(Received for publication, October 23, 1996, and in revised form, January 3, 1997)
From the Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, New York 10461
A novel cDNA, designated carboxypeptidase Z (CPZ), was identified based on its homology to known metallocarboxypeptidases. Northern blot analysis shows bands of 2.1 and/or 2.6 kilobases in all tissues examined. The major form of CPZ mRNA in human salivary gland encodes a protein with an open reading frame of 641 amino acids. In addition, three variants were found that presumably arise due to alternative intron splicing. The 641-amino acid protein contains an 18-residue signal peptide-like sequence, a 120-residue region that shows 23-29% amino acid identity with a Cys-rich domain found in frizzled proteins and in type XVIII collagen, and then a 390-residue carboxypeptidase domain with 49% amino acid identity to carboxypeptidases E and N.
The 641-amino acid form of CPZ expressed in the baculovirus system cleaves 5-dimethylaminonaphthalene-1-sulfonyl (dansyl)-Phe-Ala-Arg, although the level of enzyme activity was approximately 10-fold lower than either carboxypeptidase E or D expressed using the same viral system. The CPZ activity is more active at neutral pH than at pH 5.5 and is inhibited by active site-directed inhibitors of metallocarboxypeptidases. In summary, CPZ is a novel metallocarboxypeptidase that is active toward substrates with C-terminal basic amino acids.
Carboxypeptidases (CPs)1 perform many important functions in a variety of tissues, ranging from digestion of food (pancreatic CPA and CPB) to selective processing of bioactive peptides (CPE). Altogether, there have been ten members of the metallocarboxypeptidase gene family previously identified in mammals. These ten members can be grouped into two subfamilies based on homology and overall structure. One subfamily includes mast cell CPA, plasma CPB (also known as CPU), and pancreatic CPA1, CPA2, and CPB (1). Proteins in this subfamily are approximately 30-40 kDa and do not contain C-terminal extensions beyond the CP domain. The other subfamily includes CPE (also known as CPH and enkephalin convertase), CPM, CPN, CPD, and a protein designated AEBP1 (2). Members of this subfamily contain a CP core region that is slightly longer than the CP core of the CPA subfamily and also contain C-terminal extensions beyond the CP domain. Within each subfamily, the proteins have approximately 40-60% amino acid identity, whereas between subfamilies, the amino acid identity is typically 15-25%.
Members of the CPE subfamily are generally involved in selective processing reactions, opposed to the more degradative roles for the enzymes in the CPA subfamily (2). CPE is involved in the production of many peptide hormones and neurotransmitters by the selective removal of basic residues from the C terminus of processing intermediates (3, 4). CPN functions in plasma to remove basic residues from a variety of peptides and proteins (5). CPM is widely distributed on plasma membranes and has been proposed to process peptide hormones at the cell surface (6). CPD is a recently discovered enzyme that may play a role in the processing of peptides and proteins that transit the secretory pathway (7, 8). AEBP1 was identified as a transcription repressor with homology to metallocarboxypeptidases (9). Although AEBP1 lacks residues thought to be critical for CP activity, this protein was reported to cleave a peptide containing a C-terminal Arg residue (9).
The importance of CPE in peptide processing has been clearly demonstrated by recent studies on Cpefat/Cpefat mice (10). These mice lack functional CPE due to a point mutation within the coding region of the CPE gene (10). The absence of CPE activity leads to a 2-10-fold reduction in the levels of correctly processed peptide hormones and neurotransmitters (10-12). The reduction in peptide levels in the Cpefat/Cpefat mice is strong evidence that CPE plays a major role in peptide processing. However, since correctly processed peptides are detected in the Cpefat/Cpefat mice, it is likely that another CP is also involved with peptide processing. CPD has been proposed as a candidate peptide processing enzyme due to the similar enzymatic properties of CPD and CPE (7, 8). However, recent subcellular localization studies have found CPD to be present mainly in the Golgi and/or trans Golgi network and not in the secretory vesicles where the majority of processing is thought to occur.2 Thus, it is possible that additional CPs are involved in peptide processing.
Several cDNAs encoding a portion of a novel metallocarboxypeptidase-like protein were identified by computer homology searches. This novel member of the CP gene family has been designated carboxypeptidase Z (CPZ). The purpose of the present study was to address the possibility that the CPZ cDNA encodes an active carboxypeptidase. To accomplish this, we first isolated the full-length cDNA encoding the major form of CPZ in human salivary gland. Then, this protein was expressed in the baculovirus system and tested for CP activity. The results of this analysis demonstrate that CPZ encodes a novel enzyme with activity toward substrates with basic C-terminal residues. However, the lack of enzyme activity at pH 5.5 suggests that CPZ does not function in peptide processing in secretory vesicles, which have an acidic internal pH.
The expressed sequence tag data base was searched for homology to human CPE using the National Center for Biotechnology Information program Tblastn. This analysis revealed several novel clones that showed substantial homology to CPE but did not correspond to any of the previously identified metallocarboxypeptidases. These homologous sequences include GenBankTM accession numbers R22743[GenBank], H50744[GenBank], T88975[GenBank], and R75746[GenBank]. Two of these clones, I.M.A.G.E. Consortium CloneID 130196 (GenBankTM R22743[GenBank]) and I.M.A.G.E. Consortium CloneID 194316 (GenBankTM H50744[GenBank]), were obtained from American Type Culture Collection. The clones were sequenced in both directions using oligonucleotide primers located approximately 300 nucleotides apart. The nucleotide sequence was determined by doubled-stranded sequencing according to the dideoxy chain termination method, using a Sequenase Version 2.0 DNA sequencing kit (U. S. Biochemical Corp.) and/or by the DNA sequencing facility of the Albert Einstein College of Medicine. This facility uses AmpliTaq DNA polymerase and the cycle sequencing procedure with an Applied Biosystems Inc. model 373. The nucleotide sequence of human CPZ has been submitted to GenBankTM (accession number U83411[GenBank]).
The N-terminal 180 amino acids of CPZ (derived from the 5 end not
present in the I.M.A.G.E consortium clones) were compared with the
non-redundant GenBankTM data base using the National Center
for Biotechnology Information program tblastn. Sequences with
significant matches (smallest sum probability values less than
10
4) include the frizzled proteins with
GenBankTM accession numbers L02530[GenBank], L37882[GenBank], U43320[GenBank], U43318[GenBank],
U43205[GenBank], U43316[GenBank], U43321[GenBank], U43319[GenBank], U65589[GenBank], L43340[GenBank], U43317[GenBank], and L02529[GenBank]. A
strong match was also found (smallest sum probability approximately
0.26) with type XVIII collagen sequences U19600[GenBank] and U11637[GenBank]. Alignment
of the Cys-rich 120-amino acid domain of CPZ with the frizzled proteins
and with type XVIII collagen was performed using the Align program of
Genepro (Hoeffer Scientific).
PCR was carried out following the protocol for rapid
amplification of cDNA ends (5 RACE PCR, Life Technologies, Inc.).
cDNA was prepared from 100 ng poly(A)+ RNA of saliva
gland (CLONTECH) using 5
GGCTGTGGTGATGTCGTG as a
primer and purified by glassmax DNA isolation spin cartridge. Purified
cDNA was tailed by terminal deoxynucleotide transferase and used as
a template for PCR amplification. For the first PCR, anchor primer
(Life Technologies, Inc.) and 5
GCTCCCCCTCTTCAGGAAATTGCCTCCACA were
used. The product from the first PCR was diluted 1000 times and
subjected to a second PCR with primers UAP (Life Technologies, Inc.)
and 5
GGATGCGGGTGGTGTTGAGCAGGCGCTG. The amplification product was
cloned directly from the PCR reaction using the TA cloning kit
(Invitrogen). Positive clones were selected by color selection and
confirmed by sequencing.
To check different forms of CPZ mRNA, reverse transcription of salivary gland mRNA and the first round of PCR was performed as described above for RACE PCR. Then, oligonucleotides that bracket the 23-bp region (corresponding to the sequences SINPDGY and MQTIPFV), the 107-bp region (corresponding to the sequences LHGGDLV and LLSRAYA), or the 33-bp region were used for the second round of PCR. The resulting products were analyzed on a 5% polyacrylamide gel and were also subcloned using the TA cloning system (Invitrogen) and confirmed by sequencing. To reduce the possibility that our sequence contained PCR artifacts, multiple clones from separate PCR reactions were isolated and sequenced.
Northern Blot AnalysisThe Northern blot was purchased from CLONTECH and contained approximately 2 µg of purified human poly(A)+ RNA on a nylon membrane. The blot contained 0.24-9.5 kb of RNA ladder (Life Technologies, Inc.). Human CPZ cRNA probe labeled with 32P was generated using the standard riboprobe reaction as previously described (13). The template for the preparation of the riboprobe consisted of clone 130196, which was linearized with SacII and transcribed with SP6 RNA polymerase. Approximately 107 cpm of the labeled probe in 10 ml of standard buffer (containing 50% formamide) was incubated with the blot overnight at 68 °C. Following the incubation, the blot was washed several times with 1 × standard saline citrate buffer at 68 °C and then exposed to film for the time indicated.
Generation of ConstructsTo create the full-length cDNA
encoding the major form of human salivary gland CPZ (i.e.
without the inserts), clones 13, 32, and 130196 were pieced together as
follows. The cDNA insert from clone 130196 was subcloned into the
EcoRI and HindIII sites of pGEM7zf. This plasmid
was cut with EcoRI and SacII and ligated with the
5 EcoRI/SacII fragment of clone 32. The
resulting plasmid was digested with EcoRI and
StuI and ligated with the EcoRI/StuI fragment of clone 13. This construct was subcloned into the
EcoRI and NotI sites of the baculovirus transfer
vector pVL 1393. Dideoxynucleotide sequencing was performed to confirm
the sequence of the construct.
Recombinant baculovirus expressing the various cDNAs were generated using the Baculogold kit (Pharmingen). For this, Sf9 insect cells growing in Sf900-II serum-free media (Life Technologies, Inc.) were co-transfected with the various cDNAs in pVL1392 or pVL1393 and the Baculogold viral DNA using the calcium phosphate procedure. After culturing for 5 days, 100 µl of media were used to infect 106 Sf9 cells in a 25-cm2 flask. Then, 1 ml of the media from this second infection was used to infect 108 Sf9 cells in 50 ml of media growing in a shaker flask. After 3 days of infection, cells and media were analyzed for the expression of enzyme activity and protein. For the experiment investigating the expression of CPZ activity in intact cells, the infection was performed for 2 days.
Western Blot Analysis and Carboxypeptidase AssayThe Sf9 cells from 50 ml of the culture were recovered by centrifugation at 1000 × g for 10 min. The cells were sonicated in 10 ml of 100 mM sodium acetate buffer, pH 5.5. For Western blot analysis, 1 µl of homogenate was loaded onto the 10% SDS-PAGE gel and transferred to nitrocellulose. The nitrocellulose was then probed with a 1:1000 dilution of a rabbit polyclonal antiserum raised against a bacterially expressed glutathione S-transferase fusion protein containing the C-terminal 73 residues of CPZ (Hazelton Research Products, Denver, PA).
For the carboxypeptidase assay comparing the various proteins, 25 µl of cell homogenate was assayed for carboxypeptidase activity using 0.2 mM dansyl-Phe-Ala-Arg in either 100 mM, pH 5.5, sodium acetate buffer or 100 mM, pH 7.4, Tris-Cl buffer, as previously described (14). In the experiments testing inhibitors, the enzyme and inhibitors were preincubated for 15 min in buffer (100 mM Tris-Cl, pH 7.4) at room temperature prior to adding substrate and then incubating at 37 °C for 1 h. To investigate whether CPZ activity was expressed on the cell surface, 4 × 106 baculovirus-infected cells were centrifuged at 1000 × g, washed with phosphate-buffered saline at 8 °C, and then resuspended in 0.5 ml of the same buffer. Some of the aliquots were sonicated for 10 s, and then 0.2 mM dansyl-Phe-Ala-Arg was added and the tubes incubated at 8 °C for 4 h. The tubes were then centrifuged to remove the cells, the supernatants were combined with 200 µl of 0.5 M HCl, and then 2 ml of chloroform were added. After mixing and centrifugation for 2 min at 300 × g, the amount of product was determined by measuring the fluorescence in the chloroform phase, as previously described (14).
Several novel sequences were identified using the Tblastn program
to search the expressed sequence tag data base for homology to human
CPE. Two of these clones (Fig. 1) were obtained and
fully sequenced in both directions. The two GenBankTM
clones were identical except for a 107-nucleotide region that was found
in clone 194316 but not in clone 130196. To further study this, human
salivary gland mRNA was used since preliminary studies using
reverse transription PCR showed that this tissue expressed CPZ
mRNA. Following reverse transcription of the salivary gland
mRNA with a 3 oligonucleotide, PCR was performed using oligonucleotides that spanned the 107-bp region. The PCR product showed
two bands that differed by approximately 107 nucleotides, with the
major band being the smaller 120-bp form (Fig. 2,
lane 2). Sequence analysis of clones 130196 and 32 revealed
a second variable region upstream of the 107-nucleotide insert; this
second region differed by 23 nucleotides among the various clones (Fig. 1). PCR analysis revealed a major band at 255 bp (Fig. 2, lane 1), indicating that the major form of CPZ mRNA in the salivary gland lacks an insert in this region. To obtain the full open reading
frame of CPZ, human salivary gland mRNA was used for 5
RACE PCR.
Several clones from independent PCR reactions showed a similar 850-bp
5
extension (Fig. 1). However, some clones also contained an 883-bp
extension due to a 33-bp insert near the 5
end (Fig. 1). To evaluate
the major form, PCR was performed with oligonucleotides spanning the
33-bp region. This analysis showed that the major form of CPZ mRNA
in human salivary gland lacks the 33-bp insert (not shown). In addition
to the clones with an 850-bp 5
extension, several clones from one PCR
reaction were found with 500-bp 5
extensions (Fig. 1, clone
7).
The first ATG is present in position 40 of the longest 5 sequences
obtained from RACE PCR (Fig. 3). The sequence
surrounding this ATG matches the consensus site for translation
initiation (15). Initiation at this site would produce a protein of 641 amino acids, based on the major form of CPZ mRNA found in the salivary gland (Fig. 3). This 641-amino acid protein contains a region
with homology to the metallocarboxypeptidase domain (Fig. 3). Residues
that are known to be involved in Zn2+ binding
(His69, Glu72, and His196, using
the CPA numbering system), substrate binding (Arg145, and
Tyr248), and catalytic activity (Glu270) of
other CP are present in comparable positions in CPZ (Fig. 3). The
overall homology of this CP domain in CPZ is highest to CPE and N, with
49% amino acid identity (Table I). CPM, AEBP1, and the
second domain of CPD all have comparable homology (39-41% amino acid
identity) to CPZ, and the first domain of CPD has slightly lower
homology (Table I). The third domain of CPD, and all members of the
CPA/B subfamily of metallocarboxypeptidases show homology of less than
24% with CPZ (Table I; data not shown).
|
Preceding the CP domain is a 167-residue N-terminal extension, of which the first 18 residues resemble a signal peptide motif (16). Data base searches revealed the presence of a 120-residue Cys-rich domain with 23-29% homology to a domain in human, rat, mouse, Caenorhabditis elegans, and Drosophila melanogaster frizzled proteins (17-19). This Cys-rich region of CPZ also shows 27% amino acid identity to a similar motif in one form of type XVIII collagen (20). Importantly, all ten of the Cys residues in this "fz" motif (20) of the frizzled proteins and collagen are present in comparable positions in CPZ, with only minimal insertions or gaps to align the sequences.
The form of CPZ mRNA with the 33-bp insert would produce a protein
11 amino acids longer (Fig. 4); this insert is in the
N-terminal region preceding the fz and CP domains (Fig. 3). However,
the other two inserts would cause frameshifts that would lead to
truncation of the protein within the CP domain (Fig. 4). The form of
CPZ mRNA corresponding to clone 7 (i.e. the 500-bp 5 extension)
has an in-frame ATG located 5 nucleotides from the 5
end (Fig. 3). If
translation were to initiate at this ATG, the protein would consist of
515 residues that would contain the CP domain but not the fz
domain.
Northern blot analysis shows two strong bands of 2.1 and 2.6 kb in
human placenta (Fig. 5, left). Longer
exposure of the same blot reveals a 2.1 and/or 2.6-kb band in all
tissues examined (Fig. 5, right). In addition, one or more
bands of 4.5-5 kb are detected in all tissues examined, and some
tissues also show lower bands of 1.1 or 1.5 kb (Fig. 5,
right).
To test whether CPZ encodes an active carboxypeptidase, the cDNA
encoding the 641-residue protein was subcloned into the baculovirus expression vector pVL1393 and expressed using the Baculogold system. Also, the N-terminally truncated cDNA, which encodes a 515-residue protein, was also expressed in the baculovirus system. After 72 h
of infection with the CPZ-expressing virus, cells were harvested, and
the expression of CPZ protein was determined by Western blot analysis
using an antiserum directed against the C-terminal region of CPZ. This
analysis shows the presence of immunoreactive proteins in the cells
infected with the two CPZ constructs but not in cells infected with
wild-type virus (Fig. 6). The size of the protein corresponds to 71 kDa in cells infected with the CPZ 641 construct and
54 kDa in cells infected with the CPZ 515 construct. These sizes are in
close agreement with the predicted sizes of 70.7 and 58.6 kDa,
respectively. Immunoreactive CPZ protein was not detectable in media
samples from the infected cells (Fig. 6).
The amount of CP activity was determined in cell homogenates and in
media using the substrate dansyl-Phe-Ala-Arg. At pH 5.5, only a small
amount of CP activity was detected in the cells infected with the
CPZ-producing constructs, but this activity was not significantly different than the activity measured with wild-type virus-infected cells (Table II). However, since CPZ also shows
considerable homology to CPN, which is maximally active at neutral pH,
we also assessed the activity of the baculovirus-expressed CPZ at pH
7.4. When assayed at pH 7.4, the 641-residue form of CPZ shows activity substantially above that of the negative controls (Table II). This
difference is statistically significant using Student's t test (p < 0.01). The 515-residue form of CPZ does not
appear to have enzyme activity above background levels at either pH
tested (Table II). In addition to the wild-type virus, other negative controls were also performed using virus that makes inactive
Pro202CPE (21) or an unrelated protein (the subunit of
human translation initiation factor 2). In all cases, the cells
infected with the 641-residue form of CPZ showed considerably more
activity than the control cells when measured at pH 7.4 (Table II).
Cells expressing AEBP1 do not show activity above that of the negative
controls (Table II); AEBP1 has been reported to possess CP activity
after expression in bacteria but does not contain all of the residues thought to be required for catalytic activity (9). In contrast, cells
infected with either CPE- or CPD-expressing virus show a large amount
of activity toward the substrate at pH 5.5 and less activity at pH 7.4 (Table II). The media from cells expressing CPE and CPD also shows high
levels of enzyme activity, while the media from the CPZ-expressing
constructs showed negligible amounts of enzyme activity (Table II).
|
The lack of CPZ activity in the media was unexpected due to the presence of an N-terminal signal peptide-like sequence. To investigate the possibility that CPZ remains bound to the cell surface, as does CPM, intact cells were assayed for enzyme activity at 8 °C to prevent internalization of the substrate. For this analysis, the cells were infected for only 2 days to minimize the amount of lysed cells. Cells infected with CPZ showed significantly more enzyme activity than cells infected with wild-type virus (Table III). Cell homogenates assayed under the same low temperature conditions showed approximately three times more enzyme activity than the intact cells, suggesting that one third of the total CPZ is expressed on the surface of the infected Sf9 cells (Table III).
|
The enzyme activity in Sf9 cells infected with CPZ was tested with a variety of compounds that inhibit other members of the metallocarboxypeptidase gene family. For this analysis, cells infected with wild-type virus were also examined, and this activity was subtracted from the activity observed with the CPZ-expressing virus. The CPZ activity was inhibited by the chelating agent 1,10-phenanthroline (Table IV). The active site-directed inhibitors 2-mercaptomethyl-3-guanidinoethylthiopropanoic acid (MGTA) and guanidinoethylmercaptosuccinic acid (GEMSA) inhibit CPZ activity at 1 mM concentrations (Table IV). When tested at 1 µM, only MGTA produced a substantial inhibition of the CPZ activity. Since the dansyl-Phe-Ala-Arg substrate is also hydrolyzed by chymotryptic-like enzymes that cleave C-terminal to Phe residues,3 we tested the chymotrypsin inhibitor tosyl-phenylalanyl-chloromethyl ketone (TPCK). TPCK did not inhibit the CPZ activity (Table IV). In contrast to the results with the CPZ-expressing cells, the activity in the cells infected with wild-type baculovirus was not substantially affected by 1,10-phenanthroline, GEMSA, or MGTA (Table IV). However, TPCK caused a moderate reduction in the activity detected with dansyl-Phe-Ala-Arg in cells infected with wild-type virus (Fig. 4), suggesting that a portion of this wild-type activity is due to a chymotryptic-like cleavage. The strong inhibition of the CPZ activity by the Arg analogs (MGTA and GEMSA) and the lack of cleavage by TPCK suggests that this activity is carboxypeptidase-like and not endopeptidase-like.
|
A major finding of the present study is that CPZ is a novel member
of the CPE subfamily of metallocarboxypeptidases. All members of this
subfamily contain a 380-400 amino acid CP domain that is preceded and
followed by N- and C-terminal sequences that are unique to each CP
(Fig. 7). The overall arrangement of the domains within
CPZ is more similar to that of CPE, CPN, and CPM than to other recently
discovered members of this gene family, CPD and AEBP1 (Fig. 7). AEBP1
contains much longer N- and C-terminal variable regions than the other
family members, and CPD contains three CP domains followed by a
transmembrane domain and short cytosolic tail (Fig. 7). The C-terminal
variable region of CPE, CPN, CPM, and CPD is thought to be involved in
the interaction of the enzyme with other proteins or with membranes,
and the variability in this region reflects the distinct interactions
that each protein undergoes (1). There are no obvious features that
would enable a prediction of the function of this C-terminal region of
CPZ. This region has an abundance of Gly and Pro, which together
comprise 25% of the amino acids of the C-terminal 70 residues. Also,
there are 4 Trp within the last 20 residues, several of which are part of a 12-residue hydrophobic stretch near the C terminus. It is possible
that this region is involved with the attachment of the protein to
membranes, either as a non-conventional membrane-spanning domain or as
a glycosyl-phosphatidylinositol-linked protein.
The N-terminal domains of most metallocarboxypeptidases contain a signal peptide for targeting to the secretory pathway (Fig. 7). Except for AEBP1, which hasn't been examined, all of the other metallocarboxypeptidases are either secreted from the cell (CPE, CPN), in the secretory pathway (CPD), or attached to the cell surface (CPM), and so the signal peptide is critical for the proper targeting of the protein. CPZ contains an N-terminal sequence that resembles the signal peptide consensus sequence (16) and thus appears to be similar to the other CP family members. Although the baculovirus-expressed CPZ is not secreted into the media, a considerable amount of activity is present on the cell surface of CPZ-expressing Sf9 cells. This result suggests that CPZ is routed through the secretory pathway.
Among the known members of the metallocarboxypeptidase gene family, CPZ is unique in the presence of the fz domain (Fig. 7). This 120-residue domain has been previously found in the Drosophila "frizzled" proteins and in the numerous mammalian homologs (17-19). Also, a similar domain is found in one form of type XVIII collagen but not in two other forms of this protein (20). Although the overall amino acid identity among the fz domain of CPZ and the various other fz domains is only 23-29%, the ten Cys residues are conserved and appear in generally comparable positions as in the other fz domains, suggesting a similar structural motif (20). The function of the fz domain is not clear. The frizzled proteins themselves are receptors with seven-transmembrane-spanning domains (19). The Drosophila frizzled proteins are involved with the development of tissue polarity, and the Drosophila gene frizzled-2 has been recently found to encode the receptor for the signaling molecule designated Wingless (19). In the frizzled proteins, the fz domain is believed to be extracellular since it is located N-terminal to the seven-transmembrane segments. It has been proposed that the fz domain is involved with the binding of ligand and that this binding then changes the interaction of the fz domain with the remainder of the receptor and leads to the second messenger activation (18). It is possible that the fz domain of CPZ competes with the frizzled receptors for ligand binding, although the role of the carboxypeptidase in this model is not obvious.
The function of the fz domain of CPZ may be distinct from the function
of the CP domain. It is interesting that some of the forms of CPZ
mRNA detected upon library screening or PCR analysis would encode
proteins that contain only the fz domain and a portion of the CP domain
(i.e. the +23 and +107 variants in Figs. 1 and 4). Also, if
the clone detected by 5 RACE PCR (Fig. 1, clone 7)
corresponds to a form of CPZ mRNA that has a shorter 5
region, this would encode a protein that lacks the fz domain but contains the
full-length CP domain. The variability in the forms of CPZ mRNA is
analogous to the finding that several forms of type XVIII collagen
exist, of which only one contains the fz domain (20). However, when
this shorter form of CPZ was expressed in baculovirus, the protein was
not enzymatically active toward the substrate tested (Table II),
raising doubts as to whether this shorter protein would be a functional
carboxypeptidase.
The variants with the additional 33, 23, or 107 nucleotides presumably
arise through alternative splicing of introns; all three longer
sequences contain the intron splice acceptor site (AG) on the 3 end of
the insert (Fig. 4). The 23-nucleotide insert is present in the exact
position of an intron found in CPE, CPA, and CPB, and the
107-nucleotide insert is present in the exact position of an intron in
CPA and CPB (22, 23). Although these longer variants are minor in adult
human salivary gland mRNA, they may be more abundant in other
tissues or during development. The large number of mRNA species
detected upon Northern blot analysis (Fig. 5) suggests substantial
variability in the forms of CPZ mRNA and protein.
The carboxypeptidase activity detected when the full-length 641-residue form of CPZ was expressed in baculovirus was approximately 10-fold lower than the activity resulting from similar baculovirus expression of CPE, CPD, and CPM (Table II, and not shown). The low CP activity could be due to suboptimal conditions (substrate, buffer), or the lack of proper activation in the baculovirus system (pro peptide cleavage, or other events). Further studies are needed to purify and characterize CPZ from both the baculovirus system and from an endogenous source; these studies are beyond the scope of the present investigation. Unlike AEBP1, which is missing several of the residues thought to be essential for catalytic activity and substrate binding, CPZ contains all of the active site residues of CPA and B that are conserved in members of the CPE subfamily. Critical residues of all metallocarboxypeptidases include the Zn2+ binding ligands His69, Glu72, and His196 (using the CPA numbering system), the substrate binding Arg145 and Tyr248, and the catalytically important Glu270 (Fig. 7). Another residue that is important for CPA and B is Arg127; this residue is thought to help polarize the carbonyl group at the cleavage site (24). In CPZ, no Arg residue is present in a position comparable with this Arg127 in CPA. However, none of the other members of the CPE subfamily contain an Arg in a position comparable with Arg127, and it is likely that another residue in this subfamily contributes to the polarization of the carbonyl group. As a side point, we could not detect a significant amount of activity when AEBP1 was expressed in baculovirus, despite relatively high levels of protein expression by Western blot analysis.4 AEBP1 is predicted to encode a carboxypeptidase with low catalytic activity due to the substitution of Glu270 by a Tyr.
Our search for novel CP was directed toward secretory pathway enzymes that could substitute for the defective CPE in the Cpefat/Cpefat mice. Although CPZ is likely to be present in the secretory pathway due to the N-terminal signal peptide-like sequence, the general lack of enzyme activity at pH 5.5 would suggest that this enzyme will be inactive in the environment of the mature secretory vesicle. While it is possible that CPZ undergoes an activation step that permits activity at pH 5.5, this is considered unlikely. Thus, CPZ is not proposed as a candidate peptide processing enzyme. Based on the activity at neutral pH and on the presence of a fz domain, a domain found in the extracellular portion of other proteins, it is likely that CPZ performs an extracellular function. This function may involve the extracellular processing of peptides, as proposed for CPM and N. Previous reports have described soluble CP activities in seminal fluid, urine, and brain (25-27), and soluble metallocarboxypeptidase activity has been detected in saliva and in aqueous humor from the eye.5 Based on the enzymatic and physical properties, several of these activities appear to be distinct from the previously characterized CP. It is possible that one or more of these activities are due to CPZ. Finally, it should be noted that CPZ is not the only remaining metallocarboxypeptidase to be identified as computer searches have revealed at least two more family members, and the isolation of full-length cDNA clones for these newer members is currently in progress.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U83411[GenBank].
Yinghong Lei is gratefully acknowledged for performing the Northern blot analysis. We especially thank the following people for providing cDNAs in baculovirus expression plasmids: Dr. Frank Eng, for duck CPD cDNA; Dr. Hyo-Sung Ro, for AEBP1 cDNA; and Dr. Uma Maitra, for the cDNA encoding the gamma subunit of human translation initiation factor 2. The DNA sequencing facility of the Albert Einstein College of Medicine is supported in part by Cancer Center Grant CA13330.