Mouse angiotensin-converting enzyme is a protein composed of two homologous domains.

Angiotensin-converting enzyme (ACE) is a dipeptidyl carboxypeptidase that converts angiotensin I into the potent vasoconstrictor angiotensin II. We have used cDNA and genomic sequences to assemble a composite cDNA, ACE.315, encoding the entire amino acid sequence of mouse converting enzyme. ACE.315 contains 4838 base pairs and encodes a protein of 1278 amino acids (147.4 kDa) after removal of a 34-amino acid signal peptide. Within the protein, there are two large areas of homologous sequence, each containing a potential Zn-binding region and catalytic site. These homologous regions are approximately half the size of the whole ACE protein and suggest that the modern ACE gene is the duplicated product of a precursor gene. Mouse ACE is 83% homologous to human ACE in both nucleic acid and amino acid sequence, and like human ACE, contains a hydrophobic region in the carboxyl terminus that probably anchors the enzyme to the cell membrane (Soubrier, F., Alhenc-Gelas, F., Hubert, C., Allegrini, J., John, M., Tregear, G., and Corvol, P. (1988) Proc. Natl. Acad. Sci. U.S.A. 85, 9386-9390). Northern analysis of mouse kidney, lung, and testis RNA demonstrates that the testicular isozyme of ACE is encoded by a single, smaller RNA (2500 bases) than the two message sizes found in kidney or lung (4900 and 4150 bases), and that this testicular RNA hybridizes to the 3' portion of ACE.315.

' The abbreviations used are: ACE, angiotensin-conve~ing enzyme; bp, base pairs. tation that its inhibition would lower blood pressure in patients with high renin levels. Surprisingly, ACE inhibitors also reduce blood pressure in the 85% of essential hypertensive patients with normal or low renin levels (5). Such observations emphasize the complex nature of blood pressure regulation and the multifaceted role played by angiotensin-converting enzyme.
Dipeptidyl c a r b o x~e p t i~s e (EC.3.4.15.1) i s a single chain glycoprotein produced by many tissues including renal tubular epithelium, ciliated gut epithelium, stimulated macrophages, testis, and areas of the brain (3,6). However, it is thought that endothelium, which produce and direct this enzyme to the luminal surface, is the most significant to systemic pressure regulation (7-10). In uivo it is not known whether endothelium have varied surface expression of ACE. In cultured endothelium, however, ACE is not const~tutively produced but varies in response to such external stimuli as the state of confluence of the cells, hormone concentrations, and perhaps to ACE inhibitors (11)(12)(13).
Many reports have detailed the isolation of ACE from a variety of organs and species (3). These studies generally agree that converting enzyme consists of a single polypeptide chain with a molecular mass of approximately 140 kDa. Other studies have attempted to compare the converting enzyme isolated from different tissues using competitive radioimmunoassay, Ouchterlony double diffusion, and enzyme kinetic analyses. As reviewed by Soffer (6), these studies have generally found that with the exception of testis and epididymis, ACE proteins isolated from a variety of organs display indistin~ishable physicochemical, catalytic, and immunologic properties.
We have previously reported the purification of ACE from mouse kidney and the cloning of cDNA encoding this enzyme (14,15). Two classes of ACE cDNA were identified based on patterns of hybridizations to five ACE-specific oligonucleotide probes (15). We previously reported the sequence of ACE.31, a cDNA that encodes the NHz-terminal amino acid sequence of mouse ACE. Northern analysis of mouse kidney and lung mRNA with ACE.31 identified two hybridizing bands; genomic Southern analysis suggested that ACE.31 was encoded by a single gene within the mouse genome (15).
We now present the nucleic acid sequence of ACE.5 and ACE.ll, two independently isolated class 11 ACE cDNA.
These cDNA are virtually identical in their coding regions but differ substantially in the 3"untranslated regions. Analysis of cloned mouse genomic DNA shows that ACE.31 and both ACE.6 and ACE.11 are derived from a single gene and allow the elucidation of the entire protein sequence of mouse ACE. This protein is composed of 1278 amino acids and may be the result of a genetic duplication of a smaller precursor gene. Recently Soubrier et al. (16) published the cDNA se-  quence of human ACE and while the sequences of mouse and human ACE are highly homologous, the two regions in each molecule containing putative catalytic sites are the most highly conserved.

MATERIALS AND METHODS
mRNA was prepared from the kidneys of male NIH Swiss mice and selected by oligo(dT)chromatography (17). Mouse CD-1 lung, testis, and kidney mRNA was also purchased from Clontech Laboratories, Inc., Palo Alto, CA. The construction of a mouse kidney cDNA library, the screening of this library with ACE-specific oligonucleotide probes, the subcloning of double-stranded cDNA into the EcoRI site of Bluescript (+)(Stratagene, San Diego, CA), and the rescue and sequencing of single-stranded DNA have been described (15). During the construction of cDNA, internal EcoRI sites were not protected by methylation resulting in the separate cloning of sequences on either side of internal EcoRI sites (GAATTC). ACE.ll was completely sequenced on both strands using a combination of three approaches, subcloning RsaI restriction fragments of ACE.ll into the SmoI site of mplO M13, priming single-stranded DNA with complementary ACE oligonucleotides and unidirectional progressive deletions using EzoIII nuclease (Erase-a-Base, Promega (18). The unidirectional progressive deletions were prepared by cutting both orientations of ACE.ll in Bluescript with XbaI followed by end repair with or-phosphorothioate deoxynucleotides. The DNA was digested with BamHI, phenol extracted, and then digested with Ex0111 and recloned as described in the protocols of the Erase-a-Base kit. The sequence of ACE.5 was determined by priming single-stranded DNA with specific oligonucleotides. Where sequence was identical to that of ACE.ll, it was determined on one strand; regions of difficulty or where sequence was divergent from that of ACE.ll were determined on both DNA strands.
A mouse genomic DNA library, prepared from partially digested liver DNA in the vector EMBL3, was purchased from Clontech Laboratories. The library was screened with ACE.31 and with an oligonucleotide from the 5' portion of ACE.ll. One phage entitled G4 was plaque purified to homogeneity and DNA prepared. A 3200bp PstI restriction fragment was found to hybridize with both ACE.31 and ACE.ll probes. This was subcloned into Bluescript (KS). DNA was prepared and fragmented by sonication. Those fragments greater than approximately 600 bp were isolated after agarose gel electrophoresis. The fragments were blunt-ended using T4 DNA polymerase and then the large fragment of DNA polymerase (19). Finally, they were cloned into SmaI-cut mplO M13 (Amershaq Corp.  (20,21). All DNA was labeled with [ c Y -~P I~C T P (Amersham Corp.) using a random hexamer primer kit from Bethesda Research Laboratories (22) and was hybridized under standard conditions (15).

RESULTS
In a screening of 70,000 primary plaques of a mouse kidney cDNA library, three oligonucleotide probes, 37.21, 43.9, and 50.53, identified 26 plaques referred to as ACE group 1 cDNA (15). Nine cloned cDNA ranged in size from 2900 to 3800 bp. One additional cDNA, ACE.11, was 4700 nucleotides. cDNA ACE.5 is 3820 nucleotides. It contains one large open reading frame beginning at the 5' EcoRI site and encoding 980 amino acids ( Figs. 1 and 2). The DNA sequence of ACE.ll has also been determined and with the exception of one nucleotide at position 706 (GCC-Ala in ACE.11, ACC-Thr in ACE.5), is identical to that of ACE.5 for the 5' first 3130 nucleotides. This encompasses 2943 nucleotides of coding sequence and 187 nucleotides of 3"untranslated region (Figs. 1 and 2). Further, 3' these two cDNA are completely divergent with ACE.5 encoding a total of 873 nucleotides of 3' Ut and ACE.ll encoding 1751 nucleotides of 3' Ut ( Fig. 2).
The cDNA library used in these studies was such that internal EcoRI sites were cleaved during library construction. Thus, we asked if ACE.31 and ACE.5 were contiguous within genomic DNA. A genomic library was probed with ACE.31 and an oligonucleotide from the 5' end of ACE.5. One clone called G4 was purified, and a 3200-bp PstI fragment was subcloned and sequenced (Fig. 3). This demonstrates that the EcoRI sites at the 3' end of ACE.31 and the 5' end of ACE.5 are in fact a single restriction site within mouse genomic DNA (Fig. 3). Further evidence that ACE.31 and ACE.5 encode contiguous mRNA sequence is found in the amino acid sequence of the ACE tryptic peptide 40.1 (15). The amino terminus of this peptide, VSEgFTSLGLSP? is exactly the sequence found by abutting and translating ACE.31 and ACE.5 with the EcoRI site encoding Glu and Phe. We refer to the combined sequences of ACE.31 and ACE.5 as ACE.315. This sequence is 4838 bp and encodes a protein of 150.9 kDa composed of 1312 amino acids (Fig. 4). The mature protein without a signal peptide is predicted to contain 1278 amino acids with a molecular mass of 147.4 kDa. ACE.315 encodes

I R Y F V S F V L Q F Q F H Q A L C X E A G H Q G P L H Q C D I Y Q S B Q A G A 2 4 1 K L K Q V L Q A G C S R P W Q E V L K D
GCAGGCCACCAGGGCCCACTACACCAGTGTGACATCTACCAGTC~CCAGGCGGGGGCC 7 2 0  Deletional clones were prepared as described under "Materials and Methods." These are indicated by @. Additionally, vector-free ACE.ll was cut with RsaI and the fragments cloned into the SmaI site of M13 mplO. Single-stranded DNA was prepared and sequenced as previously described (15). Finally, ACE-specific oligonucleotides were synthesized and used to prime single-stranded DNA. These are indi-    amino acid sequence identical or highly homologous to each of nine ACE tryptic peptides previously reported (Fig. 4) (15). The method of Kyte and Doolittle (23) was used to analyze hydrophobic-hydrophilic regions of the protein encoded by ACE.315. Both the signal peptide region and the carboxyl terminus encode highly hydrophobic regions (Fig. 5). The hydrophobic carboxyl domain is followed by a highly hydrophilic region; it is probably here that ACE, a known membrane protein, is anchored to the cell membrane. Similar results were found in the study of human ACE (16).  t t a g~a g g~f g~g g a g t t t g~~g~c a g c c t g g a c~a c a t a g t g~g a c c c~g g c    portion of ACE3 (ACCAACGGATGGACGGGAGGTGGTGTGC, bp 59-87). One plaque entitled G4 was purified. DNA was prepared, cut with PstI, and a 3200-bp fragment was subcloned into Bluescript (KS). This is called G4.3200. G4.3200 was sequenced using a combination of shotgun cloning and oligonucleotides (indicated by *) as described under "Materials and Methods." Nucleotide sequence was compared with that of ACE.31 and ACE.5 to delineate introns (sequence in lowercase) and exons (sequence in uppercase). The EcoRI site present at the 3' end of ACE.31 and the 5' end of ACE.5 is underlined and in bold type.

C t t g t t t C t~a a c a G A G A G T G A C A T c M T T A c c T G c T~G A T~c c c T A G A G~T c G
A computer search of the GenBank database using the program Genepro (Riverside Scientific Enterprises, Seattle, WA) identified no homologous DNA or protein sequences. Recently, Soubrier et al. (16) published the cDNA sequence encoding human ACE. The human cDNA and ACE.315 are 83% homologous in DNA sequence. A dot matrix comparison of these two cDNA is shown in Fig. 7A. An alignment of the encoded proteins demonstrates that these are also highly homologous; 1088 of 1312 matched positions are identical (83%) while another 60 are conservative substitutions. When stringent criteria are used in assessing homology ( B ) , amino acids 625-765 of the mature protein encoded by ACE.315 is the region least conserved between human and mouse ACE. Functionally, this represents the linking amino acids between the two conserved domains present within ACE proteins.
Using thermolysin as a model system, Soubrier et al. (16,25) identified a number of amino acids potentially involved in human ACE catalytic activity. Similar amino acids are found within the two separate regions of mouse ACE and are Both the signal peptide and the carboxyl terminus are highly hydrophobic. The hydrophobic region at the carboxyl terminus is followed by a hydrophilic region, and it is probably here that ACE is anchored to a cell membrane.
highly conserved between human and mouse. X-ray analysis of thermolysin has implicated His-142, His-146, and Glu-166 in Zn binding. Within both mouse and human ACE, these may correspond to His-361, His-365, Glu-389, and His-959, His-963, Glu-987. Glu-143 and His-231 of thermolysin, thought involved in the active site of the enzyme, may correspond to Glu-362, Glu-960, and His-404, His-1002 within both mouse and human ACE. Thus, both human and mouse ACE are composed of two homologous domains each containing a potential catalytic site. Northern analysis of kidney and lung RNA with ACE.ll demonstrates a pattern identical to that observed with ACE.31 (Fig. 8A). Both probes identify two bands, one of 4900 and one of 4150 bases. In contrast when hybridized with mouse testis mRNA, ACE.ll identifies a strong band of 2500 bases and a weaker band of 1350 bases. These bands are not identified by ACE.31. Liver mRNA shows virtually no hybridization to ACE.ll (data not shown). Restriction fragments comprising ACE.ll from bp 672 to 1186 and the 5' 557 bp of ACE.31 were prepared and used to probe mouse lung and kidney mRNA. These restriction fragments identify identical bands on Northern analyses to those obtained with the full-length ACE.31 or ACE.ll ( B ) . DISCUSSION We have previously described the purification of mouse kidney ACE and the partial sequence of nine ACE tryptic peptides. Based on this information, five oligonucleotides were used to screen a mouse kidney cDNA library. The positive cDNA clones fell into two groups based on the patterns of hybridization to the oligonucleotides. The analysis of mouse genomic DNA presented here now establishes that this was artifactual, due to cleavage of an internal EcoRI site during cDNA construction. We refer to the 4838-bp fusion of ACE.31 and ACE.5 as ACE.315 and ask whether ACE.315 encodes the protein sequence of mouse angiotensin-converting enzyme. ACE.315 encodes protein sequence identical or very similar to that found in nine separate tryptic peptides from purified ACE protein. These nine sequences are found throughout the encoded protein sequence of ACE.315. Review of the amino acid chromatographic patterns for those regions where there are discrepancies between the protein microsequencing results and the predicted sequence of ACE.315 shows that these differences are due to difficulties or errors in protein sequence determination; no definitive differences are noted between the microsequencing results and the sequence predicted by cDNA analysis. In addition Northern analyses using ACE.31, ACE.ll, and smaller restriction fragments from these cDNA detects hybridizing mRNA in kidney and lung (organs rich in ACE) and very little hybridization to liver RNA (an organ poor in ACE). Perhaps the only difficulty in unconditionally accepting ACE.315 as encoding mouse converting enzyme is the unexpectedly large molecular mass of the encoded mature protein (147.4 kDa). Generally the molecular mass of ACE has been reported as being between 140-150 kDa with the assumption that a percentage of this weight is due to protein glycosylation (6). Nonglycosylated ACE has been studied by El-Dorry et al. (26) who translated rabbit lung mRNA in vitro and determined that the converting enzyme produced had a molecular mass of 129 kDa. Despite this result, there are several good reasons to believe that ACE.315 does encode ACE protein, the most convincing being that the amino acid sequences found in ACE peptides are scattered throughout the protein predicted by ACE.315. ACEB and ACE.ll, two independently isolated ACE cDNA, predict a virtually identical protein when coordinated with the DNA sequence of ACE.31. Studies in the rabbit using cDNA selected with anti-ACE antisera have identified ACE mRNA of 5000 and 2600 bp in lung and testis mRNA, respectively (27). These data are very similar to those reported in this paper and corroborate ACE.315 as encoding mouse angiotensin-converting enzyme. Finally, the recently reported sequence of human ACE is highly homologous to that of ACE.315 (16).
One difference between the results found in the rabbit and those reported here is the presence in mouse lung and kidney of two ACE mRNA bands, one of 4900 and one of 4150 bases. We do not know yet the difference in structure between these two message sizes but the results of Northern analyses with restriction fragments of ACE.31 and ACE.ll (ACE.31/557 and ACE.11/514) demonstrates that both bands contain sequences homologous to ACE.315 positions 1-557 and 1699-2213. As reported by Soubrier et al. (16), Northern analysis of RNA from cultured human umbilical vein endothelium shows only a single ACE band of 4.3 kilobases.
It is also not known at this time why ACE.ll contains such  RNA were separated using a 1% agarose gel containing formaldehyde (15). These were probed with the restriction fragment ACE.11/514 or ACE.31/557 as described in the text. The two lanes labeled K/I I are different film exposures of the same portion of the blot. The first lane ( K / I I ) was exposed to film for 3 h at -70 "C. The second and third lanes ( K / l l , L / l l ) were exposed overnight at -70 "C. The last two lanes (K/31 and L/31) were exposed for 3 days at -70 "C. a large 3' Ut. What seems clear is that the combined size of ACE.31 with ACE.ll, 5718 nucleotides, is larger than the size of the 4900-base band detected on our typical RNA analysis. One hypothesis now being investigated is that ACE.ll represents a rare form of ACE message that makes use of a different set of 3' Ut exons than that of the typical forms of ACE mRNA.
Testis expression of ACE has been somewhat of an enigma. This organ seems unique in expressing two forms of ACE that differ in their molecular weights, NH, and COOH terminal amino acids and patterns of expression (3). The larger protein appears similar to that found elsewhere in the body while the smaller protein is unique to testis and epididymis (6). We now present evidence that testis expresses large amounts of mRNA homologous to ACE.ll and not ACE.31. The RNA was sized at approximately 2500 bases, substantially smaller than kidney or lung ACE mRNA. This observation is consistent with the findings of El-Dorry et al. (26) as well as those of Roy et al. (27). As it is now clear that ACE.ll is virtually identical in sequence to the 3' portion of ACE.315, we hypothesize that the testis-specific isozyme of ACE is homologous if not iden-tical with the carboxyl-terminal portion of lung and kidney ACE.
Perhaps the most interesting aspects of the protein encoded by ACE.315 is the two internal areas of homology. Remarkably, each of these two areas of homology is roughly half the size of the parent molecule. We assume based on the work of Das and Soffer (28) that ACE has only a single substratebinding site. At present, it is impossible to assign with certainty which of the two possible catalytic sites are physiologically active. On the one hand a report by Iwata et al. (29) demonstrates that an 88-kDa amino-terminal fragment of rabbit lung ACE has enzyme activity. In contrast Northern analysis of testis RNA (Fig. SA) suggests that the testisspecific form of ACE may only contain a single catalytic site homologous to that of the carboxyl portion of kidney ACE. These data coupled with the analysis of protein structure presented here suggest that only one of the large homologous regions is sufficient for enzyme activity. A reasonable speculation is that at some point during the molecular evolution of converting enzyme, the ancestral gene encoding this protein underwent a genetic duplication. It is also possible that during evolution (or in some species even now) converting enzyme may have had two substrate-binding sites. And perhaps modern ACE can use either of its catalytic sites though not both concurrently in a single molecule.