1. Introduction
Microglia show constitutive expression of p2rx4 and p2rx7 in both rodents and humans
[1][2]. Microglial P2X4Rs play critical roles in rodent models of neuropathic pain
[3][4], experimental allergic encephalomyelitis
[5][6], and chronic migraine
[7]. In keeping with these studies, relatively low concentrations of eATP (≤100 µM) evoke desensitizing non-selective inward currents and sustained outward K
+ currents in murine microglia that most likely result from activation of P2X4Rs and P2Y12Rs, respectively
[8][9][10][11][12][13]. Surprisingly, these same concentrations of eATP fail to trigger inward membrane current or Ca
2+ influx in cultured human microglia, suggesting that either the p2rx4 is downregulated in culture, the mRNA is not translated, or the properly translated protein is not trafficked to the cell surface membrane
[14][15].
In contrast, and as expected from work on mice
[16], short applications of higher concentrations of eATP (≥1 mM) or the higher affinity ATP analog, 2′,3′-O-(benzoyl-4benzoyl)-ATP (BzATP), evoke inward currents in human microglia with properties expected of a P2X7R-mediated response (
Figure 1); the current is cation non-selective, carried in part by Ca
2+, facilitated during prolonged exposure to agonists, and blocked by P2X7R antagonists
[14]. The resting membrane potential of human microglia is unknown. In rodents, it varies with age
[17] but averages around −40 mV
[9][18]. At this potential, eATP activation of P2X7Rs causes Na
+ and Ca
2+ to rush into the cell as K
+ exits. The net result is membrane depolarization. In rodents, the inward Ca
2+ current increases the concentration of free intracellular Ca
2+ ([Ca
2+]
i), which triggers cell cycle progression
[19], the release of TNF-α
[20] and plasminogen
[21], activation of the transcription factor NFAT
[22], disruption of the cytoskeleton
[23], and the production of H
2O
2 [24]. At the same time, the outward K
+ current decreases the concentration of intracellular K
+ ([K
+]
i), leading to activation of the NLRP3 inflammasome and maturation and release of the proinflammatory cytokines, IL-1β and IL-18
[25]. While it is well accepted that the inflammasome activates when [K
+]
i drops below 90 mM
[26], more recent evidence suggests that the P2X7R is not the primary K
+ efflux pathway
[27]. Rather, the two-pore K
+ channels, THIK-1 and TWIK-2, are responsible in microglia
[9][28][29] and macrophages
[30], respectively. The data supporting a role for THIK-1 and TWIK-2 are convincing, and the conclusions are firm. However, it is unclear why K
+ efflux through the P2X7R is not sufficient by itself. In the microglial study of Madry et al., the simple fact that 2 mM ATP did not evoke a P2X7R-mediated current is enough to eliminate this receptor from consideration
[9]. The lack of response is surprising because others report robust eATP-gated currents with properties unique to P2X7Rs in murine and human microglia
[14][16][31][32], perhaps suggesting that P2X7R expression resembles P2X4R expression in its sensitivity to the choice of animal, the activation state of the microglia, and/or the experimental protocols used to study the cell. Regardless, when present, activation of P2X7Rs results in a large efflux of
86Rb
+, a proxy for K
+, in J774 macrophages and presumably in microglia
[33]. Further, eATP promotes recruitment and colocalization of microglial P2X7Rs and NLRP3 to discrete sites of the subplasmalemmal cytoplasm, suggesting that the inflammasome is positioned close enough to directly sense the P2X7R-mediated local drop in [K
+]
i [34].
Figure 1. P2X signaling in microglia. Extracellular ATP evokes influx of Na+ and Ca2+ and efflux of K+ through P2X7Rs embedded in the plasmalemmal membrane. ATP-gated efflux of K+ is also thought to involve additional channels; at present, the identity of these channels is unknown but may involve 2–pore K+ channels. The decrease in [K+]i results in influx of Cl− and the maturation and release of the proinflammatory cytokine IL–1β. Activation of P2X7Rs also increases membrane permeabilization of large organic cations like YO-PRO1, decreases phagocytosis, and initiates phosphatidylserine (PS) exposure on the cell surface. While YO-PRO1 permeates the P2X7R pore, other pathways (marked with a “?”) are also thought to play a role.
2. Membrane Permeabilization and Cell Lysis
Applications of eATP that last longer than a few seconds result in membrane permeabilization, a hallmark property of P2X7R activation
[35][36][37]. Permeabilization is the process by which eATP triggers membrane transport of hydrophilic solutes with molecular masses of <900 Da in a direction determined by their electrochemical potential. The process is reversible
[38] and does not necessarily lead to cell death. The ability of eATP to permeabilize membranes was first discovered in mouse 3T3 fibroblasts
[39], rat mast cells
[40], and mouse J774 macrophages
[33][41][42], and later described in mouse microglia
[43]. Originally thought to be a unique property of P2X7Rs, it was subsequently shown to accompany the activation of purinergic P2X2Rs, P2X3Rs and P2X4Rs
[44][45][46] as well as proton-gated TRPV1 receptors
[47]. While the physiological consequence of membrane permeabilization is the loss of cytoplasmic components such as ATP, cGMP, glutamate, and spermidine, the underlying process is typically measured as uptake of polyatomic fluorescent biomarkers such as the nucleic acid stains ethidium bromide and YO-PRO-1
[36][37]. For example, in HEK293 cells expressing recombinant P2X7Rs, BzATP triggers an increase in fluorescence that develops slowly over the course of seconds to minutes as dye enters the cell and intercalates DNA
[35]. The primary route of entry is unclear
[36]. That some of the dye transits the P2X7R pore is convincingly proven using reconstituted receptors and liposomes that lack other proteins
[48]. At the same time, eATP and P2X7Rs may activate secondary transport pathways. For example, pannexin-1 is a plasma membrane cation-selective channel related to gap junction proteins that shows a close membrane association with P2X7Rs. Pharmacologically inhibiting pannexin-1 blocks the initial phase of eATP-gated ethidium uptake in human lung alveolar macrophages, demonstrating that it is partially responsible for membrane permeabilization to cationic macromolecules in these cells
[49]. The same drugs also block P2X7R-dependent release of IL-1β, suggesting that pannexin-1 is a critical component of eATP-mediated NLRP3 inflammasome activation
[49][50]. In contrast, blocking pannexin-1 has no effect on eATP-mediated membrane permeabilization of monocyte-derived human macrophages
[51] or cultured human microglia
[14], suggesting that local paracrine signaling active in tissue microenvironments may recruit distinct permeabilization pathways in a tissue-specific manner. The hypothesis that eATP gates multiple transport portals is further supported by the finding that, in some cases, P2X7Rs increase membrane permeability to both cations (YO-PRO-1 and ethidium) and anions (glutamate and Lucifer yellow) in the same cell
[42][52][53][54]. It is unlikely that large anions travel through P2X7Rs because these channels show a strong preference for cations
[35]. In addition, the uptake of cations is temperature-dependent, whereas the uptake of anions is not
[55], again supporting the presence of multiple permeation pathways in some cells. With specific regard to human microglia, eATP does not trigger uptake of large anions in cultured cells
[14], although this may reflect the downregulation of genes encoding the necessary pathway; additional experiments on cells that more closely resemble the natural phenotype of in situ microglia are needed.
The physiological and pathophysiological outcomes of membrane permeabilization are uncertain
[36] and have not been extensively investigated in microglia. As mentioned above, the process is reversible and does not necessarily lead to cell death. For example, a 30 min application of ATP or BzATP causes significant uptake of cationic YO-PRO-1 in human microglia without inducing the release of lactate dehydrogenase, demonstrating that permeabilization is not lytic under these conditions
[14]. In contrast, longer incubations result in necrosis or apoptosis in murine microglia, a result that is prevented by pre-incubation of a P2X7R antagonist
[56][57]. Cell lysis is blocked by P2X7R antagonists and absent in cells isolated from P2X7R−/− animals, suggesting that the activation of P2X7R is a critical component of the lytic pathway
[2]. Interestingly, eATP-activation of P2X2Rs permeabilizes membranes but does not kill cells, suggesting that permeabilization and lysis proceed through different intracellular signaling pathways
[45]. If cell death is not the ultimate consequence, then what purpose does permeabilization serve? The recent work of the Grutter laboratory suggests one possibility
[46]. Spermidine is a naturally occurring intracellular polyamine that acts at extracellular sites to allosterically modulate ion channel gating
[58]. To be effective, it must be secreted. Spermidine permeates multiple subtypes of P2XRs, including P2X7Rs, suggesting that these receptors represent an eATP-dependent egress pathway for polyamines, an effect that may help to explain the ability of P2X7Rs to modulate the activity of neighboring ion channels
[37].
3. Membrane Blebbing and Microvesicular Shedding
Non-apoptotic membrane blebbing occurs when the actin cytoskeleton separates from the plasma membrane; detachment allows the hydrostatic pressure within the cell to push the membrane through the actin cortex, forming an outwardly facing membrane extrusion
[59][60]. eATP, working through P2X7Rs, is a potent initiator of non-apoptotic membrane blebbing in many cell types, including human macrophages
[51] and murine microglia
[61][62][63]. Here, blebbing occurs within minutes of P2X7R activation
[62] and reverses when the agonist is removed
[51]. As is the case of membrane permeabilization (see above), the physiological sequela of membrane blebbing is uncertain. In tumor cells, blebs facilitate cytokinesis
[64], and in streptolysin-permeabilized human embryonic kidney cells, they trap damaged membrane segments and limit further cellular damage
[65]. In THP-1 monocytes
[62] and primary mouse microglia
[61], bleb formation provides a vehicle for IL-1β release. Here, eATP activation of P2X7Rs results in rapid movement of phosphatidylserine and acid sphingomyelinase to the outer leaflet of the plasma membrane, resulting in the formation of small (40–80 nm) membrane-derived microvesicles that contain IL-1β. Surprisingly, eATP does not cause blebbing in cultured human microglia
[14].
4. Cytokines and Reactive Oxygen Species (ROS)
The ability of eATP acting on purinergic P2 receptors to function as a DAMP has long established a role for eATP in regulating neuroinflammatory immune responses
[66][67]. The immune response is characterized by a proinflammatory state, which is driven by immune cell activation and the subsequent release of proinflammatory cytokines and reactive oxygen species (ROS)
[24][68][69]. In the presence of CNS injury, infection, or neurodegeneration, copious amounts of ATP are released into the extracellular environment from stressed and dying cells
[70]. At high enough concentrations, this eATP can promote macrophage and microglial activation, which drives a cascade of P2X7R-mediated events that concludes with the time-dependent release of proinflammatory cytokines, including IL-6, IL-18, TNF-α, and IL-1β
[14][69][71][72][73]. Specifically, P2X7R-mediated IL-1β release occurs through a multi-step process that requires priming first by cellular stress or pathogens (i.e., LPS) to stimulate the production of immature pro-IL-1β
[25][74][75]. Upon accumulation in the cytosol, pro-IL-1β requires a secondary hit to promote its maturation and secretion driven by caspase-1 activity. Importantly, cells within a quiescent state store caspase-1 in an inactive form (pro-caspase-1), which requires cellular stimulation via DAMPs to undergo maturation to the active form
[25]. In the case of eATP functioning as a DAMP, P2X7R activation drives significant K
+ efflux from the intracellular environment, which subsequently promotes NLRP3 inflammasome complex formation
[67][70][75]. The NLRP3 inflammasome is composed of a primary scaffold protein (NLRP3) that recruits the accessory protein ASC, which mediates pro-caspase-1 recruitment and activation
[25][67]. Upon activation, caspase-1 drives the cleavage of pro-IL-1β into its active form, which can subsequently be secreted from the cell upon microvesicle shedding from the plasma membrane
[61][62]. Importantly, P2X7R activation in microglia also promotes reactive oxygen species (ROS) formation
[24][76]. Upon stimulation, P2X7R drives ROS production via p38 MAPK-dependent NADPH oxidase activation
[24]. Notably, both the P2X7R-mediated release of proinflammatory cytokines and ROS are highly characterized in the pathophysiology of neurodegenerative diseases
[70]. In Alzheimer’s Disease, the P2X7R is significantly upregulated adjacent to the characteristic Aβ plaques where surrounding activated microglial populations are colocalized
[24]. Importantly, Aβ stimulates microglial activation, thereby driving the release of proinflammatory cytokines and ROS, which induce pro-apoptotic gene activity, thus mediating the death of neuronal cell populations and exacerbating neuroinflammation
[70][77].
5. Tumor Microenvironment
Tumorigenesis frequently occurs at chronically inflamed tissue sites
[78][79][80], where the rate at which the tumor proliferates is largely dependent on a delicate balance of immunosuppressive and immunostimulating cell types coexisting within the tumor microenvironment (TME)
[5][81][82][83][84][85]. Underlying components of the TME include stromal cells, fibroblasts, endothelial cells, and infiltrating innate (TAMs, myeloid-derived suppressor cells, dendritic cells) and adaptive (T cells) immune cells
[78][86]. Communication between cells occurs through the release of growth factors (VEGF), cytokines (IL-6), chemokines, components of the extracellular matrix, and purines to dictate tumor growth
[78][80][87]. Tumor cells express P2X7Rs that drive proliferation by enhancing cellular metabolism and angiogenesis when the concentration of eATP is relatively low
[84][88][89]. In contrast, when eATP rises to concentrations ≥100 µM as the result of hypoxic tissue necrosis, eATP promotes tumor cell cytotoxicity through sustained membrane permeabilization
[81][84][90]. The ability of ATP to kill cancer cells justifies the development of selective P2X7R agonists as a therapeutic cancer target
[90][91].
The immune cells that infiltrate tumors also express high densities of P2X7Rs, which in part determine whether these cells work to promote or eliminate tumor cells
[79][92]. P2X7R-driven NLRP3 inflammasome complex activation and subsequent IL-1β release largely account for the immunostimulating qualities of P2X7R. Namely, IL-1β secretion from dendritic cells primes antigen-specific CD8+ T-cells, which release IFN-γ to exert their antitumor effects
[79][93]. This tumor-eradicating role is exemplified in cancer models utilizing P2X7R-deficient mice (p2rx7−/−). In the absence of P2X7R, inoculated tumors both proliferated and metastasized at a faster rate compared to those brought up in wild-type mice (p2rx7+/+)
[94]. Conversely, P2X7R activation on myeloid-derived suppressor cells fosters tumor-promotion upon the production and release of immunosuppressive factors, including reactive oxygen species, arginase-1, and TGF-β1
[79][87]. Additionally, P2X7R upregulation in glioma-associated microglia drives immunosuppression upon facilitating NLRP3 inflammasome activation and IL-1β release
[92][95]. Proinflammatory IL-1β stimulates glioma cells to produce TGF-β, which mediates subsequent upregulation of VEGF, leading to tumor proliferation via increased angiogenesis
[96][97][98]. Importantly, eATP also exhibits immunosuppressive effects upon its breakdown to adenosine via ectonucleotidases CD39 and CD73. Particularly, Tregs’ characteristic immunosuppressive activity is based on its high expression level of both ectonucleotidases. Free adenosine can then target P1 A2ARs to inhibit tumor-infiltrating cytotoxic CD8+ T cells
[90][97].
6. Cell Death and Disease
Sustained stimulation with ATP is a potent catalytic stimulus for several cell types, including microglial cells, and the available literature clearly point towards the involvement of P2X7R in ATP-induced cell death, as reviewed by Peter Illes
[99]. P2X7R has been described as a death receptor
[56][100]. Short periods of P2X7R activation are cytotoxic, and once activated, the P2X7R sets in motion an irreversible death process
[101][102]. Cells primed with inflammatory mediators (e.g., lipopolysaccharide) are particularly susceptible to the toxic actions of ATP
[103], and this priming effect may alter the distribution or activation of P2X7 receptors in cell membranes
[104].
Studies in culture strongly suggest a role for P2X7R-mediated cell death in a number of neurodegenerative diseases. Specific examples include rat microglial cell lines N9 and N13
[57], murine microglial cell line EOC13
[105], mouse primary microglia
[24][102], and enteric glia
[106].
7. Oxygen Glucose Deprivation
Oxygen glucose deprivation (OGD) is often used to study ischemic cell death. It negatively impacts microglia motility and induces microglia cell death. Upregulation of the P2X4R and P2X7R is reported to occur in N9 microglial cells deprived of oxygen
[107]. Further, the same study suggests that metabolic stress like OGD induces massive release of extracellular ATP, which in turn activates cortical P2X and P2Y receptors. Several P2 receptors (P2X1R, P2X2R, P2X3R, P2X5R, and P2Y11R) alter the homeostatic balance of Ca
2+ and Na
+ fluxes, triggering both necrotic and apoptotic pathways
[108]. In a similar fashion, P2X4 and P2X7 receptors induce the microglial release of proinflammatory cytokines
[68] and subsequent neuronal death. Blocking the receptors with the P2 antagonists PPADS and TNP-ATP reduced microglia activation and rescued cortical cells from OGD-induced cell death
[107]. OGD-induced microglial cell death has also been studied in BV2 cells, where the pharmacological inhibition of P2X7R using Brilliant Blue G (BBG) significantly reduced OGD-induced BV2 cell death. Similar results were observed in neonatal hippocampal slices. Here, OGD increases extracellular ATP, and treatments that decreased the concentration of extracellular ATP or reduced the availability of P2X7R receptors inhibited OGD-induced microglia cell death
[109]. The depletion of extracellular Ca
2+ also significantly inhibits cell death, indicating that OGD induces Ca
2+-dependent microglia cell death
[109]. Further in vivo studies performed on a middle cerebral artery occlusion rat model showed that inhibition of P2X7R expression by promoting degradation of ATP protects against the brain injury produced by OGD
[110]. However, P2X7Rs are not the sole contributors to the purine- and calcium-dependent ischemic cell death and other mechanisms remain to be discovered.