Overexpression of human procarboxypeptidase A2 in Pichia pastoris and detailed characterization of its activation pathway.

The cDNA of human procarboxypeptidase A2 has been overexpressed in the methylotrophic yeast Pichia pastoris and secreted into the culture medium by means of the alpha-mating factor signal sequence, yielding a major protein of identical size and N-terminal sequence as the wild-type form. Two other forms containing the proenzyme have also been overexpressed: one of them resulted from an incomplete processing of the signal peptide, whereas the other was a glycosylated derivative. Recombinant procarboxypeptidase A2 was purified to homogeneity, and it was shown that its mature active form displays functional properties similar to those of the enzyme directly isolated from human pancreas. The overall yield was approximately 250 mg of proenzyme or 180 mg of mature enzyme/liter of cell culture. The proteolysis-promoted activation process of the recombinant proenzyme has been studied in detail. During maturation by trypsin, the increase in activity of the enzyme is a rapid and monotonic event, which reflects the rate of the proteolytic release of the inhibitory pro-segment and the weaker nature of its interactions with the enzyme moiety compared with procarboxypeptidases of the A1 type. Three main forms of the pro-segment (96, 94, and 92 amino acids), with no inhibitory capability in the severed state, and a single mature carboxypeptidase A2 are produced during this process. No further proteolysis of these pro-segments by the generated carboxypeptidase A2 occurs, in contrast with observations made in other procarboxypeptidases (A1 and B). This differential behavior is a result of the extreme specificity of carboxypeptidase A2.

The cDNA of human procarboxypeptidase A2 has been overexpressed in the methylotrophic yeast Pichia pastoris and secreted into the culture medium by means of the ␣-mating factor signal sequence, yielding a major protein of identical size and N-terminal sequence as the wild-type form. Two other forms containing the proenzyme have also been overexpressed: one of them resulted from an incomplete processing of the signal peptide, whereas the other was a glycosylated derivative. Recombinant procarboxypeptidase A2 was purified to homogeneity, and it was shown that its mature active form displays functional properties similar to those of the enzyme directly isolated from human pancreas. The overall yield was ϳ250 mg of proenzyme or 180 mg of mature enzyme/liter of cell culture. The proteolysis-promoted activation process of the recombinant proenzyme has been studied in detail. During maturation by trypsin, the increase in activity of the enzyme is a rapid and monotonic event, which reflects the rate of the proteolytic release of the inhibitory pro-segment and the weaker nature of its interactions with the enzyme moiety compared with procarboxypeptidases of the A1 type. Three main forms of the pro-segment (96, 94, and 92 amino acids), with no inhibitory capability in the severed state, and a single mature carboxypeptidase A2 are produced during this process. No further proteolysis of these pro-segments by the generated carboxypeptidase A2 occurs, in contrast with observations made in other procarboxypeptidases (A1 and B). This differential behavior is a result of the extreme specificity of carboxypeptidase A2.
Pancreatic carboxypeptidases (CPs) 1 are digestive metalloenzymes involved in the hydrolysis of alimentary proteins and peptides from their C-terminal end. Their participation as proenzymes in the digestive cascade (promoted by limited proteolysis) is a well documented process (1)(2)(3). Also, their specificity classification between the A forms (CPA, with preference for apolar C-terminal residues) and the B forms (CPB, with preference for basic C-terminal residues) and the tertiary structures of both forms are well known (4). In recent years, there has been an increasing interest in the study of the synthesis, storage, activation, and three-dimensional structure of procarboxypeptidases (pro-CPs), the precursors of such proenzymes (4 -6).
The classification of metallocarboxypeptidases has been widened in the last few years with reports about new non-digestive pancreatic-like carboxypeptidases in different extra-pancreatic tissues and biological fluids, with the same evolutionary ancestors as pancreatic carboxypeptidases (4,(7)(8)(9)(10). Additionally, the traditional classification of pancreatic carboxypeptidases and their zymogens into the A and B forms has been expanded with the identification of the A1 and A2 isoforms in rat and humans (11,12). CPA 1 and CPA 2 differ in specificity for peptide substrates: the former (assignable to the traditional A form) shows a wider preference for aliphatic and aromatic residues, whereas the latter is more restrictive for aromatic residues; this reflects significant differences in the specificity pocket of the enzymes (13). Recently, we have reported the cloning and sequence analysis of the human pro-CPA 2 cDNA as well as its three-dimensional modeling (14). CPA 2 isoforms have also been reported in rat extra-pancreatic tissues such as brain, testis, and lung (15); these CPA 2 isoforms are shorter and have a distinct role from the pancreatic isoform. The high sequence identity found between human pro-CPA 2 and rat pro-CPA 2 (89% homology) as compared with human pro-CPA 1 (64% homology) corroborates the proposal that locates the appearance of the two isoforms by gene duplication before speciation of mammals (11).
Comparison of the prodomain structures in the family of pancreatic proenzymes shows close similarities in conformation between the A1 and A2 forms in regions assumed to be critical for their inhibition and proteolytic activation (6,14) and significant differences from the corresponding regions in the B form (5). Accordingly, the A2 proenzyme could be expected to show a bimodal and slow proteolytic activation behavior, as previously reported for the A1 form (16), and to differ from the monotonic and quick activation behavior found for the B proenzyme (17). However, earlier proteolytic activation experiments carried out on natural pro-CPA 2 isolated from human or rat pancreas (12,18) do not fit with these expectations and assumptions. Therefore, this is an issue that requires clarification.
In this work, pro-CPA 2 has been overexpressed in Pichia pastoris to produce the protein in quantities amenable to the study of the structural and functional determinants of its behavior and activation. The methylotrophic yeast P. pastoris was chosen because of its high yield and capacity of secreting heterologous proteins when linked to the appropriate secretion signal (19). The development of the system reported here to obtain large quantities of fully activable human pro-CPA 2 should facilitate not only the characterization of this form, but probably also that of other structurally related forms to which the same procedure could be applied. It could also facilitate its potential biotechnological use, such as the large-scale production of carboxypeptidases able to act as activators of antitumoral prodrugs (20,21). The efficient expression of human pro-CPA 2 in P. pastoris has allowed us to investigate the different events in the proteolytic activation and processing of this proenzyme in detail and to compare them with the processes described in other pancreatic procarboxypeptidases. An overall maturation scheme of such zymogens emerges from this study.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, T4 DNA ligase, Vent polymerase, deoxynucleotide stocks, and N-glycosidase F were purchased from Boehringer Mannheim. Salts and media for Escherichia coli and P. pastoris growth were purchased from Difco. The P. pastoris expression kit was purchased from Invitrogen. Trypsin (treated with tosylphenylalanyl chloromethyl ketone) was from Worthington. Trifluoroacetic acid, N ␣ -p-tosyl-L-lysine chloromethyl ketone (TLCK), and N-(3-(2-furyl)acryloyl)-L-phenylalanyl-L-phenylalanine (FAPP) were from Sigma. Electrophoretic studies were carried out in a Bio-Rad Mini-Protean system. HPLC studies were carried out in a Waters chromatograph.
Plasmid Constructs-DNA manipulations were carried out essentially as described by Sambrook et al. (22) using E. coli strains JM83 and MC1061 as hosts. The cDNA of human pro-CPA 2 was amplified by polymerase chain reaction from a pUC9 vector to introduce an XhoI site at the 5Ј-end and an EcoRI site at the 3Ј-end of the cDNA using the following primers: sense primer, 5Ј-GTATCTCTCGAGAAAAGACTA-GAAACATTTGTGGGAGA-3Ј; and antisense primer, 5Ј-CTAGAAT-TCATGGCTCTTGTTTCTTCC-3Ј. After restriction enzyme digestion of the polymerase chain reaction product, the cDNA was cloned and subcloned into M13 pBluescript to confirm the entire sequence and the changes made in the polymerase chain reaction. pBluescript-pro-CPA 2 was digested by XhoI and EcoRI, and the cDNA was ligated to the P. pastoris shuttle vector pPIC9 between the 5Ј-promoter and the 3Јterminator of the alcohol oxidase gene (AOX1). pPIC9 provides the ␣-mating factor signal for secretion and the HIS4 gene for selection of the recombinant yeast clones. pHIL-D2 was also used for the expression and secretion of pro-CPA 2 using its own signal sequence. In the latter case, the original pUC9-pro-CPA 2 clone was digested and ligated to the P. pastoris pHIL-D2 shuttle vector using the EcoRI site of the polylinker.
Transformation and Selection of the Productive Clones-After linearization of the corresponding P. pastoris pPIC9-pro-CPA 2 and pHIL-D2prepro-CPA 2 vectors with BglII and NotI, respectively, the P. pastoris GS115 (his4) strain was transformed either by electroporation or by the spheroplast method. After simultaneously plating the transformants in MM and MD agar (1.34% yeast nitrogen base, 0.00004% biotin, and 0.5% methanol or 1% dextrose, respectively), those clones that suffered homologous recombination with the AOX1 sequence (slow growing in MM agar) were selected. To test for the most productive clones, colonies were grown in 10 ml of buffered liquid BMGY medium (1% yeast extract, 2% peptone, 90 mM potassium phosphate, pH 6.0, 1.34% yeast nitrogen base, 0.00004% biotin, and 1% glycerol) at 30°C for 4 days. Cells were collected by centrifugation and gently resuspended in 2 ml of buffered liquid BMMY medium (same as BMGY medium but containing 0.5% methanol instead of 1% glycerol) and cultured for another 2 days to induce the expression of the recombinant protein. The production of the clones was monitored after 6, 24, and 48 h by electrophoretic analysis of the supernatant on SDS-12% polyacrylamide gels. Western blotting was carried out as described previously using 1:500 anti-human pancreatic procarboxypeptidase antiserum (14). The functionality of the expressed protein was analyzed with the synthetic substrate FAPP after activation of the proenzyme with trypsin (at a 40:1 ratio by weight).
Expression and Purification of Recombinant Human Pro-CPA 2 -1-Liter shake-flask cultures were grown at 30°C for 4 days in buffered BMGY medium (20 -30 A units at 600 nm; 20 g of cells (dry weight)/ liter). Cells were collected by centrifugation at 1500 g, gently resuspended in 200 ml of BMMY medium, and cultured for another 2 days to induce the production of pro-CPA 2 . The culture medium was separated from the cells by centrifugation, and after equilibrating its ionic strength, it was processed through a two-step chromatographic purification: hydrophobic interaction chromatography on a butyl-Toyopearl 650M column eluted with a decreasing gradient of ammonium sulfate and fast protein liquid chromatography on a preparative anion-ex-change column (TSK-DEAE 5PW) eluted with a gradient of ammonium acetate in 30 mM MES, pH 5.7. The elution of recombinant pro-CPA 2 was observed at 9% ammonium sulfate and 6 mM ammonium acetate, respectively. The identity of recombinant pro-CPA 2 was confirmed after automated Edman degradation analysis of its N-terminal sequence and MALDI-TOF spectrometry analysis.
Activation Studies of Recombinant Human Pro-CPA 2 -Recombinant human pro-CPA 2 at 1 mg/ml in 50 mM Tris-HCl and 1 M ZnCl 2 , pH 8.0, was treated with trypsin at 40:1 and 400:1 (w/w) ratios at 0°C. At given times after trypsin addition, aliquots were removed for activity measurements, for electrophoretic reverse-phase HPLC and mass spectrometry analyses, and for quantification of the released amino acids. For activity measurements, 10 l of the activation mixture were added to 190 l of aprotinin (bovine pancreas trypsin inhibitor) at 0.1 mg/ml in 20 mM Tris and 0.1 M NaCl, pH 7.5, and 10 l of this new mixture were used to carry out spectrophotometric activity measurements with the FAPP substrate at 330 nm. For electrophoretic analysis, 20 l of the activation mixture were mixed with 2 l of 22 mM TLCK in water to reach a final trypsin inhibitor concentration of 2 mM. Each sample was immediately mixed with electrophoretic loading buffer containing 1% SDS and 3% ␤-mercaptoethanol, heated at 90°C for 1 min, and stored at Ϫ20°C until analysis. Electrophoresis was carried out on Tricinepolyacrylamide gels (23). For HPLC and mass spectrometry analyses, 90-l samples were removed from the activation mixture, made 0.5% in trifluoroacetic acid to inhibit proteolysis, and immediately chromatographed or kept at Ϫ20°C for subsequent analysis. For quantitation of the amino acids released into the activation mixture, 90-l samples were taken (2 nmol of initial pro-CPA 2 ) and mixed with trifluoroacetic acid up to a final concentration of 0.5%. 1.5 nmol of norleucine were added as a quantitative reference before the addition of 3 volumes of ethanol to precipitate proteins and large peptides. The supernatant was lyophilized and analyzed for amino acid composition.
Chromatographic Analysis by Reverse-phase HPLC-Samples removed from the activation mixtures were analyzed by reverse-phase HPLC on Vydac C 4 supports. A 214TP54 column (250 ϫ 4.6 mm, 5-m particle size, 0.3 m-pore) was used, and elution was followed at 214 nm. Chromatographies were performed in 0.1% trifluoroacetic acid with an eluting linear gradient between water (solvent A) and acetonitrile (solvent B) according to the following steps: 10% solvent B from 0 to 10 min, 10 -32% solvent B from 10 to 30 min, and 32-52% solvent B from 30 to 130 min.
Mass Spectrometry-The activation mixtures were analyzed by mass spectrometry with a MALDI-TOF spectrometer (Biflex with Reflectron, Bruker). 50 pmol of each sample in 1 l were mixed with 1 l of 50% synapinic acid as a matrix and loaded.
Amino Acid Analysis-Amino acid analyses were carried out by the 4-dimethylaminoazobenzene-4Ј-sulfonyl derivatization method (24) using materials and protocols from Beckman. A reverse-phase HPLC NovaPak C 18 column was used to separate the amino acids and oligopeptides produced during the activation.
N-terminal Sequence Analysis-After analysis by SDS-PAGE or analysis and purification by HPLC, N-terminal sequence analysis of pro-CPA 2 and the activation products was performed by blotting the samples on polyvinylidene difluoride membranes, followed by direct analysis on a Beckman LF3000 Protein Sequencer.

Construction of the Expression Vector and Transformation
into P. pastoris-A 1254-base pair cDNA encoding human preprocarboxypeptidase A2 has been cloned from a pancreatic library by immunological and radioactive approaches (14). This cDNA codes for a short leader signal peptide of 16 amino acids, a pro-segment of 96 amino acids, and an active enzyme of 305 amino acids (see below). In this work, this cDNA was modified by polymerase chain reaction to add an XhoI site at the 5Ј-end and an EcoRI site at the 3Ј-end to amplify only the proenzyme. The complete human prepro-CPA 2 cDNA was also cloned into the EcoRI site of the P. pastoris pHIL-D2 expression vector to test whether the native signal sequence of pro-CPA 2 could release the proenzyme into the extracellular medium.
Transformation of the P. pastoris GS115 (his4) strain with the linearized vectors was carried out by the spheroplast and electroporation methods. Both gave a similar number of transformants, which were screened for histidinol dehydrogenase (His ϩ ) phototrophy by plating on a dextrose-based medium without histidine supplementation. Nearly 20% of the His ϩ clones from the transformation showed reduced growth on methanol as the sole carbon source (slow growing, Mut Ϫ ), indicating the integration of the expression cassette into the AOX1 gene.
Overexpression and Purification of Recombinant Human Pro-CPA 2 -The clones transformed with the ␣-MF-pro-CPA 2 fusion product secreted a dominant 45-kDa protein in the P. pastoris supernatant upon induction by methanol. Those transformed with pHIL-D2-prepro-CPA 2 did not secrete any protein into the medium. The 45-kDa protein was found to correspond to human pro-CPA 2 by Western blot analysis. Upon induction with 0.5% methanol, the Mut Ϫ phenotypes expressed more protein than the Mut ϩ phenotypes. However, a 3-fold increase in protein secretion was observed in the Mut ϩ phenotypes when induction was assayed with 5% methanol (data not shown). One of these high productivity clones was selected for large-scale production of human pro-CPA 2 .
Analysis of the production of human pro-CPA 2 in the above system by SDS-PAGE and Western blotting is shown in Fig. 1. After 2 days of induction by methanol, a strong band of the proenzyme and a faint band corresponding to the 34-kDa active form were detected in the intracellular soluble part of the culture. This observation is in accordance with previous reports about the existence of proteases with trypsin-like activity in P. pastoris (25). After 30 h of methanol induction, in addition to the dominant 45-kDa pro-CPA 2 protein, a smear of high relative molecular mass, immunodetected by human procarboxypeptidase antiserum, was visible (Fig. 1B) in the lanes corresponding to the extracellular medium. It corresponds to the ␣-MF-pro-CPA 2 fusion product, which contains N-linked glycosylations in the signal peptide (26) and is not cleaved by the P. pastoris KEX2 processing protease. After purification (see below) and treatment of the sample with N-glycosidase F, conversion of the smear into a lower molecular mass band that corresponds to the ␣-MF-pro-CPA 2 fusion product was observed (Fig. 2). The different protein products was finally identified by N-terminal sequencing of the blotted SDS-polyacrylamide bands (data not shown). Trypsin treatment gave rise in all cases to a CPA 2 (Fig. 2) and to activation peptides (not shown in the figure) with the expected molecular masses, i.e. 34 and 10 -11 kDa, respectively.
As reported above, the human pro-CPA 2 gene was cloned behind the ␣-MF peptide to accomplish secretion of the recombinant protein. N-terminal sequence analysis of secreted recombinant human pro-CPA 2 showed the occurrence of heterogeneity in the processing of the ␣-MF precursor. The deduced processing targets, located at the boundary between the ␣-MF precursor and pro-CPA 2 , are indicated in Fig. 3. Native pro-CPA 2 was the major form found in all cultures, but different unspecific cleavage targets for endopeptidases were found in every induction performed on the same clone. Since the system used has been pushed to overexpress the recombinant protein at Ͼ300 mg/liter, the processing machinery might be unable to properly process all the material.
A chromatographic approach was developed to separate the differently processed forms of human recombinant pro-CPA 2 and to isolate the native form. It was based upon a two-step purification scheme, with atmospheric hydrophobic interaction chromatography on a butyl column applied first, followed by fast protein liquid anion-exchange chromatography to separate the different forms (see "Experimental Procedures"). Starting from a cell culture of a Mut ϩ clone that achieved a cell density of 20 g of cells (dry weight)/liter, an overall yield of 250 mg of total pure pro-CPA 2 was normally obtained. The correctly processed pro-CPA 2 form was generally the principal one, with yields varying from 80 to 200 mg/liter, although in certain cultures, the glycosylated and incorrectly processed forms could account for as much as 60% of the purified material. In contrast, treatment of total pro-CPA 2 with trypsin before the anion-exchange chromatography yielded ϳ180 mg of homogeneous, correctly processed CPA 2 /liter of cell culture. Both the purified recombinant pro-CPA 2 and CPA 2 forms were fully functional, with enzymatic properties similar to those previously reported for the natural forms isolated from human pan- creas (12). Recombinant CPA 2 and the enzyme isolated from pancreas showed the same maximum specific activity of ϳ85 mol of FAPP substrate hydrolyzed per min/mg of protein.
Activation of Recombinant Human Pro-CPA 2 with Different Proteases-Analysis of the action of trypsin, chymotrypsin A, and elastase (the major active endoproteolytic counterparts in pancreatic secretion) upon correctly processed recombinant pro-CPA 2 indicates that trypsin is, by far, the most efficient activator (data not shown). At the pro-CPA 2 /endoprotease ratio used, 90% activation of the former was achieved by trypsin in ϳ10 min, whereas the same level of activation was only achieved after 125 min of treatment with elastase; in the latter time span, chymotrypsin was able to generate only 10% of active CPA 2 . This is in agreement with previous studies that reported trypsin as the main enzyme responsible for the activation of pancreatic procarboxypeptidases (3).
Trypsin Activation of Recombinant Human Pro-CPA 2 -The action of trypsin at 37°C on recombinant pro-CPA 2 at a 40:1 (w/w) pro-CPA 2 /trypsin ratio is excessively quick for detailed mechanistic analysis, in agreement with previous studies on natural human proenzymes (12). The study of the activation process was therefore carried out at 0°C and at a 400:1 ratio. Under these conditions, the appearance of carboxypeptidase activity was measured with the synthetic substrate FAPP; it followed a quick and monotonic activation curve, which can be fitted to a pseudo first-order kinetics (Fig. 4A). Using Tricine/ SDS-PAGE, the rapid proteolytic severing of the pro-segment, the concomitant appearance of CPA 2 , and a straight correlation of this proteolysis with the monotonic activity curve were observed (Fig. 4B). Nearly all pro-CPA 2 was converted into its active form in 20 min, and a maximum value in the activity curve was achieved. The generated pro-segment appeared as a single band on Tricine/SDS-PAGE (Fig. 4B), although mass spectrometry analysis indicated heterogeneity for this species (see below).
The N-terminal sequence of the electrophoretic CPA 2 band, obtained from different samples at different activation times and transferred to a polyvinylidene difluoride membrane, was always Ser-Gly-Asn-, a fact that indicates that the Arg 96 -Ser 97 peptide bond is the first point of cleavage of the proenzyme by trypsin. From this result and from the increase in the intensity of the CPA 2 band in parallel with the appearance of CPA 2 activity in the medium and the corresponding decrease in the intensity of the pro-CPA 2 band (Fig. 4B), it can be assumed that the cleavage of the Arg 96 -Ser 97 peptide bond is sufficient to generate all the carboxypeptidase activity. As a consequence, it can be concluded that the severed pro-segment does not inhibit CPA 2 , in contrast to what has been previously reported for the pro-CPA 1 forms (12,16,27).
To study the maturation process in more detail, the time course of degradation of the pro-segment was followed by reverse-phase HPLC. It was expected that this degradation would take place only at the C-terminal arginine-rich end of the pro-segment since N-terminal sequence analyses of the activation mixtures indicated the appearance of only two N termini throughout the process, one corresponding to the original proenzyme and one from the generated mature enzyme. The chromatographic analyses of the activation mixtures at different times are represented in Fig. 5, where the appearance and disappearance of three different protein fragments arising from the proteolysis of the pro-segment (labeled ␣, ␤, and ␥) can be observed. Mass spectrometry (MALDI-TOF) was used for the characterization of the molecular masses of these fragments (Table I). Their identification was facilitated by the knowledge of the sequence of pro-CPA 2 and by analysis of peptides and amino acids that were released into the medium by the action of trypsin and CPA 2 . In particular, it is worth mentioning that two dipeptides (Glu-Arg and Arg-Arg) were detected by mass spectrometry at early and late activation times and that no significant free amino acids were detected by parallel amino acid analysis of the protein-free activation medium. All of this information was used to establish and confirm the sequence of the cleavages during the proteolytic processing.
According to the above information, during the tryptic maturation of pro-CPA 2 , the first cleavage is at the C terminus of Arg 96 , producing the ␣-fragment (residues 1-96), as shown in Fig. 6. Subsequently and while some proenzyme molecules are still intact, a rapid cleavage occurs at Arg 94 of the ␣-fragment, giving rise to the ␤-fragment (residues 1-94) and to the release of a Glu-Arg dipeptide. The third cleavage (also attributable to trypsin, as are the former cleavages) occurs slowly at Arg 92 of the ␤-fragment, giving rise to the ␥-fragment (residues 1-92) and to the release of a second dipeptide, Arg-Arg. The lack of detection of free amino acids in the medium indicates the null action of generated CPA 2 on the ␣-, ␤-, or ␥-fragment. The ␥-fragment is resistant to further proteolysis under these conditions. Only a single CPA 2 species was detected throughout the proteolytic process by HPLC, N-terminal, and mass spectrometry analyses.

DISCUSSION
One of the aims of this study was the development of a highly efficient recombinant expression system for a native and activable form of human procarboxypeptidase A 2 in the P. pastoris heterologous system to study its proteolytic activation and maturation pathway in vitro. It is worth mentioning that pancreatic carboxypeptidases and their precursors are difficult to express in native and soluble forms in E. coli (28) and that reports about the expression of their precursor forms in eukaryotic cells (i.e. Saccharomyces cerevisiae) have indicated a moderate yield (Ͻ10 mg/liter) until now (21,29). In fact, the primary production of the proenzymes by recombinant approaches followed by proteolytic activation is probably the best strategy to obtain active carboxypeptidases given that it takes advantage of the high folding capability of the pro-segment in heterologous systems (29,30). The recent biotechnological interest in metallocarboxypeptidases as prodrug activators for cancer therapy (21, 31, 32) makes the availability of large-scale procedures to obtain pro-and carboxypeptidases, particularly the human forms, very useful. The efficient heterologous pro- FIG. 5. Analysis by reverse-phase HPLC of the fragments generated during tryptic activation of human pro-CPA 2 . At different times, aliquots were taken from the activation mixture under the same conditions as described in the legend to Fig. 4 and analyzed by reversephase chromatography in an HPLC system. The action of trypsin was stopped by adding 0.1% trifluoroacetic acid. 100 l of each sample were loaded on a Vydac C 4 reverse-phase column, and elution was followed at 214 nm. The ␣-fragment corresponds to residues 1-96, the ␤-fragment to residues 1-94, and the ␥-fragment to residues 1-92. PCPA2, pro-CPA 2 .

TABLE I
Analysis of the tryptic activation fragments of human pro-CPA 2 by mass spectrometry Values are the relative molecular masses derived by MALDI-TOF mass spectrometry analysis of samples taken at different times of the activation process, carried out under the same conditions as described in the legend to Fig. 4. The action of trypsin was stopped by adding 0.1% trifluoroacetic acid. The ␣-fragment corresponds to residues 1-96, the ␤-fragment to residues 1-94, and the ␥-fragment to residues 1-92. FIG. 6. Schematic representation of the cleavage points in the activation process of human pro-CPA 2 promoted by trypsin. The scheme shows the sequence of the ␣-helical connecting region between the globular activation domain and the active enzyme. Arrows indicate the tryptic cleavage points observed under the same conditions as described in the legend to Fig. 4, which released three activation fragments (␣, ␤, and ␥) of 96, 94, and 92 residues, respectively, as deduced by reverse-phase HPLC, N-terminal sequencing, and mass spectrometry studies. No cleavages were detected in other parts of the molecule. The activation sequence of pro-CPA 2 is denoted by asA 2 . duction of extra-pancreatic metalloprocarboxypeptidases, with many of them involved in important biological functions and some with a tertiary structure or domains structurally related to the pancreatic ones (4), could also facilitate their detailed characterization. Hence, the development of efficient heterologous expression systems for enzymes and proenzymes such as the one reported here is highly desirable.
The experiments designed to establish an easy and robust expression system capable of providing large amounts of soluble human pro-CPA 2 were restricted to shake-flask cultures, which rendered a production of ϳ250 mg of recombinant proenzyme/liter or 180 mg of active enzyme/liter. This production is more than enough for studies such as those presented here and for subsequent structural determinations as well as for other analytical and semipreparative purposes. However, P. pastoris can increase this production 10-fold by scaling up from shakeflask to high density fermentation (19).
The original target for the removal of the ␣-MF propeptide is the sequence KREAEAEA, which is cleaved by the yeast endopeptidase KEX2 after the dibasic peptide and undergoes subsequent elimination of Glu-Ala dipeptides by dipeptidyl aminopeptidase A (33). Thus, according to the design of the pPIC9-pro-CPA 2 expression plasmid, the pro-segment of the ␣-mating factor was expected to be cleaved after the dibasic residues Lys-Arg by a single KEX2 endopeptidase action. However, the N-terminal sequence of expressed recombinant pro-CPA 2 displayed microheterogeneity at the amino terminus, indicating that cleavage by the KEX2 endopeptidase is only partial and that other endoproteinases might be responsible for the unspecific cleavages (see Fig. 3). The microheterogeneity observed did not become an important problem since the principal product found in all of the cultures was native pro-CPA 2 . Part of the secreted fusion product appeared glycosylated; however, this product was converted to normal pro-CPA 2 after deglycosylation. In any case, all of these products can generate fully active CPA 2 by limited proteolysis.
A correctly processed ␣-MF propeptide has been reported for recombinant proteins expressed in P. pastoris without the Cterminal Glu-Ala extension (34,35); other reports indicated heterogeneity at the N terminus of the secreted proteins (i.e. aprotinin and coffee bean galactosidase), also resulting from different cleavage points in the processing of the ␣-MF prosegment (36,37). Thus, the microheterogeneity observed in some cases could be dependent on the nature of the recombinant protein expressed (33).
The trypsin-promoted maturation mechanism of human pro-CPA 2 , as derived from our studies, is shown in Fig. 6. Due to the presence of several arginines at the boundary region between the pro-segment and the enzyme moiety, a rapid tryptic cleavage at the most exposed arginine of this region and subsequent cleavages at the remaining trypsin targets should be expected. Alternatively, simultaneous cleavages could occur, but our results indicate that it is the first mechanism that takes place. In contrast to other procarboxypeptidases studied, the Arg 96 -Ser 97 peptide bond is the first target for trypsin action observed in pro-CPA 2 , releasing a pro-segment of 96 residues into the medium (␣-fragment in Figs. 5 and 6). From sequence alignments and structure comparisons (14), this peptide bond is considered to belong to the enzyme moiety in porcine pro-CPA 1 and pro-CPB, where the cleavage occurs two residues farther along N-terminally (16,17). The released primary activation fragment and mature enzyme are therefore two residues larger and shorter, respectively, in pro-CPA 2 .
The second step in the proteolytic processing is the rapid elimination of the C-terminal Glu-Arg dipeptide end from the ␣-fragment. The product of this cleavage (␤-fragment in Figs. 5 and 6) is due to a tryptic action on arginine 94 of the ␣-fragment, as shown by mass spectrometry analysis and by the lack of release of free amino acids. The action of trypsin in this conversion is not as fast as in the first cleavage, probably due to the less efficient tryptic endoproteinase activity near the C-terminal end of proteins, and takes place while some proenzyme molecules are still intact. The third and last sequence of 92 residues (␥-fragment in Figs. 5 and 6), resistant to further proteolysis under these conditions, is generated by trypsin action in the long run, releasing an Arg-Arg dipeptide into the activation medium. It is also worth commenting upon that no release of free amino acids into the medium is observed in the course of activation, in contrast to previous observations in the corresponding activation processes of pro-CPA 1 and pro-CPB (16,17,38). This indicates the high specificity shown by human CPA 2 , which is unable to trim the C-terminal arginines from its activation peptides, even at the high concentration of these species in the activation medium.
According to this study, it seems clear that the pro-segments of pro-CPA 2 released into the medium do not inhibit the activity of the active enzyme, even in its longer (primary) form (96 amino acids), since CPA 2 reaches total activity following a rapid hyperbolic activation curve at activation times when a substantial concentration of the ␣-fragment sequence is still present in solution. In contrast to this behavior, porcine pro-CPA 1 , which shows a slower and biphasic activation process, needs subsequent trimmings at the C-terminal end and a second tryptic cleavage inside the globular domain of the prosequence to achieve full carboxypeptidase activity. In this sense, pro-CPA 2 behaves in a way that is more similar to pro-CPB (17).
X-ray crystallography and modeling studies have shown the structural similarities between pro-CPA 1 and pro-CPA 2 , both sharing a long C-terminal ␣-helix at the region connecting the globular activation domain with the active enzyme, and the differences of these two proenzymes from pro-CPB, whose corresponding connecting region has a much shorter ␣-helix and is less structured overall (5,6,14). These differences were proposed to be the primary determinant responsible for the diversity observed in the rates of activation of pro-CPA 1 and pro-CPB, arguing that the larger regular structure of the connecting region in pro-CPA 1 would render the interactions with the enzyme more stable and make the structural relaxation needed for full release slower (4,39). Modeling studies with pro-CPA 2 (14) have shown that such a connecting region is also structured in a long ␣-helix, at least the same size as in porcine pro-CPA 1 . Taking into account the former evidence and the observation of a rapid and monotonic activation curve for pro-CPA 2 , it can be concluded that the folding of the connecting region of procarboxypeptidases in a long ␣-helix does not give rise to inhibitory activation fragments and to slow activation processes by itself. Other structural determinants, such as surface complementarity and electrostatic and Van der Waals interactions, should be considered to evaluate the stability of the bimolecular complex between the pro-segment and the enzyme moiety once the first proteolytic cleavage has taken place. Preliminary x-ray diffraction studies 2 on human pro-CPA 2 show that the connecting region is highly structured and that extensive interactions take place between the enzyme moiety and both the globular domain region and the connecting region of the pro-segment. These observations confirm that the latter two regions are responsible for the activation behavior of pro-CPA 2 and for its functional differences from other procarboxypeptidases.