Activation of the Native 45-kDa Precursor Form of Interleukin-1-converting Enzyme*

Active interleukin-1 (cid:98) -converting enzyme (ICE) is composed of 20- and 10-kDa polypeptides (p20 and p10) derived from the processing of a cytosolic 45-kDa precursor protein (p45). The cleavage and activation of the native p45 ICE precursor have been characterized by use of specific inhibitors and antibodies recognizing various regions of ICE. The processing of p45 in vitro in THP.1 monocytic cell cytoplasmic extracts is inhibited only by protease inhibitors that inhibit ICE and not by inhibitors of other protease classes. The addition of L-742,395, a biotinylated irreversible ICE inhibitor, to these extracts labels only p45 and simultaneously inhib- its p45 processing, demonstrating that the p45 has catalytic activity. Following a cleavage of p45 at a site that becomes the COOH terminus of p20, a more active intermediate is formed which migrates on SDS-polyacryl- amide gel electrophoresis with an molecular mass of 35 kDa (ED 50 of (cid:59) 0.1 (cid:109) M L-742,395 labeling versus 5 (cid:109) M for p45). This new more active ICE form serves both as an intermediate enzyme to cleave p45 as well as a substrate for the formation of the final active ICE (ED 50 of 1 n M L-742,395 labeling of p20 and for p22, an NH 2 -terminally extended form of p20). While initial cleavage of p45 can be found at the sites corresponding to both the NH 2 termini of p22 and p20, these fragments cannot be la- beled by L-742,395 and are hence inactive. p45 is not processed at the site corresponding to the NH 2 terminus of the p10. Less than 50% of the p45 is cleaved down to active p20 or p22 ICE as determined by band shift on SDS-polyacrylamide gel electrophoresis of the biotinylated fragments, indicating that the in vitro activation is highly inefficient. The ICE fragmentation occurs by an intermolecular process and is highly dilution sensitive. Cleavage of p45 by exogenous p20/p10 ICE differs from that of the endogenous p45 cleavage activity in that the p20/p10 activity is more salt sensitive, and it produces a different pattern of cleavage fragments, principally 35- and 12-kDa fragments. These results indicate that the nature of the ICE activity changes as p45 is processed down to the p20/p10 form of the enzyme. luminol biotinylated peroxidase-conjugated strepavidin generation of ICE peptide conjugates for immunization; Nancy Thornberry for the recombinant ICE; Ro-lando for the [ 35 S]Met pIL-1 (cid:98) and p45 and recombinant ICE fragment standards; Erin Peterson for measurement of YVAD-AMC activity; and and Livia Rosen for their critical comments on this manuscript.

IL-1 1 is a central mediator of a number of acute and chronic inflammatory diseases such as arthritis because of its stimulation of leukocyte adhesion, activation, and destruction of soft tissues and bone (1) IL-1␤, rather than its nonsecreted analog IL-1␣, is thought to be the major form of IL-1 that is involved in these effects. Synthesized as an inactive 31-kDa precursor (pIL-1␤), the active 17.5-kDa mature form (mIL-1␤) is formed by a proteolytic cleavage at Asp 116 -Ala 117 (2)(3)(4). The enzyme responsible for this cleavage, termed IL-1␤-converting enzyme (ICE), is a cytoplasmic cysteine protease found primarily in monocytic cells (5,6). When this enzyme is inhibited, IL-1␤ is released from cells as unprocessed, inactive pIL-1␤ in vitro and in vivo (7)(8)(9). In transgenic mice lacking functional ICE, only trace amounts of mIL-1␤ are released into plasma, there is a large reduction in IL-1␣ production, and the animals show increased resistance to lipopolysaccharide-mediated death (10,11). Inhibition of ICE is thus a potential therapeutic target for the treatment of inflammatory arthritis and other IL-1-mediated diseases.
ICE is the first protease member of a new family of cytoplasmic cysteine proteases, which include the Caenorhabditis proapoptotic protease Ced-3 (12), the developmentally expressed murine gene Nedd-2 (13) and its human homolog ICH-1 (14), the ICE homologs ICE rel II and ICE rel III (15), the proapoptotic homologs apopain/CPP32 (16,17), Mch2 (17), and Mch3 (18) (also known as ICE-LAP3 (19)). All of these homologs share the active site cysteine and aspartate binding residues that are essential for ICE activity (7,20,21). Active ICE consists of two polypeptides of 20 and 10 kDa (p20 and p10) associated in a 1:1 ratio, which are processed from an inactive cytoplasmic 45-kDa precursor (p45, see Fig. 1) (7,22). Likewise, the ICE homologs contain two comparable subunits derived from a common precursor (16,18,19). The activation of the p45 ICE precursor protein, which results in the loss of the precursor domain, may be important not only for ICE, but also for the other homologs. ICE rel II and ICE rel III, for example, show no proapoptotic activity until the prodomain is removed (15). Hence an understanding of the activation of ICE may be relevant for the activation of other homologs.
In the present report we trace the activation of p45 ICE through a sequential series of ICE-dependent proteolytic activation steps to generate fragments of increased catalytic activity ultimately producing p20/p10 ICE. We use a biotinylated irreversible ICE inhibitor, L-742,395, and analogs to characterize the relative activity of the ICE fragments at each cleavage step. The relative concentration of L-742,395 required to label these ICE fragments is a function of the activity of that fragment, the binding, turnover, and simultaneous covalent labeling of the ICE forms with biotin. Using this technique, we show that the p45 found in monocytic cells is weakly active on ICE peptide substrates, requiring micromolar concentrations of inhibitor for labeling and inhibition of processing. Following a cleavage at the COOH terminus of p20 (Asp 297 -Ser, Fig. 1), the resultant 35-kDa/p12 fragment is 100-fold more active than p45 and is responsible for cleavage of ICE down to the p20/p10 form. The p20/p10 ICE form has the highest affinity to L-742,395, showing the same Ͻ1 nM ED 50 for covalent labeling as was shown previously for inhibition of substrate cleavage (23). The ICE activity that cleaves p45 down to the highly active p20/p10 form is catalytically different from p20/p10: it does not cleave pIL-1␤, the activity is less sensitive to salt inhibition and is more sensitive to dilution, and its inhibition requires higher concentrations of ICE inhibitors than does p20/p10 ICE.

MATERIALS AND METHODS
Preparation of THP.1 Cell Extracts--THP.1 cells were grown in suspension in roller bottles in Iscove's modified Dulbecco's medium supplemented with 10% fetal bovine serum (BioWhittaker, Inc., Walkersville, MD) to a density of ϳ1.5 ϫ 10 6 cells/ml, subcultured weekly. Cytosolic extracts were prepared as described previously (24). Briefly, the cells were washed in phosphate-buffered saline, swollen in hypotonic buffer (containing 25 mM HEPES, pH 7.5, 5 mM MgCl 2 , and 1 mM EGTA), and broken with a Dounce homogenizer in the presence of phenylmethylsulfonyl fluoride, pepstatin, and leupeptin. Nuclei were removed, 6 mM EDTA was added to the postnuclear supernatant, and granules and microsomal membranes were subsequently removed by sequential centrifugation. The supernatants were stored at Ϫ80°C following addition of 2 mM dithiothreitol. DEAE-HPLC fractionation of cytosolic extracts were performed as described previously (24).
Measurement of ICE Activity--ICE activity was measured by either the cleavage of the protein substrates [ 35 S]Met pIL-1␤ or p45 ICE or the fluorescent peptide substrate AcYVAD-AMC (7-amino-4-methylcoumarin) as described previously (24). Protein substrate assays were typically done in a 20-l volume using 5-10 l of cytosolic extract and 0.5 l of [ 35 S]Met substrate in a buffer of 25 mM HEPES, pH 7.5, 10% sucrose, 0.1% CHAPS, and 2 mM dithiothreitol (HSCD buffer). Incubations were performed at 30°C for 60 min. The products of the [ 35 S]Met reaction were separated by SDS-PAGE on 16% gels (Novex, San Diego, CA), and the gels were transferred to polyvinylidene difluoride membranes (Immobilon, Millipore, Bedford, MA), dried, and exposed to X-Omat film (Eastman Kodak Co.) for 1-3 days. Occasionally the gels were soaked for 30 min in Amplify fluorographic enhancer (Amersham Corp.), and dried directly for autoradiography.
Biotinylation of ICE by L-742,395--The irreversible ICE inhibitor L-742,395 (Ac-Tyr-Val-Lys(biotin)-Asp-(acyloxy)-methyl-ketone), its nonbiotinylated analog L-702,066, and the reversible inhibitor L-709,049 (Ac-Tyr-Val-Ala-Asp-CHO) were prepared at Merck. Biotinylation was performed by the addition of dilutions of L-742,395 to ICE-containing samples for at least 10 min at 25°C (conditions commensurate the inactivation of Ͼ99.99% of active ICE by this irreversible inhibitor, see Ref. 23).
Immunoprecipitation, Electrophoresis, and Immunoblots--Immunoprecipitation of ICE-containing extracts was performed as described previously (22). Briefly, for immunoprecipitation of "native" samples, 1 volume of 2 ϫ RIPA buffer (1 ϫ includes 1% Triton X-100, 0.5% deoxycholate, 0.2% SDS, 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA) was added to the sample followed by 1 l of antiserum/500 l. For immunoprecipitation of denatured samples, the sample was heated with 1% SDS (diluted from a 20 ϫ stock) for 3 min, 100°C, cooled, and then mixed with 4 volumes of SDS-free RIPA prior to the addition of the indicated antiserum. Protease inhibitors as well as 10 mM iodoacetic acid were added to all samples prior to incubation with the antisera. Following incubation with the antibody at 5°C overnight, the complexed antibodies were collected with washed protein A-agarose that had been preblocked for 60 min with Superblock (Pierce). After heating with 2 ϫ SDS-PAGE sample buffer, aliquots were electrophoresed on 16% gels as described above and transferred to polyvinylidene difluoride membranes.
For immunoblotting, the membranes were dried and blocked with Superblock for 60 min at room temperature. For ICE protein detection, the blots were incubated for 60 min with a 1:5000 dilution of the primary antibody, followed by 60 min with a 1:5000 dilution of either horseradish peroxidase-conjugated donkey anti-rabbit antibody or horseradish peroxidase-conjugated protein A (both from Amersham). Bands were detected by addition of luminol reagents (ECL, Amersham) and exposure of Hyperfilm-ECL (Amersham) for 10 s to 5 min. For detection of biotinylated ICE proteins, the blocked membranes were incubated for 25 min with a 1:1000 dilution of avidin followed by 25 min with a 1:500 dilution of biotinylated anti-avidin (both from Vector Laboratories, Burlingame, CA) and a 1:10,000 dilution of horseradish peroxidase-conjugated strepavidin (Amersham).
Densitometry--To quantitate the intensity of the bands produced by luminometry or autoradiography, the films were scanned with a Quantity One densitometry system (PDI, Huntington Station, NY).

RESULTS
p45 ICE Is Processed to p22 and p20 via a 35-kDa Intermediate during Incubation in Vitro-Previously we have shown that p45 is the major form of ICE seen in monocytic cell extracts (22). The p45 band could be resolved into a mixture of two closely migrating bands with a densitometric ratio of 3:1 of the slower to more rapidly eluting band (denoted by heavy and light lines, respectively, Fig. 2A) of which the top band comigrated with p45 expressed in transfected COS cells. To observe the processing of p45, THP.1 cytoplasmic extracts were incubated at 30°C for several hours, and the ICE fragments generated were characterized by immunoblot. After a 0.5-h lag, the p45 bands were rapidly cleaved to three groups of bands: (i) a series of closely spaced bands with a major band migrating at 35 kDa, (ii) p22 and p20, and (iii) p10 ( Fig. 2A). During in vitro cleavage, the 35-kDa proteins were formed first, with maximum production during the most rapid phase of the p45 proteolysis and then a gradual disappearance as the p45 was totally broken down (Fig. 2B). The p20 and p22 were formed later in roughly equal amounts, and they were stable for several hours. This suggests that the 35-kDa bands are intermediates in the formation of p22 and p20. p10 was formed with the same time scale as p20 and p22, but its intensity relative to p20 was very low with a relative absorbance of less than 10% that of p20. In contrast, active ICE purified by a peptide inhibitor affinity column contained p20 and p10 with a relative absorbance of 1:0.3 (p20:p10; Fig. 2A). This suggests that a preferential degradation and loss of p10 versus p20 and p22 occurred during the incubation (Fig. 2, A and B).
The amount of pIL-1␤ cleavage activity observed in the incubated cytoplasmic extracts increased in parallel with the breakdown of the p45, which itself showed no pIL-1␤ cleavage activity. ICE activity reached a maximum after about 60 min of incubation ( Fig. 2C), at a time when the rate of generation of the 35-kDa intermediates and p22/p20/p10 was maximal. The pIL-1␤ cleavage activity decreased at later times despite the lack of obvious changes in the p22/p20/p10 ICE present (Fig. 2,  A and B). This loss of activity reflects the instability of active ICE, which had previously been observed in cytoplasmic extracts (24).
p45 Cleavage Is Inhibited by ICE Inhibitors-p45 cleavage was prevented by incubation with 5 mM iodoacetic acid, 1 mM N-ethylmaleimide, or specific peptide ICE inhibitors such as either the aldehyde L-709,049 (Ac-Tyr-Val-Ala-Asp-CHO) or the corresponding peptide diazomethylketone at 10 M (data not shown). In contrast, addition of such protease inhibitors as leupeptin, pepstatin, elastatinal, 3,4-dichloroisocoumarin, E64, tetrathionate, benzamidine, chymostatin, aprotinin, soybean trypsin inhibitor, ␣ 1 -protease inhibitor, orthophenanthroline, or EDTA had no effect on p45 cleavage (data not shown). This profile is consistent with cleavage by ICE or a closely related cysteine protease. The cleavage of p45 was unlikely due to the action of other ICE homologs, however. Comparison of the inhibition of p45 activation by an inhibitor of the ICE homolog apopain, Ac-Asp-Glu-Val-Asp-CHO (see Refs. 16 and 18), revealed that it required more than an order of magnitude higher concentration than L-709,049 for comparable inhibition. This relative difference in potency closely mirrored the corresponding inhibition of p20/p10 ICE by Ac-Asp-Glu-Val-Asp-CHO relative to L-709,049.  (24), and the COS-7 extracts were prepared from cells transfected with vector in the absence (Ϫ) or presence of the human ICE p45 cDNA (ϩ, see Ref. 7). Heavy solid lines denote the major ICE bands found (p45, 35 kDa, p22, and p20). Thin solid lines denote the lesser ICE bands observed, including p43 (see Fig.  1), the minor 35-kDa fragment generated during activation, and p10. B, densitometric scans of the various major ICE fragments (see part A) as visualized in immunoblots obtained as in A: q, p45; f, 35 kDa; ç, p20; å, p22; and छ, p10. C, pIL-1␤ cleavage activity of cytoplasmic extract incubated for the indicated number of minutes (from A). Each of the resultant samples (5 l) was incubated for 60 min at 30°C in a 20-l assay using [ 35 S]Met pIL-1␤ as a substrate. No additional processing of ICE occurred during the pIL-1␤ cleavage assay (see Ref. 22). An autoradiograph of the mIL-1␤ formed (17 kDa) is shown in the inset above, while a densitometric scan of the amount of the resultant mIL-1␤ is graphed below.
The detailed inhibitory effects on p45 cleavage were determined with the addition of the ICE inhibitor L-742,395, Ac-Tyr-Val-Lys(biotin)-Asp-(acyloxy)-methyl-ketone. This irreversible ICE inhibitor both inhibited p20/p10 ICE activity and irreversibly biotinylated the p20 active site Cys with an ED 50 of Ͻ1 nM (23). Addition of L-742,395 to THP.1 cytosol inhibited the proteolysis of p45 in a complex, dose-dependent fashion. Addition of 2-1000 nM L-742,395 gave a dose-dependent inhibition of the majority of p45 cleavage (Fig. 3A). There was an additional background component of the p45 cleavage that 1000 nM L-742,395 did not inhibit (Fig. 3A). This cleavage could, however, be inhibited for several hours by addition of high (Ͼ10 M) L-742,395 concentrations (Fig. 3A). Thus, it appeared that there were two activities that could cleave p45: the first was sensitive to ϳ10 -100 nM L-742,395, whereas the second required micromolar concentrations of L-742,395 for inhibition. The first of these activities was principally responsible for the formation of the p22, p20, and p10: 200 nM L-742,395 inhibited the formation of all three proteins (Fig. 3B).
When these L-742,395-inhibited extracts were tested for pIL-1␤ cleavage activity, the concentration of L-742,395 required to inhibit pIL-1␤ processing was 1 nM (Fig. 3C), indicat-ing that neither of the activities cleaving p45 cleaved pIL-1␤. By 60 min at a concentration of 1 nM, there was less than 10% inhibition of p45 breakdown and formation of p20/p22 (Fig. 3C). Hence pIL-1␤ cleavage is much more sensitive to L-742,395 than p45 processing. The observation that at 1 nM L-742,395 only the p20 and p22 polypeptides are labeled provides further evidence that only the p20/p10 or p22/p10 forms have the ability to cleave pIL-1␤ (see below).
Identification of the Active ICE Forms by Biotinylation and SDS-PAGE--Since the L-742,395 is an irreversible inhibitor, newly generated active forms of ICE are immediately inhibited and accumulate during incubation with L-742,395. The biotinylated forms can then be detected by immunoblots using a biotin detection system and compared with the amount of each protein as detected by immunoblots with an anti-ICE antibody. When this analysis was applied to the incubates described in Fig. 3, A and B, it was apparent that at very low (2 nM) L-742,395 concentrations only two bands were specifically biotinylated, corresponding to biotinylated p20 and p22 (Fig. 3D). The amounts of the biotinylated p20 and p22 forms as determined by both biotin labeling (Fig. 3D, right) and protein labeling (Fig. 3D, left) were highest in the 7-200 nM range of L-742,395, corresponding to those L-742,395 concentrations that permitted maximal p22 and p20 formation (Fig. 3B). At the highest L-742,395 concentrations where little p22 and p20 was formed (Fig. 3D, left), two higher molecular mass 35-kDa bands accumulated (Fig. 3D, left, arrowheads) that were biotinylated (Fig. 3D, right). These 35-kDa proteins represented biotinylated versions of the two unlabeled 35-kDa bands observed in the absence of inhibitor (Fig. 3D, left, heavy and light solid lines; cf. Fig. 2A). The appearance of these biotinylated 35-kDa intermediates occurred in an L-742,395 dose-dependent fashion inversely with the disappearance of the unbiotinylated forms (Fig. 3D). Most importantly, at the highest concentrations, 1000 and 200 nM, the 35-kDa biotinylated forms accumulated steadily over the several hour incubation of the experiment (not shown). These were the only biotinylated forms that accumulated with the 200-1000 nM L-742,395 concentrations under which the formation of p20 and p22 (Fig. 3B) was inhibited. This suggests that these forms comprise the enzymatic activity that was responsible for the generation of the smaller p22 and p20 forms.
At the 1000 nM concentration of L-742,395 in Fig. 3D, the p45 bands (heavy and light lines, respectively) were also slightly biotinylated (solid and open arrowheads, respectively). Incubation of cytosolic samples with 10 M L-742,395 yielded more intense staining of these two bands indicating that p45 was the major inhibited protein (data not shown). When a mixture of all the ICE forms (p45, 35 kDa, p22/p20) generated by a 60-min preincubation of cytosol was subsequently incubated with serial dilutions of L-742,395, the relative reactivity of each form could be determined (as measured by the effective dose, ED 50 , of L-742,395 labeling). The p45 forms were biotinylated in a dose-dependent fashion with an ED 50 of 5-10 M, while both 35-kDa bands were biotinylated with an increased sensitivity to L-742,395, labeling with an ED 50 of 60 nM (Fig. 4, A and C). The p20 and p22 bands showed also equal sensitivity to L-742,395, but with an ED 50 of ϳ1 nM (Fig. 4, A and C). Affinity-purified ICE containing only p20 (along with p10) also labeled with nM effectiveness (Fig. 4, B and C). Thus, the three different forms of ICE have a different affinity for the L-742,395 inhibitor.
The specificity of biotinylation of the different forms of ICE by L-742,395 was demonstrated in competition experiments with other nonbiotinylated ICE inhibitors that prevent labeling in a dose-dependent fashion. As is shown in Fig. 5, an irreversible nonbiotinylated analog of L-742,395, L-702,066 (Ac-Tyr-Val-Lys-Asp-(acyloxy)-methyl-ketone), as well as the reversible ICE inhibitor L-709,049 (Ac-Tyr-Val-Ala-Asp-aldehyde) inhibit ICE biotinylation by L-742,395. Both the 0.5 nM L-742,395 labeling of p20 and p22 (Fig. 5, A and B) and the 10 M labeling of p45, 35-kDa, and p22/p20 proteins could be inhibited (Fig.  5C). As expected, the reversible inhibitor L-709,049 was less able to inhibit the biotinylation than was the irreversible inhibitor L-702,066 (Fig. 5B). Labeling of all forms of ICE by L-742,395 was also prevented by pretreatment with 10 mM iodoacetic acid (data not shown).
Cleavage of p45 Occurs following Asp 297 to Generate an Active Intermediate--To determine the nature of the p45 fragments formed during activation, immunoprecipitation was performed with ICE antibodies specific for various regions of the protein. Incubation of p45 in the presence of 10 nM L-742,395 had little effect on the p45 cleavage and p20/p22 formation but trapped the active p20/p22 as it formed. Addition of 10 M L-742,395 midway through the incubation biotinylated all the active ICE forms and prevented further cleavage, trapping the ICE at a partially processed state. Because the active (i.e. biotinylated) forms were shifted to more slowly eluting forms on SDS-PAGE, it could be determined which cleavages led to active and which cleavages led to inactive subunits. Immunoprecipitation by antibodies to different regions of ICE was done with both native samples as well as those that had been first denatured with SDS to identify which fragments coimmunoprecipitated as part of an active complex. The antibodies used include G273 and R105, raised against the active p20/p10 complex; R1711, raised against the precursor domain; JD5, raised against the p10 protein; and R3237, raised against the peptide linking p20 and p10 in p45 (see Fig. 1). The RIPA buffer itself used for the immunoprecipitation was not denaturing, because normal processing could occur in the presence of RIPA (data not shown).
If the ICE were first denatured, however, no p10 could be immunoprecipitated as determined by antibodies recognizing p10: the R105 antibody (lanes 4 -6) and the anti-p10 antibody JD5 (lanes 22-24). This indicates that G273 recognized only ICE fragments containing p20, a result not surprising consid- ering that in the native tetrameric enzyme p10 is tightly associated with, and is essentially hidden by, the p20 polypeptides (see Refs. 20 and 21). The p20 specificity of G273 was confirmed using a p20-specific antibody (GP1523, see Fig. 1) that identified the same ICE fragments as those found in lanes 4 -6 (data not shown). In native ICE G273 also immunoprecipitated p12 (Ser 298 -COOH terminus (site 3-C), see Fig. 1), identified by the R3237 antibody against the p20-p10 linker peptide (lanes [13][14][15]. The amount of p12 present was small compared with the amount of p10 (as shown by the p10-specific antibody JD5, lanes 19 -21), but p12 is ordinarily not seen at all in the purified active enzyme.
The continuous presence of 10 nM L-742,395 resulted in the biotinylation of less than 50% of the newly generated p20 and p22 protein, because less than 50% of the p20 and p22 was shifted to the slower eluting forms (lane 2 versus lane 1, Fig. 6) that were biotinylated (lane 8 versus 7). This suggests that less than 50% of the cleavages of p45 was productive cleavages. An alternative interpretation is that less than 50% of the p45 initially was active, and the potentially inactive p45 was cleaved along with the active p45.
Immunoprecipitation by the antibody to the precursor region (R1711, Fig. 6) indicated that only two 35-kDa bands were found (lane 25) in contrast to the several bands immunoprecipitated by the G273 (lane 1): a slower migrating heavy band that comigrated with a recombinant N-3 standard and a more quickly eluting lightly stained band. Both of the R1711-immunoprecipitated 35-kDa bands were biotinylated at high L-742,395 (lane 33) and shifted to slower eluting forms on the gels (lane 27). The amount of the biotinylated 35-kDa proteins immunoprecipitated was the same for the R1711 as for the G273 immunoprecipitation (lanes 33 and 36 versus lanes 9 and  12). This suggests that all of the biotinylated 35-kDa ICE forms contained the precursor domain. This was substantiated by the lack of a biotinylated band immunoprecipitated with JD5, the p10-specfic antibody (data not shown). Furthermore, all of the biotinylated 35-kDa forms were cleaved at site 3 (Asp 297 Ser), because none contained the p20-p10 linker peptide (see lanes 37-39, cf. lanes [13][14][15]. The R1711 immunoprecipitates also contained a small amount of p12 (lanes 37-39) under native, but not denaturing, conditions (lanes 40 -42), indicating that immediately following cleavage, this fragment continues to associate with the precursor domain. It is of interest to note that p10 itself was not associated with these immunoprecipitates (lanes 43-45), indicating that the fully mature p20/p10 ICE is generally not associated with the precursor domain (see lanes [25][26][27]. (Because the JD5 antibody is not as sensitive as is the linker antibody R3237, the p12 cannot be seen in lanes 43-45). The precursor domain (NH 2 terminus-Asp 103 ) was also immunoprecipitated in the cleaved mixture of fragments as seen by R1711 blot of the R1711 immunoprecipitate (data not shown). The precursor domain was not, however, immunoprecipitated with the G273 antibody to active p20/p10 ICE (data not shown). These observations are consistent with earlier data that affinity-purified p20/p10 from THP.1 cells did not contain the precursor fragment (data not shown).
There were other 35-kDa forms of ICE immunoprecipitated by G273 that could be labeled with the R3237 linker and JD5 p10 antibodies. As seen in lanes 13-18 and 19 -24, these bands include a closely spaced doublet that migrates just faster than the N-4 marker (see solid arrowhead opposite lane 24) as well as a band migrating more rapidly than the biotinylated bands (open arrowhead). These bands were not immunoprecipitated by the R1711 prodomain antibody (lanes 37-42), but they were immunoprecipitated by the JD5 p10 antibody (data not shown). Since recombinant N-4 standard (which would generate the largest of the four potential 35-kDa fragments; see Fig. 1) migrates more slowly than all of the observed fragments, these new bands, which are precursor (R1711)-negative and p10/p12 (JD5 and R3237)-positive, may represent cleavages at sites following the Asp 103 and Asp 119 (sites 1 and 2, see Fig. 1). These bands were not biotinylated and hence inactive, as measured by the absence of a band shift at the 10 M L-742,395 concentration (lanes 15 versus 13; compare, e.g. with lanes 27 versus 25). The lack of the N-4 fragment in any of the immunoprecipitates suggests that there is normally no cleavage that occurs initially at the Asp 316 site.
Differences in the Activity of p45 Cleavage from That of p20/p10 ICE--To distinguish between the p45 cleavage activity normally observed in the cytoplasmic extracts from that of p20/p10 ICE itself, exogenously labeled p45 was used as a substrate. The addition of THP.1 cytosol to [ 35 S]Met in vitro translated p45 produced a breakdown of the labeled p45 to a 35-kDa band as well as to p22, p20, p12, and p10, the same pattern of intermediates as that found with endogenous p45 (Fig. 7A, lane 3, cf. Fig. 2A). The [ 35 S]Met-p45 itself was stable to a 2-h incubation, indicating that it could not cleave itself (Fig. 7A, lane 2). If the incubation were performed in the presence of 10 M L-702,066, the nonbiotinylated irreversible ICE inhibitor, no cleavage occurred (lane 4). A dose response of the inhibitor indicated that formation of labeled p22/p20 products occurred only when the L-702,066 concentration was below about 100 nM (Fig. 7A, lanes 5-8), results comparable for the inhibition of endogenous active p45 (see Fig. 3). Similar results were found using as a substrate inactive, native p45 that had been biotinylated with L-742,395 and then affinitypurified: when mixed with active cytosolic extracts, this labeled p45 was cleaved down to p22 and p20 fragments through a 35-kDa intermediate unless concentrations of Ն100 nM L-702,066 were first added (data not shown). In both cases of exogenously added p45, no inhibition of cleavage of the p45 substrate was observed at 10 nM L-702,066 (cf. Fig. 3A, lane 7). This indicates that p45 cleavage occurred at a concentration at which p20/p10 ICE activity was totally inhibited.
Incubation of the labeled p45 substrate with active p20/p10 directly, on the other hand, generated a different pattern of ICE cleavage products. Instead of the cleavage of the p45 substrate to a 35-kDa intermediate followed by cleavage to FIG. 6. Identification of ICE fragments and fragment associations following immunoprecipitation and immunoblot with different ICE antibodies. THP.1 cytosolic extract was incubated 30 min, 30°C, in the absence or presence of 10 nM L-742,395. One portion of the inhibited sample was then treated with 10 M L-742,395, and all three samples were incubated a second 30 min. Half of each sample was heated with 1% SDS for 3 min at 100°C and then diluted with 4 volumes into SDS-free RIPA buffer (these are the "Denatured" samples). The remaining half of the three samples were correspondingly diluted into normal RIPA buffer (these are the "Native" samples). Aliquots of the native and denatured samples of each incubation were immunoprecipitated by G273, R1711, or GP1523 (see Fig. 1). After SDS-PAGE, the resultant immunoprecipitates were blotted with R105 (lanes 1-6, 25-30, 55-60) or with strepavidin for biotin (lanes 7-12, 31-36, 61-66), R3237 (lanes 13-18, 37-42, 67-72 Aliquots of each were separated by SDS-PAGE, transferred onto PVDF, dried, and exposed to X-OMAT film overnight. The identical blot was then blocked and immunoblotted with R105 (bottom). The solid arrows on the right refer to the positions of the p45, 35-kDa intermediate (35K), p22, p20, p12, and p10 fragments. Since the L-702,066 is an irreversible ICE inhibitor, it causes the active p20 and p22 to shift to more slowly migrating bands denoted by the dashed lines on the left. B, comparison of the cleavage of [ 35 S]Met p45 by cytosol (left) with purified ICE (right). In different experiments [ 35 S]Met p45 was incubated in the presence or absence of THP.1 cytosol extract or with different amounts of purified ICE for 60 min, 30°C. Aliquots were separated by SDS-PAGE, the gels were dried, and exposed for autoradiography for 3 days. Locations of p45, 35 kDa (35K), p12, and p10 are indicated by arrows. The arrowhead refers to a band migrating at ϳ24 kDa that is formed by exogenous ICE. The p22 and p20 formed by the cytosolic extracts (see A) are not labeled. p22/p20 (Fig. 7B, lane 2), there was only cleavage to a 35-kDa intermediate and the p12 (Fig. 7B, lanes 4 -7). At high concentrations of ICE sufficient to cleave all of the labeled p45, only small amounts of p10 and ϳ24 kDa were also formed (Fig. 7B,  lane 7); there was no production of the p20 and p22 as was normally observed during p45 cleavage in cytosolic extracts. Furthermore, the p45 was a relatively poor substrate for p20/ p10 cleavage requiring concentrations of p20/p10 ICE of 2 orders of magnitude greater than those needed for cleavage of pIL-1␤. Complete cleavage of pIL-1␤ can be achieved by as little as 0.5 unit of p20/p10 ICE under these conditions (see Ref. 22).
Effects of Dilution on p45 Cleavage--Since p45 was cleaved in cytosol by an ICE form other than p20/p10, the cleavage could occur by intramolecular cleavage of p45 itself, or by intermolecular cleavage by two or more p45 molecules or fragments thereof. If cleavage were strictly intramolecular, then cleavage should be insensitive to dilution. But when the usual cellular cytoplasmic extract prepared at 10 8 cell equivalents/ml (5 mg of protein/ml; 0.5 g of p45/ml, see Ref. 22) was diluted to as little as 0.8 ϫ 10 8 cell equivalents/ml, the majority of the processing was inhibited (Fig. 8). In contrast, similar dilution of active p20/p10 ICE under these conditions results in no significant loss of activity on its pIL-1␤ substrate (see Ref. 22). Hence p45 cleavage is highly dilution-sensitive.
Effects of Salt on p45 Cleavage--It has been reported previously that the cleavage of pIL-1␤ by p20/p10 ICE is highly salt-sensitive with an IC 50 of about 40 mM (5). When p45 cleavage in cytosolic extracts was measured in the presence of increasing concentrations of NaCl, it was found to be considerably less salt sensitive, producing an IC 50 of about 100 mM (Fig.  9). pIL-1␤ cleavage activity of the p20/p10 ICE formed by a 60-min preincubation of the same extracts in the absence of salt showed an IC 50 of about 45 mM. The effects of salt on the combination of both p45 activation to p20/p10 followed by measurement of the resultant p20/p10 pIL-1␤ cleavage activity yielded an IC 50 of about 35 mM NaCl (Fig. 9). Thus, the effect of salt on the activation of p45 was minimal at the concentrations where pIL-1␤ cleavage activity was inhibited. These results further suggest that the catalytic activity that cleaves p45 varies appreciably from that of p20/p10 ICE itself. DISCUSSION The activation of native p45 ICE in extracts of monocytic cells to the resultant p20/p10 form previously identified and purified (7,24) is characterized in this paper. As is outlined in Fig. 10, p45 is found in cytosol together with smaller amounts of p43, a 2-kDa smaller form that contains all of the same epitopes identified by our ICE antibodies. p43, which is cleaved at the same rate as p45, does not appear to be a degradation product of p45. It is most likely the alternatively spliced form of ICE whose mRNA had previously been identified in monocytic cells (26 -28). We find no immunochemical evidence in monocytic cells for the expression of any smaller ICE forms that could arise from translation of other potential alternative spliced ICE mRNAs that have been reported (see Ref. 27).
While these p45 forms in cytosol have no pIL-1␤ cleavage activity, they do have catalytic activity to initiate cleavage of p45 when the cells are disrupted. In the first place, this activation is prevented by inhibitors of cysteine proteases in general and ICE in particular, while inhibitors of other common proteases classes have no such effect. Most significantly, L-742,395, an irreversible inhibitor, covalently labels with biotin p45 and p43 at the same micromolar concentrations that result in inhibition of p45 cleavage. This labeling is specific, because it is prevented by simultaneously added amounts of other nonbiotinylated ICE inhibitors. Furthermore, other ICE family homologs such as the apopain precursor are not labeled by 20 M L-742,395. Under the conditions described in this paper no other cytoplasmic protease is labeled, and all of the label is immunoprecipitable by specific ICE antibodies. The importance of p45 activity itself to initiate cleavage has also been established in recombinant systems in which the transfection of ICE containing a mutation of the active site Cys 285 of p45 prevents p45 activation (29,30, see also Ref. 31).
As diagrammed in Fig. 10, the initial activation of p45 occurs only by cleavage after Asp 297 (Fig. 1, site 3) at what is to become the COOH terminus of p20 to make a 35-kDa intermediate. The generation of the 35-kDa p12 fragment represents a critical FIG. 8. Effect of dilution on p45 autocatalysis. THP.1 cytosolic extract was incubated at 30°C at the original extract concentration of 10 8 cell equivalents/ml (100%, q) or by dilution with 25 mM HEPES, pH 7.5, 2 mM dithiothreitol to 90% (å), 80% (f), 70% (ç), 60% (E), 50% (Ç), or 33% (Ⅺ). At the indicated times aliquots from each original concentration were removed, diluted to a final dilution of 33%, heated with 2 ϫ SDS sample buffer, separated by PAGE, transferred to PVDF, and immunoblotted with JD3 antiserum. Densitometric scans were made of the p45 (together with p43) bands and normalized to the amount of p45 present in the cytosol.
FIG. 9. Salt effects of p45 autocatalysis and p20/p10 pIL-1␤ cleavage activity. THP.1 cytosol was incubated for 60 min, 30°C, in the presence of the indicated concentrations of NaCl. One aliquot was fractionated by SDS-PAGE, transferred to PVDF, immunoblotted with JD3, and the p45 was determined by densitometry. The amount of p45 cleaved in the presence of salt (f) was normalized to that cleaved in the absence of salt. A second aliquot was prepared in a buffer of the same salt concentration containing [ 35 S]Met pIL-1␤ and incubated another 60 min, 30°C to measure its activity. After SDS-PAGE, transfer to PVDF, autoradiography, and densitometry, the amount of mIL-1␤ formed was normalized to that formed in the salt-free control (å). To measure the effect of salt on the p20/p10 pIL-1␤ cleavage activity, a sample of the same cytosolic extract was preincubated for 60 min, 30°C, in the absence of salt to generate active p20/p10 ICE. The resultant active enzyme was then incubated with [ 35 S]Met pIL-1␤ and salt for 60 min, 30°C (q). In the control incubation without salt, less than 20% of the pIL-1␤ substrate was cleaved (not shown).
step in the activation of ICE. Initial cleavage at no other p45 site yields a fragment that can be biotinylated by L-742,395. This form is much more active than p45 itself. This is seen both by looking at the enzyme itself as well as its effects on the turnover of product. The 35-kDa form has a 100-fold increase in sensitivity to biotinylation by L-742,395 (at 10 -100 nM concentrations versus 5 M for p45). The concentrations of L-742,395 that titrate the 35-kDa forms are those concentrations that inhibit the majority of the p45 cleavage (Fig. 3). In addition to cleaving p45, the 35 kDa activity can also cleave the various 35-kDa intermediates to form p20 and p22, since the formation of p20 and p22 is sensitive to the same inhibitory concentrations of L-742,395 or L-702,066 (Figs. 3 and 7). The lag time associated with the onset of the rapid p45 cleavage (Fig. 3A) may be associated with the generation of this active 35-kDa form, because L-742,395 labeling and biotin blotting of extracts during this period shows a significant accumulation of labeled 35-kDa fragments (data not shown).
Cleavage of p45 at site 3 yields two 35-kDa forms generated simultaneously ( Fig. 2A) with comparable activities (Fig. 4). Found in the same relative ratio as the amount of p45 and p43, the two 35-kDa bands also differ by 2 kDa in apparent molecular mass and probably arise from the comparable processing of the p45 and p43 precursors: both share the same epitopes as judged by their reactivities with all of our ICE antibodies. Their molecular heterogeneity is clearly found at the NH 2 -terminal end of p45, since there is no evidence for such heterogeneity in the p20, p12, or p10 fragments that are subsequently formed. It is of interest to note regarding the importance of cleavage at site 3 for the activation of ICE that this cleavage site appears to be the most conserved site among all of the ICE homologs (see Ref. 32), and this is the site of cleavage leading to the activation of the proapoptotic ICE homolog apopain (16) and the Ced-3 protease itself (33). Indeed, mutation of the Asp at this site in the p45 substantially reduces the activation of ICE (34). Furthermore, a recent report on the refolding of recombinant p45 also points to the importance of cleavage at this site (31).
Not all cleavages of p45 occur, however, at the COOH terminus of p20. Based upon immunoprecipitation data (Fig. 6), there is a measurable amount of cleavage at sites corresponding to the NH 2 terminus of p20 and p22. But these forms are apparently inactive since they are not biotinylated. Because there is no accumulation of these nonproductively cleaved forms at the 35 kDa state, they must also be cleaved down to inactive p20 and p22. It is, therefore, not surprising that in vitro the efficiency of conversion of p45 to active p20/p10 ICE is low. Previously it was found that the 66,000 molecules of p45/ monocyte yielded only about 5000 active p20/p10 molecules (22,24).
In contrast to these three sites of cleavage on p45, there is no cleavage following Asp 316 (site 4). No N-4 fragments are formed initially, and there is no evidence for any 20 kDa sized forms containing the p20-p10 linker peptide. Second, whereas p12 is formed soon after initial p45 cleavage and remains associated with the cleaved 35-kDa fragment, little p10 is found and none can be immunoprecipitated with the newly generated 35-kDa fragments (Fig. 6). Third, incubation of p45 with high concen-FIG. 10. Summary of p45 autocatalysis as determined by L-742,395 labeling and inhibition of autocatalysis (see "Discussion"). The p45 (thick line) and putative translation product of the alternative exon 3 transcript, p43 (thin line), are shown with their potential sites of ICE cleavage (see Fig. 1). These forms show a modest catalytic activity as judged by the 5 M L-742,395 concentration required for labeling ("Weakly Active"). The ability of the active site Cys to be biotinylated is indicated by a solid diamond, whereas an open diamond (right) is associated with ICE forms that cannot be biotinylated. Cleavage of p45 and p43 at site 3 yields the more active 35-kDa forms at the left (thick and thin lines, respectively) together with the p12 protein that remains briefly associated with the complex. This 35-kDa form is labeled by Ͻ100 nM L-742,395 ("Moderately Active"), and it potentially cleaves p45 or the various 35-kDa forms down to the p20/p22 and p10 forms. These p22/p20/p10 forms are labeled by 1 nM L-742,395 ("Strongly Active"). p22 and p20 are formed in equal amounts, although with time p22 is converted into p20 (see Ref. 24). N-1 and N-2 fragments are released but do not remain associated with the complex (they are not immunoprecipitated by antisera to active p20/p10, Fig. 6). Since only the N-1 fragment can be found, N-2 is probably rapidly converted to N-1. p45 and p43 can also be cleaved at sites 1 or 2, but these forms cannot be biotinylated indicating that these are inactive intermediates. Since these forms do not accumulate with time, they are probably also converted down to p20 and p22 which remain inactive. The observation that at least 50% of the total p20/p22 cannot be biotinylated suggests that at least 50% of the p45 is cleaved through this pathway. Alternatively, some of the p45/p43 cleaved at site 3 may become inactivated before it is processed to p22/p20. trations of p20/p10 ICE does not yield significant amounts of p10, even though cleavage can occur at what appears to be site 3 to generate 35 kDa and p12 fragments (Fig. 7B). p10 is, however, tightly associated with p20 or p22 in the active ICE ultimately formed, since it coimmunoprecipitates with p20 (Fig. 6). This indicates that in the generation of active ICE, the p12 must be processed down to p10 just as the active 35-kDa fragments are processed down to p22 and p20 (see Fig. 10).
The data described here suggest that p45 undergoes a considerable conformational change during the process of activation. First of all, the active site changes from requiring micromolar concentrations of the tetrapeptide L-742,395 for cleavage by p45, to tens of nanomolar by the 35 kDa/p12 form, to nanomolar or subnanomolar by p20/p10. pIL-1␤ is cleaved only by p20/p10 and not by p45 or the 35-kDa form. Furthermore, the dilution sensitivity of p45 is greater than p20/p10, while its salt sensitivity is less. Second, the absence of p45 cleavage at site 4 suggests that this site is buried, even though it is exposed in the active form of ICE (20,21). Last, the 35-kDa forms generated by normal p45 activation can be quickly activated to p22 and p20, while the 35-kDa intermediate formed by p20/p10 cleavage remains stable until it is ultimately cleaved to a 24-kDa form at very high p20/p10 concentrations (Fig. 9B). Cleavage of p45 at the Asp 297 site is hence insufficient in and of itself to generate fully functional ICE processing. This suggests that the conformational changes in the 35-kDa form that permit this secondary cleavage have not occurred.