5086
A. G. Tzakos et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5084–5087
Table 2. Determination of kinetic constants for the hACE and its
Y1096F and K1087A mutants
A grid flexible docking-calculation search has been per-
formed, by mutating iteratively the enzyme carboxylate
docking partners of captopril (since the rest of the mol-
ecule shares the same architectural topology), allowing
to identify a case in which both isomers could have
equal probability to be accommodated by the enzyme.
Specifically, the Y1096F mutant resulted in comparable
free energies of binding for both states of the inhibitor
(ꢁ ꢀ6.2 kcal/mol for cis and ꢁ ꢀ6.7 kcal/mol for trans).
This can be rationalized by the fact that the carboxylate
group of the inhibitor could be selectively ‘locked’ in the
trans conformation through ionic interactions with the
side chain of K1087 (Fig. 2B), whilst the polar to hydro-
phobic alteration accomplished in the Y1096F mutant,
as also the elimination of the bulky hydroxyl group of
Y1096, open access to a strong hydrophobic packing
of the pyrrolidine ring of the inhibitor, in the cis form,
with the side chains of F1096 and Y1099 (Fig. 2D).
Interestingly, a similar intra-molecular aromatic (Tyr)–
proline interaction has been recently suggested to stabi-
lize the cis conformation of the Tyr–Pro amide bond,18
whereas in the current case the aromatic residue is pro-
vided by the enzyme (Y1099).
Enzyme
logEC50 value-1
logEC50 value-2
hACE C-domain
Y1096F
K1087A
ꢀ4.6735
ꢀ2.8469
ꢀ4.7319
—
ꢀ4.2347
—
modated by the modified enzyme, either the cis or the
trans form of the inhibitor should be engineered and re-
strained achieving a more potent drug with complemen-
tary architecture to the native enzyme binding groove.
In conclusion, the combination of pharmacophore-
hypotheses and protein-structure based virtual screening
is successful in developing lead compounds. However, as
the present study revealed, many uncertainties remain
about the details of the ligand–receptor interactions,
possibly restraining the fine-tuning of the lead com-
pound to a more potent drug. Specifically, our NMR
studies on the ACE inhibitor captopril in solution were
able to map the presence of equal population of the cis
and trans conformational states at physiological pH. A
flexible docking grid-search allowed the identification
of a point mutation (Y1096F) of the hACE-C enzyme
that could preferentially incorporate both states of the
inhibitor with the same probability. This in silico finding
has been independently validated through mutagenesis
and enzymatic studies. The results of this combinatorial
approach through NMR, flexible docking, mutagenesis,
and enzymatic studies pinpoint the importance of the
multiple conformational states of the inhibitor that
should be taken into account in the frame of the drug
design process so as to maximize the inhibitory affinity
for the enzyme and minimize the side-effects of the inter-
action. The next step in this line of research will be the
utilization of our approach toward the investigation of
the current system in a drug-refinement effort.9
In order to validate these findings, we mutated the
hACE C-domain residues Y1096 and K1087 to Phe
and Ala, respectively,19 and detailed kinetic studies have
been performed on these mutants. Indeed, Figure 3
shows that the Y1096F and K1087A mutations pro-
duced a decrease in captopril binding affinity (1/Ki),
indicating experimentally their importance for captopril
docking. However, we noticed an interesting phenome-
non: both mutants behave completely differently in their
inhibitor profile and, most importantly, data with capto-
pril inhibition for Y1096F best fit using a two-site mod-
el, whereas data for wild-type hACE C-domain and the
K1087A mutant fit on one-site model (Table 2).
Given these considerations, we could suggest that the
observed equimolar ratio of the two states (cis, trans)
of the inhibitor in solution is related to the two states
observed in our kinetic studies for the Y1096F mutant
that could accommodate both of them with the same
probability in accordance with the docking calculations.
Thus, since both states of the inhibitor could be accom-
Acknowledgments
Dr. L. Elantak, Dr. N. Locker, and Dr. A. Troganis are
gratefully acknowledged for suggestions and comments
on the manuscript. AGT acknowledges an award from
the Institute ‘Leonidas Zervas’. This research was fund-
ed by the program ‘Heraklitos’ of the Operational Pro-
gram for Education and Initial Vocational Training of
the Hellenic Ministry of Education under the 3rd Com-
munity Support Framework and the European Social
Fund. GRST PENED-99 Program (to A.G.T., I.P.G.),
and EC Access to Research Infrastructures Action of
the Improving Human Potential Programme (Contract
No. HPRI-CT-2001-00147). A.G.T. and K.C. acknowl-
edge support from the EC Marie Curie Programme
(No. HPMT-CT-2000-00137).
Supplementary data
Computational methods, details of the enzymatic stud-
ies, and thermodynamic parameters determined by
NMR. Supplementary data associated with this article
Figure 3. Inhibition of wild-type hACE C-domain, Y1096F and
K1087A mutants of hACE C-domain by the ACE inhibitor captopril.