response to the cytokine stromal-derived factor-1 (52), proliferation, adhesion, and incorporation into vascular structures (53). These deficiencies led to their inability to repair the retinal vascular damage characteristic of diabetes (54).

Thus, it appears that the diabetic retinal vasculature is being subjected to sustained stress, leading to the exhaustion of its inherent capacity for self-repair, conditions that are exacerbated by the reduction in replenishment from bone-marrow-derived endothelial precursor cells. Further work will be required to confirm that aberrant leukocyte adhesion is also responsible for the clinical pathology. One potential approach to testing this hypothesis in the clinic would be the development of agents that interfere with the ICAM-1 interaction that appears to be essential for the excess leukostasis observed in the rodent models.

INFLAMMATORY CELLS PROMOTE AND REGULATE THE DEVELOPMENT OF ISCHEMIC OCULAR NEOVASCULARIZATION

In addition to the damage to the existing retinal vasculature, DR may be accompanied by a proliferative neovascularization. While there is no animal model that accurately reproduces the course of proliferative DR, a retinopathy of prematurity model has been used in a number of studies to investigate the processes that regulate ischemic neovascularization. For these experiments, neonatal rats were raised in the presence of an elevated concentration of oxygen before returning to normal ambient oxygen levels (55). The inflammatory nature of the neovascular response to ischemia was suggested by immunohistochemical evidence demonstrating the presence of monocytes in the pathologic neovascular fronds. Further, neovascularization could be significantly suppressed by the depletion of monocyte lineage cells by intravitreous injection of clodronate liposomes (Figs. 8B,C) (55). In contrast, the physiologic vascularization that occurs during normal retinal development was affected only minimally by this treatment (Figs. 8B,D) (55). This suppression of pathologic neovascularization may reflect a role of infiltrating monocyte lineage cells in amplifying the response to ischemia (Figs. 8E–J) (55). An inherent potential for a positive feedback mechanism exists since VEGF induces activation (56, 57) and chemoattraction (56–58) of monocyte lineage cells, which in turn express and release it (59, 60), especially in conditions of hypoxia (Fig. 8K) (55). Evidence that monocytes play a role in pathologic angiogenesis has also been reported by other groups that have induced choroidal neovascularization by laser wounding; neovascularization was again suppressed by techniques that inhibited monocyte lineage cell recruitment, including their depletion with clodronate liposomes (61, 62) and genetic ablation of C-C chemokine receptor-2, the receptor for monocyte chemoattractant protein-1 (63).

In addition to the evidence implicating monocyte lineage cells in promoting pathologic vascularization, another population of inflammatory cells, T lymphocytes, was also mobilized (55). These cells were involved in the negative regulation of neovascularization, which was demonstrated by studies in which systemic injection of an antibody against CD2, a key adhesion molecule important for T lymphocyte-mediated responses, significantly increased pathologic neovascularization (Fig. 9A–C) (55). Immunocytochemical data demonstrated that these cells were positive for CD8 and CD25 antigens, characteristic of activated cytotoxic T lymphocytes (CTLs) (Fig. 9D–I) (55).

prevent the increase of endothelial cell death in these cocultures (Fig. 9J) (55). These data are consistent with the hypothesis that CTL-mediated apoptosis of endothelial cells through the Fas/FasL pathway is important in the negative regulation of pathologic neovascularization.

GROWTH FACTORS AS MEDIATORS OF INFLAMMATION

IN DIABETIC RETINOPATHY

Since McLeod et al. (33) reported that ICAM-1 expression was significantly elevated in the choroidal and retinal vasculature of diabetic patients, correlative studies have established that a variety of other inflammatory response mediators are also elevated in the ocular fluid or retinal tissue. These include VEGF (3,4), interleukin-1β (5), inter- leukin-6 (6–8), interleukin-8 (7, 8), stromal-derived factor-1 (9), angiotensin II (10), angiopoietin-1 (11), angiopoietin-2 (11, 12), erythropoietin (13, 14), tumor necrosis factor-α (TNF-α) (5, 15), monocyte chemoattractant protein-1 (16), and RANTES (Regulated on Activation, Normal T-cell Expressed and Presumably Secreted) (16). These studies, which focus on one or a few factors, are also being supplemented by more global approaches, including proteomic catalogs from the vitreous of diabetic human eyes (65, 66) as well as gene expression studies on Müller cells (27) and retinas (67) of diabetic rats. Elevations of a wide variety of proteins have been demonstrated in DR; the most recent of these reports has provided evidence that carbonic anhydrase, a heretofore unsuspected candidate, might be a viable molecular target for future therapies (66). Apart from VEGF and TNF-α, however, the evidence for the involvement of the majority of these proteins in DR remains correlative. The remainder of this discussion focuses on the evidence for the involvement of these two cytokines that have been examined in some detail in connection to their roles in DR-related inflammation. As VEGF is also the subject of other chapters in this book, this particular discussion will focus on the evidence for its proinflammatory actions.

VEGF as a Proinflammatory Factor in Diabetic Retinopathy

The involvement of VEGF in DR has been studied more extensively than that of any other single factor. This work has included numerous correlative studies reporting elevated VEGF levels in the vitreous of patients suffering from DR or diabetic macular edema (DME) (reviewed in Starita et al. (68) ). While the increases in VEGF are likely to reflect a number of processes, it is noteworthy that VEGF is upregulated by several factors that are themselves associated with DR, including reactive oxygen intermediates (69), advanced glycation end products (70), insulin-like growth factor-1 (71), and TNF-α (72). There is now increasing evidence that the actions of VEGF in promoting DR are inflammatory in nature.

Given the particular pathophysiology of DR, several properties of VEGF are especially relevant in its role as a proinflammatory agent. These include its many actions as a promoter of ocular neovascularization (73), its role as the most potent known promoter of vascular permeability (74), its expression by many retinal cell types (75, 76), and its upregulation by hypoxia (75, 77). In early studies, Tolentino et al. (78) found that intravitreous injection of VEGF in monkeys led to iris neovascularization, with the development