By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
O. Valdenaire, Inserm U460, UFR Xavier Bichat, 16 rue Henri Huchard, 75018 Paris, France. Tel: + 33 1 4485 6153, Fax: + 33 1 4485 6157, E-mail: firstname.lastname@example.org
Human endothelin-converting enzyme (ECE-1) has been shown to exist as three isoforms (ECE-1a, ECE-1b and ECE-1c) diverging in their N-terminal sequence and displaying different patterns of subcellular localization. We report here the cloning of ECE-1d, a novel isoform of 767 amino acids, which is generated from the same gene via the existence of an additional promoter located upstream from the third exon of the ECE-1 gene. ECE-1d converting activity is comparable to that of the other three isoenzymes. In contrast to ECE-1b, ECE-1d is expressed at the cell surface, although less strongly than ECE-1a. We have also shown, by identifying ECE-1b and ECE-1d in rat, that the ECE-1 diversity is conserved between human and rodent, suggesting its physiological relevance. The mRNA levels of the four isoforms were assessed in the two species in various cell types, revealing some differences. In particular, the ECE-1a isoform, strongly expressed at the plasma membrane, was found to be highly expressed in primary cultures of endothelial cells but absent from primary cultures of smooth muscle cells.
primary cultures of human umbilical vein endothelial cells
primary cultures of human umbilical vein smooth muscle cells
Endothelin (ET) consists of a family of three isopeptides, termed ET-1, ET-2 and ET-3, which are derived from distinct genes [1, 2]. The three ETs, of which ET-1 is the most abundant, mediate their various effects, including for example vasomotion or cell proliferation, through the use of two G protein-coupled receptors, called ETA and ETB[3, 4]. ET-1 is especially known for provoking a very potent and long-lasting vasoconstriction, which designates the ET system as a choice target for pharmaceutical intervention in the field of many cardiovascular pathologies. In addition, the crucial involvement of this peptidergic system in embryogenesis, particularly with respect to the development of neural crest-derived tissues, was recently demonstrated by the targeted disruption of the genes of ET-1, ET-3, ETA and ETB[5–8]. As a consequence, mutations in the genes of ET-3 and ETB are suspected to explain some cases of familial Hirschprung disease, characterized by a lack of neural crest-derived myenteric neurons [9, 10].
The precursors of ETs, called preproETs, are polypeptides of about 200 amino acid residues. After removal of their signal peptide, they are processed by dibasic pair-specific enzymatic activity  to form the inactive big ETs (38–41 residues long). The second part of the preproET transformation is more specific, as it consists of the cleavage of a Trp-Val (Trp-Ile for big ET-3) bond. The enzyme responsible for this key step has been called endothelin-converting enzyme (ECE)  and was identified in 1994 as ECE-1 [12, 13], a zinc metallopeptidase related to neutral endopeptidase  and to the Kell blood group antigen . The existence of other ECEs is regularly questioned, especially when the relatively low efficiency of ECE-1 with respect to the cleavage of big ET-2 or big ET-3 is considered. Thus ECE-2, closely related to ECE-1, was shown to process big ETs, although with a low pH optimum . The recently reported targeted disruption of the ECE-1 gene, however, indicates that ECE-1 is, if not the only, at least the main enzyme responsible for the transformation of big ETs into ETs . Indeed the resulting mice present both phenotypes of ET-1 and ET-3 knockout mice. In addition, one case of Hirschprung disease has been reported to be potentially associated with an ECE-1 inactivating mutation .
This key position of ECE-1 in the metabolism of ETs, along with the recently reported induction of ECE-1 expression in atherosclerotic proliferating smooth muscle cells , point at a beneficial role for ECE-1 inhibitors in a number of cardiovascular pathologies. ECE-1 is a type-II integral membrane protein, with a large C-terminal domain containing the enzymatic active site and a short cytoplasmic tail, and has been shown to be localized on the cell surface as well as intracellularly [20, 21]. This last feature, which could have some impact on the pharmacology of ECE-1 inhibitors, might be related to the existence of several ECE-1 isoforms. We have shown indeed that variability of the ECE-1 short N-terminal domain was responsible for the existence of different isoforms (called ECE-1a, ECE-1b and ECE-1c), which cleave big ETs with identical efficiency but differ with respect to their subcellular localization . The existence of these isoforms in the human is the direct consequence of the presence of independent promoters in the ECE-1 gene [22–24]. We present here the characterization of a fourth and novel human ECE-1 isoform, ECE-1d, and demonstrate that it is derived from a fourth promoter, located upstream from the third exon of the ECE-1 gene. ECE-1d converting activity and subcellular localization were analyzed and compared to those of the other human ECE-1 isoforms. The mRNA levels of the four isoenzymes were also assessed in various cell and tissue types. In addition, we have analyzed the situation in rat, where only two isoforms (ECE-1α and ECE-1β, homologous to human ECE-1c and ECE-1a, respectively) were known so far [25, 26]. The conservation of the four ECE-1 isoforms in both species underlines the potential importance of this diversity.
Chinese hamster ovary (CHO-K1) cells were grown in Ham F12 medium supplemented with 10% fetal bovine serum. Primary cultures of the following cells: rat aortic endothelial (RAEC), rat smooth muscle (RSMC) and rat fibroblastic (RFC), were isolated and cultured as described previously . Primary cultures of human umbilical vein endothelial cells (HUVEC) were isolated and cultured as described . Primary cultures of human umbilical vein smooth muscle cells (HSMC) were obtained as described . Transfections of CHO-K1 cells and HSMC were performed using Fugene 6 (Boehringer) according to the manufacturer’s recommendations. Transiently transfected cells were harvested 48 h after transfection. Stable CHO-K1 cell lines were selected using G418 (Life Technologies) at a concentration of 800 µg·mL−1.
The 5′ regions of rat ECE-1 mRNAs were determined using a method called SLIC . In brief, first-strand cDNA was synthesized using RSMC total RNA (5 µg) with a specific ECE-1 antisense primer (5′-GTCAGACATACCGGAGGCGTTC-3′). The purified single-stranded cDNAs were ligated at their 3′ ends to a 3′-NH2-blocked oligonucleotide (5′-NCTGCTCCTCTCGCTACCTGCTCACTCTGCGTGAC-3′) and used as a template for nested PCR with two successive sets of primers. Each set was composed of an ECE-1 specific primer (5′-GCGTCACCAGAACCACCAGCCG-3′ and 5′-AGAGTGAGTCCACCAGATCCTC-3′) and of a primer complementary to the 3′-blocked oligonucleotide (5′-GTCACGCAGAGTGAGCAGGTAG-3′ and 5′-CAGAGTGAGCAGGTAGCGAGAGGAG-3′). Amplification products of the second PCR were subcloned and sequenced. The sequence of human ECE-1d was deduced from a known region of the human ECE-1 gene which presented a very strong homology with the sequence of rat ECE-1d. The sequence of the human ECE-1d promoter was deduced from previous works  and the sequence of the rat promoter was obtained by performing PCR on genomic DNA using sense and antisense primers deduced from rat ECE-1b and ECE-1d specific sequences, respectively.
The human genomic DNA region between the ECE-1d start codon and the ECE-1b start codon (nucleotides h22/h236 on Fig. 1 A) was amplified by PCR and subcloned into the pGL3-Basic vector (Promega) upstream from the luciferase gene. The resulting plasmid was transiently transfected in CHO-K1 cells and HSMC. The pGL3-Basic (no promoter) and pGL3-Control (SV40 promoter) vectors were used as negative and positive controls, respectively. The pCH110 plasmid (Amersham Pharmacia), containing the β-galactosidase gene under the control of a SV40 promoter, was used as an internal standard. Luciferase activity was measured with the Luciferase Assay System (Promega), and was normalized with respect to β-galactosidase activity.
Total RNA was extracted with Trizol® (Life Technologies) from various human and rat cell types and tissues. Five micrograms of each RNA were reverse transcribed using M-MLV reverse transcriptase (Life Technologies) and an ECE-1 antisense specific primer (5′-CCACAGGCGTAGCTGAAGAAG-3′, common to both rat and human ECE-1 sequences) according to the manufacturer’s recommendations. The reaction mixture was completed to 50 µL with water, and 25 (30 for rat ECE-1a) cycles of PCR were then performed on 2 µL of cDNA with antisense primers common to all human (5′-GAAGAAGTCATGGCAGGGGTC-3′) or rat (5′-CTGAAGAAGTCCTGGCAGGGGTC-3′) isoforms, and sense primers specific for each species isoform (human: ECE-1a, 5′-CAGCCCTGATGCCTCTCCAG-3′; ECE-1b, 5′-CCCTGCTGTCGGCGCTGGGG-3′; ECE-1c, 5′-CGGAGCACGCGAGCTATGATG-3′; ECE-1d, 5′-ATGGAGGCGCTGAGGGAGTCC-3′; rat: ECE-1a, 5′-CCTGGTCTCACGGTCTCGCTG-3′; ECE-1b, 5′-GCTGGCCGCTTGGGGATG-3′; ECE-1c, 5′-GAGCCTTAGCGGGAGGT-GCATC-3′; ECE-1d, 5′-ATGGAGACGCTGAGGGAGTCC-3′). All PCR primers were chosen so that the genomic counterparts of amplified cDNAs encompassed at least one intron. Ten microliters of the 25 µL amplification mixture were electrophoresed on a 1% agarose gel.
Western blot analysis
The rabbit antiserum 473-17-A (called anti-(ECE-1) in this report), raised against a synthetic peptide located in the intraluminal/extracellular part of ECE-1, has been previously characterized . Cells were scraped off on ice into a lysis buffer (0.01 m NaCl/Pi, pH 7.4, containing 0.5% sodium deoxycholate and 0.5% Nonidet P40) in the presence of 200 µm phenylmethanesulfonyl fluoride (PMSF), 1 µm leupeptin and 1 µm pepstatin, then centrifuged for 10 min at 15 000 g at 4°C. Supernatants were solubilized with SDS/PAGE sample buffer. Protein concentrations were measured with the BCA Protein Assay Reagent Kit (Pierce). Proteins (25 µg) were separated by SDS/PAGE and then electrotransferred to nitrocellulose sheets . After incubation with the anti-(ECE-1) antibody, immunoreactive proteins were revealed with alkaline phosphatase-conjugated sheep anti-(rabbit IgG) antibodies (Promega).
Membranes of CHO nonexpressing or stably expressing human ECE-1a and ECE-1d were prepared as described previously . Five micrograms of solubilized membranes were incubated in 100 µL of buffer A (50 mm 2-(N-morpholino)-ethanesulfonic acid (Mes), pH 6.8, 150 mm NaCl, 0.1 µm ZnCl2, 0.1% BSA, 2 µm leupeptin and 200 µm PMSF) with big ET-1 for 30 min at 37 °C. The reaction was stopped by adding 100 µL of 5 mm EDTA and ET-1 production was measured using a commercial enzyme immunoassay kit (Cayman Chemical Co.). For determination of Km values, human big ET-1 was used at concentrations of 50, 100, 330, 500 and 667 nm, and 1, 2 and 5 µm. Each point was performed in duplicate.
Immunofluorescence performed on fixed cells
CHO-K1 cells expressing human ECE-1d were fixed in periodate/lysine/paraformaldehyde for 2 h at room temperature. They were permeabilized with 0.005% saponin, immunocytochemically stained as described previously  using anti-(ECE-1) and goat immunoglobulins against rabbit IgG labeled with tetramethylrhodamine isothiocyanate (Biosys) and examined under a Leitz epifluorescence microscope.
Electron microscope immunoperoxidase performed on fixed cells
The immunoperoxidase procedure was performed using a pre-embedding approach on CHO-ECE-1d cells grown in the culture dishes as described previously . Briefly, cells were fixed with periodate/lysine/paraformaldehyde and permeabilized with 0.005% saponin. After incubation with anti-ECE-1, cells were incubated with goat Fab against rabbit IgG conjugated with peroxidase (Biosys). After postfixation in 1% glutaraldehyde, detection of peroxidase activity and postfixation in 1% osmium tetroxide, cells were embedded in situ in Epon . Selected areas of immunoreactive cells were sectioned and ultrathin sections were examined under the electron microscope without further staining.
Immunocytochemical procedures performed on living cells
Living CHO-ECE-1d cells were preincubated for 10 min at 37°C in culture medium without serum in the presence or absence of 5 mm dithiothreitol and incubated for 1 h at 4°C in the same medium with anti-(ECE-1). After several washes at 4°C, they were incubated for 1 h at 4°C either with goat immunoglobulins against rabbit IgG, labeled with tetramethylrhodamine isothiocyanate (for immunofluorescence), or with goat Fab against rabbit IgG conjugated with peroxidase (for electron microscope immunoperoxidase). After washing at 4°C, cells were separately treated: for immunofluorescence, cells were fixed with 4% paraformaldehyde and mounted with mowiol (Hoechst); for electron microscope immunoperoxidase, cells were fixed in 1% glutaraldehyde before detection of immunoperoxidase activity, postfixation in 1% osmium tetroxide and embedding in situ in Epon.
Cloning of human and rat ECE-1 isoforms
To check the existence of the rat ECE-1b isoform, a PCR-based methodology named SLIC  was employed to clone the 5′ sequences of rat ECE-1 mRNAs from RSMC. An ECE-1c specific PCR was performed on the subcloned cDNA extremities that were obtained, and the nonresponding clones were sequenced. Most of these clones revealed a sequence as yet unknown, identical to ECE-1c downstream from its second ATG codon, but totally divergent upstream from this point which corresponds in human to the divergence between ECE-1b and ECE-1c sequences ( Fig. 2B). This new rat sequence could not be aligned with the human ECE-1b sequence, but in contrast displayed a very high homology to a short part of the human ECE-1 gene, localized immediately upstream from the third exon (exon 2, Fig. 2C) and so far considered as intronic . As this novel sequence contained a start codon 42 nucleotides upstream from its divergence with ECE-1c, we concluded that it was encoding a fourth ECE-1 isoform, termed ECE-1d and predicted to be 767 amino-acid residues long ( Fig. 2A). The existence of both human and rat ECE-1d mRNAs was checked by RT-PCR, which revealed strong signals in primary cultures of smooth muscle and endothelial cells. Additional clones of the SLIC methodology were then screened, using this time ECE-1c and ECE-1d specific amplifications as a screening process. Nonresponding clones were sequenced, a few of them displaying an ECE-1 sequence strongly homologous to the human ECE-1b sequence, and that was therefore identified as rat ECE-1b isoform. Rat ECE-1b was predicted to include 769 amino acid residues ( Fig. 2A).
Identification of the ECE-1d specific basal promoter
The human genomic regions corresponding to the major part of exon 1b and to the intron separating exon 1b from exon 2 were inserted in a plasmid upstream from the luciferase reporter gene. The resulting construct was transfected into CHO-K1 cells and HSMC. As shown in Fig. 1B, the two cell types transfected with the ECE-1d promoter plasmid displayed a strong luciferase activity, at least five times higher than the plasmid background (pGL3-basic) and reaching one-half (CHO-K1) and one-fifth (HSMC) of the activity of cells transfected with the pGL3-SV40 plasmid. The rat ECE-1d promoter sequence was obtained by performing PCR on genomic DNA, and was manually aligned with its human counterpart ( Fig. 1A).
Distribution of the four ECE-1 isoforms in human and rat tissues
The levels of the four isoform mRNAs were analyzed by RT-PCR in various cells [human: ECV304 endothelial cell line, HUVEC, HSMC, polynuclear neutrophils, MRC5 fibroblast cell line, right atrium dissociated cardiomyocytes; rat: RAEC, RSMC, RFC, simian virus 40-transformed rat aortic endothelial cells (SVAREC) endothelial cell line] and tissue (human: saphenous vein, left ventricle; rat: lung, liver) types ( Fig. 3). All the different checked samples were known to contain average to high levels of ECE-1 mRNA. As expected, the highest signals were found in endothelial cells. In human, 25 cycles of amplification enabled the detection of strong signals for the four isoforms, showing that while ECE-1a expression was the most variable (e.g. endothelial vs smooth muscle cells or fibroblasts), the three other isoforms were always present at significant levels. In rat the situation was slightly different, as at 25 cycles of PCR only ECE-1c yielded strong signals. ECE-1a mRNA could not even be detected after 25 cycles. Thus, it seems that the distribution of the four isoform mRNAs is more unbalanced in rat than in human.
Immunoblotting analysis of CHO-K1 stable cell lines expressing ECE-1a and ECE-1d
As shown in Fig. 4, under nonreducing conditions a single band could be observed at 250 kDa, corresponding to the dimeric form of ECE-1 .
Endothelin-converting activity measurements
The enzymatic properties of ECE-1d were very similar to ECE-1a. The Km value for big ET-1 was 1.26 µm (SE 0.48 µm) which is not significantly different from that of ECE-1a (1.49 µm, SE 0.38 µm, determined in the same experiment).
Subcellular localization of ECE-1d in transfected CHO-K1 cells
As revealed by immunofluorescence, a large proportion of CHO-ECE-1d cells expressed a high level of immunoreactive ECE-1. On fixed and permeabilized cells, the plasma membrane, some punctate structures dispersed in the cytoplasm and a juxtanuclear zone appeared immunofluorescent ( Fig. 5A). The specific immunolabeling of the plasma membrane was observed on living cells only after treatment with 5 mm dithiothreitol ( Fig. 5B). Indeed, no signal could be detected in the absence of dithiothreitol, probably due to the fact that the anti-(ECE-1) antibody was raised against an epitope located in the vicinity of a Cys residue which was shown to participate in the dimerization of ECE-1 through the formation of a disulfide bridge . At the electron microscopic level, the presence of ECE-1d on the plasma membrane of stably transfected CHO-K1 cells was clearly revealed on living as well as on fixed cells ( Fig. 6). In fixed and permeabilized cells, immunoreactive ECE-1d was also detected in some Golgi stacks, in clusters of vesicles dispersed in the cytoplasm and in multivesicular endosomal structures ( Fig. 6).
In the present study we report the existence of ECE-1d, a novel human ECE-1 isoform, which we have characterized and compared to the three previously known isoforms. Moreover, we have cloned two previously unknown rat ECE-1 isoforms (ECE-1b and ECE-1d) and thus shown that homologues of the four human isoenzymes are present in rat.
ECE-1 is a type-II integral membrane protein, which displays a single transmembrane stretch separating a short N-terminal cytoplasmic tail from a large C-terminal intraluminal/extracellular domain [12, 13]. This domain bears the enzyme active site, which includes the HEXXH motif characteristic of an important class of metallopeptidases . Three isoforms of ECE-1 have been previously identified in man and named ECE-1a (758 residues), ECE-1b (770 residues) and ECE-1c (754 residues) [22, 23]. They only differ in the first half (approximately) of their cytoplasmic tail and are identical over their remaining sequences, including the enzymatic catalytic site. As a logical consequence, the three human isoforms display comparable converting activity but distinct profiles of subcellular localization . Thus whereas all three are present intracellularly, ECE-1a is in addition strongly expressed at the plasma membrane, which is not the case for ECE-1b, which is almost exclusively intracellular. The third isoform ECE-1c is moderately expressed at the cell surface.
Two ECE-1 isoforms have been cloned in rat . They were termed ECE-1α and ECE-1β and are homologous to the human isoforms ECE-1c and ECE-1a, respectively . In an attempt to clone the rat ECE-1b isoform, we were able to identify two previously unknown rat ECE-1 mRNA subtypes. One of them encoded the rat ECE-1b, which encompasses 769 amino acid residues. Rat and human ECE-1b specific coding sequences were found to be 71% identical (in nucleic acid content). The other rat mRNA subtype, in contrast, did not correspond to any of the three human isoforms, but surprisingly presented a very high homology (91% of nucleic acid identity) with a short region previously considered as intronic and localized immediately upstream from the third exon of the human ECE-1 gene . The presence of an inframe start codon in both species suggested the existence of a fourth and novel ECE-1 isoform, which was logically termed ECE-1d and was predicted to be 767 amino-acid residues long ( Fig. 2A). The existence of ECE-1d in both species was confirmed by specific RT-PCR performed in various tissues from both species ( Fig. 3).
The three previously known human isoforms are derived from distinct promoters which are localized upstream from the first (ECE-1c), second (ECE-1b) and fourth (ECE-1a) exons of the ECE-1 gene [22, 23]. The ability of two of these regions (ECE-1a and ECE-1b) to promote transcription when placed upstream from a reporter gene was shown by Orzechowski et al. . The region located between exon 1b and exon 2 ( Fig. 2C) was suspected to constitute the ECE-1d specific promoter. Placed upstream from the luciferase reporter gene, the corresponding genomic fragment was indeed able to direct transcription ( Fig. 1B), either when transfected in CHO-K1 cells (which do not express ECE-1) or in HSMC (which do not express any endogenous ECE-1). This lack of tissue specificity probably reflects the fact that some cis elements of the ECE-1d promoter are not located within the examined 215-bp region. These elements may be embedded in some regions further upstream, and therefore could be shared by the three promoters of ECE-1b, ECE-1c and ECE-1d. Such a situation would be in line with the results of RT-PCR experiments obtained on human tissues ( Fig. 3A), where these three isoform mRNAs were always detected concomitantly. Both human and rat ECE-1d promoters were sequenced and aligned ( Fig. 1A), which revealed the conservation between the two species of two strong SP1 sites, probably constituting, at least in part, the ECE-1d core promoter. Indeed, some SP1 sites have been previously shown to be essential to the basal transcription of a number of genes [37, 38].
Several genes have been shown to possess alternate promoters. This is the case, for instance, for neutral endopeptidase , an enzyme related to ECE-1, or for gamma-glutamyl transpeptidase . Most of the time, however, the resulting diversity is confined to the mRNA 5′ untranslated regions. The hypothesis which is then often formulated is that the presence of alternate promoters participates in the overall regulation of the protein expression, allowing a higher plasticity of the system. Indeed, independent promoters presumably respond differently to a given stimulus, and also probably do not display identical time or tissue specificity.
To examine the cell specificity of the four ECE-1 mRNAs, we have assessed by RT-PCR their levels in various cell types, and effectively found some differences ( Fig. 3). In particular, a major difference was found between primary cultures (derived from umbilical cords) of endothelial and smooth muscle cells with respect to the ECE-1a isoform, which was highly expressed in HUVEC but undetectable in HSMC. Smooth muscle cell expression of ECE-1 has been examined by Minamino et al. , who were able to detect ECE-1 mRNA in a rat vascular injury model and in human coronary atherosclerosis but not in nonpathologic vascular media. The relatively low signals that were detected in normal saphenous vein, for instance, probably only reflect the ECE-1 expression of the endothelial cell monolayer. HSMC are in culture and therefore in a proliferating state, which could explain the observed strong ECE-1 signals. ECE-1a is highly expressed at the cell surface . Its absence from atherosclerotic smooth muscle cell, if confirmed, would plead for an intracellular conversion of big ETs and therefore would be in line with an autocrine role for the mitogenic ET-1 in smooth muscle cell proliferation.
A similar RT-PCR analysis in rat revealed a striking difference with the human situation. Indeed, significant mRNA levels could be detected only for ECE-1c after 25 cycles of PCR. Five more cycles were needed to reveal significant ECE-1a signals ( Fig. 3B). Thus it appears that the four isoforms are more equally represented in man than in rat. This probably reflects some promoter differences, and suggests that although both human and rat ECE-1 display four isoforms, any parallelism between the two species in this respect should be considered with caution.
In addition to allowing the fine regulation of the protein expression, the presence of alternate promoters can also, in a few rare cases including ECE-1, lead to the production of distinct isoproteins. The biochemical properties of ECE-1d did not differ from those of ECE-1a, which was previously shown to be identical to those of ECE-1b and ECE-1c . These results are not surprising considering that the catalytic domain of the enzymes is located in the intraluminal/extracellular portion of the protein (identical in the four isoforms) and, moreover, are in agreement with the fact that a recombinant soluble ECE-1 was shown to hold full ECE-1 activity, inhibitor recognition and substrate specificity [31, 41].
As human ECE-1a, ECE-1b and ECE-1c, when expressed in CHO cells, displayed differences in their subcellular distribution , we analyzed by immunofluorescence and electron microscope cytochemistry the localization of ECE-1d. In transfected CHO-K1 cells, the fourth human isoform was detected both at the cell surface and intracellularly in Golgi and endosomal structures. Immunolabeling, detected at the level of the Golgi stacks and vesicles, most probably corresponds to the visualization of ECE-1d during its biosynthetic pathway to the plasma membrane. The localization in endosomal structures suggests that these compartments could be involved in the degradative pathway of ECE-1d or in its recycling pathway or in both. Comparison of the subcellular localization of ECE-1d with that of the three other isoforms  reveals that ECE-1d is more strongly expressed at the cell surface than ECE-1c but less than ECE-1a. Furthermore, the presence of ECE-1d in endosomal structures is in accordance with the previously reported localization of human ECE-1b . Taken together, the subcellular localization of the four isoforms constitute the distribution pattern which has been described for ECE-1 using a common antibody. In HUVEC, for instance, ECE-1 has been shown to be present on the plasma membrane and also in some specific endothelial cell compartments, the Weibel–Palade bodies . Whether this specific localization concerns the four isoforms should be investigated.
In conclusion, we have analyzed and characterized ECE-1 diversity in both human and rat species, and have shown in both cases the existence of four isoforms, differing by their N-terminal extremities and produced from the same gene via the use of four distinct promoters. This will clarify a confused situation where, so far, the incomplete description of the ECE-1 isoform system has been at the origin of partially inexact reports, in which, for instance, conclusions have been formulated for a specific isoform using antibodies which recognized several isoforms. Differences have already been observed with respect to the subcellular and cellular localizations of the four isoforms and should be further investigated. Indeed, the fact that this multiplicity is conserved between rat and man suggests that it definitely has a physiological importance.
We would like to acknowledge C. Badier and Dr J.-J. Mercadier for the gift of RNAs, F. Mongiat for excellent technical assistance, E. Etienne for photographic work and Dr M. Osborne-Pellegrin for critical reading of the manuscript. This work was supported by grants from the Fondation pour la Recherche Médicale (O.V.) and the Fondation Simone et Cino del Duca (G.E.).
Enzymes: ECE-1 (EC 188.8.131.52), neutral endopeptidase (EC 184.108.40.206)Note: The novel nucleotide sequence data reported here have been submitted to the GenbankTM/EMBL DataBank with accession numbers AJ130826, AJ130827 and AJ130828.