Design and preliminary structure-activity relationship of redox-silent semisynthetic tocotrienol analogues as inhibitors for breast cancer proliferation and invasion

Article abstract of DOI:10.1016/j.bmc.2009.11.054

Vitamin E (VE) is a generic term that represents a family of compounds composed of various tocopherol and tocotrienol isoforms. Tocotrienols display potent anti-angiogenic and antiproliferative activities. Redox-silent tocotrienol analogues also display potent anticancer activity. The ultimate objective of this study was to develop semisynthetically C-6-modified redox-silent tocotrienol analogues with enhanced antiproliferative and anti-invasive activities as compared to their parent compound. Examples of these are carbamate and ether analogues of α-, γ-, and δ-tocotrienols (1-3). Various aliphatic, olefinic, and aromatic substituents were used. Steric limitation, electrostatic, hydrogen bond donor (HBD) and hydrogen bond acceptor (HBA) properties were varied at this position and the biological activities of these derivatives were tested. Three-dimensional quantitative structure-activity relationship (3D QSAR) studies were performed using Comparative Molecular Field (CoMFA) and Comparative Molecular Similarity Indices Analyses (CoMSIA) to better understand the structural basis for biological activity and guide the future design of more potent VE analogues.

Full text of DOI:10.1016/j.bmc.2009.11.054

Bioorganic & Medicinal Chemistry 18 (2010) 755–768
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Design and preliminary structure–activity relationship of redox-silent
semisynthetic tocotrienol analogues as inhibitors for breast cancer
proliferation and invasion
*
Ahmed Y. Elnagar, Vikram B. Wali, Paul W. Sylvester, Khalid A. El Sayed
Department of Basic Pharmaceutical Sciences, College of Pharmacy, University of Louisiana at Monroe, Monroe, LA 71201, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Vitamin E (VE) is a generic term that represents a family of compounds composed of various tocopherol
and tocotrienol isoforms. Tocotrienols display potent anti-angiogenic and antiproliferative activities.
Redox-silent tocotrienol analogues also display potent anticancer activity. The ultimate objective of this
study was to develop semisynthetically C-6-modified redox-silent tocotrienol analogues with enhanced
antiproliferative and anti-invasive activities as compared to their parent compound. Examples of these
Received 2 October 2009
Revised 22 November 2009
Accepted 24 November 2009
Available online 27 November 2009
are carbamate and ether analogues of a-, c-, and d-tocotrienols (1–3). Various aliphatic, olefinic, and aro-
Keywords:
matic substituents were used. Steric limitation, electrostatic, hydrogen bond donor (HBD) and hydrogen
bond acceptor (HBA) properties were varied at this position and the biological activities of these deriva-
tives were tested. Three-dimensional quantitative structure–activity relationship (3D QSAR) studies were
performed using Comparative Molecular Field (CoMFA) and Comparative Molecular Similarity Indices
Analyses (CoMSIA) to better understand the structural basis for biological activity and guide the future
design of more potent VE analogues.
Anti-invasive
Antiproliferative
Breast cancer
Carbamoylation
CoMFA
CoMSIA
3D QSAR
Etherification
Redox-silent
Tocotrienols
Vitamin E
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
stances corresponding to d-tocotrienol P
c-tocotrienol > a-tocotri-
enol > d-tocopherol > -, and
c
a
-tocopherol.5,6
Vitamin E (VE) is a generic term used for describing a family of
eight different naturally occurring phenolic compounds.1 The VE
family members are further subdivided into two subgroups, toc-
opherols and tocotrienols.2 Tocopherols contain a saturated phytyl
side chain at C-2 while tocotrienols have an unsaturated side chain,
VE appears to play an important role in reducing risk of cancer
and cardiovascular diseases.7–9 VE members have documented po-
tent anti-angiogenic and antiproliferative activities.7–9
a-Tocoph-
erol exerts good bioavailability due to its selective recognition by
transporter proteins, (e.g., alpha-tocopherol transfer (a-TT) and al-
0
0
0
with all E D3 ,7 ,11 system. The four members of each family differ by
pha-tocopherol-associated proteins).7–9 Some cancer cells can
inactivate VE members by rapidly metabolizing them to the corre-
sponding carboxyethyl-hydroxychromans. Semisynthetic modifi-
cations of VE analogues targeted the improvement of their
potency or to increase water solubility via the esterification of C-
the number and placement of methyl groups on the chromanol ring,
resulting in a-, c
-, d-tocopherols and tocotrienols.3,4 However, indi-
vidual tocopherol and tocotrienol isoforms display a wide-range of
antioxidant potencies.4 This activity shows a poor correlation to
the biopotency of many antioxidant-independent actions of these
compounds.4 Evidence suggests that the level of phytyl chain satura-
tion and/or chroman ring methylation is critical in determining the
differential biopotencies demonstrated by individual VE isoforms.1–
6 phenolic group.10,11 Amides derived from
a- and d-tocopheram-
ines recently reported and demonstrated improved anticancer
activity.12,13 The redox-silent analogue of 1, 6-O-carboxypropyl-
a
-tocotrienol, showed improved anti-carcinogenic and anti-inva-
4
Direct comparisons between the two VE subclasses showed that
sive activities against lung cancer A549 cells through the suppres-
sion of hypoxia adaptation of these cells by the inhibition of
hypoxia-induced activation of Src signaling.14
tocotrienols are significantly more potent than tocopherols in most
biological actions, and display a consistent relationship in most in-
The ultimate objective of this study is to design semisyntheti-
cally C-6-modified redox-silent tocotrienol analogues with en-
hanced anticancer effects. The accessible C-6 phenolic group was
utilized to investigate the steric limitations, electronic properties
* Corresponding author. Tel.: +1 318 342 1725; fax: +1 318 342 1737.
E-mail address: elsayed@ulm.edu (K.A. El Sayed).
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.11.054

756
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
and HBD and HBA characters at this position to probe and correlate
the surrounding chemical tolerance and biological activity. Various
aliphatic, olefinic, and aromatic substituents were selected. This
study reports several new tocotrienol carbamate and ether ana-
logues with potent antiproliferative and anti-invasive activities
along with 3D QSAR analyses.
R1
R2
R3
Compound
4
5
6
7
8
CH3
H
H
CH3
H
H
CH3
H
H
CH3
H
H
CH3
H
H
CH3
H
H
CH3
CH3
H
CH3
CH3
H
CH3
CH3
H
CH3
CH3
H
CH3
CH3
H
CH3
CH3
H
a
a
a
b
b
b
c
c
c
d
d
d
e
e
e
f
9
10
11
12
13
14
15
16
17
18
19
20
21
22
2. Results and discussion
2.1. Semisynthesis of tocotrienol analogues
Tocotrienols 1–3 were isolated from a palm oil-derived tocotri-
enol rich fraction (TRF) generously provided by First Tech Interna-
tional Ltd (Hong Kong) using normal phase vacuum liquid
chromatography (VLC, hexane–dioxane, isocratic 99:1), followed
by preparative HPLC (C-18 RP, isocratic H2O–MeOH, 5:95).15
f
f
g
12
R1
4
H
H
HO
9
16`
CH3
15`
CH3
13`
CH3
11
6
The HRESIMS suggested the molecular formula C37H51NO5S. 1H and
13C NMR data of 7 indicated the presence of 6-O-methyl tosylcarba-
mate (Tables S1 and S2). The benzylic methyl group at H3-800 (dH
2.44) showed a 2J-HMBC correlation with the quaternary aromatic
carbon C-500 (dC 135.6) and 3J-HMBC correlations with the methine
carbons C-400and C-600 (dC 128.6 and 129.1, respectively). Thus, com-
pound 7 was determined to be (R)-2,5,7,8-tetramethyl-2-(30E,70E)-
40,80,120-trimethyltrideca-30,70,110-trienyl) chroman-6-yl tosylcar-
bamate. Similarly, analysis of the 1H and 13C NMR data of com-
pounds 8 and 9 (Tables 1 and 2) confirmed their identity as the
chroman-6-yl tosylcarbamates of 2 and 3, respectively.
Reaction of 1 with 4-chlorophenylsulfonylisocyante afforded
compound 10. The HREIMS data of 10 suggested the molecular for-
mula C36H48ClNO5S with the M and M + 2, 3:1, isotopic cluster
characteristic pattern for a monochlorinated compound. The 1H
and 13C NMR (Tables 1 and 2) data showed that the 4-chloro-
phenylsulfonylcarbamoyl moiety has been introduced at C-6. This
was based on the downfield shift of carbons C-5, C-7, and C-8 and
the presence of the new carbamoyl carbonyl C-100at d 150.2. There-
fore, compound 10 was proved to be (R)-2,5,7,8-tetramethyl-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman-6-yl-
400-chlorophenylsulfonylcarbamate. Similarly, the NMR data of 11
and 12 (Tables S3 and S4) confirmed their identity as the C-6-400-
chlorophenylsulfonylcarbamate analogues of 2 and 3, respectively.
The HRMS, 1H, and 13C NMR data of 13 (Tables S3 and S4) sug-
gested the molecular formula C36H49NO3 and possible C-6 phenyl
carbamoylation of 1. The aromatic protons H-400and H-600at dH
7.32 were assigned based on their 3J-HMBC correlation with the
quaternary aromatic carbon C-200 (dC 145.2, Table S4). Protons H-
400/H-600 showed COSY coupling with protons H-300/H-700 (dH 7.47)
as well as with H-500 (dH 7.08). Protons H-300/H-700 showed a 3J-
HMBC correlation with the aromatic methine carbon C-500. Thus,
compound 13 was determined to be (R)-2,5,7,8-tetramethyl-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman-6-yl
phenylcarbamate. Similarly, compounds 14 (Tables S3 and S4) and
15 (Tables S5 and S6) were found to be the C-6 phenylcarbamate
analogues of 2 and 3, respectively.
10'
13
R2
O
2
10
CH3
CH
8
4'
8'
12'
3
1
1'
14`
14
R1 R2
α-tocotrienol (1) CH3 CH3
γ-tocotrienol (2) H CH3
δ-tocotrienol (3) H
H
Reaction of diverse aliphatic, olefinic, and aromatic isocyanates
with tocotrienols 1–3 afforded their corresponding C-6-carbamates
4–22. The carbamate functionality offered alternative HBD group in
these compounds, replacing the C-6 phenol moiety, in addition to
the HBA carbonyl group. The HREIMSa data of 4 suggested the
molecular formula C36H49NO5S and possible C-6 benzenesulfonyl
carbamoylation. The 1H and 13C NMR data (Tables 1 and 2) further
supported this conclusion. The aromatic protons H-400 and H-600 at
dH 7.54 (Table 1) were assigned based on their 3J-HMBC correlations
with the quaternary aromatic carbon C-200 (dC 139.6, Table 2). Pro-
tons H-400 and H-600 showed COSY coupling with H-300/H-700 (dH
8.01) as well as with H-500 (dH 7.65). Proton H-500 also showed 3J-
HMBC correlation with the aromatic methine carbons C-300/700 (dC
128.6). Thus, compound 4 was determined to be (R)-2,5,7,8-tetra-
methyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chro-
man-6-yl phenylsulfonylcarbamate. C-6 Carbamoylation induced
the downfield shift of C-5 (+5.1 ppm) and H-5 (+0.13) in 5, com-
pared to its parent 2 (Tables S1 and S2).14 Similar downfield shift
was also observed in 6 for C-5/H-5 in addition to +10.8 and
0.08 ppm shift for C-7/H-7, respectively, compared to its parent
compound 3 (Tables S1 and S2).16
O
R1
H
N
2``
S
1``
O
CH3
CH3
4``
CH3
CH3
R3
O
O
O
R2
6``
a
CH3
O
O
S
S
The HRESIMS data of 16 suggested the molecular formula
C37H51NO3. The 1Hand 13C NMRdata of 16 (Tables 1 and 2)suggested
C-6 benzyl carbamate of 1. The benzylic methylene protons H-200 (dH
4.46) showed 3J-HMBC correlations with H-400/C-800 (dC 127.6) C-800
(dC 45.4) and the carbamate carbonyl C-100 (dC 155.6). Thus, com-
pound 16 was determined to be (R)-2,5,7,8-tetramethyl-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman-6-yl
benzylcarbamate. Similarly, compounds 17 and 18 (Tables S5 and
S6) have been proven to be (R)-2,7,8-trimethyl-2-((30E,70E)-
O
O
5``
Cl
8``
b
c
d
2``
4``
10``
11``
2``
8``
O
2``
3``
O
4``
8``
6``
e
f
g

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
757
Table 1
1H NMR data of compounds 4, 8–10, and 16a
Position
dH
4
8
9
10
16
3
4
5
7
1.77, m
2.52, m
—
—
1.21, s
1.76, s
1.76, s
2.02, s
2.02, m
1.75, m
2.63, m
6.51, s
—
1.23, s
—
1.74, m
2.67, m
6.58, d (6.6)
6.58, d (6.6)
1.21, s
—
—
2.08, s
1.95, m
2.00, m
5.11, m
1.95, m
2.07, m
5.11, m
1.95, m
2.07, m
5.11, m
1.57, s
1.66, s
1.57, s
1.57, s
—
7.91, d (8.4)
7.33, d (8.4)
—
7.33, d(8.4)
7.94, d (8.4)
2.44, s
1.73, m
2.53, dd (6.6, 6.6)
—
1.77, m
2.59, dd (6.6, 6.6)
—
—
1.25, s
2,03, s
2.07, s
2.10, s
1.95, m
2.13, m
5.10, m
1.95, m
2.11, m
5.10, m
1.95, m
2.11, m
5.10, m
1.59, s
1.67, s
1.59, s
1.59, s
4.46, d (6.2)
—
7.30, d (6.9)
7.34, dd (5.8, 5.8)
7.32, dd (5.8, 5.8)
7.34, dd (5.8, 5.8)
7.30, d (6.9)
—
11
12
13
14
10
1.95, m
1.22, s
1,79, s
1.79, s
2.08, s
2.13, m
5.10, m
1.96, m
2.11, m
5.10, m
1.95, m
2.11, m
5.10, m
1.58, s
1.66, s
1.58, s
1.58, s
—
7.94, d (8.1)
7.34, d (8.1)
—
7.33, d (8.1)
7.93, d (8.1)
—
1.84, s
2.04, s
1.95, m
2.12, m
5.11, m
1.95, m
2.12, m
5.11, m
1.95, m
2.12, m
5.11, m
1.63, s
1.69, s
1.59, s
1.59, s
—
20
2.05, m
30
5.10, t (6.9)
1.94, m
2.00, m
5.10, t (6.9)
2.02, m
2.07, m
5.10, t (6.9)
1.67, s
1.58, s
1.58, s
1.58, s
—
8.01, d (7.7)
7.54, dd (7.7, 7.7)
7.65, dd (7.5, 7.5)
7.54, dd (7.7, 7.7)
8.01, d (7.7)
—
50
60
70
90
100
110
130
140
150
160
200
300
400
500
600
700
800
7.94, d (8.4)
7.32, d (8.1)
—
7.32, d (8.1)
7.94, d (8. 4)
2.44, s
a
In CDCl3, 400 MHz. Coupling constants (J) are in hertz.
40,80,120-trimethyltrideca-30,70,110-trienyl)chroman-6-yl benzylcar-
bamate and (R)-2,8-dimethyl-2-((30E,70E) 40,80,120-trimethyltri-
deca-30,70,110-trienyl)chroman-6-yl benzylcarbamate, respectively.
The HREIMS data of 19 suggested the molecular formula
C37H49NO5 and possible C-6 phenylformate carbamoylation. The
1H and 13C NMR data (Tables S5 and S6) further supported this con-
clusion. The aromatic protons H-500/H-700 at dH 7.40 (Table S5)
showed a 3J-HMBC correlation with the quaternary oxygenated
aromatic carbon C-300 (dC 149.8, Table S6). Protons H-500/H-700
showed COSY couplings with H-400/H-800 (dH 7.21) as well as with
H-600 (dH 7.26). Thus, compound 19 was determined to be (R)-phe-
nyl 2,5,7,8-tetramethyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,
110-trienyl)-chroman-6-yl iminodicarbonate. Similarly, compounds
20 and 21 (Tables S7 and S8) were shown to be (R)-2,7,8-tri-
methyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chro-
man-6-yl iminodicarbonate and (R)-2, 8-dimethyl-2-((30E,70E)-40,
80,120-trimethyltrideca-30,70,110-trienyl)-chroman-6-yl iminodicar-
bonate, respectively.
The HREIMS data of 22 suggested the molecular formula
C38H47NO3 and possible C-6 naphthyl carbamoylation. The 1H and
13C NMR data (Tables S7 and S8) further supported this conclusion.
ThearomaticprotonsH-400 (dH 7.51) andH-900 (dH 7.93)were assigned
based on their 3J-HMBC correlations with the quaternary aromatic
carbons C-200 and C-1100 (dC 135.0 and 134.0, respectively, Table S8).
Proton H-900 was COSY-coupled to H-800 (dH 7.51) which showed a
3J-HMBC correlation with C-600 (dC 128.9). Furthermore, H-400 showed
COSY coupling with H-300 and H-500 which showed 3J-HMBC correla-
tions with C-1000 (dC 132.3). Thus, compound 22 was determined to
be (R)-2,8-dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-
trienyl) chroman-6-yl naphthalen-100-yl carbamate.
(dH 3.87) showed COSY coupling with H-200 (dH 2.44). Proton H-100
3
showed a J-HMBC correlation with C-300 (dC 110.1). Proton H-300
3
(dH 5.22) showed J-HMBC correlations with C-500 and C-600 (dC
25.7 and dC 17.8, respectively) and 2J-HMBC correlation with the
quaternary carbon C-400 (dC 134.3). Thus, compound 24 was deter-
mined to be (R)-2,7,8-trimethyl-6-(400-methylpent-300-enyloxy)-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl) chroman.
R1
5``
O
CH3
CH3
CH3
CH3
R3
6``
R2
O
1``
a
CH3
8``
OCH3
14``
15``
12``
4``
5``
1``
3``
9``
1``
b
c
R
1``
R
d H
e OCH3
Compound
R1
R2
R3
23
24
25
26
27
28
29
30
31
32
H
H
H
CH3
H
H
CH3
H
CH3
CH3
CH3
CH3
CH3
H
CH3
CH3
CH3
CH3
CH3
a
b
c
c
c
d
d
e
CH3
H
Reaction of 1–3 with alkyl or aryl halides in THF in presence of
NaH afforded the corresponding ethers 23–32. The HREIMS, 1H,
and 13C NMR data of 24 (Tables S7 and S8) suggested the molecular
formula C34H52O2 and possible C-6 etherification of 2 with a 4-
methylpent-3-enyl moiety. The methylene protons H2-100
e
The HREIMS, 1H, and 13C NMR data of 25 (Tables S7 and S8) sug-
gested the molecular formula C33H52O2 and possible C-6 isopentyl

758
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
ether analogue of 2. The methylene protons H2-100 (dH 3.87) showed
a 3J-HMBC correlation with C-300 (dC 25.2). The methyl doublets H3-
400 and H3-500 (dH 0.95) showed 2J-HMBC correlations with C-300 and a
3J-HMBC correlation with C-200 (dC 22.7). Thus, compound 25 was
determined to be (R)-6-(isopentyloxy)-2,7,8-trimethyl-2-((30E,70E)-
40,80,120-trimethyltrideca-30,70,110-trienyl)chroman.
The HREIMS data of 29 suggested the molecular formula
C37H52O3. The 1H and 13C NMR data (Tables S11 and S12) suggested
that compound 29 is a C-6 4-methoxybenzyl ether analogue of 1.
The benzylic methylene protons H2-100 (dH 4.66) showed 3J-HMBC
correlations with the quaternary carbon C-6 (dC 147.9) and the aro-
matic methine carbons C-300 and C-700 (dC 119.9 and 113.3, respec-
tively), connecting the new 4-methoxybenzyl ether with the
chroman ring of 1. The aromatic proton singlet H-300 (dH 7.03)
showed 3J-HMBC correlations with the aromatic methine carbons
C-500 (dC 113.1) and C-700. The proton doublet of doublet H-600 (dH
7.30) showed 3J-HMBC correlations with the quaternary carbons
C-200 and C-400at (dC 139.8 and 160.2, respectively). A double dou-
blet proton H-600 showed the COSY coupling with H-700 at dH 7.30.
The methoxy singlet H3-800 (dH 3.86) showed a 3J-HMBC correlation
with C-400. Thus, compound 29 was determined to be (R)-6-(400-
methoxybenzyloxy)-2,5,7,8-tetramethyl-2-((30E,70E)-40,80,120-trim-
ethyltrideca-30,70,110-trienyl)chroman. Compound 30 was assigned
in a similar manner based on the 1H and 13C NMR data (Tables S11
and S12) and was determined to be (R)-6-(400-methoxybenzyloxy)-
2,7,8-trimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-
trienyl)chroman.
The HREIMS data of 31 suggested the molecular formula
C38H54O4. The 1H and 13C NMR data (Tables S11 and S12) was closely
similar to those of 29 with C-6 (400, 600-dimethoxybenzyloxy moiety.
The methoxy proton singlets H3-800 and H3-900 (dH 3.81) showed 3J-
HMBCcorrelations with carbons C-400 and C-600 (dC 160.9). Thus, com-
pound 31 was determined to be (R)-6-(400,600-dimethoxybenzyloxy)-
2,5,7,8-tetramethyl-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-
trienyl)chroman. Similarly, compound 32 was assigned based on the
1H and 13C NMR data (Table S13) and was determined to be (R)-6-
(400,600-dimethoxybenzyloxy)-2,7,8-trimethyl-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl) chroman.
The HREIMS data of 26 suggested the molecular formula
C44H68O2. The 1H and 13C NMR data of 26 (Tables S9 and S10) sug-
gested C-6 farnesyl ether of 1. The methylene protons doublet H2-
100 (dH 4.18) showed a 3J-HMBC correlation with the quaternary oxy-
genated carbon C-6 (dC 148.0). Protons H2-100 showed COSY coupling
with H-200 (dH 5.58). Proton H-100 also showed a 3J-HMBC correlation
with C-300 (dC 140.6). The methyl protons singlet H3-1500 (dH 1.60)
3
showed J-HMBC correlations with C-200 and C-400 (dC 124.2 and
39.8, respectively) and a 2J-HMBC correlation with C-300, connecting
the isoprenyl segment C-100-C-400/C-1500. Similarly, the other two iso-
prenyl segments C-500-C-800/C-1400 and C-900-C-1300 were confirmed.
Thus, compound 26 was determined to be (R)-2,5,7,8-tetramethyl-
6-(200E,600E)-300,700,1100-trimethyldodeca-200,600,1000-trienyloxy)-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman. Simi-
larly, compounds 27 and 28 (Tables S9 and S10) have been proven
to be (R)-2,7,8-trimethyl-6-(200E,600E)-300,700,1100-trimethyldodeca-
200,600,100-trienyloxy)-2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110
-trienyl)chromanand(R)-2,8-dimethyl-6-((200E,600E)-300,700,1100-trim-
ethyldodeca-200,600,1000-trienyloxy)-2-((30E,70E)-40,80,120-trimethyl-
trideca-30,70,110-trienyl)chroman.
Table 2
13C NMR data of compounds 4, 8–10, and 16a
Position
dC
4
8
9
10
16
2
75.0, qC
76.0, qC
76.1, qC
75.1, qC
76.1, qC
3
4
5
6
7
8
9
31.0, CH2
20.5, CH2
117.5, qC
138.4, qC
125.3, qC
126.9, qC
123.3, qC
149.9,qC
23.8 CH3
11.7, CH3
12.5, CH3
11.8, CH3
39.8, CH2
22.2, CH2
124.4, CH
135.0, CH2
39.8, CH2
26.6, CH2
124.2, CH
135.0, qC
39.8, CH2
26.8, CH2
124.2, CH
131.3, qC
25.8, CH3
17.8, CH3
16.1, CH3
16.0, CH3
149.1, qC
139.6, qC
128.6,CH
129.1,CH
134.1,CH
129.1,CH
128.6,CH
—
30.9, CH2
22.2, CH2
118.7, qC
140.6, qC
126.1, qC
127.1, qC
118.5, qC
149.3,qC
24.1 CH3
11.7, CH3
12.5, CH3
11.8, CH3
40.0, CH2
22.2, CH2
124.1, CH
135.3, CH2
39.7, CH2
26.6, CH2
124.2, CH
135.3, qC
39.7, CH2
26.8, CH2
124.2, CH
131.3, qC
25.8, CH3
17.8,CH3
16.1, CH3
16.0, CH3
150.0, qC
145.3, qC
128.6, CH
129.7, CH
135.0, qC
129.7, CH
128.6, CH
21.7 CH3
30.9, CH2
22.2, CH2
120.8, CH
141.8, qC
118.8, CH
127.8, qC
121.0, qC
150.4,qC
24.1, CH3
—
31.0, CH2
22.2, CH2
123.6, qC
140.0, qC
125.5, qC
127.1, qC
117.5, qC
149.2, qC
24.1, CH3
11.7, CH3
11.8, CH3
12.5, CH3
39.7, CH2
20.5, CH2
124.4, CH
135.5, qC
39.7, CH2
26.6, CH2
124.2, CH
135.3, qC
39.7, CH2
26.8, CH2
124.1, CH
131.2, qC
25.7, CH3
17.9, CH3
16.0, CH3
15.9, CH3
150.2, qC
145.2,qC
129.7, CH
128.6, CH
135.6, qC
128.6, CH
129.7, CH
—
31.2, CH2
20.6, CH2
117.3, CH
140.9, qC
125.5, qC
125.9, qC
123.0, qC
149.6, qC
23.9, CH3
11.9, CH3
13.2, CH3
11.9, CH3
39.8, CH2
22.2, CH2
124.4, CH
135.2, qC
39.7, CH2
26.6, CH2
124.2, CH
135.0, qC
39.7, CH2
26.8, CH2
124.4, CH
131.8, qC
25.8, CH3
18.0, CH3
16.4, CH3
16.3, CH3
155.6, qC
45.4, CH2
139.1, qC
127.6, CH
129.4, CH
136.7, qC
129.4, CH
127.6, CH
2.2. Biological activity
Tocotrienols 1–3 are significantly more potent than tocopherols
in suppressing tumor cell growth and viability. The anticancer ef-
fects of tocotrienols are observed using treatments doses that have
little or no effect on normal cell function or viability.4,7 Recent find-
10
11
12
13
14
10
ings also suggest that the antiproliferative c-tocotrienol may occur
upstream of the PI3K/PDK-1/AKT mitogenic signaling pathway at
—
the level of the EGF-receptor.4,14 The antiproliferative effects of
14.1, CH3
40.0, CH2
22.2, CH2
124.1, CH
135.3, CH2
39.7, CH2
26.6, CH2
124.2, CH
135.3, qC
39.7, CH2
26.8, CH2
124.2, CH
131.3, qC
25.8, CH3
17.8, CH3
16.1, CH3
15.9, CH3
150.8, qC
145.1, qC
128.5, CHb
129.7, CHc
135.0, qC
128.6, CHc
129.6, CHb
21.7, CH3
c-tocotrienol on neoplastic +SA mammary epithelial cells are med-
20
iated through the reduction in ErbB3 tyrosine phosphorylation and
subsequent activation of the PI3K/PDK-1/AKT mitogenic signaling
pathway.4,14 Treatment with a growth inhibitory, but not cytotoxic
30
40
50
dose of c-tocotrienol, results in approximately a 50% reduction in
60
70
EGF-dependent +SA cell growth, and these effects are associated
with a corresponding decrease in phospho-AKT (active) levels in
these cells, and this lead finally to apoptosis and cell death.4,13,17
The antiproliferative and anti-invasive activities of 1–32 were
evaluated using MTT and cell invasion assay kits, respectively.
The antiproliferative activity has been evaluated against the highly
malignant +SA mouse mammary epithelial cells and the human
breast cancer MCF-7 and MDA-MB-231 cell lines. The later cell line
was also used for the assessment of the anti-invasive activity. With
its documented antiproliferative activity, 2 was used as a positive
control in all biological assays.13 MTT-assay detects the living,
but not dead cells, in quantitative colorimetric fashion utilizing cel-
lular ability to reduce the MTT to insoluble purple formazan dye.17
Therefore, this method is routinely used to assess the cytotoxicity
and proliferation inhibitory activities.
80
90
100
110
120
130
140
150
160
100
200
300
400
500
600
700
800
While many of the prepared semisynthetic tocotrienols re-
tained the activity, some of them were more active than the start-
ing natural products. Compound 8 was the most active against +SA
a
In CDCl3, 100 MHz. Carbon multiplicities were determined by DEPT135° or APT
experiments. qC = quaternary, CH = methine, CH2 = methylene, CH3 = methyl
carbons.
b,c
Interchangeable within the same column.
mouse mammary epithelial cells with an IC50
2 lM, which repre-

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
759
Table 3
Antiproliferative activity of tocotrienols 1–21 against the human breast cancer cell
lines MCF-7 and MDA-MB-231
4
3
2
1
Compound
IC50
l
M MDA-MB-231
IC50
>20
11.4 1.53
15.5 0.40
4.5 1.02
>20
6.1 1.40
6.1 1.67
2.4 0.35
>20
5.1 0.45
>20
>20
>20
>20
>20
>20
>20
>20
>20
lM MCF7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
33
34
35
>10
>10
>10
>10
>10
>10
>10
>10
6.6 1.70
>10
>10
>10
>10
>10
>10
7.2 0.56
>10
>10
>10
ND
ND
ND
0
1
2
3
4
5
6
Conc in μM
Figure 1. Antiproliferative activity of the most active analogue 8 against the highly
malignant mouse +SA mammary epithelial cells.
>20
5.6 0.56
12
>368
sents nearly twofold the activity of its parent
c-tocotrienol (2) (
Fig. 1). Compounds 4, 6, 8, 10, and 21 showed improved activities
against the ER-positive human breast cancer cells MCF7 compared
to the starting natural products 1–3 (Fig. 2). Tocotrienols 4, 8, and
2
ND
ND
329 12
10 with IC50 5–10 lM were the most active (Table 3). Compounds 4
and 10 are eightfold more active than their parent 1. Compound 8
was nearly threefold as active as its parent 2 (Fig. 2). Compounds 9,
13, 16 and 17 showed improved activities against the ER-negative
human breast cancer cells MDA-MB-231 (Fig. 3). The most active
assay kit against the highly metastatic MDA-MB-231 cells.21,22 This
assay employs a simplified Boyden Chamber design with a poly-
ethylene terephthalate membrane (8 lm). Detection of cell inva-
compound was 9 with an IC50 range of 5–10
l
M (Table 3).
sion is quantified using calcein–acetomethylester (calcein–
AM).21,22 Cells internalize calcein–AM and intracellular esterases
cleave the acetomethylester moiety to generate free calcein. Free
calcein fluoresces brightly, and this fluorescence may be used to
quantify the number of cells that have invaded across BME.21,22
A number of studies discussing the ability of tocotrienols to
inhibit tumor cells invasion and migration were reported.18 The
anti-migratory activity was usually correlated with the anti-
angiogenic activity, especially for d-tocotrienol.19 Recently, the
anti-invasive and anti-migratory activities of 2 against the hu-
man gastric adenocarcinoma, melanoma, and prostate cancer
cells were reported.20
The most active analogue 16, a-tocotrienol-6-benzylcarbamate, al-
lowed only 8% of MDA-MB-231 cells to migrate (Fig. 4), which sug-
gests that introduction of a bulky aromatic moiety at an optimum
distance from the 6-oxygen atom along with carbamate moiety
(benzyl carbamate) can enhance the anti-invasive activity of 1. This
The anti-invasive activities of tocotrienols 1–32 were measured
using a 96 well basement membrane extract (BME) cell invasion
Figure 2. Antiproliferative activity of tocotrienol analogues against MCF-7 cells.

760
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
Figure 3. Antiproliferative activity of tocotrienol analogues 9, 13, 16, and 17 against MDA-MB-231 cells.
was further supported by the improved activity of 17 (
c
-tocotri-
cases, indicator variables have been attributed to differentiate
racemates and active enantiomers.23
enol-6-benzylcarbamate) and 9 (d-tocotrienol-6-tosylcarbamate)
compared to their parent compounds 2 and 3 (Fig. 4). Meanwhile,
all redox-silent ethers 23–32 did not show any significant activity,
which proves the importance of the carbamate moiety and the
requirement of at least a HBD functionality at an optimized dis-
tance of C-6 oxygen for the activity.
Since there is no previous report for 3D QSAR model for tocot-
rienols, it is important to build a QSAR model to predict and opti-
mize the properties and activities of future untested tocotrienol
analogues and determine key structural requirements for en-
hanced activity. Accordingly, 3D QSAR models for tocotrienols
were designed using the most widely used computer-based meth-
odologies, CoMFA and CoMSIA.23
2.3. 3D QSAR studies
In order to widen the range of bioactivity, the previously pub-
lished tocopherol esters 33 and 34 were considered for their
known antiproliferative activities against MCF7.24 The actual and
predicted (by CoMFA and CoMSIA) pIC50 for the tocotrienol and
tocopherol series are listed in Tables 4 and 5. Partial least square
(PLS), the statistical method used in deriving the 3D QSAR models,
is a variation of principal component regression in which the origi-
nal variables are replaced by a small set of linear combinations.25
The generated latent variables were used for multivariate regres-
sion, maximizing the communality of predictor and response var-
iable blocks. Initial leave-one-out (LOO) cross-validated PLS
analyses were used to determine the optimum number of compo-
nents to be used in the final QSAR models.
Classical QSAR correlates biological activities of drugs with
physicochemical properties or indicator variables which encode
certain structural features.23 In addition to lipophilicity, polariz-
ability, and electronic properties, steric parameters are also fre-
quently used to describe different size of substituents. In some
2.3.1. CoMFA models
PLS analysis of the multifit alignment of all compounds in the
training set using CoMFA-STD resulted in a CoMFA QSAR model
with a good q2 value of 0.601 (Table 6). The PLS stdev coefficient
*
contour maps for the CoMFA model are shown in Figures 5 and 6.
Green regions indicate areas where steric bulk is predicted to en-
hance biological activity, whereas yellow contours indicate regions
where steric bulk is predicted to be detrimental to the activity. The
bulky benzene or p-toluenesulfonyl moieties are located in the
green contour area (Fig. 5), which justifies the enhanced activity
of these analogues. Blue-colored regions indicate areas where elec-
tropositive groups are predicted to enhance the biological activity
while red regions are areas where electronegative groups are
Figure 4. Anti-invasive activities of 5
and 23 against MDA-MB-231 cells using Cultrex™ assay kit.
lM dose of tocotrienol analogues 12, 13, 16,

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
761
Table 4
Actual, predicted, and residual of predictions of the test set by the CoMFA and CoMSIA models
Compound
Actual pIC50
Predicted CoMFA
pIC50
CoMFA residuals
Predicted CoMSIA
pIC50
CoMSIA residuals
4
2
25
32
24
1.651
2.051
2.690
2.690
2.872
1.529
1.934
2.682
2.752
2.961
0.121
0.117
0.008
1.711
2.001
2.631
2.688
2.994
ꢁ0.060
0.050
0.059
0.002
ꢁ0.122
ꢁ0.062
ꢁ0.089
predicted to favor activity (Fig. 6). The most active analogue, 8, pos-
sesses an electropositive charge around the NH-moiety near the
blue contour, and an electronegative charge around the sulfone
and benzene groups, near the red contour justifying improved
activity. Lower energy cutoff values of 25 and 20 kcal/mol were
also investigated but they all led to a decrease in the q2 value.
CH3
RO
CH3
CH3
CH3
CH3
H3C
CH3
O
R
R
R
O
H3N
HO
34
H
2.3.2. CoMSIA models
35
33
CoMSIA is less affected by changes in molecular alignment and
provides smoother and better interpretable contour maps as a re-
sult of employing Gaussian type distance dependence with the
molecular similarity indices it uses. In addition to the steric and
electrostatic fields, CoMSIA defines explicit hydrophobic and HBD
and HBA descriptor fields, which are not available in standard
CoMFA. CoMSIA also offers more fields of activity prediction. Sev-
eral trials were attempted using default SYBYL CoMSIA parameters
and all compounds in the training set (Table 5) with the field fit
molecular alignment resulted in a PLS model with a better q2 value
of 0.651, compared to a q2 value of 0.601 for CoMFA, and an r2 va-
lue of 0.993 for 10 PLS components (Table 6). Hydrophobic, HBD
and HBA, electrostatic, and steric descriptors were used. Only
hydrophobic, steric, and electrostatic descriptors were proven to
be better descriptors for tocotrienols SAR, Figures 7–9, because
whenever other descriptors were used a lower q2 value (0.305)
was obtained (Table 6).
H3N
O
O
R2
R1
R3
H
N
O
O
CH3
CH3
CH3
CH3
S
O
O
O
H3C
CH3
Compound
R1
R2
R3
36
37
36
39
40
OH
OH
OH
COOH
NH2
H
H
H
OH
H
H
OH
OH
H
H
The CoMSIA-derived hydrophobic fields (Fig. 7, yellow, hydro-
phobic group favored; white, hydrophobic disfavored) correlate
the biological activity and hydrophobicity features of tocotrienols.
Hydrophobic properties around benzene ring is more favorable but
not favorable around the NH or sulfone moiety in the most active
compound 8. CoMSIA also suggested the importance of hydropho-
bicity around the isoprenyl side chain of tocotrienols. Similar to
CoMFA, green regions indicate areas where steric bulk is predicted
to enhance the antiproliferative activity, whereas yellow contours
indicate regions where steric bulky groups are expected to reduce
the activity (Fig. 8). The green contour in case of compound 8 sug-
gests that bulky groups like p-toluene and sulfone will increase the
antiproliferative effect of tocotrienols. Blue and red contours indi-
cate areas where electropositive and electronegative groups,
respectively, are expected to enhance the activity ( Fig. 9). This
was illustrated in 8 where the sulfone moiety was located in the
red area while the NH-moiety was near to the blue one.
3. Conclusions
Tocotrienols are potent antiproliferative and anti-invasive natu-
ral products. To probe the chemical space around the C-6 phenol
group, several redox-silent carbamate and ether analogues were
prepared. It has been shown that a HBD at an optimized distance
of C-6 oxygen is required for the activity. 3D QSAR analyses were
performed for the first time to tocotrienols and their models pro-
vide a guide for future design of more potent analogues. 3D QSAR
models have good cross-validated q2 values, suggesting a good pre-
dictive ability. The 3D contour plots derived in this study may as-
sess the future development of tocotrienol-based antiproliferative
and anti-invasive entities.
4. Material and methods
Based on 3D QSAR results, new analogues (36–40) with en-
hanced activity have been designed to confirm the predictivity of
the model. Compound 8, the most active analogue, was considered
as a scaffold. Antiproliferation IC50 values of 36–40 against MCF-7
cell line were calculated using the CoMSIA model (Table 7). This
was based on the fact that introducing a hydrophilic group in p-po-
sition of benzenesulfonyl carbamate moiety will enhance the activ-
ity. Compound 36 with a 400, 500, and 600-trihydroxy substitution was
the most active. Compounds 39 with a 400-carboxy and 37 with 400
and 500-dihydroxy substitutions were the next most potent. This
clearly demonstrates the preference of hydroxyl and carboxyl
groups over the amino groups since compound 40 with C-500 amino
group was the least potent.
4.1. General experimental procedures
Optical rotations were measured on a Rudolph Research Analyt-
ical Autopol III polarimeter. IR spectra were recorded on a Varian 800
FT-IR spectrophotometer. The 1H and 13C NMR spectra were re-
corded in CDCl3, using TMS as an internal standard, on a JEOL Eclipse
NMRspectrometeroperatingat400 MHzfor1Hand100 MHzfor13C.
The HREIMS experiments were conducted at the University of Mich-
igan on a Micromass LCT spectrometer. Analytical HPLC analyses
were performed on a DIONEXÒ Summit II instrument on a Phenom-
enex Luna 250 ꢀ 4.6 mm, RP-C18 column using MeOH–H2O (9.5:0.5)

762
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
Table 5
Actual, predicted, and residual of predictions of the training set by the CoMFA and CoMSIA models
Compound
Actual pIC50
Predicted CoMFA
pIC50
CoMFA residuals
Predicted CoMSIA
pIC50
CoMSIA residuals
8
10
21
6
7
33
3
19
13
22
11
17
14
28
20
27
26
29
30
31
12
1
1.380
1.707
1.748
1.780
1.786
2.081
2.190
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.690
2.693
2.795
2.870
3.003
3.040
3.517
3.565
1.383
1.715
1.725
1.733
1.794
2.081
2.122
2.73
2.711
2.692
2.719
2.673
2.732
2.742
2.797
2.687
2.695
2.693
2.74
2.648
2.718
2.724
2.904
2.977
2.929
3.442
3.571
ꢁ0.003
ꢁ0.008
0.023
1.383
1.695
1.725
1.773
1.82
ꢁ0.003
0.012
0.023
0.007
ꢁ0.034
0
0.047
ꢁ0.008
0
2.081
2.122
2.698
2.711
2.692
2.719
2.673
2.732
2.742
2.797
2.687
2.695
2.693
2.74
2.648
2.693
2.724
2.904
3
2.929
3.442
3.571
0.068
0.068
ꢁ0.008
ꢁ0.021
ꢁ0.002
ꢁ0.029
0.017
ꢁ0.042
ꢁ0.052
ꢁ0.107
0.003
ꢁ0.005
ꢁ0.003
ꢁ0.05
0.042
0
0.071
ꢁ0.034
0.003
0.111
0.075
ꢁ0.006
ꢁ0.04
ꢁ0.021
ꢁ0.002
ꢁ0.029
0.017
ꢁ0.042
ꢁ0.052
ꢁ0.107
0.003
ꢁ0.005
ꢁ0.003
ꢁ0.05
0.042
ꢁ0.025
0.071
9
16
5
35
34
ꢁ0.034
0.026
0.111
0.075
ꢁ0.006
Table 6
PLS statistics of CoMFA and CoMSIA 3D QSAR (multifit alignment)
isocratic elution with UV detection set at 295 nm to verify the purity
of each compound. A purity of >95% has been established for all com-
pounds except 11 and 12, where 90% purity has been established.
PLS statistics
CoMFA model
CoMSIA model
TLC analysis was carried on precoated Si Gel 60 F254 500 lm TLC
plates (EMD Chemicals), using n-hexane–EtOAc (8:2) or CHCl3–
MeOH (9:1) as developing systems. For column chromatography,
Si Gel 60 (Natland, 63–200 lm), n-hexane–EtOAc (9.5:0.5), and n-
hexane–CHCl3 (6:4) systems were used.
4.2. Chemical reactions
q2
0.601
0.989
0.022
272.061
0.000
7
0.651
0.993
0.039
251.359
0.000
10
r2 Conventional
Std. error
F value
P value
PLS components
Field contribution
Steric
Electrostatic
Hydrophobic
0.540
0.460
0.209
0.458
0.333
(A) Carbamoylation of 1–3.26 To solutions of 1–3 in toluene
(2 mL), different isocyanates and 10 lL of triethylamine (Et3N)
q2, cross-validated correlation coefficient; r2, correlation coefficient; F, F-test value.
Figure 5. CoMFA steric contours plots; green contours indicate regions where bulky groups increase activity, whereas yellow contours indicate regions where bulky groups
decrease activity.

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
763
*
Figure 6. CoMFA stdev coeff electrostatic contour plots; Blue contours indicate regions where positive groups increase activity, whereas red contours indicate regions
where negative charge increases activity.
*
Figure 7. CoMSIA stdev coeff hydrophobic contour plots; yellow contours indicate regions where hydrophobic (lipophilic) groups increase activity, whereas white contours
indicate regions where hydrophobic (lipophilic) groups decrease activity.
Figure 8. CoMSIA steric contours plots; green contours indicate regions where bulky groups increase activity, whereas yellow contours indicate regions where bulky groups
decrease activity.

764
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
*
Figure 9. CoMSIA stdev coeff electrostatic contour plots; blue contours indicate regions where positive groups increase activity, whereas red contours indicate regions
where negative charge increases activity.
1358 cmꢁ1 1H and 13C NMR, see Tables S1 and S2; HRESIMS m/z
;
Table 7
616.3068 [M+Na]+ (calcd for C35H47NO5SNa, 616.3073).
Predicted IC50 values of new tocotrienol analogues designed according to multifit-
aligned 3D QSAR studiesa
4.2.3. (R)-2,8-Dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
Compound
Predicted IC50/lM
30,70,110-trienyl)-chroman-6-yl phenylsulfonylcarbamate (6)
36
37
36
39
40
2.03
1.82
1.30
1.70
2.31
A reaction of 50 mg of 3 with 11.70 lL benzenesulfonyl isocya-
nate was carried out to give 6, 40% yield: yellow oil, ½a D25
ꢂ
+30 (c 0.1,
CHCl3); IR mmax (CHCL3) 2927, 1760, 1450, 1355, 1159, 1089,
880 cmꢁ1 1H and 13C NMR, see Tables S1 and S2; HRESIMS m/z
;
a
602.2917 [M+Na]+(calcd for C34H45NO5SNa, 602.2916).
Predicted activity presents the CoMSIA predictions.
4.2.4. (R)-2,5,7,8-Tetramethyl-2-((30E,70E)-40,80,120-trime
thyltrideca-30,70,110-trienyl)chroman-6-yl tosylcarbamate (7)
were added. Each solution was separately stirred and refluxed for
3 h. Water (10 mL) then added and each reaction mixture was ex-
tracted with EtOAc (3 ꢀ 10 mL). Each EtOAc extract was dried over
anhydrous Na2SO4 and concentrated under vacuum to produce
after further column chromatography compounds 4–22.
A reaction of 50 mg of 1 with 17.98 lL p-toluenesulfonyl isocy-
anate was carried out give 7, 50% yield: yellow oil, ½a D25
ꢂ
+28 (c 0.35,
CHCl3); IR mmax (CHCL3) 2928, 1757, 1355, 1152, 1088, 890 cmꢁ1
;
1H and 13C NMR, see Tables S1 and S2; HRESIMS m/z 644.3369
(B) Etherification of 1–3.27 Solutions of 1–3 (0.15 mmol) in dry
THF were added drop-wise to a cooled (5 °C) oil-free suspension of
NaH (0.16 mmol) in 5 mL dry THF. Different alkyl or aryl bromides
or chlorides(0.16 mmol) were added drop-wise tothe mixture. After
hydrogen evolution ceased, the solution was warmed to 23 °C and
stirred for 2 h. Each solution was then poured into 1 N NaOH and
the mixture was extracted with CHCl3 (3 ꢀ 10 mL). Each organic
layer was then washed with brine (1 ꢀ 10 mL) and dried over anhy-
drous Na2SO4. Crude mixtures were then purified on Si Gel 60 using
isocratic CHCl3–methanol (9.9:0.1) system. Evaporation of fractions
containing target compounds afforded 23–32 in 70–80% yields.
[M+Na]+(calcd for C37H51NO5SNa, 644.3386).
4.2.5. (R)-2,7,8-Trimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl)-chroman-6-yl tosylcarbamate (8)
A reaction of 100 mg of 2 with 37.2 lL p-toluenesulfonyl isocy-
anate was carried out to get 8, 60% yield: yellow oil, ½a D25
ꢁ71.6 (c
ꢂ
0.06, CHCl3); IR mmax (CHCl3) 2935, 1758, 1458, 1407, 1512, 1152,
1088 cmꢁ1 1H and 13C NMR, see Tables 1 and 2; HRESIMS m/z
;
630.3244 [M+Na]+ (calcd for C36H49NO5SNa, 630.3229).
4.2.6. (R)-2,8-Dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl)-chroman-6-yl tosylcarbamate (9)
4.2.1. (R)-2,5,7,8-Tetramethyl-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl)chroman-6-yl
phenylsulfonylcarbamate (4)
A reaction of 100 mg of 3 with 38.5 lL p-toluenesulfonyl isocy-
anate was carried out to get 9, 55% yield: yellow oil, ½a D25
ꢁ15.0 (c
ꢂ
0.7, CHCl3); IR mmax (CHCl3) 2930, 1753, 1350, 1158, 1089,
909 cmꢁ1 1H and 13C NMR, see Tables 1 and 2; HRESIMS m/z
;
A reaction of 20 mg of 1 with 6.30 lL of benzenesulfonyl isocy-
anate was carried out give 4, 40% yield: yellow oil, ½a D25
ꢂ
+18.6 (c
616.3070 [M+Na]+ (calcd for C35H47NO5SNa, 616.3073).
0.07, CHCl3); IR mmax (CHCl3) 2929, 1758, 1450, 1405, 1156, 1088,
909 cmꢁ1 1H and 13C NMR, see Tables 1 and 2; HRESIMS m/z
;
4.2.7. (R)-2,5,7,8-Tetramethyl-2-((30E,70E )-40,80,120-
trimethyltrideca-30,70,110-trienyl)-chroman-6-yl-400
chlorophenylsulfonylcarbamate (10)
630.3218 [M+Na]+ (calcd for C36H49NO5SNa, 630.3229).
4.2.2. (R)-2,7,8-Trimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
A reaction of 100 mg of 1 with 35.17
nyl isocyanate was carried out to get 10 in a 45% yield: yellow oil,
ꢁ8.3 (c 0.6, CHCl3); IR mmax (CHCl3) 2928, 2360, 1757, 1458,
1154, 988 cmꢁ1 1H and 13C NMR(Tables 1 and 2); HRESIMS m/z
664.2840 [M+Na]+ (calcd for C36H48ClNO5SNa, 664.2835).
lL 4-chlorobenzenesulfo-
30,70,110-trienyl)chroman-6-yl phenylsulfonylcarbamate (5)
a 2D5
Reaction of 100 mg of 2 with 32.63 lL of benzenesulfonyl isocy-
½ ꢂ
anate was proceeded to give 5, 40% yields: yellow oil, ½a D25
ꢂ
+16 (c
;
0.1, CHCl3); IR mmax (CHCL3) 2928, 1765, 1602, 1450, 1409,

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
765
4.2.8. (R)-2,7,8-Trimethyl-2-((30E,70E )-40,80,120-trimethyltrideca-
4.2.16. (R)-2,5,7,8-Tetramethyl-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl)chroman-6-yl
benzoylcarbamate (19)
30,70,110-trienyl)-chroman-6-yl-400-chlorophenylsulfonylcarbamate
(11)
A reaction of 100 mg of 2 with 36.4
l
L 4-chlorobenzenesulfonyl
A reaction of 100 mg of 1 with 31.2 lL phenylisocyanatofor-
isocyanate was carried out to get 11, 40% yield: yellow oil, ½a D25
ꢂ
+5
mate was carried out to get 19, 60% yield: yellow oil, ½a D25
ꢁ4 (c
ꢂ
(c 1, CHCl3); IR mmax (CHCl3) 2931, 2360, 1761, 1154, 1091,
1.6, CHCl3); IR mmax (CHCl3) 1823, 1748, 1478, 1143, 1085,
893 cmꢁ1
;
1H and 13C NMR, see Tables S3 and S4; HRESIMS m/z
1005 cmꢁ1 1H and 13C NMR, see Tables S5 and S6; HRESIMS m/z
;
650.2668 [M+Na]+ (calcd for C35H46ClNO5SNa, 650.2683).
610.3506 [M+Na]+ (calcd for C37H49NO5Na, 610.3508).
4.2.9. (R)-2,8-Dimethyl-2-((30E,70E )-40,80,120-trimethyltrideca-30,
70,110-trienyl)-chroman-6-yl-400-chlorophenylsulfonylcarbamate
(12)
4.2.17. (R)-2,7,8-Trimethyl-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl)-chroman-6-yl
benzoylcarbamate (20)
A reaction of 100 mg of 3 with 37.60
l
L 4-chlorobenzenesulfo-
A reaction of 50 mg of 2 with 32.2 lL phenylisocyanatoformate
nyl isocyanate was carried out to get 12 in a 50% yield: yellow oil,
was carried out to get 20, 70% yield. Yellow oil, ½a D25
ꢂ
+3.75 (c 0.16,
½
a 2D5
ꢂ
ꢁ9.0 (c 0.1, CHCl3); IR mmax (CHCl3) 2929, 2360, 1758, 1602,
CHCl3); IR mmax (CHCl3) 2935, 1822, 1740, 1487, 1141, 1085 cmꢁ1
;
1457, 1360, 978 cmꢁ1; 1H and 13C NMR, see Tables S3 and S4; HRE-
1H and 13C NMR, see Tables S7 and S8; HRESIMS m/z
SIMS m/z 636.2529 [M+Na]+ (calcd for C34H44ClNO5SNa, 636.2526).
596.3333[M+Na]+ (calcd for C36H47NO5Na, 596.3352).
4.2.10. (R)-2,5,7,8-Tetramethyl-2-((30E,70E)-40,80,120-trimethyl
4.2.18. (R)-2,8-Dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
trideca-30,70,110-trienyl)chroman-6-yl phenylcarbamate (13)
30,70,110-trienyl)-chroman-6-yl benzoylcarbamate (21)
A reaction of 20 mg of 1 with 4.52
l
L phenylisocyanate was car-
A reaction of 100 mg of 3 with 33.4 lL phenyl isocyanatofor-
ried out to get 13 in a 50% yield: yellow oil, ½a D25
ꢂ
+11 (c 0.2, CHCl3);
mate was carried out to get 21, 65% yield: yellow oil, ½a D25
ꢂ
+4 (c
IR mmax (CHCl3) 2929, 1731, 1602, 1454, 1104, cmꢁ1
;
1H and 13C
0.5, CHCl3); IR mmax (CHCl3) 2930, 1824, 1471, 1132, 1056 cmꢁ1
;
NMR, see Tables S3 and S4; HRESIMS m/z 566.3610 [M+Na]+ (calcd
1H and 13C NMR, see Tables S7 and S8; HRESIMS m/z
for C36H49NO3Na, 566.3610).
582.3193[M+Na]+ (calcd for C35H45NO5Na, 582.319).
4.2.11. (R)-2,7,8-Trimethyl-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl)-chroman-6-yl
phenylcarbamate (14)
4.2.19. (R)-2,8-Dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl) chroman-6-yl naphthalen-100-ylcarbamate (22)
A reaction of 100 mg of 3 with 37.8 lL 1-naphthylisocyanate
A reaction of 40 mg of 2 with 12.00
l
L phenylisocyanate was
was carried out to get 22, 70% yield: yellow oil, ½a D25
ꢂ
+2 (c 1.9,
carried out to get 14, 50% yield: yellow oil, ½a D25
ꢂ
+55 (c 0.1, CHCl3);
CHCl3); IR mmax (CHCl3) 2928, 1743, 1472, 1348, 1176, 1140,
IR
m
max (CHCl3) 3019, 2400, 1522, 1424, 929 cmꢁ1; 1H and 13C NMR,
1000 cmꢁ1 1H and 13C NMR, see Tables S7 and S8; HRESIMS m/z
;
see Tables S3 and S4; HRESIMS m/z 552.3521[M+Na]+ (calcd for
588.3460[M+Na]+ (calcd for C38H47NO3Na, 588.3454).
C35H47NO3Na, 552.3520).
4.2.20. (R )-2,7,8-Trimethyl-6-(400-methylpent-300-enyloxy)-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman (24)
4.2.12. (R)-2,8-Dimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl)-chroman-6-yl phenylcarbamate (15)
About 61.5 mg of 2 in 214
l
L 5-bromo-2-methylpent-2-ene gave
ꢁ30 (c 0.35, CHCl3); IR mmax (CHCl3)
24, 70% yield: yellow oil, ½a D25
ꢂ
A reaction of 100 mg of 3 with 27.40
l
L phenylisocyanate to get
15, 50% yield: yellow oil, ½a D25
ꢂ
+1.8 (c 3.6, CHCl3); IR mmax (CHCl3)
2931, 1452, 1379, 1098, 986 cm1; 1H and 13C NMR, see Tables S7
2929, 2854, 1744, 1604, 1442, 1139, 1021 cmꢁ1 1H and 13C
;
and S8; HREIMS m/z 492.3955 [M+] (calcd for C34H52O2, 492.3967).
NMR, see Tables S5 and S6; HRESIMS m/z 538.3271[M+Na]+ (calcd
4.2.21. (R)-6-(Isopentyloxy)-2,7,8-trimethyl-2-((30E,70E)-
for C34H45NO3Na, 538.3297).
40,80,120-trimethyltrideca-30,70,110-trienyl)chroman (25)
4.2.13. 13(R)-2,5,7,8-Tetramethyl-2-((30E,70E)-40,80,120-trimethy
Reaction of 61.5 mg of 2 with 191
l
L 1-bromo-3-methylbutane
ltrideca-30,70,110-trienyl)-chroman-6-yl benzylcarbamate (16)
gave 25, 80% yield: yellow oil, ½a D25
ꢂ
+2.2 (c 0.85, CHCl3); IR mmax
(CHCl3) 2931, 2870, 1466, 1379, 1098, 983 cmꢁ1 1H and 13C NMR,
;
A reaction of 40 mg of 1 with 11.63 lL benzylisocyanate was
carried out to get 16, 40% yield: yellow oil, ½a D25
ꢂ
+43 (c 0.09, CHCl3);
see Tables S7 and S8; HREIMS m/z 480.3952 [M+] (calcd for
IR mmax (CHCl3) 2927, 1745, 1603, 1443, 1097, cmꢁ1 1H and 13C
;
C33H52O2, 480.3967).
NMR, see Tables 1 and 2; HRESIMS m/z 580.3766 [M+Na]+ (calcd
for C37H51NO3Na, 580.3767).
4.2.22. (R)-2,5,7,8-Tetramethyl-6-((200E,600E)-300,700,1100-trimethyl-
dodeca-200,600,1000-trienyloxy)-2-((30E,70E)-40,80,120-trimethyltrid-
eca-30,70,110-trienyl)chroman (26)
4.2.14. (R)-2,7,8-Trimethyl-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl)-chroman-6-yl benzylcarbamate (17)
Reaction of
ethyldodeca-2,6,10-triene (433
ꢁ8 (c 0.5, CHCl3); IR mmax (CHCl3) 2930, 2856, 1455, 1378,
1
(64 mg) with (2E,6E)-1-bromo-3,7,11-trim-
A reaction of 50 mg of 2 with16.31 lL benzylsulfonyl isocyanate
l
L) gave 26, 85% yield: yellow oil,
was carried out to get 17, 40% yield: yellow oil, ½a D25
ꢂ
+6.0 (c 0.1,
½ ꢂ
a 2D5
CHCl3); IR mmax (CHCl3) 2929, 1744, 1603, 1443, cmꢁ1
;
1H and 13C
1086, 976 cmꢁ1 1H and 13C NMR, see Tables S9 and S10; HREIMS
;
NMR, see Table S5 and S6; HRESIMS m/z 566.3610 [M+Na]+ (calcd
m/z 628.5276 [M+] (calcd for C44H68O2, 628.5219).
for C36H49NO3Na, 566.3610).
4.2.23. (R)-2,7,8-Trimethyl-6-((200E,600E)-300,700,1100-trimethyldo-
deca-200,600,100-trienyloxy)-2-((30E,70E)-40,80,120-trimethyltrideca-
30,70,110-trienyl)chroman (27)
4.2.15. (R)-2,8-Dimethyl-2-((30E,70E) 40,80,120-trimethyltrideca-
30,70,110-trienyl)chroman-6-yl benzylcarbamate (18)
A reaction of 100 mg of 3 with 31.0 lL benzylsulfonyl isocya-
nate was carried out to get 18, 40% yield: yellow oil, ½a D25
ꢂ
+3.0 (c
Reaction of
2
(61.5 mg) with (2E,6E)-1-bromo-3,7,11-trim-
L) afforded 27, 80% yield: yellow
+1.4 (c 1.07, CHCl3); IR mmax (CHCl3) 2931, 2856, 2360,
1H and 13C NMR, see Tables
3.55, CHCl3); IR mmax (CHCl3) 2930, 1733, 1471, 1154, 911 cmꢁ1
;
ethyldodeca-2,6,10-triene (433
l
oil, ½a 2D5
ꢂ
1H and 13C NMR, see Tables S5 and S6; HRESIMS m/z 552.3444
1453, 1397, 1379, 1097, 986 cmꢁ1
;
[M+Na]+ (calcd for C35H47NO3Na, 552.3454).

766
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
S9 and S10; HREIMS m/z 614.5076 [M+] (calcd for C43H66O2,
614.5063).
After a four-day culture period, viable +SA cell number was deter-
mined by the MTT colorimetric assay as described previously.23
The cells growth of MCF7 and MDA-MB-231 cell lines was mea-
sured using MTT kit (TACS™, TrevigenÒ, Inc.). After passing the
cells for 3–4 times, growing cells were incubated in a 96 well plate
at a density of 8 ꢀ 103 cells per well, and allowed to attach for 24 h.
4.2.24. (R)-2,8-Dimethyl-6-((200E,600E)-300,700,1100-
trimethyldodeca-200,600,1000-trienyloxy)-2-((30E,70E)-40,80,120-
trimethyltrideca-30,70,110-trienyl)chroman (28)
Reaction of
3
(59 mg) with (2E,6E)-1-bromo-3,7,11-trim-
L) afforded 28, 80% yield: yellow
+3.2 (c 1.55, CHCl3); IR mmax (CHCl3) 2933, 2859, 2340,
Complete growth medium was then replaced with 100
of serum free medium (GIBCO-Invitrogen, NY) containing various
doses (20, 10, 5, 2 M) of the specific compound and culture will
continued at 37 °C under 5% CO2. After 96 h, the incubated cells
were treated with MTT solution (10 L/well) at 37 °C for 4 h. The
color reaction stopped by the addition of solubilization/stop solu-
tion (100 L/well), and the incubation at 37 °C continued to com-
lL of RPMI
ethyldodeca-2,6,10-triene (433
l
oil, ½a 2D5
ꢂ
l
1457, 1397, 1379, 986 cmꢁ1 1H and 13C NMR, see Tables S9 and
;
S10; HREIMS m/z 600.4902 [M+] (calcd for C42H64O2, 600.4906).
l
4.2.25. (R)-6-(400-Methoxybenzyloxy)-2,5,7,8-tetramethyl-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman
(29)
l
pletely dissolve the formazan product. Absorbance of the samples
was determined at 570 nm with an ELISA plate reader (BioTek,
VT). The number of cells per well was calculated against a standard
curve prepared by plating various concentration of cells, as deter-
mined by hemocytometer, at the start of each experiment. IC50 for
the compound was calculated using nonlinear regression (curve
fit) of log concentration versus number of cells/well implemented
in GraphPad Prism 5.0.
Growth curves were determined to ensure that cells used in
experiments were within the exponential growth phase. Cell pro-
liferation was assessed by monitoring the conversion of MTT to for-
mazan. The reduction of MTT is catalyzed by mitochondrial
dehydrogenase enzymes and is therefore a measure for cell viabil-
Reaction of 1 (64 mg) with 1-(bromomethyl)-3-methoxyben-
zene (462
l
L) gave 29, 70% yield: yellow oil, ½a D25
ꢂ
+38 (c 0.26,
CHCl3); IR mmax (CHCl3) 2938, 2839, 2359, 1457, 1370, 1091,
987 cmꢁ1 1H and 13C NMR, see Tables S11 and S12; HREIMS m/z
;
544.3916 [M+] (calcd for C37H52O3, 544.3916).
4.2.26. (R)-6-(400-Methoxybenzyloxy)-2,7,8-trimethyl-2-
((30E,70E)-40,80,120-trimethyl-trideca-30,70,110-trienyl)chroman
(30)
Reaction of 2 (61.5 mg) with 1-(bromomethyl)-3-methoxyben-
zene (462
l
L) afforded 30, 75% yield: yellow oil, ½a D25
ꢂ
+2.4 (c 1.19,
CHCl3); IR mmax (CHCl3) 2930, 2855, 2360, 1717, 1604, 1491, 1457,
ity. Briefly, cells (100 lL/well) were seeded at seeding densities of
1378, 1270, 1099, 909 cmꢁ1
;
1H and 13C NMR, see Tables S11 and
1 ꢀ 105, 1 ꢀ 104 or 1 ꢀ 103 cells/ml into 96 well microtitre plates
S12; HRESIMS m/z 553.3651 [M+Na]+ (calcd for C36H50O3Na,
and allowed to adhere for 24 h. Cell viability was assessed on a dai-
553.3658).
ly basis by adding 10
lL of filter sterilized MTT (5 mg/mL in PBS) to
a each well. Following a 4 h incubation period with MTT, 100
l
L of
4.2.27. (R )-6-(400,600-Dimethoxybenzyloxy)-2,5,7,8-tetramethyl-
2-((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl)chroman
(31)
solubilizing agent was added to each well, and the blue formazan
crystals trapped in cells dissolved at 37 °C for 2–4 h or at room
temperature for over night in dark place. The absorbance at
550 nm was measured with a plate reader.
Reaction of 1 (64 mg) with 1-(bromomethyl)-3,5-dimethoxy-
benzene (37 mg) afforded 31, 80% yield: yellow oil, ½a D25
ꢁ14.8 (c
ꢂ
0.27, CHCl3); IR mmax (CHCl3) 2931, 2856, 1599, 1458, 1370, 1155,
1089, 993 cmꢁ1; 1H and 13C NMR, see Tables S11 and S12; HREIMS
m/z 574.4015 [M+] (calcd for C38H54O4, 574.4022).
4.5. Cell invasion assay
Anti-invasive activities were measured using Trevigen’s Cult-
rexÒ Cell Invasion Assay.29,30 About 50
l
L of (BME 1X) coat was
added per well. After incubation for overnight at 37 °C in a 5%
CO2, 50,000/50 L of MDA-MB-231 cells in serum free RPMI med-
ium were added per well to top chamber containing the tested
compound at treatment concentration (5 M). About 150 L of
4.2.28. (R )-6-(400,600-Dimethoxybenzyloxy)-2,7,8-trimethyl-2-
((30E,70E)-40,80,120-trimethyltrideca-30,70,110-trienyl) chroman
(32)
l
Reaction of 2 (61.5 mg) with 1-(bromomethyl)-3,5-dimethoxy-
l
l
benzene (37 mg) afforded 32, 85% yield: yellow oil, ½a D25
ꢂ
1.03 (c
RPMI medium was added to lower chamber containing 10% FBS
(chemoattractant) and penicillin/streptomycin and using fibronec-
tin (1 lL/mL) and N-formyl-met-leu-phe (10 nM) as chemoattrac-
tants. Cells were allowed to migrate to lower chamber at 37 °C in
CO2 incubator. After 24 h, top and bottom chambers were aspirated
and washed with wash buffer supplemented with the kit. About
1.65, CHCl3); IR mmax (CHCl3) 2939, 2841, 1599, 1461, 1157, 1099,
1068, 993, 838 cmꢁ1 1H and 13C NMR, see Tables S13; HRESIMS
;
m/z 583.3753 [M+Na]+ (calcd for C37H52O4Na, 583.3763).
4.3. Cell culture
100 lL of cell dissociation/calcein–AM solution was added to the
Breast cancer cell lines, MCF7 and MDA-MB-231 were pur-
chased from ATCC (Manassas, VA). The cell lines were grown in
10% fetal bovine serum (FBS) and RPMI 1640 (GIBCO-Invitrogen,
bottom chamber and incubated at 37 °C in CO2 incubator for 1 h.
The cells internalize calcein–AM, and the intracellular esterases
cleaved the AM moiety to generate free calcein. Fluorescence of
the samples was determined at 485 nm excitation, 520 nm emis-
sion using plate reader (BioTek Synergy 2, VT). The numbers of
cells that have invaded through the BME coat were calculated
using a standard curve.
NY) supplemented with 2 mmol/L glutamine, 100
lg/mL penicillin
G, and 100 g/mL streptomycin, at 37 °C under 5% CO2.
l
4.4. Proliferation assay
The antiproliferative effects of tocotrienol analogues were
tested in culture on the highly malignant +SA mouse mammary
epithelial cell line maintained on serum free media and containing
4.6. Chemicals and reagents
All isocyanates reagents have been purchased from Sigma–Al-
drich. (TRF) 50 g (Palm TRF 70%, low in tocopherol from First Tech
International Ltd, Hong Kong) was fractionated using Si Gel 60 VLC
with n-hexane–ethyl acetate (gradient elution) as a mobile phase.
10 ng/mL EGF and 10 lg/mL insulin as mitogens, as described pre-
viously in detail.28 Cells were plated at a density of 5 ꢀ 104 cells/
well (6 wells/group) in 24-well culture plates and fed media con-
taining various concentrations (0.01–1000 lM) of each compound.

A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
767
4.7. CoMFA and CoMSIA
idation method might lead to high q2 values, which do not neces-
sarily reflect a general predictiveness of the models. Therefore,
we have performed cross-validation using two groups (leave-
half-out) in CoMFA study. In this method 50% of the compounds
were randomly selected and a model was generated, which was
then used to predict the activity of the remaining 50% of com-
pounds. The random formation of the cross-validation groups
may have an effect on the results; hence cross-validation was per-
formed 100 times for all the analysis.33,34
Three-dimensional structure building was performed using the
SYBYL program package, version 8.1, installed on DELL desktop
workstations equipped with a 1.8 GHz IntelÒ XeonÒ processor run-
ning the Red Hat Enterprise Linux (version 5.0) operating system.
For the calculating of charges, the Gasteiger-Hückel method was
used as implemented in SYBYL
CoMSIA analysis was performed using the QSAR module in SYBYL
8.1. The five similarity indices in CoMSIA (steric (S), electrostatic
(E), hydrophobic (H), H-bond donor (D), and H-bond acceptor (A)
descriptors) were calculated using the probe atom Csp3 + with a ra-
dius of 1 Å and a +1.0 charge placed at the lattice points of the
same region of grid as it was used for the CoMFA calculations.
CoMSIA similarity indices (AF) for a molecule j with atom i at a grid
31
.
The conventional CoMFA and CoMSIA descriptors derived above
were used as explanatory variables, and pIC50 (ꢁlog IC50) values
(Tables 4 and 5) were used as the target variable in PLS regression
analyses to derive 3D QSAR models using the implementation in
the SYBYL package. The predictive value of the models was
evaluated by LOO cross validation with SAMPLS. The cross-vali-
dated coefficient, q2, was calculated using the equation:q2 =
point q are calculated by the equation:
P
P
P
a
riq2
AF;kqðjÞ ¼ ꢁ xprobe; kxik eꢁ
, where k represents the follow-
1 ꢁ (Ypred ꢁ Yactual)2/ (Yactual ꢁ Ymean)2, where Ypred, Yactual, and
ing physicochemical properties: steric, electrostatic, hydrophobic,
HBD, and HBA. A Gaussian type distance dependence was used be-
tween the grid point q and each atom i of the molecule. A default
value of 0.3 was used as the attenuation factor (R). Here, steric
indices are related to the third power of the atomic radii, electro-
static descriptors are derived from atomic partial charges, hydro-
phobic fields are derived from atom-based parameters.32
Ymean are predicted, actual, and mean values of the target property
P
2
(pIC50), respectively.
(Ypred ꢁ Yactual
)
is the predictive sum of
squares (PRESS). The number of components giving the lowest
PRESS value or the optimal number of components (ONC) was used
to generate the final PLS regression models. The conventional cor-
relation coefficient r2 and its standard error, s, were subsequently
computed for the final PLS models.
In order to have sufficient number of compounds with activity
variation for PLS analysis, the previously published tocopherol ana-
logues 33–35 have been added to a database set of 29 tocotrienol
derivatives.35,36 This dataset of total number 32 tocotrienol and
tocopherol analogues were used for both CoMFA and CoMSIA stud-
ies based on the antiproliferative effect against MCF-7 breast can-
cer cell line using MTT-assay. The pIC50 (ꢁlog IC50), which is
responsible for 50% inhibition of cells growth, was used in these
studies. The 32 compounds were divided to training set comprising
27 compounds (Table 5), 84% of database set, and test set, five com-
pounds, 16% of the set (Table 4).
For deriving the CoMFA and CoMSIA descriptor fields, the mol-
ecules from the training set were placed in a three dimensional cu-
bic lattice with a spacing of 2 Å and extending 4 Å units beyond the
aligned molecules in all directions. CoMFA descriptors were calcu-
lated using an sp3 carbon probe atom with a van der Waals radius
of 1.52 Å and a charge of +1.0 with a distance-dependent dielectric
to generate steric (Lennard-Jones 6–12 potential) field energies
and electrostatic (Columbic potential) fields with a distance-
4.8. Partial least-squares (PLS) analysis
The statistical method used in deriving the 3D QSAR models, is a
variation of principal component regression33 in which the original
variables are replaced by a small set of linear combinations thereof.
The latent variables so generated are used for multivariate regres-
sion, maximizing the communality of predictor and response var-
iable blocks. PLS has several attractive features, such as: (i) the
ability to handle multivariate regression analysis in cases where
the number of independent variables is greater than the number
of compounds as found in CoMFA and CoMSIA 3D QSAR analyses;
(ii) the ability to perform well even when inter-descriptor correla-
tions exist that would pose a problem for traditional multiple lin-
ear regression; (iii) the reduction of the risk of chance
correlations.33,34
Initial the leave-one-out (LOO) cross-validated PLS analyses
were used to determine the optimum number of components to
be used in the final 3D QSAR models. The leave-one-out cross-val-
Figure 10. Superposition of tocotrienols training and test sets using field fit alignment.

768
A. Y. Elnagar et al. / Bioorg. Med. Chem. 18 (2010) 755–768
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.11.054.
References and notes
1. Sen, C. K.; Khanna, S.; Roy, S.; Packer, L. J. Biol Chem. 2000, 275, 13049.
2. Crowell, P. L.; Chang, R. R.; Ren, Z. B.; Elson, C. E.; Gould, M. N. J. Biol. Chem.
1991, 266, 17679.
3. Elson, C. E.; Yu, S. G. J. Nutr. 1994, 124, 607.
4. McIntyre, B.; Briski, K. P.; Gapor, A.; Sylvester, P. W. Proc. Soc. Exp. Biol. Med.
2000, 224, 292.
5. Sporn, M. B.; Suh, N. Nat. Rev. Cancer 2002, 2, 537.
6. Shukla, S.; Gupta, S. Nutr. Cancer 2005, 53, 18.
7. Nesaretnam, K.; Guthrie, N.; Chambers, A. F.; Carroll, K. K. Lipids 1995, 30, 1139.
8. Qureshi, A. A.; Sami, S. A.; Salser, W. A.; Khan, F. A. Atherosclerosis 2002, 161,
199.
9. Khanna, S.; Roy, S.; Ryu, H.; Bahadduri, P.; Swaan, P. W.; Ratan, R. R. J. Biol.
Chem. 2003, 278, 43508.
10. Akaho, N.; Takata, J.; Fukushima, T.; Matsunaga, K.; Hattori, A.; Hidaka, R.;
Fukui, K.; Yoshida, M.; Fujioka, T. R.; Karube, Y.; Imai, K. Drug Metab. Dispos.
2007, 35, 1502.
11. Takata, J.; Hidaka, R.; Yamasaki, A.; Hattori, A.; Fukushima, T.; Tanabe, M.;
Matsunaga, K.; Karube, Y.; Imai, K. J. Lipid Res 2002, 43, 2196.
12. Mazzini, F.; Netscher, T.; Salvadori, P. Eur. J. Org. Chem. 2009, 13, 2063.
13. Tomic-Vatic, A.; EyTina, J.; Chapman, J.; Mahdavian, E.; Neuzil, J.; Salvatore, B.
Int. J. Cancer 2005, 117, 188.
14. Kashiwagi, K.; Harada, K.; Yano, Y.; Kumadaki, I.; Hagiwara, K.; Takebayashi, J.;
Kido, W.; Virgona, N.; Yano, T. Biochem. Biophys. Res. Commun. 2008, 365, 875.
15. Ali, H.; Nazzal, S. J. Pharm. Biomed. Anal. 2009, 49, 950.
16. Ohnmacht, S.; West, R.; Simionescu, R.; Atkinson, J. Magn. Reson. Chem. 2008,
46, 287.
17. Shah, S. J.; Sylvester, P. W. Exp. Biol. Med. 2005, 230, 235.
18. Liu, H.-K.; Wang, Q.; Li, Y.; Sun, W.-G.; Liu, J.-R.; Yang, Y.-M.; Xu, W.-L.; Sun, X.-
R.; Chen, B.-Q., J. Nutr. Biochem., in press. doi:10.1016/j.jnutbio.2008.11.004).
19. Chang, P.; Yap, W.; Lee, W. D.; Ling, M. T.; Wong, Y. C.; Yap, Y. Nutr. Cancer
2009, 61, 357.
20. Yap, W. N.; Chang, P. N.; Han, H. Y.; Lee, D. T. W.; Ling, M. T.; Wong, Y. C.; Yap, Y.
L. Br. J. Cancer 2008, 99, 1832.
21. Tamilarasan, K. P.; Kolluru, G. K.; Rajaram, M.; Indhumathy, M.; Saranya, R.;
Chatterjee, S. B. M. C. Cell Biol. 2006, 7, 17.
22. Borghesani, P. R.; Peyrin, J. M.; Klein, R.; Rubin, J.; Carter, A. R.; Schwartz, P. M.;
Luster, A.; Corfas, G.; Segal, R. A. Development 2002, 6, 1435.
23. Kubinyi, H., 5th ed.. In Burger’s Medicinal Chemistry; Wolff, M. E., Ed.; Wiley:
New York, 1995; Vol. I, pp 497–571.
24. Arya, P.; Alibhai, N.; Qin, H.; Batist, G.; You, S.; Alaoui-Jamali, M. Bioorg. Med.
Chem. Lett. 1998, 8, 2433.
Figure 11. Prediction curves for CoMFA (a) and CoMSIA (b) model-predicted pIC50
for the test set of MCF-7 proliferation inhibition.
dependent dielectric at each lattice point. An energy cutoff of
30 kcal/mol was applied. The CoMFA steric and electrostatic fields
37
were scaled by the CoMFA-STD method in SYBYL
.
Tables 4 and 5
show the actual and predicted activity for both test and training
sets respectively. The predicted activities of training and test set
values were obtained by multiplying the values of each descriptors
for a particular row generated by CoMFA, CoMSIA by their corre-
sponding coefficient from the PLS equation in the model and this
can be translated through prediction curves for CoMFA and CoM-
SIA model-predicted pIC50 for the test set of MCF-7 proliferation
inhibition as shown in Figure 11.
25. Pearlman, R. S. 3D QSAR in Drug Design. Theory, Methods, and Applications;
ESCOM Science: Leiden, The Netherlands, 1993.
4.9. Alignment rule
26. El Sayed, K.; Khanfar, M.; Shallal, H.; Muralidharan, A.; Awate, B.; Youssef, D.;
Shah, G. J. Nat. Prod. 2008, 71, 396.
27. Pearce, B.; Parker, R.; Deason, M.; Qureshi, A.; Wright, K. J. Med. Chem. 1992, 35,
3595.
28. Denizot, F.; Lang, R. J. Immunol. Methods 1986, 89, 271.
29. Kleinman, H. K.; Martin, G. R. Semin. Cancer Biol. 2005, 15, 378.
30. Koblinski, J. E.; Kaplan-Singer, B. R.; van Osdol, S. J.; Wu, M.; Engbring, J. A.;
Wang, S.; Goldsmith, C. M.; Piper, J. T.; Vostal, J. G.; Harms, J. F.; Welch, D. R.;
Kleinman, H. K. Cancer Res. 2005, 65, 7370.
31. Gupte, A.; Buolamwini, J. Bioorg. Med. Chem. Lett. 2009, 19, 314.
32. Viswanadhan, V. N.; Ghose, A. K.; Revankar, G. R.; Robins, R. K. J. Chem. Inf.
Comput. Sci. 1989, 29, 163.
33. Wold, S.; Rhue, A.; Wold, H.; Dunn, W. J. SAIM J. Sci. Stat. Comput. 1994, 5, 735.
34. Wold, S.; Albano, C.; Dunn, W. J., III; Edlund, U.; Esbensen, K.; Geladi, P.;
Hellberg, S.; Johanson, E.; Lindberg, W.; Sjostrom, M. NATO Sci. Ser., Ser. C 1984,
138 (Chemometrics), 17.
35. Nikolic, K.; Agababa, D. J. Mol. Graph. Model. 2009, 27, 777.
36. Nikolic, K. J. Mol. Graph. Model 2008, 26, 868.
The alignment rule is one of the most sensitive input areas for
CoMFA studies. The SYBYL QSAR rigid body field fit command was
used for this alignment. Field uses a simplex algorithm in SYBYL that
minimizes the differences in steric and electrostatic fields averaged
over all the lattice grid points to find the best fit. The most active
tocotrienol 8 was used as a reference and there is also a skeleton
structure, without hydrogen atoms, was used as common com-
pound on which other molecules were aligned as showed in Figure
10. Values of the steric and electrostatic fields were truncated at
30 kcal/mol.
Acknowledgment
This work was supported in part by a grant from First Tech
International Ltd.
37. Matter, H.; Will, D.; Nazare, M.; Schreuder, H.; Laux, V.; Wehner, V. J. Med.
Chem. 2005, 48, 3290.