Macular edema

Macular edema is one of the leading causes of visual impairment in patients with uveitis or diabetes, and may also occur after cataract surgery [1–3]. Retinal ischemia, oxidative stress, and inflammation are pathogenic factors implicated in the development of macular edema [4–8]. Though promising new therapeutical approaches arose from the significantly increased knowledge regarding the factors that cause retinal edema and neovascularization [8–10], the pathogenic mechanisms underlying the development of macular edema are still not fully understood. One major pathogenic event in the development of edema is an increase in the permeability of the blood retinal barrier formed by vascular endothelial and retinal pigment epithelial cells [11]. The leakage from perifoveal vessels and pigment epithelium causes fluid accumulation in the foveal tissue and subretinal space, respectively, resulting in serous macular detachment and increased thickness of the retinal tissue due to diffuse or cystoid edema. The fluid accumulation contributes to the degeneration of retinal neurons and to a decrease in visual acuity, as it results in compression of neurons, nerve fibers, and vessels (which exacerbates ischemic conditions), and in an elongation of the routes for the diffusion of metabolic substrates and oxygen. Among the various vasoactive factors that induce vascular leakage (e.g., prostaglandins, interleukin-1ß, tumor necrosis factor [12]), the vascular endothelial growth factor (VEGF) is thought to be the major cytokine causing the breakdown of the blood-retinal barrier [8, 13–15]. However, the development of chronic edema depends on two factors: the rate of fluid entry from leaky vessels into the retinal tissue, and the rate of fluid absorption from the retina back into the blood. An impairment of the fluid absorption from the retinal tissue should be considered as a major pathogenic mechanism of edema formation, for example in patients which display macular edema without angiographic vascular leakage. In the preclinical stage of diabetic retinopathy, two types of increased retinal thickness exist that are or not associated with vascular leakage [16]. It has been shown that clinically significant diabetic macular edema develops only when (in addition to vascular leakage) the active transport mechanisms of the blood-retinal barrier are dysfunctional [17]. Obviously, any anomalies in vessel permeability need to be accompanied by ineffective edema-resolving mechanisms to cause chronic edema [18].

Macular edema-a Müller cell disease?

Müller cells, the principal glial cells of the retina, constitute a functional link between neurons and vessels. They support neurons with blood-derived nutrients, remove metabolic waste, and are responsible for maintaining the homeostasis of the retinal extracellular milieu (ions, water, neurotransmitter molecules, and pH) [19]. Retinal capillaries are closely ensheathed by glial cell processes arising from astrocytes and Müller cells. Müller cells become activated upon virtually all pathogenic stimuli [19]. In the retina of diabetic animals, for example, Müller cells become reactive at an early stage owing to disruption of the blood-retinal barrier [20, 21]. Normally, Müller cells participate in the establishment of the blood-retinal barrier [22]; however, under hypoxic conditions they impair the barrier function [23]. Under hypoxic, inflammatory or glucose-deprivation conditions, Müller cells secrete factors such as VEGF that increase the vascular permeability [24–27]. Müller cells are also a source of matrix metalloproteinases [28, 29] which impair the barrier function of retinal endothelial cells, by proteolytic degradation of the tight junction protein occludin [30]. Even under normoxic conditions, Müller cells secrete factors that decrease the barrier permeability, such as the pigment epithelium-derived factor (PEDF) [31] which downregulates the expression of VEGF. Under hypoxic conditions, the expression of PEDF is reduced in the retina and Müller cells [31, 32].

In addition to the effects on the barrier state of the retinal vasculature, Müller cells may also contribute to edema development by another way. There are electron microscopic studies carried out in the early 1980s which suggest that—in addition to ischemic changes in the retinal microvasculature—swelling (i.e., intracellular edema) of Müller cells may contribute to the development of cystoid macular edema, with the cysts being formed by swollen and necrotic Müller cells [33, 34]. In contrast, other authors did not find Müller cell swelling [35]. In the brain, swelling of astrocytes usually occurs concomitantly in vascular edema, and represents a major mechanism of edema formation under ischemic and various other conditions such as hyponatremia [36]. There are further data suggesting that dysfunctional Müller cells may be a causative factor in edema development. Dominantly inherited cystoid macular edema was suggested to represent a primary disease of Müller cells since degenerated Müller cells were found to be located around a virtually intact retinal vascular endothelium [37]. In an animal model of retinal hypoxia, it has been shown that vascular leakage is associated with cellular edema of Müller cells [38]. In the present review we discuss data which support the assumption that an impairment in the fluid absorption from the retinal tissue normally carried out by Müller cells may represent one causative factor in edema development. Under distinct circumstances, Müller cells may even swell.

Müller cells: fluid absorption from the retina

The fluid absorption from the subretinal space and from retinal tissue into the blood is normally carried out by pigment epithelial and glial cells. This is mediated via transcellular water transport that is osmotically coupled to the fluxes of osmolytes, particularly of ions [39, 40]. Fluid absorption is carried out by both cell types under normal conditions to redistribute water which accumulates in the retinal tissue and subretinal space due to various processes (Fig. 1): the influx of water from the blood into the retinal parenchyma which is coupled to the uptake of metabolic substrates such as glucose, and the formation of water within the retinal tissue which is associated with the aerobic energy production (per glucose molecule, 42 molecules of water are formed). The water fluxes through the cells are facilitated by the polarized intramembranous expression of specialized water transporting proteins, the aquaporins. Aquaporins facilitate bidirectional water movements across membranes, in dependence on the transmembranal osmotic gradient and hydrostatic pressure, and are involved in the maintenance of the ionic and osmotic balance in the tissue [41, 42]. Pigment epithelial cells express aquaporin-1 [43] whereas Müller cells express aquaporin-4 (Fig. 2a) [44].

Fig. 1
figure 1

Water fluxes through the retina. Under normal conditions, water accumulates in the neural retina and subretinal space due to influx from the blood (coupled to the uptake of nutrients such as glucose) and the oxidative synthesis of adenosine 5′-triphosphate (ATP) that generates carbon dioxide and water. The excess water is redistributed into the blood by transcellular water transport through Müller cells (yellow) and pigment epithelial cells (RPE). The transmembranous water transport is facilitated by aquaporin (AQP) water channels. RPE cells express AQP1, whereas Müller cells express AQP4. The transcellular water transport is osmotically coupled to the transport of osmolytes, especially of potassium and chloride ions. The ion fluxes across the cellular membranes are facilitated by transporter molecules and ion channels. In Müller cells, the Kir4.1 potassium channel is co-localized with AQP4 in membranes that surround the vessels, and at both limiting membranes (compare Fig. 2)

Full size image
Fig. 2
figure 2

Experimental diabetes in rats alters the retinal expression of Kir4.1 potassium channels. a Localization of Kir4.1 and AQP4 proteins in slices of the neural rat retina. The Kir4.1 potassium channel is co-localized with AQP4 water channels in Müller cell membranes that surround the vessels (arrows), and at both limiting membranes (arrowheads). AQP4 is additionally expressed by Müller cells in both plexiform layers, and in the ganglion cell layer. b The prominent expression of the Kir4.1 protein around the vessels and at both limiting membranes (control retina) is absent in the retina of a diabetic animal. c View on the deep vascular plexus in the inner nuclear layer (wholemount preparation). The vessels are surrounded by AQP4 and Kir4.1 proteins in the control retina, but only by AQP4 in the retina of the diabetic animal. GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; ONL, outer nuclear layer. Bars, 20 μm. With permission from ref. 52

Full size image

The transcellular water transport is tightly coupled to fluxes of osmolytes, in particular of potassium and chloride ions (Fig. 1) [39, 40, 45]. Activated neurons release potassium ions. To avoid potassium-induced depolarization of neurons which may cause neuronal hyperexcitation resulting in excess release of neurotransmitters and glutamate toxicity, Müller cells take up excess potassium from the extracellular space especially in the synaptic (plexiform) layers of the retina, and release a similar amount of potassium into spaces outside of the neural retina, especially into the blood and the vitreous [46]. This “spatial buffering” of extracellular potassium is mediated predominantly by passive currents through potassium channels localized in Müller cell membranes. Müller cells express different types of potassium channels. Under normal conditions, inwardly rectifying potassium (Kir) channels of the Kir2.1 subtype are expressed in neuron-abutting membranes through which Müller cells take up excess potassium. Kir4.1 channels are expressed in membranes which are in close contact with spaces outside of the neural retina, i.e., in Müller cell processes enwrapping blood vessels and at both limiting membranes of the retina (Fig. 2a) [40, 47, 48]. Kir2.1 channels mediate only inward potassium currents into the Müller cells, whereas Kir4.1 channels mediate bidirectional currents between the extra-retinal tissues and the Müller cell interior. The co-localization of Kir4.1 potassium channels and aquaporin-4 water channels around vessels and at the limiting membranes of the retina [40] indicates a coupling of water transport to potassium currents in Müller cells. Moreover, aquaporin-4 channels are expressed by Müller cells in both plexiform layers and in the ganglion cell layer (Fig. 2a) where the cells take up excess potassium through Kir2.1 channels.

Müller cells: impairment of fluid absorption

Since the Müller cell-mediated fluid absorption from the retinal tissue predominantly relies on channel-mediated co-transport of potassium ions and water, any dysfunction of potassium channels should result in an impairment of the retinal water clearance and, therefore, will contribute to the development of chronic edema. Indeed we found in various animal models of retinopathies, e.g., after transient ischemia of the retina [49, 50], during ocular inflammation [51], in the retina of diabetic animals [52], and after retinal detachment [53, 54], that a major response of Müller cells under pathological conditions is an alteration in the expression of Kir4.1 channels. The Kir4.1 channel protein is redistributed from its prominent expression sites around the vessels and at the limiting membranes of the retina, and displays an even distribution in the retina at low level (Fig. 2b,c). This disclocation of Kir4.1 may be associated with a decrease in retinal expression of Kir4.1 protein [49], and causes a general decrease in the potassium currents flowing through Müller cell membranes [50–54]. On the other hand, the localization of Kir2.1 channels does not change after retinal ischemia [48]. Thus, Müller cells can take up potassium ions (via Kir2.1 channels), but the extrusion of excess potassium by Müller cells into extra-retinal spaces is impaired under pathological conditions. Hence, this may lead to an intracellular potassium overload and an increase in the osmotic pressure of the Müller cell interior in the diseased retina. Disruption of glial potassium homeostasis due to the downregulation of Kir4.1 channels may contribute to the death of nerve cells in the retina under pathological conditions [55], and should disturb the transport of water through Müller cells. It has been shown that Müller cells of patients with proliferative diabetic retinopathy display a strong reduction of their potassium conductance [56], suggesting that the downregulation of potassium channels may represent a feature of Müller cells in the human retina, as well as in animal models.

Fig. 3
figure 3

Experimental diabetes in rats affects the osmotic swelling characteristics of Müller cells, suggesting an impaired water transport across Müller cell membranes. Triamcinolone acetonide inhibits the osmotic Müller cell swelling via stimulation of endogenous adenosine signaling. a Müller cell bodies in retinal slices from control animals do not change their size upon application of a hypoosmolar solution (60% of control osmolarity). In contrast, Müller cell bodies in retinal slices from diabetic animals increase their size under osmotic stress (see the insets for an example; bar, 5 μm). b The osmotic swelling of Müller cell bodies in retinas from diabetic animals is inhibited in the presence of triamcinolone acetonide (triam; 100 μM). c The swelling-inhibitory effect of triamcinolone is prevented by inhibitors of adenosine transporters (NBTI) and adenosine A1 receptors (DPCPX), respectively. With permission from ref. 52

Full size image
Fig. 4
figure 4

Putative mechanisms of Müller cell swelling and the effect of triamcinolone acetonide in the diabetic retina. a Under normal conditions, Müller cells mediate the fluid absorption from the retinal tissue into the blood by a co-transport of water (facilitated by AQP4 water channels) and osmolytes, especially potassium ions (facilitated by Kir4.1 channels). b In the diabetic retina, vascular leakage occurs due to the action of inflammatory mediators and VEGF. The Müller cells downregulate the expression of functional Kir4.1 channels. This downregulation impairs the release of potassium from Müller cells into the blood, and may result in an accumulation of potassium ions within the cells. The increase in the intracellular osmotic pressure results in an osmotic gradient across the plasma membrane that draws water into the Müller cells facilitated by AQP4 water channels. An influx of sodium ions into the cells, evoked by inflammatory mediators such as arachidonic acid (AA) and prostaglandins (PGs) which are formed in response to oxidative stress, contributes to the increase in intracellular osmotic pressure. The uptake of extravasated serum proteins by Müller cells may further increase the osmotic pressure of the cell interior. c Triamcinolone reduces edema by both inhibition of the fluid inflow from the vessels (through blocking the effects of inflammatory mediators and VEGF) and re-activation of the fluid clearance function of Müller cells. The latter action is mediated by triggering a release of endogenous adenosine and subsequent activation of adenosine A1 receptors which results in the opening of potassium channels (likely two pore domain [TASK] channels). The efflux of potassium ions is associated with a water transport out of the cells and, therefore, with a shrinkage of the cells. The triamcinolone-evoked stimulation of the potassium clearance by Müller cells facilitates the fluid absorption from the retinal tissue

Full size image

An increase in intracellular osmotic pressure and an impairment of the rapid release of ionic osmolytes will favor swelling of Müller cells under distinct conditions. Indeed, we found that the swelling characteristics of Müller cells from diseased retinas were altered. Under normal conditions, Müller cells do not swell when isolated retinas are perfused with a hypoosmolar solution (Fig. 3a). However, Müller cells in pathologically altered retinas display a time-dependent swelling of their cell bodies in a hypoosmolar environment. Osmotic Müller cell swelling was observed under all circumstances which are associated with a downregulation of Kir4.1 channels, i.e., in retinas of diabetic animals (Fig. 3a) [52], after retinal ischemia [49] or detachment [54], and during endotoxin-induced ocular inflammation [51]. There is a correlation between the extent of osmotic swelling and the decrease of potassium currents in Müller cells of detached retinas [54]. The swelling of Müller cells indicates that the water transport across Müller cell membranes in response to a transmembranal osmotic gradient is impaired due to the restriction of a rapid release of potassium ions after downregulation of functional Kir4.1 channels. Moreover, a causal relationship between the downregulation of functional potassium channels and the osmotic swelling of Müller cells is suggested by the observation that a blockade of potassium channels in Müller cells of control retinas, e.g., by barium ions, induces cellular swelling under hypoosmolar conditions [49]. Both, the low potassium channel expression and the inefficient cell volume regulation, resemble properties of undifferentiated Müller cells in the young postnatal retina [57], suggesting that Müller cells dedifferentiate under pathological conditions.

In addition to the decreased expression of functional potassium channels, two further pathogenic factors are implicated in the osmotic Müller cell swelling: inflammatory mediators and oxidative stress. Inhibition of enzymes which produce arachidonic acid and prostaglandins (phospholipase A2 and cyclooxygenase) prevents the osmotic swelling of Müller cells in pathologically altered retinas [52, 54, 58]. Likewise, inhibition of oxidative stress by application of a reducing agent impedes the swelling of Müller cells under hypoosmolar conditions. Conversely, the application of arachidonic acid or prostaglandin E2, or oxidative stress evoked by H2O2, induces osmotic swelling of Müller cells in control retinas [52, 54, 58]. Local inflammation and oxidative stress are pathological factors in retinal diseases known to be associated with edema, e.g., diabetic retinopathy [6, 8, 59–65]. Under normal conditions, Müller cell bodies in the porcine retina do not express immunoreactivity for cyclooxygenase-2, whereas after experimental retinal detachment, they express this enzyme [54], suggesting that (in addition to retinal neurons) Müller cells increase the expression of enzymes that produce inflammatory mediators. A similar up-regulation of inflammatory-related proteins in Müller cells has been described during experimental diabetes [64]. Retinas of diabetic animals display an increased expression of cyclooxygenase-2, and an increased production of prostaglandin E2 [66, 67]. Arachidonic acid and its metabolites, especially prostaglandin E2, are major mediators of macular edema [5, 7]. It is known that oxidative stress evoked by acute application of H2O2 mimicks the edema-inducing effect of retinal ischemia-reperfusion [68]. Therefore, radical scavengers which inhibit vascular leakage in the ischemic retina [69] should also inhibit Müller cell swelling.

Presumably, the osmotic swelling of Müller cells in injured retinas is caused by oxidative stress, resulting in activation of phospholipase A2 and cyclooxygenase, subsequent arachidonic acid- and prostaglandin-evoked lipid peroxidation, and intracellular sodium overload. Simultaneously, the compensatory channel-mediated efflux of potassium might be inhibited (Fig. 4b). Oxidative stress can initiate lipid peroxidation, resulting in the release of arachidonic acid from the cell membrane [70, 71]. Ischemia in the neural tissue causes formation of free polyunsaturated fatty acids, particularly arachidonic acid, from membrane phospholipids, mainly by the action of phospholipase A2 [72–75]. Free radicals and hydroperoxides produced during ischemia-reperfusion stimulate the activity of lipoxygenase and cyclooxygenase [76]. Arachidonic acid induces both vascular and cellular edema in the brain [77, 78] and evokes swelling of cultured astrocytes [79]. The increase in intracellular sodium, being an effect of arachidonic acid and prostaglandins, is mainly due to inhibition of the sodium pump activity (Fig. 4b) [80, 81]. A similar increase in intracellular sodium occurs during swelling of Müller cells [82]. However, the intracellular sodium overload leads to cellular swelling only when the cells lost their capability to release osmolytes. Under normal conditions, retinal glial cells release potassium ions via Kir4.1 channels and, thus, avoid swelling despite of their increased sodium content. However, when the channel-mediated release of potassium is disturbed, they swell. It is known that arachidonic acid blocks voltage-gated potassium channels in Müller cells [83] which are the only channels that (under normal conditions) mediate outward potassium currents across the Müller cell membrane after downregulation of Kir4.1 [84–86]. Arachidonic acid may also inhibit the swelling-activated efflux of other osmolytes such as amino acids and chloride ions [87].

The induction of osmotic Müller cell swelling, which is not observed in cells from control tissues, suggests that the water transport across Müller cell membranes in response to osmotic gradients is impaired under pathological conditions; a disturbance of this water transport should prevent the resolution of retinal edema. The data suggest that an insufficient fluid absorption by Müller cells, and even Müller cell swelling under distinct conditions associated with osmotic imbalances between retinal and extra-retinal tissues, may represent an important factor contributing to the development of edema.

Stimulation of fluid absorption

Chronic edema is caused by an imbalance between the fluid influx into the tissue and the fluid absorption from the tissue. Edema can be resolved by inhibition of vascular leakage or by stimulation of the fluid clearance. Hitherto, most of the research regarding clinically used edema-resolving therapies was focused on the inhibition of vascular leakage. Reduction in fluid extravasation can be obtained by inhibition of the release, or of the effects of the vessel-permeabilizing factors VEGF and inflammatory factors. VEGF inhibitors, e.g., bevacizumab, reduce the level of free VEGF. Ruboxistaurin blocks protein kinase-ß which is an intracellular mediator involved in VEGF-induced increase in vascular permeability [88]. The anti-inflammatory corticosteroid, triamcinolone acetonide, inhibits vascular leakage [89, 90], reduces the vitreal level of VEGF [91], the secretion of VEGF by retinal cells [92–94], and the cellular effects of VEGF, e.g., the VEGF-evoked secretion of matrix metalloproteinases (own unpublished results). However, it is not known whether the edema-resolving therapies used clinically stimulate the fluid absorption from the retinal tissue concomitantly. This is likely since anti-inflammatory steroids are also effective in resolution of retinal edema in cases which are not associated with angiographic vascular leakage.

Stimulation of the fluid clearance function of pigment epithelial and Müller cells may represent an approach to the development of novel edema-resolving drugs, as it has been shown in experimentally detached retinas. Here, activation of purinergic P2Y2 receptors stimulates the ion transport and, therefore, the fluid absorption from the subretinal space by the pigment epithelium, resulting in proper attachment of the sensory retina [95, 96]. We found that triamcinolone inhibits the osmotic swelling of Müller cells. The swelling-inhibitory effect was observed in animal models of retinal ischemia-reperfusion and detachment, and in retinas of diabetic animals (Fig. 3b) [52, 54, 58]. Moreover, triamcinolone inhibits the swelling of Müller cells in control retinas observed in the presence of the potassium channel blocker barium, the inflammatory mediators arachidonic acid or prostaglandin E2, as well as under oxidative stress [52, 54, 58]. We found that triamcinolone inhibits osmotic swelling of Müller cells by stimulation of an autocrine purinergic signaling cascade (Figs. 3c, 4c) [58]. Application of triamcinolone stimulates the nucleoside transporter-mediated release of adenosine from the cells. Adenosine activates adenosine A1 receptors which leads to the opening of barium- and arachidonic acid-insensitive potassium channels (likely, two pore-domain channels [97]), as well as of chloride channels, in the Müller cell membrane [54, 82]. The extrusion of ions draws water out of the cells thereby preventing cellular swelling under hypoosmolar conditions (Fig. 4c). The action of adenosine on ion channels does not require intracellular calcium but is mediated by cyclic adenosine monophosphate, protein kinase A, and phosphatidylinositol-3 kinase (own unpublished results) [54, 82].

Two pore-domain potassium channels may function as an osmolyte extrusion pathway that helps to maintain proper Müller cell volumes when Kir4.1 channels are downregulated under pathological conditions. Since the potassium currents drive the water transport through Müller cells, the activation of two pore-domain channels by adenosine should facilitate the potassium clearence and the water absorption from the edematous retinal tissue. Stimulation of two pore-domain channels, either indirectly by A1 receptor activation or directly by channel openers, may represent a method to dissolve retinal edema. It is known that activation of A1 receptors has protective effects against retinal ischemic injury [98, 99]; a stimulatory effect on the fluid clearance through Müller cells may contribute to this protective effect of A1 receptor activation.

Conclusions: retinal edema- a dysregulation at the glio-vascular interface

Retinal edema develops during ischemia/hypoxia and inflammation; under both conditions, Müller cells alter their potassium conductance and swelling characteristics [49–51]. The osmotic swelling of Müller cells indicates that the fluid transport through Müller cell membranes is impaired under conditions which are clinically associated with macular edema (ischemia, ocular inflammation, diabetes). In addition to vascular leakage, dysregulation of the water movements across the glio-vascular interface may represent a major causative factor of edema. Most likely, there are different subtypes of macular edema in individual patients, with differences in the relative contribution of vascular leakage and dysfunction of Müller cells. Additive effects of systemic diseases such as hypertension [100] may increase the variability of mechanisms for edema formation. In the presence of an osmotic gradient across the glio-vascular interface, e.g., when the blood becomes hypoosmolar due to hyponatremia or hypoalbuminemia associated with renal or hepatic failures, or when neuronal hyperexcitation results in a strong increase of the retinal osmotical pressure in comparison to the blood and vitreous, Müller cells may swell similar as described in brain astrocytes [101]. Pharmacological stimulation of the ion and water clearance function of Müller cells should aid in resolution of both vascular and cellular edema. It is suggested that triamcinolone—via stimulation of endogenous adenosine signaling in Müller cells—facilitates the fluid absorption from the edematous tissue. Adenosine A1 receptors may represent a promising target for the development of novel drugs that stimulate the absorption of excesss fluid in edema.