Fig. 1. Neural retina and the surrounding vasculature. The retina has two separate vascular systems: retinal and the choroidal vessels. The retinal vessels have tight endothelial barriers as also seen in the vessels of the brain, constituting the inner blood retinal barrier (BRB). In comparison, the outer BRB is comprised of the retinal pigment epithelium (RPE) together with the Bruch’s membrane, separating the leaky choroidal vessels from the neural retina.

cells in the blood. The BRB acts as an active regulatory interface, where transport of fluids, proteins, and cells in both directions takes place [3]. The integrity of BRB is essential for retinal neuronal health, and a compromised BRB is seen in various ocular diseases. The inner BRB is formed by normal retinal vessels, while the outer BRB is made by the retinal pigment epithelium (RPE) (Fig. 1). Cumulatively, these barriers regulate the flow of fluid, proteins, and cells into the extracellular space of the neural retina. Active transport mechanisms in the RPE result in a net fluid flow out of the neural retina [4]. Even under pathological conditions, RPE function can compensate for part of the leakage of vessels into the extracellular environment and reduce fluid accumulation in the outer retina.

THE INNER AND THE OUTER BRB

The inner BRB of the retinal vessels is similar to that in brain microvessels (Fig. 2). Various cellular components are needed to form such a barrier [5]. A milestone was the discovery that astrocyte end feet surround microvessels and that their connection to the endothelium induces various unique barrier properties in the endothelial cells [5]. These properties include high-resistance tight junctions between the capillary endothelial cells that impede the passive diffusion of solutes from the blood into the extracellular space [5]. Since then, much of the insight gained about vascular barriers comes from cell

Fig. 2. Schematic of the neurovascular barrier. This is a schematic showing the tight apposition of endothelial cells lining blood vessels in the brain. This is characteristic of the selective blood–brain barrier, which separates the circulation from brain parenchyma. Pericytes sheath the basement membrane covering the vascular endothelium.

culture models, in which endothelial cells are co-cultured with astrocytes or sometimes also with pericytes.

Changes of BRB in diabetes has long been of central interest. In DR, BRB breakdown causes protein and fluid extravasation, possibly leading to acute macular edema or longerterm neuronal damage. Therefore, elucidating the factors that compromise the BRB might lead to new therapeutic approaches for DR or diabetic macular edema, which is the main cause of visual loss in diabetic patients. BRB investigations in vivo are commonly studied in the streptozotocin (STZ)-induced diabetes in rats [6]. STZ, an antibiotic produced from Streptomyces achromogenes, enters the cytoplasm via glucose transporter (GLUT) 2, which is the b-cell’s GLUT in the pancreas [7], and reduces insulin secretion through b-cell toxicity [8]. STZ-injected animals rapidly develop hyperglycemia, resembling the conditions found in type 1 diabetes, and develop diabetic retinal vasculopathy, making them a convenient tool in the study of early diabetic changes. These animals develop some earlier vascular changes, such as increased retinal leukostasis, vascular leakage, or elevated cytokine expression. However, STZ-injected animals do not exhibit the entire pathology of the human DR. For instance, they do not show retinal neovascularization. Furthermore, the following metabolic disarray, including insulin resistance, dyslipidemia, and adipokine changes, is not truly reflected in STZ-induced diabetes. The recently introduced model of spontaneously occurring type 2 diabetes in the Nile grass rat (NGR) shows many pertinent characteristics of the human condition [9]. The hyperglycemia in NGR is accompanied by dyslipidemia and insulin resistance. Hope is great that with the help of such realistic models of human diabetes, effective mechanistic explorations as well as therapeutic advances will take place.

Due to the growing importance of age-related diseases, a large amount of interest lies in understanding the physiological changes of vascular barrier function during aging [10]. Recent work indicates a gradual and continuous decline in vascular barrier function with physiologic aging and that immune cells contribute to this process [11]. This indicates that the barrier-privileged vessels of the body, similar to other organs, are subject to changes resulting from age.

A plausible explanation for how physiologic aging might impact vascular barrier function comes from the observation that deficiency of a cholesterol transport protein, the apolipoprotein E (apoE), in mice substantially accelerates the barrier decay with age [11]. Since apoE−/− mice are prone to chronic vascular inflammation, such as accelerated atherosclerosis [12] and neurodegeneration [13], this indicates that chronic inflammation compromises vascular barrier privilege. Analogously, in normal animals, constitutive inflammatory processes during aging cause cumulative damage to the vasculature, which can be a prelude to age-related vascular diseases [11].

To investigate retinal vascular leakage in vivo, for instance in diabetes, many investigators use protein leakage assays, of which various modifications exist. These assays commonly quantify the passage of plasma albumin into the parenchyma. To do so, dyes such as Evans blue (EB) are injected into the circulation [14, 15]. Under controlled conditions, these techniques allow quantitative assessment of inner BRB leakage. However, due to the low amount of retinal tissue and the large variability between animals, albumin/protein-based leakage assays have limitations both in terms of sensitivity and in the large variability of the outcome. Therefore, there is currently a great need for more sensitive in vivo assays that can reliably quantify subtle leakage.

The outer BRB is primarily comprised of the RPE, a cellular layer that causes a tight epithelial barrier. The healthy RPE forms not only the outer BRB but also actively removes subretinal fluid, thus regulating fluid accumulation in the subretinal space. RPE function is essential to maintaining a balanced outer retinal environment. Moreover, the RPE is a principal source of angiogenic and antiangiogenic factors and also expresses the receptors for these agents.

Both acute and chronic inflammation disrupt the (BRB), as in uveitis or diabetic retinopathy, respectively [16]. These facts have led to the hypothesis that barrier changes in physiologic aging or in acute or chronic inflammation are related. Indeed, certain immune cells in the peripheral blood, neutrophils and macrophages, contain a highly potent permeability factor, azurocidin (AZ), that these cells release when interacting with the activated endothelium.

Inflammation and BRB Permeability

Leukocyte accumulation in retinal vessels is a critical early event in the pathogenesis of DR. Firm adhesion of neutrophils to the inflamed endothelium causes vascular leakage [17–19]. However, the molecular details are only beginning to be understood. Leukocyte accumulation on the inflamed endothelium of retinal vessels follows the general principles of cascade-like recruitment [20]. Leukocyte rolling, the initial step in the recruitment cascade, is followed by leukocyte activation, firm adhesion, and transmigration into the interstitial tissue [20]. The endothelium sequentially expresses adhesion molecules, such as selectins, integrins, and immunoglobulins, and presents

Fig. 3. Steps of inflammatory leukocyte recruitment. The transition from rolling to firm adhesion is achieved by endothelial intracellular adhesion molecule (ICAM)-1 that interacts with its leukocyte ligand, CD18 [23]. The retinal endothelium of diabetic animals expresses ICAM-1, which binds to leukocyte b2-integrins, LFA-1 (CD18CD11a) and Mac-1 (CD18CD11b), mediating firm leukocyte adhesion. Leukocytes use their integrins to extravasate through the extracellular matrix (ECM) [103].

chemoattractants to the free-flowing leukocytes to orchestrate each stage of the recruitment process [20, 21] (Fig. 3).

Selectins mainly mediate the first steps of the leukocyte-endothelial interaction [20]. Through their lectin domain, the selectins bind to other carbohydrates presented by mucins [22]. P-selectin is the first adhesion receptor transiently upregulated on the endothelium during inflammation, which initiates leukocyte rolling [21].

Leukocyte adhesion to the retinal vessels is critical for DR pathology, as inhibition of leukocyte adhesion through intracellular adhesion molecule (ICAM)-1 or b2-integrin blockade effectively suppresses vascular endothelial growth factor (VEGF)-induced and diabetic BRB breakdown, establishing the link between leukocyte adhesion and increased retinal vascular leakage [23, 24]. However, the molecular pathways involved in BRB breakdown downstream of leukocyte adhesion are only beginning to be understood.

When neutrophils and monocytes, two leukocyte subtypes, interact via their b2- integrins with ICAM-1 on activated endothelium, they release the content of their azurophilic granulae. One of the protein contents of these granulae, AZ, is a potent permeability factor [25]. Interestingly, b2-integrin expression on peripheral blood neutrophils is higher in diabetic animals [24]. Under these conditions, leukocytes are more prone to release AZ.