The proinflammatory IL-1 cytokines IL-1α, IL-1β, and IL-18 are key mediators of the acute immune response to injury and infection. Mechanisms underlying their cellular release remain unclear. Activation of purinergic P2X7 receptors (P2X7R) by extracellular ATP is a key physiological inducer of rapid IL-1β release from LPS-primed macrophage. We investigated patterns of ATP-mediated release of IL-1 cytokines from three macrophage types in attempts to provide direct evidence for or against distinct release mechanisms. We used peritoneal macrophage from P2X7R−/− mice and found that release of IL-1α, IL-18, as well as IL-1β, by ATP resulted exclusively from activation of P2X7R, release of all these IL-1 cytokines involved pannexin-1 (panx1), and that there was both a panx1-dependent and -independent component to IL-1β release. We compared IL-1-release patterns from LPS-primed peritoneal macrophage, RAW264.7 macrophage, and J774A.1 macrophage. We found RAW264.7 macrophage readily release pro-IL-1β independently of panx1 but do not release mature IL-1β because they do not express apoptotic speck-like protein with a caspase-activating recruiting domain and so have no caspase-1 inflammasome activity. We delineated two distinct release pathways: the well-known caspase-1 cascade mediating release of processed IL-1β that was selectively blocked by inhibition of caspase-1 or panx1, and a calcium-independent, caspase-1/panx1-independent release of pro-IL-1β that was selectively blocked by glycine. None of these release responses were associated with cell damage or cytolytic effects. This provides the first direct demonstration of a distinct signaling mechanism responsible for ATP-induced release of pro-IL-1β.
The IL-1 family of cytokines comprises 11 members whose aberrant production leads to chronic inflammation. IL-1α, IL-1β, IL-1Ra, and, more recently, IL-18, have been the most studied of the IL-1 cytokines (1, 2). The synthesis, processing, and secretion of the proinflammatory IL-1 cytokines, IL-1α, IL-1β, and IL-18 are tightly regulated by several mechanisms (3). Although much is now known about the biosynthesis and posttranslational processing of these cytokines (3, 4, 5, 6, 7, 8), much remains unknown about their secretion. TLR engagement by inflammatory stimuli, such as LPS and other bacterial endotoxins, leads to cellular accumulation of these cytokines mainly via activation of NF-κB- and/or MAPK-signaling cascades. Intracellular IL-1β and IL-18 are synthesized as inactive proforms that are processed by activated caspase-1 to produce the active, mature forms. In contrast, IL-1α, which is bioactive in its unprocessed state, can also be posttranslationally processed by the calcium-dependent protease, calpain (9, 10, 11). Although LPS acting via TLR4 potently induces synthesis and accumulation of large amounts of these IL-1 cytokines intracellularly, in the absence of secondary stimuli very little is secreted and most is degraded within the cell (3, 12). There are two key physiological secondary stimuli currently known to induce rapid processing and extensive release of bioactive IL-1 cytokines: engagement by cytotoxic T cells and activation of purinergic P2X7 receptors (P2X7R) by extracellular ATP (3, 5, 6, 7, 8, 12, 13).
P2X7Rs in unactivated monocytes and macrophage show little or no functional activity but are up-regulated and become functional in response to LPS and other inflammatory stimuli (3, 14). P2X7R activation then leads to the rapid assembly and activation of a specific caspase-1-dependent multiprotein complex, the cryopyrin inflammasome, which includes the adapter protein, apoptosis-associated speck-like protein containing a C-terminal caspase-activating recruiting domain (ASC),3 and the pyrin domain-containing protein cryopyrin/Nalp3 (15, 16, 17). The assembly and activation of this cryopyrin inflammasome results in release of fully processed IL-1β and IL-18. However, mechanisms of the release process subsequent to inflammasome activation remain enigmatic.
With the exception of the secreted form of IL-1Ra, IL-1 cytokines lack a signal peptide and do not follow the classical endoplasmic reticulum/Golgi secretory pathway. Several models of IL-1β release from activated monocytes/macrophage or microglia in response to ATP have been proposed (3, 4, 5, 6, 7, 8). Because prolonged stimulation with high concentrations of ATP were long known to lead to cell death (18, 19), a result we now know to be due to P2X7R activation (20), the original notion was that active cell death processes were responsible for the initiation of processing and release of mature proinflammatory IL-1 cytokines. Evidence has been presented that these processes may include lysosomal localization and release of IL-1β (8, 21, 22), or direct transport across the membrane via as yet unidentified membrane transporters (7). More recently, it has become clear that brief P2X7R activation induces substantial release of mature IL-1β and IL-18 in the absence of cell damage or lysis (3, 5, 23). Shedding of plasma membrane microvesicles containing IL-1β (5, 24, 25), and/or release of a separate population of IL-1β-containing exosomes (8) have been proposed as noncytolytic mechanisms for release of IL-1 cytokines. It is possible that this multiplicity of proposed mechanisms may be explained by the existence of multiple distinct release processes. This could explain apparently discordant results being obtained by different groups, or even within the same research laboratory. For example, ATP-mediated release of IL-1β from different macrophage and microglia has been variously observed to have a strict requirement for extracellular calcium or to have little or no dependence on extracellular calcium (8, 26). Further support for the existence of multiple release processes for IL-1 cytokines comes from examination of Western blots of IL-1β release from a variety of LPS-primed immune cells in response to P2X7R activation (e.g., Refs. 7 , 11 , 23); it is clear different forms of IL-1β can be released in widely differing proportions, with one extreme example being release of pro-, but not mature, IL-1β from RAW264.7 macrophage (23).
Although commented upon previously (27), scant notice has been paid to the possible functional sequelae of pro-IL-1β release, or to the possibility of caspase-1-independent, noncytolytic mechanisms underlying such release. We have recently identified pannexin-1 (panx1) as a key component in P2X7R-mediated activation of the caspase-1 cascade and release of mature IL-1β and found that inhibition of panx1 abolished ATP-induced activation of caspase-1 and release of mature, but not pro-, IL-1β (28, 29). Therefore, the aim of the present study was first to ask whether panx1 may also be involved in the release of other proinflammatory IL-1 cytokines and second whether selective inhibition of panx1-dependent IL-1 release may provide direct evidence for distinct release pathways. In this study, we show that panx1 is involved in a component of ATP-mediated release of all three proinflammatory IL-1 cytokines and, most importantly, we provide the first direct evidence for the existence of a distinct release pathway for pro-IL-1β that involves neither cytolytic processes nor caspase-1-mediated events.
Materials and Methods
ATP, glycine, nigericin (NIG), and Escherichia coli LPS were obtained from Sigma-Aldrich, maitotoxin (MTX) was obtained from Alexis, and BAPTA-AM was obtained from Calbiochem. The pannexin-1 mimetic peptides 10panx1 (WRQAAFVDSY) and 14panx1 (SGILRNDSTVPDQF) were synthesized with a purity >95% by Alta Biosciences. Capture and biotinylated Abs used for ELISA were obtained from R&D System for mouse IL-1β, IL-1α, and TNF-α and from MBL for mouse IL-18.
Mice and cell culture
Adult male C57BL/6 and P2X7R−/− mice (30) from 35- to 45-days old were sacrificed by rising CO2 concentrations in accordance with the regulations of the U.K. Animal Scientific Procedures Act of 1986. In some experiments (where indicated), BALB/c mice were used according to the same procedures and protocols. The peritoneal cavity was lavaged twice with 5 ml of Dulbecco's PBS without calcium, magnesium (Invitrogen). The recovered buffer from two to three mice was pooled; cells were collected by centrifugation (250 × g, 5 min) and plated in 12-well plates at a density of 1.5 × 106 cell/well in RPMI 1640 medium (Invitrogen) supplemented with 10% FCS (Invitrogen), 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen). The macrophages were allowed to adhere overnight (37°C, 5% CO2) and washed with fresh medium to remove unattached cells before use. Murine macrophages cell line RAW264.7 (American Type Culture Collection (ATCC) TIB-71) and J774 A.1 (ATCC TIB-67) were cultured in DMEM (Invitrogen) and HEK293 cells were cultured in F12 medium, all supplemented with 10% FCS. HEK293 cells were transiently transfected using Lipofectamine 2000 (Invitrogen).
Total RNA was isolated using an RNeasy Mini kit (Qiagen), followed by reverse transcription using SuperScript III (Invitrogen) RNase H-reverse transcriptase with oligo-dT. Specific primers for mouse panx1, IL-1β, caspase-1, ASC, NALP3, or β-actin (Table I) were used in PCR with BIOTAQ DNA Polymerase (Bioline). Amplification reactions were performed in a JMBS 0.2G (Hybaid) for one cycle of 94°C for 2 min, 30 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, followed by one cycle of 72°C for 7 min. PCR products were resolved in 1% agarose (Sigma-Aldrich) gel containing 0.5 μg/ml ethidium bromide (Sigma-Aldrich), the obtained product sizes for the different amplifications were as expected from their mRNA sequences (Table I).
|Name .||Sequence (5′ to 3′) .||Fragment .||Gene ID .|
|Name .||Sequence (5′ to 3′) .||Fragment .||Gene ID .|
Dye uptake experiments were performed as previously described (28, 29). Briefly, macrophages were plated on glass coverslips the day before and incubated 3 min with 20 μM ethidium (Sigma-Aldrich) in a standard physiological extracellular solution consisting of: 147 mM NaCl, 10 mM HEPES, 13 mM glucose, 2 mM CaCl2, 1 mM MgCl2, and 2 mM KCl. Fluorescent images were recorded on a Nikon confocal microscope under a ×20 objective at 5-s intervals for 10 min during superfusion at 37°C with 5 mM ATP (Sigma-Aldrich). For each experiment, the time course of ethidium fluorescence was measured for 30 isolated cells and then averaged to obtain the mean fluorescent signal, the slope of the fluorescent signal vs the time was used as the most accurate and consistent measurement for comparisons. Digitonin (100 μM; Calbiochem) was used at the end of the experiments to induce maximum dye uptake.
RAW264.7 and J774 A.1 macrophages were seeded at 2 × 106 cells/well in a 6-well dish; primary peritoneal macrophages were seeded as described above on the day before experiment. Where indicated, cells were treated with 100 μM of the cell-permeable, irreversible inhibitor of caspase-1 (Ac-YVAD-2,6-dimethylbenzoyloxymethyl ketone (AOM), caspase-1 inhibitor IV; Calbiochem) and then stimulated with LPS (0.1–1 μg/ml) for 4 h, washed twice with the same solution used for dye uptake experiments (except when high K+ extracellular solution was used, where 130 mM KCl was used replacing NaCl) and incubated for 20–30 min with 3–5 mM ATP, 0.5–2 nM MTX, or 5–20 μM NIG, in the presence or absence of 5 mM glycine or the indicated concentrations of panx1 mimetic peptides. In experiments examining calcium dependence, extracellular solution contained 0 MgCl2/0.2 mM CaCl2 or cells were preincubated with 1 mM BAPTA-AM for 20 min before, and during, ATP stimulation. Cells were lysed in RIPA buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl2, and 1 mM CaCl2) supplemented with 1% Triton X-100 and Complete protease inhibitor mixture (Roche) for 1 h at 4°C and centrifuged to remove particulate matter. Cell supernatants were clarified by brief centrifugation and 60–85% of the volume was concentrated using 10-kDa nominal molecular mass cutoff filters (Millipore). A total of 10–15 μg of total protein and the concentrated supernatants were resolved in 4–15% polyacrylamide gels, transferred to polyvinylidene fluoride membranes (Millipore) by electroblotting and immunoblotted with 1/1000 anti-IL-1β Ab (3ZD; Biological Resources Branch, National Cancer Institute), with 1/500 anti-caspase-1 p10 Ab (M-20; Santa Cruz Biotechnology), with 1/1000 anti-ASC (AL177; Alexis), or with 1/2000 anti-P2X7R Ab (C-terminal; Alomone), following HRP-conjugated secondary Ab (DakoCytomation) at a 1/2000 dilution and detection using the ECL-plus kit (Amersham) and Kodak Bio-Max MS film (Sigma-Aldrich). β-actin-HRP conjugated Ab (Santa Cruz Biotechnology) was used to check protein loading and to normalize protein content between lanes. Blots were analyzed by densitometry measurements using NIH ImageJ software (http://rsb.info.nih.gov/ij/).
Lactate dehydrogenase (LDH) release assay
The presence of LDH in the medium was measured in all the experiments using the Cytotoxicity Detection kit (Roche) following the manufacturer's instructions and expressed as the percent of the total amount of LDH in the cells. Average results from triplicate samples are expressed as the mean ± SEM from 3 to 10 independent experiments per condition. Results reported in this study were obtained from experiments in which LDH release changed by <3% from control values (which ranged from 0.3 to 0.9% of total cellular LDH content). Experiments which resulted in LDH levels greater than this were excluded from this study in order to exclude, as much as possible, the involvement of cytotoxic actions on the release processes; ATP caused >25% LDH release in 4 of 30 experiments and these four experiments were excluded from analysis.
Average results are expressed as the mean ± SEM from the number of assays indicated. Data were analyzed by an unpaired Student t test to determine the difference between groups using Instat (GraphPad) software (∗∗, p < 0.001).
P2X7R-induced release of IL-1 cytokines from mouse peritoneal macrophage is panx1 dependent
Initial experiments were performed to determine whether panx1 is likely to be involved in the release of the other proinflammatory IL-1 cytokines, IL-1α and IL-18, by activation of P2X7R. We first directly demonstrated that it is the P2X7R, and not other purine receptors, that is responsible for ATP-mediated release of IL-1α and IL-18, as well as IL-1β, by measuring levels of these three cytokines in peritoneal macrophage obtained from wild-type and P2X7R−/− mice. There were no differences in the up-regulation of IL-1α, IL-1β (Fig. 1,A), or IL-18 (data not shown) by LPS, demonstrating that the presence of P2X7R is independent of TLR signaling to up-regulation of genes encoding for IL-1 cytokines. In agreement with previous studies (3, 5, 6, 7, 8, 12), in wild-type peritoneal macrophage insignificant amounts of these cytokines were released into the medium during the 4-h treatment with LPS while 20 min application of a maximum concentration of ATP (5 mM) resulted in IL-1α, IL-1β, and IL-18 release (Fig. 1, B, D, and E, see also Fig. 3,A). ATP evoked no release of IL-1α, IL-1β, or IL-18 (Fig. 1, B, D, and E) in peritoneal macrophage from P2X7R−/− mice. In contrast, peritoneal macrophage from wild-type and P2X7R−/− mice showed comparable release of TNF-α in response to LPS and ATP had no effect on release of this TLR-associated proinflammatory cytokine (Fig. 1 C), indicating a specific involvement of P2X7R in release of IL-1 cytokines.
The panx1-mimetic inhibitory peptide 10panx1 (28, 29) was as effective in inhibiting ATP-evoked release of IL-1α and IL-18 as it was in inhibiting IL-1β (Fig. 1,F) with half-maximal inhibitory concentrations (IC50 values) of 0.55, 0.58, and 0.57 mM for IL-1β, IL-1α, and IL-18, respectively (Fig. 1,H). Another panx1-mimetic peptide, 14panx1 (28, 29), had no significant effect on ATP-induced release of these cytokines even when used at very high concentrations (Fig. 1, F and H). We previously found the 10panx1 peptide blocked caspase-1 activation as an upstream step in the release of mature IL-1β in response to extracellular ATP (28), therefore we tested a cell-permeable caspase-1 inhibitor (Ac-YVAD-AOM) in the ATP-induced release of the other IL-1 cytokines. The release of IL-1β, IL-1α (Fig. 1,G), and IL-18 (data not shown) was significantly blocked by caspase-1 inhibition, but, as expected, had no effect on the release of TNF-α (Fig. 1 G). We also examined whether 10panx1 might directly block caspase-1 activity in a standard in vitro caspase-1 assay and found the same enzyme activity in the presence or absence of 10panx1 (caspase-1 activity (relative units) of 381 ± 28; 308 ± 42; 311 ± 37; and 320 ± 37 in the presence of 0, 400, 800, and 1600 μM peptide, respectively). Thus, this panx1-mimetic inhibitory peptide does not interfere with caspase-1 enzymatic activity per se.
We examined the kinetics of 10panx1 actions on ATP-induced IL-1 cytokine release from LPS-primed peritoneal macrophage (Fig. 2). There were no significant differences in the amount of inhibition by 10panx1 when it was present for up to 30 min before ATP challenge or when coapplied with ATP (Fig. 2,A). Macrophage incubated for 30 min with 10panx1 and then washed for 60 min before ATP challenge resulted in 50% inhibition of IL-1β release (Fig. 2,B). This slow washout indicates that the peptide binds strongly to its target. Moreover, 10panx1 significantly inhibited ATP-induced IL-1α and IL-1β release even when it was added after ATP had been present for 10 min (Fig. 2, C and D), indicating that it rapidly inhibits the release mechanism itself. An important observation was that high concentrations of 10panx1 (≥1 mM) applied for periods >3–4 h were clearly toxic, as evidenced by substantial loss of cells and very high levels of LDH release at 24 h (>32%). We have not investigated these toxic effects of long-term application of 10panx1 in this study and have limited its application to ≤30 min in all additional experiments.
Macrophage show differential IL-1β-processing/release mechanisms
We noted a considerable variability in the inhibition of ATP-mediated IL-1β by 10panx1 in these ELISA measurements (range: 50–95% inhibition with maximum 10panx1 concentrations). Because the Ab used in the above experiments recognizes all forms of IL-1β, we investigated whether this variability may be due to differential release of different forms of IL-1β with 10panx1 inhibition possibly selective for one form of release. Therefore, we compared IL-1β immunoblots from three types of mouse macrophage: peritoneal macrophage, J774A.1, and RAW264.7 macrophage cell lines. We first compared cellular and released IL-1β from peritoneal macrophages obtained from wild-type BALB/c and C57BL/6 as it has been reported that several downstream effects of P2X7R activation (Ca2+ influx, dye uptake, and phosphatidylserine exposure) are markedly diminished in T cells from C57BL/6 mice (31), which is the background species of the P2X7R−/− mice used in our study. However, Western blot analysis showed similar levels of LPS-induced synthesis of intracellular 35-kDa pro-IL-1β (Fig. 3,A, second lanes) as well as ATP-induced release of mature 17-kDa IL-1β (Fig. 3,A, third lanes) from LPS-primed peritoneal macrophage obtained from either mouse species. These results are in accord with previous studies which have not observed a difference between C57BL/6 and BALB/c when examining either P2X7R-induced membrane currents (32) or IL-1β release from bone marrow-derived macrophage (23). We further confirmed that all ATP-mediated IL-1β release was due solely to activation of P2X7R because no release of any form of IL-1β was observed in ATP-treated LPS-primed peritoneal macrophages obtained from P2X7R−/− mice (Figs. 1,B and 3 A and Refs. 3 , 8 , 30 , 33).
Both J774A.1 and RAW264.7 macrophage have been shown to have functional P2X7Rs but while ATP induces the release of mature IL-1β from J774A.1 macrophage, it has been observed to release only pro-IL-1β from RAW264.7 macrophage (23, 28). As expected, no forms of IL-1β were detected intracellularly or in the supernatant in the absence of LPS priming (Fig. 3,B, lane 1 in each panel). Incubation with LPS for 4 h resulted in intracellular expression of pro-IL-1β (35-kDa form) and a ∼20-kDa form (p20) in all macrophage types with the p20 form more prominent in RAW264.7 and J774A.1 cells than in peritoneal macrophage (Figs. 3, A and B, lanes 2). Presence of a p20 form has often been observed previously but not investigated (e.g., Refs. 25 and 34). ATP induced the release of 35-kDa pro-IL-1β from RAW264.7 macrophage but never resulted in release of p20 or mature (p17) IL-1β (Fig. 3,B, lanes 3 and 4, n = 14 preparations). In contrast, J774A.1 macrophage and peritoneal macrophage released high levels of mature IL-1β in response to the same stimulation (Fig. 3, A and B, lanes 3); peritoneal macrophage were significantly more efficient than J774A.1 cells in releasing IL-1β as virtually no intracellular IL-1β remained after ATP stimulation of these cells (Fig. 3,A). In contrast to the obvious impairment of mature IL-1β release from RAW264.7 macrophages, IL-1α release from these cells was qualitatively similar to that measured from peritoneal macrophages (Fig. 3 C).
We investigated the nature of the p20 form in LPS-primed J774A.1 macrophage (Fig. 3,D). Maximal stimulation of P2X7R with 5 mM ATP completely depleted the cell-associated p35 and p20 forms with a concomitant strong increase in the fully processed p17 form in the supernatant (Fig. 3,D, left panels). Incubation with the irreversible, cell-permeable caspase-1 inhibitor (Ac-YVAD-AOM) before LPS priming did not alter cellular accumulation of either p35 or p20 forms of IL-1β, although, as expected, it completely blocked release of processed IL-1β (Fig. 3 D, right panels). Additionally, no mRNA splice variants were found for the IL-1β transcript in either RAW264.7 or J774A.1 macrophage that could explain an alternative smaller form (data not shown). Taken together, these results demonstrate that while caspase-1 is not responsible for the processing of pro-IL-1β (p35) to the p20 form, it can effectively process the p20 form to fully mature p17 IL-1β, presumably via N-terminal cleavage.
RAW264.7 macrophage do not process IL-1β because they do not express ASC
We first asked whether defects in P2X7R, panx1, and/or P2X7R-panx1 interactions might be responsible for the inability of RAW264.7 macrophage to process IL-1β. However, protein levels of P2X7R were significantly higher in RAW264.7 cells than in J774A.1 macrophage (Fig. 4,A), and mRNA levels of panx1 were similar in both macrophage types (Fig. 4,B). Functional coupling between P2X7R and panx1 was also similar in all three macrophage types as assayed by ATP-induced dye uptake (Fig. 4, C and D). We then asked whether the inability of RAW264.7 macrophage to process IL-1β was specific to P2X7R signaling or to caspase-1 mechanisms in general. Therefore, we examined the effects of two nonphysiological but often used inducers of IL-1β processing and release, MTX and NIG (4, 17, 29). Again, no mature 17-kDa IL-1β was detected in the supernatant of RAW264.7 cells even with high concentrations of these toxins, although the expected robust release of mature 17-kDa IL-1β was observed in the other two macrophage types (Fig. 5). This result suggested a general defect in caspase-1 levels or its activity. Western blot analysis of caspase-1 revealed similar levels of the 45-kDa procaspase-1 in all three macrophage types, and significant amounts of active p10 caspase-1 after ATP stimulation in J774A.1 and peritoneal macrophage, but no accumulation of active p10 caspase-1 was detected in RAW264.7 macrophage (Fig. 6,A). We next used RT-PCR to examine expression levels of three key components to the P2X7R-activated inflammasome, caspase-1, ASC, and cryopyrin/Nalp3 (15, 16, 17). Strikingly, we found the complete absence of ASC mRNA in RAW264.7 macrophage with no differences in expression levels of caspase-1 or cryopyrin among the three macrophage types (Fig. 6,B). Selective absence of ASC in RAW264.7 macrophage was substantiated at the protein level using immunoblot analysis with an ASC-specific Ab (Fig. 6 C).
Differentiation of caspase-1-dependent and -independent IL-1β-release pathways
Having determined that RAW264.7 macrophage lack a functional caspase-1 inflammasome, we used these cells as a valuable tool to investigate whether P2X7R can couple to distinct release pathways. We compared the actions of glycine, which has been shown to block ATP-induced release of p35 pro-IL-1β without altering caspase-1 activation (23), with 10panx1, which blocks caspase-1 activation and IL-1β processing in J774A.1 and peritoneal macrophage (28, 29, 35). We confirmed these previous results (Fig. 7,A) and further found that while glycine prevented the release of pro-IL-1β from both peritoneal macrophage and RAW264.7 macrophage, 10panx1 did not alter release of 35-kDa pro-IL-1β from either of these macrophage types, although it completely abolished release of mature IL-1β and active caspase-1 from peritoneal macrophage (Fig. 7,A). Semiquantitative densitometry measurements were used to estimate proportions of pro- and mature IL-1β released from peritoneal macrophages during the 20-min stimulation period with ATP; the average ratio of p35/p17 was 0.43 ± 0.09 (n = 24, range 0.03–1.07). Thus, ∼40% of total IL-1β released is in the unprocessed form. In the presence of glycine, no proform could be detected, thus all released IL-1β was the mature form. Surprisingly, glycine treatment significantly reduced the release of active caspase-1 (p10) with a small increase of intracellular caspase-1 p10 (Fig. 7,A, left panels). Similarly, incubation in a high potassium external solution, which prevents caspase-1 inflammasome activation via a panx1-independent pathway (29, 36, 37), did not prevent release of pro-ILβ (Fig. 7,B). Moreover, the caspase-1 inhibitor, Ac-YVAD-AOM, did not prevent ATP-induced release of pro-IL-1β from either peritoneal or RAW264.7 macrophages (Fig. 7,C) although, as expected, it completely prevented release of mature IL-1β from peritoneal macrophages (Fig. 7,C, lane 2). The caspase-1-independent release of pro-IL-1β was also found to be calcium-independent because it was not altered by removing extracellular divalent cations (Fig. 7,D), or by strong chelation of intracellular calcium by use of high concentrations of BAPTA-AM (Fig. 7 E).
Both IL-1β-release pathways are independent of cell death processes
We, and others, have previously provided several lines of evidence that little, or none, of the caspase-1/panx1-dependent release of mature 17-kDa IL-1β by brief activation of P2X7R (e.g., conditions used in this study) can be attributed to active cell death processes (Refs. 3 , 5 , 23 and see Discussion). LDH released into the medium was measured in all experiments to assess cell damage/lysis; in keeping with previous results (29), LDH release increased from <1% total cellular LDH in the basal state (range 0.3–1.9%, n = 12) to only 2–4% in the presence of maximum ATP concentrations (1.2–4.8%, n = 12). This very small increase in LDH release cannot account for the large amounts of IL-1α, IL-18, or IL-1β released (e.g., Figs. 1 and 3). Indeed, this 2–4% increased LDH release by ATP is caspase-1 activation independent because this same small increase was observed in RAW264.7 macrophage as well as in J774A.1 and peritoneal macrophage. Moreover, the caspase-1-independent release of 35-kDa pro-IL-1β can also be shown to be cell damage/death independent not only by the lack of correlation between the very low levels of LDH and very high levels of pro-IL-1β released, but also by three further results: 1) 10panx1 actually reduced both basal and ATP-induced levels of LDH (control basal/ATP levels: 0.6 ± 0.08/3.1 ± 0.4; 10panx1 basal/ATP levels: 0.3 ± 0.04/0.9 ± 0.08, n = 6 for all conditions) but did not prevent release of pro-IL-1β from RAW264.7 or peritoneal macrophage (Fig. 7,A); 2) high external potassium solution reduced release of pro-IL-1β from RAW264.7 macrophage by ∼50% (Fig. 7,B) but did not alter the 2% increase in LDH release by ATP (n = 3); and 3) glycine completely prevented release of pro-IL-1β, but not mature IL-1β, from peritoneal macrophage and reduced by ∼80% its release from RAW264.7 macrophage (Fig. 7 A) with LDH release decreased by <3% (from 3.8 ± 0.2% to 1.05 ± 0.2%, n = 4).
From the time of the first studies linking caspase-1 to IL-1β processing and release, virtually all studies of IL-1β processing and release have focused on this caspase-1 cascade of events (1, 3, 12, 27). The present study provides the first direct demonstration that pro-IL-1β can be released from macrophage via a signaling pathway distinct from the caspase-1/inflammasome pathway and distinct from any cell death pathway. This study also provides new insight into P2X7R-mediated processing and release of IL-1 family members (IL-1α, IL-1β, and IL-18).
Several lines of evidence have led to the conclusion that extracellular ATP acts exclusively via P2X7Rs to induce release of the proinflammatory IL-1 cytokines IL-1α, IL-1β, and IL-18 from activated macrophage (3, 27, 38). This has been directly demonstrated for IL-1β using peritoneal macrophage from P2X7R knockout mice (8, 30, 33), a finding we have repeated in the present study and extended to both IL-1α and IL-18. We found that P2X7R-induced release of IL-1α and IL-18 are also both dependent on panx1, as previously demonstrated for IL-1β (28, 29). The 10panx1-mimetic peptide inhibited ATP-induced release of all three cytokines over the same concentration range, and with the same rapid kinetics. These results strongly indicate that the same mechanism underlies its inhibition of each of these cytokines. The present study, and previous studies (28, 35), showed that 10panx1-mimetic inhibitory peptide blocks caspase-1 activation; thus, it is reasonable to conclude that the panx1-dependent release of all these IL-1 cytokines is mediated via caspase-1/cryopyrin inflammasome assembly and function. This is not surprising for either IL-1β or IL-18 as these are well-established substrates for cleavage by activated caspase-1 (3), and nor is it unexpected for IL-1α which is not a substrate for caspase-1 (11) but which has been shown to be reduced in macrophage from mice deficient in either caspase-1 (39, 40) or the inflammasome adapter protein ASC (16). How this panx1-dependent caspase-1 cascade ultimately controls IL-1α release remains to be determined but likely involves cross-talk between caspase-1 and calpain, the major processing enzyme for IL-1α (41). In contrast, our results also indicate there is a caspase-1-independent release of IL-1α from RAW264.7 macrophages which is not observed in peritoneal macrophage (e.g., Fig. 3 C). Thus, it is likely there remain to be identified distinct release mechanisms for IL-1α that may overlap those of IL-1β under some circumstances. For example, it has recently been shown that IL-1α, but not IL-1β or caspase-1, is a critical component of the sterile inflammatory response in mice to accumulation of necrotic host cells (42).
How and where panx1 interacts with caspase-1 and/or other components of the inflammasome remains unknown (28, 29, 35), but the kinetics of inhibition by 10panx1 observed in the present study suggest panx1 activation, or interaction, is tightly coupled to the caspase-1-dependent release process per se. This is suggested by the finding that 10panx1 was able to inhibit IL-1β release even when it was added in the middle of a 20-min stimulation with ATP. The selectivity of 10panx1 for panx1 has been questioned in a recent study where panx1 currents or connexin-46 currents were measured from oocytes heterologously overexpressing either panx1 or connexin-46 (43). In that study, concentrations of 10panx1 (1–2 mM), which abolish processing and release of mature IL-1β (e.g., Fig. 7 and Refs. 28 , 29 , 35), reduced connexin-46 currents by 10–12% and panx1 currents by >70%; conversely, similar concentrations of connexin-46 mimetic inhibitory peptides or a scrambled 10panx1 peptide inhibited overexpressed panx1 currents by ∼12%. We consider such results to support, rather than refute, the general selectivity of 10panx1 in blocking panx1. Moreover, we found in the present study that even very high concentrations (2 mM) of another panx1-mimetic peptide (14panx1) were without effect on the ATP-mediated release of IL-1α, IL-1β, or IL-18. Nevertheless, it was clear that long-term (>3–4 h) application of high 10panx1 (but not 14panx1) concentrations is toxic. Further investigations into mechanisms of inhibition of ATP-mediated IL-1 release by brief application of 10panx1, and particularly the generation of smaller peptide mimetics effective against panx1, will be critical to our understanding of the function of this new player in IL-1 cell biology.
Fully processed 17-kDa IL-1β is the most potent activator of IL-1Rs and hence the most potent proinflammatory IL-1 cytokine; it is ∼5-fold more potent than an often observed p20 (∼20 kDa) form, and >10-fold more potent than pro-IL-1β (34). Nevertheless, these larger forms are weakly biologically active and are readily accessible to cleavage into more potent forms by extracellular proteases (34). In particular, a recent study found that the serine proteases, proteinase 3 and neutrophil elastase, which, along with cathepsin G, are the key proteases released at sites of infection by neutrophils, cleaved pro-IL-1β into a mature, bioactive form from macrophage as well as from heterologous expression systems (44). Thus, it is of physiological significance to understand release mechanisms of unprocessed or not fully processed IL-1β and to determine whether caspase-1-independent processes are involved. TLR4 engagement by LPS results in intracellular appearance of pro-IL-1β and, more variably, a p20 form. This intermediate form can be seen in a number of studies in macrophage and/or microglia (e.g., Refs. 25 , 34 , 45 , 46) but has not been previously investigated in any detail. We found the p20 form in all RAW264.7 and J774A.1 macrophage and variably (∼40% of preparations) in peritoneal macrophage. Our results demonstrate that this intracellular p20 form results from a caspase-1-independent N-terminal processing of pro-IL-1β rather than to any possible alternative splicing. Several constitutively active intracellular proteases, such as proteinase 3, cathepsin G, chymase, collagenase, or elastase I, have been reported to process pro-IL-1β (34, 44), and are likely to be responsible. P2X7R activation (this study), as well as hypotonic challenge (47) results in release of this p20 form in peritoneal macrophage in those preparations where it is found intracellularly.
We found that RAW264.7 macrophage lack ASC and so are devoid of caspase-1-mediated processing and release of mature IL-1β. RAW264.7 macrophage have previously been observed to release pro-, but not mature, IL-1β in response to P2X7R activation (23). Similarly, LPS-primed peritoneal macrophage isolated from ASC-deleted mice release pro-IL-1β (but not the mature form) in response to ATP (16). We note a separate study on RAW264.7 macrophage where intracellular mature IL-1β has been reported (48), but it is possible that the caspase-1-independent p20 form (which we found to be prominent in RAW264.7 cells) may have been mistaken for the mature 17-kDa form. In any event, the absence of functional caspase-1 inflammasome activity in RAW264.7 macrophage has provided an important tool to address the question of P2X7R-mediated caspase-1 independent pro-IL-1β release.
P2X7R activation in RAW264.7 macrophage couples to all cellular processes associated with signaling of this receptor (i.e., cationic current, dye uptake, membrane blebbing, and cytoskeletal rearrangements) (49, 50, 51, 52, 53) with the exception of release of mature IL-1β (23) and we further confirmed in the present study its presence and function and also that of panx1. This allows us to conclude that there is no defect in P2X7R signaling per se and it is only the absence of ASC in these cells that prevents caspase-1 processing and release mechanisms. However, P2X7R activation produced robust release of pro-IL-1β from these cells, as well as release of pro-IL-1β from peritoneal macrophage. In both these macrophage types, the release was caspase-1 independent: in RAW264.7 cells because there is no functional inflammasome and in peritoneal macrophage because inhibition of caspase-1 signaling by high potassium solutions, by inhibition of panx1, or by caspase-1 inhibition, prevented release of mature, but not pro-, IL-1β. In contrast, glycine markedly reduced or abolished release of pro-IL-1β from both macrophage types without concomitant inhibition of either caspase-1 activation or release of mature IL-1β from peritoneal macrophage. This caspase-1-independent release of pro-IL-1β was also independent of extracellular calcium, or changes in intracellular calcium. Taken together, these results clearly define two distinct pathways for P2X7R-mediated IL-1 release: the well-established caspase-1-dependent cascade leading to processing and release of mature IL-1β and a novel, as yet unknown, mechanism leading to release of pro-IL-1β. This mechanism is independent of caspase-1, panx1, and calcium, and can be inhibited by glycine.
How then might glycine act to inhibit P2X7R-mediated release of pro-IL-1β? Glycine has long been known to be neuroprotective in neurons but, despite numerous studies, there are currently no general explanations for its effects to prevent neuronal cell death (54). Dubyak and colleague (23) have previously shown glycine to prevent the release of pro-, but not mature, IL-1β from macrophage, and have attributed this to an anticytolytic action of glycine, and therefore have assumed the glycine-sensitive release of pro-IL-1β is due to cell damage/lysis induced by P2X7R activation. There is no doubt that sufficient activation of P2X7R leads to inevitable cell death in macrophage via apoptosis and/or necrosis with a combination of factors determining what constitutes the “cytolytic stimulus”; these factors include concentration and duration of ATP application, type and activation state of the macrophage, external ionic environment, and tyrosine phosphorylation/phosphatase modulation of P2X7R (3, 23, 36, 55, 56). However, while glycine is quite likely to protect against P2X7R-mediated cell death, there are several reasons why little of the caspase-1-independent release of pro-IL-1β observed in the present study can be attributed to active cell death processes. Cell lysis as assayed by LDH release or trypan-blue exclusion never rose >2% in response to the relatively brief ATP application (15–20 min) we used; this cannot mathematically account for the levels of release measured in the present study. Inhibition of panx1 using 10panx1-mimetic inhibitory peptide did significantly decrease these low levels of LDH release but had no effect, or even increased, the amount of pro-IL-1β released. Overall, we can confidently conclude that we have defined a distinct pathway, independent of caspase-1 and independent of active cell death, for the release of pro-IL-1β via P2X7R activation (Fig. 8). The availability of RAW264.7 macrophage as a tool to study this release pathway in isolation from the more prominent caspase-1 cascade should provide the means to determine the signaling pathway(s) involved.
One of the most significant general outcomes of the present study is that it now provides a basis for consolidation of the various models that have been proposed for the release of IL-1β (see Introduction). Active cell death processes have long been known to lead to IL-1 release (4, 19). Noncytolytic caspase-1/inflammasome activation is now well-established as a key IL-1 release pathway (3, 5, 6, 8, 15, 16, 17, 23). We show here a distinct noncytolytic, noncaspase-1, process for the release of pro-IL-1 (Fig. 8). We also find evidence for a separate pathway leading to the release of a p20 form of IL-1β. One may expect further distinct release pathways to be delineated based on specific differences, including immune cell type and/or activation state, presence, and activity of proteolytic enzymes, and stimulus for release. It is reasonable to expect that distinct release pathways will follow equally distinct mechanisms for release. Thus, the question does not seem to be “what is the correct model of release of IL-1β,” but rather “which model corresponds to which release state?”
We thank E. Martin and I. Hui for molecular biology and tissue culture support. Prof. G. R. Dubyak (Case Western Reserve University, Cleveland, OH) supplied the mouse ASC expression vector.
The authors have no financial conflict of interest.
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.
This work was supported by The Wellcome Trust and a postdoctoral fellowship (to P.P.) from AstraZeneca Charnwood, U.K.
Abbreviations used in this paper: ASC, apoptosis-associated speck-like protein containing a C-terminal caspase-activating recruiting domain; panx1, pannexin-1; AOM, 2,6-dimethylbenzoyloxymethyl ketone; LDH, lactate dehydrogenase; NIG, nigericin; MTX, maitotoxin.