Differential Effect toward Inhibition of Papain and Cathepsin C by Recombinant Human Salivary Cystatin SN and Its Variants Produced by a Baculovirus System

Archives of Biochemistry and Biophysics Vol. 380, No. 1, August 1, pp. 133–140, 2000 doi:10.1006/abbi.2000.1909, available online at http://www.idealibrary.com on

Differential Effect toward Inhibition of Papain and Cathepsin C by Recombinant Human Salivary Cystatin SN and Its Variants Produced by a Baculovirus System Ching-Chung Tseng,* ,1 Ching-Ping Tseng,† Michael J. Levine,‡ and Libuse A. Bobek‡ *Divisions of Basic Sciences and of Restorative and Prosthodontic Sciences, Room 1027 S, College of Dentistry, New York University, 345 East 24th Street, New York, New York 10010-4086; †School of Medical Technology, Chang Gung University, Tao-Yuan, Taiwan, ROC; and ‡Department of Oral Biology and Research Center in Oral Biology, State University of New York at Buffalo, Buffalo, New York 14214

Received February 17, 2000, and in revised form May 3, 2000

Human salivary cystatin SN (CsnSN) is a member of the cystatin superfamily of cysteine proteinase inhibitors. In this study we used a baculovirus expression system to produce a full-length unaltered CsnSN and its variants. The variants were constructed with the changes in the three predicted proteinase-binding regions: the N-terminus (variant N 12–13, G12A–G13A), ␤-hairpin loop I (variant L56 –58, Q56G–T57G–V58G) and ␤-hairpin loop II (variant L 106 –107, P106G–W107G). The secreted CsnSNs were purified using sequential spiral cartridge ultrafiltration and DE-52 radial flow chromatography. The purified proteins were examined for papainand cathepsin C-inhibition. The wild-type CsnSN, and variants N 12–13 and L 106 –107 bound tightly to papain (K i < 10 pM), whereas mutation in the loop I reduced binding affinity 5700-fold (K i ⴝ 57 nM). On the other hand, the wild-type CsnSN bound to cathepsin C less tightly (K i ⴝ 100 nM). The mutation in the N-terminus or loop I reduced binding affinity by 16 (K i ⴝ 1.6 ␮M)- and 19-fold (K i ⴝ 1.9 ␮M), respectively, while mutation in loop II resulted in an ineffective cathepsin C inhibitor (K i ⴝ 14 ␮M). Collectively, these results suggest that the N-terminal G12–G13 residues of CsnSN are not essential for papain inhibition but play a role in cathepsin C inhibition; residues Q56 –T57–V58 in the loop I are essential for both papain and cathepsin C inhibitions, and residues P106 – W107 in the loop II are not important for papain inhibition but essential for cathepsin C inhibition. These results demonstrated that CsnSN variants have different effects toward different cysteine proteinases. © 2000 Academic Press

1 To whom correspondence should be addressed. Fax: (212) 9954087. E-mail: [email protected].

0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

Key Words: mutagenesis; radial flow chromatography; circular dichroism; structure–function relationship; cysteine proteinase inhibitor.

Cysteine proteinases (CysPs) 2 comprise a family of proteolytic enzymes that use a reactive cysteine residue at their catalytic site. These enzymes include papain, cathepsins, ficin, calpain, and gingipain. Among these enzymes, papain is the best-characterized member of the CysP family both structurally and mechanistically (1). CysPs have been a target for drug development because they participate in a wide range of functions in health and disease. They are involved in intracellular protein degradation (2), proenzyme and prohormone processing (3), and the pathogenicity of malignant cells (4) and microorganisms (5). In dentistry, CysPs are considered as possible pathogenic agents for adult periodontitis because of their proteolytic activity against oral tissues (6, 7). Cystatins (Csns) are naturally occurring inhibitors of CysPs. They inhibit CysPs by forming tight but reversible complexes with their target enzymes. Several studies have demonstrated that various Csns and their synthetic derivatives are capable of blocking bacterial or viral replication (8, 9). In addition, Csns were 2

Abbreviations used: CysP, cysteine proteinase; Csn, cystatin; CsnC, cystatin C; rCsnSN, recombinant CsnSN; BEVS, baculovirus expression vector system; AcNPV, Autographa californica nuclear polyhydrosis virus; rBV, recombinant baculovirus; p.i., postinfection; PTH, phenylthiohydantoin; CD, circular dichorism; Bz-Arg-AMC, N ␣-benzoyl-L-arginine-7-amido-4-methylcoumarin; DTT, dithiothreitol; H-Gly-Phe-AMC, glycine-phenylalanine-7-amido-4-methylcoumarin; NBS, N-bromosuccinimide; HNB, 2-hydroxy-5-nitrobenzyl. 133

134

TSENG ET AL.

shown to inhibit the chemotactic response and the phagocytosis-associated respiratory burst, indicating that they may modulate the immune response in mucosa tissues (10). Human saliva contains at least eight Csn isoforms (11). These include one neutral cystatin SN, three isoforms of moderately acidic cystatin SA, three or four isoforms of more acidic cystatin S, and a more recently identified cystatin D. Salivary Csns are multifunctional. They may modulate the mineralization process at the saliva– enamel interface by inhibiting transformation of dicalcium phosphate dihydrate (12), influencing hydroxyapatite growth kinetics (13), and participating in the formation of acquired enamel pellicle (14). In addition, they may play a role in protection of periodontal tissues (15) and inhibition of CysPs of bacterial (16) and viral (17, 18) origin. Therefore, salivary Csns are considered one of the important nonimmune defense factors of saliva and might be developed into therapeutic agents. Salivary Csns share about 54% amino acid sequence homology with other members of the Csn family, such as human cystatin C (CsnC, also known as ␥ trace protein) and chicken egg white Csn. The shared sequences include the N-terminal G9 residue (chicken Csn numbering), and the two ␤-hairpin loop structures (loop I and loop II). Studies based on X-ray crystallography and computer docking modeling suggested that these regions form a wedge-shaped edge complementary to the CysP active site (19). Since Csns from different species share sequence homology and all exhibit CysP-inhibitory activity, it is conceivable that the similarities in biological function are due, in part, to the shared sequence homologies. To explore the role of these regions of salivary Csns, we have previously produced recombinant CsnS as well as CsnSN and its variants in Escherichia coli pGEX-2T expression system (20, 21). In that system, Csns are produced as fusion proteins with glutathione S-transferase and in an insoluble form. Isolation of the rCsns, thus, involved solubilization of the inclusion bodies with 8.0 M urea and refolding of the proteins by dialysis. Nevertheless, the purified recombinant CsnSN (rCsnSN) exhibited bioactivity and secondary structure very similar to that of the natural CsnSN. One of the variants, ⌬12–16, has been crystallized (22). Another variant, ⌬56 – 60, however, could not be solubilized, most likely due to its improper refolding. To improve production of the CsnSN protein quantitatively and qualitatively, we describe the generation and purification of the full-length CsnSN and three CsnSN variant recombinant proteins using the baculovirus expression vector system (BEVS) (23). The activities of these recombinant proteins have also been characterized by analyzing the inhibitory effects toward the proteases papain and cathepsin C. In sum-

mary, we have generated useful tools for future characterization of CsnSN that should assist us in the understanding of the functional roles of various domains or residues involved in the CsnSN–CysP interaction. MATERIALS AND METHODS Construction of recombinant shuttle vector (pVL-SN) carrying fulllength CsnSN. The EcoRI–NotI cDNA fragment encoding the leader and secreted peptide of CsnSN was excised from a previously described construct in the ␭gt11 Sfi–Not vector (24). The cDNA was ligated to vector pVL1393 to create the recombinant plasmid pVLSN. The presence of the correct insert was confirmed by restriction enzyme analysis and by DNA sequencing. Construction of recombinant shuttle vectors carrying CsnSN mutations. Polymerase chain reaction (PCR)-based mutagenesis strategy was used to generate CsnSN variants with specific amino acid replacements in CsnSN highly conserved regions predicted to be essential for CysP inhibition. These include the N-terminal variant N 12–13 (G12A–G13A), loop I variant L 56 –58 (Q56G–T57G–V58G), and loop II variant L 106 –107 (P106G–W107G). The N 12–13 PCR employed a BanII/N 12–13 mutagenic primer (5⬘-CTG GCC TGG AGC CCC AAG GAG GAG GAT AGG ATA ATC CCG GCT GCC ATC TAT AAC GC-3⬘) designed to anneal to the CsnSN sequence at the BanII site (underlined) 5⬘ to the mutation site, and a polyhedrin reverse primer (5⬘-GTC CAA GTT TCC CTG-3⬘) designed to anneal to the vector sequence beyond the 3⬘ end of the cDNA. The L 56 –58 PCR employed a L 56 –58/Asp 700 mutagenic primer (5⬘-TAC GTC GAA GAA GTA ATT CAC CCC CCC ACC GCC CCC TTG CCT GGC-3⬘) designed to anneal to the CsnSN sequence at the Asp 700 site (underlined) 3⬘ to the mutation site, and a baculovirus forward primer (5⬘-TTT ACT GTT TTC GTA ACA GTT TTG-3⬘) designed to anneal to the vector sequences beyond the 5⬘ end of the cDNA. The L 106 –107 PCR employed a BglII/L 106 –107 mutagenic primer (5⬘-TCT TTC GAG ATC TAC GAA GTT GGC GGG GAG AAC AGA AG-3⬘) designed to anneal to the CsnSN sequence at the BglII site (underlined) 5⬘ to the mutation site, and a polyhedrin reverse primer (same as above). All PCR products were end-trimmed with respective restriction enzymes. In addition, corresponding supplementary CsnSN sequences (pVL-SN sequence minus the mutagenic PCR region and vector sequence) were isolated from pVL-SN and concomitantly ligated with mutagenic PCR products into either pVL1392 (for N 12–13) or pVL1393 (for L 56 –58 and L 106 –107) to create intact mutant plasmids. All clones were verified by cycle sequencing reactions using the CircumVent Phototope Kit (New England Biolabs, Beverly, MA). Cell lines and viruses. The frozen Sf9 cell line was initially grown at 28°C in a monolayer culture in Sf-900 II serum-free medium (Life Technologies, Grand Island, NY). For a monolayer culture, cells were seeded at a density of 5 ⫻ 10 5 cells/mL in a T-25 culture flask and incubated at 28°C in accordance with established procedures (25). For a shaker culture, cells were seeded at a density of 3 ⫻ 10 5 cells/mL in 100- or 500-mL Erlenmeyer flasks, and the culture was maintained at 28°C on an orbital shaker apparatus at 130 rpm. Virus stocks were prepared by infection with 1 pfu/cell. Infections for production of rCsnSN were performed using 10 pfu/cell. Baculovirus expression of rCsnSN and variants in Sf9 cells. Transfer of the CsnSN or variant cDNAs into the baculovirus Autographa californica nuclear polyhedrosis virus (AcNPV) genome was accomplished by cationic liposome-mediated transfection using Bsu36I-digested BacPAK6 viral DNA (Clontech Laboratories, Palo Alto, CA). The recombinant baculoviruses (rBVs) harboring the native or variant CsnSN cDNAs were plaque-purified and were amplified twice in Sf9 cells to generate high titer viral stocks of at least 10 8 pfu/ml. For protein expression, log-phase Sf9 cells were infected with high-titer rBVs at 10 pfu/cell. The kinetics of rCsnSN expression was

THE ACTIVITY OF BACULOVIRUS-EXPRESSED HUMAN SALIVARY CYSTATIN SN determined by collecting 22-␮l aliquots of infection supernatant at various time points and analyzed by immunoblotting using native CsnSN as a control (data not shown). Protein bands on developed blots were quantified by an LKB 2202 Ultrascan Laser Densitometer (Pharmacia LKB, Piscataway, NJ) (data not shown). For subsequent large-scale preparation, culture media was harvested at the time of optimal rCsnSN production as determined by the time course studies. Purification of rCsnSN and variants from culture media. The rCsnSN and variants were purified from Sf9 culture media by sequential ultrafiltration and radial flow chromatography. All procedures were carried out at 4°C except where indicated otherwise. Prior to chromatography, the 96-h postinfection (p.i.) medium (1 L) was centrifuged at 2000g for 15 min to remove cell debris, and then at 96000g for 1 h to remove viral particles. The supernatant was then adjusted to pH 8.5 with NaOH to precipitate some of the medium components. The precipitate was removed by centrifugation at 10000g for 15 min and filtering through a 0.2-␮m filter. The 14-kDa rCsnSN was ultrafiltrated through a 30-kDa spiral cartridge. The resulting filtrate was concentrated with a 3-kDa spiral cartridge, diafiltrated with 10 mM Tris–HCl, pH 8.0 (equilibration buffer), and applied to a Superflo 100 radial flow column (Sepragen, San Leandro, CA) of DEAE-52 cellulose at room temperature (Fig. 1A). The column was washed with 1.2 L equilibration buffer at 24 mL/min. Bound materials were eluted using a linear gradient consisting of 500 mL each of equilibration buffer and 100 mM Tris–HCl, pH 8.0, followed by washing with 500 mL of 100 mM Tris–HCl, pH 8.0, and a linear salt gradient consisting of 200 mL of 100 mM Tris–HCl, pH 8.0, and 200 mL of 1 M NaCl in 100 mM Tris–HCl, pH 8.0. Fractions of 12 mL were collected and analyzed. Based on SDS–PAGE/immunoblot analysis using rabbit anti-CsnSN antiserum (26) (Fig. 1B), fractions containing pure rCsnSN were pooled, dialyzed against distilled water, and lyophilized. SDS–PAGE, anionic PAGE, and immunoblotting. For SDS– PAGE, the discontinuous gel system of Laemmli (27) with a 4% stacking gel at pH 6.8 and 12% separating gel at pH 8.8 was used. For anionic PAGE, the procedure of Ornstein (28) was used. Proteins were visualized by staining with a 0.1% solution of Coomassie brilliant blue R250 and/or by silver staining using a silver staining kit (Bio-Rad, Richmond, CA). For immunoblotting, proteins separated on the SDS–PAGE were transferred to the Immobilon-P membrane. The membranes were blocked in 5% BSA in 10 mM Tris–HCl, pH 8.0, and 150 mM NaCl (TBS). Rabbit anti-Csn antibody (1:15,000) in 3% BSA (26) was used as a primary antibody and alkaline phosphatase-conjugated goat anti-rabbit IgG (Fc) as a secondary antibody. Antigens were revealed with alkaline phosphatase color development solution containing 5-bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium. Chemical analyses of CsnSN. The human salivary CsnSN was purified from submandibular–sublingual saliva as described (29). Both the rCsnSN and native salivary CsnSN were analyzed as follows. For amino acid analysis, samples were hydrolyzed with 6 N HCl for 24 h at 110°C in evacuated sealed tubes. Hydrolysates were analyzed by a Beckman 6300 amino acid analyzer using ␣-amino-␤guanidino propionic acid (Pierce, Rockford, IL) as an internal standard. N-terminal microsequencing was performed on an automated Model 471A protein sequencer (Applied Biosystems, Foster City, CA). Phenylthiohydantoin (PTH)-amino acids were monitored at 270 nm with an in-line variable-wavelength detector interfaced to a Hewlett-Parkard HP3394A integrator. Circular dichroism (CD) spectroscopy. Samples were constituted into 10 mM stock solutions with 20 mM sodium phosphate, pH 7.1. CD spectra were recorded from 185 to 250 nm in triplicate on a Jasco J600 spectropolarimeter (Japan Spectroscopic Co., Tokyo, Japan) at room temperature. Protein concentrations used for the calculation of

135

FIG. 1. Chromatography of rCsnSN on a Superflo 100 radial flow column of DEAE-52 cellulose. (A) Column elution profile. A, gradient of 5 to 100 mM Tris–HCl, pH 8.0; B, 100 mM Tris–HCl, pH 8.0; C, gradient of 100 mM Tris–HCl, pH 8.0, with 0 to 1 M NaCl; D, 100 mM Tris–HCl, pH 8.0, with 1 M NaCl. (B) Analysis of column fractions by immunoblotting. Aliquots of 20 –300 ␮l (each calculated to an absorbance of 0.01 at 280 nm) were lyophilized and subjected to 12% SDS–PAGE/immunoblotting using anti-salivary cystatin antiserum. Lanes 1 to 13, fractions 31, 40, 45, 60, 90, 105, 115, 126, 135, 143, 148, 154, and 165, respectively; lane 14, natural CsnSN. [␪] were determined by amino acid analysis. Secondary structures were estimated using the CONTIN program (30). Papain and cathepsin C inhibition assays. The amount of CsnSN was determined by the BCA protein assay (Pierce, Rockford, IL) according to the manufacturer’s instruction. The papain inhibition assay was performed using N ␣-benzoyl-L-arginine-7-amido-4-methylcoumarin (Bz-Arg-AMC; Sigma, St. Louis, MO) as a substrate as previously described (29). Briefly, 0 –2 nmol of CsnSN or variants in 250 ␮l of buffer (0.4 M sodium phosphate, pH 6.0, 8 mM DTT, and 4 mM EDTA) was incubated at 40°C for 10 min with 200 mU of papain (EC 3.4.22.2; Sigma, St. Louis, MO) (in 500 ␮l of 0.1% Brij 35/H 2O). After the incubation, 250 ␮l of 20 ␮M substrate in H 2O was added and the reaction was incubated for another 10 min. The enzyme reaction was then terminated by the addition of 1.0 mL of 0.1 M sodium monochloroacetate in 0.1 M sodium acetate, pH 4.3. The release of 7-amido-4-methylcoumarin was monitored on a fluorescence spectrophotometer (Perkin-Elmer Model 650-40, Norwalk, CT) using excitation wavelength of 345 nm and emission wavelength of 438 nm.

136

TSENG ET AL.

The cathepsin C inhibition assay was performed using glycine-phenylalanine-7-amido-4-methylcoumarin (H-Gly-Phe-AMC; Bachem Feinchemikalien, AG) as a substrate (31). Briefly, 0 –2 nmol of CsnSN or variants in 125 ␮l of buffer (10 mM sodium phosphate buffer, pH 7.0, containing 5 mM cysteine–HCl) was incubated at 37°C for 10 min with 3.2 mU of cathepsin C (EC 3.4.14.1; Sigma) (in 250 ␮l of the same buffer). After the incubation, 125 ␮l of 80 ␮M substrate in the same buffer was added and incubated further for 10 min. The enzyme reaction was terminated by 500 ␮l of 0.1 M sodium monochloroacetate in 0.1 M sodium acetate, pH 4.3. The release of 7-amido-4-methylcoumarin was monitored fluorometrically as above. Determination of equilibrium inhibition constants (K i). Equilibrium data for CsnSN–papain or CsnSN– cathepsin C interactions were obtained by measurement of progress curves for substrate hydrolysis in the presence of an inhibitor (32, 33). CsnSN–papain interactions were examined in 1 mL buffer composed of 100 mM sodium phosphate, pH 6.0, 2 mM DTT, 1 mM EDTA, and 0.05% Brij 35/H 2O. Uninhibited rate (v 0) of substrate hydrolysis was obtained with 10 ␮M Bz-Arg-AMC and 10 nM papain. The inhibited rate (v i) was obtained in the presence of 2–12 nM inhibitor, except for variant L 56 –58 where the inhibitor concentration was from 0.4 –1 ␮M. CsnSN– cathepsin C interactions were examined in 1 mL buffer composed of 10 mM sodium phosphate, pH 7.0, and 5 mM cysteine–HCl. The uninhibited rate (v 0) of substrate hydrolysis was obtained with 20 ␮M H-Gly-Phe-AMC and 5.3 nM cathepsin C. The inhibited rate (v i) was obtained in the presence of 1– 6 ␮M inhibitor, with exception of natural CsnSN, where the inhibitor concentration was 93– 465 nM. In all experiments, measurements started about 10 s after enzyme addition. Product formation was monitored by fluorescence using the excitation and emission wavelengths of 345 and 438 nm, respectively. Substrate hydrolysis never exceeded 5% of the total substrate used. Steady-state velocities before (v 0) and after (v i) addition of inhibitor were obtained by linear regression analysis of the progressive curve data. Apparent K i values (K i(app)) were calculated as the 1/slope of the plot of v 0/v i ⫺ 1 versus [I]. The inhibition constant, K i, was calculated from the relation of K i ⫽ K i(app)/1 ⫹ [S]/K m , where [I] is the inhibitor concentration, [S] is the substrate concentration, and K m the Michaelis-Menten constant. The K m values of papain and cathepsin C for the substrate, under the given conditions, were estimated by similar measurements without inhibitor with substrate concentration varying from 0 to 123 ␮M for papain and from 0 to 80 ␮M for cathepsin C. The data were analyzed by nonlinear regression to the Michaelis-Menten equation using the UltraFit nonlinear curve-fitting program (Biosoft, Ferguson, MO).

FIG. 2. Electrophoretic analyses of rCsnSN. (A) 12% SDS–PAGE. The gel was stained with Coomassie blue/silver stain. Lane 1, lowmolecular-weight markers; lane 2, purified baculovirus-expressed rCsnSN; lane 3, natural salivary CsnSN. (B) 7.5% Anionic PAGE. The gel was stained with Coomassie blue. Lane 1, rCsnSN; lane 2, natural salivary CsnSN.

tration prior to the final purification with a DE-52 ion exchanger in a radial flow Superflo 100 column (Fig. 1A). SDS–PAGE/immunoblotting of the column fractions is shown in Fig. 1B. The natural human salivary CsnSN (lane 14) served as a positive control. The higher molecular weight band seen in lanes 3 and 4 (fractions 45 and 60, respectively) represented an aggregated CsnSN, as also seen with the natural Csn (lane 14). SDS–PAGE/immunoblotting of the column fractions indicated that the first peak contained rCsnSN, and SDS–PAGE/silver staining indicated that only the descending portion of the first peak contained pure rCsnSN (data not shown). Fractions containing pure protein were lyophilized and characterized. Approximately 30 mg of rCsnSN could be produced by 2 ⫻ 10 9 Sf9 cells in 1 L of culture.

RESULTS

Characterization of rCsnSN

Expression of the Full-Length rCsnSN by Sf9 Cells Using BEVS

On SDS–PAGE, the purified rCsnSN yielded a single band of ⬃14 kDa (Fig. 2A, lane 2) with a mobility similar to the natural CsnSN (Fig. 2A, lane 3). Anionic PAGE also demonstrated that purified rCsnSN (Fig. 2B, lane 1) had an electrophoretic mobility identical to natural CsnSN (Fig. 2B, lane 2). The N-terminal 15-aa sequence of purified rCsnSN was found to be identical to that of natural CsnSN (34), indicating that insect cells are able to recognize and cleave mammalian signal peptide as previously reported (35). The secondary structure of both natural and rCsnSN were compared by CD spectroscopy using aqueous conditions. Both molecules have virtually identical CD patterns, with a ␭ min at ⬃205 nm (data not shown). The CONTIN secondary structure estimations (30) indicated a similar composition of 12% ␣-helix, 44 –50% ␤-sheet, and 37– 44% other structures. Both natural

The expression of rCsnSN by Sf9 cells as a function of p.i. time was analyzed by SDS–PAGE/immunoblotting of the culture medium using rabbit anti-CsnSN antiserum. Densitometric analysis of the developed blot indicated that a significant level of rCsnSN was secreted in the culture medium 36 h p.i. and continued to accumulate up to 115 h (data not shown). No immunoreactive protein was visible in noninfected or wildtype AcNPV-infected cultures. Purification of rCsnSN The strategy we employed to facilitate the purification of secretory rCsnSN was to express the protein in serum-free medium, and to enrich rCsnSN by ultrafil-

THE ACTIVITY OF BACULOVIRUS-EXPRESSED HUMAN SALIVARY CYSTATIN SN

FIG. 3. Cysteine protease (papain) inhibitory activity of purified natural CsnSN and rCsnSN. The data are presented as the percentage of residual papain activity. The amount of CsnSN inhibitor used was determined by the BCA protein assay.

and rCsnSN demonstrated parallel dose–response curves in a papain-inhibition assay using Bz-Arg-AMC as substrate (Fig. 3). These results collectively indicate that rCsnSN is biochemically and biophysically comparable to natural CsnSN. Construction, Purification, and Characterization of CsnSN Variants The recombinant shuttle vectors carrying CsnSN mutations were constructed as described under Materials and Methods. The variant proteins (N 12–13, L 56 –58, and L 106 –107) were each purified from the culture supernatants by sequential ultrafiltration and radial flow chromatography as described above for the rCsnSN. SDS–PAGE/silver staining demonstrated that all purified variant proteins have a mobility similar to the natural CsnSN (data not shown).

FIG. 4. Comparison of the papain-inhibitory activities of natural and variant forms of rCsnSN. 䊐, Natural CsnSN; }, N 12–13 (G12– G13 3 A–A); F, L 56 –58 (Q56 –T57–V58 3 GGG); ⌬, L 106 –107 (P106 – W107 3 GG); ⫻, N-truncated CsnSN. See text for details.

The equilibrium inhibition constants (K i) of the complexes between natural human salivary CsnSN or recombinant CsnSN variants and papain were summarized in Table I. We have shown that the K i value for CsnSN binding to papain was extremely small, at a level of ⬍10 pM. The K i values of two variants, N 12–13 and L 106 –107, for papain were found to be similar to that of the natural CsnSN (K i ⬍ 10 pM). On the other hand, the K i value of variant L 56 –58 for papain was 57 nM, representing ⬃5700-fold reduction of papain inhibition. Inhibition of Cathepsin C The bioactivities of the variant rCsnSN and the natural CsnSN were compared using a cathepsin C-inhibition assay (Fig. 5). The dose–response curves showed that both natural and variant CsnSNs inhibited cathepsin C less efficiently. The natural CsnSN exhibited

Inhibition of Papain The bioactivities of the rCsnSN variants and the natural CsnSN were first compared using a papain inhibition assay (Fig. 4). The dose–response curves showed that at 0.25 nmol, the natural CsnSN, the N-terminal variant (N 12–13) and the loop II variant (L 106 –107) exhibited 100% papain inhibition (0% residual papain activity), whereas the same amount of the loop I variant (L 56 –58) showed only 12% papain inhibition (88% residual papain activity). Even at 2.0 nmol, L 56 –58 exhibited only 72% papain inhibition (28% residual papain activity).

137

TABLE I

Equilibrium Inhibition Constants (K i) of the Complexes between CysP and CsnSN or Variants

CsnSN N 12–13 L 56–58 L 106–107

Papain

Cathepsin C (␮M)

⬍10 pM ⬍10 pM 57 ⫾ 3.6 nM ⬍10 pM

0.1 ⫾ 0.01 1.6 ⫾ 0.2 1.9 ⫾ 0.1 14.0 ⫾ 0.8

Note. Measured values are given with their standard errors. N 12–13, G12–G13 replaced with A12–A13; L 56 –58, Q56 –T57–V58 replaced with G56 –G57–G58; L 106 –107; P106 –W107 replaced with G106 –G107.

138

TSENG ET AL.

FIG. 5. Comparison of the cathepsin C-inhibitory activities of natural and variant forms of rCsnSN. 䊐, Natural CsnSN; }, N 12–13 (G12–G13 3 A–A); F, L 56 –58 (Q56 –T57–V58 3 GGG); ⌬, L 106 –107 (P106 –W107 3 GG); ⫻, N-terminal truncated Csn. See text for details.

only 50% inhibition at 0.25 nmol, 75% at 0.5 nmol, and almost 100% at 2.0 nmol. All three variants had still weaker inhibitory activity, with the L 106 –107 being the least effective inhibitor. At 0.25 nmol, all three variants exhibited only 3–10% inhibition (90 –97% residual cathepsin C activity), as opposed to the 50% seen with the natural CsnSN. At 1.0 nmol, the natural CsnSN exhibited 94% inhibition (6% residual activity), while variants N 12–13, L 56 –58 and L 106 –107 exhibited 28, 36, and 0% inhibition (72, 64, and 100% residual activity), respectively. At 2.0 nmol, the natural CsnSN exhibited 100% inhibition, while the three variants exhibited 40, 66, and 14% inhibition, respectively. The K i values of the complexes between natural CsnSN or its variants and cathepsin C are also summarized in Table I. The K i value of natural CsnSN for cathepsin C was 0.1 ␮M, indicating that CsnSN binds to cathepsin C much less tightly than to papain. The K i values of two variants, N 12–13 and L 56 –58, were 1.6 and 1.9 ␮M, respectively, representing a 16- and 19-fold increase in the K i when compared to the natural CsnSN K i value. The mutation at the loop II region produced a very ineffective cathepsin C inhibitor, with a K i value of 14 mM, representing a 140-fold reduction in inhibitory effectiveness. DISCUSSION

Our present and previous studies (36, 37) have demonstrated that production of salivary proteins in BEVS offers an advantage over the E. coli system. In contrast

to the bacterial system, the same protein expressed in BEVS does not form insoluble aggregates. Traditionally, the Sf9 insect cells used in BEVS are maintained in media requiring supplementation with fetal bovine serum. In the present study, we have cultured the Sf9 cells in the protein-free Sf-900 II medium. The rCsnSN and variants were isolated and concentrated using spiral cartridges, and finally purified with a DE-52 ion exchanger in a radial flow Superflo 100 column (Fig. 1A). These strategies helped reduce protein-processing time and increase the yield of the end products. Indeed, approximately 30 mg of rCsnSN could be purified from 1 L of culture containing 2 ⫻ 10 9 Sf9 cells. Protein characterization indicates that the Sf9-expressed rCsnSN is biochemically and biophysically comparable to natural CsnSN. To assess the functional importance of individual contact regions of CsnSN in the context of its affinity and specificity of binding to different cognate enzymes, we tested the natural and variant CsnSNs for papainand cathepsin C-inhibition. Both the inhibition assays (Figs. 4 and 5) and K i measurements (Table I) suggest that CsnSN discriminates strongly between papain and cathepsin C. We have shown that CsnSN possesses a K i value of ⬍10 pM for papain and 0.1 ␮M for cathepsin C. Therefore, CsnSN is at least a 1000-fold weaker inhibitor of cathepsin C than of papain. We have also shown that a mutation at either the N-terminal (N 12–13) or the loop II (L 106 –107) region did not significantly affect the K i values for the two variants binding to papain (Table I). These results suggest that neither the N-terminal G12–G13 nor the loop II P106 – W107 residues contributed significantly to the CsnSN– papain interaction. Furthermore, although mutation in the loop I region results in at least a 5700-fold reduction in papain inhibition, the fact that L 56 –58 still possesses a K i value within the nM-range suggests that papain is relatively insensitive to this CsnSN mutation. The role of the N-terminals of other Csn members in CysP inhibition has been previously studied. For example, it was shown that replacing CsnC conserved G11 (equivalent to CsnSN G12) with larger E and W residues caused a 2000-fold reduction in CsnC binding affinity to both the tight- (papain and ficin) and weak-binding (actinidin and cathepsin B) enzymes, whereas G11S and G11A mutations caused only a 20fold decrease in binding affinity (39). It is likely that the bulkier side chains of E and W caused steric hindrance and prevented Csn from binding efficiently with papain. Similarly, the G-to-A mutation in our N-terminal variant (N 12–13) may be too conservative to significantly alter CsnSN-papain binding behavior. Indeed, computer molecular modeling has demonstrated that G12 to A12 mutation only caused a minor increase of the intermolecular distance of 3.7 to 4.8 Å, respectively, to the papain reactive C25 (data not shown).

THE ACTIVITY OF BACULOVIRUS-EXPRESSED HUMAN SALIVARY CYSTATIN SN

Our result with the loop I variant (L 56 –58) substantiates Bode’s model of Csn–papain interaction (19). The model proposed that Csns loop I could interact with the active site cleft of papain-like enzymes. The present study has demonstrated that loop I region plays an important role in papain inhibition by CsnSN. This result also coincides with a previous study that showed that Q53E, Q53N, V55D, G57A, and other combination variants possess reduced inhibitory activity against papain, actinidin, and cathepsin B by 10- to 1000-fold (40). Little is known about the contribution of the loop II region to Csn–papain complex formation. Previous studies using W-modifying agents such as N-bromosuccinimide (NBS) (41) and 2-hydroxy-5-nitrobenzyl (HNB) (42) indicated that W104 of chicken Csn (equivalent to W107 of CsnSN) may be located in or near Csn’s CysP-binding site. However, our result indicates that P106 –W107 of CsnSN is not essential for papain inhibition. In fact, our result coincides well with a previous study on the inhibitory activities of individual segments of kininogens. Kininogens are members of Csn, which contain three divergent copies of type II Csn sequences. It has been demonstrated that the second segment of kininogens, which contained the N-terminal GC and loop I GVVAG but lacked the loop II PW sequence, still inhibited papain, cathepsin L, and calpain (43). More significantly, we have observed interesting differences between papain and cathepsin C inhibition by the three CsnSN variants. While double G-to-A replacements at G12 and G13 had no effect on papain inhibition (Fig. 4), they did have profound effects on cathepsin C inhibition (Fig. 5). This suggests that the G12–G13 region may participate in binding to cathepsin C-reactive site. Another interesting observation was that an N-terminal (aa 1–16)-truncated CsnSN variant (21) possessed better cathepsin C-inhibitory activity than the N 12–13 variant (Fig. 5). It is likely that truncation of the N-terminal 16-aa residues may allow the rest of the CsnSN polypeptide, particularly the loop II P106 –W107 region, to get closer to and block the cathepsin C-reactive site. Thus, the subtle increase in bulk created by the G-to-A mutation, which has no effect in papain inhibition, may have been sufficient to affect loop II blocking of the cathepsin C-reactive site. A mutation at the loop I QTV region has a similar effect as that of the N-terminal GG (Fig. 5). These results collectively suggest that loop II containing P106 –W107 is the most important functional domain of CsnSN for cathepsin C inhibition, while both the G12–G13 and the Q56 –T57–V58 regions may aid in “shielding” off the cathepsin C reactive site, facilitating loop II blockage of the cathepsin C reactive site. In summary, the present study showed that all three variants examined (N 12–13, L 56 –58, and L 106 –107) exhibited different effects toward their target enzymes, papain

139

and cathepsin C. This behavior indicated that the three highly conserved regions (i.e., the N-terminus, loop I, and loop II) of human salivary CsnSN had differential contributions to the interaction mechanism with papain and cathepsin C, presumably as a result of differences in the active-site structures of these CysPs. Such distinct effects of the inhibitor variants toward different CysPs can ultimately be utilized in the development of saliva-based therapeutic agents in that specific inhibitors can be designed to target against specific disease-related CysPs. ACKNOWLEDGMENTS This study was supported by NRSA Postdoctoral Training Grant 5 T32 DE07034 and USPHS Grants DE07034, DE07585, DE08240, and DE09820.

REFERENCES 1. Brocklehurst, K., Kowlessur, D., O’Driscoll, M., Patel, G., Quenby, S., Salih, E., Templeton, W., Thomas, E. W., and Willenbrock, F. (1987) Biochem. J. 244, 173–181. 2. Barrett, A. J., and Kirschke, H. (1981) Methods Enzymol. 80, 535–561. 3. Taugner, R., Buhrle, C., Nobiling, R., and Kirschke, H. (1985) Histochemistry 88, 102–108. 4. Sloane, B., and Honn, K. (1984) Cancer Metastasis Rev. 3, 249 – 263. 5. Kapur, V., Topouzis, S., Majesky, M. W., Li, L. L., Hamrick, M. R., and Hamill, R. J. (1993) Microbial. Pathogen. 15, 327–346. 6. Lah, T., Skaleric, U., Babnik, J., and Turk, V. (1985) J. Periodont. Res. 20, 458 – 466. 7. Pavloff, N., Potempa, J., Pike, R. N., Prochazka, V., Kiefer, M. C., Travis, J., and Barr, P. J. (1995) J. Biol. Chem. 270, 1007–1010. 8. Korant, B., Brzin, J., and Turk, V. (1985) Biochem. Biophys. Res. Commun. 127, 1072–1076. 9. Bjorck, L., Akesson, P., Bohus, M., Trojnar, J., Abrahamson, M., Olafsson, I., and Grubb, A. (1989) Nature 337, 385–386. 10. Leung-Tack, J., Tavera, C., Gensac, M. C., Martinez, J., and Colle, A. (1990) Exp. Cell. Res. 188, 16 –22. 11. Bobek, L. A., and Levine, M. J. (1992) Crit. Rev. Oral Biol. Med. 3, 307–332. 12. Shomers, J. P., Tabak, L. A., Levine, M. J., Mandel, I. D., and Hay, D. I. (1982) J. Dent. Res. 61, 397–399. 13. Johnsson, M., Richardson, C. F., Bergey, E. J., Levine, M. J., and Nancollas, G. H. (1991) Arch. Oral. Biol. 366, 631– 636. 14. Al-Hashimi, I., and Levine, M. J. (1989) Arch. Oral Biol. 34, 289 –295. 15. Skaleric, V., Babnik, J., Curin, V., Lah, T., and Turk, V. (1989) Arch. Oral Biol. 34, 301–305. 16. Obenauf, S. D., Fisher, R. H., Cowman, R. A., and Fitzgerald, R. J. (1984) Infect. Immun. 46, 797– 801. 17. Bergey, E. J., Gu, M., Collins, A. R., Bradway, S. D., and Levine, M. J. (1993) Oral Microbiol. Immunol. 8, 89 –93. 18. Gu, M., Haraszthy, G. G., Collins, A. R., and Bergey, E. J. (1995) Oral Microbiol. Immunol. 10, 54 –59. 19. Bode, W., Engh, R., Musil, D., Thiele, U., Huber, R., Karshikov, A., Brzin, J., Kos, J., and Turk, V. (1988) EMBO J. 7, 2593–2599. 20. Bobek, L. A., Wang, X., and Levine, M. J. (1993) Gene 123, 203–210.

140

TSENG ET AL.

21. Bobek, L. A., Ramasubbu, N., Wang, X., Weaver, T. R., and Levine, M. J. (1994) Gene 151, 303–308. 22. Ramasubbu, N., Weaver, T., Tseng, C.-C., Bobek, L. A., and Levine, M. J. (1996) Acta Crystallogr. D52, 869 – 870. 23. Fraser, M. J. (1992) Curr. Top. Microbiol. Immunol. 158, 131– 172. 24. Bobek, L. A., Aguirre, A., and Levine, M. J. (1991) Biochem. J. 278, 627– 635. 25. Summers, M. D., and Smith, G. E. (1987) A Manual of Methods for Baculovirus Vector and Insect Cell Culture Procedures, Texas Agricultural Experiment Station Bulletin 1555, College Station, TX. 26. Aguirre, A., Testa-Weintraub, L. A., Banderas, J. A., Dunford, R., and Levine, M. J. (1992) Arch. Oral Biol. 37, 355–361. 27. Laemmli, U. K. (1970) Nature 277, 680 – 685. 28. Ornstein, L. (1964) Ann. NY Acad. Sci. 121, 321–349. 29. Ramasubbu, N., Reddy, M. S., Bergey, E. J., Haraszthy, G. G., Soni, S. D., and Levine, M. J. (1991) Biochem. J. 280, 341–352. 30. Provencher, S. W. (1982) Comp. Physics Commun. 27, 229 –242. 31. McDonald, J. K., Zeitman, B. B., Reilly, T. J., and Ellis, S. (1969) J. Biol. Chem. 244, 2693–2709. 32. Nicklin, M. J. H., and Barrett, A. J. (1984) Biochem. J. 223, 245–253. 33. Salvesen, G., and Nagase, H. (1989) in Proteolytic Enzymes. A

34. 35.

36. 37.

38. 39. 40.

41. 42. 43.

Practical Approach (Beynon, R. J., and Bond, J. S., Eds), pp. 83–104, IRL Press, Oxford. Al-Hashimi, I., Dickinson, D. P., and Levine, M. J. (1988) J. Biol. Chem. 263, 9381–9387. Smith, G. E., Ju, G., Ericson, B. L., Moschera, J., Lahm, H. W., Chizzonite, R., Summers, M. D. (1985) Proc. Natl. Acad. Sci. USA 82, 8404 – 8408. Tseng, C.-C., Tseng, C.-P., Lupovici, J., Yip, J., and Chou, J. (1998) J. Dent. Res. 77(B), 850. Tseng, C.-C., Miyamoto, M., Ramalingam, K., Hemavathy, K. C., Levine, M. J., and Ramasubbu, N. (1999) Arch. Oral Biol. 44, 119 –127. Lindahl, P., Abrahamson, M., and Bjork, I. (1992) Biochem. J. 281, 49 –55. Bjork, I., Brieditis, I., and Abrahamson, M. (1995) Biochem. J. 306, 513–518. Auerswald, E. A., Genenger, G., Assfalg-Machleidt, I., Machleidt, W., Engh, R. A., and Fritz, H. (1992) Eur. J. Biochem. 209, 837– 845. Lindahl, P., Ariksson, E., Jornvall, H., and Bjork, I. (1988) Biochemistry 27, 5074 –5082. Nycander, M., and Bjork, I. (1990) Biochem. J. 271, 281–284. Salvesen, G., Parkes, C., Abrahamson, M., Grubb, A., and Barrett, J. (1986) Biochem. J. 234, 429 – 434.