INTRODUCTION

Serine proteinase inhibitors or serpins are a ubiquitous superfamily of homologous proteins that resemble ₁-proteinase inhibitor in overall structure and include antithrombin III, ₂-antiplasmin (₂-AP),¹ and plasminogen activator inhibitor-1 (1). In general, serpins contain a highly exposed reactive site loop near the carboxyl terminus of the molecule that interacts as a pseudosubstrate for the target protease. The inhibitory specificity of the serpin is in part defined by the P₁ amino acid residue in the reactive site loop, e.g. proteases of trypsin-like specificity attack efficiently peptide bonds after arginine and lysine residues, whereas chymotrypsin favors large hydrophobic amino acids in the P₁ position (2). Interaction of the P₁ amino acid with the active site of the target protease triggers a conformational change in the serpin that results in a stoichiometric 1:1 inhibitory complex with the protease generally detectable by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (3). In this manner, serpins play crucial roles in regulating proteolytic enzymes that are involved in a wide variety of vital processes including blood coagulation, fibrinolysis, complement activation, inflammation, and cell migration (3). In addition to the serpins that regulate protease activity, several members of this superfamily lack a protease-inhibitory capability and have other physiological roles. These latter proteins were originally identified as serpins by data base searching and include thyroxine binding globulin, angiotensinogen, and ovalbumin (1, 4). Ovalbumin represents the prototype of a unique family of proteins within the serpin superfamily, which have several structural features in common, including the absence of a typical cleavable signal sequence, but have been found to reside extracellularly, intracellularly, or both (5). Initial members of this ovalbumin family of serpin proteins (ov-serpins) include the human proteins plasminogen activator inhibitor-2 (6), monocyte/neutrophil elastase inhibitor (7), and squamous cell carcinoma antigen (8). More recently identified members of the human ov-serpins include a tumor suppressor called maspin (9), which is presumably not a protease inhibitor (10, 11), a bone marrow-associated inhibitory serpin, designated bomapin (12), and cytoplasmic antiproteinase or CAP (13), also known as protease inhibitor 6 (14) or as placental thrombin inhibitor (15).

CAP is an intracellular, probably cytoplasmic serpin that is expressed in a wide variety of human tissues including platelets (13, 15, 16, 17, 18, 19). Alignment of the deduced amino acid sequence of CAP with other serpins indicated that the P₁ reactive site loop residue in this inhibitor is Arg³⁴¹ (13, 15), consistent with the observation that CAP forms SDS-stable complexes with and inhibits several trypsin-like proteases including thrombin, trypsin, urokinase, plasmin, and factor Xa (16, 20, 21). Cytoplasmic proteases play a key role in a variety of cellular functions (22, 23), and our recent observations indicate that platelet CAP may interact with an endogenous protease upon platelet activation (19). However, the identity of intracellular target proteases for CAP or any other ov-serpin remain currently elusive. In order to obtain sufficient quantities of CAP for analyzing its interaction with different classes of proteases, we decided to generate large quantities of this inhibitor in recombinant form. This report details the expression of functionally active CAP in bacterial cells, and we demonstrate that CAP rapidly inhibits chymotrypsin at Met³⁴⁰, suggesting that this inhibitor may have evolved separate reactive sites for the specific regulation of different proteolytic activities. Moreover, in contrast to the previously observed dissociation of complexes between CAP and trypsin-like proteases, no release of active chymotrypsin from the chymotrypsin/CAP complex was detected.

EXPERIMENTAL PROCEDURES

The cDNA encoding CAP was isolated using the polymerase chain reaction as described previously (19). Briefly, RNA was extracted from phorbol 12-myristate 13-acetate-activated human erythroleukemic cells, reverse transcribed, and amplified utilizing CAP-specific primers that were designed based on the published nucleotide sequence (13), i.e. GTA CTG CTC GAG GTC TGC CAT CAT GGA TGT TC (sense; base position 179-198) and GAT GAC GAA TTC CTG CCC TGT CCT CAC GGA GA (antisense; base position 1330-1311). Primers included 12-base pair 5 overhangs containing the unique restriction sites XhoI and EcoRI, respectively. Amplification products were cloned into pBluescript SK(+) (Stratagene, La Jolla, CA), and both strands of the CAP coding region were sequenced completely by the dideoxy-mediated chain termination method (24). Subsequently, the entire coding region for CAP was subcloned into the bacterial expression vector pTrcHisB (Invitrogen, San Diego, CA) utilizing the XhoI and EcoRI restriction sites. This construct leads to the synthesis of CAP fused to a 39-amino acid N-terminal peptide containing six consecutive histidine residues as an affinity tag for purification (i.e. MGGSH HHHHH GMASM TGGQQ MGRDL YDDDD KDPSS RSAI). After transformation into Escherichia coli XL-1 blue, rCAP expression was induced by the addition of 1 mM isopropyl-thio--D-galactoside to a plasmid-carrying clone growing in log phase (optical density at 600 nm = 1.0) in the presence of 100 µg/ml ampicillin at 37 °C. The bacterial cells were grown for an additional 4 h (37 °C), harvested by centrifugation at 4 °C, and resuspended in 25 ml of ice-cold sonication buffer (300 mM NaCl, 50 mM sodium phosphate, pH 8.0, 2 mM -mercaptoethanol) per liter of culture volume, frozen in liquid nitrogen, slowly thawed in cold water, and kept at 4 °C for all subsequent purification steps. The freeze-thawed cells were extensively sonicated (Astrason Ultrasonic Processor XL; Heat Systems, Farmingdale, NY), cellular debris was pelleted by centrifugation for 20 min at 10,000 × g, and the supernatant was loaded onto a column containing 1 ml of Ni²⁺ charged Sepharose (Invitrogen) per liter of culture volume. The column was washed at a flow rate of 30 ml/h with 100 ml of sonication buffer followed by 500 ml of wash buffer (300 mM NaCl, 50 mM sodium phosphate, pH 6.0, 2 mM -mercaptoethanol, 10% glycerol) containing 0.1% Triton X-100 and 50 ml of wash buffer without Triton X-100. Bound proteins were subsequently eluted utilizing a 30-ml gradient of 0-0.5 M imidazole in wash buffer, optical density at 280 nm in 2-ml column fractions was determined, the fractions were subjected to SDS-PAGE followed by either silver staining or immunoblotting (described below), and the fractions containing recombinant CAP were pooled. After dialysis against phosphate-buffered saline (PBS; 137 mM NaCl, 6.4 mM Na₂HPO₄, 2.7 mM KCl, 1.5 mM KH₂PO₄, pH 7.33) containing 2 mM -mercaptoethanol, aliquots of the rCAP preparation were stored at 80 °C. Consistent with earlier reports (15, 16), we observed that rCAP activity is stabilized in the presence of 2 mM -mercaptoethanol.

High purity human -thrombin (Sigma T6759; 3000 units/mg protein), high molecular weight human urokinase (American Diagnostica Inc., Greenwich, CT), and 3 × crystallized bovine -chymotrypsin (Sigma C3142, Type VII) were used in the kinetic studies. Proteases were radiolabeled with ¹²⁵I utilizing immobilized chloramine T (Iodo-Beads; Pierce) according to the manufacturer's protocol. Specific radioactivities between 0.8 × 10⁴ and 1.4 × 10⁴ cpm/ng of protein were routinely obtained. Suspensions of washed human platelets were prepared as described previously (19). Briefly, blood from normal individuals was collected into acid-citrate-dextrose, and platelets were isolated from the platelet-rich plasma by centrifugation followed by three washing steps (19).

Samples were incubated in the presence of 2% SDS and 100 mM dithiothreitol (DTT) either at 100 °C for 3 min or at 65 °C for 5 min, as stated in the figure legends. Proteins were resolved by electrophoresis as described by Laemmli (25) using 4% stacking and 9% separating polyacrylamide slab gels unless stated otherwise. Protein molecular weight standards were obtained from Life Technologies, Inc. Following electrophoresis, the gels were analyzed by silver staining, dried, and exposed to XAR-5 film (Eastman Kodak Co.) for 24-48 h utilizing intensifying screens as described (19). Alternatively, protein A-purified polyclonal rabbit anti-CAP immunoglobulin G was prepared and used for immunoblotting as described previously (19). Peroxidase-conjugated donkey anti-rabbit immunoglobulin (Amersham Corp.) was utilized as the secondary antibody followed by detection using the enhanced chemiluminescence system (Amersham).

The chromogenic substrates S-2238 (for thrombin and trypsin) and S-2586 (for chymotrypsin) were obtained from Pharmacia Biotech Inc. Catalytic constants for hydrolysis of these chromogenic substrates were determined (Lineweaver-Burke plots) in PBS containing 0.01% bovine serum albumin (BSA; Sigma A-7906) at 23 °C. K_m values under these conditions were 10 µM for thrombin and S-2238, 34 µM for trypsin and S-2238, and 37 µM for chymotrypsin and S-2586. Thrombin and trypsin were active site-titrated with p-nitrophenyl p-guanidinobenzoate (26), and chymotrypsin was titrated with p-nitrophenyl acetate (27). The concentration of active inhibitor in preparations of rCAP was determined by incubating 25 nM thrombin, trypsin, or chymotrypsin in a total volume of 100 µl of PBS, 0.01% BSA with increasing concentrations of rCAP for 60 min at 23 °C in individual wells of a 96-well microtiter plate. The wells were in all experiments previously blocked with 200 µl of PBS, 0.1% BSA for 60 min at 23 °C. Residual amidolytic activity was analyzed after the addition of 50 µl 0.5 mM S-2238 (for thrombin and trypsin) or S-2586 (for chymotrypsin) in PBS, 0.01% BSA by monitoring the rate of p-nitroaniline release for 10 min at 405 nm using a V_max kinetic microplate reader (Molecular Devices Inc., Menlo Park, CA). The rate of substrate hydrolysis was plotted as a function of the molar amount of rCAP added to the individual reaction wells, and extrapolation to the x-intercept allowed the calculation of the precise concentration of rCAP capable of forming an inhibitory complex with the specific protease. Overall inhibition rates of thrombin, trypsin, and chymotrypsin by rCAP were measured under second order rate conditions by allowing equimolar concentrations of protease and inhibitor (0.5 nM) in 100 µl of PBS, 0.01% BSA to react for increasing periods of time before the addition of 50 µl of S-2238 or S-2586 in PBS, 0.01% BSA to a final concentration of 1 mM, which stops the association reaction and measures the residual amidolytic activity (28). The variation of the residual free protease concentration [E] with time (t) obeys the equation,

(Eq. 1)

where [E]₀ is the initial protease concentration. Estimates for the apparent second order rate constants k_a were determined by fitting the observed data to Eq. 1 by nonlinear regression analysis using the SigmaPlot software package (Jandel, San Rafael, CA). All experiments were performed in duplicate wells and included controls where protease or rCAP had been replaced with assay buffer. A continuous assay method was utilized to analyze the interaction of rCAP with chymotrypsin in more detail (13, 29, 30). The onset of inhibition was monitored by adding chymotrypsin (final concentration, 0.35 nM) to a mixture of rCAP and S-2586 (final concentration, 0.33 mM). The generation of p-nitroaniline in the presence of different rCAP concentrations (0-24 nM) was measured over time (t) for up to 8 h. Values of the determined absorbance at 405 nm (A) were fitted to the integrated rate equation for slow binding inhibition (29),

(Eq. 2)

by nonlinear regression analysis, generating values for v₀ (initial rate), v_S (final steady-state rate), k (apparent pseudo-first order rate constant for the transition from v₀ to v_S) and A₀ (the initial absorbance at 405 nm) for each of the inhibition curves. Kinetic constants for the inhibition of chymotrypsin by rCAP were derived from these values by various methods (see ``Results'').

Automated Edman degradation was carried out using the Procise Protein Sequencing System (Applied Biosystems, Uppsala, Sweden) with on-line analysis of the phenylthiohydantoins by microgradient high pressure liquid chromatography. Mass spectra were acquired by matrix-assisted laser desorption ionization time-of-flight spectrometry in a PerSeptive Biosystems (Framingham, MA) Voyager Elite linear time-of-flight mass spectrometer equipped with a 2-m flight tube and a 337-nm nitrogen laser. Spectra were sampled at a laser repetition rate of 20 Hz and a sampling rate of 250 MHz using sinapinic (trans-3,5-dimethoxy-4-hydroxycinnamic) acid as matrix.

RESULTS

Because current information indicates that native CAP is expressed as a cytoplasmic, presumably nonglycosylated protein without intramolecular disulfide bonds (13, 15), we investigated the applicability of utilizing a prokaryotic expression system for the preparation of large quantities of this inhibitor. The complete coding region of CAP (13) was polymerase chain reaction-amplified using CAP-specific primers and reverse transcribed RNA from phorbol 12-myristate 13-acetate-stimulated human erythroleukemia cells as the template. Polymerase chain reaction products were cloned into the prokaryotic expression vector pTrcHisB, and DNA sequencing confirmed sequence identity of a polymerase chain reaction product with the published CAP sequence (13) preceded in frame by the coding region for an N-terminal peptide containing six consecutive histidine residues. Following growth and induction of E. coli XL-1 blue carrying this pTrcHisB/CAP expression construct, the cells were lysed, and recombinant fusion proteins were purified by affinity chromatography on Ni²⁺-charged Sepharose. Approximately 500 µg of purified rCAP were obtained from 1 liter of bacterial culture, and the preparation exhibited a single major band at 45 kDa upon SDS-PAGE followed by silver staining (Fig. 1A; Fig. 3, lane 1). The apparent molecular mass of this protein corresponds to the combined calculated masses of the N-terminal fusion peptide (4.1 kDa) and CAP (42.6 kDa). A minor additional band at 42 kDa may represent reactive site cleaved rCAP. In order to determine the functional activity of rCAP, we tested its ability to form SDS-stable complexes with two known target proteases. Fig. 1B (lanes 4 and 7) demonstrates formation of appropriately sized complexes following incubation of rCAP with either ¹²⁵I-thrombin or ¹²⁵I-urokinase, respectively. We have previously shown (19) that human platelets contain relatively large quantities of CAP and that complexes of native platelet CAP and several trypsin-like proteases including thrombin and urokinase unfold to a higher apparent molecular mass upon boiling in the presence of SDS (Fig. 1B, compare lanes 1 and 2). ¹²⁵I-thrombin·rCAP and ¹²⁵I-urokinase·rCAP complexes undergo a similar temperature-dependent unfolding transition (Fig. 1B, lanes 3 and 4 and lanes 6 and 7, respectively), suggesting similar conformational stability of native CAP- and rCAP-containing complexes.

CAP is known to form SDS-stable complexes with a broad spectrum of trypsin-like proteases (16, 20, 21), and consistent with these observations arginine has been identified by sequence alignment as the reactive site loop P₁ residue in this inhibitor (13, 15). To understand the interaction of CAP with other classes of proteases, we investigated the interaction of CAP with chymotrypsin, a protease that cleaves substrate proteins preferentially after large hydrophobic amino acids, but not after arginine residues (2). Purified rCAP (Fig. 2, A and C) and whole platelet lysates (Fig. 2B) were incubated with increasing concentrations of ¹²⁵I-chymotrypsin, subjected to reducing SDS-PAGE, and transferred to a membrane. The membrane was first analyzed by Western blotting utilizing specific antibodies against CAP (19) and a chemiluminescent detection system (Fig. 2, right lanes), followed by autoradiography to detect ¹²⁵I-chymotrypsin-containing bands (Fig. 2, left lanes). In this manner, CAP and chymotrypsin were localized independently on the same membrane, allowing unambiguous identification of SDS-stable complexes between these proteins. Active chymotrypsin contains either two (1.3 + 23.7 kDa) or three (1.3 + 13.5 + 10.1 kDa) disulfide-linked polypeptide chains with the active site serine residue located in the C-terminal 23.7-kDa and 10.1-kDa fragments of the two- and three-chain forms, respectively (31). Fig. 2 demonstrates that rCAP (panel A) and native platelet CAP (panel B) form SDS-stable complexes with chymotrypsin. Other SDS-stable complexes with ¹²⁵I-chymotrypsin in platelet lysates (panel B) may contain ₁-proteinase inhibitor and ₂-AP, which are known to be present in platelets (32, 33). Chymotrypsin-CAP complexes unfold only upon boiling to an apparent molecular mass corresponding to the combined molecular masses of the serpin and the C-terminal chymotrypsin chains (panel C), as observed for complexes of trypsin-like proteases and either native or recombinant CAP (Fig. 1B).

We next analyzed whether chymotrypsin is actually inhibited in the complex with CAP. Active site titrated proteases were used to determine the concentration of rCAP in our preparation with inhibitory activity. Based upon the total protein content, 97, 93, and 75% of the rCAP inhibited the amidolytic activities of thrombin, trypsin, and chymotrypsin, respectively. These data indicate that the rCAP preparation is predominantly in an active form and that rCAP not only forms SDS-stable complexes with chymotrypsin but also inhibits this protease. However, the reduced inhibitory activity of the rCAP preparation against chymotrypsin suggested that this protease may also be capable of cleaving a portion of rCAP into an inactive form. To examine this possibility, the molecular species of rCAP that result following its incubation with increasing concentrations of chymotrypsin were compared with the profile following the incubation of rCAP with comparable concentrations of trypsin. Fig. 3 indicates that the amounts of chymotrypsin-rCAP (lanes 2-5) and trypsin-rCAP (lanes 6-9) complexes detected by SDS-PAGE decrease as the enzyme:inhibitor ratio approaches or exceeds 1:1 on a molar basis. More specifically, incubation of chymotrypsin and rCAP at 0.6:1 and 0.8:1 ratios (lanes 2 and 3) results in the appearance of 55- and 69-kDa complexes and the detection of a 42-kDa cleavage product, whereas incubation of chymotrypsin-rCAP at 1.0:1 and 1.2:1 ratios (lanes 4 and 5) results in the formation of lower molecular mass species (i.e. 44 and 39 kDa). Similarly, increasing the trypsin:rCAP ratio to 1.2:1 (lane 9) results in a loss of 67-kDa trypsin-rCAP complexes and the appearance of a number of lower molecular mass species. Thus, incubation of rCAP with a molar excess of either chymotrypsin or trypsin results in a decrease in the detection of SDS-stable complexes and the appearance of lower molecular mass species, presumably representing proteolytic degradation of protease-rCAP complexes. Importantly, these observations are consistent with our analysis of inhibition of amidolytic activity (reported above) indicating that approximately 80% of the rCAP with trypsin-inhibiting activity inhibits chymotrypsin under the current assay conditions.

To further define the interaction of chymotrypsin with rCAP, second order rate constants were calculated following the incubation of rCAP with equimolar concentrations of chymotrypsin as well as with equimolar concentrations of trypsin or thrombin. The reactions were allowed to react for increasing lengths of time (up to 10 min) prior to analysis of the residual amidolytic activity. Determination of the overall second order association rate constants (Table I) indicated that rCAP inhibits chymotrypsin rapidly and that the interaction of this recombinant serpin with thrombin and trypsin proceeds with similar overall rate constants as reported earlier for native CAP (13, 16). Because the slow binding kinetic approach for enzyme-inhibitor interactions (29) has been previously employed to study the inhibition of proteases by serpins (30, 34, 35), including the interaction of CAP with bovine trypsin and human factor Xa (13), we attempted to utilize this method to dissect the interaction of rCAP with chymotrypsin in more detail. Fig. 4 shows the progress of chymotrypsin inhibition in a continuous assay system using different concentrations of rCAP in a representative experiment. The data from each progress curve were fitted to Eq. 2 (see ``Experimental Procedures''), yielding values for the initial velocity of substrate hydrolysis v₀, the steady state velocity v_S, and the apparent rate constant k for the transition from v₀ to v_S. All progress curves reached a plateau where no further substrate hydrolysis could be detected (v_S < 2 × 10⁸ s¹ for all rCAP concentrations), indicating that chymotrypsin is not detectably released in active form over time from the tight complex with rCAP. Fig. 5 shows that v₀ was inversely related to the inhibitor concentration (panel A) and that k increased with increasing rCAP concentration (panel B), indicating that formation of the tight chymotrypsin-rCAP complex (EI*) involves two steps with the initial formation of a loose complex (EI) that subsequently isomerizes to a more stable form according to the reaction scheme, (13, 29, 30). A dissociation constant K_i(k₁/k₁) for the initial loose chymotrypsin-rCAP complex of 2.7 nMwas calculated from the slope of the linear regression line in Fig. 5A. The data in Fig. 5Bcould be fitted well to the hyperbolic equation expected for the proposed reaction scheme (29), and a forward isomerization constant k₂ = 0.013 s¹for the formation of the tight complex was calculated (k₂/K_i = 4.8 × 10⁶M¹s¹). To corroborate these results, a plot of 1/k versus1/[rCAP] was utilized to estimate K_i = 4.2 nMand k₂ = 0.014 s¹(k₂/K_i = 3.3 × 10⁶M¹s¹) from the intercepts of the linear regression line on the xand yaxis, respectively (36), in reasonable agreement with the data determined above (Fig. 5C). These results and the previously reported corresponding values for the interaction between trypsin and CAP (13) are summarized in Table Iand indicate that both proteases form an initial complex with CAP, which converts to a more stable form with similar rate constants. However, the reverse isomerization rate k₂of the trypsin-CAP complex has been found to be relatively high (5.0 × 10⁵s¹) (13), whereas no dissociation of the chymotrypsin-rCAP complex was detected in our assay system. To compare the release of active protease from trypsin-rCAP and chymotrypsin-rCAP complexes directly, we prepared complexes between these proteases and rCAP by incubating mixtures of 1 µMprotease and 1.5 µMrCAP for 1 h on ice. The preformed complexes were then quickly diluted 1:2000 into PBS, 0.01% BSA containing 1 mMS-2238 (for trypsin) or 1 mMS-2586 (for chymotrypsin) to give final protease concentrations of 0.5 nM, and the amidolytic activity was monitored over 5 h at 23 °C against a substrate blank. Whereas regeneration of trypsin amidolytic activity was readily detected with similar kinetic parameters as published previously (13), no amidolytic activity was demonstrated upon incubation of the chymotrypsin-rCAP complexes (data not shown). For an irreversible inhibition, k₂/K_ican be derived directly using the observed plateau values (A_plateau) in Fig. 4of the absorption at different inhibitor concentrations (Fig. 5D) (29). From the slope of a linear regression line through the origin of the plot of 1/A_plateauagainst inhibitor concentration, k₂/K_iwas calculated as 3.8 × 10⁶M¹s¹, in good agreement with the values determined above.

Table I. Kinetic constants for the interaction of CAP with thrombin, trypsin, and chymotrypsin Overall second order association rate constants for inhibition by rCAPa 106 · M1 · s1 Thrombin 0.12  ± 0.02 Trypsin 9.7  ± 2.0 Chymotrypsin 6.0  ± 1.4 Slow binding kinetic constants Ki (k1/k1) k2 k2 k2/Ki nM 102 · s1 105 · s1 106 · M1 · s1 Chymotrypsin (rCAP)b 3.3 ± 0.6 1.4 ± 0.2 4.3 ± 0.6  Trypsin (native CAP)c 2.0 0.57 5 2.85 a Overall inhibition rates were determined in a discontinuous assay using equal initial concentrations of protease and inhibitor (0.5 nM) as described under ``Experimental Procedures'' (Equation 1). The numbers are the average ± S.D. of 10 independent measurements. b The values were determined by the continuous assay method in the presence of competing chromogenic substrate (Fig. 4) and represent the mean results from three independent experiments obtained by analysis of the data as described in Fig. 5. c Data from Morgenstern et al. (13).

2

2

Because chymotrypsin prefers large hydrophobic amino acids in the P₁ position (2), reaction mixtures of rCAP and chymotrypsin were subjected to amino-terminal amino acid sequencing to identify the P₁ residue in the CAP reactive site loop that is responsible for the interaction with chymotrypsin (Table II). An amino acid sequence beginning with Arg³⁴¹ of CAP was identified, indicating that chymotrypsin cleaves CAP after Met³⁴⁰ (Fig. 6). This amino acid sequence was absent if similar mixtures of rCAP and chymotrypsin pretreated with the serine protease inhibitor phenylmethylsulfonyl fluoride were analyzed (data not shown). It should be noted that amino-terminal sequencing confirmed the efficient removal of the initiator methionine from rCAP by methionyl aminopeptidase in E. coli (37). To corroborate our observation of a distinct reactive site residue in CAP that interacts with chymotrypsin, chymotrypsin and thrombin were separately incubated with a 2-fold molar excess of rCAP, and the reaction mixtures were analyzed by mass spectrometry to determine the molecular mass of N-terminal rCAP fragments after interaction with protease (Table III). Masses corresponding to unreacted rCAP (without the initiator methionine) and to rCAP fragments derived by cleavage after Arg³⁴¹ (for thrombin) and Met³⁴⁰ (for chymotrypsin) were observed, indicating that complexes between rCAP and either thrombin or chymotrypsin are at least partially disrupted under the conditions used for mass spectrometry. Thus, the mass spectrometric analyses of rCAP fragments generated by incubation with chymotrypsin (Table III) are in agreement with the data obtained by Edman degradation (i.e. Met³⁴⁰ as the P₁ residue; Table II), and comparable studies with rCAP and thrombin mixtures (Table III) confirm the expected cleavage site Arg³⁴¹ for trypsin-like proteases (Fig. 6).

Table II. Edman degradation of reaction mixtures of rCAP and chymotrypsin rCAP (1 µM) was incubated with chymotrypsin (0.5 µM) in a volume of 50 µl of PBS for 15 min at 23 °C, residual proteolytic activity was blocked by adding phenylmethylsulfonyl fluoride to a final concentration of 200 µM, and samples were frozen for subsequent analysis by N-terminal amino acid sequencing as described under ``Experimental Procedures.'' Cysteine residues are not detected in this system. N-terminal residue number 1 2 3 4 5 6 rCAP, N terminus without initiator Met G G S H H H rCAP cleaved after Met340, C-terminal peptide R A R F V Chymotrypsin A-chain G V P A I Chymotrypsin B-chain I V N G E E Chymotrypsin C-chain A N T P D R

Table III. Mass spectrometric analysis of rCAP fragments after incubation with thrombin or chymotrypsin rCAP (1 µM) was incubated with thrombin (0.5 µM) or chymotrypsin (0.5 µM) in a volume of 50 µl (30 mM NaCl, 5 nM Na2HPO4, pH 7.0) for 15 min at 23 °C and chilled on ice, and samples were analyzed within 2 h by mass spectrometry as described under ``Experimental Procedures.'' Fragment Calculated mass rCAP uncleaved 46,699.7 Da rCAP cleaved after Arg341 (N-terminal fragment) 42,672.5 Da rCAP cleaved after Met340 (N-terminal fragment) 42,516.4 Da Observed mass rCAP + thrombin (n = 3) 46,747  ± 40 Da 42,649  ± 23 Da rCAP + chymotrypsin (n = 3) 46,735  ± 33 Da 42,522  ± 42 Da

DISCUSSION

In this report, we have shown that recombinant CAP can be expressed in E. coli and purified to apparent homogeneity as a fusion protein with an N-terminal peptide containing six consecutive histidine residues (Fig. 1A). Most importantly, this rCAP is functionally active, with similar characteristics as native CAP (Fig. 1B) and with overall association rate constants for the inhibition of thrombin and trypsin (Table I) similar to previously estimated values for native CAP (16). We demonstrate that native CAP and rCAP also form SDS-stable complexes with chymotrypsin (Figs. 2 and 3), which permitted us the opportunity of utilizing this recombinant protein to study the interaction with chymotrypsin in more detail. The results indicate that chymotrypsin is inhibited by rCAP with an overall second order association rate constant of 6.0 × 10⁶ M¹ s¹ (Table I), similar to the rate constant of 5.9 × 10⁶ M¹ s¹ reported for chymotrypsin inhibition by ₁-proteinase inhibitor, the most rapidly acting serpin previously known to interact with this protease (28). Our data clarify a recent report investigating the interaction between a series of proteases, including chymotrypsin, and yeast-expressed CAP in which SDS-stable chymotrypsin/yeast-expressed CAP complexes were not detected following their incubation at a 1:1 molar ratio (21). This observation resulted in these investigators concluding that cleavage products resulting from incubating chymotrypsin with yeast-expressed CAP followed by SDS-PAGE were mediated by nonspecific proteolysis of this serpin's reactive site loop. In light of these authors' observations that the neutralization of 14 nM trypsin required 15 nM yeast-expressed CAP, a 1:1 molar mixture between these two reagents would contain an excess of protease. Our current analysis of the interaction between chymotrypsin and rCAP indicates that this situation would result in the neutralization of the majority of chymotrypsin's activity but a loss of SDS-stable complexes mediated at equimolar or excess enzyme-to-inhibitor ratio (Fig. 3).

CAP interacts with chymotrypsin at Met³⁴⁰ as the reactive site P₁ residue corresponding to P₂ in the previously published alignments (13, 15), whereas thrombin interacts as expected at Arg³⁴¹ (Tables II and III, Fig. 6). ₂-AP has been previously shown to form inhibitory complexes with different proteases at separate reactive site residues (38), i.e. Arg³⁷⁶ for plasmin and trypsin and Met³⁷⁷ for chymotrypsin (numbering according Enghild et al. (39)) (Fig. 6), and thus our data suggest that this feature may be more common in the serpin superfamily than previously envisioned. Chymotrypsin binds with high affinity to rCAP, resulting in an initial reversible complex (K_i = 3.3 nM) that isomerizes subsequently more slowly to the final tight complex (k₂ = 0.014 s¹) (Figs. 4 and 5, Table I). These kinetic constants for the formation of the tight complex are similar to the reported values for the interaction of CAP with trypsin (Table I) and ₂-AP with plasmin and chymotrypsin (K_i = 8 nM, k₂ = 0.006 s¹; and K_i = 6.6 nM, k₂ = 0.009 s¹, respectively) (30). However, trypsin-CAP, plasmin·₂-AP, and especially chymotrypsin·₂-AP complexes have been shown to dissociate with relatively high k₂ values of 5.0 × 10⁵ s¹, 1.7 × 10⁶ s¹, and 1.1 × 10⁴ s¹, respectively (13, 30), whereas we could not detect any release of active chymotrypsin from the chymotrypsin-rCAP complex (Fig. 4). Furthermore, in ₂-AP chymotrypsin processes the N-terminal region and inactivates a significant portion of this inhibitor by cleavage after Met³⁷⁴ in the reactive site loop (39) (Fig. 6). These observations indicate that ₂-AP is a much better overall inhibitor of plasmin than of chymotrypsin. Thus, CAP is the first serpin that has been shown to interact with similar efficiency at separate reactive site residues with different proteases. The capability of two neighboring amino acids in the serpin reactive site loop to interact rapidly with similar kinetic constants with the substrate binding pockets of proteases strongly supports the hypothesis of a serpin's reactive site loop in a highly mobile and flexible conformation. The structure of the linkage between serpins and proteases in native complexes is currently not completely understood, and evidence supporting a tetrahedral intermediate (40, 41) as well as an acyl enzyme intermediate (42) has been reported. One possibility to explain these discrepant observations would be that the linkage between the protease and serpin can vary and is dependent upon the interacting molecules. Our observation that complexes between rCAP and trypsin, as well as complexes between rCAP and chymotrypsin, are at least partially disrupted either upon SDS-PAGE (Fig. 3) or under the conditions used for mass spectrometry argues against a covalent acyl enzyme intermediate form of these complexes. In addition to the chemical nature of the linkage between a serpin's reactive loop P₁ residue and the active site serine of the protease, other factors have been observed to contribute to the stability of serpin-protease complexes (43).

The rapid overall association rate and the absence of detectable release of active protease suggest that chymotrypsin may resemble a cognate target protease for this inhibitor. The observation that human pancreas contains the highest relative abundance of CAP mRNA in a survey of tissue samples (15) raises the possibility that CAP plays a role in the protection against active chymotrypsin and trypsin that might enter the cellular cytoplasm either from the extracellular milieu or from storage organelles. However, the wide distribution of this intracellular serpin suggests that pancreatic proteases are not the only targets of this serpin. Thus, the recent observation of high levels of CAP expression in the cytosol of endothelial and especially epithelial cells (e.g. cells lining the excretory ducts of exocrine glands) (18) appears to be consistent with the more general notion of CAP as a protective factor in cells that line organ or vessel lumina. In addition, proteolytic activities with chymotrypsin-like specificity are normally expressed in the cytoplasm of all cells (e.g. by the proteasome (44)), and our data suggest that these activities should be included in the search for physiological targets of CAP. Wright and Scarsdale (45) recently proposed that serpins may have evolved not only to inhibit certain target proteases irreversibly but to act as a buffer for proteolytic activities that might be generated rapidly during physiological processes. The elaborate mechanism of protease inhibition by serpins, including a major conformational change in the serpin and highly divergent rate constants of different protease serpin pairs for dissociation of the initial Michaelis-Menten complex (K_i), for isomerization to a tight complex (k₂), and for dissociation of a tight complex (k₂) may serve to provide a specific time course of activity for different proteases in the presence of a serpin. In this manner, CAP may have evolved separate reactive sites for the specific regulation of several distinct cytoplasmic proteases.

The recent cloning of the mouse CAP gene (14) indicates that the reactive site P₁-P₁ residues (arginine-cysteine) are the same in this inhibitor as in its human counterpart, suggesting that CAP inhibits a related target protease at this position in both species. However, considerable divergence of the primary structure of the reactive site loop at other positions (e.g. valine at P₂, methionine at P₂) was observed. These data may explain a lower or absent efficacy of mouse CAP for chymotrypsin inhibition, which was observed in the study of Kirschner et al. (46) analyzing a trypsin-binding factor isolated from mouse cells. It has been proposed (47, 48) that interspecies hypervariability in reactive site loops of extracellular serpins is caused by positive evolutionary selection for functional differences between them, with a key factor being their interaction with extrinsic proteases used by parasites to facilitate their spread throughout the host (47). Thus, human CAP may have acquired a physiologically relevant different inhibitory profile from mouse CAP following its divergence from a common ancestor. This information coupled with our current data raise the possibility that one selective force exerted on the reactive site loop of this inhibitor may have been the defense against proteases (e.g. chymotrypsin-like) (49, 50) expressed by viruses or other pathogens after invading host cells by access through organ lumina or from the blood.