Chromatin degradation in the death of thymic lymphocytes induced by radiation or dexamethasone: the necessity for a stage of preliminary proteolysis

Proteolytic activity in a protein fraction of a rat thymocyte nuclear matrix was found to increase 1-2 h after gamma-irradiation or administration of dexamethazone. Cycloheximide did not prevent the observed protease activation. Neither histons nor thymocyte nuclear matrix proteins were subjected to proteolysis after exposure to radiation or the hormone. Such proteolysis inhibitors as phenylmethylsulfonyl fluorine, trasilol, and partly leupeptine inhibited nuclear DNA degradation in irradiated and dexamethazone treated thymus lymphocytes. In all appearance, this effect was not due to Ca/Mg-dependent endonuclease inactivation. The same was observed in the system of autolytic chromatin degradation in isolated thymocyte nuclei.     

Methionine sulfoxide and proteolytic cleavage contribute to the inactivation of cathepsin G by hypochlorous acid: an oxidative mechanism for regulation of serine proteinases by myeloperoxidase

Using myeloperoxidase and hydrogen peroxide, activated neutrophils produce high local concentrations of hypochlorous acid (HOCl). They also secrete cathepsin G, a serine protease implicated in cytokine release, receptor activation, and degradation of tissue proteins. Isolated cathepsin G was inactivated by HOCl but not by hydrogen peroxide in vitro. We found that activated neutrophils lost cathepsin G activity by a pathway requiring myeloperoxidase, suggesting that oxidants generated by myeloperoxidase might regulate cathepsin G activity in vivo. Tandem mass spectrometric analysis of oxidized cathepsin G revealed that loss of a peptide containing Asp108, which lies in the active site, associated quantitatively with loss of enzymatic activity. Catalytic domain peptides containing Asp108 were lost from the oxidized protein in concert with the conversion of Met110 to the sulfoxide. Release of this peptide was blocked by pretreating cathepsin G with phenylmethylsulfonyl fluoride, strongly implying that oxidation introduced proteolytic cleavage sites into cathepsin G. Model system studies demonstrated that methionine oxidation can direct the regiospecific proteolysis of peptides by cathepsin G. Thus, oxidation of Met110 may contribute to cathepsin G inactivation by at least two distinct mechanisms. One involves direct oxidation of the thioether residue adjacent to the aspartic acid in the catalytic domain. The other involves the generation of new sites that are susceptible to proteolysis by cathepsin G. These observations raise the possibility that oxidants derived from neutrophils restrain pericellular proteolysis by inactivating cathepsin G. They also suggest that methionine oxidation could render cathepsin G susceptible to autolytic cleavage. Myeloperoxidase may thus play a previously unsuspected role in regulating tissue injury by serine proteases during inflammation.     

O-glycosylation regulates autolysis of cellular membrane type-1 matrix metalloproteinase (MT1-MMP)

MT1-MMP is a key enzyme in cancer cell invasion and metastasis. The activity of cellular MT1-MMP is regulated by furin-like proprotein convertases, TIMPs, shedding, autoproteolysis, dimerization, exocytosis, endocytosis, and recycling. Our data demonstrate that, in addition to these already known mechanisms, MT1-MMP is regulated by O-glycosylation of its hinge region. Insignificant autolytic degradation is characteristic for naturally expressed, glycosylated, MT1-MMP. In turn, extensive autolytic degradation, which leads to the inactivation of the protease and the generation of its C-terminal membrane-tethered degraded species, is a feature of overexpressed MT1-MMP. We have determined that incomplete glycosylation stimulates extensive autocatalytic degradation and self-inactivation of MT1-MMP. Self-proteolysis commences during the secretory process of MT1-MMP through the cell compartment to the plasma membrane. The strongly negatively charged sialic acid is the most important functional moiety of the glycopart of MT1-MMP. We hypothesize that sialic acid of the O-glycosylation cassette restricts the access of the catalytic domain to the hinge region and to the autolytic cleavage site and protects MT1-MMP from autolysis. Overall, our results point out that there is a delicate balance between glycosylation and self-proteolysis of MT1-MMP in cancer cells and that when this balance is upset the catalytically potent MT1-MMP pool is self-proteolyzed.     

The role of subunit autolysis in activation of smooth muscle Ca2+-dependent proteases

Ca2+-dependent proteases isolated from chicken gizzard and bovine aortic smooth muscle were compared with respect to subunit autolysis and the role of autolysis in modulating enzyme activity. The protease isolated from chicken gizzard was a heterodimer consisting of 80,000- and 30,000-dalton subunits. The protease isolated under identical conditions from bovine aorta consisted of 75,000- and 30,000-dalton subunits. In the presence of Ca2+, both enzymes underwent autolysis of their 30,000-dalton subunits with conversion to an 18,000-dalton species. In addition, the 80,000-dalton subunit of the gizzard protease was degraded to a 76,000-dalton form. The Ca2+ concentrations required for autolysis of the 30,000-dalton subunits were different for the two enzymes (i.e. gizzard: K0.5 Ca2+ = 335 microM; aortic: K0.5 Ca2+ = 1,250 microM) although in both cases, stimulation of autolysis by Ca2+ exhibited positive cooperativity. When compared with respect to kinetics of substrate degradation, the native forms of the smooth muscle Ca2+-dependent proteases (gizzard, GIIa = 80,000/30,000-dalton heterodimer; bovine aortic, IIa = 75,000/30,000-dalton heterodimer) exhibited a lag phase in product appearance. On the other hand, the autolyzed forms (gizzard, GIIb = 76,000/18,000-dalton heterodimer; bovine aortic, IIb = 75,000/18,000-dalton heterodimer) exhibited linear rates of substrate degradation. These results were analyzed in terms of autolysis of the 30,000-dalton subunits as determined by the conversion of this subunit to its 18,000 dalton form. For both enzymes, the time course for the autolytic transition, 30,000----18,000 daltons, and Ca2+-dependence of the apparent rate constants for this transition were found to correlate well with the lag phase in enzymatic activity. No such correlation could be established for the 80,000----76,000 dalton autolytic transition of the high molecular mass subunit of the gizzard protease. Our results suggest that catalytic activity of the Ca2+-dependent proteases isolated from gizzard and bovine aortic smooth muscle requires autolysis of the 30,000-dalton subunit. The native or unautolyzed forms of these enzymes appear to be proenzymes that can be activated by autolysis.     

Further stabilization of earthworm serine protease by chemical modification and immobilization

Earthworm serine protease is more stable and is less affected by organic solvents and detergent than other proteases. However, it is inactivated, probably by autolysis, at 60 degrees C or above under alkaline conditions. Further stabilization was managed by chemical modification of the enzyme with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide and phenylglyoxal to protect the activity from the autolytic inactivation. Stabilization was possible also under acidic conditions, in which the stability of the enzyme was rather low, by immobilization with folded sheet mesoporous material. Thus, further stabilization of the enzyme has been achieved by chemical modification or immobilization.     

N-linked glycans of the human insulin receptor and their distribution over the crystal structure

The human insulin receptor (IR) homodimer is heavily glycosylated and contains a total of 19 predicted N-linked glycosylation sites in each monomer. The recent crystal structure of the IR ectodomain shows electron density consistent with N-linked glycosylation at the majority of sites present in the construct. Here, we describe a refined structure of the IR ectodomain that incorporates all of the N-linked glycans and reveals the extent to which the attached glycans mask the surface of the IR dimer from interaction with antibodies or other potential therapeutic binding proteins. The usefulness of Fab complexation in the crystallization of heavily glycosylated proteins is also discussed. The compositions of the glycans on IR expressed in CHO-K1 cells and the glycosylation deficient Lec8 cell line were determined by protease digestion, glycopeptide purification, amino acid sequence analysis, and mass spectrometry. Collectively the data reveal: multiple species of complex glycan at residues 25, 255, 295, 418, 606, 624, 742, 755, and 893 (IR-B numbering); multiple species of high-mannose glycan at residues 111 and 514; a single species of complex glycan at residue 671; and a single species of high-mannose glycan at residue 215. Residue 16 exhibited a mixture of complex, hybrid, and high-mannose glycan species. Of the remaining five predicted N-linked sites, those at residues 397 and 906 were confirmed by amino acid sequencing to be glycosylated, while that at residue 78 and the atypical (NKC) site at residue 282 were not glycosylated. The peptide containing the final site at residue 337 was not recovered but is seen to be glycosylated in the electron density maps of the IR ectodomain. The model of the fully glycosylated IR reveals that the sites carrying high-mannose glycans lie at positions of relatively low steric accessibility.     

Histone proteolysis in fish erythrocyte nuclei

The erythrocyte histones of trout, carp, white sucker, and chicken are subject to very different levels of autolytic activity. Carp erythrocyte histones extracted from typical nuclear preparations suffer extensively from degradation; histones 1, 5, and 3 (H1, H5, and H3 respectively) are preferentially cleaved and characteristic peptides designated P1, P2, and P3 appear during the course of proteolysis. Generally, erythrocytes from different fish species yield highly disparate proportions of H1 and H5, but this is not a consequence of the variable levels of proteolytic activity in these species. Phenylmethylsulfonyl fluoride (PMSF) (1.0 mM) was found to be superior to 50 mM sodium bisulfite as a protease inhibitor and was well suited for use in media employed for cell washes and the isolation of nuclei. Nonetheless, in carp erythrocytes residual protease activity (qualitatively the same as the uninhibited activity) persists even in the presence of PMSF. It is activated during cell lysis and remains associated with the nuclear fraction of the lysate during subsequent washes. The isolation of intact nuclei is important for the ultimate extraction of undegraded histone, especially from sources in which the risks of autolysis are high or unknown.     

Mechanism of pneumococcal cell wall degradation in vitro and in vivo.

We compared the products of autolytic amidase-catalyzed wall degradation in vivo (in penicillin-induced lysis) and in vitro. Pneumococci labeled in their cell wall stem peptides by radioactive lysine were treated with penicillin, and the nature of wall degradation products released to the medium during lysis of the bacteria was determined. At early times of lysis (20% loss of wall label), virtually all the radioactive peptides released (greater than 94%) were of high molecular size and were still attached to glycan and teichoic acid. At times of more extensive bacterial lysis (56%), progressively larger and larger fractions of the released peptides became free, i.e., detached from glycan and teichoic acid. Analysis of the nondegraded residual wall material by high-resolution high-pressure liquid chromatography revealed that this in vivo-triggered autolysis did not involve selective hydrolysis of some of the chemically distinct stem peptides. Parallel in vitro experiments yielded completely different results. Purified pneumococcal cell walls labeled with radioactive lysine were treated in vitro with low concentrations of pure amidase, and the nature of wall degradation products released during limited hydrolysis and after more extensive degradation was determined. In sharp contrast to the in vivo experiments, the main products of in vitro hydrolysis were free peptides. After a short treatment with amidase (resulting in a 20% loss of label), the material released was enriched for the monomeric stem peptides. At all times of hydrolysis (including the time of extensive degradation), only a relatively small fraction of the released wall peptides was covalently attached to glycan and teichoic acid components (17% as compared with 40% in the intact cell wall). We propose that the in vivo-triggered amidase activity first attacks the amide bonds in some strategically located (or unprotected) stem peptides that hold large segments of cell wall material together. The observations indicate that the in vivo activity of the pneumococcal autolysin is under topographic constraints.     

Expression of functional hypodermin A,a serine protease from Hypoderma lineatum (Diptera,Oestridae), in Schneider 2 cells

Hypodermin A (HA) is a serine protease secreted by first-instar Hypoderma lineatum larvae (Oestridae, Diptera). It plays a crucial role in induced immunosuppression during hypodermosis. This report describes the production of recombinant HA in Drosophila melanogaster Schneider 2 cells, its purification and its characterization, and compares it with the protease extracted form parasite larvae. The recombinant protein and the native HA have similar biochemical and biological features. Activity of the recombinant protease on the lymphocyte proliferation inhibition and on membrane antigen cleavage was tested and shown to be similar to the native one. Tunicamycin treatment of the recombinant HA shows that the two putative glycosylation sites carry glycan residues. Unglycosylated recombinant HA has the same enzymatic activity as the fully glycosylated protein, indicating that glycosylation is not important for the protease activity of HA.     

Protease domain glycans affect oligomerization, disulfide bond formation, and stability of the meprin A metalloprotease homo-oligomer

The meprin A homo-oligomer is a highly glycosylated, secreted zinc metalloprotease of the astacin family and metzincin superfamily. This isoform of meprin is composed of disulfide-bonded dimers of alpha subunits that further associate to form large, secreted megadalton complexes of 10 or more subunits. The aim of this study was to determine the sites of glycan attachment and to assess their ability to affect the formation and stability of the homo-oligomer. Nine of the ten potential N-linked glycosylation sites (Asn-41, Asn-152, Asn-234, Asn-270, Asn-330, Asn-426, Asn-452, Asn-546, and Asn-553) were found to be glycosylated in recombinant mouse meprin A using chemical and enzymatic deglycosylation methods and electrospray ionization mass spectrometry. Chemical cross-linking demonstrated that carbohydrates are at or near the noncovalent subunit interface. The removal of two glycans in the protease domain at Asn-234 and Asn-270, as well as one in the tumor necrosis factor receptor-associated factor domain at Asn-452, by a deglycosidase under nondenaturing conditions decreased the chemical and thermal stability of the homo-oligomer without affecting quaternary structure. Site-directed mutagenesis demonstrated that no single glycan was essential for oligomer formation; however, the combined absence of the glycans at Asn-152 and Asn-270 in the protease domain hindered intersubunit disulfide bond formation, prevented noncovalent associations, and abolished enzymatic activity. These studies provide insights into the role of glycans in the biosynthesis, activity, and stability of this extracellular protease.     

The latent form of macropain (high molecular weight multicatalytic protease) restores ATP-dependent proteolysis to soluble extracts of BHK fibroblasts pretreated with anti-macropain antibodies

Specific immunoadsorption of the high molecular weight multicatalytic protease, macropain, from postmicrosomal extracts of BHK fibroblasts inhibited ATP-dependent proteolysis of exogenous protein substrates. The immunoprecipitated macropain represented the latent (L) form of the protease because it had low protease activity but was activated by methods that activate purified macropain L. Reconstitution of the antibody-treated extracts with purified macropain L, but not macropain A, from bovine heart or human erythrocytes, completely restored ATP-dependent proteolysis, even though ATP did not directly activate either purified macropain L or the immunoprecipitated protease. Reconstituted ATP-dependent proteolysis was saturable with respect to added macropain and never exceeded the level of proteolysis present in the original extract. These results indicate that macropain L plays a key role in ATP-dependent proteolysis but suggest that the protease may require interaction with or modification by another cellular component to demonstrate this effect.     

Site-selective glycosylation of subtilisin Bacillus lentus causes dramatic increases in esterase activity

Using site directed mutagenesis combined with chemical modification, we have developed a general and versatile method for the glycosylation of proteins which is virtually unlimited in the scope of proteins and glycans that may be conjugated and in which the site of glycosylation and the nature of the introduced glycan can be carefully controlled. We have demonstrated the applicability of this method through the synthesis of a library of 48 glycosylated forms of the serine protease subtilisin Bacillus lentus (SBL) as single, pure species. As part of our ongoing program to tailor the activity of SBL for use in peptide synthesis, we have screened these enzymes for activity against the esterase substrate succinyl-Ala-Ala-Pro-Phe-S-benzyl. Gratifyingly, 22 enzymes displayed greater than wild type (WT) activity. Glycosylation at positions 62, in the S2 pocket, resulted in five glycosylated forms of SBL that were 1.3- to 1.9-fold more active than WT. At position 217, in the S1' pocket, all glycosylations increased kcat/KM up to a remarkable 8.4-fold greater than WT for the glucosylated enzyme L217C-S-beta-Glc(Ac)3. Furthermore, the ratio of amidase to esterase activity, (kcat/KM)esterase/(kcat/KM)amidase (E/A), is increased relative to wild type for all 48 glycosylated forms of SBL. Again, the most dramatic changes are observed at positions 62 and 217 and L217C-S-beta-Glc(Ac)3 has an E/A that is 17.2-fold greater than WT. The tailored specificity and high activity of this glycoform can be rationalized by molecular modeling analysis, which suggests that the carbohydrate moiety occupies the S1' leaving group pocket and enhances the rate of deacylation of the acyl-enzyme intermediate. These glycosylated enzymes are ideal candidates for use as catalysts in peptide synthesis as they have greatly increased (kcat,KM)esterase and severely reduced (kcat/KM)amidase and will favor the formation of the amide bond over hydrolysis.     

Structure of the glycosyl-phosphatidylinositol membrane anchor of the Leishmania major promastigote surface protease

In common with many other plasma membrane glycoproteins of eukaryotic origin, the promastigote surface protease (PSP) of the protozoan parasite Leishmania contains a glycosyl-phosphatidylinositol (GPI) membrane anchor. The GPI anchor of Leishmania major PSP was purified following proteolysis of the PSP and analyzed by two-dimensional 1H-1H NMR, compositional and methylation linkage analyses, chemical and enzymatic modifications, and amino acid sequencing. From these results, the structure of the GPI-containing peptide was found to be Asp-Gly-Gly-Asn-ethanolamine-PO4-6Man alpha 1-6Man alpha 1-4GlcN alpha 1-6myo-inositol-1-PO4-(1-alkyl-2-acyl-glycerol). The glycan structure is identical to the conserved glycan core regions of the GPI anchor of Trypanosoma brucei variant surface glycoprotein and rat brain Thy-1 antigen, supporting the notion that this portion of GPIs are highly conserved. The phosphatidylinositol moiety of the PSP anchor is unusual, containing a fully saturated, unbranched 1-O-alkyl chain (mainly C24:0) and a mixture of fully saturated unbranched 2-O-acyl chains (C12:0, C14:0, C16:0, and C18:0). This lipid composition differs significantly from those of the GPIs of T. brucei variant surface glycoprotein and mammalian erythrocyte acetylcholinesterase but is similar to that of a family of glycosylated phosphoinositides found uniquely in Leishmania.     

Interplay between acid phosphatase and cysteine proteases in mediating vitellin degradation during early embryogenesis of Periplaneta americana

In this work, we characterized the activities of two classes of proteases and AcP during early embryogenesis of Periplaneta americana. AcP activity was first detected at day 6 and reached a maximum level at day 10 of development. Using phosphoamino acids, phosphatase activity was shown to be directed only against phosphotyrosine at day 6 while at day 10 it was also active against phosphoserine. In parallel, two classes of proteases were detected and located within yolk granules: a clan CA-cysteine protease, which was inhibited by E-64, insensitive to CA 074 and activated by acidic pH at day 3; and a neutral serine protease, which was inhibited by aprotinin at day 6. Assays of vitellin (Vt) degradation evidenced that incubations at neutral pH induced slight proteolysis, while the incubations at acidic pH did not result in Vt degradation. However, pre-incubations of Vt with AcP increased the levels of Vt acidic proteolysis and this could be inhibited by the addition of phosphatase inhibitors. On the other hand, the same pre-incubations showed no effects on the profile of degradation at neutral pH. We propose that AcP and cysteine protease cooperate to assure Vt breakdown during early embryogenesis of P. americana.     

The N Domain of Human Angiotensin-I-converting Enzyme: THE ROLE OF N-GLYCOSYLATION AND THE CRYSTAL STRUCTURE IN COMPLEX WITH AN N DOMAIN-SPECIFIC PHOSPHINIC INHIBITOR, RXP407*

Angiotensin-I-converting enzyme (ACE) plays a critical role in the regulation of blood pressure through its central role in the renin-angiotensin and kallikrein-kinin systems. ACE contains two domains, the N and C domains, both of which are heavily glycosylated. Structural studies of ACE have been fraught with severe difficulties because of surface glycosylation of the protein. In order to investigate the role of glycosylation in the N domain and to create suitable forms for crystallization, we have investigated the importance of the 10 potential N-linked glycan sites using enzymatic deglycosylation, limited proteolysis, and mass spectrometry. A number of glycosylation mutants were generated via site-directed mutagenesis, expressed in CHO cells, and analyzed for enzymatic activity and thermal stability. At least eight of 10 of the potential glycan sites are glycosylated; three C-terminal sites were sufficient for expression of active N domain, whereas two N-terminal sites are important for its thermal stability. The minimally glycosylated Ndom389 construct was highly suitable for crystallization studies. The structure in the presence of an N domain-selective phosphinic inhibitor RXP407 was determined to 2.0 Å resolution. The Ndom389 structure revealed a hinge region that may contribute to the breathing motion proposed for substrate binding.     

Autolytic transition of mu-calpain upon activation as resolved by antibodies distinguishing between the pre- and post-autolysis forms.

A novel method to observe the autolytic activation of a mammalian cytoplasmic calcium protease, mu-calpain, was developed using a set of antipeptidic antibodies capable of distinguishing between the pre- and post-autolysis forms of the enzyme. Antibodies raised against synthetic peptides designed to match the N-terminal sequences of the pre- and post-autolysis forms of the mu-calpain large subunit reacted specifically with the corresponding form of calpain and not with the other. The antibodies were specific and sensitive enough to detect the antigens in crude cell lysates. The relevance of the immunochemical detection of calpain activation was confirmed by the observation that proteolysis of a substrate protein by purified mu-calpain paralleled autolysis at various pCa as probed by these antibodies and that autolysis preceded substrate proteolysis. We also observed calcium-dependent autolysis of calpain accompanying subsequent proteolysis of substrate in intact cells using the antibodies. The method will provide a novel approach to assess the physiological targets of the enzyme by determining the local intracellular sites of calpain activation.     

Degradation of oxidatively denatured proteins in Escherichia coli

When exposed to oxidative stress, by oxygen radicals or H2O2, E. coli exhibited decreased growth, decreased protein synthesis, and dose-dependent increases in protein degradation. The quinone menadione induced proteolysis when cells were incubated in air, but was not effective when cells were incubated without oxygen. Anaerobically grown cells also exhibited significantly lower proteolytic capacity than did cells that were grown aerobically. Xanthine plus xanthine oxidase (which generate O2- and H2O2) caused a stimulation of proteolysis which was inhibitable by catalase, but not by superoxide dismutase: Indicating that H2O2 was responsible for the increased protein degradation. Indeed, H2O2 alone was effective in inducing increased intracellular proteolysis. Two-dimensional polyacrylamide gel electrophoresis of [3H]leucine labeled E. coli revealed greater than 50% decreases in the concentrations of 10-15 cell proteins following H2O2 or menadione exposure, while several other proteins were less severely affected. To test for the presence of soluble proteases, we prepared cell-free extracts of E. coli and incubated them with radio-labeled protein substrates. E. coli extracts degraded casein and globin polypeptides at rapid rates but showed little activity with native proteins such as superoxide dismutase, hemoglobin, bovine serum albumin, or catalase. When these same proteins were denatured by exposure to oxygen radicals or H2O2, however, they became excellent substrates for degradation in E. coli extracts. Studies with albumin revealed correlations greater than 0.95 between the degree of oxidative denaturation and proteolytic susceptibility. Pretreatment of E. coli with menadione or H2O2 did not increase the proteolytic capacity of cell extracts; indicating that neither protease activation, nor protease induction were required.(ABSTRACT TRUNCATED AT 250 WORDS)     

Identification and characterization of site‐specific N‐glycosylation in the potassium channel Kv3.1b

The potassium ion channel Kv3.1b is a member of a family of voltage‐gated ion channels that are glycosylated in their mature form. In the present study, we demonstrate the impact of N‐glycosylation at specific asparagine residues on the trafficking of the Kv3.1b protein. Large quantities of asparagine 229 (N229)‐glycosylated Kv3.1b reached the plasma membrane, whereas N220‐glycosylated and unglycosylated Kv3.1b were mainly retained in the endoplasmic reticulum (ER). These ER‐retained Kv3.1b proteins were susceptible to degradation, when co‐expressed with calnexin, whereas Kv3.1b pools located at the plasma membrane were resistant. Mass spectrometry analysis revealed a complex type Hex₃HexNAc₄Fuc₁ glycan as the major glycan component of the N229‐glycosylated Kv3.1b protein, as opposed to a high‐mannose type Man₈GlcNAc₂ glycan for N220‐glycosylated Kv3.1b. Taken together, these results suggest that trafficking‐dependent roles of the Kv3.1b potassium channel are dependent on N229 site‐specific glycosylation and N‐glycan structure, and operate through a mechanism whereby specific N‐glycan structures regulate cell surface expression.     

The roles of surface loop insertions and disulfide bond in the stabilization of thermophilic WF146 protease

Thermophilic WF146 protease possesses four surface loop insertions and a disulfide bond, resembling its psychrophilic (subtilisins S41 and S39) and mesophilic (subtilisins SSII and sphericase) homologs. Deletion of the insertion 3 (positions 193-197) or insertion 4 (positions 210-221) of WF146 protease resulted in a significant decrease of the enzyme stability. In addition, substitution of the residues Pro211 and Ala212 or residue Glu221 which localized in the vicinity of a Ca(2+) binding site of the enzyme by the corresponding residues in subtilisin S41 remarkably reduced the half-life of the enzyme at 70 degrees C, suggesting that the three residues contributed to the thermostability of the enzyme, probably by enhancing the affinity of enzyme to Ca(2+). In the presence of dithiothreitol, the WF146 protease suffered excessive autolysis, indicating that the Cys52-Cys65 disulfide bond played a critical role in stabilizing the WF146 protease against autolysis. The autolytic cleavage sites of the WF146 protease were identified to locate between residues Asn63-Gly64 and Cys65-Ala66 by N-terminal amino acid analysis of the autolytic product. It was noticed that the effect of the autolytic cleavage at Asn63-Gly64 could be compensated by the disulfide bond Cys52-Cys65 under non-reducing condition, and the disulfide bond cross-linked autolytic product remained active. The apparent stabilization effect of the disulfide bond Cys52-Cys65 in the WF146 protease might provide a rational basis for improving the stability of subtilase against autolysis by protein engineering.