Evaluation of Proteoforms of the Transmembrane Chemokines CXCL16 and CX3CL1, Their Receptors, and Their Processing Metalloproteinases ADAM10 and ADAM17 in Proliferative Diabetic Retinopathy

Abstract

The transmembrane chemokine pathways CXCL16/CXCR6 and CX3CL1/CX3CR1 are strongly implicated in inflammation and angiogenesis. We investigated the involvement of these chemokine pathways and their processing metalloproteinases ADAM10 and ADAM17 in the pathophysiology of proliferative diabetic retinopathy (PDR). Vitreous samples from 32 PDR and 24 non-diabetic patients, epiretinal membranes from 18 patients with PDR, rat retinas, human retinal Müller glial cells and human retinal microvascular endothelial cells (HRMECs) were studied by enzyme-linked immunosorbent assay, immunohistochemistry and Western blot analysis. In vitro angiogenesis assays were performed and the adherence of leukocytes to CXCL16-stimulated HRMECs was assessed. CXCL16, CX3CL1, ADAM10, ADAM17 and vascular endothelial growth factor (VEGF) levels were significantly increased in vitreous samples from PDR patients. The levels of CXCL16 were 417-fold higher than those of CX3CL1 in PDR vitreous samples. Significant positive correlations were found between the levels of VEGF and the levels of CXCL16, CX3CL1, ADAM10 and ADAM17. Significant positive correlations were detected between the numbers of blood vessels expressing CD31, reflecting the angiogenic activity of PDR epiretinal membranes, and the numbers of blood vessels and stromal cells expressing CXCL16, CXCR6, ADAM10 and ADAM17. CXCL16 induced upregulation of phospho-ERK1/2, p65 subunit of NF-κB and VEGF in cultured Müller cells and tumor necrosis factor-α induced upregulation of soluble CXCL16 and ADAM17 in Müller cells. Treatment of HRMECs with CXCL16 resulted in increased expression of intercellular adhesion molecule-1 (ICAM-1) and increased leukocyte adhesion to HRMECs. CXCL16 induced HRMEC proliferation, formation of sprouts from HRMEC spheroids and phosphorylation of ERK1/2. Intravitreal administration of CXCL16 in normal rats induced significant upregulation of the p65 subunit of NF-κB, VEGF and ICAM-1 in the retina. Our findings suggest that the chemokine axis CXCL16/CXCR6 and the processing metalloproteinases ADAM10 and ADAM17 might serve a role in the initiation and progression of PDR.

Keywords: ADAM10; ADAM17; CX3CL1; CXCL16; chemokines; metalloproteinases; proliferative diabetic retinopathy.

Figure 1

Detection of CXCL16, ADAM10, and ADAM17 in vitreous fluid. The expression of CXCL16, ADAM10, and ADAM17 in equal volumes (15 μl) of vitreous fluid samples from patients with proliferative diabetic retinopathy (PDR) (n=8) and from non-diabetic patients with rhegmatogenous retinal detachment (RD) (n=8) was determined by Western blot analysis. Immunoreactive proteoforms are indicated in kilodaltons (kDa) on the basis of a size standard preparation. Representative sets of samples are shown.

Figure 2

Detection of pathologic new blood vessels, leukocytes and myofibroblasts in proliferative diabetic retinopathy epiretinal fibrovascular membranes. No labeling was observed in the negative control slide (the same procedure without the primary antibody) (A). Immunohistochemical staining for the endothelial cell marker CD31 showing pathologic new blood vessels in an epiretinal membrane from a patient with active neovascularization (B) and in a membrane from a patient with inactive involuted disease which is composed mostly of fibrous tissue (arrows) (C). Immunohistochemical staining for the leukocyte common antigen CD45 showing infiltrating leukocytes in the stroma (D). Immunohistochemical staining for α-smooth muscle actin (α-SMA) showing immunoreactivity in spindle-shaped myofibroblasts (E) (scale bar, 10 μm).

Figure 3

Characterization of cells expressing CXCL16 and CX3CL1 in proliferative diabetic retinopathy epiretinal fibrovascular membranes. Immunohistochemical staining for CXCL16 (A) and CX3CL1 (C) showing immunoreactivity in vascular endothelial cells (arrows). Immunoreactivity for CXCL16 was also detected in stromal spindle-shaped cells (arrows) (B). Double immunohistochemistry for CD45 (brown) and CXCL16 (red) (D) or CX3CL1 (red) (E) demonstrated co-expression in stromal leukocytes (arrows). No counterstain was applied in panels (D, E) (scale bar, 10 μm).

Figure 4

Characterization of CXCR6- and CX3CR1-expressing cells in proliferative diabetic retinopathy epiretinal fibrovascular membranes. Immunoreactivities for CXCR6 (A) and CX3CR1 (B) were detected in vascular endothelial cells. Immunoreactivity for CXCR6 was also detected in stromal spindle-shaped cells (A) (scale bar, 10 μm).

Figure 5

Characterization of ADAM10- and ADAM17-expressing cells in proliferative diabetic retinopathy epiretinal fibrovascular membranes. Immunoreactivities for ADAM10 (A, C) and ADAM17 (D, F) were detected in vascular endothelial cells (arrows) in membranes from patients with active neovascularization (A, D) and in membranes from patients with inactive involuted disease (C, F). Immunoreactivities for ADAM10 (B) and ADAM17 (E) were also detected in stromal spindle-shaped cells (arrows) (scale bar, 10 μm).

Figure 6

Characterization of ADAM10- and ADAM17-expressing cells in proliferative diabetic retinopathy epiretinal fibrovascular membranes. Double immunohistochemistry for CD45 (brown) and ADAM10 (red) (A) or ADAM17 (red) (B) showed co-expression in leukocytes (arrows). No counterstain to visualize the cell nuclei was applied in panels (A, B) (scale bar, 10 μm).

Figure 7

Detection of myofibroblasts and leukocytes in proliferative vitreoretinopathy epiretinal fibrocellular membranes. No staining was observed in the negative control slide (A). Immunohistochemical staining for α-smooth muscle actin (α-SMA) showing immunoreactivity in spindle-shaped myofibroblasts (B). Immunohistochemical staining for CD45 showing infiltrating leukocytes (C) (scale bar, 10 μm).

Figure 8

Characterization of cells expressing transmembrane chemokines and their receptors in proliferative vitreoretinopathy epiretinal fibrocellular membranes. Immunohistochemical stainings for CXCL16 (A), CXCR6 (B), CX3CL1 (C), and CX3CR1 (D) showing immunoreactivities in spindle-shaped myofibroblasts. Double immunohistochemistry for CD45 (brown) and CXCL16 (red) (E) or CX3CL1 (red) (F) showed co-expression in leukocytes (arrows). No counterstain to visualize the cell nuclei was applied in panels E and F (scale bar, 10 μm).

Figure 9

Characterization of ADAM10- and ADAM17-expressing cells in proliferative vitreoretinopathy epiretinal fibrocellular membranes. ADAM10 (A) and ADAM17 (B) immunoreactivities were detected in spindle-shaped myofibroblasts. Double immunohistochemistry for CD45 (brown) and ADAM10 (red) (C) or ADAM17 (D) demonstrated co-expression in leukocytes (arrows). No counterstain to visualize the cell nuclei was applied in panels (C, D) (scale bar, 10 μm).

Figure 10

CXCL16 induces vascular endothelial growth factor (VEGF) expression and activates ERK1/2 and NF-κB pathways in Müller cells. Müller cells were left untreated or treated with CXCL16 for 24 h. Levels of VEGF were quantified in the culture media by enzyme-linked immunosorbent assay (ELISA). Protein expression of phospho-ERK1/2 and the p65 subunit of NF-κB in the cell lysates was determined by Western blot analysis (representative Western blots are depicted on top of the graphs). The box plots (median and interquartile range) show results from three different experiments performed in triplicate. (*p < 0.05; Mann-Whitney test).

Figure 11

The proinflammatory cytokine tumor necrosis factor-alpha (TNF-α) induces the expression of CXCL16 and ADAM17 in Müller cells. Müller cells were left untreated or treated with TNF-α (50 ng/ml) for 24 h. Levels of CXCL16 were quantified in the culture media by ELISA. Protein expression of CXCL16, CXCR6, ADAM10, and ADAM17 in cell lysates was determined by Western blot analysis. Results are expressed as median (interquartile range) from three different experiments performed in triplicate. (*p < 0.05; Mann-Whitney test).

Figure 12

Human retinal microvascular endothelial cells (HRMECs) express CXCL16, CXCR6, CX3CL1, and CX3CR1. HRMECs were left untreated or treated with interleukin-1 beta (IL-1β) (50 ng/ml) or tumor necrosis factor-alpha (TNF-α) (50 ng/ml) for 24 h. Protein expression of CXCL16 (A), CXCR6 (B), CX3CL1 (C), and CX3CR1 (D) in cell lysates was determined by Western blot analysis (representative Western blots are depicted on top of the graphs). The same loading control (β-actin) was used for quantitation of the relative band intensity of both CXCL16 and CXCR6. Levels of CX3CL1 were quantified in the culture media by ELISA (E). The box plots (median and interquartile range) show results from three different experiments performed in triplicate. Kruskal-Wallis test and Mann-Whitney tests were used for comparisons between three groups and two groups, respectively. *P < 0.05 compared with values obtained from untreated cells. ^#p < 0.05 compared with IL-1β-treated cells.

Figure 13

CXCL16 induces leukocyte adhesion to human retinal microvascular endothelial cells (HRMECs). Adhesion of fluorescently-labeled monocytic cells to a HRMEC monolayer treated with tumor necrosis factor‐α (TNF‐α) (50 ng/ml) or CXCL16 (50 ng/ml) was assessed. Bar graphs represent three independent experiments (control: 11 wells, TNF‐α: 11 wells and CXCL16: 11 wells). RFU, relative fluorescent unit. One-way ANOVA and independent t-tests were used for comparisons between the three and two groups, respectively (A). HRMECs were left untreated or treated with CXCL16 (50 ng/ml) for 24 h. Protein expression of intercellular adhesion molecule-1 (ICAM-1) in cell lysates was determined by Western blot analysis (B). Results are expressed as mean ± standard deviation from three different experiments performed in triplicate. (*p < 0.05; independent t-test).

Figure 14

CXCL16 induces proliferation, phosphorylation of ERK1/2 and angiogenic sprouting in human retinal microvascular endothelial cells (HRMECs). HRMECs were stimulated with control medium alone or control medium with varying concentrations of CXCL16 (1 to 10 ng/ml) or with 10 ng/ml VEGF as a positive control. The number of metabolically active cells was quantified after 48 h using the ATPlite assay and expressed relative to control. (n=2 to 4, in quadruplicate) (A). The amount of phospho-ERK1/2 in HRMECs stimulated with CXCL16 (1 or 5 ng/ml) or VEGF (30 ng/ml) for 15 min was measured through enzyme-linked immunosorbent assay (ELISA). Relative phospho-ERK1/2 levels compared to unstimulated control are shown. (n=4, in duplicate or triplicate) (B); HRMEC spheroids were treated with medium (control), 0.3 or 3 ng/ml of CXCL16. After 14 h of incubation, spheroid sprouting was assessed in two independent experiments. Representative images of a control spheroid and a spheroid stimulated with 3 ng/ml CXCL16 are shown (C). For each condition (control, n=17; 0.3 ng/ml, n=17; 3 ng/ml, n=14), the number of sprouts per spheroid was counted (D). All results are expressed as median (interquartile range) ((*p < 0.05, **p < 0.01,***p < 0.001, ****p < 0.0001, Mann-Whitney test).

Figure 15

Western blot analysis of rat retinas. Rats received intravitreal injection of human CXCL16 (5 ng/5 μl) in the right eye and phosphate-buffered saline (PBS) in the left eye. The expression levels of phospho-ERK1/2 (A), the p65 subunit of NF-κB (B), intercellular adhesion molecule-1 (ICAM-1) (C) and vascular endothelial growth factor (VEGF) (D) were assessed by Western blot analysis after 24 h (A, B) or 4 days (C, D). Results are expressed as mean ± standard deviation. *p < 0.05 (independent t-test) compared to the values obtained from PBS-injected eyes.