Inflammation involving vessels and neural tissue occurs early in diabetic retinopathy, while excessive and persistent microglial activation is believed a major contributor to the inflammatory responses [2, 36, 37]. Microglial activation occurs via a complex regulatory system involving multiple signals; however, the mechanism remains to be determined.
Amadori-glycated proteins and the early nonenzymatic glycation of proteins are increased in diabetes and have been independently associated with microvascular complications [14, 15, 38]. As the breakdown of the blood-retinal barrier (BRB) is a characteristic early event, the increased GA in plasma may penetrate from the compromised BRB and accumulate in retinal parenchyma, influencing the types of retinal cells and molecular mediators that subsequently promoting the development of diabetic retinopathy [17, 18, 39, 40]. Our data suggested that in vitro GA activate microglia, producing proinflammatory factors. This event may be expected to occur in vivo, thereby aggravate the pathological inflammatory process. It was supported by the recent investigation that inhibiting the formation of GA would ameliorate the development of diabetic retinopathy .
The GA used in this study was produced commercially less than 1 week prior and purified to exclude residual contamination with AGEs as described by Baynes et al . In contrast to AGEs, only a few studies have emphasized the role of GA, despite the fact that the Amadori product is a major form of glycated proteins and the concentration of GA exceeds that of AGEs [14, 42]. Amadori-modification is structurally distinct from AGEs, and can bind to monocytes/macrophages via specific receptors [19, 20]. These receptor proteins have been characterized and the amino acid sequence homologies were not found in any of the AGE receptors [19–21, 43]. Therefore, it is reasonable that binding of GA to microglia via specific receptors rather than receptors of AGEs, induced microglial activation.
Accumulating evidences showed M-CSF is a key cytokine in the regulation of the microglial activation, proliferation and migration in CNS [23–25]. Moreover, the high-affinity binding of M-CSF to CSF-1R plays a role in mononuclear/macrophage-associated diabetic complications, including diabetic retinopathy [32, 33, 44, 45]. Our previous study provided the evidence that the vigorous expression of M-CSF/CSF-1R occurred in the early diabetic retina and a robust induction of CSF-1R was observed on the activated microglia . Furthermore, as shown in the current study, primary retinal microglial cells express either CSF-1R or M-CSF, albeit at relatively low baseline levels. While they were exposed to gradient concentrations of GA, the enhanced expression of CSF-1R signaling was detected in activated microglia, not only at the mRNA but also at the protein level. This up-regulation is consistent with that as seen in vivo, and may be integral to diabetic microglial activation . In light of the above findings, we hypothesize that M-CSF signaling is a possible molecular pathway in diabetic retinopathy.
To further our understanding, we measured the amounts of TNF-α and IL-1β, two important proflammatory cytokines secreted by activated microglia, in GA-stimulated microglial cultures treated with or without additional M-CSF. Our results show that M-CSF enhances the production of proflammatory cytokines in GA-activated microglia. M-CSF alone, however, did not induce the proinflammatory responses but proliferate microglia as shown by MTT assay. Although theoretically, more cells due to M-CSF treatment may contribute to more cytokine production, given the fact that this inducible proinflammation could be significantly inhibited by antibody neutralization of M-CSF or CSF-1R, and the fact that GA enhanced expression of M-CSF/CSF-1R by microglia, it is reasonable that this is quite far from the actual synergistic effects of the two agents. Taken all together, we concluded that M-CSF may be a co-stimulator with GA and that M-CSF/CSF-1R signaling exerts a synergistic effect with GA on the production of proinflammatory cytokines, not simply due to effects on cell proliferation.
Astrocytes, a major source of M-CSF in CNS, can be induced to release M-CSF by proinflammatory cytokines [47–50]. However, note that M-CSF is also secreted by activated microglial cells [26, 28]. Moreover, as determined by immunocytochemistry and ELISA, the level of M-CSF elevates in GA-stimulated-microglia. Elevated M-CSF expression could cause further reactive microgliosis, phagocytosis, and the release of inflammatory cytokines such as IL-1β, IL-6, and M-CSF [23–25]. Thus, the present study provides evidence for the roles of M-CSF and its signaling cascade in activated microglia in response to GA stimulation, suggesting that M-CSF is an important cross-talk mediator, involved in astrocytes, neurons, and microglia.