Purines in the eye: recent evidence for the physiological and pathological role of purines in the RPE, retinal neurons, astrocytes, Müller cells, lens, trabecular meshwork, cornea and lacrimal gland

Abstract

This review highlights recent findings that describe how purines modulate the physiological and pathophysiological responses of ocular tissues. For example, in lacrimal glands the cross-talk between P2X₇ receptors and both M₃ muscarinic receptors and α_1D-adrenergic receptors can influence tear secretion. In the cornea, purines lead to post-translational modification of EGFR and structural proteins that participate in wound repair in the epithelium and influence the expression of matrix proteins in the stroma. Purines act at receptors on both the trabecular meshwork and ciliary epithelium to modulate intraocular pressure (IOP); ATP-release pathways of inflow and outflow cells differ, possibly permitting differential modulation of adenosine delivery. Modulators of trabecular meshwork cell ATP release include cell volume, stretch, extracellular Ca²⁺ concentration, oxidation state, actin remodeling and possibly endogenous cardiotonic steroids. In the lens, osmotic stress leads to ATP release following TRPV4 activation upstream of hemichannel opening. In the anterior eye, diadenosine polyphosphates such as Ap₄A act at P2 receptors to modulate the rate and composition of tear secretion, impact corneal wound healing and lower IOP. The G_q11-coupled P2Y₁-receptor contributes to volume control in Müller cells and thus the retina. P2X receptors are expressed in neurons in the inner and outer retina and contribute to visual processing as well as the demise of retinal ganglion cells. In RPE cells, the balance between extracellular ATP and adenosine may modulate lysosomal pH and the rate of lipofuscin formation. In optic nerve head astrocytes, mechanosensitive ATP release via pannexin hemichannels, coupled with stretch-dependent upregulation of pannexins, provides a mechanism for ATP signaling in chronic glaucoma. With so many receptors linked to divergent functions throughout the eye, ensuring the transmitters remain local and stimulation is restricted to the intended target may be a key issue in understanding how physiological signaling becomes pathological in ocular disease.

1. Introduction

The purines ATP, adenosine, and diadenosine polyphosphates (Fig. 1), along with their metabolites and pyrimidine analogs, mediate a large number of functions in the eye. While some of these activities are required for the daily maintenance of ocular tissue, there is a growing recognition that dysregulated stimulation can contribute to disease. This review discusses recent developments in several ocular tissues involving both basic signaling and contribution to disease.

Figure 1. Basic structure of endogenous purinergic agonists.

Structures of A. Adenosine 5′ triphosphate (ATP); B. Uridine 5′triphosphate (UTP); C. Adenosine and; D. Diadenosine tetraphosphate (Ap4A).

The fundamentals of purinergic signaling in the eye are common throughout the body. While ATP can be released by nerves as a classic neurotransmitter, purinergic signaling systems are particularly interesting because nearly all cell types release ATP. The use of vesicular, non-vesicular or a mixture of both release mechanisms by these cells enables local delivery of the transmitter and allows the transmitter release to be more responsive to local cues (Fitz, 2007). This released ATP is dephosphorylated into ADP, AMP and adenosine by ectonucleotidases and ectoATDPases. ATP acts at ionotropic P2X receptors or metabotropic P2Y receptors, while adenosine acts at metabotropic receptors (Burnstock, 2006). Critically, the locally released ATP acts in an autocrine and paracrine fashion to maintain homeostasis as well as respond to insults (Corriden and Insel, 2010).

While ocular tissues have been known for over a decade to possess purinergic receptors and respond to stimulation, the field has moved far beyond receptor identification in the past few years. Below we highlight recent key advances in ocular purine research and stress the complexities and nuances of purinergic signaling. From interactions with other transmitter systems and differential effects of splice variants on the ocular surface, different mechanisms of ATP release from lens and trabecular meshwork, and the effects of diadenosine polyphosphates in the anterior eye, to the manipulation of Müller cell volume, purinergic contributions of ATP to visual processing in the inner and outer retina, survival of retinal ganglion cells and manipulation of lysosomal pH and lipofuscin levels in RPE cells, the review draws attention to recent advances that push our understanding of how purinergic signaling influences the eye. A graphical overview of the ocular regions discussed in this review is given in Fig. 2.

Figure 2. Sites of purinergic influence in the eye that are discussed in this review.

Purinergic signaling in the conjunctiva, lacrimal gland and cornea is expanded in Fig. 3, in the trabecular meshwork is expanded in Fig 4, in the lens in Fig. 5, in the retina in Fig. 6, and Müller cells in Fig. 7.

This review represents interesting recent work by some of the key researchers in the field of ocular purines; it is not meant an exhaustive review of the field. There are several exciting areas of research in ocular purines that are not discussed including development (Pearson et al., 2005), glial cells (Newman, 2004), pericytes (Sugiyama et al., 2005), photoreceptor survival (Notomi et al., 2011), A₁ adenosine receptors and outflow (Crosson et al., 2005), and many others. It is hoped that future reviews will be able to capture the excitement of these additional fields.

2. Recent findings regarding purines in the eye

2.1. Purinergic signaling in the lacrimal gland: cross-talk between P2X₇, M₃ muscarinic, and α_1D-adrenergic receptors

A decrease in the amount or change in the composition of lacrimal gland fluid secretion can lead to dry eye disease. Lacrimal gland secretion consists of proteins, electrolytes, and water that is regulated predominantly by activation of parasympathetic and sympathetic nerves (Dartt, 2009). An additional stimulus for secretion is ATP acting at purinergic receptors (Hodges et al., 2009; Novak et al., 2010) suggesting neuromodulatory or autocrine function of purines influence lacrimal secretion. Freshly isolated rat lacrimal gland acini have been used to determine the cellular interactions between purinergic receptors, M₃ muscarinic receptors (Dartt and Hodges, 2011a) and α_1D-adrenergic receptors (Dartt and Hodges, 2011b), and how they influence protein secretion (Fig. 3).

Figure 3. Purines in the lacrimal gland and cornea.

Schematic of ocular surface with mechanisms of purinergic, muscarinic, and α-adrenergic pathway interactions in lacrimal gland acini (top) and growth faction and nucleotide interaction to cause corneal epithelial wound healing (bottom). Ach-acetylcholine, M3 AChR- muscarinic type 3 receptor, PLC-phospholipase C, PKC-protein kinase C, PLD-phospholipase D, ERK1/2-extracellular regulated kinase 1/2. ATP-adenosine triphosphate, P2X7R- purinergic receptor X7, α1D-AR -α1D adrenergic receptor, norepi-norepinephrine, NO-nitric oxide, cGMP-cyclic guanosine monophosphate, EGF-epidermal growth factor, EGFR-epidermal growth factor receptor, P2X3-purinergic receptor X3, TGFα-transforming growth factor α.

In the lacrimal gland ATP activates P2X₃ and P2X₇ receptors to increase intracellular calcium ([Ca²⁺]_i) and stimulate protein secretion (Hodges et al., 2011; Hodges et al., 2009; Novak et al., 2010). When the interaction between P2X₇ and M₃ muscarinic receptors was investigated, cholinergic agonists were found to release ATP from lacrimal gland pieces, but not isolated acini, suggesting that cholinergic agonists may stimulate ATP release from efferent nerve endings rather than from the acinar and duct cells themselves (Dartt and Hodges, 2011a). Using muscarinic and P2X₇ agonists and antagonists on acinar cells, muscarinic agonists were found to activate P2X₇ receptors to increase [Ca2+]_i and stimulate protein secretion, whereas P2X₇ agonists did not activate M₃ muscarinic receptors. As the latter measurements were made in acini, muscarinic agonists are not activating P2X₇ receptors by releasing ATP (Dartt and Hodges, 2011a); rather muscarinic agonists are activating an intracellular signaling pathway. The signaling components of this pathway remain to be identified, but activation of protein kinase C (PKC) α or ε has been eliminated.

The interaction between P2X₇ and α_1D-adrenergic receptors was subsequently investigated (Dartt and Hodges, 2011b). This interaction was different from the P2X₇/M₃ muscarinic receptor interaction. First the α_1D-adrenergic receptor agonist phenylephrine caused ATP release from both lacrimal gland pieces and acini. Second, using agonists and antagonists, activation of α_1D-adrenergic receptors was also found to activate P2X₇ receptors. This caused only an increase in [Ca²⁺]_i, but not in protein secretion. Third, activation of P2X₇ receptors also activated α_1D-adrenergic receptors causing an increase in [Ca2+]_iI, but not in protein secretion (Dartt and Hodges, 2011b).

To determine if the interaction of α_1D-adrenergic receptors with P2X₇ receptors was relevant for protein secretion, P2X₇ receptor knockout mice were examined. In the absence of P2X₇ receptors α_1D-adrenergic agonist-stimulated protein secretion was decreased. This suggests an important interaction between α_1D-adrenergic and P2X₇ receptors. In the same knockout mice, M₃ muscarinic agonist stimulated secretion was unchanged. This finding is consistent with M₃ muscarinic agonists stimulating P2X₇ receptors, but not the reverse. The results with P2X₇ receptor knockout mice in which protein secretion was decreased or unchanged suggests that activation of P2X₇ receptors is beneficial for lacrimal gland secretion and that P2X₇ receptors are not deleterious in the lacrimal gland.

In conclusion, M₃ muscarinic and α_1D-adrenergic receptors each interact with P2X₇ receptors to increase [Ca²⁺]_i and stimulate protein secretion in the lacrimal gland, however the cellular mechanisms of interaction differ (Fig. 3). M₃ muscarinic receptors use an intracellular pathway to stimulate P2X₇ receptors, whereas α_1D-adrenergic receptors release ATP from acinar cells and α_1D-adrenergic and P2X₇ receptors activate each other. P2X₇ receptor activation is not deleterious in the lacrimal gland.

2.2. Purine signaling in the cornea: a role in wound repair

The avascular and highly innervated cornea is divided into 3 regions by 2 basement membranes: epithelium, a matrix-rich stroma, and endothelium. Trigeminal cultures express the full complement of P2Y and P2X receptors while corneal epithelial cells express select P2Y (Boucher et al., 2011; Boucher et al., 2010; Oswald et al., 2012) and P2X receptors (Mankus et al., 2012; Mankus et al., 2011; Mayo et al., 2008). The expression of these receptors may provide a manner of interdependent communication between the neurons and the corneal cells in relation to wound healing; epithelium injured in vitro releases ATP to mobilize a Ca²⁺ wave to neighboring cells, but sensory nerves are also required for proper wound closure in vivo. The potential cross-talk between corneal epithelial cells and trigeminal nerves was demonstrated in co-cultures under normoxic and hypoxic conditions, as there was a change in Ca²⁺ dynamic patterns in epithelial cells when stimulated by wounded trigeminal nerves exposed to hypoxic conditions (Lee et al., 2014).

Activation of purinergic receptors in corneal epithelial cells results in distinct post-translational modification of the epidermal growth factor receptor (EGFR) and structural proteins to link the stimulus and signaling pathways leading to wound repair (Boucher et al., 2011). P2Y₂ mRNA expression increased during onset of cell migration after injury and its knockdown attenuates Ca²⁺ mobilization, phosphorylation of downstream molecules and cell migration (Fig. 3)(Boucher et al., 2010; Kehasse et al., 2013). For example EGF stimulation yielded prototypical tyrosine phosphorylation on EGFR, but ATP stimulation caused a 2- to 15-fold increase over control in phosphorylation of EGFR Y974, Y1086, and Y1148 residues. Interestingly, ATP induced only minimal phosphorylation intensity on EGFR Y1173 compared to that induced by EGF. Cells cultured in stable isotope labeled amino acids followed by knockdown of P2Y₂ receptor revealed a significant decrease in Src Y421 and paxillin Y118 with no change in EGFR Y1173. Together these data indicate the far reaching importance of the P2Y₂ receptor (Kehasse et al., 2013).

While the P2Y₂ receptor appears to be critical in modulating downstream wound healing, the P2X₇ receptor also modulates integrity of the cornea. Altered stromal structure was detected in the Pfizer P2X7^−/− mice (Mayo et al., 2008), with collagen fibrils individually thinner and lacking organization. While the two most common collagen types, collagen α1(I) and collagen α3(V), were decreased, collagen α1(III), a marker of corneal injury and scarring, was increased. Likewise, lysyl oxidase expression increased, which may contribute to the observed smaller fibril diameters. Corneal stromas in the P2X7^−/− mice also showed decreased expression of the proteoglycans, decorin, keratocan, and lumican, which usually facilitate collagen organization, and a decrease in overall proteoglycan sulfation. In contrast, perlecan, a heparan sulfate proteoglycan associated with wounding, increased throughout the unwounded P2X7^−/− stroma. While in the wildtype, perlecan is present along the basement membrane it was absent in the P2X7^−/− corneas, which may explain the fragile epithelial-stromal interface. Digestion with keratanase I and chondroitinase ABC removed more glycosaminoglycan side chains in the wildtype cornea, indicating that the levels of undigested proteoglycans, which have heparin sulfate side chains, are reciprocally regulated by P2X₇ (Mankus et al., 2012). Altogether, these data suggest that the lack of P2X₇ holds the cornea in a pseudo-wounded state.

The human corneal epithelium expresses a splice variant in addition to the full-length transcript of P2X₇. While the variant form is more prominent in subconfluent epithelium, it decreases as the full-length form increases with polarization and stratification. This switch, accompanied by decreased proliferating cell nuclear antigen (PCNA) and increased apical cell dye uptake, indicate that corneal epithelial cells may respond to environmental changes through P2X₇ regulation (Mankus et al., 2011).

In conclusion, purines play a major regulatory role in the cornea in cell communication and wound repair (Fig. 3). Future studies to investigate how diabetes compromises purine-regulated cell migration are currently underway and promise to add to the complexities of pathophysiological actions of purines in the cornea.

2.3. Purine signaling in the trabecular meshwork: a modulator of intraocular pressure

The purines ATP and adenosine have long been acknowledged to modulate the flow of aqueous humor. For example, adenosine modulates intraocular pressure (IOP) by modifying both inflow and outflow pathways of aqueous humor. Activation of A₃ adenosine receptors stimulates Cl⁻ channels of nonpigmented ciliary epithelial cells, increasing inflow (Kiel et al., 2011) and IOP. Knockout of A₃ adenosine receptors lowers IOP (Avila et al., 2002). In contrast, A₁ adenosine receptor activation of trabecular meshwork (TM) cells stimulates matrix metalloproteinase-2 and -9 (MMP-2,9) release, reducing resistance through the trabecular outflow and lowering IOP (Fig. 4). Inhibiting MMP activity abolishes A₁-triggered resistance reduction (Crosson et al., 2005). Adenosine can be delivered to inflow and outflow cells by release and ectoenzymatic conversion of ATP to adenosine. Six ecto-ATPases in native and immortalized human TM cells have been identified (Li et al., 2012b). These enzymes are functional since ATP addition stimulates release of zymographically-measured MMP-2 and MMP-9 (Li et al., 2012a). Interrupting ATP conversion to adenosine abolishes ATP-triggered MMP release.

Figure 4. ATP release by trabecular meshwork cells.

At least six modulators modify ATP release through PX1, Cx and P2X₇ channels (see section 2.3). Enzymatic conversion of released ATP to adenosine through ecto-nucleotide pyrophosphatase/phosphodiesterases, ecto-nucleotide triphosphate diphosphohydrolases and ecto-5′-nucleotidase initiates sequential: activation of A₁ adenosine receptors, release of MMP-2 and -9, and reductions of outflow resistance and IOP.

Three major pathways of ATP release have been identified by the luciferin-luciferase method during hypotonic swelling of TM cells with/without inhibitors as shown in Fig. 4. Pannexin-1 (PX1) hemichannels, connexin (Cx) hemichannels and P2X₇ ATP ionotropic receptors (P2RX₇) all are (Li et al., 2012b). Simultaneously blocking PX1, Cx and P2RX₇ abolishes swelling-activated ATP release. Pharmacalogical evidence has been supplemented with PX1-knockdown studies in TM cells by lentivirus-mediated RNA interference. Partial knockdown reduced total swelling-activated ATP release and reduced the efficacy of the PX1 blocker probenecid, while enhancing the efficacy of Cx and P2RX₇ blockers, in inhibiting ATP release. These knockdown results support the pharmacologic results (Li et al., 2012b).

The ciliary epithelial cells responsible for aqueous humor inflow also release ATP through PX1 and Cx hemichannels. However, vesicular release, not P2RX₇, provides the third major pathway for swelling-activated ATP release (Li et al., 2012b). This difference may permit differential delivery of ATP and adenosine to inflow and outflow cells to alter IOP.

Six factors that can modulate TM-cell ATP-release have been identified (Fig. 4). Like cell swelling, graded stretch of human TM cells produces graded ATP release, with identical pharmacologic profile (Li et al., 2012b). As in other cells, the reducing agent dithiothreitol inhibits PX1 and thereby ATP release, whereas nonphysiologic reductions in extracellular Ca²⁺ increase Cx hemichannel activity and ATP release (Li et al., 2012b). Increasing intracellular Ca²⁺ does not activate PX1-mediated ATP release in TM or ciliary epithelial cells. In contrast, actin remodeling and cardiotonic steroids (CTS) strongly alter ATP release. Incubation (≥14 days) with dexamethasone markedly polymerizes actin, increases cross-linked actin networks (CLANs) and inhibits swelling/stretch-activated ATP release, whereas brief incubation with cytochalasin-D depolymerizes actin and enhances swelling/stretch-activated ATP release (Li et al., 2012b). The cytoskeleton-dependent changes in ATP release are correlated with changes in cell volume regulation. Finally, inhibiting Na,K ATPase with CTS reduces swelling-activated ATP release, which is dissociable from inhibiting Na/K-exchange through the Na⁺ pump, and is likely mediated by the enzyme’s scaffolding/signaling functions (Li et al., 2012a). This inhibition may be physiologically relevant since humans produce ouabain-like factors and express high-affinity α2 Na,K ATPases in TM cells. Furthermore, selectively rendering α2 Na⁺ pumps ouabain-resistant produces a knock-in mouse phenotype (Schaefer et al., 2011).

2.4. Purines and the lens: response to osmotic stress

Lens transparency is the result of its unique cells. All but the youngest lens fibers lack organelles. Having no nucleus, ribosomes, endoplasmic reticulum or mitochondria, fully differentiated lens fibers rely on the monolayer of epithelial cells that covers the anterior surface of the fiber mass. Na,K-ATPase activity in epithelial cells at the lens at the equator is critical to establishing a circulating flow of ions that underpins homeostasis of the entire fiber cell mass (Mathias et al., 2007). Earlier studies point to the ability of purinergic agonists ATP and UTP to increase Na,K-ATPase activity in lens epithelium by a mechanism that hinges on activation of Src family tyrosine kinases (Tamiya et al., 2007). The lens itself is capable of releasing ATP and does so when subjected to hyposmotic stress. In studies with intact porcine lens, osmolarity was reduced from 300 to 200 mOsm and within 10 min the concentration of ATP in the bathing solution more than doubled (Shahidullah et al., 2012a). The same hyposmotic stress also doubled Na,K-ATPase activity in the lens epithelium and the change of Na,K-ATPase activity in the epithelium was suppressed by maneuvers which interfered purinergic receptors or stopped accumulation of extracellular ATP in the medium.

Studies on the mechanism of ATP release from the lens suggest it occurs because connexin and pannexin hemichannels open. The relative contribution of these two pathways has not been resolved: a combination of 18α-glycyrrhetinic acid (AGA) (a connexin blocker) added together with probenecid (PROB) (a pannexin blocker) completely eliminated the ATP release but added alone, AGA or PROB blocked only partially. The notion of a hemichannel mechanism is reinforced by the finding that hyposmotic stress opens a pathway for propidium iodide to gain entry to the lens epithelium and this is prevented by AGA+PROB. With a molecular weight of 668, PI would be excluded by most channels but not hemichannels. It is interesting that both increased PI uptake and ATP release in hyposmotic solution were blocked by RN 1734, a very different channel blocker that interferes with TRPV4 channels (Shahidullah et al., 2012b). Moreover, it was found that PI entry into the lens epithelium was increased by more than twofold in lenses exposed to a TRPV4 agonist GSK1016790A (GSK). We interpreted these findings to signify activation of TRPV4 channels is a trigger for hemichannel opening. In support of this idea, GSK was found to trigger ATP release from the lens and the response was prevented by AGA+PROB. The observations are consistent with hyposmotic stress-induced TRPV4 channel activation which triggers hemichannel-mediated ATP release (Fig. 5). The results point to a critical role for TRPV4 activation as trigger for hemichannel opening and ATP release that serves to activate purinergic receptor-dependent, Src kinase-dependent stimulation of Na,K-ATPase activity. This chain of events may enable the lens to sense and respond to hyposmotic stress.

Figure 5. Purinergic signaling in the lens.

Swelling activates a TRPV4 receptor which in turn activates ATP release from pannexin/connexin hemichannels. The released ATP stimulates a P2Y receptor which activates the Na/K ATPase pump on the lens epithelium to maintain lens transparency.

2.5. Diadenosine polyphosphates and the anterior eye: additional endogenous agonists to modulate IOP, corneal wound healing and tear formation

While the above sections have outlined the multiple effects of ATP and adenosine signaling on the anterior eye, diadenosine polyphosphates (abbreviated as ApnA) are also capable of stimulating P2 receptors and have been demonstrated to modulate several physiological processes in the anterior segment. Diadenosine polyphosphates are a distinct group of dinucleotide agonists formed by two adenosine moieties linked by their ribose 5′-ends to a number of phosphates, which can vary from 2 to 7 (see Fig. 1). They are naturally occurring substances that can be released to the extracellular medium by means of different mechanisms. The action of these compounds is by means of P2 receptors as occurs with other nucleotides. The most representative dinucleotides are diadenosine tetraphosphate and diadenosine pentaphosphate, Ap₄A and Ap₅A respectively.

It is known that ApnA are present in tears and that they can stimulate tear secretion to 60% above normal tear values. This effect occurs via P2Y₂ receptors. Diadenosine polyphosphates are released as a consequence of the shear stress which occurs during blinking (Peral et al., 2006). In addition, Ap₄A is able to modify protein tear composition. For instance, the levels of lysozyme increase about 95% above normal values in rabbit tears when Ap₄A is instilled. Considering that lysozyme is one of the first defence mechanisms against bacterial infection, Ap₄A is providing protection against pathogen invasion (Guzman-Aranguez et al., 2011).

Ap₄A participates in other processes such as corneal wound healing. In vivo experiments showed that Ap₄A increased the rate of healing by 130%. Consistent with this finding, the dinucleotide also accelerated the rate of migration in primary cultures of rabbit corneal epithelial cells. In both in vivo and in vitro cases, the actions occur via P2Y₂ receptors (Crooke et al., 2008). Ap₄A is also a marker for dry eye in pathologies such as evaporative and non-evaporative dry eye, Sjögren syndrome and aniridia among other conditions (Carracedo et al., 2010).

One interesting function of Ap₄A is its ability to lower IOP, with reductions of nearly 30% found (Guzman-Aranguez et al., 2007). Pharmacological studies suggest that this hypotensive effect was mediated by a P2X₂ receptor. Moreover, denervation studies and experiments with anticholinergic agents localized the P2X₂ receptor to the cholinergic nerve terminals that innervate and control the ciliary processes. Ap₄A activates these P2X₂ receptors, facilitating the release of more acetylcholine, which contracts the muscle pulling the scleral spur, opening the irido-corneal angle and reducing hydrodynamic resistance to the outflow. Nevertheless, the hypotensive action of Ap₄A is not limited to these nerve endings and a direct effect of Ap₄A on the trabecular meshwork has also been demonstrated. Ap₄A increased trabecular outflow facility in bovine ocular anterior segments by P2Y₁ receptors activation. This indicates that Ap₄A facilitates the drainage of the aqueous humor through the trabecular meshwork, contributing to an IOP reduction (Guzman-Aranguez et al., 2013). Another function has been recently proposed for Ap₄A. The application of Ap₄A preserved the sympathetic terminals innervating the ciliary body from 6-hydroxydopamine-induced degeneration, indicating that this molecule has a neuroprotective role that may be of interest in neurodegenerative diseases.

At present there is no direct evidence supporting any role for diadenosine polyphosphates in the retina. Nonetheless, the dinucleotide deoxycytidine tetraphosphouridine is able to enhance the rate of subretinal fluid reabsorption via P2Y₂ receptor activation in rodent models of retinal detachment (Guzman-Aranguez et al., 2013), suggesting additional roles for the compounds in the posterior eye are possible.

2.6. P2X receptors in retinal signaling: a role in modulating the visual output

There is emerging evidence that purines can contribute to neuromodulation in both the inner and outer retina (Fig. 6). Expression of purinergic receptors, enzymes important for purine degradation and the vesicular nucleotide transporter have all been demonstrated (Puthussery and Fletcher, 2004, 2006, 2007). Double labeling of P2X receptors with known markers of retinal neurons has provided valuable information regarding the possible involvement of purinergic receptors in retinal signaling. For example, amacrine cells have either GABA or glycine as their predominate amino acid neurotransmitters. Extensive colocalization of P2X₂, P2X₃ and P2X₇ receptors with GABA has been reported suggesting that P2X receptors modulate information processing by GABAergic amacrine cells (Puthussery and Fletcher, 2004, 2006, 2007). Moreover, different P2X receptors are segregated to specific circuits within the inner retina. Notably, P2X₂ receptors are localized to putative amacrine cells postsynaptic to cone bipolar cells and not rod bipolar cells, while P2X₃ and P2X₇ are localized to neurons postsynaptic to both cone and rod bipolar cells. The neuronal circuitry subserving scotopic vision is well understood. In contrast to the cone pathways (cones to cone bipolar cells to ganglion cells) the rod pathway involves rod bipolar cells forming a synapse onto two amacrine cells. With respect to purinergic receptors, A17 GABAergic amacrine cell processes display immunoreactivity for both P2X₃ and P2X₇ and these receptors are localized at both conventional and reciprocal synapses (Puthussery and Fletcher, 2004, 2007). Although it is not known whether the same neuron expresses both P2X receptor types, the results imply that GABA modulation of the rod bipolar cell is more complex than hitherto thought, and that purines acting on P2X₃ and/or P2X₇ receptors, may be important for potentiating the release of GABA from A17 amacrine cells onto rod bipolar cells.

Figure 6. Purines in the retina in signaling and pathology.

Schematic diagram illustrating the mechanisms highlighted in the present review. On the left, pathways involved in the transmission of the visual signal, with P2X₇ receptors both increasing the a-wave and modulating the response to light by RGCs (see section 2.6). Purines help Müller cells maintain their volume and the fluid balance of the retina, contributing to both signal transmission and pathophysiological outcomes (see Section 2.7 and Fig. 7). With regards to the role of purines in retinal pathophysiology, ATP acting at P2X₇ receptors can raise Ca2+ and kill RGCs, while conversion to adenosine can temper this (Section 2.8). Likewise, ATP released from RPE cells can autostimulate P2X₇ receptors to raise lysosomal pH and increase lipofuscin-like autofluorescence while adenosine acting at A_2A receptors can reacidify compromised lysosomes and reduce autofluorescence (Section 2.9).

P2X₃ and P2X₇ receptors have also been immunolocalized to the outer plexiform layer. P2X₇ has been localized presynaptically to rod and cone pedicles, and functional studies have shown that BzATP increases the amplitude of the a-wave (Puthussery and Fletcher 2004). These results imply that P2X₇ receptors have a role in regulating photoreceptor function. The expression of P2X₇ receptors by photoreceptors may have important implications for photoreceptor integrity. Indeed, intravitreal injection of ATP causes rapid loss of photoreceptors, that can be reduced by co-injection with the P2X antagonist, PPADS (Puthussery and Fletcher, 2009). Treatment of rd1 mice with PPADS reduced photoreceptor loss by approximately 30% (Puthussery and Fletcher, 2009). This suggests that ATP release from dying photoreceptors could affect the integrity of neighboring photoreceptors, potentiating photoreceptor loss.

Purinergic signaling can also modulate visual signaling in the inner retina. The complex synaptic processing potentially has an outcome that affects visual information flow, as RGC responses to visual stimulation are modulated by P2X₇ receptor activation with BzATP or blockade by A438079 (Chavda et al., 2013). Whether this reflects presynaptic P2X₇ receptors or postsynaptic P2X₇ receptors located on retinal ganglion cells, or a combination of both, is still unclear at present. This represents an avenue for further work, as does the finding that these effects are paralleled by changes in microglial morphology (Chavda et al. 2013).

2.7. Purinergic signaling in Müller cells: a role in volume regulation

Müller cells are the principal macroglial cells of the retina, with numerous fine processes emanating from central stems that span the whole thickness of the retina. These processes place them in intimate contact both with neurons and non-neural structures such as blood vessels, the vitreous and the subretinal space (Fig. 7). This intimate interaction makes Müller cells ideally located to fulfill several crucial functions in maintaining the integrity of retinal tissue. For example, Müller cells mediate neurovascular coupling, perform neurotransmitter recycling, support the neuronal energy metabolism and maintain the retinal ion- and volume homeostasis (Reichenbach and Bringmann, 2013).

Figure 7. The contribution of purines to Müller cell volume regulation.

Growth factors acting directly on Müller glia trigger sequential glutamate (via exocytosis) and ATP (via hemichannels) release. ATP acts at P2Y₁ receptors after dephosphorylation to ADP by ectonucleotidase NTPDase2. The adenosine released stimulated A₁-receptors and initiates the opening of Cl⁻ and K⁺ channels; this efflux of ions osmotically draws water out of the cell, reducing cell volume and balancing retinal fluid levels (See section 2.7). Transport across the RPE and spatial buffering of K⁺ across the inner retina add to maintaining fluid balance. BV, blood vessel; NT, nucleoside transporter, 5′NT, 5′-ecto-nucleotidase, mGluR, metabotropic glutamate receptor.

Recent studies point to a key role of purinergic signaling in Müller cell function (Wurm et al., 2011). Retinal neurons and glial cells are endowed with various purinergic receptors enabling them to react to extracellular purines such as ATP, ADP or adenosine. The principal ATP-sensitive receptor on Müller cells is the P2Y₁-receptor, coupling to G_q11, while ionotropic P2X receptors appear to be largely absent from the Müller cells (Wurm et al., 2009). P2Y₁-activation is primarily responsible for elevations in intracellular calcium levels seen in response to ATP application in juvenile and differentiated adult Müller cells. Moreover, a complex glutamatergic-purinergic signaling cascade involving the P2Y₁-receptor, metabotropic glutamate receptors, ATP release, the P2Y₁ receptor and adenosine A₁ receptors enables Müller cells to maintain their cell volume. In contrast, bipolar cells do not possess comparable mechanisms of volume regulation and swell under hypoosmotic conditions; purines have no effect on the swelling behavior of bipolar cells (Vogler et al., 2013).

Tight cellular volume regulation is a prerequisite for Müller cells to mediate transcellular ion- and fluid fluxes from the extracellular space of the retina to reservoirs like the vitreous body or blood vessels; this in turn enables the spatial buffering of potassium the and maintenance of retinal fluid homeostasis (Reichenbach and Bringmann, 2013). While volume regulation is limited in Müller cell progenitors and in dedifferentiated gliotic Müller cells found in pathologically altered retina (after ischemia or diabetic retinopathy), this regulation can be improved by stimulation of P2Y₁ receptors or A₁ adenosine receptors (Wurm et al., 2009; Wurm et al., 2008). Loss of volume regulation from Müller cells may contribute to retinal edema, as can occur in the diabetic retina. As such, the restoration of regulatory capabilities by activation of purinergic receptors should be considered when thinking about new therapeutic strategies to resolve retinal edema.

2.8. P2X₇ receptors and neurodegeneration of retinal ganglion cells

Retinal ganglion cells (RGCs) express multiple P2 and adenosine receptors. This abundance of receptors in the healthy eye indirectly implies an important physiological contribution from the purinergic signaling system. While our understanding of these physiological roles is emerging, (as discussed in section 2.6. above), much of the research in relation to purines and RGCs has focused on the P2X₇ receptor and its potential role in neurodegeneration. This is a key area of interest since RGC degeneration is central to the pathophysiology of glaucoma, and neuroprotection is a major therapeutic goal.

Stimulation of the P2X₇ receptor with the agonist BzATP induced death of isolated RGCs via an increase in intracellular Ca²⁺ and activation of caspase (Zhang et al., 2005), while intravitreal injection of BzATP resulted in RGCs loss in vivo (Hu et al., 2010). In both cases RGC death was prevented by P2X₇ receptor antagonists. More recently, P2X₇-mediated RGC death was demonstrated in an in vivo optic nerve crush model (Kakurai et al., 2013) and in human retinal explants (Niyadurupola et al., 2011; 2013). Together, these experiments strongly suggest a role for the P2X₇ receptor in the pathological loss of RGCs (Fig. 6). Of particular note is the ability of P2X₇ receptor antagonists to prevent RGC death triggered by simulated ischemia (oxygen and glucose deprivation) in a human retinal explant model (Niyadurupola et al., 2013). This suggests that metabolic stress may cause release of ATP which can trigger P2X₇ receptor-mediated excitotoxicity.

The adenosine produced by dephosphorylation of extracellular ATP may balance the pathological effects of the P2X₇ receptor on RGCs. Adenosine directly modulates RGC function and confers general neuroprotection via the A₁ receptors (Hartwick et al., 2004; Newman, 2004), with A₃ adenosine receptors specifically preventing the damage induced by the P2X₇ receptor (Zhang et al., 2006). Recent evidence for the mechanosensitive release of ATP from optic nerve head astrocytes via pannexin hemichannels, coupled with the ability of stretch to increase expression of several pannexin isoforms both in vitro and in the optic nerve head of 8 month old Tg-Myoc^Y437H mice provides a mechanism for amplifying levels of extracellular ATP in chronic glaucoma (Beckel et al., 2014).

2.9. Purinergic signaling and lipofuscin in RPE cells

In RPE cells, as in other ocular tissues, the mechanisms leading to ATP release, and the regulation of enzymes responsible for dephosphorylating ATP into adenosine, provide the goalposts that ultimately control the extent and duration of purinoreceptor stimulation. The relative levels of purines and pyrimidines determined by release and degradation ultimately dictate the availability of agonists capable of stimulating the P2X, P2Y and adenosine receptors present on RPE cells (Mitchell and Reigada, 2008). While ATP release and subsequent dephosphorylation to ADP, AMP and adenosine seem to follow the general patterns, two observations may be of relevance to signaling in RPE cells.

The first point relates to the rapid release of ATP from swollen RPE cells (Mitchell, 2001). While cell swelling is a standard tool used to study ATP release, RPE cells are generally not associated with large changes in cell volume. However, RPE cell hypertrophy is widespread in several retinal degenerations and swelling could lead to a chronic increase in ATP release. The increased expression of a marker for extracellular ATP in RPE cells of the ABCA4^−/− mouse model of Stargardt’s retinal degeneration is consistent with this theory (Guha et al., 2013).

Secondly, the balance between extracellular ATP and adenosine may alter the lysosomal activity of RPE cells and thus influence the production of lipofuscin (Fig. 6). Stimulation of the P2X₇ receptor raises lysosomal pH in RPE cells and slowes autophagic turnover, modifying levels of p62 and the LC3Bll/LC3Bl ratio (Guha et al., 2013). RPE cells fed photoreceptor outer segments displayed a lipofuscin-like autofluorescence upon lysosomal alkalinization, but blockage of the P2X₇ receptor reduced this autofluorescence. The ability of lysosomal alkalinization to increase levels of lipid oxidation, and of P2X₇ antagonism to reduce this lipid oxidation are of particular interest. The action of P2X₇ antagonists to reduce levels of both lipofuscin and oxidized lipid is particularly striking as no exogenous agonist was added. This suggests that the release of endogenous ATP from RPE cells can autostimulate P2X₇ receptors to alkalinize lysosomes, oxidize lipids and increase lipofuscin levels.

In contrast to the actions of the P2X₇ receptor, the adenosine A_2A receptor can reacidify compromised lysosomes and reduce levels of photoreceptor outer segment autofluorescence (Liu et al., 2008). The elevation of cytoplasmic cAMP following stimulation of the adenosine A_2A receptor likely underlies this restoration of acidic lysosomal pH. Together, this implies that conversion of released ATP into adenosine by ectoenzymes may influence the production of lipofuscin and lipid oxidation in RPE cells (Guha et al, 2014). Whether the balance between ATP and adenosine influences the progression of diseases like age-related macular degeneration remains to be seen.

3. Conclusion

It is clear that ATP, adenosine and diadenosine polyphosphates have complex actions in the eye, and newly discovered functions are expected to add to this complexity. Even if most of the transmitter responsible for activating the receptors comes from local release, it is still remarkable that the actions of released purines remain confined to the desired target tissue. While the ectoATPDases, ectonucleotidases and other enzymes are highly effective at keeping purine levels low in most experimental conditions, the presence of so many tissues capable of ATP release combined with a plethora of purinergic receptors in the eye suggest a possibility for off-target actions, particularly if excessive release occurs under pathological or inflammatory conditions. Future directions may indicate whether interactions between the purinergic signaling systems of distinct ocular regions discussed above can contribute to problems in eye disease.