Reversed-phase TLC and HPLC retention data in correlation studies with in silico molecular descriptors and druglikeness properties of newly synthesized anticonvulsant succinimide derivatives

Article abstract of DOI:10.1021/mp100373d

The properties relevant to pharmacokinetics of two series of newly synthesized succinimide derivatives have been studied. The properties under consideration have been either determined empirically, by reversed-phase liquid chromatography (TLC and HPLC technique), or calculated with the use of established theoretical medicinal chemistry/drug design software. Chromatographic techniques allowed determination of the retention constants R_M⁰ and log k_w, which characterize lipophilicity of compounds. Considering potential pharmaceutical importance of succinimide derivatives, we (i) examined the retention behavior in the reversed-phase liquid chromatographic (RP LC) systems, in both planar and column LC, and (ii) determined the relationships between chromatographic data and selected structural features of analytes that are believed to markedly affect their processes of absorption, distribution, metabolism, excretion and toxicity (ADMETox). Significant relationships were found between the retention constants, R_M⁰ and log k_w, and the in silico calculated bioactivity descriptors, in particular HIA (human intestinal absorption) and PPB (plasma protein binding) parameters. The R_M⁰ and log k_w values of the investigated compounds have been recommended for description of their lipophilicity and evaluating pharmacokinetic properties. In view of results of this study the newly synthesized succinimide agents meet pharmacokinetic criteria of preselection of drug candidates and hence qualify for pharmacodynamic phase of antiepileptic drug development. Best compromising human intestinal absorption and plasma protein binding features appear to be compounds A4, A5, A10 and A11.

Full text of DOI:10.1021/mp100373d

ARTICLE
pubs.acs.org/molecularpharmaceutics
Reversed-Phase TLC and HPLC Retention Data in Correlation Studies
with in Silico Molecular Descriptors and Druglikeness Properties of
Newly Synthesized Anticonvulsant Succinimide Derivatives
Nada Perisic-Janjic,† Roman Kaliszan,*,‡,§ Pawez Wiczling,‡ Natasa Milosevic,|| Gordana Uscumlic,^ and
Nebojsa Banjac^
†Department of Chemistry, Faculty of Sciences, University of Novi Sad, Trg Dositieja Obradovica 3, 21000 Novi Sad, Serbia
‡Department of Biopharmaceutics and Pharmacodynamics, Medical University of Gdaꢀnsk, Gen. J. Hellera 107, 80-416 Gdaꢀnsk, Poland
§Department of Biopharmacy, Collegium Medicum, Nicolaus Copernicus University, Skzodowskiej-Curie 9, 85-094 Bydgoszcz, Poland
Department of Pharmacy, Faculty of Medicine, University of Novi Sad, Hajduk Veljkova 1-3, 21000 Novi Sad, Serbia
^Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, 11000 Belgrade, Serbia
ABSTRACT: The properties relevant to pharmacokinetics of two series
of newly synthesized succinimide derivatives have been studied. The
properties under consideration have been either determined empirically,
by reversed-phase liquid chromatography (TLC and HPLC technique),
or calculated with the use of established theoretical medicinal chemistry/
drug design software. Chromatographic techniques allowed determina-
0
tion of the retention constants RM and log kw, which characterize
lipophilicity of compounds. Considering potential pharmaceutical im-
portance of succinimide derivatives, we (i) examined the retention behavior in the reversed-phase liquid chromatographic (RP LC)
systems, in both planar and column LC, and (ii) determined the relationships between chromatographic data and selected structural
features of analytes that are believed to markedly affect their processes of absorption, distribution, metabolism, excretion and toxicity
(ADMETox). Significant relationships were found between the retention constants, RM0 and log kw, and the in silico calculated
bioactivity descriptors, in particular HIA (human intestinal absorption) and PPB (plasma protein binding) parameters. The RM0 and
log kw values of the investigated compounds have been recommended for description of their lipophilicity and evaluating
pharmacokinetic properties. In view of results of this study the newly synthesized succinimide agents meet pharmacokinetic criteria
of preselection of drug candidates and hence qualify for pharmacodynamic phase of antiepileptic drug development. Best
compromising human intestinal absorption and plasma protein binding features appear to be compounds A4, A5, A10 and A11.
KEYWORDS: ADMETox prediction, anticonvulsant agents, drug design, HPLC, lipophilicity, liquid chromatography, molecular
descriptors, pharmacokinetics evaluations, SAR, structure-activity relationships, retention parameters, succinimides, TLC
’ INTRODUCTION
that two hydrophobic regions and two electron-rich centers,
interconnected with similar relative orientations, represent a
major molecular requirement for antiepileptic activity. Hence,
having the pharmacophore system defined can allow one to
concentrate on physicochemical properties of the agents which
would provide their proper pharmacokinetics, which is necessary
for In Vivo activity. Knowledge of physicochemical properties
of drug candidates at an early phase of drug development is
crucial to reduce attrition rates due to poor biopharmaceutical
properties.
Lipophilicity is one of the physicochemical properties which
influence a drug’s permeation ability (amount and rate of
intestinal absorption and/or blood-brain barrier crossing), as
well as its distribution properties (e.g., degree of plasma protein
binding).5-8 Lipophilicity is usually quantified as the logarithm
Succinimide derivatives exhibit a pronounced anticonvulsant
activity, combined with some antimuscarinic effects.1 The succi-
nimide moiety provides a pharmacophore for potential antiepi-
leptic agents. Thus, the requested molecular pharmacology of
potential new drugs from the group of succinimide derivatives
seems to be structurally guaranteed. However, a safe and effective
therapeutic agent must also possess proper pharmacokinetic
features to attain (and maintain for some time) its effective
concentrations at the site of action after administration to a living
organism. In other words, a drug candidate must exhibit phys-
icochemical properties providing its proper absorption, distribu-
tion, metabolism, excretion and toxicity (ADMETox).
Nitrogen-containing five membered rings have several applications
in medicine, biology and chemistry. Succinimides and hydantoins
show antimuscarinic and anticonvulsant activity, and the 3,3-diphenyl
derivatives of those compounds are potent anticonvulsants.1-3
Presently, development of new drugs involves research on
structure-activity relationships (SAR). It has been suggested4
Received: November 4, 2010
Accepted: January 18, 2011
Revised:
December 22, 2010
Published: January 18, 2011
r
2011 American Chemical Society
555
dx.doi.org/10.1021/mp100373d Mol. Pharmaceutics 2011, 8, 555–563
|

Molecular Pharmaceutics
ARTICLE
Table 1. Structure of Compounds Investigated
chromatographic systems, was to establish the relationships
between the chromatographic data and the selected structural
properties of succinimides that are related to the ADMETox
parameters, obtained by the established computational medicinal
chemistry methods. To that aim chromatographic lipophilicity
parameters for the newly synthesized compounds have been
determined experimentally by two independent methods. On the
other hand, such druglikeness properties of the agents, like
human intestinal absorption (HIA) and plasma protein binding
(PPB), were evaluated theoretically.
Importance of a proper HIA value for the quality of the orally
administered drugs, like succinimide derivatives, is obvious. A
good absorption is the main prerequisite for reproducible plasma
levels after individual equimolar drug doses.
As regards binding of drugs to plasma proteins, a high PPB
(above 95%) brings a risk of a highly varying free (unbound)
fraction of a drug in plasma. Small changes in that free fraction,
which is active pharmacokinetically, can cause dramatic changes
in bioactivity. For example, if 2 molecules of 98 previously bound
to plasma protein are released (e.g., by a competing xenobiotic),
then the bound to protein fraction of a drug changes minimally
(from 98 to 96%), but the active unbound fraction increases by
100% (from 2 to 4 molecules per 100). In the case of a little less
strongly bound drug, say 94%, releasing 2% molecules increases
free fraction from 6 to 8%, i.e., for 25% only. Therefore, drugs not
that strongly binding to plasma proteins (<95%) allow for an
easier control of dosing.
compd
R1
R2
compd
R1
-H
R2
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
-H
-H
-H
-H
-H
-H
-H
-H
-H
-CH3
-Cl
-Br
B1
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-CN
-OH
-Cl
-CH3
-OCH3
-Cl
B2
-CH3
-OCH3
-OH
-NO2
-CN
-F
B3
B4
-Br
B5
-OH
-NO2
-OCOCH3
-H
B6
B7
B8
-Cl
B9
-Br
-H
B10
B11
B12
B13
B14
-I
-H
-COCH3
-H
-H
-H
of the water-octanol partition coefficient for partitioning of
the agent between the two immiscible solvent phases. The
1-octanol-water system is a widely accepted reference system
for the determination of lipophilicity expressed as log P.4
Lipophilicity plays an important role in the transport of com-
pounds through a biological system, and it may also influence the
interaction between a drug compound and a pharmacological
receptor. The key aspect of the biological activity of a majority of
compounds is their ability to get through biological membranes
by passive diffusion. This process involves a series of partitioning
steps, in combination with diffusion through several regions, i.e.,
partitioning between aqueous intra- and extracellular media and
phospholipid membranes.
The drugs, in particular those requiring a long-lasting therapy,
like anticonvulsives, should be devoid of or have only negligible
adverse effects. For that reason, drug candidates are expected not
to bind markedly to irrelevant pharmacological receptors or
enzymes. For the anticonvulsive succinimides studied, an affinity
to GPCR, kinases, ion channels and nuclear receptors was
modeled according to the established procedures based on the
agents’ molecular structure.
Based on the structure-activity relationships (SAR) derived, a
selection of best drug candidates for bioactivity studies has been
rationally guided, compromising lipophilicity affecting HIA, PPB
and adverse effects.
Processes of drug absorption, distribution and excretion in the
pharmacokinetic phase of drug action, as well as drug-receptor
interactions in the pharmacodynamic phase, are dynamic in
nature, as are the chromatographic separation processes. There-
fore, chromatographic data are often used to model pharmaco-
kinetics of drugs and other xenobiotics.6 Owing to its simplicity,
as well as efficiency, reversed-phase thin-layer chromatography
(TLC) appears especially attractive for lipophilicity determi-
nation.9,10 However, the newly proposed gradient HPLC
approach11,12 allows fast and convenient determinations of log
kw parameter (along with pKa values), which is considered an
even more reliable chromatographic lipophilicity parameter.
Also, the slope, S, of the linear relationship between log kw and
volume percent of organic modifier in the aqueous eluent can
readily be calculated. All that can be obtained from two gradient
HPLC runs instead of 5-6 chromatographic runs required in
classical isocratic mode.
’ RESULTS AND DISCUSSION
Retention Behavior of the Compounds Investigated. Two
groups of newly synthesized succinimide derivatives (Table 1)
were subjected to initial chemical screening of their retention
behavior and evaluation of their lipophilicity and pharmaco-
kinetic properties.
The reversed-phase HPTLC and HPLC was carried out on
octadecyl-modified silica stationary phases. Aqueous mobile
phases, with an organic modifier (methanol, acetone and diox-
ane), were used. In TLC, the Rf values employed were averages of
at least three measurements. For subsequent analyses the mean
RM values were used, as calculated from the equation14,15
RM ¼ logð1=Rf - 1Þ
ð1Þ
In HPLC the corresponding parameter was log k, calculated as a
mean of 4-5 determinations from
Correlations of the properly determined log kw data with the
reference lipophilicity parameter log P have been firmly
established.6,9,13
k ¼ ðtr - t0Þ=t0
ð2Þ
The objective of this study, apart from examination of
the retention behavior in TLC and HPLC reversed-phase
where tr is retention time and t0 is the so-called dead time
determined by injection of a nonretained marker.
556
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
Table 2. Data for Linear Correlation, eqs 3 and 4, between RM and log k Values and the Volume Fraction of Organic Modifier, u, in
the Mobile Phase along with the Slopes S and S0, the Correlation Coefficients, r, and the Standard Errors of Estimation, SD
water-methanol
water-acetone
water-dioxane
water-methanol
0
0
0
compd
RM
S
r
SD
RM
S
r
SD
RM
S
r
SD
log kw
S0
A1
1.718
2.273
1.647
2.476
2.538
1.239
2.162
1.411
2.096
2.562
2.643
2.588
3.422
2.972
2.595
3.435
3.009
2.941
3.516
4.076
3.496
2.848
2.931
2.605
4.248
-2.757
-3.420
-2.646
-3.537
-3.582
-2.521
-3.105
-2.393
-3.159
-3.581
-3.607
-3.465
-4.369
-3.867
-4.274
-4.217
-3.906
-3.780
-4.321
-5.086
-4.143
-3.817
-3.841
-3.743
-5.363
0.993
0.994
0.994
0.993
0.995
0.986
0.995
0.996
0.991
0.990
1.000
0.992
0.991
0.993
0.993
0.996
0.992
0.991
0.994
0.999
0.992
0.992
0.990
0.995
0.996
0.031
0.033
0.026
0.037
0.032
0.038
0.029
0.022
0.040
0.047
0.008
0.045
0.061
0.048
0.052
0.040
0.053
0.050
0.049
0.018
0.056
0.051
0.058
0.040
0.044
1.615
1.889
1.621
2.143
2.317
1.179
2.184
1.508
1.833
2.251
2.318
2.546
3.253
3.056
2.194
3.374
2.915
2.932
3.326
3.368
3.504
2.665
2.748
2.292
3.024
-2.828
-3.144
-2.887
-3.412
-3.651
-2.482
-3.547
-2.764
-3.100
-3.571
-3.634
-3.938
-4.927
-4.730
-4.386
-5.027
-4.461
-4.419
-4.865
-4.912
-5.070
-4.141
-4.174
-3.771
-4.423
0.993
0.998
0.999
0.998
0.995
0.995
0.994
0.998
0.998
0.996
0.995
0.995
0.992
0.991
0.996
0.997
0.997
0.997
0.999
0.995
0.994
0.994
0.992
0.994
0.995
0.034
0.020
0.015
0.021
0.038
0.026
0.042
0.020
0.021
0.033
0.038
0.037
0.058
0.058
0.034
0.035
0.030
0.030
0.003
0.045
0.050
0.040
0.049
0.039
0.041
1.273
1.613
1.363
2.065
2.256
1.101
1.968
1.381
1.695
2.128
2.212
2.443
2.729
2.603
1.875
2.994
2.836
2.786
2.871
3.025
3.110
2.368
2.544
1.882
2.820
-2.391
-2.834
-2.604
-3.496
-3.768
-2.621
-3.415
-2.746
-3.012
-3.590
-3.700
-3.923
-4.169
-4.105
-4.296
-4.570
-4.482
-4.273
-4.256
-4.480
-4.570
-3.855
-4.022
-3.289
-4.169
0.999
0.986
0.992
0.994
0.988
0.988
0.978
0.993
0.978
0.999
0.992
0.999
0.993
0.999
0.999
0.998
0.995
0.997
0.992
0.991
0.991
0.993
0.995
0.991
0.995
0.012
0.044
0.030
0.036
0.053
0.038
0.066
0.031
0.059
0.016
0.044
0.016
0.044
0.018
0.017
0.026
0.041
0.030
0.049
0.055
0.055
0.041
0.038
0.042
0.039
2.73
3.19
2.92
3.33
3.42
2.39
3.65
2.86
3.06
3.29
3.55
nda
-4.29
-4.67
-4.57
-4.70
-4.74
-4.42
-4.94
-4.70
-4.53
-4.69
-5.02
nd
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
B1
B2
4.35
4.14
4.15
4.35
4.00
4.22
4.54
4.61
4.73
4.04
4.02
3.74
4.50
-5.34
-5.30
-5.49
-5.47
-5.21
-5.38
-5.52
-5.54
-5.62
-5.36
-5.34
-5.31
-5.53
B3
B4
B5
B6
B7
B8
B9
B10
B11
B12
B13
B14
a Not determined.
The relationships between the retention values (RM or log kw)
of the compound and the volume fraction (j) of organic
modifier in the mobile phase were, as expected, linear and
expressed by the well-known equations
eluent. These differences become more explicit when a more
polar modifier is used. Hence, it can be assumed that reversed
phase liquid chromatographic retention constants, RM0 and log
kw, of the compounds investigated reflect their lipophilicity, the
more so that increments to lipophilicity due to the phenyl
substituent are by no means identical.
0
RM ¼ RM þ Sj
ð3Þ
0
Correlations between Retention Constants, RM and log
kw, and log P Values. Quantitative structure-retention rela-
tionships (QSRR) are among the most extensively studied
procedures by which molecular chemical structure is quantita-
tively correlated with a well-defined physicochemical property of
analytes, such as chromatographic retention. Chromatographic
retention depends on the net effect of intermolecular interactions
between the analyte, the stationary phase and the mobile phase.
Linear relationships between the retention factor, RM0 or log kw,
and the standard lipophilicity parameter, log P, can be expected
because retention of compounds in reversed-phase liquid chro-
matography is governed by hydrophobic interactions.
log k ¼ log kw þ S0j
ð4Þ
0
where RM and log kw (intercepts) are the extrapolated values
corresponding to j = 0%, and S and S0 are the slopes of the linear
plot. Equations 1-4 served for deriving data for the further
QSAR studies.
Equations 3 and 4 were applied separately for each modifier
and each compound. The coefficients of the linear relationships
between retention and the volume fraction of organic modifier in
the mobile phase (eqs 3 and 4) are listed in Table 2.
0
The calculated RM and log kw values (intercepts) were
different for individual compounds due to different substituents.
Comparison of the corresponding parameters for compounds of
group A with those of group B, which have the same substituent
attached to N-phenyl, had shown that the RM0 and log kw data
were generally higher for compounds of group B than for
compounds of group A. It is reasonable as the additional phenyl
group increases the ability of the compounds of the group B to
participate in nonpolar attractive interactions with the stationary
phase to a higher extent than with the polar molecules of the
In this study RM0 and log kw data, determined on octadecylsi-
lica stationary phases were correlated against log P values,
calculated by the use of several softwares16 (Table 3). The best
relationships obtained are between ACD/log P and the retention
constants RM0 and log kw (Figure 1). Using different values of log
P from computational chemistry other significant correlations
were also obtained (Table 4). The present results indicate that
the theoretically calculated partition coefficients can be used
to predict retention constants of succinimide derivatives on
557
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
C18-silica stationary phases. That is as expected because the
empirical chromatographic lipophilicity parameters are known to
agree with the log P values.6,9,13 However, such a correlation
needs an empirical verification for individual sets of congeners.
Having the correlation proved for the succinimides studies,
instead of measuring the log P values by a tedious slow
Table 3. Different log P Computational Values Calculated for
Investigated Compounds
compd ACD/log P log PKOWWIN milog P log P X2 log P X3
A1
1.71
2.17
1.62
2.79
2.48
0.97
1.44
1.05
2.17
2.64
2.48
3.34
3.8
1.79
2.34
1.87
2.44
2.68
1.31
2.01
1.39
2.34
2.44
2.68
3.46
4.01
3.54
2.98
3.68
3.01
3.66
4.11
4.35
4.63
3.14
3.01
2.98
4.11
2.418
2.867
2.475
3.096
3.228
1.939
2.377
1.970
2.843
3.072
3.204
3.61
2.23
2.66
2.14
2.85
3.03
1.82
2.12
2.01
2.66
2.85
3.03
4.39
4.83
4.31
3.98
4.28
4.11
4.55
5.01
5.19
5.46
4.23
4.11
3.98
5.01
1.92
2.76
2.37
3.03
3.09
2.04
2.23
2.15
2.76
3.03
3.09
3.88
4.25
3.85
3.53
3.71
3.6
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
B1
B2
4.06
B3
3.26
2.61
3.07
2.78
3.4
3.67
B4
3.13
B5
3.57
B6
3.37
B7
3.77
3.98
4.51
4.57
4.53
3.57
3.6
B8
4.42
4.12
4.38
2.79
2.78
2.61
4.27
4.29
B9
4.42
B10
B11
B12
B13
B14
4.69
3.51
3.34
Figure 1. Relationships between retention constants, RM0 (a) and log
kw (b) and ACD/log P.
3.11
3.53
4.51
4.26
Table 4. Relationships between Retention Constants RM0 and log kw and log P Values, Determined by Using Different
Computational Programs
RM0 and log kw vs log P
modifier
equation
r
SD
RM0-ACD/log P
methanol
acetone
RM0 = 0.697 þ 0.723 ACD/log P
RM0 = 0.791 þ 0.612 ACD/log P
RM0 = 0.725 þ 0.547 ACD/log P
RM0 = 0.400 þ 0.777 log PKOWWIN
RM0 = 0.463 þ 0.652 log PKOWWIN
RM0 = 0.453 þ 0.603 log PKOWWIN
RM0 = -0.520 þ 0.978 milog P
RM0 = -0.289 þ 0.842 milog P
RM0 = -0.246 þ 0.755 milog P
RM0 = 0.475 þ 0.612 log P X2
RM0 = 0.518 þ 0.540 log P X2
RM0 = 0.501 þ 0.478 log P X2
RM0 = -0.205 þ 0.863 log P X3
RM0 = -0.008 þ 0.740 log PX3
RM0 = 0.003 þ 0.259 log P X3
log kw = 2.10 þ 0.599 ACD/log P
log kw = 1.75 þ 0.677 log PKOWWIN
log kw = 1.06 þ 0.818 milog P
log kw = 1.75 þ 0.552 log P X2
log kw = 1.23 þ 0.750 log P X3
0.932
0.907
0.889
0.940
0.952
0.922
0.940
0.932
0.915
0.911
0.927
0.898
0.933
0.922
0.907
0.896
0.954
0.916
0.953
0.942
0.282
0.283
0.281
0.263
0.206
0.238
0.265
0.244
0.247
0.319
0.252
0.270
0.278
0.260
0.259
0.300
0.299
0.271
0.205
0.227
dioxane
RM0-log PKOWWIN
methanol
acetone
dioxane
RM0-milog P
methanol
acetone
dioxane
RM0-log P X2
RM0-log P X3
methanol
acetone
dioxane
methanol
acetone
dioxane
log kw-ACD/log P
log kw-log PKOWWIN
log kw-milog P
methanol
methanol
methanol
methanol
methanol
log kw-log P X2
log kw-log P X3
558
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
Table 5. Relationships between Slopes, S and S0, and log P
Values, Determined by Using Different Computational
Programs
Table 6. Molinspiration Set of Descriptors
compd milogPa TPSA natoms MW nON nOHNH nrotb volume
A1
2.42 37.38
2.87 37.38
2.47 46.61
19 251.28
20 265.31
21 281.31
20 285.73
20 330.18
20 267.28
3
3
4
3
3
4
6
5
3
3
3
3
3
4
4
6
4
3
3
3
3
4
4
4
3
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
2
2
3
2
2
2
3
4
2
2
2
3
3
4
3
4
3
3
3
3
3
4
3
3
3
228.99
S vs
A2
245.55
log P
modifier
equation
r
SD
N
A3
254.536
242.527
246.88
S-ACD
log P
methanol S = -1.876 - 0.651
ACD log P
0.898 0.319 25
0.886 0.370 25
0.836 0.376 25
0.896 0.321 25
0.943 0.265 25
0.878 0.329 25
A4
3.1
37.38
A5
3.23 37.38
1.94 57.61
acetone
S = -1.938 - 0.710
ACD log P
A6
237.01
A7
2.38 83.204 22 296.28
252.32
dioxane
S = -2.066 - 0.577
ACD log P
A8
1.97 63.85
2.84 37.38
3.072 37.38
23 309.32
20 265.31
20 285.73
20 330.18
25 327.383
26 341.41
273.52
A9
245.55
S-log
methanol S = -1.632 - 0.691
A10
A11
B1
242.53
PKOWWIN
logPKOWWIN
3.2
37.38
246.88
acetone
dioxane
S = -1.824 - 0.804
logPKOWWIN
3.61 37.38
4.06 37.38
300.075
316.636
425.611
308.093
323.409
316.935
305.007
313.611
317.916
324.065
335.62
B2
S = -1.755 - 0.644
logPKOWWIN
B3
3.67 46.614 27 357.41
3.13 57.608 26 343.382
3.57 83.024 28 372.38
3.37 61.172 27 352.393
B4
S-milog P methanol S = -0.800 - 0.874 milogP 0.899 0.319 25
B5
acetone
dioxane
S = -0.676 - 0.980 milogP 0.913 0.326 25
S = -1.019 - 0.803 milogP 0.868 0.341 25
B6
B7
3.77 37.38
4.29 37.38
4.42 37.38
4.69 37.38
26 345.373
26 361.828
26 406.279
26 453.279
S-log P X2 methanol S = -1.692 - 0.546 logPX2 0.871 0.355 25
B8
acetone
dioxane
S = -1.569 - 0.642 logPX2 0.927 0.300 25
S = -1.792 - 0.514 logPX2 0.882 0.348 25
B9
B10
B11
B12
B13
B14
S-log P X3 methanol S = -1.053 - 0.780 logPX3 0.903 0.311 25
3.51 54.451 28 369.42
3.34 61.172 27 252.393
3.11 57.608 26 343.382
acetone
dioxane
S = -0.964 - 0.874 logPX3 0.916 0.321 25
S = -1.249 - 0.712 logPX3 0.832 0.336 25
316.935
308.093
313.611
S0-ACD
log P
S0-log
methanol S = -4.138 - 0.340
ACD log P
0.806 0.253 24
4.26 37.38
26 361.828
a milogP, calculated logarithm of octanol-water partition coefficient;
TPSA, total polar surface area; natoms, number of atoms; MW, molecular
weight; nOH, number of hydrogen bond acceptors; nOHNH, number of
hydrogen bond donors; nrotb, number of rotatable bonds; volume,
molecular volume. Number of violations, nviolations, was zero in every case.
methanol S = -3.901 - 0.398
logPKOWWIN
0.888 0.196 24
PKOWWIN
S0-milog P methanol S = -3.541 - 0.466 milogP 0.828 0.239 24
S0-log P X2 methanol S = -3.850 - 0.339 logPX2 0.926 0.161 24
S0-log P X3 methanol S = -3.552 - 0.454 logPX3 0.903 0.183 24
equations, the slopes can be considered alternative to intercepts
lipophilicity parameters as proposed by Cserhati.17
equilibrium “shake-flask” method, the chromatographic reten-
tion constants, readily determined on the RP-C18 stationary
phases, could be successfully applied to predict lipophilicity of
that class of compounds.
Correlations of Retention Parameters, RM and log kw
(intercepts), and S and S0 (Slopes), of the RP LC Equations.
The slopes, S and S0, of the RP LC equations were correlated
with log P from calculation chemistry.11 These correlations are
shown in Table 5. It is evident that the slopes of RP LC equations
may be applied for lipophilicity expression of the compounds
investigated with similar reliability like the intercepts (Table 4).
That observation well agrees with a report by Valko.18
Correlations of Retention Constants, RM0 and log kw, and
Pharmacokinetic Descriptors. Calculations of physicochemical
properties of the chemical structures studied by using the
Molinspiration online program19 had shown that none of the
compounds investigated violates Lipinski’s rule of five (Table 6).
The total polar surface area (TPSA) molecular descriptor had
been shown to be a good predictor of drug absorption feasibility,
including intestinal absorption and general bioavailability, as well
as the blood-brain barrier penetration ability. The number of
rotable bonds (nrotb) parameter is considered a good descriptor
of oral bioavailability of drugs. Based on the results obtained, a
good intestinal absorption can be assumed for all the succinimide
compounds investigated.
0
Highly significant linear relationship between retention con-
0
stants, RM and log kw (intercepts), and S and S0 (slopes), of
the RP LC equations (Table 2) were obtained:
0
RM ðmethanolÞ ¼ - 1:183 - 1:055SðmethanolÞ
ð5Þ
r ¼ 0:985; SD ¼ 0:131; n ¼ 25
0
RM ðacetoneÞ ¼ - 0:787 - 0:838SðacetoneÞ
ð6Þ
r ¼ 0:996; SD ¼ 0:0063; n ¼ 25
0
RM ðdioxaneÞ ¼ - 0:995 - 0:886SðdioxaneÞ
ð7Þ
r ¼ 0:990; SD ¼ 0:086; n ¼ 25
log kwðmethanolÞ ¼ - 2:778 - 1:289SðmethanolÞ
ð8Þ
Plasma protein binding (PPB) and human intestinal absorp-
tion (HIA) predictors (Table 7) were estimated for the agents
studied by the ChemSilico program.20
r ¼ 0:961; SD ¼ 0:223; n ¼ 25
Taking into account quite a satisfactory quality (in terms of
statistical characteristics) of the obtained relationships between
The comparison of HIA values with RM0 and log kw values of
the RP LC equation is shown in Figure 2. The RM0 and log kw
0
slopes, S and S0, and intercepts, RM and log kw, of the RP LC
559
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
Table 7. Predictors of Binding to Receptors Calculated by the
ChemSilico Program20
kinase
nuclear
inhibitor receptor
compd HIAa PPB GPCR potency affinity
ion channel
binding
1A
96.3 76.395 -0.57 -0.59
96 93.901 -0.59 -0.58
96.8 88.928 -0.51 -0.53
96 91.493 -0.53 -0.55
-0.67
-0.68
-0.52
-0.7
-0.51
-0.56
-0.55
-0.47
-0.54
-0.44
-0.56
--
2A
3A
4A
5A
95.9 94.196 -0.64 -0.52
95.4 58.765 -0.49 -0.52
92.2 84.43 -0.65 -0.66
96.5 88.069 -0.72 -0.75
-0.84
-0.4
6A
7A
-0.62
-0.82
-0.68
-0.89
-0.98
-0.22
-0.27
-0.18
-0.06
-0.29
-0.27
-0.16
-0.29
-0.39
-0.2
8A
9A
96
96
93.87 -0.6
-0.58
-0.58
-0.48
-0.53
-0.31
-0.36
-0.35
-0.27
-0.37
-0.24
-0.25
-0.29
-0.34
-0.45
-0.32
-0.25
-0.29
-0.29
10A
11A
1B
91.564 -0.53 -0.53
95.9 94.332 -0.64 -0.46
96.1 92.876 -0.2 -0.2
2B
95.2 96.829 -0.23 -0.22
96.9 97.891 -0.18 -0.19
95.8 82.949 -0.16 -0.17
3B
4B
5B
94.1 98.18 -0.3
-0.31
6B
94.9 97.695 -0.13 -0.14
95.9 97.765 -0.13 -0.12
7B
8B
95
97.287 -0.18 -0.19
9B
94.2 96.544 -0.27 -0.17
Figure 2. Relationships between retention constants RM0 (a) and log kw
(b) and the HIA predictor. Open triangles denote agents excluded from
regression as outliers. Open circles refer to a subgroup of nitro, nitrile
and acetyl derivatives.
10B
11B
12B
13B
14B
94.1 96.241 -0.2
94.9 98.05 -0.2
-0.16
-0.27
-0.23
-0.27
-0.2
94.9 97.651 -0.14 -0.14
95.8 82.519 -0.16 -0.16
93.935 -0.19 -0.18
concluded from Figure 3b, the strongest binding to plasma
protein (above 95%) can be expected for agents having log kw
above 4. That would correspond to log P above 3.2. Hence, it
would be significantly higher than log P optimal for intestinal
absorption, namely 2-2.5 units.
-0.43
a HIA, human intestinal absorption; PBP, plasma protein binding;
GPCR, G-protein coupled receptor binding.
values of compounds containing nitro groups, nitrile group and
acetyl group (A7, B5, B6, B11, B12) formed one parabolic
correlation, while RM and log kw values of fourteen other
Finally, the Molinspiration online program was used for
druglikeness calculation. Compounds were analyzed as potential
GPCR ligands, kinase inhibitors, ion channel modifiers and
nuclear receptor ligands (Table 7). The compounds investigated
were not expected to bind for GPCR, ion channels nor nuclear
receptors, and they should not inhibit kinase. The results of the
calculations confirmed these presumptions. Comparisons of
0
compounds, as seen in Figure 2, determined the other curvilinear
dependence. Statistics for the relationships are given in Table 8.
They can be considered satisfactory. However, to arrive at them
four agents (A1, A3, A8 and B3) had to be excluded from the
regression analysis as evident outliers. The maximum of the main
curves in Figure 2 corresponds to a RM0 value between 2.25 and
2.50 and log kw values between 3.25 and 3.50. That, in turn,
according to the equations given in Table 4, would correspond to
a log P value between 2 and 2.5.21 That is a lipophilicity range
which is considered optimal for drug intestinal absorption. The
most commonly used anticonvulsive drugs have log P around
that value range, e.g., carbamazepine 2.45, phenytoin 2.47,
clonazepam 2.41, methsuximide 2.38.22 Therefore, the agents
studied appear to be promising candidates for antiepileptic drugs.
Based on data from the ChemSilico program, all the newly
synthesized compounds are supposed to bind to plasma proteins
at a relatively high degree. The relationships between PPB values
0
average RM values from the systems employing methanol,
acetone and dioxane as mobile phase modifiers, and log kw from
the methanol-modified eluent systems, with the parameters of
succinimide binding to GPCR, ion channels and nuclear recep-
tors, and kinase inhibition are shown in Figures 4, 5, 6, and 7. The
plots are grouped in two clusters depending on the overall
lipophilicity of the compounds. It can be noted that the weak
activities considered are even less pronounced in the case of
compounds which are less lipophilic (compounds of the group
A) as compared to the compounds which are more lipophilic
(compounds of the group B). One can argue that there is a jump
in binding properties when moving from group A to group B of
agents, despite a partial overlapping in lipophilicity parameters.
That would suggest that other parameters, apart from lipophili-
city, are involved. That could perhaps be true, but the binding
properties are very low for all the agents studied and hence the
error in their evaluation might be hardly comparable to that of
lipophilicity determination. For sure, both groups of new
0
and both RM and log kw values of the RP LC equations are
parabolic (Figure 3), except compounds A3, A8 and A9 in
0
relationship PPB vs RM and compounds A7 and B13 in the
relationship PPB vs log kw. The statistical parameters of the
curvilinear relationship of PPB are very good (Table 9). As can be
560
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
Table 8. Correlations between the HIA Predictor and Lipophilicity Parameters, RM0 and log kw
modifier
equation
r2
SD
N
0 2
RM0 methanol
HIA = 93.131 þ 2.535 RM0 - 0.561 (RM
)
0.970
0.984
0.864
0.998
0.821
0.856
0.867
0.999
0.101
0.070
0.260
0.026
0.299
0.221
0.262
0.008
14
6
0
2
2
2
HIA = 80.604 þ 10.252 RM0 - 1.834 (RM
)
)
)
0 2
RM0 acetone
HIA = 92.194 þ 3.648 RM0 - 0.866 (RM
)
15
5
0
HIA = 78.282 þ 12.053 RM0 - 2.182 (RM
0
2
RM0 dioxane
log kw methanol
HIA = 91.500 þ 4.759RM0 -1.238 (RM
)
15
5
0
HIA = 76.414 þ 14.904 RM0 - 2.989 (RM
HIA = 85.986 þ 6.042 log kw - 0.905 (log kw)2
15
5
HIA = -2.382 þ 49.286 log kw - 6.178 (log kw)2
Chromatography. TLC. Precoated RP-18W/UV254 plates
(Machery-Nagel GmbH and Co., D€uren, Germany) were used
for TLC analysis. The following mobile phases were used:
water - methanol ðjmethanol ¼ 0:55 - 0:75% v=vÞ
water - acetone ðjacetone ¼ 0:45 - 0:7% v=vÞ
water - acetonitrile ðjacetonitrile ¼ 0:45 - 0:75% v=vÞ
The plates were developed by ascending technique without
previous saturation of the chromatographic chambers with the
solvent. All measurements were carried out at ambient tempera-
ture. The investigated compounds were dissolved in acetone (2
mg/mL) and 0.2 μL solutions were separately spotted onto the
plates. Following development and drying of the plates, the spots
were observed under a UV light at λ = 254 nm. All the other
reagents used were of analytical grade purity.
HPLC. Experiments were done using a Merck-Hitachi La-
Chrome (Darmstadt, Germany; San Jose, CA, USA) apparatus
equipped with a diode array detector, autosampler and thermo-
stat. Chromatographic data were collected using a D-7000 HPLC
System Manager, version 3.1 (Merck-Hitachi). The system dwell
volume Vd equaled 1.74 mL. The extra column volume equaled
0.59 mL. It served to find the extra column time that has been
subtracted from all the retention times prior to any calculation.
XTerra MS C-18 column, 150 ꢀ 4.6 mm, particle size 5 μm
(Waters Corporation, Milford, MA, USA) was used. 1% thiourea
was injected to determine the column dead volume Vo, which was
1.44 mL.
Phosphate buffer of pH = 7.4 at concentration 0.01 M
was used.
Chromatographic measurements were done at 25 ꢀC with
eluent flow-rate of 1.0 mL/min.
Two chromatographic runs were conducted for each analyte
with different gradient development times, tg1 = 20 min and,
tg1 = 60 min, and with a wide methanol concentration range from
5% to 100%. The retention time for organic modifier gradients is
given by11,12
Figure 3. Relationships between retention constants RM0 (a) and log kw
(b) and PPB predictor. Open symbols denote agents excluded from
regression as outliers.
succinimide derivatives are devoid of significant receptor and
enzyme activity which would interfere with their main antic-
onvulsant mechanism of action.
’ EXPERIMENTAL SECTION
Synthesis of Succinimide Derivatives. All of the investi-
gated 1-(4-substituted-phenyl)-3-phenylpyrrolidine-2,5-diones
and 1-(4-substituted-phenyl)-3,3-diphenylpyrrolidine-2,5-diones
were synthesized from the corresponding succinic anhydrides
using a modified literature procedure.2,23 The succinic anhydride
was treated with the corresponding substituted aniline in reflux-
ing acetone for 2 h and concentrated, and the succinimic acid was
taken in acetic anhydride with sodium acetate. The mixture was
heated at 70 ꢀC for 2 h. Acetic anhydride was removed under
vacuum, and the crude material was purified by crystallization.
The chemical structures and the purities of the synthesized
8
0
R
>
t þ t þ t k for t e t
>
e
0
0
0
d
>
ꢀ
ꢀ
ꢁ
ꢁ
>
>
>
t0
td
>
>
t þ t0 þ td þ log 2:303bk0 1 -
þ 1
> e
>
b
t0k0
>
<
for td < t0 e tG þ td
tR
¼
R
ꢀ
ꢁ
>
>
>
t0
t0
>
>
>
t þ t þ t þ t -
þ
t0k0 - td þ
e
0
d
G
>
>
2:303b
2:303b
>
>
ꢀ10-bt =t for t0 > tG þ td
>
:
G
0
R
1
succinimides were confirmed by melting points and the H
NMR, 13C NMR, FT-IR and UV spectra.
ð9Þ
561
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
Table 9. Correlations between the PPB Predictor and Lipophilicity Parameters, RM0 and log kw
modifier
equation
r2
SD
N
0
2
RM0 methanol
PPB = 2.507 þ 56.408 RM0 - 8.262 (RM
)
0.982
1.373
0.855
2.149
0.743
2.915
0.980
3.327
21
4
0
PPB = 76.929 þ 7.683 RM
0
2
RM0 acetone
PPB = -4.496 þ 67.379 RM0 - 11.084 (RM
)
0.957
0.919
0.939
21
4
0
PPB = 61.798 þ 17.164 RM
0
2
RM0 dioxane
log kw methanol
PPB = 6.831 þ 62.659RM0 -10.850 (RM
)
21
4
0
PPB = 63.679 þ 18.184 RM
PPB = -111.646 þ 105.592 log kw - 13.197 (log kw)2
20
Figure 6. Relationship between retention constants RM0 (a) and log kw
Figure 4. Relationship between retention constants RM0 (a) and log kw
(b) and kinase inhibitor predictor.
(b) and GPCR binding predictor.
Figure 7. Relationship between retention constants RM0 (a) and log kw
(b) and the nuclear receptor binding predictor.
Figure 5. Relationship between retention constants RM0 (a) and log kw
(b) and ion channel binding predictor.
Two unknowns (log kw and S) were determined by solving a
system of two equations of the type of eq 9 using Matlab.
where te is an extra column time; t0 is a hold-up time; td is dwell
time, b is a steepness parameter equal to b = Sβt0; β is the
steepness of the gradient equal β = (jf - j0)/tG; j0 and jf are
the initial and final organic modifier contents in the mobile phase,
and k0 is a retention factor corresponding to the initial mobile
phase composition. The log kw is given by logkw = logk0þSj0.
’ CONCLUSION
A series of 25 new succinimide derivatives synthesized in the
project are potential anticonvulsant drug candidates. The
562
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563

Molecular Pharmaceutics
ARTICLE
compounds possess a characteristic for antiepileptic drug struc-
tural feature (pharmacophore). Their lipophilicity, both deter-
mined experimentally by two high performance reversed-phase
chromatographic techniques (HPLC and TLC) and calculated
from structural formula by the established drug design softwares,
is within the range providing efficient intestinal absorption. At
the same time, the plasma protein binding ability of the agents
has been evaluated as moderate and the side effects due to
interactions with selected receptors and enzymes can be expected
negligible. Certainly, the TLC method of determination of
lipophilicity of drug candidate is simple and inexpensive. How-
ever, HPLC provides the data which correlate better with the
pharmacokinetically relevant compound descriptors, in particu-
lar, as regards the plasma protein binding, PPB.
(11) Snyder, L. R.; Kirkland, J. J.; Glajch, J. L., Eds.; Practical HPLC
Method Development, 2nd ed.; Wiley-Interscience: New York, 1997.
(12) Kaliszan, R.; Wiczling, P.; Markuszewski, M. J. pH gradient
reversed-phase HPLC. Anal. Chem. 2004, 76, 749–760.
(13) Guillot, A.; Henchoz, Y.; Moccand, C.; Guillarme, D.; Veuthey,
J.-L.; Carrupt, P.-A.; Martel, S. Lipophilicity determination of highly
lipophilic compounds by liquid chromatography. Chem. Biodiversity
2009, 6, 1828–1836.
(14) Bate-Smith, E. C.; Westall, R. G. Some physical properties of
aqueous solution of tetramethylammonium bromide and tetramethyl-
ammonium iodide. Biochim. Biophys. Acta 1950, 4, 427–427.
(15) Soczewiꢀnski, E. Mechanistic molecular model of liquid-solid
chromatography retention-eluent composition relationship. J. Chroma-
togr., A 2002, 965, 109–116.
(16) http://146.107.217.178/lab/alogps/start.html.
(17) Cserhati, T. Lipophilicity determination of some monoamine
oxidase inhibitors: the effect of methanol and ammonium chloride. J. Liq.
Chromatogr. 1993, 16, 1805–1817.
(18) Valko, K. Measurments of physical properteis for drug desing in
industry. In Separation methods in Drug Synthesis and Purification; Valko,
K., Ed.; Elsevier: Amsterdam, 2000; pp 535-583.
(19) http://www.molinspiration.com/services/properties.html.
(20) http://www.chemsilico.com/.
Of the set of 25 compounds considered for further drug
development studies the agents A4, A5, A10 and A11 are
proposed as the best compromising the human intestinal absorp-
tion and plasma protein binding properties.
’ AUTHOR INFORMATION
Corresponding Author
(21) Dennis, M. J., Absorption processes. In Comprehensive Medic-
inal Chemistry Vol. 5; Taylor, J. B., Ed.; Pergamon Press: Oxford, 1990;
pp 9-11.
(22) Craig, P. N. Drug compendium. In Comprehensive Medicinal
Chemistry Vol. 6; Drayton, C. J., Ed.; Pergamon Press: Oxford, 1990; pp
237-965.
*R.K.: Medical University of Gdaꢀnsk, Al. Gen. Hallera 107, 80-
416, Gdaꢀnsk, Poland. Phone: þ 48 58 349 3261. Fax: þ48 58
349. E-mail: roman.kaliszan@gumed.edu.pl.
(23) Miller, C. A.; Schol, H. J.; Long, L. M. Anticonvulsants. II. A
study of N-R-R,β-substituted succimides. J. Am. Chem. Soc. 1951,
73, 5608–5610.
’ ACKNOWLEDGMENT
This work was performed within the framework of the
research projects No. OI 172013, supported by the Ministry of
Science of Serbia, and No. NN405630038, supported by the
Ministry of Science and Higher Education of Poland.
’ REFERENCES
(1) Goodman, L. S.; Gilman, A. Succinimides. In Goodman and
Gilman’s the Pharmacological Basis of Therapeutics, 10th ed.; Hardman,
J. G., Limbird, L. E., Eds.; McGraw-Hill: New York, 2001; pp 535-536.
(2) Katritzky, R. A.; Nair, K. S.; Witeck, M. R.; Hutchins, M. S.
Synthesis of 3,3-diarylpyrrolidines from diaryl ketones. ARKIVOC, Issue
in Honor of Prof. Gabor Bernath, 2003, 9-18.
(3) Hudkins, L. R.; DeHaven-Hudkins, L. D.; Doukas, P. Design of
dual acting anticonvulsant-antimuscarinic succinimide and hydantoin
derivatives. Bioorg. Med. Chem. Lett. 1997, 7, 979–984.
(4) Poupaert, H. J.; Vandervorst, D.; Guiot, P.; Moustafa, M. M. M.;
Dumont, P. Structure-activity relationships of phenytoin-like anti-
convulsant drugs. J. Med. Chem. 1984, 27, 76–78.
(5) Henchoz, Y.; Bard, B.; Guillarme, D.; Carrupt, P. A.; Veuthey,
J. L.; Martel, S. Analytical tools for the physicochemical profiling of drug
candidates to predict absorption/distribution. Anal. Bioanal. Chem.
2009, 394, 707–729.
(6) Kaliszan, R. QSRR Quantitative structure-(chromatographic)
retention relationships. Chem. Rev. 2007, 107, 3212–3246.
(7) Di, L.; Kerns, E. H. Pharmaceutical profiling in discovery
research. Curr. Opin. Chem. Biol. 2003, 7, 402–408.
(8) Kerns, E. H.; Di, L. Pharmaceutical profiling in drug discovery.
Drug Discovery Today 2003, 8, 316–322.
(9) Nasal, A.; Siluk, D.; Kaliszan, R. Chromatographic retention
parameters in medicinal chemistry and molecular pharmacology. Curr.
Med. Chem. 2003, 10, 381–426.
(10) Janjiꢀc-Periꢁsiꢀc, U. N.; Djakoviꢀc-Sekuliꢀc, L. T.; Stojanoviꢀc, Z. S.;
Penov-Gaꢁsi, M. K. HPTLC chromatography of androstene derivates.
Application of normal phase thin-layer chromatographic retention data
in QSAR studies. Steroids 2005, 70, 137–144.
563
dx.doi.org/10.1021/mp100373d |Mol. Pharmaceutics 2011, 8, 555–563