Despite advances in treatment, atherosclerotic cardiovascular disease remains the leading cause of death in patients with diabetes. Even when risk factors are mitigated, the disease progresses, and thus, newer targets need to be identified that directly inhibit the underlying pathobiology of atherosclerosis in diabetes. A single-cell sequencing approach was used to distinguish the proatherogenic transcriptional profile in aortic cells in diabetes using a streptozotocin-induced diabetic Apoe−/− mouse model. Human carotid endarterectomy specimens from individuals with and without diabetes were also evaluated via immunohistochemical analysis. Further mechanistic studies were performed in human aortic endothelial cells (HAECs) and human THP-1–derived macrophages. We then performed a preclinical study using an activator protein-1 (AP-1) inhibitor in a diabetic Apoe−/− mouse model. Single-cell RNA sequencing analysis identified the AP-1 complex as a novel target in diabetes-associated atherosclerosis. AP-1 levels were elevated in carotid endarterectomy specimens from individuals with diabetes compared with those without diabetes. AP-1 was validated as a mechanosensitive transcription factor via immunofluorescence staining for regional heterogeneity of endothelial cells of the aortic region exposed to turbulent blood flow and by performing microfluidics experiments in HAECs. AP-1 inhibition with T-5224 blunted endothelial cell activation as assessed by a monocyte adhesion assay and expression of genes relevant to endothelial function. Furthermore, AP-1 inhibition attenuated foam cell formation. Critically, treatment with T-5224 attenuated atherosclerosis development in diabetic Apoe−/− mice. This study has identified the AP-1 complex as a novel target, the inhibition of which treats the underlying pathobiology of atherosclerosis in diabetes.

Article Highlights
  • The cell-specific transcriptional profile of vascular cells in diabetes-associated atherosclerosis is not known.

  • Single-cell RNA sequencing analysis not only defined the transcriptional profile of vascular cells in diabetes but also identified the activator protein-1 (AP-1) complex as one of the important transcription factor complexes in flow-mediated endothelial cell activation and foam cell formation.

  • Although previous studies have implicated AP-1 in atherosclerosis, our study has defined the role of AP-1 as a central regulator of a gene expression program linked to endothelial dysfunction and foam cell formation in diabetes.

  • Importantly, inhibition of AP-1 transcription activity with T-5224 attenuated atherosclerosis development in vivo in diabetes.

Atherosclerosis, as manifested clinically by myocardial infarction, stroke, and peripheral vascular disease, is a major contributor to cardiovascular disease, the leading cause of death in patients with diabetes (1). Currently, the primary approach for the management of atherosclerotic cardiovascular diseases (CVDs) targets systemically LDL cholesterol. However, the residual atherosclerotic burden persists as reflected by the elevated atherosclerotic CVD–related mortality worldwide, with rates remaining higher in patients with diabetes (2–4). Thus, newer approaches are needed to directly inhibit the underlying pathobiology of atherosclerosis.

Atherosclerotic plaque development is a progressive process involving endothelial cell activation, monocyte adhesion, and foam cell formation (5–8). These key pathways are prominent in diabetes, where there are associated regional hemodynamic abnormalities and accelerated atherosclerosis (4). These cellular events occur primarily at the regions of blood vessels exposed to turbulent blood flow (TBF) and low shear stress (LSS), such as vascular bends and bifurcations (9). Atherosclerosis preferentially develops at vascular regions exposed to TBF with LSS (10–12).

Macrophages in the atherosclerotic plaque are characterized by impaired cholesterol homeostasis that leads to foam cell formation (13,14). Gene expression of scavenger receptors and cholesterol transporters that play an important role in cholesterol homeostasis can be regulated by multiple factors, including transcription factors (TFs) (15). Therefore, targeting the upstream TFs involved in major cholesterol homeostatic pathways could be an effective novel therapeutic strategy to combat atherosclerotic CVD.

Recent single-cell sequencing studies showed endothelial reprogramming by TBF and confirmed upregulation of many flow-sensitive TFs, including activator protein-1 (AP-1) (16,17). However, gene targets and the mechanisms underlying AP-1 activation in vascular cells are not known. T-5224 is a highly specific bioavailable drug that blocks AP-1 DNA binding activity, with this drug having been shown to prevent and resolve rheumatoid arthritis in mice (18,19). This drug has advanced to a phase II clinical trial in rheumatoid arthritis with an excellent safety profile (20); however, its role in CVD specifically in the context of diabetes is unknown. In addition, endothelial cell heterogeneity has been reported in a study where endothelial cells collected from heart and aorta were subjected to single-cell RNA sequencing (scRNA-seq) in LDLR−/− mice fed a diabetogenic high-fat diet with cholesterol (21). Furthermore, studies using cytometry by time-of-flight or scRNA-seq approaches performed in murine models of atherosclerosis suggested the existence of multiple macrophage populations, including lipid-loaded foamy macrophages within the intima (22–24). However, cell-specific transcriptional changes in the whole diabetic aorta are not known.

We first used scRNA-seq to define the transcriptional profile of vascular cells in diabetes-associated atherosclerosis. We then identified the AP-1 complex as one of the important TF complexes involved in both flow-mediated endothelial cell activation and foam cell formation. Although, previous studies have implicated AP-1 as a mechanosensitive TF in endothelial cells and in atherosclerosis (17,25–27), the role of AP-1 in endothelial dysfunction and foam cell formation in the context of diabetes is not known. Thus, we pursued AP-1 as a novel target in diabetes-associated atherosclerosis.

Ethics

All animal experiments were approved by the Alfred Medical Research Education Precinct Animal Services ethics committee (E/2039/2020/B and P8458 V1.3), and studies were conducted according to Australian National Health and Medical Research Council guidelines in line with international standards. Human studies were approved by the Alfred Human Research Ethics Committee (authorization no. 24/07). Individuals were recruited at the Alfred Hospital, Melbourne, Victoria, Australia. Written informed consent for proper and ethical use of human tissue was obtained from all participants.

Mice

To induce diabetes, 6-week-old male apolipoprotein E–deficient (Apoe−/−) mice were rendered diabetic by daily intraperitoneal injections of streptozotocin (STZ) (55 mg/kg body weight/day) or citrate vehicle (control) for 5 consecutive days and followed for 10 weeks. Mice were fed a standard chow diet (SF00-100; Specialty Feeds, Glen Forrest, WA, Australia). After 10 weeks of diabetes, mice were sacrificed using CO2 inhalation, and aortas were isolated for further analysis. Metabolic data are listed in Supplementary Table 1.

Aorta Isolation and Processing

Aortas were stripped of adventitial fat under a dissecting microscope and either fixed in 10% neutral buffered formalin for en face analysis or digested for scRNA-seq. Assessment of plaque area was undertaken using en face analysis, after staining with Sudan IV-Herxheimer solution (BDH, Poole, U.K.) as previously described (28).

To obtain a single-cell suspension, aortas from four control and four diabetic mice were collected in DMEM containing 10% FCS on ice. Aortas were then individually cut into small pieces and digested for 1 h at 37°C on a thermomixer in PBS containing Liberase (0.1 mg/mL; Sigma-Aldrich), CaCl2 (1.4 mmol/L), hyaluronidase (60 units/mL; Sigma-Aldrich), and DNase I (120 units/mL; QIAGEN). The aortic suspension after digestion was filtered through a 40-µm cell strainer and washed with FCS containing DMEM. Cells from the experimental groups (control and diabetes) were pooled and resuspended in the FCS-containing DMEM and incubated at 37°C with 5% CO2 in a cell culture incubator for 30 min. Aortic cells were resuspended in FACS buffer containing 1× Hanks’ balanced salt solution, 0.05 mmol/L EDTA, and 5 g BSA and incubated with fluorochrome-coupled antibodies (7AAD-PE/Cy5 and calcein-FITC) to sort viable and metabolically active cells using FACSAria Fusion (nozzle size 100 µm; BD Biosciences) (29).

scRNA-Seq

Viable aortic cells were prepared for single-cell transcriptomic analysis using the droplet-based Chromium Controller platform and Chromium Single Cell 3′ Reagent Kits v2 (10x Genomics, Pleasanton, CA) according to the manufacturer’s instructions. In brief, to target recovery of 8,000 cells per sample, individual master mix suspensions containing approximately 15,000 cells each were prepared and loaded into sample wells of a chromium chip prior to capture in droplets or Gel bead in EMulsions (GEMs; 10x Genomics). Unique functional oligos on each 10x gel bead incorporate a poly(dT) capture sequence, a unique 16-nt 10x bead barcode, and a 10-nt unique molecular identifier to enable capture and barcoding of polyadenylated mRNA transcripts from GEM-encapsulated cells. Following GEM reverse transcription incubation, barcoded full-length cDNA was amplified using 11 PCR cycles to generate sufficient material for library construction. Following enzymatic fragmentation, size selection, and adaptor ligation, 12 PCR cycles and unique sample indexes supplied from Single Index Kit T Set A (10x Genomics) were used in the construction of final libraries. Completed libraries were quantified and assessed using a Qubit fluorometer (Invitrogen, Waltham, MA) and D5000 high-sensitivity assay for TapeStation (Agilent Technologies, Santa Clara, CA). Libraries passing quality control were pooled for sequencing on the Illumina NovaSeq 6000 platform using S1 Reagent Kits (100 cycles).

scRNA-Seq Data Analysis

The single-cell analysis package Cell Ranger 3.1.0 was used to demultiplex raw data, extract cell barcodes, and generate cell feature count matrices. Specifically, sample FASTQ files were extracted from Illumina binary base call files using the cellranger mkfastq function, followed by cell barcode extraction, reference genome alignment (cellranger refdata mm10 3.0.0), and cell feature counting using cellranger count with default settings. Count matrices generated with Cell Ranger were transferred to the Partek Flow genomic analysis suite for further analysis. Low-quality cells, doublets, and potentially dead cells were removed based on the percentage of mitochondrial genes, total counts, and number of genes expressed in each cell (inclusion criteria: total counts 1,000–23,000, expressed genes 300–3,000, mitochondrial reads 0–10%). The remaining data were normalized, log-transformed, and subjected to principal component analysis to reduce dimensionality. t-Distributed stochastic neighbor embedding was used to visualize data in two dimensions, while graph-based clustering was performed to identify cells with similar transcriptomic profiling. Finally, cluster identities were assigned based on expression of canonical markers. For subclustering, cells were filtered and reclustered using graph-based default settings. The gene-specific analysis (GSA) function in Partek Flow was used to identify differential gene expression (DGE) profiles between cell populations and experimental groups, while minimum thresholds for false discovery rate (FDR) (0.05) and fold change (3×) were used to define differentially expressed genes (DEGs). Cell-cell communication was investigated using NicheNet as previously described (30).

In Vivo Preclinical Studies

Apoe−/− male mice were either administered STZ for diabetes induction or citrate vehicle for controls as stated above. Five weeks after diabetes induction, mice were randomized into four groups (control + vehicle, control + T-5224, diabetes + vehicle, diabetes + T-5224). Mice were administered daily either vehicle (polyvinylpyrrolidone [PVP]) or T-5224 dissolved in PVP (30 mg/kg body weight) by oral gavage for 5 weeks. Tissues were collected at the end of the study for further analysis.

Human Carotid Endarterectomy

Carotid tissue was removed during the carotid endarterectomy procedure, and the specimens were fixed in 4% formalin before embedding in paraffin. The specimens were then decalcified with nitric acid (10% vol/vol) prior to cutting and subsequent immunohistochemistry. Study participant data are listed in Supplementary Table 2.

Statistical Analysis

Data are presented as mean ± SEM. The statistical analysis for in vitro experiments was performed using GraphPad Prism 9.0.1 software. The Mann-Whitney test for nonparametric data was used to determine significance between two groups. One-way ANOVA with Tukey multiple comparison test and Kruskal-Wallis with Dunn multiple comparison tests were used for comparison among several groups of parametric and nonparametric data, respectively. A P < 0.05 was considered statistically significant.

Data and Resource Availability

The authors declare that all data are available with the article and the data supplement. All sequencing data sets in this article are deposited in the international public repository Gene Expression Omnibus under accession no. GSE211216.

scRNA-Seq in Diabetes-Associated Atherosclerosis

Cellular Clustering and Identification

We used a state-of-the-art scRNA-seq approach not previously studied in diabetes-associated atherosclerosis to distinguish a proatherogenic transcriptional profile specifically in aortic cells. As a well-validated example of accelerated atherosclerosis, insulin-deficient diabetes was induced with low-dose STZ in Apoe−/− mice and followed for 10 weeks to minimize the toxic effects of STZ (Fig. 1A). This is the preferred animal model for atherosclerosis studies as reported by the JDRF and National Institutes of Health Consortium on Animal Models of Diabetic Complications (31). STZ-induced diabetic Apoe−/− mice displayed significantly lower body weights and higher blood glucose and HbA1c levels compared with nondiabetic control Apoe−/− mice, representing a preferable animal model of hyperglycemia-induced atherosclerosis (Supplementary Table 1). As expected, diabetes increased atherosclerotic plaque area as assessed by Sudan IV staining of the arch of the STZ-treated Apoe−/− mice versus nondiabetic control mice (Fig. 1B). Viable cells from digested aortas of control and diabetic mice were FACS sorted (Supplementary Fig. 1) and subjected to scRNA-seq using the 10x Genomics platform and Illumina sequencing. After extraction of raw data using the 10x Cell Ranger pipeline, quality filtering of cell count matrixes yielded 3,054 cells from control and 2,518 cells from diabetic aortas for inclusion in further analysis using Partek Flow software. Gene expression data from all vascular cells were aligned and projected in a two-dimensional space through t-stochastic neighbor embedding to allow identification of cellular populations from both control and diabetic aortas (Fig. 1C). An unsupervised graph-based clustering algorithm grouped cells into cellular clusters with similar gene expression profiles from aortas of control vehicle-treated and diabetic STZ-treated mice (Fig. 1C). Classification of cell identities was achieved through comparison of well-known gene markers unique to each cluster (Fig. 1D and Supplementary Fig. 2).

Figure 1

Identification of aortic cell populations in diabetes-associated atherosclerosis. A: Schematic diagram of the experimental design for scRNA-seq in diabetes-associated atherosclerosis. B: Combined dot plot and bar graph with representative images showing percentage of Sudan IV–positive area of total lesion area in the aortic arch in control (vehicle [Veh])–treated and STZ–treated Apoe−/− mice. Data are mean ± SEM. P value was determined using two-tailed Mann-Whitney test. C: t-Distributed stochastic neighbor embedding showing all identified cellular clusters of aortas from both control and diabetic mice. D: Heat map showing cell markers differentially expressed between all detected aortic cellular clusters (FDR step up <0.05). Four aortas from Apoe−/− mice per group were included in the pool of aortic cells. *P < 0.05. F, foam; M, monocyte; Mac, macrophage.

Figure 1

Identification of aortic cell populations in diabetes-associated atherosclerosis. A: Schematic diagram of the experimental design for scRNA-seq in diabetes-associated atherosclerosis. B: Combined dot plot and bar graph with representative images showing percentage of Sudan IV–positive area of total lesion area in the aortic arch in control (vehicle [Veh])–treated and STZ–treated Apoe−/− mice. Data are mean ± SEM. P value was determined using two-tailed Mann-Whitney test. C: t-Distributed stochastic neighbor embedding showing all identified cellular clusters of aortas from both control and diabetic mice. D: Heat map showing cell markers differentially expressed between all detected aortic cellular clusters (FDR step up <0.05). Four aortas from Apoe−/− mice per group were included in the pool of aortic cells. *P < 0.05. F, foam; M, monocyte; Mac, macrophage.

Close modal

Expression of canonical gene markers Tagln, Acta2, and Myh11 defined vascular smooth muscle cells (VSMCs) that were further subdivided into two subpopulations based on graph-based cellular clustering (Supplementary Fig. 2). A cluster of SMCs (VSMC I) showed increased expression of genes originally described in the contractile SMC phenotype (higher expression of Myh11, Acta2, Tagln, Cnn1, and Tpm2), indicating that these cells are potentially contractile SMCs (Supplementary Fig. 3A and Supplementary DEG Data). The other SMC cluster (VSMC II) showed a transcriptional profile of synthetic-like SMCs with reduced expression of contractile genes and increased expression of extracellular matrix genes (Col15a1, Col4a6, Col6a1, Col6a2, and Sfrp2) (Supplementary DEG Data). Fibroblast cells were identified with the expression of Serpinf1, Lum, and Dcn (32) (Supplementary Fig. 2). Graph-based clustering subdivided these cells into three subpopulations with distinct transcriptional profiles from aortas of both control and diabetic mice (Fib I, Fib II, and Fib III) (Supplementary Fig. 3B).

Gene markers Pecam1, Cdh5, and Cldn5 were used to identify the endothelial cell cluster (Supplementary Fig. 2). Graph-based unsupervised subclustering of filtered endothelial cells identified four endothelial cell subpopulations (EC1–EC4) (Fig. 2A). The EC1 subpopulation showed a gene expression profile of endothelial cells exposed to TBF. In agreement with a previous single-cell sequencing study (32), expression of several mechanosensitive genes was elevated in the EC1 subpopulation including Icam1, Vcam1, Gata3, and Gata6 (Fig. 2A). Other endothelial cell subpopulations showed higher expression of Klk10, Stmn2, and Mgp (EC2); Sptbn1, Ece1, and Stx12 (EC3); and Cd36, Fabp4, and Meox2 (EC4).

Figure 2

Cellular subclustering and DGE. A: Heat map showing expression of biomarkers of endothelial subclusters. B: Heat map showing expression of biomarkers of foam cells, monocyte-derived DCs (MoDCs), and monocytes (Mono). C: Heat map showing expression of biomarkers of macrophage subclusters. D: Histogram showing numbers of up- and downregulated genes by diabetes in each cell type. E: Volcano plot showing significantly dysregulated genes in aortic EC1 in diabetic compared with control Apoe−/− mice. TFs and Icam1 are labeled, and AP-1 members Fos, Jun, and Junb are highlighted in bold (FDR step up <0.05). F: Volcano plot showing DEGs with log2 fold change of −1.5/1.5 (FDR step up <0.05) in aortic foam cells in diabetic mice vs. control Apoe−/− mice. TFs are labeled, and AP-1 members Fos, Atf3, Atf4, and Junb are highlighted in bold.

Figure 2

Cellular subclustering and DGE. A: Heat map showing expression of biomarkers of endothelial subclusters. B: Heat map showing expression of biomarkers of foam cells, monocyte-derived DCs (MoDCs), and monocytes (Mono). C: Heat map showing expression of biomarkers of macrophage subclusters. D: Histogram showing numbers of up- and downregulated genes by diabetes in each cell type. E: Volcano plot showing significantly dysregulated genes in aortic EC1 in diabetic compared with control Apoe−/− mice. TFs and Icam1 are labeled, and AP-1 members Fos, Jun, and Junb are highlighted in bold (FDR step up <0.05). F: Volcano plot showing DEGs with log2 fold change of −1.5/1.5 (FDR step up <0.05) in aortic foam cells in diabetic mice vs. control Apoe−/− mice. TFs are labeled, and AP-1 members Fos, Atf3, Atf4, and Junb are highlighted in bold.

Close modal

Expression of gene markers for various lymphocyte lineages allowed us to identify B cells (Cd79a, Cd79b, Cd19), T cells (Cd3e, Cd3g, Ccr9), and natural killer T (NKT) cells (Cxcr6, Icos, Il7r) (Supplementary Fig. 2). Gene expression of markers for myeloid lineages identified macrophages (CD68, Adgre1, Cd163), monocytes (F10, Ms4a4c, Chil3), and monocyte-derived dendritic cells (DCs) (Cd209a, Trip3, Plbd1) and foam cells (Trem2, Ctsb, CD9) (22–24) (Fig. 2B and Supplementary Fig. 2). Further subclustering of filtered macrophages identified four macrophage subpopulations (Mac-1–Mac-4) (Fig. 2C). Several markers of inflammatory macrophages, including CD14, CD81, and Cxcl2, showed a higher expression in Mac-1, indicating that these are inflammatory macrophages consistent with a previous study (24) (Fig. 2C). Trbc, CD3g, and Tmsb10 were the top biomarkers for Mac-2, while Ctsb, Ctss, and Srgn were the top biomarkers for Mac-3 (Fig. 2C). Mac-4 exhibited higher expression of gene markers for resident-like macrophages, including Fcna, Lyve1, and Pf4 (Fig. 2C), leading us to hypothesize that this subpopulation has resident macrophages consistent with a previous study (24). These data have established the selective transcriptomic profiles of aortic cellular populations.

DGE

After assigning cellular identities, cells were classified into groups and pooled. GSA with its default settings in Partek Flow identified DGE profiles of cellular clusters and subcluster in diabetes-associated atherosclerosis (Fig. 2D and Supplementary DEG Data). The number of DEGs varied among cellular clusters and subclusters (Fig. 2D). The GSA identified a proatherogenic gene expression profile of B cells (59 upregulated, 44 downregulated), T cells (53 upregulated, 257 downregulated), and NKT cells (90 upregulated, 66 downregulated) in diabetes-associated atherosclerosis (Fig. 2D and Supplementary DEG Data). We observed that B-cell receptors CD79a and Nfkbiα were downregulated in B cells from aortas of diabetic mice compared with that of control mice (Supplementary DEG Data).

GSA identified the number of genes dysregulated in each endothelial cell subpopulation (Fig. 2D and Supplementary DEG Data). Fifty-three genes were upregulated and 93 were downregulated in the EC1 cluster that we hypothesized comprised endothelial cells exposed to TBF (Fig. 2E and Supplementary DEG Data). Both VSMC subclusters showed variation in DEG numbers in diabetes-associated atherosclerosis (71 upregulated, 38 downregulated in VSMC I and 86 upregulated, 10 downregulated in VSMC II) (Fig. 2D and Supplementary DEG Data). The GSA test identified proatherogenic transcriptional profiles of all three fibroblast cell subpopulations, with a varying degree of DEGs (Fig. 2D and Supplementary DEG Data). Many antiatherogenic genes, including Plat, Thbd, and Txn1, were downregulated by diabetes in the Fib III cell population (Fig. 2D and Supplementary DEG Data).

GSA also identified the DGE profile of myeloid cells, including monocyte-derived DCs and monocytes with >300 genes dysregulated in DCs, whereas >400 genes were dysregulated in monocytes by diabetes (Fig. 2D and Supplementary DEG Data). Macrophage subpopulations (Mac-1–Mac-4) also show a variable number of genes dysregulated in diabetes-associated atherosclerosis (Fig. 2D and Supplementary DEG Data). A total of 198 genes were upregulated and 125 genes were downregulated in foam cells (Fig. 2F and Supplementary DEG Data).

Cell-Cell Communication

To assess cell-cell communication, we set out to identify the signal imprinting of the vascular cell populations on all cell populations identified in our scRNA-seq data using the computational method NicheNet (30). This method predicted a list of ligands in each cell population (sender cells) that could induce changes in expression of gene targets in receiver cells (Fig. 3 and Supplementary Figs. 421). EC1, as a receiver cell population, was enriched for many gene targets with strong regulatory potential for top ligands, including Il6 and Tnf (Fig. 3). Predicted gene targets in EC1 with strong regulatory potential included AP-1 members Junb, Icam1, ler3, Irf7, Maff, and Cavin2 for the ligand Tnf that showed higher expression in foam cells, Mac-1, and Mac-2 (Fig. 3). Diabetes increased Tnf expression in foam cells and monocytes (Fig. 3). AP-1 members Fos, Jun, and Atf4 were also among the predicted gene targets in EC4 for ligands Agrn and Jam2. Both Agrn and Jam2 showed higher expression in EC1 and EC2, with diabetes increasing the expression of Agrn in EC1 and Jam2 in EC2 (Supplementary Fig. 4). Both Fos and Jun were downregulated and Atf4 upregulated by diabetes in EC4 (Supplementary DEG Data). The AP-1 member Junb was also one of the predicted gene targets in Mac-4 for the ligand Bmp2, which showed higher expression in Fib III (Supplementary Fig. 5). Bmp2 was also identified as a ligand in foam cells with multiple predicted targets genes, including Zbtb1, ZmiZ1, and Ssbp2 (Supplementary Fig. 6). Fos was enriched in the list of predicted target genes in Fib I and Junb in Fib II, with a higher regulatory potential for the ligands Plau and Tgfb1, respectively (Supplementary Figs. 7 and 8). Expression of Fos was reduced in Fib I, and expression of Junb was elevated in Fib II in diabetes (Supplementary DEG Data).

Figure 3

Cell-cell communication. Ligand activity: NicheNet-identified prioritized ligands with increasing target gene prediction ability (area under the precision-recall [AUPR] curve). Expression in sender: Bubble heat map showing expression of ligands in sender cells. LFC (log fold change) in sender: Heat map showing LFC in expression of ligands by diabetes in sender cells. Predicted gene targets in: EC1: List of predicted gene targets in targeted EC1 cells, with color intensity showing increasing regulatory potential.

Figure 3

Cell-cell communication. Ligand activity: NicheNet-identified prioritized ligands with increasing target gene prediction ability (area under the precision-recall [AUPR] curve). Expression in sender: Bubble heat map showing expression of ligands in sender cells. LFC (log fold change) in sender: Heat map showing LFC in expression of ligands by diabetes in sender cells. Predicted gene targets in: EC1: List of predicted gene targets in targeted EC1 cells, with color intensity showing increasing regulatory potential.

Close modal

AP-1 Complex in Endothelial Cells

Four endothelial subpopulations were identified in this study (EC1–EC4). We hypothesized that EC1 is from an atheroprone aortic region exposed to proatherogenic TBF as evident with the higher expression of Vcam-1, Icam-1, Gata3, and Gata6 and lower expression of lipid-handling genes CD36 and Fabp4 compared with other endothelial cell clusters (Fig. 2A). Endothelial cells exposed to TBF show specific structural and transcriptional changes that are known to contribute to atherosclerotic plaque development. Thus, we focused on EC1, and GSA identified 93 genes downregulated and 53 upregulated in the EC1 cluster in diabetes compared with that of control mice (Fig. 2E and Supplementary DEG Data). We then filtered these genes for mouse TFs and identified nine that were dysregulated in EC1 in diabetes compared with that of control mice (Fig. 2E and Supplementary DEG Data). Of these nine TFs, three were the AP-1 members Fos (FBJ osteosarcoma oncogene), Jun, and Junb. Fos and Jun were exclusively upregulated in EC1 in diabetes (Fig. 2E and Supplementary DEG Data). Importantly, Fos showed the highest statistical significance (FDR) in DGE analysis among all differentially expressed TFs (Fig. 2E and Supplementary DEG Data). This analysis led us to further investigate the role of the AP-1 TF complex in diabetes-associated atherosclerosis.

We first assessed the abundance of FOS in human carotid endarterectomy specimens from participants with and without diabetes. Immunohistochemical analysis with an antibody against FOS showed a more than twofold increase in FOS expression in carotid endarterectomy specimens from participants with type 2 diabetes compared with those from participants without diabetes (Fig. 4A and B). We then investigated the regional heterogeneity of Fos-expressing endothelial cells by performing immunofluorescent assays in the arch and thoracic regions of aortas from control and diabetic mice using antibodies against CD31 and Fos. The arch was used as a representative region for LSS and TBF. We observed more Fos-expressing endothelial cells in the arch compared with the thoracic region in both the control and diabetic setting, indicating that blood flow mediated endothelial-specific regulation of Fos expression (Fig. 4C and D). There was a further increase in the number of Fos-expressing endothelial cells in the arch of the insulin-deficient mice compared with that of control mice, indicating a hyperglycemia- and TBF-specific role for the AP-1 complex (Fig. 4C and D).

Figure 4

AP-1 TFs in TBF-mediated endothelial cell activation in diabetes-associated atherosclerosis. A and B: Representative images and combined dot plot and bar graph of FOS immunostaining in human carotid enterectomy sections from participants without diabetes (n = 5) and participants with diabetes (n = 5). Arrow heads indicate FOS. Scale bars are 100 µm. 3× digital zoom for zoomed in area. Data are mean ± SEM. P value was determined using two-tailed Mann-Whitney test. C and D: Representative images and combined dot plot and bar graph of Fos immunofluorescence staining in the aortic thoracic and arch regions of control (n = 4) and diabetic mice (n = 4). Data are mean ± SEM. P value was determined using Kruskal-Wallis with Dunn multiple comparison tests. E: Schematic diagram for in vitro experiment using microfluidics in HAECs where cells were culture for 24 h under LSS and HSS in the presence and absence of HG (25 mmol/L) and the AP-1 inhibitor T-5224 (10 µmol/L). After 24 h, either nuclear fractions or RNA were extracted, and AP-1 activity assay and RT-PCR were performed for gene expression analysis. F: AP-1 activity was determined in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) with or without AP-1 inhibitor T-5224 (n = 5). Data are mean ± SEM. P value was determined using two-way ANOVA with Tukey multiple comparison test. G: ICAM1, FOS, and JUNB mRNA expression as determined by RT-PCR in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) (n = 5). Data are mean ± SEM. P value was determined using two-way ANOVA with Tukey multiple comparison test. H: Predicted AP-1 binding sites at CAVIN2 and TSC22D3 gene promoters. I: CAVIN2, TSC22D3, and PDIA6 mRNA expression as determined by RT-PCR in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) with or without AP-1 inhibitor T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01. Ar, arch; Th, thoracic; UTR, untranslated region.

Figure 4

AP-1 TFs in TBF-mediated endothelial cell activation in diabetes-associated atherosclerosis. A and B: Representative images and combined dot plot and bar graph of FOS immunostaining in human carotid enterectomy sections from participants without diabetes (n = 5) and participants with diabetes (n = 5). Arrow heads indicate FOS. Scale bars are 100 µm. 3× digital zoom for zoomed in area. Data are mean ± SEM. P value was determined using two-tailed Mann-Whitney test. C and D: Representative images and combined dot plot and bar graph of Fos immunofluorescence staining in the aortic thoracic and arch regions of control (n = 4) and diabetic mice (n = 4). Data are mean ± SEM. P value was determined using Kruskal-Wallis with Dunn multiple comparison tests. E: Schematic diagram for in vitro experiment using microfluidics in HAECs where cells were culture for 24 h under LSS and HSS in the presence and absence of HG (25 mmol/L) and the AP-1 inhibitor T-5224 (10 µmol/L). After 24 h, either nuclear fractions or RNA were extracted, and AP-1 activity assay and RT-PCR were performed for gene expression analysis. F: AP-1 activity was determined in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) with or without AP-1 inhibitor T-5224 (n = 5). Data are mean ± SEM. P value was determined using two-way ANOVA with Tukey multiple comparison test. G: ICAM1, FOS, and JUNB mRNA expression as determined by RT-PCR in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) (n = 5). Data are mean ± SEM. P value was determined using two-way ANOVA with Tukey multiple comparison test. H: Predicted AP-1 binding sites at CAVIN2 and TSC22D3 gene promoters. I: CAVIN2, TSC22D3, and PDIA6 mRNA expression as determined by RT-PCR in HAECs exposed to HSS and LSS in the presence and absence of HG (25 mmol/L) with or without AP-1 inhibitor T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01. Ar, arch; Th, thoracic; UTR, untranslated region.

Close modal

To further confirm these findings and to identify AP-1 target genes linked to LSS, we used microfluidics technology to culture human aortic endothelial cells (HAECs) under two shear stress levels of 10 dynes/cm2 (referred to high shear stress [HSS]) and 2 dyne/cm2 (referred to as LSS). This approach enabled us to mimic the atheroprotective and atherogenic shear stress conditions that endothelial cells experience in arteries (33,34). We exposed HAECs to LSS and HSS in normal glucose (5 mmol/L) (NG)– and high glucose (25 mmol/L) (HG)–containing media in the presence and absence of the AP-1–specific small molecule inhibitor T-5224 before extraction of RNA and the nuclear fraction for gene expression and the AP-1 activity assay (Fig. 4E). DNA binding activity of AP-1 was elevated in HG-stimulated endothelial cells under low- and high-shear conditions (Fig. 4F). The presence of T-5224 in the media blocked both the HG- and low shear–induced DNA binding activity of AP-1 (Fig. 4F). We observed upregulation of ICAM1 mRNA expression as well as AP-1 members FOS and JUNB in an LSS environment in both NG and HG conditions (Fig. 4G). Cells exposed to LSS in the presence of HG showed a further increase in FOS expression compared with LSS alone (Fig. 4G). Cavin2 and Tsc22d3 were downregulated in aortic EC1 in diabetic mice as shown in the scRNA-seq data (Fig. 2E and Supplementary DEG Data). When scanned for AP-1 TF binding sites using the JASPAR open access database that stores manually curated TF binding profiles, promoter regions of both genes were enriched for AP-1 binding sites (Fig. 4H). Expression of both genes CAVIN2 and TSC22D3 was also downregulated by LSS and HG in HAECs (Fig. 4I). CAVIN2, TSC22D3, and PDIA6 are known to be important for endothelial function. The AP-1 inhibitor T-5224 prevented the LSS and HG-induced downregulation of CAVIN2 and TSC22D3, strengthening the postulate of AP-1 inhibition by T-5224 being a potential atheroprotective treatment (Fig. 4I). Gene expression of PDIA6 validated our scRNA-seq data with reduced expression in HAECs exposed to LSS with HG. However, we observed no change in PDIA6 expression when these cells were stimulated with T-5224, indicating that PDIA6 is regulated independently of the AP-1 complex in this setting (Fig. 4I).

To determine the role of the AP-1 complex in monocyte-endothelial interactions, which is one of the initial steps in atherosclerotic plaque development, we performed in vitro static leukocyte cell adhesion assays. We observed that addition of tumor necrosis factor-α (TNF-α) induced a robust increase of approximately twofold in THP-1 adhesion to HAECs isolated from participants with diabetes compared with unstimulated cells (Supplementary Fig. 22A and B). AP-1 inhibition with T-5224 attenuated TNF-α–induced monocyte adhesion to the level seen in the unstimulated diabetic endothelial cells (Supplementary Fig. 22A and B). Collectively, these data suggest that the AP-1 complex regulates transcriptional programs linked to TBF and hyperglycemia-mediated pathways that lead to inflammatory processes, including leukocyte-endothelial interactions in diabetes-associated atherosclerosis.

AP-1 Complex in Foamy Macrophages

Increased expression of genes involved in cholesterol metabolism and transportation and downregulation of genes linked to inflammatory response suggested that a small cluster of cells within myeloid cells identified with gene markers Trem2, Ctsb, and CD9 are indeed lipid-loaded foamy macrophages (22) (Supplementary Fig. 2). More than 400 genes were dysregulated in diabetic foam cells compared with foam cells from control mice (FDR <0.05) (Fig. 2F and Supplementary DEG Data). When filtered for mouse TFs, we identified 16 TFs dysregulated in aortic foam cells of diabetic origin (Fig. 2F). All AP-1 members (Fos, Junb, Atf3, and Atf4) were upregulated in aortic foam cells in diabetes compared with that of control mice (Fig. 2F). We also observed diabetes-induced upregulation of Fos in foam cells, Mac-4 and elevated Junb expression in foam cells, and Mac-4 and DCs compared with all other immune cell types (Supplementary DEG Data). Furthermore, Atf3 was only upregulated in foamy macrophages compared with all other immune cell types, whereas Atf4 was also upregulated in monocytes and B cells (Supplementary DEG Data). Foam cell formation is the hallmark of atherosclerotic lesion development, and several AP-1 members were upregulated in these cells in diabetes. Thus, we explored further the role of AP-1 TF activity in foam cell formation.

To complement these findings and to further investigate the role of AP-1 in foam cell formation, human THP-1–derived macrophages were grown in HG and oxidized LDL (ox-LDL) containing media with or without T-5224 for 24 and 48 h, with foam cell formation and gene expression assessed with Oil Red O (ORO) staining and RT-PCR, respectively (Fig. 5A). THP-1 cell differentiation into macrophages was confirmed with gene expression assessed by RT-PCR for the monocyte marker (CCR2) and the macrophage marker (CD163) (Fig. 5B and C). Assessment of lipid content with ORO staining showed ox-LDL alone and ox-LDL with HG-induced foam cell formation in THP-1–derived macrophages at both the 24- and 48-h time points, with a significant increase in lipid content at the 48-h compared with the 24-h time point (Fig. 5D). There was a further increase in lipid content in cells exposed to ox-LDL and HG compared with ox-LDL alone, though only at the 48-h time point (Fig. 5D). Gene expression of AP-1 members FOS, JUNB, ATF3, and ATF4 was upregulated in ox-LDL and HG-induced foam cells (Fig. 5E–H). Foam cell transformation was prevented by the AP-1 inhibitor T-5224 in both ox-LDL alone and ox-LDL-HG conditions at both the 24- and 48-h time points (Fig. 6A and B). AP-1 activity was induced by ox-LDL and HG, which was attenuated by T-5224 (Fig. 6C). Genes involved in cholesterol metabolism and transportation, including Lipa, Abcg1, and Fabp5, were highly expressed in foam cells compared with other myeloid cells. Since AP-1 is well known for its TF activity, we assessed mRNA expression of genes involved in cholesterol metabolism and transportation. Gene expression assessed by RT-PCR showed that expression of both LIPA and ABCG1 was initially elevated at the 24-h time point and was downregulated at the 48-h time point (Fig. 6D–E). The change in gene expression of ABCG1 was blunted only in cells stimulated with both ox-LDL and HG at the 48-h time point (Fig. 6D). Stimulation of cells with T-5224 prevented the change in gene expression of LIPA at the 24- but not at the 48-h time point (Fig. 6E). Gene expression of fatty acid transporter FABP5 was elevated at both the 24- and 48-h time points in ox-LDL and ox-LDL-HG–stimulated THP-1–derived macrophages compared with control unstimulated cells (Fig. 6F). Treatment of these cells with T-5224 downregulated FABP5 to normal levels at both the 24- and 48-h time points (Fig. 6F). Collectively, these data suggest an important role for the AP-1 complex in cholesterol homeostasis in macrophages in diabetes-associated atherosclerosis and indicate that AP-1 inhibition could represent a novel therapeutic strategy to combat atherosclerotic CVD complications in patients with diabetes.

Figure 5

Role of the AP-1 TFs in foam cell formation. A: Schematic diagram for in vitro experiments using human THP-1–derived macrophages. B: CCR2 mRNA expression determined by RT-PCR in THP-1 monocytes compared with THP-1–derived macrophages, confirming monocyte differentiation into macrophages (n = 5). Data are mean ± SEM. P value was determined using a Mann-Whitney test. C: CD163 mRNA expression determined by RT-PCR in THP-1 monocytes vs. THP-1–derived macrophages, confirming monocyte differentiation into macrophages (n = 5). Data are mean ± SEM. P value was determined using a Mann-Whitney test. D: Combined dot plot and bar graph with representative images of the ORO-stained THP-1–derived macrophages grown in media supplemented with HG (25 mmol/L) and ox-LDL (50 µg/mL). Lipid content as measured by ORO staining normalized for the number of cells is shown as percentage of control. Scale bars are 10 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. EH: AP-1 members FOS, JUNB, ATF3, and ATF4 mRNA levels as determined by RT-PCR in foam cells (THP-1–derived macrophages stimulated with ox-LDL and ox-LDL with or without HG) compared with control cells (THP-1–derived macrophages) (n = 5). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001. Mono, monocyte; Mϕ, macrophage.

Figure 5

Role of the AP-1 TFs in foam cell formation. A: Schematic diagram for in vitro experiments using human THP-1–derived macrophages. B: CCR2 mRNA expression determined by RT-PCR in THP-1 monocytes compared with THP-1–derived macrophages, confirming monocyte differentiation into macrophages (n = 5). Data are mean ± SEM. P value was determined using a Mann-Whitney test. C: CD163 mRNA expression determined by RT-PCR in THP-1 monocytes vs. THP-1–derived macrophages, confirming monocyte differentiation into macrophages (n = 5). Data are mean ± SEM. P value was determined using a Mann-Whitney test. D: Combined dot plot and bar graph with representative images of the ORO-stained THP-1–derived macrophages grown in media supplemented with HG (25 mmol/L) and ox-LDL (50 µg/mL). Lipid content as measured by ORO staining normalized for the number of cells is shown as percentage of control. Scale bars are 10 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. EH: AP-1 members FOS, JUNB, ATF3, and ATF4 mRNA levels as determined by RT-PCR in foam cells (THP-1–derived macrophages stimulated with ox-LDL and ox-LDL with or without HG) compared with control cells (THP-1–derived macrophages) (n = 5). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001. Mono, monocyte; Mϕ, macrophage.

Close modal
Figure 6

AP-1 inhibition in foam cell formation. A: Representative images of the ORO-stained THP-1–derived macrophages grown in media supplemented with ox-LDL (50 µg/mL), with or without HG (25 mmol/L), and with or without T-5224 (10 µmol/L) at 24- and 48-h time points. Lipid content as measured by ORO staining normalized for the number of cells is shown as the percentage of control (NG + vehicle [Veh]). Scale bars are 10 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. B: Combined dot plot and bar graph showing the lipid content in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. C: AP-1 activity determined in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. DF: Expression of ABCG1, LIPA, and FABP5 mRNA as determined by RT-PCR in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 6

AP-1 inhibition in foam cell formation. A: Representative images of the ORO-stained THP-1–derived macrophages grown in media supplemented with ox-LDL (50 µg/mL), with or without HG (25 mmol/L), and with or without T-5224 (10 µmol/L) at 24- and 48-h time points. Lipid content as measured by ORO staining normalized for the number of cells is shown as the percentage of control (NG + vehicle [Veh]). Scale bars are 10 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. B: Combined dot plot and bar graph showing the lipid content in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. C: AP-1 activity determined in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. DF: Expression of ABCG1, LIPA, and FABP5 mRNA as determined by RT-PCR in THP-1–derived macrophages stimulated with ox-LDL with or without HG in the presence and absence of T-5224 (n = 5). Data are mean ± SEM. P value was determined using a two-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal

AP-1 Inhibition With T-5224 in In Vivo Preclinical Studies

Diabetic Apoe−/− mice exhibited elevated blood glucose levels and lower body weight, which was assessed weekly, compared with nondiabetic control mice (Fig. 7A and B). HbA1c levels were elevated in the diabetic mouse cohorts (Fig. 7C). T-5224 treatment had no effects on blood glucose levels, body weight, or HbA1c levels in diabetic and nondiabetic mice (Fig. 7A–C). T-5224 treatment had no effect on the lipid profile (cholesterol, triglycerides, HDL, and LDL) under diabetic or nondiabetic conditions, suggesting that any effects of treatment are lipid and glucose independent (Supplementary Table 4). As expected, diabetes increased atherosclerotic total plaque, with significant increases observed in the arch, thoracic, and abdominal aortic regions. T-5224 treatment had no effects in the nondiabetic Apoe−/− mice but significantly attenuated atherosclerosis development in diabetic Apoe−/− mice (Fig. 7D–H). Further plaque characterization in the aortic arch showed that T-5224 treatment significantly reduced the necrosis in the diabetic plaques (Fig. 7I–J) without affecting collagen content as assessed by picrosirius red staining (Supplementary Fig. 23).

Figure 7

AP-1 inhibition with T-5224 in diabetic Apoe−/− mice. A and B: Weekly blood glucose and body weight for the duration of 10 weeks. C: HbA1c levels at the end of the study. A high blood glucose reading was given a value of 33.3 mmol/L. A below detection reading for HbA1c levels was given a value of 4.0% (n = 6–10 per group). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. D: Representative images of Sudan IV–stained aortas from control + vehicle (PVP), control + T-5224, diabetic + vehicle (PVP), and diabetic + T-5224 Apoe−/− mice. E: Percent total plaque area in the aorta. FH: Percent plaque area in aortic arch, thoracic, and abdominal regions (n = 6–10 per group). I and J: Representative images and quantification of percentage of necrosis from hematoxylin-eosin–stained plaques within the aortic arch region of control and diabetic Apoe−/− mice treated either with vehicle or T-5224 (n = 6 per group). Necrotic area highlighted with dotted lines. Scale bars are 100 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001. Veh, vehicle.

Figure 7

AP-1 inhibition with T-5224 in diabetic Apoe−/− mice. A and B: Weekly blood glucose and body weight for the duration of 10 weeks. C: HbA1c levels at the end of the study. A high blood glucose reading was given a value of 33.3 mmol/L. A below detection reading for HbA1c levels was given a value of 4.0% (n = 6–10 per group). Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. D: Representative images of Sudan IV–stained aortas from control + vehicle (PVP), control + T-5224, diabetic + vehicle (PVP), and diabetic + T-5224 Apoe−/− mice. E: Percent total plaque area in the aorta. FH: Percent plaque area in aortic arch, thoracic, and abdominal regions (n = 6–10 per group). I and J: Representative images and quantification of percentage of necrosis from hematoxylin-eosin–stained plaques within the aortic arch region of control and diabetic Apoe−/− mice treated either with vehicle or T-5224 (n = 6 per group). Necrotic area highlighted with dotted lines. Scale bars are 100 µm. Data are mean ± SEM. P value was determined using a one-way ANOVA with Tukey multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001. Veh, vehicle.

Close modal

Despite novel therapies to manage traditional risk factors, such as hyperlipidemia, CVD remains the major cause of mortality. Diabetes is an independent risk factor for atherosclerosis. With the ongoing diabetes epidemic and despite improved management of various cardiovascular risk factors, including hypertension, dyslipidemia, and glycemic control, there remains a high residual cardiovascular risk. Furthermore, despite recent advances in our understanding of atherosclerosis, we still lack definitive evidence to identify appropriate specific therapeutic strategies that can directly target the intrinsic pathobiology of atherosclerotic plaque development. This study represents several novel aspects, including the use of an scRNA-seq approach to identify the TF AP-1 not only in endothelial dysfunction but also in foam cell formation, directly targeting the underlying pathobiology of atherosclerosis. In agreement with previous studies, scRNA-seq analysis identified endothelial cell subpopulations, including endothelial cells from aortic regions exposed to TBF. In this study, we used a specific inhibitor of T-5224 for the first time to target endothelial dysfunction and foam cell formation in vitro, suggesting AP-1 inhibition as a potential vasculoprotective therapy in cardiovascular complications of diabetes. Multiple agents have been proposed to inhibit AP-1, including curcumin and doxycycline; however, T-5224 used in this study is a highly specific small-molecule inhibitor of DNA binding activity of AP-1 (18,19). This drug has advanced to a phase II clinical trial for rheumatoid arthritis with an excellent safety profile, but its role in CVD specifically in diabetes is unknown (20). Our preclinical study confirmed that AP-1 inhibition with T-5224 attenuated atherosclerosis particularly in diabetes, with no effect in control mice and, thus, should be considered as a novel therapy for atherosclerosis in diabetes. Importantly, our data show reduced necrotic core size after T-5224 treatment. In addition, T-5224 treatment maintains collagen content of the plaque. Taken together, T-5224 treatment appears to stabilize the plaque. Moreover, mice treated with T-5224 showed no effect of the drug on the lipid profile, indicating that the reduction in atherosclerosis may be at more discrete levels, such as reverse cholesterol transport, that would not necessarily change the lipid levels as evident from our in vitro experiments performed in THP-1 cells. Future studies examining the plaque directly may shed further light on these and related mechanisms. The STZ-induced diabetic Apoe−/− mouse is a well-established model of diabetes-associated atherosclerosis studied by us and others (35–37). Indeed, it has been suggested to be the most appropriate model to study diabetes-accelerated atherosclerosis (31). Therefore, although there is an increase in lipids in diabetic mice, this is not considered the major driver of atherosclerosis in this model but, rather, the diabetic milieu associated with a glucose-mediated increase in the local renin-angiotensin system, receptor for advanced glycation end products, and NADPH oxidase–derived reactive oxygen species (35–37). Indeed, interventions such as the ACE inhibitor perindopril and genetic deletion of receptor for advanced glycation end products or the NADPH oxidase isoform NOX1 reduce atherosclerosis in diabetes without significantly affecting lipid levels (35–37).

The AP-1 complex is a heterodimer composed of the FOS, JUN, and ATF (activating transcription factor) family of TFs (38). The regulation of AP-1 occurs at multiple levels, including dimer composition, and at the transcriptional and posttranslational level. Mitogen-activated kinases are a serine/threonine kinase superfamily consisting of the extracellular signal-regulated kinase, Fyn-related kinase, and c-Jun N-terminal kinase (JNK) family of kinases that upon external stimulation, can activate FOS, JUN, and ATF. Therefore, in some studies, JNKs (which phosphorylate c-Jun) have been targeted either using a JNK inhibitor (SP600125 for JNK-1) or genetic deletion of JNK2 to assess the role of AP-1 in vascular disease and atherosclerosis (25,27). However, those studies did not target the AP-1 complex directly either by genetic deletion of AP-1 members, such as the Jun or Fos family of proteins, or using a specific inhibitor of DNA binding activity of AP-1, such as T-5224 used in this study. Furthermore, a study has shown AP-1 activation throughout all stages of human atherosclerosis by quantifying phosphorylated c-Jun nuclear translocation using immunohistochemistry (26). In that study, an intervention treatment was performed using the broad-spectrum antibiotic doxycycline in 12 high-risk individuals. It was hypothesized that by inhibiting JNK1 and JNK2 activity of c-Jun phosphorylation, doxycycline may inhibit AP-1 activity, resulting in reduced inflammation and endothelial function (26). The study concluded that AP-1 inhibition by doxycycline did not reduce the inflammation (as assessed by plasma hs-CRP, IL-6, and IL-8) or influenced the endothelial function (as assessed by flow-mediated vasodilation) (26). However, this conclusion was not supported by the results since AP-1 activity or expression was not assessed in these individuals after doxycycline treatment. In our study, we first observed upregulation of mRNA of the AP-1 members Fos and Junb exclusively in an endothelial cell subpopulation that potentially reflects endothelial cells from aortic regions exposed to TBF and foamy macrophages in the aortas of diabetic Apoe−/− mice. We also show flow-mediated induction of AP-1 activity in HAECs that was prevented by T-5224 treatment. We then focused on transcriptional activity of AP-1 using a specific inhibitor of its DNA binding activity in regulating the transcription of molecules that are important in endothelial function and cholesterol efflux in macrophages in vitro.

Studies have shown that Fos:Jun dimers have stronger affinity for DNA binding and show stronger transcription stimulating activity (39). HG has been shown to increase the AP-1 activity on multiple genes, including TGF-β and TET1 in mesangial cells, suggesting a potential role of the AP-1 complex in diabetic nephropathy (40,41). However, the role of the AP-1 in macrovascular disease in diabetes, including atherosclerosis, is not known. Moreover, studies have shown that FOS and JUN expression and nuclear localization are elevated by LSS in cultured endothelial cells (42,43). Our study is the first in our knowledge to report the activation of downstream gene targets of the AP-1 in LSS and concomitant HG conditions as discussed below. In addition, a limited number of studies have only shown that the expression of the AP-1 members FOS and JUN was elevated in foam cells in atherosclerosis (44,45). However, using T-5224 to inhibit foam cell formation, our study is the first to identify the role of the AP-1 complex in regulating the expression of genes related to cholesterol metabolism and transportation in macrophages in diabetes.

Role of the AP-1 in Endothelial Cell Activation

In agreement with recent single-cell studies, our scRNA-seq data identified endothelial cell subpopulations. When we compared the transcriptomic profile of endothelial cell subpopulations with a previous single-cell study (32), our findings indicated that EC1 is potentially from an atheroprone aortic region exposed to proatherogenic TBF, as evidenced by the higher expression of Vcam-1, Icam-1, Gata6, and Gata3 and lower expression of lipid-handling genes CD36 and Fabp4 in EC1 compared with other endothelial cell clusters, thereby justifying our focus on EC1. DGE analysis identified multiple genes linked to the inflammatory status of endothelial cells in diabetes. For example, G-protein subunit α 2 (Gnai2) was upregulated in EC1 in diabetes. GNAI2 has been recently shown to activate endothelial inflammation in diabetes-associated atherosclerosis (46). Moreover, genes known to be important for endothelial functions, such as Cavin2 and Tsc22d3 (Gilz) were significantly downregulated in EC1, suggesting that hyperglycemia plays a significant role in endothelial cell activation by targeting those genes that are important for endothelial maintenance and function (47,48). Nine TFs were dysregulated in EC1, and members of AP-1 Fos and Jun were exclusively upregulated by diabetes in EC1.

Fluid mechanical forces generated by blood flow are known to cause transcriptional changes in the vascular endothelium (10,11). TBF occurs at regions where arteries branch, bifurcate, or bend and is known to cause LSS on the vascular endothelium. These vascular regions with TBF and LSS are predisposed to atherosclerotic plaque development. Since AP-1 components FOS and JUNB are known to be mechanosensitive and have been shown to be upregulated in atherosclerosis, we further investigated their role in endothelial activation in the setting of diabetes (44,45). First, we confirmed abundance of Fos-expressing endothelial cells as a representative of the AP-1 complex in the aortic arch region, which is known to be atheroprone and is an arterial region exposed to TBF with LSS.

Extending these findings, in vitro microfluidic experiments showed that LSS, an important feature of TBF, not only upregulated members of the AP-1 complex but also elevated AP-1 activity in HAECs. Elevated expression of ICAM1 in cells exposed to LSS and HG validated the microfluidic experimental conditions. As expected, expression of FOS and JUNB was elevated both in LSS and in LSS with HG media but not in HG alone. Further gene expression analysis showed reduced expression of CAVIN2 in cells grown under LSS and HG conditions consistent with our scRNA-seq data where aortic endothelial cells of diabetic mice showed reduced expression of Cavin2, suggesting compromised endothelial function in diabetic mice. This reduction in CAVIN2 expression, caused by LSS and HG, was attenuated by ∼30% by T-5224. In addition to its function in caveolae biogenesis, CAVIN2 has been shown to regulate endothelial cell function via regulation of endothelial nitric oxide synthase activity (47). A recent study profiled a single-cell transcriptomic signature of TBF-associated cells by performing scRNA-seq in left carotid arteries with and without partial carotid ligation. That study showed that CAVIN2 can be used as a marker for endothelial cells, confirming the importance of CAVIN2 in endothelial cell function (16). Another study also profiled a gene expression signature of endothelial cells exposed to TBF (17). Both studies identified genes responsive to TBF, including Icam1, Jun, Fos, and Klf4, indicating a proatherogenic gene expression profile associated with TBF. In agreement with those studies, our scRNA-seq data showed that these genes were upregulated in aortic endothelial cells from diabetic mice, implying that diabetes facilitates a proatherogenic gene expression profile. In addition, our complementary in vitro experiments performed in HAECs isolated from participants with diabetes not only showed that AP-1 inhibition attenuated TNF-α–induced monocyte adhesion but also identified AP-1–responsive genes in TBF conditions. In addition to CAVIN2, our study identified that TSC22D3, another gene important in endothelial function, was regulated by the AP-1 complex. TSC22D3 overexpression has been shown to inhibit endothelial cell adhesion function (49). TSC22D3 was downregulated in aortic EC1 from diabetic mice and in HAECs exposed to LSS and HG. AP-1 inhibition blocked the increase in AP-1 activity and the change in gene expression and increased expression of TSC22D3 back to normal levels. These mechanistic studies clearly strengthen the postulate that the AP-1 complex is an important mechanosensitive transcriptional regulator in the setting of diabetic atherosclerosis.

Role of AP-1 in Foam Cell Formation

Disturbed cholesterol homeostasis in macrophages is central to foam cell formation in atherosclerotic CVD (13,14). Although several single-cell studies have reported transcriptional changes in macrophages, including foamy macrophages, in atherosclerosis, the key mechanism of transcriptional regulation linked to cholesterol homeostasis in macrophages in the context of diabetes has not been identified previously. A previous study has profiled gene expression changes in aortic leukocytes, including monocytes, macrophages, and B and T cells, in high-fat diet–fed Apoe−/− mice (23). Two of the top 20 DEGs (Fcmr1 and Bank11) in B cells were also enriched in our data. None of the top 20 DEGs in monocytes were enriched in our data set; however, the DEGs in macrophages showed that 3–12 DEGs of the top 20 were also enriched in four subclusters of macrophages in our scRNA-seq data (C1qb, C1qa, C1qc, Csfr1, Trf, C3ar1, Adgre1, Fcgr3, Pf4, Ms4a7, Cd14, and Cd63). Since both studies were performed in an Apoe−/− mouse model, this comparison has defined the effects of diabetes per se on gene expression changes in addition to the effects seen as a result of hyperlipidemia (23). In agreement with the previous scRNA-seq studies in nondiabetic hyperlipidemic mice, foamy and nonfoamy macrophages were clearly distinguishable based on their transcriptomic profiles as identified by scRNA-seq (22–24). Our cell-specific gene expression data showed elevated expression of genes related to the inflammatory response (Cd163, Ccl2, Il1b, Nfkb1, Ccl24, Ccl6, Birc3, and Ccl7) and decreased expression of genes linked to cholesterol metabolism and transportation (Abcg1, Lipa, Fabp5, and Lpl) in the nonfoamy macrophage cell population. In contrast, foamy macrophages showed decreased expression of inflammatory genes and elevated expression of genes linked to cholesterol uptake, efflux, and processing consistent with the previous studies (22–24). Although, previous studies have shown upregulation of FOS and JUN in foam cells (44,45), we performed in vitro experiments in THP-1–derived macrophages to gain molecular insight into the therapeutic effects of AP-1 inhibition in restoring disturbed cholesterol homeostasis. Accumulation of cytoplasmic lipid droplets, as assessed by ORO staining, showed increased lipoprotein uptake in cells incubated with ox-LDL and HG. Moreover, at the 48-h time point, cells incubated with ox-LDL and HG showed a further increase in lipid accumulation compared with cells incubated with ox-LDL alone, thereby identifying the additional effects of hyperglycemia on cholesterol homeostasis in human-derived macrophages. Gene expression analysis showed increased expression of the AP-1 members ATF3, ATF4, FOS, and JUNB consistent with the findings of our murine scRNA-seq data. Both LIPA and ABCG1 were initially upregulated at the 24-h time point, with subsequent downregulation at the 48-h time point indicating decreased capacity in lipid efflux in macrophages that are transforming into foam cells. AP-1 inhibition had limited effects on the expression of both LIPA and ABCG1. However, FABP5, a fatty acid transporter that showed a consistent increase in expression at 48 h, was downregulated by AP-1 inhibition to the level observed in unstimulated macrophages. Foamy macrophages in our scRNA-seq analysis showed increased expression of Fabp5 compared with nonfoamy macrophages consistent with previous scRNA-seq studies (22). The abundant expression of FABP5 in foamy macrophages is consistent with a potential role for FABP5 in foam cell formation (50). Our study identified the AP-1 complex as a transcriptional regulator of FABP5. Indeed, AP-1 complex inhibition halted the transformation of macrophages into foam cells via potentially downregulating the expression of FABP5.

Intercellular communication can be studied using computational tools such as NicheNet. When applied to single-cell data, NicheNet analysis can predict ligand-receptor interactions linked to gene expression regulation in target cells. Given that AP-1 members are well-known nuclear TFs, their absence as NicheNet-predicted ligands in the analysis of our scRNA-seq data were unsurprising. However, multiple AP-1 members were identified as predicted gene targets in EC1, EC4, Mac-4, Fib I, and Fib II, with increased regulatory potential for ligands such as Tnf, Bmp2, Plau, Il6, and Tgfb1. These ligands showed higher expression in other cell types, including foam cells, macrophages, and fibroblasts, indicating that cell-cell communication may play a role in controlling gene expression changes linked to the AP-1 pathway in diabetes-associated atherosclerosis. Although this analysis proposes a role for cell-cell communication in driving the expression of some of the AP-1 members and potentially downstream AP-1 target genes, further experimental evidence is required to validate this analysis. In addition, AP-1 is a complex made of multiple proteins, and thus, NicheNet may not predict the full potential of the activity of AP-1 in this setting.

In conclusion, using the unbiased state-of-the-art technology of scRNA-seq, the AP-1 complex was identified as a central gene regulatory factor in endothelial activation and foam cell formation in diabetes-induced atherosclerosis. Furthermore, experimental evidence as reported in this study suggested a key role for the AP-1 complex in regulating the transcriptional program involved in the development of atherosclerosis. Thus, our study has identified an upstream master regulator of this gene expression program, specifically the AP-1 complex, which is involved in endothelial cell activation and cholesterol homeostasis. We acknowledge some limitations of this study, including scRNA-seq data analysis of a single pool of aortas of control and diabetic mice and not using flavopiridol in our aortic dissociation protocol. Furthermore, we have only studied an insulin-deficient STZ-induced diabetic mouse model; therefore, future studies should also include insulin-resistant diabetic models of atherosclerosis. Since AP-1 is a complex made of multiple TFs, we could not perform genetic deletion studies in vitro or in vivo but, instead, used a specific inhibitor. Female mice show resistance to STZ and show less atherosclerosis development; thus, only male mice were used in this study. With respect to the involvement of flow-induced endothelial cell dysfunction and foam cell formation, small-molecule AP-1 inhibitors like T-5224 as assessed in our preclinical study could represent a major advance in managing diabetes-associated atherosclerosis. Therefore, it is proposed that targeting this complex could afford superior therapeutic options to combat atherosclerotic CVD especially in diabetes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.26060317.

Acknowledgments. The authors thank the Alfred Research Alliance sequencing platform for sequencing the scRNA libraries.

Funding. This study was supported by National Health and Medical Research Council grant App1163233 (to K.A.M.J.-D.), Australian Research Council grants LP190100728 and DO200101248 (to S.B.), and National Heart Foundation of Australia fellowships 102492 (to A.W.K.) and 105631 (to A.C.C.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.W.K. wrote the manuscript. A.W.K., M.A., and A.S. performed the in vitro experiments and analyzed data. A.W.K., M.K.S.L., K.C.S., A.D., A.M.D.W., and Y.Z. performed the in vivo experiment and analyzed data. A.W.K., M.K.S.L., and S.M. performed the scRNA-seq. A.W.K. and S.M. analyzed the scRNA-seq data. A.W.K. and S.B. performed the microfluidics experiments. A.W.K. and K.A.M.J.-D. developed the concept. A.M.D.W., S.M., M.E.C., A.C.C., A.J.M., S.B., and K.A.M.J.-D. revised the manuscript. M.E.C. and K.A.M.J.-D. provided resources. All authors read and edited the manuscript. A.W.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Pyörälä
K
,
Laakso
M
,
Uusitupa
M
.
Diabetes and atherosclerosis: an epidemiologic view
.
Diabetes Metab Rev
1987
;
3
:
463
524
2.
Sabatine
MS
,
Giugliano
RP
,
Keech
AC
, et al.;
FOURIER Steering Committee and Investigators
.
Evolocumab and clinical outcomes in patients with cardiovascular disease
.
N Engl J Med
2017
;
376
:
1713
1722
3.
Schwartz
GG
,
Steg
PG
,
Szarek
M
, et al.;
ODYSSEY OUTCOMES Committees and Investigators
.
Alirocumab and cardiovascular outcomes after acute coronary syndrome
.
N Engl J Med
2018
;
379
:
2097
2107
4.
La Sala
L
,
Prattichizzo
F
,
Ceriello
A
.
The link between diabetes and atherosclerosis
.
Eur J Prev Cardiol
2019
;
26
:
15
24
5.
Libby
P
.
The changing nature of atherosclerosis: what we thought we knew, what we think we know, and what we have to learn
.
Eur Heart J
2021
;
42
:
4781
4782
6.
Soehnlein
O
,
Libby
P
.
Targeting inflammation in atherosclerosis - from experimental insights to the clinic
.
Nat Rev Drug Discov
2021
;
20
:
589
610
7.
Botts
SR
,
Fish
JE
,
Howe
KL
.
Dysfunctional vascular endothelium as a driver of atherosclerosis: emerging insights into pathogenesis and treatment
.
Front Pharmacol
2021
;
12
:
787541
8.
Maguire
EM
,
Pearce
SWA
,
Xiao
Q
.
Foam cell formation: a new target for fighting atherosclerosis and cardiovascular disease
.
Vascul Pharmacol
2019
;
112
:
54
71
9.
Hsu
P-L
,
Chen
J-S
,
Wang
C-Y
,
Wu
H-L
,
Mo
F-E
.
Shear-induced CCN1 promotes atheroprone endothelial phenotypes and atherosclerosis
.
Circulation
2019
;
139
:
2877
2891
10.
Chiu
J-J
,
Chien
S
.
Effects of disturbed flow on vascular endothelium: pathophysiological basis and clinical perspectives
.
Physiol Rev
2011
;
91
:
327
387
11.
Wang
L
,
Luo
J-Y
,
Li
B
, et al
.
Integrin-YAP/TAZ-JNK cascade mediates atheroprotective effect of unidirectional shear flow
.
Nature
2016
;
540
:
579
582
12.
Patel
M
,
Savvopoulos
F
,
Berggren
CC
, et al
.
Considerations for analysis of endothelial shear stress and strain in FSI models of atherosclerosis
.
J Biomech
2021
;
128
:
110720
13.
Chistiakov
DA
,
Bobryshev
YV
,
Orekhov
AN
.
Macrophage-mediated cholesterol handling in atherosclerosis
.
J Cell Mol Med
2016
;
20
:
17
28
14.
Yan
J
,
Horng
T
.
Lipid metabolism in regulation of macrophage functions
.
Trends Cell Biol
2020
;
30
:
979
989
15.
Khan
AW
,
Paneni
F
,
Jandeleit-Dahm
KAM
.
Cell-specific epigenetic changes in atherosclerosis
.
Clin Sci (Lond)
2021
;
135
:
1165
1187
16.
Li
F
,
Yan
K
,
Wu
L
, et al
.
Single-cell RNA-seq reveals cellular heterogeneity of mouse carotid artery under disturbed flow
.
Cell Death Discov
2021
;
7
:
180
17.
Andueza
A
,
Kumar
S
,
Kim
J
, et al
.
Endothelial reprogramming by disturbed flow revealed by single-cell RNA and chromatin accessibility study
.
Cell Rep
2020
;
33
:
108491
18.
Aikawa
Y
,
Morimoto
K
,
Yamamoto
T
, et al
.
Treatment of arthritis with a selective inhibitor of c-Fos/activator protein-1
.
Nat Biotechnol
2008
;
26
:
817
823
19.
Uchihashi
S
,
Fukumoto
H
,
Onoda
M
,
Hayakawa
H
,
Ikushiro
S-i
,
Sakaki
T
.
Metabolism of the c-Fos/activator protein-1 inhibitor T-5224 by multiple human UDP-glucuronosyltransferase isoforms
.
Drug Metab Dispos
2011
;
39
:
803
813
20.
Ye
N
,
Ding
Y
,
Wild
C
,
Shen
Q
,
Zhou
J
.
Small molecule inhibitors targeting activator protein 1 (AP-1)
.
J Med Chem
2014
;
57
:
6930
6948
21.
Zhao
G
,
Lu
H
,
Liu
Y
, et al
.
Single-cell transcriptomics reveals endothelial plasticity during diabetic atherogenesis
.
Front Cell Dev Biol
2021
;
9
:
689469
22.
Kim
K
,
Shim
D
,
Lee
JS
, et al
.
Transcriptome analysis reveals nonfoamy rather than foamy plaque macrophages are proinflammatory in atherosclerotic murine models
.
Circ Res
2018
;
123
:
1127
1142
23.
Winkels
H
,
Ehinger
E
,
Vassallo
M
, et al
.
Atlas of the immune cell repertoire in mouse atherosclerosis defined by single-cell RNA-sequencing and mass cytometry
.
Circ Res
2018
;
122
:
1675
1688
24.
Cochain
C
,
Vafadarnejad
E
,
Arampatzi
P
, et al
.
Single-cell RNA-seq reveals the transcriptional landscape and heterogeneity of aortic macrophages in murine atherosclerosis
.
Circ Res
2018
;
122
:
1661
1674
25.
Osto
E
,
Matter
CM
,
Kouroedov
A
, et al
.
c-Jun N-terminal kinase 2 deficiency protects against hypercholesterolemia-induced endothelial dysfunction and oxidative stress
.
Circulation
2008
;
118
:
2073
2080
26.
Meijer
CA
,
Le Haen
PAA
,
van Dijk
RA
, et al
.
Activator protein-1 (AP-1) signalling in human atherosclerosis: results of a systematic evaluation and intervention study
.
Clin Sci (Lond)
2012
;
122
:
421
428
27.
Wang
J
,
An
FS
,
Zhang
W
, et al
.
Inhibition of c-Jun N-terminal kinase attenuates low shear stress-induced atherogenesis in apolipoprotein E-deficient mice
.
Mol Med
2011
;
17
:
990
999
28.
Watson
AMD
,
Li
J
,
Samijono
D
, et al
.
Quinapril treatment abolishes diabetes-associated atherosclerosis in RAGE/apolipoprotein E double knockout mice
.
Atherosclerosis
2014
;
235
:
444
448
29.
Neri
S
,
Mariani
E
,
Meneghetti
A
,
Cattini
L
,
Facchini
A
.
Calcein-acetyoxymethyl cytotoxicity assay: standardization of a method allowing additional analyses on recovered effector cells and supernatants
.
Clin Diagn Lab Immunol
2001
;
8
:
1131
1135
30.
Browaeys
R
,
Saelens
W
,
Saeys
Y
.
NicheNet: modeling intercellular communication by linking ligands to target genes
.
Nat Methods
2020
;
17
:
159
162
31.
Hsueh
W
,
Abel
ED
,
Breslow
JL
, et al
.
Recipes for creating animal models of diabetic cardiovascular disease
.
Circ Res
2007
;
100
:
1415
1427
32.
Kalluri
AS
,
Vellarikkal
SK
,
Edelman
ER
, et al
.
Single-cell analysis of the normal mouse aorta reveals functionally distinct endothelial cell populations
.
Circulation
2019
;
140
:
147
163
33.
Baratchi
S
,
Khoshmanesh
K
,
Woodman
OL
,
Potocnik
S
,
Peter
K
,
McIntyre
P
.
Molecular sensors of blood flow in endothelial cells
.
Trends Mol Med
2017
;
23
:
850
868
34.
Chatterjee
S
.
Endothelial mechanotransduction, redox signaling and the regulation of vascular inflammatory pathways
.
Front Physiol
2018
;
9
:
524
35.
Candido
R
,
Jandeleit-Dahm
KA
,
Cao
Z
, et al
.
Prevention of accelerated atherosclerosis by angiotensin-converting enzyme inhibition in diabetic apolipoprotein E-deficient mice
.
Circulation
2002
;
106
:
246
253
36.
Candido
R
,
Allen
TJ
,
Lassila
M
, et al
.
Irbesartan but not amlodipine suppresses diabetes-associated atherosclerosis
.
Circulation
2004
;
109
:
1536
1542
37.
Soro-Paavonen
A
,
Watson
AMD
,
Li
J
, et al
.
Receptor for advanced glycation end products (RAGE) deficiency attenuates the development of atherosclerosis in diabetes
.
Diabetes
2008
;
57
:
2461
2469
38.
Trop-Steinberg
S
,
Azar
Y
.
AP-1 expression and its clinical relevance in immune disorders and cancer
.
Am J Med Sci
2017
;
353
:
474
483
39.
Bejjani
F
,
Evanno
E
,
Zibara
K
,
Piechaczyk
M
,
Jariel-Encontre
I
.
The AP-1 transcriptional complex: local switch or remote command?
Biochim Biophys Acta Rev Cancer
2019
;
1872
:
11
23
40.
Weigert
C
,
Sauer
U
,
Brodbeck
K
,
Pfeiffer
A
,
Häring
HU
,
Schleicher
ED
.
AP-1 proteins mediate hyperglycemia-induced activation of the human TGF-beta1 promoter in mesangial cells
.
J Am Soc Nephrol
2000
;
11
:
2007
2016
41.
Tan
Y
,
Cao
H
,
Li
Q
,
Sun
J
.
The role of transcription factor Ap1 in the activation of the Nrf2/ARE pathway through TET1 in diabetic nephropathy
.
Cell Biol Int
2021
;
45
:
1654
1665
42.
Bao
X
,
Clark
CB
,
Frangos
JA
.
Temporal gradient in shear-induced signaling pathway: involvement of MAP kinase, c-fos, and connexin43
.
Am J Physiol-Heart C
2000
;
278
:
H1598
H1605
43.
Nagel
T
,
Resnick
N
,
Dewey
CF
,
Gimbrone
MA
.
Vascular endothelial cells respond to spatial gradients in fluid shear stress by enhanced activation of transcription factors
.
Arterioscler Thromb Vasc Biol
1999
;
19
:
1825
1834
44.
Thomas
AC
,
Eijgelaar
WJ
,
Daemen
MJAP
,
Newby
AC
.
Foam cell formation in vivo converts macrophages to a pro-fibrotic phenotype
.
PLoS One
2015
;
10
:
e0128163
45.
Trusca
VG
,
Fuior
EV
,
Kardassis
D
,
Simionescu
M
,
Gafencu
AV
.
The opposite effect of c-Jun transcription factor on apolipoprotein E gene regulation in hepatocytes and macrophages
.
Int J Mol Sci
2019
;
20
46.
Chao
M-L
,
Luo
S
,
Zhang
C
, et al
.
S-nitrosylation-mediated coupling of G-protein alpha-2 with CXCR5 induces Hippo/YAP-dependent diabetes-accelerated atherosclerosis
.
Nat Commun
2021
;
12
:
4452
47.
Boopathy
GTK
,
Kulkarni
M
,
Ho
SY
, et al
.
Cavin-2 regulates the activity and stability of endothelial nitric-oxide synthase (eNOS) in angiogenesis
.
J Biol Chem
2017
;
292
:
17760
17776
48.
Caligiuri
G
.
Mechanotransduction, immunoregulation, and metabolic functions of CD31 in cardiovascular pathophysiology
.
Cardiovasc Res
2019
;
115
:
1425
1434
49.
Cheng
Q
,
Fan
H
,
Ngo
D
, et al
.
GILZ overexpression inhibits endothelial cell adhesive function through regulation of NF-κB and MAPK activity
.
J Immunol
2013
;
191
:
424
433
50.
Umbarawan
Y
,
Enoura
A
,
Ogura
H
, et al
.
FABP5 is a sensitive marker for lipid-rich macrophages in the luminal side of atherosclerotic lesions
.
Int Heart J
2021
;
62
:
666
676
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.