CD44 Drives M1 Macrophage Polarization in Diabetic Retinopathy

Abstract Purpose Diabetic retinopathy is a typical complication of diabetes, which can facilitate the risk of blindness in severe cases. We sought to determine the function of CD44 in inflammatory responses of human retinal microvascular endothelial cells (HRMECs) and macrophage polarization during diabetic retinopathy (DR). Methods The hub genes were tested based on two datasets from the Gene Expression Omnibus database. Gene Ontology and pathway enrichment analysis was conducted on the base of differentially expressed genes (DEGs). The infiltration score and infiltration of the immune cells were assessed, and the link between key genes and macrophages was analyzed. The role of CD44 in HRMECs and macrophage polarization was determined by quantitative reverse transcription polymerase chain reaction, western blot, cell counting kit-8, Enzyme-linked immunosorbent assay, flow cytometry, and immunofluorescence. Results DEGs were enriched in several pathways linked to DR, such as cellular response to retinoic acid, retinol metabolic process, retina homeostasis, PI3K-AKT signaling pathway, and leukocyte transendothelial migration. A total of 144 DEGs were identified by up-regulation both in GSE102485 and GSE160306. Moreover, the infiltration of macrophages was greater in the DR group than that in the control group. We highlighted an obvious increase in the expression of CD44 and CD86 in patients with DR, and distinct positive associations were found between levels of macrophages and levels of CD44 and CD86. Furthermore, CD44 expression was substantially increased in HRMECs under high glucose (HG) conditions and CD44 knockdown markedly inhibited HG-induced inflammatory responses of HRMECs. HG-induced HRMECs remarkably influenced M1 polarization of macrophages, but CD44 knockdown significantly nullified this effect. Conclusions CD44 influenced the advancement of DR via meditating M1 polarization of macrophages. Our findings could enhance the understanding of the mechanism of DR, which might offer a therapeutic target for DR patients.


Introduction
Diabetic retinopathy (DR) is one of the most prevalent and special microvascular complications of diabetes mellitus (DM). DR remains the leading cause of vision loss globally, which is linked to retinal vascular damage and abnormalities. 1,2 It is estimated that diabetes patients will increase from 537 million in 2011 to 643 million in 2030. 3 What matters is that the process of DR is undetectable, which is accompanied by irreversible retinal pathological changes. 4 Therefore, a comprehensive study of the mechanism influencing DR is urgently required to investigate an alternative therapeutic clue for patients with DR.
Inflammation, the central link to pathogenic processes in DR and metabolic syndrome, particularly involves innate immunity in the progression of complications. 5 Long-term or excessive stimulation will keep tissue cells in a state of inflammation for a long time, promoting continuous tissue damage. 6 For example, chronic stimulation of hyperglycemia in DR patients gives rise to long-term ischemia and hypoxia in retinal tissues, thereby stimulating the body to produce an immune inflammatory response. 7,8 Several studies have demonstrated that in the serum, vitreous and retinal tissues of DR patients, there are leukocyte adhesion stasis, increased neutrophils, T, B lymphocytes, and monocytes/macrophages, etc. [9][10][11] Among these immune cells, activated macrophages are generally differentiated into pro-(M1) or anti-inflammatory (M2). 12 The activated macrophages are implicated in a range of biological processes such as cell growth and differentiation, which are linked to other signals such as cytokines, hormones, and signal transduction pathways. 13 PPAR-a agonist fenofibrate relieved diabetic nephropathy by mitigating M1 macrophages via improving endothelial cell function in db/db mice. 14 Nonetheless, the specific mechanism of the immune cells involved in the course of the development of DR is still unclear.
In the present study, two datasets (GSE102485 and GSE160306) from the Gene Expression Omnibus (GEO) database were employed to identify candidate immune cellassociated biomarkers in DR and determine the immune landscape of DR. Furthermore, we evaluated the impact of CD44 on human retinal microvascular endothelial cells (HRMECs) and macrophage polarization. This study may provide a potential therapeutic clue for DR.

Data download
Two RNA seq datasets (GSE102485 and GSE160306) were acquired from the GEO database. [15][16][17] A total of 22 retinal samples of patients with proliferative DR (PDR) and 3 normal retinal samples from GSE102485, and the macula region and the retinal periphery tissue samples of 10 patients with nonproliferative DR (NPDR) and 11 patients with diabetic macular edema (DME) from GSE160306 were enrolled in our research.

Screening for differentially expressed genes
To identify the differentially expressed genes (DEGs) between DR and control tissues, or between NPDR and DME, the false discovery rate (FDR) <0.05 andlog 2 fold change (FC) -!1 were deemed as the threshold value. The DEGs in the datasets were extracted and determined via the "limma" package, and then visualization was carried out by employing the "pheatmap" and "ggplot2" packages.

Functional analyses
The functional enrichment analyses, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, were executed to decipher differences in the biological processes using the clusterProfiler package in R language. A P-value less than 0.05 was deemed to be statistically significant.

Protein-protein interaction network and retrieval of key genes
The connection between DEGs and immune cells was visualized by employing Cytoscape software. Subsequently, the degree algorithm of the CytoHubba plug-in was applied to determine major genes. The genes in the Protein-protein interaction (PPI) network were deemed hub genes in the process of DR.

Establishment of high glucose-induced cell model
Human retinal microvascular endothelial cells (HRMECs, Pricella, CP-H130) were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco, Grand Island, NY, USA) containing 10% fetal bovine serum (FBS, Gibco), 100 IU/ml penicillin, and 100 lg/ml streptomycin (Gibco). HRMECs were incubated in an incubator comprising 5% CO 2 at 37 C. The medium was altered once every 2 days. Passage treatment was carried out when the cell fusion reached approximately 90%. Confluent cells were incubated in 30 mmol/L D-glucose for 96 h to derive high glucose (HG)induced cells. Furthermore, cells were cultured in 5 mmol/L D-glucose regarded as the normal glucose (NG) condition.

Cell transfection
CD44 siRNA and normal control siRNA (Genepharma, China) were transfected into HRMECs by applying Lipofectamine 2000 (Thermo Fisher Scientific, USA) following the manufacturer's protocol. At 6 h after transfection, the medium was substituted by fresh medium. At 24 h after transfection, cell supernatants were procured for subsequent experiments.

Macrophage induction and cell culture
Human monocytes (THP-1 cells) were incubated in a DMEM medium containing penicillin-streptomycin (100 U/ml) and FBS at 37 C in an incubator comprising 5% CO2. To induce monocyte-to-macrophage differentiation, THP-1 cells were treated with 100 ng/ml phorbol 12-myristate 13-acetate for 24 h. Subsequently, the supernatants of NC and CD44 knockdown HRMECs were added for incubating the treated THP-1 cells. At the same time, the treated THP-1 cells were stimulated with 20 ng/ml IFN-c and 100 ng/ml LPS for 24 h.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
Total RNA was extracted with Trizol reagent (Invitrogen, USA). The RevertAid First Strand cDNA Synthesis Kit (Thermo, USA) was utilized to synthesize first-strand cDNA. qRT-PCR was conducted using the ABI Q6 (Applied Biosystems Inc., USA) and FastStart Universal SYBR Green Master (Roche, USA) following the manufacturer's directive. The reaction conditions were 95 C for 10 min, and 40 cycles of 95 C for 15 s and 60 C for 60 s. The relative mRNA expression was determined using the 2 -DDCt method. GAPDH was regarded as an endogenous control. The primer sequences were displayed in Table S1.

Enzyme-linked immunosorbent assay
The levels of tumor necrosis factor-a (TNF-a), interleukin (IL)-1b, and interleukin (IL)-6 were evaluated by corresponding Enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Bioengineering Institute, China) in light of the manufacturer's protocol. The absorbance at 450 nm of wells was tested with a microplate reader (SpectraMAX Plus384, USA).

Cell counting kit-8 assay
Cell proliferation was detected by employing cell counting kit-8 (CCK-8) (Beyotime, China) at 0, 24, 48, 72, and 96 h. HRMECs were seeded in 96-well plates with 2 Â 10 3 cells in each well and cultured at 37 C. Subsequently, each well was supplemented with 10 ml CCK-8 solution, and cells were cultured for 1 h at 37 C. A microplate reader (BioRad, Hercules, CA, USA) was applied to examine the absorbance of each well at 450 nm wavelength.

Apoptosis assay
The apoptosis rate was determined by employing the Annexin V-FITC/propidium iodide (PI) Apoptosis Detection Kit (Beyotime, China) following the manufacturer's directive. After HG-induced HRMECs were transfected with CD44 siRNA, the cells were adopted, rinsed with PBS, and resuspended in 1 Â Binding Buffer. After the addition of Annexin V-FITC and PI, the cells were cultured at room temperature for 15 min in the dark. Finally, cells were analyzed by flow cytometry (Beckman Coulter, Germany).

Statistical analysis
Statistical analysis was carried out with SPSS 23.0 and GraphPad Prism 9.0. All data are displayed as mean ± standard deviation (SD). The differences between groups were calculated by two-side Student's t-test or one-way analysis of variance (one-way ANOVA) followed by Tukey's test. Pvalue <0.05 was deemed statistically significant.

Data preprocessing and DEGs analysis
To discriminate key genes associated with the process of DR, DEGs in retinal tissues were screened from the GEO database GSE102485 (DR vs control) and GSE160306 (DME vs NPDR) using the limma package. With FDR < 0.05 and -log2FC -!1 as the cut-off value, we obtained 1,534 DEGs from the GSE102485 dataset, with 840 DEGs that were featured by up-regulation and 694 DEGs that were identified by down-regulation in patients with DR compared with control ( Figure 1(A)). Additionally, 538 DEGs were procured from the GSE160306, including 451 DEGs that demonstrated a significant increase and 87 DEGs with an obvious decrease in patients with DME compared with NPDR patients (Figure 1(A)). The bidirectional hierarchical clustering heatmaps of DEGs were depicted in Figure 1(B). After that, GO analysis and pathway enrichment analysis were performed to interrogate potential signaling pathways related to these DEGs in DR. GO analysis indicated that the DEGs could regulate a series of biological processes that might be linked to DR, such as cellular response to retinoic acid, retinol metabolic process, retina homeostasis, etc (Figure 1(C)). In addition, it was discovered that DEGs were associated with leukocyte transendothelial migration, cell adhesion molecules, and the PI3K-Akt signaling pathway (Figure 1(D)).

Immune cell infiltration is related to DR progression
Furthermore, to narrow the pool of candidate genes, overlapping DEGs from the GSE102485 and GSE160306 data sets were acquired with Venn diagram and we discovered a total of 144 DEGs were identified by up-regulation both in these two datasets (Figure 2(A)). Next, the PPI network of overlapping DEGs was appraised by STRING, which was visualized utilizing Cytoscape (Figure 2(B)). Genes with a degree of more than 10 in the PPI network were determined as hub genes, and we eventually derived 20 hub genes, including CD44 and CD86 (Figure 2(C)). Moreover, the results of gene function enrichment analysis indicated that the overlapped DEGs were mainly enriched in the macrophage markers, microglia pathogen phagocytosis signaling pathway, complement activation, validation response, and other immune signaling pathways (Figure 2(D)). As a result, we proposed that immune cell infiltration was linked to DR progression.

The immune landscape in DR
To determine the immune landscape in DR, the immucellAI online database was utilized to analyze the immune cell infiltration and abundance of immune cells in each sample. As depicted in Figure 3(A), the infiltration score of the DR group was evidently boosted compared with the control group. In addition, the infiltration ratio of B cell, CD8 þ T cell, and CD8 naive cells was significantly decreased, while the infiltration of monocyte, macrophage, NK cells Tr1 cells, nTreg, and effect memory cells was elevated in the DR group relative to the control group (Figure 3(B)). Similarly, it was found that the infiltration score in the DME group was markedly augmented relative to the NPDR group (Figure 3(C)). The infiltration ratio of B cell, neutrophil, and cd cells was lowered, whereas the infiltration ratio of DC cells, monocyte, macrophage, and NK cells were increased in the DME group relative to the NDPR group (Figure 3(D)). Notably, the abundance of B cells, monocyte, and macrophage cells was significantly different between the control and DR groups in the two datasets.

The relationship between DEGs and immune cells
To assess the genes that regulate the B cells and macrophage cells, Pearson correlation analysis was carried out and a regulatory network of immune-associated genes was displayed in Figure 4(A). A Venn diagram was then employed to determine overlapping genes from the immune-associated genes and hub genes and only two key genes (CD44 and CD86) were derived (Figure 4(B)). Moreover, prominent positive correlations were observed between levels of macrophage and levels of CD44 and CD86 (Figure 4(C)). Furthermore, the expression of CD44 and CD86 was increasingly elevated in the DR group compared with the control group (Figure 4(D and E)). Also, it was also discovered that the expression of CD44 and CD86 was upregulated in the DME group relative to the NPDR group (Figure 4(D  and E)). In the two molecules, CD86 is already a wellknown marker of M1 macrophage, so it is not meaningful for us to verify it. Therefore, we chose CD44 for the next research.

CD44 is increasingly expressed in HRMECs under HG condition
We determined the levels of CD44 and inflammatory factors in HRMECs. qRT-PCR assay implied that the mRNA expression of CD44, TNF-a, IL-1b, and IL-6 in HRMECs was markedly elevated in the HG group relative to the NG group ( Figure 5(A)). Furthermore, the ELISA assay also demonstrated that HG evidently enhanced the levels of TNF-a, IL-1b, and IL-6 in HRMECs ( Figure 5(B)). Western blot assay revealed that HG remarkably promoted the protein expression of CD44 in HRMECs ( Figure 5(C)). Overall, these data indicated that HG promoted the expression of CD44 and inflammatory factors in HRMECs.

CD44 regulates HG-induced inflammatory responses in HRMECs
We investigated the function of CD44 on inflammatory responses in HG-induced HRMECs. We discovered that the mRNA and protein expression of CD44 was significantly decreased in HG-induced HRMECs after CD44 knockdown ( Figure 6(A and B)). Interference of CD44 dramatically repressed the levels of TNF-a, IL-1b, and IL-6 in HGinduced HRMECs (Figure 6(C)). CCK8 assay implied that CD44 knockdown remarkably increased the proliferation of HG-induced HRMECs (Figure 6(D)). Flow cytometry indicated apoptosis of HG-induced HRMECs was significantly suppressed after CD44 knockdown (Figure 6(E)). These results revealed that knockdown of CD44 inhibited inflammatory responses in HG-induced HRMECs.

HRMECs promote M1 polarization of macrophages through CD44
We explored whether CD44 regulated M1 polarization of macrophages. As displayed in Figure 7(A), the levels of TNF-a, IL-1b, and IL-6 were markedly elevated in macrophages incubated by the supernatant of HG-induced HRMECs. Knockdown of CD44 in HRMECs dramatically restrained the levels of TNF-a, IL-1b, and IL-6 in The infiltration score in the control and DR groups in GSE102485 was examined. The infiltration score in the DR group was considerably higher than that in the control group. (B) The infiltration ratio of immune cells was assessed in GSE102485. The infiltration score was obviously increased in the DR group compared with the control group. (C) The infiltration score in nonproliferative diabetic retinopathy (NPDR) and diabetic macular edema (DME) groups in GSE160306 was appraised. The infiltration score in the DME group was markedly higher than that in the NDPR group. (D) The infiltration ratio of immune cells was examined in GSE160306. In comparison with the NDPR group, the infiltration score was higher in the DME group. The data was displayed as mean þ SD. Ã p < 0.05; ÃÃ p < 0.01. The network map showed the genes that control B cells and macrophages. CD44 was shown to regulate macrophage and B cells. (B) A Venn diagram was used to identify overlapping genes from the immune-associated genes and hub genes. Two common genes were discovered between immuneRDs and hub genes. (C) The levels of CD44 and CD86 were positively correlated with macrophage. (D, E) The expression of CD44 and CD86 was quantified in GSE102485 and GSE160306 in patients with diabetic retinopathy (DR), nonproliferative diabetic retinopathy (NPDR) and diabetic macular edema (DME). It indicated that CD44 and CD86 were both increased in patients with DR and patients with DME. The data was displayed as mean þ SD. Ã p < 0.05; ÃÃÃ p < 0.001.

Figure 5. CD44 is highly expressed in high glucose (HG)-induced human retinal microvascular endothelial cells (HRMECs). (A)
The mRNA expression of CD44 and inflammatory factors in normal glucose (NG) and HG-induced HRMECs was evaluated by qRT-PCR. It indicated that the mRNA expression of CD44, TNF-a, IL-1b, and IL-6 was considerably elevated in HG-treated HRMECs. (B) ELISA was used to measure the levels of inflammatory factors in NC and HG-induced HRMECs. It revealed that the concentration of TNF-a, IL-1b, and IL-6 was remarkably enhanced in HRMECs after HG treatment. (C) The protein expression of CD44 in NC and HG-induced HRMECs was detected by western blot. It revealed that CD44 protein expression was markedly increased in HG-induced HRMECs. The data was displayed as mean þ SD. Ã p < 0.05; ÃÃ p < 0.01. macrophages. Western blot assay implied that HG-induced HRMECs remarkably promoted the protein expression of M1 markers, NOS2, and TNF-a in macrophages, whereas CD44 knockdown inhibited their expression (Figure 7(B)). Immunofluorescence results showed that HG-induced HRMECs enhanced the expression of CD86 in macrophages, but this impact was abrogated by CD44 knockdown (Figure 7(C)). The above results demonstrated that HRMECs promote M1 polarization of macrophages via CD44.

Discussion
DR, the chronic progressive complication of patients with DM, is one of the leading causes of vision loss, but the precise pathogenic mechanisms have not been fully elucidated. 18 Unfortunately, current therapeutic remedies for DR-related blindness are not satisfactory. 19 A recent study revealed that immune cells and immune factors are involved in DR progression. 20 The present study demonstrated that DR patients expressed high levels of CD44 and CD86, and the expression of CD44 was obviously enhanced in HG-treated HRMECs. We also discovered that CD44 induced the secretion of inflammatory cytokines in HG-HRMECs, and then activated the M1 polarization of macrophages. The results contribute to the further understanding of the pathogenesis of DR.
CD44, a momentous factor in the process of blood vessel formation, is a single-transmembrane glycoprotein. 21,22 The CD44 protein is encoded by 19 exons, 10 of which are successive across all isoforms in human cells. 23,24 It has been well-known that CD44 is prevalently found in the membrane of a variety of cells, which is responsible for multiple physiological and pathological activities, including pathological angiogenesis, intercellular communication, the process of DR, etc. 25 As an example, CD44-epidermal growth factor receptor interaction increases the production of MMP-2, which strengthens the potential of cell motility in melanoma. 26 Also, CD44 is implicated in ligand-receptor interactions between profibrotic microglia and cytokines upregulated in the proliferative DR vitreous. 27 In addition, the transcription levels of CD44 in the DR mouse model group were featured by an apparent increase as against the control group. 25 Intriguingly, we found a remarkable increase in the expression of CD44 in DR patients relative to the control group in the present study. Furthermore, cellular function experiments displayed that knockdown of CD44 reduced inflammatory responses of HG-induced HRMECs.
CD86, also known as B7-2, is an important member of the growing B7 family. 28 Since it was initially cloned in the early 1990s, B7-2 was thought to be a factor that could stimulate the growth of decidual stromal cells. 29 Interestingly, B7-2 is primarily located on class II human leukocyte antigen þ decidual cells and helps to stimulate allogeneic T cells. 29 CD86 can interact with CD28 and CD152/cytotoxic T lymphocyte-associated antigen 4 molecules on T cells, which influences T cell survival and response via co-stimulatory and co-inhibitory pathways. 30,31 The T cell receptor complex first recognizes the antigen peptide expressed on antigen-presenting cells and binds to MHC, and then the co-stimulatory molecule CD86/CD80 on APC interacts with CD28 on naive T cells to generate stimulatory signals for T cells. 32 Diabetes contributed to B cell activation and increased the expression of CD80 and CD86. 33 Here, it was discovered that the expression of CD86 was characterized by significant up-regulation in the DR group relative to the control group.
As a crucial component of the inflammatory microenvironment, macrophages are the major acting cells of the immune system in humans. 34 Under normal circumstances, macrophages function in monitoring infection, regulating cell renewal, wound repair, and tissue remodeling. 35 Studies have shown that plasticity and versatility are the basic characteristics of macrophages, and macrophages in tissues respond to environmental signals to acquire different functional phenotype responses to various signals of the body. 36,37 Recent studies have discovered that M1 macrophages have the potential to facilitate retinal pathological neovascularization. 38,39 Herein, we unveiled that interference of CD44 suppressed HG-induced inflammatory responses of HRMECs, and HRMECs promoted M1 polarization of macrophages through CD44. Therefore, CD44 plays a pivotal role in the development of DR through M1 macrophage polarization.
In summary, CD44 and CD86 were augmented in DR and were positively linked to the level of macrophages. Additionally, CD44 induced the secretion of inflammatory cytokines in HG-treated HRMECs and then promoted M1 macrophage polarization. This research might provide an innovative viewpoint on the development of DR.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by Key Laboratory of Guangdong Higher Education Institutes (No.2021KSYS009) and Guangzhou Key Laboratory of Biological Targeting Diagnosis and Therapy (No.202201020379).

Data availability statement
The data used and/or analyzed during the current study are available from the corresponding author on reasonable request.