Extracellular ATP is a mediator of intercellular communication and a danger signal. Release of this and other nucleotides modulates microglia responses via P2Y and P2X receptors, among which the P2X7 subtype stands out for its proinflammatory activity and for up-regulation in a transgenic model of Alzheimer disease and in brains from Alzheimer disease patients. Here we show that amyloid β (Aβ) triggered increases in intracellular Ca2+ ([Ca2+]i), ATP release, IL-1β secretion, and plasma membrane permeabilization in microglia from wild-type but not from P2X7-deleted mice. Likewise, intra-hippocampal injection of Aβ caused a large accumulation of IL-1β in wild-type but not in P2X7−/− mice. These observations suggest that Aβ activates a purinergic autocrine/paracrine stimulatory loop of which the P2X7 receptor is an obligate component. Identification of the P2X7 receptor as a non-dispensable factor of Aβ-mediated microglia stimulation may open new avenues for the treatment of Alzheimer disease.
Microglia are the main immune cells in the brain where they play a key role in changes occurring during inflammatory and neurodegenerative disorders, Alzheimer disease included (1). The key pathogenetic event in Alzheimer is accumulation of amyloid β (Aβ)3 ; however, how this protein, whether in the soluble or fibrillar form, triggers neuroinflammation is as yet an open question (2, 3, 4). Participation of microglia to Aβ-mediated neuronal loss is debated as this cell type may have a protective effect by phagocytosing amyloid fibrils (5, 6, 7), but on the other hand, there is also evidence that inflammatory cytokines or oxidant species released from microglia might be detrimental for neurons (8, 9, 10). Despite unequivocal evidence for Aβ-mediated activation of microglia, the plasma membrane receptors involved are largely unknown. Different receptors, among which CD36 (9), α6β1 integrins (11), CD14 (12), the formyl peptide receptor-like protein (13), TLR2 (14) are ligated by Aβ, thus leading to microglia activation, generation of inflammatory mediators and Aβ internalization. Recent findings raised the possibility that soluble Aβ oligomers, or protofibrils are the real pathogenic species, but the identity of the cellular receptors mediating the biological effects of soluble Aβ peptide is obscure (2). The intracellular signal transduction cascade is also poorly understood despite circumstantial evidence for a role of tyrosine kinases and cytoplasmic Ca2+ ([Ca2+]i) increases (15, 16).
In recent years the P2X7 receptor has attracted increasing interest as a possible player in neuroinflammation (17, 18, 19, 20, 21). More recently it has been shown that stimulation of microglia P2X7 receptor causes neuronal damage via release of activated oxygen species (10), that blockade of this receptor attenuates inflammatory brain damage caused by intrastriatal injection of bacterial endotoxin (22), and that Aβ may cause ATP release from microglia (23). This observation put ATP and P2 receptors at the heart of research in neurodegenerative diseases (24). However, the possible participation of the P2X7 receptor in Aβ-mediated brain damage, despite some intriguing recent data pointing in this direction (25, 26, 27), has not been investigated in depth. Likewise, the interplay between extracellular ATP, Aβ, and [Ca2+]i, as well as the role of ATP and P2 receptors in extracellular amyloid processing have been largely overlooked, exception made for a few pioneeristic studies (28, 29, 30). In the present work we show that the P2X7 receptor is an obligate participant in microglia activation by Aβ. This observation may open new avenues for the elucidation of the molecular mechanisms of brain damage in Alzheimer disease and for the development of novel therapies.
Materials and Methods
Cells and solutions
Mice deleted of the P2X7 receptor (P2X731): 125 mM NaCl, 5 mM KCl, 1 mM MgSO4, 1 mM NaH2PO4, 20 mM HEPES, 5.5 mM glucose, 5 mM NaHCO3, 1 mM CaCl2, pH 7.4. When indicated, CaCl2 was omitted and 500 μM EGTA added. Primary mouse microglial cells were isolated from 2- to 4-day-old postnatal mice as described by Ciccarelli et al. (32). More than 98% of the cultured cells were identified as microglia using a macrophage cell-specific F4/80 biotinylated mAb (Serotec) followed by staining with Oregon Green 488 goat anti-rat IgG (Molecular Probes) (33). Cultures were used for experiments 24 h after plating. Aβ25–35, Aβ35–25 (iAβ), and oxidized ATP (oATP) were purchased from Sigma-Aldrich; Aβ1–42 and Aβ42–1 (iAβ) were purchased from Bachem. All experiments were performed at 37°C.2+-containing saline solution, as described by Falzoni et al. (
Measurement of ATP release
Microglial cells were seeded in poly-Lys-pretreated microtiter plastic dishes in 100 μl of culture medium, placed directly in the test chamber of a luminometer (FireZyme), and tested online with the Enliten luciferase/luciferin kit (Promega).
Measurement of enzymatic activity and IL-1β release
Lactate dehydrogenase (LDH) and ecto-ATPase activity were measured as described previously (31
Changes in [Ca2+]i and in plasma membrane permeability
Changes in [Ca2+]i3134). Fluorescence was analyzed with an Olympus IMT-2 inverted fluorescence microscope equipped with a 40× objective (Olympus Optical). Alternatively, increases in plasma membrane permeability were monitored with ethidium bromide at the wavelength pair 360–580 nm as earlier described (31).
This study was conducted in wild-type (wt) and P2X7 KO mice weighing 35–40 g. They were housed at constant room temperature (22 ± 1°C) and relative humidity (55 ± 5%) under a 12-h light/dark cycle (lights on from 7:00 AM to 7:00 PM). Food and water were freely available. Procedures involving animals and their care were conducted in conformity with the institutional guidelines that are in compliance with national (D. L. No. 116, G. U., Suppl. 40, 18 Febbraio1992, Circolare No. 8, G. U., 14 Luglio 1994) and international laws and policies (EEC Council Directive 86/609, OJ L 358, 1, December 12, 1987; Guide for the Care and Use of Laboratory Animals, U.S. National Research Council, 1996). Protocol were reviewed and approved by the Ethical Committe of the Universities of Ferrara and Foggia. All efforts were made to minimize the number of animals used and their suffering.
The Aβ peptides (Aβ1–42 and iAβ42–1) were obtained from Bachem. All solutions were freshly prepared. Mice were anesthetized with 3 ml/kg Equithesin i.p. (composition: 1.2 g of sodium pentobarbital, 5.3 g of chloral hydrate, 2.7 g of MgSO4, 49.5 ml of propylene glycol, 12.5 ml of ethanol, and 58 ml of distilled water), and secured in a stereotaxic frame (David Kopf Instruments). The skin was shaved, disinfected, and cut with a sterile scalpel to expose the skull. Then, animals were injected with Aβ peptide (2.2 μM in 1 μl of artificial CSF (aCSF)) or plain aCSF into the right dorsal hippocampus, using the following coordinates relative to the bregma: anteroposterior: −2.0; medial lateral: +1.8; dorsoventral: −2.3. The injections were delivered at an infusion rate of 0.5 μl/min for a duration of 2.0 min. The injection needle was left in place for 5 min before withdrawal to allow diffusion from the tip and prevent reflux of the solution. The injection placement of needle track was visible and was verified at the time of dissection. Animals were kept on a warming pad until they had fully recovered from the anesthetic and hosted in individual cages to prevent damage to the scalp sutures until they were killed for tissue processing. Starting on the 7th day after surgery, animals were killed by cervical dislocation and brains were removed. Right and left hippocampi were collected by tissue dissection and immediately frozen on dry ice. Tissues were stored frozen at −80°C until analysis of IL-1β release was performed.
Data analysis and statistics
All results are expressed as averages ± SD. Statistical analysis was performed with the ANOVA Tukey’s test. Statistically significant differences from controls are indicated by ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. Statistically significant differences between treated samples are indicated by #, p < 0.05; ##, p < 0.01; ###, p < 0.001. Where not shown, bars are smaller than symbols.
Aβ triggers [Ca2+]i increase and ATP release from wt but not P2X7-deficient microglia
There is evidence that Aβ may affect [Ca2+]i homeostasis by causing release from intracellular Ca2+ stores and by increasing the plasma membrane permeability to Ca2+ (15, 35). In Fig. 1⇓ we show that stimulation of wtN13 microglial cells with Aβ25–35 increased [Ca2+]i dose-dependently. The increase, unlike that previously shown in THP-1 cells (15) was rapid, sustained and almost fully abrogated by removal of extracellular Ca2+ (Fig. 1⇓B, trace a). We have selected and characterized in the past several N13 microglial cell clones lacking the P2X7 receptor (36), then named ATP-resistant or R-N13 cells. As shown in Fig. 1⇓, D and E, Aβ25–35 was unable to elevate [Ca2+]i in these clones. The scrambled, inactive Aβ35–25 peptide (iAβ35–25) did not raise [Ca2+]i in the N13 cells, whether expressing P2X7 or not (Fig. 1⇓, C and F).
Several proinflammatory factors cause non-lytic release of ATP from mononuclear or polymorphonuclear phagocytes (37). ATP in turn feeds back onto P2Y receptors to trigger the [Ca2+]i increase. Thus we tested whether Aβ25–35 triggered ATP release from microglia. Fig. 2⇓A shows that Aβ25–35 did induce a dose-dependent release of ATP from wtN13 but not from R-N13 cells. Release of ATP was fully abrogated by the P2X blocker oATP. Maximal ATP release occurred at 40 μM Aβ25–35, but even an Aβ25–35 concentration of 20 μM triggered near maximal ATP secretion. In both wtN13 and R-N13, there was a basal ATP release in the absence of Aβ25–35 stimulation, slightly higher in R-N13 than in wtN13. Incubation of wtN13 in the presence of Aβ25–35 caused a modest release of LDH, that never rose above 10% of total LDH cell content even with 40 μM Aβ25–35, the highest concentration tested (not shown). On the contrary, in R-N13 cells Aβ25–35-dependent release of LDH was undetectable (not shown). Fig. 2⇓B shows a time course of ATP release from N13 cells in the presence of 10 μM Aβ25–35. The increase in the extracellular ATP concentration in the wtN13 supernatants was rather slow and steady over the 24 h of incubation in the presence of Aβ25–35, whereas basically no increase was observed in the supernatants of Aβ25–35-challenged R-N13 cells, or in those of control cultures incubated in the absence of Aβ25–35. Increase in extracellular LDH was negligible at all time points (not shown). The Aβ25–35 peptide is the shortest fragment that retains the toxicity of full length Aβ. Although the biological effects of the short and full length Aβ peptide overlap, we also investigated the activity of Aβ1–42. Fig. 2⇓C shows that Aβ1–42 was more potent than the 25–35 fragment, as ATP release was nearly maximal already at a concentration of 10 μM. A slight, not statistically significant, increase in ATP release was also seen in the R-N13 cells, very likely due to residual expression of this receptor by some of the cells. Although N13 cells are usually considered a reliable model of microglia, they may not faithfully reproduce all the features of primary microglia, thus we checked whether Aβ was also able to trigger ATP release from primary cultures. Fig. 2⇓D shows that in primary microglia Aβ25–35 caused a statistically significant ATP release already at a concentration of 2.5 μM, reaching plateau at 20 μM. Aβ1–42 was also an effective ATP-releasing agent from primary microglia (not shown). On the contrary, Aβ25–35 was fully inactive in microglia isolated from mice deleted of the P2X7 receptor. Interestingly, like in the R-N13 clone, basal level of extracellular ATP was significantly higher in microglia from P2X7 KO mice compared with wt.
The effect of Aβ on the extracellular ATP concentration might be indirect in that this peptide might inhibit ecto-ATPase activity, and thus facilitate accumulation of extracellular ATP. To clarify this issue, we measured ecto-ATPase activity of microglial cells in the absence and presence of the 25–35 Aβ peptide. However, the peptide had no statistically significant effect on ecto-ATPase activity, whether in wt or R-N13 microglia (not shown). Interestingly, ecto-ATPase activity was much lower in the R-N13 than in the wtN13 cells. Lower ecto-ATPase activity might explain why R-N13 cells maintained higher basal levels of extracellular ATP.
Aβ permeabilizes the plasma membrane of wt but not P2X7-deficient microglia
Ability of Aβ to trigger Ca2+ influx and ATP release only in cells expressing the P2X7 receptor suggested to us that Aβ might interact with target cells via this receptor, and prompted an investigation of Aβ effects on other P2X7 receptor-dependent functions. Large P2X7 pore formation was assessed by means of the standard YO-PRO uptake measurement. An incubation time of 6 h in the presence of Aβ25–35 was chosen because at this time point Aβ25–35- stimulated ATP release was near maximal. Incubation in the presence of Aβ25–35 made wtN13 cells permeable to YO-PRO (Fig. 3⇓). Permeabilization was prevented by coincubation in the presence of apyrase, and did not occur with iAβ35–25. Permeabilization was also fully prevented by a 2 h incubation in the presence of oATP. The R-N13 cells were as expected resistant to Aβ25–35-stimulated permeabilization, because less than 5% took up YO-PRO, and only after a 24 h incubation in the presence of Aβ25–35. As in wtN13 cells, Aβ25–35 induced full permeabilization in primary microglia, which was prevented by apyrase (Fig. 4⇓, C–F). The iAβ35–25 peptide was inactive (Fig. 4⇓, G and H). No YO-PRO uptake was induced by Aβ in microglia isolated from P2X7 KO mice (Fig. 4⇓, I–L). The intriguing P2X7-stimulating activity of the β amyloid peptide prompted us to investigate closer the kinetic of membrane permeabilization. Fig. 5⇓ shows ethidium bromide uptake from wtN13 microglial cells stimulated with Aβ25–35 or ATP or a combination of the two agonists. The challenge with Aβ25–35 triggered an increase in ethidium bromide uptake after a lag of ∼30–60 s. Compared with the initial uptake rate induced by ATP, that caused by Aβ25–35 was slower whereas maximal uptake measured 10 min after the addition of the stimulant was slightly larger, but not statistically different than that caused by ATP alone. The combined addition of Aβ25–35 and ATP caused a larger effect, which was not truly synergic but rather additive. The iAβ35–25 peptide by itself did not cause an increased ethidium bromide uptake nor did it enhance the ATP-stimulated uptake. Average ethidium uptake values from three separate determinations are shown in the inset.
Activation of the P2X7 receptor is known to cause swelling and eventual cell death (38). Fig. 6⇓ shows phase contrast pictures of wtN13 and R-N13 cells incubated in the presence of Aβ25–35, ATP, or Aβ25–35 plus ATP. The Aβ25–35 fragment caused morphological changes similar to those caused by ATP and potentiated the effects of this nucleotide. The R-N13 cells were refractory to the effect of ATP and Aβ, whether alone or in combination. Despite membrane-permeabilizing effect, the Aβ25–35 peptide had no cytolytic effect for incubations up to 24 h but strongly potentiated the ATP-dependent cytotoxicity (Fig. 7⇓). R-N13 cells were fully refractory to the combination of Aβ25–35 plus ATP.
Aβ triggers IL-1β release from wt but not P2X7-deficient microglia
It is thought that release of inflammatory cytokines is a key pathogenetic mechanism in Aβ-mediated cell damage; thus we investigated the IL-1β-releasing activity of Aβ. Aβ25–35 per se caused no IL-1β release over a 6 h incubation (not shown), nor was IL-1β release induced by the combined addition of Aβ25–35 and ATP (not shown). Incubation in the presence of LPS produced a modest release in the 20–30 pg/ml/106 cells range (Fig. 8⇓). Addition of Aβ25–35 to LPS-primed microglia triggered a large IL-1β secretion that reached 150 pg/ml/106 cells at a concentration of 60 μM Aβ25–35. Release of IL-1β from LPS-primed microglia was also potently stimulated by the addition of extracellular ATP, up to a level of 400 pg/ml/106 cells in response to 0.6 mM ATP. The finding that Aβ25–35 caused ATP release (see Fig. 2⇑) suggested that Aβ might trigger IL-1β release via an autocrine/paracrine ATP loop. To test the hypothesis, we checked the effect of apyrase on LPS-primed Aβ25–35-challenged cells. Addition of this ATP/ADPase fully prevented Aβ25–35-stimulated IL-1β release. On the other hand, apyrase had no effect on the already low IL-1β secretion stimulated by LPS alone and did not affect IL-1β release from unprimed cells. In support of a non-dispensable role of the P2X7 receptor, Aβ25–35 did not trigger IL-1β secretion from R-N13 cells (Fig. 8⇓B). The full length Aβ1–42 peptide was a more potent stimulus for IL-1β release than the Aβ25–35 fragment and, like the shorter peptide, was fully inactive in R-N13 cells (Fig. 8⇓C). Apyrase reduced but did not fully obliterate Aβ1–42-stimulated IL-1β release. Primary microglia from wt mice also released IL-1β in response to Aβ25–35 (Fig. 8⇓D) or Aβ1–42 (not shown), whereas microglial cells from the P2X7 KO mice were fully unresponsive. Interestingly, ATP was a much more potent stimulus in primary microglia than in the N13 cells, as a nucleotide concentration of 0.4 mM triggered a IL-1β release about 4-fold larger in the former than in the latter. Aβ-stimulated IL-1β secretion from primary microglia was fully blocked by apyrase. The iAβ35–25 or iAβ42–1 peptides were inactive, whether in wtN13 or wt primary microglia.
Aβ causes intrahippocampal IL-1β accumulation in wt but not P2X7−/− mice
Finally, we tested the in vivo effect of Aβ in the P2X7−/− mice. As shown in Fig. 9⇓, intrahippocampal injection of Aβ1–42 caused a large accumulation of IL-1β in the hippocampus of wt mice. On the contrary, IL-1β release in KO mice was about 3-fold (ipsilateral hippocampus) or 8-fold (contralateral hippocampus) lower. Injection of vehicle (aCSF) had no effect in wt or KO mice. Likewise, injection of the iAβ42–1 peptide was without effect (not shown).
There are few doubts that production of inflammatory mediators in the brain is one of the key feature of Alzheimer’s disease, although it is not really clear whether this has a detrimental or beneficial effect on disease progression (4, 7). This is especially true for IL-1β (39, 40), a cytokine which is massively accumulated in the brains of Alzheimer disease patients (41). Although astrocytes can also secrete this cytokine in response to Aβ, the main source is microglia. IL-1β secretion is a two-step process, the first step consisting of transcription of the IL-1β gene and accumulation of the pro-cytokine in the cytoplasm, and the second step consisting of pro-IL-1β processing by caspase-1 and release of the mature cytokine (42). For many years, the mechanism by which caspase-1 is activated to cause IL-1β release has remained mysterious, but very recent studies now show that caspase-1 activation occurs within the inflammasome in response to extracellular ATP and P2X7 receptor activation (43, 44). The tight association with the key intracellular apparatus responsible for IL-1β processing and (probably) release puts P2X7 at the very heart of inflammation.
The physiological agonist of the P2X7 receptor is ATP. For several years, the role of this nucleotide as an extracellular messenger has been neglected, but now an impressive amount of studies has shown beyond any possible doubt that ATP has a crucial extracellular messenger role in the CNS (45, 46, 47, 48), thus providing sound experimental support to the original Burnstock’s hypothesis (for review, see Ref. 49). A key issue, only partially clarified to date, has been the measurement of the actual extracellular ATP concentration. It is now clear that the ATP concentration in the extracellular space is in the low nanomolar range, but at sites of inflammation, tissue traumas, or intensive cell stimulation, its level can reach the low or even high micromolar range (45). Of course these levels refer to the average concentration in the bulk solution, whereas it is understood that actual ATP levels in the vicinity of the plasma membrane or at sites of close cell-to-cell contact can be much higher (50, 51, 52). This suggests that ATP concentrations sufficient to activate even the low affinity P2X7 receptor may build up in vivo. Role of this receptor in extracellular ATP homeostasis is intriguing; recent data suggest that it is a target as well as conduit for ATP translocation across the plasma membrane (51). Our finding that microglia lacking P2X7 maintains an extracellular ATP concentration about twice as high as that of wt cells, but on the other hand is unable to further release ATP in response to Aβ (Fig. 2⇑), suggests that the P2X7 receptor might be necessary to support stimulated but not basal ATP release.
An additional intriguing finding of our study is the ability of Aβ to trigger permeabilization of the plasma membrane, i.e., to activate the large conductance P2X7 pore. There is now an indication that the large conductance P2X7 pore is a molecular entity separated from the receptor itself, i.e., pannexin-1 (53). Thus, Aβ might permeabilize the plasma membrane by at least two mechanisms: 1) indirectly, by triggering ATP release followed by P2X7 activation; 2) directly, by activating P2X7 or pannexin-1. However, the inhibitory effect of apyrase on Aβ-mediated permeabilization and the ability of Aβ to trigger ATP release indicate that an indirect mechanism is more likely. In this regard, we were intrigued by the long lasting accumulation of extracellular ATP caused by the amyloid β peptide. In rat microglia, Kim and coworkers (23) found that stimulation with 1 μM Aβ1–42 triggered a peak ATP release 60 min after stimulation, which slowly declined over the following 12 h. On the contrary, in mouse microglia Aβ activates a mechanism that supports a continuous release for at least 25 h. Several hypothetical release pathways have been put forward, one of the most likely being pannexin-1. The identity of pannexin-1 as the ATP release pathway activated by Aβ is obviously consistent with the ability of the Aβ peptide to cause a long lasting opening of the P2X7 pore.
Due to its proinflammatory activity, ATP is increasingly considered an early inflammatory mediator, or a “danger signal” (44, 54, 55). As such, it may be included in the growing family of those molecules that are released by damaged tissues and modulate the early phase of inflammation, also known as damage-associated molecular patterns (DAMPs) (44, 56). Our experiments show that ATP release also has a role in Aβ-mediated microglia activation because in the presence of apyrase, a soluble ecto-nucleotidase, release of IL-1β is dramatically reduced. This finding can be explained by two non-exclusive interpretations: 1) Aβ triggers ATP release from microglia via an as yet unknown mechanism and ATP in turn activates P2X7; or 2) Aβ is an incomplete stimulus for IL-1β processing and secretion and as such it needs the synergistic effect of ATP acting at P2X7. It is possible that both Aβ- and ATP-dependent pathways converge at the level of the inflammasome. This is of course a testable hypothesis for future studies. As our experiments clearly show, soluble Aβ by itself has no IL-1β-releasing activity in vitro, but it needs cell priming with LPS, a feature that it shares with ATP. However, the picture is different in vivo. Here the mere intrahippocampal injection of Aβ1–42 triggered a large accumulation of IL-1β. This raises the intriguing issue of the identity of the priming stimulus that replaces LPS under these conditions. Considering the rather long (7 days) postinjection time at which intrahippocampal IL-1β levels were measured in vivo it is likely that Aβ caused the generation of additional endogenous proinflammatory factors facilitating il-1β gene expression and intracellular pro-IL-1β accumulation. In any case, whether ATP is the sole mediator of Aβ-dependent IL-1β release, or Aβ triggers secretion of several inflammatory mediators that cooperate in this process, our findings suggest that this nucleotide might be a crucial local soluble mediator of inflammation in Alzheimer disease.
The identification of extracellular ATP and P2X7 as key factors in Aβ-dependent microglia activation unveil a non-conventional mechanism in neuroinflammation and suggest new possible pharmacological targets.
F.D.V. acts as Consultant of Duska Therapeutics and Affectis AG, biotech companies involved in the development of ATP-based drugs.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by grants by the Italian Ministry of Education, University and Scientific Research, the Italian Association for Cancer Research, the Italian Space Agency, Telethon of Italy (GGP06070), the Commission of European Communities (7th Framework Program HEALTH-F2-2007-202231), the Regione Emilia Romagna (Research Programs “Innovative approaches to the diagnosis of inflammatory diseases” and “Moniter”), and local funds from the University of Ferrara.
↵2 Address correspondence and reprint requests to Dr Francesco Di Virgilio, Department of Experimental and Diagnostic Medicine, Section of General Pathology, University of Ferrara, Via Borsari 46, 44100 Ferrara, Italy. E-mail address:
↵3 Abbreviations used in this paper: Aβ, amyloid β; iAβ, scrambled amyloid β; [Ca2+]i, intracellular Ca2+ concentration; oATP, oxidized ATP; LDH, lactic dehydrogenase; R-N13, ATP-resistant N-13 cells; wtN-13, wild-type N-13 cells; aCSF, artificial CSF.
- Received October 27, 2008.
- Accepted January 20, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.