Inhibitory Potency and Specificity of Subtilase-like Pro-protein Convertase (SPC) Prodomains*

The SPCs (subtilisin-like pro-protein convertases) are a family of enzymes responsible for the proteolytic processing of numerous precursor proteins of the constitutive and regulated secretory pathways. SPCs are them-selves synthesized as inactive zymogens. Activation of SPCs occurs via the intramolecular autocatalytic re-moval of the prodomain. SPC prodomains have been proposed as templates in the development of potent and specific SPC inhibitors. In this study, we investigated the specificity and potency of complete prodomains and short C-terminal prodomain peptides of each SPC on highly purified, soluble enzyme preparations of human SPC1, SPC6, and SPC7. Progress curve kinetic analysis of prodomain peptides and complete prodomains showed competitive inhibitory profiles in the low nanomolar range. Complete prodomains were 5–100 times more potent than C-terminal prodomain peptides, sug-gesting that N-terminal determinants are involved in the recognition process. However, complete prodomains and prodomain peptides exhibit only a partial specificity toward their cognate enzyme. SPC The were subcloned the pTrcHis A vector (Invit-rogen). The resulting protein is expressed in frame with a short N- terminal tag (Xpress) and a His 6 tag sequence that permits affinity purification using a Ni 2 (cid:4) column. One Shot TOP 10 cells (Invitrogen) were transformed with the cDNA constructs, and selected cells were amplified in log phase growth ( A 600 of 0.6 at approx. 2.5 h). Expression of the recombinant protein was induced with 1 m M isopropyl-1-thio- (cid:2) D -galactopyranoside. The collected cells were lysed, and the recom- binant protein purified by nickel affinity chromatography using the Xpress Protein Purification Kit (Invitrogen). Appropriate fractions were further purified by reversed-phase HPLC, and protein content was assessed Bradford protein assay, Coomassie Blue gel and mass spectrometry. Spectros- Recombinant-expressed proteins MALDI-mass spectrometer equipped delayed extraction and a 337-nm MALDI calibrated externally utilizing horse heart myoglobin. Average molecular determined using the dou- bly and singly charged molecular ions. Assays— C-terminal potent inhibitors (cid:1) Online progress curves 4-parameter

Proteolytic processing is a post-translational modification by which a cell diversifies and controls the protein products of its genes. In mammalian species, endoproteolytic activation of many secretory protein precursors is carried out by the SPCs 1 (for a review see Ref. 1). The catalytic domains of the SPCs have a high structural similarity to the bacterial subtilisins and yeast kexin (2). The SPC family of enzymes consists of seven distinct members named, using the unified nomenclature of Chan et al. (3), SPC1 (furin/PACE), SPC2 (PC2), SPC3 (PC1/PC3), SPC4 (PACE4), SPC5 (PC4), SPC6 (PC5/PC6), and SPC7 (LPC/PC7/PC8). Each SPC contains at least five well conserved domains: 1) an N-terminal signal peptide, responsible for directing proteins into the secretory pathway; 2) a prodomain acting as a putative intramolecular chaperone for facilitated transportation, folding, and regulation of enzymatic activity (4 -7); 3) a catalytic domain responsible for substratespecific interactions and cleavage; 4) a P-domain with a conserved RGD motif essential for enzyme structural cohesion (8) and activity and that also regulates stability, calcium, and pH dependence (9,10); and finally 5) a C-terminal-specific domain that contains membrane attachment sequences, Cys-rich regions, and intracellular sorting signals (1,11).
The substrate cleavage specificity of SPCs is recognized to be C-terminal to either single or paired basic residues, with the Lys-Arg motif being the most common. The substrate specificity of SPC1 has been thoroughly studied, with a minimal recognition sequence for catalysis being RXXR. In general, it is evident that SPCs favor the presence of basic residues at subsites P 1 , P 2 , and, in many cases, P 4 for efficient catalysis. The presence of an Arg residue at P 4 is not mandatory for the activity of all SPCs but various studies indicate that the presence of an Arg residue in P 2 , P 4 , or even P 6 enhances cleavage. The apparent close similarity of cleavage specificity between SPCs leads to the notion of possible redundant processing functions. Indeed, different SPCs have been shown to process various precursors at the same cleavage sites. Extensive mapping studies have demonstrated the frequent occurrence of overlapping cellular SPC expression patterns (12)(13)(14)(15)(16). Knockout studies in mice have also provided support for some level of redundant functions, depending on the precursor studied (17).
SPCs are involved in many important biological processes, including zymogen activation (6,7), peptide hormone processing (18 -21), complement activation (22), clot formation and lysis (23), angiogenesis (24), and tissue remodeling (25,26). These proteases have also been implicated in a number of pathophysiologies, thus raising the possibility that SPC inhibitors may become useful therapeutic agents (1,(27)(28)(29). However, the use of SPC inhibitors as pharmacological agents is highly dependent on a clear understanding of their redundant/ distinct functions. Furthermore, the development of highly potent and specific SPC inhibitors requires a better knowledge of the molecular determinants of catalytic activity that distinguish each member of the SPC family.
Endogenous inhibitors are often a good starting point in the development of pharmacological compounds. With regards to the SPCs, 7B2 CT (7B2 C-terminal) peptide and proSAAS are the only two endogenous inhibitors identified (30,31) as they specifically inhibit SPC2 and SPC3, respectively. However, regulation of enzymatic activity can also be carried out by intramolecular mechanisms. Indeed, SPCs, like many other proteinases, are synthesized as zymogens (32). The inhibitory mechanism often involves the presence of a prodomain, whose function is to prevent premature enzymatic activity (6). Each SPC contains a distinct prodomain that acts in cis to regulate that enzyme's activity. Based on the fact that SPC prodomains exhibit low levels of overall homology to each other, it has been suggested that they could be used in trans as potent and specific inhibitors of their cognate enzymes (33). Some evidence in support of the high inhibitory potency of prodomains has been provided. However, their degree of specificity remains somewhat unclear. The present study is an extensive comparative analysis of the inhibitory characteristics of SPC prodomains to address the issue of specificity. Our studies reveal that prodomains are potent inhibitors of SPCs in the low nanomolar range, that inhibition is highly dependent on the C-terminal structure of each prodomain, but also that some N-terminal elements within the prodomains are required for maximal inhibition. However, our results show that prodomains are not highly specific inhibitors of their cognate SPCs and that further structural modifications will be required to achieve higher levels of specificity.
Expression of Recombinant hSPC1⌬, hSPC6A, and hSPC7⌬-Drosophila Schneider 2 cells (S2 cells, Invitrogen) were grown in complete DES expression medium supplemented with 10% heat-inactivated fetal bovine serum, 50 units/ml penicillin G, 50 g/ml streptomycin, and 2 mM L-glutamine to a density of 6 -20 ϫ 10 6 cells/ml at 22-24°C. Stable cell lines were established by co-transfecting the enzyme cDNA/ pAC5.1/V5 HisA plasmid with pCoHYGRO plasmid (Invitrogen) at a weight ratio of 19:1 g, respectively, by calcium-phosphate precipitation and selecting stable transformants with 300 g/ml hygromycin B over a period of 3 weeks. The established resistant cell lines were then adapted to serum-free medium IPL-41-containing lipid supplement, 4 g/L ultrafiltrated yeastolate, 2 mM L-glutamine, 50 units/ml penicillin G, and 50 g/ml streptomycin but no hygromycin B by successive passages of medium with 5, 2.5, 1, and 0% of fetal bovine serum. The stable cell cultures were then scaled up to a volume of 600 ml in shake flasks spinning at 120 rpm supplying them with pluronic acid F68 to a final concentration of 0.1%. Typically, conditioned medium was collected every 6 -7 days and submitted to purification procedures.
Enzymatic Titration-Enzyme titration was performed for all enzyme preparations using the irreversible inhibitor dec-RVKR-CH 2 Cl. Assuming that all of the enzyme molecules were active, 1 nM of hSPC1⌬, hSPC6A, and hSPC7⌬ preparations, as estimated by Bradford protein assay, was incubated with increasing concentrations of Dec-RVKR-CH 2 Cl (0 -1 M) at room temperature for 15 min in a microtiter plate. Saturating concentrations of pERTKR-MCA (100 M for hSPC1⌬ and hSPC6A, 250 M for hSPC7⌬) were then added to evaluate the residual activity after 1 h of incubation at 37°C, which was recorded on a SpectraMax Gemini XS spectrofluorometer (Molecular Devices) (37).
Enzymatic and Protein Assays-Enzyme activity was routinely evaluated during the expression and purification procedures as well as for inhibition assays. The enzymatic assays of hSPC6A and hSPC7⌬ were carried out in 20 mM Bis-Tris, 1 mM CaCl 2 , pH 6.5 (2 g/l bovine serum albumin were added for hSPC6A). The hSPC1⌬ enzymatic assay was done in 100 mM HEPES, pH 7.5, 1 mM CaCl 2 , 1 mM ␤-mercaptoethanol, 0.5 g/l bovine serum albumin. Unless otherwise stated, all initial rate enzymatic assays were performed at saturating conditions of fluorogenic substrate pERTKR-MCA (250 M for hSPC7⌬ and 100 M for the hSPC1⌬ and hSPC6A), in the presence or absence of inhibitor at 37°C for 1 h or less in a total volume of 100 l in 96-well microtiter plates. The resulting fluorescence due to released MCA was measured online as progress curves on a SpectraMax Gemini XS spectrofluorometer with SoftMaxPro 3.3.1 software (Molecular Devices) at EX/EM wavelengths of 370/460 nm with a 435-nm cut-off filter, and the amount of MCA released was calculated by reference to corresponding MCA standard curves. Protein level was also routinely determined by colorimetric assay using Bradford reagent with bovine serum albumin as the standard protein.
Determination of hSPC1⌬, hSPC6A, and hSPC7⌬ Steady State Kinetic Constants-For hSPC7⌬, these experiments were repeated for two different preparations. The enzyme's pre-steady and steady states were first evaluated by performing time course experiments in saturating substrate conditions from 5 to 120 min in enzymatic assay conditions. The hydrolysis rate of pERTKR-MCA, measured by the amount of fluorescence versus time was plotted. Based on this, the Michaelis-Menten constant was determined by adding varying amounts of fluorogenic substrate (5 nM-1 M) to 1-26 units of enzyme in activity buffer. The reactions were incubated and monitored online for 30 min at 37°C. The data, plotted as the rate of hydrolysis activity versus pERTKR-MCA concentration, was fit to a standard pseudo-first order equation using ENZFITTER (BioSoft Corp.) to determine the Michaelis-Menten constant (K m ), the limiting rate (V max ), the turnover number (K cat ), and the K cat /K m ratio for the fluorogenic peptide pERTKR-MCA.
Production and Purification of Prodomains-Six prodomains were prepared using PCR amplification of full-length cDNAs of hSPC1, hSPC2, mSPC3, mSPC5, hSPC6, and hSPC7. The following sense and antisense primers were used: for hSPC1: cagaaggtcttcaccaacacg and tcaccgtttagtccgtcgctttgc; for hSPC2: gagcgaccggtcttcacgaat and tcatcgcttttttcggtcaaatcc; for mSPC3: aagaggcagtttgttaatgaa and tcaacgtttacttctctctttttc; for mSPC5: caggcccccatctatgtcagc and tcagcgtttcacccggcgcctcaa; for hSPC6: cgcgtctacaccaaccactgg and tcacctctttgtccgcttttttac: and for hSPC7: ctggcagggacaggtgggcct and tcagcgcttggcccgccttagcag. The amplified fragments contained the prodomain sequence of each SPC from the signal peptide cleavage site to the primary cleavage site. The PCR fragments were subcloned into the pTrcHis A vector (Invitrogen). The resulting protein is expressed in frame with a short Nterminal tag (Xpress) and a His 6 tag sequence that permits affinity purification using a Ni 2ϩ column. One Shot TOP 10 cells (Invitrogen) were transformed with the cDNA constructs, and selected cells were amplified in log phase growth (A 600 of 0.6 at approx. 2.5 h). Expression of the recombinant protein was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside. The collected cells were lysed, and the recombinant protein purified by nickel affinity chromatography using the Xpress Protein Purification Kit (Invitrogen). Appropriate fractions were further purified by reversed-phase HPLC, and protein content was assessed by Bradford protein assay, Coomassie Blue gel and mass spectrometry.
Matrix-assisted Laser Desorption Ionization (MALDI)-Mass Spectroscopy-Recombinant-expressed proteins were analyzed on a MALDImass spectrometer equipped with delayed extraction and a 337-nm nitrogen laser (Tof Spec 2E, Micromass, UK). Proteins were dissolved in water and diluted to 5 pmol/l in matrix solution (␣-4-hydroxycinnamic acid dissolved in 50% aqueous acetonitrile containing 0.1% trifluoroacetic acid) before application of 0.5 l onto an ultra thin matrix layer. (38). All MALDI spectra were calibrated externally utilizing horse heart myoglobin. Average molecular weights were determined using the doubly and singly charged molecular ions.
Inhibition Assays-The analysis of inhibition by complete prodomains, short C-terminal prodomain peptides, Ala-substituted peptides, and others was done in duplicate and is presented as the mean of three or more independent experiments. Concentrations of potent inhibitors ranging from 0.1 nM to 100 M were used. Inhibitors and saturating concentrations of substrate were mixed in appropriate buffer, and then active enzyme was added. Online progress curves were monitored by measuring the amount of MCA released at 37°C for an hour and were graphically analyzed by fitting data to 4-parameter plots to obtain the IC 50 . Since the value of IC 50 is not very informative, because its numerical value depends on the concentration of the enzyme, the true inhibition constant values, K i , were also determined. The K i values were calculated assuming behavior as competitive inhibitors and by correcting the values for the effect of substrate concentration (39) with the Cheng and Prusoff equation (40): where IC 50 is the concentration of inhibitor required to reduce enzyme velocity to 50% of maximal velocity, S is the concentration of substrate, and K m (Michaelis-Menten constant) is the substrate concentration (in the absence of inhibitor) at which the velocity of the reaction is half-maximal.
Circular Dichroism Analysis of SPC Complete Prodomains-Circular dichroism measurements were performed with a Jasco J-810 spectropolarimeter. The instrument was routinely calibrated with an aqueous solution of d-10-(ϩ)-camphorsulfonic acid at 290.5 nm. All samples of prodomain were analyzed in 20 mM Bis-Tris, pH 6.5, and 1 mM CaCl 2 buffer. The samples were loaded in quartz cells with pathlengths of 1 mm and 0.02 mm depending on the prodomain concentration, which ranged from 17 to 57 M. Far-UV wavelength scans were recorded at 5°C from 190 to 250 nm with 0.1 nM increments. The average of 20 wavelength scans is presented. The ellipticity results were expressed as mean residue ellipticity, [], in degrees cm 2 ⅐dmol Ϫ1 .

RESULTS
Characterization of Recombinant Prodomains-Six of seven SPC prodomains were prepared for this study using a bacterial expression system and characterized using MALDI-mass spectrometry. A representative MALDI spectrum of SPC3 is shown (Fig. 1) while all the results are summarized in Table I.
Preparation of Recombinant hSPC1⌬, hSPC6A, and hSPC7⌬ Using Schneider 2 Cells-To investigate the structure/activity relationship of the SPCs, enzymatically pure preparations of selected enzymes are an important condition for accurate analysis. Previously, numerous expression systems, including vaccinia virus and baculovirus, were used to produce enzymatically active recombinant SPCs. In the present study, we used a non-viral and highly efficient expression system employing Schneider 2 insect cells to produce soluble forms of hSPC1, hSPC6, and hSPC7. The recombinant enzymes were expressed continuously in serum-free media and then submitted to extensive purification procedures. First, we established three cell lines by transfecting S2 cells with plasmids containing the cDNA-encoding hSPC1⌬, hSPC6 (A isoform), and hSPC7⌬. Western blot analysis of the conditioned media of these selected stable lines, using hSPC1-, hSPC6-, and hSPC7-specific antibodies, showed strong immunoreactive signals of 83 and 80 kDa for hSPC1⌬ (34), 90 and 75 kDa for hSPC6A (not shown), and 89 and 86 kDa for hSPC7⌬ (Fig. 2). As a control, wild-type S2 cells did not show any hSPC1⌬, hSPC6A, and hSPC7⌬ immunoreactivity. The same purification procedures were employed to obtain pure active recombinant hSPC1⌬, hSPC6A, and hSPC7⌬. Table II summarizes the results for hSPC7⌬'s purification. Only one major proteolytically active and immunoreactive peak was detected after the MonoQ HR 10/10 anion exchange column, which was a 28-fold purification compared with the starting media. This active peak was further purified on a hydrophobic interaction column followed by size exclusion fractionation. After four steps of purification (including ultrafiltration), a single enzyme-specific peak was still observed, resulting in a 320-fold purification. Analysis of active peaks at all purification steps by Western blotting demonstrated that proteolytic activity coeluted with specific immunoreactivity throughout the entire procedure (Fig. 2, A and B). This provided strong evidence that the immunoreactive signal consisting in a doublet of 86 and 89 kDa is responsible for the enzymatic activity observed and corresponds to two isoforms of hSPC7⌬. These two isoforms may be the result of differential N-glycosylation of the four putative asparagine-linked glycosylation sites or of a minor truncation at the C-terminal of hSPC7⌬. The final yield of protein was 1.84 mg (from 1.6 liters of conditioned media) with a 20% recovery. Therefore, extrapolation of this result suggests that the expression system established can produce up to 10 mg of active enzyme in the crude preparation and hence, more than 5 mg per liter of media.  When aliquots of each purification step were subjected to SDS-PAGE and overexposed silver staining, multiple protein bands were apparent including an 86 -89-kDa band corresponding to the immunoreactive protein. Throughout the purification, this band was enriched compared with the non-immunoreactive proteins to a purity of greater than 80%, based on silver stain gel (Fig. 2C) and densitometry analysis.
In the case of hSPC1⌬, a single active peak with a strong immunoreactive signal at 81-83 kDa was also detected through the entire purification, which was purified 21.7-fold over the starting media. Again, a doublet of 83 and 81 kDa was observed by Western blotting analysis (34). A final yield of 2.81 mg was achieved. For hSPC6, two immunoreactive proteins of 90 kDa and 75 kDa coeluted in a single proteolytic activity peak. A purification of 16-fold and a final yield of 400 g was achieved.
The high purity preparations of hSPC1⌬, hSPC6A, and hSPC7⌬ and the precise knowledge of their active content allowed us to accurately establish the steady state kinetic parameters using pERTKR-MCA as the fluorogenic substrate. Very distinct values were obtained for K m , K cat and K m /K cat (Table III). These comparative in vitro behaviors indicate their distinct affinity for the substrate as well as their variable activity efficiency. The lower affinity and cleavage efficiency of hSPC7⌬ and hSPC6A for the fluorogenic substrate, as compared with that of hSPC1, was taken into consideration for the determination and comparison of the inhibition constants as described under "Inhibition Assays." Short and Long Propeptides Inhibitory Potency and Specificity-Previous studies indicate that prodomains could serve as potent and specific inhibitors that could be used in trans (33). These studies and others (6 -7, 38) indicate that the C-terminal region of the prodomain is critical for inhibition. In the present studies, we explored the specificity of each prodomain by using three different strategies, including 1) short synthetic C-terminal propeptides with their sequence terminating at the primary prodomain cleavage site, 2) longer synthetic C-terminal propeptides that included six amino acids in the PЈ positions, and 3) complete prodomains (Fig. 3). Based on an alignment of all seven prodomains, it can be observed that the highest homologous regions are in the C-terminal region, nearest to the cleavage site. This alignment suggests that if each prodomain has specificity for its cognate convertase, subtle differences in these C-terminal regions would be responsible or that N-terminal differences would be an important factor to determine specificity. The following studies tested these assumptions using in vitro kinetic analysis.
We first compared the progress curves of three short prodomain peptides and three complete prodomains on their respective enzymes (Fig. 4). Each peptide was a potent inhibitor in the nanomolar range, although complete prodomains were more potent. More importantly, these progress curves showed that short propeptides have a competitive behavior and that addition of the complete N-terminal region did not result in any significant changes in the kinetic profiles. The competitive inhibition behavior of the short propeptides and of the longer prodomains has been confirmed in saturation kinetic experiments that showed only a change in the affinity (i.e. higher K m app ) of the enzymes for the substrate, typical of competitive inhibitors that only bind to the catalytic site of the free enzyme (data not shown).
The next series of experiments investigated the specificity of each SPC short propeptide on the three enzyme preparations (Fig. 5). We tested the seven different short propeptides on hSPC1⌬ (Fig. 5A) and noted that four of the seven peptides tested had similar inhibitory potency, equivalent to the K i of the hSPC1 propeptide which was 184 nM. This result suggests that each short propeptide is not a highly specific inhibitor. Similar results were obtained for hSPC6A, where five of the seven propeptides tested were determined to be potent inhibitors. In the case of hSPC7⌬, only the hSPC7 and the mSPC5 propeptides displayed significant inhibition (K i values of 69 nM and 228 nM, respectively). Interestingly, the hSPC2 and hSPC3 short propeptides had very little inhibitory potency on any of the three enzyme preparations (Fig. 5). Furthermore, hSPC6A was the most sensitive to inhibition by the propeptides in general, while hSPC7⌬ had the most specific response.
The lack of specificity of each short propeptide may in fact not be so surprising considering that the C-terminal regions of each prodomain has a high degree of homology (Fig. 3). We therefore investigated if longer peptides that included residues in the PЈ region could improve the specificity of each propeptide (Table IV). A comparative analysis revealed that extending the short propeptides at the C-terminal had essentially no effect on potency and specificity.
Ala-scan Structural Analysis of hSPC1 and hSPC7 Short Propeptides-While our results indicated that short propeptides are generally nonspecific, some distinctions were observed between the hSPC1 and hSPC7 short propeptides in their capacity to inhibit hSPC1⌬ and hSPC7⌬. Indeed, the hSPC1 short propeptide is a 6-fold better inhibitor for hSPC1⌬ as compared with hSPC7⌬ (i.e. K i 184 nM for hSPC1⌬ and 850 nM for hSPC7⌬, see Table IV). Also, the hSPC7 short propeptide is at least a 15-times more potent inhibitor of hSPC7⌬ than of hSPC1⌬ (i.e. K i 69 nM for hSPC7⌬ and Ͼ1 M for hSPC1⌬, Table IV). To understand why hSPC1 and hSPC7 short propeptides had such distinct inhibition properties on both enzymes, we designed hSPC1 and hSPC7 Ala-substituted short propeptides (Fig. 6). We did not substitute the P 1 and P 2 C-terminal basic residues since it has already been shown that such substitutions completely abolish inhibitory potency (33). The results showed that inhibition of hSPC1⌬ decreased when basic residues in positions P 4 (Ala-9), P 5 (Ala-8), and P 6 (Ala-7) were substituted in the hSPC1 short propeptide (Fig. 6A). This is in good agreement with previous studies indicating that multiple basic residues are an essential requirement of inhibition for hSPC1 (41). In contrast, the Ala-substituted short propeptides designed for hSPC7⌬ revealed that the most important positions for inhibition were the basic residues in the P 4 (Ala-9) and P 5 (Ala-8) positions, but also the hydrophobic residues (Leu) in the P 6 (Ala-7) and P 7 (Ala-6) positions (Fig. 6B). When Alamodified hSPC7 short propeptides were tested on the hSPC1⌬ preparation, or when Ala-modified hSPC1 short propeptides were tested on the hSPC7⌬ preparation, no significant improvement in K i was observed (Table V).
Complete Prodomains Inhibitory Potency and Specificity-The short C-terminal region of prodomains ending with the required KR is sufficient for potent inhibition of the SPCs (Fig.  5, Table IV). However, their lack of specificity toward their cognate enzyme triggered us to investigate whether the Nterminal extension of the complete prodomains would provide a greater degree of specificity and possibly higher inhibitory potency. These complete prodomains were tested for their potency and specificity of inhibition of hSPC1⌬, hSPC6A, and hSPC7⌬ (Table VI). The results showed that each of the prodomains, except for the SPC2 prodomain, were highly potent inhibitors of the three enzymes tested (0.1-150 nM). These data also demonstrated that addition of N-terminal region of each prodomain did not increase the specificity of inhibition. However, a comparison of the inhibitory potency of the complete prodomains with their related short propeptides did reveal an increase in potency (Fig. 7). In general complete prodomains were 5-100 times more potent than the corresponding short propeptides.
Prodomain Structural Analysis-Recently, the global fold of mSPC3 prodomain was determined by NMR and CD spectroscopy and shown to closely correspond to the structure of the prodomain of subtilisin BPNЈ (42). The secondary structure of the mSPC3 prodomain was found to be a mixture of ␣-helix (22%), ␤-strands (30%), and random coils and turns. To evaluate the secondary structure of the prodomains that we produced, we measured their far-UV CD spectra (Fig. 8). The CD spectra of prodomains hSPC1, mSPC3, mSPC5, and hSPC7 are quite similar in shape, and closely related to the CD spectrum of mSPC3 reported (42). This suggests that these prodomains have similar content of secondary structure. As for the hSPC2 prodomain, the CD spectrum is indicative of an unfolded protein. On the other hand, the structure of hSPC6 prodomain seems to contain more ␣-helical structure, as depicted by the double minimum at 223 and 210 nm and the maximum at 192 nm. These analyses strongly suggest that the prodomains that we have expressed and purified are folded, with the exception of hSPC2. DISCUSSION Previous studies have established the ability of prodomains to act as intramolecular chaperones that are essential for the correct folding of their parent enzyme (43). For a number of peptidyl hydrolases including serine proteases, the prosegment has also been shown to be a potent inhibitor for its associated protease (4 -7). Some well studied prosegments include those of the bacterial subtilases, such as subtilisin E (43,44). These studies have demonstrated the high inhibitory potency of the prosegment. Thus, prodomains can also serve as a key component of the proteolytic machinery to regulate the intracellular processing of precursors. These observations have led to the proposal that prosegments could potentially be used in a strategy to regulate the enzymatic activity. As a strategy to develop potent inhibitors used in trans, prodomain sequences could be useful to develop lead compounds or molecular tools that can help in the study of SPC function.
Alignment of the SPC prodomain regions shows low levels of overall homology, but the most conserved region is found in their C-terminals (Fig. 3). Initial studies using the prosegment of SPC1, have demonstrated a potent inhibition for SPC1 (K i ϳ 14 nM) when tested in vitro (6). Others have also shown the potent inhibitory effect of the prosegment of SPC3 on SPC3 as well as on SPC1 (7). These observations thus raised the question as to the specificity of each prosegment when used as an inhibitor in trans. In a preliminary attempt to address this issue, the prodomains of SPC1 and SPC7 as well as peptide fragments thereof were tested on concentrated media from cells infected with recombinant SPC vaccinia virus (33). This study concluded that SPC prosegments were indeed potent, but also specific inhibitors when used in trans. The sum of these data imply that each prodomain would in fact be a highly specific inhibitor of its cognate enzyme. In light of the implication of these observations, we carried out a detailed comparative analysis of at least six of the seven prodomains on three distinct highly purified enzyme preparations.
Since the C-terminal of the prodomain is essential for enzymatic inhibition (6 -7, 38), we prepared seven short C-terminal propeptides (i.e. 12 aa) for initial experiments. These peptides proved to retain a high inhibitory potency when tested against their cognate enzyme (Table IV). For example, hSPC1-S inhibited hSPC1⌬ with a K i of 184 nM, hSPC6-S inhibited hSPC6A with a K i of 23 nM and hSPC7-S inhibited hSPC7⌬ with a K i of 69 nM. However, these propeptides were also potent inhibitors of the other SPC enzymes as well. For example, the hSPC1-S propeptide was a more potent inhibitor of hSPC6A (K i ϳ27 nM) than of hSPC1⌬ (K i ϳ184 nM) itself. Furthermore, hSPC1-S, hSPC4-S, mSPC5-S, and hSPC6-S were also equipotent inhibitors of hSPC1⌬ (K i values ranging from 123-184 nM) as well as of hSPC6A (K i values ranging from 12-27 nM). In contrast hSPC2-S and mSPC3-S (derived from the neuroendocrine cellspecific hSPC2 and mSPC3) were very poor inhibitors with K i values greater than 1 M. The reason for this lack of inhibitory potency may be due to the lack of basic residues at the P 5 or P 6 positions (Fig. 3). In general, these data suggest that the short  propeptides are not very specific SPC inhibitors.
There are several reasons that explain the lack of inhibitory specificity of each short propeptide. Some of the molecular determinants may be absent in the chosen sequences. In fact, some specificity could potentially be found in the PЈ positions of the prodomain primary cleavage site (Fig. 3). We therefore conducted a similar analysis on our three enzyme preparations using long propeptides (L) extended at the C-terminal. Even though the long propeptides have six amino acids in the PЈ position as compared with the short propeptides, identical results were obtained, both in terms of potency and specificity (Table IV). Thus, our results suggest that the amino acids in the PЈ positions do not contribute to the potency and specificity of inhibition of SPCs. However, it is noted that since these extended propeptides are equipotent to the short propeptides, they could in fact be useful in strategies in which such inhibi-tors are used ex vivo or in vivo. The amino acids in the PЈ positions would thus serve to protect the peptide from carboxypeptidases that would remove the critical C-terminal basic residues required for SPC inhibition. Indeed, our assays using the short propeptides were all conducted in vitro, without the presence of contaminating carboxypeptidases. The addition of carboxypeptidase E or carboxypeptidase B to our assays resulted in a completely abolished inhibition of the short propeptides (data not shown). This is consistent with previous reports showing that substitution of the C-terminal basic amino acids results in a complete loss of inhibitory potency by the prodomains (33,45).
The lack of specificity of our short propeptides could also be due to the absence of an extended N-terminal sequence. We therefore tested six different complete prodomains on the three enzyme preparations. Our results demonstrate that complete prodomains are highly potent inhibitors, with K i values often in the low nanomolar range (0.1-25 nM). However, no significant improvement in specificity was observed as compared with the short prodomain peptides (Fig. 7). Our data indicate that inhibitory specificity is not determined by the N-terminal region of the prodomains. However, in comparison to the short propeptides, complete prodomains were more potent inhibitors, sometimes 100-fold better. Two possible reasons could explain the higher potency of the complete prodomains. First, the short prodomain peptides (i.e. only 12 amino acids) adopt a random coil structure (analysis by NMR, data not shown), and thus addition of the N-terminal region may add structure and con-formation to the polypeptide to provide a better fit at the enzyme subsites (i.e. improves affinity, but not necessarily specificity). In support of this we have shown by CD analysis that most of the prodomains, except for hSPC2 prodomain, do have extensive secondary structure (Fig. 8). Furthermore, the recent studies on the stability and global fold of the mSPC3 prodomain (1,(27)(28)(29) suggested that an N-terminal ␤-strand can stabilize the C-terminal region into a ␤-strand through the formation of a ␤-sheet. Secondly, secondary binding sites could be present in the N-terminal region of the prodomains. These   6. Inhibitory potency of Ala-scan-substituted short propeptides. Ala-substituted hSPC1 and hSPC7 short propeptides were synthesized and tested on hSPC1⌬ and hSPC7⌬ enzyme preparations, respectively. The K i values were plotted on a logarithmic scale. A shows the inhibition potency of each modified hSPC1 short propeptides on hSPC1⌬. Basic residues are essential for hSPC1⌬ inhibition by the hSPC1 prodomain. B shows the inhibition potency of each modified hSPC7 short propeptides on hSPC7⌬. Hydrophobic and basic residues are essential for hSPC7⌬ inhibition by the hSPC7 propeptide.  sites may produce allosteric interactions that could improve the affinity of the interaction of the C-terminal region within the catalytic pocket. Alternatively, a portion of the N-terminal extension could be involved in binding to an exosite that provides additional interactions. Indeed, the determined experimental structure of the bacterial prodomain in complex with the subtilisin BPNЈ shows interactions between the ␤-sheet of the prodomain and two helices near the catalytic domain (46,47). In sum, the present data do not permit us to distinguish between these different possibilities; however, we can state that specificity was not improved by the addition of the Nterminal extension of the prodomains. The sum of our results indicates that the C-terminal region contains the primary information necessary to produce potent inhibition of SPCs. We therefore conducted an Ala-scan experiment of two of the short propeptides, hSPC1-S and hSPC7-S, in an attempt to delineate the molecular determinants important for their action, other than the essential C-terminal basic residues. We chose to study these two peptides since they showed a significant difference in specificity when comparing their actions on hSPC1⌬ and hSPC7⌬. Ala-substitutions of the hSPC1-S propeptide revealed that the basic residues in the P 4 (Ala-9), P 5 (Ala-8), or P 6 (Ala-7) positions are critical to inhibit hSPC1⌬. These data are in good agreement with recent data showing that polyarginine peptides are very potent inhibitors of mSPC1 (41). Indeed the C-terminal hSPC1 prodomain contains five basic residues out of six in the P 1 -P 6 positions. Ala-substitutions of the hSPC7-S propeptide also revealed that the basic residues in the P 4 (Ala-9) and P 5 (Ala-8) positions were very important. However, we also noted that the hydrophobic residues (Leu) in the P 6 (Ala-7) and P 7 (Ala-6) positions had a significant effect on the inhibitory potency for hSPC7⌬. These two hydrophobic residues most likely provide a certain degree of specificity between the hSPC1-S and hSPC7-S propeptides for the two parent enzymes. Further support for the importance of hydrophobic residues at the P 6 or P 7 positions is provided by our data showing that the short propeptide mSPC5-S is the second best inhibitor of hSPC7⌬ (Fig. 5C). In contrast to the other short propeptides, the mSPC5-S short propeptide has a Leu at the P 7 position.
Taken together, the experiments described demonstrated that although prodomains were very potent inhibitors (low nM range), they did not display highly specific properties within the SPC family of enzymes. In light of our present understanding of the importance of the C-terminal regions of the prodomains, this is not necessarily surprising since these regions have the highest degree of similarity. While our studies were entirely conducted in vitro, the data suggested that the use of prodomains as inhibitors in trans, possibly using expression systems, could result in inhibition of more than one SPC at a time. This is particularly important since most, if not all, cell lines examined to date express at least two or more SPCs endogenously (1,(27)(28)(29). Inhibition of an endogenous SPC using an expressed prodomain will also be difficult to control since various cellular expression systems do not control the exact levels of expression nor necessarily control the intracellular targeting of the expressed prodomain. It is clear that having a specific inhibitor for each SPC would have a tremendous usefulness to better dissect the function of each convertase within cellular systems or in vivo. Having the ability to inhibit a specific convertase with at least three log units more potency than another convertase would resolve the issue of cellular expression levels. The development of more specific inhibitors may be possible using the C-terminal sequences; however, various modifications would be required, and the success of this approach is not assured especially if SPCs are in fact very similar in their catalytic functionality. Alternatively, such potent inhibitors with broader specificity profiles may be more appropriate to target enzymes directed to the constitutive secretory pathway. This could be advantageous in the case where redundant SPC cleavage functions have been reported, such as that recently described for the ␤-secretase enzyme (48). FIG. 7. Complete prodomains are more potent inhibitors than short propeptides but do not display significant specificity. Graphical comparison of the inhibition potency and specificity of the hSPC1, hSPC6, and hSPC7 short propeptides and complete prodomains on hSPC1⌬, hSPC6A, and hSPC7⌬. A presents the K i values of hSPC1-S, hSPC6-S, and hSCP7-S propeptides on hSCP1⌬ (in black bars), hSPC6A (in white bars), and hSPC7⌬ (in gray bars). B represents the K i values of complete prodomains hSPC1, hSPC6, and hSPC7 on hSPC1⌬ (in black bars), hSPC6A (in white bars), and hSPC7⌬ (in gray bars). . The hSPC1, mSPC3, mSPC5, and hSPC7 prodomains have almost identical structure profiles, while the hSPC6 prodomain exhibits a high degree of ␣ helical structure. The hSPC2 prodomain is distinctly different from the other SPC prodomains and shows very little secondary structure.