Voss et al. - 2006 - Identification of potent anticancer activity in Xi

Identification of potent anticancer activity in Ximenia americana aqueous
extracts used by African traditional medicine
Cristina Voss, Ergu¨l Eyol, Martin R. Berger*
Unit of Toxicology and Chemotherapy, Deutsches Krebsforschungszentrum Heidelberg, E100, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
Received 17 March 2005; revised 13 May 2005; accepted 31 May 2005
Available online 11 July 2005
Abstract
The antineoplastic activity of a plant powder used in African traditional medicine for treating cancer was investigated by analyzing the
activity of various extracts in vitro. The most active, aqueous extract was subsequently subjected to a detailed investigation in a panel of 17
tumor cell lines, showing an average IC50 of 49 mg raw powder/ml medium. The sensitivity of the cell lines varied by two orders of
magnitude, from 1.7 mg/ml in MCF7 breast cancer cells to 170 mg/ml in AR230 chronic-myeloid leukemia cells. Immortalized, non-
tumorigenic cell lines showed a marginal sensitivity. In addition, kinetic and recovery experiments performed in MCF7 and U87-MG cells
and a comparison with the antineoplastic activity of miltefosine, gemcitabine, and cisplatinum in MCF7, U87-MG, HEp2, and SAOS2 cells
revealed no obvious similarity between the sensitivity profiles of the extract and the three standard agents, suggesting a different mechanism
of cytotoxicity.
The in vivo antitumor activity was determined in the CC531 colorectal cancer rat model. Significant anticancer activity was found
following administration of equitoxic doses of 100 (perorally) and 5 (intraperitoneally) mg raw powder/kg, indicating a 95% reduced activity
following intestinal absorption.
By sequencing the mitochondrial gene for the large subunit of the ribulose bis-phosphate carboxylase (rbcL) in DNA from the plant
material, the source plant was identified as Ximenia americana.
A physicochemical characterization showed that the active antineoplastic component(s) of the plant material are proteins with galactose
affinity. Moreover, by mass spectrometry, one of these proteins was shown to contain a stretch of 11 amino acids identical to a tryptic peptide
from the ribosome-inactivating protein ricin.
D 2005 Elsevier Inc. All rights reserved.
Keywords: New antineoplastic agent; Plant origin; Selective cytotoxicity; CC531 isogenic rat model
Introduction
For many centuries, plants have been a main source for
drug development. Early examples of anticancer agents
developed from higher plants are the antileukemic alkaloids
vinblastine and vincristine, which were both obtained from
the Madagascar periwinkle (Vinca rosea). Since the early
1950s, the identification and development of new lead
compounds for anticancer chemotherapy has been partially
driven by broad plant screening programs. This way, new
antineoplastic agents like the taxane derivative paclitaxel,
later known as taxol (from Taxus brevifolia) and the alkaloid
camptothecin (from the Chinese tree Camptotheca acumi-
nata) were identified (De Smet, 1997). Interestingly, an
analysis of plant materials that had been studied at the US
NCI for discovering new anticancer drugs showed that if
ethnopharmacological information had been used, the yield
of plants harboring antineoplastic activity would have been
significantly increased (Spjut and Perdue, 1976). The drug
0041-008X/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.taap.2005.05.016
Abbreviations: NCI, National Cancer Institute (USA); MTT, (3-(4,5-
dimethyldiazol-2-yl)-2,5 diphenyl tetrazolium bromide); IC10, inhibitory
concentration 10; IC50, inhibitory concentration 50; IC90, inhibitory
concentration 90; U, unit for growth-inhibiting activity; T/C%, treated
over control ratio (in percent); TA, total growth-inhibiting activity; rbcL,
ribulose bis-phosphate carboxylase, large subunit; MW, molecular weight;
RIP, ribosome inactivating protein.
* Corresponding author. Fax: +49 6221 423313.
E-mail address: m.berger@dkfz.de (M.R. Berger).
Toxicology and Applied Pharmacology 211 (2006) 177 – 187
www.elsevier.com/locate/ytaap

potential of Chinese plants, where a written tradition of
using medical herbs has existed for thousands of years, is
currently being exploited. Several natural compounds with
antineoplastic activity have been identified. Some of them
have already been tested in clinical trials, like the alkaloids
homoharringtonine from Cephalotaxus species (Kantarjian
et al., 2001) and indirubin from a mixture from 11 Chinese
plants traditionally used to treat chronic myelocytic leuke-
mia (Xiao et al., 2002). Many other natural compounds of
Chinese origin, belonging to substance classes as diverse as
alkaloids, diterpenes, flavones, polysaccharides, and pro-
teins, were reported to show significant antineoplastic
activity in various in vitro or in vivo model systems (Tang
et al., 2003a, 2003b).
The same concept of exploiting the knowledge of
traditional medicine was followed in characterizing the
biological activity of a plant material with an African
ethnopharmacological background: case reports from a
cancer clinic in Tanzania described an unexpected improve-
ment of patients who had been terminally ill from advanced
prostate cancer. On the clinician’s inquiry, these patients
reported that they had been treated with a powder used in
African traditional medicine. This plant material was offered
by the clinician to our group for an initial analysis for
anticancer activity. The surprisingly high antiproliferative
efficacy found in two cell lines prompted a broad
experimental investigation, including assessment of the
antineoplastic activity in a cell line screen, comparison with
standard chemotherapeutic drugs and investigation of the
anticancer activity in vivo. In addition, a detailed phys-
icochemical analysis of the nature of the active compound(s)
was started. Moreover, the source of the material was
identified.
Methods
Materials and extracts
Plant material of African origin was obtained from a
local traditional healer in Tanzania. The powdered material
was supplied in three batches (designated A, B, and C)
within a period of 3 years. Each batch was tested for
antineoplastic activity in vitro in two cell lines (MCF7 and
HT29). For the main cell line screen and the in vivo
experiments, batch B was used.
Extracts were prepared by suspending 50 mg dry powder
per ml solvent and shaking for 10 min. The undissolved
material was separated by centrifugation (2500 Â g, 5 min)
and the supernatant stored at À20 -C. All extract concen-
trations are given in Ag/ml, with ‘‘Ag’’ representing the
amount of dry raw material which had been used for
preparing the extract.
Primary extracts were prepared from the raw material in
water and water-based buffers (PBS, Tris–HCl) as well as in
organic solvents (ethanol, methanol, and acetone) and
admixtures of water and organic solvents. Secondary
extracts were prepared from pre-extracted dried pellets.
To compare the amount of biological activity which
could be extracted by a specific solvent, a biological activity
unit (U) was defined as the amount of extract (in Al) able to
inhibit the growth of MCF7 cells by 50%. The specific
activity of an extract was described by the value U/ml. The
total activity (TA, U/g dry raw material) of an extract was
subsequently calculated by multiplying the specific activity
(U/ml) with the total volume of the respective extract (ml)
and dividing by the amount of raw substance (g) which had
been used for extraction.
Other chemicals were obtained as follows: media for cell
culture from Sigma-Aldrich (Mu¨nchen, Germany), MTT
from Serva (Heidelberg, Germany), miltefosine from H.J.
Eibl, Max Plank Institute for Biophysical Chemistry
(Go¨ttingen, Germany), gemcitabine from Eli-Lilly (Bad
Homburg, Germany), cisplatin from Bristol-Meyers (Ber-
gisch-Gladbach, Germany).
In vitro experiments
The cell line panel included 17 tumor cell lines originat-
ing from human (n = 16) and rat (n = 1) neoplasias as well as
four non-tumorigenic, immortalized cell lines. The tumor
cell lines were the human chronic-myeloid leukemia cell
lines AR230 (Wada et al., 1995), BV173 (DSMZ), CML-T1
(DSMZ), K562 (DSMZ), LAMA84 (DSMZ), the human
acute-myeloid leukemia cell line HL60 (DSMZ), the human
acute-lymphoblastic leukemia cell line SKW-3 (DSMZ), the
human breast cancer cell lines MCF7 (estrogen-receptor
positive, DSMZ) and MDA-MB-231 (estrogen-receptor
negative, ATCC), the human glioma cell line U333 (kindly
provided by Dr. Langbein, Dept. of Cell Biology, DKFZ,
Germany), the human glioblastoma/astrocytoma cell line
U87-MG (ATCC), the human epidermoid larynx carcinoma
cell line HEp2 (ATCC), the human large cell lung carcinoma
cell line NCI-H460 (ATCC), the human osteosarcoma cell
line SAOS2 (DSMZ), the human prostate carcinoma (meta-
stasis to bone) cell line PC3 (ATCC), the human colon
adenocarcinoma cell line HT29, and the rat colon carcinoma
cell line CC531 (Saenger et al., 2004; Wittmer et al., 1999).
The non-tumor cell lines included the human breast
epithelial cell line MCF10 (ATCC), the canine kidney
epithelial cell line MDCK (ATCC), the mouse fibroblast
cell line NIH/3T3 (ATCC), and the human prostate cell line
PNT-2 (ECACC). Cells were grown as recommended by the
suppliers in an incubator under standard culture conditions
(humidified atmosphere, 37 -C and 5% CO2 in air). For
experiments, the cells were distributed into 96-well (adher-
ently growing cells) or 24-well (cells growing in suspension)
flat bottom microtiter plates. The initial cell density was
selected for each cell line in a way that logarithmic growth
was maintained throughout the experiment.
The viable cell count was measured by the MTT [3-(4,5-
dimethyldiazol-2-yl)-2,5 diphenyl tetrazolium bromide]
C. Voss et al. / Toxicology and Applied Pharmacology 211 (2006) 177–187178

assay as described by Konstantinov et al. (1998) with some
modifications. In short, cells growing in microtiter plates
were exposed to the treatment as described in Table 1. For
all extracts prepared in organic solvents, a vehicle control
was run in parallel covering the respective range of
concentrations, not exceeding 2% (ethanol and methanol)
or 0.2% (acetone).
For adherent cells, 0.1 Â volume MTT solution (10 mg/
ml in PBS) was added to each well at the end of the
incubation period except for the blank. After 3 h incubation,
the culture medium containing the excess MTT was
removed. The formazan crystals were dissolved in 200 Al
isopropanol containing 0.4 M HCl. Cells growing in
suspension were pelleted, washed in PBS, re-suspended in
1 ml fresh medium, and distributed into 96-well flat bottom
microtiter plates (100 Al/well). After adding 10 Al MTT
solution and incubating 3 h, the formazan crystals were
dissolved by 100 Al isopropanol containing 0.4 M HCl.
Extinctions were measured by a microplate reader at 540 nm
(reference 690 nm). Before treating a specific cell line, the
range of linearity for the MTT assay (extinction vs. cell
number) was defined. In a single experiment, 8 wells were
used for each concentration. Each experiment was repeated
two to three times.
In vivo experiments
For determining the effect of the aqueous extract in a rat
liver metastasis model, the rat colon cancer cell line CC531-
lac-Z was used, as described before (Saenger et al., 2004;
Wittmer et al., 1999). In short, 4 Â 106
CC531-lac-Z cells
were implanted intraportally into male Wag Rij rats (day 0).
Tumor-bearing rats were treated perorally or intraperito-
neally with the aqueous extract starting on day 1, as shown
in Table 2. Three weeks later (day 21), the experiment was
terminated, the liver of the animals was removed, weighed,
and kept at À80 -C until analysis. The number of tumor
cells per liver was determined by the h-galactosidase assay
(Applied Biosystems, Weiterstadt, Germany), as compared
to a standard curve established with a mixture of healthy
liver tissue and rising numbers of tumor cells.
Identification of the source plant
DNA was extracted from the plant material by the
cetyltrimethylammonium bromide (CTAB) method and the
chloroplastic gene for the large subunit of ribulose bis-
phosphate carbixilase (rbcL) was amplified and sequenced
as described by Treutlein and Wink (2002). For the resulting
DNA sequence, a phylogenetic analysis was carried out
with the program ‘‘Path’’ of the program package HUSAR
(DKFZ Heidelberg). The identity of the plant was confirmed
by isolating DNA from fresh plant material and sequencing
of the rbcL gene.
Identification of the active compound
pH-dependency. Aqueous extracts were prepared in differ-
ent buffers (100 mM Na acetate, pH = 4.5; 100 mM Tris–
HCl, pH = 7.0; NaCO3, pH = 9.5) as well as in 100 mM HCl
Table 2
Antineoplastic effect of the aqueous extract of X. americana in the CC531-lac-Z rat liver metastasis model
Group no. No. of animals Treatment Liver weight Tumor cell no./liver
Dosage (mg/kg) Route Total dosea
(mg/kg) Mean T SD (g) T/C% Mean T SD T/C%
1 11 Control – – 44.5 T 7.2 – 4.8 T 2.0 Â 109
–
2 6 100 p.o. 1000 15.1 T 9.1 34.0b
1.5 T 1.1 Â 109
31.2b
3 9 Control – – 37.6 T 11.0 – 10.8 T 2.5 Â 109
–
4 6 5 i.p. 50 23.4 T 12.5 62.2 3.7 T 2.5 Â 109
34.3b
a
Administration started on day 1 and was continued every second day until day 21.
b
P < 0.05 (Wilcoxon rank sum test).
Table 1
Design of in vitro experiments
Type of assay Agent Concentration range Duration of exposure (days) Post-exposure period (days)
MTT assay for
antiproliferative activity
Primary/secondary extract in water
or organic solvents
0.1–1000 Ag/ml 3 –
Organic solvent 0.02–2%
Standard MTT assay
for cell line screen
Primary extract in water 0.1–1000 Ag/ml 3 À
Variation in exposure duration Primary extract in water 1–100 Ag/ml 2, 3, 4 À
Recovery after
standard exposure
Primary extract in water 1–100 Ag/ml 3 1, 2, 3, 4, 5
Positive controls Miltefosinea
0.32–160 AM 3 À
Gemcitabine 0.001–160 AM
Cisplatinum 0.32–80 Ag/ml
a
Hexadecylphosphocholine (HPC).
C. Voss et al. / Toxicology and Applied Pharmacology 211 (2006) 177–187 179

and 100 mM NaCl. After a short (15 min) or long (2 h)
incubation at room temperature, the growth-inhibiting
activity of the extracts was determined in MCF7 cells.
Sensitivity to increased temperature. The aqueous extract
was incubated for 10 or 30 min at 50 -C and for 10 min
at 90 -C and its antineoplastic activity was subsequently
analyzed in MCF7 cells.
Precipitation. Precipitation experiments were performed
by adding 1–4 volumes of ethanol, isopropanol, or
acetone to 1 volume of aqueous extract and subsequent
separating the pellet by centrifugation. The pellet was re-
dissolved in the original volume of water and both pellet
and supernatant were analyzed for antineoplastic activity
in MCF7 cells.
Ultrafiltration. Ultrafiltration of the aqueous extract was
performed on Microcon YM10 (MW cutoff 10 kDa), YM50
(MW cutoff 50 kDa), and YM100 (MW cutoff 100 kDa)
membranes (Millipore, Schwalbach, Germany). Filtrates as
well as supernatants were analyzed for antineoplastic
activity in MCF7 cells.
HPLC assay for tannins. Tannins were identified in
various extract samples with or without preceding alka-
line/acid hydrolysis, by liquid-chromatography electro-
spray–ionization mass spectrometry (Owen et al., 2003).
SDS-PAGE analysis. SDS-PAGE of various extracts was
performed with the NuPAGE SDS-polyacrylamide gel-
electrophoresis system from Invitrogen (Karlsruhe, Ger-
many), under reducing or non-reducing conditions. For
protein detection, gels were stained with silver (Silver
staining kit SDS PAGE, Serva, Heidelberg, Germany) or
Coomassie blue (SimplyBlue, Invitrogen, Karlsruhe, Ger-
many). To detect glycosylated compounds, a periodic acid-
Schiff’s reagent-staining (PAS) was performed (Zacharius et
al., 1969).
Galactose affinity. A matrix containing free terminal
galactose was prepared by partial hydrolysis of Sepharose
4B, as described in Adam and Becker (2000). The
secondary aqueous extract was adjusted to 20 mM
Tris–HCl, 500 mM NaCl, pH = 7.0 and incubated with
hydrolyzed Sepharose for 15 min at room temperature
under gentle shaking. The hydrolyzed Sepharose gel was
subsequently separated by centrifugation and the super-
natant was analyzed for antineoplastic activity in MCF7
cells, as well as by SDS-PAGE. The separated gel was
washed three times with the binding buffer (20 mM Tris–
HCl, 500 mM NaCl, pH = 7.0) and incubated at room
temperature for 15 min with elution buffer (20 mM Tris–
HCl, 500 mM NaCl, 100 mM galactose, pH = 7.0). After
a brief centrifugation, the gel was discarded and the
supernatant containing proteins with galactose affinity was
collected and analyzed for antineoplastic activity in MCF7
cells and by SDS-PAGE.
Mass spectroscopy. After separating the proteins from the
fraction with galactose affinity by SDS-PAGE and staining
with Coomassie blue, the desired protein band was cut out
from the gel and an in-gel digestion was performed with
trypsin as described in Kinter and Sherman (2000). The
eluted tryptic peptides were analyzed by electrospray–
ionization mass spectrometry on a hybrid Q-TOF mass
spectrometer type Q-Tof2 (Waters Micromass, Manchester).
For the obtained mass spectra, a database search was
performed using the program Mascot search from Matrix
science (Perkins et al., 1999).
Evaluation and statistics
Means and respective standard deviations were calcu-
lated from the microtiter plate readings. For replicate
experiments, the results were averaged and characterized
by the standard error. Dose–response curves were plotted
for each cell line. Treatment effects were given as percent of
control (T/C%). The sensitivity of each cell line was
characterized by the values IC10 (concentration inhibiting
the growth by 10%), IC50 (concentration inhibiting the
growth by 50%), and IC90 (concentration inhibiting the
growth by 90%). As for all concentrations in this paper,
the IC values are given in Ag/ml, with ‘‘Ag’’ representing
the amount of dry raw material which had been used for
preparing the extract.
A mean IC50 was calculated for the cell panel used. For
comparison of cell lines, a ratio between individual and
mean IC50 values was calculated for each cell line.
Data from the in vivo experiments (liver weight and
tumor cell number/liver) were presented as mean T standard
deviation. A non-parametric rank sum test was used to
compare treated vs. control groups (Wilcoxon rank sum test,
ADAM statistics program, DKFZ, Heidelberg). P values <
0.05 were considered significant.
Results
Antineoplastic activity in various solvents
The material to be investigated appeared as a reddish-
brown powder of plant origin. Primary and secondary
extracts prepared in various solvents or solvent–water
admixtures were tested for antineoplastic activity in MCF7
breast und HT29 colon cancer cell lines (Fig. 1). In both cell
lines, the highest cell growth-inhibition levels were
achieved by extracts prepared in water or water-based
buffers. Primary extracts in organic solvents (methanol,
ethanol and acetone) contained distinctly less activity: in
HT29 cells, the methanol extract showed a higher growth-
inhibitory activity than the corresponding ethanol extract

(Fig. 1a). In MCF7 cells, methanol and ethanol extracts
showed slight growth-stimulatory effects, which were
antagonized by a growth-inhibitory activity at high concen-
trations only (Fig. 1b). The antineoplastic activity of
primary extracts in water and methanol admixtures
decreased with increasing methanol content (Fig. 1c). The
acetone extract showed only stimulatory activity in the
concentration range tested. Interestingly, this effect was
even increased for the primary extract prepared in an
admixture of 70% acetone in water (Fig. 1d). The influence
of the respective solvents was insignificant, as their effects
did not differ from normal experimental variation (T6%).
The total activity of the primary aqueous extract was
0.6 Â 106
U/g. Secondary extracts in water or water-
based buffers prepared following extraction with ethanol
and acetone showed an almost identical activity (Fig. 1b
and d, 0.5 Â 106
U/g each). The activity of the
secondary extract following methanol extraction was
diminished by 88% (Fig. 1b, 0.07 Â 106
U/g) and that
following 70% acetone extraction was increased by 143%
(Fig. 1d, 1.4 Â 106
U/g).
Antineoplastic activity in a panel of tumor and non-tumor
cell lines
For the cell panel screen, a primary extract in water was
used throughout. An overview on the cell growth
inhibition is given in Table 3. The cell lines tested showed
different sensitivities, as illustrated in Fig. 2 for both solid
tumor (a) and leukemia-derived cell lines (b). The IC50
values ranged by a factor of 100, from 1.7 Ag/ml (MCF7
cells) to 170 Ag/ml (AR230 cells). The origin of a tumor
cell line did not predict its sensitivity, as evidenced by
BV173 (IC50 = 1.8 Ag/ml) and CML-T1 (IC50 = 160 Ag/
ml) chronic myeloid cells. The shape of the respective
concentration–activity curve is reflected by the ratio
between the IC90 and IC10 values (Table 3). A low ratio
represents a dose–response curve with a steep slope, a
high ratio a curve with a lower slope. There was no
obvious relationship between the IC50 value and the slope
of the dose–response curve.
The average IC50 over the whole cell line panel was 49
Ag/ml. A classification of the cell lines as sensitive or less
sensitive was based on the ratio of the individual IC50 to the
average IC50 (Fig. 3). The majority of cell lines (11 out of
17) were classified as sensitive with a degree of sensitivity
that varied over two orders of magnitude. Three of these
sensitive cell lines (MCF7 breast cancer, BV173 CML, and
CC531 rat colon carcinoma) showed a particularly high
sensitivity, with ratios lower than 0.1 of the average IC50.
The IC50 of non-tumor cell lines ranged from 20
(PNT-2) to over 100 (MCF10) Ag/ml (Table 3). Based on
the classification for tumor cell lines, the immortalized
cell lines should be grouped as marginally to less
Fig. 1. Concentration–effect curves of various primary and secondary extracts of the plant powder in HT29 (a) and MCF7 (b–d) cells. The cytotoxic activity
was determined by MTT assay. Each concentration was tested in six replicates. Vertical bars denote standard deviation of the mean.

sensitive (Fig. 3). The growth inhibition of these cell lines
strongly depended on the initial cell density, as shown in
Fig. 2c for MDCK II cells, with low cell densities being
distinctly more susceptible to the extract than high cell
densities.
Variation in exposure duration and recovery
When comparing the dose–response curves after 2, 3,
and 4 days of exposure, the ratio of surviving cells (treated/
control) decreased with increasing exposure duration for
MCF7 breast cancer cells over the whole time period
investigated (Fig. 4a). For U87-MG glioblastoma/astrocy-
toma cells, the treated over control ratio decreased for up to
3 days but remained constant thereafter (Fig. 4b).
In the recovery experiments, concentrations effecting
more than >70% growth inhibition in MCF7 (Fig. 4c) or
!80% in U87-MG (Fig. 4d) cells were associated with
permanent growth suppression. Lower concentrations were
inversely related to an increasing degree of cell prolifer-
ation, indicating recovery.
Comparison with positive controls
A comparison of the antineoplastic activity of the extract
with that of three clinically used agents is given in Table 4.
The cytotoxicity profiles of four cell lines are illustrated by
Table 3
Antiproliferative activity of an aqueous extract from X. americana in 16
human and one rodent tumor cell lines, as well as in four immortalized non-
tumor cell lines
Cell line IC10
a
(Ag/ml
medium)
IC50
b
(Ag/ml
medium)
IC90
c
(Ag/ml
medium)
IC90/IC10
d
(ratio)
Tumor cell lines
MCF7 0.6 1.7 10 16.7
BV173 0.4 1.8 7 17.5
CC531 0.8 3.3 12 15.0
U87-MG 1.0 9 100 100
K562 5 11 180 36
SKW-3 3.1 20 700 226
HEp2 5 21 100 20
NCI-H460 4 21 150 38
PC3 3.5 26 >1000 >300
MDA-MB231 5 33 100 20
HT29 8 40 350 44
U333 7 65 300 43
SAOS2 20 66 1000 50
LAMA84 10 90 600 60
HL60 30 90 1000 33
CML-T1 2.5 160 1000 400
AR230 17 170 700 41
Non-tumor cell lines
MCF10 35 >100 >100 >2
MDCK II 12 27 60 5
NIH/3T3 2 33 >100 >50
PNT-2 2 20 >100 >50
a
Inhibitory concentration 10 (concentration inhibiting the cell growth by
10%), as assessed by MTT assay.
b
c
d
Ratio of IC90 and IC10 values.
Fig. 2. The aqueous extract of the plant powder in adherent (a) and
leukemic (b) tumor cell lines. Four cell lines were selected, respectively,
including those with highest and lowest sensitivity. For comparison,
concentration–effect curves are shown for a non-tumorigenic immortalized
cell line (MDCK) seeded at different cell densities (c). The cytotoxic
activity was determined by MTT assay. Each concentration was tested in six
replicates and each experiment was repeated one or two times. Vertical bars
denote standard error of the mean.

the respective IC10, IC50, and IC90 values, as well as by the
corresponding IC90 to IC10 ratio, describing the slope of the
concentration–effect curve. Most prominently, the ranking
in sensitivity differed between the extract and the positive
controls. In variance to the extract, which resulted in the
lowest IC50 and IC90/IC10 ratio in MCF7 cells, miltefosine
Fig. 3. Sensitivity plot of all cell lines. The ratio of individual and average (based on tumor cell lines only) IC50 was used to rank the panel into more (ratio < 1)
and less (ratio > 1) sensitive cell lines. Cell lines were grouped according to their origin as well as their tumorigenicity. The ratio of the MCF10 cells (*) is an
underestimate, since the IC50 was not reached at 100 Ag/ml.
Fig. 4. Kinetic and recovery experiments based on the aqueous extract of the plant powder in MCF7 and U87-MG cells. (a,b) Concentration–effect curves in
response to the varied exposure duration in MCF7 (a) and U87-MG (b) cells. The cytotoxic activity was determined by MTT assay. Each concentration was
tested in six replicates and each experiment was repeated one or two times. Vertical bars denote standard error of the mean. (c,d) Growth curves of MCF7 (c)
and U87-MG (d) cells following replacement of the extract containing medium after 3 days of exposure with standard growth medium. The absorbance was
determined by MTT assay. Each concentration was tested in six replicates. Vertical bars denote standard deviation of the mean.

and cisplatinum caused the lowest IC50 and IC90/IC10 ratio
in HEp2 cells. Similar to the extract, the lowest IC50
following gemcitabine exposure was seen in MCF7 cells.
However, this agent differed from all others by its lack in
effecting 90% growth inhibition in three cell lines, including
the most sensitive MCF7 cells. The only cells, in which
gemcitabine induced 90% growth inhibition, were the HEp2
cells; notably, these cells were most resistant to this agent. In
contrast, SAOS2 cells were found to be most resistant to the
extract as well as to miltefosine and cisplatinum.
In vivo experiments
As shown in Table 2, the aqueous extract was adminis-
tered to male tumor-bearing Wag Rij rats via the peroral and
the intraperitoneal route. Preliminary experiments had
shown that the maximum tolerated single dose was 30
mg/kg following i.p. administration. Higher dosages caused
weight loss, hemorrhage, and acute death within 24 h. In
contrast, no toxicity was observed following dosages of up
to 100 mg/kg given p.o.
Peroral administration of 100 mg/kg every second day
effectively reduced the increase in liver weight seen in
untreated tumor-bearing rats (T/C% = 34.0, P < 0.05, Table
2). Concomitantly, the number of tumor cells was
significantly lower in treated rats as compared to the
respective controls (T/C% = 31.2, P < 0.05, Table 2).
Interestingly, intraperitoneal administration of 5 mg/kg
every second day was found to be less effective regarding
the liver weight (T/C% = 62.0, Table 2), but similarly
effective with regard to tumor cell number (T/C% = 34.0,
P < 0.05, Table 2).
Physicochemical characterization of the biological activity
The variation in antineoplastic activity between three
batches of plant material was determined in MCF7 breast
cancer cells. The most recent batch (C) was characterized by
the lowest IC50 value (0.9 Ag/ml, TA = 1.1 Â 106
U/g),
followed by the first batch (A, IC50 = 1.1 Ag/ml, TA = 0.9 Â
106
U/g) and the second batch (B, IC50 = 1.7 Ag/ml; TA =
0.6 Â 106
U/g). Storage of the dry material for up to 2 years
at 4 -C did not change the level of the antineoplastic
activity.
The aqueous extract was stable for up to 2 years at
À20 -C but lost 20% of its antineoplastic activity after 24 h
at 4 -C, as determined in HT29 colon carcinoma and MCF7
breast cancer cells.
Incubation for 10 or 30 min at 50 -C resulted in a marked
loss of activity as shown by IC50 values of 2.5 and 5.0 Ag/
ml, respectively, compared to 1.2 for the untreated extract
(as determined in MCF7 breast cancer cells). A complete
loss of the antineoplastic activity was observed after 10 min
incubation at 90 -C.
Alterations in pH had also a time-dependent effect on the
biological activity of the extract. Short-term exposure (10
min) to pH values of 1.5–12 had no effect. After 2 h, the
extracts exposed to pH values < 5 lost part of their activity
whereas that of the alkaline extracts remained unchanged.
Exposure for 24 h to both acids and alkalis led to a complete
loss of the antineoplastic activity.
Precipitation caused by addition of organic solvents
(methanol, ethanol, isopropanol) to the aqueous extract was
associated with pelleting of the entire biologic activity.
The filtrate obtained after aqueous extract ultrafiltration
through membranes with 10 kDa cutoff was devoid of any
activity. The antineoplastic activity was completely found in
filtrates obtained by ultrafiltration through a 100-kDa cutoff
membrane and partially in that through a 50-kDa cutoff
membrane.
Confining the active compound(s) to a substance class
SDS-PAGE of the raw aqueous extract with subsequent
staining for protein (Coomassie blue) showed this to contain
proteins as well as a Coomassie-negative smear in the high
molecular weight domain. This smear as well as some of the
protein bands were colored by Schiff’s reagent, demonstrat-
ing the presence of polysaccharides in the smear as well as
of glycosylated proteins. Incubation of the raw aqueous
extract with trypsin or proteinase K, however, did not result
in a diminished biological activity.
Tannins were also identified in the raw aqueous extract
by reverse-phase HPLC and were most efficiently
extracted by 70% acetone. This extract showed no
antineoplastic activity in MCF7 cells (Fig. 1d). The
secondary extract, prepared after pre-extraction with 70%
acetone, was devoid of tannins and showed a >2-fold
increased biological activity as compared to the raw
aqueous extract (TA = 1.4 Â 106
and 0.6 Â 106
U/g for
the secondary and primary aqueous extracts, respectively).
Incubation of this secondary extract with trypsin or
proteinase K led to a 2-fold (trypsin) to 3-fold (proteinase
Table 4
Cytotoxicity profiles of the extract and three standard antineoplastic agents
in a subpanel of cell lines
Cell line Treatment IC10 IC50 IC90 IC90/IC10
MCF7 Extract (Ag/ml) 0.6 1.8 10 16.7
Miltefosine (AM) 6.5 40 80 12.3
Cisplatinum (Ag/ml) 0.22 2.2 10 45
Gemcitabine (AM) 0.001 0.012 >100 >105
U87-MG Extract (Ag/ml) 1.0 9 100 100
Miltefosine (AM) 4.7 27 70 14.9
Gemcitabine (AM) 0.002 0.014 >100 >5 Â 104
HEp2 Extract (Ag/ml) 5 21 100 20
Miltefosine (AM) 1.2 2.8 8 6.7
Cisplatinum (Ag/ml) 0.09 0.4 1.4 15.6
Gemcitabine (AM) 0.2 0.47 17 85
SAOS2 Extract (Ag/ml) 20 66 1000 50
Miltefosine (AM) 5.0 40 120 24
Gemcitabine (AM) 0.007 0.034 >100 >104

K) decrease in antineoplastic activity in MCF7 cells (24
T/C% for trypsin, 38 T/C% for proteinase K vs. 12 T/C%
for the undigested extract, for an extract concentration of
10 Ag/ml medium).
Affinity to galactose
Hydrolysis of the cross-linked galactose polysaccharide
Sepharose 4B was used to prepare an affinity matrix
containing free terminal galactose residues. Incubation of
the aqueous extract with this matrix resulted in an almost
complete depletion of the biological activity from the
supernatant (T/C = 80% after Sepharose incubation vs.
18% before, as determined for an extract concentration of 2
Ag/ml in MCF7 cells). This finding led to the assumption
that the active component(s) had bound to the matrix.
Indeed, after separating the supernatant and washing the
hydrolyzed Sepharose gel, an active fraction could be eluted
by using buffer with 100 mM galactose. A non-reducing
SDS-PAGE analysis of this fraction revealed the presence of
two proteins with a molecular weight (MW) of about 50–60
kDa (Fig. 5a). Under reducing conditions, four bands were
seen within the MW range 25–35 kDa (Fig. 5b), hinting at a
two-chain structure for both proteins. The protein with the
higher mass was subsequently processed for mass-spectro-
metric analysis.
A database search based on the mass-spectroscopic
results failed to identify a known protein in the database.
However, one tryptic peptide was identified by its identity to
a peptide contained by the ribosome inactivating protein
ricin with the sequence SNTDANQLWTLK.
Identification of the source plant
In a phylogenetic analysis, the new rbcL sequence was
found to cluster together with the rbcL sequence from
Heisteria parvifolia from the Olacaceae family. A compar-
ison with unpublished data (courtesy of Dr. D.L. Nickrent,
Department of Plant Biology, Southern Illinois University,
Carbondale, IL, USA) showed a close homology to the
Ximenia americana rbcL sequence. After obtaining fresh
plant material from X. americana of US origin as well as
DNA from X. americana of African origin and sequencing
the respective rbcL genes, this plant was confirmed to be the
source of the powder used in this study.
Biological activity of X. americana extracts
Aqueous extracts prepared from powdered dry leaves
from X. americana of US origin showed no antineoplastic
effect in MCF7 breast cancer cells. Conversely, a strong
growth inhibition of MCF7 cells was seen (IC50 = 33 ng/ml,
TA = 30.3 Â 106
U/g) following treatment with aqueous
extracts prepared from powdered dry nuts from X. ameri-
cana of US origin. As for the African plant material, a
higher biological activity was seen for the secondary
aqueous extract prepared from US X. americana nut
powder, pre-extracted with 70% acetone (IC50 = 23 ng/ml,
TA = 43.5 Â 106
U/g).
Discussion
A lot of work is currently being invested into exploiting
the drug potential of Chinese plants, particularly of those
used in Chinese traditional medicine. However, information
about plants used in African traditional medicine is still
scarce. This is related to different ways of propagating
medical experience: whereas the Chinese ethnopharmaco-
logical knowledge has been passed on for thousands of
years based on written information, the traditional African
medicine’s mode of transmission by word of mouth has
hindered systematic scientific investigation (Okpako, 1999).
Case reports on the improvement of cancer patients after
treatment with a plant material used in African traditional
medicine prompted in vitro tests of this powder for antineo-
plastic activity. Since the patients had used the powder by
suspending it in water and ingesting the supernatant, a water
extract was initially used. A surprisingly high growth
inhibition upon extract treatment was found in two tumor-
derived cell lines. This observation gave rise to an extended
study on the aqueous extract’s effect in 17 tumor cell lines
originating from eight different organs/tissues.
Although no typical sensitivity of a certain organ was
discernible, the differences in sensitivity between cell lines
ranked as ‘‘highly sensitive’’ (MCF7 and BV173) and those
classified as ‘‘less sensitive’’ (AR230, CML-T1) point to the
fact that the cytotoxic activity of the aqueous extract is
selective. However, the molecular basis for this selectivity is
so far unknown and is not directly related to the tumor-
igenicity of cell lines, since some of the immortalized, non-
tumorigenic cell lines were ranked as marginally sensitive.
Interestingly, immortalized, non-tumorigenic cells showed a
Fig. 5. SDS-PAGE of the cytotoxic, affinity-purified protein fraction (lanes
G) compared to the raw aqueous extract (lane E) under reducing (a) or non-
reducing (b) conditions. M—molecular weight marker (kDa).

3- to 4-fold higher sensitivity when seeded as single cells as
compared with higher densities. We speculate that this
phenomenon is related to the proliferation rate, since the
slowly proliferating MCF10 cells were the most resistant of
this subgroup.
The slope of the dose–response curves, as reflected by
the IC90/IC10 ratio, differed also by factor of 25. Moreover,
the shape of these curves showed a shoulder in some cases
(AR230) or a growth stimulation at low concentrations
(SAOS2). Both observations could reflect such diverse
properties as stimulation concomitant with cytocidal effects
both caused by extract components, or subpopulations of
more resistant cells within the same cell line.
The IC50 averaged over all cell lines was surprisingly
low, as compared with other plant extracts tested in our
laboratory (unpublished results). Remarkably, the aqueous
extract prepared from X. americana nuts showed an IC50
almost 100-fold lower than that of the powder extract in
MCF7 cells. This finding leads to the assumption that the
powder obtained from Tanzania was blended with plant
parts containing low cytotoxic activity.
To asses the kinetics of the cytotoxic activity, exposure
duration and recovery experiments were performed in two
cell lines. The decreasing ratio of surviving cells with in-
creasing exposure duration describes an improved effective-
ness of the extract treatment with time. This effect might be
related to various cellular processes triggered by the aqueous
extract, which finally lead to cell death or growth delay. The
highly sensitive MCF7 cells showed this effect for a longer
period than the less sensitive U87-MG cells. In addition, both
MCF7 and U87-MG cells showed no recovery after being
exposed to sufficiently high extract concentrations.
The ‘‘signature’’ of the aqueous extract’s activity in four
cell lines differed from that of three standard chemo-
therapeutic drugs used as positive controls. This implies
that the mechanism of action differs from that of these
agents, including alkylating agents, antimetabolites, and
signal transduction modifiers.
In addition to the antineoplastic activity, a growth
stimulation was seen following exposure to the primary
aqueous extract in three cell lines (MCF7, SAOS2, and
HL60). In MCF7 cells, this stimulating effect was increased
in response to extracts prepared in ethanol, methanol, and
acetone. The highest increase in MCF7 cell proliferation
was observed after treatment with an extract containing a
high tannin concentration (70% acetone primary extract).
Moreover, when separating by size, the stimulating compo-
nents passed through a 10-kDa cutoff membrane, proving
the stimulating components to be physically distinct from
the cytotoxic components. The fact that none of these
extracts had any stimulatory activity in HT29 cells leads us
to speculate that the growth stimulation observed in MCF7
cells was caused by compounds with hormonal activity,
such as phyto-estrogens (De Naeyer et al., 2005).
Concomitantly with the cell line screen, the stability of
the plant material and that of the aqueous extract was
monitored. The biological activity of the powdered plant
material was stable throughout the period of the cell line
screen, as shown by the reproducibility of the results.
Moreover, different batches of plant material effected
comparable degrees of growth inhibition. Aqueous extracts,
however, lost activity in a time-dependent manner when
stored at 4 -C and were therefore always stored at À20 -C.
Since the rat colon carcinoma cell line CC531 was ranked
among the three most sensitive cell lines in the panel, it was
tempting to determine whether any antineoplastic activity
would be discernible in a rat model based on this cell line.
The significant antineoplastic activity observed following
peroral administration was surprising since no concomitant
toxicity was detected. This observation is furthermore in line
with the case reports from Tanzania. A 20-fold lower, equi-
toxic dosage administered i.p. was less effective in moderat-
ing the liver weight increase, but equally effective in
controlling the increase in tumor cell number. Several factors
could be responsible for this difference, including degrada-
tion within the gastrointestinal tract, incomplete gastro-
intestinal absorption, differences in tissue distribution,
metabolism, and excretion. Future pharmacokinetic studies
are needed to determine the reason for the observed differ-
ence between peroral and parenteral administration routes.
The promising results of both the in vitro screen and the
in vivo experiments triggered studies aiming at the identi-
fication of the source plant and the active component(s).
The phylogenetic analysis used for plant identification,
allowing the identification of the presumed plant family,
Olacaceae, took advantage of the broad information
available on the rbcL gene sequence in public databases.
The assumed source species, X. americana, was confirmed
by comparison to rbcL sequences from fresh X. americana
plant material, as well as by finding a high cytotoxic activity
in plant parts subsequently obtained from this species.
In order to define the substance class of the active
component(s), experiments were carried out on their
physicochemical properties, including solubility, stability at
varying pH and temperature, size estimation, and precip-
itation. In this process, lipids and lipophilic plant secondary
metabolites could be excluded, since the biological activity
was only extracted by strongly polar solvents.
Large amounts of tannins were identified in the aqueous
extract. However, extracts prepared in methanol or 70%
acetone, both solvents known to efficiently extract tannins
from plant materials, had only a low (methanol) or no (70%
acetone) cytotoxic activity. Remarkably, depletion of tannins
by pre-extraction with 70% acetone led to the release of a
higher total biological activity in the secondary aqueous
extract, when compared to a respective primary extract. This
implies that the biological activity is partially inhibited or
hidden by tannins or other components extracted by 70%
acetone.
Molecules smaller than 10 kDa were excluded by
ultrafiltration experiments, which also allowed estimating
the MW of the active components to be in the range of 50 T

20 kDa. Out of the known classes of plant cell macro-
molecules, DNA and RNA were not found in the aqueous
extract and digestion experiments with DNase or RNase had
no effect on the biological activity. However, proteins and
polysaccharides were shown to be present in the primary
and the secondary aqueous extracts and could not be further
separated by physicochemical methods. Finally, digestion
experiments with trypsin and proteinase K hinted at a
protein being responsible for the cytotoxic activity.
A well-defined family of cytotoxic plant proteins is that
of the type II ribosome-inactivating proteins (RIPs; Battelli,
2004). These proteins with a molecular weight of about 60
kDa consist of two polypeptide chains, termed A- and B-
chain, with an MW of about 30 kDa each, being held
together by a disulphide bridge. The A-RIP chain has the
function of an RNA N-glycosidase and is able to inhibit
protein synthesis in a cell-free lysate. The B-chain has the
function of a lectin with affinity for galactose or other sugar
residues. Binding of the B-chain to oligosaccharides on the
cell membrane surface elicits internalization of the RIP and
results in cytotoxicity (Barbieri et al., 1993).
Based on this background, the biologically active extract
components were tested for their affinity to galactose and
found to bind to Sepharose containing free galactose ends.
When eluted with galactose-containing buffer, the galactose-
binding fraction was associated with high cytotoxic activity.
This fraction consisted of only two proteins (Fig. 5).
The cytotoxic effects, the MW, as well as the two-chain
structure of the proteins in the affinity-purified fraction can
be related to the molecular characteristics of known members
of the type II RIP family. In keeping with this assumption,
one of the mass-spectrometrically sequenced tryptic pep-
tides, originating from one of these proteins, showed identity
with a tryptic peptide in the B-chain of the type II RIP ricin.
However, the lack of further hits when comparing the MS
spectrum with the database implies that the protein is largely
unknown. This is corroborated by the fact that no RIPs are
known to date to exist in the plant X. americana, which has
been identified as the source of the powder used. In
conclusion, this cumulative evidence strongly suggests that
the active components of the plant material are so far
unknown proteins belonging to the type II RIP family.
Acknowledgments
We are indebted to Dr. D.L. Nickrent, Department of
Plant Biology, Southern Illinois University, Carbondale, IL,
USA, for comparing our rbcL sequence with his unpub-
lished data and for providing the rbcL sequence of Ximenia
americana collected in the Bahamas.
We also thank Dr. W. Lehmann, Central Spectroscopy
Unit, DKFZ, Heidelberg, for the mass spectroscopical
analysis of the protein bands and Dr. B. Owen, Department
of Toxicology and Cancer Risk Factors, DKFZ, Heidelberg,
for analyzing our samples by HPLC for tannins.
Finally we are grateful to Captain A. Kurzenhauser
(retired, US Navy), Tampa, FL, USA, for his invaluable help
in tracing and collecting Ximenia americana.
References
Adam, K.P., Becker, H., 2000. Peptide–proteine. In: Adam, K.P., Becker,
H. (Eds.), Analytik Biogener Arzneistoffe. Wissenschaftliche Verlags-
gesellschaft, Stuttgart, pp. 470–474.
Barbieri, L., Battelli, M.G., Stirpe, F., 1993. Ribosome-inactivating proteins
from plants. Biochim. Biophys. Acta 1154, 237–282.
Battelli, M.G., 2004. Cytotoxicity and toxicity to animals and humans of
ribosome-inactivating proteins. Min. Rev. Med. Chem. 4, 513–521.
De Naeyer, A., Pocock, V., Milligan, S., De Keukeleire, D., 2005.
Estrogenic activity of a polyphenolic extract of the leaves of Epimedium
brevicornum. Fitoterapia 76, 35–40.
De Smet, P.A., 1997. The role of plant-derived drugs and herbal medicines
in healthcare. Drugs 54, 801–840.
Kantarjian, H.M., Talpaz, M., Santini, V., Murgo, A., Cheson, B., O’Brien,
S.M., 2001. Homoharringtonine: history, current research, and future
direction. Cancer 92, 1591–1605.
Kinter, M., Sherman, N.E., 2000. An in-gel digestion protocol. In:
Desiderio, D.M., Nibbering, M.M. (Eds.), Protein Sequencing and
Identification Using Tandem Mass Spectrometry. Wiley-Interscience,
New York, pp. 152–160.
Konstantinov, S.M., Topashka-Ancheva, M., Benner, A., Berger, M.R.,
1998. Alkylphosphocholines: effects on human leukemic cell lines and
normal bone marrow cells. Int. J. Cancer 77, 778–786.
Okpako, D.T., 1999. Traditional African medicine: theory and pharmacol-
ogy explored. Trends Pharmacol. Sci. 20, 482–485.
Owen, R.W., Haubner, R., Hull, W.E., Erben, G., Spiegelhalder, B.,
Bartsch, H., Haber, B., 2003. Isolation and structure elucidation of the
major individual polyphenols in carob fibre. Food Chem. Toxicol. 41,
1727–1738.
Perkins, D.N., Pappin, D.J., Creasy, D.M., Cottrell, J.S., 1999. Probability-
based protein identification by searching sequence databases using mass
spectrometry data. Electrophoresis 20, 3551–3567.
Saenger, J., Leible, M., Seelig, M.H., Berger, M.R., 2004. Chemoemboli-
zation of rat liver metastasis with irinotecan and quantification of tumor
cell reduction. J. Cancer Res. Clin. Oncol. 130, 203–210.
Spjut, R.W., Perdue Jr., R.E., 1976. Plant folklore: a tool for predicting
sources of antitumor activity? Cancer Treat. Rep. 60, 979–985.
Tang, W., Hemm, I., Bertram, B., 2003a. Recent development of antitumor
agents from chinese herbal medicines: Part I. Low molecular com-
pounds. Planta Med. 69, 97–108.
Tang, W., Hemm, I., Bertram, B., 2003b. Recent development of antitumor
agents from Chinese herbal medicines: Part II. High molecular
compounds. Planta Med. 69, 193–201.
Treutlein, J., Wink, M., 2002. Molecular phylogeny of cycads inferred from
rbcL sequences. Naturwissenschaften 89, 221–225.
Wada, H., Mizutani, S., Nishimura, J., Usuki, Y., Kohsaki, M., Komai, M.,
Kaneko, H., Sakamoto, S., Delia, D., Kanamaru, A., 1995. Establish-
ment and molecular characterization of a novel leukemic cell line with
Philadelphia chromosome expressing p230 BCR/ABL fusion protein.
Cancer Res. 55, 3192–3196.
Wittmer, A., Khazaie, K., Berger, M.R., 1999. Quantitative detection of lac-
Z-transfected CC531 colon carcinoma cells in an orthotopic rat liver
metastasis model. Clin. Exp. Metastasis 17, 369–376.
Xiao, Z., Hao, Y., Liu, B., Qian, L., 2002. Indirubin and meisoindigo in the
treatment of chronic myelogenous leukemia in China. Leuk. Lymphoma
43, 1763–1768.
Zacharius, R.M., Zell, T.E., Morrison, J.H., Woodlock, J.J., 1969.
Glycoprotein staining following electrophoresis on acrylamide gels.
Anal. Biochem. 30, 148–152.

Voss et al. - 2006 - Identification of potent anticancer activity in Xi

More Related Content

What's hot (20)

Viewers also liked (15)

Similar to Voss et al. - 2006 - Identification of potent anticancer activity in Xi (20)

Voss et al. - 2006 - Identification of potent anticancer activity in Xi