Molecular cloning and expression of rat brain endopeptidase 3.4.24.16.

We have isolated by immunological screening of a ZAPII cDNA library constructed from rat brain mRNAs a cDNA clone encoding endopeptidase 3.4.24.16. The longest open reading frame encodes a 704-amino acid protein with a theoretical molecular mass of 80,202 daltons and bears the consensus sequence of the zinc metalloprotease family. The sequence exhibits a 60.2% homology with those of another zinc metallopeptidase, endopeptidase 3.4.24.15. Northern blot analysis reveals two mRNA species of about 3 and 5 kilobases in rat brain, ileum, kidney, and testis. We have transiently transfected COS-7 cells with pcDNA containing the cloned cDNA and established the overexpression of a 70-75-kDa immunoreactive protein. This protein hydrolyzes QFS, a quenched fluorimetric substrate of endopeptidase 3.4.24.16, and cleaves neurotensin at a single peptide bond, leading to the formation of neurotensin(1, 2, 3, 4, 5, 6, 7, 8, 9, 10) and neurotensin (11, 12, 13). QFS and neurotensin hydrolysis are potently inhibited by the selective endopeptidase 3.4.24.16 dipeptide blocker Pro-Ile and by dithiothreitol, while the enzymatic activity remains unaffected by phosphoramidon and captopril, the specific inhibitors of endopeptidase 3.4.24.11 and angiotensin-converting enzyme, respectively. Altogether, these physicochemical, biochemical, and immunological properties unambiguously identify endopeptidase 3.4.24.16 as the protein encoded by the isolated cDNA clone.

Endopeptidase 3.4.24.16 is a metalloendopeptidase ubiquitously distributed in the central nervous system and in peripheral organs of mammals (1). This enzyme was first detected (2) and later purified (3) on the basis of its ability to cleave the Pro 10 -Tyr 11 bond of the tridecapeptide neurotensin, leading to the formation of the biologically inactive catabolites, neurotensin (1-10) and neurotensin (11)(12)(13). Studies on neurotensin catabolism in vitro by membrane fractions or cell lines of central or peripheral origin indicated that endopeptidase 3.4.24. 16 was the only peptidase that ubiquitously contributed to the inactivation of this neuropeptide (4). Several lines of evidence later suggested that endopeptidase 3.4.24.16 indeed participated to neurotensin inactivation in vivo in the gastrointestinal tract (5). Thus, by means of a vascularly perfused model of dog ileum, we showed that the dipeptide Pro-Ile, a fully selective blocker of endopeptidase 3.4.24.16 (6), inhibited the formation of one of the major catabolites, i.e. neurotensin (1-10) (5), leading to a drastic protection of neurotensin from degradation. In the central nervous system, we recently showed that mixed inhibitors of endopeptidases 3.4.24.16 and 3.4.24.15 potently enhanced the neurotensin-induced analgesia in the hot platetested mice (7). Altogether, this indicates that endopeptidase 3.4.24.16 contributes to the catabolism of neurotensin in vivo in the periphery but also likely in the central nervous system.
The characterization of the biochemical and pharmacological properties of endopeptidase 3.4.24.16 indicated that the enzyme behaved as a 70 -75-kDa monomer that was inhibited by metal chelators and dithiothreitol (3). Several studies suggested that endopeptidase 3.4.24.16 resembles another metallopeptidase, endopeptidase 3.4.24.15. Particularly, studies on the specificity of endopeptidase 3.4.24.15 showed that the enzyme cleaved several neuropeptides at peptidyl bonds that were reminiscent of those targeted by endopeptidase 3.4.24.16 (8,9). However, several aspects that included the nature of the cleavage site on neurotensin, the sensitivity to dipeptide inhibitors and dithiothreitol, as well as immunological data clearly distinguished the two peptidases (10). The present paper reports on the molecular cloning and expression of rat brain endopeptidase 3.4.24.16 and establishes that the two peptidases are related but clearly distinct molecular entities.

Screening of a Rat Brain cDNA Library and Isolation of the Full-length cDNA
A rat brain cDNA library constructed in the ZAPII vector (Stratagene) was screened with a polyclonal antibody raised against the rat brain endopeptidase 3.4.24.16 (1). Approximately 6 ϫ 10 5 recombinant phages were plated and incubated at 42°C for 3.5 h; then each plate was overlaid with a nylon filter previously saturated in 10 mM isopropyl-␤-D-thiogalactopyranoside and incubated at 37°C for 3 h. The filters were blocked overnight at 4°C in TBST buffer (150 mM NaCl, 0.05% Tween in 10 mM Tris-HCl, pH 8) containing 5% of fat milk, then incubated for 8 h at 4°C with a 1/1000 dilution of the primary antibody in TBST containing 1% of fat milk. Filters were washed three times (5 min each) in TBST and incubated for 1 h at room temperature in 1% fat milk TBST containing a 1/2500 dilution of a goat anti-rabbit antibody conjugated with horseradish peroxidase. The antibody-antigen complexes were revealed with the chromogenic substrate 4-chloro-1-naph-* This work was supported by CNRS and INSERM. 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.
‡ Recipient of a grant from the Association pour la Recherche contre le Cancer.
The cDNAs of 14 isolated positive clones were subcloned into pBluescript by in vivo excision according to the manufacturer's procedures (Stratagene). A clone, 7a with an insert of 1806 bp, 1 was sequenced and showed an open reading frame of 1613 bp, lacking the 5Ј-region coding for the N-terminal domain of the protein. Using two synthetic oligonucleotides, a polymerase chain reaction fragment of 1390 bp was derived from the 7a clone, labeled with 32 P by random-priming (Appligene), and used as a probe to screen 6 ϫ 10 5 clones of the above ZAPII cDNA library. Hybridization was carried out overnight at 65°C in 6 ϫ SSC, 0.1% SDS, 5 ϫ Denhardt's solution, and 0.2 mg/ml heatdenatured herring sperm DNA. The filters were washed in 3 ϫ SSC, 0.1% SDS at room temperature and autoradiographied. A clone, B1 containing an insert of 2158 bp, was isolated. This clone encompassed 1516 bp of clone 7a ( Fig. 1) but lacks the complete 3Ј-region as illustrated by the absence of a stop codon. Therefore the full-length cDNA was reconstituted with these two overlapping cDNAs by ligating a 380-bp NcoI-EcoRI fragment of 7a with a 2120-bp KpnI-NcoI fragment of B1. The resulting insert, 7aB1, was subcloned in pBluescript previously digested with KpnI-EcoRI. This construction allowed us to confirm the whole sequence of the cDNA and to verify that the ligation of the two fragments occurred without introduction of errors in the coding phase.

cDNA Sequencing
The automated sequencing was performed by means of the dideoxy chain termination method (11) on both strands by walking along the cDNA using synthetic oligonucleotides according to the strategy described in Fig. 1.

Northern Blot Analysis
Poly(A) ϩ mRNAs were prepared from rat tissues by purification on oligo(dT) columns. 5 g of poly(A) ϩ mRNAs were electrophoresed on a 1% formaldehyde/agarose gel, transferred onto a nylon membrane, and hybridized with the polymerase chain reaction probe derived from the 7a clone described above, in 50% formamide, 5 ϫ SSC, 10% dextran sulfate, 1 ϫ Denhardt's solution, and 0.2 mg/ml heat-denatured herring sperm DNA for 15 h at 42°C. The filter was washed two times at room temperature in 1 ϫ SSC, 0,1% SDS and two times at 50°C in 0,5 ϫ SSC, 0,1% SDS. Autoradiography was performed at Ϫ70°C for 3 days.

Transient Expression of Endopeptidase 3.4.24.16 in COS-7 Cells
The 7aB1 fragment was excised from pBluescript with KpnI-EcoRI and subcloned into the KpnI-EcoRI site of the eukaryotic expression vector pcDNA 3 (Invitrogen). Semi-confluent COS-7 cells, grown in 100-mm cell culture dishes, were transfected with 1 g of pcDNA 3 -7aB1 by the DEAE-dextran precipitation method (12). Negative control was performed in the same conditions with 1 g of pcDNA 3 vector (mocktransfected COS-7 cells). Approximately 48 h after transfection, the cells were collected, washed with 25 mM Tris-HCl, pH 7.5, buffer containing 250 mM sucrose and 1 mM EDTA, centrifuged for 10 min at 5000 rpm, and homogenized in 5 mM Tris-HCl, pH 7.5. Protein concentrations were determined by the Bradford method with ␥ globulin as standard (13).

Endopeptidase 3.4.24.16 Assays
Fluorimetric Analysis of QFS Hydrolysis-QFS (50 M) was incubated for various times at 37°C with COS-7 cells protein homogenate (final concentration of 0.1 mg/ml of protein) in a final volume of 100 l of 50 mM Tris-HCl, pH 7.5, in the absence or in the presence of different concentrations of inhibitors. Incubations were stopped with 2.5 ml of sodium formate, pH 3.7, and fluorimetrically monitored at 345 nm and 405 nm as described previously (14).
HPLC Analysis of Neurotensin and QFS Hydrolysis-Neurotensin (2 nmol) and QFS (5 nmol) were incubated for various times at 37°C with COS-7 cells protein homogenate (final concentration of 0.1 mg/ml of protein) in a final volume of 100 l of 50 mM Tris-HCl, pH 7.5, in the absence or in the presence of inhibitors. Reactions were stopped, centrifuged for 5 min at 10,000 ϫ g, and then supernatants were HPLC analyzed with the triethylamine/trifluoroacetic acid/acetonitrile chromatographic system previously described (4).

Western Blot Analysis
Samples were analyzed on a 8% SDS-polyacrylamide gel, and proteins were transferred onto a nylon membrane. Hybridization conditions with a 1/1000 dilution of the IgG-purified fraction of the polyclonal antiserum developed toward rat brain endopeptidase 3.4.24.16, and the revelation of the IgG-antigen complexes were as described previously (1).

Molecular
Cloning of Rat Brain Endopeptidase 3.4.24.16-We previously purified endopeptidase 3.4.24.16 from rat brain by means of various chromatographic steps and preparative SDS-PAGE (for review see Ref. 14). This allowed us to obtain sufficient amounts of purified material to raise a polyclonal antiserum, the purified IgG fraction of which specifically recognized the native and denaturated forms of endopeptidase 3.4.24.16 (1). This immunological tool allowed us to finally obtain a 2448-bp-long cDNA (clone 7aB1, see "Experimental Procedures" and Fig. 1), of which the longest open reading frame of 2112 bp encoded a 704-amino acid polypeptide of theoretical molecular mass of 80,202 daltons and bore the HEXXH consensus sequence (residues 497-501, Fig. 2) that is the structural signature of members of the zinc metalloprotease family (15). Hydropathic profile (Fig. 3) did not delineate a putative signal peptide but indicated several hydrophobic stretches around the consensus sequence (Fig. 3). A cluster of charged amino acid residues (residues 331-335 and 341-348) and three putative Asn-X-Ser/Thr consensus sequences for glycosylation sites were also identified on the sequence (Fig. 2). The 3Ј-end of the cDNA of clone 7aB1 was 193 bp long and did not reveal a polyadenylation signal, indicating that the 3Ј-noncoding region was likely not complete.
Northern blot analysis performed with rat brain, ileum, kidney, and testis poly(A) ϩ mRNAs consistently revealed two mRNA species of about 3 and 5 kilobases, the lower molecular weight label always being prominent (Fig. 4). It is noticeable that the label was lower in brain than in other tissues (Fig. 4). This is in agreement with our data concerning the relative endopeptidase 3.4.24.16 activities detected in various tissues (14). It is not yet clear as to whether the discrepancy observed 1 The abbreviations used are: bp, base pair(s); PAGE, polyacrylamide gel electrophoresis; HPLC, high pressure liquid chromatography. between the sizes of these two mRNA species reflects a variable length of their non-coding 3Ј-region, consequently, to two distinct polyadenylation sites. An alternative hypothesis could be that the higher molecular weight mRNA represents an intermediate immature form of the mRNA. Finally, the possibility that the mRNAs encode two distinct proteins could be evoked. However, such a hypothesis is not sustained by our previous data indicating that in whole rat brain homogenate, a tissue that would be expected to contain all the various putative molecular forms of the peptidase, we consistently detected a single immunolabeled protein migrating with the apparent molecular weight of the recombinant peptidase (1).
Clone 7aB1 Encodes Endopeptidase 3.4.24.16-7aB1 cDNA was subcloned into the eukaryotic expression vector pcDNA 3 and transiently transfected in COS-7 cells. Western Poly(A) ϩ RNA (5 g) from various rat tissues was fractionated on a 1% formaldehyde/agarose gel, blotted on a nylon membrane, and hybridized with the 32 P-labeled polymerase chain reaction fragment derived from a 7a clone as described under "Experimental Procedures." RNA molecular weight markers are indicated in kilobases (kb) on the left. blot analysis indicated that these whole cells overexpressed a 70 -75-kDa protein that was recognized by anti-rat brain endopeptidase 3.4.24.16 (Fig. 5A, lane 2), while mock-transfected COS-7 cells did not (Fig. 5A, lane 1). The label appears mainly associated with the soluble fraction prepared from the transfected cells, although a minor membrane-associated immunoreactive protein was also observed (Fig. 5B). Deglycosylation experiments did not affect the apparent molecular weight of the immunoreactive proteins, whatever the subcellular fraction that was examined (Fig. 5B). The enzymatic activity of the expressed protein was examined by means of a quenched fluorimetric substrate (QFS) that was previously shown to behave as a good substrate of endopeptidase 3.4.24.16 (16). An important QFS-hydrolyzing activity was monitored in pcDNA 3 -7aB1 transfected COS-7 cells (specific activity of 60 -80 nmol/h/mg of proteins) but not in mock-transfected cells (Fig. 6A). HPLC analysis indicated that one of the cleavage products eluted with the retention time of 3-carboxy-7-methoxycoumarin-Pro-Leu (Fig. 6B). This agreed well with data obtained with purified endopeptidase 3.4.24.16 (16). The QFS hydrolysis was dose-dependently and fully prevented by the endopeptidase 3.4.24.16 dipeptide blocker, Pro-Ile (Fig. 6C) and dithiothreitol (Fig. 6D) with K i values of about 180 M and 0.4 mM, respectively, in good agreement with the K i values reported for the purified endopeptidase 3.4.24.16. Furthermore, the activity was totally abolished by 1 mM of o-phenanthroline but remained unaffected by saturating concentrations of captopril, phosphoramidon, and phosphodiepryl 20, the fully specific inhibitors of angiotensin-converting enzyme (17), endopeptidase 3.4.24.11 (18), and endopeptidase 3.4.24.15, respectively (data not shown). Transfected COS-7 cells readily cleaved neurotensin at a single peptide bond as indicated by the HPLC analysis that revealed the only formation of neurotensin (1-10) and neurotensin (11)(12)(13) (Fig. 7B), the production of which appeared fully prevented by Pro-Ile (Fig. 7C) and by dithiothreitol (Fig. 7D). These fragments were not generated by mock-transfected COS-7 cells (Fig. 7A). DISCUSSION The immunological approach used in the present study has led us to isolate a cDNA clone that unambiguously encodes endopeptidase 3.4.24.16. First, the protein overexpressed in transfected COS-7 cells is recognized by the IgG-purified fraction of a specific polyclonal antibody developed toward rat brain endopeptidase 3.4.24.16 (1). Second, transfectant cells hydrolyze two peptides (QFS and neurotensin) at peptide bonds identical with those targeted by purified endopeptidase 3.4.24.16 (3,16). Third, the catalytic activity of the protein produced by transfectants is fully inhibited by Pro-Ile, a dipeptide that selectively blocks endopeptidase 3.4.24.16 (6), and by dithiothreitol (19) in agreement with the pharmacological spectrum previously established for rat endopeptidase 3.4.24.16 (14). Furthermore, the activity remains insensitive to the specific inhibitors of angiotensin-converting enzyme, endopeptidase 3.4.24.11, and endopeptidase 3.4.24.15.
It is interesting to note that transfected cells cleave neurotensin at a single peptide bond, giving rise to neurotensin (1-10) and neurotensin (11)(12)(13). It was recently suggested that endopeptidase 3.4.24.16 of porcine origin targeted the same peptide bond but also triggered a minor production of neurotensin (1-8) and neurotensin (9 -13) (20). The present study clearly indicates that the production of neurotensin (1-8) and neurotensin (9 -13) by the purified enzyme from porcine sources was indeed due to a peptidase distinct from endopeptidase 3.4.24.16. This agrees well with our recent work showing that such neurotensin (1-8) and neurotensin (9 -13) production derived from the participation of contaminating endopeptidase 3.4.24.15. 2 The purification of endopeptidase 3.4.24.16 from various rat organs allowed us to establish that the apparent molecular mass of the enzyme corresponded to 70 -75 kDa (3,21). This appeared to be slightly lower than the molecular mass deduced from the longest open reading frame. Apparently, none of the nucleotidic sequences that abut to the ATG initiation codons fulfilled the structural features that unambiguously identify the eukaryotic Kozak sequence usually required to modulate the initiation of the translation (22). However, according to the fact that a purinergic base appears generally required at the position 3 upstream to the initiation codon (22), the second ATG codon appears the best candidate to initiate the genuine open reading frame encoding the enzyme. This will raise a protein of 77,724 daltons in good agreement with the reported molecular mass of the purified enzyme (3).
We previously established that endopeptidase 3.4.24.16 was predominantly recovered in majority in a soluble form in the brain (23). This also appears to be the case in transfected COS-7 cells as illustrated in Fig. 5B, which indicates a major soluble form of the protein, in agreement with the recovered activity in the two subcellular fractions (not shown). Previous immunological data clearly indicate that a minor fraction of endopeptidase 3.4.24.16 could exist in a genuine membraneassociated form in the brain (24). This hypothesis was reinforced by light and electron microscopic analysis of the localization of endopeptidase 3.4.24.16 in rat mesencephalon (25). Thus, it was shown that endopeptidase 3.4.24.16-like immunoreactivity could be characteristically associated with restricted zones of the plasma membrane of a subpopulation of neurons in the rat substantia nigra and ventral tegmental area (25). Biochemical analysis of the type of association of endopeptidase 3.4.24.16 with the membrane of kidney microvilli indicated that the enzyme was not attached to the membrane by a glycosyl-phosphatidylinositol anchor (21) but partitioned in the detergent phase after Triton X-114 phase separation (21), a physicochemical behavior that appears to be common to various intrinsic membrane proteins (26).
Sequence analysis of endopeptidase 3.4.24.16 does not reveal the clearcut structural requirements generally fulfilled by intrinsic membrane-bound proteins. First, it is not possible to clearly delineate a N-terminal signal peptide that could serve as a membrane anchor, as has been shown for endopeptidase 3.4.24.11 (27). In agreement with this observation, it is noticeable that membrane-associated and soluble forms of endopeptidase 3.4.24.16 comigrate after SDS-PAGE and Western blot analysis experiments (Fig. 5B). Second, although there exist three putative glycosylation sites, deglycosylation experiments performed with the whole homogenate of transfected cells as well as with the membrane-associated and soluble enzymes did not affect the apparent molecular weight of the peptidase (Fig.  5B), in agreement with our previous biochemical data showing that endopeptidase 3.4.24.16 did not bind to various sugarlinked resins (3). However, it is interesting to note that several clusters of hydrophobic residues can be deduced from the hydropathic profile of the protein that could be responsible for some protein-protein interactions. Furthermore, one can underline the presence of a stretch of charged residues at amino acids 331-335 and 341-348. This could be of importance with respect to a previous work showing that carboxypeptidase E displayed a similar domain rich in charged amino acids (28) that was shown to be responsible for the attachment of the "membrane-bound" carboxypeptidase E counterpart to the plasma membrane (29). Mutagenesis analysis experiments should allow us to examine whether the above possibilities could account for the anchoring of the "membrane-bound" form of endopeptidase 3.4.24.16.
Purified endopeptidase 3.4.24.16 is sensitive to metal chelators such as EDTA and o-phenanthroline (3). We showed that the activity of the apoenzyme could be restored upon incubation with various divalent cations, the most efficient recovery being obtained with zinc (19). The sequence of endopeptidase 3.4.24.16 reveals the presence of an HEFGH sequence that confirms that the enzyme belongs to the zinc metalloprotease family (15,30).
The current knowledge of the biochemical and physicochemical features of endopeptidases 3.4.24.16 and 3.4.24.15 and their specificity toward various neuropeptides underlined that the two enzymatic activities share some similar properties (8,9). On the other hand, the two peptidases can be distinguished by their distinct cleavage sites for neurotensin (3,8), their sensitivity to dipeptide inhibitors (16,23), and by the lack of 2 B. Vincent, J. P. Vincent, and F. Checler, submitted for publication. Neurotensin (2 nmol) was incubated for 2 h at 37°C with 10 g of protein homogenate from pcDNA 3 (A) and pcDNA 3 -7aB1 transfected COS-7 cells (B-D) in absence (B) or in the presence of 10 mM Pro-Ile (C) or 5 mM dithiothreitol (D). HPLC analysis was performed as described under "Experimental Procedures." Arrows indicate the elution times of synthetic peptides run in the same HPLC conditions. recognition of endopeptidase 3.4.24.15 by the IgG-purified fraction of the antiserum raised against rat brain endopeptidase 3.4.24.16 (1,10). Furthermore, Orlowski et al. (8) reported on the activation of endopeptidase 3.4.24.15 by low concentrations of dithiothreitol while this peptidase appeared inhibited by higher concentrations of such agents. This appeared not to be the case for endopeptidase 3.4.24.16, which is never activated by dithiothreitol whatever the concentration that were examined (19). According to the above considerations, it is therefore not unexpected to find that the sequence of endopeptidase 3.4.24.16 displays a 60.2% homology with that of endopeptidase 3.4.24.15 (31).
It is interesting to note that endopeptidase 3.4.24.15 exhibits a 35.5% homology with proteinase YscD (32). This protein is encoded by the PRD1 gene borne by the chromosome III of yeast and was therefore claimed to be the yeast analog of endopeptidase 3.4.24.15 (32). However, the purified yeast enzyme generated neurotensin (1-10) from neurotensin and was not activated by dithiothreitol (33), two properties reminiscent of endopeptidase 3.4.24.16. The fact that endopeptidase 3.4.24.16 displayed a 35.7% identity with the yeast enzyme strongly suggests the possibility that YscD could indeed be the yeast counterpart of endopeptidase 3.4.24.16.
Recently, the complete sequence of a microsomal metalloendopeptidase from rabbit liver was established (34) and shows 90.3% identity with that of endopeptidase 3.4.24.16. Sequence homology strongly suggests that microsomal metalloendopeptidase corresponds to the rabbit counterpart of rat endopeptidase 3.4.24.16. However, very limited information exists on the specificity of this enzyme toward natural neuropeptides since Kawabata and Davie (35) only reported on the ability of microsomal metalloendopeptidase to cleave a synthetic peptide that mimics the amino acid sequence encompassing the processing site of vitamin K-dependent proteins. Further studies are clearly needed to document the specificity of microsomal metalloendopeptidase with respect to the known properties of endopeptidase 3.4.24.16. The sequence of endopeptidase 3.4.24.16 exhibits 24.2 and 25.6% homology with those of a rat liver mitochondrial intermediate peptidase (36,37) and a dipeptidyl carboxypeptidase from Escherichia coli (38), respectively. Finally, the enzyme did not align with the sequences of endopeptidase 3.4.24.11 (39,40) and angiotensin-converting enzyme (41)(42)(43).
The isolation of the cDNA clone of endopeptidase 3.4.24.16 should allow us to express a high amount of the recombinant protein. The recent design of highly potent inhibitors of endopeptidase 3.4.24.16 and their use to affinity purify the enzyme should allow us to obtain high quantities of pure enzyme. This tool should be of importance to examine the detailed structural features of the enzyme and to envision crystallographic experiments. The cDNA should also prove useful to delineate the putative biological signals that could modulate the level of expression of the peptidase as it seems to occur during the differentiation processes of primary cultured neurons (24).