Interleukin-1 beta (IL-1 beta)-converting enzyme cleaves the IL-1 beta precursor to mature IL-1 beta, an important mediator of inflammation. The identification of the enzyme as a unique cysteine protease and the design of potent peptide aldehyde inhibitors are described. Purification and cloning of the complementary DNA indicates that IL-1 beta-converting enzyme is composed of two nonidentical subunits that are derived from a single proenzyme, possibly by autoproteolysis. Selective inhibition of the enzyme in human blood monocytes blocks production of mature IL-1 beta, indicating that it is a potential therapeutic target.

Interleukin 1 (IL-1) is a lymphokine secreted by monocytes in response to a variety of inflammatory stimuli. IL-1fB the predominant form of IL-1 produced by human monocytes, is synthesized as an inactive precursor of 31 kDa and is cleaved at Asp"'6-Ala"17 to yield a 17.5-kDa extracellular form. The exact cellular site of cleavage and mechanism of secretion is at present unknown. We have prepared cell-free postnuclear extracts from freshly isolated human monocytes as well as THP. (1) provided the first substantive evidence that, in a mouse monocyte cell line, IL-1 was synthesized as a cell-associated precursor that could be chased into an extracellular 17-kDa form. Subsequently, reports emerged which suggested that a 31-kDa form of IL-1i3 was associated with human monocytes (2-5) and that this material was cleaved to release the mature form (2, 4, 5). These studies were corroborated by cDNA sequence data from a number of species which indicated that IL-1 mRNA encodes a larger protein than that identified as mature secreted IL-1 (6-10). As precursor IL-183 (pre-IL-1,j)is unable to bind to IL-1 receptors and is biologically inactive (11), some form of proteolytic processing is apparently required to release active IL-1p8. While the kinetics of IL-1 synthesis and secretion has been analyzed in some detail, little has been uncovered about the mechanism by which IL-1 is synthesized, processed, and secreted. Analysis ofthe predicted amino acid sequence from pre-IL-1,3 cDNA has not revealed the presence of a unique hydrophobic signal sequence domain, common to most secreted proteins (6)(7)(8)(9)(10)12). The N-terminal amino acid of mature monocyte IL-1p from humans has been sequenced by a number of investigators as Ala"17 (6, 13), suggesting that a cleavage site exists between Asp'6 and Ala"7. While the first 116 residues may be considered a signal sequence of sorts, it is not recognized as such by otherwise competent endoplasmic reticulum membranes (G.L., unpublished observation). Young et al. (14) showed that mature pre-IL-1P was not secreted from hamster fibroblasts that were stably transformed with pre-IL-1,i cDNA. Instead, large amounts of the precursor accumulated in the cytoplasm of the cell (14). Lomedico et al. (12) The processing of IL-1f3 has recently been investigated by using purified recombinant precursor as a substrate (5, 17).Hazuda et al. (5) showed that pre-IL-1f3, when added to intact human blood monocytes, was not cleaved or processed in any fashion, arguing against an extracellular site of processing. In another report, a potential pre-IL-1,8 cleavage activity was identified in a pelletable compartment of KG-1 cells, a neutrophil-like cell line. This enzymatic activity was able to generate IL-1 activity of similar size to authentic IL-1 from a partially purified pre-IL-1f3 substrate (17). However, the products were not sequenced and the site ofcleavage was not identified.In this report, we describe an in vitro processing system in which mature 17.

SummaryInterleukin-1 [3 (IL-l[3)-converting enzyme (ICE) is a novel cysteine protease that cleaves the 31-kD inactive cytoplasmic IL-113 precursor into active extracellular 17-kD IL-113. The ICE gene product is a 45-kD proenzyme that requires proteolytic processing to activate ICE. Active ICE is a heterodimer consisting of equal amounts of p20 and p l0 subunits. Generation of active ICE is affected by the removal of an 11-kD NH2-terminal precursor domain (pl 1) and an internal 19-amino acid sequence that separates the 20-and 10-kD subunits. Immuno-electron microscopy was performed on human monocytes with immunoglobulins recognizing the active (p20) or precursor (p11) domains of ICE. Elutriated monocytes were stimulated with 50 pM lipopolysaccharide followed by heat-killed Staphylococcus aureus under conditions that induce maximal rates of IL-113 secretion. Ultrathin cryosections were cut from fixed frozen pellets of these monocytes and were immunogold labeled with either antibody. Active and precursor domain ICE epitopes were localized in the cytoplasmic ground substance, but they were not detected within the endoplasmic reticulum, the Golgi apparatus, and secretory granules of activated or inactive monocytes. Importantly, numerous ICE p20 epitopes were also observed on the extracellular surfaces of the cell membrane, and were concentrated on the microvilli. Very similar patterns of ICE localization were obtained with unstimulated blood monocytes. In contrast, ICE pl I epitopes were not detected on the surfaces of these rnonocytes. Likewise, labeling of fixed ultrathin cryosections ofmonocytes with a biotinylated irreversible ICE inhibitor [Ac-Tyr-Val-Lys(biotin)-Asp-(acyloxy)-methyl-ketone] showed that the compound localized on the outer cell surface as well, and to a lesser extent, within the cytoplasmic ground substance. Furthermore, antipeptide antibodies specific for either the mature or precursor domains of IL-113 were both localized upon the cell membrane after stimulation of IL-113 secretion. Lipopolysaccaride-primed monocytes that synthesized, but did not secrete IL-113, exhibited only cytoplasmic staining. The data suggests that mature IL-l[3 is generated via cleavage of the 31-kD inactive cytoplasmic IL-lJ3 precursor by ICE after association with the plasma membrane during secretion.

The subcellular location of IL-1 beta was determined using a postsectioning immunoelectron microscopic method on ultrathin frozen sections of human monocytes stimulated with LPS. This methodology permits access of antibody probes to all sectioned intracellular compartments, and their visualization at high resolution. Staining was performed with a rabbit antibody that specifically recognized amino acids 197-215 in the 33-kD IL-1 beta precursor molecule, followed by affinity-purified goat anti-rabbit IgG conjugated to 10 nm colloidal gold particles. Approximately 90% of the IL-1 beta antigens were localized in the ground substance of the cytoplasm at 4 or 20 h after activation, when both intracellular and extracellular accumulation of IL-1 beta was well underway. No significant IL-1 beta staining was observed on the outer cell membrane, nor within the lumens of the endoplasmic reticulum (ER), the Golgi apparatus, or secretory vesicles. In contrast, lysozyme was localized in the ER and dense secretory granules using these methods. Our results suggest that IL-1 beta is not anchored on the plasma membrane, and that its secretion occurs by a novel mechanism that does not use a secretory leader sequence, nor the classical secretory pathway involving the ER and Golgi apparatus.

The pI-6.8 species of normal human interleukin 1 (IL-1) has been isolated by ion-exchange and reverse-phase high-performance liquid chromatography. The isolated material had a molecular weight of 18,000, and had a specific bioactivity of 1.7 X 10(7) half-maximal U/mg in the murine thymocyte proliferation assay, values similar to those obtained for murine P388D1-derived IL-1 (12), and human IL-1 isolated by a previously published purification protocol (15). Amino-terminal sequence analysis revealed a single N-terminal, and resulted in the identification of 30 of the first 35 amino acid residues. Sequence of three CNBr cleavage fragments of purified IL-1 resulted in the identification of an additional 38 residues. All of the sequences agree exactly with those deduced from complementary DNA (cDNA) by Auron, et al. (18), demonstrating that this cloned cDNA, though considerably different from the cDNA reported for murine IL-1 (12), nevertheless codes for the pI-6.8 species of human IL-1. The evidence also shows that the precursor protein for human IL-1 is largely processed at the N-terminal end. Little or no processing occurs at the carboxy-terminal end. Sequence homology with interferon-inducing factor (26) suggests that the pI-6.8 species of human IL-1 is a member of a gene family. Although equally potent in the murine thymocyte proliferation assay, murine IL-1 and the pI-6.8 species of human IL-1 are structurally distinct. Further study will answer the interesting question as to the relationship of the other charged species of human IL-1 to these distinct IL-1 classes.