Proteinase Inhibition by Proform of Eosinophil Major Basic Protein (pro-MBP) Is a Multistep Process of Intra- and Intermolecular Disulfide Rearrangements*

The metzincin metalloproteinase pregnancy-associated plasma protein A (PAPP-A, pappalysin-1) promotes cell growth by the cleavage of insulin-like growth factor-binding proteins-4 and -5, causing the release of bound insulin-like growth factors. The proteolytic activity of PAPP-A is inhibited by the proform of eosinophil major basic protein (pro-MBP), which forms a covalent 2:2 proteinase-inhibitor complex based on disulfide bonds. To understand the process of complex formation, we determined the status of cysteine residues in both of the uncomplexed molecules. A comparison of the disulfide structure of the reactants with the known disulfide structure of the PAPP-A·pro-MBP complex reveals that six cysteine residues of the pro-MBP subunit (Cys-51, Cys-89, Cys-104, Cys-107, Cys-128, and Cys-169) and two cysteine residues of the PAPP-A subunit (Cys-381 and Cys-652) change their status from the uncomplexed to the complexed states. Upon complex formation, three disulfide bonds of pro-MBP, which connect the acidic propiece with the basic, mature portion, are disrupted. In the PAPP-A·pro-MBP complex, two of these form the basis of both two interchain disulfide bonds between the PAPP-A and the pro-MBP subunits and two disulfide bonds responsible for pro-MBP dimerization, respectively. Based on the status of the reactants, we investigated the role of individual cysteine residues upon complex formation by mutagenesis of specific cysteine residues of both subunits. Our findings allow us to depict a hypothetical model of how the PAPPA·pro-MBP complex is formed. In addition, we have demonstrated that complex formation is greatly enhanced by the addition of micromolar concentrations of reductants. It is therefore possible that the activity in vivo of PAPP-A is controlled by the redox potential, and it is further tempting to speculate that such mechanism operates under pathological conditions of altered redox potential.

The 1547-residue pregnancy-associated plasma protein A (PAPP-A, pappalysin-1) 1 belongs to the metzincin superfamily of metalloproteinases (1). Known substrates of PAPP-A include insulin-like growth factor-binding proteins (IGFBPs)-4 and -5, which function to inhibit the biological activities of IGF-I and -II, and cleavage by PAPP-A releases bioactive IGF (2,3). However, the growth-promoting activity of PAPP-A is antagonized by the proform of eosinophil major basic protein (pro-MBP) of 206 residues (4,5), originally known in its mature form as a cytotoxic protein from the granules of the eosinophil leukocyte (6). Pro-MBP inhibits the proteolytic activity of PAPP-A by the formation of an inactive, 2:2 proteinase-inhibitor complex, denoted PAPP-A⅐pro-MBP, covalently bound by disulfides (4,7).
The 500-kDa PAPP-A⅐pro-MBP complex was first isolated from the circulation of pregnant women (8), and both PAPP-A and pro-MBP were found to be synthesized in the placenta (9). However, PAPP-A and pro-MBP expression has since been demonstrated in several other tissues (10), and data indicate that the proteolytic activity of PAPP-A is involved in a number of normal and pathological processes, such as fetal development (11)(12)(13), ovarian follicular development (14), human implantation (15), wound healing (16), atherosclerosis (17), and cancer (18).
In the placenta, PAPP-A and pro-MBP are known to be synthesized by different cell types (9), and about 99% of circulating PAPP-A is present as a tetrameric complex with pro-MBP (5), showing that in vivo complex formation occurs outside the cell. In agreement with this, in vitro formation of the disulfide-bonded PAPP-A⅐pro-MBP complex after separate synthesis was recently demonstrated and shown to be required for inhibition of the proteolytic activity (4). Other proteinase inhibitors known to bind covalently to their target proteinase are the serpins (19) and ␣ 2 -macroglobulin (20), but, to the best of our knowledge, there are no prior examples of a proteinase inhibitor, in which the inhibitory mechanism is based on the formation of intermolecular disulfide bonds.
Disulfide bonds in nascent proteins are formed by the oxidation of cysteine residues in the endoplasmic reticulum. Because of the oxidizing nature of the extracellular milieu, disulfides have been considered to be inert. However, several recent reports suggest that the reduction or formation of specific disulfide bonds function in reversible and irreversible switches in the regulation of extracellular protein function. For example, the reduction of specific disulfide bonds in platelet integrin ␣IIb␤3 causes conformational changes in both integrin sub-units and leads to exposure of ligand binding sites (21). Also, the binding of platelets to multimers of the blood protein von Willebrand factor (vWf) is indirectly controlled by thrombospondin-1, which facilitates reversible reduction of vWf intersubunit disulfide bonds (22). Additionally, several animal viruses (e.g. HIV-1 and murine leukemia virus) (23,24) depend on the rearrangements of specific disulfides for virus-membrane fusion. Two systems of protein regulation based on disulfide switches relate to proteolysis. First, the reduction of two disulfide bonds in the fifth kringle domain of plasmin by phosphoglycerate kinase makes plasmin a substrate for proteolysis (25). Second, the serine proteinase inhibitor, plasminogen activator inhibitor type 2 (PAI-2) uses the disruption or the formation of a disulfide bond to switch between a stable, inhibitory monomeric conformation and a conformation that readily forms inactive polymers (26).
We have recently reported the connectivity of the 188 cysteine residues in the 2:2 PAPP-A⅐pro-MBP complex (7). We here show that oxidation does not promote complex formation, indicating that the formation of intermolecular disulfide bonds is not a reaction between free thiol groups and molecular oxygen. To understand the process of complex formation, we determined the status of cysteine residues in both of the uncomplexed molecules and further performed mutagenesis of individual cysteines found to change their status upon complex formation.
Tissue Culture and Transfection-Human embryonic kidney 293T cells (293tsA1609neo) (28) were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM glutamine, nonessential amino acids, and gentamicin (Invitrogen, Life Technologies). Cells were plated onto 6-cm culture dishes, and were transfected 18 h later by calcium phosphate co-precipitation (29) using 10 g of plasmid DNA. The cells were transfected with either a pro-MBP or a PAPP-A expression vector. After another 48 h, the supernatants were harvested, cleared by centrifugation, and replaced by serumfree medium (CD293, Invitrogen) for another 48 h to facilitate Western blotting or purification. All PAPP-A mutants expressed at the level of wild-type PAPP-A, and all pro-MBP mutants, except for pro-MBP(C89S) and pro-MBP(C128S) (4), at the level of wild-type pro-MBP, as determined by ELISA.
In Vitro Formation of the PAPP-A⅐pro-MBP Complex-The PAPP-A⅐pro-MBP complex was formed by mixing culture supernatants from 293T cells transfected separately with PAPP-A or pro-MBP cDNAs. The concentrations of PAPP-A and pro-MBP in the reaction mixtures were 20 and 200 nM, respectively, reflecting physiological concentrations (30). The mixtures were incubated at 37°C, and samples were taken out and frozen at defined time points between 0 and 48 h. In some reactions, defined concentrations of hydrogen peroxide (H 2 O 2 ), reduced glutathione (GSH), dithiothreitol, or ␤-mercaptoethanol were added. Complex formation was visualized by Western blotting, and/or quantitatively monitored by an ELISA specific for the complex (see below).
Western Blotting-Visualization of PAPP-A and pro-MBP was done by Western blotting following separation by SDS-PAGE in 3-8% precast Tris acetate gels (Invitrogen). After electrophoresis, the protein was blotted onto a polyvinylidene difluoride membrane and blocked in 2% skimmed milk powder diluted in 50 mM Tris, 500 mM sodium chloride, 0.1% Tween 20, pH 9.0 (TST). After washing and equilibration in TST, the membrane was incubated with primary antibodies (mAb 234-2 for PAPP-A or mAb 234-10 for pro-MBP), diluted to 1 g/ml in TST containing 2% skimmed milk powder, and incubated for 1 h at room temperature. Incubation with peroxidase-conjugated secondary antibodies (P0260, DAKO), diluted in TST containing 2% skimmed milk powder, was done for 0.5 h at room temperature. The blots were developed using enhanced chemiluminescence (ECL Plus, Amersham Biosciences), and images were captured with a KODAK Image Station 1000.
Enzyme-linked Immunosorbent Assay-PAPP-A and pro-MBP concentrations were measured by sandwich ELISAs, in which polyclonal rabbit anti-(PAPP-A⅐pro-MBP) (31) was used for capture, and monoclonal antibodies against PAPP-A (234-2) (32) or pro-MBP (234-10) (33) followed by peroxidase-conjugated anti-(mouse IgG) (P0260, DAKO) were used for detection. The wells were blocked by incubation with phosphate-buffered saline (PBS) containing 2% bovine serum albumin. Antibodies were diluted in PBS with 0.01% Tween-20 (PBST) and 1% bovine serum albumin. PBST was used for washing. The PAPP-A⅐pro-MBP complex formation was monitored over time using a complexspecific double monoclonal ELISA: The PAPP-A specific monoclonal antibody VRPA-1A 3 was used as the catching antibody, and the PAPP-A⅐pro-MBP complex was detected with biotinylated, pro-MBP-specific monoclonal antibody VRPM-5A 3 followed by incubation with peroxidase-conjugated avidin (P0347, DAKO). In the latter assay, sample dilution and washing after sample incubation were carried out using PBST to which 800 mM sodium chloride was added, to avoid noncovalent association. The amount of complex formed was expressed as a percentage of the amount of complex formed upon completion of the reaction (100%). Dilution series of the PAPP-A⅐pro-MBP complex purified from pregnancy serum (31) were used to establish standard curves.
Measurement of IGFBP-4 Proteolytic Activity-The IGFBP-4 proteolytic activity of PAPP-A-containing samples was measured as described (3), following incubation with pro-MBP or pro-MBP mutants for 48 h, or as indicated. In brief, PAPP-A (0.6 nM, 0.12 g/ml, corresponding to a ϳ30-fold dilution of the sample), 125 I-labeled IGFBP-4 (10 nM, 0.30 g/ml), and IGF-II (50 nM, 0.35 g/ml) (DSL) in 50 mM Tris, 100 mM sodium chloride, 1 mM calcium chloride, pH 7.5 were incubated at 37°C. Samples of the reaction mixtures were taken out at time points from 0 to 30 min and separated by non-reducing SDS-PAGE (10 -20% Trisglycine gels). The degree of cleavage was assessed by measuring band intensities with a PhosphorImager (Molecular Dynamics) and plotted as a function of time after subtraction of background (3).
Deduction of Cysteine Connectivity by S-Cyanylation-2-Nitro-5-thiocyanobenzoic acid (NTCB, Sigma) specifically reacts with free cysteines at pH 8.0, and the peptide bond is subsequently cleaved N-terminal to the cyanylated cysteine at pH 9.0 (34). To avoid interference by fragments originating from PAPP-A autoproteolysis (1), the S-cyanylation experiments were carried out using proteolytically inactive PAPP-A, mutant E483Q (1), or inactive PAPP-A variants, PAPP-A(C381A/ E483Q), and PAPP-A(C652A/E483Q). Culture supernatant with pH adjusted to 8.0 by the addition of 1 M Tris, pH 8.0 was incubated with 2 mM NTCB for 30 min at 37°C. N-Ethylmaleimide was then added to the mixture to a final concentration of 20 mM, the pH was raised to 9.0 by the addition of 3 M Tris, pH 9.0, and the sample was incubated for 2 The numbering of the 1547-residue mature PAPP-A polypeptide is used in this report. Glu-1 of mature PAPP-A corresponds to position 81 of prepro-PAPP-A (Swiss-Prot accession number Q13219). The numbering of the 222-residue prepro-MBP polypeptide (Swiss-Prot accession number P13727) is used in this report. 16 h at room temperature. NTCB cleavage profiles were visualized by Western blotting using polyclonal anti-PAPP-A (31) following separation by reducing SDS-PAGE (10 -20%). The sizes of the resulting fragments were compared with a well-characterized fragment of PAPP-A autoproteolysis, representing PAPP-A residues 1-386 (1), to C-terminally truncated, recombinant PAPP-A variants (representing PAPP-A residues 1-950 (mutant PA1-950) and 1-1129 (mutant PA1-1129) (35)), or to an N-terminally truncated PAPP-A variant representing residues 1133-1547 (mutant PA1133-1547). 4 Amino Acid Analysis-Amino acids were quantified by cation exchange after hydrolysis of peptide bonds at 110°C for 18 h in 6 M HCl, 0.1% phenol, 5% thioglycolic acid (36). For detection of chromatographic fractions containing disulfide-bonded peptides, paired and unpaired cysteine residues were oxidized to cysteic acid (Cya) by incubation of lyophilized samples with performic acid prepared by reacting hydrogen peroxide (30%) with formic acid (1:9 (v/v)) on ice for 10 min. Following incubation on ice for 2 h, the samples were lyophilized again and then hydrolyzed.
Generation of Pro-MBP Peptides and Chromatographic Separation-Recombinant pro-MBP (1 mg), purified by affinity chromatography (4), was digested with thermolysin (type X protease, Sigma) for 18 h at 55°C using an enzyme:substrate ratio of 1:25 (w/w). Thermolytic peptides were separated by reversed-phase high-pressure liquid chromatography (RP-HPLC) on a 4 ϫ 250 mm column packed with Nucleosil C18 100 -5 (Macherey-Nagel). Gradients were formed from 0.1% (v/v) trifluoroacetic acid (solvent A) and 0.075% (v/v) trifluoroacetic acid in 90% (v/v) acetonitrile (solvent B), increasing the amount of solvent B to 90% over 90 min at a flow rate of 1 ml/min. The column was operated at 50°C, the eluate was monitored at 218 nm, and fractions of 0.5 ml were collected. One peptide was further digested with trypsin (1 h at 37°C, 1:25 (w/w)) prior to mass spectrometry.
Mass Spectrometry and N-terminal Sequence Analysis-Mass spectra were acquired with a Voyager-DE PRO matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) instrument (Applied Biosystems). Samples were prepared by co-crystallization of the peptide analytes with ␣-cyano-4-hydroxycinnamic acid (Sigma), and of intact pro-MBP with sinapinic acid (Sigma). The masses of reduced peptides were acquired by on-target reduction using dithiothreitol. Spectra were obtained by averaging 50 -100 single shot spectra and were externally calibrated. Edman degradation was performed on an Applied Biosystems 477A sequencer equipped with an on-line HPLC. For sample loading, isolated peptides or protein (50 -200 pmol) were pipetted onto Polybrene-coated glass filters.

Reductant, but Not Oxidant Promotes Covalent Complex
Formation-To understand the process of PAPP-A⅐pro-MBP complex formation, we first analyzed the ability of reducing and oxidizing agents to influence the reaction. Culture supernatants from human embryonic kidney 293T cells transfected separately with PAPP-A and pro-MBP cDNAs, were mixed to final PAPP-A and pro-MBP concentrations of 20 and 200 nM, respectively, reflecting physiological concentrations (30). Under these conditions, about 10% of PAPP-A is found covalently complexed to pro-MBP after 1 h of incubation, while close to 100% is complexed after 48 h (4). The status of complex formation after 1 h of incubation was analyzed by Western blotting with a PAPP-A-specific antibody following non-reducing SDS-PAGE: The addition of H 2 O 2 (100 M) to the mixture did not promote complex formation (Fig. 1A, lanes 1 and 2). However, the addition of reduced glutathione (GSH) (100 M) caused a large fraction of PAPP-A to migrate in two discrete bands at 450 and 500 kDa, indicating the presence of both 2:1 and 2:2 PAPP-A⅐pro-MBP complexes (Fig. 1A, lane 3). Incubation of PAPP-A alone in the presence of GSH did not cause a shift in its electrophoretic mobility (Fig. 1A, lane 4). This interpretation was confirmed by analyzing the same samples with a pro-MBP specific antibody, showing that PAPP-A and pro-MBP comigrated after 1 h of incubation in the presence of GSH (Fig. 1B).
Similar experiments were carried out at different concentrations of reductant or oxidant, and the amount of PAPP-A⅐pro-MBP complex formed after 1 h was measured quantitatively using an ELISA specific for the complex (Fig. 1C). Incubation of PAPP-A and pro-MBP in the presence of 10 M GSH caused a 4-fold increase in the amount of complex formed, and incubation with 100 M GSH led to an 8-fold increase. In the presence of 1 mM GSH, all PAPP-A was found in the complexed form after 1 h of incubation. Similar results were obtained using the reductants dithiothreitol and ␤-mercaptoethanol (data not shown). Interestingly, the presence of oxidant appeared to inhibit the process (Fig. 1C), as did concentrations of GSH higher than 5 mM (data not shown).
PAPP exchange reactions, including protein-disulfide isomerase (PDI) and thioredoxin (37), but has also been reported to function as a redox switch in a number of extracellular proteins unrelated to protein folding, as in integrin ␣IIb␤3 and murine leukemia virus Env (24,38). Recombinant pro-MBP migrates in nonreducing SDS-PAGE as a monomer (4), but since it exists as a disulfide-bonded dimer in the PAPP-A⅐pro-MBP complex through cysteine residues Cys-104 PM and Cys-107 PM (7), these residues are necessarily involved in exchange reactions during complex formation. To test the hypothesis that pro-MBP residues Cys-104 PM and Cys-107 PM , which are part of a Cys-Xxx-Xxx-Cys motif, function to catalyze complex formation, we constructed a plasmid encoding the mutant pro-MBP(C104S/ C107S), in which both cysteines are replaced by serine residues. However, the ability of mutant pro-MBP(C104S/ C107S) to form a covalent complex with PAPP-A or to inhibit its proteolytic activity could not be distinguished from wildtype pro-MBP (Fig. 2, A and B).
Cysteine residues Cys-125 PM and Cys-128 PM are also part of a Cys-Xxx-Xxx-Cys motif. Because a pro-MBP mutant, in which Cys-128 PM is substituted with serine, forms a covalent complex with PAPP-A and also inhibits its activity as well as wild-type pro-MBP (4), we also conclude that these residues do not function to catalyze complex formation.
PAPP-A⅐pro-MBP Complex Formation Occurs Independently of Pro-MBP Dimerization-To exclude that the above results are caused by the absence of dimerization of wild-type pro-MBP in the recombinant PAPP-A⅐pro-MBP complex, we examined the stoichiometry of mutant PAPP-A(C1130A), in complex with wild-type pro-MBP. Unlike wild-type PAPP-A, which is secreted as a dimer, mutant PAPP-A(C1130A) is secreted as a monomer, because the single disulfide bridge between the PAPP-A subunits is absent (7). As expected, the PAPP-A(C1130A) mutant migrated at 200 kDa (Fig. 2C, lane 1). In contrast, the PAPP-A(C1130A)⅐pro-MBP complex migrated as the wild-type 2:2 covalent complex at 500 kDa (Fig. 2C, lane 2), necessarily formed via pro-MBP dimerization. Thus, this experiment unequivocally shows that the pro-MBP dimer is formed in the recombinant complex.
Deduction of Cysteine Connectivity in Uncomplexed PAPP-A-To describe the events leading to the formation of the tetrameric PAPP-A⅐pro-MBP complex, knowledge of the status of cysteine residues in both of the reactants and in the product is required. The status of cysteines in the PAPP-A⅐pro-MBP complex, a total of 188, was recently delineated (7), but it would not be feasible to determine the status of cysteines in the uncomplexed PAPP-A dimer by a direct method. Instead, we used an indirect approach based on chemical modification of unpaired cysteine residues by S-cyanylation with NTCB. S-Cyanylation of exposed cysteines at pH 8 and subsequent spontaneous chemical cleavage of the peptide bond N-terminal to the modified cysteine at pH 9, results in the formation of an N-terminal peptide fragment and a C-terminal 2-iminothiazolidine-4-carboxylyl (ITC) peptide derivative (34). Hence, the number of fragments and their sizes in the NTCB cleavage profile of PAPP-A will be indicative of the number and location of free cysteines. Previously characterized fragments and truncated variants of the 1547-residue recombinant PAPP-A subunit were used to estimate the location of the NTCB cleavage sites, and candidate cysteines were probed by mutagenesis: If a given cysteine residue is free, a mutation into alanine eliminates one cleavage site, causing two fragments to disappear from the NTCB cleavage profile. Alternatively, if the cysteine residue is engaged in disulfide pairing, substitution into alanine results in one additional reactive cysteine, adding fragments to the cleavage profile. At the sites of modified cysteines, only partial cleavage, typically less than 10% is observed (39,40). Thus, all bands observed are N-or C-terminal fragments, as fragments arising from two cleavages within the same polypeptide chain are quantitatively negligible.
The NTCB cleavage profile of PAPP-A was visualized by Western blotting using polyclonal anti-PAPP-A following separation by reducing SDS-PAGE. Six bands were observed, indicating the presence of three unpaired cysteines in uncomplexed PAPP-A (Fig. 3A). Considering their migration, we conclude that the uppermost band (band 1) pairs with the lower band (band 6). Accordingly, band 2 pairs with band 5, and band 3 with band 4.
Band 6 migrated close to PAPP-A fragment 1-386 (Fig. 3B,  lanes 1 and 2), suggesting Cys-381 PA as a candidate free cysteine, as it binds to pro-MBP in the PAPP-A⅐pro-MBP complex (7). We consequently expressed a PAPP-A mutant, PAPP-A(C381A), with this cysteine substituted into alanine. Upon NTCB treatment of PAPP-A(C381A), bands 1 and 6 were absent from the cleavage profile (Fig. 3C, lane 1), demonstrating that these bands originated from cleavage at Cys-381 PA . Band 3 migrated slightly faster than PAPP-A fragment 1-950 (Fig. 3C, lanes 2 and 3), leaving at least 14 cysteine residues in the sequence stretch from approximately residue 600 -950 as possible candidates (7). As Cys-652 PA also binds to the pro-MBP subunit in the PAPP-A⅐pro-MBP complex, we probed this residue by the expression of mutant PAPP-A(C652A). Upon NTCB cleavage, bands 3 and 4 were absent from its profile (Fig. 3D), demonstrating that Cys-652 PA is unpaired in uncomplexed PAPP-A.
Band 2 comigrated with a PAPP-A variant representing residues 1-1129 (Fig. 3C, lanes 3 and 4), and band 5 comigrated with a C-terminal PAPP-A fragment of residues 1133-1547 (Fig. 3B, lanes 2 and 3), pointing at Cys-1130 PA as a possible candidate. In the NTCB cleavage profile of mutant PAPP-A(C1130A), bands 2 and 5 were not observed (data not shown), demonstrating that cleavage did occur at Cys-1130 PA , which in wild-type PAPP-A is responsible for dimerization through the Cys-1130 PA -Cys-1130 PA disulfide bond (7). However, a minor fraction of wild-type PAPP-A (about 10%) is observed to migrate as a monomer in nonreduced Western blots (data not shown), indicating that, in this fraction, Cys-1130 PA is unpaired and particular susceptible to S-cyanylation. We have now unequivocally shown that the NTCB cleavage profile of wild-type PAPP-A originates from cleavage at Cys-381 PA , Cys-652 PA , and Cys-1130 PA . In the PAPP-A⅐pro-MBP complex, Cys-381 PA and Cys-652 PA form intermolecular disulfide bonds to pro-MBP residues Cys-51 PM and Cys-169 PM , respectively (Fig. 4).
The Disulfide Structure of Uncomplexed Pro-MBP-As the pro-MBP subunit is smaller, we used a direct approach to determine the status of pro-MBP cysteine residues. Prior to purification of pro-MBP by affinity chromatography (4), potentially free cysteine residues were modified by alkylation with iodoacetamide. The mass spectrum of the purified protein showed a broad peak with an average mass of 28 kDa, representing the pro-MBP monomer, and short, broad peaks at 14 and 56 kDa, representing M2H ϩ and 2MH ϩ species, respectively (Fig. 5). By SDS-PAGE, reduced pro-MBP migrated at 38 kDa, while non-reduced pro-MBP migrated at 36 kDa indicating the presence of disulfide bonds (Fig. 5, gel inset).
The alkylated pro-MBP was treated with performic acid, and the content of Cya was determined by amino acid analysis to 9.9 Ϯ 0.3 residues per molecule. No difference in Cya content was found for pro-MBP modified by iodoacetamide in the presence of 8 M urea and EDTA after purification. We conclude that 10 cysteine residues of alkylated pro-MBP were susceptible to performic acid oxidation and therefore engaged in disulfide bonding.
A thermolytic digest of pro-MBP was then separated by reversed-phase high-pressure liquid chromatography (RP-HPLC) (Fig. 6). The eluate was screened for carbamidomethylated cysteine (CAM-Cys) and the content of Cya following performic acid oxidation, and selected fractions were analyzed by mass spectrometry, as detailed in the legend of Fig. 7. The isolated peptides were also subjected to N-terminal sequence analysis (not shown).
A total of five disulfide bonds were located in uncomplexed pro-MBP (Table I) (Table I). In striking contrast, only two of these intramolecular disulfide bonds, Cys-125 PM -Cys-220 PM and Cys-197 PM -Cys-212 PM , are present in the pro-MBP subunit of the PAPP-A⅐pro-MBP complex (Fig. 4B). Thus, in the course of complex formation, the three pro-MBP disulfide bonds Cys-51 PM -Cys-169 PM , Cys-89 PM -Cys-128 PM , and Cys-104 PM -Cys-107 PM , which all connect the acidic propiece with the basic MBP region, are disrupted.
The Effect of Cysteine Substitutions on the PAPP-A⅐pro-MBP Complex Formation-We have now established a set of cysteines residues in both PAPP-A and pro-MBP that all change their status during the process of complex formation (Fig. 4, A  and B). Based on this, we analyzed the role of individual cysteine residues upon complex formation by mutagenesis of specific cysteine residues in both subunits.
Two of the cysteines engaged in an intramolecular disulfide bond of pro-MBP, Cys-51 PM and Cys-169 PM , form intermolecular disulfide bonds with PAPP-A cysteines Cys-381 PA and Cys-652 PA , respectively, when complexed (Fig. 4). We expressed and analyzed two substitution mutants, in which this disulfide was disrupted: Pro-MBP(C169S), in which Cys-51 PM is unpaired, was unable to form a covalent complex and inhibit the proteolytic activity of PAPP-A (Fig. 8, A and B and Table  II). In contrast, pro-MBP(C51S), in which Cys-169 PM is unpaired, formed a covalent complex severalfold faster than wildtype pro-MBP (Fig. 8A).
Additional cysteine residues, here shown to change their status upon complex formation were analyzed in a similar manner (Table II). Pro-MBP(C89S) and pro-MBP(C128S) formed a covalent complex at a rate similar to wild-type pro-MBP, indicating that disruption of the Cys-89 PM -Cys-128 PM disulfide does not occur prior to the formation of the first intermolecular disulfide bond between a PAPP-A and a pro-MBP subunit. PAPP-A(C381A) also formed the complex at a rate similar to wild-type PAPP-A, while the ability of PAPP-A(C652A) to form a covalent complex was several fold reduced.
The Formation of One Intermolecular Disulfide Bond outside the Proteolytic Domain Inhibits the Proteolytic Activity of PAPP-A-Our findings suggest that the first covalent interaction between PAPP-A and pro-MBP is the formation of the Cys-169 PM -Cys-652 PA disulfide bond. To determine the effect of this disulfide bond on the proteolytic activity of PAPP-A, we compared the proteolytic activity with the degree of covalent complex formation for PAPP-A incubated with wild-type pro-MBP or pro-MBP(C51S). The degree of inhibition and the degree of covalent complex formed correlated well for both the wild-type PAPP-A⅐pro-MBP and the PAPP-A⅐pro-MBP(C51S) complex (Fig. 8, C and D), suggesting that the formation of the Cys-169 PM -Cys-652 PA intermolecular disulfide bond leads to immediate inhibition of the proteolytic activity of PAPP-A, independent of the formation of the Cys-51 PM -Cys-381 PA disul-fide. This finding indicates that the inhibitory mechanism of pro-MBP is based on the formation of one disulfide bond between the PAPP-A and pro-MBP subunits, located ϳ70 residues outside the proteolytic domain (1) (Fig. 4A). DISCUSSION We here show that formation of a covalent PAPP-A⅐pro-MBP complex is accelerated by low concentrations of reductant, but inhibited by oxidant. Concentrations of GSH comparable to the level in blood plasma (10 -20 M) (41) enhanced the rate of PAPP-A⅐pro-MBP complex formation about 5-fold, while 1 mM GSH enhanced the rate 10-fold (Fig. 1C). Our findings indicate that the redox potential of the local environment in tissues may function to control PAPP-A⅐pro-MBP complex formation and hence, in turn, IGF activity. Importantly, this may be relevant under pathological conditions, in which the redox potential is altered.
To delineate the events leading to the formation of the PAPP-A⅐pro-MBP complex, we first analyzed the cysteine connectivity of the reactants, uncomplexed PAPP-A and pro-MBP. By Scyanylation of unpaired cysteine residues in wild-type PAPP-A and subsequent chemical cleavage N-terminal to the modified cysteine, we showed that uncomplexed PAPP-A contains reactive cysteine residues. Using recombinant PAPP-A fragments to evaluate the size of the cleavage fragments and further  b The numbers correspond to the RP-HPLC fractions analyzed by mass spectrometry in Fig. 7. c The first residue of each sequence is numbered. Cysteine residues are shown in bold. d Numbers indicate pairing of cysteine residues of the isolated peptides. An asterisk indicates that the given cysteine residue is carbamidomethylated.
e Experimentally determined masses (m/z) of MH ϩ species are given with the deviations from the calculated peptide masses indicated in parentheses.
f Variants of TL-1 with one or both histidines modified by iodoacetamide were found in fractions 34 and 35. g A variant of TL-3 was found in fraction 65 with the methionine modified by iodoacetamide. h A shorter variant of this peptide, 99 VGIPGCQTCRY, was found in fraction 88. i TL-6 is a longer variant of TL-1 and contains a carbamidomethylated cysteine. mutagenesis to probe candidate residues, we demonstrated that cysteines Cys-381 PA and Cys-652 PA are free in uncomplexed PAPP-A (Fig. 4A). The disulfide structure of purified pro-MBP was determined directly by proteolytic digestion followed by the analysis of peptides by amino acid analysis and mass spectrometry after chromatographic separation. We found that 10 of 12 pro-MBP cysteine residues are engaged in five disulfide bonds (Fig. 4B): Two disulfide bonds, Cys-125 PM -Cys-220 PM and Cys-197 PM -Cys-212 PM , are also found in the structure of mature MBP isolated from the granules of the eosinophil leukocyte (42,43). In addition, pro-MBP contains three disulfide bonds, Cys-51 PM -Cys-169 PM , Cys-89 PM -Cys-128 PM , and Cys-104 PM -Cys-107 PM , which anchor the acidic propiece to the basic MBP domain, leaving only cysteines, Cys-147 PM and Cys-201 PM , unpaired. In striking contrast, five cysteine residues are unpaired in mature MBP purified from the granules of eosinophil leukocytes (Fig. 4B).
The extracellular formation of disulfide-bonded complexes has been studied for the interaction of thrombospondin and vitronectin with thrombin-antithrombin III (44 -46). The proposed mechanism, based on the disulfide structures of the reactants, is the isomerization of a free cysteine in thrombospondin or vitronectin with a disulfide bridge in thrombin, leading to the formation of an intermolecular disulfide bond. We found that the target cysteines in PAPP-A are unpaired, whereas the reacting cysteines of pro-MBP are connected to each other by a disulfide bond (Fig. 4, A and B). However, a comparison of the disulfide structure of the reactants with that of the PAPP-A⅐pro-MBP complex, shows that six cysteine residues of the pro-MBP subunit (Cys-51 PM , Cys-89 PM , Cys-104 PM , Cys-107 PM , Cys-128 PM , and Cys-169 PM ) and two cysteine residues of the PAPP-A subunit (Cys-381 PA , and Cys-652 PA ) change their status from the uncomplexed to the complexed states (Fig. 4, A and B). Thus, the covalent changes during formation of the PAPP-A⅐pro-MBP complex are far more extensive than expected considering the status of the reactants. The covalent binding of pro-MBP to PAPP-A serves to inhibit its proteolytic activity. However, the extensive rearrangement of disulfide bonds upon complex formation likely causes conformational changes, which may not be related directly to the control of proteolysis.
The set of cysteine residues here shown to change their status, enabled us to probe the role of individual cysteine residues in the process of complex formation by mutagenesis. We found that disruption of the pro-MBP disulfide Cys-51 PM -Cys-169 PM positions the free cysteines so that Cys-169 PM  readily interacts with Cys-652 PA , as the rate of complex formation for pro-MBP(C51S) was accelerated 4-fold. In contrast, the position of Cys-51 PM is not favorable for initial interaction with Cys-381 PA , as the rate of complex formation for pro-MBP(C169S) was reduced severalfold compared with the wildtype (Fig. 8A). The substitution of additional pro-MBP cysteines had no effect on the rate of complex formation suggesting that the initial changes of pro-MBP during complex formation are the disruption of the disulfide bond Cys-51 PM -Cys-169 PM and the reaction of Cys-169 PM with Cys-652 PA (Table II).
Micromolar levels of reductant may function to reduce disulfide bonds particularly susceptible as a consequence of proteinprotein interactions between PAPP-A and pro-MBP. A priori, we are unable to predict the location of such disulfides in either the PAPP-A or the pro-MBP subunit. However, it is tempting to speculate that the redox potential of the disulfide bond Cys-51 PM -Cys-169 PM is altered upon the initial interaction between PAPP-A and pro-MBP, as the effect of substituting cysteine residue Cys-51 PM with a serine residue, thereby disrupting the disulfide bond, are similar to the effect of micromolar amounts of reductant. A similar effect of reductant and substitution of cysteine residues has been observed with integrin ␣IIb␤3, in which disruption of specific disulfide bonds in the cysteine-rich domain of the ␤-subunit caused the exposure of ligand binding sites (47)(48)(49). Thus, in analogy with this, the covalent interaction between PAPP-A and pro-MBP may require the disruption of specific disulfide bonds. Such disruption may be favored by the initial interaction between the subunits.
Unlike PAPP-A, uncomplexed pro-MBP is a monomer, but is present in the PAPP-A⅐pro-MBP complex as a dimer based on two disulfides, Cys-104 PM -Cys-104 PM and Cys-107 PM -Cys-107 PM . Upon complex formation, these two bridges form from the intrachain disulfide Cys-104 PM -Cys-107 PM , present in each pro-MBP monomer. Because mutant pro-MBP(C104S/C107S) forms a fully inhibited 2:2 complex with PAPP-A at a rate similar to the rate for wild-type pro-MBP (Fig. 2, A and B), we conclude that the process of pro-MBP dimerization occurs independently of other disulfide rearrangements. Importantly, however, dimerization of pro-MBP does require an interaction between PAPP-A and pro-MBP, as pro-MBP does not form dimers in the absence of PAPP-A. The presence of intermediate PAPP-A⅐pro-MBP 2:1 complexes (Fig. 1A, lane 3) shows that a pro-MBP subunit, once complexed to PAPP-A, is able to dimerize with a second pro-MBP subunit. Additionally, we have ruled out that Cys-104 PM and Cys-107 PM function to catalyze disulfide bonding between PAPP-A and pro-MBP, although they conform to the Cys-Xxx-Xxx-Cys motif, known from the active site of enzymes involved in disulfide exchange reactions (37).
Based on our findings, we have depicted the formation of the PAPP-A⅐pro-MBP complex in a hypothetical model (Fig. 9). We suggest that first, the pro-MBP disulfide bond Cys-51 PM -Cys-169 PM is broken causing exposure of a reactive Cys-169 PM , which then forms the first interchain disulfide, Cys-169 PM -Cys-652 PA (step 1). Upon disruption of the pro-MBP disulfide Cys-89 PM -Cys-128 PM , Cys51 PM forms the second interchain disulfide, Cys-51 PM -Cys-381 PA (step 2). Last, independently of these events, pro-MBP dimerizes through formation of the two interchain bonds, Cys-104 PM -Cys-104 PM and Cys-107 PM -Cys-107 PM, from two Cys-104 PM -Cys-107 PM intrachain disulfides (step 3). This process may involve a hypothetical catalytic activity of the Cys-Xxx-Xxx-Cys motif, to which these residues belong. Importantly, the total redox balance upon complex formation is maintained, as each bond broken is accompanied by another bond formed. In vivo, this series of disulfide bond rearrangements may be mediated by a specific or a nonspecific catalyst, e.g. a disulfide reductase or a low molecular weight thiol compound.
In conclusion, we have determined the status of cysteine residues in the reactants, uncomplexed PAPP-A and pro-MBP. A total of six pro-MBP and two PAPP-A cysteine residues have been shown to change their status upon formation of the covalent PAPP-A⅐pro-MBP complex. Based on the status of the reactants, we investigated the role of individual cysteine residues upon complex formation by mutagenesis of specific cysteine residues, and we have proposed a model of how the PAPP-A⅐pro-MBP complex is formed. We have also found that the rate of complex formation is greatly influenced by the addition of micromolar concentrations of reductants. It is therefore possible that the activity in vivo of PAPP-A is controlled by the redox potential, and it is further tempting to speculate that such mechanism operates under pathological conditions of altered redox potential to control the biological activity of IGF. FIG. 9. Hypothetical model of the PAPP-A⅐pro-MBP complex formation. The pro-MBP monomer and the PAPP-A subunit with numbered cysteine residues (C) involved in the process of complex formation are depicted. Disulfide bonds are shown as connecting lines and free cysteines as spheres. Except for the final PAPP-A⅐pro-MBP complex, only one of the two subunit of the dimeric PAPP-A is shown. Further details are given in Fig.  4 and in the main text.