Báo cáo khoa học: Combined use of selective inhibitors and fluorogenic substrates to study the specificity of somatic wild-type angiotensin-converting enzyme docx

Combined use of selective inhibitors and fluorogenic
substrates to study the specificity of somatic wild-type
angiotensin-converting enzyme
Nicolas D. Jullien
1
, Philippe Cuniasse
1
, Dimitris Georgiadis
2
, Athanasios Yiotakis
2
and Vincent Dive
1

1 CEA, De
´
partement d’Inge
´
nerie et d’Etudes des Prote
´
ines, Gif ⁄ Yvette, France
2 Department of Chemistry, Laboratory of Organic Chemistry, University of Athens, Greece
Angiotensin-converting enzyme (ACE) in vertebrates
is a zinc metallopeptidase involved in the release of
angiotensin II and the inactivation of bradykinin, two
peptide hormones that play a key role in blood pres-

sure regulation and renal and cardiovascular function
[1–4]. ACE inhibitors have been on the market for
more than 20 years, with successful applications for
conditions ranging from mild hypertension to post-
myocardial infarction [5,6]. Somatic ACE is a very
unusual enzyme which contains two active sites on the
same polypeptide chain [7]. Since this discovery, there
has been much speculation about the functional
significance of the presence of two active sites in the
same enzyme [8–13]. At the biochemical level, the pres-
ence of two active sites in the same enzyme has
hampered the full characterization of substrate and

inhibitor selectivity of somatic ACE. To circumvent
these limitations, mutants of human ACE containing a
single functional active site [14,15] or isolated ACE
domains have been utilized to perform these studies
[10,11]. Alternative approaches to study directly the
properties of both domains in somatic form of ACE
are still lacking. The development of the first highly
N-domain-specific inhibitor, RXP407 [16], and the
recent identification of a C-domain-specific inhibitor,
RXPA380 [13], of human ACE may provide such an
alternative strategy for studying inhibitor and substrate
selectivity of any form of somatic ACE (Scheme 1). To

demonstrate its interest, this approach was used to
study the selectivity of somatic ACE purified from
Keywords
active site; angiotensin-converting enzyme
(ACE); fluorogenic substrates; phosphinic
inhibitors
Correspondence
V. Dive, CEA, De
´
partement d’Inge
´
nerie et

d’Etudes des Prote
´
ines, 91191 Gif ⁄ Yvette
Cedex, France
Fax: +33 169089071
Tel: +33 169083585
E-mail:
(Received 19 December 2005, revised 16
February 2006, accepted 21 February 2006)
doi:10.1111/j.1742-4658.2006.05196.x
Somatic angiotensin-converting enzyme (ACE) contains two homologous
domains, each bearing a functional active site. Studies on the selectivity of

these ACE domains towards either substrates or inhibitors have mostly
relied on the use of mutants or isolated domains of ACE. To determine
directly the selectivity properties of each ACE domain, working with wild-
type enzyme, we developed an approach based on the combined use of
N-domain-selective and C-domain-selective ACE inhibitors and fluorogenic
substrates. With this approach, marked differences in substrate selectivity
were revealed between rat, mouse and human somatic ACE. In particular,
the fluorogenic substrate Mca-Ala-Ser-Asp-Lys-DpaOH was shown to be a
strict N-domain-selective substrate of mouse ACE, whereas with rat ACE
it displayed marked C-domain selectivity. Similar differences in selectivity
between these ACE species were also observed with a new fluorogenic sub-
strate of ACE, Mca-Arg-Pro-Pro-Gly-Phe-Ser-Pro-DpaOH. In support of

these results, changes in amino-acid composition in the binding site of these
three ACE species were pinpointed. Together these data demonstrate that
the substrate selectivity of the N-domain and C-domain depends on the
ACE species. These results raise concerns about the interpretation of func-
tional studies performed in animals using N-domain and C-domain sub-
strate selectivity data derived only from human ACE.
Abbreviations
ACE, angiotensin-converting enzyme; DpaOH, N
3
-(2,4-dinitrophenyl)-L-2,3-diaminopropionyl); Mca, (7-methoxycoumarin-4-yl)acetyl.
1772 FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS
mouse and rat, as compared with human recombinant

somatic ACE, in cleaving the Mca-Ala-Ser-Asp-Lys-
DpaOH substrate (Mca-Ala), as well as a new
fluorogenic substrate of this enzyme, Mca-Arg-Pro-
Pro-Gly-Phe-Ser-Pro-DpaOH (Mca-BK
(1)8)
). A few
years ago, we reported that there are slight variations
in the potency of RXP407 toward the N-domain of
human, mouse and rat [17]. The present study extends
our previous observations by showing unexpected dif-
ferences between the N-domain and C-domain sub-
strate selectivity of these three ACE species. Models of

the N-domain and C-domain of these ACE enzymes
were developed based on the crystal structure of germi-
nal human ACE [18]. From these, the differences in
substrate selectivity between the human, rat and mouse
enzymes can tentatively be explained by the presence
of discrete amino-acid substitutions in the active site.
Results
Titration of somatic ACE by RXP407 using
Mca-Ala substrate
The profile of inhibition of human somatic ACE by
RXP407, using the Mca-Ala substrate, typically exhib-
its two asymmetrical parts (Fig. 1A, closed circle). As

discussed in previous papers [16,19], each part reflects
first the binding of RXP407 to the N-domain at low
inhibitor concentration, followed at higher concentra-
tions by inhibitor binding to the C-domain. The loca-
tion of the inflection points in this profile provides a
direct measure of the IC
50
value (concentration that
causes 50% inhibition) for both domains. In addition,
the value of the inhibition percentage, determined after
inhibitor binding to the active site showing the highest
affinity, was demonstrated to depend on the selectivity

with which the substrate is processed by the N-domain
and C-domain (indicated by the arrow in Fig. 1A).
Thus, information on the selectivity of both the sub-
strate and the inhibitor for the N-domain and
C-domain can be determined from such profiles.
Remarkably, with the same substrate, RXP407 inhi-
bition profiles of somatic ACE purified from mouse
(Fig. 1A, open circle) and rats (closed square) are quite
different from that observed for human ACE. For
mouse ACE, the shape of the inhibition profile sug-
gests the titration of a single active site to which
RXP407 binds with high affinity. Such a profile can be

expected for a substrate that is mostly cleaved by only
one active site, probably the N-domain given the
RXP407 selectivity. In contrast with mouse and human
ACE forms, titration of rat ACE by RXP407 yields a
profile that is shifted to the right. The shape of this
profile is consistent with the binding of RXP407 to the
N-domain, at low concentration, and to the C-domain
at higher concentration. The differences observed
between the human and rat profiles are best explained
by the fact that the affinity of RXP407 for the rat
N-domain is lower than that for the human N-domain,
and Mca-Ala selectivity differs between rat and human

ACE. In the human enzyme, the binding of RXP407
to the N-domain led to 80% inhibition of its activity
(arrow in Fig. 1A, closed circle), whereas in the rat
enzyme binding of RXP407 to the N-domain only gave
20% inhibition (closed square, Fig. 1A). The second
part of the inhibition profile, from 20% to 100%,
observed for rat ACE represents RXP407 binding to
the C-domain and highlights that this substrate is more
efficiently cleaved by the C-domain of rat ACE.
The above experiments suggest that the selectivity
by which the Mca-Ala substrate is cleaved by the
N-domain and C-domain of somatic ACE depends on

the ACE species. To support this hypothesis, mouse
and rat ACE forms were titrated with the RXPA380,
an ACE inhibitor previously shown to exhibit high
selectivity for the C-domain of somatic human ACE.
Titration of somatic ACE by RXPA380 using
Mca-Ala substrate
The titration of human ACE by RXPA380, using
Mca-Ala substrate (Fig. 1B, closed circles), gave a pro-
file that differs from that obtained by RXP407 titra-
tion (Fig. 1A, closed circles). Binding of RXPA380 to
the human ACE C-domain only reduced the enzyme
activity by 20% (left part of the profile). This observa-

P
O
-
H
N
CH
2
H
N
O
CH
3

NH
2
O
O
N
H
COO
-
O
H
3
C

O
H
3
C
RXP407
P
O
H
N
CH
2
C

H
N
OCH
2
OH
O
HN
O
O
OH
RXPA380
Scheme 1.

N. D. Jullien et al. ACE substrate specificity
FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS 1773
tion is in agreement with the suggestion made above
that Mca-Ala is more efficiently cleaved by the human
ACE N-domain. In fact, titration of the N-domain
using higher RXPA380 concentration (right part of the
profile) raised the inhibition from 20% to 100%.
In rat, on the other hand, ACE binding of
RXPA380 to the C-domain reduced the enzyme activ-
ity by 85% (closed square, Fig. 1B), as compared with
the 20% inhibition observed with human ACE. Thus,
titration of rat ACE by both RXP compounds led to a

similar proposal, the higher catalytic efficiency of the
C-domain in cleaving the Mca-Ala substrate.
Titration of the mouse ACE C-domain using low
concentrations of RXPA380 (Fig. 1B, open circle) did
not affect enzyme activity. Inhibition of enzyme activ-
ity only occurred when RXPA380 started to bind to
the N-domain of ACE, at very high concentration.
This observation is in agreement with the RXP407
inhibition profile of mouse ACE, suggesting that the
Mca-Ala substrate is mostly cleaved by the N-domain
of mouse ACE.
To conclude this part, the inhibition profiles

obtained with both RXP inhibitors confirm that the
catalytic efficiency by which Mca-Ala is processed by
the N-domain and C-domain varies according to the
ACE species. For rat and human enzymes, estimation
of IC
50
values for the binding of the inhibitor to the
N-domain and C-domain can be deduced from the
inhibition profiles established with the Mca-Ala sub-
strate. For mouse ACE, as this substrate is mostly
cleaved by the N-domain, only IC
50

values for inhib-
itor binding to the N-domain can be obtained from
the inhibition profile. Thus, affinity of the inhibitor for
the mouse ACE C-domain cannot be determined using
this substrate. This limitation has been overcome by
developing another fluorogenic substrate of ACE.
Titration of somatic ACE with RXP407 and
RXPA380 using Mca-BK
(1)8)
substrate
The presence of two distinct parts in all inhibition pro-
files (Fig. 1C,D) clearly indicates that the Mca-BK

(1)8)
AB
D
C
Fig. 1. Inhibition profiles (% inhibition) of human (d), mouse (s) and rat (n) somatic ACE with RXP407 (A) or RXPA380 (B) when the substrate
used was Mca-Ala-Ser-Asp-Lys-DpaOH (Mca-Ala, 8 l
M). Arrows show the inhibition of human and rat N-domain by RXP407 (A) or human and
rat C-domain by RXPA380 (B). Inhibition profiles of the three ACE forms with RXP407 (C) or RXPA380 (D) when the substrate was Mca-Arg-
Pro-Pro-Gly-Phe-Ser-Pro-DpaOH [Mca-BK
(1)8)
,5lM]. Arrows show the inhibition of human, mouse and rat N-domain by RXP407 (C) or human,
mouse and rat C-domain by RXPA380 (D). All the assays were carried out in 50 m

M Hepes buffer (pH 6.8) ⁄ 200 mM NaCl, at 25 °C. Continu-
ous lines display the simulated profiles obtained with the
DYNAFIT program that represent the best fit of the experimental data.
ACE substrate specificity N. D. Jullien et al.
1774 FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS
substrate is cleaved by the two ACE active sites. The
higher levels of inhibition observed after binding of
RXPA380 to the ACE C-domains (Fig. 1D), as com-
pared with the inhibition levels reached by RXP407
binding to the N-domain (Fig. 1C), suggest that Mca-
BK
(1)8)

displays C-domain selectivity, the degree of
which depends on the ACE species. Titration of the
ACE C-domain with RXPA380 promoted higher ACE
inhibition (55% to 90% inhibition, depending on the
ACE species, Fig. 1D) than blockade of the N-domain
by RXP407 (10% to 45% inhibition, depending on the
ACE species, Fig. 1C). Comparison of the profiles in
Fig. 1C,D shows that Mca-BK
(1)8)
displays the highest
C-domain selectivity toward rat ACE. In contrast with
Mca-Ala, IC

50
values for the binding of the RXP com-
pounds to both the N-domain and C-domain of mouse
ACE can be estimated using the Mca-BK
(1)8)
sub-
strate. From the similarities between the different
profiles, it can be concluded that the affinity and selec-
tivity of both inhibitors for the N-domain and
C-domain of different ACE species are conserved,
except for RXP407 affinity for the N-domain of rat
ACE (see below).

Catalytic efficiencies of the ACE N-domain and
C-domain in cleaving the Mca-Ala and
Mca-BK
(1)8)
substrates
The inhibition profiles described above indicate the
approximate inhibitor concentration required to block
only the N-domain or C-domain of the different ACE
species. Specific blockade of only one ACE active site
by RXP compounds may be a way to determine the
catalytic properties of the second ACE active site, free
of inhibitor, assuming that the two active sites function

independently. To check this proposal, the catalytic
parameters K
m
and k
cat
of the different ACE species
were determined in the absence or presence of one
RXP compound. In these experiments, the concentra-
tion of each RXP compound was chosen in order to
inhibit mostly just one ACE active site. The results
reported in Table 1 suggest that, within experimental
error, the catalytic efficiencies of the N-domain and

C-domain in cleaving the Mca-Ala substrate, in the
presence of RXP compounds, are in good agreement
with the catalytic efficiency determined for the enzyme
without inhibitor. Specifically, for each ACE species,
the catalytic efficiency of the free enzyme corresponds
to the sum of the catalytic efficiencies determined for
the N-domain and C-domain in the presence of inhib-
itor. These results lend credence to the hypothesis that,
for this substrate, each ACE active site functions inde-
pendently. Moreover, these data confirm that each
ACE species processes this substrate with a particular
selectivity. From these experiments, the Mca-Ala sub-

strate turns out to be an N-domain-selective substrate
of mouse ACE, showing the highest catalytic efficiency
in cleaving this substrate, as compared with human
and rat N-domain. The two active sites of human
ACE hydrolyze this substrate, but the N-domain is
more efficient than the C-domain in catalyzing this
reaction. The reverse is observed for rat ACE, this
substrate being better cleaved by the C-domain. Inter-
estingly, although significant differences between the
catalytic efficiency of either the N-domain or
C-domain are observed between these ACE species,
the overall activity of both domains is constant for the

three species. Thus, in this case, the gene duplication
of ACE may be a way to keep the catalytic efficiency
of the somatic enzyme intact, while allowing variations
in the N-domain and C-domain catalytic properties.
Overall, these data on interspecies Mca-Ala substrate
Table 1. Kinetic parameters for the hydrolysis of the Mca-Ala substrate by rat, mouse and human somatic ACE. Kinetic parameters k
cat
and
K
m
were obtained by inhibiting the activity of the C-domain of ACE with RXPA380 (active N-domain) or by inhibiting the N-domain activity
with RXP407 (active C-domain)(see Experimental procedures). N + C is the sum of the k

cat
⁄ K
m
values reported for the N-domain and
C-domain. k
cat
and K
m
were obtained by the direct linear plot method; the confidence limits are indicated in parentheses. k
cat
⁄ K
m

values
were calculated using the experimental k
cat
and K
m
values. ND, not determined, no detectable activity.
Parameter Rat Mouse Human
N-domain k
cat
(s
)1
) 2.8 (2.6–3.0) 13.8 (12.8–15.2) 14.0 (12.9–14.8)

K
m
(lM) 45.9 (37.8–52.9) 26.1 (19.9–33.6) 35.7 (26.4–40.8)
k
cat
⁄ K
m
(s
)1
ÆlM) 0.06 0.53 0.39
C-domain k
cat

(s
)1
) 19.1 (17.9–20.0) ND 4.2 (37.–4.7)
K
m
(lM) 49.5 (43.1–53.3) ND 64.6 (52.4–77.1)
k
cat
⁄ K
m
(s
)1

ÆlM) 0.39 ND 0.06
N+C k
cat
⁄ K
m
(s
)1
ÆlM) 0.45 – 0.45
Somatic ACE k
cat
(s
)1

) 24.7 (22.5–26.3) 14.2 (13.6–15.8) 16.5 (15.1–17.7)
K
m
(lM) 46.9 (38.4–52.5) 24.6 (17.9–30.8) 34.8 (27.2–40.9)
k
cat
⁄ K
m
(s
)1
ÆlM) 0.53 0.57 0.47
N. D. Jullien et al. ACE substrate specificity

FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS 1775
selectivity confirm our interpretation of the inhibition
profiles reported in Fig. 1.
Similar experiments were performed using the Mca-
BK
(1)8)
substrate (Table 2). For this substrate, the
activity of the free enzyme also corresponds to the sum
of the activities determined for each domain in the
presence of an RXP compound. This substrate was
cleaved by the two ACE active sites of different spe-
cies, the selectivity of cleavage varying with the source

of ACE. In particular, this substrate exhibits a marked
C-domain selectivity toward rat ACE, whereas it dis-
plays only a very slight C-domain selectivity when tes-
ted with mouse ACE. In addition, this substrate is
cleaved 10 times faster by rat ACE than mouse ACE.
As compared with Mca-Ala, the activity of the somatic
enzyme varies between the three ACE species.
Determination of RXP407 and RXPA380 K
i
values
for ACE species
Inhibition profiles as reported in Fig. 1 can be used to

determine the K
i
values of the RXP compounds for
the N-domain and C-domain of each ACE species.
Assuming that each ACE active site functions inde-
pendently, we previously showed that simulated inhibi-
tion profiles that best reproduce the experimental data
can provide access to K
i
values [16,19]. Such simula-
tions rely on the use of inputs, notably the catalytic
parameters displayed by each ACE domain in cleaving

the substrate used in the experiments. The results
reported above, in the presence of RXP inhibitor
blocking only one active site, provide approximate val-
ues of the catalytic parameters displayed by each ACE
domain for cleaving the Mca-Ala and Mca-BK
(1)8)
substrates. These values were thus tentatively used to
simulate inhibition profiles able to reproduce the
experimental inhibition profiles. As shown in Fig. 1,
excellent fits between experimental data and simula-
ted curves were generally observed. The K
i

values as
determined by this approach (Table 3) indicate that
RXP407 and RXPA380 are, respectively, highly
N-domain-selective and C-domain-selective inhibitors
of the three ACE species, whatever the substrate used.
For human ACE, the K
i
values determined by this
fitting approach were in close agreement with those
previously reported, using mutants of this enzyme con-
taining only one functional active site [16]. With both
substrates, RXP407 appears slightly less potent toward

the N-domain of rat ACE, whereas RXPA380 displays
similar affinity toward the C-domain of each ACE spe-
cies, showing three orders of magnitude in selectivity.
Inhibitor and substrate selectivity relationships
with the ACE sequences
In a previous paper, we proposed active-site residues
contributing to inhibitor selectivity based on a model of
Table 2. Kinetic parameters for the hydrolysis of the Mca-BK
(1)8)
substrate by rat, mouse and human somatic ACE. Kinetic parameters k
cat
and K

m
were obtained by inhibiting the activity of the C-domain of ACE with RXPA380 (active N-domain) or by inhibiting the N-domain activ-
ity with RXP407 (active C-domain)(see Experimental procedures). N + C is the sum of the k
cat
⁄ K
m
values reported for the N-domain and
C-domain. k
cat
and K
m
were obtained by the direct linear plot method; the confidence limits are indicated in parentheses. k

cat
⁄ K
m
values
were calculated using the experimental k
cat
and K
m
values.
Parameter Rat Mouse Human
N-domain k
cat

(s
)1
) 38.4 (28.9–58.6) 8.9 (7.9–10.5) 17.8 (16.6–20.3)
K
m
(lM) 58.0 (22.9–65.6) 30 (18.8–36.4) 26.1 (18.9–31.3)
k
cat
⁄ K
m
(s
)1

ÆlM) 0.66 0.30 0.68
C-domain k
cat
(s
)1
) 110.9 (105.6–119.3) 4.1 (3.7–4.5) 15.7 (15.1–16.0)
K
m
(lM) 18.9 (14.8–21.3) 12.1 (8.5–14.2) 9.2 (8.3–10.3)
k
cat
⁄ K

m
(s
)1
ÆlM) 5.87 0.34 1.71
N+C k
cat
⁄ K
m
(s
)1
ÆlM) 6.53 0.64 2.39
Somatic ACE k

cat
(s
)1
) 119.0 (108.7–133.3) 12.4 (11.3–14.3) 38.8 (37.3–40.0)
K
m
(lM) 20.2 (14.2–25.4) 21.0 (13.1–24.2) 13.9 (12.5–14.9)
k
cat
⁄ K
m
(s

)1
ÆlM) 5.89 0.59 2.79
Table 3. Potency and selectivity of RXP407 and RXPA380 toward
the N-domain and C-domain of human, mouse and rat somatic
ACE. K
i(app)
values were determined from the simulations that best
reproduced the inhibition profiles reported in Fig. 1. ND, not deter-
mined, no detectable activity.
Domain
K
i(app)

values (nM)
Mca-Ala Mca-BK
(1)8)
Rat Mouse Human Rat Mouse Human
RXP 407 N 55 8 10 30 13 16
C 8500 ND 4000 10000 6000 8000
RXP A380 N 7000 3500 5500 10000 5500 5500
C 2.6 ND 5 7 8 10
ACE substrate specificity N. D. Jullien et al.
1776 FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS
RXPA380 in complex with human ACE [20]. In partic-
ular, two bulky and hydrophobic residues of the human

ACE C-domain, Val955 and Val956 located in the S
2
¢
subsite, were proposed to provide favorable interactions
with the tryptophan side chain of RXPA380. These
interactions should be lost in the N-domain, as valine
residues are replaced by smaller polar residues (Ser357
and Thr358). These differences were suggested to con-
tribute to RXPA380 selectivity. As depicted in Table 4,
rat and mouse ACE C-domain exhibit bulky hydropho-
bic residues at positions 955 and 956 and smaller polar
residues at positions 357 and 358 in the N-domain. This

conservation in the S
2
¢ binding pocket of ACE fits
with the similar potency and selectivity displayed by
RXPA380 towards the different ACE species. Follow-
ing the same strategy, a model of RXP407 interaction
with the human ACE N-domain has been developed to
identify residues of the S
2
,S
1
,S

1
¢ and S
2
¢ pockets that
may influence either inhibitor or substrate selectivity.
This model reveals (Fig. 2) that, among the residues
defining the S
2
,S
1
,S
1

¢ and S
2
¢ pockets, two residues
may greatly influence the RXP407 potency and selectiv-
ity. In fact, in the N-domain, the aspartyl side chain
of RXP407 is observed to interact with Tyr369 and
Arg381 through hydrogen-bond contacts. Indeed, short
distances between the O
d1
and O
d2
atoms of the

RXP407 aspartyl residue and Tyr369 ⁄ O
g
atom on one
side and Arg381 ⁄ N
f
,N
g
atoms on the other side are
observed in this model (Fig. 2). Similar interactions
cannot take place in the C-domain, as these two resi-
dues are replaced by Phe967 and Glu979. As the aspar-
tyl residue in RXP407 is the key residue controlling

inhibitor selectivity, we suggest that the mutations
observed in the 369 ⁄ 967 and 381 ⁄ 979 positions may
contribute to RXP407 selectivity. Rat and mouse ACE
display the same feature as is observed in human ACE,
which is consistent with the potency and selectivity dis-
played by RXP407 toward the rat and mouse enzymes.
It should be mentioned that a model of RXP407 inter-
action with human N-domain was previously reported
that differs from our model [6]. In that model, the
Table 4. Residues located in the S
2
,S

1
,S
1
¢ and S
2
¢ subsites of ACE enzymes possibly implicated in the substrate selectivity. Residues that
change between ACE species are in bold and those that change between the N-domain and C-domain are in italics.
N-domain C-domain
Rat Mouse Human Rat Mouse Human
S
2
Tyr369 Tyr369 Tyr369 Phe967 Phe967 Phe967

Arg381 Arg381 Arg381 Glu979 Glu979 Glu979
S
1
Ser39 Ser39 Ser39 Asn642 Asn642 Asn642
Ser119 Ser119 Ser119 Glu719 Glu719 Glu719
Ser494 Asn494 Asn494 Ala1092 Ala1092 Ser1092
Val495 Val495 Val495 Asn1093 Asn1093 Ser1093
Thr496 Thr496 Thr496 Val1094 Val1094 Val1094
S
1
¢ Ala332 Ala332 Ala332 Ala930 Pro930 Ala930
S

2
¢ Ser260 Ser260 Ser260 Thr858 Thr858 Thr858
Asp354 Glu354 Asp354 Glu952 Glu952 Glu952
Ser357 Ala357 Ser357 Val955 Val955 Val955
Thr358 Thr358 Thr358 Ile956 Ile956 Val956
Glu431 Glu431 Glu431 Asp1029 Asp1029 Asp1029
Fig. 2. Detail of the human ACE N-domain active in the interaction
with RXP407. The active-site helix carrying the HEXXH sequence is
colored yellow. RXP407 is colored by atom type; the carbon atoms
are in green. Active-site residues located at a distance less than
5A
˚

from RXP407 are in light blue when they are observed to
change between ACE species (in either the N-domain or C-domain).
Three residues interacting with RXP407 in the model are displayed
in orange; these residues are conserved between the N-domain of
the three ACE species (N-domain numbering). Distances in ang-
stroms between atoms of Arg381, Tyr369 and the aspartate atoms
of RXP407 are reported. Hydrogens and zinc atom are omitted. The
figure was prepared with
PYMOL software.
N. D. Jullien et al. ACE substrate specificity
FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS 1777
Tyr369 and Arg381 residues were observed to interact

with the acetyl group of RXP407. As mentioned above,
our model seems to better explain the key role of the
aspartyl residue in inhibitor selectivity. As shown in
Fig. 2 (see also Table 4), the model of RXP407 interac-
tion with the N-domain reveals that several positions in
the ACE active site, covering the S
1
to S
2
¢ subsites,
change between the three ACE species, in either the
N-domain or C-domain (494, 354, 357, 930, 956, 1092

and 1093). These mutations may contribute to the dif-
ferences in Mca-Ala and Mca-BK
(1)8)
selectivity
observed with the three ACE species. As the major dif-
ferences in catalytic properties between the N-domain
and C-domain of the three ACE species is reflected by
changes in k
cat
but not K
m
, it might be suggested that

all the aforementioned residues participate in the stabil-
ization of the transition state.
Discussion
The use of highly selective inhibitors of ACE makes it
possible to inhibit one enzyme active site and derive
catalytic parameters for the other active site, free of
inhibitor. The good agreement between the catalytic
parameters obtained by blocking one ACE active site
and those determined with the free somatic form using
inhibitor profiles (Tables 1 and 2) validates the use of
selective inhibitors as convenient tools to study the spe-
cificity of the somatic form of ACE. Thus, with this

approach, the catalytic property of each domain
towards the hydrolysis of a substrate can be determined
without the need to produce mutants or isolate ACE
domains. The results of this approach are also consis-
tent with the shape of the inhibition profiles observed
for each ACE species. Intuitive interpretation of the
inhibition profiles of mouse and rat somatic ACE with
RXP407 and RXPA380 (Fig. 1A,B) suggests that the
Mca-Ala substrate is a strict N-domain-selective sub-
strate of mouse ACE, whereas it is almost cleaved by
the C-domain of rat ACE. This interpretation is con-
firmed by the kinetic parameters reported in Table 1.

Overall, this work based on the study of two syn-
thetic substrates and three different ACE species
reveals that the N-domain and C-domain substrate
selectivity is not conserved between different ACE spe-
cies, implying that the functional role played by each
domain may change from one species to another. Thus,
any conclusion on the N-domain and C-domain sub-
strate selectivity based on a single ACE species could
be misleading. In this respect, it is worth mentioning
that the N-domain and C-domain selectivity towards
physiological substrates of ACE (angiotensin I, brady-
kinin and N-acetyl-seryl-aspartyl-lysyl-proline) has been

established only for the human enzyme [21], because of
the availability of a mutant form of the human enzyme
containing a single functional active site. As far as sub-
strate selectivity is concerned, data obtained in animal
models could be misinterpreted if the properties of the
human enzyme are used. The approach presented in
this study, which determines the selectivity of any sub-
strate whatever the source of ACE, could be used to
resolve this important issue. The study of mouse and
rat ACE selectivity toward natural substrates, such as
N-acetyl-seryl-aspartyl-lysyl-proline, angiotensin I and
bradykinin, should be possible using the same

approach, as the potency and selectivity of the RXP
inhibitors did not appear to vary to any great extent
toward these ACE enzymes. Such studies are relevant
in the light of the differences exhibited by rat and
mouse for the hydrolysis of Mca-Ala and Mca-BK,
which can be viewed as mimics of he natural N-acetyl-
seryl-aspartyl-lysyl-proline and BK
1-8
peptides. It
should be noted that, until the development of inhibi-
tors able to block specifically only one ACE active site,
the question of ACE domain selectivity was not so crit-

ical. The correct interpretation of in vivo data obtained
after animal treatment with RXP inhibitors requires a
good knowledge of the specificity of the domain, free
of inhibitor, towards the physiological substrates of
ACE. This requirement also applies to the natural an-
giotensin-(1–7) peptide, which was reported to act as a
C-domain selective inhibitor of ACE [11]. This know-
ledge is essential to appreciate whether ACE domain-
specific inhibitors represent a new class of inhibitors
showing particular pharmacological profiles of medical
interest. Another major concern is that, according to
the data reported in this study, evaluation of these

domain-selective ACE inhibitors may vary between rat
and mouse models, rendering extrapolation of these
results to the human situation problematic.
In support of our experimental results highlighting
differences in N-domain and C-domain substrate selec-
tivity between ACE species, our nonexhaustive compar-
ison of mouse, rat and human ACE sequences around
the active site of these enzymes identifies several resi-
dues that change (Table 4). In human ACE, it has
already been reported that Asn494 occurs in an N-gly-
cosylation sequon (NTV), that is unique to the N-
domain [6]. Any glycosylation of this residue, which is

located in the active site, is expected to greatly influence
enzyme activity. Interestingly, this asparagine is
replaced by serine in the rat N-domain. Whether this
mutation results in no glycosylation or O-glycosylation
of Ser494 in rat ACE is not known [22], but, in any
case, it can be expected to affect enzyme activity.
The low efficiency of the rat N-domain in cleaving the
ACE substrate specificity N. D. Jullien et al.
1778 FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS
Mca-Ala substrate, compared with the N-domain of
the mouse and human enzymes, may in part be due to
this mutation. The resolution of the 3D structure of the

germinal form of ACE has provided a strong impetus
for further studies aimed at understanding, at the
molecular level, how the two active sites of somatic
ACE function [18]. The data reported in this study pro-
vide supplementary information about the residues that
should be considered in any ACE models intended to
explain the selectivity of this enzyme. These residues
should be considered in future mutagenesis studies
designed to map the residues involved in the specificity
of both the N-domain and C-domain of ACE. Gene
duplication leading to the presence of several protease
domains within a single polypeptide chain is a rare

event, so far only observed in ACE, carboxypeptidase
D and polyserases [23,24]. The putative functional
advantages that may result from such complex protein
assembly remain elusive. Rat and mouse ACE were
observed to cleave the Mca-Ala peptide with similar
efficiency, but surprisingly the contribution of each
domain to this cleavage varies considerably in these
ACE species. In this case, independent evolution of
each domain does not affect the global enzyme activity.
Beside complementary enzyme activity, many other
functional advantages have been proposed for these
multidomain proteases, such as positive or negative co-

operativity [23–26]. Clearly, a better understanding of
the functional role played by each domain in these
intriguing proteases will require additional studies.
Experimental procedures
Chemicals
RXP407 and RXPA380 (Scheme 1) were synthesized as des-
cribed previously in [16] and [13], respectively. Quinaprilat
was kindly provided by Professor P. Corvol of the Institut
National de la Sante
´
et de la Recherche Me
´

dicale, Unite
´
36,
Paris, France. Mca-Arg-Pro-Pro-Gly-Phe-Ser-Pro-DpaOH
(Mca-BK
(1)8)
) substrate [Mca, (7-methoxycoumarin-4-
yl)acetyl; DpaOH, N
3
-(2,4-dinitrophenyl)-l-2,3-diaminopro-
pionyl)] was prepared by following the procedure described
for Mca-Ala-Ser-Asp-Lys-DpaOH (Mca-Ala) [16].

Enzymes
Human wild-type somatic ACE was obtained by stable
expression in Chinese hamster ovary cells transfected with
appropriate ACE cDNA [14]. This material was kindly pro-
vided by P. Corvol (Colle
`
ge de France, Paris, France).
Expression and purification of ACE were performed as pre-
viously described [14]. Mouse and rat somatic ACE were
purified by affinity chromatography as described previously
[17], from lung homogenates obtained from C57BL ⁄ 6 mice
and Lewis rats (Charles River France, L’arbresle, France).

ACEs purified by this method appeared as homogeneous
single bands on SDS ⁄ PAGE.
Enzyme assays
Continuous assays were performed by recording the fluores-
cence increase at 405 nm (e
ex
¼ 320 nm) induced by the
cleavage of Mca-Ala and Mca-BK
(1)8)
substrates by ACE,
using black, flat-bottomed, 96-well nonbinding surface
plates (Corning-Costar, Schiphol-Rijk, the Netherlands).

Assays were carried out in triplicate, at 25 °C, in 50 mm
Hepes (pH 6.8) ⁄ 200 mm NaCl. Fluorescence signals were
monitored using a Fluoroscan Ascent photon counter spec-
trophotometer (Thermo-Labsystems, Courtaboeuf, France)
equipped with a temperature control device and a plate sha-
ker. The substrate and enzyme concentrations for the
experiments were chosen so as to remain well below 10%
of substrate utilization and to observe initial rates. Concen-
trations of substrate solutions were determined spectropho-
tometrically using e
328nm
¼ 12 900 m

)1
Æcm
)1
. The site of
cleavage of Mca-BK
(1)8)
by ACE enzymes was determined
by HPLC analysis coupled with MS. For the three ACE
species, a unique cleavage was observed at Phe-Ser.
Determination of kinetic parameters
The apparent kinetic parameters K
m

and k
cat
for the hydro-
lysis of the substrates Mca-Ala and Mca-BK
(1)8)
by the
N-domain or the C-domain of human, mouse, or rat ACE
were determined by blocking the activity of the C-domain
by 150 nm RXPA380 or the activity of the N-domain by
350 nm RXP407, except for rat ACE for which the
N-domain was inhibited using 750 nm RXP407. At these
concentrations, RXP compounds mainly inhibit the activity

of one active site, allowing the determination of the kinetic
parameters of the other active site, free of inhibitor. For
each species and each inhibitor, ACE was incubated with
the inhibitor for 45 min before substrate addition. The kin-
etic parameters were determined using the direct linear plot
method [27–30] and substrate concentration ranges of 10–
122 lm for Mca-Ala and 2–71 lm for Mca-BK
1-8
. Concen-
trations of ACE in these experiments were determined by
titration of the enzyme with quinaprilat.
Inhibition studies

For inhibition studies, all inhibitors were preincubated for
45 min before initiation of the reaction by substrate addition.
The substrate concentration used was 8 lm for Mca-Ala and
5 lm for Mca-BK
(1)8)
. Data were collected every 30 s over a
period of 25 min. Inhibitor concentrations were selected in
N. D. Jullien et al. ACE substrate specificity
FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS 1779
order to observe a full range of inhibition percentages. With
the wild-type somatic ACE from the three species (human,
mouse, rat), K

i(app)
values were estimated from simulations.
Inhibition profiles of ACE by the different inhibitors with
Mca-Ala and Mca-BK
(1)8)
substrates were simulated using
the program dynafit from P. Kuzmic [31], as described in
[16,19]. A two-active-site model was used to fit these profiles,
except for the titration profiles of mouse ACE using the
Mca-Ala substrate, for which a one-active-site model was
used. In some cases, in order to achieve a better fit between
simulated and experimental data, k

catN ⁄ C
values were varied
within the limit of the experimental confidence values of
these parameters (Tables 1 and 2), keeping the selectivity
(k
cat
⁄ K
m
)
N
⁄ (k
cat

⁄ K
m
)
C
ratio constant.
Identification of the residues covering the
S
2,
S
1,
S
1

¢ and S
2
¢ binding sites in ACE
A model of RXP407 interaction with the human ACE N-
domain has been developed to identify residues of the S
2
,
S
1
,S
1
¢ and S

2
¢ pockets that could influence either inhibitor
or substrate selectivity in this enzyme. The conservation of
these residues was then checked, using aligned sequences of
the N-domain and C-domain of human, mouse and rat
enzymes. The model of the N-domain of human ACE
(ACE-N) was achieved by homology modeling using the
3D structure of the human germinal form of this enzyme
reported in complex with lizinopril (pdbcode 1086) [18].
The initial structure of the N-domain of ACE was obtained
with the program modeller 6v2 [32] by alignment with the
germinal form of ACE (55.2% identity in 583 amino acids

overlap). The resulting structure was then used to build an
initial structure of the complex N-domain–RXP407. The
structure of RXP407 was constructed with the insight ii
software (Accelrys Inc., Sandiego, CA, USA). To obtain
the initial structure of the N-domain–RXP407 complex, we
used the previously reported model of C-domain–RXPA380
[20]. The N-domain structure was superimposed on the
C-domain structure and the RXP407 inhibitor was superim-
posed on RXPA380 by minimizing the r.m.s.d. on the posi-
tion of the common backbone atoms. The resulting
N-domain–RXP407 structure was then refined using a pro-
tocol of molecular dynamics with the program charmm

(v.27) [33]. The charmm force field version 22 was used
[34]. Geometrical and nonbonded parameters for the phos-
phinic inhibitor RXP407 were derived from ab initio quan-
tum calculations with the program gaussian98 (Gaussian
Inc., Pittsburgh, PA, USA). These calculations were per-
formed at the MP2 level of theory using a 6–31 + G(d,p)
basis set. To preserve the structure of the protein during
the relaxation of the complexes, harmonic restraints were
applied to the atomic position of several sets of atoms.
The harmonic constants were set to 100, 5 and 0.5 kcalÆ
mol
)1

ÆA
˚
)2
for the ions and their chelating residues, the
atoms of the protein situated at a distance greater than 5 A
˚
of the inhibitor, and the backbone and Cb atoms of the
inhibitor and those of the protein located at a distance
smaller than 5 A
˚
of the inhibitor, respectively. No har-
monic restraints were applied to hydrogen atoms. The ini-

tial step of the relaxation protocol consists of an initial
2000 cycles of Adopted Basis Newton-Raphson energy min-
imization. Then 100 000 steps of molecular dynamics using
the Verlet algorithm were undertaken. The integration step
was set to 0.0004 ps. The temperature was gradually
increased by 25 K each 1000 steps to reach 300 K. This
molecular dynamics was followed by 5000 cycles of energy
minimization. During the calculations, the nonbonded
interactions were modeled using a Lennard-Jones function
and a coulombic electrostatic term with a nonbonded cut-
off of 16 A
˚

. The dielectric constant was set to 1. The result-
ing structure was then analyzed with the program pymol
(Delano Scientific Inc., San Francisco, CA, USA).
Acknowledgements
This work was supported by the CEA ( Commissariat a
`
l’Energie Atomique). N.J. was funded through a fellow-
ship from French Government and Servier Institut.
References
1 Erdos EG (1990) Angiotensin I converting enzyme and
the changes in our concepts through the years. Lewis K.
Dahl memorial lecture. Hypertension 16, 363–370.

2 Bhoola KD, Figueroa CD & Worthy K (1992) Bioregu-
lation of kinins: kallikreins, kininogens, and kininases.
Pharmacol Rev 44, 1–80.
3 Gavras H (1994) Corcoran Lecture. Angiotensin-con-
verting enzyme inhibition and the heart. Hypertension
23, 813–818.
4 Dzau VJ (2001) Theodore Cooper Lecture: tissue angio-
tensin and pathobiology of vascular disease: a unifying
hypothesis. Hypertension 37, 1047–1052.
5 Zaman MA, Oparil S & Calhoun DA (2002) Drugs tar-
geting the renin-angiotensin-aldosterone system. Nat
Rev Drug Discov 1, 621–636.

6 Acharya KR, Sturrock ED, Riordan JF & Ehlers MR
(2003) ACE revisited: a new target for structure-based
drug design. Nat Rev Drug Discov 2, 891–902.
7 Soubrier F, Alhenc-Gelas F, Hubert C, Allegrini J, John
M, Tregear G & Corvol P (1988) Two putative active
centers in human angiotensin I-converting enzyme
revealed by molecular cloning. Proc Natl Acad Sci USA
85, 9386–9390.
8 Rousseau A, Michaud A, Chauvet MT, Lenfant M &
Corvol P (1995) The hemoregulatory peptide N-acetyl-
Ser-Asp-Lys-Pro is a natural and specific substrate of
the N-terminal active site of human angiotensin-convert-

ing enzyme. J Biol Chem 270, 3656–3661.
ACE substrate specificity N. D. Jullien et al.
1780 FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS
9 Deddish PA, Wang LX, Jackman HL, Michel B, Wang J,
Skidgel RA & Erdos EG (1996) Single-domain angioten-
sin I converting enzyme (kininase II): characterization
and properties. J Pharmacol Exp Ther 279, 1582–1589.
10 Deddish PA, Jackman HL, Skidgel RA & Erdos EG
(1997) Differences in the hydrolysis of enkephalin
congeners by the two domains of angiotensin converting
enzyme. Biochem Pharmacol 53, 1459–1463.
11 Deddish PA, Marcic B, Jackman HL, Wang HZ, Skid-

gel RA & Erdos EG (1998) N-domain-specific substrate
and C-domain inhibitors of angiotensin-converting
enzyme: angiotensin-(1–7) and keto-ACE. Hypertension
31, 912–917.
12 Junot C et al. (2001) RXP 407, a selective inhibitor of
the N-domain of angiotensin I-converting enzyme,
blocks in vivo the degradation of hemoregulatory pep-
tide acetyl-Ser-Asp-Lys-Pro with no effect on angioten-
sin I hydrolysis. J Pharmacol Exp Ther 297, 606–611.
13 Georgiadis D, Beau F, Czarny B, Cotton J, Yiotakis A
& Dive V (2003) Roles of the two active sites of somatic
angiotensin-converting enzyme in the cleavage of angio-

tensin I and bradykinin: insights from selective inhibi-
tors. Circ Res 93, 148–154.
14 Wei L, Alhenc-Gelas F, Corvol P & Clauser E (1991)
The two homologous domains of human angiotensin
I-converting enzyme are both catalytically active. J Biol
Chem 266, 9002–9008.
15 Jaspard E, Wei L & Alhenc-Gelas F (1993) Differences
in the properties and enzymatic specificities of the two
active sites of angiotensin I-converting enzyme (kininase
II). Studies with bradykinin and other natural peptides.
J Biol Chem 268, 9496–9503.
16 Dive V, Cotton J, Yiotakis A, Michaud A, Vassiliou S,

Jiracek J, Vazeux G, Chauvet MT, Cuniasse P & Corvol
P (1999) RXP 407, a phosphinic peptide, is a potent
inhibitor of angiotensin I converting enzyme able to dif-
ferentiate between its two active sites. Proc Natl Acad
Sci USA 96, 4330–4335.
17 Vazeux G, Cotton J, Cuniasse P & Dive V (2001)
Potency and selectivity of RXP407 on human, rat, and
mouse angiotensin-converting enzyme. Biochem Pharma-
col 61, 835–841.
18 Natesh R, Schwager SL, Sturrock ED & Acharya KR
(2003) Crystal structure of the human angiotensin-
converting enzyme-lisinopril complex. Nature 421 , 551–

554.
19 Cotton J, Hayashi MA, Cuniasse P, Vazeux G, Ianzer
D, De Camargo AC & Dive V (2002) Selective inhibi-
tion of the C-domain of angiotensin I converting
enzyme by bradykinin potentiating peptides. Biochemis-
try 41, 6065–6071.
20 Georgiadis D, Cuniasse P, Cotton J, Yiotakis A & Dive
V (2004) Structural determinants of RXPA380, a potent
and highly selective inhibitor of the angiotensin-convert-
ing enzyme C-domain. Biochemistry 43, 8048–8054.
21 Michaud A, Williams TA, Chauvet MT & Corvol P
(1997) Substrate dependence of angiotensin I-converting

enzyme inhibition: captopril displays a partial selectivity
for inhibition of N-acetyl-seryl-aspartyl-lysyl-proline
hydrolysis compared with that of angiotensin I. Mol
Pharmacol 51, 1070–1076.
22 Christlet THT & Veluraja K (2001) Database analysis
of O-glycosylation sites in proteins. Biophys J 80, 952–
960.
23 Novikova EG, Eng FJ, Yan L, Qian YM & Fricker LD
(1999) Characterization of the enzymatic properties of
the first and second domains of metallocarboxypepti-
dase D. J Biol Chem 274, 28887–28892.
24 Cal S, Quesada V, Llamazares M, Diaz-Perales A,

Garabaya C & Lopez-Otin C (2005) Human polyserase-
2, a novel enzyme with three tandem serine protease
domains in a single polypeptide chain. J Biol Chem 280,
1953–1961.
25 Binevski PV, Sizova EA, Pozdnev VF & Kost OA
(2003) Evidence for the negative cooperativity of the
two active sites within bovine somatic angiotensin-con-
verting enzyme. FEBS Lett 550, 84–88.
26 Woodman ZL, Schwager SL, Redelinghuys P, Carmona
AK, Ehlers MR & Sturrock ED (2005) The N domain
of somatic angiotensin-converting enzyme negatively
regulates ectodomain shedding and catalytic activity.

Biochem J 389, 739–744.
27 Eisenthal R & Cornish-Bowden A (1974) The direct
linear plot. A new graphical procedure for estimating
enzyme kinetic parameters. Biochem J 139, 715–720.
28 Cornish-Bowden A & Eisenthal R (1974) Statistical con-
siderations in the estimation of enzyme kinetic para-
meters by the direct linear plot andother methods.
Biochem J 139, 721–730.
29 Cornish-Bowden A & Eisenthal R (1978) Estimation of
Michaelis constant and maximum velocity from the
direct linear plot. Biochim Biophys Acta 523, 268–272.
30 Cornish-Bowden A, Porter WR & Trager WF (1978)

Evaluation of distribution-free confidence limits for
enzyme kinetic parameters. J Theor Biol 74, 163–175.
31 Kuzmic P (1996) Program DYNAFIT for the analysis
of enzyme kinetic data: application to HIV proteinase.
Anal Biochem 237, 260–273.
32 Sali A & Blundell TL (1993) Comparative protein mod-
elling by satisfaction of spatial restraints. J Mol Biol
234, 779–815.
33 Brooks BR, Bruccoleri RE, Olafson BD, States DJ,
Swaminathan S & Karplus M (1983) charm: a program
for macromolecular energy, minimization, and calculat-
ions. J Comput Chem 4, 187–217.

34 MacKerell AD Jr, Bashford D, Bellott M, Dunbrack
RL Jr, Evanseck JD, Field MJ, Fischer S, Gao H, Ha
S, Joseph-McCarthy D, et al. (1998) All-atom empirical
potential for molecular modeling and dynamics studies
of proteins. J Phys Chem B 102, 3586–3616.
N. D. Jullien et al. ACE substrate specificity
FEBS Journal 273 (2006) 1772–1781 ª 2006 The Authors Journal compilation ª 2006 FEBS 1781