Conformational and solvation studies via computer simulation of the novel large scale diastereoselectively synthesized phosphinic MMP inhibitor RXP03 diluted in selected solvents

Article abstract of DOI:10.1021/jp903830v

Structure-activity relationship studies, regarding the influence of side chains of phosphinic pseudotripeptidic inhibitors on matrix metalloproteinases (MMPs), provided potent and selective inhibitors for this family of structurally and functionally related proteases. Among them, phosphinic pseudopeptide CbzPheψ [P(O)(OH)CH₂] phenylpropyl TrpNH₂, known as RXP03, has been extensively used for in vivo and in vitro studies so far. The large quantities of RXP03 required for in vivo studies, as well as the necessity for diastereoisomeric purity, motivated us to further explore and develop an efficient synthetic methodology, which allows separation of the four diastereoisomers of RXP03 based on the astonishing observed differences in solubility of the four isomers in various solvents. This fact prompted us to examine theoretically the conformational differences of these four isomers via computer simulations in the solvents used experimentally. Given the fact that the four examined diastereoisomeric forms of the phosphinic peptides exhibit different behavior in terms of potency and selectivity profiles toward zinc-metalloproteases, this theoretical study provides valuable information on the conformation of phosphinic inhibitors and therefore improves the design and synthesis of active structures. The differences in solubility of RXP03 diastereoisomers in the used solvents were examined in terms of intra- and intermolecular structure. It is found that the different solubility of the RRS and RSS diastereoisomers in EtOH is a result of the different number of hydrogen bonds formed by each isomer with EtOH molecules. In the case of SRS and SSS in Et₂O, their different solubility might be attributed to the different intramolecular hydrogen bonds formed on these diastereoisomers.

Full text of DOI:10.1021/jp903830v

J. Phys. Chem. B 2010, 114, 421–428
421
Conformational and Solvation Studies via Computer Simulation of the Novel Large Scale
Diastereoselectively Synthesized Phosphinic MMP Inhibitor RXP03 Diluted in Selected
Solvents
Magdalini Matziari,† Dimitris Dellis,‡ Vincent Dive,§ Athanasios Yiotakis,*,† and
Jannis Samios*,‡
Department of Chemistry, Laboratory of Organic Chemistry, UniVersity of Athens, Panepistimiopolis Zografou
15771, Athens, Greece, Department of Chemistry, Laboratory of Physical Chemistry, UniVersity of Athens,
Panepistimiopolis Zografou 15771, Athens, Greece, and CEA, SerVice D’Inge´nierie Mole´culaire des Prote´ines
(SIMOPRO), Bat 152, CE-Saclay, Gif/YVette Cedex 91191, France
ReceiVed: April 26, 2009; ReVised Manuscript ReceiVed: September 12, 2009
Structure-activity relationship studies, regarding the influence of side chains of phosphinic pseudotripeptidic
inhibitors on matrix metalloproteinases (MMPs), provided potent and selective inhibitors for this family of
structurally and functionally related proteases. Among them, phosphinic pseudopeptide CbzPheψ
[P(O)(OH)CH2] phenylpropyl TrpNH2, known as RXP03, has been extensively used for in vivo and in vitro
studies so far. The large quantities of RXP03 required for in vivo studies, as well as the necessity for
diastereoisomeric purity, motivated us to further explore and develop an efficient synthetic methodology,
which allows separation of the four diastereoisomers of RXP03 based on the astonishing observed differences
in solubility of the four isomers in various solvents. This fact prompted us to examine theoretically the
conformational differences of these four isomers via computer simulations in the solvents used experimentally.
Given the fact that the four examined diastereoisomeric forms of the phosphinic peptides exhibit different
behavior in terms of potency and selectivity profiles toward zinc-metalloproteases, this theoretical study
provides valuable information on the conformation of phosphinic inhibitors and therefore improves the design
and synthesis of active structures. The differences in solubility of RXP03 diastereoisomers in the used solvents
were examined in terms of intra- and intermolecular structure. It is found that the different solubility of the
RRS and RSS diastereoisomers in EtOH is a result of the different number of hydrogen bonds formed by
each isomer with EtOH molecules. In the case of SRS and SSS in Et2O, their different solubility might be
attributed to the different intramolecular hydrogen bonds formed on these diastereoisomers.
1. Introduction
Phosphinic peptides (Figure 1a) have gained significant
attention during the past decade due to their inhibitory effect
toward proteases, mainly zinc- and aspartyl proteases, which
are implicated in numerous physiological and pathological
conditions.1 Although the phosphinyl group is a relatively weak
Zn-binding group, optimized structures gain not only in potency
but also mostly in selective inhibition of these enzymes.2 Acting
as transition state analogues, phosphinic peptides exhibit
interesting properties for drug candidates:3 They are devoid of
toxicity, stable in vivo, and are bound to be elongated by both
sides of the phosphinyl group, allowing thus optimization of
the inhibitor selectivity by diversification of both primed and
unprimed (P1 and P1′) positions (Figure 1a).4 Note also that
phosphinic peptides act as potent inhibitors of many metallo-
proteases such as angiotensin-converting enzyme (ACE),5
matrixins metalloproteinases (MMPs),6,7 aminopeptidase A
(APA),8 leucine aminopeptidase,9 etc.
Figure 1. (a) Generic structure of the phosphinic inhibitors; (b) RXP03
structure. For the inhibitory potencies of RXP03 toward some MMPs,
see ref 3.
* Corresponding author. E-mail: isamios@chem.uoa.gr, yiotakis@
Phosphinic peptides can be synthesized by using either
parallel or combinatorial chemistry strategies10,11 both in solution
and on solid support.12,13 The phosphinic pseudopeptide
CbzPheψ[P(O)(OH)CH2] phenylpropyl TrpNH2 (RXP03) (Fig-
ure 1b) has been shown to be a potent inhibitor of MMP-2,
chem.uoa.gr.
† Department of Chemistry, Laboratory of Organic Chemistry, University
of Athens.
‡ Department of Chemistry, Laboratory of Physical Chemistry, University
of Athens.
§ CEA.
10.1021/jp903830v  2010 American Chemical Society
Published on Web 12/16/2009

422 J. Phys. Chem. B, Vol. 114, No. 1, 2010
Matziari et al.
MMP-8, MMP-9, MMP-11, and MMP-14 but not of MMP-1
and MMP-7. On the basis of these results, RXP03 was used in
an evaluation study of the expression of the aforementioned
MMPs and their TIMPs (tissue inhibitors of matrix metallo-
proteinases) in a rat model of liver ischemia/reperfusion.14 In a
later study, the in vivo disposition and antitumor efficacy of
RXP03 was studied, revealing a dosing and scheduling influence
of RXP03 on the growth of primary tumors.15 In addition, the
3D crystal structure of the complex of the phosphinic inhibitor
RXP03 in interaction with the catalytic domain of MMP-11 has
been recently resolved and the results obtained provided valuable
information on the conformation of this enzyme’s active site.16
Given the fact that the four aforementioned isomers of RXP03
exhibit different inhibitory potencies against MMPs, we devel-
oped a new diastereoselective method to synthesize each of these
four isomers. In addition, the fact that each isomer was found
to exhibit different solubility properties motivated us to further
investigate the behavior of these isomers in the solvents
experimentally used, namely, ethanol (EtOH) and diethyl ether
(Et2O). To obtain insight of the different solubility of the isomer
pairs, namely, RRS, RSS in EtOH, and SRS and SSS in Et2O,
obviously one needs a thorough investigation of the presumably
different strength of the inter- and intramolecular interactions
among solute and solvent molecules. Aiming at this goal, in
the framework of the present study, we decided to employ
appropriate computational techniques to explore theoretically
the underlying molecular mechanisms in detail. To realize the
target of this study, we used quantum mechanical calculations
and molecular dynamics simulation techniques. Quantum me-
chanical calculations were used to optimize the geometry of
each isolated isomer and further to obtain the distribution of
the electrostatic charges on their atoms. On the other hand, the
molecular dynamics simulation of the diluted isomer solutions
led us to extract useful information concerning properties of
interest.
SCHEME 1: Synthetic Procedure and Resolution of the
Four Diastereoisomers I-IV of RXP03: (a) Resolution of
1 to 1a and 1b; (b) Michael-Type Addition to 2; (c)
Saponification of 2a and 2b; (d) Coupling of 3a and 3b
with TrpNH2; (e) Resolution of I and II by
Recrystallization in Absolute EtOH (I Insoluble, II
Soluble); (f) Resolution of III and IV by Recrystallization
in Et2O (III Insoluble, IV Soluble)
The remainder of this paper is organized as follows. Section
2 is devoted to the experimental part of the study. In section 3,
we outline the computational methods used, while, in section
4, we present and discuss the results obtained. Finally, the last
section of the paper contains the main conclusions.
2. Experimental Section
2.1. Synthetic Method. As previously reported,6 RXP03 has
been synthesized using a solid-phase peptide synthesis protocol.
Note however that the method proposed therein is not suitable
for the requirements of a large-scale synthesis of RXP03
isomers. In order to overcome this hurdle, a new synthetic
method has been developed in the framework of the present
study, which according to Scheme 1 allows the separation of
the four isomers of RXP03 by simple recrystallization in
appropriate solvents. It should be mentioned here that the
previously reported method allowed their separation only by
using cumbersome chromatographic (RP-HPLC) techniques.
The synthetic protocol presented herein allows the synthesis of
the four diastereoisomers of RXP03 in gram scale and high
yields. In detail, R (1a) and S (1b) Cbz-protected phosphinic
acid analogues of phenylalanine were synthesized as a mixture
of enantiomers (1), and they were separated using (R)-(+)-R-
methylbenzylamine for 1a and (S)-(-)-R-methylbenzylamine
for 1b.17 Michael type addition of 1a and 1b to R-subsituted
phenylpropyl acrylic ethylester6 leads to the phosphinic dipep-
tides 2a and 2b, respectively. Removal of the ethylester group
under basic conditions produced 3a and 3b. Coupling with (S)-
TrpNH2 provided the phosphinic pseudotripeptides 4a and 4b
as mixtures of two pairs of isomers.10 To our astonishment, it
has been observed that each isomer proved to have different
solubility properties. When, for instance, 4a was treated with
absolute ethanol (EtOH), a solid precipitated, which proved to
be isomer I, while the filtrate consisted exclusively of isomer
II. On the other hand, when 4b was treated with diethyl ether

Phosphinic MMP Inhibitor RXP03
J. Phys. Chem. B, Vol. 114, No. 1, 2010 423
(Et2O), isomer III was found to be insoluble, while IV was
soluble in this solvent. The purity of each isomer was 99%, as
obtained by using the well-known experimental techniques RP-
HPLC and NMR. Finally, assignment of the absolute config-
uration of each isomer was performed according to methods
described in detail previously.10,18
We found that at B3LYP/6-31G* or HF/6-31G** the obtained
partial charges converge. The basis set dependence of obtained
partial charges of each RXP03 atom of RRS isomer is presented
in Figure S2 of the Supporting Information. All QM DFT
calculations were performed using the GAMESS package.20
3.3. Molecular Dynamics Simulations. Molecular dynamics
simulation was used to study the properties of the four RXP03
stereoisomers in the solvent environment, namely, RRS and RSS
in EtOH and SRS and SSS in Et2O at room temperature and
the density of each system corresponding to an atmospheric
pressure of 1 atm. The Gromos96 force field21 was used to
describe intra- and intermolecular interactions between atoms
of the RXP03 and solvent molecules. The Gromos forcefield
includes intra- and intermolecular interactions. Intramolecular
interactions include bond stretching, angle bending, proper and
improper dihedral torsions, van der Waals, as well as electro-
static interactions between sites separated by three or more
bonds. Intermolecular interactions include van der Waals and
electrostatic interactions. The functional form of these terms is
presented in eqs 1-6.
3. Computational Methods
3.1. General Considerations. As mentioned in the Introduc-
tion, the different solubility properties of the four RXP03
isomers obtained in various solvents in conjunction with the
different inhibitory effect of each isomer toward Zn-metallo-
proteases10 motivated us to investigate the conformational
differences exhibiting these isomers, due to the different strength
of the intra- and intermolecular interactions with solvents. The
main question here is therefore under investigation: how the
molecular shape of each isomer and its dynamical folding affects
the characteristic parts of the isomer that are exposed to solvent
molecules, leading presumably to different solubilities. We recall
here that, to make clear the origin of such a behavior, the isolated
molecular shape of each of the four isomers was investigated
by employing a well-known geometry optimization procedure
using quantum mechanical calculations. On the basis of the
quantum mechanical results, we also employed the molecular
dynamics (MD) simulation technique in order to make possible
the interpretation of, among others, the main factors that could
affect the solvation of the four isomers in the solvents used.
Thus, by employing accurate potential models to describe intra-
and interatomic interactions, we were able to investigate possible
changes of the molecular shape of the four isomers in these
solvents and, specifically, how it is affected by the presence of
the solvent molecules. In addition, it was possible to reveal
which parts of the isomer studied are close exposed to solvent
molecules, and what kind of interactions on the atomic level
are responsible for the solubility or not of the solute in the
solvent under investigation. Further, possible hydrogen bonding
(HB) between atoms of the diastereoisomer (intra-HB) as well
as between isomer and protic solvent (inter-HB) have been
systematically analyzed and interpreted.
3.2. Geometry Optimization. Geometry optimization of the
isolated RXP03 isomer molecules, namely, RRS, RSS, SRS,
and SSS, was performed at the DFT/B3LYP level of theory
with the 6-31G* basis set. Further to this and based on the
optimized molecular geometry of each of the four RXP03 isomer
molecules, we calculated the distribution of the partial atomic
charges of each system by using the well-known CHELPG
method.19 In summary, the CHELPG method employs a least-
squares fitting procedure to determine the set of atomic partial
charges of the isolated molecules that reproduces the quantum
mechanical (QM) electrostatic potential at selected grid points.
In this study, the constructed grid was extended to 3 Å from
any of the atomic centers on the molecules and the grid spacing
was set to be equal to 0.1 Å. The partial atomic charges on
RXP03 molecules were calculated for all atoms of the isolated
molecules except all of the -CH2- and -CH- aliphatic groups
and phosphinic OH that were treated as single united interaction
sites. The charge on these united interaction sites was located
on the center of mass of each site. The fitting procedure was
subject to the constraint that the sum of the partial charges
should be equal to zero for each isomer. The optimized geometry
and atomic charge distribution of each isolated isomer of RXP03
obtained is summarized in Tables S1-S4 of the Supporting
Information. The basis set dependence of obtained partial
charges was examined by applying the same procedure using
various basis sets and the HF or DFT/B3LYP levels of theory.
12
6
σij
σij
VvdW(rij) ) 4εij
-
(1)
(2)
( )
( )
[
]
rij
rij
qiqj
Vcoul(rij) )
4πε0rij
1
4
2
2 2
Vstretch(rij) ) kstretch(rij - r0 )
(3)
(4)
(5)
1
Vbend(ϑ) ) kbend(cos(ϑ) - cos(ϑ0))2
2
1
Vimpr(ꢀ) ) kimpr(ꢀ - ꢀ0)2
2
4
1
Vtors(ꢁ) )
(ktors,m[1 - (-1)m ×
∑
2 m)1
cos(mꢁ)]) (6)
As mentioned above, the partial charges on interaction sites were
those obtained by the QM calculations of the isomer molecules.
In all cases, the cutoff distance of the nonbonded interactions
was set equal to 14 Å. The long-range electrostatic interactions
were described by the well-known particle mesh (PME)
method.22,23
For each simulated molecular system, its initial configuration
was prepared by placing the QM optimized structure of the
RXP03 isomer molecule in the center, of a previously equili-
brated solvent box. The number of the added solvent molecules
was such to ensure the simulation box length to be at least 40
Å with no overlaps between solute and solvent molecules.
Depending on the RXP03 isomer and solvent, this results to
710-712 EtOH and 441-494 Et2O molecules around the solute
in the simulation box. This initial solute-solvent configuration
was further optimized by minimization of the simulation box
potential energy using the L-BFGS minimizer,24 followed by a
1 ns constant temperature and a pressure of 1 bar equilibration
run using the Berendsen thermostat and barostat25 with relaxation
times of 0.2 and 1 ps, respectively. The resulting final config-
uration of the system was employed as the initial one for 10 ns
NVT at 298 K extended runs using a Nose-Hoover thermostat26,27
with a relaxation time of 0.2 ps, on which the properties of
interest were calculated. The equations of motion were integrated
using the leapfrog scheme28 with a time step of 1 fs, while the
intramolecular one was handled with the SHAKE method.29 All
of the systems were simulated under periodic boundary condi-

424 J. Phys. Chem. B, Vol. 114, No. 1, 2010
Matziari et al.
Figure 2. Two-dimensional RXP03 molecular structure with marked
atoms.
tions, and the center of mass motion was removed every 0.1
ps. In addition, for each system and in order to improve the
results for the potential energy surface,30 a few 1 ns long runs
were performed starting from a given configuration of RXP03
isomer prepared from a randomly perturbed initial structure.
The perturbation of the initial structure was achieved by a
random change of the equilibrium backbone dihedral angles up
to 10°. For the two asymmetric carbon atoms of interest (see
Figure 2), the possible isomerization was examined by the
calculation of the dihedral angle distribution between the first
side chain connected to an asymmetric carbon atom and the
plane defined by the two backbone atoms connected to these
carbon atoms. It was found that the distribution of these angles
is centered near the equilibrium dihedral angle of each isomer,
with a small width distribution. The dihedral angle distribution
is presented in Figure S3 of the Supporting Information. Finally,
all of the MD simulations were performed using the GROMACS
package.31
Figure 3. Calculated O-OEtOH and O-HEtOH RDFs. The peak at short
distances is the O-HEtOH RDF, and the peak at higher distances is the
O-OEtOH RDF.
RRS and RSS in EtOH. Since EtOH is a well-known
hydrogen bonding fluid at liquid32 and supercritical states,33 we
first investigate the RDFs between O, N and H of amino group
atoms of the RRS and RSS isomers with O and H of EtOH.
Due to the strong electrostatic interactions, it is expected that
the EtOH molecules interact with O and NHx sites of RXP03,
obviously if they are exposed to solvent molecules. On the other
hand, the aromatic groups on the solute interact mainly via van
der Waals interactions with the nonpolar parts of the solvent
(CH3), if they are exposed to solvent. In Figure 2, we present
the notation for O, N, and H atoms on the RXP03 molecules.
From these RDFs, those that exhibit interesting features,
namely, the O1, O2, O4, and O5 oxygen atoms of RRS and
RSS isomers with O and H atoms of EtOH, are presented in
Figure 3. The RDFs of O3 and O6 with O and H atoms of
EtOH show typical liquid-like behavior, similar for both isomers.
By carefully inspecting the O-OEtOH and O-HEtOH RDFs (O1,
O2, O4, O5) shown in Figure 3, we may observe the charac-
teristic behavior of RDFs typical for hydrogen bonded fluids.
Concretely speaking, the O-HEtOH RDFs for both isomers
exhibit first peaks of high amplitude located at a very short
distance of 1.5-1.6 Å, while the O-OEtOH RDFs reveal a lower
first peak at larger correlation distances in the range 2.4-2.6
Å. However, by comparing these correlations to each other, we
easily observe that the oxygen atoms O1 and O2 with O and H
atoms of the EtOH molecule reveal a similar behavior for both
isomers RRS and RSS. Therefore, we may conclude that RRS
and RSS isomers form strong hydrogen bonds with EtOH
molecules quite close to the O1 and O2 atoms.
In the case of the phosphinic oxygen atom O4, the calculated
RDFs for both isomers in EtOH, exhibit first peaks of lower
amplitude compared to those of O1 and O2 oxygen atoms. Note
also that the correlations corresponding to the RRS isomer are
higher than those of the RSS one.
It is of particular interest to discuss at this point the results
obtained for the RDFs corresponding to the oxygen atom O5
(P2 position of the molecule, see Figure 1). The corresponding
RDFs show that the RRS isomer exhibits a relatively high first
peak at low correlation distances in the O5-OEtOH and
O5-HEtOH correlations, while, in the case of RSS isomer, EtOH
molecules are almost absent at distances lower than 4 Å. In
other words, the oxygen atom O5, as a potential candidate site
to form hydrogen bonds with EtOH molecules at this position
of RSS isomer, seems not to be in this particular case because
4. Results and Discussion
4.1. Intermolecular Structure. The structure of RXP03
isomer in the EtOH and Et2O environments was investigated
using the site-site radial distribution functions g(r) (RDF)
between the isomers and solvent molecules. Due to the large
number of site-site combinations, we will present below those
that exhibit interesting features, namely, those that could support
the differences in the solvation process of isomers in EtOH and
Et2O. The corresponding RDFs obtained from the simulations
using the perturbed RXP03 isomer initial configurations were
found to be close to those obtained from the 10 ns simulations.
This is an indication that the initial configuration has a small
effect on the calculated RDFs between certain sites of the four
RXP03 isomers with solvents. This might be expected due to
the relatively small RXP03 structure that can quickly relax close
to the equilibrium structure and the fact that we deal with
noncharged molecules, where the initial position of the coun-
terions could affect the relaxation. For comparison, a few of
these RDFs, namely, those that exhibit differentiation between
these runs, are presented in Figure S1 of the Supporting
Information.
The extent of RXP03 isomer structural changes in the solvent
environment was further investigated by the calculation of the
root mean squared deviation (rmsd) with respect to the initial
configuration of the simulation. This initial configuration of the
simulation is the configuration that results after energy mini-
mization and equilibration in the NPT ensemble to achieve
atmospheric pressure. In all cases, the backbone rmsd was found
to be in the range 0.08-0.14 nm. This means that the structure
of RXP03 isomers is quite stable during simulation in the solvent
environment. The time dependence of the four RXP03 isomer’s
rmsd’s in solution is presented in Figure S4 of the Supporting
Information.

Phosphinic MMP Inhibitor RXP03
J. Phys. Chem. B, Vol. 114, No. 1, 2010 425
TABLE 1: Solvent Accessible Area and Its Hydrophobic
and Hydrophilic Parts of RRS and RSS Isomers in EtOH
and SRS and SSS in Et2O
EtOH
RRS
Et2O
RSS
SRS
SSS
total area (nm2)
10.42
7.95
2.47
8.74
6.25
2.49
10.43
8.37
2.06
9.53
7.82
1.71
hydrophobic area (nm2)
hydrophilic area (nm2)
functions are higher in the case of SSS and lower for SRS. On
the other hand, in the case of the N4/H4-O RDFs, we see a
large differentiation between the two isomers SRS and SSS.
These RDFs are much higher in the case of the SRS isomer
compared to those of the SSS isomer. This fact suggests that,
in the case of SRS isomer, more Et2O molecules are able to be
located at short distances from the N4 and H4 atoms, while, in
the case of SSS isomer, it seems clearly that N4 and H4 are
relatively protected from solvent molecules.
Figure 4. Calculated N-OEt O and H-OEt OH RDFs. The peak at short
2
distances is the H-OEt OH RDF, and the pe2ak at higher distances is the
2
N-OEt O RDF.
2
4.2. Solvent Accessible Surface. The solvent accessible
surface (SAS) of the RXP03 isomers in each solvent studied
has been systematically investigated following a well-established
computational procedure.34-37 By definition, the summarized
SAS of a RXP03 molecule is the outer surface that results if
one draws a sphere of van der Waals radius for each atom of
the solute. Thus, the corresponding SAS of each atom is a part
of its sphere surface and in principle it constitutes a very small
part of the whole molecular outer surface. Further, by using a
well-known computational technique, we have thoroughly
determined the hydrophilic as well as hydrophobic area of the
aforementioned outer surface for each isomer. To this point,
for a given atom on the solute molecule, we define its exposed
area as hydrophilic if the absolute value of its charge is higher
than 0.35|e|. This value for the charge ensures that aromatic
rings and aliphatic sites are characterized as hydrophobic. The
solvent accessible areas obtained for RRS and RSS in EtOH
and SRS and SSS in Et2O are summarized in Table 1.
From the data in Table 1, we may extract some information
concerning the structure of the isomers in the solvent environ-
ment that could give insight into the different solubilities of
the isomers in EtOH and Et2O, respectively. For instance, it is
found that RRS has a higher SAS than RSS in EtOH. The
difference in SAS between RRS and RSS isomers in EtOH
comes mainly from the increased hydrophobic part of SAS. At
the atomic level, the O5 atoms of RRS and RSS isomers have
an atomic SAS of 0.18 and 0.11 nm2, respectively. This is
another indication that the O5 atom of RRS isomer is more
exposed than that of RSS to the EtOH molecules. On the other
hand, SRS has a higher SAS than SSS isomer with higher both
hydrophobic and hydrophilic SAS. At the atomic level, the N4/
H4 atoms of SRS and SSS isomers, that exhibit different RDFs
with the O atom of Et2O, have an atomic SAS of 0.045/0.060
and 0.022/0.027, respectively. To conclude, the behavior of the
SAS obtained in the framework of this study was found to be
in agreement with the experimental solubility identified at the
last stage of synthesis.
our results show that this atom is probably protected from
solvent molecules by aromatic rings due to the isomer three-
dimensional configuration. Thus, from the above results, it is
seen that the P2 position of the RRS and RSS isomers (see Figure
1) in EtOH is affected in a different way from the solvent
molecules. Concretely, the P2 position on the RRS isomer seems
to be unfolded, allowing thus the oxygen atom O5 to directly
interact with EtOH solvent molecules, while in the case of the
RSS isomer this position is not directly exposed to the solvent.
SRS and SSS in Et2O. As in the case of RRS and RSS in
liquid EtOH, the fact that SRS is soluble in the aprotic and not
hydrogen bonding Et2O solvent is also unclear, while this is
not the case for SSS. As a first step toward elucidating this
problem, we investigated the site-site RDFs between O atoms
(O1-O6) and amino groups (N/H 1-4) and the center of mass
of the aromatic rings of SRS and SSS isomers and the O atom
and Me group on the Et2O molecules. By carefully inspecting
all of these functions, it is found that each kind of the
aforementioned RDFs calculated for both isomers exhibits
similar behavior except for the RDFs for the amino groups NHx
and O atom on the Et2O molecules. Most of these RDFs show
typical liquidlike behavior with small first peaks with amplitude
in the range 0.8-1.5 located at moderate correlation distances,
3.6-6 Å, without unusual features. On the other hand, the
calculated RDFs between the atoms of the amino groups N, H,
and O on Et2O presented in Figure 4 exhibit strong first peaks
located at short correlation distances and they are different in
amplitude for the two SRS and SSS isomers.
From the RDFs in Figure 4, it is easily seen that all of the
H-OEt O RDFs have their peak at a distance of 1.8-2 Å, while
2
the N-OEt O RDFs in the range 2.9-3 Å. This is an indication
2
that all of the amino groups of both isomers are surrounded by
Et2O molecules oriented with their O atom toward the H atoms.
From the same figure, the N3-O RDFs as well as H3-O are
almost the same for both isomers, suggesting that this part in
both isomers is similarly exposed to Et2O solvent molecules.
The RDFs of the rest of the N and H atoms are similar in shape,
but they are different in height for each isomer. The peaks of
N1/H1-O RDFs of SRS and SSS have relatively small
amplitudes with those of SSS being higher. The N2/H2-O
4.3. Hydrogen Bonding. 4.3.1. Solute-SolWent Hydrogen
Bonding. From the RDFs between the O and H atoms of EtOH
and the O atoms of RRS and RSS isomers, it seems that RRS
isomer forms hydrogen bonds with O1, O2, O4, and O5 atoms
with EtOH. The RSS isomer seems to have the O4 atom
protected from the O and H atoms in EtOH, while the rest of
the O atoms seem to form hydrogen bonds with EtOH. In order
to get detailed information concerning the hydrogen bonds (HBs)
formed by the RRS and RSS isomers with EtOH solvent
RDFs have lower peak heights than the other N/H-OEt O RDFs
2
for both isomers, suggesting that N2 and H2 are relatively
protected from the solvent molecules probably due to the
intramolecular structure. Note also that the peaks of these

426 J. Phys. Chem. B, Vol. 114, No. 1, 2010
Matziari et al.
TABLE 2: Simulated Mean Number of Hydrogen Bonds
per O Atom, 〈nHB〉, of RRS and RSS Isomer O Atoms with
EtOH
4.3.2. Intramolecular Hydrogen Bonding. The existence or
not of intramolecular HBs on the isomers was investigated by
the calculation of the time dependent distance between O and
amino H atoms separated by three or more bonds that could
form intramolecular HBs. From these functions, those that
suggest the existence of intramolecular HBs are the H1-O4,
H1-O2, and H2-O4 of SRS in Et2O. We have to note here
that for all four isomers in EtOH and Et2O, transient configura-
tions with intramolecular HBs are present, but the frequency
of appearance and the duration in time of such events is found
to be quite small. For instance, in the case of SRS in Et2O
environment, we found two intramolecular HBs, namely,
between H1 and O2 and between H2 and O4. From the
calculated time dependent distance, we have estimated the
percentage of time where the involved intramolecular HBs
between the O and H atoms are separated no more than 2.6 Å.
The selection of this distance comes from the geometric criterion
used in the calculation of HBs of RRS and RSS isomers in
EtOH.32,33,38,39 The percentage of time when the H and O atoms
are involved in HBs is presented in Table 3 for the cases where
it exceeds 5%. The interesting feature of these HBs is the part
of the isomers where the hydrogen bonded atoms is formed.
When these HBs are formed, they force a part of the backbone
to form a kind of ring. This affects the orientation of the large
hydrophobic side chains. This finding supports the experimental
solubility of SRS isomer in Et2O. When HBs are present, the
isomer is oriented with its hydrophobic side chains toward the
solvent, leading thus to increased solubility in the aprotic solvent
Et2O. Some characteristic configurations of RRS and RSS in
EtOH as well as SRS and SSS in Et2O are presented in
Figure 5.
〈nHB
〉
system
O1
O2
O4
O5
total
RRS-EtOH
RSS-EtOH
1.72
1.00
1.87
0.97
1.32
0.87
1.49
0.14
6.40
2.98
TABLE 3: Percentage of Simulation Time Where the O and
H Atoms of RXP03 Isomers Involved in Intramolecular
Hydrogen Bonds Are Separated by a Distance No More
Than 2.6 Åa
isomer
solvent
O-H pair
% of time
RRS
RSS
SRS
SRS
SRS
SSS
SSS
SSS
EtOH
EtOH
Et2O
Et2O
Et2O
Et2O
Et2O
Et2O
O4-H1
O4-H2
O2-H1
O4-H1
O4-H2
O2-H1
O4-H2
O2-H4
39.5
16.1
7.6
68.8
28.5
21.4
35.0
18.8
a Only the cases where this percentage is higher than 5% are
presented.
molecules, we calculated the mean number of hydrogen bonds
for the above-mentioned RXP03 O atoms. For the HB calcula-
tion, we employed the geometric definition38,39 according to
which an RXP03 O atom forms a HB with a neighboring EtOH
molecule if the O-OEtOH and O-HEtOH distances are less than
3.6 and 2.5 Å, respectively, and the angle O-O-HEtOH is less
than 30°. The mean number of HBs per O atom as well as the
total number of HBs per isomer molecule obtained is presented
in Table 2. From Table 2, it is clearly seen that the RRS isomer
forms more HBs than the RSS one. For the O1, O2, and O4
atoms, the mean number of HBs of RRS is by a factor of 2
greater than the corresponding one of RSS. The mean number
of HBs of the O5 atom is 1.49 for RRS while only 0.14 for the
RSS isomer. It means that the position of the RSS isomer is
protected from the EtOH molecules.
5. Conclusions
In the present study, the large quantities of RXP03 required
for in vivo studies, as well as the necessity for diastereoisomeric
purity, motivated us to develop an efficient synthetic method,
which allows separation of the four diastereoisomers of RXP03
based on the observed differences in solubility of the four
isomers in various solvents. Thus, in the framework of this
Figure 5. Characteristic structures of RRS/RSS isomers in EtOH and SRS/SSS in Et2O.

Phosphinic MMP Inhibitor RXP03
J. Phys. Chem. B, Vol. 114, No. 1, 2010 427
effort, a new synthetic methodology has been undressed, which
provided the four diastereoisomers of RXP03 in a large scale,
high yield, and high purity. Resolution of the Cbz-protected
aminophosphinic acid analogues of phenylalanine, Michael-type
addition to phenylpropyl acrylic acid ethyl ester, saponification,
and coupling with TrpNH2 provided the four isomers, by means
of recrystallization, based on the remarkable difference in
solubility of the four isomers of RXP03. This fact leads us to
examine theoretically the conformational differences of these
four isomers via computer simulations in the solvents used
experimentally.
Organic Chemistry, Special Account for Research Grants of
National Athens University (NKUA) and by the CEA (SIMO-
PRO).
Supporting Information Available: Figures showing cal-
culated radial distribution functions, ESP partial charges,
dihedral angle distributions, and time evolutions, tables showing
optimized geometries and fractional charges, an explanation of
the general considerations, a list of abbreviations, and detailed
experimental procedures. This material is available free of
charge via the Internet at http://pubs.acs.org.
The differences in solubility of the four RXP03 diastereoi-
somers in the used solvents were examined in terms of intra-
and intermolecular structure in order to provide information
regarding the influence of a possible conformation of the solutes
on their characteristic solvation process.
References and Notes
(1) Dive, V.; Georgiadis, D.; Matziari, M.; Makaritis, A.; Beau, F.;
Cuniasse, P.; Yiotakis, A. Cell. Mol. Life Sci. 2004, 61, 2010–2019.
(2) Cuniasse, P.; Devel, L.; Makaritis, A.; Beau, F.; Georgiadis, D.;
Matziari, M.; Yiotakis, A.; Dive, V. Biochimie 2005, 87, 393–402.
(3) Andarawewa, D. V., K. L.; Boulay, A.; Matziari, M.; Beau, F.;
Guerin, E.; Rousseau, B.; Yiotakis, A.; Rio, M. C. Int. J. Cancer 2001,
113, 775–781.
(4) Schechter, I.; Berger, A. Biochem. Biophys. Res. Commun. 1967,
27, 157–162.
(5) Georgiadis, D.; Beau, F.; Czarny, B.; Cotton, J.; Yiotakis, A.; Dive,
V. Circ. Res. 2003, 93, 148–154.
(6) Vassiliou, S.; Mucha, A.; Cuniasse, P.; Georgiadis, D.; Lucet-
Levannier, K.; Beau, F.; Kannan, R.; Murphy, G.; Knauper, V.; Rio, M. C.;
Basset, P.; Yiotakis, A.; Dive, V. J. Med. Chem. 1999, 42, 2610–2620.
(7) Schiodt, C. B.; Buchardt, J.; Terp, G. E.; Christensen, U.; Brink,
M.; Berger Larsen, Y.; Meldal, M.; Foged, N. T. Curr. Med. Chem. 2001,
8, 967–976.
(8) Georgiadis, D.; Vazeux, G.; Llorens-Cortes, C.; Yiotakis, A.; Dive,
V. Biochemistry 2000, 39, 1152–1155.
(9) Grembecka, J.; Mucha, A.; Cierpicki, T.; Kafarski, P. J. Med. Chem.
2003, 46, 2641–2655.
(10) Makaritis, A.; Georgiadis, D.; Dive, V.; Yiotakis, A. Chem.sEur.
Using QM calculations, the geometry of each isolated isomer
molecule was first optimized and then the corresponding charge
distribution was obtained. Further, the different solubility of the
examined diastereoisomers in EtOH and Et2O was investigated
by means of the molecular dynamics simulation of each
molecular system. The intermolecular structure between the
RXP03 diastereoisomers and the solvent molecules obtained
supports the experimentally found different solubilities among
the isomers. All of the calculated O-OEtOH and O-HEtOH RDFs
of O atoms of the RRS isomer exhibit high first peaks at very
short distances with the characteristic shape showing hydrogen
bonded behavior. The corresponding RDFs for the RSS isomer
are similar to each other except for the P2 position of O atom,
which shows that the EtOH molecules are almost absent at short
distances at this part of the isomer. The RRS isomer was found
to form about six HBs with the surrounding EtOH molecules,
distributed in all molecular positions that make this isomer
soluble. On the other hand, the RSS isomer forms only about
three HBs with EtOH but the P2 positions of the polar atoms
that potentially could form HBs with the solvent species are
not accessible to them due to the fact that they are protected by
the hydrophobic aromatic rings. In addition, the RRS isomer
exhibits higher solvent accessible surface in EtOH compared
with that for the RSS isomer. On the other hand, the SRS isomer
in Et2O has been found to form intramolecular HBs, while this
does not seem to be the case for the SSS isomer. More clearly,
the SRS isomer has its polar atoms protected from the nonpolar
solvent, leaving their hydrophilic aromatic and aliphatic exposed
to the nonpolar solvent. On the other hand, the SSS isomer does
not provide such a kind of configuration because in this molecule
its polar parts are protected from the solvent. As in the case of
RRS/RSS in EtOH, the soluble in Et2O SRS isomer has a higher
solvent accessible area. These findings might explain the fact
that the SRS isomer is soluble in Et2O while the SSS is not.
J. 2003, 9, 2079–2094.
(11) Buchardt, J.; Schiodt, C. B.; Krog-Jensen, C.; Delaisse, J. M.; Foged,
N. T.; Meldal, M. J. Comb. Chem. 2000, 96, 624–638.
(12) Dive, V.; Cotton, J.; Yiotakis, A.; Michaud, A.; Vassiliou, S.;
Jiracek, J.; Vazeaux, G.; Chauvet, M. T.; Cuniasse, P.; Corvol, P. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 4330–4335.
(13) Jiracek, J.; Yiotakis, A.; Vincent, B.; Checler, F.; Dive, V. J. Biol.
Chem. 1996, 271, 19606–19611.
(14) Cursio, R.; Mari, B.; Louis, K.; Saint-Paul, M. C.; Rostagno, P.;
Giudicelli, J.; Bottero, V.; Anglard, P.; Yiotakis, A.; Dive, V.; Gugenheim,
J.; Auberger, P. FASEB J. 2002, 16, 93–95.
(15) Andarawewa, D. V., K. L.; Boulay, A.; Matziari, M.; Beau, F.;
Guerin, E.; Rousseau, B.; Yiotakis, A.; Rio, M. C. Int. J. Cancer 2001,
113, 775–781.
(16) Gall, A.; Ruff, M.; Kannan, R.; Cuniasse, P.; Yiotakis, A.; Dive,
V.; Rio, M.; Basset, P.; Moras, D. J. Mol. Biol. 2001, 307, 577–586.
(17) Baylis, E. K.; Campbell, C. D.; Dingwall, J. D. J. Chem. Soc.,
Perkin Trans. 1984, 1, 2845–2853.
(18) Seco, J. M.; Quinoa, Riguera, E. Chem. ReV. 2004, 104, 17–117.
(19) Breneman, C. N.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361.
(20) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.;
Gordon, M. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Su, K. S. J.;
Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993,
14, 1347.
In general, we may conclude that the different solubility of
the RRS and RSS diastereoisomers in EtOH is a result of the
different number of HBs formed by each isomer with EtOH
molecules. In the case of SRS and SSS in Et2O, their different
solubility might be attributed to the different intramolecular HBs
formed on these diastereoisomers.
(21) van Gunsteren, W. F.; Billeter, S. R.; Eising, A. A.; Hunenberger,
P.; Kruger, M. A. E.; Scott, W. R. P.; Tironi, I. G. Biomolecular Simulation:
The GROMOS96 manual and user guide; Hochschulverlag AG an der ETH
Zurich: Zurich, Switzerland,1996.
(22) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089.
(23) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.;
Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577.
(24) Byrd, R. H.; Lu, P.; Nocedal, J. SIAM J. Sci. Stat. Comput. 1995,
16, 1190.
(25) Berendsen, H. J. C.; Postma, J. P. M.; DiNola, A.; Haak, J. R. J.
Chem. Phys. 1984, 81, 3684.
(26) Nose`, S. Mol. Phys. 1984, 52, 255.
Finally, in this study, the diastereoselective synthesis of the
potent phosphinic inhibitor RXP03 was achieved and confor-
mational as well as solvation studies of the RXP03 diastereoi-
somers were successfully performed.
(27) Hoover, W. G. Phys. ReV. A 1985, 31, 1695.
(28) Hockney, R. W.; Goel, S. P.; Eastwood, J. J. Comput. Phys. 1974,
14, 148.
(29) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys.
1977, 23, 327.
Acknowledgment. This work was supported by the European
Commission (FP5RDT, QLK3-CT02-02136 and FP6RDT,
LSHC-CT-2003-503297), by funds from the Laboratory of

428 J. Phys. Chem. B, Vol. 114, No. 1, 2010
Matziari et al.
(30) Masella, M.; Borgis, D.; Cuniasse, P. J. Comput. Chem. 2008, 29,
1707.
(35) Eisenhaber, F.; Lijnzaad, P.; Argos, P.; Sander, C.; Scharf, M.
J. Comput. Chem. 1995, 16, 273.
(31) van der Spoel, D.; Lindahl, E.; Hess, B.; van Buuren, A. R.; Apol,
E.; Meulenhoff, P. J.; Tieleman, D. P.; Sijbers, A. L. T. M.; Feenstra, K. A.;
van Drunen, R.; Berendsen, H. J. C. Gromacs User Manual Version 3.3,
2005; http://www.gromacs.org/.
(32) Saiz, L.; Padro´, J. A.; Guardia, E. Mol. Phys. 1999, 97, 897.
(33) Dellis, D.; Chalaris, M.; Samios, J. J. Phys. Chem. B 2005, 109,
18575.
(36) Connolly, M. L. J. Appl. Crystallogr. 1983, 16, 548.
(37) Richmond, T. J. J. Mol. Biol. 1984, 178, 63.
(38) Jo¨rgensen, W. L. J. Phys. Chem. 1986, 90, 1276.
(39) Ferrario, M.; Haughney, M.; McDonald, I. R.; Klein, M. L. J. Phys.
Chem. 1990, 93, 5156.
(34) Shrake, A.; Rupley, J. A. J. Mol. Biol. 1973, 79, 351.
JP903830V