VISTOLI ET AL.
contribution of a thioether function is globally underestimated
diastereoisomeric peptides that suitably cover the accessible
chemical space, (2) rationalizing the experimentally deter-
mined physicochemical differences between diastereoisomers,
and (3) developing correlative models exploiting the property
space parameters to conveniently predict the lipophilicity of
dipeptides also accounting for their configurational factors.
The results presented herein confirm that the accuracy of
conformer-dependent predictive methods can be enhanced
by exploiting property space parameters that can take both
property variability and diastereoisomeric differences into ac-
count. One doubts that versatile equations could be derived
for heterogeneous molecules. In contrast, it now appears
possible to develop targeted relationships for homogeneous
classes of compounds accurate enough to satisfactorily deal
with the challenge of diastereoisomeric peptides. Finally,
the obtained results suggest that property space parameters
can also support the prediction of ionization constants, al-
ways within homogeneous datasets, thus underlining a
similar relevance of the property space concept in the whole
physicochemical profiling.
by a value of 3.02, a result that further confirms the need of a
better reparameterization of sulfur-containing functional
groups.
To further assess the predictive power of eq. 3 and despite
the low number of included data, the examined dipeptides
were randomly subdivided into a training set (n = 12) and an
external test set (n = 6). Equation 4 was developed consider-
ing only the training set and appears very similar to the
previous one, thus confirming a substantial stability of the
predictive model that is almost independent on the included
data. The relation between experimental and predicted log
D7.4 values (eq. 5) for the external test set affords an encour-
aging validation for the predictive power of eq. 4. Equation 5
is indeed a satisfactory one given its statistics and its slope
being close to 45ꢀ, whereas the analysis of all residuals
(training and test sets) shows that there is no correlation with
the log D7.4 values suggesting that eq. 4 is equally predictive
for polar and apolar dipeptides (plot not shown).
In general, similar correlative studies involving experimental
pK values afforded worst relationships. Nevertheless, eq. 6
unveils the possibility to predict pK1 values with satisfactory
statistics. In detail, eq. 6 includes the number of rotors, PSA
means, and PSA ranges, thus emphasizing that the N-terminus
ionization depends on molecular flexibility (as encoded by
rotors) and intermolecular interactions (as encoded by PSA
parameters). Noticeably, both PSA parameters appear with
negative coefficients indicating that the exposure of amino
group destabilizes its protonated state.
LITERATURE CITED
1. Pliska V, Testa B, van de Waterbeemd H, In Lipophilicity in drug action
and toxicology, VCH Weinheim, New York, 1996.
2. Testa B, Vistoli G, Pedretti A A fresh look at molecular structure and
properties. In (Manhold R, Ed.) Molecular drug properties. Wiley-VCH,
Weinheim: Germany, 2007, p 3–23.
3. Liu X, Testa B, Fahr A. Lipophilicity and its relationship with passive drug
permeation. Pharm Res 2011;28:962–77.
4. Henchoz Y, Bard B, Guillarme D, Carrupt PA, Veuthey JL, Martel S.
Analytical tools for the physicochemical profiling of drug candidates to
predict absorption/distribution. Anal Bioanal Chem 2009;394:707–29.
CONCLUSIONS
5. Meanwell NA. Improving drug candidates by design:
a focus on
physicochemical properties as a means of improving compound disposi-
tion and safety. Chem Res Toxicol 2011;24:1420–56.
6. Wang J, Skolnik S. Recent advances in physicochemical and ADMET
profiling in drug discovery. Chem Biodivers 2009;6:1887–99.
7. Ritchie TJ, Ertl P, Lewis R. The graphical representation of ADME-related
molecule properties for medicinal chemists. Drug Discov Today
2011;16:65–72.
8. Kerns EH, Di L, Carter GT. In vitro solubility assays in drug discovery.
Curr Drug Metab 2008;9:879–85.
9. Pagliara A,Caron G,Lisa G,Fan W,Gaillard P,Carrupt PA,Testa B,
Abraham MH. Solvatochromic analysis of di-n-butyl ether/water partition
coefficients as compared to other solvent systems. J Chem Soc, Perkin
Trans 1997;2:639–43.
Although peptides find ever increasing therapeutic
applications, their physicochemical profiling has not received
the attention it deserves. Similarly, the computational
approaches to predict the main physicochemical properties
have been mainly focused on small organic molecules,
whereas the studies of peptides were performed sporadically
and failed to yield encouraging results. Lipophilicity is not an
exception, and indeed, the best fragmental approaches have
intrinsic difficulties to successfully predict the log P of
peptides.39
Besides the scarcity of computational studies purposely
focused on peptides, such a failure can find two main justifica-
tions. First, peptides are quite complex molecules in which
flexibility (of both backbone and side chains) and the simulta-
neous presence of more interacting functional groups concur
to hamper a successful log P prediction. Second, the few
available data concern peptides that differ largely in length,
capping, and polarity, a heterogeneity that further compli-
cates the development of accurate predictive methods. As
stated by Thompson and coworkers,39 such problems would
be resolved by producing a proper dataset composed by
homogenous peptides that exhaustively cover the available
chemical space. Moreover, although several computational
approaches and many chirality descriptors have been pro-
posed in the last years to predict enantioselective binding,
asymmetric reactivity, or chiral separations,40,41 the physico-
chemical profiling of diastereoisomers has been scantly
investigated and very few studies have tried to predict their
differences.
10. Carrupt PA, Testa B, Gaillard P Computational approaches to lipophilicity:
methods and applications. Rev. Comput. Chem. 1997;11:241–315.
11. Mannhold R, Poda GI, Ostermann C, Tetko IV. Calculation of molecular
lipophilicity: State-of-the-art and comparison of log P methods on more
than 96,000 compounds. J Pharm Sci 2009;98:861–93.
12. Leo A. Calculating log Poct from structures. Chem Rev 1993;93:1281–1306
13. Efremov RG, Chugunov AO, Pyrkov TV, Priestle JP, Arseniev AS, Jacoby
E Molecular lipophilicity in protein modeling and drug design. Curr Med
Chem 2007;14:393–415.
14. Kraszni M, Bányai I, Noszál B Determination of conformer-specific
partition coefficients in octanol/water systems.
2003;46:2241–2245.
J
Med Chem
15. Cruciani G, Mannhold R, Kubinyi H, Folkers G. Molecular interaction
fields: applications in drug discovery and ADME prediction. Weinheim:
Wiley, 2005.
16. Vistoli G, Pedretti A, Testa B. Chemodiversity and molecular plasticity:
recognition processes as explored by property spaces. Future Med Chem.
2011;3:995–1010.
17. Vistoli G, Pedretti A, Testa B Assessing drug-likeness – what are we
missing?. Drug Discov Today 2008;13:285–293.
Accordingly, this study was undertaken with a view to fill
these noteworthy gaps by (1) preparing a homogeneous set of
Chirality DOI 10.1002/chir
18. Goldreich O. In Computational complexity: a conceptual perspective.
Cambridge: Cambridge University Press, 2008.