Purinergic Modulation of Interleukin-1β Release from Microglial Cells Stimulated with Bacterial Endotoxin

N9 and N13 microglial cell lines were a kind gift of Dr. Paola Ricciardi-Castagnoli (University of Milano, Italy) and were grown in RPMI 1640 medium (PAA, Linz, Austria) supplemented with 2 mM glutamine and 10% (heat-inactivated) FCS (Life Technologies Ltd., Paisley, Scotland), 100 U/ml penicillin, and 100 μg/ml streptomycin as described previously (2). Human monocytes were isolated from buffy coats by one-step gradient (Percoll; Pharmacia Biotech SpA, Cologno Monzese, Italy) or by adherence on plastic Petri dishes. After isolation, cells were kept in culture for 5 d in RPMI medium containing 2 mM glutamine, 5% human serum, 100 U/ml penicillin, and 100 μg/ml streptomycin. IL-1β and IL-6 in the supernatant of LPS (Sigma Chemical Co., St. Louis, MO) treated cells were measured with the Intertext-1βX mouse IL-1β ELISA kit and Intertext-6X mouse IL-6 ELISA kit, respectively (Genzyme srl, Cinisello Balsamo, Italy). All reagents used were dissolved in endotoxin-free water (Sigma) and checked for endotoxin contamination. Periodate-oxidised ATP (oATP) was synthesized by Dr. S. Hanau as described in reference 13.

Microglial cells (25 × 10³/ well) were plated in microtiter plastic wells in culture medium and incubated in a CO₂ incubator at 37°C in the absence or presence of LPS for 24 h. At the end of this incubation, the monolayers were thoroughly rinsed with saline solution and supplemented with 100 μl of a special diluent buffer (FireZyme Ltd., San Diego, CA) to stabilize extracellular ATP and directly placed in the test chamber of a luminometer (FireZyme). Then, 100 μl of luciferin-luciferase solution (FireZyme) was added, and light emission was recorded. As a control, total cellular ATP content was also routinely monitored. Under resting conditions, extracellular ATP amounted to ∼10–15% of total.

Fig. 1 shows that a 24-h incubation in the presence of 10 μg/ml LPS triggers release of IL-1β and that this is blocked by pretreatment with the selective P2Z/P2X₇ inhibitor (13) oATP. To show that the effect of oATP is not due to a nonspecific inhibition of cell responses, we have also monitored IL-6 release, which is much less affected. As further proof that oATP does not have nonspecific effects, we show that IL-1β release is restored in LPS-treated, oATPinhibited cells by the K⁺ ionophore nigericin, an agent known to cause IL-1β release through a receptor-independent pathway (4, 10).

Autocrine/paracrine stimulation of purinergic receptors can also in principle be prevented by exogenously added ATP-consuming enzymes such as apyrase or hexokinase. Fig. 2,A shows that apyrase completely inhibits LPS-dependent IL-1β release (the inactivated enzyme has no such effect). Surprisingly, hexokinase does not inhibit but rather stimulates IL-1β release. The main difference between apyrase and hexokinase is that the first hydrolyzes ATP and ADP, thus generating AMP, whereas hexokinase uses ATP as phosphorus donor to phosphorylate glucose, thus generating glucose 6 phosphate and ADP. It is known that ADP is an agonist at P2Z/P2X₇ receptor, though less potent than ATP (12). Thus we checked whether the potentiating effect of hexokinase is mediated by stimulation of the P2Z/ P2X₇ receptor by accumulated ADP. This seems to be the case because pretreatment with oATP blocks IL-1β secretion due to the combined addition of LPS and hexokinase (Fig. 2,A), and more importantly, exogenous ADP (ADP_e) is a much more potent stimulus than ATP (Fig. 2 B). These experiments suggest that IL-1β release could be modulated by ATP_e and ADP_e, probably released by the inflammatory cells themselves under LPS stimulation.

An obvious sine qua non of this hypothesis is that microglial cells must release ATP in response to LPS. Fig. 3,A shows that microglial cells chronically stimulated with LPS release ATP. Because the incubation medium is changed right before ATP determination, extracellular ATP measured in this experiment is very likely not accumulated in the bulk phase but continuously generated by the microglial cells. In support of this interpretation, we consistently found very little extracellular ATP in the cell-free supernatant (not shown). This observation is consistent with that previously reported by Filippini et al. (14) in T lymphocytes. The LPS dose– response curve for ATP release closely matched that for IL-1β release, as shown in Fig. 3,B. It has been shown previously that ATP is a powerful stimulus for IL-1β secretion from macrophages (10, 11), thus suggesting that this nucleotide might also have a role in autocrine/paracrine stimulation of these cells. In support of this hypothesis, Fig. 4 shows that ATP is released by human macrophages isolated from three different subjects after stimulation with LPS.

The mechanism of IL-1β processing and release is a key issue in immunology (5–9, 15). Rather surprisingly, recent evidence points to a decrease in cytoplasmic K⁺ as a pivotal stimulus for ICE activation and IL-1β maturation (4, 10). However, LPS itself does not directly activate plasma membrane K⁺ channels, and mouse microglial cells express inwardly but not outwardly rectifying K⁺ channels (16), thus raising the issue of the mechanism responsible for lowering the cytoplasmic K⁺ concentration. It has been suggested that this might be achieved by a LPS-dependent increase in the number of voltage-dependent K⁺ channels in the macrophage plasma membrane (4), but typical K⁺ channel inhibitors blocked IL-1β release only at concentrations far above those necessary to inhibit these channels (4). The P2Z/P2X₇ receptor is a good candidate to mediate cytoplasmic K⁺ depletion. This receptor is typically expressed in macrophages and macrophage-like cells (2, 17, 18), and it is modulated by inflammatory cytokines (17, 19). A brief stimulation with ATP_e triggers massive K⁺ efflux (12) and release of processed IL-1β (2, 10, 11), whereas a sustained activation causes cell death (1, 17, 20). Our data suggest that IL-1β release from microglial cells requires a double stimulation: first, LPS-dependent transcription of the IL-1β gene and cytoplasmic accumulation of proIL-1β; second, paracrine/autocrine activation of the P2Z/P2X₇ receptor that causes release of the mature cytokine. Adenine nucleotides can originate from many different sources: (a) the microglial cells themselves can release ATP_e, either spontaneously or under LPS stimulation; (b) injured or damaged cells certainly release significant amounts of this nucleotide, a process likely to occur in vivo at sites of inflammation; (c) in the central nervous system, ATP_e can be released by neurons that establish close contact with the microglial cells.

It might seem paradoxical that ADP is a better IL-1β– releasing agent than ATP, although notoriously it is a less potent stimulus at the P2Z/P2X₇ receptor. However, this is not unexpected because ADP, in contrast with ATP, is devoid of cytotoxic activity, and data from our laboratory show that release of IL-1β is optimal in response to a submaximal, noncytotoxic stimulation of the P2Z/P2X₇ receptor, such as that due to ADP_e (D. Ferrari and F. Di Virgilio, manuscript in preparation).

Involvement of the P2Z/P2X₇ purinergic receptor in LPS-dependent IL-1β release may allow the development of new pharmacological antagonists (i.e., oATP and derivatives) to modulate the in vivo production of this cytokine in pathological conditions such as septic shock or chronic inflammatory diseases.

We are grateful to Dr. Paola Ricciardi-Castagnoli (National Research Council and University of Milano) for the gift of N9 and N13 mouse microglial cell lines.