The Proinflammatory and Proangiogenic Macrophage Migration Inhibitory Factor Is a Potential Regulator in Proliferative Diabetic Retinopathy

The macrophage migration inhibitory factor (MIF)/CD74 signaling pathway is strongly implicated in inflammation and angiogenesis. We investigated the expression of MIF and its receptor CD74 in proliferative diabetic retinopathy (PDR) to reveal a possible role of this pathway in the pathogenesis of PDR. Levels of MIF, soluble (s)CD74, soluble intercellular adhesion molecule-1 (sICAM-1) and vascular endothelial growth factor (VEGF) were significantly increased in the vitreous from patients with PDR compared to nondiabetic control samples. We detected significant positive correlations between the levels of MIF and the levels of sICAM-1 (r = 0.43; p = 0.001) and VEGF (r = 0.7; p < 0.001). Through immunohistochemical analysis of PDR epiretinal membranes, significant positive correlations were also found between microvessel density (CD31 expression) and the numbers of blood vessels expressing MIF (r = 0.56; p = 0.045) and stromal cells expressing MIF (r = 0.79; p = 0.001) and CD74 (r = 0.59; p = 0.045). Similar to VEGF, MIF was induced in Müller cells cultured under hypoxic conditions and MIF induced phosphorylation of ERK1/2 and VEGF production in Müller cells. Intravitreal administration of MIF in normal rats induced increased retinal vascular permeability and significant upregulation of phospho-ERK1/2, NF-κB, ICAM-1 and vascular cell adhesion molecule-1 expression in the retina. MIF induced migration and proliferation of human retinal microvascular endothelial cells. These results suggest that MIF/CD74 signaling is involved in PDR angiogenesis.


INTRODUCTION
Ischemia-induced retinal angiogenesis and excessive deposition of extracellular matrix lead to the formation of fibrovascular membranes at the vitreoretinal interface in proliferative diabetic retinopathy (PDR). This outgrowth of fibrovascular tissue, composed of new blood vessels, leukocytes and α-smooth muscle actin (α-SMA)-expressing myofibroblasts (1)(2)(3)(4), often causes serious vision loss due to recurrent vitreous hemorrhage and/or traction retinal detachment.
Several studies support the paradigm that inflammation, neovascularization and fibrosis are critical mechanisms for PDR initiation and progression as the authors showed overexpression of inflammatory, angiogenic and fibrogenic factors inducing those processes (1)(2)(3)(4)(5).
In PDR, hypoxia seems to drive neovascularization through upregulation of angiogenic factors (6,7). In particular vascular endothelial growth factor (VEGF), upregulated in retinal cells in response to oxygen deprivation (8,9), plays a pivotal role in promoting retinal neovascularization and vascular leakage (10,11). In addition to angiogenesis, recruitment of leukocytes occurs in the ocular microenvironment of patients with PDR (1)(2)(3)(4)(5). Recent data support a causal relationship between persistent inflammation and angiogenesis (12,13) and this interplay might also be critical for PDR development and progression. Accordingly, some of the signaling molecules of the inflammatory response, such as cytokines, chemokines and their receptors might play an essential role in PDR angiogenesis and progression. Therefore, a new challenge in PDR research is the identification of the molecular links between inflammation and angiogenesis.
Macrophage migration inhibitory factor (MIF) is a widely expressed proinflammatory cytokine originally discovered as a product isolated from the culture medium of activated T lymphocytes that inhibited the random migration of cultured macrophages in vitro (14). Today a wide spectrum of biological properties has been attributed to MIF. MIF is closely involved in autoimmune and inflammatory diseases (15)(16)(17). The biological effects of MIF are mediated through its primary receptor CD74, which is the major histocompatibility class II-associated invariant chain (18). The binding of MIF to its receptor CD74 leads to the activation of extracellular signal regulated kinase (ERK) 1 and 2 and the proinflammatory transcription factor nuclear factor-κB (NF-κB) (15). Recently, it was demonstrated that the MIF/CD74 signaling pathway promotes macrophagemediated inflammation in type 1 diabetes (19). In addition, the chemokine receptors CXCR2 and CXCR4 were identified as functional receptors for MIF. By activating CXCR2 and CXCR4, MIF displays chemokine-like functions and stimulates leukocyte chemotaxis (20).
Given the key roles of the MIF/CD74 signaling pathway in angiogenesis and inflammation, we hypothesized that this pathway may be involved in the pathogenesis of PDR. To test this hypothesis, we investigated the expression of MIF and CD74 in the ocular microenvironment of patients with PDR and correlated their levels with the angiogenic activity in epiretinal fibrovascular membranes and the vitreous levels of VEGF and the inflammatory biomarker soluble intercellular adhesion molecule 1 (sICAM-1). We examined the effect of intravitreal administration of MIF on the retinas from normal rats. We analyzed in vitro the expression of MIF in human retinal Müller glial cells following exposure to hydrogen peroxide (H 2 O 2 )-induced oxidative stress and the hypoxia mimetic agent cobalt chloride (CoCl 2 ) and monitored expression of VEGF in Müller cells following exposure to MIF. Finally, we investigated the response of human retinal microvascular endothelial cells (HRMECs) to treatment with MIF.

Patient Samples
Undiluted vitreous fluid samples were obtained from 36 patients with PDR during pars plana vitrectomy for the treatment of tractional retinal detachment, and/or nonclearing vitreous hemorrhage and processed as described previously (1)(2)(3)(4)(5). The diabetic patients were 23 males and 13 females whose ages ranged from 27 to 74 years with a median [IQR] of 54 [44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59]. The PDR group consisted of 20 patients who had insulin-dependent diabetes mellitus and 16 patients who had non-insulin-dependent diabetes mellitus. Vitreous fluid samples obtained from 20 patients who had undergone vitrectomy for the treatment of rhegmatogenous retinal detachment with no proliferative vitreoretinopathy (PVR) were used as the control samples. Clinical check-up confirmed that control subjects were free from diabetes or other systemic disease. The controls were 14 males and 6 females whose ages ranged from 26 to 73 years with a median [IQR] of 55 . The ages (p = 0.526; Mann-Whitney test) and male/female ratios (p = 0.547; Chi-Square test) did not differ significantly between nondiabetic control patients and PDR patients.
Fourteen patients with PDR undergoing pars plana vitrectomy for the repair of tractional retinal detachment donated epiretinal fibrovascular membranes. At the time of the procedure, using previously published criteria, retinal neovascular activity was clinically graded (36). We made a distinction between active neovascularization (visible perfused new vessels on the retina or optic disc present within epiretinal membranes) and inactive involuted disease (nonvascularized, white fibrotic epiretinal membranes). For comparison, epiretinal fibrocellular membranes were obtained from ten patients without diabetes undergoing vitreoretinal surgery for the treatment of retinal detachment complicated by PVR. The epiretinal membranes were processed as previously described (1)(2)(3)(4).
The study was conducted according to the tenets of the Declaration of Helsinki. Before undergoing vitrectomy, all patients signed a preoperative informed written consent and approved the use of the excised epiretinal membranes and aspirated vitreous fluid for further analysis and clinical research. The Research Center and Institutional Review Board of the College of Medicine, King Saud University approved the study design and protocol. To identify the phenotype of cells expressing MIF and CD74, sequential double immunohistochemistry was performed. The sections were first incubated with anti-CD45, followed by treatment with peroxidase-conjugated secondary antibody and 3, 3 ′ -diaminobenzidine tetrahydrochloride substrate. Next, the second primary antibodies (anti-MIF or anti-CD74) were added and detected by alkaline phosphatase-conjugated secondary antibody and Fast Red reactions. No counterstain was applied. In negative controls, the incubation step with primary antibody was omitted from the protocol and only the ready-to-use antibody diluent (Cat No 52022; Dako) was applied.

Immunohistochemical Staining of Human Epiretinal Membranes and Quantitations
Immunoreactive blood vessels and cells were counted in five representative fields, using an eyepiece calibrated grid in combination with the 40x objective as previously described (1)(2)(3)(4). The level of vascularization in epiretinal membranes was determined by immunodetection of the vascular endothelium marker CD31.

Enzyme-Linked Immunosorbent Assays
Enzyme-linked Immunosorbent Assay (ELISA) kits for human MIF (Cat No DMF00B), human VEGF (Cat No SVE00) and human sICAM-1 (Cat No SCD540) were purchased from R&D Systems (Minneapolis, MN, USA). Levels of human MIF, VEGF and sICAM-1 in vitreous fluid and MIF and VEGF in culture medium were determined using those ELISA kits according to the manufacturer's instructions. The minimum detection limits for MIF, VEGF and sICAM-1 ELISA kits were 0.016 ng/ml, 9 pg/ml and 0.096 ng/ml, respectively.

Intravitreal Injection of MIF
Intravitreal injection into the eyes of Sprague Dawley rats (220-250 g) was performed as previously described (37). While the animals were kept under deep anesthesia, 5 µl of sterilized solution containing 5 ng recombinant MIF or sterile phosphate buffer saline (PBS) was injected into the right or left eye, respectively. Four days after intravitreal administration the rats were sacrificed, retinas were carefully dissected, snap frozen in liquid nitrogen, and stored at −80 • C.

Measurement of Blood-Retinal Barrier Breakdown
Blood-retinal barrier (BRB) breakdown in excised retinas was evaluated 4 days after intravitreal injection as previously described (37). Briefly, deeply anesthetized rats were intravenously injected with 50 mg/kg fluorescein isothiocyanate (FITC)-conjugated dextran (3-5 kDa, Sigma-Aldrich Corp., St. Louis, MO, USA). After 30 min, a blood sample was collected, and each rat was then perfused with PBS. The retinas were carefully excised, weighed and homogenized to extract the FITC-conjugated dextran. Fluorescence was measured using a spectra Max Gemini-XPS microplate reader (Molecular Devices, Sunnyvale, CA, USA) with excitation and emission wavelengths of 485 and 538 nm, respectively, with PBS as a blank. A correction for autofluorescence was made by subtracting the autofluorescence of retinal tissue from non-treated rats. The concentration of FITC-conjugated dextran in each retina was calculated from a standard curve of FITC-conjugated dextran in water. For normalization, the retinal FITC-conjugated dextran amount was divided by the retinal weight and by the concentration of FITC-conjugated dextran in the plasma. BRB breakdown was calculated using the following equation, with the results being expressed in µl/(g * h).

Western Blot Analysis of Human Vitreous Fluid, Müller Cell Lysates and Rat Retinas
Retina and cell lysates were homogenized in western blot lysis buffer [30 mM Tris-HCl; pH 7.5, 5 mM EDTA, 1% Triton X-100, 250 mM sucrose, 1 mM Sodium vanadate, and a complete protease inhibitor cocktail from Roche (Mannheim, Germany)]. After centrifugation of the homogenates (14,000 × g; 15 min, 4 • C), protein concentrations were measured in the supernatants (DC protein assay kit; Bio-Rad Laboratories, Hercules, CA). Equal amounts (50 µg) of the protein extracts were subjected to SDS-PAGE and transferred onto nitrocellulose membranes.

Chemotaxis Assay
Chemotaxis of HRMECs was evaluated with an xCELLigence system (ACEA Biosciences, San Diego, CA) as described before (38). Migration was evaluated in response to different concentrations of recombinant human MIF (Cat No 289-MF, R&D Systems), 10 ng/ml of VEGF (Cat No 583702, Biolegend, San Diego, CA) or dilution medium. First, 160 µl of stimulus diluted in MCDB131 medium (Gibco/ThermoFisher Scientific, Waltham, MA, USA) supplemented with 0.4% (v/v) fetal calf serum were added to the wells of the lower chamber of a Cell Invasion/Migration (CIM)-Plate (ACEA Biosciences). After assembly of the lower and upper chamber, 50 µl of serum-free MCDB131 medium was added in the upper wells. After equilibration of the plate (1 h at 37 • C), HRMECs were added in the upper chamber at 4 × 10 4 cells in 100 µl/well. After an additional incubation period (30 min at room temperature) to allow settling of the cells, migration was monitored every minute for 15 h in the xCELLigence apparatus. Cell migration from the upper to the lower compartment was recorded as changes in electrical impedance. These changes were converted into cell indices, as a measure of cell migration.

Proliferation Assay
To assess the proliferative effect of MIF, HRMECs were seeded at 5 × 10 3 cells in 100 µl/well of a 96-well plate in culture medium (vide supra). The next day, cells were washed with serum-free MCDB131 medium and stimulated in proliferation medium [MCDB131 medium supplemented with 2mM GlutaMAX TM (Gibco), 30 µg/ml Gentamicin and 3% fetal calf serum]. Either different concentrations of MIF, 10 ng/ml VEGF or proliferation medium were added to the wells. After 48 h, cell proliferation was assessed using the ATPlite Luminescence Assay kit (PerkinElmer, Waltham, MA) according to the manufacturer's instructions.

Statistical Analysis
Statistical analyses of the data were performed using SPSS version 21.0. Normal distribution of the data was verified using the Shapiro-Wilk (S-W) test and normal Q-Q plots. Normally distributed data were presented as mean ± SD and an independent t-test was used to compare the groups. For normally distributed data, Pearson correlation coefficients were calculated. Non-parametric tests (Kruskal-Wallis test, Mann-Whitney test and Spearman's correlation coefficients) were performed for not normally distributed data, which were presented as median and interquartile range [IQR; Q1-Q3]. Proportions were compared using the Chi-Square test. The level of statistical significance was set at 0.05.

Analysis of Angiogenic and Inflammatory Activities and the Expression of the Myofibroblast Marker α-SMA in Epiretinal Fibrovascular Membranes From Patients With PDR
As a negative control, the experimental staining procedure was performed with omission of the primary antibody and no staining was observed ( Figure 1A). Subsequently, we used staining for the vascular endothelial cell marker CD31, the leukocyte common antigen CD45 and α-SMA to evaluate ongoing angiogenesis, inflammation, and fibrosis, respectively. All membranes showed neovessels positive for CD31 (Figures 1B,C). Representative examples of CD31 staining in membranes with active ( Figure 1B) and involuted ( Figure 1C) disease are shown. Furthermore, leukocytes expressing CD45 (Figure 1D), as well as spindleshaped cells expressing α-SMA ( Figure 1E) were detected in all membranes. Immunohistochemical staining for CD31 showing pathologic new blood vessels expressing this endothelial cell marker in a membrane from a patient with active neovascularization (B) and in a membrane from a patient with involuted PDR (C). Note that the membrane from the patient with involuted PDR is composed mostly of fibrous tissue. Immunohistochemical staining for the leukocyte common antigen CD45 showing numerous leukocytes in the stroma (D). Immunohistochemical staining for α-smooth muscle actin (α-SMA) showing immunoreactivity in myofibroblasts (arrows) (E). (Scale bar, 10 µm).

Expression of MIF and Its Receptor CD74 in Epiretinal Fibrovascular Membranes From Patients With PDR
Next, immunohistochemical analysis was used to reveal whether MIF, an inflammatory and angiogenic molecule associated with tumor pathology, is also expressed in the context of PDR. MIF immunoreactivity was observed in all membranes. Figure 2 shows representative images of membranes derived from patients with active (Figure 2A) or involuted (Figure 2B) disease. Immunoreactivity for MIF was noted in both endothelial cells lining blood vessels (Figures 2A,B) and stromal cells (Figure 2C). In the stroma, MIF expression was detected in spindle-shaped cells, as well as in CD45-expressing leukocytes. In serial sections, the distribution and morphology of spindle-shaped cells expressing MIF ( Figure 2C) were similar to those of myofibroblasts expressing α-SMA (Figure 1E). Double immunostaining confirmed co-expression of CD45 and MIF in stromal and intravascular cells (Figures 2D,E).

Correlations Between Microvessel Density and the Expression of MIF and CD74 in Epiretinal Fibrovascular Membranes From Patients With PDR
Quantification of CD31-positive vessels in tumors is a standard method of measuring intra-tumoral microvessel density (MVD). Several studies reported that the level of MVD reflects the angiogenesis process in tumor tissues (22-24, 30, 31).
The mean number of blood vessels expressing CD31 was significantly higher in membranes from patients with active PDR than in membranes from patients with involuted PDR ( Table 1) (Figures 1B,C). The numbers of blood vessels expressing MIF and stromal cells expressing MIF and CD74 were significantly higher in membranes from patients with active PDR than in membranes from patients with inactive PDR ( Table 1) (Figures 2A,B).

Levels of MIF, VEGF, and sICAM-1 in Vitreous Samples
In addition, we used ELISA to compare MIF levels in vitreous samples from 36 patients with PDR to those of 20 nondiabetic   The angiogenic biomarker VEGF was detected in 9 of 20 (45%) vitreous samples from nondiabetic controls and in 33 of 36 (91.6%) vitreous samples from PDR patients. The proinflammatory biomarker sICAM-1 was detected in 18 of 20 (90%) vitreous samples from nondiabetic controls and in all vitreous samples from patients with PDR. The levels of VEGF and sICAM-1 were significantly higher in PDR (p < 0.001 for both comparisons; Mann-Whitney test) ( Table 3).
Significant positive correlations (Spearman's correlation coefficient) were found between vitreous fluid levels of MIF and levels of VEGF (r = 0.70; p < 0.001) and sICAM-1 (r = 0.43; p = 0.001). In addition, a significant positive correlation was observed between vitreous fluid levels of VEGF and the levels of sICAM-1 (r = 0.32; p = 0.023). Although  the correlations between MIF and sICAM-1 and between VEGF and sICAM-1 were weak, yet they were statistically significant (Figure 6).

Detection of sCD74 in Vitreous Samples
Using Western blot analysis, we demonstrated the presence of sCD74 in vitreous samples from patients with PDR. In agreement with previous studies (39,40), sCD74 protein was detected as two protein bands with molecular weights of around 30 and 40 kDa. Densitometric analysis of the bands demonstrated a significant increase in both the 30 kDa band (p = 0.002; Mann-Whitney test) and the 40 kDa band (p = 0.006; Mann-Whitney test) intensities in samples from PDR patients (n = 8) compared to samples from nondiabetic control patients (n = 8) (Figure 7).

MIF Induces Upregulation of VEGF and Phospho-ERK1/2 in Retinal Müller Cells
As we observed a positive correlation between the vitreous fluid levels of MIF and VEGF, we performed short-term induction experiments on Müller cells with MIF as an inducer of VEGF production. At 100 ng/ml, MIF significantly enhanced the levels of VEGF in the culture medium (Figure 8). However, 2 and 20 ng/ml MIF did not affect the expression of VEGF as compared to untreated control (Figure 8). Western blot analysis demonstrated that treatment of Müller cells with MIF (100 ng/ml) induced significant upregulation of the protein levels of phospho-ERK1/2 ( Figure 9A). In contrast, expression of the p65 subunit of NF-κB was not significantly altered (p = 0.057; Mann-Whitney test) ( Figure 9B).

The Hypoxia Mimetic Agent CoCl 2 Induces Upregulation of MIF and VEGF in Retinal Müller Cells
Retinal Müller cells are not only a major source of VEGF (41), these cells also produced enhanced amounts of MIF in addition to VEGF after treatment with 300 µM of CoCl 2 (Figure 10). However, oxidative stress, mimicked by addition of 100 µM of H 2 O 2 to the Müller cell cultures did not affect the expression of MIF as compared to untreated control (Figure 10).

MIF Induces Migration and Proliferation of Human Retinal Microvascular Endothelial Cells
As MIF has been described to be involved in pathological neovascularization of tumors, we tested in the following experiments whether HRMECs are also responsive to MIF. When added to the lower compartment of a chemotaxis chamber, MIF rather potently induced migration of the HRMECs. Indeed, the minimal effective dose required to trigger a significant chemotactic response was only 0.1 ng/ml. VEGF, however, was a more efficient chemoattractant ( Figure 11A). Furthermore, MIF also induced proliferation of HRMECs. When 0.01 ng/ml or higher concentrations of MIF were added to the endothelial cell cultures, proliferation increased on average to 60% above background ( Figure 11B).

In vivo Effect of Intravitreal Administration of MIF
Finally, 5 ng of MIF was injected in the eyes of normal rats. Fluorescein isothiocyanate-conjugated dextran was used to determine the subsequent change in vascular permeability. Figure 12A shows that intravitreal administration of MIF (n = 12) significantly increased retinal vascular permeability by about two-fold compared with vehicle (PBS)-injected eyes (n = 11). Furthermore, MIF induced significant upregulation of the protein levels of phospho-ERK1/2 (Figure 12B), the p65 subunit of NF-κB (Figure 12C), ICAM-1 (Figure 12D), VCAM-1 (Figure 12E), and VEGF ( Figure 12F) in the retinas, compared to the values obtained from the contralateral eyes that received PBS alone.

DISCUSSION
In the current study, we detected for the first time coexpression of MIF and its receptor CD74 in endothelial cells, leukocytes and myofibroblasts in epiretinal fibrovascular membranes from patients with PDR. In addition, there were  and CD74 expression by tumor-associated fibroblasts, leukocytes and endothelial cells (27). Several studies proposed that the MIF/CD74 signaling pathway is an important regulator in pathological tumor-associated angiogenesis (21)(22)(23)27) and showed that MIF expression levels correlated with tumor angiogenesis (30,31). In animal models of cancer, it was demonstrated that MIF had the potential to promote tumor growth and tumor-associated angiogenesis and that anti-MIF antibodies suppressed angiogenesis in these models (33). Several reports demonstrated that MIF directly and potently induces angiogenesis in multiple in vitro and in vivo models (24)(25)(26). In the present study, we demonstrated that MIF induced HRMEC migration and proliferation, early key steps during angiogenesis. These findings implied that MIF and CD74 might play roles as a ligand-receptor complex in PDR angiogenesis. Previously, it has been shown that MIF expression was also upregulated in an animal model of corneal neovascularization and that MIFdeficient mice had less neovascularization (42). In a mouse model of oxygen-induced ischemic retinal neovascularization, MIF deficiency reduced pathological preretinal angiogenesis and the expression of proinflammatory and proangiogenic factors (43).
In this study, we also demonstrated that MIF and sCD74 were significantly upregulated in the vitreous fluid from patients with PDR, and that vitreous MIF levels were significantly higher in PDR eyes with active neovascularization compared with eyes with quiescent disease. Our findings are in agreement with a previous study that demonstrated upregulated expression of MIF in the vitreous fluid from patients with PDR (44). Our analysis showed significant positive correlations between the vitreous fluid levels of MIF and those of the inflammatory biomarker sICAM-1 and the angiogenic biomarker VEGF. In addition, we demonstrated that intravitreal injection of MIF in normal rats induced significant upregulation of ICAM-1 and VCAM-1 in the retina. Our findings are consistent with previous reports documenting the role of MIF in upregulating ICAM-1 and VCAM-1 in different types of cells (45,46). In an animal model of retinal detachment, MIF was identified by a proteomics screen to be the most important cytokine upregulated in retinal detachment. Administration of a MIF inhibitor blocked pathological damage responses by protecting photoreceptors and reducing gliosis (47).
Among the proangiogenic factors, VEGF is considered as the most potent one with a pivotal role in PDR (10,11). To corroborate the findings at the cellular level, stimulation with MIF caused upregulation of VEGF in Müller cells. Müller cells are considered to contribute to pathological retinal neovascularization by being the principle VEGF-producing cell type (41). In addition, Matsuda et al. demonstrated expression of MIF in rat Müller cells (48). To our knowledge, the present study is the first to report the capability of MIF to target Müller cells and to induce upregulation of phospho-ERK1/2 and the synthesis and secretion of VEGF. Additionally, intravitreal injection of MIF induced a significant upregulation of VEGF, phospho-ERK1/2 and the p65 subunit of NF-κB in the retina of rats. This is in line with previous studies documenting the capacity of MIF to induce VEGF and phosphorylation of ERK1/2 in tumor cells (23,(28)(29)(30). These findings suggest that one possible mechanism of MIF-induced angiogenesis in PDR is related to the upregulation of VEGF. In addition, upregulation of VEGF is a major contributor to BRB breakdown in diabetes (49)(50)(51). Our findings also suggest that MIF-induced BRB breakdown might be related to upregulation of VEGF. Furthermore, we showed that stimulation with the hypoxia mimetic agent CoC1 2 promoted VEGF and MIF expressions in Müller cells. Similarly, previous studies demonstrated that MIF was induced by hypoxia in cancer cells (30,52,53).
In addition to its well characterized role in inflammation and angiogenesis, MIF is upregulated in fibrotic disorders, such as idiopathic pulmonary fibrosis (54,55) and systemic sclerosis (56). Additionally, MIF expression increased during the wound healing of rat skin injured by excision and anti-MIF antibodies induced a delay in wound healing (57). In the present study, immunohistochemical analysis FIGURE 12 | Macrophage migration inhibitory factor (MIF) induces blood-retinal barrier (BRB) breakdown and in vivo activation of inflammatory pathway. MIF (n = 12) was injected intravitreally at the dose of 5 ng in 5 µl in one eye and the same volume of phosphate buffer saline (PBS; n = 11) was injected in the contralateral eye of normal rats. BRB breakdown was quantified with the FITC-conjugated dextran technique (A). As demonstrated by Western blot analysis of rat retinas, intravitreal administration of MIF induced a significant upregulation of the expression of phospho-ERK1/2 (B), p65 subunit of NF-κB (C), intercellular adhesion molecule-1 (ICAM-1) (D), vascular cell adhesion molecule-1 (VCAM-1) (E) and vascular endothelial growth factor (VEGF) (F) compared with intravitreal administration of PBS. Results are expressed as median (interquartile range) (*p < 0.05 compared to the values obtained from PBS-injected eyes).
demonstrated MIF and CD74 localization in myofibroblasts, the key cellular mediator of fibrosis (58), in epiretinal membranes from patients with PDR and PVR. Similarly, dermal fibroblasts from skin wound lesions (57) and systemic sclerosis (56) produced higher amounts of MIF than normal dermal fibroblasts. It was demonstrated in vitro that transient exposure of fibroblasts to MIF induced fibroblast activation and promoted fibroblast proliferation, migration and replenishment of cell monolayers after scratching (59). In a recent study, an interesting genetic association of MIF with epiretinal membrane formation was found, suggesting a potential contribution of MIF to formation of the membranes (60).
In conclusion, the proinflammatory and proangiogenic cytokine MIF and its receptor CD74 are upregulated in the intraocular microenvironment of patients with PDR, particularly in patients with active angiogenesis. Additionally, intravitreal administration of MIF significantly increased retinal vascular permeability in rats. MIF stimulation of Müller cells induced increased secretion of VEGF and CoC1 2 induced the production of MIF from Müller cells in vitro. Stimulation of endothelial cells isolated from human retinas with MIF induced migration and proliferation, confirming its reported angiogenic effects. Therefore, the MIF/CD74 signaling pathway might play an important role in PDR angiogenesis and progression and could become a primary therapeutic target for improving the vascular function in patients with PDR.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The study was conducted according to the tenets of the Declaration of Helsinki. Before undergoing vitrectomy, all patients signed a preoperative informed written consent and approved the use of the excised epiretinal membranes and aspirated vitreous fluid for further analysis and clinical research. The Research Centre and Institutional Review Board of the College of Medicine, King Saud University approved the study design and protocol.

AUTHOR CONTRIBUTIONS
AMA designed the manuscript, supplied funding, interpreted the data, and wrote the manuscript. AA, MS, EA, and AD performed experiments and interpreted the data. PG analyzed the data. GD designed, supervised, and interpreted IHC stainings. JV designed experiments. GO provided funding and designed experiments. SS provided funding, designed experiments, interpreted the data, and wrote the manuscript. All authors read and approved the final manuscript.