Heterologous expression of human cholecystokinin in Saccharomyces cerevisiae. Evidence for a lysine-specific endopeptidase in the yeast secretory pathway.

Precursors of the human regulatory peptide cholecystokinin (CCK) have been expressed in Saccharomyces cerevisiae, and the post-translational processing of secreted CCK-related products analyzed. Recombinant plasmids expressing native human prepro-CCK and a hybrid molecule encompassing the prepro leader of the yeast alpha-mating pheromone fused to pro-CCK were examined. The latter construct resulted in considerably higher levels of pro-CCK secretion and was therefore analyzed in more detail. Two of the protein modifications essential for CCK bioactivity, C-terminal alpha-amidation and tyrosyl sulfation, were not detected in S. cerevisiae. Proteolytic cleavage of pro-CCK occurred C-terminally of three basic sites; (i) Arg105-Arg106 which, upon exposure to carboxypeptidase activity, leads to the production of glycine-extended CCK; (ii) Arg95 to produce CCK-8 related processing intermediates; and (iii) Lys81 resulting in CCK-22 related products. To elucidate which protease(s) are involved in these endoproteolytic cleavage events, pro-CCK was expressed in yeast mutants lacking various combinations of the Mkc7, Yap3, and Kex2 proteases. Only in S. cerevisiae strains deficient in Kex2 function was any of the above mentioned pro-CCK cleavages abolished, namely processing at the Arg105-Arg106 and Arg95 sites. This suggests that mammalian Kex2-like serine proteases may process pro-CCK at single arginine residues. Our data suggests that an as yet uncharacterized endopeptidase(s) in the S. cerevisiae secretory pathway is responsible for the lysine-specific cleavage of pro-CCK.

lytic processing (Rehfeld et al., 1988). Bioactive CCKs possess the same amidated C terminus which results from endoproteolytic cleavage C-terminally of the dibasic Arg 105 -Arg 106 residues followed by removal of these basic amino acids by carboxypeptidase digestion, exposing Gly 104 to peptidyl-monooxygenase activity (Fig. 1A). In contrast, endoproteolytic processing of pro-CCK is far more heterogeneous at the N terminus with examples of  having been identified in various tissues or plasma samples ( Fig. 1A) (Rehfeld and Hansen, 1986;Rehfeld et al., 1988;Eberlein et al., 1992). Many of these forms of CCK are produced via cleavage of pro-CCK C-terminally of single arginine residues (Rehfeld and Hansen, 1986;Eberlein et al., 1992). However, CCK-22, the predominant hormonal form of CCK, is produced by endoproteolytic processing at Lys 81 . Consequently this lysine processing event is of crucial significance for understanding the biology of CCK (Liddle et al., 1984(Liddle et al., , 1985Cantor and Rehfeld, 1987;Rehfeld, 1994;Paloheimo and Rehfeld, 1995).
Many foreign proteins have been expressed in yeast (Hadfield et al., 1993). Often these proteins are directed through the yeast secretory pathway, either by the endogenous N-terminal signal sequence of the protein or via production of a hybrid protein which includes the leader region of a secreted yeast preproprotein. Passage through the secretory pathway exposes proteins to a number of post-translational capabilities, including disulfide bridge formation, glycosylation and proteolytic processing (Hadfield et al., 1993). The best characterized yeast endopeptidase is the Kex2 serine protease, which was identified for its ability to cleave precursors of the ␣-mating pheromone and killer toxin at dibasic sites (Julius et al., 1984). Overexpression studies in kex2 mutants resulted in the isolation of the yeast aspartyl protease (Yap3), which is able to rescue the mating deficiency of kex2 deficient Saccharomyces cerevisiae (Egel-Mitani et al., 1990). More recently a third endopeptidase, Mkc7, with considerable homology to Yap3 has been detected (Komano and Fuller, 1995). This protease, which was isolated as a multicopy suppressor of the cold sensitive growth phenotype of kex2 mutants, has the ability to cleave an internally quenched flurogenic substrate C-terminally of a Lys-Arg site (Komano and Fuller, 1995).
In the present study, we have examined the extent of posttranslational processing of human pro-CCK expressed and secreted from S. cerevisiae. Evidence was found for an, as yet, uncharacterized endopeptidase in yeast which is able to cleave CCK at a single lysine residue to produce CCK-22, a predominant form of circulating CCK in mammals.

MATERIALS AND METHODS
Yeast Strains, Transformations, and Growth Conditions-Yeast strains used in this study are listed in Table I. S. cerevisiae grown in YPD medium were transformed with plasmid DNA using the lithium acetate procedure (Ito et al., 1983). Transformants were selected and * 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. Section 1734 solely to indicate this fact.
Plasmid Constructs-A full-length cDNA clone encoding human prepro-CCK was generously supplied by Karin Pedersen. The S. cerevisiae ␣-mating pheromone 1 gene (MF␣1) was kindly provided by Ira Herskowitz (Kurjan and Herskowitz, 1982). Two pro-CCK expression plasmids were constructed in the URA3-2 plasmid pJ399, in which gene transcription is driven by the yeast glyceraldehyde-3-phosphate dehydrogenase promoter (Bitter and Egan, 1984) and terminated by the yeast TPI1 termination signal (Alber and Kawasaki, 1982). Plasmid pJ399 was constructed from pYES2.0 (Invitrogen).
The coding region of human prepro-CCK was amplified by PCR using the Pwo DNA polymerase (Boehringer Mannheim). The forward primer (5Ј-ACGTGAATTCTAAACTAAAAATGAACAGCGGCGTGTGCCTGTG-CGTG-3Ј) resulted in an EcoRI site 10 nucleotides upstream of the start codon, while the reverse primer (5Ј-ACGTTCTAGACTAGGAGGGGTA-CTCATACTCCTCGGC-3Ј) introduced an XbaI site immediately downstream of the TAG stop codon. The PCR product was purified after agarose gel electrophoresis using the Sephaglas Band Prep procedure (Pharmacia), cleaved with EcoRI and XbaI (Boehringer Mannheim), and cloned into pJ399.
To ensure high secretion levels of CCK, we constructed a second expression plasmid where the sequence encoding the prepropeptide of the MF␣1 yeast ␣-factor (Kurjan and Herskowitz, 1982) was fused to the coding region of pro-CCK (amino acids 31-115) (Fig. 1B). This expression plasmid was constructed by a two-step process. First, the yeast MF␣1 gene was amplified by PCR (forward primer, 5Ј-ACGTGA-ATTCTAAACTAAAAATGAGATTTCCTTCAATTTTTACTGCAG-3Ј; reverse primer, 5Ј-ACGTTCTAGATTAGTACATTGGTTGGCCGGGTTT-TACA-3Ј) and cloned into pJ399 as described above. The resulting plasmid was cleaved with HindIII and XbaI to remove the coding region of the four tandem ␣-factor repeats, but leaving the sequence encoding the MF␣1 prepropeptide. The region encoding pro-CCK was amplified using PCR (forward primer, 5Ј-ACGTTCTAGACTAGGAGG GGTACT-CATACTCCTCGGC-3Ј; reverse primer was the same as used for amplification of the coding region of prepro-CCK) resulting in a product with a 5Ј HindIII site and a 3Ј XbaI site that could be ligated into the pJ399 plasmid containing the coding region of prepro-MF␣1. The resulting plasmid encodes a hybrid protein that consists of the first 89 amino acids of prepro-MF␣1 and the last 85 amino acids of prepro-CCK starting at Ser 31 (Fig. 1B). It was expected that pro-CCK would be liberated in the yeast secretory pathway by Kex2 cleavage C-terminally of the Lys-Arg site in the ␣-factor prepropeptide, followed by two dipeptidyl aminopeptidase digestion steps to remove the Glu-Ala-Glu-Ala sequence ( Fig. 1B) (Hadfield et al., 1993). To ensure the coding nucleotides had not been altered during the PCR steps, plasmid inserts were exposed to double stranded DNA sequencing with the dideoxy chain termination method using the Sequenase 2.0 DNA sequencing procedure (U. S. Biochemical Corp.).
Radioimmunoassays-Radioimmunoassay (RIA) analysis was used to detect and partially characterize pro-CCK products secreted from yeast transformants. Four monospecific antisera raised in rabbits which binds various processing intermediates of CCK were utilized; Ab 2609, Ab 3208, Ab 7270, and Ab 89009 (Table II). Measurements were performed as described previously (Rehfeld, 1978a;Hilsted and Rehfeld, 1986;Paloheimo and Rehfeld, 1994), either directly on the liquid media used for propagation of yeast transformants or on fractions collected from gel chromatography. The molecules used as tracers and standards in RIA experiments are presented in Table II.
The presence of CCK which had not been processed at the Arg 105 -Arg 106 site, referred to as C-terminally extended CCK, was detected by RIA with Ab 3208 after the samples had been treated with trypsin and carboxypeptidase B (T/C). Trypsin cleaves C-terminally of the Arg 105 -Arg 106 site, and carboxypeptidase B removes these residues leaving glycine-extended CCK. This procedure involves incubating the sample with an equal volume of 2 mg/ml trypsin (Boehringer Mannheim) in 0.1 mM sodium phosphate buffer (pH 7.5) at room temperature for 30 min. The reaction was terminated by boiling for 10 min, after which the samples were cooled to room temperature. Carboxypeptidase B (Boehringer Mannheim) was then added to a final concentration of 100 g/ml, the reaction was incubated at room temperature for 30 min, and terminated by a second boiling step. Processing products of pro-CCK larger than CCK-22 were detected by RIA analysis with Ab 89009 after samples were exposed to trypsin as described above.
Gel Chromatography-Yeast transformants grown to late exponential phase were centrifuged at 3,000 ϫ g for 5 min to collect the cells and 100 -500 l of the culture supernatant was loaded directly onto a Sephadex G-50 superfine (Pharmacia) column (1 ϫ 100 cm) at 4°C. The sample was eluted in VBA buffer (20 mM barbital buffer, 0.6 mM thiomersal, and 0.11% bovine serum albumin) at a flow rate of 2.4 ml/h, with 800-l fractions being collected every 20 min. The column was calibrated with trace amounts of [ 125 I]albumin and 22 NaCl to indicate the void and total volumes, respectively.
Anion exchange chromatography was conducted on a 5 ϫ 50-mm Mono-Q HR/R anion exchange column (Pharmacia) which had been pre-equilibrated with 20 mM sodium phosphate buffer (pH 8). The column was eluted with a gradient from 0 to 20% acetic acid over 70 min, with 1-ml fractions being collected every minute.  (Takahashi et al., 1986). The sites of known basic amino acid directed proteolytic cleavage events, the species of CCK produced, as well as processing site to remove the N-terminal hydrophobic signal sequence, are indicated (Rehfeld and Hansen, 1986;Cantor and Rehfeld, 1987;Rehfeld et al., 1988;Eberlein et al., 1992). B, diagram of the hybrid protein encoded by the prepro-MF␣1-pro-CCK expression construct. This molecule consists of the first 89 amino acids of prepro-MF␣1 and the C-terminal 85 amino acids of pro-CCK. Relevant pro-CCK proteolytic processing sites are shown. Numbering is relative to the residues in prepro-CCK and not the hybrid protein. MAT ␣can1-100 ade2-101 his2-11,-15 leu2-3-112 trp1-1 ura3-1 mkc7 ⌬::HIS3 yap3⌬::LEU2 kex2⌬::TRP1 Komano and Fuller (1995) a YNG318 is a derivative of strain JC482 described by Cannon and Tatchell (1987).
were used for calibration. Samples were dried under vacuum and resuspended in VBA buffer before RIA analysis.
Purification of CCK Secreted from Yeast Transformants-A five-step procedure was utilized to purify CCK secreted from transformants of S. cerevisiae.
Step 1: the wild type yeast YNG318 transformed with the plasmid encoding the hybrid prepro-MF␣1-pro-CCK construct was grown in 200 ml of selective media to late exponential phase and cells removed by centrifugation at 3,000 ϫ g for 5 min. The supernatant was adjusted to pH 3 with trifluoroacetic acid and centrifuged at 10,000 ϫ g for 10 min. The cleared supernatant was loaded onto five Sep-Pak plus C18 columns (Waters) connected in series, which had been pre-equilibrated with 10 ml of 0.5% trifluoroacetic acid in acetonitrile followed by 10 ml of 0.5% trifluoroacetic acid in H 2 O. Material was passed through the columns at a flow rate of 2 ml/min. The column was eluted sequentially with 10-ml solutions of 0.5% trifluoroacetic acid in 0, 10, 30, 50, 70, 90, and 100% acetonitrile. The fractions collected were diluted in distilled water and analyzed by RIA using Ab 3208.
The same procedure was utilized with the mkc7 yap3 (HKY24) and kex2 yap3 (ME938) double mutants expressing the prepro-MF␣1-pro-CCK construct. However, the starting quantity was 2 liters of culture medium and CCK-immunoreactive fractions were detected using RIA analysis with Ab 89009 and/or Ab 3208 after the samples had been treated with trypsin and carboxypeptidase B.
Step 2: immunoreactive fractions from Sep-Pak purification were concentrated under vacuum to a final volume of 1 ml, adjusted to pH 8 with NaOH and loaded onto a Sephadex G-50 superfine column as described previously.
Step 3: the next stage of purification involved anion exchange chromatography on a Mono-Q column and was performed as outlined above.
Step 4: the fractions containing CCK immunoreactivity were further processed by reversed phase high performance liquid chromatography using a C 8 column (2.1 ϫ 150 mm) (Vydac) at 50°C, eluted at 0.2 ml/min with a gradient from 0.1% trifluoroacetic acid in acetonitrile (0.5%/min). Fractions were collected every 0.5 min.
Step 5: selected immunoreactive fractions from Step 4 were subjected to a further reversed phase high performance liquid chromatography step on a Vydac C 18 column using the same solvents but employing a gradient of 0.2%/min. Peak fractions were collected manually (monitored at 214 nm).
Mass Spectrometry and Peptide Sequencing-Molecular masses of purified peptides were determined by matrix-assisted laser desorption mass spectrometry performed in a Biflex instrument (Bruker-Franzen) using ␣-cyano-4-hydroxy cinnamic acid as a matrix. Both positive and negative ions were analyzed in the linear mode using external calibration to give an accuracy of 0.1%. Individual peptides were subjected to amino acid sequence analysis using an automatic protein sequencer (Procise 494A, Applied Biosystems) equipped with an on-line high performance liquid chromatography system for detection of the amino acid phenylthiohydantoins.
Metabolic Labeling and Immunoprecipitation-Transformants of the YNG318 strain expressing the prepro-MF␣1-pro-CCK fusion peptide were grown overnight at 30°C in selective media. The next morning the cells were collected by centrifugation and resuspended in 500 l of synthetic complete media lacking uracil and methionine, and supplemented with 500 Ci of L-[ 35 S]methionine (Amersham). After the cells were incubated for a further 30 min at 30°C they were washed twice in 50 mM Tris-HCl (pH 7.4) before being resuspended in 50 mM Tris-HCl (pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride, 50 g/ml aprotinin, 20 g/ml leupeptin, and 2 g/ml pepstatin. An approximate equal volume of pre-chilled acid washed glass beads (Sigma) was added and the sample vortexed 10 times for 30 s with cooling on ice between each vortexing. SDS was then added to a final volume of 1% and the cell lysate boiled for 3 min. An equal volume of 2 ϫ sample buffer (100 mM Tris-HCl (pH 8.3), 380 mM NaCl, 12 mM EDTA (pH 8), 5% Triton X-100) was included and a cleared supernatant prepared by centrifugation at 13,000 ϫ g for 10 min at 4°C.
Immunoprecipitations were performed at 4°C for 2 h following the addition of 5 l of Ab 7270 which binds to glycine extended CCK (Table  II). Immune complexes were bound to Protein A-Sepharose (Pharmacia) during a second 2-h incubation at 4°C with gentle agitation. The sample was then centrifuged at 13,000 ϫ g for 10 s and the pellet washed 8 times in 1 ϫ sample buffer, boiled for 5 min, and loaded directly onto a Sephadex G-50 superfine column. Radiolabeled CCK was detected in the column fractions by a combination of liquid scintillation counting and RIA analysis utilizing Ab 3208.

Secretion of CCK from Yeast Transformants-Transformants
of the YNG318 strain expressing prepro-CCK and the prepro-MF␣1-pro-CCK fusion protein were grown to late exponential phase and the culture media was subjected to RIA analysis for the presence of glycine-extended, C-terminally extended, and ␣-amidated CCK. In each case two independent transformants were analyzed. No major differences in CCK immunoreactivity were detected with transformants harboring the same plasmid constructs (Table III).
Secretion of glycine-extended CCK was determined by immunoreactivity to Ab 3208 (Table III). Both constructs resulted in the secretion of glycine-extended CCK from yeast transformants. However, approximately hundredfold higher levels of secretion were seen when CCK was expressed as a fusion to the ␣-factor prepropeptide compared with the native prepro-CCK. Treatment of the samples with trypsin and carboxypeptidase B increased the quantity of glycine-extended immunoreactivity at least 2-fold indicating that considerable quantities of secreted CCK was C-terminally extended (Table III).
In contrast to glycine and C-terminally extended CCK, no ␣-amidated CCK was found in media supernatants of yeast expressing pro-CCK, as indicated by the absence of immunoreactivity to Ab 2609 in media in which yeast expressing the prepro-CCK or the prepro-MF␣1-pro-CCK constructs had been TABLE III Expression levels of CCK secreted from transformants of the yeast strain YNG318 Secretion levels of each construct, prepro-CCK and prepro-MF␣1-pro-CCK, were determined for two independent transformants. T/C indicates the samples were treated with trypsin and carboxypeptidase B before radioimmunoassay analysis, CCK-Cterm refers to C terminally extended CCK. Data are presented as means Ϯ S.E. for six independent cultures. No background immunoreactivity from yeast proteins was detected for any of the antibodies used.   Paloheimo and Rehfeld (1994) cultured (Table III). Analysis of the Proteolytic Processing of CCK Expressed in Yeast-To further characterize the extent of endoproteolytic processing of CCK produced in S. cerevisiae, secreted material of strain YNG318 expressing the prepro-MF␣1-pro-CCK fusion protein were subjected to gel chromatography. Analysis of the fractions collected for glycine-extended CCK uncovered two major peaks of CCK immunoreactivity, one at an elution constant (K D ) of 0.75, the other 1.15 (Fig. 2A). These K D values are similar to the known elution constants of CCK-22-Gly and CCK-8-Gly, respectively, determined by the same procedures (Cantor and Rehfeld, 1987). To confirm the identity of the CCK-immunoreactive products they were purified to homogeneity and subjected to mass spectrometry and protein sequence analysis. As predicted, the latter immunoreactive peak was identified as CCK-8-Gly and the earlier as CCK-22-Gly (Fig. 3,  A and B). It should be noted that the peptides isolated during this study were oxidized to varying degrees (Fig. 3). All human CCK peptides contain methionine and tryptophan residues. It is a common observation that methionine residues are prone to oxidation in the presence of oxygen. In addition, we often find that tryptophan containing peptides undergo oxidation in the mass spectrometer. 2 Both processes appear to be more pronounced in short peptides and at low peptide concentrations.
Since the initial RIA data indicated that a considerable quantity of CCK secreted from yeast was not processed at the C terminus (Table III), the column fractions were treated with trypsin and carboxypeptidase B before RIA analysis with Ab 3208. This resulted in the detection of two additional CCKimmunoreactive peaks not present in untreated fractions, one at a K D of 0.6 and the other at a K D of 1 (Fig. 2B). Thus, it would appear the new peaks represented C-terminally extended FIG. 3. Mass spectra of some of the purified CCK molecules secreted from yeast transformants. A, CCK-8-Gly (calculated molecular mass 1121) secreted from YNG318 expressing the prepro-MF␣1pro-CCK construct, the four last peaks stem from increasing oxidations of the parent molecule, as often seen with peptides containing methionine and tryptophan residues. B, CCK-22-Gly, including two oxidized residues (calculated molecular mass: 2800) secreted from YNG318 expressing the prepro-MF␣1-pro-CCK construct. C, C-terminally extended CCK-22 (calculated molecular mass: 4136, including 4 times oxidation (also compare with A) calculated molecular mass becomes 4136 ϩ (4 ϫ 16) ϭ 4200) secreted from mkc7 yap3 double mutant HKY24 expressing the prepro-MF␣1-pro-CCK construct. D, C-terminally extended CCK-22 (calculated molecular mass 4136) secreted from the kex2 yap3 double mutant ME938 expressing the prepro-MF␣1-pro-CCK construct. In each case values are m/z of the molecular ions. Data in panel A were obtained in the negative mode, hence value(s) are for M Ϫ ion(s), while the others were obtained in the positive mode and thus show values for the MH ϩ ions. a.i., accumulated ions. CCK-22 and CCK-8, respectively. Fractions collected from Sephadex G-50 gel chromatography were also analyzed with Ab 89009 which binds the N terminus of CCK-22. Two immunoreactive peaks were observed, one corresponding to CCK-22-Gly and the other at K D 0.6 ( Fig. 2C), strongly suggesting this elution constant represents C-terminally extended CCK-22. A similar experiment was performed using an antibody which binds the N terminus of non-sulfated CCK-8 (Ab 94179). 3 The results showed two immunoreactive peaks at K D values of 1 and 1.15, the latter being CCK-8-Gly, and the former being C-terminally extended CCK-8 (data not shown).
Expression of the pro-CCK fusion in strain MT960 (Table I), which is the wild type for the PEP4 protease, but defective in the BAR1 encoded "barrier" protease, an exported pepsin-like protease which cleaves ␣-factor (Mackay et al., 1988), resulted in the formation of CCK-22-Gly and CCK-8-Gly in a manner similar to that of strain YNG318 (data not shown). Thus, neither protease A nor the barrier protease seems to be responsible for the observed proteolytic cleavages of pro-CCK.
Proteolytic Processing of Human Pro-CCK Expressed in Yeast Endoprotease Mutants-The prepro-MF␣1-pro-CCK plasmid was transformed into S. cerevisiae strains deficient in the function of known yeast secretory pathway endopeptidases in an attempt to determine which enzyme(s) were responsible for the endoproteolytic processing of pro-CCK. Analysis of the extracellular products from transformants of the mkc7 yeast mutant HKY21 by gel chromatography and RIA analysis with Ab 3208 showed that both CCK-22-Gly and CCK-8-Gly were produced (Fig. 4A). Pro-CCK processing was then analyzed in yeast deficient in both the Mkc7 and Yap3 aspartyl proteases (HKY24). However, the three proteolytic cleavage events occurring in wild type yeast persisted in the mkc7 yap3 double mutant (Fig. 4B). Treatment of the column fractions with trypsin and carboxypeptidase B uncovered two additional immunoreactive peaks (Fig. 4B) corresponding to C-terminally extended versions of CCK-22 and CCK-8. Processing at Lys 81 to produce C-terminally extended CCK-22 was confirmed in the HKY24 strain by the purification of the CCK-immunoreactive material eluting at a K D of 0.6 followed by protein sequencing and mass spectrometry (Fig. 3C).
When the prepro-MF␣1-pro-CCK construct was expressed in a kex2 mutant, ME598, the pattern of pro-CCK endoproteolytic processing was altered (Fig. 4C). No glycine extended products were detected, and a single major CCK-immunoreactive peak was found eluting at a K D of 0.6 after the Sephadex G-50 column fractions had been exposed to trypsin and carboxypeptidase B treatment (Fig. 4C). A similar situation was observed when the fusion protein was expressed in the kex2 yap3 double mutant ME938 (Fig. 4D). Immunoreactive material in the medium of this double mutant was purified, exposed to protein sequencing and mass spectrometry (Fig. 3D), and its identity shown to be C-terminally extended CCK-22. However, several other CCK species were recovered during the large scale preparation of pro-CCK processing products secreted by the ME938 transformant, each constituting a minor fraction (less than 5%) of total secreted CCK. Most were N-or C-terminally truncated fragments, indicating that a low level of degradation by aminoand carboxypeptidases had occurred (data not shown). Although, a small amount of C-terminally extended CCK-39, N-terminally truncated by one residue, was also identified (data not shown).
CCK processing patterns were also examined in the triple protease mutant strain HKY25 (Table I). Analysis of secreted pro-CCK products revealed a large degree of variation between experiments in molecular forms of CCK produced, however, cleavage at Lys 81 persisted (data not shown).
Intracellular Processing of Pro-CCK in Yeast Transformants-Metabolic labeling and immunoprecipitation experiments were performed to determine which, if any, of the proteolytic cleavage events of pro-CCK occurred within the yeast cell. Both radiolabeled and CCK-Gly immunoreactive material was found to elute at K D values of 0.75 and 1.15 (Fig. 5), previously shown to correspond to CCK-22-Gly and CCK-8-Gly, respectively (Figs. 2 and 3, A and B). Since no larger forms of CCK were detected it appears that only the smaller forms of CCK are glycine extended in the yeast cell. In addition, when CCK-33-NH 2 was incubated for 5 h at 30°C with S. cerevisiae, as well as with media in which the yeast had been grown, only CCK-33-NH 2 was recovered (data not shown), indicating that the cleavage at Lys 81 is occurring entirely within the cell.
Analysis for Tyrosyl Sulfation of CCK Secreted from Yeast Transformants-Anion exchange chromatography was em-3 J. F. Rehfeld, unpublished data. ployed to determine if the Tyr 97 residue of CCK was sulfated in the yeast secretory pathway. Transformants of YNG318 expressing the prepro-MF␣1-pro-CCK fusion construct were grown in a 5-ml culture to late exponential phase. The supernatant was then treated with trypsin and carboxypeptidase B to convert all CCK-related peptides into CCK-8-Gly which contains the potential sulfation site. The digest was loaded on a Sephadex G-50 superfine column from which a single peak of CCK immunoreactivity was detected with Ab 7270 at the elution position known for CCK-8-Gly (data not shown). The immunoreactive material was subjected to anion exchange chromatography where it eluted at a position similar to the nonsulfated control (Fig. 6), indicating that yeast were unable to sulfate CCK. In addition, all purified CCK species possessed the same molecular mass when measured in the positive and negative mode, confirming that these peptides are not sulfated (Talbo and Roepstorff, 1993).

DISCUSSION
Human prepro-CCK has been expressed in S. cerevisiae, resulting in the secretion of numerous processing intermediates of pro-CCK. However, many native signal sequences are inefficient at directing foreign proteins through the yeast secretory pathway (for review see; Hadfield et al. (1993)). When pro-CCK was expressed as a fusion protein, N-terminally linked to the prepro region of MF␣1, a hundredfold increase in the secretion of partially cleaved CCK was observed. Due to the higher expression levels, yeast expressing the prepro-MF␣1pro-CCK fusion construct were utilized to further analyze pro-CCK processing in the yeast secretory pathway.
Peptide ␣-amidation, a widespread protein modification occurring in invertebrates and vertebrates (Eipper et al., 1993), is essential for the biological activity of CCK (Martinez et al., 1986). The substrate for ␣-amidation is invariably a C-terminal glycine, which is processed to an ␣-amide group by the bifunctional protein, peptidyl-␣-amidating monooxygenase, located in the late Golgi and/or secretory vesicles (Eipper et al., 1993). Amidated CCK was not found in media supernatants of yeast expressing native prepro-CCK or the prepro-MF␣1-pro-CCK construct. Consequently it appears that S. cerevisiae lack the ability to ␣-amidate proteins.
Typically CCK is synthesized in the tyrosyl-sulfated (Tyr 97 ) form, which is essential for binding to the CCK A receptor in the gastrointestinal tract and some regions of the brain (Dourish and Hill, 1987). Analysis of CCK-Gly secreted from yeast transformants by both anion exchange chromatography and mass spectrometry revealed only the non-sulfated forms. The inabil-ity of S. cerevisiae to sulfate tyrosine residues was also observed for the leech anticoagulant, hirudin, which requires tyrosyl sulfation for full activity (Riehl-Bellon et al., 1989).
Analysis of the proteolytic processing products of CCK secreted from transformants of wild type yeast strains proficient in known secretory pathway endopeptidases uncovered four pro-CCK products, glycine-extended CCK-8 and CCK-22, as well as C-terminally extended CCK-8 and CCK-22. Therefore S. cerevisiae is able to cleave pro-CCK at three sites: (i) Cterminally of the dibasic Arg 105 -Arg 106 residues, a cleavage which is an essential step in production of bioactive CCK in vertebrates, (ii) between the Arg 95 -Asp 96 amino acids producing CCK-8 related products, the most common form of CCK in the mammalian brain (Rehfeld, 1978b;Rehfeld and Hansen, 1986); and (iii) between the Lys 81 -Asn 82 residues producing CCK-22 related products, which is a predominant form of CCK in human plasma and plasma from most other mammals (Liddle et al., 1984(Liddle et al., , 1985Cantor and Rehfeld, 1987;Rehfeld, 1994;Paloheimo and Rehfeld, 1995). Processing at the Lys 81 residue is extremely efficient, abolishing detection of endoproteolytic cleavage events which may have occurred N-terminally of Lys 81 with the antisera utilized in this study. However, a small quantity of C-terminally extended CCK-39, lacking one amino acid at the N terminus, was purified from the kex2 yap3 double mutant ME938 expressing the pre-proMF␣1-pro-CCK construct.
In an attempt to identify which yeast endopeptidase(s) were responsible for the above mentioned cleavage events, the processing patterns were analyzed in yeast strains deficient in known secretory pathway endopeptidases. When pro-CCK processing was examined in strains lacking the structurally and functionally related Mkc7 and Yap3 proteases, all forms of CCK secreted from wild type yeast were produced by the aspartyl protease-deficient strains. Transformants of the mkc7 yap3 double mutant seemed to secrete more C-terminally extended forms of CCK than glycine extended forms when compared with the wild type strain. However, since yeast lacking FIG. 5. Gel chromatogram of in vivo labeled and immunoprecipitated CCK-Gly produced by the S. cerevisiae YNG318 strain expressing the prepro-MF␣1-pro-CCK construct. Radioactivity (left y axis) and CCK-Gly immunoreactivity utilizing Ab 3208 (right Y axis) were determined for each column fraction.
FIG. 6. Determination of the level of tyrosyl sulfation of human CCK produced in yeast. Cholecystokinin secreted from yeast expressing the prepro-MF␣1-pro-CCK construct was cleaved with trypsin and the resulting CCK-8-Gly partially purified by gel chromatography. The sample was applied to a 5 ϫ 50-mm Mono-Q HR5/5 anion exchange column (Pharmacia), which had been calibrated with non-sulfated and sulfated CCK-8-Gly, and eluted with a gradient from 0 to 20% acetic acid. Fractions collected were analyzed by RIA with Ab 7270 which binds glycine-extended CCK. Elution positions of non-sulfated and sulfated CCK-8-Gly are shown (red.), as well as elution position of CCKs in which a methionine residue had been oxidized (ox.).
Kex2, but proficient in Mkc7 and Yap3 function, do not produce any CCK-Gly this result suggests that pro-CCK expressed in yeast is not cleaved at Arg 105 -Arg 106 by either the Mkc7 or Yap3 proteases within the yeast cell. Incubating CCK-33 with purified Yap3 in vitro resulted in the liberation of both CCK-22 and CCK-8 (Cawley et al., 1996). Since processing to produce these forms of CCK persists in a mkc7 yap3 double mutant it is presently difficult to determine if either of the proteases are able to cleave pro-CCK at Lys 81 or Arg 95 in vivo.
Two pro-CCK endoproteolytic processing events were abolished in a yeast strain lacking the Kex2 serine protease, namely the dibasic Arg 105 -Arg 106 directed cleavage and processing between Arg 95 -Asp 96 , suggesting that Kex2 is responsible for both processing events. However, the possibility that the intracellular trafficking of pro-CCK may be altered in the kex2 mutant, thus directing the pro-CCK molecule to a distinct area of the secretory pathway where it cannot be cleaved at Arg 105 -Arg 106 or Arg 95 cannot be ruled out. Recently the first unequivocal evidence for at least two distinct vesicles in the yeast secretory pathway has been reported, where it appears that one of these vesicles may represent proteins bound for the plasma membrane, with the other vesicle type bound for secretion from the cell (Harsay and Bretscher, 1995).
The Kex2 enzyme is widely known to be responsible for processing of both yeast and foreign proteins at dibasic sites (Fuller et al., 1988;Hadfield et al., 1993). Thus, it was not surprising that pro-CCK processing was abolished at the Arg 105 -Arg 106 residues in a kex2 mutant. As seen in other instances where Kex2 has been responsible for post-translational processing, not all of the proprotein was processed at this site (Miyajima et al., 1986;Thim et al., 1986;Zsebo et al., 1986;Moody et al., 1987;Driedonks et al., 1995). Notably, even when low amounts of pro-CCK were passed through the yeast secretory pathway, as was the case when prepro-CCK was expressed in yeast, there were still relatively high quantities of C-terminally extended CCK secreted. Since large amounts of glycineextended CCK were secreted from S. cerevisiae transformants expressing the prepro-MF␣1-pro-CCK construct, it seems improbable that yeast expressing the prepro-CCK construct lack the capacity to cleave all pro-CCK present in the secretory pathway at the Arg 105 -Arg 106 site.
The second pro-CCK proteolytic processing event which occurred in wild type yeast, but not the kex2 mutant, was the single arginine directed cleavage resulting in CCK-8 related products. There is an increasing number of instances where Kex2 and related serine proteases of the mammalian secretory pathway have been shown to preferentially process proproteins at single arginine residues (Zhu et al., 1992;Bourbonnais et al., 1994;Dupuy et al., 1994;Galanopoulou et al., 1995;Vollenweider et al., 1995). For instance, prohormone convertase 1, but not prohormone convertase 2, was shown to cleave rat prodynorphin at a single arginine residue when co-expressed in mouse AtT-20 cells (Dupuy et al., 1994). Our data raises the possibility that some members of the mammalian Kex2-like proteases may also play a role in the processing of pro-CCK at the Arg 95 -Asp 96 site to release CCK-8. A non-serine endopeptidase which displayed a high selectivity for cleavage of CCK-33, resulting in CCK-8, has been partially purified from rat brain synaptosomes (Viereck and Beinfeld, 1992), although there is currently no direct evidence to suggest that this protease cleaves pro-CCK in vivo. Experiments similar to those performed on prodynorphin by Dupuy and co-workers (1994) would be interesting to determine if any of the known mammalian Kex2-like proteases are able to process pro-CCK at the Arg 95 -Asp 96 site. It is possible that the tissue specificity observed in the N-terminal processing of pro-CCK in mammalian tissues is a result of differential expression of a single argininespecific endopeptidase and Kex2-like serine proteases, some of which may have the ability to process pro-CCK to CCK-8.
The ability to cleave pro-CCK at Lys 81 to produce CCK-22 related products was maintained in a variety of recombinant yeast strains deficient in three known putative endopeptidases of the S. cerevisiae secretory pathway. Hence our data suggests that another endopeptidase is present in the secretory pathway of yeast. Previously, two groups have suggested that an uncharacterized protease was cleaving the human ␤-amyloid precursor after the lysine residue in the His-Gln-Lys-Leu-Val sequence, when expressed and secreted in yeast (Hines et al., 1994;Zhang et al., 1994). However, one of these studies did not examine kex2 or yap3 mutant yeast strains (Hines et al., 1994), whereas the other did not express ␤-amyloid precursor in yap3 mutants . The lysine-directed processing of ␤-amyloid precursor in yeast is identical to that performed by ␣-secretase in the mammalian brain, to produce the ␤/A4 peptide found in senile deposits of Alzheimer's disease patients (Hines et al., 1994;Selkoe, 1994;Zhang et al., 1994). At present it is debatable if the same protease is responsible for the lysinedirected processing of pro-CCK and ␤-amyloid precursor in yeast. This can only be determined when the gene encoding the endopeptidase responsible for the lysine-directed cleavage in pro-CCK is identified.
In conclusion, the expression of pro-CCK in S. cerevisiae has uncovered important features of the yeast secretory pathways ability to post-translationally process secreted peptides/proteins. First, it appears that yeast lack the tyrosyl sulfation and ␣-amidation protein modifying capabilities. Second, our evidence suggests that the Kex2 protease has the ability to cleave Arg-Asp bonds. Finally, we provide evidence that a novel lysine-specific endopeptidase is present in the yeast secretory pathway. Attempts to isolate and characterize the gene encoding this putative endopeptidase are currently underway.