Background
Glaucoma is the third leading cause of visual impairment and the second cause of blindness worldwide [
1]. It is defined as a group of chronic degenerative optic neuropathies, characterized by the irreversible and progressive loss of retinal ganglion cells (RGCs) and damage of the optic nerve (RGC axons). Although glaucoma is a multifactorial disease, elevated intraocular pressure (IOP) is a major risk factor and the current treatments are mainly focused on reducing IOP [
2]. However, many patients continue to lose vision despite the control of IOP, and neuroprotective strategies aimed to prevent RGC loss are necessary [
3].
Increasing evidence has shown that neuroinflammation has an important role in the pathogenesis of glaucoma [
4‐
6]. Accordingly, microglial cells display an activated amoeboid-like morphology at the early stages of glaucoma [
7‐
10]. In parallel, there is an increased expression and release of pro-inflammatory cytokines [e.g. tumour necrosis factor (TNF), interleukin-1β (IL-1β)] and nitric oxide (NO) in the glaucomatous eye [
11‐
14]. The importance of this microglia-associated neuroinflammation in glaucoma is underscored by the observation that the control of microglia activation [
15‐
17] or of pro-inflammatory cytokine expression [
4,
18] can prevent the loss of RGC in animal models of glaucoma.
Microglia-associated neuroinflammation is also involved in different brain disorders [
19]. Adenosine is a neuromodulator, which can control inflammatory reactions [
20,
21] and microglia reactivity [
22‐
24] mainly through the activation of its G-protein-coupled receptor of the A
2A receptor (A
2AR) subtype [
25]. Accordingly, A
2AR antagonists afford robust neuroprotection upon ischemia, epilepsy or Alzheimer’s or Parkinson’s disease [
25].
All these evidence prompt the hypothesis that A2AR antagonists may also control the microglia-associated neuroinflammation and loss of RGC in animal models of glaucoma. Therefore, the main aim of this work was to investigate whether A2AR blockade modulates retinal microglia reactivity, neuroinflammation and loss of RGC triggered by lipopolysaccharide (LPS) or elevated hydrostatic pressure (EHP).
Discussion
The present work demonstrates that the blockade of A2AR prevented retinal neuroinflammation and death of RGC in an ex vivo model of glaucoma. We exposed retinal organotypic cultures to LPS and EHP, which bolstered microglia reactivity, increased neuroinflammatory response and loss of RGCs. These two noxious conditions up-regulated the A2AR system, as typified by an increase in the extracellular levels of ATP and increased expression and density of A2AR in microglia. Concomitantly, the A2AR system critically contributed to the neuroinflammation and RGC death, since A2AR blockade prevented the activation of microglia, the production of pro-inflammatory cytokines and the death of RGCs.
We took advantage of retinal organotypic cultures, a suitable model to evaluate cellular and molecular signalling mechanisms in which retinal anatomy is maintained [
26] and which has been established as a convenient model for screening potential neuroprotective drugs in the retina [
41]. This in vitro system enabled us to demonstrate that EHP changed microglia morphology towards an amoeboid-like form, similar to that caused by LPS, which has been extensively used as a microglial activator. Activation of microglial cells is observed as an early event in animal models of glaucoma [
9,
42], in which increased IOP is a main risk factor [
2]. In retinal organotypic cultures, the observed EHP- and LPS-induced microglia reactivity was paralleled by an increased expression and release of the pro-inflammatory cytokines IL-1β and TNF. Likewise, an increased production of TNF [
11,
43] and IL-1β [
44,
45] has been observed in glaucomatous animal models and in human glaucoma. Furthermore, the ability of anti-IL-1β and anti-TNF antibodies to prevent EHP-induced RGC death provided critical evidence that the death of RGCs upon exposure to EHP or LPS in retinal organotypic cultures actually resulted from the impact of pro-inflammatory cytokines. This is in agreement with previous reports demonstrating that the control of microglia reactivity [
15‐
17] or of pro-inflammatory cytokines [
4,
18,
46] prevents the loss of RGC in animal models of glaucoma. Nevertheless, the release of IL-6 by astrocytes and microglia triggered by EHP was reported to protect RGCs [
27], although the authors used purified cultures of microglia, astrocytes and RGCs and did not evaluate the possible interactions between these cells in a more complex in vitro model, as the retinal organotypic culture. The globally deleterious role of microglia-associated pro-inflammatory status is further heralded by the report that minocycline, an inhibitor of microglia activation, reduced microglia activation and improved RGC axonal transport and integrity [
15]. Overall, this evidence indicates that microglia reactivity is a precocious event and contributes to the pathophysiology of glaucoma by impairing the viability of RGCs.
The main conclusion of this study was the critical role of A
2AR in the control of EHP- or LPS-induced microglia activation, production of pro-inflammatory cytokines and RGC death in retinal organotypic cultures. Indeed, we observed that the blockade of A
2AR prevented the EHP- or LPS-induced modification of the production of pro-inflammatory cytokines and of NO as it was previously observed in the rodent hippocampus [
24]. Accordingly, it was already demonstrated that activation of A
2AR potentiates NO release from reactive microglia in culture, an effect that was associated with microglia neurotoxicity, and A
2AR antagonist was suggested as a potential neuroprotective drug [
22]. Moreover, we observed that A
2AR blockade prevents EHP-induced microglia morphological alterations, in agreement with recent findings that A
2AR antagonism reduces the retraction of processes in LPS-activated microglia [
47].
These conclusions seems to contradict previous studies reporting that the activation of A
2AR reduces microglia reactivity using primary retinal microglia cultures exposed either to LPS, hypoxia or amadori-glycated albumin [
48‐
50]. Several factors may explain this discrepancy: (1) while others used cultures of microglial cells, we used an organotypic retinal culture in which all retinal cells are present, and thus, an additional contribution from other glial cells cannot be excluded [
51,
52]; this is particularly important given that the control by A
2AR of microglia reactivity can be shifted from inhibitory to excitatory by the presence of increased extracellular levels of glutamate [
53]; (2) the insults triggering microglia activation are different and the LPS concentrations and time points were different; and (3) CGS 21680, the A
2AR agonist, at the concentration used in those studies (20 and 40 μM) is no longer selective, being proposed to bind also to A
1R [
54,
55]. The A
1R is coupled to G
i/o-proteins and often inhibitory, whereas the A
2AR is usually coupled to G
s-proteins, enhancing cAMP accumulation and PKA activity [
56]. Nevertheless, the observation of different responses in different models should be taken into account due to the dual role of adenosine receptors and different responses of microglia, which can be elicited with different stimuli and environmental conditions [
57]. In fact, in the brain, it is the blockade rather than the activation of A
2AR than reduce microglia activation and neuroinflammation upon different noxious stimuli [
58,
24]. This probably contributes to the neuroprotection afforded by A
2AR antagonists in brain diseases with a neuroinflammatory involvement such as ischemia, epilepsy, traumatic brain injury, multiple sclerosis or Alzheimer’s or Parkinson’s disease (reviewed in [
25]). Accordingly, we also observed that A
2AR blockade prevented the LPS- and the EHP-induced RGC death in retina organotypic cultures. This might result from the ability of A
2AR to control the activation of microglia and the production of pro-inflammatory cytokines that we showed to be sufficient and necessary to trigger RGC death, but it may also involve an ability of A
2AR to directly control neuronal viability. In fact, neuronal A
2AR can directly affect the degeneration of mature neurons upon exposure to different stimuli (e.g. [
59,
60]), namely to pro-inflammatory cytokines [
61], whereas they have an opposite effect in immature neurons [
62,
63] and during neurodevelopment [
64].
In our work, the relevance of the A
2AR modulation system in the control of RGC death through a control of neuroinflammation in the retina is further underscored by the observed up-regulation of this system in retinal organotypic cultures exposed either to LPS or to EHP. In fact, LPS and EHP caused an increase in the extracellular levels of ATP. The cellular source of this extracellular ATP is not clear, but it can be released from different cells in the retina, such as RGCs [
65], microglia [
66] and Müller cells [
67]. Moreover, recent work demonstrated that astrocytes present in the optic nerve head can also release ATP through pannexin channels in response to a mechanical strain, suggesting this mechanism as a source of extracellular ATP under chronic mechanical strain, as occurs in glaucoma [
68]. Actually, elevated levels of extracellular ATP have been reported in the retina as a response to an acute rise in ocular pressure [
69,
70], and the ATP levels are elevated in the aqueous humour of patients with primary acute and chronic angle closure glaucoma, which presents evidence for a contribution of the purinergic signaling in this disease [
71,
72]. The increased levels of ATP can function as a danger signal [
73] and can either activate P2 receptors, namely P2X7 receptors in the retina [
74,
65,
75,
76], or be extracellular catabolized by ecto-nucleotidases into extracellular adenosine that preferentially activates A
2AR [
77,
36]. Remarkably, EHP and LPS not only bolstered the source of adenosine activating A
2AR but also triggered an increased expression of A
2AR, which was translated into an increased density of A
2AR in microglia. This is in accordance with the up-regulation of A
2AR that is observed upon different noxious conditions (reviewed in [
78,
25]), namely in microglia [
79,
24,
80]. Thus, noxious stimuli such as LPS or EHP triggered an up-regulation of the A
2AR system in retinal microglia, which critically contributes to the development of neuroinflammation and RGC death. We cannot rule out the role of A
2AR present in other cell types of the retinal organotypic culture, but in the GCL, A
2AR was found to be mainly located in microglia. Furthermore, additional studies will be required to determine if A
2AR blockade only affords a prophylactic benefit or may also be therapeutically effective.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MHM, RAC, AFA and ARS conceived and designed the experiments. MHM, FE, RB and FQG performed the experiments. MHM, FE, RB, FQG, RAC, AFA and ARS analyzed the data. RAC, AFA and ARS contributed the reagents/materials/analysis tools. MHM, FE, RB, RAC, AFA and ARS wrote the paper. All contributing authors have read and approved the final version of the manuscript.