Inhibitory Specificity and Potency of proSAAS-derived Peptides toward Proprotein Convertase 1*

Prohormone convertase 1 (PC1), mediating the proteolytic processing of neural and endocrine precursors, is thought to be regulated by the neuroendocrine protein proSAAS. The PC1 inhibitory sequence is mostly confined within a 10–12-amino acid segment near the C terminus of the conserved human proSAAS and contains the critical KR 244 dibasic motif. Our results show that the decapeptide proSAAS-(235–244) 235 VLGALLRVKR 244 is the most potent reversible competitive PC1-inhibitor ( K i (cid:1) 9 n M ). The C-terminally extended proSAAS-(235– 246) exhibits a 5–6-fold higher K i ( (cid:1) 51 n M ). The addi-tional LE sequence at P1 (cid:1) -P2 (cid:1) , resulted in a competitive substrate cleaved by PC1 at KR 244 2 LE 246 . Systematic alanine scanning and in some cases lysine scanning tested the contribution of each residue within proSAAS-(235–246) toward the PC1-inhibition’s

Proprotein convertases (PCs), 1 a family of Ca 2ϩ -dependent mammalian subtilases, are known to mediate the proteolytic processing at selected sites of many precursor proteins into their functionally active forms (1,2). These sites are generally composed of a pair of basic amino acids within the consensus sequence of R/K/H-X n -R2, where n ϭ 0, 2, 4, or 6 and X represents any amino acid except cysteine. Numerous potential substrates have so far been identified for PCs. These include hormonal peptides and growth factors, their receptors, cell surface proteins, bacterial toxins, envelope viral glycoproteins, enzymes, transcription factors, and others (1,2). The delicate balance between cleaved functional proteins and their precursors is critical for normal growth, function, metabolism, and development as well as in pathophysiologic conditions (1)(2)(3)(4)(5).
All PCs are initially synthesized as inactive zymogens that must be proteolytically activated through the autocatalytic cleavage of their inhibitory N-terminal prosegment. A number of studies have already revealed this unique property of prodomains in the regulation of enzymatic activity (6 -9). However, at least for the neuroendocrine convertases PC1 and PC2, their cellular activity is controlled by endogenous inhibitors. Thus, co-localization, in situ hybridization, and other biochemical studies revealed that the production of enzymatically active PC2 requires the presence of a binding protein 7B2 (10), which also serves as a specific temporal endogenous inhibitor of this enzyme (for reviews, see Refs. 1, 2, and 11). 2 Subsequent deletion and alanine-scanning studies identified the inhibitor segment as a 16-aa fragment of the 31-aa C-terminal domain of 7B2 (13).
Recently, a granin-like 26-kDa (258 -260-aa) neuroendocrine secretory protein, called proSAAS, was identified as a specific PC1 inhibitor (14). It is interesting to note that while proSAAS and 7B2 are not homologous, they are of similar size, with an N-terminal proline-rich region; both contain several pairs of basic amino acids and are broadly expressed in neural and endocrine tissues. Very recently, proSAAS was shown to specifically inhibit PC1 and not furin, PC2, PC5, or PC7 (15). The processing profile of proSAAS was recently reported and revealed that it is cleaved in a tissue-specific fashion at its C terminus into smaller inhibitory peptides (15)(16)(17). The inhibitory segment was mapped to a short 6 -12-aa sequence near its C terminus that contains a critical Arg 244 (human proSAAS nomenclature) located at the processing site 235 VLGALLRVKR2LE 246 (15,17). Interestingly, the peptide LLRVKR was previously identified from a combinatorial peptide library screen aimed at identifying specific PC1 inhibitors (18). The in vitro inhibition of PC1 by these short peptides was at least 17-fold better than with the * This work was supported by Canadian Institutes of Health Research (formerly Medical Research Council (MRC)) Program Grant PG-11474 (to N. G. S. and M. C.) and by the Government 1 The abbreviations used are: PC, proprotein/prohormone convertase; aa, amino acid(s); Fmoc, N-(9-fluorenyl)methoxycarbonyl; RP-HPLC, reversephase high performance liquid chromatography; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight; BTMD, before transmembrane domain; Mes, 2-(N-morpholino)ethanesulfonic acid; AMC, 7-amino 4-methylcoumarin; MCA, 4-methyl 7-aminocoumarinamide. full-length proSAAS (15). Furthermore, it was reported that the 66-kDa processed form of PC1 is better inhibited by proSAAS than is its 87-kDa precursor (15) (Fig. 1).
In this article, we present detailed kinetic studies on the specificity and potency of PC1 inhibition by the above 12-mer proSAAS peptide and its mutants. Data obtained for the wild type sequence as well as from alanine and lysine scans allowed the identification of the critical aa within this sequence. The specificity and potency of these peptides was also tested against other PCs such as the mammalian furin, PACE4, PC5, PC7, and yeast kexin. The 12-mer proSAAS peptides were compared with that of rat proSAAS-(221-254), also known as little PenLen, which is one of the major processing forms of proSAAS in AtT20 cells (14). Finally, circular dichroism and molecular modeling studies were used to correlate the secondary structure of the inhibitory peptides to their potency of PC1 inhibition.
Production of Recombinant PCs-All of the recombinant forms of PCs were produced by using the vaccinia virus constructs of soluble human furin-BTMD (before transmembrane domain): human PACE4, mouse PC1, mouse PC5A, rat PC7-BTMD, and yeast kexin-BTMD. The enzymes were recovered from serum-free culture media as reported previously (15,19). The recombinant mPC1 used in the present study is obtained from expression of its full-length cDNA, but it contains mostly the C-terminal truncated and enzymatically more active 66-kDa form and some 74-and 87-kDa form as well.
Enzyme Assay-All enzyme assays were performed with the fluorogenic substrate pyroglutamyl-Arg-Thr-Lys-Arg-4-methyl-coumaryl-7amide (Peptides International, Louisville, KY) at pH of either 7.4 (furin) or 6.5 (for other PCs). The assay buffer in all cases was composed of 25 mM Tris, 25 mM Mes, 2.5 mM CaCl 2 . The concentration of the substrate was maintained at 100 M unless otherwise mentioned. The amounts of enzymes used were adjusted so as to give approximately the same hydrolytic activity (3.9 -4.5 nmol of AMC released/h of incubation) in an aliquot of 5 l. The release of fluorescence was monitored for 6 h using a spectrofluorometer (Gemini, Molecular Probes, Inc., Eugene, OR) at excitation and emission wavelengths of 370 and 460 nm, respectively). Enzymatic activities were measured either from raw fluorescence readings in end time assay or from the progress curves.
Progress Curve for Enzyme Inhibition-The fluorescence of released AMC was measured on-line with a spectrofluorometer every 60 s up to 60 min, and the slope of each curve was assessed with the computergenerated highest point fit.
Determination of Inhibition Constant, K i , and IC 50 -For K i determination, various peptide concentrations (0.12 nM to 700 M) were incubated at 37°C with respective enzyme (5 l) in the above described buffer (100 l) in the presence of at least two different concentrations of fluorogenic substrate, pERTKR-MCA (100, 50, 25 or 12.5 M) (7,15). The precise concentration range was selected so as to produce an inhibition of 20 -80% of initial activity. For all kinetic measurements, the peptides were preincubated with enzyme for 15 min prior to the addition of the substrate. All assays were performed in duplicate for two independent experiments on a 96-well microplates (flat bottom, black; Dynatec). K i was estimated from Dixon plots, while for IC 50 value, a double reciprocal Lineweaver-Burk plot was used (15,17,19). Both initial rate and end time assays were used, and the values obtained for the measured parameter were averaged.
Cleavage of proSAAS Peptides by PCs-Each proSAAS peptide including the little PenLen (20 M) was incubated for 6 h or overnight at 37°C with PC1, furin, PACE4, kexin, PC5, or PC7 (5 l) in buffer (100 l) as described. The enzyme digests were separated by RP-HPLC using a C 18 analytical column with a diode ray detector (Varian, Prostar, CA), and the peaks were analyzed by MALDI-TOF mass spectrometry.
Comparative Analysis of PC Inhibition by proSAAS Peptides-Each proSAAS peptide (750 nM or 25 M) was preincubated with enzyme (5 l) for 15 min in buffer (100 l) before the addition of fluorogenic substrate, pERTKR-MCA (100 M). Fluorescence readings were measured following overnight incubation and compared with the control experiment run in parallel without the peptide.
Effect of Preincubation Time on Enzyme Inhibition-For this study, a representative peptide (proSAAS-(235-244), Table I) (20 M) was preincubated with PC1 (5 l) for 0, 5, 10, 15, and 30 min in buffer (100 l) before the addition of substrate pERTKR-MCA (100 M). Control experiments were run in parallel without the peptide, and the fluorescence readings were measured after 6 h of reaction.
CD Analysis-All CD measurements were carried out with a JASCO J-810 spectropolarimeter instrument (Easton, MD) using a 100-l solution of each peptide in distilled water (concentration 0.2-0.5 mg/ml) in a quartz cell (1-mm path length) as described (20). The final corrected CD spectra were obtained by subtracting the spectrum obtained with control water from those of crude samples.
Molecular Modeling Study-Three-dimensional theoretical structures of proSAAS peptides and their mutants were generated by computer software hyperchem (version 5.0; Hypercube) with Robek-Polard energy minimization carried out at ambient temperature.

Inhibition Constant (K i ) of proSAAS Peptides against PCs-
Based on earlier studies (15), we selected the 12-mer human/ rat proSAAS 235 VLGALLRVKRLE 246 (same as in proSAAS-(233-242) in mouse sequence) as a model peptide for detailed kinetic analysis and positional scanning. As shown in Fig. 1, the N-terminal 10 aa of this inhibitory peptide represent the C terminus of the natural processing product obtained by PC1 cleavages at PRRLRR 220 and LLRVKR 244 and contain the critical Arg 244 (15,17). Table I lists all of the 19 peptides synthesized, including little PenLen, and analyzed in this work. These include the wild type sequence and its 11 Ala derivatives; three Lys mutants at P1, P4, and P1 ϩ P4; and a P3 ϩ P5 double Ala mutant. Finally, in order to compare the effects of PЈ residues on inhibition, we also synthesized the wild type proSAAS-(235-244) 235 VLGALLRVKR 244 as well as its all-dextro derivative.
The measured K i of all of the above proSAAS peptides against PC1 and in some cases against furin, PC5, and PC7 are presented in Table II. The competitive nature of inhibition of PC1 by both wild type proSAAS-(235-246) ( Fig. 2A) and its mutants (not shown) as well as proSAAS-(235-244) (Fig. 2B) was demonstrated by Dixon plots conducted at four different concentrations of the fluorogenic substrate pERTKR-MCA. The near linear regression and the presence of a single point of intersection are evident. However, a close examination of the graphs revealed a slight hyperbolic nature. This may perhaps be due to some artifact, to the presence of multiple PC1 forms in the enzyme preparation, or alternatively to the adsorption of the peptide or enzyme to the sample tubes. The competitive nature of this inhibition was further confirmed by an observed linear increase of IC 50 with substrate concentration (not shown) (21)(22)(23).
Overall, the data indicated that the 10-mer 235 VLGALL-RVKR 244 lacking any PЈ residues was the most powerful PC1 inhibitor (K i ϳ9 nM). This value is 5.7-fold lower than the ϳ50.5 nM calculated for the K i of the C-terminally extended 12-mer 235 VLGALLRVKRLE 246 (Table II). Although both peptides are selective PC1 inhibitors, the C-terminal Leu-Glu extension resulted in an enhancement of selectivity for PC1 inhibition by 27-and 57-fold as compared with furin and PC5, respectively. Interestingly, the ϳ100-fold lower inhibitory potency toward PC7 was not affected by the dipeptide insertion. Noticeably, Ala mutation of either the P1Ј Leu or the P2Ј Glu resulted in a drastic loss of potency and selectivity toward PC1 inhibition. This means that the nature of these residues is critical for the observed high PC1-selective inhibitory property of proSAAS-(235-246). Finally, it is worth mentioning that the Ala mutation of P1Ј Leu resulted in a complete loss of PC1 selectivity as compared with furin, whereas the P2Ј Glu to Ala mutation had a 14-fold lesser effect (Table II). This is reminiscent of the observed critical importance of P1Ј Leu for the processing of prorenin that is cleaved by PC1 but not at all by furin (24). Finally, we note that the loss of selectivity toward PC1 versus PC5 and PC7 is somewhat similar for either Ala mutants but not as drastic as the P1Ј mutation for furin.
In order to extend those data and include other convertases such as human PACE4 and yeast kexin, in Fig. 3 we present in a bar graph format the extent of enzyme inhibition by 750 nM proSAAS peptides and their selected mutants. For this purpose, we used a 4-h stop time assay in which the level of each enzyme used was adjusted so as to give similar initial pER-TKR-MCA activity. In agreement with the K i data (Table II), the shorter 10-mer is more potent than the 12-mer proSAAS peptide on PC1, furin, PC5, and PC7 (Fig. 3, first row). In contrast, PACE4 is not inhibited by either peptide even at this high concentration, whereas kexin is ϳ50% inhibited by the 12-mer peptide and not at all by the 10-mer one. The chirality of inhibition is evident by the fact that the all-dextro derivative of the 10-mer peptide does not significantly affect the activity of any enzyme tested. Although not shown, Ala mutations at P1Ј and P2Ј follow the expected pattern from Table II (i.e. they are both important for the selective inhibition of PC1).
The critical importance of the three basic residues at P1, P2, and P4 is also evident, since their Ala mutants lost most of their inhibitory properties against all enzymes (Table II, Fig. 3,  second row). Furthermore, Arg to Lys mutations at either P1, P4, or P1 ϩ P4 resulted in a drastic loss of PC1 inhibition, especially for the P4 or P1 ϩ P4 mutations (Table II). These data attest to the critical importance of Arg at the P1 and P4 positions and a basic residue at P2.

TABLE I
List of various proSAAS-derived and -related peptides All amino acids are in L-configuration except for peptide 14, where all amino acids are in D-configuration (indicated by the front letter "d" and the lower case letters for each amino acid residue (*)), all mutations of amino acids are indicated in boldface type and underlined. "h" and "r" prefixes represent human and rat, respectively. MW, molecular weight. In addition, the aliphatic residues Val at P3 and P10, and Leu at P5, P6, and P9 also contribute to the inhibitory potency and selectivity of proSAAS-(235-246) toward PC1, with the P9 Leu and P3 Val positions being the most critical (Table II, Fig.  3, third and fourth rows). We also tested whether double P3 Val and P5 Leu to Ala mutations could further influence the inhibitory selectivity of proSAAS-(235-246) toward PC1 versus other convertases. Interestingly, the double mutant is a better inhibitor of furin (K i ϳ127 nM) as compared with PC1, PC5, and PC7 (Table II).
Digestion of proSAAS-(235-246) and Its Ala Mutants by PCs-Previously, it was demonstrated that radioiodinated Tyr ϩ mouse SAAS 219 -258 was internally cleaved by PC1 (17), suggesting that PC1 is inhibited by proSAAS by a mechanism similar to that of PC2 by 7B2 (13,25). However, the exact cleavage site of the PenLen peptide was not determined in this previous study (17). To determine this cleavage site as well as to examine whether human proSAAS-(235-246) behaves as a competitive substrate, it was incubated for 18 h with PC1 in the presence of the fluorogenic substrate pERTKR-MCA. The RP-HPLC of the crude digest and the mass spectral data of the isolated peaks are shown in Fig. 4. Unlike the fluorogenic peptide, which was partially cleaved by PC1, the peptide proSAAS-(235-246) (R t ϭ 33.7 min; (M ϩ H) ϩ ϭ 1,369) was completely digested, giving the expected decapeptide VLGALL-RVKR 244 2 (R t ϭ 21.7 min; (M ϩ H) ϩ ϭ 1,123). The C-terminal dipeptide LE product eluted with the injection peak and was not analyzed. The two other peaks at R t ϭ 23.1 and 15.1 min were characterized by mass spectrometry as the unreacted fluorogenic substrate pERTKR-MCA ((M ϩ H) ϩ ϭ 830) and its N-terminal product pERTKR-OH ((M ϩ H) ϩ ϭ 668).
Using a similar approach, we tested the cleavage of the Ala mutants of the 12-mer proSAAS-(235-246) peptides as well as the PenLen at 20 M by PC1, furin, PC5, PACE4, and PC7, under identical initial pERTKR-MCA activities (Table III). Overall, the data show that, following 6-h incubation, PCs cleaved the individual proSAAS-(235-246) derivatives only at amino acid P1 Arg 244 (except for its Ala mutant) with varying efficiencies. The wild type 12-mer sequence is best cleaved by PACE4 (100%) and least so by PC1 (20%), but PenLen containing this peptide in its internal sequence is most efficiently cleaved by PC1, poorly (15%) by PACE4, and not at all by either  proSAAS-(235-244) (B). The graphical plots were obtained with data using mPC1 (5 l) and pERTKR-MCA as substrate at concentrations 12.5 (S1), 25 (S2), 50 (S3), and 100 M (S4). The inhibition study was conducted with a 15-min preincubation between the enzyme and the inhibitor, following which the substrate pERTKT-MCA was added. The fluorescence readings were measured after a 6-h reaction at 37°C (see "Experimental Procedures"). hproSAAS, human proSAAS. furin, PC5, or PC7. Interestingly, the P2Ј Ala mutant was found to be a better substrate compared with the wild type for all PCs except PACE4, while the P1Ј Ala mutant was a worst substrate for all enzymes. As expected, the P1 Arg to Ala mutant was not cleaved by any convertase. In contrast, the P4 Arg to Ala mutant, although not cleaved by PC1 and less cleaved by furin, PC5, and PC7, is still quite efficiently cleaved by PACE4 (Table III). The last observation is in accord with a report that shows that a basic residue at P4 (e.g. Arg) is not an essential requirement for PACE4's ability to cleave fluorogenic substrates (26). The P2 Lys to Ala mutant, although leading to a better substrate for PC1 and furin, is paradoxically a worse one for the other enzymes.

. Dixon plots for inhibition of mPC1 by proSAAS-(235-246) (A) and
The Ala mutations of the P8 Gly and the aliphatic residues at P3, P5, P9, and P10 demonstrated that, except for PC5, some of them are critical for the ability of the other PCs to selectively process these peptides (Table IV). Particularly, PACE4 is very sensitive to these Ala mutations at all of the above residues, including a double P3, P5 one. Furthermore, P3 Val, P5 Leu, and P8 Gly generally enhance processing by PC1, furin, and PC7, while the P9 Leu is not favorable for PC1 processing and has no effect on either furin, PC5, or PC7. Finally, P10 Leu enhances processing by furin. Thus, for PC1, among the aliphatic and/or neutral residues only three selectively affect its ability to process proSAAS-(235-246), with P3 Val and P8 Gly enhancing processing and P9 Leu diminishing it as compared with their Ala mutants. tion. Accordingly, for a fixed inhibitor concentration, a 15-30min preincubation period of enzyme and proSAAS-(235-244) is necessary before achieving maximal PC1 inhibition (Fig. 5,  inset). This indicates a slow and tight binding inhibition of PC1 by proSAAS-(235-244) (7,8). Using a 15-min preincubation, progress curves of PC1, furin, PACE4, and PC7 activity in the presence of various concentrations of the 10-mer proSAAS-(235-244) are depicted in Fig. 6. Accordingly, at a fixed inhibitor concentration, the degree of PC1, furin, PACE4, and PC7 (Fig. 6) as well as PC5 and kexin (not shown) inhibition increases linearly with time. The slow tight binding inhibition is also deduced from the characteristic observation that IC 50 of proSAAS-(235-244) increased linearly with the enzyme amount (not shown) (22,23).
Circular Dichroism and Molecular Modeling-The secondary structures of proSAAS-(235-244) and proSAAS-(235-246) was investigated by circular dichroism (Fig. 7, A and B). The data indicated that these potent PC1 inhibitors (Table II) exhibited a predominantly poly-L-proline II type conformation (27) in aqueous solution at pH 6.8, whereas this structure was completely destroyed in both the inactive P1 Arg to Ala mutant (Fig. 7C) and in the all-dextro proSAAS-(235-244) peptide (Fig.  7D). The presence of some helical structure is also noticeable. Studies with the other Ala mutants of proSAAS-(235-246) revealed varying degrees of poly-L-proline II type conformations nearly in proportion to the extent of PC1-inhibitory potency deduced from the K i values in Table II (not shown). In order to extend this conclusion, we performed molecular modeling studies, using minimum energy calculations (Hyperchem software, version 5) aimed at predicting the possible three-dimensional structures of the various proSAAS peptides. The theoretical structures obtained suggest that while proSAAS-(235-246) is predominantly found in an extended conformation, the inactive P1, P2, and P4 mutants mostly exhibit ␤-turn structures (not shown). The above structural features, namely poly-L-proline II type and extended conformations may thus be important parameters affecting the ability of proSAAS to inhibit PC1. DISCUSSION The most important finding of this study is that the highly conserved 10-or 12-mer proSAAS peptides, namely proSAAS-(235-244) or proSAAS-(235-246) encompassed within the sequence 235 VLGALLRVKRLE 246 are highly potent and selective inhibitors of PC1. Alanine scanning of this peptide at all positions and lysine mutations at selected sites revealed the inhibitory profile played by the various P and PЈ aa. For Ala mutations, substitution of P1 Arg was most critical followed by P2 Lys and P4 Arg, respectively. These data agree with those predicted from modeling studies by Lipkind et al. (28) and Siezen et al. (29), whereby specific Glu and Asp residues in the S1, S2, and S4 subsites of PC1 are thought to intimately contact these basic residues. Among the aliphatic residues at P3, P5, P6, P9, and P10, all except P6 made significant contributions toward PC1 inhibition (Table II). Although generally agreeing with the results of positional scanning synthetic peptide combinatorial 6-mer peptide library studies (18), our data differ with respect to the critical importance of P6 Leu. Thus, our data reveal that within the context of the 12-mer proSAAS-(235-246) peptide, P6 Leu is not critical for selective PC1 inhibition and/or recognition ( Fig. 3 and Table II). The difference might be attributed to the 12-mer peptide used in this study, which is extended both N-terminally (up to P10) and C-terminally (up to P2Ј), as compared with the N-acetylated 6-mer used by Apletalina et al. (18). Among the aliphatic residues, the P9 Leu was found to be the most critical since its substitution by Ala led to a ϳ15-fold decrease in PC1 inhibition. Interestingly, alignment of the inhibitory prosegment of the PCs revealed a strict conservation of a P9 Gln among all members, possibly implicating this site in PC inhibition (30). The next critical hydrophobic residue is the P3 Val, since its replacement by Ala led to a ϳ7-fold increase in K i . Interestingly, the P3 position is occupied by Val in the inhibitory prosegment of PC1 and PC4 and by Ala in that of furin and PC7 (30). Ala replacements at P5 and P10 resulted in only 3-fold increase in K i , suggesting a relatively lesser contribution of these residues toward PC1 inhibition. Thus, the P9 Leu in proSAAS-(235-246) plays a critical discriminating role in PC1 recognition, as opposed to other PCs. In addition, our data also predict that hydrophobic/neutral residues at positions P3, P5, and even at distant positions such as P8 and P10 may further enhance the selective recognition of proSAAS-(235-246) by PC1.
It should be pointed out that the recombinant mPC1 used in the present study and obtained from expression of its fulllength cDNA, contains mostly the C-terminally truncated and enzymatically more active 66-kDa form but that some 74-and 87-kDa forms are also present. The use of homogeneous and pure 66-kDa PC1, which is more likely the functional form in neuroendocrine granules, may lead to quantitative differences that may differ somewhat from those presented in this study.
Another interesting outcome of this study is that the elimination of PЈ residues from proSAAS-(235-246) led to an increased (ϳ6-fold) potency but decreased selectivity of PC1

PC1 Inhibition by proSAAS Peptides
inhibition. Thus, while proSAAS-(235-244) inhibited PC1 780-, 1,980-, and 94-fold more efficiently than furin, PC5, and PC7, proSAAS-(235-246) inhibited PC1 29-, 35-, and 103-fold more efficiently than the above convertases, respectively (Table II). This suggested that the aa occupying the P1Ј and P2Ј positions as well as the P9 one (see above) play major discriminatory roles in the potent selective inhibition of PC1. In addition, our data suggest that the P1Ј Leu in proSAAS-(235-246) is very critical to discriminate between furin and PC1 (Table II and Fig. 3), an observation already reported for human prorenin processing (24), and would be predicted for the leptin proreceptor (1). This information may be further exploited in the future in order to engineer more potent and specific PC1 inhibitors and/or substrates. The other interesting observation is that the little PenLen peptide which comprises aa 221-254 of rat proSAAS sequence exhibited a strong but less efficient inhibition of PC1 compared with the conserved shorter proSAAS-(235-244) and proSAAS-(235-246). Moreover, PenLen was found to be less discriminatory toward PC1 than either proSAAS-(235-244) or proSAAS-(235-246). Thus, whereas PenLen is selectively cleaved by PC1 (Table III), it is not as selective an inhibitor of PC1 as compared with the 12-mer proSAAS-(235-246), especially with respect to PC5 (Table II). Finally, the observation that the 34-mer proSAAS-(221-254) (PenLen) is more efficiently processed than the 12-mer proSAAS-(235-246) (Table III) suggests that residues before and after the 12-mer sequence may enhance the ability of PC1 to process the KR 244 2. This reinforces the notion that the potency and selectivity of PC1 inhibition of proSAAS is mostly located around the dibasic site KR 244 , representing the junction between Pen and Len (16). Interestingly, this motif is found between two other dibasic sites, RR 220 and RR 256 , the processing of which leads to the formation of PenLen. The latter peptide was found to be one of the processed forms of pro-SAAS in AtT20 cells (14), and it is likely that PenLen is also a processing intermediate in rat brain and pituitary (16). Recent data showed that in Cpe fat /Cpe fat mice, the processing of proSAAS is slightly impaired relative to wild-type mice, resulting in the accumulation of partially processed peptides (16). Our in vitro data showing the inhibitory function of PenLen toward PC1 may rationalize the observed reduction in the level of PC1 activity in Cpe fat /Cpe fat mice.
Another important outcome of the present study is derived from circular dichroism data on various proSAAS peptides that suggested that a distinguishing poly-L-proline type IIlike conformation could be critical for the selective and potent inhibition of PC1. The implication of this secondary structural motif in PC1 inhibition in relation to pH and possible association with trace metal ions found in the secretory pathway such as Zn 2ϩ and Cu 2ϩ are currently under investigation.
The identification of the highly specific and potent PC1 inhibitors proSAAS-(235-246) and proSAAS-(235-244) may be useful in the development of an effective affinity procedure for the purification of PC1. Methodologies using peptide based inhibitors have been applied in the past to purify a number of serine proteases (31). So far, PC1 has only been purified partially using commonly used multiple chromatographic steps (15,32). Other applications of these inhibitors may include derivatization with either fluorescent or radioactive moieties as specific molecular markers of PC1 (33,34). In view of the potential implications of PC1 in early embryogenesis, preimplantation, and obesity (12), 3 development of selective inhibitors may help to define the role of PC1 in vivo. This is especially needed, since so far no viable PC1 (Ϫ/Ϫ) mice could be obtained. 4 In this respect, more work is needed to improve the cellular permeability and/or delivery of the designed inhibitors. Future work on the targeting of pro-SAAS peptides or expression of hybrid constructs inside the cell should provide further insights on the biological functions of this regulated critical convertase. FIG. 7. Circular dicroism spectra of various proSAAS peptides. The spectra were recorded in a 1-mm path length quartz cell in water (100 l), pH 6.8.