cells [64]. We also demonstrated that glial cells in the retina show constitutive expression of GDNF, suggesting that retinal glia potentially regulates the permeability of the BRB [65]. In addition, AGEs increase the expression of VEGF while simultaneously decreasing GDNF expression from glial cells [4]. Additionally, they induce apoptosis in pericytes in diabetic retinopathy. These findings suggest that AGE-mediated phenotypic alterations of glial cells in hyperglycemia result in an increase of the vascular permeability of the BRB in vitro and lead causally to BRB breakdown in the diabetic retina [4].

APKAP12

A-kinase anchor protein 12 (APKAP12) is a putative tumor suppressor linked with protein A, and protein kinase C serves as a scaffolding protein in signal transduction. Src-suppressed C-kinase substrate (SSeCKS), the rodent ortholog of human AKAP12, is identified to be important for mouse brain homeostasis by regulating BBB formation [66]. Recently, VEGF has been reported to be downregulated by A-kinase anchor protein 12 (APKAP12), which in turn causes upregulation of angiopoietin-1 in glia cells [67]. Thus, it is suggested that APKAP12 may be involved in the BRB formation through antiangiogenesis and barriergenesis during the retinal development, and its defect can lead to a loss of tight junction components resulting in BRB dysfunctions.

IL-6

IL-6 is a cytokine that functions in inflammation and the maturation of B cells. IL-6 is primarily produced at sites of acute and chronic inflammation, where it is secreted into the serum and induces a transcriptional inflammatory response through the IL-6 receptor alpha. The functioning of IL-6 is implicated in a wide variety of inflammationassociated disease states, such as diabetes mellitus and systemic juvenile rheumatoid arthritis [47]. Similar to TNF-a, intravitreal injection of IL-6 has been reported to induce an ocular inflammation by breaking the BRB [68].

A POSSIBLE TREATMENT OF THE RETINOPATHY WITH RETINOIC ACID ANALOGUES

Retinoic acid (RA) is an established signaling molecule that is involved in a variety of neuronal functions, such as the development, regeneration, and maintenance of the nervous system [69, 70]. Such RA signaling is thought be assessed by binding to a transcription factors comprising the heterodimer of the RA receptor (RAR) and retinoic X receptor (RXR). In each receptors, three genes (a, b, and g) are present, and together, the heterodimeric pair binds to a DNA sequence termed as a retinoic acid–response element (RARE). In addition to ligand binding, phosphorylation of the receptors and recruitment of coactivator or cosuppressors are required for the induction or suppression of gene transcription [71]. At present, more than 500 genes have been identified as RA-responsive [72].

Thang et al. reported that RA also plays a pivotal role in the induction of GDNF expression and its responsiveness in rat superior cervical ganglia [73]. This allows us to speculate that RA may also enhance GDNF expression in the retina and affect the barrier

function of TJ in the BRB resulting in suppression of the vascular permeability. Consistent with this hypothesis, real-time PCR, semiquantitative RT-PCR, and ELISA demonstrated significant upregulation of GDNF and downregulation of VEGF by all-trans RA (atRA), a RAR pan-agonist and Am580 (4(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl- 2-naphtamido) benzoic acid) in glial cells. In contrast, such effects were not observed by 9-cis-RA, an RXR agonist, or RAR or RXR antagonists. In addition, RARa agonists enhanced the expression of glial fibrillary acidic protein (GFAP), an intermediate filament protein that is thought to be specific for glial cell in central nervous system and glial cells in the retina (Fig. 2).

We recently demonstrated that GDNF secreted from glia cells plays an important role in the regulation of vascular permeability of the BRB and the BBB in a biological unit comprised of capillary endothelial cells and glial cells [5, 6]. As shown in Fig. 3, recombinant GDNF and RARa stimulants significantly enhanced the TER and inhibited the flux through endothelial cells, which indicates enhancement of the permeability of the BRB. Furthermore, these effects were affected by the addition of GDNF-specific siRNA, which selectively silenced the constitutive expressed GDNF in glial cells. Upon systemic administration of RARa stimulants to a mouse model with diabetic retinopathy, vascular leakage of the mouse retina was significantly reduced (Fig. 4). Taken together, this RARa-mediated enhancement of the barrier function of the BRB is sufficient for significant reductions of vascular leakage and angiogenesis in the diabetic retina, suggesting that RARa significantly antagonizes the loss of TJ integrity induced under diabetes. As expected, upon administration with RARa stimulants, the expression levels of endothelial TJ proteins such as claudin-5, a major determinant of vascular permeability; occludin; and ZO-1 were markedly increased, indicating that RARa stimulants regulate barrier functions through modulation of expression of a number of TJ-associated genes [21]. Thus, it is very likely that RAs upregulate expression of GDNF in glial cells and GDNF then induces the TJ-associated gene-expression alterations in endothelial cells.

Regarding possible molecular mechanisms of RA-dependent upregulation of GDNF, it has been reported that RARa transcriptionally may stimulate GDNF expression through the p300/CREB-binding protein (CBP)–signal transducer and the activator of the transcription 3 (STAT3) pathway [21]. Consistently, as indicated in Fig. 4, we found that the treatments with atRA and Am580 remarkably increased the levels of p300/CBP, STAT3, smad1, Notch, Hes-1, and Hes-5 mRNA in glial cells. To confirm this possible mechanism responsible for the RAs-mediated GDNF upregulation, a ~1.8 kb putative promoter fragment including the transcription initiation codon was isolated and made into a deletion mutant (~1.2 kb) that lacked putative p300-binding motifs for promoter assay. atRA and Am580 significantly enhanced the promoter activity of GDNF, whereas a deletion mutant showed a marked decrease of the promoter activities. Furthermore, p300 was selectively recruited to the GDNF promoter after treatments with RAs, indicating that the expression of GDNF is exclusively regulated through the recruitment of an RARa-driven trans-acting coactivator to the ~1.8 kb 5¢-flanking fragment of the GDNF promoter.

Fig. 3. Glial cell–derived cytokines regulate the vascular permeability in vitro. (A) Semiquantitative RT-PCR analysis showing that expression of GDNF and VEGF is modulated in human astrocytes after treatments with 100 nM atRA and 10 nM Am580. RAs such as atRA and RARa stimulants Am580 upregulate GDNF mRNA expression and conversely decrease VEGF. (B) atRA and Am580-mediated gene-expression alteration is sufficient to promote endothelial barrier function. Primary cultures of bovine brain microvascular endothelial cell were grown to confluence on transwell semipermeable membranes (pore size, 0.4 mm). In our coculture experiments, glial cells cultured in the lower chamber of the transwell were treated with 100 nM atRA or 10 nM Am580 for 8 h and cocultured with endothelial cells that were grown to confluence on transwell membranes in the upper chamber. Transendothelial electrical resistance (TER) was measured using an EVOM voltohmmeter, and electric resistance was expressed in standard units of W cm2. Paracellular tracer flux was measured by applying [14C]-mannitol at 1 × 105 dpm/well and [14C]-inulin at 5 × 105 dpm/well onto an endothelial monolayer in the apical compartment, and the samples were collected from the basolateral compartment in a time-dependent manner. Radioactivity of [14C] was counted by scintillation counter. Group 1: cells treated with vehicle only; Group 2: cells treated with atRA; Group 3: cells treated with Am580. #: p < 0.05, vs. cells treated with vehicle.

CONCLUSION

In this chapter, we described the BRB under physiological and diabetic conditions. Three conclusions reached are as follows: (1) The BRB is composed of glia and endothelial cells. The relationship between these cells is deeply functional as well as anatomical.

(2) The barrier function of endothelial tight junctions, in terms of permeability of the BRB, is predominantly regulated by cytokines derived from glial cells. This fact clearly shows that glial cells are a promising therapeutic target of diabetic retinopathy, even at an

Fig. 4. RARa-mediated phenotypic transformation of glial cells antagonizes the loss of TJ integrity induced under diabetes. C57BL/6 male mice (5 weeks old) were intraperitoneally injected with 40 mg/kg streptozotocin for 5 consecutive days. Fourteen weeks after the verification of diabetes, mice were treated with 1.0 mg/kg atRA every day or 3.75 mg/kg Am580 every other day for 1 week. To examine the leakage of retinal vessels, we injected 50 mg/kg fluorescein isothiocyanate (FITC) dextran dissolved in saline into mice via the vena cava, and the mice were sacrificed and the bilateral eyes enucleated 5 min after the FITC injection under general anesthesia. FITC concentration was measured using right eye. Left eyes were flat mounted, and the FITC dextran–perfused retinas were analyzed by laser-scanning confocal analysis. To provide a quantitative control, the FITC concentration in cardiac blood of each mouse was calculated. (A) Blood sugar (BS) and urinal sugar (US) were increased in diabetic mice. US was assessed as follows: score 0, negative (−); score 1, slightly positive (±); score 2, weakly positive (+); score 3, moderately positive (++); and score 4, strongly positive (+++). Note that RAs did not affect these parameters, indicating evidence that RA is not a drug for diabetes. (B) Western blot analysis to demonstrate the increase of GDNF and decrease of VEGF expression in the mouse eye by the treatment of RAs. (C, D) FITC leakage from diabetic retina was assessed by quantification of FITC (C) and laser-scanning confocal microscope (D). FITC leakage is clearly observed in diabetic mice; however, phenotypic alterations mediated by RARa were sufficient for inhibiting the vascular leakage to maintain vascular integrity in the retinal microenvironment. Scale bars, 100 mm. Group 1: control animals; Group 2: diabetic mice without the treatment; Group 3: atRA-treated diabetic mice; and Group 4: Am580-treated diabetic mice. *: p < 0.05, vs. control animals; #: p < 0.05, vs. animals treated without RAs.