Skip to main content
NCBI home page
BMJ Open Diabetes Research & Care logoLink to BMJ Open Diabetes Research & Care
. 2024 Jul 16;12(4):e004189. doi: 10.1136/bmjdrc-2024-004189

Single-cell transcriptomic analysis reveals the antiangiogenic role of Mgarp in diabetic retinopathy

Ling Ren 1,0, Jiao Xia 2,0, Chang Huang 1,0, Yun Bai 3, Jin Yao 4,*, Dan Li 1,*, Biao Yan 2,✉
PMCID: PMC11268071  PMID: 39013633

Abstract

Introduction

Diabetic retinopathy (DR) is a common vascular complication of diabetes mellitus and a leading cause of vision loss worldwide. Endothelial cell (EC) heterogeneity has been observed in the pathogenesis of DR. Elucidating the underlying mechanisms governing EC heterogeneity may provide novel insights into EC-specific therapies for DR.

Research design and methods

We used the single-cell data from the Gene Expression Omnibus database to explore EC heterogeneity between diabetic retinas and non-diabetic retinas and identify the potential genes involved in DR. CCK-8 assays, EdU assays, transwell assays, and tube formation assays were conducted to determine the role of the identified gene in angiogenic effects.

Results

Our analysis identified three distinct EC subpopulations in retinas and revealed that Mitochondria-localized glutamic acid-rich protein (Mgarp) gene is potentially involved in the pathogenesis of DR. Silencing of Mgarp significantly suppressed the proliferation, migration, and tube formation capacities in retinal endothelial cells.

Conclusions

This study not only offers new insights into transcriptomic heterogeneity and pathological alteration of retinal ECs but also holds the promise to pave the way for antiangiogenic therapy by targeting EC-specific gene.

Keywords: angiogensis, Diabetes Complications, Diabetic Retinopathy

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Retinal endothelial heterogeneity has been observed in the pathogenesis of diabetic retinopathy (DR) and represents a promising therapeutic target for the disease.

WHAT THIS STUDY ADDS

  • We identified an endothelial cell (EC)-specific gene, Mitochondria-localized glutamic acid-rich protein (Mgarp), involved in pathological angiogenesis in DR. Silencing of Mgarp effectively suppressed retinal endothelial angiogenic effects in vitro.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • This study provides new insights into the transcriptomic heterogeneity of retinal vascular ECs and highlights Mgarp as a potential target for DR therapy.

Introduction

Diabetic retinopathy (DR) emerges as one of the most prevalent complications of diabetes mellitus and represents a leading cause of visual impairment in developed nations.1,3 The increasing prevalence of DR is anticipated to affect 191 million individuals by the year 2030, posing a great burden on global public health.4 DR is characterized by retinal neurodegeneration and vascular dysfunction, encompassing increased retinal vasopermeability, vascular inflammation, and pathological neovascularization.5 Among these pathological processes, abnormal retinal angiogenesis stands out as the most detrimental manifestation of DR, giving rise to vision-threatening complications such as vitreous hemorrhage and retinal detachment.6,8 As a result, antiangiogenesis emerges as a primary strategy for DR treatment.

Endothelial cells (ECs), constituting the inner lining of blood vessels, play a pivotal role in vasculogenesis, angiogenesis, and blood vessel homeostasis.9 10 Notably, the dysfunction of retinal vascular ECs serves as an early event and a key driver in the pathogenesis of DR, as well as in the context of pathological retinal neovascularization.11 Given their significant role, retinal microvascular ECs are regarded as a primary therapeutic target for mitigating the progression of DR. It is noteworthy that ECs display remarkable heterogeneity across various vessel types, adapting to the specific functional demands of the surrounding tissues or organs.12,14 Consequently, liver sinusoidal ECs (LSECs) are equipped with fenestrae, lack an organized basement membrane, and demonstrate a high receptor-mediated endocytic capacity. These structural and functional features of LSECs aim to facilitate efficient transfer between the bloodstream and hepatocytes.15 In recent years, there has been an increasing acknowledgment of significant functional heterogeneity displayed by ECs within vascular beds of the same organs or tissues. In the classic retinal vessel sprouting model, retinal ECs exhibit heterogeneity, where navigating tip ECs lead the sprout, while proliferating stalk ECs elongate it.16 17 The analysis of EC heterogeneity offers a comprehensive perspective to understand cell function and the underlying mechanisms in the diseased condition. However, despite the importance of ECs in the pathogenesis of DR, the heterogeneity of ECs in diabetic retina has not yet been fully elucidated.

Single-cell RNA sequencing (scRNA-Seq) represents a breakthrough by offering single-cell resolution in transcriptomic analysis.18 19 This technology presents an opportunity to investigate the internal heterogeneity of ECs in the retina at the gene expression level, thereby identifying new molecular targets for the treatment of DR. In this study, we used single-cell transcriptome data obtained from diabetic retinas and non-diabetic retinas to construct a single-cell atlas, aiming to investigate the changes in the composition of EC subclusters in DR. Our findings revealed three transcriptionally heterogeneous EC subclusters, with subcluster 0 showing a strong association with pathological angiogenesis in diabetic retina. Particularly noteworthy was the high expression of Mitochondria-localized glutamic acid-rich protein (Mgarp) in this subcluster, significantly upregulated in the diabetic group. Additionally, silencing of Mgarp effectively mitigated the angiogenic capacity of ECs. This study provides novel insights into the transcriptomic heterogeneity of retinal vascular ECs and underscores Mgarp as a potential therapeutic target for the treatment of DR.

Research design and methods

scRNA-Seq data source

The single-cell transcriptomic data used in this study were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). The specific dataset is identified by the GEO accession code GSE178121 and was generated using the GPL24247 platform.20 The samples subjected to sequencing consisted of retinal tissues obtained from two distinct groups: three non-diabetic C57BL/6 mice and three C57BL/6 mice with streptozotocin-induced diabetes. The diabetic mice, characterized by an elevated blood glucose concentration (>300 mg/dL) for a duration of 25 weeks, were sacrificed for subsequent sample collection and scRNA-Seq.

scRNA-Seq data analysis

The R package Seurat (V.4.3.0) was used for quality control and clustering, using unique molecular identifiers (UMI) to eliminate PCR duplicates. Cells expressing fewer than 200 or more than 8000 genes, as well as those with over 10% mitochondrial UMI reads, were excluded, resulting in 6398 cells for subsequent analysis. The top 2000 hypervariable genes served as the inputs for principal component (PC) analysis. The first 16 PCs were selected using the ElbowPlot function. Batch correction was performed using the “harmony” R package to mitigate batch effects.

Subsequently, the shared nearest neighbor modularity optimization-based clustering algorithm was applied to group cells into different clusters with a resolution of 1.6. Visualization of cells in a two-dimensional space was achieved using t-Distributed Stochastic Neighbor Embedding (t-SNE). The FindClusters algorithm was then employed to identify EC subclusters with a resolution of 0.25. For functional annotation, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) terms were obtained from the DAVID Bioinformatics Database (DAVID Bioinformatics Resources, https://david.ncifcrf.gov/).21 22

Cell culture and transfection

Human retinal microvascular ECs (HRMECs) were cultured in Endothelial Cell Medium (Sciencell, USA), supplemented with 10% fetal bovine serum (FBS), 20 µg/mL EC growth supplement, and 1% penicillin/streptomycin at 37°C in a humidified incubator with 5% CO2. Transfection procedures of Mgarp knockdown involved the use of small-interfering RNAs (siRNAs) obtained from Ribobio, China, and followed the manufacturer’s protocol for Lipofectamine RNAiMAX (Invitrogen, USA). A scramble (Scr) siRNA served as a negative control. The sequence of Scr siRNA was 5′-GGAGGAAGUAUAUGUGAAGAA-3′ and that of Mgarp siRNA2 was 5′-GGTGAGAAGTAGAGAGATATA-3′. For Mgarp overexpression, HRMECs were transfected with null vector pcDNA3.1 plasmid and Mgarp overexpression plasmid (pcDNA3.1-Mgarp, Ribobio, China) using Lipofectamine 3000 (Invitrogen, USA). Transfection efficiency was assessed through quantitative real-time PCRs (qRT-PCRs).

Quantitative real-time PCRs (qRT-PCRs)

Total RNAs were extracted from HRMECs using the FastPure Cell/Tissue Total RNA Isolation Kit (Vazyme, China). cDNA synthesis was carried out using 1 µg of RNA and the HiScript III RT SuperMix (Vazyme, China) by one cycle of 15 min at 37°C and 5 s at 85°C, according to the manufacturer’s instructions. qRT-PCRs were performed with the ChamQ Universal SYBR qPCR Master Mix (Vazyme, China). The samples were held at 95 °C for 5 min followed by 40 cycles of 10 s at 95°C and 30 s at 60°C. Relative mRNA expression of target gene was calculated using the relative standard curve method (2−ΔΔCt) and normalized to β-actin. The forward primer sequence of Mgarp is 5′- GCTCGGAAAGGACGCATCTC-3′ and the reverse primer sequence is 5′- TGACTGTGACGCCTACAACC-3′. The forward and reverse primer sequences of β-actin were 5′-CATGTACGTTGCTATCCAGGC-3′ and 5′- CTCCTTAATGTCACGCACGAT-3′, respectively.

Cell viability assay

Cell viability was determined using a Cell Counting Kit-8 (CCK-8, Beyotime, China). After transfection with Scr siRNA, Mgarp siRNAs, vector plasmid pcDNA3.1, pcDNA3.1-Mgarp or being left untreated for 24 hours, cells were seeded in a 96-well plate at 5000 cells per well and cultured for an additional 24 hours. Subsequently, 10 µL of CCK-8 solution was added to each well and incubated at 37°C for 2 hours. Cell viability was determined by measuring the absorbance at 450 nm using a spectrophotometer (Tecan, China). The wells containing culture medium without cells served as the blanks for baseline measurement.

EdU assay

Cell proliferation was measured using a BeyoClick EdU Cell Proliferation Kit (Beyotime, China), following the manufacturer’s protocol. Briefly, both control and transfected cells were seeded in a 24-well plate and incubated with 10 µM of EdU at 37°C for 2 hours, then fixed with 4% paraformaldehyde and permeated with 0.3% Triton X-100. Subsequently, the cells were stained by Azide Alexa Fluor 555 reaction solution for 30 min and Hoechst 33 342 for 10 min. Images were captured using a fluorescent microscope (Olympus, Japan). Cell proliferation was determined by calculating the percentages of EdU-positive cells.

Transwell assay

Transwell assay was conducted to examine the migratory capacity of HRMECs. The cells transfected with siRNAs, plasmids or left untreated were suspended in 500 µL of serum-free medium and seeded in the upper chamber of each insert with 8 µm pore size (Corning, America). The lower chamber was filled with 500 µL of culture medium containing 10% FBS. After incubation at 37°C for 16 hours, the cells on the upper surface of the insert were removed with a cotton swab, while the cells on the bottom surface were fixed with 4% paraformaldehyde for 10 min and stained with 0.2% crystal violet for 30 min. Images were captured using an inverted light microscope (Olympus, Japan).

Tube formation assay

Tube formation assays were conducted to assess tube formation ability of HRMECs. Briefly, each well of a precooled 24-well plate was coated with 50 µL of growth-factor-reduced Matrigel (BD Biosciences, America), which was allowed to polymerize at 37 °C for 30 min. The non-transfected and transfected cells were then seeded onto the Matrigel and incubated in the full culture medium at 37°C for 4–6 hours. Subsequently, the wells were imaged using an inverted light microscope (Olympus, Japan). The Angiogenesis Analyzer plugin in ImageJ was employed to quantify the formation of tube-like structures.

Statistical analysis

Statistical analysis was carried out using GraphPad Prism software (GraphPad Software, USA, V.9.4.1). The presented data represent mean±SEM with each “n” value corresponding to an independent experiment. Statistical significance was assessed using either a two-tailed Student’s t-test or one-way analysis of variance. P<0.05 was considered statistically significant.

Results

A single-cell atlas of non-diabetic retinas and diabetic retinas

The sequencing data from both the non-diabetic retinas and diabetic retinas were combined for subsequent analysis. After passing through quality control filters, 3290 cells from non-diabetic retinas and 3108 cells from diabetic retinas were obtained, respectively. These cells were then grouped into 30 transcriptionally distinct clusters, as depicted in figure 1A. By using established cellular markers, 11 cell types were identified (figure 1A), including Müller glial cells (Glul, Pax6), retinal ganglion cells (RGCs) (Prkca), photoreceptor cells (Nrl), rod bipolar cells (Grm6), cone bipolar cells (Lhx4), ECs (Pecam1), pericytes (Pdgfrb), amacrine cells (Slc6a9, Pax6), astrocytes (Pax6, Aqp4), RPE (Rpe65) and microglial cells (Csf1r) (figure 1B). The proportions of each cell types were calculated and compared between non-diabetic retinas and diabetic retinas (figure 1C). The notable decrease in the proportion of photoreceptor cells was attributed to the removal of a significant portion of photoreceptor cells during sample processing stage.

Figure 1. A single-cell atlas of non-diabetic retinas and diabetic retinas. (A) t-Distributed Stochastic Neighbor Embedding (t-SNE) plots display 30 original clusters and the identification of different cell types. Each color represents a specific cluster or cell type, including Müller glial cells (Müller), retinal ganglion cells (RGCs), photoreceptor cells (PCs), bipolar cells (BCs), endothelial cells (ECs), pericytes (Peri), amacrine cells (ACs), astrocytes (Ast), retinal pigment endothelial cells (RPE), and microglial cells (MCs). (B) Violin plots illustrate the expression of selected cell-type-specific markers used for validating each cluster. (C) The proportions of different retinal cell types in non-diabetic and diabetic samples are depicted. (D) Feature plot shows the number of differentially expressed genes (DEGs) in different cell types.

Figure 1

Open in a new tab

Moreover, we conducted an analysis of differentially expressed genes (DEGs) across various retinal cells in non-diabetic and diabetic mice. The extent of gene expression variation in each cell type was indicated by varying shades of red, representing the number of DEGs. The results demonstrated that gene profiles of RGCs, astrocytes, microglial cells and ECs underwent the most significant alterations in diabetic retinas (figure 1D).

DEGs in ECs between diabetic retinas and non-diabetic retinas

Through the analysis of DEGs in each retinal cell type, it becomes evident that ECs are significantly impacted in the context of DR compared with all other cell types in the retina. The activation and injury of ECs induced by hyperglycemia play a crucial role in the progression of diabetic vascular complications. Therefore, we specifically examined gene expression changes in ECs.

The distribution of non-diabetic and diabetic samples was compared in the t-SNE plots (figure 2A), and DEGs between non-diabetic and diabetic samples were depicted in the volcano plot. Significance was defined as p<0.05 and |log2(FC)|>1. The volcano plot revealed that the top 10 significantly upregulated genes in ECs were Cirbp, Igfbp3, Pcp2, Apoe, Mgarp, Hist1h1c, Mt1, Guca1a, Mir124-2hg, and Lcn2 (figure 2B). Notably, Cirbp, previously reported as the most globally altered gene across all cell types in DR,20 was also found to be the most upregulated in ECs. Cirbp is a cold-inducible RNA-binding protein triggered by stressors such as cold and hypoxia, functioning to protect cells from damage.23

Figure 2. Differentially expressed genes in endothelial cells between diabetic retinas and non-diabetic retinas. (A) t-Distributed Stochastic Neighbor Embedding (t-SNE) plot illustrates the single-cell mapping of endothelial cells in both non-diabetic retinas and diabetic retinas. (B) The volcano plot displays differentially expressed genes in endothelial cells between non-diabetic retinas and diabetic retinas. FC, fold change. (C) Dot plot depicts Gene Ontology-biological process (GO-BP) enrichment of the top 1000 differentially expressed genes in endothelial cells between non-diabetic retinas and diabetic retinas. (D) Dot plot shows Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment of the top 1000 differentially expressed genes in endothelial cells between non-diabetic and diabetic retinas.

Figure 2

Open in a new tab

The GO-biological process (GO-BP) analysis of the top 1000 DEGs revealed significant enrichment in terms associated with vascular dysfunction, including “Angiogenesis,” “Positive regulation of cell migration,” and “Positive regulation of cell proliferation.” Other enriched terms such as “Positive regulation of apoptotic process,” “Regulation of cell cycle,” “Response to hypoxia,” and “Cell adhesion,” were also found to be related to pathological processes in DR (figure 2C). KEGG analysis further verified that the enriched signaling pathways were associated with vascular dysfunction such as “MAPK signaling pathway,” “PI3K-Akt signaling pathway,” “HIF-1 signaling pathway,” “Notch signaling pathway,” “TGF-beta signaling pathway,” and “VEGF signaling pathway” (figure 2D).

EC subcluster 0 is potentially associated with pathological angiogenesis in diabetic retina

To enhance the understanding of the transcriptomic heterogeneity within ECs during pathological angiogenesis, we conducted Uniform Manifold Approximation and Projection (UMAP) reduction analysis. We identified three EC subclusters, with subcluster 0 having the most substantial population (figure 3A). Subsequently, we compared the proportions of ECs between the non-diabetic retinas and diabetic retinas within each subcluster. Remarkably, a significant increase in cell number was observed in the diabetic group of subcluster 0, approximately 1.6 times higher than that in the non-diabetic group (figure 3B).

Figure 3. Endothelial cell subcluster 0 is potentially associated with pathological angiogenesis in diabetic retina. (A) Uniform Manifold Approximation and Projection (UMAP) plots show three different subclusters of endothelial cells in non-diabetic retinas and diabetic retinas. (B) Bar graph displays the proportions of endothelial cells from non-diabetic retinas and diabetic retinas in different subclusters. (C) Heatmap shows the expression patterns of the top five highly expressed genes in each subcluster. (D) Expression levels of the top two highly expressed genes in subcluster 0, 1, and 2 are shown in violin plots.

Figure 3

Open in a new tab

The top two most abundant genes in subcluster 0 were Clu and Mgarp, both related to the regulation of cell proliferation, cell apoptosis, and hypoxia-induced mitochondrial movement. Subcluster 1 exhibited high expression of genes associated with cell response to environmental stress and inflammatory regulation, including Ier3, Jun, and Hspa1b, implying that subcluster 1 is mainly involved in inflammatory response. Subcluster 2 was characterized by Rgs5 and Vtn. Rgs5 is recognized as a proapoptotic/antiproliferative protein involved in hemodynamic regulation, pericyte recruitment, as well as vascular morphology and remodeling, while Vtn participates in the regulation of extracellular matrix formation and cell adhesion (figure 3C and D). Notably, subcluster 2 highly expressed pericyte-related markers such as Rgs5 and Myl9, suggesting that this subcluster is potentially responsible for EC–pericyte crosstalk and interaction, thereby playing a collaborative role in the maintenance of normal vascular function alongside pericytes. Based on the results, we can infer that subcluster 0 is potentially linked to pathological angiogenesis in diabetic retina.

Mgarp is dominantly expressed in subcluster 0 and demonstrates a noteworthy upregulation in diabetic retina

To uncover the potential functions of subcluster 0, we conducted a GO-BP enrichment analysis using the top 200 highly expressed genes within this subcluster (figure 4A). The analysis revealed the enrichments in genes associated with mitochondrial function, including “ATP synthesis coupled proton transport,” “Aerobic respiration,” “Protein import into mitochondrial matrix,” and “Cellular response to hypoxia.” Moreover, significant enrichments were found in the processes related to “Cell cycle,” “Regulation of growth,” and “Angiogenesis,” suggesting a correlation between subcluster 0 and cellular proliferation in the context of angiogenesis. These findings imply that ECs within subcluster 0 play an important role in energy metabolism and contribute to pathological angiogenesis in the retina.

Figure 4. Mgarp is dominantly expressed in subcluster 0 and demonstrates a noteworthy upregulation in diabetic retina. (A) Dot plot shows Gene Ontology-biological process (GO-BP) enrichment of top 200 highly expressed genes in subcluster 0. (B) Volcano plot shows differentially expressed genes between diabetic retinas and non-diabetic retinas in subcluster 0. (C,D) Expression level of Mgarp in different subclusters from diabetic retinas and non-diabetic retinas are shown in Violin plot (C) and FeaturePlot (D). DR, diabetic retinopathy; FC, fold change; UMAP, Uniform Manifold Approximation and Projection.

Figure 4

Open in a new tab

To explore transcriptional changes in diabetic retina, we compared gene expression differences between the diabetic retinas and non-diabetic retinas within subcluster 0 by using the significance threshold of p<0.05 and |log2(FC)|>1 (figure 4B). A total of 82 genes were differentially expressed, with 57 upregulated and 25 downregulated. Notably, Mgarp, identified as one of the most abundantly expressed genes in subcluster 0, exhibited a significant upregulation in diabetic retinas compared with non-diabetic retinas. This observation was further supported by Violin plot and FeaturePlot analyses, highlighting that Mgarp expression is predominantly concentrated in diabetic retinas within subcluster 0 (figure 4C,D).

Mgarp silencing exerts antiangiogenic effects in ECs

Based on the abovementioned analysis, Mgarp is shown as a potential pivotal factor in the process of pathological angiogenesis. siRNAs were used to investigate the impact of Mgarp silencing on endothelial angiogenic effects in vitro. HRMECs were transfected with either Scr siRNA, three different Mgarp-specific siRNAs, or left untreated for 24 hours. Among the three Mgarp-specific siRNAs tested, siRNA2 demonstrated the highest knockdown efficiency (figure 5A). Consequently, siRNA2 was used in subsequent cellular experiments. As illustrated in figure 5A, the transfection of Mgarp siRNA2 substantially decreased the expression level of Mgarp gene, leading to a reduced cell viability compared with non-transfected HRMECs and Scr siRNA-transfected HRMECs (figure 5B). Moreover, a lower number of EdU-positive cells were observed in the Mgarp-silenced group, revealing a significant suppression of the proliferation ability of HRMECs (figure 5C). Transwell assays revealed that HRMECs transfected with Mgarp siRNA2 displayed diminished migratory capacity compared with the cells in the non-transfected group and Scr siRNA-transfected group (figure 5D). Furthermore, tube formation assays demonstrated that Mgarp silencing impaired the ability of HRMECs to form tube-like structures on Matrigel (figure 5E). By contrast, overexpression of Mgarp enhanced cell viability, proliferation ability, migration ability, and tube formation ability of HRMECs in vitro (online supplemental figure S1). Taken together, these findings suggest that Mgarp is involved in the regulation of endothelial angiogenic effects in vitro.

Figure 5. Mgarp silencing exerts antiangiogenic effects in endothelial cells. (A) Human retinal microvascular endothelial cells (HRMECs) were transfected with scramble (Scr) siRNA, Mgarp siRNA, or left untreated (Ctrl) for 24 hours. The levels of Mgarp expression were detected by qRT-PCRs. n=4, *p<0.05. (B) Cell viability was determined using CCK-8 assays. n=4, *p<0.05. (C) Cell proliferation was detected using EdU staining assays. n=4, *p<0.05, scale bar: 20 µm. (D) Transwell assays were conducted to detect cell migration ability. n=4, *p<0.05, scale bar: 50 µm. (E) Tube formation assays were conducted to determine the tube formation ability of HRMECs. n=4, *p<0.05, scale bar: 100 µm.

Figure 5

Open in a new tab

Discussion

Cellular heterogeneity is a pervasive characteristic found in various tissues across organisms.24 25 Even within the same cell type, difference may arise in surface molecules, structures, and functions, affected by intrinsic factors like genetic modification and extrinsic factors such as surrounding microenvironment.26,28 scRNA-Seq emerges as a powerful tool enabling the acquisition of comprehensive transcriptomic expression data at the individual cell level. While widely applied in fields like oncology and neuroscience,29 30 the utilization of scRNA-Seq technology in ophthalmology remains in its early stages.

In recent years, efforts have concentrated on delineating distinct cell types in the choroid and retina at the single-cell level, constructing single-cell atlases of the retina, investigating disease pathogenesis, and identifying potential therapeutic targets. For instance, Van Hove et al explored the heterogeneity of macroglial cells and immune cells using scRNA-Seq on retinal tissues from wild-type and Akimba mice, categorizing macroglial cells into four subgroups and identifying three immune cell subgroups.31 Lv et al applied single-cell sequencing to identify an activated microglia group in Streptozotocin (STZ)-induced diabetic mice.32 Additionally, a previous study used scRNA-Seq for identifying specific genes contributing to DR pathogenesis. Notably, the upregulation of RLBP1 expression in Müller glial cells emerged as a potential target for alleviating neurovascular degeneration.33

It is well known that ECs play a crucial role in maintaining vascular homeostasis and dysfunction of retinal ECs is a key contributor to the progression of DR. Despite the pivotal role of ECs in the pathogenesis of DR, the heterogeneity of ECs in the diabetic retina remains incompletely understood. Therefore, we used single-cell transcriptomic data from the GEO database to delve into the heterogeneity of ECs in diabetic retinas. Through UMAP reduction analysis, we initially categorized ECs into three distinct subclusters. Notably, gene expression profiles of EC subcluster 2 exhibited significant differences when compared with ECs in subclusters 0 and 1, as visualized in the UMAP plot. This subcluster exhibited high expression levels of pericyte markers, collaborating with pericytes to maintain normal vascular function. In EC subcluster 1, there is a pronounced overexpression of genes linked to cellular response to environmental stresses and inflammatory regulation, suggesting active participation of EC subcluster 1 in inflammatory processes, a crucial pathological feature of DR.5

EC subcluster 0, being the largest among the three subclusters, exhibited a significantly higher cell count in diabetic group compared with non-diabetic group. This observation sparked the speculation that subcluster 0 may be intricately linked to pathological angiogenesis in diabetic retinas. Subsequent GO-BP enrichment analysis further supported this notion, indicating that EC subcluster 0 is notably active in energy metabolism and cell proliferation within the retina. Motivated by these findings, we conducted a differential gene expression analysis of EC subcluster 0 and uncovered a surprising result that Mgarp, predominantly expressed in EC subcluster 0, was significantly upregulated in diabetic retinas. This suggests a potential association between Mgarp and EC function in diabetic retinas.

Mgarp, initially identified through a large-scale ovary screen,34 is a mitochondria-localized, glutamic acid-rich protein found in steroidogenic organs such as testes, ovaries, and adrenal glands.35 36 Except for involvement in steroid synthesis, Mgarp also plays a role in regulating mitochondria distribution and motility in neurons.37 Notably, Mgarp exhibits high expression in mouse retinas, particularly in mitochondria-rich layers.38 Positioned in the inner mitochondrial membrane and cristae, where vital energy conversion and ATP production occur,39 downregulation of Mgarp in mouse adrenal Y1 cells is tightly associated with reduced cell proliferation, causing mitochondrial swelling and reduced membrane potential.35 Based on the abovementioned studies, we know that Mgarp plays a crucial role in maintaining normal mitochondrial function, aligning with the enrichment of mitochondrial function and cell proliferation terms observed in EC subcluster 0. The upregulation of Mgarp in diabetic retinas within EC subcluster 0 suggests a close association between Mgarp and pathological angiogenesis in diabetic retinas. Functional studies reveal that Mgarp silencing significantly inhibits the proliferation, migration, and tube-formation abilities of HRMECs. Mgarp may affect endothelial angiogenic effects through mitochondrial function, prompting the need for further study to unveil the underlying mechanism.

In conclusion, our study has identified an EC subcluster marked by heightened Mgarp expression, intricately linked to pathological angiogenesis in diabetic retinas. Within this subcluster, Mgarp exhibits a substantial upregulation in diabetic retinas, and Mgarp silencing exerts antiangiogenic effects in vitro. This study offers a comprehensive single-cell atlas illustrating EC heterogeneity in DR, presenting new prospects for antiangiogenic treatments by targeting specific EC subpopulation.

Supplementary material

online supplemental file 1
bmjdrc-12-4-s001.pdf (318.4KB, pdf)
DOI: 10.1136/bmjdrc-2024-004189

Footnotes

Funding: This study was funded by the grants from National Natural Science Foundation of China (no. 81770945 to BY).

Provenance and peer review: Not commissioned; externally peer reviewed.

Patient consent for publication: Not applicable.

Ethics approval: Not applicable.

Data availability statement

Data are available in a public, open access repository. Data are available on reasonable request.

References

  • 1.Miloudi K, Binet F, Wilson A, et al. Truncated netrin-1 contributes to pathological vascular permeability in diabetic retinopathy. J Clin Invest. 2016;126:3006–22. doi: 10.1172/JCI84767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Stitt AW, Curtis TM, Chen M, et al. The progress in understanding and treatment of diabetic retinopathy. Prog Retin Eye Res. 2016;51:156–86. doi: 10.1016/j.preteyeres.2015.08.001. [DOI] [PubMed] [Google Scholar]
  • 3.Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med. 2012;366:1227–39. doi: 10.1056/NEJMra1005073. [DOI] [PubMed] [Google Scholar]
  • 4.Cecilia O-M, José Alberto C-G, José N-P, et al. Oxidative stress as the main target in diabetic retinopathy pathophysiology. J Diabetes Res. 2019;2019:8562408. doi: 10.1155/2019/8562408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Robinson R, Srinivasan M, Shanmugam A, et al. Interleukin-6 trans-signaling inhibition prevents oxidative stress in a mouse model of early diabetic retinopathy. Redox Biol. 2020;34:101574. doi: 10.1016/j.redox.2020.101574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Chen J, Smith LEH. Retinopathy of prematurity. Angiogenesis. 2007;10:133–40. doi: 10.1007/s10456-007-9066-0. [DOI] [PubMed] [Google Scholar]
  • 7.Sapieha P, Hamel D, Shao Z, et al. Proliferative retinopathies: angiogenesis that blinds. Int J Biochem Cell Biol. 2010;42:5–12. doi: 10.1016/j.biocel.2009.10.006. [DOI] [PubMed] [Google Scholar]
  • 8.Sapieha P, Joyal J-S, Rivera JC, et al. Retinopathy of prematurity: understanding ischemic retinal vasculopathies at an extreme of life. J Clin Invest. 2010;120:3022–32. doi: 10.1172/JCI42142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Tang DG, Conti CJ. Endothelial cell development, vasculogenesis, angiogenesis, and tumor neovascularization: an update. Semin Thromb Hemost. 2004;30:109–17. doi: 10.1055/s-2004-822975. [DOI] [PubMed] [Google Scholar]
  • 10.Basagiannis D, Zografou S, Goula E, et al. Chemical inhibitors of dynamin exert differential effects in VEGF signaling. Cells. 2021;10:997. doi: 10.3390/cells10050997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Fan C, Ma Q, Xu M, et al. Ginsenoside Rb1 attenuates high glucose-induced oxidative injury via the NAD-PARP-SIRT axis in rat retinal capillary endothelial cells. Int J Mol Sci. 2019;20:4936. doi: 10.3390/ijms20194936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Aird WC. Phenotypic heterogeneity of the endothelium: II. Representative vascular beds. Circ Res. 2007;100:174–90. doi: 10.1161/01.RES.0000255690.03436.ae. [DOI] [PubMed] [Google Scholar]
  • 13.Coppiello G, Collantes M, Sirerol-Piquer MS, et al. Meox2/Tcf15 heterodimers program the heart capillary endothelium for cardiac fatty acid uptake. Circulation. 2015;131:815–26. doi: 10.1161/CIRCULATIONAHA.114.013721. [DOI] [PubMed] [Google Scholar]
  • 14.Géraud C, Koch P-S, Zierow J, et al. GATA4-dependent organ-specific endothelial differentiation controls liver development and embryonic hematopoiesis. J Clin Invest. 2017;127:1099–114. doi: 10.1172/JCI90086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Poisson J, Lemoinne S, Boulanger C, et al. Liver sinusoidal endothelial cells: physiology and role in liver diseases. J Hepatol. 2017;66:212–27. doi: 10.1016/j.jhep.2016.07.009. [DOI] [PubMed] [Google Scholar]
  • 16.Potente M, Gerhardt H, Carmeliet P. Basic and therapeutic aspects of angiogenesis. Cell. 2011;146:873–87. doi: 10.1016/j.cell.2011.08.039. [DOI] [PubMed] [Google Scholar]
  • 17.Potente M, Mäkinen T. Vascular heterogeneity and specialization in development and disease. Nat Rev Mol Cell Biol. 2017;18:477–94. doi: 10.1038/nrm.2017.36. [DOI] [PubMed] [Google Scholar]
  • 18.Jaitin DA, Kenigsberg E, Keren-Shaul H, et al. Massively parallel single-cell RNA-seq for marker-free decomposition of tissues into cell types. Science. 2014;343:776–9. doi: 10.1126/science.1247651. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Soneson C, Robinson MD. Bias, robustness and scalability in single-cell differential expression analysis. Nat Methods. 2018;15:255–61. doi: 10.1038/nmeth.4612. [DOI] [PubMed] [Google Scholar]
  • 20.Sun L, Wang R, Hu G, et al. Single cell RNA sequencing (scRNA-Seq) deciphering pathological alterations in streptozotocin-induced diabetic retinas. Exp Eye Res. 2021;210:108718. doi: 10.1016/j.exer.2021.108718. [DOI] [PubMed] [Google Scholar]
  • 21.Sherman BT, Hao M, Qiu J, et al. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update) Nucleic Acids Res. 2022;50:W216–21. doi: 10.1093/nar/gkac194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4:44–57. doi: 10.1038/nprot.2008.211. [DOI] [PubMed] [Google Scholar]
  • 23.Lleonart ME. A new generation of proto-oncogenes: cold-inducible RNA binding proteins. Biochim Biophys Acta. 2010;1805:43–52. doi: 10.1016/j.bbcan.2009.11.001. [DOI] [PubMed] [Google Scholar]
  • 24.Schelker M, Feau S, Du J, et al. Estimation of immune cell content in tumour tissue using single-cell RNA-Seq data. Nat Commun. 2017;8:2032. doi: 10.1038/s41467-017-02289-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Li Y, Shan Y, Desai RV, et al. Noise-driven cellular heterogeneity in circadian periodicity. Proc Natl Acad Sci USA. 2020;117:10350–6. doi: 10.1073/pnas.1922388117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Jin X, Kim LJY, Wu Q, et al. Targeting glioma stem cells through combined BMI1 and EZH2 inhibition. Nat Med. 2017;23:1352–61. doi: 10.1038/nm.4415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Raser JM, O’Shea EK. Noise in gene expression: origins, consequences, and control. Science. 2005;309:2010–3. doi: 10.1126/science.1105891. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Raj A, van Oudenaarden A. Nature, nurture, or chance: stochastic gene expression and its consequences. Cell. 2008;135:216–26. doi: 10.1016/j.cell.2008.09.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Zhang Y, Wang D, Peng M, et al. Single-cell RNA sequencing in cancer research. J Exp Clin Cancer Res. 2021;40:81. doi: 10.1186/s13046-021-01874-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Zeng Z, Miao N, Sun T. Revealing cellular and molecular complexity of the central nervous system using single cell sequencing. Stem Cell Res Ther. 2018;9:234. doi: 10.1186/s13287-018-0985-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Van Hove I, De Groef L, Boeckx B, et al. Single-cell transcriptome analysis of the Akimba mouse retina reveals cell-type-specific insights into the pathobiology of diabetic retinopathy. Diabetologia. 2020;63:2235–48. doi: 10.1007/s00125-020-05218-0. [DOI] [PubMed] [Google Scholar]
  • 32.Lv K, Ying H, Hu G, et al. Integrated multi-omics reveals the activated retinal microglia with intracellular metabolic reprogramming contributes to inflammation in STZ-induced early diabetic retinopathy. Front Immunol. 2022;13:942768. doi: 10.3389/fimmu.2022.942768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Niu T, Fang J, Shi X, et al. Pathogenesis study based on high-throughput single-cell sequencing analysis reveals novel transcriptional landscape and heterogeneity of retinal cells in type 2 diabetic mice. Diabetes. 2021;70:1185–97. doi: 10.2337/db20-0839. [DOI] [PubMed] [Google Scholar]
  • 34.Kinouchi R, Kinouchi T, Hamamoto T, et al. Distribution of CESP-1 protein in the corneal endothelium and other tissues. Invest Ophthalmol Vis Sci. 2006;47:1397–403. doi: 10.1167/iovs.05-0602. [DOI] [PubMed] [Google Scholar]
  • 35.Matsumoto T, Minegishi K, Ishimoto H, et al. Expression of ovary-specific acidic protein in steroidogenic tissues: a possible role in steroidogenesis. Endocrinology. 2009;150:3353–9. doi: 10.1210/en.2008-1584. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Zhou M, Wang Y, Qi S, et al. The expression of a mitochondria-localized glutamic acid-rich protein (MGARP/OSAP) is under the regulation of the HPG axis. Endocrinology. 2011;152:2311–20. doi: 10.1210/en.2011-0050. [DOI] [PubMed] [Google Scholar]
  • 37.Li H, Jiang Y, Liu M, et al. DNA methylation-mediated inhibition of MGARP is involved in impaired progeny testosterone synthesis in mice exposed to DBP in utero. Environ Toxicol. 2023;38:914–25. doi: 10.1002/tox.23734. [DOI] [PubMed] [Google Scholar]
  • 38.Qi S, Wang Y, Zhou M, et al. A mitochondria-localized glutamic acid-rich protein (MGARP/OSAP) is highly expressed in retina that exhibits a large area of intrinsic disorder. Mol Biol Rep. 2011;38:2869–77. doi: 10.1007/s11033-010-9948-x. [DOI] [PubMed] [Google Scholar]
  • 39.Zhang S. MGARP is ultrastructurally located in the inner faces of mitochondrial membranes. Biochem Biophys Res Commun. 2019;516:138–43. doi: 10.1016/j.bbrc.2019.06.028. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

online supplemental file 1
bmjdrc-12-4-s001.pdf (318.4KB, pdf)
DOI: 10.1136/bmjdrc-2024-004189

Data Availability Statement

Data are available in a public, open access repository. Data are available on reasonable request.

Articles from BMJ Open Diabetes Research & Care are provided here courtesy of BMJ Publishing Group

RESOURCES