Conserved cysteine and tryptophan residues of the endothelin-converting enzyme-1 CXAW motif are critical for protein maturation and enzyme activity.

The neprilysin (NEP)/endothelin-converting enzyme (ECE) family of metalloproteases contains a highly conserved carboxyl-terminal tetrapeptide sequence, CXAW, where "C" is cysteine, "X" is a polar amino acid, "A" is an aliphatic residue, and "W" is tryptophan. Although this sequence strongly resembles a prenylation motif, human ECE-1 did not appear to be prenylated when labeled in vivo using various isoprenoid precursors in cell lines expressing ECE-1. We used site-directed mutagenesis to investigate the role of the CXAW motif and determined that the conserved cysteine residue of the CXAW motif in ECE-1, Cys(755), is critical for proper folding of the enzyme, its export from the endoplasmic reticulum, and its maturation in the secretory pathway. In addition, site-directed mutagenesis revealed that the conserved tryptophan residue of the sequence CEVW appears to be important for endoplasmic reticulum export and is essential for enzyme activity. Deletion of Trp(758) or substitution with alanine greatly slowed maturation of the enzyme, and resulted in more than a 90% loss of enzyme activity relative to the wild type. Conservative substitution of the tryptophan with phenylalanine did not reduce activity, whereas replacement with tyrosine, methionine, or leucine reduced enzyme activity by 50%, 75%, and 85%, respectively. Together, these data indicate that the conserved CEVW sequence does not serve as a prenylation signal and that both the conserved cysteine and tryptophan residues are necessary for proper folding and maturation of the enzyme. Furthermore, the conserved tryptophan appears to be critical for enzyme activity.

The carboxyl terminus of each of these metalloproteases terminates in a highly conserved tetrapeptide sequence CXAW. Most frequently a charged (E/R) or uncharged polar residue (Q/S) follows the cysteine, and is in turn followed by a hydrophobic residue (V/L/I) preceding the Trp. The CXAW sequence resembles a possible CAAX prenylation motif where "C" is cysteine, "A" is an aliphatic residue, and "X" is any amino acid. A GenBank™/EMBL Data bank search for proteins terminating in CAAX, and containing the conserved zinc-binding motif of metalloproteases, HEXXH, revealed only members of the NEP/ECE family. The conservation of this sequence suggests that it might play an important role in these proteins.
We wanted to investigate the function of the CXAW motif for this family of proteases using ECE-1 as an example. ECE-1 is responsible for the final proteolytic processing step in the biosynthesis of endothelins (ETs) (5) and may be involved in the synthesis and/or degradation of other peptide hormones as well (12,13). ECE-1 cleaves the biologically inactive big ET between Trp 21 and Val 22 /Ile 22 to generate the 21-amino acid mature peptide ET, a potent vasoconstrictor (4,5). The enzyme therefore plays a key role in maintaining vascular tone by catalyzing the production of ET, making it an important target for the treatment of pathological conditions such as cardiovascular and renal diseases.
One potential role for the CXAW sequence is that it serves as a novel prenylation site, possibly involving a unique prenyltransferase. Prenylation, a post-translational lipid modification necessary for the association of proteins with membranes as well as for specific protein-protein interactions, involves the covalent attachment of a 15-carbon farnesyl or a 20-carbon geranylgeranyl isoprenoid moiety through a thioether linkage to the CAAX motif cysteine (for review, see Refs. 14 and 15). The nature of the isoprenyl group is primarily dependent upon the amino acid in the X position. Well characterized CAAX motifs include the sequences CAAM, CAAS, or CAAA, which confer the addition of a farnesyl moiety to proteins, and CAAL and CAAI, specific for geranylgeranyl addition (reviewed in Refs. 14 and 15). The topology of the NEP/ECE family proteins suggests that the CXAW sequence would be an unusual location for prenylation to occur since they are type II transmembrane proteins with a lumenal carboxyl terminus. The question of prenylation was particularly interesting because there are no known examples of prenylation of a lumenal CAAX motif. In addition, we wanted to investigate the importance of the conserved tryptophan residue in this motif. The sequence CAAW has not been identified as a common prenyltransferase substrate, yet the tryptophan in this tetrapeptide is completely conserved throughout the NEP/ECE family of proteases In the present study, we wished to analyze the tetrapeptide CXAW sequence of ECE-1, CEVW, and determine a possible role for the motif. In order to investigate whether the sequence serves as a prenylation signal, in vitro and in vivo prenylation assays were developed to assess prenylation of ECE-1. Additionally, site-directed mutations of the CXAW motif of ECE-1 were constructed to characterize the role of the highly conserved cysteine and tryptophan residues. Cell Culture-Chinese hamster ovary (CHO-K1) cells were cultured in DMEM/F-12 (Life Technologies, Inc.) and supplemented with 10% fetal bovine serum (Life Technologies, Inc.). The stable CHO cell line expressing ECE-1a, CHO/ECE-1a, was maintained in the same medium as above including 1 mg/ml G418. Pooled human umbilical vein endothelial cells (HUVEC) were grown in EGM-2 growth medium supplied by the manufacturer (Clonetics).

Materials
Plasmids and Mutagenesis of ECE-1-The cDNA for full-length ECE-1a was provided by M. Yanagisawa and transfected into CHO cells to make a stable cell line expressing ECE-1 (5). A flag-tagged construct of ECE-1a was constructed by placing the flag epitope (DYKDDDDK) at the amino terminus of full-length ECE-1a immediately after the start methionine. This construct was then subcloned into the mammalian expression vector pZeoSV (Invitrogen). Mutagenesis was performed on the wild type flag-ECE-1a cDNA using the QuikChange site-directed mutagenesis kit from Stratagene. All mutations were confirmed by DNA sequencing, and cassettes containing each mutation were subcloned using SpeI and KpnI into a wild type pZeoSV/flag-ECE-1a plasmid that had not been subjected to mutagenesis In Vitro Prenylation Assays-Biotinylated hexapeptides HKCEVW, TKCVIM, TKCVLS, and TKCVIL were incubated in reaction buffer with CHO cell lysate as a source of prenyltransferase enzymes. CHO cells were grown to confluence, trypsinized, and resuspended in homogenization buffer (50 mM Tris-HCl, pH 7.4, 0.1 M NaCl, 0.1 M sucrose, 1 mM Pefabloc, 1 g/ml pepstatin A, and 50 g/ml leupeptin), Douncehomogenized, and centrifuged at 2,000 rpm for 5 min. Cell lysate and hexapeptides were incubated with 0.67 M [ 3 H]farnesyl pyrophosphate or 1 M [ 3 H]geranylgeranyl pyrophosphate in assay buffer (50 mM HEPES-KOH, pH 7.5, 5 mM MgCl 2 , 0.1% polyethylene glycol 8000, 20 M ZnCl 2 ) with or without 5 mM dithiothreitol at 37°C for 1 h. The reaction was stopped by the addition of acetic acid and streptavidin beads, and prenylation of the peptides analyzed by a Microbeta counter (Wallac).
In Vivo Prenylation and Immunoprecipitation of ECE-1-To facilitate cellular uptake of mevalonolactone, CHO cells and the stable cell line CHO/ECE-1a were transfected with 1 g of a plasmid encoding the mevalonate transporter, pMev (ATCC), using LipofectAMINE 2000 following the manufacturer's protocol (Life Technologies, Inc.). Two days following transfection, cells were incubated with 20 M lovastatin for 1 h. The cells were then incubated with 150 Ci of [ 3 H]mevalonolactone (50 -60 Ci/mmol) (ARC) in the presence of 40 M lovastatin for 16 h. Plates were washed with cold PBS, and cells were scraped into PBS and pelleted. The cells were resuspended in 10 mM NaH 2 PO 4 buffer, pH 8.0, 1% Triton X-100, 1 mM Pefabloc, 1 g/ml pepstatin A, and 50 g/ml leupeptin, Dounce-homogenized, and passed through a 25-gauge needle to lyse cells. The cells were solubilized for 1 h at 4°C, and the lysate was precleared by incubating with protein A Staphyloccus aureus (PAS) beads (Sigma) for 1 h at 4°C. The beads were pelleted, and the precleared lysate was incubated with the monoclonal anti-ECE-1 antibody ECE-6 (a kind gift from Dr. B.-M. Loffler, F. Hoffman-La Roche, Basel, Switzerland) or with a polyclonal anti-Ras antibody (Upstate Biotechnology) overnight at 4°C. Antibody conjugates were immunoprecipitated with PAS beads and incubated for 1 h at 4°C. Beads were then washed three times in 10 mM NaH 2 PO 4 , pH 8.0, 1% Triton X-100 buffer, one time with 100 mM NaH 2 PO 4 , pH 8.0, and one time with 10 mM NaH 2 PO 4 without detergent. Beads were resuspended in Laemmli sample buffer, and samples were subjected to 4 -20% or 8% SDS-PAGE (Novex). Gels were fixed in 30% methanol, 7.5% acetic acid, soaked in Enlightening (PerkinElmer Life Sciences), dried, and exposed to Biomax MR film (Eastman Kodak Co. Plates were washed with cold PBS, and cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.5, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 10 mM EDTA, 1 mM Pefabloc, 1 g/ml pepstatin A, and 50 g/ml leupeptin). After preclearing samples with PAS, immunoprecipitation and analysis of ECE and Ha-Ras was carried out as described above. 35 S Metabolic Labeling and Immunoprecipitation of ECE-1-CHO cells were transfected with the wild type and mutant pZeoSV/flag-ECE-1a constructs using LipofectAMINE 2000 following the manufacturer's protocol (Life Technologies, Inc.). Two days after transfection, cells were incubated for 1 h in methionine-free DMEM (Life Technologies, Inc.) containing 10% dialyzed fetal bovine serum, and then labeled with Tran 35 S-label methionine (50 Ci/ml) for 5 h. Plates were washed with cold PBS, and cells were lysed in RIPA buffer. Lysates were vortexed, and a post-nuclear supernatant was prepared by centrifugation at 13,000 ϫ g. The post-nuclear supernatant was precleared by incubating with PAS beads for 1 h at 4°C. Flag-ECE-1a was immunoprecipitated with an anti-flag M2-agarose affinity gel (Sigma) overnight at 4°C. Beads were then washed three times in RIPA buffer and resuspended in Laemmli sample buffer. Samples were subjected to 8% SDS-PAGE and analyzed by autoradiography as described above. To monitor expression levels of flag-ECE-1a proteins, 10-l aliquots of the samples were subjected to SDS-PAGE, transferred to a nitrocellulose filter (Novex), immunoblotted with a polyclonal anti-ECE antibody, and detected by ECL following the manufacturer's protocol (Amersham Pharmacia Biotech).
Pulse-Chase Labeling-CHO cells were transfected with the WT and mutant pZeoSV/flag-ECE-1a constructs as described above. Two days following transfection, cells were incubated for 1 h in methionine-free DMEM (Life Technologies, Inc.) containing 10% dialyzed fetal bovine serum, and labeled with Tran 35 S-label (100 Ci/ml) for 20 min at 37°C. Plates were washed twice with PBS and chased in DMEM/F-12 with 10% fetal bovine serum containing excess cold methionine and cysteine at 3 mM each. At the appropriate time points, plates were washed with PBS and cells were lysed with RIPA buffer. A control sample for each was continuously labeled with 50 Ci/ml Tran 35 S-label for 4 h. Cell lysates were prepared with an anti-flag M2-agarose affinity gel, subjected to 8% SDS-PAGE, and analyzed by autoradiography as described above.
PNGase F and Endo H Treatment-Wild type flag-ECE-1a and the ECE-1a mutant, SEVW, were immunoprecipitated from transfected CHO cells 48 h after transfection using an anti-flag M2-agarose affinity gel (Sigma) as described above. The immune complexes were heated in denaturing buffer (0.5% SDS, 1% ␤-mercaptoethanol) at 100°C for 10 min. Samples were then incubated with and without 3,000 units of endo H or 3,000 units of PNGase F overnight at 37°C following the manufacturer's protocol (New England Biolabs). The reaction was stopped by the addition of sample buffer, and samples were run on 8% Novex SDS-PAGE, transferred to nitrocellulose, immunoblotted with a poly-clonal anti-ECE antibody, and detected by ECL following the manufacturer's protocol (Amersham Pharmacia Biotech).
ECE-1 Membrane Preparation-CHO cells were transfected with the wild type and mutant pZeoSV/flag-ECE-1a constructs as described above. Three days after transfection, cells were washed and scraped into PBS, collected by centrifugation at 2,000 ϫ g, resuspended in homogenization buffer (50 mM Tris-HCl, pH 7.4, 0.1 M NaCl, 0.1 M sucrose, 1 mM Pefabloc, 1 g/ml pepstatin, and 50 g/ml leupeptin), Dounce homogenized, and centrifuged at 100,000 ϫ g for 1 h. The pellet was resuspended in homogenization buffer and again centrifuged at 100,000 ϫ g for 1 h. The pellet was solubilized in solubilization buffer (50 mM Tris-HCl, pH 7.4, 0.1 M NaCl, 0.25% C 12 E 10 , 1 mM Pefabloc, 1 g/ml pepstatin A, and 50 g/ml leupeptin) at 4°C overnight. Unsolubilized membranes were centrifuged at 100,000 ϫ g for 1 h. Aliquots were subsequently used for activity assays.
ECE-1 Activity Assays-Solubilized membrane fractions were diluted with 20 mM Tris-HCl, pH 7.4, 0.1% C 12 E 10 , and incubated with 0.1 M big ET-1-(1-38) in reaction buffer (100 mM MES-KOH, pH 6.8, 0.1% C 12 E 10, 1 mM Pefabloc, 50 g/ml leupeptin, and 1 g/ml pepstatin A). Reactions were carried out at 37°C for 1 h and stopped by adding EDTA to give a final concentration of 5 mM. The final mixture was then analyzed for the amount of ET-1 by enzyme-linked immunosorbent assay following the manufacturer's protocol (Amersham Pharmacia Biotech).

In Vitro Prenylation of the Hexapeptide HKCEVW-The
NEP/ECE subfamily of metalloproteases has considerable amino acid sequence homology, and all the members terminate in a unique tetrapeptide CXAW. Table I illustrates the CXAW sequence homology between the seven known mammalian family members as well as a metalloprotease found in Caenorhabditis elegans, 2 which has 32% amino acid sequence identity to ECE-1. The conservation of this sequence indicated that it might play an important role in these proteins, possibly as a novel prenylation site. To test this, we performed in vitro prenylation assays using as substrate a biotinylated hexapeptide corresponding to the native sequence, HKCEVW. Briefly, biotinylated hexapeptides were incubated with CHO cell lysate, as a source of prenyltransferase enzymes, and either 3 [H]farnesyl pyrophosphate or 3 [H]geranylgeranyl pyrophosphate. Hexapeptides containing known prenylation sequences were included as positive controls (TKCVIM and TKCVLS for farnesyl transferase, and TKCVIL for geranylgeranyl transferase). A small percentage of the HKCEVW peptide (less than 5% relative to the control hexapeptide) was radiolabeled when incubated with 3 [H]geranylgeranyl pyrophosphate (data not shown). The weak prenylation suggested that the sequence can serve as a prenyltransferase substrate. We therefore assessed whether ECE-1 could be prenylated in vivo.
In Vivo Labeling of ECE-1a with Isoprenyl Precursors-CHO cells stably expressing the ECE-1 isoform, ECE-1a (CHO/ECE-1a), were metabolically labeled with the isoprenoid precursor [ 3 H]mevalonolactone. To facilitate the cellular uptake of the mevalonolactone, a vector encoding the mevalonate trans-porter, pMev (16), was transfected into the cells. As a positive control for prenylation, a plasmid encoding Ha-Ras was transiently co-transfected with the pMev plasmid into CHO cells. In addition, CHO cells transfected with pMev alone served as a negative control. Approximately 48 h after transfection, cells were preincubated with lovastatin at 40 M (a concentration known to inhibit hydroxymethylglutaryl-CoA reductase; Ref. 17) to deplete the intracellular pool of mevalonate and its metabolites. Cells were then incubated with [ 3 H]mevalonolactone for 16 h in the presence of lovastatin. As described under "Experimental Procedures," cell lysates were prepared, and ECE-1a and Ha-Ras were immunoprecipitated and then analyzed by SDS-PAGE and autoradiography. Fig. 1A shows the autoradiograph of the labeled immunoprecipitates. Labeling and prenylation of Ha-Ras was evident, as shown by a single band of the approximate molecular mass of 23 kDa (Fig. 1A, lane 2). By contrast, labeling of ECE-1a, which has a molecular mass of ϳ130 kDa when analyzed by SDS-PAGE, was not observed (Fig. 1A, lane 3). Exposure of the gel for 40 days also revealed no labeling. Western blot analysis of the immunoprecipitate complex using a polyclonal antibody to ECE-1 confirmed the presence of ECE-1a (Fig. 1B, lane 2). These results indicated that ECE-1a cannot be metabolically 2 GenBank™/EMBL Data Bank accession no. AAC46806.1.  labeled in vivo by a mevalonolactone derivative.
Four isoforms of human ECE-1 (1a, 1b, 1c, and 1d) have been cloned. All of these are encoded by one gene and share a common carboxyl-terminal portion, but each is expressed from one of four distinct promoters, which regulate expression of four unique amino termini (18 -21). Although the carboxyl terminus containing the CEVW sequence is identical in each of the isoforms, the amino-terminal sequences may be responsible for differences in tissue specificity of expression as well as differences in subcellular localization. The different isoforms are localized either to the cell surface or an intracellular compartment, possibly the Golgi (5,19,(21)(22)(23)(24)(25)(26).
In order to eliminate the possibility that prenylation of ECE-1 is isoform-specific, HUVECs, which are known to express messages encoding each ECE-1 isoform (19,21), were used for labeling. Cells were metabolically labeled with [ 3 H]farnesol or [ 3 H]geranylgeraniol as precursors for farnesyl pyrophosphate and geranylgeranyl pyrophosphate, respectively (27,28). Unlike mevalonolactone, the alcohol derivatives are readily taken up by the cells, providing a method to metabolically label the primary HUVECs without the need to transfect the pMev transporter. Cells were labeled overnight with [ 3 H]farnesol or [ 3 H]geranylgeraniol in the presence of lovastatin, lysed, ECE-1-immunoprecipitated, and analyzed by SDS-PAGE and autoradiography as described under "Experimental Procedures." Immunoprecipitation of endogenous Ha-Ras was used as a positive control for labeling and prenylation from HUVECs labeled with [ 3 H]farnesol. As shown in Fig. 1C, only Ha-Ras appears to be prenylated migrating as a single band of ϳ23 kDa (Fig. 1C, lane 2). ECE-1 did not appear to be labeled by either [ 3 H]farnesol (Fig. 1C, lane 3) or [ 3 H]geranylgeraniol (data not shown), indicating that it may not be modified by prenylation.

Effect of CXAW Motif Mutations on ECE-1a Protein
Expression-Site-directed mutagenesis was used to assess the role of the CXAW motif and its effect on protein modification and enzyme activity. To simplify discussion of the mutants, the nomenclature used refers to the amino acid substitution in the CXAW sequence as summarized in Table II. The ECE-1a mutant, SEVW, refers to the replacement of Cys 755 of the CXAW motif with serine. ⌬CEVW refers to deletion of the CEVW sequence by introducing a stop codon at Cys 755 . The mutants CEVL and CEVM refer to mutations constructed to create possible prenylation sites where the tryptophan residue has been replaced by a preferred residue, either leucine for geranylgeranylation (CEVL mutant) or methionine for farnesylation (CEVM mutant). To further evaluate the role of the conserved Trp 758 , the CEV mutant was constructed in which the tryptophan residue was deleted. Conservative substitutions of Trp 758 were also produced with phenylalanine (CEVF) and tyrosine (CEVY). Finally, the aromatic residue was removed by replacing Trp 758 with alanine (CEVA).
A construct of ECE-1a containing a flag epitope tag at the amino terminus was used for the mutagenesis. Constructs were subcloned into the mammalian expression vector pZeoSV and transfected into CHO cells. Two days following transfection, ECE-1a expression was determined by both Western blot analysis of cell lysates using a polyclonal antibody to ECE-1 (data not shown), and by metabolic labeling of the cells with [ 35 S]methionine and immunoprecipitation of ECE-1a (Fig. 2). Cells were transiently transfected with the WT and mutant flag-ECE-1a constructs, labeled with [ 35 S]methionine for 5 h, cell lysates prepared, and flag-ECE-1a immunoprecipitated using an anti-flag M2-agarose affinity gel. The immunoprecipitates were then analyzed by SDS-PAGE and autoradiography.
With expression of WT, two 35 S-labeled bands at 130 and 110 kDa were immunoprecipitated ( Fig. 2A, lane 1). The identity of the two bands as ECE-1a was confirmed by Western blot analysis (data not shown). The ECE-1a mutants CEVL, CEVM, CEVF, and CEVY ( Fig. 2A, lanes 3, 4, 7, and 8) all showed near WT expression levels of both bands, but the CEV and CEVA mutants ( Fig. 2A, lanes 5 and 9) showed decreased levels of the upper 130-kDa band. In contrast, the two mutants in which the cysteine was replaced, SEVW and ⌬CEVW, showed only the lower molecular weight form of the enzyme ( Fig. 2A, lanes 2  and 6).
We wanted to determine whether the presence of the 130-kDa band was due to post-translational modification. In addition, we wanted to determine whether the 110-kDa band represented an immature form of the protein rather than a degradation product. The recent crystal structure of human NEP at 2.1-Å resolution shows that the cysteine of the CXAW motif in NEP, Cys 746 , is involved in an intramolecular disulfide bond with Cys 620 (29). This result suggests that the homologous cysteine of the CXAW motif of ECE-1a, Cys 755 , may also be involved in an intramolecular disulfide bond important for proper folding of the enzyme. To test whether the 110-and 130-kDa forms of ECE-1a represent differences in protein folding or post-translational modification, two conserved cysteine  residues other than Cys 755 of ECE-1a, Cys 632 and Cys 743 , were changed to serine creating two mutants (C632S and C743S) in addition to the SEVW (C755S) mutant. By sequence alignment with NEP, Cys 632 in ECE-1a corresponds to the cysteine paired with Cys 755 . Cys 743 is a conserved cysteine upstream of Cys 755 and is believed to be disulfide-bonded with Cys 110 . Metabolic labeling and immunoprecipitation of ECE-1a from these mutants also resulted in a single band of ϳ110 kDa (Fig. 2B, lanes  2 and 3). These results suggest strongly that the conserved cysteine residues in ECE-1a are necessary for protein folding or processing.

FIG. 2. Effect of mutations of ECE-1 CXAW motif on protein expression in CHO cells as assessed by immunoprecipitation from 35 S-labeled cells. CHO cells were transiently transfected with
Pulse-Chase Labeling of ECE-1a-To further determine whether the conserved cysteine and tryptophan residues of the CEVW motif in ECE-1a are important for proper processing of the enzyme in the secretory pathway, we performed pulsechase labeling of the WT enzyme and the ECE-1a mutants SEVW and CEV. Fig. 3 shows the results of the pulse-chase labeling in CHO cells transfected with WT, SEVW, or CEV. Briefly, cells were pulsed with 500 Ci of [ 35 S]methionine for 20 min and then chased for 0, 15, 30, 45, 60, 120, and 240 min in the presence of excess amounts of unlabeled methionine and cysteine as described under "Experimental Procedures." A control sample for each was continuously labeled with 250 Ci of [ 35 S]methionine for 4 h. Cells were lysed at the indicated time points and immunoprecipitated with an anti-flag M2-agarose affinity gel. After the pulse, a single band of 110 kDa was observed in each cell line (Fig. 3, A-C, lane 1). During the chase the WT enzyme undergoes post-translational modification, resulting in a mature form of the enzyme of 130 kDa by 30 min (Fig. 3A, lane 3). However, labeling of the SEVW mutant results in only the 110-kDa form of the enzyme even after 4 h of chase or in the control sample with continuous labeling (Fig.  3B). This result indicates that Cys 755 of the CXAW motif is critical for proper processing and maturation of ECE. Additionally, post-translational processing of the ECE-1a mutant without the conserved Trp 758 , CEV, appears to be slower than for the WT, with the mature form of the enzyme not appearing until 120 min (Fig. 3C).
Endoglycosidase Treatment of ECE-1a-Treatment of WT and the SEVW mutant with glycosidases confirmed that the 110-kDa band is an immature form of the enzyme (Fig. 4). Treatment of glycoproteins with endo H removes immature but not medial Golgi-processed N-linked oligosaccharide side chains (30). Immunoprecipitated WT was treated overnight with endo H, and mobility of ECE-1a was analyzed by SDS-PAGE and immunoblot using a polyclonal anti-ECE-1 antibody. The lower 110-kDa band was endo H-sensitive and shifted to ϳ80 kDa, consistent with the predicted molecular mass of ECE-1a (Fig. 4, lanes 1 and 2). The upper 130-kDa band was endo H-resistant, indicating that the two bands represent an immature endoplasmic reticulum-associated form and a mature form of the enzyme. Treatment of the mutant protein SEVW with endo H also resulted in a molecular mass shift from 110 to 80 kDa (Fig. 4, lanes 3 and 4). PNGase F, an enzyme that removes all N-linked oligosaccharide side chains, was used as a positive control for deglycosylation of ECE-1a. Treatment of both the WT and the SEVW mutant with PNGase F overnight completely deglycosylated both the mature and immature forms of ECE-1a and yielded one band of ϳ80 kDa (Fig. 4, lanes 5-8). These results further demonstrate that Cys 755 is critical for proper processing of the enzyme through the secretory pathway.
ECE-1a Activities on Purified Membranes from ECE-1a CXAW Motif Mutants-Enzyme activity assays were performed on the flag-ECE-1a mutants to determine what effect mutations of the CEVW sequence have on enzyme activity. ECE-1a activities were measured on purified membranes from CHO cells expressing WT and mutant flag-ECE-1a proteins using big ET-1 as substrate as described under "Experimental Procedures." The amount of ECE-1a was normalized for the 130-kDa mature form of the protein by Western blot. Fig. 5 shows the relative ECE-1a activities of the WT and the CXAW mutants. Activities are expressed relative to the WT enzyme, which had a specific activity of 1.9 ϫ 10 Ϫ4 mol/min/mg of the total membrane fraction defined as 100%. The CEVF mutant yielded a fully active enzyme, and CEVY had ϳ50% activity of the WT enzyme. Surprisingly, deletion of the tryptophan residue, mutant CEV, or replacement with alanine, CEVA, resulted in more than a 95% loss in activity. The mutants CEVM

FIG. 3. Effect of pulse-chase labeling of WT and ECE-1 mutants. CHO cells were transiently transfected with pZeo/flag-ECE-1a
constructs, WT, and the mutants SEVW and CEV. Two days after transfection, cells were incubated in methionine-free medium for 1 h, pulsed with Tran 35 S-label for 20 min, and chased for various times with DME/F-12 medium containing excess methionine and cysteine. Cells were lysed, flag-ECE-1a immunoprecipitated using an anti-flag M2agarose affinity gel, and samples subjected to SDS-PAGE and autoradiography performed as described under "Experimental Procedures." Pulse-chase labeling for WT is shown in panel A, SEVW in panel B, and CEV in panel C .   FIG. 4. Effect of glycosidase treatment of flag-ECE-1a. Membranes were prepared from CHO cells transiently transfected with pZeo/flag-ECE-1a constructs, WT and SEVW. Flag-ECE-1a was immunoprecipitated using an anti-flag M2-agarose affinity gel, denatured, and treated with endo H (lanes 1-4) or PNGase F (lanes 5-8) as described under "Experimental Procedures." Samples were subjected to 8% SDS-PAGE, transferred to nitrocellulose, and immunoblotted using a polyclonal anti-ECE-1 antibody. and CEVL had ϳ25% and 15% activity of WT levels, respectively. These two mutant proteins express both the 110-and 130-kDa bands at comparable levels to those of WT. These results suggest that an aromatic residue at the terminal position of the CXAW motif is essential for activity, and that a hydrophobic residue only partially restores activity.
To determine whether these mutations affected binding to the active site, K i determinations for two competitive ECE-1a inhibitors, phosphoramidon and PD 069185 (31), were performed. As shown in Table III, the K i values for the mutants were not significantly different from WT for either compound. Only one mutant, CEVY, demonstrated a lower K i than the other mutants for phosphoramidon, and the K i for PD 069185 was reduced ϳ3-fold for mutant CEVL, suggesting a small increase in affinity for these compounds. Further kinetic work using purified enzyme will be needed to better characterize the function of the tryptophan residue on enzyme activity. DISCUSSION Currently there is no information concerning the role of the conserved terminal tetrapeptide CXAW sequence of the NEP/ ECE family of metalloproteases. The highly conserved sequence of cysteine-charged (E/R) or uncharged polar residue (Q/S)-hydrophobic residue (V/I/L)-tryptophan in this family of metalloproteases led us to investigate the function of this sequence. The sequence resembles a CAAX prenylation motif, making it a possible candidate for a novel prenylation signal. An aromatic residue in the carboxyl-terminal position of the CAAX motif is not a common feature, although the sequence CCIF, which is found at the carboxyl terminus of bovine brain, Ras-related small G protein, G25K, is modified by all-transgeranylgeranyl-Cys methyl ester (32). In addition, substrate specificity studies of farnesyltransferase modification of the precursor of the yeast a-mating factor demonstrated that the sequence CVIW can be farnesylated, although to a much lesser degree than the wild type sequence CVIA (33). We wanted to determine whether one function of the CXAW motif is to serve as a prenylation signal for the NEP/ECE family that might involve a novel prenyltransferase.
The wild type sequence HKCEVW was weakly geranylgeranylated in vitro using the biotinylated hexapeptide as substrate for prenyltransferases. However, ECE-1 was not labeled in vivo by isoprenoid precursors in either of two cell lines tested: CHO cells stably expressing ECE-1a or HUVECs expressing endogenous isoforms of ECE-1. ECE-1 isoforms 1a, 1b, and 1c were shown to be palmitoylated (34), demonstrating that the enzyme undergoes post-translational modifications other than glycosylation. Nonetheless from the data presented in this study, it appears that ECE-1 is not prenylated.
In addition to investigating prenylation of ECE-1, we wished to further characterize the role of the CXAW motif through mutagenic analysis. Site-directed mutagenesis of the CEVW sequence demonstrated that the conserved Cys 755 is critical for proper folding and maturation of the enzyme, possibly through intramolecular disulfide bond formation. Substitution of Cys 755 with serine (mutant SEVW) or removal of the tetrapeptide by introduction of a stop codon at Cys 755 (mutant ⌬CEVW) resulted in a misfolded protein that possibly remained in an early secretory compartment, presumably the endoplasmic reticulum, as demonstrated by both pulse-chase labeling and endo H glycosidase treatment. The recent x-ray crystal structure of human NEP complexed with the inhibitor phosphoramidon shows that NEP contains 12 cysteine residues, all involved in disulfide bridges (29). Based on the NEP crystal structure and sequence alignment between NEP and ECE-1, Cys 755 of ECE-1a is predicted to be disulfide-bonded with Cys 632 . Sitedirected mutagenesis of Cys 632 as well as Cys 743 , the cysteine immediately upstream of Cys 755 , resulted in an immature form of the enzyme only. These results indicate that the conserved cysteines in ECE-1a, including Cys 755 of the CEVW motif, are important for proper folding of the protein possibly through disulfide bonds making Cys 755 unavailable for prenylation or modification.
Mutations of the conserved Trp of the CEVW sequence, Trp 758 , suggest that an aromatic or hydrophobic residue as the terminal amino acid of the enzyme is important for protein processing and for enzyme activity. The conservative mutation CEVF had little effect on enzyme activity, whereas the CEVY mutation reduced activity by 50%. A hydrophobic residue only partially substituted for Trp with the CEVL and CEVM mutations, resulting in 15% and 25% enzyme activity compared with wild type, respectively. These effects on activity do not appear to be due to changes in substrate affinity as K i determinations for two competitive inhibitors for ECE-1, phosphoramidon and PD 069185, did not show significant differences. Removal of FIG. 5. Effect of mutations on ECE-1a activity on isolated membranes. CHO cells were transiently transfected with pZeo/flag-ECE-1a constructs. Three days after transfection, cells were harvested, Dounce homogenized, and solubilized with 0.25% Lubrol as described under "Experimental Procedures." Aliquots were incubated at 37°C for 1 h with 0.1 M big ET-1 in reaction buffer containing 100 mM MES-KOH, pH 6.8, 0.25% C 12 E 10 , and protease inhibitors. The amount of ET-1 generated was analyzed by enzyme-linked immunosorbent assay. Trp 758 in the CEV mutant or replacement with alanine, mutant CEVA, resulted in a completely inactive enzyme but appeared to affect protein processing and stability. These data suggest that an aromatic residue as the terminal amino acid of the protein is crucial for enzyme activity.
The recently released coordinates of human NEP reveal that the corresponding tryptophan, Trp 749 , resides in a hydrophobic pocket, which may be necessary to anchor the carboxyl terminus. Proper orientation of the carboxyl terminus would be critical to ensure that Cys 746 is accessible for disulfide bond formation. In addition, Trp 749 appears to hydrogen-bond with the surrounding helices containing the histidine residues that complex with the Zn 2ϩ ion. An aromatic or large hydrophobic residue may be required as the terminal amino acid to fit into the hydrophobic pocket and stabilize the surrounding helices. Trp 749 may therefore play an important role in the organization of the active site and in stabilizing the protein. 3 It is likely that Trp 758 of ECE-1a plays a similar structural role for the enzyme and that the effects observed on enzyme maturation and activity are the consequence of conformational changes rather than direct effects at the active site. This is consistent with the mutagenesis data, where the largest effects on protein maturation and activity were observed with the removal of the aromatic or hydrophobic residue. However, although NEP and ECE-1 share significant homology, it is important to note that the two enzymes differ in substrate specificity, sensitivity to inhibitors, pH optimum, and tissue distribution. ECE-1 also functions as a dimer, whereas NEP does not. Such distinctions indicate that considerable structural and functional differences may exist between these proteins. Previous mutagenesis studies have demonstrated that Arg 747 of NEP in the conserved CXAW sequence, CRVW, is critical for substrate and inhibitor interaction (35). From the structure, Arg 747 forms a salt bridge with the carboxyl terminus. The corresponding residue in ECE-1 and ECE-2 is the glutamic acid residue of the conserved CEVW sequence, Glu 752 in ECE-1a. Mutation of this residue to glutamine, aspartic acid, or arginine did not result in significant changes in kinetic parameters, indicating that Glu 752 does not play a critical role in substrate binding or catalysis (36). These data suggest that there may be structural differences between the two enzymes. Mutations at Trp 758 of ECE-1a may exert their effects through conformational changes or the residue may be interacting with the active site. Further kinetic work using purified ECE-1a will be needed to better characterize the function of the tryptophan residue on enzyme activity.
We have demonstrated that the conserved CXAW sequence of ECE-1a does not appear to be a prenylation signal, but that the conserved Cys 755 and Trp 758 residues are both critical for proper folding and maturation of the enzyme through the secretory pathway. We have also provided evidence for the importance of the terminal amino acid of ECE-1a, Trp 758 , in enzyme activity. It is clear from this study that the conserved CXAW motif plays an important role in enzyme function. This is the first work that identifies residues outside of the zincbinding consensus sequence and catalytic site that influence activity, possibly giving new evidence for residues involved in formation or stability of the active site. Further work will be required to define more completely the function of these residues in the control of enzyme activity.