Characterization of the Enzymatic Properties of the First and Second Domains of Metallocarboxypeptidase D*

Carboxypeptidase D (CPD) contains three domains with homology to other metallocarboxypeptidases. To further characterize the various domains, we constructed a series of point mutants with a critical active site Glu of duck CPD converted to Gln. The proteins were expressed in the baculovirus system, purified to homogeneity, and characterized. Point mutations within both the first and second domains eliminated enzyme activity, indicating that the third domain is inactive toward dansyl-Phe-Ala-Arg. CPD removed only the C-terminal Lys or Arg from peptides, with the first domain more efficient toward Arg and the second domain more efficient toward Lys. Peptides containing Pro in the penultimate position were poorly cleaved by either domain. Cleavage of a peptide with Ala in the penultimate position was most efficient, with the relative order Ala ≥ Met > Ser, Phe > Tyr > Trp > Thr ≥ Gln, Asp, Leu, Gly ≫ Pro for CPD with both domains active. There were only minor differences between the first and the second domains regarding the influence of the penultimate amino acid. The first domain was optimally active at pH 6.3–7.5, whereas the second domain was optimally active at pH 5.0–6.5. Thus, the first and second carboxypeptidase domains have complementary enzyme activities. Furthermore, the finding that CPD with both domains active shows a broad activity to a wide range of substrates is consistent with a role for this enzyme in the processing of many proteins that transit the secretory pathway.

Many proteins that are secreted from cells require proteolytic processing to generate the biologically active form. Examples include peptide hormones, neurotransmitters, growth factors, and viral proteins (1). In addition, proteolytic processing is also required for the production of cell surface membrane proteins such as the insulin receptor and several growth factor receptors (1). In general, the proteolytic sites are composed of basic amino acids. A variety of endopeptidases that cleave at basic amino acids have been described within the secretory pathway (2)(3)(4)(5)(6)(7). Some of these enzymes, such as prohormone convertase 1 (PC1, 1 also known as PC3) and prohormone convertase 2 (PC2) are largely restricted to neuroendocrine cells, and are predominantly involved in the processing of peptide hormones and neurotransmitters (8 -10). Within the neuroendocrine cells, PC1 and PC2 are present in the peptide-containing secretory vesicles (11,12). Other endopeptidases, such as furin, are broadly expressed in many cell types and are primarily localized to the trans Golgi network (1,13). Furin and related endopeptidases are thought to process proteins that do not enter the regulated secretory pathway, such as the insulin receptor, albumin, and other proteins (1, 14 -16).
PC1, PC2, furin, and other related enzymes cleave to the C-terminal side of the basic residues (17,18). Thus, one or more of the products of endopeptidase activity will contain C-terminal basic residues. For many peptide hormones and neurotransmitters, these basic residues need to be removed before the peptide has biological activity. Carboxypeptidase E (CPE) is the primary enzyme involved in the removal of C-terminal basic residues from a large number of peptide intermediates (19 -21). CPE is present in neuropeptide-containing secretory vesicles along with PC1 and PC2 (22,23). CPE also has a neuroendocrine-specific tissue distribution (24 -26), and so it is unlikely that this enzyme plays a role in the processing of non-neuroendocrine proteins cleaved by furin and related enzymes.
CPD was recently discovered in a search for CPE-like enzymes that could contribute to peptide processing in the Cpe fat / Cpe fat mouse (27). These mice lack active CPE due to a point mutation that converts a Ser into a Pro (28). The mutant CPE is inactive and is rapidly degraded prior to transport into the Golgi (29). Without active CPE, the Cpe fat /Cpe fat mice have greatly elevated levels of peptide processing intermediates with C-terminal basic residues, and reduced levels of the fully processed peptides (28,30,31). However, low levels of mature peptides are detected, indicating that another enzyme is able to partially compensate for the absence of CPE activity. A search for novel CPE-like enzymes led to the identification of CPD, carboxypeptidase Z, and proteins designated CPX-1 and CPX-2 (27,(32)(33)(34). In addition, another group identified a protein designated AEBP1 as a novel member of the carboxypeptidase gene family (35). Of these proteins, only CPD and carboxypeptidase Z demonstrate activity toward standard CPE substrates (27,(32)(33)(34). The cellular distribution of carboxypeptidase Z is restricted to specific cell types, such as the leptomeningeal cells in brain (36). In contrast, CPD has a broad tissue distribution and is present in many cell types in each tissue (37)(38)(39)(40). Also, CPD is present along with furin in the trans Golgi network and immature secretory vesicles (41,42). Thus, CPD has the right cellular and subcellular distribution to participate to some extent in the processing of neuroendocrine peptides as well as playing a major role in the processing of proteins initially cleaved by furin.
Human, rat, and duck CPD consists of three carboxypeptidase-like domains, a transmembrane domain, and then a short cytosolic tail (38,43,44), whereas the Drosophila CPD homolog contains two (45) and the Aplysia homolog contains four carboxypeptidase-like domains (46). Thus, the general feature of multiple carboxypeptidase-like domains is highly conserved. The purpose of the present study was to investigate whether the various domains perform complementary functions. A previous study investigated the activity of individual domains using a deletion approach and found that only the first two domains possessed enzymatic activity toward standard CPE substrates (47). However, removal of hundreds of residues can potentially cause large changes in the structure of the protein.
In the present study, we used a more selective mutagenesis approach and converted a single Glu residue in the first and/or second domains into a Gln (Fig. 1). This Glu, which corresponds to Glu 270 in carboxypeptidase A and B and Glu 300 in CPE, has been previously been shown to be essential for enzyme activity but not substrate binding (48). Thus, this mutation would not be predicted to have a large impact on the structure of the protein. Using these point mutants, we confirmed that the third domain is devoid of enzyme activity toward the standard substrates. We also compared the enzymatic properties of the first and second domains using both standard substrates and a novel assay using a peptide mixture. The results of these analyses suggest that the two domains of CPD complement each other with regard to pH and specificity for C-terminal basic residues but show overlapping substrate specificities regarding the penultimate amino acid.

MATERIALS AND METHODS
Construction of Mutants-The construction of gp170 and gp75 within the baculovirus expression vector pVL1392 (Pharmingen) were previously described (47), resulting in vectors pVL170 and pVL75. For mutagenesis, portions of gp170 were first inserted into the vector pAlter (Promega). For mutation of the second domain (gp170(Q 771 )), a 1.2kilobase pair StuI/SnaBI fragment was inserted into pAlter at its SmaI site. For mutation of the first domain (gp170(Q 353 )), a 750-base pair HindIII/SacI fragment was inserted into pAlter. Site-directed mutagenesis was then performed using the Quick-change mutagenesis kit (Stratagene). In each case, a GAG codon was converted to a CAG codon generating a new PvuII site. Following mutagenesis, a 670-base pair BamHI/DraIII fragment carrying the Gln 771 mutation and a 750-base pair HindIII/SacI fragment carrying the Gln 353 mutation were engineered back into the expression vectors pVL170 and pVL75. The plas-mids were confirmed by dideoxynucleotide sequencing of the regions used for mutagenesis and subcloning.
Expression of Proteins in Baculovirus and Purification-Sf9 cells were co-transfected with 5 g of baculovirus expression plasmid and 0.25 g Baculo-Gold DNA (Pharminogen), and virus was amplified as described previously (32). For purification of gp170, gp170(Q 353 ), and gp170(Q 771 ), 1 liter of medium from cells infected for 2-3 days was collected after centrifugation at 30,000 ϫ g for 30 min. The supernatant was adjusted to pH 5.5 with 0.1 M sodium acetate and then applied to a 5-ml p-aminobenzoyl-Arg-Sepharose affinity resin. The column was washed with 100 mM NaAc buffer, pH 5.5, containing 1 M NaCl, and 1% Triton X-100. Proteins were eluted with 50 ml of 50 mM Tris HCl, pH 8.0, 100 mM NaCl, 0.01% Triton X-100, and 25 mM Arg. These eluates were concentrated to approximately 1 ml by filtration dialysis using a Centriplus 30 (Amicon). To decrease Arg concentration below 0.1 mM, the filtration was repeated two times with 15 ml of elute buffer lacking Arg. For studies using mass spectrometry, the Triton X-100 in the wash and elute buffers was replaced with 0.5 and 0.01% CHAPS, respectively.
Carboxypeptidase Assays with Fluorescent Substrates-Enzyme activity was typically assayed with 0.2 mM dansyl-Phe-Ala-Arg in 100 mM, pH 6.4, Tris acetate buffer in a 250-l final volume. After 30 min at 37°C, the reaction was terminated with 100 l of 0.5 M HCl, and then 2 ml of chloroform were added. After mixing and centrifugation for 2 min at 300 ϫ g, the amount of product was determined by measuring the fluorescence in the chloroform phase (excitation 350 nm, emission 500 nm). The pH optimum of purified enzymes was determined with 0.2 mM dansyl-Phe-Ala-Arg in 0.1 M Tris acetate at the indicated pH. To examine the effect of site-directed inhibitors, purified enzymes were added to a mixture of buffer, substrate, and inhibitor to give a final concentration of 100 mM Tris acetate, pH 6.4, 100 M dansyl-Phe-Ala-Arg, and 10-fold dilutions of inhibitor ranging from 100 M to 1 nM. To examine the effect of general enzyme inhibitors and metals on gp170 and mutants, the compounds were preincubated with purified enzymes for 1 h at 4°C and then assayed with 200 M dansyl-Phe-Ala-Arg at pH 6.4. For kinetic analysis, purified enzyme was combined with substrate (final concentration 6.25, 12.5, 25, 50, 100, and 200 M) and 100 mM Tris acetate pH 6.4, 0.01% Triton X-100 buffer in a final volume of 1 ml. After 30 min at 37°C the reaction was terminated with the addition of 100 l of 2 M HCl, and then 2 ml of chloroform was added. The amount of enzyme chosen was that which hydrolyzed a maximum of 20% of the peptide. The amount (nmol) of product was determined from standard curves for each peptide. The kinetic parameters were evaluated by fitting the data to y ϭ (m 1 ϫ X)/(m 2 ϩ X), using the KaleidaGraph program (where y represents velocity, X is substrate concentration, m 1 ϭ V max , and m 2 ϭ K m ).
Carboxypeptidase Assays with Mixed Peptides-The mixed peptide Tyr-Glu-Pro-Gly-Ala-Pro-Ala-Ala-Gly-X-Arg, where X represents Gly, Ala, Ser, Pro, Thr, Leu, Asp, Gln, Met, Phe, Tyr, or Trp, was synthesized by the Laboratory for Macromolecular Analysis, Albert Einstein College of Medicine. Purified enzymes were serially diluted 1:3 with 50 mM ammonium acetate, pH 6.4, 0.01% acetylated bovine serum albumin and combined with 0.1 mM mixed substrate (calculated from the average molecular mass of the mixture) in a final volume of 30 l. After 8 h at 37°C, the reactions were frozen, lyophilized, and reconstituted in 50 l of 50% acetonitrile, 50% H 2 O containing 1% acetic acid. Each sample was analyzed by electrospray ionization mass spectrometry on a Finnigan quadrupole ion trap mass spectrometer. The sample solution (20 l) was introduced with a syringe pump into the mass spectrometer ion source at a flow rate of 3 l/min. The mass spectrometer performance was optimized with the autotuning feature on the instrument using the standard peptides des-Arg 1 -bradykinin ((M ϩ 2H) 2ϩ ϭ m/z 452.5) and Arg insulin ((M ϩ 4H) 4ϩ ϭ m/z 1492.5). Each spectrum represented an average of approximately 100 scans collected over 2 min, with a scan range of m/z 500 -1400. The data were processed with Bioexplore software. The percentage of product formed for each dilution of enzyme was calculated by the formula (Hp ϩ HpNa)/(Hp ϩ HpNa ϩ Hs) ϫ 100 (where Hp is the peak height of the ion of the product, HpNa is the peak height of the ion of the product plus sodium, and Hs is the peak height of the ion of the substrate).

RESULTS
Previous studies had established that duck CPD (gp180) lacking the C-terminal transmembrane domain was soluble and secreted from cells into the medium (47). The enzymatic properties of this 170-kDa form of duck CPD (termed gp170) were found to be identical to those of full-length gp180, so we chose to use the soluble form for further analysis of the individual domains. Wild type gp170 and the various point mutants were expressed at high levels in Sf9 cells using the baculovirus expression system ( Fig. 2A). In addition to the gp170 constructs, we also prepared a point mutation of Glu 353 to Gln within CPD containing only the first domain (termed gp75); this was a necessary control for the binding to the affinity column (described below). The domain 1-specific form containing the point mutation, gp75(Q 353 ), was also expressed at high levels in the Sf9 cells using baculovirus ( Fig. 2A). All of the constructs were also found at high levels in the media from infected Sf9 cells ( Fig. 2A). Cells infected with the wild type gp170 expressing baculovirus and the medium from these cells show levels of carboxypeptidase activity greater than 100-fold above that of control virus (Table I). Media and cells infected with gp170(Q 353 ) or gp170(Q 771 ) also show considerable carboxypeptidase activity. In contrast, the medium and cells infected with the double mutant show only background levels of enzyme activity (Table I). This result suggests that the mutation of Glu 353 or Glu 771 eliminates the enzyme activity of each domain and that the third domain of CPD does not have appreciable enzyme activity (Ͻ0.5% of the wild type enzyme). As a control, the mutation of Glu 353 to Gln in the construct containing only the first domain of CPD shows no appreciable enzyme activity (Table I).
All mutants are able to bind to the p-aminobenzoyl-Arg-Sepharose substrate affinity column and are eluted with 25 mM Arg (Fig. 2B). Following this purification, only a single band is detected for each construct on denaturing polyacrylamide gels (Fig. 2B). Following the purification, carboxypeptidase activity above background levels was only detected for the wild type gp170, the gp170(Q 353 ) mutant, and the gp170(Q 771 ) mutant, consistent with the results from the crude media and cell extracts. The double mutant and the p75(Q 353 ) mutant had no activity detectable above the background obtained after subjecting the medium from control virus-infected cells to the same purification procedure.
The pH optimum of wild type gp170 is 5-7 (Fig. 3, bottom), consistent with previous studies (47). In contrast, the pH optimum of the construct with only the active first domain is 6.3-7.5, and the construct with only the active second domain is 5.0 -6.5 (Fig. 3, top). Thus, the pH optimum of the protein with both domains active is a combination of the two individual domains. The construct with mutations in both the first and second domains had no detectable enzyme activity at any of the pH values examined (Fig. 3, top).
The enzymatic activities of the various constructs were examined in the presence of several different ions and enzyme inhibitors. In response to a 1 mM concentration of various ions, domains 1 and 2 showed differences in their activation or inhibition (Table II). Whereas both domains were strongly inhibited by the chelating agent 1,10-phenanthroline, other chelating agents (EDTA and EGTA) inhibited mainly domain 1 (Table II) The kinetic parameters for dansyl-Phe-Ala-Arg hydrolysis were evaluated for wild type gp170 and the two active constructs (Table III). The k cat value for domain 2 was approximately twice that of domain 1, and the sum of the two k cat values were close to the k cat for wild type gp170. The substrates dansyl-Pro-Ala-Arg and dansyl-Phe-Gly-Arg also showed higher k cat values for domain 2 than domain 1, whereas dansyl-Phe-Phe-Arg had comparable k cat values for the two domains. K m values for the four dansyl peptide substrates ranged from 9 to 50 M. To evaluate the kinetic parameters of cleavage of C-terminal Arg versus Lys, two enkephalin precursors were tested, and the amount of product was determined using HPLC. Wild type gp170 cleaved the two enkephalin-related peptides with comparable k cat values but with a lower K m for the Lys-containing peptide (Table III). The Arg-containing peptide was cleaved with a higher k cat and lower K m by the construct with only the first domain active, compared with the construct with the second domain active. The opposite result FIG. 2. A, Western blot analysis of cells and media after infection with wild type baculovirus (wt vir) or baculovirus expressing one of the constructs indicated in Fig. 1. Proteins were electrophoretically transferred to nitrocellulose from a denaturing polyacrylamide gel. Blots were probed with a 1:1000 dilution of polyclonal rabbit antiserum raised against gp170 (47). Following exposure of the blot to primary antiserum, the enhanced chemiluminescence method (Amersham Pharmacia Biotech) was used to detect bound antiserum. The positions and molecular masses in kDa of prestained protein standards (Bio-Rad) are indicated. B, analysis of protein after purification on a p-aminobenzoyl-Arg affinity column. Aliquots of concentrated eluates were analyzed on a denaturing polyacrylamide gel followed by silver staining (56). The positions and molecular masses in kDa of prestained protein standards (Bio-Rad) are indicated. was observed for the Lys-containing peptide (Table III). In order to further characterize the specificity of CPD toward a wider range of substrates, we used a peptide mixture of the sequence Tyr-Glu-Pro-Gly-Ala-Pro-Ala-Ala-Gly-X-Arg, where X is one of 12 amino acids that are distinguishable by mass spectrometry. This sequence is based on the pro region of rat CPE, which is removed in the late Golgi and is therefore a likely physiological substrate of CPD (49). Analysis of the mixture using electrospray ionization mass spectrometry shows all of the expected masses for the individual components of the mixture. When incubated with CPD, the peptide mixture showed new peaks corresponding to the expected masses of the products and reduced levels of the peaks corresponding to the substrates. Quantitation of the peak height of the product and substrate was determined for reactions containing various dilutions of each enzyme, and the percentage of product relative to the total substrate plus product was calculated. Of all the peptides present in the mixture, the one where X is Ala is cleaved most efficiently by the wild type gp170 (Fig. 4, top). The peptide where X is Pro is not efficiently cleaved, even with 1000-fold more enzyme than that required for cleavage of the peptide with Ala in the penultimate position (Fig. 4, top). Excluding the peptide where X is Pro, there is only a 30-fold range in the amount of enzyme required for the conversion of 50% of each substrate to product (Fig. 4, top). The pattern of cleavage of the mixed peptide by the second domain is virtually identical to that with wild type gp170 (Fig. 4, bottom), and the pattern with the first domain is very similar (Fig. 4, middle). In contrast, the pattern with CPE (not shown) is substantially different from the patterns with any of the active CPD constructs. The only differences observed between domains 1 and 2 of CPD were for peptides where X represented Tyr, Phe, Leu, or Gly. The individual domains 1 and 2 showed less than 100-fold differences in the amount of enzyme needed to cleave 50% of the various peptides (excluding the peptide where X represented Pro). DISCUSSION A small number of enzymes have multiple catalytic domains; examples include angiotensin-converting enzyme, peptidyl-glycine-␣-amidating monooxygenase, and CPD. In the case of peptidyl-glycine-␣-amidating monooxygenase, the two domains perform subsequent steps in the production of C-terminal amide residues on neuroendocrine peptides (50). For enzymes where the two domains have similar activities (angiotensinconverting enzyme and CPD), the rationale for distinct domains is less clear. The two domains of angiotensin-converting enzyme are differentially affected by chloride and have different k cat values for substrates (51). It is likely that both the first and second carboxypeptidase domains of CPD perform important functions, since these two domains are conserved in all species examined, including mammals, Drosophila, and Aplysia (38,(43)(44)(45)(46). In addition, the high degree of conservation of the third domain of CPD among human, rat, and duck implies an important function for this domain as well.
One major finding of the present study is that the third domain of CPD has no detectable enzymatic activity. This result is consistent with a previous study investigating a deletion construct that lacked the first two domains of duck CPD (47). However, it is hard to interpret the results of such a large deletion, since it is likely that protein folding is affected. The present approach using point mutations to inactivate the first and second domains is advantageous to the deletion approach. The conversion of the putative active site Glu into Gln is not expected to substantially alter the folding of the protein; this is consistent with the observation that the mutant proteins are able to bind to the substrate affinity column despite the absence of activity. A previous study of Glu 300 in CPE (the equivalent residue) also found that the mutation to Gln eliminated activity but not substrate binding or routing to the regulated secretory pathway in a neuroendocrine cell line (48). In contrast, mutants that are severely misfolded are not properly   FIG. 3. Effect of pH on purified gp170 (bottom) and point mutations shown in Fig. 1 (top). Carboxypeptidase (CP) activity was determined using 200 M dansyl-Phe-Ala-Arg in 100 mM Tris acetate buffer at the indicated pH at 37°C for 30 min. Activity was normalized to the maximal activity detected at the optimal pH for each construct and represents the average of two determinations with less than 10% variation. routed through the cell and do not bind to a substrate affinity column (29). Due to the ability of the Glu to Gln mutants to bind to the substrate affinity column, it is not possible to determine whether the third domain of CPD also binds to this column using the present approach. This third domain is predicted to be inactive due to the naturally occurring loss of the equivalent active site Glu. Other members of the metallocarboxypeptidase gene family that lack a corresponding Glu (CPX2 and AEBP1) are also inactive toward standard CPE substrates (32,34), thus leading to the proposal that these family members are binding proteins rather than active enzymes. It is possible that the third domain of CPD also functions as a binding protein; from the present study, this third domain is not a functional carboxypeptidase. (Note that it is unlikely that the third domain of CPD cleaves C-terminal residues other than Arg, since the studies with the mixed peptide showed only cleavage of the C-terminal Arg with no further cleavage of the peptide despite the presence of an intact third domain in all three constructs tested.) The only substantial differences found between the first and second domains of CPD that would have physiological implications are the pH optima and specificity for C-terminal Lys versus Arg (Fig. 3 and Table III). For proteins that are present in the secretory pathway, the pH range between 7 and 5 is important in regulating activity in the various intracellular compartments. The internal pH of the Golgi is neutral, as is the pH of the extracellular environment. In contrast, the internal pH of the trans Golgi network is slightly acidic, and the pH drops even further in the immature and mature secretory ves-icles to a final value of 5-5.5 (52, 53). CPD is predominantly present in the trans Golgi network and is also present in immature secretory vesicles and a small fraction of vesicles that morphologically resemble mature vesicles (41,42). The membrane-bound CPD is transiently expressed on the cell surface and is rapidly internalized (41,54). Thus, CPD exists in many compartments with distinct pH values. The first domain of CPD would be more active in the Golgi and extracellular environment, while the second domain would be more active at the acidic pH values of the trans Golgi network and secretory vesicles. The preference of the first domain for C-terminal Arg fits with the pH optimum of this domain and suggests that this domain will initially cleave peptides following the action of the endopeptidases. The consensus site for endopeptidase cleavage is most frequently Lys-Arg, so carboxypeptidase-directed cleavage of the C-terminal Arg would precede that of the Lys.
In addition to the differences in the pH optima and substrate specificity of the first and second domains of CPD, these domains also show large differences in the sensitivity to thiol reagents and some chelating agents. The physiological significance of these differences is not clear. The increased sensitivity of domain 1 to thiol reagents may reflect the presence of Cys residues near the active site in domain 1 (Cys 195 and Cys 357 ) that are not present in domain 2. The differential sensitivity of domains 1 and 2 to chelating agents may reflect the binding of metals other than the active site Zn 2ϩ or to differences in the affinity for the active site Zn 2ϩ . Previously, CPE was shown to bind Ca 2ϩ and to be regulated to a small extent by the presence or absence of this ion (55). However, CPE was substantially inhibited by 1 mM EGTA, EDTA, or 1,10-phenanthroline (55), like the first domain of CPD (Table II). Both CPE and the second domain of CPD contain a region that is homologous to the Ca 2ϩ binding region of carboxypeptidase T, and it is not clear if this domain influences the sensitivity to EGTA and EDTA.
Although conventional substrates provide kinetic information, which is not possible from the mixed peptide/mass spectrometric approach, the in vivo situation more closely resembles the mixed peptides in that the enzyme is presented with a range of potential substrates. The peptides that are cleaved with the shortest incubation times or with the lowest amount of enzyme are those that would presumably be cleaved first in vivo. The results with the mixed peptides are generally similar to those obtained with the limited number of dansyl peptides used in the conventional assay (Table III). For example, the k cat /K m for gp170 with dansyl-Phe-Ala-Arg is approximately 3-fold higher than the k cat /K m with dansyl-Phe-Phe-Arg and 14-fold higher than the k cat /K m with dansyl-Phe-Gly-Arg. This is in close agreement with the requirement for approximately 4-and 30-fold more enzyme to achieve 50% hydrolysis of mixed peptides with penultimate Phe and Gly, respectively, compared with Ala-containing substrates. The mass spectrometric approach only permitted residues that differed by 2 or more mass units to be tested. Thus, Leu, Ile, and Asn could not be tested in the same substrate; nor could Lys, Gln, and Glu. Additionally, Lys, Arg, and His were excluded from the penultimate position, since the product of the first cleavage could serve as a substrate for a second cleavage, thus complicating the analysis of the substrate/product ratio. Despite these disadvantages, the mixed peptide approach provides a rapid and sensitive assay to screen a large number of peptides. The mixed peptide method is also highly reproducible, with identical results obtained when the same enzyme was used in two separate experiments (not shown).
In summary, the first domain of CPD is most likely to be involved with cleavage of C-terminal Arg from peptides in the FIG. 4. Relative amount of product formed from each peptide present in the mixture after incubation of Tyr-Glu-Pro-Gly-Ala-Pro-Ala-Ala-Gly-X-Arg with various amounts of enzyme for 8 h. Samples were analyzed by mass spectrometry as described under "Materials and Methods." The relative amount of product formed from each peptide present in the mixture after incubation with various amounts of enzyme was calculated by the formula (Hp ϩ HpNa)/(Hp ϩ HpNa ϩ Hs) ϫ 100 (where Hp represents the peak height of the ion of the product, HpNa is the peak height of the ion of the product plus sodium, and Hs is the peak height of the ion of the substrate Golgi, trans Golgi network, and the extracellular environment and/or endocytic vesicles where the pH is close to neutral. The second domain also cleaves C-terminal Arg, but it has a preference for C-terminal Lys. Based on the pH optimum of the second domain, it is likely that this domain functions primarily in the trans Golgi network and immature secretory vesicles. The combination of the two domains produces an enzyme that is maximally active in a variety of cellular compartments and toward a broad range of substrates. Taken together, these results are consistent with the hypothesis that CPD functions in the processing of many proteins and peptides following the action of furin and related secretory pathway endopeptidases.