Biol. Chem., Vol. 390, pp. 931–940, September 2009 • Copyright � by Walter de Gruyter • Berlin • New York. DOI 10.1515/BC.2009.105

2009/149

Article in press - uncorrected proof

Catalytic properties of recombinant dipeptidylcarboxypeptidase from Escherichia coli: a comparativestudy with angiotensin I-converting enzyme

Carlos Eduardo L. Cunha1,a, Helena de FatimaMagliarelli1,a, Thaysa Paschoalin1,2, Aloysius T.Nchinda3, Jackson C. Lima4, Maria A. Juliano1,Paulo B. Paiva4, Edward D. Sturrock3, LuizR. Travassos2 and Adriana K. Carmona1,*1 Department of Biophysics, Federal University of SaoPaulo, Rua 3 de Maio 100, 04044-020 Sao Paulo, SP,Brazil2 Department of Microbiology, Immunology andParasitology, Federal University of Sao Paulo, RuaBotucatu 862, 8oI andar, 04023-062 Sao Paulo, SP,Brazil3 Division of Medical Biochemistry, Institute of InfectiousDisease and Molecular Medicine, University of CapeTown, Observatory 7925, Cape Town, South Africa4 Department of Information Technology in Health,Federal University of Sao Paulo, Rua Botucatu 862,04023-062 Sao Paulo, SP, Brazil

* Corresponding authore-mail: adriana@biofis.epm.br

Abstract

Dipeptidyl carboxypeptidase from Escherichia coli(EcDcp) is a zinc metallopeptidase with catalytic prop-erties closely resembling those of angiotensin I-convert-ing enzyme (ACE). However, EcDcp and ACE areclassified in different enzyme families (M3 and M2,respectively) due to differences in their primary sequen-ces. We cloned and expressed EcDcp and studied indetail the enzyme’s S3 to S19 substrate specificity usingpositional-scanning synthetic combinatorial (PS-SC)libraries of fluorescence resonance energy transfer(FRET) peptides. These peptides contain ortho-amino-benzoic acid (Abz) and 2,4-dinitrophenyl (Dnp) as donor/acceptor pair. In addition, using FRET substrates devel-oped for ACE wAbz-FRK(Dnp)P-OH, Abz-SDK(Dnp)P-OHand Abz-LFK(Dnp)-OHx as well as natural ACE substrates(angiotensin I, bradykinin, and Ac-SDKP-OH), we showthat EcDcp has catalytic properties very similar to humantestis ACE. EcDcp inhibition studies were performed withthe ACE inhibitors captopril (Kis3 nM) and lisinopril(Kis4.4 mM) and with two C-domain-selective ACEinhibitors, 5-S-5-benzamido-4-oxo-6-phenylhexanoyl-L-tryptophan (kAW; Kis22.0 mM) and lisinopril-Trp (Kis0.8 nM). Molecular modeling was used to provide thebasis for the differences found in the inhibitors potency.

These authors contributed equally to this work.a

The phylogenetic relationship of EcDcp and relatedenzymes belonging to the M3 and M2 families was alsoinvestigated and the results corroborate the distinct ori-gins of EcDcp and ACE.

Keywords: angiotensin I-converting enzyme; dipeptidylcarboxypeptidase; inhibitors; molecular modeling;phylogeny; specificity studies.

Introduction

The regulation of the life cycle in all types of cells is deter-mined by a balance of antagonistic processes, such asprotein synthesis and proteolysis (Maurizi, 1992; Gottes-man, 1996). This balance is generally called protein turn-over and plays a major role in the control of the cell cycle(Wright et al., 1996; Grunenfelder et al., 2001). Many bac-terial peptidases involved in these processes havealready been described, including some that hydrolyzesmall peptides instead of high molecular mass proteins(Miller, 1975). One of these peptidases is a dipeptidyl car-boxypeptidase (Dcp) present in Escherichia coli whichwas first described by Yaron et al. (1972). The enzyme isa zinc metallopeptidase and consists of 680 amino acidresidues, forming an active monomer with a molecularmass of 77.5 kDa (Henrich et al., 1993). Due to its struc-tural characteristics, Dcp from E. coli (here termedEcDcp) was grouped in the M3 family which also includesneurolysin and thimet oligopeptidase (Barrett, 2004). Thedistinguishing feature of the MA(E) subclan of metallo-peptidase family is their highly conserved metal ligandmotif HEXXH (Chu and Orlowski, 1984).

The catalytic properties of EcDcp closely resemble themammalian angiotensin I-converting enzyme (ACE)(Yaron et al., 1972; Yaron, 1976). ACE is a zinc metallo-peptidase that is important in circulatory homeostasisdue to its capability to hydrolyze angiotensin I producingthe potent vasoconstrictor angiotensin II (Skeggs et al.,1956). In addition, the enzyme also inactivates the vaso-depressor peptide bradykinin (Yang et al., 1970). ACE isexpressed as a somatic isoform (150–180 kDa) in endo-thelial, epithelial, and neuroepithelial cells and as asmaller isoform (90–110 kDa) only in male germinal cells.Somatic ACE (sACE) contains two large homologousdomains (N- and C-domains), both being catalyticallyactive (Wei et al., 1991). Testicular ACE (tACE) containsa single active site and corresponds to somatic ACEC-domain with the exception of a short N-terminalsequence (Ehlers et al., 1989; Kumar et al., 1989; Lattionet al., 1989; Natesh et al., 2003).

EcDcp removes dipeptides from the free C-termini ofpeptides, N-blocked tripeptides and unprotected tetra-

932 C.E.L. Cunha et al.

Article in press - uncorrected proof

Figure 1 Homology of the bacterial Dcp with M2 and M3 pep-tidases in the phylogenetic tree.Inferred unrooted phylogenetic tree of M2 and M3 metallopep-tidases. EcDcp and enzymes related to M2 and M3 familiesgroup into two monophyletic groups. The sequences and acces-sion numbers used for the phylogenetic analysis were: ACN1,Caenorhabditis elegans ACE-like non-peptidase (AAD28560);XcACE, Xanthomonas axonopodis pv. citri ACE (YP_244195);ACEN, human N-domain ACE (NP_000780); ACEC, human C-domain ACE (NP_690043); tACE, testicular ACE; ACE2, humanACE2 (AAS57725); AnCE, Drosophila melanogaster AnCE(NP_477046); ACEr, D. melanogaster ACEr (Q9VLJ6); Elito,Erythrobacter litoralis predicted dipeptidyl-carboxypeptidase(ZP_00376302); Gviol, Gloeobacter violaceus predicted dipepti-dyl-carboxypeptidase (NP_926089); TOP, human thimet oligo-peptidase (P52888); NEU, human neurolysin (NP_065777);OPDA, Escherichia coli oligopeptidase A (AP_004295); EcDcp,E. coli dipeptidyl carboxypeptidase II (CAA41014); LdDcp, Leish-mania donovani dicarboxypeptidase (AAV80217); MIP, mitochon-drial intermediate peptidase (NP_005923).

peptides (Yaron et al., 1972). The enzyme hydrolyses theACE substrates angiotensin I (Yaron et al., 1972), bra-dykinin (Yaron, 1976), and Hip-His-Leu (Hip: hippuricacid; Yaron, 1976). As with ACE, peptides containing Proat P19 (nomenclature by Schechter and Berger, 1967) orwith blocked C-terminus are also resistant to EcDcpcleavage (Yaron, 1976). In addition, EcDcp is inhibited bythe ACE-specific inhibitor captopril (Deutch and Soffer,1978; Turner and Hooper, 2002). Despite their catalyticsimilarity, however, EcDcp (EC 3.4.15.5) and ACE (EC3.4.15.1) are classified in different enzyme families (M3and M2, respectively) due to differences in their primarysequence (Barrett, 2004).

Comellas-Bigler et al. (2005) described the crystallo-graphic structure of EcDcp and provided evidence thattACE and EcDcp bear some topological similarity. Bothhave similar size, shape, and secondary structures.Docking experiments with the ACE inhibitor lisinoprilhave shown that it could bind to EcDcp in a similararrangement to that in the lisinopril-tACE crystal structure(Comellas-Bigler et al., 2005). A single ACE homologdomain, ACE2 that exhibits carboxypeptidase activity(Donoghue et al., 2000; Tipnis et al., 2000), also showeda similar topology with EcDcp, based on their crystallo-graphic structures, despite the low level of sequenceidentity (Comellas-Bigler et al., 2005).

Recently, Goyal et al. (2006) described a dipeptidylcarboxypeptidase in Leishmania donovani (LdDcp) thatalso belongs to the M3 family but does not displaysequence homology with ACE. LdDcp was able to hydro-lyze the synthetic ACE substrates Hip-His-Leu and 2-furylacryloyl-phenylalanyl-glycyl-glycine (FAPGG) but notthe natural substrate angiotensin I. Moreover, it was alsoshown that the specific ACE inhibitor, captopril, was ableto inhibit LdDcp catalytic activity (Goyal et al., 2006).More recently, an ACE-like enzyme belonging to the M2family was cloned and expressed in Xanthomonasaxonopodis pv. citri (XcACE) (Riviere et al., 2007). Thecharacterization of XcACE revealed that it has catalyticproperties very similar to those already described for Dcpfrom E. coli (Yaron et al., 1972). However, an importantdifference was observed for XcACE in that it was able tohydrolyze angiotensin I, Hip-His-Leu, and the ACE N-domain specific substrate AcSDK(Dnp)P-OH. In addition,XcACE is inhibited by ACE specific inhibitors (Riviereet al., 2007).

Due to the increasing interest in the study of ACE-likeenzymes, we cloned and expressed recombinant EcDcpand elucidated its substrate specificity requirementsusing positional-scanning synthetic combinatorial (PS-SC) libraries of fluorescence resonance energy transfer(FRET) peptides. EcDcp catalytic activity using FRETsubstrates developed for ACE (Araujo et al., 2000; Ber-sanetti et al., 2004) as well as natural ACE substrates(angiotensin I, bradykinin, and Ac-SDKP-OH) was alsotested. The results indicated that the enzyme has hydro-lytic properties very similar to human tACE, reinforcingthe structural results which show that Dcp is composedof a single domain similar to the C-domain of mammalianACE (Goyal et al., 2006). In addition, EcDcp inhibitionstudies were performed with the ACE inhibitors captopriland lisinopril and with the two recently described C-domain-selective ACE inhibitors, kAW (5-S-5-Benzami-

do-4-oxo-6-phenylhexanoyl-L-tryptophan; Nchinda et al.,2006a; Watermeyer et al., 2008) and lisinopril-Trp wN2-(R,S)-(1-carbonyl-3-phenylpropyl)-L-lysyl-L-tryptophan;Nchinda et al., 2006bx. Further, the phylogenetic relation-ship of EcDcp and related enzymes belonging to M2 andM3 families was inferred.

Results

Phylogenetic analysis

The phylogenetic results were obtained using the sub-stitution model WAG which best befits the data (Whelanand Goldman, 2001). The inferred phylogeny (Figure 1)

Characterization of dipeptidyl carboxypeptidase from E. coli 933

Article in press - uncorrected proof

presented a consensual topology between the differentprograms. Sequences of EcDcp and enzymes related toM2 and M3 families suggest the existence of two mono-phyletic groups, one containing EcDcp and M3 familyenzymes. LdDcp and EcDcp are monophyletic with MIP.Although these samples come from species of Bacteriaand Eucarya, which are phylogenetically distant (DeLongand Pace, 2001), the inferred phylogeny suggests thatthe samples share the same ancestor. The bacterial ACEfrom the phytopathogenic bacterium X. citri (XcACE),which has features in common with both N- and C-domain of mammalian somatic ACE (Riviere et al., 2007),is phylogenetically distant.

Recombinant EcDcp enzymatic activity

The cloning, expression, and purification of EcDcp weresuccessfully carried out and a single band of approxi-mately 77 kDa could be visualized by sodium dodecylsulfate polyacrylamide gel electrophoresis analysis (datanot shown). The catalytic activity of the recombinantenzyme was tested with the FRET substrate Abz-FRK(Dnp)P-OH, developed for ACE (Araujo et al., 2000).EcDcp cleaved the peptide at the R-K(Dnp) bond con-firming the enzyme’s carboxydipeptidase activity and itscatalytic similarity to mammalian somatic ACE. In addi-tion, the enzyme was also 100% inhibited by captopril(1 mM).

Screening of peptide libraries

PS-SC fluorogenic libraries were used to map the aminoacid preferences of the EcDcp subsites S3, S2, S1, andS19. Initially, the peptide library Abz-GXXZXK(Dnp)-OHwas screened to define the specificity of the P1 position.Figure 2 shows the relative values expressed as the per-centage of initial rate normalized by the best substrate ineach series. The S1 subsite has a strong preference forbasic amino acids with Arg as the best accepted residueon P1 position. The hydrophobic residues Phe and Leuwere also favored. On the other hand, the acidic residueAsp is not tolerated at this subsite as demonstrated bythe resistance to hydrolysis of the sequence Abz-GXXDXK(Dnp)-OH.

Thus, we fixed Arg in the P1 position and preparedthree other libraries with general sequences Abz-GZXRXK(Dnp)-OH, Abz-GXZRXK(Dnp)-OH, and Abz-XXRZK(Dnp)-OH to study the specificity of the subsitesS3, S2, and S19, respectively. Figure 2 shows that peptidescontaining Lys and Arg at P19 position were very suscep-tible to hydrolysis by EcDcp. The pH used (pH 8.0) is nothigh enough to keep His protonated, explaining why thepeptide Abz-XXRHK(Dnp)-OH is poorly hydrolyzed byEcDcp when compared to the sequences containingthe other basic residues in the P19 position. As expected,the S19 subsite did not tolerate a Pro residue with thesequence Abz-GXXRPK(Dnp)-OH being resistant tohydrolysis. The S2 pocket of EcDcp is more restrictiveand shows remarkable selectivity for Phe at P2 as shownby the efficient hydrolysis of Abz-GXFRXK(Dnp)-OH. Onthe other hand, the S3 subsite tolerated a broad range ofamino acids suggesting that this position is less impor-tant for enzyme activity.

These results are very similar to those obtained in thespecificity studies with PS-SC fluorogenic libraries forwild-type recombinant sACE (Bersanetti et al., 2004). Aninteresting difference in the substrate specificity of ACEand EcDcp was observed for Abz-GXXDXK(Dnp)-OH andAbz-GXXIXK(Dnp)-OH, which were hydrolyzed by ACEwith low catalytic efficiency (Bersanetti et al., 2004), butcompletely resisted hydrolysis by EcDcp.

EcDcp catalytic activity on FRET peptides

EcDcp activity was assayed with the ACE FRET sub-strates Abz-FRK(Dnp)P-OH, Abz-LFK(Dnp)-OH, and Abz-SDK(Dnp)P-OH (Table 1) and the cleavage sites analyzedby liquid chromatography/electrospray ionization massspectrometry (LC/ESI-MS; Figure 3). The enzyme wasable to hydrolyze Abz-FRK(Dnp)P-OH at the R-K bondwith a kcat/Km value of 2080 mM-1 s-1. The ACE C-domain-specific substrate Abz-LFK(Dnp)-OH was hydrolyzed atthe L-F bond with lower catalytic efficiency (kcat/Kms550 mM-1 s-1), while the ACE N-domain-specific sub-strate Abz-SDK(Dnp)P-OH was resistant to hydrolysis.

EcDcp was not able to hydrolyze the ACE2 substrateAbz-APK(Dnp)-OH (Figure 3) due to the proline residueat P19 (Table 1), thus confirming the selectivity shown inthe library screening (Figure 2). The cathepsin B sub-strate Abz-FRAK(Dnp)-OH (Cotrin et al., 2004), which hasa free carboxyl group, was cleaved by EcDcp at the R-A bond (Figure 3) with a kcat/Km value of 1680 mM-1 s-1

(Table 1). As expected, its analog Abz-FRAK(Dnp)-NH2

was resistant to hydrolysis by EcDcp due to the blockedC-terminus.

A model peptide, containing the most favorable aminoacid at P2, P1, and P19 positions, as defined by the PS-SC fluorogenic libraries, was synthesized. The resultingsubstrate Abz-FRKK(Dnp)-OH was hydrolyzed by EcDcpwith the highest kcat/Km value among the FRET substratestested (Table 1).

Hydrolysis of natural ACE substrates by EcDcp

Three natural ACE substrates, namely angiotensin I, bra-dykinin, and AcSDKP, were incubated with recombinantEcDcp for 1 h and the cleavage sites defined by LC/ESI-MS (Figure 4). Angiotensin I (DRVYIHPFHL) washydrolyzed by EcDcp at the F-H bond generating angio-tensin II (m/z 1045) and the double-charged peak at m/z522 provided further confirmation of this product. Bra-dykinin (RPPGFSPFR) was cleaved by EcDcp, at sequen-tial cleavage sites releasing the C-terminal dipeptidesF-R and S-P, as seen with ACE. Not surprisingly, the spe-cific substrate for the ACE N-domain Ac-SDKP-OH wasnot hydrolyzed, due to the presence of an acidic residueat P1 that prevents EcDcp hydrolysis as demonstratedusing the PS-SC fluorogenic libraries (Figure 2).

Influence of pH and NaCl on EcDcp catalyticactivity

The effect of pH on the hydrolysis of Abz-FRK(Dnp)P-OHby recombinant EcDcp was studied over a range of 4.0to 10.0 (Figure 5). A typical bell-shaped curve could befitted to the data with the maximum catalytic efficiencyat pH 7.5. The influence of NaCl was also determined

934 C.E.L. Cunha et al.

Article in press - uncorrected proof

Figure 2 Combinatorial fluorescent-quenched peptide libraries for scanning the subsites S3-S19.Libraries with general sequences Abz-GXXZXK(Dnp)-OH (P1), Abz-GXZRXK(Dnp)-OH (P2), Abz-GZXRXK(Dnp)-OH (P3), and Abz-GXXRZK(Dnp)-OH (P19) were incubated with recombinant EcDcp as described in the materials and methods section. The activity wasdetermined measuring the initial velocity of hydrolysis of each run. The y-axis provides the relative catalytic efficiency values (%)normalized by the best substrate in each series. The x-axis shows the specific amino acids (represented by the one-letter codes)present in the different positions that were investigated (P3 to P19). The errors are shown at the tops of the bars, each of whichrepresents the average of three determinations. According to Schechter and Berger (1967), P1, P2, and P3 are designated amino acidresidues in the N-terminal direction and P19 in the C-terminal direction from the scissile bond.

Table 1 Kinetic parameters for hydrolysis of FRET peptides byEcDcp.

Peptide Km (mM) kcat (s-1) kcat/Km

(mM-1 s-1)

Abz-FRKK(Dnp)-OH 0.26"0.008 1.62"0.06 6230Abz-FRK(Dnp)P-OH 0.29"0.01 0.81"0.03 2793Abz-LFK(Dnp)-OH 2.55"0.1 1.4"0.7 538Abz-SDK(Dnp)P-OH ResistantAbz-FRAK(Dnp)-OH 0.66"0.03 1.1"0.4 1667Abz-FRAK(Dnp)-NH2 ResistantAbz-APK(Dnp)-OH Resistant

The assays were performed at 378C in 0.1 M Tris-HCl, pH 8.0.Measurements were made as described in the materials andmethods section.

using Abz-FRK(Dnp)P-OH as substrate (Figure 5, inset).No significant chloride anion activation was detected,even at high salt concentration. The effect of NaCl on thehydrolysis of Abz-FRKK(Dnp)-OH was also tested (datanot shown) and the result was very similar to thatobtained with Abz-FRK(Dnp)P-OH.

Effect of ACE inhibitors and molecular modelingof EcDcp

Using the substrate Abz-FRK(Dnp)P-OH, we determinedthe inhibition constants for the ACE inhibitors captopriland lisinopril (Table 2). The Ki value determined forcaptopril was 3 nM, indicating that this important ACEinhibitor is also very potent for recombinant EcDcp. In

contrast, the Ki value for lisinopril was a thousand-foldhigher (4.4 mM) demonstrating a striking difference in theaffinity of this inhibitor for EcDcp. Due to the catalyticand structural similarities between EcDcp and tACE, wetested two C-domain-selective ACE inhibitors, namelythe keto-ACE (5-S-5-benzamido-4-oxo-6-phenylhexa-noyl-L-proline; Deddish et al., 1998) derivative kAW inwhich a Trp replaced the C-terminal Pro (5-S-5-benza-mido-4-oxo-6-phenylhexanoyl-L-tryptophan; Nchinda etal., 2006a; Watermeyer et al., 2008) and lisinopril-Trp(Nchinda et al., 2006b). EcDcp was inhibited by lisinopril-Trp with a Ki value (0.8 nM) ten thousand-fold lower thanthat determined for kAW (Kis22.0 mM).

Molecular modeling allowed us to suggest a basis forthe differences in inhibition potency. The modality ofbinding of the four inhibitors is similar (Figure 6). Theactive site zinc is coordinated by residues on helices a16and a17 as described in the crystal structure (Comellas-Bigler et al., 2005). The zinc coordination geometry issimilar to that of ACE and ACE2 where the zinc ion istetra-coordinated by the amide groups of two His resi-dues (His 469 and His 473 for EcDcp), the carboxyl oxy-gen of a conserved Glu located 24 residues downstream,and the inhibitor mimicking the carbonyl oxygen of thescissile peptide bond. The P1 Phe and P19 Lys of lisinoprilalign closely with the equivalent groups of the lisinopril-Trp derivative. The P1 Phe of lisinopril stacks snuglybetween Trp425 and His 601 and the NH of the P19 lysylgroup makes contact with the backbone carbonyls ofAsn447 and Gly423. The lisinopril-Trp has increased

Characterization of dipeptidyl carboxypeptidase from E. coli 935

Article in press - uncorrected proof

Figure 3 Determination of the FRET peptide cleavage site.The FRET peptides Abz-FRAK(Dnp)-NH2 (A), Abz-FRAK(Dnp)-OH (B), Abz-FRK(Dnp)P-OH (C), and Abz-LFK(Dnp)-OH (D) were incu-bated for 1 h with recombinant EcDcp at 378C in 0.1 M Tris-HCl, pH 8.0, and analyzed by LC/ESI-MS.

Figure 4 Determination of cleavage sites on ACE natural substrates.Angiotensin I (A), bradykinin (B), and AcSDKP-OH (C) were incubated for 1 h with EcDcp at 378C in 0.1 M Tris-HCl, pH 8.0, and thecleavage sites determined by LC/ESI-MS.

Figure 5 pH dependence and NaCl effect (inset) on Abz-FRK(Dnp)P-OH hydrolysis by EcDcp.The hydrolysis conditions are described in the materials and methods section.

936 C.E.L. Cunha et al.

Article in press - uncorrected proof

Table 2 Inhibition constants for recombinant EcDcp.

Inhibitor Ki (M)

Captopril (3"0.08)=10-9

Lisinopril (4.4"0.1)=10-6

Lisinopril-Trp (0.8"0.07)=10-9

kAW (22"0.9)=10-6

The assays were performed at 378C in 0.1 M Tris-HCl, pH 8.0,using Abz-FRK(Dnp)P-OH as substrate. Measurements weremade as described in the materials and methods section.

Figure 6 Molecular docking of inhibitors in the DCP active site.(A) Alignment of the docked structures of EcDcp with captopril(green sticks), lisinopril (yellow sticks), lisinopril-Trp (purplesticks), and kAW (gray sticks). Structures are shown by pairwisealignment of the a-carbons of the different EcDcp models. Thezinc ion is shown as a blue-gray sphere. (B) A surface represen-tation of the EcDcp active site with lisinopril-Trp (purple sticks)docked into the active site. The zinc ion is shown as a greensphere. The Figure was prepared using Pymol (DeLano Scien-tific, Palo Alto, CA, USA).

contact with the S29 pocket via its Trp-like P29 group whichis surrounded by a cage of tyrosine and leucine residues(Leu550, Tyr611 and Tyr614). Moreover, Tyr611 is withinhydrogen bonding distance of the indole nitrogen of theP29 Trp. The P2 aminobenzoyl group of kAW is con-strained between Trp425 and Ala600 in the S2 pocket.

Discussion

Although EcDcp and ACE have very different primarysequences, crystallographic studies provided evidence

for striking similarities in their three-dimensional struc-tures (Comellas-Bigler et al., 2005). In the present work,we describe several catalytic properties that are sharedby both enzymes and demonstrate that, despite belong-ing to different enzyme families and having low sequencehomology, they display common hydrolytic characteris-tics. The inferred phylogeny shows a clear distinctionbetween two monophyletic groups of sequences fromthe M2 and M3 families. LdDcp, EcDcp, and MIP whichare found in phylogenetically distant species are mono-phyletic the M3 family, suggesting a common ancestor.This phylogeny corroborates the distinct origin of bothEcDcp and ACE.

We tested a combinatorial fluorogenic substrate librarypreviously used for ACE studies (Bersanetti et al., 2004)to define the specificity of EcDcp. Our results show thatEcDcp and sACE have very similar preferences at P3, P2,P1, and P19 positions. Both enzymes showed a preferencefor a lysine residue at P19 and both were unable to hydro-lyze a substrate with proline in this position. We wereable, however, to identify some selectivity in EcDcp cata-lytic activity owing to its inability to hydrolyze substrateswith an Asp residue in P1 position, unlike sACE that wasable to hydrolyze them (Bersanetti et al., 2004).

A number of enzymes related to EcDcp have beendescribed recently. LdDcp is able to hydrolyze Hip-His-Leu and is inhibited by captopril (Goyal et al., 2006) andXcACE is able to hydrolyze angiotensin I and the N-selective ACE substrate Ac-SDKP (Riviere et al., 2007).In the present paper, we demonstrate that EcDcp is ableto hydrolyze angiotensin I, as also shown for XcACE(Riviere et al., 2007). In addition, we have shown thatEcDcp cleaves bradykinin as previously described byYaron (1976). Unlike XcACE (Riviere et al., 2007), how-ever, EcDcp is unable to hydrolyze the natural ACEN-domain-specific substrate Ac-SDKP-OH and its fluo-rogenic analog Abz-SDK(Dnp)P-OH due to the acidic res-idue in its P1 position. Related to the FRET substratesAbz-FRK(Dnp)P-OH and Abz-LFK(Dnp)-OH, EcDcp fol-lows the same pattern of hydrolysis presented by recom-binant ACE C-domain and testis ACE, but cleaved thepeptides with at least one order of magnitude lowerkcat/Km values due to the low kcat presented. The valuesfor hydrolysis of Abz-FRK(Dnp)P-OH by the recombinantACE C-domain and testis ACE were determined to be25.6 mM-1 s-1 (Araujo et al., 2000) and 20.5 mM-1 s-1 (Ber-sanetti et al., 2004), respectively. Bersanetti et al. (2004)described the kcat/Km values for the hydrolysis of Abz-LFK(Dnp)-OH by recombinant ACE C-domain and testisACE to be 36.7 mM-1 s-1 and 16.5 mM-1 s-1, respectively.

By using inhibition studies we demonstrated that thetwo classical ACE inhibitors lisinopril and captopril wereable to inhibit EcDcp catalytic activity. However, unlikethat described for recombinant ACE (Wei et al., 1992) andtestis ACE (Grinshtein et al., 1999), which are inhibitedby both in nanomolar range, captopril was a much betterEcDcp inhibitor (Kis3 nM) than lisinopril (Kis4.4 mM). TwoC-domain-selective ACE inhibitors kAW and lisinopril-Trpwere also tested with EcDcp. The Ki value obtained forkAW (22.0 mM) was almost four orders of magnitudegreater than that found for lisinopril-Trp (Kis0.8 nM).Molecular modeling was used to examine the basis for

Characterization of dipeptidyl carboxypeptidase from E. coli 937

Article in press - uncorrected proof

the differences in inhibition potency. The S1 pocket ofEcDcp is quite narrow and the P1 Phe of lisinopril fitssnugly stacking between Trp425 and His 601. The lisi-nopril-Trp inhibitor has a P29 Trp that makes more exten-sive contact with the S29 pocket (Figure 6B) explaining itsvastly improved Ki value. In contrast, the significantlyhigher Ki value determined for the ketomethylene inhibi-tor kAW could be due to its weaker zinc binding groupand the steric hindrance caused by its bulky amino ben-zoyl group and Trp425 in the constrained S2 pocket. Cap-topril makes the least contact with the obligatory bindingsite of all four inhibitors and relies on its strong interac-tion with zinc via the sulfhydryl group, and its ionic inter-action between the prolyl carboxylate and Arg593.Lys511 in tACE is the positively charged counterpart inthe S29 pocket that anchors lisinopril and captopril(Natesh et al., 2003).

In conclusion, we have shown that EcDcp and ACEhave many enzymatic similarities, despite their differenc-es in primary sequence. Moreover, the data from ourkinetic study are in good agreement with the resultsobtained from molecular docking experiments regardingthe close similarity of EcDcp and tACE, in spite of theirphylogenetically distance.

Materials and methods

Cloning of the Dcp gene

The expression vector pHis3-EcDcp was constructed by inser-tion of the Dcp gene into pHis3 plasmid, a modified pET vector.The Dcp gene was cloned by PCR from total DNA of E. colistrain BL21 (DE3) pLysS, extracted with PureLink Genomic DNAPurification Kit (Invitrogen, Carlsbad, CA, USA). The PCR mix-ture consisted of 200 mM deoxynucleoside triphosphates, 2 mM

MgCl2, 50 mM KCl, 20 mM Tris-HCl (pH 8.4), 1 U Taq DNA poly-merase, and 50 pmol of each of the primers (59-ATG ACA ACAATG AAT CCT TTC CT-39 and 59-TTA TAT GTT CAA GCC ACGATG TT-39). The PCR temperature cycle was 948C for 5 min,followed by 35 cycles at 948C for 30 s, 508C for 1 min, 728C for2 min, and finally 10 min at 728C. The amplified fragment of2100 bp was recovered from 1% agarose gel using BIOCLEANfor purification of DNA bands (Biotools, Sao Paulo, Brazil) andcloned on pGEM-T easy vector (Promega, Madison, WI, USA).The cloned gene fragment was excised from the plasmid bydigestion with EcoRI (Invitrogen), ligated into the EcoRI site ofpHis3 vector and sequenced. For expression, BL21 (DE3) pLysSwas transformed by heat shock with pHis3-EcDcp and thelysate from grown bacteria was prepared as described below.

Expression and purification of EcDcp recombinantenzyme

Briefly, BL21 (DE3) pLysS bacteria transformed by heat shockwith plasmid pHis3-EcDcp were grown at 378C for 16 h, withshaking, in Luria-Bertani medium, with ampicillin (100 mg/ml)and chloramphenicol (50 mg/ml). BL21(DE3) pLysS/pHis3-EcDCP cells were reinoculated in fresh medium and grown withantibiotic selection to A600s0.6, when the expression of therecombinant protein was induced with 0.5 mM isopropyl b-D-thiogalactoside (IPTG), holding for 4 h at 308C. Bacterial cultureswere centrifuged at 1075 g for 20 min at 48C, resuspended in50 mM Tris-HCl, pH 7.4, then lysed in a French Press at 1.3=108

Pa. After removal of the bacterial debris by centrifugation, thesupernatants were used in the kinetic assays and the recombi-

nant proteins purified in a Hi-Trap Ni-column (GE Healthcare,Uppsala, Sweden). The protein concentration was determinedas previously described (Bradford, 1976) using bovine serumalbumin as standard. Bacterial lysates and the recombinant pro-tein preparations were analyzed by polyacrylamide gel electro-phoresis after staining with Coomassie Blue R-250 (Bio-RadLaboratories, Richmond, CA, USA).

Synthesis of peptide libraries

PS-SC libraries containing ortho-aminobenzoic (Abz) and dini-trophenyl (Dnp) as fluorescence donor/acceptor pair were syn-thesized as previously described (Bersanetti et al., 2004; Cotrinet al., 2004). In preliminary assays, a library with the generalstructure Abz-GXXZXK-(Dnp)-OH was used, in which the Z posi-tion was successively replaced by one of the 19 natural aminoacids (cysteine excluded) and the X represents a randomly incor-porated residue introduced by coupling a balanced mixture of19 amino acids (Herman et al., 1996). Three other libraries wereprepared with general structures Abz-GXXRZK-(Dnp)-OH, Abz-GXZRXK(Dnp)-OH, and Abz-GZXRXK-(Dnp)-OH, in which the P1

position was pre-fixed as Arg and a second fixed position(ZsP19, P2, and P3, respectively) included one of the 19 naturalamino acids, the other positions being randomized (Cys exclud-ed). The stock solution of each peptide mixture was prepared indimethyl sulfoxide (DMSO), and the concentrations determinedusing the absorbance of the Dnp group, with a molar extinctioncoefficient at 365 nm of 17 300 M-1 cm-1.

Synthesis of FRET peptides

The FRET peptides containing the Dnp group coupled to the´-NH2 of a Lys residue were synthesized by solid phase meth-odology using Fmoc-Lys(Dnp)-OH (Araujo et al., 2000). The pep-tides were purified by semi-preparative high pressure liquidchromatography (HPLC), monitored by absorbance at 220 nmand by fluorescence emission at 420 nm following excitationat 320 nm. The molecular mass of purified peptides waschecked by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (TofSpec-E, Micromass,Manchester, UK). All peptides were stocked in DMSO and theconcentrations measured using the molar extinction coefficientof Dnp group (´365s17 300 M-1 cm-1).

Screening of peptide libraries

The hydrolysis of library solutions by EcDcp was performed in0.1 M Tris-HCl buffer (pH 8.0). Enzymatic activity was continu-ously monitored measuring the fluorescence in a Hitachi F-2000fluorimeter (lexs320 nm; lems420 nm) with controlled temper-ature (378C), in 1 ml final volume. The concentration of the pep-tide library solutions was calibrated to perform all assays at100 nM. The reactions were followed over 300 s and the initiallinear increment of fluorescence with time was taken as thevelocity of hydrolysis as previously described (Bersanetti et al.,2004; Cotrin et al., 2004). Measurements were made in triplicateand differences were less than 5%.

Determination of kinetic parameters

Assays with FRET peptides were performed at 378C in 0.1 M

Tris-HCl (pH 8.0). Enzymatic activity was continuously monitoredin a Hitachi F-2000 fluorimeter, measuring the fluorescence atlems420 nm and lexs320 nm. The enzyme concentration waschosen allowing 5% of the substrate present to be hydrolyzed.The slopes were converted into micromoles of substrate hydro-lyzed per minute. The inner filter effect was corrected using anempirical equation as described by Araujo et al. (2000). The

938 C.E.L. Cunha et al.

Article in press - uncorrected proof

kinetic parameters Km and kcat were calculated by the non-linearregression data analysis Grafit program (Leaterbarrow, 2001).The kcat/Km values were calculated as the ratio of these twoparameters.

Optimum pH and NaCl influence on EcDcpcatalytic activity

The pH dependence in EcDcp hydrolysis was determined usingAbz-FRK(Dnp)P-OH as substrate in a pH range of 4.0–10.0. Afour-component buffer comprising 25 mM glycine, 25 mM aceticacid, 25 mM 2-(N-morpholino)ethanesulfonic acid (MES) and75 mM Tris (Polgar, 1999) was used. The enzymatic activity wasmonitored at 378C using the fluorimetric assay described aboveand the apparent second rate constant kcat/Km was determinedat low substrate concentration where the reaction followed thefirst-order conditions. To study the influence of NaCl (0–0.5 M)on the catalytic activity of EcDcp, the enzyme was incubatedwith Abz-FRK(Dnp)P-OH, at 378C in 0.1 M Tris-HCl (pH 8.0). Thehydrolysis was monitored as described above and the apparentsecond rate constant kcat/Km was determined.

Determination of the cleavage sites

For determination of the scissile bonds, EcDcp was incubatedfor 1 h with natural ACE substrates and FRET peptides in 0.1 M

Tris-HCl (pH 8.0), at 378C, in 300 ml final volume. The reactionwas stopped by addition of trifluoroacetic acid (TFA; 10%). Theincubation medium was analyzed by LC/ESI-MS using aShimadzu apparatus, model 2010 with a SPD-20A ShimadzuUV/vis detector (Tokyo, Japan) and RF-10AXL fluorescencedetector (Shimazu, Tokyo, Japan) coupled with an UltrasphereC-18 column (5 mm, 4.6=250 mm). A linear gradient of 10–80%of solvent B was run for 20 min after 8 min of isocratic flow.Solvent A: 0.1% TFA/H2O; solvent B: 0.1% TFA in CH3CN/H2O(75:25).

Inhibition studies

EcDcp inhibition studies were performed using the substrateAbz-FRK(Dnp)P-OH and appropriate concentrations of the ACEinhibitors captopril and lisinopril. Two C-domain-selective ACEinhibitors, kAW w5-S-5-Benzamido-4-oxo-6-phenylhexanoyl-L-tryptophanx (Nchinda et al., 2006a; Watermeyer et al., 2008) andlisinopril-Trp wN2-(R,S)-(1-carbonyl-3-phenylpropyl)x-L-lysyl-L-tryptophan (Nchinda et al., 2006b) were also tested. The assayswere carried out in 0.1 M Tris-HCl (pH 8.0), at 378C, after 5 minpre-incubation of the enzyme with the inhibitor. Fluorescenceemission was continuously measured and apparent inhibitionconstant (Kiapp) values were obtained using the equation:

v wIx0s1qv K1 iapp

where Vo and Vi are the velocity of less than 2% of substratehydrolysis in the absence and in the presence of different inhib-itor concentrations wIx, respectively. The assays were performedin duplicate and the Ki parameters were obtained from theequation:

KiappKsi wSx1q

Km

The Ki values for the ACE inhibitor were calculated by thetight-binding titration data analysis Grafit program (Leaterbar-row, 2001).

Phylogeny studies

Sequence data from M2 and M3 families of metallopeptidaseswere obtained from NCBI (v161) and aligned using three distinctprograms, MAFFTv6.05 (Katoh et al., 2005), MUSCLE (Edgar,2004), and Probcons (Do et al., 2005), in order to obtain a con-sensus alignment. The phylogeny of this aligned sequence datawas inferred with HYPHYv0.99 (Pond et al., 2005) using Likeli-hood and the General Reversible model (REV) with fixedsubstitution rates. A comparison of 19 different amino acid sub-stitution models, using HYPHY, suggested the Akaike Informa-tion Criterion (Akaike, 1974) as the most appropriate for the data.A new phylogenetic inference was performed with the parame-ters obtained from the comparison of substitution models. Thephylogenetic tree topology was verified by inferring the phylog-eny using two other methodologies with the same substitutionmodel. MrBayes (Huelsenbeck and Ronquist, 2001; Ronquistand Huelsenbeck, 2003) and TreePuzzle (Schmidt et al., 2002)inferred phylogenies were compared in order to build a consen-sual tree.

Modeling studies

Molecular modeling experiments were carried out using the DIS-COVER module of INSIGHT II package (Version 98.0; AccelrysInc., San Diego, CA, USA) on a Silicon Graphics Octane 1 work-station. The starting structure was the X-ray crystal structure ofE. coli dipeptidyl carboxypeptidase (EcDcp) complexed with thedipeptide as reported by Comellas-Bigler et al. (2005). Afterremoving the crystallographic water molecules and addinghydrogens, the Consistent Valence Force Field (cvff) and theExtensible Valence Force Field (esff, metal adapted) were usedin all energy minimizations and dynamic runs. Initially, the con-jugate gradient algorithm was used to run 1000 steps of mini-mization. Then 3000 cycles of molecular dynamics in a constanttemperature/constant volume (NVT) ensemble at 300 K was per-formed and the final conformation from molecular dynamics wasminimized another 3000 steps to produce the final structure. Allcalculations were carried out in a dielectric constant of 1.00 andcut-off distance of 9.50 A. The structures of the ACE inhibitorscaptopril, lisinopril, lisinopril-Trp, and kAW were generated withstandard bond lengths and angles using the builder tool ofINSIGHT II software (Accelrys Inc.) and then minimized. The ini-tial position of the ACE inhibitors in the active sites of the EcDcpwas obtained by superimposing and positioning the importantpharmacophoric groups on the corresponding atoms of thedipeptide in the EcDcp-dipeptide complex. After removal of thereference dipeptide, the structure of the EcDcp-ACE inhibitorcomplex was refined by running energy minimizations andmolecular dynamics.

Acknowledgments

This work was supported by the Brazilian Agencies Fundacao¸de Amparo a Pesquisa do Estado de Sao Paulo (FAPESP) andConselho Nacional de Desenvolvimento Cientıfico e Tecnologico(CNPq) and by the Wellcome Trust, UK and National ResearchFoundation.

References

Akaike, H. (1974). A new look at the statistical model identifi-cation. IEEE Trans. Autom. Contr. 19, 716–723.

Araujo, M.C., Melo, R.L., Cesari, M.H., Juliano, M.A., Juliano, L.,and Carmona, A.K. (2000). Peptidase specificity characteri-zation of C- and N-terminal catalytic sites of angiotensin I-converting enzyme. Biochemistry 39, 8519–8525.

Characterization of dipeptidyl carboxypeptidase from E. coli 939

Article in press - uncorrected proof

Barrett, A.J. (2004). Bioinformatics of proteases in the MEROPSdatabase. Curr. Opin. Drug Discov. Dev. 7, 334–341.

Bersanetti, P.A., Andrade, M.C., Casarini, D.E., Juliano, M.A,Nchinda, A.T., Sturrock, E.D., Juliano, L., and Carmona, A.K.(2004). Positional-scanning combinatorial libraries of fluores-cence resonance energy transfer peptides for defining sub-strate specificity of the angiotensin I-converting enzyme anddevelopment of selective C-domain substrates. Biochemistry50, 21–43.

Bradford, M.M. (1976). A rapid and sensitive method for thequantification of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72, 248–254.

Chu, T.G. and Orlowski, M. (1984). Active site directed N-car-boxymethyl peptide inhibitors of a soluble metalloendopep-tidase from rat brain. Biochemistry 23, 3598–3603.

Comellas-Bigler, M., Lang, R., Bode, W., and Maskos, K. (2005).Crystal structure of the E. coli dipeptidyl carboxypeptidaseDcp: further indication of a ligand-dependent hinge move-ment mechanism. J. Mol. Biol. 349, 99–112.

Cotrin, S.S., Puzer, L., de Souza Judice, W.A., Juliano, L., Car-mona, A.K., and Juliano, M.A. (2004). Positional-scanningcombinatorial libraries of fluorescence resonance energytransfer peptides to define substrate specificity of carboxy-dipeptidases: assays with human cathepsin B. Anal. Bio-chem. 335, 244–252.

Deddish, P.A., Marcic, B., Jackman, H.L., Wang, H.Z., Skidgel,R.A., and Erdos, E.G. (1998). N-domain-specific substrateand C-domain inhibitors of angiotensin-converting enzyme:angiotensin-(1–7) and keto-ACE. Hypertension 31, 912–917.

DeLong, E.F. and Pace, N.R. (2001). Environmental diversity ofbacteria and archaea. Syst. Biol. 50, 470–478.

Deutch, C.E. and Soffer, R.L. (1978). Escherichia coli mutantdefective in dipeptidyl carboxypeptidase. Proc. Natl. Acad.Sci. USA 75, 5998–6001.

Do, C.B., Mahabhashyam, M.S., Brudno, M., and Batzoglou, S.(2005). ProbCons: probabilistic consistency-based multiplesequence alignment. Genome Res. 15, 330–340.

Donoghue, M., Hsieh, F., Baronas, E., Godbout, K., Gosselin, M.,Stagliano, N., Donovan, M., Woolf, B., Robison, K., Jeyasee-lan, R., et al. (2000). A novel angiotensin-converting enzyme-related carboxypeptidase (ACE2) converts angiotensin I toangiotensin 1–9. Circ. Res. 87, E1–E9.

Edgar, R.C. (2004). MUSCLE: a multiple sequence alignmentmethod with reduced time and space complexity. BMCBioinformatics 5, 113.

Ehlers, M.R.W., Fox, E.A., Strydom, D.J., and Riordan, J.F.(1989). Molecular cloning of human testicular angiotensin-converting enzyme: the testis isozyme is identical to the C-terminal half of endothelial angiotensin-converting enzyme.Proc. Natl. Acad. Sci. USA 86, 7741–7745.

Gottesman, S. (1996). Proteases and their targets in Escherichiacoli. Annu. Rev. Genet. 30, 465–506.

Goyal, N., Duncan, R., Selvapandiyan, A., Debrabant, A., Baig,M.S., and Nakhasi, H.L. (2006). Cloning and characterizationof angiotensin converting enzyme related dipeptidylcarboxy-peptidase from Leishmania donovani. Mol. Biochem. Para-sitol. 145, 147–157.

Grinshtein, S.V., Binevski, P.V., Gomazkov, O.A., Pozdnev, V.F.,Nikolskaya, I.I., and Kost, O.A. (1999). Inhibitor analysis ofangiotensin I-converting and kinin-degrading activities ofbovine lung and testicular angiotensin-converting enzyme.Biochemistry (Moscow) 64, 938–944.

Grunenfelder, B., Rummel, G., Vohradsky, J., Roger, D., Lange-nand, H., and Jenal, U. (2001). Proteomic analysis of the bac-terial cell cycle. Proc. Natl. Acad. Sci. USA 98, 4681–4686.

Henrich, B., Becker, S., Schroeder, U., and Plapp, R. (1993). dcpgene of Escherichia coli: cloning, sequencing, transcriptmapping and characterization of the gene product. J. Bac-teriol. 175, 7290–7300.

Herman, L.W., Tarr, G., and Kates, S.A. (1996). Optimization ofthe synthesis of peptide combinatorial libraries using a one-pot-method. Mol. Diversity 2, 147–155.

Huelsenbeck, J.P. and Ronquist, F. (2001). MRBAYES: Bayesianinference of phylogenetic trees. Bioinformatics 17, 754–755.

Katoh, K., Kuma, K., Miyata, T., and Toh, H. (2005). Improvementin the accuracy of multiple sequence alignment programMAFFT. Genome Inform. 16, 22–33.

Kumar, R.S., Kusari, J., Roy, S.N., Soffer, R.L., and Sen, G.C.(1989). Structure of testicular angiotensin-convertingenzyme. A segmental mosaic isozyme. J. Biol. Chem. 264,16754–16758.

Lattion, A.L., Soubrier, F., Allegrini, J., Hubert, C., Corvol, P., andAlhenc-Gelas, F. (1989). The testicular transcript of the angi-otensin I-converting enzyme encodes for the ancestral, non-duplicated form of the enzyme. FEBS Lett. 252, 99–104.

Leaterbarrow, R.J. (2001). GraFit version 5. Erytacus SoftwareLtd, Staines, UK.

Maurizi, M.R. (1992). Proteases and protein degradation inEscherichia coli. Experientia 48, 178–201.

Miller, C.G. (1975). Peptidases and proteases of Escherichia coliand Salmonella typhimurium. Annu. Rev. Microbiol. 29,485–504.

Natesh, R., Schwager, S.L., Sturrock, E.D., and Acharya, K.R.(2003). Crystal structure of the human angiotensin-convertingenzyme-lisinopril complex. Nature 421, 551–554.

Nchinda, A.T., Chibale, K., Redelinghuys, P., and Sturrock, E.D.(2006a). Synthesis of novel keto-ACE analogues as domain-selective angiotensin-I converting enzyme inhibitors. Bioorg.Med. Chem. Lett. 16, 4612–4615.

Nchinda, A.T., Chibale, K., Redelinghuys, P., and Sturrock, E.D.(2006b). Synthesis and molecular modeling of a lisinopril-tryptophan analogue inhibitor of angiotensin I-convertingenzyme. Bioorg. Med. Chem. Lett. 16, 4616–4619.

Polgar, L. (1999). Oligopeptidase B: a new type of serine pepti-dase with a unique substrate-dependent temperature sensi-tivity. Biochemistry 38, 15548–15555.

Pond, S.L., Frost, S.D., and Muse, S.V. (2005). HyPhy: hypoth-esis testing using phylogenies. Bioinformatics 21, 676–679.

Riviere, G., Michaud, A., Corradi, H.R., Sturrock, E.D., RaviAcharya, K., Cogez, V., Bohin, J.P., Vieau, D., and Corvol, P.(2007). Characterization of the first angiotensin-convertinglike enzyme in bacteria: ancestor ACE is already active. Gene399, 81–90.

Ronquist, F. and Huelsenbeck, J.P. (2003). MrBayes 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19, 1572–1574.

Schechter, I. and Berger, A. (1967). On the size of the active sitein proteases. I. Papain. Biochem. Biophys. Res. Commun.27, 157–162.

Schmidt, H.A., Strimmer, K., Vingron, M., and von Haeseler, A.(2002). TREE-PUZZLE: maximum likelihood phylogeneticanalysis using quartets and parallel computing. Bioinforma-tics 18, 502–504.

Skeggs, L.T., Kahn, J.R., and Shumway, N.P. (1956). Preparationand function of the hypertensin converting enzyme. J. Exp.Med. 103, 295–299.

Tipnis, S.R., Hooper, N.M., Hyde, R., Karran, E., Christie, G., andTurner, A.J. (2000). A human homolog of angiotensin-con-verting enzyme. Cloning and functional expression as a cap-topril-insensitive carboxypeptidase. J. Biol. Chem. 275,33238–33243.

Turner, A.J. and Hooper, N.M. (2002). The angiotensin convertingenzyme gene family: genomics and pharmacology. TrendsPharmacol. Sci. 23, 177–183.

Watermeyer, J.M., Kroger, W.L., O’Neill, H.G., Sewell, B.T., andSturrock, E.D. (2008). Probing the basis of domain-depend-ent inhibition using novel ketone inhibitors of angiotensin-converting enzyme. Biochemistry 47, 5942–5950.

Wei, L., Alhenc-Gelas, F., Soubrier, F., Michaud, A., Corvol, P.,and Clauser, E. (1991). Expression and characterization ofrecombinant human angiotensin I-converting enzyme. Evi-

940 C.E.L. Cunha et al.

Article in press - uncorrected proof

dence for a C-terminal transmembrane anchor and for a pro-teolytic processing of the secreted recombinant and plasmaenzymes. J. Biol. Chem. 266, 5540–5546.

Wei, L., Clauser, E., Alhenc-Gelas, F., and Corvol, P. (1992). Thetwo homologous domains of human angiotensin I-convertingenzyme interact differently with competitive inhibitors. J. Biol.Chem. 267, 13398–13405.

Whelan, S. and Goldman, N. (2001). A general empirical modelof protein evolution derived from multiple protein familiesusing a maximum-likelihood approach. Mol. Biol. Evol. 18,691–699.

Wright, R., Sthephens, C., Zweiger, G., Shapiro, L., and Alley,M.R. (1996). Caulobacter Lon protease has a critical role in

cell-cycle control of DNA methylation. Genes Dev. 10,1532–1542.

Yang, H.Y.T., Erdos, E.G., and Levin, Y. (1970). A dipeptidyl car-boxypeptidase that converts angiotensin I and inactivatesbradykinin. Biochim. Biophys. Acta 214, 374–376.

Yaron, A. (1976). Dipeptidyl carboxypeptidase from Escherichiacoli. Methods Enzymol. 50, 599–610.

Yaron, A., Mlynar, D., and Berger, A. (1972). A dipeptidyl car-boxypeptidase from E. coli. Biochem. Biophys. Res.Commun. 47, 897–902.

Received February 19, 2009; accepted May 13, 2009