Inhibitory mechanism of a cross-class serpin, the squamous cell carcinoma antigen 1.

The squamous cell carcinoma antigen (SCCA) 1 and its homologous molecule, SCCA2, belong to the ovalbumin-serpin family. Although SCCA2 inhibits serine proteinases such as cathepsin G and mast cell chymase, SCCA1 targets cysteine proteinases such as cathepsin S, K, L, and papain. SCCA1 is therefore called a cross-class serpin. The inhibitory mechanism of the standard serpins is well characterized; those use a suicide substrate-like inhibitory mechanism during which an acyl-enzyme intermediate by a covalent bond is formed, and this complex is stable against hydrolysis. However, the inhibitory mechanism of cross-class serpins remains unresolved. In this article, we analyzed the inhibitory mechanism of SCCA1 on a cysteine proteinase, papain. SCCA1 interacted with papain at its reactive site loop, which was then cleaved, as the standard serpins. However, gel-filtration analyses showed that SCCA1 did not form a covalent complex with papain, in contrast to other serpins. Interaction with SCCA1 severely impaired the proteinase activity of papain, probably by inducing conformational change. The decreased, but still existing, proteinase activity of papain was completely inhibited by SCCA1 according to the suicide substrate-like inhibitory mechanism; however, papain recovered its proteinase activity with the compromised level, when all of intact SCCA1 was cleaved. These results suggest that the inhibitory mechanism of SCCA1 is unique among the serpin superfamily in that SCCA1 performs its inhibitory activity in two ways, contributing the suicide substrate-like mechanism without formation of a covalent complex and causing irreversible impairment of the catalytic activity of a proteinase.

The serpins (serine proteinase inhibitors) are a superfamily of proteinase inhibitors characterized by a conserved structure and employing a suicide substrate-like inhibitory mechanism (1,2). The structure of the serpins consists of three ␤ sheets (A-C), nine ␣ helices (A-I), and the reactive site loop (RSL) 1 composed of ϳ17 amino residues (1). The inhibitory mechanism of the serpin is well characterized (2). The exposed RSL of the serpin is recognized by the proteinase, and an initial noncovalent Michaelis encounter complex is formed. Then, in the inhibitory pathway, a "bait" peptide bond (P1-P1Ј) that mimics the normal substrate of the proteinase is attacked by the active serine residue of the proteinase, subsequently forming an acylenzyme intermediate linked by an oxy-ester bond. In the cleaved form, the P side of the RSL inserts into the body of the protein, which dramatically changes the conformations of the serpin and the proteinase, making it impossible for the ester bond to hydrolyze (3). In the non-inhibitory or substrate pathways, the serpin is cleaved by the proteinase just as the substrate of the proteinase after the Michaelis encounter complex is formed. It has been revealed that the serpins are involved in various kinds of biological functions: fibrinolysis, coagulation, inflammation, tumor cell invasion, cellular differentiation, and apoptosis (1).
The squamous cell carcinoma antigen (SCCA) 1 (SERPINB3) and SCCA2 (SEPINB4) belong to the ovalbumin-serpin family, and these proteins are 91% identical at the amino acid level (4). Both genes locate at 18q21.3 very closely, generating a cluster of serpins together with plasminogen activator inhibitor type 2 and maspin, suggesting that either the SERPINB3 or the SEPINB4 gene could arise from the other by gene duplication (5). SCCA1 was originally purified from squamous cell carcinoma of uterine cervix (6), and it turned out later that SCCA1 and SCCA2 were co-expressed broadly in normal tissues: the epithelium of tongue, tonsil, esophagus, uterine cervix, vagina, and the conducting airways; Hassall's corpuscles of the thymus; and some areas of the skin (7). The biological functions of SCCA1 and SCCA2 still remain obscure. It has been reported that these proteins confer resistance against tumor necrosis factor-␣-or radiation-inducing apoptosis (8 -10). We have recently shown that expression of both SCCA1 and SCCA2 is up-regulated by two related Th2-type cytokines, IL-4 and IL-13, in bronchial epithelial cells and that SCCA expression is also augmented in bronchial lesions and in peripheral blood of bronchial asthma patients (11). These results shed light on the probable novel pathophysiological roles of SCCA1 and SCCA2.
Although SCCA1 and SCCA2 are very homologous, SCCA1 has unique properties as a serpin; SCCA1 inhibits cysteine proteinases such as cathepsin K, L, S, and papain, whereas SCCA2 inhibits serine proteinases such as cathepsin G and human mast cell chymase (HMC) (4,12,13). Although target proteinases for most serpins are the chymotrypsin family, very few serpins inhibit cysteine proteinases: for example, cytokine response modifier A (CrmA) produced by cowpox virus and proteinase inhibitor 9 (PI9, SERPINB9) are known to inhibit caspase proteins (cysteine proteinases) (14 -16). Such a proteinase inhibitor is defined as a cross-class inhibitor, and thus far, CrmA, PI9, and SCCA1 are all obvious cross-class serpins (1). The specificities of SCCA1 and SCCA2 are due to a difference in the RSL sequences because only 7 amino acid residues among 13 (54%) were identical in the RSL regions (P7 to P6Ј) of these proteins (17). SCCA2 employs the typical suicide substrate-like inhibitory mechanism of the serpins. Through this mechanism, SCCA2 and serine proteinases are linked by the covalent oxy-ester bond, which is SDS-resistant (4). In contrast, the inhibitory mechanism of cross-class serpins, including SCCA1 on cysteine proteinases, remains unclear. It has been reported that SCCA1 and cathepsin S form an SDSresistant complex, although its amount is very small (13). Other cross-class serpins, CrmA and PI9, do not form SDSresistant complexes with caspase proteins, although they do so with a serine proteinase, granzyme B (14,16,18,19). It has not yet been answered how cysteine proteinases lose their catalytic activities, although the thiol-ester bonds between the crossclass serpins and cysteine proteinases are unstable.
In this article, we analyze the inhibitory mechanism of SCCA1 on papain. To retain the native association between SCCA1 and papain, we used gel-filtration system analyses. It turned out that the association between SCCA1 and papain was non-covalent and that the interaction with SCCA1 severely impaired the proteinase activity of papain. These results indicated the unique inhibitory mechanism of SCCA1 among the serpin superfamily.
Generation of the SCCA1 and SCCA2 Protein-SERPINB3 and SER-PINB4 cDNA, prepared as reported before (9), were incorporated into pGEX(-KG)-4T (Amersham Biosciences). SCCA2 mutants were generated by PCR-based site-directed mutagenesis, using two complementary primers (Proligo Japan, Kyoto, Japan), designed to introduce a single codon mutation by substituting for the corresponding SCCA1 residue at the RSL. Standard PCR amplification was performed using the SCCA2 cDNA as a template and a mixture of primers. Resulting amplified fragments were treated with DpnI (New England Biolabs, Beverly, MA). Isolated plasmid DNAs were digested with StuI/XbaI and then ligated into the StuI/XbaI site of the pGEX-KG-SCCA2 plasmid. The RSL replacement mutants of SCCA1 and SCCA2 were similarly generated by digestion and ligation into their StuI/XbaI site.
Glutathione S transferase (GST)-fused SCCA1 and SCCA2 proteins were expressed in an Escherichia coli strain, BL21, and isolated by using glutathione-Sepharose 4B beads (Amersham Biosciences). Purity of the generated proteins was greater than 95% as estimated by SYPRO Ruby staining (Molecular Probes, Eugene, OR). Concentrations of the proteins were determined by Protein Assay (Bio-Rad).
Enzyme Assays-The substrates used for enzyme assays of papain and cathepsin L were benzoyl-Arg-7-amino-4-methylcoumarin (Bz-Arg-MCA) and benzyloxycarbonyl-Phe-Arg-methylcoumarin (Z-FR-MCA), respectively, and succinyl-Ala-Ala-Pro-Phe-methylcoumarin (Suc-AAPF-MCA) was used for cathepsin G and HMC, all purchased from Peptide Institute Inc. The molar concentrations of papain were determined by active site titration with E-64. The indicated concentration of each enzyme was incubated with indicated concentrations of GST-fused SCCA1 or SCCA2 for 30 min at 25°C in activity-measuring buffer. The buffer was composed of 50 mM sodium acetate (pH 5.5), 4 mM dithiothreitol, 1 mM EDTA, and 1% bovine serum albumin for papain, or 50 mM sodium acetate (pH 5.5), 4 mM dithiothreitol, 1 mM EDTA, and 0.001% bovine serum albumin for cathepsin L, or phosphate-buffered saline containing 0.001% bovine serum albumin for cathepsin G and HMC as described previously (4). Upon addition of the substrate to the reaction mixture, the residual enzyme activity was measured by continuous monitoring using excitation and emission wavelengths of 380 and 460 nm, respectively.
Separation of the Incubated SCCA1 and Papain-After the indicated concentrations of papain and 54 M GST-fused SCCA1 were incubated in the acetate reaction buffer for 30 min at 25°C, the reaction mixture was applied to a high pressure-liquid chromatography system equipped with ProteinPak 300SW (Waters, Milford, MA). Then, the subjected samples were eluted with the acetate buffer (50 mM sodium acetate; pH 5.5), and each fraction was subjected to SDS-PAGE or the enzyme assay. The amounts of the proteins on the gels were estimated by protein assay.
Western Blotting-The samples were applied to SDS-PAGE and then electrophoretically transferred to polyvinylidene difluoride membranes (Amersham Biosciences). The membranes were blotted by either antipapain polyclonal antibody (Ab) (Rockland, Gilbertsville, PA) or anti-GST monoclonal Ab (Upstate Biotechnology, Lake Placid, NY). The proteins were visualized by enhanced chemiluminescence (ECL, Amersham Biosciences).
Amino Acid Sequencing Analysis-The reaction mixture was applied to SDS-PAGE followed by staining with Coomassie Blue R250 in 50% methanol, and then transferred to polyvinylidene difluoride membranes (Bio-Rad). Amino acid sequence analysis of the transferred peptides was performed using an Applied Biosystems 477A/120A protein sequencer (ABI Applied Biosystems, Foster City, CA).

RESULTS
Expression and Purification of Functional SCCA1 and SCCA2-To perform functional analyses of SCCA1 and SCCA2, we expressed and purified recombinant proteins of GST-fused SCCA1 and SCCA2. Purity of these two proteins was more than 95%, as estimated by SYPRO Ruby staining (Fig. 1A). We first confirmed proteinase-inhibitory effects of these two proteins. We examined the inhibitory effects of SCCA1 and SCCA2 on papain, cathepsin L, cathepsin G, and HMC. The K m and k cat values of papain used in the experiments were estimated as 23.8 Ϯ 2.29 M and 7.95 Ϯ 0.218 s Ϫ1 (mean Ϯ S.D., n ϭ 3), respectively, which were compatible with those reported in the literature (20). SCCA1 inhibited cysteine proteinase activities of papain and cathepsin L but not the serine proteinase activities of cathepsin G and HMC, whereas SCCA2 showed the opposite effects (Fig. 1B), as reported previously (4,13). SCCA1 inhibited the catalytic activity of papain in a dose-dependent manner, and the stoichiometry of inhibition value was estimated as 4.6 Ϯ 0.1 at 10 nM SCCA1 (n ϭ 3, Fig. 1B).
It has been demonstrated that SCCA2 and cathepsin G form an SDS-resistant complex by a covalent bond (4). We next analyzed how SCCA1 generated a complex with papain similar to that done by SCCA2 with cathepsin G. SCCA2 formed an SDS-resistant complex with cathepsin G but not with papain, whereas SCCA1 did not show any formation of a complex with either papain or cathepsin G (Fig. 1C). These results raised a possibility that SCCA1 performs its proteinase-inhibitory activity without generation of a covalent bond.
Non-covalent Bond of SCCA1 and Papain-Although an SDS-resistant complex between SCCA1 and papain was not detected, it would be possible that the covalent bond between these two proteins is unstable to SDS, like the complex between chymotrypsin and ␣ 2 -antiplasmin (21). To retain the native association between SCCA1 and papain, we employed a gelfiltration system. We subjected three different samples, whose I 0 /E 0 ratios were 4.2, 5, and 6, for this analysis. It was con-firmed that proteinase activities of papain were completely inhibited in the samples of I 0 /E 0 ϭ 6 and 5, but not 4.2 (data not shown). When these samples were applied to the gel-filtration column, most SCCA1 and papain were eluted according to their molecular weights (Fig. 2, A-C). However, an additional peak with a faster retention time was detected in the samples of I 0 /E 0 ϭ 6 and 5, but not 4.2, on the chart, although it was not detected when SCCA1 alone was applied ( Fig. 2D and data not shown). Together with this peak on the chart, a band of low molecular weight the same as papain appeared on the SDS-PAGE gel in the samples of I 0 /E 0 ϭ 6 and 5, but not 4.2, as discussed later (Fig. 2, A-C). These results demonstrated that SCCA1 inhibited the catalytic activity of papain without forming a covalent bond.
Determination of the Cleavage Site in SCCA1 and Loss of Its Inhibitory Effect-It has been shown that the inhibitory serpins exert their activities by the suicide substrate-like inhibitory mechanism, in which the serpins interact with their target proteinases at their RSLs, which are then cleaved (2). We next investigated whether SCCA1 also used the inhibitory mechanism against papain as other serpins do. When papain-treated SCCA1 was eluted from the gel-filtration column, doublet bands appeared with molecular sizes of 62 and 55 kDa (Fig. 2, A-C). Both bands were recognized by anti-GST Ab, indicating that the upper and lower bands corresponded to intact and C-terminal truncated SCCA1, respectively (data not shown). We then determined the cleavage site of SCCA1 by papain, using amino acid sequencing and MALDI-TOF mass spectrometry. When we applied the visible band of Ͻ10 kDa to the amino acid sequencing, two sequences derived from SCCA1 were read out; the major one started at Ser-354, and the minor one started at Phe-352 (Fig. 3A). The measurement of papaincleaved peptides in SCCA1 by MALDI-TOF mass spectrometry also revealed the existence of the peptide corresponding from Ser-354 to the C terminus and the peptide from Phe-352 to the C terminus, which was less than the peptide starting at Ser-354 (Fig. 3B). These results clearly demonstrated that the bond between Gly-353 and Ser-354 was the main cleavage site of SCCA1 and that the one between Gly-351 and Phe-352 was minor, as reported previously (13).
The intensities of the upper bands corresponding to intact SCCA1 became weaker as the I 0 /E 0 ratio dropped, and those of the lower bands corresponding to truncated SCCA1 and of more degraded bands showed the opposite tendency (Fig. 2, A-C). If SCCA1 used the suicide mechanism, the inhibitory actions in these solutions would be parallel to the amount of intact SCCA1, irrespective of the amount of truncated SCCA1. To explore this possibility, after papain-treated SCCA1 was eluted by the gel-filtration column, we analyzed its inhibitory activity on freshly prepared papain. Consequently, the inhibitory activity of papain-treated SCCA1 increased parallel to the amount of intact SCCA (Fig. 4), which clearly supported the finding that the inhibitory manner of SCCA1 fit the suicide mechanism.
Association of Truncated SCCA1 and Papain-Although SCCA1 and papain did not form a covalent complex (Fig. 2,  A-C), it was possible that the truncated SCCA1 would associate with papain non-covalently in the solution. Actually, an additional peak with a faster retention time than monomers or dimers of SCCA1 was detected by gel-filtration analysis, and a band of the same low molecular weight as papain appeared on the SDS-PAGE gel in the fractions corresponding to this peak (Fig. 2). It was reasonable to think that this fraction would correspond to co-migration of SCCA1-associated papain and truncated SCCA1. To explore this possibility, we first examined Western blotting showed that the fractions corresponding to the peak with a faster retention time indicated the existence of papain (Fig. 5A). Both the peak on the chart and the band corresponding to papain in the Western blotting disappeared in the presence of an irreversible cysteine proteinase inhibitor, E-64 (Fig. 5, B and C). However, surprisingly, the molecular size of the complex was estimated as about 1100 kDa, judging from the retention time of the gel-filtration column (Fig. 5D), which is far larger than ϳ90 kDa, the expected molecular size of the 1:1 complex of papain and GST-fused SCCA1. These results suggested that although SCCA1 and papain formed a non-covalent complex dependent on the proteinase activity of papain, this complex was a huge molecule composed of several oligomers of SCCA1 and papain.
Although it was confirmed that papain and SCCA1 generated the relatively firmly associated complex of very large molecular weight, it was still possible that SCCA1-treated papain and truncated SCCA1 would form a transient, more loosely associated complex at 1:1 stoichiometry. To confirm the existence of such a complex, we next chemically cross-linked these two molecules by BS 3 , which elicited the appearance of a 97-kDa band corresponding to a 1:1 complex composed of papain and truncated SCCA1 in the absence of E-64 (Fig. 6). This complex was recognized by both anti-papain and anti-GST Abs. A much fainter band with a bigger molecular size correspond-ing to the complex composed of papain and intact SCCA1 appeared in the presence of E-64. These results clearly suggested that SCCA1-treated papain and truncated SCCA1 formed a 1:1 complex dependent on the proteinase activity of papain in the reaction solution.
Preference of Amino Acids in RSL of SCCA1 for Papain-It is reasonable to assume that the distinct properties of SCCA1 and SCCA2 regarding the inhibitory effects on papain are due to the differences of their RSL sequences because only 7 amino acid residues among 13 (54%) are identical in the RSL regions (P7 to P6Ј) of these proteins (17). Actually, swapping the RSL of SCCA1 for that of SCCA2, or vice versa, revealed that the inhibitory effect on papain was dependent on the RSL of SCCA1 (SCCA1 RSL2, SCCA2 RSL1: Table I). We then exchanged each amino acid specific for the RSL of SCCA2 with that corresponding to SCCA1 and analyzed its inhibitory effect on papain (Table I). When Glu-353 was replaced with Gly, the mutated type showed the inhibitory effect at the same level as native SCCA1 (SCCA2 mut3). However, when Val-351, Val-352, or Leu-354 was replaced with Gly, Phe, or Ser, respectively, none of the mutated types exhibited the inhibitory effect (SCCA2 mut1, SCCA2 mut2, SCCA2 mut4). Furthermore, switching of both Ser-356 and Pro-357 with Pro and Thr, respectively, also did not recover the inhibitory activity (SCCA2 mut5). These results demonstrated that Gly-353 was critical for SCCA1 to exert its inhibitory effect on papain.
Irreversible Inhibition of SCCA1-treated Papain-We next examined whether SCCA1 treatment affects the catalytic ac- FIG. 4. The catalytic activity of truncated SCCA1. In A, a mixture of papain and SCCA1 prepared as described in the legend for Fig.  2 was applied to the gel-filtration column. Ten nM eluted SCCA1 (fraction 5 at I 0 /E 0 ϭ 5 and 6 and fraction 4 at I 0 /E 0 ϭ 4.2) was incubated with freshly prepared papain at the indicated I 0 /E 0 ratio. The residual enzyme activity at the indicated I 0 /E 0 ratio is depicted. In B, the relationship between the amount of intact SCCA1 of each sample and the intensity of the inhibitory activity is plotted. The amount of intact SCCA1 and the intensity of the inhibitory activity are estimated as the ratios of those of non-treated SCCA1 (squares). tivity of papain. To address this question, we analyzed the proteinase activity of SCCA1-treated papain eluted by the gelfiltration column. We found that the proteinase activity of SCCA1-treated papain was severely impaired as compared with non-treated papain but was still present (15 Ϯ 4.2% at I/E ϭ 6, 17 Ϯ 6.1% at I/E ϭ 5, 20 Ϯ 9.0% at I/E ϭ 4.2, n ϭ 3: Fig.  7A). The activity of E-64-treated papain was completely inhibited (2.7 Ϯ 2.4%, n ϭ 3). The K m value of SCCA1-treated papain was more than that of non-treated papain (36.2 Ϯ 1.35 M, n ϭ 3, p ϭ 0.001), and the k cat value of SCCA1-treated papain was less than that of non-treated papain (5.14 Ϯ 0.202 s Ϫ1 , n ϭ 3, p ϭ 0.00008), which confirmed the impairment of the catalytic activity.
It is possible that, due to unexpected modification of the active cysteine residue of papain, SCCA1 treatment caused a significant decrease in its proteinase activity. To exclude this possibility, we compared modification of the active cysteine residue of SCCA1-treated or non-treated papain by biotin-conjugated maleimide. When non-treated papain was incubated with 77 M maleimide for 2 h, its modification was not saturated (data not shown). Under this condition, modification of SCCA1-treated papain by biotin-conjugated maleimide was almost at the same level as non-treated papain (Fig. 7B). These results suggested that the active cysteine residue of papain was intact even after SCCA1 treatment. Taken together, these results suggest that SCCA1 treatment probably induced irreversible conformational change of papain, which severely impaired its proteinase activity.
Although SCCA1-treated papain still sustained its catalytic activity with the compromised level, all activity was abolished in the reactive solution (Figs. 1B and 7A). This may be due to the suicide substrate-like inhibition on the residual activity of papain by intact SCCA1. If this were the case, the catalytic activity of papain would be completely inhibited, whereas intact SCCA1 would remain. To explore this possibility, we analyzed the time-dependent profile of the proteinase activity of papain and the digestion pattern of SCCA1 (Fig. 8). Incubation gel blotted by anti-papain Ab is depicted. B, samples prepared as described in the legend for Fig. 2 in the presence of E-64 were applied to the gel-filtration column. The gel stained with SYPRO Ruby is depicted. The arrows indicate intact SCCA1 (*) and papain (**). In C, the chart displaying the intensity of the absorbance at 280 nm is depicted. In D, retention time and molecular size of each protein are plotted. The closed circle (*) indicates the retention time and the estimated molecular weight of the oligomers of SCCA1 and papain. BSA, bovine serum albumin.
FIG . 5. Co-migration of the SCCA1-treated papain and truncated SCCA1. A, samples prepared as described in the legend for Fig.  2 in the absence of E-64 were applied to the gel-filtration column. The with SCCA1 led very quickly to complete inhibition of papain activity, lasting up to 2 h; however, the proteinase activity of papain started to recover at 4 h and then reached about 21% of the original activity at the same level as modified papain. In concert with the recovery of proteinase activity, the intact (62 kDa) and truncated (55 kDa) types of SCCA1 were completely degraded. These results indicated that if intact SCCA1 was completely cleaved by papain, the inhibitory effect of SCCA1 by the suicide substrate-like inhibition was no longer retained, allowing papain to recover its proteinase activity up to the lowest initial level.

DISCUSSION
In this article, we examined the inhibitory mechanism of a cross-class serpin, SCCA1, on papain. We propose the inhibition mechanism of SCCA1 as described in Fig. 9, as compared with the standard serpins (1,2). In the standard serpins, a proteinase (E) and a serpin (I) initially form a non-covalent Michaelis-like complex (EI) followed by an acyl-enzyme intermediate (EI # ) linked by an oxy-ester bond. The inhibitory mechanism of SCCA1 on papain would share this pathway. In contrast, the acyl-enzyme intermediate (EI # ) is processed to either a covalent complex (EI ϩ ), or a cleaved serpin (I*) and a free proteinase (E) in the standard serpins. However, in the case of SCCA1, the acyl-enzyme intermediate (EI # ) linked by a thiol-ester bond would easily hydrolyze into the non-covalent complex (E*I*) composed of cleaved SCCA1 (I*) and modified papain (E*). Modified papain (E*) loses most of its proteinase activity as compared with intact papain. Furthermore, a part of the non-covalent complex (E*I*) forms a more firmly bound complex composed of oligomers (E* m I* n ).
In this model, we first suggested that although SCCA1 used the suicide substrate-like mechanism as other serpins do, SCCA1 did not generate a covalent complex, in contrast to other serpins. The notion that SCCA1 used the suicide substrate-like mechanism was confirmed by the following results: 1) the RSL of SCCA1 was cleaved at the predicted site by papain (Fig. 3). 2) The inhibitory activity of SCCA1 on papain was dependent on intact, but not truncated, SCCA1 (Fig. 4). 3) Although intact SCCA1 existed, papain with a compromised level of proteinase activity lost its activity completely (Fig. 8). These results proved that the exposed RSL of SCCA1 was recognized by papain and that cleaved SCCA1 (I*) was inactive, which was in line with the typical suicide substrate-like mechanism. In contrast, a unique property of SCCA1 as a serpin has been also revealed. Most SCCA1 and papain were eluted according to their molecular weights by the gel-filtration column (Fig. 2), and truncated SCCA1 and papain formed a 1:1 complex dependent on the proteinase activity of papain (Fig. 6). These results indicated that the acyl-enzyme intermediate (EI # ) would be processed to formation of a non-covalent complex (E*I*) but not a covalent complex (EI ϩ ). In the cleaved form of standard serpin, the insertion of RSL causes drastic conformational changes of the serpin and the proteinase so that the histidine residue of the catalytic triad was too far from the serine residue to let the ester bond hydrolyze (3). In the case of SCCA1, the partners of the catalytic center might still be close to the cysteine residue, making it possible for the ester bond to hydrolyze. It has been reported that other cross-class serpins, CrmA and PI9, also do not form SDS-resistant complexes with caspase proteins (14,16) as SCCA1 was unable to do with papain (Fig. 1D). Furthermore, we have recently observed that SCCA2 inhibited cysteine proteinase activity of a major mite allergen, Der p 1 and Der f 1, without forming a covalent complex. 2 Taking these results together, the suicide substratelike mechanism, without forming a covalent complex, may be a common property of cross-class serpins. Thus far, it remains unresolved how stable the non-covalent complex (E*I*) would be. If this complex is not transient, but stable, formation of this complex would also contribute to the inhibition of free modified papain (E*) as the inhibitors of apoptosis protein family do for caspase proteins by partially substrate-like inhibition or as thrombin inhibitors do for thrombin by exosite binding inhibition (22,23).
In this model, we next suggested that irreversible impairment of the catalytic activity of papain (E*) by SCCA1 contributed to the inhibitory mechanism of SCCA1 on papain, in addition to the suicide substrate-like mechanism (Fig. 9). This mechanism was verified by results that eluted papain showed compromised inhibitory activity (Fig. 7A) and that a longer exposure to SCCA1 allowed papain to recover its proteinase activity, but only up to the lowest initial level (Fig. 8). The contribution of the irreversible impairment to the whole inhibitory mechanism was significantly high (ϳ85%: Fig. 7A). However, the result that incorporation of a thiol-residue modifying reagent, maleimide, into the SCCA1-treated papain was invariable with non-treated papain (Fig. 7B) suggested that SCCA1treated papain still kept its conformation sufficiently for maleimide to access the catalytic cysteine residue. The drastic conformational change of a proteinase by engagement with its inhibitor is reported with the crystallographic structure between trypsin and ␣1-antitrypsin (3). The interaction of these two molecules causes a 37% loss of structure in trypsin; in particular, plucking of the ester-linked catalytic active center, Ser-195, from its catalytic partners prevents hydrolysis of the covalent bond, which sustains the complex. The result SCCA1 caused irreversible impairment of the catalytic activity of papain indicated that SCCA1 also disrupted the papain structure, as did the standard serpins. However, we also showed that incorporation of maleimide into papain was not affected, in addition to the result that the thiol-ester bond between SCCA1 and papain was unstable. These two results suggested that the distortion of papain induced by SCCA1 was not so complete (as is the case with standard serpins) that hydrolysis of the thiolester bond in cooperation with catalytic partners might still be possible.
Unexpectedly, in addition to a 1:1 complex, a part of truncated SCCA1 and SCCA1-treated papain formed a large complex composed of oligomers (E* m I* n ) (Fig. 5). This complex was predicted to contain more than 10 molecules of SCCA1 and papain, although its precise components were unclear. As the formation of this complex disappeared with addition of E-64, intact SCCA1 (I) and papain (E) could not generate this complex (Fig. 5, B and C). It was speculated that truncated SCCA1 and SCCA1-treated papain changed their conformations so that the aberrant association between these two molecules might be induced.
We demonstrated that Gly-353 in the RSL was critical for the inhibitory effects of SCCA1 and that residue was sufficient for SCCA2 to achieve the same inhibitory effect as SCCA1 (Table  I). The significance of the P1 residue of SCCA molecules is variable among the target proteinases. Glu-353 in SCCA2 was important for the inhibitory effect on cathepsin G, whereas Gly-353 in SCCA1 was not required for the inhibitory effect on cathepsin S (17). When SCCA1 and papain interact, Gly-353 locates deep at the cleft of papain and forms a thiol-ester bond with Cys-25 of papain. It may be that the ionic strength of Glu interrupts the interaction of the RSL and papain, whereas Gly at the P1 site stabilizes the interaction.
As it is widely known that proteinases play important roles in the homeostasis of the body, proteinase inhibitors have great potential as novel therapeutic reagents. Cathepsin S degrades the invariant chain, important for antigen presentation of major histocompatibility class II molecules so that cathepsin Sdeficient mice show diminished susceptibility to collagen-induced arthritis (24,25). Cathepsin K is involved in bone remodeling by degrading bone matrix proteins such as type I and type II collagen, and the loss-of-function mutations in the cathepsin K gene evoke pycnodysostosis, characterized by osteosclerosis and short stature (26 -28). Cathepsin L is reported to have a critical role for degrading invariant chain as well as cathepsin L (29); however, the analyses of cathepsin L-deficient mice show that unexpectedly, cathepsin L is important for epidermal homeostasis and hair follicle morphogenesis (30). Because SCCA1 has an ability to inhibit all of these proteinases, the compounds that mimic the inhibitory effects of SCCA1 have the potential to be applied to autoimmune diseases, osteoporosis, and epidermal disorder diseases (30,31). Furthermore, we have recently demonstrated that expression of SCCA1 is related to bronchial asthma, although it has remained unresolved whether it acts as a worsening or preventing factor (11). It would be of use for developing a therapeutic reagent against these diseases to clarify the precise mechanism of the inhibitory mechanism of SCCA1. In conclusion, we show in this article that SCCA1 inhibits the catalytic activity of papain in two ways, contributing the suicide substrate-like mechanism without formation of a covalent complex and causing irreversible impairment of papain.