THEJOURNALOF BIOLOGICAL
CHEMISTRY
Vol. 267, No. 19, Issue of July 5, pp. 13398-13405, 1992
Printed in U.S.A.
0 1992 by The American Society for Biochemistry and Molecular Biology, Inc.
The TwoHomologous Domains of Human Angiotensin I-converting
Enzyme Interact Differently withCompetitive Inhibitors*
(Received for publication, December 23,
1991)
Lei Wei, Eric Clauser, Frangois
Alhenc-Gelas, and PierreCorvol
From the Institut de la Sante et de la Recherche M6dicale Unit6 36, 3 rue d’Ulm, 75005 Paris, France
13398
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
TheendothelialangiotensinI-convertingenzyme
is an unusual zinc-metallopeptidase in that it is activatedby
(ACE; EC 3.4.15.1) has recentlybeen shown to contain chloride and lacks a narrow in vitrosubstrate specificity (3).
two large homologous domains (called here the N and ACE is a widely distributed peptidase, predominantly exC domains), each being a zinc-dependent dipeptidyl
pressed as a membrane-bound ectoenzyme in vascular endocarboxypeptidase. To further characterize the two ac-thelial cells and also in several other cell types including
tive sites of ACE, we have investigated their interac- absorptiveepithelial cells, neuroepithelial cells, and male
tion with four competitive ACE inhibitors, which are germinal cells (4-6).
all potent antihypertensive drugs. The bindingof r3H]
Inhibition of ACE is a widely used approach in the treattrandolaprilat to the two active sites was examined
using the wild-type ACE and four ACE mutants each ment of hypertension. The firstavailable competitive inhibitors of ACE were the naturally occurring peptides in snake
containing only one intact domain, the other domain
being either deleted or inactivated by point mutation venom. Clinical studies using the nonapeptide teprotide, the
of the zinc-coordinatinghistidines. In contrast with allmost efficient of these snake venom peptides in uiuo, demonthe previous studies, which suggested the presenceof strated the potential of ACE inhibitors as antihypertensive
a single high affinity inhibitor binding site
in ACE, the drugs (7).Highly potent inhibitorsof ACE which can be taken
present study shows that both the
N and C domainsof orally have subsequently been developed. The first of these,
ACE contain a high affinity inhibitor binding site(Kn captopril, was designed with the help of a theoretical model
= 3 and 1 X 10”’ M, respectively, at pH 7.5, 4 “C, and of the active site of ACE, which was based on its presumed
known active site of carboxypeptidase A and
100 mM NaCl). Chloride stabilizes the enzyme-inhibi- similarity to the
tor complex for each domain primarily by slowingits also withreference to the C-terminalsequences of the venom
dissociation rate, as the k l values of the N and C peptides which compete with substrates (8). In this model,
domains are markedly decreased (about30-and 1100- captopril coordinates strongly thezinc atom in theactive site
fold, respectively) by 300 mM NaCl. At high chloride of ACE by its sulfhydrylgroup, which is coupled tothe
concentrations, the chloride effect
is much greater for dipeptide Ala-Pro responsible for the specific side chain inthe C domain than for the N domain resulting in a teractions of this inhibitor with ACE (8). Numerous other
higher affinity of this inhibitor for the C domain. In inhibitors have subsequently been synthesized usingthe same
addition, the inhibitory potency of captopril (C), ena- model but differing in the natureof their zinc-binding ligands
laprilat (E), and lisinopril
(L) for each domain was and other interaction groups. Enalaprilat and lisinopril, for
assayed by hydrolysis of Hip-His-Leu. Their Ki values example, contain acarboxylategroup for coordinating the
forthetwo
domains are allwithinthenanomolar
zinc atom and are analogous to Phe-Ala-Pro and Phe-Lysrange, indicating that they are highly
all
potent inhibPro,
respectively (9). The effectiveness of these inhibitors in
itors forboth domains. However, their relativepotenof hypertension and congestive heart failure
the
treatment
cies are different for the
C domain (L > E > C) and the
has
been
demonstrated
in numerousclinical studies (10). The
N domain (C > E > L). The different inhibitor binding
availability
of
high
affinity
ACE inhibitors has also provided
properties of the two domains observed in the present
studyprovidestrongevidenceforthepresence
of powerful tools for studing the structure, function, and tissular
structural differences between the two active sitesof distribution of this enzyme.
Several studies have been undertaken to examine the moACE.
lecular mechanisms of ACE-inhibitor interactions by enzymatic kinetic (11) and radiolabeledligandbinding (12-14)
approaches. All thesestudies suggested the presence of a
Angiotensin I-converting enzyme (kininase 11; EC 3.4.15.1) single class of high affinitybindingsitein
ACE, witha
(ACE)’ plays an important role in blood pressure regulation. stoichiometry of one, a result in agreementwith the previous
I t is a dipeptidyl carboxypeptidase whichconverts angiotensin observation of onezinc atom per ACE molecule (15, 16).
I intothepotent
vasopressor peptideangiotensin I1 and However, recently, the primary structure of human endotheinactivates the vasodepressor peptide bradykinin (1, 2). ACE lial ACE (17) and then thatof mouse kidney ACE
(18)have
been
determined
by
cDNA
cloning,
disclosing
the
presence
of
* This work was supported by the Institut National de la Sante et
two large homologous domains. Each domain contains the
de la Recherche MGdicale and by a grant from the Bristol-MyersSquibb Institute for Medical Research. The costs of publication of consensus sequence of zinc-metallopeptidases His-Glu-X-Xthis article were defrayed in part by the payment of page charges.
His and thereforea putative catalytic center.By site-directed
This articlemust therefore be hereby marked“aduertisement”in
mutagenesis, we have demonstrated that both domains
(called
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
here
the
N
and
C
domains)
1) bear an independently func’ The abbreviations used are: ACE, angiotensin I-converting enzyme; Hip, hippuryl; Hepes, 4-(2-hydroxyethyl)-l-piperazineethane-tional active site, 2) possess a zinc-dependent dipeptidyl carsulfonic acid; Fa, 2-furanacryloyl; LH-RH, luteinizing hormone-re- boxypeptidaseactivity, and 3) are sensitive to competitive
leasing hormone.
ACE inhibitors (19). This study also established that thetwo
Inhibitor Binding Sites
of Angiotensin I-converting Enzyme
13399
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
pril was a gift from the Bristol-Myers-Squibb Institute, and enalaprilat and lisinopril were obtained from Merck Sharp and Dohme
Research Laboratories.
Enzyme InhibitionAssays-Enzyme activities were measured using
Hip-His-Leu (Biochem, Switzerland) as substrate as previously described (20). The concentration of the full-length wild-type or mutated enzymes was determined by direct radioimmunoassay and also
by calculation from enzyme activity measurements using the previously determined kinetic parameters for each enzyme (19, 20). The
concentration of the truncated mutants was calculated from enzyme
activity measurements by comparison with the full-length mutants,
as the catalytic properties
of the truncated mutants are indistinguishable from those of the corresponding full-length mutants (19).
The potency of each inhibitor toward each
enzyme was determined
by establishing dose-dependent inhibition curves a t equilibrium. As
all the compounds tested are
competitive slow tight binding inhibitors
of ACE, enzyme and inhibitor were incubated together for 1 h at
37 "C before the addition of substrate toallow a steady-state equilibrium between the enzyme and inhibitor to be established. The incubations were conductedinduplicateorintriplicate
withvarious
concentrations of inhibitor in 225 pl of 100 mM potassium phosphate
buffer, pH 8.3, containing 10 p~ ZnS04, 20 or 300 mM NaC1, 1 mg/
MATERIALSANDMETHODS
ml bovine serum albumin, and 1 X 10"' or 0.1 X 10"' M of enzyme.
Enzymes-Wild type and mutated ACE were produced in stable
A 1-h incubation was found to be sufficient to reach equilibrium for
transfected Chinese hamster ovary cells, as previously described (19,
each enzyme and inhibitor tested, a t all inhibitor concentrations.
20). The wild-type recombinant ACE is enzymatically indistinguishEnzyme activities in the absence of inhibitor did not change during
able from human kidney ACE (20). The two truncated mutants (N
control incubation of a t least 3 h indicating that the enzyme was
N or C domain, the residues
and C fragments) contain either the
stable during theincubation.
738-1277 or 5-571 of ACE being deleted, respectively(the numbering
To determine the residual free enzyme concentration, the enzyof ACE residues is according to Soubrier etal. (17)) (Fig. 1).The two
matic reaction was then started by adding 25 p1of the substrate
full-length mutants ( A C E K ~
365,~ ACEK,,,
~,
963) contain only one funcsolution to give a final concentration of 250 pM (corresponding to
tional domain, the two zinc-binding histidines of either the N or C
0.12-0.15 K,,,). Initial velocities were measured during the hydrolysis
domain being substituted by lysine residues (Fig. 1). The study was
of the first 5% of substrate. Under these conditions, the
enzyme
carried outusing purified enzymesexcept in thecase of the truncated
concentration in the assaywas too low (c0.2 K,) to induce significant
mutant, N fragment, where the dialyzed serum-free medium of the
free inhibitor depletion, and the influence of substrate and dilution
transfected cells was used (19). The purified wild-type recombinant
can be considered negligible. ICso. theconcentration of inhibitor
ACE and the full-length mutants have an apparent molecular mass
needed to produce 50% inhibition, can thus be considered to correof 170 kDa on sodium dodecyl sulfate-polyacrylamide gel electrophospondto K,, theinhibitionconstant,inthe
case of competitive
resis (19, 20). The molecular mass of the N and C fragments is 135
inhibition (22).
and 100 kDa, respectively (19).
PHITrandolaprilat Binding Assay-The dissociation constant(s)
Inhibitors-Trandolaprilat or RU44 403 is the diacid form of the
orally active ACE inhibitor trandolapril or RU44 570 (21). Trando- Kr, and stoichiometry of [3H]trandolaprilat binding for each enzyme
(i.e. the binding parameters a t equilibrium) were determined by
laprilat contains a carboxylate group as zinc-bindingligand and was
equilibriumdialysis.Dialysis tubing (Polylabo, France) containing
derived from enalaprilat by substituting the proline ring by a saturated indole ring (21). [3H]Trandolaprilat with a specific activity of enzyme in 0.5 ml of buffer A (100 mM Hepes, pH 7.5, 10 p~ ZnS04,
1 mg/ml lysozyme) with orwithoutNaCl
was immersed intube
66 Ci/mmol was prepared: The purity of the radiolabeled compound
prefilled
with
30
ml
of
the
same
buffer
containing
[3H]trandolaprilat.
was greater than 95% when analyzed by thin silica-gel chromatograSodium chloride concentrations varying from 0 to 300 mM were used
phy using chloroform/methanol/acetic acid (90:5:5) as solvant and
by high-performance liquid chromatography using an analytical col- to examine theeffects of chloride on the binding parametersof each
umn of Hypersil ODS 5 pm (25 X 0.46 cm, inside diameter) (Bioch- enzyme. KJ, values were measured a t one enzyme concentration and
produce mole
rom, France) and acetonitrile/lO mM potassium phosphate buffer, a t 10 differentinhibitorconcentrationschosento
and 90%. After 20 h a t 4 "C
p H 3.8 (25:75) as solvant. Both trandolaprilat and [3H]trandolaprilat fractions of free enzyme between 10
were kindly provided by Dr. Hamon (Roussel-Uclaf, France). Capto- under continuous shaking, aliquots of 0.2 ml were taken from both
the dialysis bagand the surrounding bath and counted in scintillation
vials after addition of PICO-FLUOR 15 scintillationfluid (Packard).
All measurements were performed in duplicate. Enzyme activities in
ACE
the dialysis tubing after 20-h incubation a t 4 "C in the absence of
inhibitor were measured to ensure the
absence of enzyme degradation.
The nonspecific binding estimated in thepresence of 1p~ unlabeled
trandolaprilat remained below 1%of the binding in the absence of
unlabeled inhibitor.
N fragment
C fragment
The binding parameters were determined by the computer method
described by Claire et al. (231, which comprises several iterations to
obtain the best-fitted curves for interaction of a ligand with one or
two specific and nonspecific binding sites. The standard deviations
for the binding parameters estimated from the experimental data
by
the computer were less than 15% of the mean values. The results
FIG. 1. Diagram of ACE constructions. The wild-type ACE were represented according to Scatchard (24).
(ACE)containsthe signal peptide (left black box), thetwo large
Determination of Association and Dissociation Kinetic Constantshomologous domains (shadedboxes), and the transmembrane domain Association kinetic experiments were performed in triplicate by in(right black box) near the C terminus. The sequence containing the cubating 5 X lo-' M [3H]trandolaprilatwith 2.5 X 10"" or 5 X 10""
two zinc-binding histidines is shown under each domain. The trun- M enzyme in 0.2 ml of buffer A with or without NaC1, a t 4 "C for
cated mutant, N fragment, contains the signal peptide and the N various periods of time from 1 to 120 min. The enzyme-inhibitor
domain.Thetruncatedmutant,
C fragment,containsthe
signal complex was separated from the free inhibitor by addition of 1 ml of
peptide,the C domainandtheC-terminal
region including the cold ethanol (-20 "C), followed by centrifugation. The pellet was
transmembrane and intracellular domains. For the
two full-length
resuspended in bufferA and counted. Approximately 80% of the
mutants, the two zinc-binding histidines have been substituted by enzyme-inhibitor complex was precipitated, while less than 2% of the
lysines. Unchangedamino acids arerepresented by dashes. The free inhibitor was precipitated in the absence of enzyme. The nonposition of the mutated histidinesis indicated in the mutantname.
specific binding estimatedinthe
presence of 1 p~ of unlabeled
histidines located in the consensus sequence His-Glu-X-XHis of each domain are involved in zinc binding of the
corresponding domain.
In thepresent study, the interactions of four ACE inhibitors
with each active site of the enzyme were examined using the
wild-type recombinant ACE and a series ofACE mutants
each containingonly one functional domain, the otherdomain
being deleted or inactivated by point mutation of the two
zinc-binding histidines. Radiolabeled trandolaprilat, a compound relatedto enalaprilat, was chosen for thebinding study
in order to make comparison with a previous analysis of
human kidney ACE using the same compound (14). We also
compared the inhibitory potency of three widely used ACE
inhibitors, captopril, enalaprilat, and lisinopril for each of the
two active sites by steady-state kinetic approach. Our results
demonstrate that ACE contains two high affinity binding
sites with different properties.
Inhibitor Binding Sites of Angiotensin I-converting Enzyme
13400
trandolaprilat remained below 1%of the total binding in the absence
of unlabeled inhibitor. The results were therefore corrected for the
nonspecific precipitation of free ligand and for the incomplete precipitation of bound ligand, both of which were measured in each experiment.
As less than 10% of radiolabeled inhibitor was bound a t equilibrium, the apparent association kinetic constant k+, can be determined
from the pseudo-first order equation:
In[B,/(B, - B)1 = kob.t
(1)
where Be and B are the concentration of the enzyme-inhibitor complex at equilibrium and attime t, respectively, and kd, is the slope of
the pseudo-first order plot. The k,, values can be calculated from kobs
by the following equation:
k+I
=
kobs[Br/(l)T(E)T]
(11)
ln(B/B,) = -k-]t
(111)
RESULTS
Stoichiometry of pH]Trandolaprilat Binding-The inhibitor-binding studywas first performed on the wild-type recombinant ACE by equilibrium dialysis. The radiolabeled inhibitor bound to thewild-type recombinant ACE in aspecific and
saturable manner. The Scatchardrepresentation of a typical
saturation curve obtained in thepresence of 100 mM NaCl is
shown in Fig. 2. The experimental data best fit a two-specific
binding site model. The dissociation constants of these two
sites were, respectively, 1.0 f 0.2 X 10"' M and 3.0 f 0.3 X
P
.-a
.n.
Bound [3HlTrandolaprilat (nM)
FIG.2. Scatchard plot of [SH]trandolaprilatbinding to the
wild-type enzyme (e),
the N fragment (W), and the C fragment
(A).The binding of 1 X lo-" to 2.5 X lo-' M of [3H]trandolaprilatto
3 X 10"' M of enzyme was determined by equilibrium dialysis in 100
mM Hepes, pH 7.5, at 4 'C and in the presence of 100 mM NaCl as
described under "Materials and Methods." The solid lines are the
best-fitted theoretical curves to theexperimental data with the model
of one (N fragment or C fragment) or two (wild-type ACE) specific
binding sites. The dissociation constant (KO)and maximal concentration of specific binding site
are, respectively, KO = 3.2 X
10"' M and B,,, = 2.9 X 10"' M for the N fragment, KO= 1.2 x 10"'
M and B , = 3.1 X 10"' M for the C fragment, KO = 3.3 X 10"' and
1.2 X 10"' M for the two binding sites of the wild-type ACE with
&,ax
= 3.0 X 10"' and 3.2 X 10"' M, respectively.
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
where ( 1 )is~the total concentration of inhibitor and ( E ) T is the total
concentration of enzyme (25).
The dissociation kinetic constant k-, was determined by preincubating 5 X 10"' M enzyme with 1 X lo-' M [3H]trandolaprilat in
buffer A with or without NaCl for 4 h at 4 "C to allow equilibrium,
and then by adding an excess of unlabeled inhibitor to give a final
concentration of 1PM. Triplicate aliquots were then taken at different
times, ethanol-precipitated, and counted. The k-, values were calculated from the following equation.
10"' M under these experimental conditions (Table I). The
determination of the inhibitor-binding stoichiometry, performed with threeseparate preparations of the enzyme,
showed that the wild-type recombinant enzyme bound 2.0 5
0.3 (n = 6)mol of trandolaprilat per mol of enzyme.
Parallel measurements were conducted on the two truncated mutants (N andC fragments) containingonly the N or
C domain. Only one class of high affinity binding site was
detected in each molecule (Fig. 2). The KO values determined
in the presence of 100 mM NaCl were 3.1 f 0.2 X 10"' and
1.1 f 0.1 X 10"' M for the N and C fragments, respectively,
which are consistent with the KO values of the two binding
sites of the wild-type enzyme determined under the same
conditions (Table I). These results suggest that the affinities
of the N and C domains in the truncated mutants for [3H]
trandolaprilat are identical to those of the N and C domains
in the intact molecule.
Removal of the zinc atom from ACE has been shown to
decrease its affinity for enalaprilat by 20,000-fold (12). Alteration of the zinc-binding site of either the N or C domain
should result in a dramatic reduction of the affinity of the
corresponding domain for [3H]trandolaprilat. Indeed, the two
full-length mutants ACEK361.365,
A C E K ~ in
~ which
~ , ~ ~the
~ zinc,
binding site of either the N or C domain was altered by point
mutation, contain only one high affinity binding site. The KO
value of the remaining intact domain was in each case identical to that of the corresponding domain in the truncated
mutant or in the intact ACE molecule (Table I). Thus ACE
possesses two different high affinity binding sites for [3H]
trandolaprilat; one is located on each of the two homologous
domains.
Effects of Chloride on the Kinetic and Equilibrium Constants
of PHITrandolaprilat Binding-Previous inhibition studies
have shown that chloride influences inhibitor binding to ACE,
especially the dissociation rate of the enzyme-inhibitor complex (14, 26). This chloride effect required re-examination
because the interpretations of all previous results assumed
the presence of one binding site in ACE. Moreover, some
results such as thebiphasic pattern of the dissociation of the
complex at high chloride concentrations (12, 14) may result
from the presence of two binding sites. It was thus important
to investigate the effects of chloride on inhibitor binding to
each domain.
The association of the enzyme-inhibitor complexes follows
a monophasic process for each domain. Fig. 3 shows the
association curves obtained at 300 mM NaC1. The kinetic
constants were determined from the slope of the pseudo-first
order plot (Fig. 3). The association rate was only slightly
influenced by chloride for both domains (Table 11). Identical
k,, values were obtained for the N and C domains in the
absence of chloride. The k+i value of the N domain was not
altered and that of the C domain decreased only 2-fold when
the chloride concentration was increased from 0 to 300 mM
(Table 11). The fact that the k+, values for the two domains
are similar explains why the association of trandolaprilat to
ACE follows a monophasic kinetic in the previous study (14)
and also in the present study (data not shown).
The dissociation of the enzyme-inhibitor complex is also
monophasic for each of the two domains. The dissociation
curves obtained at 300 mM NaCl are shown in Fig.4. In
contrast with the association rates, chloride has a dramatic
effect on the dissociation rate of the enzyme-inhibitor complex for each of the two domains (Table 11).The k-1 value of
the N domain fell about 30-fold when the chloride concentration was increased from 0 to 20 mM, with no further decrease
at higher chloride concentrations. The effect of chloride ap-
Inhibitor
Binding
Sites
of Angiotensin
I-converting
Enzyme
13401
TABLE
I
Binding parameters of ['Hltrandolaprilat to various ACE mutants
Values were determined by equilibrium dialysis in the presence of 100 mM NaCl at pH 7.5 and 4 "C, BN and Bc are the total binding sites
corresponding to the N and C domains, KDNand KDcare the equilibrium constants of the N and C domains.
KDN
Bc
KDC
3.2
1.5
1.2
0.86
['HITrandolaprilat
bound/mol of enzyme
[Enzyme]
BN
Wild-type
3.0
1.5
3.0
1.3
3.3
2.7
N fragment
5.0
3.0
5.2
2.9
2.9
3.2
1.04
0.97
ACEK~~S,S~X~
5.0
1.0
5.3
0.85
3.4
2.8
1.06
0.85
C fragment
5.0
3.0
Enzyme
moles
Xlo", M
2.07
1.87
1.o
1.2
0.88
1.03
4.4
1.8
1.1
0.94
0.88
0.90
Tlrne
0.5
0
60
I20
(rnln)
180
240
3 '0
1
0.8
0.6
04
*
m
,
m
02
0
I
1
I
I
I
IO
20
30
40
50
f
Tlrne ( r n l n )
FIG. 3. Association of ['Hltrandolaprilat to the N fragment
(W) and the C fragment (A).The kinetics of specific binding of
[SHItrandolaprilat (5 X lo-' M) to the enzyme (5 X 10"' M) were
determined after incubation in 100 mM Hepes, pH 7.5, at 4 "C and in
the presence of 300 mM NaCl for the indicated times. Inset, replot of
the data according to equation (I) (see "Materials and Methods"). E,
and E are, respectively, the concentration of the enzyme-inhibitor
complex at equilibrium (60 min) and that at time t. B,/(B, - B ) is
plotted on a logarithmic scale. The values of k+, calculated from the
slope of the line by using equation (11) are given in Table 11.
TABLEI1
Effect of chloride on therate constants of [3H]trandolaprilat binding
to the N and C domains
Assays were performed in 100 mM Hepes buffer, pH 7.5, at 4 "C.
k+, and k-, values were determined as described in Figs. 2 and 3 and
are the average of two to four independent determinations which
varied by less than 15%.
k-l
k+,
ICl-1
mM
9.9
0
5
20
100
300
N framnent
C
frament
x ~ o - M'
~ , s-l
9.8
8.3
8.0
8.1
8.6
8.7
5.6
4.5
4.3
N fraement
C framnent
x 106,
's
770
36
27
25
24
1604
69
20
5.2
1.4
pears to be greater for the C domain as its k-l value continuously decreased to about 1100-foldwhen the chloride concentration was increased from 0 to 300 mM. These resultsindicate
that the affinity of trandolaprilat for each domain, especially
01
FIG. 4. Dissociation of [3H]trandolaprilat from the wildtype enzyme,).( the N fragment (W), and theC fragment (A).
Enzyme (5 X 10"' M ) was incubated with [3H]trandolaprilat (1 X
lo-' M ) for 4 h in 100 mM Hepes, pH 7.5, at 4 "C and in the presence
of 300 mM NaCI. The dissociation kinetic experiment was started by
adding
M of unlabeled trandolaprilat. The ratio of the bound
[3H]trandolaprilat at time t ( E )to that attime 0 ( B e )is plotted on a
logarithmic scale. The values of k-1 calculated from equation (111) are
given in Table 11. The dashed line represents the theoretical curve of
the inhibitor dissociation from the wild-type enzyme, assuming that
35% of the bound inhibitor is on the N domain and 65% on the C
domain, with the k-l values of the two domains determined in the
same experimental conditions with the two truncated mutants.
the C domain, is greatly enhanced by chloride, which essentially slows the dissociation rate of the enzyme-inhibitor
complex.
The calculated dissociation constants obtained by dividing
the dissociation rate constants by the association rate constants are in good agreement with the KO values determined
by equilibrium dialysis (Table 111). The affinities of trandolaprilat for the two domains are very different (22-fold) only
at high (2100 mM) or low ( 5 5 mM) chloride concentrations.
Thus, thebinding assay is only able to reveal the presence of
two different binding sites inthe wild-type ACE at these high
or low chloride concentrations. Here again, the K D values of
the N and C domains in the wild-type ACE are similar to
those of the corresponding intact domains in the truncatedor
full-length mutants(data notshown),as
observed in the
presence of 100 mM NaCl (Table I).
At high chloride concentrations, where the dissociation rate
of the C domain was lower than that of the N domain, the
dissociation of the wild-type enzyme-[3H]trandolaprilatcom-
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
4.4
3.1
Inhibitor
Binding
Sites
13402
of Angiotensin
I-converting
Enzyme
TABLE111
Effect of chloride on the equilibrium constantsof [3H]trandolaprilat
binding to theN a n d C domains
KD = k-,/k+Ib
KU0
[CU
mM
N fragment
C fragment
Xlo'', M
N fragment
X
C fragment
10", M
162
0
7997
167
5
8.9
4.2
20
2.8
3.5
1.1 100
3.2
1.2
3.1
0.32 300
2.9
0.33
2.8
' Values were determined by equilibrium dialysis in 100 mM Hepes,
pH 7.5, at 4 "C as described under "Materials and Methods" and are
the average of twoor three independent determinations which varied44
by less than 15%.
Values were derived from the rate constants presented in Table
11.
TABLEIV
Inhibition of theN a n d C domains by captopril, enalaprilat,
lisinopril, and trandolaprilat
Assays were performedin 100 mM potassium phosphate buffer, pH
8.3, at 37 "C with Hip-His-Leuas substrate. Values of Kt were
determined from the dose-dependent inhibition curves in thepresence
of varying concentrations of inhibitor as described under "Material
and Methods." The results are the means of two to four independent
determinations which varied by less than 15%.
Enzyme
K,
[CI-]
Captopril Enalaprilat Lisinopril Trandolaprilat
X1o1', M
mM
Wild-type
300
0.45
N 26
fragment
8.9
300
20
9.1
133.9
6.5
3.1
3.3
31
C fragment
6.3 14 300
111
78
27
0.29
2.2
2.4
plex appeared to be biphasic. The dissociation curve obtained
in the presence of 300 mM NaCl is shown in Fig. 4.The first chloride dependent: the Kivalues for the C domain increased
rapid dissociation step, involving about 35% of the bound about 10-fold for all these inhibitors when the chloride coninhibitor, corresponds mainly to the dissociation of the N centration was lowered from 300 to 20mM; whereas the K,
domain. The second slow phase corresponds to the dissocia- values for the N domain are unchanged. Thus, at low chloride
tion of the C domain. The experimental data are inagreement concentration, the difference in affinities to the two domains
with the calculated values of the bound inhibitor on the N was reduced for lisinopril, trandolaprilat, and enalaprilat but
and C domains (35 and 65%, respectively) according to the not for captopril. Captopril indeed bound much better to the
K,) values of the N and C domains under the conditions of N domain than to theC domain at low chloride concentration
this experiment. Furthermore, the experimental data are also and was even a slightly better inhibitor of the N domain at
in accordance with the theoretical dissociation curve (Fig. 4) high chloride concentration (Table IV). The relative potency
assuming that 35 and 65% of the bound inhibitor were on the of these inhibitorsfor the C domain,as assessed by comparing
the K; values, is trandolaprilat > lisinopril > enalaprilat >
N and C domains, respectively, with k l values of 2.4 X
and 1.4 x 10-5/s, as determined using the truncated mutants. captopril, whereas for the N domain, trandolaprilat > captoInhibitory Potency of Competitive ACE Inhibitors forthe N pril > enalaprilat > lisinopril, at 300 mM as well as 20 mM
and C Domuins-All the previous studies using captopril, NaCl. These results indicate that thedifference in the potenenalaprilat, lisinopril, and trandolaprilat suggested the pres- cies of these inhibitors to the N and C domains depends both
ence of a single high affinity binding site inthe ACE molecule. on the chloride concentration and also on the structureof the
Since two high affinity binding sites for [3H]trandolaprilat inhibitor (Table V).
were detected, it was of importance to determine the inhibiDISCUSSION
tory potency of other inhibitors for each domain of ACE.
Experiments were performed at 300 mM NaC1, because at this
We have recently demonstrated by site-directed mutagenconcentration, the difference in affinities for trandolaprilat esis that ACE possesses two independent active sites and that
between the two domains is maximum. Additional experi- their activities are inhibited by captopril and enalaprilat (19).
ments were performed at 20 mM NaC1, since the effect of This finding was unexpected since the results of all the
chloride on trandolaprilat binding to theN domain reaches a previous studies of the molecular interaction ofACE with
competitive inhibitors suggested the presence of a single high
plateau at thisconcentration.
The inhibitory potency of these inhibitors was determined affinity binding site in the ACE molecule (11-14).The present
for the wild-type recombinant ACE and the truncated or
full- study establishes that both domains of ACE contain a high
length mutants by enzymatic assay using Hip-His-Leu as affinity binding site for these inhibitors and that these two
substrate. The K; values of trandolaprilat for the N and C domains display different properties and different chloride
domains (3.1 X 10"' and 0.29 X 10"' M) (Table IV),measured sensitivities for inhibitor binding.
The presence of two different high affinity binding sites in
by enzyme inhibition assay at 300 mM NaCl in potassium
phosphate buffer, pH 8.3, at 37 "C, were consistent with the ACE is supported by several lines of evidence: 1) [3H]tranKr, values determined by [3H]trandolaprilat binding assay dolaprilat binding to the wild-type ACE reveals the presence
with Hepes buffer, pH 7.5, at 4 "C (2.9 X 10-l' and 0.32 X of two classes of high affinity binding site and a stoichiometry
10"' M) (Table 111). Similar result was obtained at 20 mM of two molecules of inhibitor bound per enzyme molecule; 2)
NaCl (Tables I11 and IV). For all the inhibitors tested, the K , the two truncated mutants,lacking either the N or C domain,
values of the truncated mutants
were identical to those of the each contain only one class of binding site and present a
stoichiometry of one; 3) the binding affinities of the N and C
corresponding full-length mutants (data not shown).
The results, shown in Table IV, indicate that captopril, domains for [3H]trandolaprilat in the truncated mutants are
enalaprilat, and lisinopril are all potent inhibitors for both consistent with those of the N and C domains in the wildthe N and C domains with K; values within the nanomolar type enzyme, suggesting that each domain independently
range. Lisinopril, trandolaprilat, enalaprilat, but not captopril binds an inhibitor molecule; 4) alteration of the zinc-binding
inhibit preferentially the C domain more than theN domain, site in either the N or C domain destroys the corresponding
at least at high chloride concentration, as the K, values of inhibitor binding site. In addition, we demonstrate that the
these inhibitors were, respectively, about 18-,lo-, 4-fold lower catalytic activitiesof both the N and C domains are inhibited
for the C domain than for the N domain at 300 mM NaC1. by captopril, enalaprilat, lisinopril, and trandolaprilat with K;
However, this difference in affinities to the two domains was values within the nanomolar range.
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
20
42
Inhibitor
Binding
Sites
of Angiotensin
I-converting
Enzyme
13403
TABLEV
Comparison of the inhibitor binding propertiesof the N and C domains
Effect of chloride on
apparent
affinity'
Relative
inhibitory
Trandolaprilat"
Kr,
XIO1", M
N domain
C domain
2.9
0.32
k+l
X
IO-^, M Is-'
8.6
4.3
K,b
k-1
XI@,
s-'
24
1.4
20 mM
300 mM
28
60
33
522
potencyd
X10", M
3.1
0.29
T>C>E>L
T>L>E>C
" Parameters were determined in the presence of 300 mM NaCl.
* Values were measured using Hip-His-Leu as substrate.
' The K,, value of trandolaprilat determined without chloride addition dividedby that at 20 or 300 m M NaCl.
T, trandolaprilat; L, lisinopril; E,enalaprilat;C, captopril. Relative inhibitorypotency was determined at 20 and 300 mM NaCl. The same
order was obtained.
therefore crucial in determining the stoichiometry affinity
and
of inhibitorbinding for ACE by the equilibriumdialysis
method. In the present study, this was greatly facilitated by
separate determination of the affinities of the two domains
using ACE mutants, each containingonly one intactdomain.
The availability of these mutants is
also essential to determine
the inhibitory potency of competitive inhibitors for each of
the two domains of ACE by steady-state kinetic method. The
results presented in TableIV show that theK, values of four
inhibitors tested for the wild-type ACE are very similar to
those for the C domain at 300 mM NaCl using Hip-His-Leu
as substrate, because under these conditions the catalytic
activity of the C domain is greater than thatof the N domain
and accountsfor the major part of ACE activity, as previously
observed (19).Since theprevious steady-state inhibition studies investigating the potencyof these inhibitors for ACE (1114, 27) were generally performed under conditionswhere the
activity of the C domain is greater than that
of the N domain,
they reported mainly the potency of these inhibitors for the
C domain, whereas their potency for the N domain could not
beassessedusing
the intact ACE molecule. However, the
potencies of ACE inhibitors for the N domain, which were
determined using ACE mutants containing only the intact N
domain, can be assumed to be equivalent to those of the N
domain in thewild-type ACE onlyif the two domains of ACE
function independently. This most
is likely the case, since the
binding of trandolaprilat to either domaindoes not seem to
affect the binding to the other domain as shown by the fact
that the binding affinities
of the two domains in the truncated
or full-length mutants are
essentially the same as those
of the
two domains in the wild-type enzyme (Table I). These findings are in good agreement with ourprevious results showing
that the wild-type enzyme activity was always equal to the
sum of the activities of the two domains assessed separately
(19), and further support concept
the
that thetwo active sites
of ACE function independently.
The present study also shows that the two high affinity
bindingsitesin ACE display differentbindingproperties,
especially different chloride sensitivities. The bindingof [3H]
trandolaprilat to both domains
was greatly enhanced by chloride, as the KO values were decreased by about 30- and 500fold for the N and C domains, respectively, by the presence
of 300 mM NaC1. The major effect of chloride is to decrease
the dissociation rate of the enzyme-inhibitorcomplex for both
domains, since the association rates of the two domains are
onlyslightlyinfluenced
by chloride (Table 11). Themain
difference between the two domains is that the stabilizing
effect of chloride is much greater for the C domain (about
1100-fold decrease in its k l value when the chloride concentration increases from 0 to 300 mM) than for the N domain
(about 30-fold). Moreover, this effect of chloride reaches a
plateau a t a much lower concentration for the N domain (20
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
Therefore, the present observations differ from previous
findings and, in particular, from
a previous study of our group
on [''H]trandolaprilat binding to human kidney ACE. This
report suggested a single binding site inACE with a Kn value
of 4.4 X 10"' M, when determined at 300 mM NaCl and at
room temperature after a 24-h incubation time (14). In the
present study,two binding sites withKOvalues of 2.9 x 10"'
and 0.32 X 10"' M were observed. The discrepancy between
these two studies may be due to degradationof the enzyme in
the conditions of the previous experiment. Indeed, we observed a 30 to 70% decrease in enzyme activities of human
kidney ACE and recombinant ACE when incubated for 24 h
at room temperature (data not shown). No degradation was
observed a t 4 "C. In addition, the detection of two different
binding sites in ACE requires a n appropriate range of inhibitor concentrations toproduce mole fractions of free enzyme
between 10 and 90%. If the inhibitor concentration range is
too narrow, only one binding sitewill be identified.
Given the low K, values of the N and C active sites for
captopril, enalaprilat, and lisinopril, previous studies using
the corresponding radiolabeled inhibitors should have been
able to detect the presence of two high affinity binding sites
in ACE, whereas theydetected onlyone (12, 13).Several
explanations for this discrepancy can be proposed. First, the
two sites present similar affinities for the inhibitors under
some experimental conditions, leading to the detection of a
single class of binding site in the wild-type enzyme, as was
the case for [3H]trandolaprilat at 20 mM NaCl (Table 111).
Another possibility is that even if the affinities of the two
binding sites are
very different, as in the
case of trandolaprilat
at high ( Z 100 mM) or low (55 mM) chloride concentrations
(Table 111), a single class of binding site will be detected if
the various concentrations of inhibitor are not appropriately
chosen,as discussed above. Our resultsindicatethatthe
difference in affinities between the two domains is greater a t
high chloride concentration (300 mM) for trandolaprilat, enalaprilat, and lisinopril, whereas for captopril, the difference
is larger at low chloride concentration (20 mM) (Table IV).
Finally, the stoichiometry of 1 mol of inhibitor bound per
mol of enzyme found in the previous studies (11-14) may be
due to inaccurate determination
of the enzyme concentration,
t o partial occupation of the two binding sites
because of a too
narrow range of inhibitor concentrations, or to partialdegradation of enzyme during incubation, as discussed above. A
recent study on the binding of lisinopril (27) detected two
binding sites in human kidney and rabbit lung ACE, and a
single binding site in human and rabbit testis
ACE which
contain only the C domain (28-30). However,only one of
these two sites in somaticACE appeared tobe a high affinity
bindingsite,andtheaffinity
of the second site was not
determined.
The choice of the appropriate experimental conditions is
13404
Inhibitor
Binding
Sites
of Angiotensin
I-converting
Enzyme
phenylalanyl residue in the PI position of enalaprilat may
interact better with theC domain than with the
N domain or
that the N domain mayprefera
strong zinc-coordinating
inhibitor than the C domain. In our previous study, we suggested that the two active sites of ACE may have different
functions since they
display different catalytic properties
(19).
The presence of structural differences between the two active
sites revealed by interaction with different inhibitorsprovides
additional evidence to support this hypothesis. It has been
recently reported that endothelial
ACE but not testicular
ACE
efficiently hydolyzes the N-terminal tripeptideof LH-RH, an
in vitro ACE substrate, suggesting that theN domain may be
specifically involved inthis endoproteolytic cleavage (27).
Using ACE mutants, we found that LH-RH
was indeed better
hydrolyzed by the N domain than by the C domain.’ The N
domain may therefore have aspecific i n vivo function. Further
experiments witha large number of ACE inhibitors may help
to elucidate the structural and functional
differences between
the two active sites and therebyhelp to design specific inhibitors for each active site which will be useful to investigate
the physiological functions of each of the two active sites of
ACE.
In conclusion, the present study demonstrates that ACE
contains two different high affinity binding sites for competitive inhibitors. Thedifference in thepotencies of an inhibitor
for the two domains depends both on the
chloride concentration and on the structure of the inhibitor. The difference in
the characteristicsof trandolaprilat bindingbetween the two
domains at 300 mM NaCl and the other marked differences
in the inhibitor binding properties summarized in Table V
provide strong evidence for the existence of structural differences between the two active sites of ACE. The availability
of ACE mutants, each containing only one intact domain,
provides powerful tools for studying the structural andfunctional differences of the two active sitesandtherebyto
understand the catalytic mechanism
of ACE at molecular
level.
Acknowledgments-We are grateful to Dr. Bernard P. Roques and
Dr. Edgar Haber for critical reading of the manuscript, to Dr. MarieThbrGse Chauvet and Dr. Christine Hubert for many helpful discussions during this work, and to Annie Michaud for purifying human
kidney ACE.
REFERENCES
1. Skeggs, L. T., Jr., Kahn, J. R., and Shumway, N. P. (1956) J.
Exp. Med. 103,295-299
2. Yang, H. Y. T., Erdos, E. G., and Levin, Y. (1970) Biochim.
Biophys. Acta 2 1 4 , 374-376
3. Erdos, E. G., and Skidgel, R. A. (1987) Lab. Znuest. 56, 345-348
4. Caldwell, P. R., Seegal, B. C., Hsu, K. C., Das, M., and Soffer, R.
L. (1976) Science 191, 1050-1051
5. Defendini, R., Zimmerman, E. A., Weare, J. A., Alhenc-Gelas, F.,
and Erdos, E. C.. (1983) Neuroendocrinology 3 7 , 32-40
6. El-Dorry, H. A., Bull, H. G., Iwata, K., Thornberry, N. A., Cordes,
E. H., and Soffer, R.L. (1982) J . Biol.Chem. 2 5 7 , 1412814133
7. Gavras, H., Brunner, H. R., Laragh, J. H., Sealey, J. E., Gavras,
I., and Vukovich, R. A. (1974) N. Engl. J . Med. 291,817-821
8. Ondetti, M. A., Rubin, B., and Cushman, D. W. (1977) Science
196, 441-444
9. Patchett, A. A., Harris, E., Tristam, E. W., Wyvratt, M. J., WU,
M. T., Taub, D., Peterson, E. R., Ikeler, T. J., ten Broeke, J.,
Payne, L. G., Ondeyka, D. L., Thorsett, E. D., Greenlee, W. J.,
Lohr, N. S., Hoffsommer, R. D., Joshua, H., Ruyle, W.V.,
Rothrock, J. W., Aster, S. D., Maycock, A. L., Robinson, F. M.,
Hirschmann, R., Sweet, C. S., Ulm, E. H., Gross, D. M., Vassil,
T. C., and Stone, C. A. (1980) Nature 288, 280-283
10. Waeber, B., Nussberger, J., and Brunner, H. S.(1990) Hyperten-
E. Jaspard, L. Wei, and F. Alhenc-Gelas, unpublished results.
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
mM) than for the C domain (2300 mM). Chloride activation
of substrate hydrolysis by the N and C domains showed a
similar phenomenon (19). A consequenceof these differences
between the two domains is that therelative binding affinity
of trandolaprilat toward the two domains depends on the
chloride concentration: its binding affinity for the C domain
becomes higher than that for theN domain at chloride concentrations higher than 20 mM,because of the greaterchloride
effect on the C domain (Table 111). Moreover, the difference
in the binding affinities of the two domains is mainly due to
the differences in theirdissociation rates, since the association
rates of the two domains are very similar (Table 11).Another
consequence of the different effects of chloride on the dissociation ratesof the two domains is that the
dissociation kinetic
of the wild-type enzymeappears tobe biphasica t high chloride
concentrations (Fig. 4).
The different chloride sensitivities of the two domains for
trandolaprilatbindingcan probably be extendedtoother
inhibitors such as captopril, enalaprilat, and lisinopril, since
the Kt values of these inhibitors for theN and C domains are
also differently influenced by chloride (Table IV). Moreover,
the association kinetics of these inhibitors to the native
ACE
were found to be monophasic (12,13), suggesting that thetwo
domains of ACE display similar association kinetics. In the
previous dissociation kinetic studies, where it was assumed
that ACE contained a single active site, it was observed that
enalaprilat, lisinopril, and trandolaprilat had biphasic dissociationkinetics (12, 14). Theseobservationscan
now be
interpreted asreflecting the presence of two binding siteswith
different dissociationkinetics.
The previous studies also
showed that chloride enhanced inhibitor binding to
ACE (1214, 26) by decreasing the dissociation rate of the ACE-inhibitor complex (14, 26). The present results indicate that this
property is conserved for the two active sites of ACE, suggesting that the mechanism of enzyme-chloride interaction
for thetwo active sites isprobably similar.
The observed difference in chloride sensitivities for inhibitor binding between the two domains raises thequestion
whether the two active sites of ACE are equally inhibited
under physiological conditions(about 100 mM chloride in
plasma) after drug intake. The present i n vitro study shows
that the Ku values of trandolaprilat for the two domains are
within the 10”’ M range at 100 mM NaCl, and suggests that
trandolaprilat binds both active sites after drug intake. Similarly, captopril, enalaprilat, and
lisinopril are all high affinity
binding inhibitors for the two domains at high chloride concentrations (220mM) (Table IV), indicating that both
active
sites should be readily inhibited in vivo by these inhibitors
for hydrolysis of angiotensin I.
Interestingly, severalstructural differences betweenthe two
active sites appearto be independent of chloride binding. The
relative potency of the four inhibitors tested is different for
the two domains (trandolaprilat > lisinopril > enalaprilat >
captopril for the C domain and trandolaprilat > captopril >
enalaprilat > lisinopril for the N domain), at 300 mM as well
as 20 mM NaC1. The fact that the replacement
of Ala in
enalaprilat by Lys in lisinopril in the P’l position results in
enhanced binding for the C domain, but decreased binding
for the N domain suggests structural difference of the two
domains at their S’l substrate binding subsite. In addition,
captopril ismore potent than enalaprilat for inhibiting the
N
domain, but less potent than enalaprilat for the C domain.
Captopril and enalaprilat differ in that enalaprilat contains
a n additional residue (Phe) in the PI position and a weaker
zinc-coordinating ligand (a carboxylate group, captopril contains a sulfhydryl group).The present resultssuggest that the
Inhibitor
Binding
Sites
of
Angiotensin
I-converting
Enzyme
20. Wei, L., Alhenc-Gelas, F., Soubrier, F., Michaud, A., Corvol, P.,
and Clauser, E. (1991) J . Biol. Chem. 266,5540-5546
21. Brown, N. L., Badel, M.-Y., Benzoni, F., Zanirato, J., Vincent,
J.-C., Fichelle, J., and Worcel, M. (1988) Eur. J. Pharmacol.
148,79-91
22. Naqui, A. (1983) Biochem. J. Lett. 215,429-430
23. Claire, M., Rafestin-Oblin, M. E., Michaud, A., Corvol, P., Venot,
A,, Roth-Meyer, C., Boisvieux, J. F., and Mallet, A. (1978)
FEBS Lett. 88,295-299
24. Scatchard, G. (1949) Ann. N . Y. Acad. Sci. 51,660-672
25. Weiland, G. A,, and Molinoff, P. B. (1981) Life Sci. 29, 313-330
26. Shapiro, R., and Riordan, J. F. (1984) Biochemistry 23, 52345240
27. Ehlers, M. R. W., and Riordan, J. F. (1991) Biochemistry 30,
7118-7126
28. Lattion, A.-L., Soubrier, F., Allegrini, J., Hubert, C., Corvol, P.,
and Alhenc-Gelas, F. (1989) FEBS Lett. 252, 99-104
29. Ehlers, M. R. W., Fox, E. A., Strydom, D. J., and Riordan, J. F.
(1989) Proc. Natl. Acad. Sci. U. S. A . 8 6 , 7741-7745
30. Kumar, R. S., Kusari, J., Roy, S. N., Soffer, R. L., and Sen, G. C.
(1989) J. Biol. Chem. 264, 16754-16758
Downloaded from www.jbc.org at Dana Medical Library, University of Vermont on October 8, 2006
sion: Pathophysiology, Diagnosis and Management (Laragh, J.
H., and Brenner, B. M., eds) pp. 2209-2232, Raven Press, New
York
11. Shapiro, R., and Riordan, J. F. (1984) Biochemistry 23, 52255233
12. Bull, H. G., Thornberry, N. A., Cordes, M. H. J., Patchett, A. A.,
and Cordes, E. H. (1985) J. Biol. Chem. 260,2952-2962
13. Strittmatter, S. M., De Souza, E. B., Lynch, D. R., and Snyder,
S. H. (1986) Endocrinology 118, 1690-1699
14. Cumin, F., Vellaud, V., Corvol, P., and Alhenc-Gelas, F. (1989)
Biochem. Biophys. Res. Commun. 163, 718-725
15. Das, M., and Soffer, R. L. (1975) J. Biol. Chem. 250, 6762-6768
16. Biinning, P., and Riordan, J. F. (1985) J. Inorg. Biochem. 24,
183-198
17. 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
18. Bernstein, K. E., Martin, B. M., Edwards, A. S., and Bernstein,
E. A. (1989) J. Biol. Chem. 264, 11945-11951
19. Wei, L., Alhenc-Gelas, F., Corvol, P., and Clauser, E. (1991) J.
Biol. Chem. 266,9002-9008
13405