Chemokine mediated monocyte trafficking into the retina: role of inflammation in alteration of the blood-retinal barrier in diabetic retinopathy

Abstract

Inflammation in the diabetic retina is mediated by leukocyte adhesion to the retinal vasculature and alteration of the blood-retinal barrier (BRB). We investigated the role of chemokines in the alteration of the BRB in diabetes. Animals were made diabetic by streptozotocin injection and analyzed for gene expression and monocyte/macrophage infiltration. The expression of CCL2 (chemokine ligand 2) was significantly up-regulated in the retinas of rats with 4 and 8 weeks of diabetes and also in human retinal endothelial cells treated with high glucose and glucose flux. Additionally, diabetes or intraocular injection of recombinant CCL2 resulted in increased expression of the macrophage marker, F4/80. Cell culture impedance sensing studies showed that purified CCL2 was unable to alter the integrity of the human retinal endothelial cell barrier, whereas monocyte conditioned medium resulted in significant reduction in cell resistance, suggesting the relevance of CCL2 in early immune cell recruitment for subsequent barrier alterations. Further, using Cx3cr1-GFP mice, we found that intraocular injection of CCL2 increased retinal GFP+ monocyte/macrophage infiltration. When these mice were made diabetic, increased infiltration of monocytes/macrophages was also present in retinal tissues. Diabetes and CCL2 injection also induced activation of retinal microglia in these animals. Quantification by flow cytometry demonstrated a two-fold increase of CX3CR1+/CD11b+ (monocyte/macrophage and microglia) cells in retinas of wildtype diabetic animals in comparison to control non-diabetic ones. Using CCL2 knockout (Ccl2-/-) mice, we show a significant reduction in retinal vascular leakage and monocyte infiltration following induction of diabetes indicating the importance of this chemokine in alteration of the BRB. Thus, CCL2 may be an important therapeutic target for the treatment of diabetic macular edema.

Figure 1. Growth factors and chemokines are upregulated in diabetic rats.

PCR gene arrays of 4 week (A and C) and 8 week (B) diabetic rats demonstrate significant upregulation of a subset of angiogenesis and chemokine genes in comparison to non-diabetic animals. CCL2 is the most remarkably increased in diabetic animals. * Significantly greater than control non-diabetic animals (p<0.05). CCL2: Chemokine Ligand 2; VEGFα: Vascular Endothelial Growth Factor alpha; TNF: Tumor Necrosis Factor; Ang2: Angiopoietin-2; CCL5: Chemokine Ligand 5; CCL7: Chemokine Ligand 7.

Figure 2. Diabetes or intraocular injection of CCL2 increases F4/80 expression in rat retinas.

Levels of the monocyte/macrophage associated marker F4/80 measured by qRT-PCR in 4 week non-diabetic or diabetic rat retinas and rats receiving an intraocular injection of either PBS or CCL2 (n = 4 animals in each group). * Significantly greater than control non-diabetic (p = 0.01). ** Significantly greater than PBS injected (p = 0.035).

Figure 3. Diabetes or intraocular injection of CCL2 increases monocyte/macrophage infiltration and activates retinal microglia.

Representative confocal images of retinal whole mounts from Cx3cr1-GFP mice. Retinas were counterstained with TRITC-labeled GS-IB4 isolectin (red) for vessel and activated monocyte/macrophage identification. (A) Normal non-diabetic mice showed isolectin labeled vessels and GFP⁺ retinal microglia (green) (arrowheads) uniformly distributed with ramified processes. (B) Retina from a 4 week diabetic animal demonstrates several GFP⁺/isolectin⁺ round cells closely associated with the external surface of the vessels (arrows). In addition, GFP⁺ microglial cells exhibit altered morphology typical of an activated state (arrowheads). (C) High magnification image of the retina from a diabetic animal demonstrates co-localization of GFP and isolectin to several of the extravascular monocytes/macrophages within the retinal tissue (arrows). As the isolectin stains activated monocytes in addition to blood vessels, some of the round cells (monocytes/macrophages) also stain red. Further, several GFP⁺ monocytes/macrophages presumably in the process of diapedesis are seen (double arrowheads). (D) Mouse retina 16 hours after intraocular injection of purified CCL2 (5 ng). Numerous GFP⁺/isolectin⁺ activated monocytes/macrophages (arrows) and microglia (arrowheads) are present within the extravascular space, with an altered morphology.

Figure 4. Diabetic mice retinas have significantly higher numbers of CX3CR1⁺/CD11b⁺ macrophage/microglia cells.

Representative dot plots of flow cytometry analysis of unstained (A), stained control non-diabetic (B) and 8 week diabetic wildtype mice (C) retinal cell suspensions. Gating was done to include monocyte/macrophage and microglia cells labelled CX3CR1⁺/CD11b⁺. Graphic representation of the total number of CX3CR1⁺/CD11b⁺ cells in diabetic mice demonstrates a two-fold increase in comparison to non-diabetic (D). * Significantly greater than control non-diabetic (p<0.05).

Figure 5. Human retinal endothelial cells (HRECs) express and secrete CCL2 in response to hyperglycemic conditions.

HRECs were incubated in normal glucose (5.5 mM), the osmotic control mannitol (25.5 mM), high glucose (30.5 mM) or glucose flux for 7 days. CCL2 gene expression was analyzed by qRT-PCR (A), and protein secretion by ELISA (B). High glucose and glucose flux significantly upregulated mRNA and protein levels of CCL2 in HRECs. * Significantly greater than low glucose or mannitol-treated cells (p<0.003).

Figure 6. Macrophage conditioned media, not CCL2, alters the integrity of the retinal endothelial barrier.

Preformed monolayers of HRECs were incubated with increasing concentrations of CCL2 and VEGF (A); or human macrophage (PBMCs) conditioned medium (B) and the normalized resistance was determined. No significant change was seen in response to treatment with the different concentrations of CCL2, however, cells incubated with VEGF demonstrated a significant drop in resistance over time. When macrophage conditioned medium was added to endothelial cells, barrier resistance started to decrease significantly after an initial period of 14 hours.

Figure 7. Diabetic Ccl2^−/− mice show a reduction in retinal vascular permeability.

The modified Evans Blue dye technique was performed in wildtype and Ccl2^−/− mice with 6 weeks of diabetes. Wildtype mice demonstrate a nearly 4-fold increase in retinal vascular permeability compared to non-diabetic. The degree of permeability is significantly reduced in Ccl2^−/− mice. *Significantly greater than wildtype non-diabetic (p = 0.001). **Significantly less than wildtype diabetic animals and significantly greater than non-diabetic controls (p<0.05).

Figure 8. Diabetic retinas of Cx3cr1-GFP/Ccl2^−/− mice show significant reduction of activation of microglia and monocyte/macrophage infiltration.

Representative confocal images of retinal whole mounts from Cx3cr1-GFP (A) and Cx3cr1-GFP/Ccl2^−/− mice with 6 weeks of diabetes (B) counterstained with TRITC-labeled GS IB4 isolectin for vessel and activated macrophage identification (red). (A) Retinas of diabetic Cx3cr1-GFP mice show activated microglia with ameboid morphology (typical of an activated state) and numerous GFP⁺/isolectin⁺ round monocytes/macrophages (arrows). (B) Diabetic Cx3cr1-GFP/Ccl2^−/− mice show a significant reduction in monocyte/macrophage infiltration. Very few monocytes/macrophages are seen in these retinas (arrows) as opposed to numerous quantities typically seen in diabetic animals (Cx3cr1-GFP). Retinas show GFP⁺ microglia uniformly distributed with ramified processes, similar to inactivated cells.