Next, we isolated bone marrow-derived macrophages (BMDMs), cardiomyocytes, fibroblasts, and ECs were isolated from normal hearts. IL-4 stimulation promoted the expression and secretion of CCL28 in BMDMs (Fig. S2A, 2B). Hypoxia mediated the expression and secretion of CCL28 in fibroblasts, but not in cardiomyocytes or ECs (Figure S2C-2F).
We evaluated the expression of the CCL28 receptor, chemokine C-C motif receptor (CCR) 3, and CCR10 in ischemic diseases. The mRNA levels of Ccr3 and Ccr10 were upregulated in the ischemic myocardium and gastrocnemius muscle, and the increase in Ccr10 expression was more significant than that in Ccr3 (Fig. 2A, 2B). The protein and mRNA expression levels of CCR10 peaked on the 7 days post-HI and 3 days post-MI (Fig. 2A-2D). The gating strategy for CCR10 cells and their cell subsets were presented in Figure S2G, including immunocytes (CCR10CD45 cells), fibroblasts (CCR10CD45CD140a cells), and ECs (CCR10CD45CD31 cells). CCR10 cells were the most abundant on 7 days post HI and 3 days post MI (Fig. 2E). Additionally, the largest percentage of CCR10CD45CD31 ECs was observed on 7 days post HI among the CCR10 cell subsets. Meanwhile, CCR10CD45 immunocytes were the most prevalent on 3 days post MI, followed by CCR10CD45CD31 ECs (Fig. 2F). Immunofluorescence staining displayed that CCR10 was highly expressed in ECs on 7 days post HI and 3 days post MI (Fig. 2G). Immunofluorescence staining for isolectin b4 (IB4, red) and CCR10 (green) showed that CCR10 was also located in the capillaries (Fig. S2H). These findings suggest that the CCL28-CCR10 axis is involved in EC function in HI and MI, and that CCR10 ECs may play an important role in ischemic diseases.
Flow cytometry was performed to evaluate Ki67 expression levels in CCR10 and CCR10 ECs on 7 days post HI and 3 days post MI (Fig. S3A). The mean fluorescence intensity (MFI) of KI67 was higher in CCR10 ECs than that in CCR10 ECs, indicating that CCR10 ECs have a strong proliferate ability (Fig. 3A, 3B). Next, we used flow cytometry to sort CCR10 and CCR10 ECs and then examined their endothelial function (Fig. 3C). Annexin V/propidium iodide (PI) staining was utilized to evaluate the anti-apoptotic ability of CCR10 and CCR10 ECs after 24 h of hypoxia treatment. The results of flow cytometry demonstrated that the apoptosis level of CCR10 ECs was significantly lower than that of CCR10 ECs (Fig. 3D). Wound healing assays showed that CCR10 ECs had a stronger migration ability than CCR10 ECs (Fig. 3E). We performed a spheroid-based sprouting assay to study angiogenesis. Interestingly, a significant increase in both sprout number and length was observed in CCR10 ECs compared with that in CCR10 ECs (Fig. 3F, 3G). Tube formation assay showed that more tube numbers were observed in CCR10 ECs to CCR10 ECs (Fig. 3H). Finally, we constructed an EC aging model using hydrogen peroxide and found that CCR10 ECs could resist the aging stimuli (Fig. 3I). In addition, RT-qPCR analysis revealed that the expression of angiogenic programs in CCR10 ECs was higher than that in CCR10 ECs, including IGF-1, VEGF-A and FGF-2 (Fig. 3J). In addition, we detected the response of CCR10 ECs to CCL28 and found that CCL28 promoted angiogenesis in CCR10 endothelial cells (Fig. S3B). These in vivo and in vitro experiments showed that CCR10 ECs were characterized by an enhanced proangiogenic capacity.
Given the consistent correlation between the number of CCR10 ECs and CCL28 expression in myocardial and hind limb ischemia, CCL28 was considered to be promoting the expression of CCR10 in ECs. The recombinant CCL28 protein (rCCL28) upregulated the levels of CCR10 mRNA and protein in a concentration-dependent manner (Fig. 4A, Fig. S3C). We also found that 10 ng/mL recombinant CCL28 protein (rCCL28) increased CCR10 protein levels in BEND.3 cells (Fig. S3C). The flow cytometry results also confirmed that rCCL28 increased the proportion of CCR10 ECs in a concentration-dependent manner (Fig. 4B, S3D). Ccl28 global knockout mice (Ccl28-KO) were generated (Fig. S3E) and Ccl28 was successfully deleted on 3 days post MI, as confirmed by RT-qPCR and Western blot assay (Fig. S3F, 3G). H&E staining and echocardiography revealed no differences between the WT and Ccl28-KO mice under physiological conditions (Fig. S3H). RNA-seq was performed using CD45CD31 ECs magnetic bead sorted from WT and Ccl28-KO mice (Fig. S4A). Gene set enrichment analysis (GSEA) showed that Ccl28 deletion inhibited angiogenesis and blood vessel remodeling, and promoted oxidative stress and redox pathways (Fig. 4C). A volcano plot of the differential RNA is shown in Fig. S4B, from which the three most significantly downregulated transcription factors were the transcription factor CP2-like 1 (Tfcp2l1), double homeobox B-like 1 (Duxbl1), and sex determining region Y-box protein 5 (Sox5) (Fig. S4C). RT-qPCR data revealed that si-Sox5, rather than si-Tfcp2l1 and si-Duxbl1, decreased the mRNA level of Ccr10 (Fig. S4D). These genes were involved in mitogen-activated protein kinase (MAPK) signaling, chemokine signaling, and inflammatory response (Fig. S4E-S4G).
SOX5 involves in angiogenesis and EC function. MAPK/extracellular-signal-regulated kinase (ERK) signaling is the main regulator of angiogenesis. It is assumed that the CCL28-CCR10 axis mediates angiogenesis through MAPK signaling and SOX5. We found that rCCL28 upregulated the protein expression of p-ERK1/2, VEGFA, SOX5, and CCR10, whereas the ERK1/2 inhibitor Ravoxertinib and CCR10 inhibitor BI-6901 blocked CCL28-mediated increases in these proteins (Fig. 4D and S4H). SOX5 knockdown resulted in a reduction of CCR10 expression during both control and rCCL28 stimulation (Fig. 4E and S4I). The JARSPAR database predicted the presence of SOX5 binding sites in the Ccr10 promoter region (site1: -2000 to -1400; site 2: -1000 to -400; site 3: -400 to +200) (Fig. S5A). To confirm this, a chromatin immunoprecipitation (ChIP)-qPCR assay was performed on ECs using either an anti-SOX5 antibody or an immunoglobulin G (IgG) control (Fig. 4F). Ccr10 promoter region 3 (-400 to +200) was preferentially enriched after immunoprecipitation with the anti-SOX5 antibody compared to the IgG control (Fig. 4G and S5B). Dual-luciferase assays confirmed that SOX5 knockdown or Ravoxertinib inhibited the transcriptional activity of pGL3-promoter containing the Ccr10 promoter region (-400 to +200). Additionally, rCCL28 stimulation promoted its transcriptional activity and this increase was reversed by SOX5 knockdown or Ravoxertinib (Fig. 4H and S5C). Finally, the role of SOX5 in EC function was evaluated. The ability of ECs to sprout and resist aging was suppressed by SOX5 knockdown. Furthermore, SOX5 knockdown blocked the enhancement of sprouting and anti-aging abilities induced by rCCL28 (Figs. 4I-K and S5D). In addition, we explored the crosstalk between macrophages and endothelial cells using in vitro co-cultures. The results showed that M2 macrophages promoted the expression of CCR10 and tube formation ability, while the CCL28 antibody inhibited these effects (Fig. S5E, F). We investigated the effect of CCL28 on human endothelial cells (HMEC-1). The results showed that rCCL28 promote the expression of CCR10 and the ability of tube formation in HMEC-1 (Fig. S5G, H). Taken together, CCL28 enhances endothelial ability by activating CCR10/ERK/SOX5 positive feedback signaling (Fig. 4L).
The hindlimb ischemia model mice were administered with an intraperitoneal injection of rCCL28 (50 μg/kg, every four days) or vehicle. We found the increased CCR10 expression (Fig. S6G-I) and the proportion of CCR10CD31 ECs (Fig. S6J) in the rCCL28 group compared to the vehicle group on 7 days post HI. Laser Doppler imaging showed the improved blood flow recovery after femoral artery ligation following treatment with rCCL28 (Fig. 6A). Further staining results demonstrated that rCCL28 increased vascular numbers (Fig. 6B and S6E) and reduced gastrocnemius fibrosis (Fig. 6C). Further, we evaluate the therapeutic potential of rCCL28 (50 μg/kg, every four days) in MI. An increase in CCR10 expression (Fig. S6G-I) and the proportion of CCR10CD31 ECs (Fig. S6J) was observed under the rCCL28 treatment on 3 days post MI. The intervention of rCCL28 promoted cardiac angiogenesis (Fig. 6D and S6K) and improved cardiac function, remodeling (Fig. 6E-G and S6L), and coronary permeability (Fig. 6H).
We then investigated whether rCCL28 could be used to treat Ccl28-knockout mice. Treatment with rCCL28 improved blood flow recovery (Fig. S7A), increased vascular numbers (Fig. S7B), and reduced gastrocnemius fibrosis (Fig. S7D) following hind limb ischemia in KO mice. Similar therapeutic effects of rCCL28 were also demonstrated in KO mice with myocardial infarction (Fig. S7C and S7E, F).
Considering the CCL28-CCR10 regulatory axis, we speculated that CCL28 promotes angiogenesis in CCR10 endothelial cells. AAV9-Tie1-sh-CCR10 (2 × 10 vg/mouse) was transfected into endothelial cells via tail vein injection. After 4 weeks, we examined the expression of CCR10 and found that the mRNA level of CCR10 in ECs from the hearts and gastrocnemius (Fig. S8A) and the proportion of CCR10CD45CD31 ECs were significantly reduced by AAV9-Tie1-sh-CCR10 compared to AAV9-Tie1-sh-Control (Fig. S8B, C).
Mice subjected to tail vein injections of AAV9-Tie1-sh-CCR10 or AAV9-Tie1-sh-Control for 4 weeks underwent femoral artery ligation in one lower limb. Hindlimb ischemia model mice were administered with an intraperitoneal injection of rCCL28 (50 μg/kg, every four days) or vehicle. Laser Doppler imaging showed that the downregulation of CCR10 resulted in impaired blood flow recovery and blocked the therapeutic effect of rCCL28 (Fig. S8D). Further staining that demonstrated AAV9-Tie1-sh-CCR10 decreased vascular numbers (Fig. S8E) and increased gastrocnemius fibrosis (Fig. S8F) compared to AAV9-Tie1-sh-Control. The effect of rCCL28 on angiogenesis and fibrosis is blocked by AAV9-Tie1-sh-CCR10 (Fig. S8E, F).
We determined the effect of AAV9-Tie1-sh-CCR10 on MI. Intervention with AAV9-Tie1-sh-CCR10 inhibited cardiac angiogenesis (Fig. S9A) and impaired remodeling (Fig. S9B), cardiac function (Fig. S9C), and coronary permeability (Fig. S9D). In addition, tail vein injection of AAV9-Tie1-sh-CCR10 blocked the therapeutic effect of rCCL28 in MI (Fig. S9A-D). Taken together, these results indicate that CCR10 is crucial for endothelial angiogenesis, and the pro-angiogenic effect of CCL28 depends on CCR10 ECs.
Diabetes impairs endothelial function and angiogenesis. Here, we investigated the role of the CCL28-CCR10 axis in myocardial and hindlimb ischemia in diabetes. Western blot analysis showed that CCL28 and CCR10 expression was reduced in HI and MI of db/db mice compared to that in WT mice on 7 days post HI and 3 days post MI (Figs. 7A-D and S10A). The proportion of CCR10CD31 ECs decreased in diabetic mice (Fig. 7E-H). Further, we evaluate the therapeutic potential of rCCL28 (50 μg/kg, every four days) in HI and MI of db/db mice. Immunofluorescence analysis indicated increased vascular density in the diabetic ischemic gastrocnemius and heart treated with rCCL28 (Fig. 7I-L and Fig. S10B). Treatment with rCCL28 significantly improved blood flow recovery after femoral artery ligation, as shown by Laser Doppler imaging (Fig. 7M). And gastrocnemius fibrosis, and cardiac infarct size and fibrosis were mitigated by treatment of rCCL28 (Fig. S10C-D). Angiogenesis is related to wound healing and we observed that rCCL28 accelerated wound healing caused by lower limb femoral artery ligation in db/db mice (Fig. S10E).
This nested case-control study was conducted in a cohort of 493 consecutive patients with stable angina and CTO between February 2019 and September, 2023. Seventy-seven patients were excluded for the following reasons: pregnancy, lactation or malignant tumor with<1-year life expectancy (9), chronic heart failure (22), pulmonary heart disease (6), immune system disorders (2), history of coronary artery bypass grafting (3), percutaneous coronary intervention within the prior 3 months (31), and intermittent claudication (4). Finally, 416 patients were enrolled in the study (Fig. 8A). Rentrop scores of 0, 1, 2, and 3 were observed in 24, 133, 145, and 114 patients, respectively. Baseline characteristics according to Rentrop scores are detailed in Table S1. Serum levels of CCL28 gradually increased across the Rentrop scores of groups 0, 1, 2, and 3 (p < 0.05) (Fig. 8B). The levels of CCL28 were significantly lower in patients with a poorer CCV cluster (Rentrop score 0 and 1 groups) than in those with a better CCV cluster (Rentrop score 2 and 3 groups) [582 (521-666) vs. 629 (571-792) pg/mL, p < 0.001]. CCL28 levels were significantly correlated with the Rentrop score (Spearman's r = 0.2967, p < 0.0001). The association between CCL28 and Rentrop scores was investigated using univariate and multivariate logistic regression analyses. In Model 1, after adjusting for age and sex, CCL28 was correlated with Rentrop score (OR, 2.373; 95% CI, 1.582-3.586; p < 0.0001). In Model 2, after adjusting for sex, age, body mass index, smoking, hypertension, diabetes, prior myocardial infarction, and serum creatinine (sCr), the association between CCL28 and Rentrop score remained significant (OR, 2.133; 95% CI, 1.392-3.289; p = 0.0005) (Table 1). In the receiver operating characteristic (ROC) curve analysis, the area under the curve (AUC) was 0.650 (95% CI 0.595-0.705, p < 0.0001) for serum levels of CCL28 in predicting poor CCV (Rentrop scores 0 and 1), with an optimal cutoff point of 605 pg/mL (sensitivity, 59.1%; specificity, 63.1%) (Fig. 8C). CCL28 significantly improved the prediction efficacy for the Rentrop score in addition to traditional CTO risk factors including sex, age, body mass index, smoking, hypertension, diabetes, prior myocardial infarction, and sCr. ROC analysis showed AUC of 0.657 (95%CI 0.604-0.711) and 0.695 (95%CI 0.642-0.746) for traditional CTO risk factors and traditional CTO risk factors+CCL28, respectively. The CTO risk model+CCL28 had higher AUC values than the CTO risk model alone (0.695 vs.0.657, p = 0.0115) (Fig. 8D). Moreover, the addition of CCL28 to the conventional model improved risk reclassification (0.388, 95%CI: 0.205-0.572, p < 0.0001) and integrated discrimination improvement (0.035, 95%CI: 0.020-0.049, p < 0.0001) for the Rentrop score (Table 2).