Quantitative structure inter-activity relationship (QSInAR). Cytotoxicity study of some hemisynthetic and isolated natural steroids and precursors on human fibrosarcoma cells HT1080

Article abstract of DOI:10.3390/molecules16086603

Combined experimental and quantitative structure inter-activity relationship (QSIAR) computation methods were advanced in order to establish the structural and mechanistic influences that steroids and triterpenes, either as newly synthesized or naturally isolated products, have on human HT1080 mammalian cancer cells. The main Hansch structural indicators such as hydrophobicity (LogP), polarizability (POL) and total energy (Etot) were considered and both the structure-projected as well as globally computed correlations were reported; while the inter-activity correlation of the global activity with those projected on structural information was revealed as equal to the direct structural-activity one for the trial sets of compounds, the prediction for the testing set of molecules reported even superior performances respecting those characteristic for the calibration set, validating therefore the present QSInAR models; accordingly, it follows that the LogP carries the most part of the cytotoxic signal, while POL has little influence on inhibiting tumor growth-A complementary behavior with their earlier known influence on genotoxic carcinogenesis. Regarding the newly hemisynthetic compounds it was found that stigmasta-4,22-dien-3-one is not adapted for cell membrane diffusion; it is recommended that aminocinnamyl chlorohydrate be further modified in order to acquire better steric influence, while aminocinnamyl-2,3,4,6-O- tetraacetyl-a-D-glucopyranoside was identified as being inhibited in the tumor cell by other molecular mechanisms-here not revealed-although it has a moderate-high anti-cancer structurally predicted activity.

Full text of DOI:10.3390/molecules16086603

Molecules 2011, 16, 6603-6620; doi:10.3390/molecules16086603
OPEN ACCESS
molecules
ISSN 1420-3049
www.mdpi.com/journal/molecules
Article
Quantitative Structure Inter-Activity Relationship (QSInAR).
Cytotoxicity Study of Some Hemisynthetic and Isolated Natural
Steroids and Precursors on Human Fibrosarcoma Cells HT1080
Mihai V. Putz 1,*, Marius Lazea 1 and Louis P. Sandjo 1,2
1
Laboratory of Computational and Structural Physical Chemistry, Chemistry Department,
West University of Timişoara, Pestalozzi Street No.16, Timişoara, RO-300115, Romania
2
Department of Organic Chemistry, University of Yaoundé 1 P.O. Box 812 Yaoundé, Cameroon
* Author to whom correspondence should be addressed; E-Mails: mvputz@cbg.uvt.ro or
mv_putz@yahoo.com; Tel.: +40-256-592-633; Fax: +40-256-592-620.
Received: 30 June 2011; in revised form: 28 July 2011 / Accepted: 29 July 2011 /
Published: 5 August 2011
Abstract: Combined experimental and quantitative structure inter-activity relationship
(QSIAR) computation methods were advanced in order to establish the structural and
mechanistic influences that steroids and triterpenes, either as newly synthesized or
naturally isolated products, have on human HT1080 mammalian cancer cells. The main
Hansch structural indicators such as hydrophobicity (LogP), polarizability (POL) and total
energy (Etot) were considered and both the structure-projected as well as globally
computed correlations were reported; while the inter-activity correlation of the global
activity with those projected on structural information was revealed as equal to the direct
structural-activity one for the trial sets of compounds, the prediction for the testing set of
molecules reported even superior performances respecting those characteristic for the
calibration set, validating therefore the present QSInAR models; accordingly, it follows
that the LogP carries the most part of the cytotoxic signal, while POL has little influence
on inhibiting tumor growth—A complementary behavior with their earlier known influence
on genotoxic carcinogenesis. Regarding the newly hemisynthetic compounds it was found
that stigmasta-4,22-dien-3-one is not adapted for cell membrane diffusion; it is
recommended that aminocinnamyl chlorohydrate be further modified in order to acquire
better steric influence, while aminocinnamyl-2,3,4,6-O-tétraacétyl-α-D-glucopyranoside
was identified as being inhibited in the tumor cell by other molecular mechanisms–here not
revealed–although it has a moderate-high anti-cancer structurally predicted activity.

Molecules 2011, 16
6604
Keywords: steroids and triterpenes; organic synthesis; HT1080 cancer cell lines;
cytotoxicity activity; Hansch indices; mono and multi-linear correlations
1. Introduction
Steroids and their triterpene precursors are both natural product classes widespread in plants and
possessing a broad spectrum of biological activities. This family of secondary metabolites is still a
research target since many of its compounds such as betulinic, oleanolic and ursolic acids have been
reported as antitumor agents [1], while stigmasterol which is one of the biosynthetic descents of
triterpenes have significant anticancer activity [2], analgesic [3] and hypoglycemic effects [4]. Moreover,
many previous studies have shown that some phytocomponents related to cholesterol presented
antimalarial activities [5], or induced apoptosis [6]. However, most of steroids presented in the literature
as cytotoxic compounds are polyhydroxylated or contain a α,β-unsaturated ketone function [7-10].
Besides, these terpenoids caused cell death when they are presented as sulfate [11] or quaternary
ammonium salts [12].
On the other side, apart of (hemi)synthesis and isolation of natural products the identification of
their anti-cancer molecular mechanism represents one of the main current research concerns [13,14]. In
this regard, chemoinformatics in general [15] and the recent spectral and residual variants of
quantitative structure-activity relationship (QSAR) in particular [16,17], may offer a reliable in silico
(or computational) analysis for establishing whether a series of compounds may indeed constitute a
fruitful direction of clinical investigations. Basically, the cytotoxic molecular mechanism may be
decomposed into three elementary steps: The compound effect on membrane structure, on the signal
transduction pathways, and on sterical induction signal towards the cancer cells; these effects can be
structurally modeled by the so called Hansch indices’ [18] such as hydrophobicity (LogP),
polarizability (POL) and optimized total energy (Etot) influence on the globally observed cytotoxic
activity (A), respectively. However, having to account on the cumulative effect of each of these steps
in producing the recorded output of programmed cancer cell death, a sort of inter-activity QSAR
should be investigated, i.e., by correlating the overall cytotoxic activity with its projection on
structurally-based independent activities for the above Hansch indicators. Such a study is to be in the
present report unfolded upon a congeneric class of steroids and triterpenes, among which some new
hemisynthetic ones were also included.
2. Chemistry and Cytotoxicity Assay
With the aim to discovering new bioactive components, chemical transformations of stigmasterol
were carried out following the route reaction described by Syamala et al. [19] (Scheme 1) while the
oxidation reaction with pyridinium chlorochromate (PCC) presented was performed as reported by
Li et al. [20]. For instance, the preparation of aminocinnamyl chlorohydrate was done following the
reaction of Scheme 2 [21], while the α-glucosylation reaction is new and presented in Scheme 3.
From the compounds of Table 1, the semisynthetic three pentacyclic triterpenes [lupeol (1),
oleanolic acid (2) and maslinic acid (10)] and two steroids (stigmasterol and sitosterol 3-O-β-D-gluco-

Molecules 2011, 16
6605
pyranoside) isolated from the stems of cultivated Triumfetta cordifolia (Tiliaceae) were tested against
HT1080 cancer cell lines, together with as well as the hemisynthetic steroids 3/G3, 4/G4 and 18 and
two cinnamyl derivatives 5/G5, 11/G11.
Scheme 1. Δ5 epoxidation of steroids.
a
O
O
HO
b
d
c
O
O
O
O
O
O
Reaction Conditions: (a) (H3CCO)2O/C5H5N (1/1) v/v, r.t., 1 h; (b) CH2Cl2, PCC/Al2O3 (10 equiv.),
under N2, 24 h; (c) CH2Cl2, PCC/Al2O3 (10 equiv.), under N2, 36 h; (d) A mixture of KMnO4 (1.6 g)
and CuSO4·5H2O (0.8 g), Water (0.3 mL), CH2C12 (20 mL), and stigmasteryl acetate (0.37 mmol)
followed by tert-butyl alcohol (1 mL) for a while and then cooled down after stirring for 2 h at r.t.
Scheme 2. Preparation of aminocinnamyl chlorohydrate.
Br
NH2.HCl
Zn/NH4Cl
N3
NaN3/DMF/NH4Cl
reflux/15h
EtOH-H2O (3:1)/
Room Temp. (10°C)/3h
Scheme 3. α-Glucosylation.
OH
OAc
OAc
NH2.HCl
C5H5N/(H3CCO)2O (1:1)
Room temp. (10°C)/1h
O
O
O
HO
HO
AcO
AcO
AcO
AcO
C5H5N/70°C/12h
OH
OAc
OAc
OH
NH
OAc
These cells are able to undergo physiological cell death, a process termed apoptosis [22], and are
sensitive to a wide variety of cytotoxic drugs such as etoposide, cisplatin, staurosporine or TNF-α. In
order to avoid a putative toxicity of the solvents, the maximum used volume of the compound solution
never exceeded 1/100 of the total volume of culture medium. Controls with analogous concentrations
of solvents were always carried out in parallel to check their lack of toxicity. Six compounds were
prepared at 100 mM allowing testing concentrations as high as 1 mM, whereas the other five
compounds showed lower solubility, consequently eliminating the possibility to test such concentrations.

Molecules 2011, 16
6606
Table 1. Some hemisynthetic and isolated natural products along their IC50 cytotoxic activity on human fribosarcoma cell HT1080 and the
computed [23] QSAR properties for the Hansch parameters, i.e., hydrophobicity, polarizability and the (PM3) total energy.
IC50 (mM)
QSAR properties
No.
Name
Chemical
Formula
(observed hours)
[A = log10(1/IC50)]
Structure
LogP
Pol
[Å3]
−Etot
/G/NG *
(Hemisynthetic, if any)
(kcal/mol)
1/G1
lup-20(29)-en-3β-ol
C30H50O
1
8.03
52.4
106582.9141
(Lupeol)
(48)
[0.000]
HO
2/G2
3/G3
4/G4
3β-hydroxyolean-12-en-28-oic acid
C30H48O3
C31H50O3
C29H46O
0.7
(48)
7.32
6.30
8.19
53.12 119358.2422
54.96 122840.5859
50.60 102422.1563
(Oleanolic acid)
OH
[0.155]
O
HO
3-O-acetyl-5,6
epoxystigmast-22-ene-3-ol
(Hemisynthetic)
0.54
(48)
[0.268]
O
O
O
Stigmasta-4,22-dien-3-one
0.3
(Hemisynthetic)
[0.523]
O

Molecules 2011, 16
6607
Table 1. Cont.
No.
Name
Chemical
Formula
Structure
IC50 (mM)
QSAR properties
/G/NG *
(Hemisynthetic, if any)
(observed hours)
[A = log10(1/IC50)]
LogP
Pol
−Etot
(kcal/mol)
[Å3]
5/G5
6/G6
Chlorhydrate
of Aminocinnamyl
(Hemisynthetic)
Acovenosigenin A
or
C9H12ClN
C23H34O5
0.1
2.82
19.03 39170.58984
NH
ClH
2
(48)
[1.000]
0.0012
[2.921]
O
O
2.07
1.89
41.55 108835.2969
1β,3β,14β-trihydroxycard-20(22)-
OH
enolide
OH
HO
O
O
7/G7
Periplogenin
C23H34O5
0.00081
[4.092]
41.55 108837.7656
or
3β,5β,14β-trihydroxycard-20(22)-
enolide
OH
HO
OH
O
8/G8
Periplogenin
or
C29H44O10
0.00016
[3.796]
0.67
54.97 162652.2813
OH
O
3-O-β-D-glucopyranosyl-5β,14β-
dihydroxycard-20(22)-enolide
HO
OH
HO
O
OH
OH

Molecules 2011, 16
6608
Table 1. Cont.
IC50 (mM)
QSAR properties
No.
Name
Chemical
Formula
(observed hours)
[A = log10(1/IC50)]
Structure
LogP
Pol
[Å3]
−Etot
(kcal/mol)
/G/NG *
(Hemisynthetic, if any)
O
O
9/G9
17α-H-digitoxigenin
C23H34O4
0.0015
[2.823]
3.04
40.91 102065.0234
or
3β,14β-dihydroxycard-20(22)-
enolide
OH
HO
10/G10
11/G11
1α,2β-dihydroxyolean-12-en-28-oic
C30H48O4
0.075
(48)
6.55
1.13
53.76 126133.4297
acid
OH
(Maslinic acid)
[1.125]
HO
HO
O
NH
Aminocinnamyl-2,3,4,6
-O-tetraacetyl-
C23H29NO9
0.88
(48)
45.54 137839.3281
O
O
O
O
O
α-D-glucopyranoside
(Hemisynthetic)
[0.056]
O
O
O
O
HO
12/NG1
Ginsenoside
C30H52O3
0.07678
[1.115]
6.29
54.45 120871.4531
OH
20(S)-protopanaxadiol
or
3β,12β,20S-trihydroxydammar-24-
HO
ene

Molecules 2011, 16
6609
Table 1. Cont.
IC50 (mM)
QSAR properties
No.
Name
Chemical
Formula
(observed hours)
[A = log10(1/IC50)]
Structure
LogP
Pol
[Å3]
−Etot
(kcal/mol)
/G/NG *
(Hemisynthetic, if any)
O
O
13/NG2
Acovenosigenin
A digitoxoside
or
C29H44O8
0.0011
[2.958]
2.25
53.70 149114.3281
OH
HO
HO
O
O
3-O-β-D-digitoxopyranosyl-1β,14β-
dihydroxycard-20(22)-enolide
Periplogenin
OH
O
O
14/NG3
15/NG4
C29H44O8
0.000093
[4.032]
2.07
2.35
53.70 149112.1719
Digitoxoside
or
HO
HO
O
3-O-β-D-digitoxopyranosyl-5β,14β-
dihydroxycard-20(22)-enolide
OH
O
OH
O
O
17α-H-periplogenin-
3-O-β-D-cymaroside
or
C30H46O8
0.000096
[4.018]
55.53 152546.9375
HO
O
3-O-β-D-cymarosyl-5β,14β-
dihydroxycard-20(22)-enolide
OH
O
O
OH
16
Tormentic acid
C30H48O5
Toxicity < 50%
for max 100 µM
dose
HO
OH
HO
O
HO

Molecules 2011, 16
6610
Table 1. Cont.
IC50 (mM)
(observed hours)
[A = log10(1/IC50)]
Toxicity < 50%
for max 500 µM
dose
QSAR properties
No.
/G/NG*
Name
Chemical
Formula
Structure
LogP
Pol
−Etot
(kcal/mol)
(Hemisynthetic, if any)
[Å3]
17
stigmasterol
C29H48O
HO
18
19
Stigmasta-4,22-dien-3,6-dione
C29H44O2
Toxicity < 50%
for max 10 µM
dose
(Hemisynthetic)
O
O
3-O-β-D-glucopyranoside
C35H60O6
Toxicity < 50%
for max 100 µM
dose
OH
O
of sitosterol
HO
HO
O
OH
* Of Gaussian (G) or Non-Gaussian (NG) type, see Figure 1.

Molecules 2011, 16
6611
Among these compounds (see Experimental section), four of them (molecules 16–19 of Table 1)
induced less that 50% cell death after 48 h of treatment at maximum concentrations, rending
impossible the determination of an IC50: This was the case for tormentic acid (Cmax = 100 μM),
stigmasterol (Cmax = 500 μM), stigmasta-4,22-dien-3,6-dione (Cmax = 10 μM) and 3-O-β-D-
glucopyranoside of sitosterol (Cmax = 100 μM).
The poor solubility of some of these compounds, such as stigmasta-4,22-dien-3,6-dione (Cmax = 10 μM)
was undoubtedly a great disadvantage to test their putative cytotoxicity. However, maslinic acid
induced 50% of cell death after 48 h of treatment (IC50) at 0.075 mM, thus showing a powerful and
specific cytotoxic effect. Among the six compounds which were prepared at 100 mM, it has been
possible to determine an IC50 value for each of them, from the most to the less efficient (Table 1):
Aminocinnamyl chlorohydrate (IC50 = 0.1 mM), stigmasta-4,22-dien-3-one (IC50 = 0.3 mM),
3-O-acetyl-5,6-epoxystigmast-22-ene-3-ol (IC50 = 0.54 mM), oleanolic acid (IC50 = 0.7 mM),
amino-cinnamyl-2,3,4,6-O-tetraacetyl-α-D-glucopyranoside (IC50 = 0.88 mM) and lupeol (IC50 = 1 mM).
Thus, three of the tested compounds appear to have both good solubility and high specificity
(IC50 ≤ 0.3 mM) to induce an efficient cellular toxicity in fibrosarcoma cells: Maslinic acid,
aminocinnamyl chlorohydrate and stigmasta-4,22-dien-3-one. It is probable for the other compounds
that the lack of one or both of these properties would be a disadvantage to trigger an effective in vivo
cytotoxicity.
3. Inter-Activity Models and Discussion
The analyzed molecules of Table 1 are grouped on Gaussian (G) and Non-Gaussian (NG) sets,
according with their activity ordering, see Figure 1; they will correspond with the trial/calibration and
the test/predicting QSAR series, respectively.
3.1. Calibration Quantitative Structure Inter-Activity Relationship (QSInAR)
For Gaussian (trial) compounds one has the basic QSAR analysis: The correlation equations of the
recorded activities with the Hansch parameters of Table 1, along the statistical indicators as the
Pearson correlation (R) and the standard error of estimation (SEE), either as individual structure
projected parameter-activity expressions:
AQSAR = 3.1098 − 0.3634 LogP , R1 = 0.6732, SEE1 = 1.2264
(1)
1
A2QSAR = 2.6528− 0.0244 POL , R2 = 0.1648, SEE2 = 1.6360
(2)
A3QSAR = 0.5600 − 0.000009 Etot , R3 = 0.1647, SEE3 = 1.6360
(3)
as well as the global structure-activity multi-regression equation:
A
QSAR =1.4045−1.6039LogP+ 0.6226POL+ 0.00019Etot , RQSAR = 0.8358, SEEQSAR = 1.0326
(4)

Molecules 2011, 16
Figure 1. Representation of the cytotoxic activities A = log10(1/IC50) for the molecules of
6612
Table 1 on Gaussian (G) and Non-Gaussian (NG) curves as corresponding to the trial and
testing sets of compounds in Table 1, respectively.
Equation (4) represents the ordinary QSAR predictive equation; note that the small correlation of
the total energy in Equation (3) and with the very small coefficient in (4) is the indication that while
the physicochemical properties of a small molecule (like steroids are) are definitely important (e.g., for
transport), the adverse effect may stem from the interaction of such compound with a macromolecule
(as the actual endpoints are very complex).
Such an observation further suggests that another sort of total-energy related quantity should be
considered in multi-regressions; as such, not the energy itself but its projected activity given by
Equation (3) is to be next considered; while doing the same with the other parameters too, one arrives
to provide the so called inter-activity correlation or the quantitative structure-inter-activity relationship
(QSInAR), that is correlating the activity of the Gaussian molecules of Table 1 with the projected
activities on the Hansch structural parameters of Equations (1)–(3), and reported in Table 2.
Accordingly, the obtained multi-linear inter-activity regression of global activities of Table 1 with
projected activities of Table 2 is unfolded for the training set with the form:
AQSInAR = 67.3578 + 4.4138AQSAR − 25.5178A2QSAR − 21.4027A3QSAR
1
(5)
RQSInAR = 0.8358, SEEQSInAR = 1.0326
It is immediate that Equation (5) displays two major features, namely:
• It does not reproduce Equation (4) in regards to the computed coefficients of the structurally-
projected activities, while
• Displaying equal correlation respecting the direct structural-parameters-activity correlation of
Equation (4).
The quantitative structure inter-activity relationship (QSInAR) may thus be used for further
modeling and prediction on equal footing with the original QSAR modeling. The external prediction
test is unfolded next.

Molecules 2011, 16
Table 2. The activities for the Gaussian (trial) molecules of Table 1 as computed by
6613
Equations (1)–(3).
QSAR
QSAR
QSAR
Index
G1
A1
A2
A3
0.191
0.449
0.820
0.133
2.085
2.358
2.423
2.866
2.005
0.729
2.699
1.372
1.355
1.310
1.416
2.188
1.638
1.638
1.310
1.653
1.339
1.540
1.473
1.583
1.613
1.438
0.90
G2
G3
G4
G5
G6
1.493
1.493
1.954
1.435
1.641
1.741
G7
G8
G9
G10
G11
3.2. Predictive QSInAR
However, in order to validate the present inter-activity approach, both QSAR and QSInAR-tests are
now performed for the Non-Gaussian/NG (test) molecules of Table 1 by employing the Gaussian
correlation Equations (4) and (5) to produce the data reported in Table 3, respectively. Finally, the
correlations between the observed activities and those structurally either as ordinary QSAR or by
actual QSInAR prediction analysis are presented in Equations (6) and (7) by employing the data of
Tables 1 and 3 for the Non-Gaussian/NG molecules, respectively as:
AOBS = −5.76267 + 3.04278AQSAR , RQSAR = 0.9982, SEEQSAR = 0.10153
(6)
AOBS = −5.39564 + 3.35479AQSInAR , RQSInAR = 0.9991, SEEQSInAR = 0.07114
(7)
The QSInAR comparative statistics between trial and test series of Table 1 and of Figure 1 are
graphically represented in Figure 2 based on the basic Equations (5) and (7), respectively; the
improvement of correlation for the test vs. trial correlation is evident, while being QSInAR superior
with respect to ordinary QSAR analysis.
Table 3. The activities for the Non-Gaussian/NG (or test) molecules of Table 1 as
computed upon Equations (1)–(3), alongside the predicted and computed activities as
prescribed by Equations (4) and (5), respectively.
QSAR
QSAR
QSAR
Index A1
A2
A3
AQSAR AQSInAR
NG1 0.824
NG2 2.292
NG3 2.358
NG4 2.256
1.324
1.343
1.343
1.298
1.648 2.251
1.902 2.898
1.902 3.187
1.933 3.224
1.938
2.500
2.787
2.822

Molecules 2011, 16
Figure 2. Scatter-plots representations of the observed vs. predicted correlations for the
6614
ordinary QSAR and inter-activity correlation QSInAR for both trial (Gaussian/G), the
central figure, and for the test (Non-Gaussian/NG) sets, the two diagonal extreme insets,
according with the regression Equations (4), (5) and (6), (7), respectively. Note that for the
trial molecules the QSAR and QSInAR correlations are identically, see Equations (4) and
(5), while for the test molecules the data of Table 3 were used.
Overall, one can draw the main features revealed by the present analysis:
• The tested correlation performances are notably superior to those recorded for the calibration
set for both QSAR and QSInAR frameworks; this way giving more confidence to the present
models as reliable predictive activity tools for future designed steroids and triterpenes;
• Although the interactivity QSInAR model is equal with the direct QSAR one in the trial-
correlation set, see Equations (4) and (5), it is slightly superior for the externally tested one, see
Equations (6) and (7);
• The consecrated triplet of Hansch parameters (LogP, POL and Etot) is sufficiently powerful to
provide significant and reliable correlations of the structure with the measured activities of
steroids and triterpenes, yet their projected activities are even more efficient in prediction when
considered together in the inter-activity correlations;
• For hemisynthetic compounds 4 one record in Table 2 shows that it reaches the minimum
(LogP) projected computed (QSAR) activities in the Gaussian series; the extreme values are
also seen for the hemisynthetic compound 5 for POL and Etot projected (QSAR) computed
maximum and minimum activities, respectively; this means that these compounds are not
predicted as active through the membrane transgression (compounds 4) and by the steric
interaction (compound 5) but only by signal transduction (compound 5);
• Instead, the hemisynthetic compound 11, that was globally observed with the lowest
cytotoxicity activity in Table 1, it is predicted as having moderate anti-cancer activity
especially by membrane diffusion, but also through polarizability and steric effects on tumoral

Molecules 2011, 16
6615
cells; this means that this compound is especially adapted for membrane transduction, while
being afterwards somehow inhibited by other (by this approach not revealed) molecular
mechanisms during its interaction with tumor cell; further studies may improve upon replacing
the total energy indicator, that is strongly depending on the size of the molecule and the number
of hydrogen-bonding partner it features, with a key energy information relaying on the binding-
site models (atomistic, quasi-atomistic, virtual) which might serve to estimate the strength of
the interaction;
• Polarizability has little influence on the recorded activity, see eq. (2), this way eliminating, in
principle, the electrophilic mechanism specific to modeling genotoxic carcinogenity [24]; this
is naturally since, in fact, the present study reveals the cytotoxic series that due to their induced
cancer cell apoptosis acts contrarily to those producing cancer.
4. Experimental
4.1. General
1
A Bruker DRX-400 MHz spectrometer was used for 1D H-NMR studies. Melting points were
measured in a Stuart Scientific Melting Point apparatus SMP3 and they are uncorrected. The matrix
MALDI-TOFMS were a saturated 2,5-dihydroxybenzoic acid (2,5-DHB) solution in 50% methyl
alcohol, 0.1% trifluoroacetic acid (TFA) and 2,5-dihydroxybenzoic acid (2,5-DHB, 1 M) solution in
50% acetonitrile / 50% ultrapure water / 0.1% trifluoroacetic acid (TFA), respectively. All the depositions
were made using the dried-droplet method. MALDI-TOFMS measurements were possible using a
Bruker Reflex IV time-of-flight mass spectrometer (TOFMS) (Bruker-Daltonics, Bremen, Germany)
equipped with the SCOUT 384 probe ion source, using a nitrogen pulsed laser (337 nm, model
VSD-337ND, Laser Science Inc., Boston, MA, USA) with energy output of 400 μJ/pulse. Vacuum column
chromatography, column chromatography and thin layer chromatography were performed respectively
on silica gel 60H (particle size 90% < 45µm), 200–300 mesh silica gel, and silica gel GF254.
4.2. Plant Material
T. cordifolia A. Rich was collected from the central Yaoundé province of Cameroon and a
specimen (N° 12830SRF Cam) has been deposited in the National Herbarium of Yaoundé, Cameroon.
4.3. Isolation
Dried stems of wild species of Triumfetta cordifolia A. Rich (2.55 Kg) were cut into small pieces
and powdered. Powder (250 g) was extracted with methanol by heating for 8 h. Methanol extract (60
g) was concentrated under vacuum and partitioned between water and ethyl acetate. Both phases were
concentrated under vacuum to give 35 g of organic phase and 24 g of aqueous extract. The ethyl
acetate extract was subjected to flash chromatography (silica gel, cyclohexane, cyclohexane-EtOAc
3:1-1:1-1:3, EtOAc, in order of increasing polarity) yielding five fractions. The cyclohexane-EtOAc
(3:1) fraction was purified by column chromatography with different mixtures of cyclohexane-EtOAc
yielding 217 fractions. The fractions 22–32 eluted with the mixture cyclohexane-EtOAc (9:1) yielded
32.0 mg of lupeol (1). The ones (40–52) eluted with the mixture cyclohexane-EtOAc (17:3) yielded 45

Molecules 2011, 16
6616
mg of oleanolic acid (2). The flash chromatography fraction from mixture of cyclohexane-EtOAc (1:1)
yielded 125.0 mg of stigmasterol (17) in the mixture cyclohexane-EtOAc (17:3),
2α-Hydroxyoleanolic acid (32.5 mg) was obtained from mixture of cyclohexane-EtOAc (3:1). The
fraction obtained from mixture of cyclohexane-EtOAc (1:3) was eluted with CH2Cl2-MeOH mixtures
of increasing polarity. The fractions 54–59 eluted with the 19:1 CH2Cl2-MeOH mixture gave the
β-D-glucopyranoside of β-sitosterol (11.0 mg). The dried aqueous partition was extracted with acetone
yielding 20 mg of extract. This was eluted with a CH2Cl2-MeOH mixture of increasing polarity to
yield 114 fractions. Fractions 74–80 eluted with 37:3 mixture of CH2Cl2-MeOH yielded tormentic acid
(30 mg).
4.4. Cellular Viability
Given that solvents are often toxic to mammalian cells, the compounds were dissolved at high
concentration (100 mM). When the solubility did not allow us to reach 100 mM, we tried to prepare
the most concentrated solutions possible. Lupeol, oleanolic acid, 3-O-acetyl-5,6-epoxystigmast-22-
ene-3-ol, stigmasta-4,22-dien-3-one and aminocinnamyl-2,3,4,6-O-tetraacetyl-β-D-glucopyranoside
were prepared at 100 mM in THF. Aminocinnamyl chlorohydrate was prepared at 100 mM in DMSO.
Stigmasterol and stigmasta-4,22-dien-3,6-dione were prepared at 50 mM and 1 mM, respectively, in
THF. Finally, maslinic acid and both tormentic acid and the 3-O-β-D-glucopyranoside of sistosterol
were prepared in DMSO at 50 mM and 10 mM, respectively.
The human HT1080 fibrosarcoma adherent cell line was cultured at 37 °C in a humidified
atmosphere containing 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM/F12) supplemented
with 10% fetal bovine serum together with penicillin (100 μg/mL), streptomycin (100 U/mL) and
glutamax (1% v/v) from Invitrogen. The cells were seeded in 12 well plates (5 × 104 cells/well). After
24 h, the medium was replaced in each well by 1 mL of complete medium with the appropriate
concentrations of the tested drugs. Corresponding controls with analogous concentrations of solvent
were carried out in parallel. Cells were then incubated for 48 h and cellular viability was determined
by flow cytometric analysis. Since type I (apoptosis), type II (autophagy) and type III (necrosis) cell
death trigger permeabilization of mitochondria [25], overall cell death was then determined with the
cationic lipophilic DiOC6(3) dye (Invitrogen) which specifically probes mitochondrial membrane
potential (ΔΨm) [26]. A decrease in forward light scattering (FSC) was also checked to confirm the
cell death process [27]. After drug treatment, the media from each well were kept in centrifuge tubes.
The adherent cells were detached using trypsine, pooled with the corresponding media, centrifuged and
resuspended in complete medium. Cells were then loaded with 100 nM DiOC6 (3) and incubated for
30 min at 37 °C. Finally, cells were kept on ice prior flow cytometric analysis. Flow cytometric
measurements were performed using a XL3C flow cytometer (Beckman-Coulter). Fluorescence was
induced by the blue line of an argon ion laser (488 nm) at 15 mW. Green fluorescence of DiOC6 (3)
was collected with a 525-nm band pass filter. Analyses were performed on 104 cells.

Molecules 2011, 16
6617
4.5. Compounds Chosen from the Literature
The compound 12 was obtained from the Research Institute of Luye Pharmacy, Yantai, China and
its cytotoxic activity on human fibrosarcoma was tested [28]. Its melting point was measured on the
Fisher micromelting point apparatus (hot-stage) and recorded uncorrected [29]. The compounds 6–9,
and 13–15 were isolated and identified from Streptocaulon juventas [30] and their cytotoxic properties
investigated [31,32].
5. Conclusions
Since nowadays there is increasing interest in public health through various environmental
interactions, the design of chemical compounds with little toxic effects on organisms or with anti-cancer
effects remains at the forefront of actual molecular research either by experimental synthesis and
computational modeling. In this context, the present work sought to combine both aspects
(experimental and computational) in reporting new molecules in a series with pre-defined cytotoxicity
activities. As such, the chemical synthesis of two new stigmast-en-ol/one (compounds 3 and 4 of Table 1)
and aminocinnamyl- (compounds 5 and 11 of Table 1) derivatives were described and their
cytotoxicity influence on human fibrosarcoma cells HT1080 analyzed by means of structural projected
activities within the so called quantitative structure inter-activity relationship (QSIAR) modeling. The
results are quite relevant, and are summarized below:
• The cytotoxicity of congeneric (both trial and test) series was considered and modeled as being
projected on three fundamental Hansch parameters such as LogP, POL and Etot; it was found
that the hydrophobicity (LogP) is the main mechanism of interaction, while the polarizability
(POL) has little influence on anti-tumoral mechanism—Rejecting this way the eventual
generalization of Miller’s electrophilic theory of carcinogenicity to its reversible evolution
though cancer cells’ apoptosis; in fact, while LogP was found to have little influence on
carcinogenity [17], it appears here as having the main influence; one may conclude therefore
that indeed the carcinogenity (formation of cancer cells) and the cytotoxicity (death of cancer
cells) are complementary and not symmetrical with respect to QSAR analysis;
• The presented inter-activity QSInAR correlation equations show overall slightly better
statistical prediction with respect to the custom QSAR counterpart, while providing equal
performance for our trial set of newly hemisynthesyzed and isolated natural steroids and
precursors triterpenes; this behavior assures the present reliable QSInAR analysis and the
conclusions that result from it, namely:
o The newly hemisynthesized compounds are not all well suited to accounting for
cytotoxicity of HT1080 cells, while their anti-cancer action is more relevant through
the projection on some particular stages of molecular mechanism of ligand-receptor
interaction, as following:
o The compound stigmasta-4,22-dien-3-one (compound 4 in Table 1) is particularly
not adapted for membrane transduction and therefore should be omitted from further
clinical studies;

Molecules 2011, 16
6618
o The compound aminocinnamyl chlorohydrate (compound 5 in Table 1) is less
effective in tumoral apoptosis by steric signals but potentially active by polarization
and therefore it should be further developed into another derivatives with improved
structural projected activities;
o Aminocinnamyl-2,3,4,6-O-tetraacetyl-α-D-glucopyranoside (compound 11 in Table
1) although it recorded the lowest observed overall molecular activity it is predicted
to have moderately high cyt-activities as projected on all hydrophobic, polarizability
and steric influences; such a behavior suggest that it undergoes successive molecular
interaction within the cancer cells where some unrevealed inhibition interaction
impedes its structurally-projected activities from being globally manifest towards
tumor apoptosis.
While the ideal case will be the equal cytotoxicity performances at both local (structurally-
projected) and global (observed) activities, one may nevertheless rely on the inter-activity analysis and
make further recommendations for new synthesis and molecular interactions and mechanism as far as
it provides statistical relevant results for both trial and test congeneric molecules when
physicochemical working structural indicators are included. The present work provides the first step in
this combined experimental-QSInAR analysis and there is hope that it will be followed by other
similar dedicated studies.
Acknowledgements
This work was supported by Romanian CNCS-UEFISCDI (former CNCSIS-UEFISCSU) project
TE16/2010-2011 within the PN II-RU-TE-2009-1 framework, and partly by the grant POSDRU
/21/1.5/G/38347. Louis P. Sandjo thanks Vincent Rincheval from Laboratory of Genetics and Cellular
Biology of University of Versailles for support on the cytotoxicity assays and AUF-Eugen Ionescu for
the mobility grant to West University of Timisoara. The anonymous referees are kindly thanked for
their constructive comments on the paper.
References and Notes
1. Ma, C.-M.; Cai, S.-Q.; Cui, J.-R.; Wang, R.-Q.; Tu, P.-F.; Hattori, M.; Daneshtalab, M.
The cytotoxic activity of ursolic acid derivatives. Eur. J. Med. Chem. 2005, 40, 582-589.
2. Awad, A.B.; Fink, C.S. Phytosterols as anticancer dietary components: Evidence and mechanism
of action. J. Nutr. 2000, 130, 2127-2130.
3. Villaseñor, I.M.; Angelada, J.; Canlas, A.P.; Echegoyen, D. Bioactivity studies on β-Sitosterol
and its glucoside. Phytother. Res. 2002, 16, 417-421.
4. Jamaluddin, F.; Mohamed, S.; Lajis, Md.N. Hypoglycaemic effect of Parkia speciosa seeds due to
the synergistic action of β-sitosterol and stigmasterol. Food Chem. 1994, 49, 339-345.
5. Saez, J.; Cardona, W.; Espinal, D.; Blair, S.; Mesa, J.; Bocar, M.; Jossang, A. Five new steroids
from Solanum nudum. Tetrahedron 1998, 54, 10771-10778.

Molecules 2011, 16
6619
6. Itoh, H.; Ito, H.; Hibasami, H. Blazein of a new steroid isolated from Agaricus blazei Murrill
(himematsutake) induces cell death and morphological change indicative of apoptotic chromatin
condensation in human lung cancer LU99 and stomach cancer KATO III cells. Oncol. Rep. 2008,
20, 1359-1361.
7. Wang, Y.-S.; Yang, J.-H.; Luo, S.-D.; Zhang, H.-B.; Li, L. New cytotoxic steroid from Stachyurus
himalaicus var. himalaicus. Molecules 2006, 11, 536-542.
8. Samadi, A.K.; Tong, X.; Mukerji, R.; Zhang, H.; Timmermann, B.N.; Cohen, M.S. Withaferin A,
a cytotoxic steroid from Vassobia breviflora, induces apoptosis in human head and neck
squamous cell carcinoma. J. Nat. Prod. 2010, 73, 1476-1481.
9. Shang, X.-Y.; Li, J.-J.; Liu, M.-T.; Li, S.; Liu, Y.; Wang, Y.-F.; Huang, X.; Jin, Z.-L. Cytotoxic
steroids from Monascus purpureus-fermented rice. Steroids 2011, 76, 1185-1189.
10. Amagata, T.; Doi, M.; Tohgo, M.; Minoura, K.; Numata, A. Dankasterone, a new class of
cytotoxic steroid produced by a Gymnascella species from a marine sponge. Chem. Commun.
1999, 1321-1322.
11. Zhang, H.-J.; Sun, J.-B.; Lin, H.-W.; Wang, Z.-L.; Tang, H.; Cheng, P.; Chen, W.-S.; Yi, Y.-H.
A new cytotoxic cholesterol sulfate from marine sponge Halichondria rugosa. Nat. Prod. Res.
2007, 21, 953-958.
12. Brassart, B.; Gomez, D.; De Cian, A.; Paterski, R.; Montagnac, A.; Qui, K.-H.; Temime-Smaali, N.;
Trentesaux, C.; Mergny, J.-L.; Gueritte, F.; et al. A new steroid derivative stabilizes
G-quadruplexes and induces telomere uncapping in human tumor cells. Mol. Pharmacol. 2007,
72, 631-640.
13. Rickardson, L.; Fryknäs, M.; Dhar, S.; Lövborg, H.; Gullbo, J.; Rydåker, M.; Nygren, P.;
Gustafsson, M.G.; Larsson, R.; Isaksson, A. Identification of molecular mechanisms for cellular
drug resistance by combining drug activity and gene expression profiles. Br. J. Cancer 2005, 93,
483-492.
14. Caraglia, M.; Santini, D.; Marra, M.; Vincenzi, B.; Tonini, G.; Budillon, A. Emerging anti-cancer
molecular mechanisms of aminobisphosphonates. Endocr. Relat. Cancer 2006, 13, 7-26.
15. Comprehensive Chemometrics: Chemical and Biochemical Data Analysis; Brown, S.D.,
Tauler, R., Walczak, B., Eds.; Elsevier: Amsterdam, The Nederland, 2009.
16. QSAR & SPECTRAL-SAR in Computational Ecotoxicology; Putz, M.V., Ed.; Apple Academics:
Ontario, Canada, 2011.
17. Putz, M.V. Residual-QSAR. Implications for genotoxic carcinogenesis. Chem. Central J. 2011, 5, 29.
18. Hansch, C.; Kurup, A.; Garg, R.; Gao, H. Chem-bioinformatics and QSAR: A review of QSAR
lacking positive hydrophobic terms. Chem. Rev. 2001, 101, 619-672.
19. Syamala, M.S.; Das, J.; Baskaran, S.; Chandrasekaran, S. A novel and highly β-selective
epoxidation of Δ5-unsaturated steroids with permanganate ion. J. Org. Chem. 1992, 57, 1928-1930.
20. Li, S.-H.; Li, T.-S. Steroidal 5-en-3-ones, intermediates of the transformation of steroidal 5-en-3β-
ols to steroidal 4-en-3,6-diones oxidized by pyridinium dichromate and pyridinium chlorochromate.
Steroids 1998, 63, 76-79.
21. Lin, W.; Zhang, X.; He, Z.; Jin, Y.; Gong, L.; Mi, A. Reduction of azides to amines or amides
with zinc and ammonium chloride as reducing agent. Syn. Comm. 2002, 32, 3279-3284.

Molecules 2011, 16
6620
22. Chowdhury, B.T.; Bhat, G.K. Current concepts in apoptosis: The physiological suicide program
revisited. Cell. Mol. Biol. Lett. 2006, 11, 506-525.
23. Program Package, HyperChem 7.01; Hypercube, Inc.: Gainesville, FL, USA, 2002.
24. Miller, E.C.; Miller, J.A. Searches for ultimate chemical carcinogens and their reactions with
cellular macromolecules. Cancer 1981, 47, 2327-2345.
25. Bras, M.; Queenan, B.; Susin, S.A. Programmed cell death via mitochondria: Different modes of
dying. Biochemistry (Moscow) 2005, 2, 231-239.
26. Galluzi, L.; Zamzami, N.; de La Motte-Rouge, T.; Lemaire, C.; Brenner C.; Kroemer, G. Methods
for the assessment of mitochondrial membrane permeabilization in apoptosis. Apoptosis 2007, 12,
803-813.
27. Darzynkiewicz, Z.; Juan, G.; Li, X.; Gorczyca, W.; Murakami, T.; Traganos, F. Cytometry in cell
necrobiology: analysis of apoptosis and accidental cell death (necrosis). Cytometry 1997, 27, 1-20.
28. Li, G.; Wang, Z.; Sun, Y.; Liu, K.; Wang, Z. Ginsenoside 20(S)-protopanaxadiol inhibits the
proliferation and invasion of human fibrosarcoma HT1080 cells. Basic Clin. Pharmacol. Toxicol.
2006, 98, 588-592.
29. Im, K.S.; Chang, E.H.; Je, N.G. A modified alkaline hydrolysis of total ginsenosides yielding
genuine aglycones and prosapogenols. Arch. Pharm. Res. 1995, 18, 454-457.
30. Ueda, J.-Y.; Tezuka, Y.; Banskota, A.H.; Tran, Q.L.; Tran, Q.K.; Saiki, I.; Kadota, S. Constituents
of the Vietnamese medicinal plant Streptocaulon juventas and their antiproliferative activity
against the human HT-1080 fibrosarcoma cell line. J. Nat. Prod. 2003, 66, 1427-1433.
31. Karle, I.L.; Karle, J. The crystal structure of digitoxigenin, C23H34O4. Acta Cryst. 1969, 25, 434-442.
32. Khine, M.M. Isolation and characterization of phytoconstituents from myanmar medicinal plants.
In Dissertation; Martin-Luther-Universität Halle-Wittenberg: Halle, Germany, 2006; pp. 114-119.
Sample Availability: Samples of the compounds are available from the author Louis P. Sandjo.
© 2011 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/3.0/).