A 36-Residue Peptide Contains All of the Information Required for 7B2-mediated Activation of Prohormone Convertase 2*

The prohormone convertases (PCs) are serine proteinases responsible for the processing of secretory protein precursors. PC2 is the only member of this family whose activation requires intracellular interaction with a helper protein, the neuroendocrine protein 7B2. In order to gain a better understanding of the mechanism of proPC2 activation, we have characterized the structural determinants of 7B2 required for proPC2 activation. We had already identified a proline-rich binding determinant in the 21-kDa domain, the portion of 7B2 responsible for proPC2 activation. We have now investigated the function of the weakly conserved amino-terminal portion of 21-kDa 7B2 by sequential deletions. Mutant proteins were analyzed in four assays: binding to proPC2, facilitation of proPC2 maturation, and activation of proPC2 in vivo and in vitro. We found that the amino-terminal half of 7B2 is not involved in proPC2 activation, and we identified an active 36-residue peptide that contains the previously characterized proline-rich sequence as well as an α-helix and the only disulfide bond of 7B2. Mutation of the α-helix and of the cysteines demonstrated that these determinants are absolutely required for PC2 activation. Thus, the 186-residue full-length 7B2 rat protein can be functionally reduced to an internal segment of only 36 residues.

Endoproteolytic processing is one of the major post-translational modifications that hormones and neuropeptides precursors must undergo during their biosynthesis. A family of mammalian subtilisin-like endoproteases responsible for these processing events has been recently identified, the prohormone convertases (PCs) 1 (reviewed in Refs. 1 and 2). PC1 and PC2 are the prohormone convertases specific for neuroendocrine cells; both enzymes are active late in the secretory pathway, i.e. the trans-Golgi network (TGN) and the secretory granules.
Precursors such as proinsulin or proopiomelanocortin are processed sequentially, first by PC1, then by PC2 (reviewed in Ref. 3). In agreement with the ordered activity of these enzymes, PC1 and PC2 have different activation pathways (reviewed in Ref. 4). Whereas the propeptide of proPC1 is first processed in the endoplasmic reticulum (ER) (5)(6)(7), as are those of furin (8,9), PC5 (10), and LPC/PC7/PC8 (11), the proPC2 propeptide is processed only in the acidic compartments of the TGN/secretory granules. In addition, proPC2 is the only PC that specifically interacts with a helper protein, the neuroendocrine-specific protein 7B2 (12)(13)(14). We have demonstrated that this interaction is absolutely required for proPC2 activation, both in vivo in transfected AtT-20 cells (14) and in 7B2 null mice (16), as well as in vitro (17).
The neuroendocrine protein 7B2 was originally purified from pituitary extracts (18). It is an 185-residue precursor that is cleaved at a pentabasic site at residue 155 ( Fig. 1), most probably by furin (19 -21). The two domains generated by this cleavage have distinct functions; the amino-terminal domain of 21 kDa is sufficient for PC2 activation (14), and the COOHterminal domain of 31 residues is a potent inhibitor of PC2 activity (22,23). Based on the homology between the 90 aminoterminal residues with members of the chaperonin 60 family, 7B2 was originally proposed to act as a molecular chaperone specific for proPC2 (12). We have already demonstrated that the 90 amino-terminal residues alone cannot bind to proPC2 (14); they could nonetheless be involved in PC2 activation. Thus, the function of the amino-terminal half of 7B2 is not yet known. In order to analyze the role of the amino-terminal half of 7B2 and to better understand the involvement of 7B2 in PC2 activation, we have investigated the effect of amino-terminal sequential deletions. Deleted 7B2s were tested in four functional assays: binding to proPC2; facilitation of its maturation in stably transfected AtT-20 cells; and activation of PC2 both in vivo, by transient transfection in CHO cells, and in vitro, using Golgi subcellular fractions enriched in proPC2. As these experiments demonstrated that the 86 amino-terminal residues of 7B2 are not required for PC2 activation, we further investigated the minimum structural determinants of 7B2. We were able to identify a 36-residue peptide that contains the PPNPCP motif, as well as an ␣-helix and a disulfide bridge, which are all absolutely required for binding to proPC2 and for proPC2 activation.

MATERIALS AND METHODS
Cell Culture-AtT-20 cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies Inc.) containing 10% Nuserum (Becton Dickinson, Mountain View, CA), 2.5% fetal calf serum (Irvine Scientific, Santa Ana, CA), and the appropriate selection agents, 200 g/ml G418 (Life Technologies Inc.) and/or 100 g/ml hygromycin (Sigma). The AtT-20 cells stably expressing PC2 (AtT-20/PC2) were provided by R. E. Mains (Johns Hopkins University School of Medicine, Baltimore, MD) (24). The control AtT-20 cells that co-express PC2 and 7B2 have already * This work was supported in part by National Institutes of Health Grant DK49703 (to I. L.) and by the Fundacao de Amparo Pesquisa do Estado de Sao Paulo and Conselho Nacional de Desenvolvimento Cientifico e Technologico-PADCT Project 620498/98-6. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S. C been described (14). CHO cells were maintained in ␣-minimum essential medium without nucleosides, containing 10% dialyzed fetal calf serum (Irvine Scientific) and 50 M methotrexate (Sigma). CHO cells stably expressing PC2 (CHO/PC2), which have been amplified by the dihydrofolate reductase method, have already been characterized (25).
Peptide Synthesis-An automated bench-top simultaneous multiple solid-phase peptide synthesizer (PSSM 8 system from Shimadzu) was used for the synthesis of all the peptides by the Fmoc procedure in NovaSyn TGR resin (Novabiochem, San Diego, CA). Therefore, all peptides were obtained with the carboxyl-terminal carboxyl group in amide form. A trityl group was used as protection for the ␤-thiol group of cysteine, except for the synthesis of the peptide 7B2 (86 -121) containing methylated cysteines. In this case we used Fmoc Cys(S-Me)-OH. The final peptides were deprotected and cleaved from the resin with trifluoroacetic acid containing 0.5% triisopropylsilene, and purified by semipreparative HPLC on an Econosil C-18 column (10 m, 22.5 ϫ 250 mm) using a two-solvent system: A, trifluoroacetic acid/H 2 O (1:1000); and B, trifluoroacetic acid/acetonitrile/H 2 O (1:900:100). The column was eluted at a flow rate of 5 ml/min with a 10 (or 30) to 50 (or 60) % gradient of solvent B over 30 or 45 min. The molecular weight and purity of synthesized peptides were verified by matrix assisted laser desorption ionization-time of flight mass spectrometry (TofSpec-E, Micromass). For formation of the disulfides, the peptides were dissolved in water (at a final concentration of about 1 mM), the pH maintained at 7.5, and air oxidation was allowed to proceed with slow stirring. Aliquots were collected at different periods of time to monitor the oxidation of the two cysteines by HPLC and by Ellman's assay.
Purification of Recombinant 7B2s-Recombinant His-tagged 7B2s were prepared using the QIAexpress system (Qiagen Inc., Chatsworth, CA). The DNAs coding for the deleted 7B2s were generated by PCR using the following primers: a common carboxyl-terminal primer (5Ј-GCGGCAAGCTTCTACTGTCCTCCCTTCATCTT-3Ј) and the following amino-terminal primers: for the 30 -150 construct, 5Ј-CGCGGATC-CCCACGTGTGGAGTACCCA-3Ј; for the 68 -150 construct, 5Ј-GCCG-GATCCATCGTGGCAGAGTTG-3Ј; for the 86 -150 construct, 5Ј-CGCG-GATCCTACCCAGACCCTCCAAAT-3Ј. The PCR fragments were cloned in pQE30 at the BamHI and HindIII restriction sites. All sequences were verified by DNA sequencing. Proteins were expressed in Escherichia coli XL1-Blue (Stratagene, La Jolla, CA) and purified with the guanidine HCl method as described previously for 21-kDa 7B2 (26). The formation of an actual disulfide bridge within recombinant 7B2 was verified by performing SDS-polyacrylamide electrophoresis under reducing and non-reducing conditions. A distinct molecular weight shift between these two conditions was detected, indicating the presence of a disulfide bridge in recombinant 7B2. Furthermore, the fact that we observed only one band under non-reducing conditions indicates that 100% of the recombinant 7B2 contains this disulfide bond.
Metabolic Labeling and Immunoprecipitation-In order to investigate the binding of 7B2 to PC2, AtT-20 cells were metabolically labeled with 0.5 mCi/ml [ 35 S]methionine/cysteine (Promix, Amersham Pharmacia Biotech). The cells were then lysed in 1% Triton X-100 in 25 mM Tris, pH 7.4, containing 100 mM NaCl, 5 mM EDTA, 0.5 mM p-chloromercuriphenylsulfonic acid (Sigma) and 1 mM phenylmethanesulfonyl fluoride (Roche Molecular Biochemicals Corp.). 7B2 and PC2 were immunoprecipitated from clarified extracts with LSU13 and LSU18 antisera as described previously (17). Immune complexes were boiled in Laemmli sample buffer and proteins were separated by SDS-PAGE on 8.8% or 15% gels. Gels were analyzed with a Storm PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
In Vitro Activation of PC2-Golgi-enriched subcellular fractions were prepared from CHO/PC2 cells as described previously (17). Golgi fractions (1-3 g of protein) were incubated in 100 mM sodium acetate, pH 5, in the presence of 1% Triton X-100, 5 mM CaCl 2 and an inhibitor mixture composed of 1 M pepstatin, 0.28 mM tosylphenylalanine chloromethyl ketone, 1 M trans-epoxysuccinic acid (E64), and 0.14 mM tosyllysyl chloromethyl ketone. Incubations were conducted at 37°C for the indicated periods of time. PC2 activity was estimated using 200 M Pyr-Glu-Arg-Thr-Lys-Arg-methylcoumarin amide as a substrate. Fluorescence was measured with a Cambridge Technology (Cambridge, MA) fluorometer and the amount of released aminomethylcoumarin was calculated by reference to a free aminomethylcoumarin standard. In some experiments the incubation was performed in the presence of 1 M CT-peptide, a PC2-specific inhibitor that corresponds to the carboxyl terminus of 7B2 (22). Experiments were reproduced at least three times. The results presented are the mean (Ϯ S.D.) of triplicate measurements of PC2 activity from a representative experiment.
PC2 Activity Determination-For transient expression experiments, overnight medium (Opti-MEM, Life Technologies, Inc.) was collected 48 h after transfection and centrifuged in order to pellet floating cells. PC2 activity from 25 l of medium was measured in 100 mM sodium acetate, pH 5, in the presence of 5 mM CaCl 2 , 0.4% octyl glucoside, and the inhibitor mixture indicated above, using the same substrate. The identity of the enzymatic activity as PC2 was confirmed by measuring the extent of inhibition with 1 M CT-peptide. Three independent transfections were performed on duplicate wells, and PC2 activity was assayed in triplicate for each well. The results presented are the mean (Ϯ S.D.) of the six measurements of PC2 activity from a representative experiment.

RESULTS
The 86 Amino-terminal Residues of 7B2 Are Not Involved in PC2 Activation-We analyzed the role of the amino-terminal half of 7B2 in proPC2 activation by serial deletion (Fig. 1A). As we have already demonstrated that the capacity of 7B2 to activate proPC2 resides in the 150-residue amino-terminal domain referred to as 21-kDa 7B2 (14), we terminated the constructs at residue 150. In order to target the truncated 7B2s to the secretory pathway, we introduced cDNAs coding for the various proteins lacking signal peptides into the pSecTag2 HygroA vector. This plasmid encodes the signal peptide from the murine Ig protein upstream of the multicloning site; the truncated 7B2s are thus directed to the secretory pathway by a different signal peptide than that of wild type 7B2. The efficiency of this plasmid in delivering the truncated 7B2s to the secretory pathway was verified by the detection of these proteins in the secretion medium (data not shown).
We first tested the interaction of the deleted 7B2s with proPC2 in stably transfected AtT-20 cells by co-immunoprecipitation after metabolic labeling. The three mutants bound to proPC2 as efficiently as the 21-kDa 7B2 (Fig. 1B). In AtT-20 cells, 7B2 binds to proPC2 in the ER, which increases the transport rate of proPC2 to the TGN/secretory granules. This increased transport in turn generates a higher processing rate of the propeptide (14,17). We tested the capacity of the truncated 7B2s to facilitate the maturation rate of proPC2. Coexpression of 7B2 30 -150 and 7B2 68 -150 increased the proteolytic processing of proPC2 as efficiently as the wild type 7B2 (Fig.  1C). On the other hand, processing of proPC2 in AtT-20 cells co-expressing PC2 and the shortest 7B2 86 -150 construct was identical to processing of the propeptide in cells not cotransfected with 7B2 (Fig. 1C).
As these three truncated 7B2s were able to bind to proPC2, we tested their capacity to activate proPC2 in vivo. For this purpose, we transiently expressed different 7B2 constructs in CHO/PC2 cells. These cells express very high levels of proPC2, as this gene has been amplified using the dihydrofolate reductase method (25). In addition, CHO/PC2 cells do not express 7B2 and therefore possess no endogenous PC2 activity (25). After transient transfection of mutant 7B2 constructs, PC2 activity was tested in the overnight conditioned medium. In line with their binding to proPC2, the three deleted 7B2s were able to activate PC2 ( Fig. 2A); however, 7B2 86 -150 was not as efficient as 7B2 30 -150 and 7B2 68 -150 .
The capacity of these deleted 7B2s to activate proPC2 was confirmed using an in vitro activation assay, which is based upon the use of purified Golgi membranes from the same CHO/ PC2 cells described above. We have already characterized this cell-free assay and have demonstrated that addition of recombinant 21-kDa 7B2 to these membranes is sufficient to achieve activation of proPC2 (17). The recombinant 7B2 30 -150 and 7B2 68 -150 proteins were able to activate proPC2 in vitro (Fig.  2B). We were not able to obtain the recombinant protein corresponding to the shortest of the truncated 7B2s since it was not expressed in bacteria, perhaps due to its short size (64 residues).
Taken together, these experiments demonstrate that the 86 amino-terminal residues of 7B2 are not required for 7B2 binding to proPC2 or for activation of proPC2. In addition, the analysis of proPC2 conversion kinetics shows that the 68 -87residue segment of 7B2 is required for the increase of proPC2 conversion rate generated by 7B2.
A Peptide Corresponding to Residues 86 -121 Is Sufficient for ProPC2 Activation-We have previously demonstrated that carboxyl-terminal deletion of the 21-kDa domain following residue 109 prevents its interaction with proPC2, whereas deletion after residue 121 does not prevent binding to proPC2 (26). We combined these results with the results of the amino-terminal deletions described above, and synthesized a set of peptides corresponding to residues 86 -121 or to shorter constructs.
The 86 -121 peptide contains the PPNPCP motif that we have already identified as a binding determinant (26), as well as the only disulfide present in mammalian and Xenopus 7B2s, and a COOH-terminal putative ␣-helix. The various deletions in the 86 -121 peptide were thus designed to selectively remove these three determinants (Fig. 3A). The peptides were tested in the in vitro Golgi membrane proPC2 activation assay. The synthetic peptides (5 M final concentration) were added to CHO/

. The amino-terminal half of 7B2 is not required for binding to proPC2 but is involved in the facilitation of its maturation.
A, schematic representation of the 7B2 amino-terminal deletions. CT corresponds to the CT-peptide, the carboxyl-terminal domain of 7B2 that specifically inhibits PC2 activity. The proline-rich sequence PPNPCP and the disulfide bond are represented. The gray box corresponds to the putative ␣-helix. The first residue of each 7B2 construct is indicated. B and C, binding of 7B2s to proPC2 and effects on proPC2 maturation. AtT-20 cells expressing PC2 alone (Ϫ) or co-expressing PC2 and wild type 7B2 (WT) or 7B2 30 -150 (30), 7B2 68 -150 (68), or 7B2 86 -150 (86) were radiolabeled for 20 min and chased for either 30 min (B) or 120 min (C). PC2 was immunoprecipitated under native (B) or denaturing (C) conditions, and proteins were separated by SDS-PAGE. The proPC2 and PC2 bands were quantitated after a 120-min chase, and the ratio of proPC2/(proPC2 ϩ PC2) was calculated and shown on the bottom of the gel (C). Note that the position of endogenous 7B2 is shown (21 kDa 7B2); this protein is barely detectable in cells not expressing exogenous constructs. PC2 cell Golgi membranes, and the pH was lowered to 5 for measurement of PC2 activity. The 86 -121 sequence was the only peptide that could activate proPC2 (Fig. 3B). All of the other peptides were not able to effect PC2 activation, including those with deletions within the proline-rich sequence (95-121) and those with deletions within the ␣-helix (86 -112). On a molar basis, the 86 -121 peptide was almost as efficient as the entire amino-terminal 7B2 domain in activating proPC2 in vitro (Fig. 3C).
The ␣-Helix Present in the 86 -121 Peptide Is Required for PC2 Activation-We then investigated the involvement of the ␣-helix in PC2 activation. The recent cloning of 7B2 from Caenorhabditis elegans (27) and Drosophila melanogaster 2 showed that this putative ␣-helix is conserved, even though the overall conservation of 7B2 through evolution is quite poor. In order to test the requirement for the ␣-helix, we introduced two mutations: the replacement of an arginine by a proline (R116P), which should disrupt the helical structure, and the insertion of an alanine at position 115 (ϩA115), which should preserve the helical structure but modify the distribution of the residues on its surface (Fig. 4A). The helical wheel projection of residues 108 -121 shows that two conserved phenylalanines (one of which is a tyrosine in Xenopus laevis and C. elegans) are located on the same face of the helix (Fig. 4B). Three negatively charged residues neighbor the phenylalanines, but these negatively charged residues are not conserved in other species. The insertion of an alanine between these two phenylalanines (ϩA115) results in the shift of one of the two phenylalanines on the other side of the helix (Fig. 4B). The predicted effects of these mutations were confirmed by different secondary structure prediction programs (28 -30). Mutated 7B2s were tested for their ability to promote PC2 activation in vivo using transient transfection in CHO/PC2 cells (Fig. 4C). Both ␣-helix mutations prevented PC2 activation, confirming that both the ␣-helix structure and the distribution of the residues at the surface of the helix are important. To confirm these results, we synthesized a corresponding 86 -121 peptide mutated in the helix by insertion of an alanine at the same position (ϩA115). In addition, we mutated either one (F114A) or both (F114A/ F118A) phenylalanine residues into alanine. As a positive control, we also replaced arginine-116 by an alanine (R116A); this mutation, unlike R116P, is not predicted to disrupt the helical structure (28 -30). These synthetic peptides were tested for their capacity to activate PC2 in vitro using the Golgi membrane assay (Fig. 4D). The ϩA115 mutant and the two phenylalanine mutants could only activate PC2 at 20 -30% of the activity of the control 86 -121 peptide (Fig. 4D). This result suggests that the presence of the two hydrophobic residues on one face of the helix constitutes a determinant directly involved in proPC2 binding or PC2 activation. The R116A mutation had no effect on PC2 activation (Fig. 4D), thus confirming that the effect of the R116P mutation was due to the disruption of the helical structure and not to the role of the arginine residue itself.
The Disulfide Bond Is Required for PC2 Activation-Another feature of the 86 -121 peptide is the presence of the sole two cysteine residues in mammalian 7B2s, which are also conserved in all of the invertebrate 7B2s cloned thus far (27,31). 2 These cysteines form a disulfide bond that is apparently not required for correct processing and for secretion of 7B2 (17,32). However, in vivo reduction of the disulfides does prevent binding of 7B2 to proPC2 (17). In order to determine whether this lack of binding is a direct result of the reduction of the disulfide bond, we mutated cysteine 104 into alanine (C104A). This mutant was unable to bind to proPC2 (Fig. 5A) and was unable to facilitate its maturation in AtT-20 cells (Fig. 5B). The C104A mutant was also not able to activate proPC2 in transiently transfected CHO/PC2 cells (Fig. 5C). Finally, the requirement for the 7B2 disulfide bond in proPC2 activation was confirmed with synthetic peptides in which the cysteines were blocked by methylation. Our data clearly show that, whereas the 86 -121 peptide could activate PC2 in vitro, the peptide containing methylated cysteines could not (Fig. 5D). Thus four different assays demonstrate the necessity for the 7B2 disulfide bond for binding to proPC2 as well as for activating proPC2.

DISCUSSION
The production of enzymatically active PC2 requires the interaction of its proenzyme with the neuroendocrine protein 7B2 (14). Whereas we have been able to delineate the cellular mechanism of the interaction between proPC2 and 7B2 (14,17), the analysis at a molecular level remains puzzling (reviewed in Ref. 4). Interestingly, although intracellular encounter with 7B2 is required for proPC2 activation, the lack of interaction between these two proteins does not prevent the cleavage of the PC2 propeptide. However, the mature enzyme generated under these conditions (lack of intracellular interaction with 7B2) exhibits absolutely no catalytic activity, and this particular maturation process has thus been referred to as 2 D. Siekhaus and I. Lindberg, unpublished data. "unproductive" propeptide cleavage (16,17,33). In order to gain a better understanding of the molecular mechanism of 7B2-mediated activation of proPC2, we have investigated the structural requirements of 7B2. We were able to identify a 36-residue peptide that contains a proline-rich motif, a disulfide bridge, and an ␣-helix. These three determinants are necessary and sufficient to generate a proPC2 species capable of actual activation. This 36-residue peptide (position 86 -121) is not localized in the amino-terminal half of 7B2, which is homologous to the chaperonin 60 GroEL (12). The lack of importance of this chaperone-like domain for proPC2 activation is in agreement with its weak conservation through evolution. Whereas the sequences of X. laevis (34) and mammalian 7B2s (35)(36)(37) share more than 90% homology, the recent cloning of invertebrate 7B2s, from Lymnea stagnalis (31), C. elegans (27), and D. melanogaster 2 shows no significant conservation in the amino-terminal half of these 7B2s. Whereas 7B2 86 -150 , which excludes the 7B2 chaperone-like domain, could not facilitate proPC2 maturation, the 7B2 68 -150 construct was active. These results support the idea that, even though the 7B2 chaperonelike domain is not involved in PC2 activation, the segment corresponding to residues 68 -86 of this domain is involved in the facilitation of proPC2 maturation. The role of the NH 2terminal half of 7B2 therefore remains unclear. It is conceivable that, in addition to its role in proPC2 activation, 7B2 is involved in another regulatory function. Our recent analysis of 7B2 null mice showing that this null exhibits a much more severe phenotype than the PC2 null supports the notion that 7B2 may have non-PC2-related roles, which relate to the regulation of either the development or the secretory activity of the intermediate pituitary (16). Such roles could potentially be carried out by the NH 2 -terminal portion of the molecule, although this idea is purely speculative at present.
In agreement with our previous study (26), our new peptide data show that the P 91 PNPCP 96 motif is required for proPC2 activation. Furthermore, the fact that a peptide which begins at cysteine 95 was unable to activate proPC2 is in agreement with the essential role of proline 94 already demonstrated (26). In this previous study, we had also shown the involvement of residues 109 -121 in binding to proPC2. We have now demonstrated the essential role of the helical structure of this section using mutations that either disrupt the helix (R116P), or modify the distribution of the residues on the surface of the helix (ϩA115) (Fig. 4B). In addition to providing evidence for the role of the ␣-helix, we have identified important residues in this structure. The helical wheel projection of residues 108 -121 shows the presence of three negatively charged residues next to two hydrophobic residues (phenylalanine), on the same face of the helix. Whereas the charged residues are not conserved among species, the hydrophobic phenylalanines are, which suggests that they could be functionally important. In agreement with this conservation, mutation of one or two phenylalanine residue(s) (F114A and F114A/F118A) demonstrates their essential role in activation of proPC2. We thus propose that these two hydrophobic residues constitute an important site of interaction between the 7B2 ␣-helix and proPC2.
The third structural element required for binding to proPC2 and for PC2 activation is 7B2's only disulfide. This disulfide involves the cysteine present in the PPNPCP motif. We had already shown that exchanging the position of this cysteine with the fourth proline residue of the motif prevented binding to proPC2 and PC2 activation. We had thus proposed that this effect resulted from the disorganization of the proline-rich motif. Further work had then demonstrated that, even though the formation of 7B2 disulfide is not required for 7B2 maturation and secretion (17,32), the in vivo reduction of disulfide bridges in newly synthesized proteins prevents the interaction between 7B2 and proPC2 (17). We have now demonstrated that the presence of the 7B2 disulfide is required for effective interaction with proPC2. The most probable hypothesis that explains the effect of the cysteine 104 mutation is that the disulfide bridge stabilizes the proline-rich motif in a conformation nec- essary for binding to proPC2. We cannot, however, rule out the possibility that the 7B2 disulfide bridge is more directly involved in the interaction with proPC2 or in PC2 activation, for example by providing a disulfide isomerase activity to 7B2.
The presence of both the proline-rich motif and the ␣-helix in a relatively short peptide suggests that these determinants are involved in both binding to proPC2 and in activation of PC2. In order to test the possibility that one determinant would be responsible for binding only, we performed competition experiments using the control 86 -121 peptide and the deleted peptides that contain either the proline-rich motif alone (86 -100), or this motif and the disulfide (86 -107) or the disulfide and the ␣-helix (95-121). None of these deleted peptides was able to significantly reduce the PC2 activation induced by the 86 -121 peptide, even when used at a concentration in 100-fold excess (data not shown). These results suggest that the 86 -121 peptide must acquire a specific conformation that involves all three structural determinants, or that these three determinants bind efficiently to proPC2 when they interact with different sites on proPC2. The previous studies that were aimed at identifying PC2 determinants responsible for binding 7B2 concluded that this interaction most probably involves several domains of proPC2 (33,38,39). The involvement of different 7B2 structural determinants is in agreement with such a conclusion.
There are, however, still no data available concerning the crystal structure of any PC, and the structural study of these enzymes is based on the model of the related enzyme subtilisin (40 -42).
ProPC2 exits from the ER as a proenzyme that is autocatalytically processed in the acidic compartments of the TGN and secretory granules (5,7,25,43,44), whereas the propeptides of the other PCs are initially processed in the ER. A recent study demonstrated that furin-cleaved propeptide remains associated with the mature enzyme (15). These authors also suggested that this interaction could be required for furin transport out of the ER. If such a hypothesis also holds true for proPC2, it must occur without actual cleavage of the propeptide. The binding of 7B2 to several proPC2 determinants could thus provide an alternative for the formation of a transportcompetent propeptide/mature enzyme complex. 7B2 binding to the proenzyme could protect proPC2 either from premature unproductive processing, or from inactivation of the proenzyme due to exposure to an acidic pH. Indeed, acidification of proPC2-enriched Golgi fractions prior to addition of recombinant 7B2 completely blocks proPC2 activation (17). 7B2 could fulfill such a passive role, or it could promote proPC2 activation more actively. In both cases, bringing together several domains of proPC2 through binding to different 7B2 determinants could constitute the actual mechanism of 7B2-mediated activation of proPC2. Our new data suggest that a productive interaction between proPC2 and 7B2 requires a specific conformation not only for proPC2, as previously proposed, but also for 7B2. The characterization of a 36-residue peptide that contains this conformation and that is sufficient for effecting proPC2 activation in a cell-free system should prove helpful for the final molecular analysis of the proPC2 activation mechanism.