Increased Mature Interleukin-1β (IL-1β) Secretion from THP-1 Cells Induced by Nigericin Is a Result of Activation of p45 IL-1β-converting Enzyme Processing*

Perregaux and Gabel (Perregaux, D., and Gabel, C. A. (1994) J. Biol. Chem. 269, 15195–15203) reported that potassium depletion of lipopolysaccharide-stimulated mouse macrophages induced by the potassium ionophore, nigericin, leads to the rapid release of mature interleukin-1β (IL-1β). We have now shown a similar phenomenon in lipopolysaccharide-stimulated human monocytic leukemia THP-1 cells. Rapid secretion of mature, 17-kDa IL-1β occurred, in the presence of nigericin (4–16 μm). No effects on the release of tumor necrosis factor-α, IL-6, or proIL-1β were seen. Addition of the irreversible interleukin-1β-converting enzyme (ICE) inhibitor, Z-Val-Ala-Asp-dichlorobenzoate, or a radicicol analog, inhibited nigericin-induced mature IL-1β release and activation of p45 ICE precursor. The radicicol analog itself did not inhibit ICE, but markedly, and very rapidly depleted intracellular levels of 31-kDa proIL-1β. By contrast, dexamethasone, cycloheximide, and the Na+/H+ antiporter inhibitor, 5-(N-ethyl-N-isopropyl)amiloride, had no effect on nigericin-induced release of IL-1β. We have therefore shown conclusively, for the first time, that nigericin-induced release of IL-1β is dependent upon activation of p45 ICE processing. So far, the mechanism by which reduced intracellular potassium ion concentration triggers p45 ICE processing is not known, but further investigation in this area could lead to the discovery of novel molecular targets whereby control of IL-1β production might be effected.

Interleukin-1␤ (IL-1␤) 1 is produced as an inactive 31-kDa precursor protein through the enzymatic cleavage of IL-1␤converting enzyme (ICE), which cleaves the IL-1␤ precursor between Asp-116 and Ala-117 (1). ICE itself is produced as a 45-kDa precursor, which has recently been shown to be converted autocatalytically to an active p10/p20 heterodimer (2). The physiological control of ICE processing, and hence IL-1␤ conversion and secretion, is still unknown. Studies by Perregaux et al. (3,4) suggest that IL-1␤ processing is controlled by intracellular potassium ion concentration. Mouse peritoneal macrophages stimulated with LPS produce massive amounts of cell-associated, 31-kDa IL-1␤. Upon addition of the K ϩ /H ϩ ionophore, nigericin, rapid and complete processing of intracellular IL-1␤ occurred with the appearance of mature 17-kDa IL-1␤ in the medium. Similar effects were reported using human peripheral blood monocytes. Although in these studies marked leakage of the cytoplasmic enzyme, lactic acid dehydrogenase (LDH) occurred, suggesting substantial cell damage, it was argued that the effect of nigericin was not due simply to lysis, inasmuch as, unlike the effects of hypotonic shock, at no time were significant levels of proIL-1␤ detected in the culture medium. Furthermore, the nigericin-induced 17-kDa IL-1␤ was shown to have the expected N-terminal sequence. These results, together with studies by Walev et al. (5) showing that high extracellular concentrations of K ϩ or combinations of K ϩchannel blockers prevented the physiological release of IL-1␤, suggest that a net reduction of intracellular K ϩ ion concentration is necessary for the processing of proIL-1␤. Both Perregaux et al. (4) and Walev et al. (5) speculated that a reduction of K ϩ ion concentration might activate ICE or promote the processing of pro-ICE. Alternatively, it was suggested that nigericin-induced K ϩ depletion alters the cytoplasmic compartmentalization of ICE and IL-1␤. So far, however, there has not been any direct evidence that nigericin-induced release of IL-1␤ is ICE-dependent.
Cytokine Production by THP-1 Cells and Biochemical Assays-Cells from the human monocytic leukemia cell line, THP-1, were grown in RPMI medium supplemented with 110 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine, and 2 g/liter NaHCO 3 . Heat-treated fetal bovine serum (5%) was added before use. The cells were grown to a density of 5 ϫ 10 5 /ml and then stimulated with interferon-␥ (100 units/ml). Three hours later, LPS (5 g/ml) was added. This time point was designated time 0. Incubation continued for an additional 40 h. The media were then removed and clarified by centrifugation at 1000 ϫ g for 10 min. LDH measurements were performed immediately (8). Cytokine assays were performed using commercially available enzyme-linked immunosorbent assay kits (IL-1␤, Cayman, Ann Arbor, MI; proIL-1␤, Cistron, Biotechnology, Pine Brook, NJ; IL-6 and TNF-␣, Innogenetics, Zwijndrecht, Belgium). DNA was assayed fluorimetrically using the * 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.
In Vitro Refolding and Activation of ICE Processing-Refolding and the induction of autoprocessing was carried out as described before (2) with the exception that no glutathione (GSH) was present during the dialysis step. A 3-h incubation at room temperature in the presence of 25 mM GSH following dialysis led to the induction of autocatalytic processing. Radicicol analog A was added at 5 M final concentration to this last step. Cations, where mentioned, were present at the indicated concentrations in the refolding mixture, the dialysis buffer, as well as during the last incubation at room temperature. Western blot analysis was performed using anti-N-terminal p45 ICE or anti-p10 ICE subunit antibodies raised in our laboratories and shown to cross-react with p45 ICE. Detection was performed using an anti-rabbit IgG-POD (Sigma) with the chemiluminescence detection system of Boehringer Mannheim. Western blots were analyzed with a Molecular Dynamics Computing Densitometer 300A using Image Quant software.
IL-1␤ Convertase Activity Determinations-A fluorogenic Z-Val-Ala-Asp-aminomethyl coumarin (Z-VAD-AMC) substrate was used to assay activity. Free AMC, which is cleaved off directly by ICE, was detected using an excitation wavelength of 365 nm and monitoring the emission at 450 nm. Assay conditions were as described in Refs. 1 and 6.

Nigericin-induced IL-1␤ Secretion from THP-1 Cells-Pre-
liminary experiments showed that when THP-1 cells, prestimulated with 5 g/ml LPS for 39.5 h, were exposed for 30 min to nigericin, a consistent, rapid, concentration-dependent release of IL-1␤ into the medium occurred ( Fig. 2A). This increase in total cumulative IL-1␤ in the medium varied by 2-5-fold in different experiments. Measurement of IL-1␤ levels at 30, 39.5, and again at 40 h in control cultures (no nigericin), showed that secreted IL-1␤ levels were at their peak and that IL-1␤ release over this time was negligible. Nigericin thus stimulated a massive and rapid release of IL-1␤ over and above the normal steady-state levels. When the ICE inhibitor Z-VAD-DCB was added to the cultures at 39 h (30 min before nigericin), it was found to substantially block the nigericin-induced IL-1␤ release ( Fig. 2A and Table I). The effect of nigericin was not caused by cytotoxicity because, as shown in Fig. 2B, even at the highest concentration used (16 M), there was no increase in LDH leakage over the 30 min period of exposure. Because longer exposure to nigericin eventually does lead to signs of cytotoxicity, the 30-min exposure was adhered to for all experiments. The specificity of the effect on IL-1␤ is further indicated in Fig. 2B, as TNF-␣ levels were unaltered by nigericin even at the highest concentration. Additional studies (results not shown) indicated that nigericin does also not affect the amount of IL-6 secreted by THP-1 cells.
Effects of IL-1 Inhibitors on Nigericin-induced IL-1␤ Secretion-In a second series of experiments we compared the effects of Z-VAD-DCB, dexamethasone, cycloheximide and radicicol analog A, a compound previously demonstrated to reduce IL-1␤ production by causing mRNA instability (7, 10), on nigericininduced IL-1␤ release. Tables I and II show that, whereas Z-VAD-DCB was able to inhibit nigericin-induced release of IL-1␤, dexamethasone or cycloheximide were without effect. The radicicol analog also blocked the effects of nigericin. Intracellular levels of unprocessed 31-kDa IL-1␤ were measured in cell lysates. Table I shows that Z-VAD-DCB had no significant effect on intracellular levels of proIL-1␤. By contrast, both the translational inhibitor, cycloheximide (Table II), and dexamethasone (Table I)  control and nigericin-treated cells. Although no increase in LDH leakage was detected, we wished to determine whether any proIL-1␤ was released from the cells, which would indicate that the nigericin-induced release of 17-kDa IL-1␤ was simply a result of cellular membrane damage. The results also show that, in control cells, only very small amounts of proIL-1␤ are released into the medium and that none of the test compounds alone affected the amount released. Nigericin, however, more than doubled the concentration of proIL-1␤ in the medium, suggesting that a degree of non-physiological leakage of proIL-1␤ did occur. The release of proIL-1␤ was unaffected by either the ICE inhibitor, Z-VAD-DCB, cycloheximide or dexamethasone; however, radicicol analog A caused a marked reduction in release.
Activation of p45 ICE Processing by Nigericin-Because the release of mature 17-kDa IL-1␤ appeared to be dependent on ICE activity, we next investigated whether there was evidence that p45 ICE precursor was being activated by nigericin. THP-1 cells were treated with a combination of 100 units/ml interferon-␥ and 5 g/ml LPS as described under "Experimental Procedures." Z-VAD-DCB (1 M) or radicicol analog A (1 M) were given 30 min prior to nigericin (39 h after LPS addition). Following the addition of 16 M nigericin (39.5 h after LPS addition) for the final 30 min of incubation, a marked and statistically significant decrease in the amount of p45 ICE as determined by Western blotting was observed, suggesting that processing of p45 ICE had indeed been induced (Fig. 3, A and  B). This correlated well with the stimulation of IL-1␤ release by 16 M nigericin in these experiments (Fig. 3C). In the presence of the ICE inhibitor, which we had shown previously to inhibit autocatalytic processing of p45 ICE in a cell-free system (2), the effect of nigericin was reversed. The same was the case for radicicol analog A. Radicicol analog A can also affect the autocatalytic processing of p45 ICE. As Fig. 4A shows, radicicol analog A (5 M) when given at the same time as glutathione (GSH, 25 mM), which induces autocatalysis (2), prevents autoprocessing and as expected, leads to the absence of ICE activity (Fig. 4B). Unlike Z-VAD-DCB, however, radicicol analog A does not inhibit active, processed ICE in an isolated enzyme assay (results not shown).
Effects of Cations on ICE Activation and Activity-Because the main effect of nigericin is to decrease the intracellular concentration of K ϩ ions, it was possible that ICE activity or ICE processing could be directly affected by K ϩ . We therefore measured recombinant ICE activity in the presence of a range of K ϩ ion concentrations from zero to approximately 2-fold intracellular. It was found that 200 mM K ϩ causes 50% inhibition of ICE activity (results not shown). However, this was not ion-specific, and similar effects were seen with Na ϩ and Ca 2ϩ at comparable ionic strengths. We next looked at the effect of cations on ICE processing. Using recombinant p45 ICE under reducing conditions where autocatalytic processing is known to take place (see Ref. 2 and "Experimental Procedures"), the appearance of the p10 subunit and the resulting ICE activity in the presence of various cations was measured. Again an inhibitory effect was observed (Table III). This result clearly demonstrates that K ϩ ions do inhibit p45 ICE autoprocessing. In contrast to cellular systems, where monensine, A23187 (3), and amiloride (see below) have no effect, inhibition of autoprocessing in vitro is not ion-specific.
Effects of Amiloride on Nigericin-induced IL-1␤ Release-Based on findings that 5-dimethyl amiloride was able to suppress IL-1␤ secretion from LPS-activated monocytes with an IC 50 ϭ 3.5 M, it was reported that extracellular Na ϩ and high intracellular pH was required for IL-1␤ secretion (12). It was possible, therefore, that the effects of nigericin were caused by

Effects of inhibitors on IL-1␤ secretion and intracellular and extracellular proIL-1␤ accumulation
THP-1 cells were treated as described in Fig. 2  a secondary compensatory influx of Na ϩ ions. If this were indeed so, one would expect 5-dimethyl amiloride to reverse or neutralize the effects of nigericin. We thus tested IL-1␤ secretion from THP-1 cells in the presence of both nigericin and 5-dimethyl amiloride given at the same time. No reversal of the effects of nigericin by 5-dimethyl amiloride up to 30 M was observed, suggesting that nigericin exerts its effects directly through K ϩ efflux. It may well be that the efflux of K ϩ is a stronger signal for ICE activation than extracellular Na ϩ levels or changes in intracellular pH. It is also quite possible that THP-1 cells react differently to these changes, because 5-dimethyl amiloride, when given alone at 30 M before LPS stim-ulation, decreased mature IL-1␤ secretion only slightly, which is in contrast to the strong inhibition reported on monocytes (12). DISCUSSION This study expands previous findings that nigericin is able to specifically induce the release of mature IL-1␤ from mononucleic cells to THP-1 cells. The process was extremely rapid, with 2-5 times as much IL-1␤ being released in 30 min as the cumulative release of IL-1␤ over the previous 39.5 h. Unlike the studies of Perregaux et al. (3, 4), there was no evidence of cytotoxicity over the time course used, as indicated by the lack  3. Effect of nigericin on p45 ICE activation. Western blot analysis of cellular ICE. Cells were stimulated as described under "Experimental Procedures" and harvested after 40 h of incubation in the presence of LPS. Nigericin was added 39.5 h after LPS, and Z-VAD-DCB was added 30 min prior to nigericin. Secreted mature IL-1␤ was measured by enzymelinked immunosorbent assay. To control for possible uneven blotting, equal aliquots of cells were loaded randomly onto SDS gels and levels of p45 ICE were analyzed by Western blotting using rabbit anti-human N-terminal p45 ICE antibodies that cross-react with p45 ICE. Digitized images of the autoradiographs allowed relative band intensities to be measured. Given are means Ϯ S.E. (n ϭ 6 for samples with nigericin, n ϭ 12 for the control, and n ϭ 3 for all other treatments). One-way analysis of variance statistical analysis was carried out followed by Bonferroni multiple comparison test against the control (a) and the nigericintreated sample (b). of any increased LDH in the medium. On a molar basis, over 80% of the IL-1␤ released by LPS-stimulated cells is in the processed 17-kDa form. Under the influence of nigericin, the ratio of released 17-kDa to 31-kDa IL-1␤ in the medium remains the same, further substantiating the conclusion that the increase is not a result of cell lysis or simple leakage. Furthermore, the lack of effect on other cytokines such as TNF-␣ and IL-6 (latter not shown) demonstrate that the effects are specific.
When comparing intracellular levels of proIL-1␤ at 39 and 40 h after LPS stimulation no measurable increase could be observed (results not shown). Steady-state synthesis of proIL-1␤ at this time is apparently occurring at a very slow rate. Intracellular levels of proIL-1␤ were only affected by nigericin in the presence of inhibitors of de novo synthesis (dexamethasone and radicicol analog A, Table I; cycloheximide,  Table II). The most likely explanation for a lack of an effect with nigericin alone is the presence of a homeostatic mechanism designed to maintain constant levels of intracellular proIL-1␤. This would also explain why the presence of an ICE inhibitor does not increase intracellular proIL-1␤ levels. Because ICE inhibitors do not lead to a decrease in proIL-1␤ levels inside the cell, no signal is generated to induce an increase in the rate of proIL-1␤ synthesis.
The addition of an irreversible ICE inhibitor, Z-VAD-DCB, substantially blocks the nigericin-induced release of mature IL-1␤, suggesting that nigericin is dependent upon mecha-nisms that operate during the physiological release of IL-1␤. Analyses of the cellular levels of p45 ICE by Western blotting clearly show that nigericin induces the autocatalytic processing of p45 ICE (Fig. 3). The nigericin-induced processing of p45 ICE is prevented in the presence of the ICE inhibitor, which is consistent with the observations that this inhibitor prevents the autocatalysis of recombinant p45 ICE in a cell-free system (2). Furthermore, although Z-VAD-DCB also inhibits other caspases, ICE, with the exception of caspase 4 (which cleaves proIL-1␤ 250-fold less effectively), is the only caspase known to correctly cleave proIL-1␤ to its mature form (13). Also, unlike other caspases, again with the exception of caspase 4, no enzyme has so far been described to process p45 ICE other than ICE itself whereas in vitro, ICE can also cleave pro-caspase 4. This, together with the inhibition of p45 ICE autoprocessing, but not p10/p20 ICE activity by radicicol analog A (see below), further reduces the likelihood of a nigericin-induced activation of an enzymatic cascade upstream of ICE.
Not surprisingly, 1 nM dexamethasone (a concentration shown to give Ͼ80% inhibition of IL-1␤ secretion if added before LPS stimulation) and cycloheximide do not prevent nigericin-induced IL-1␤ processing indicating that nigericin induced secretion of IL-1␤ comes from a pre-existing pool of proIL-1␤.
Radicicol analog A had a profound effect on IL-1␤ levels. A 1-h exposure led to a dramatic reduction in intracellular proIL-1␤. Previous studies (10) have shown that radicicol analog A induces the rapid degradation of cytokine mRNAs (including IL-1␤), which have in common the AUUUA instability motif in the 3Ј-untranslated region. In contrast to dexamethasone and cycloheximide, the effect of radicicol analog A on released proIL-1␤ from nigericin-treated cells therefore reflects a drastically reduced pool of proIL-1␤ in the cells that is available for processing (Tables I and II). Because over the time period measured in our experiments there is no detectable increase in IL-1␤ secretion, or proIL-1␤ leakage in control cells, it is not surprising that no effect is seen with radicicol analog alone on the secretion of mature IL-1␤ or on extracellular levels of proIL-1␤. The effect of radicicol analog A on the nigericininduced release of mature IL-1␤ is twofold. First, as mentioned  a Activities were assayed at dilutions resulting in ion concentrations which had no effect on ICE activity as determined in separate experiments (results not shown). before, radicicol analog A induces rapid degradation of IL-1␤ mRNA and possibly also inhibits transcription and translation. Thus, in the presence of radicicol analog A, synthesis of proIL-1␤ is decreased. Because the rate of proIL-1␤ production is decreased and the rate of proIL-1␤ consumption either by export of processed IL-1␤ or intracellular degradation of proIL-1␤, is unaffected by radicicol analog A, intracellular proIL-1␤ levels are expected to decrease whereas the levels of mature IL-1␤ secreted remain at the control levels because radicicol analog A does not inhibit pre-existing mature ICE. Therefore, also with the addition of nigericin, extracellular proIL-1␤ levels do not increase. Second, because radicicol analog A also blocks p45 ICE processing (Fig. 4) but does not inhibit ICE activity (results not shown), the increase in IL-1␤ secretion in the presence of nigericin is a result of the presence of more active ICE, a result of increased p45 ICE processing triggered by the lowering of K ϩ levels by nigericin. Radicicol analog A blocking p45 ICE processing inhibits the effects of nigericin, whereas cycloheximide, which also inhibits protein synthesis but does not inhibit ICE activity or ICE processing (results not shown), does not influence the effects of nigericin.
So far, however, there are no clues as to how a reduction in K ϩ ion concentration may induce p45 ICE processing. Autoprocessing of p45 ICE, in vitro is not sensitive to K ϩ alone (Table  III). Ionic strength-dependent inhibition of ICE activity in vitro was also seen (results not shown). Mature ICE was sensitive to salt concentrations equivalent to the intracellular concentration of potassium. Such concentrations caused approximately 50% inhibition of ICE activity using the synthetic substrate Z-VAD-AMC. However, this effect was not ion-specific, so it is unlikely that this explains the specific effects of reducing K ϩ ion concentration in whole cells. We have also eliminated the possibility that nigericin has a direct effect on ICE activity in a cell-free system (results not shown). Although it remains unclear how nigericin induces ICE processing, this effect does not appear to require metabolically active cells because nigericininduced proIL-1␤ processing continued in azide-treated cells (results not shown).
Whether K ϩ ion flux plays a role in the physiological processing of ICE and IL-1␤ secretion in response to pro-inflamma-tory stimuli, is not clear. Walev et al. (5) showed that a variety of manipulations that resulted in reduced intracellular levels of K ϩ could trigger IL-1␤ processing. Furthermore, high extracellular concentrations of K ϩ could reverse these effects. Combinations of tetraethylammonium and 4-aminopyridine (potassium channel blockers) could also inhibit the physiological release of LPS-induced IL-1␤. Because no single channel blocker was effective, even at high concentration, this suggests that multiple channels are involved and therefore potassium channel blockers are unlikely to be leads in the search for cytokine release inhibitors.
Taken together, our results show that nigericin-induced K ϩ efflux induces rapid p45 ICE processing leading to an increase in active ICE, which in turn results in a higher secretion of mature IL-1␤. Because nigericin treatment specifically leads to mature IL-1␤ release, a better understanding of the mechanism by which K ϩ ions control p45 ICE activation and proIL-1␤ processing might lead to the identification of new anti-inflammatory drug targets.