM171A1, RELN, EPB41L3, PLCG2, CD55, MPZL2, ANGPT2, JAG1, FN1, SORBS2, LYVE1, BST1, MYO1B, MARCKS, FABP5, ITGA10, TNFSF4, CHMP2B	0.0003	1.582
integral component of plasma membrane	CD63, SCARB1, JAG1, LRP1, GYPC, HHIP, FLT4, TAP1, HLA-A, LYVE1, ICAM1, PROCR, THBD, CDH2, TNFSF4, SIRPA, TSPAN3	0.011	1.980
apical plasma membrane	DPP4, JAG1, CDH2, CTSL, ANXA13, FN1, SORBS2	0.033	2.904
membrane	LRP1, GYPC, MGST1, PTTG1IP, RASAL3, CLU, ICAM1, LRRC1, DPP4, SELENBP1, LAMP2, SIRPA, TSPAN3, JAG1, MGMT, ITGA1, PLA2G4C, ANXA13, TAP1, SORBS2, HLA-A, LYVE1, WDR11, BST1, RSL1D1, MARCKS, COL4A2, PDE3A, PSME2, MXRA7, FTL	0.033	1.427
lysosomal membrane	DPP4, SCARB1, CD63, LRP1, ANPEP, LAMP2, CHMP2B, HLA-A, WDR11	0.003	3.761
membrane raft	DPP4, SRC, ANXA13, ITGA1, PLCG2, CD55, ICAM1	0.004	4.758
cytosol	TRIO, SRC, FLT4, FHL1, HTRA1, GIMAP4, PYGL, RCSD1, RASAL3, HNMT, NDRG1, CLU, PSMB10, NRGN, LRRC1, SMG6, SELENBP1, SEC14L2, AKAP12, SULT1E1, DPYSL3, EPB41L3, PLCG2, COTL1, PGM5, ANKS1A, PPP1R14A, TPM2, PLA2G4C, AFAP1, SERPINB8, PALD1, TUBA4A, PSMB9, WDR11, NFKB2, SULT1B1, BDH2, FABP4, FABP5, DPYD, CHMP2B, RBP1, PDE3A, PSME2, TRIM16, FTL	0.006	1.414
focal adhesion	LRP1, SRC, FHL1, ITGA1, AFAP1, SORBS2, SYNE2, AIF1L, ICAM1, PDLIM1, DPP4, AKAP12, PROCR, MARCKS, CDH2, CDH13, PGM5	5.25E-09	6.629
cell surface	CD63, SCARB1, HHIP, ITGA1, HLA-A, CLU, ICAM1, DPP4, PROCR, THBD, CDH2, TNFSF4, SIRPA, CD55	0.0001	3.606
actin filament	MYO1B, SRC, TPM2, COTL1, RCSD1, AIF1L	0.0003	10.257
cytoskeleton	PDLIM1, AKAP12, LRP1, BASP1, TPM2, EPB41L3, CLU, TUBA4A, PFN2, KANK3, MPZL2	0.002	3.342
neuron projection	RELN, CDH2, SRC, ITGA1, CDH13, ANKS1A, PTGS1	0.023	3.168
adherens junction	PDLIM1, JAG1, CDH2, CDH13, PGM5, NDRG1	0.0047	5.488
cell junction	CLDN11, DPP4, CDH2, BASP1, SRC, EPB41L3	0.012	4.302

Bold: higher in HDMVECs, regular: lower in HDMVECs. Fold Enrichment refers to the enrichment of a term for the gene list we identified as compared to the background population of genes. The higher the Fold enrichment, the more proteins are involved in a term in the gene list we identified compared to the whole human genome background. The P-values refer to the probability of a term in which the gene list we identified is specifically enriched as compared to the whole human genome background. The smaller the P-Value, the less likely a term is significantly enriched just by chance.

Table 3:

Top 20 molecular functions and their corresponding proteins.

Molecular Functions	Genes	P-Value	Fold Enrichment
receptor binding	DPP4, ANGPT2, SRC, TNFSF4, ITGA1, FN1, HLA-A, A2M, CLU	0.0040	3.511
signaling receptor activity	PROCR, THBD, LRP1, ANPEP, LYVE1, ICAM1	0.0150	4.121
actin binding	PDLIM1, EPS8, MARCKS, MYO1B, TPM2, GMFG, EPB41L3, COTL1, AFAP1, SYNE2, PFN2	0.0001	5.170
actin filament binding	MARCKS, MYO1B, TPM2, COTL1, RCSD1, SYNE2, AIF1L	0.0032	4.829
collagen binding	VWF, CTSL, ITGA10, ITGA1, FN1, CTSS	0.0001	14.183
growth factor binding	FLT4, HTRA1, A2M	0.0253	12.055
structural constituent of cytoskeleton	TUBB8, EPB41L3, SORBS2, TUBA4A	0.0300	5.899
extracellular matrix structural constituent	COL4A2, COL5A1, VWF, FN1, FBN1	0.0108	5.782
cadherin binding	RSL1D1, MYO1B, CDH2, SRC, CHMP2B, CDH13, NDRG1	0.0144	3.516
enzyme binding	SRC, PLSCR4, LAMP2, COTL1, FN1, A2M, SOD2	0.0325	2.915
integrin binding	COL5A1, VWF, SRC, ITGA10, ITGA1, FN1, ICAM1, FBN1	0.0001	8.037
proteoglycan binding	COL5A1, CTSL, FN1, CTSS	0.0001	45.925
protein phosphatase binding	CDH2, FLT4, ITGA1, SIRPA	0.0188	7.065
protease binding	DPP4, VWF, FN1, A2M	0.0328	5.690
lipoprotein particle receptor binding	RELN, LRP1	0.0183	107.158
calcium ion binding	RCN3, THBD, JAG1, LRP1, CDH2, MGMT, PLSCR4, ANXA13, PLA2G4C, CDH13, AIF1L, FBN1	0.0071	2.551
collagen binding involved in cell-matrix adhesion	ITGA10, ITGA1	0.030	64.294
fatty acid binding	FABP4, FABP5, RBP1	0.0207	13.395
virus receptor activity	DPP4, SCARB1, ANPEP, CD55, ICAM1	0.0015	10.046
SH3 domain binding	PLSCR4, DPYSL3, SH3BGRL2, SIRPA, AFAP1	0.0082	6.279

Bold: higher in HDMVECs, regular: lower in HDMVECs. Fold Enrichment refers to the enrichment of a term for the gene list we identified as compared to the background population of genes. The higher the Fold enrichment, the more proteins involved in a term in the gene list we identified compared to the whole human genome background. The P-values refer to the probability of a term in which the gene list we identified is specifically enriched as compared to the whole human genome background. The smaller the P-Value, the less likely a term is significantly enriched just by chance.

Table 4:

Top 10 Proteins based on fold change and p-values.

Protein Name	Gene Name	Fold Change ± STDEV	P-value
Superoxide dismutase [Mn], mitochondrial	SOD2	3.197 ± 0.693	0.0000210
Tryptophan-tRNA ligase, cytoplasmic	WARS1	2.728 ± 0.901	0.001
Unconventional myosin-Ib	MYO1B	2.228 ± 0.124	0.000000218
Leucine-rich repeat-containing protein 1	LRRC1	2.071 ± 0.316	0.000251
Cytosolic phospholipase A2 gamma	PLA2G4C	1.903 ± 0.260	0.000246
Neurogranin	NRGN	0.385 ± 0.112	0.000583
Profilin	PFN2	0.390 ± 0.076	0.0000239
Brain acid soluble protein 1	BASP1	0.394 ± 0.151	0.000469
Reticulocalbin-3	RCN3	0.411 ± 0.069	0.0000927
Latexin	LXN	0.525 ± 0.102	0.000917

Bold: higher in HDMVECs, Regular: lower in HDMVECs. Fold change represents the protein levels in HDMVECs compared to the HAECs where above 1.5 is higher and below 0.67 is lower. P-values were calculated from unpaired t-tests between HDMVECs and HAECs.

HAECs and HDMVECs experience dysfunction under high glucose conditions

To compare the functional changes of HAECs and HDMVECs when exposed to stressful conditions, we treated the cells with high glucose exposure (25mM, 72 hours) using osmotically adjusted normal glucose exposure (5mM glucose plus 20mM mannitol, 72 hours) as control. Functional assays for proliferation, migration, and permeability were conducted. We found that although HG did not impact the proliferation of either HAECs or HDMVECs, HDMVECs demonstrated increased proliferation compared to HAECs under both NG and HG conditions (Figure 3A). While HG impaired the migration of both HAECs and HDMVECs, the HDMVECs had worse migration compared with HAECs under both NG and HG conditions (Figure 3B). Meanwhile, while HG increased permeability of both HAECs and HDMVECs (Figure 3C), the two cell types had comparable permeability under NG or HG conditions.

Figure 3: High glucose harms EC functions.

HAECs and HDMVECs were treated with high glucose exposure (HG, 25mM, 72 hours) using osmotically adjusted normal glucose exposure (NG, 5mM glucose plus 20mM mannitol, 72 hours) as control. Functional assays for proliferation (BrdU incorporation), migration (Transwell Assay), and permeability (FITC-Dextran Diffusion Assay) were conducted. A. Cell proliferation of HAECs and HDMVECs under NG and HG conditions. B. EC layer permeability assay in HAECs and HDMVECs. C. Cell migration of HAECs and HDMVECs under NG and HG conditions. Representative images of migrated cells were shown in the right panel. Scale bar=200 μm. In all the experiments, HAECs were from 4 donors, and HDMVECs were from 5 donors. Assays were conducted a minimum of 3 times. Data were expressed as mean ± SD. The Kruskal-Wallis test with the post hoc Mann-Whitney U test was used in Figure 3A. The one-way ANOVA with post hoc 2-sample t tests was used in Figure 3B and 3C. For all the figures, P values were adjusted with Benjamini Krieger, and Yekutieli’s adjustment *p < 0.05, NS, not significant.

SOD2 and WARS1 abundance change in HDMVECs under hyperglycemic conditions.

Our proteomic studies have identified the top 10 proteins with differences between HAECs and HDMVECs, including SOD2, WARS1, MYO1B, LRRC1, PLA2G4C, NRGN, PFN2, BASP1, RCN3, and LXN. After investigating their biological functions, LRRC1, PLA2G4C, NRGN, and PFN2 were put to a lower priority for investigation because of the low abundance or the large variations within the HDMVEC group, and we further investigated SOD2, WARS1, MYO1B, BASP1, RCN3 and LXN. To confirm that the mannitol treatment did not affect protein abundance, we compared treated cells to cells grown in the basal medium using western blotting. For all proteins analyzed, there was no significant difference in protein abundance between mannitol and basal medium conditions (Figure S1). Our results suggested that SOD2 was increased (Figure 4A) while WARS1 was decreased (Figure 4B) under HG conditions when compared with NG conditions. In the meantime, the isoforms of WARS1, WARS2, and the other target proteins, including MYO1B, BASP1, RCN3, and LXN, were comparable between the HG and NG groups (Figures 4C–4G).

Figure 4: SOD2, WARS1, and WARS2 protein abundance increased under high glucose.

(A-G) Western blot analysis of SOD2, WARS1, WARS2, MYO1B, BASP1, RCN3, LXN, and β-actin abundance in NG and HG-treated HDMVECs. n = 5 per group. Representative bands are shown below. The p-values were determined by paired t-tests between the two groups. In all the dot-plot figures, horizontal lines show mean ± SD. *p < 0.05. ***p < 0.001. NS, not significant.

Knockdown of WARS1 and SOD2 overexpression recovers HDMVEC function

Because WARS1 abundance was increased (Figure 4A) while SOD2 was decreased (Figure 4B) following high glucose treatment, we performed WARS1 knockdown (via transfection of WARS1 siRNA) and SOD2 overexpression (via transfection of a plasmid carrying the human SOD2 gene) in HDVMECs to explore if reversing these changes could restore the EC function. Quantitative real-time polymerase chain reactions verified the reduction of WARS1 (Figure 5A) and overexpression of SOD2 in HDMVECs (Figure 5D). The functional assay results demonstrated that, under normal conditions, WARS1 knockdown does not affect migration (Figure 5B) or permeability (Figure 5C). However, the knockdown of WARS1 after high glucose exposure reversed the dysfunction. Similarly, the overexpression of SOD2 increased migration and reduced permeability under high glucose conditions (Figures 5E and 5F). However, this change was not as remarkable as the WARS1 knockdown. Furthermore, the combination of WARS1 knockdown and SOD2 overexpression significantly improved HDMEC migration (Figure 5H) and reduced permeability under high glucose conditions (Figure 5I). The real-time PCRs revealed that manipulating either of them did not alter the gene expression of the other (Figure 5G). Indeed, our results suggested that siWARS1+SOD2 achieved comparable EC migration and permeability as the ECs under normal glucose conditions (Figure 5H, 5I). We speculated that reversing WARS1 or SOD2 affects different but signaling pathways, because the siWARS1 achieved comparable EC migration and permeability as the combination therapy did (Figure 5H, 5I), while SOD2 expression achieved comparable EC migration as the combination therapy did (Figure 5H), but not permeability (Figure 5I). These observations suggested that WARS1 and SOD2 modulate different aspects of EC behaviors during angiogenesis.

Figure 5: WARS1 Knockdown recovers EC migration and reduces EC permeability under high glucose.

A. Real-time quantitative polymerase chain reaction (real-time PCR) confirmed the reduced expression levels of WARS in HDMVECs treated with NG and HG conditions. B. Cell migration of HDMVECs with WARS1 knockdown and NG or HG treatment. C. EC layer permeability assay in HDMVECs with WARS1 knockdown and NG or HG treatment. D. Real-time quantitative polymerase chain reaction (real-time PCR) confirmed the increased expression levels of SOD2 in HDMVECs treated with NG and HG conditions. E. Cell migration of HDMVECs with SOD2 overexpression and NG or HG treatment. F. EC layer permeability assay in HDMVECs with SOD2 overexpression and NG or HG treatment. G. real-time PCR confirmed the levels of WARS1 and SOD2 in HDMVECs treated with HG conditions. H. Cell migration of HDMVECs with WARS1 knockdown and SOD2 overexpression in HG treatment. I. EC layer permeability assay in HDMVECs with WARS1 knockdown and SOD2 overexpression in HG treatment. Assays were conducted a minimum of 3 times. Representative images of migrated cells are shown in the right panel. Scale bar=200 μm. n = 5 for all experiments with HDMVECs. Data were expressed as mean ± SD. For all the figures, one-way ANOVA with 2-sample t test was used, and P values were adjusted with Benjamini Krieger, and Yekutieli’s adjustment. *p < 0.05, **p < 0.01, NS, not significant.

Combination therapy of Wars1 knockdown and SOD2 overexpression accelerated wound closure and enhanced wound angiogenesis in hyperglycemia.

In a full-thickness excisional wound in a type 2 diabetic mouse model, the db/db mouse, we applied topical WARS1 knockdown and SOD2 expression therapy to the wound immediately after wounding via the transfection of WARS1 siRNA and SOD2 plasmid. The transfection of the scrambled and empty plasmid served as a control. The results indicated that the combined therapy accelerated wound closure in diabetic mice starting from day 2 when compared with wounds receiving the scramble oligo and control plasmid (Scr+Ctrl) (Figure 6A). In the meantime, measurement of capillary density in the wound bed using CD31 immunohistology suggests that the combined WARS1 siRNA + SOD2 therapy enhanced wound angiogenesis for the healing process (Figure 6B). Taken together, we believe that the combination therapy of Wars1 knockdown and SOD2 overexpression would be able to fully recover wound angiogenesis while partly rescuing the delayed wound closure in animals with hyperglycemia.

Figure 6: Combined WARS1 knockdown and SOD2 overexpression improved wound healing under hyperglycemic conditions in vivo.

A. The wound closure curves in a mouse model of male type 2 diabetes (db/db) receiving mouse Wars1 siRNA and human SOD2 plasmid (siWars1+SOD2) transfection, using Scramble oligo and control plasmid (Scr+Ctrl) transfection as controls in db/db mice (diabetic controls) and their in-colony healthy control db/+ mice. Representative images of the wounds are shown below the curves. n = 5 per group. *p < 0.05 vs. db/db - Scr+Ctrl; #p < 0.05 vs. db/+ - Scr+Ctrl. Two-way repeated-measures ANOVA, followed by Bonferroni post hoc testing and Benjamini Krieger, and Yekutieli’s adjustment for p values to compare both differences between treatments at different time points. B. The quantifications of capillary density as measured by CD31 immunohistology staining in wound sections. The tissues positive for CD31 were stained brown (red arrow). Representative images of wound sections are shown below the scatter plot. Scale bar=200 μm. n = 5 per group. *p < 0.05 vs. db/db - Scr+Ctrl. The Kruskal-Wallis test was used across all the groups, and the pairs of interest were tested using the Mann-Whitney U test with Benjamini, Krieger, and Yekutieli’s adjustment for multiple comparisons.

DISCUSSION

While they are critical in maintaining vascular homeostasis, ECs are key communicators for other cell types and send signals to initiate inflammation, an immune response, or angiogenesis in the event of injury. In this study, we observed cellular behavioral differences between microvascular ECs and macrovascular ECs, suggesting heterogeneity of EC functions at different anatomic locations. The proteomic analysis provides novel insights into unique protein profiles for EC subtypes. Whilst both cell types experience similar dysfunction, such as impaired migration when exposed to high glucose conditions, HDMVECs’ permeability is more likely to increase in response to high glucose-induced impairment compared with HAECs.

The proteomic profiling analysis showed that, out of the 4004 proteins detected in both HAECs and HDMVECs (Figure 2A), 126 had significantly different abundance levels, which might be associated with the functional differences observed in Figure 1. The Spectral Abundance Factor (SAF) in Table S1 represents the estimated protein abundance in a sample, where a higher SAF indicates higher abundance. The proteins with the highest SAF but comparable between the two cell types include GAPDH (5.15), CFL1 (3.19), LGALS1 (3.17), MYL6 (2.97), and ANXA2 (2.61). The proteins with the highest SAF and p-value < 0.05 in HAECs and HDMVECs, but not meeting the fold change criteria, include ENO1 (2.50), PKM (2.11), TUBA1C (2.10), and CALM2 (2.04). FABP4 had the highest fold change in HDMVECs vs. HAECs and was statistically significant. However, we did not look into this molecule due to the large variation within HDMVEC samples that could contribute to the high fold changes. Speculated based on their biological functions, these proteins are essential to maintaining the homeostasis of the cells and are less likely to be impacted by the anatomic locations. On the other hand, KEGG pathway analysis indicates that proteins with higher levels in HAECs (than HDMVECs) are enriched in pathways that regulate the actin cytoskeleton, fluid sheer stress, and atherosclerosis, whereas proteins with higher levels of in HDMVECs (than HAECs) are enriched in pathways participating in inflammation, cell adhesion, and virus infection, reflecting the specialized functions that have been acclimated to the microenvironment of the tissue or organs (Figure 2E). Upon testing the top candidates that may account for the functional differences between HDMVECs and HAECs, we identified WARS1 and SOD2 as major molecules for HDMVECs functional loss under high glucose.

WARS1, also known as Tryptophanyl-tRNA Synthetase 1 (TrpRS), catalyzes the aminoacylation of tRNA (trp) with tryptophan. We demonstrated that WARS1 mRNA (Figure 5A) and protein (Figure 4B) levels increased upon exposure to high glucose, suggesting transcriptional activation of WARS1 expression by high glucose. WARS1 has been reported to be a downstream target of the unfolded protein response branch, PERK-ATF4, in regulating protein homeostasis (21). Given the activation of PERK-ATF4 under hyperglycemia occurs in multiple tissues and cell types, including ECs (7, 19, 39), it is possible that hyperglycemia induces WARS1 expression via PERK-ATF4 axis. Another question that the current study cannot address is how WARS1 affects HDMVEC migration and permeability. It is unclear why a tRNA synthetase can modulate cellular behavior, as there has been limited information on this enzyme in ECs. A previous report showed that full-length WARS1 had no angiostatic effect under physiologic conditions whereas mini-WARS1 (mini-TrpRS) inhibited EC proliferation and migration (24). Although the current study was unable to distinguish whether the modulation of EC function was due to full-length or mini-WARS1, it will be an interesting area to explore in future studies. Notably, Studies have shown that WARS1 can be secreted to the extracellular space and act as an endogenous ligand for toll-like receptors (TLR)2/4 (25, 26). The latter, when activated, induces genes that signify inflammatory responses, increases EC permeability, and either promotes or inhibits EC migration in a context-dependent manner (20, 37). The aforementioned interesting observations and speculations warrant further investigations into the role of WARS1 in regulating angiogenesis in cardiometabolic disorders.

SOD2 is a protein located in the mitochondria whose primary function is to process superoxide into less harmful products, making it a critical regulator of oxidative stress (22). Decreased SOD2 expression would allow the accumulation of free radicals, thus contributing to EC dysfunction. While SOD2 expression under HG helped recover EC migration (Figure 5E) and permeability (Figure 5F), the effect was less drastic compared to WARS1 knockdown. The combination treatment of siWARS1+SOD2 achieved comparable EC migration and permeability as the ECs under normal glucose conditions (Figure 5H, 5I). We speculated that reversing WARS1 or SOD2 would affect distinct signaling pathways, because the siWARS1 achieved comparable EC migration and permeability as the combination therapy did, but SOD2 overexpression achieved comparable EC migration (Figure 5H), but not permeability (Figure 5I), as the combination therapy did. These observations suggest that WARS1 and SOD2 modulate different aspects of EC behavior, possibly through distinct but interacting pathways. For example, the secreted WARS1 acts as a ligand for TLR2/4 in the inflammatory response, while SOD2 neutralizes toxic superoxide radicals in the mitochondria. Both inflammation and oxidative stress are crucial features of diabetes-associated vascular disorders, contributing to EC dysfunction (35).

We did not observe significant sex-related differences in EC proliferation, migration, and permeability in either HAEC or HDMVECs, based on our years of experience working with these two cell types. This was possibly because we chose donors aged between 45 and 70, an age range in which the benefit of estrogen on the cardiovascular system diminishes(27, 28). In relation to the clinical scenario, the delayed wound closure is usually seen in middle-aged and senior patients with a long history of type 2 diabetes, whereas young patients with diabetes have a much lower incidence. In this study, due to the limited pool of HDMVECs from the vendor, we only had female donors at comparable ages to choose from. We observed from the proteomic analysis that HDMVECs exhibited significantly different levels of HSD17B2 (higher) and SULT1E1 (lower) compared to HAECs. HSD17B2 converts estradiol to estrone, while SULT1E1 catalyzes the sulfation of estrogens to inactivate them (15, 38). We speculated that the difference in these two molecules in proteomic studies was because the HDMVECs were from female donors. However, to specify sex-specific differences, future studies using young donors are needed.

We recognize that there were limitations to this study regarding our donor demographics. Although we would have preferred that all our donors were similar in age, BMI, and sex, we have a limited pool of donors. The reason for using HDMVECs from female donors in experiments has been discussed above. The primary human ECs can only be used between passages 4–10, so we have limited cells from each donor. We started the experiments presented in Figures 1 and 2 using HAECs from 5 donors. However, the HAECs from one donor were exhausted from our pool and the company’s inventory after proteomic studies. We were unable to find another suitable donor at the time, as the donors in the inventory had comorbidities or were of an unmatched age, which excluded them from being selected. Therefore, for the function assays and Western blot under high glucose, we only had 4 HAEC donors. We are looking to perform this analysis in future projects and include a larger set of donors. Future comparisons of microvascular ECs from different anatomic locations will help us better understand which aspects of EC heterogeneity cause differences in function.

In conclusion, the present study demonstrates that heterogeneity in protein abundance in ECs from large versus small vessels impacts EC function. Our observations highlight the importance of testing multiple cell types based on their anatomical location to avoid generating side effects when manipulating protein targets. Furthermore, we highlighted proteins that may be critical for EC function and require further testing to evaluate their therapeutic potential.

Supplementary Material

SUPPLEMENTAL MATERIAL

Supplemental Figure. S1 and Supplemental Tables S1-S2 can be accessed at: http://dx.doi.org/10.6084/m9.figshare.30273070

GLOSSARY

ECs

Endothelial cells

HDMVECs

Human dermal microvascular endothelial cells

HAECs

Human aortic endothelial cells

DM

Diabetes Mellitus

BASP1

Brain Abundant Membrane Attached Signal Protein 1

SOD2

Superoxide dismutase [Mn], mitochondrial

WARS1

Tryptophanyl-TRNA Synthetase 1

RCN3

Reticulocalbin 3

LXN

Latexin

MYO1B

Unconventional Myosin-Ib

NG

Normal glucose

HG

High glucose

DATA AVAILABILITY

Data will be available upon reasonable request.