Molecular docking and pharmacogenomics of vinca alkaloids and their monomeric precursors, vindoline and catharanthine.

Abstract

Vinblastine and vincristine are dimeric indole alkaloids derived from Catharanthus roseus (formerly: Vinca rosea). Their monomeric precursor molecules are vindoline and catharanthine. While vinblastine and vincristine are well-known mitotic spindle poisons, not much is known about vindoline and catharanthine. Vindoline and catharanthine showed weak cytotoxicity, while vinblastine, vincristine, and the semisynthetic vindesine and vinorelbine revealed high cytotoxicity towards cancer cells. This may reflect a general biological principle of poisonous plants. Highly toxic compounds are not only active towards predators, but also towards plant tissues. Hence, plants need mechanisms to protect themselves from their own poisons. One evolutionary strategy to solve this problem is to generate less toxic precursors, which are dimerized to toxic end products when needed. As shown by in silico molecular docking and biochemical approaches, vinblastine, vincristine and vinorelbine bound with high affinity to α/β-tubulin and inhibited tubulin polymerization, whereas the effects of vindoline and catharanthine were weak. Similarly, vinblastine produced high fractions of mono- and multipolar mitotic spindles, while vindoline and catharanthine did only weakly affect bipolar mitotic spindle formation. Here, we show that vinblastine contributes to cell death by interference with spindle polarity. P-glycoprotein-overexpressing multidrug-resistant CEM/VCR1000 cells were highly resistant towards vincristine and cross-resistant to vinblastine, vindesine, and vinorelbine, but not or only weakly cross-resistant to vindoline and catharanthine. In addition to tubulin as primary target, microarray-based mRNA signatures of responsiveness of these compounds have been identified by COMPARE and signaling pathway profiling.

Full text

Page 1

Accepted Manuscript
Title: Molecular docking and pharmacogenomics of Vinca
alkaloids and their monomeric precursors; vindoline and
catharanthine
Authors: Serkan Sertel, Yujie Fu, Yuangang Zu, Blanka
Rebacz, Badireenath Konkimalla, Peter K. Plinkert, Alwin
Kr¨ amer, J¨ urg Gertsch, Thomas Efferth
PII:
DOI:
Reference:
S0006-2952(11)00013-X
doi:10.1016/j.bcp.2010.12.026
BCP 10794
To appear in:
BCP
Received date:
Revised date:
Accepted date:
17-11-2010
23-12-2010
24-12-2010
Please cite this article as: Sertel S, Fu Y, Zu Y, Rebacz B, Konkimalla B, Plinkert PK,
Kr¨ amer A, Gertsch J, Efferth T, Molecular docking and pharmacogenomics of Vinca
alkaloids and their monomeric precursors; vindoline and catharanthine, Biochemical
Pharmacology (2010), doi:10.1016/j.bcp.2010.12.026
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
Themanuscriptwillundergocopyediting,typesetting,andreviewoftheresultingproof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
peer-00672298, version 1 - 21 Feb 2012
Author manuscript, published in "Biochemical Pharmacology 81, 6 (2011) 723"
DOI : 10.1016/j.bcp.2010.12.026

Page 2

Page 1 of 42
Accepted Manuscript
Abbreviations: ABC transporter, ATP-binding cassette transporter; HNSCC, head and neck
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Molecular docking and pharmacogenomics of Vinca alkaloids and
their monomeric precursors, vindoline and catharanthine

Serkan Sertel a, e, g, Yujie Fu b, c, Yuangang Zu b, c, Blanka Rebacz d, Badireenath Konkimallae,
Peter K. Plinkert a, Alwin Krämer d, Jürg Gertsch e and Thomas Efferth e, f, g, *

a Department of Otorhinolaryngology, Head and Neck Surgery, University of Heidelberg,
Heidelberg, Germany
b Key Laboratory of Forest Plant Ecology, Ministry of Education, Northeast Forestry
University, Harbin, China
c Engineering Research Center of Forest Bio-Preparation, Ministry of Education, Northeast
Forestry University, Harbin, China
d Clinical Cooperation Unit Molecular Hematology/Oncology, German Cancer Research
Center and Department of Internal Medicine V, University of Heidelberg, Heidelberg,
Germany
e Pharmaceutical Biology (C015), German Cancer Research Center, Heidelberg, Germany
f Institute of Biochemistry and Molecular Medicine, University of Bern, Bern, Switzerland
g Department of Pharmaceutical Biology, Institute of Pharmacy and Biochemistry, University
of Mainz, Mainz, Germany


* Corresponding Author: Department of Pharmaceutical Biology, Institute of Pharmacy and
Biochemistry, University of Mainz, Staudinger Weg 5, 55128 Mainz, Germany.
Phone: +49-6131-3925751; Fax: +49-6131-3923752; E-mail: efferth@uni-mainz.de


Running Title: Pharmacogenomics of Vinca Alkaloids

Keywords: centrosomal clustering, molecular docking, multidrug resistance,
pharmacogenomics, vinca alkaloids

squamous cell carcinoma; IC50, 50% inhibition concentration; RMSD, root mean square
deviations; mRNA, messenger RNA; NCI, national cancer institute
peer-00672298, version 1 - 21 Feb 2012

Page 3

Page 2 of 42
Accepted Manuscript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

2
Abstract

Vinblastine and vincristine are dimeric indole alkaloids derived from Catharanthus roseus
(formerly: Vinca rosea). Their monomeric precursor molecules are vindoline and
catharanthine. While vinblastine and vincristine are well-known mitotic spindle poisons, not
much is known about vindoline and catharanthine. Vindoline and catharanthine showed weak
cytotoxicity, while vinblastine, vincristine, and the semisynthetic vindesine and vinorelbine
revealed high cytotoxicity towards cancer cells. This may reflect a general biological
principle of poisonous plants. Highly toxic compounds are not only active towards predators,
but also towards plant tissues. Hence, plants need mechanisms to protect themselves from
their own poisons. One evolutionary strategy to solve this problem is to generate less toxic
precursors, which are dimerized to toxic end products when needed. As shown by in silico
molecular docking and biochemical approaches, vinblastine, vincristine and vinorelbine
bound with high affinity to -tubulin and inhibited tubulin polymerization, whereas the
effects of vindoline and catharanthine were weak. Similarly, vinblastine produced high
fractions of mono- and multipolar mitotic spindles, while vindoline and catharanthine did
only weakly affect bipolar mitotic spindle formation. Inhibition of centrosomal clustering
represents a treatment novel strategy leading to multipolar spindle formation and apoptosis.
This has been shown for vinblastine by us for the first time. P-glycoprotein-overexpressing
multidrug-resistant CEM/VCR1000 cells were highly resistant towards vincristine and cross-
resistant to vinblastine, vindesine, and vinorelbine, but not or only weakly cross-resistant to
vindoline and catharanthine. In addition to tubulin as primary target, microarray-based mRNA
signatures of responsiveness of these compounds have been identified by COMPARE and
signaling pathway profiling.
peer-00672298, version 1 - 21 Feb 2012

Page 4

Page 3 of 42
Accepted Manuscript
less toxic precursor molecules for self-protection, whereas the final synthesis of highly toxic
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

3
1. Introduction

Plants produce secondary metabolites as defense weapons against microbial infections by
viruses, bacteria, or protozoa and parasites such as insects or worms as well as against
herbivores. Many plants are poisonous, while others can serve as medicinal plants with
pharmacological activity. As shown in a previous survey conducted by the National Cancer
Institute (NCI), USA, more than two thirds of all anticancer drugs established in anticancer
therapy are natural products, derivatives of natural products or mimic bioactive principles of
natural products [1].
Among the clinically established natural products with anticancer activity are the
Vinca alkaloids vinblastine and vincristine and more recently, the semi-synthetic derivatives
vindesine and vinorelbine, which are highly useful drugs for the treatment of certain
malignancies.

Vinca alkaloids arrest tumor cells during mitosis by binding to tubulin and
depolymerization of microtubules [2]. This leads to cell cycle arrest in mitosis [3]. Besides
interaction of Vinca alkaloids with tubulins, other mechanisms upstream (e.g. membrane-
bound drug efflux transporters) and downstream (e.g. signal transduction pathways,
programmed cell death) also account to the drugs’ efficacy towards cancer cells.
Vinblastine and vincristine are dimeric indole alkaloids derived from Catharanthus
roseus (formerly: Vinca rosea). Their monomeric precursor molecules are vindoline and
catharanthine. While there is clear evidence for the action of vinblastine and vinblastine as
mitotic spindle poisons, not much is known about the monomers vindoline and catharanthine.
Both precursor molecules are less cytotoxic than their dimeric drugs, vinblastine and
vincristine. The question arises, whether this reflects a biological principle of poisonous
plants. Poisonous natural products such as vinblastine and vincristine are effective defence
mechanisms against herbivores and other predators. However, these compounds may also
reveal toxicity to the plants themselves. Hence, they may generate and store large amounts of
end products occurs only upon appropriate external stimulation.
In the present investigation, we hypothesized that different cytotoxicities of
monomeric precursors and dimeric end products should affect binding to the primary target of
Vinca alkaloids, the microtubules. In addition, dimeric second-generation drugs, the
semisynthetic vindesine and vinorelbine have been included in the study. A comparative
analysis of functional effects of the above mentioned compounds on microtubule formation
peer-00672298, version 1 - 21 Feb 2012

Page 5

Page 4 of 42
Accepted Manuscript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

4
(effect on α/β-tubulin polymerisation) has been carried out in vitro. The relative binding
affinities of vindoline and catharanthine were estimated from Dixon plots assuming that all
compounds either directly or indirectly (allosteric modulation) interfere with the [3H]-
vinblastine binding sites in α/β-tubulin. The experimental data have been compared to
molecular modelling studies. The binding of Vinca alkaloids to tubulin may not only cause
inhibition of microtubule elongation, but may also affect mitotic spindle formation. The
formation of multipolar mitotic spindles by inhibition of centrosomal coalescence has been
anticipated as novel treatment strategy [4,5]. Therefore, we have analyzed the capacity of
Vinca alkaoids to induce multipolar mitotic spindles. Finally, we have analyzed the role of
drug resistance mechanisms for momomeric and dimeric Vinca alkaloids. We first analyzed
cross-resistance of vincristine-resistant CEM/VCR1000 leukemia cells towards vindoline and
catharanthine in comparison to vinblastine, vindesine, and vinorelbine. Then, we have
analyzed other determinants of responsiveness towards Vinca alkaloids in the cell line panel
of the NCI by means of COMPARE-analyses of microarray-based transcriptome-wide mRNA
expression.


peer-00672298, version 1 - 21 Feb 2012

Page 6

Page 5 of 42
Accepted Manuscript
catharanthine, or vinblastine were added to the cell culture medium for 24 h. In all
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

5
2. Material and Methods

2.1. Compounds

Vindoline and cantharanthine were isolated from Cataranthus roseus as described [6].
Vinblastine sulphate, vincristine sulphate, vindesine sulphate salt, and vinorelbine ditartrate
salt vindesine were obtained from Sigma-Aldrich (Taufkirchen, Germany). Vindoline and
catharanthine are precursor molecules in the biosynthesis route, while vinblastine and
vincristine are end products (Figure 1). The entire biosynthesis pathway has previously been
elucidated [7]. Vindoline and catharanthine were isolated from Cataranthus roseus by two of
the authors (YF and YZ). Vindesine and vinorelbine are semi-synthetic derivatives and were
obtained from Sigma-Aldrich (Taufkirchen, Germany).

2.2. Cell lines

Human CCRF-CEM leukemia cells were maintained in RPMI medium (Gibco, Eggenstein,
Germany) supplemented with 10% fetal calf serum in a humidified 7% CO2 atmosphere at
37°C. Cells were passaged twice weekly. All experiments were performed with cells in the
logarithmic growth phase. The multidrug resistance gene 1 (ABCB1, MDR1)-expressing
CEM/VCR1000 subline was maintained in 1000ng/mL vincristine. The establishment of the
resistant subline has been described [8]. Sensitive and resistant cells were kindly provided by
Dr. A. Sauerbrey (Dept. of Pediatrics, University of Jena, Jena, Germany).
The HNSCC cell line SCC114 (oral squamous cell carcinoma) cells were cultured in
Dulbecco’s modified Eagle’s Medium (DMEM, Gibco, Invitrogen, Karlsruhe, Germany)
supplemented with 10 % FCS (Biochrom AG, Berlin, Germany). When indicated, vindoline,
experiments, the final DMSO concentration was <1%.
The panel of 60 human tumor cell lines of the Developmental Therapeutics Program
of the NCI consisted of leukemia, melanoma, non-small cell lung cancer, colon cancer, renal
cancer, and ovarian cancer cells, cells of tumors of the central nervous system, prostate
carcinoma, and breast cancer. Their origin and processing have been previously described [9].

peer-00672298, version 1 - 21 Feb 2012

Page 7

Page 6 of 42
Accepted Manuscript
adjusted to pH 6.9 with KOH) plus ATP (1.5 mM final concentration) and glycerol (98%) ad
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

6

2.3. Sulforhodamine B assay

The determination of drug sensitivity in the NCI cell lines by the sulforhodamine B assay has
been reported [10]. The 50% inhibition concentration (IC50) values for vinblastine, vincristine,
catharanthine, and vindoline have been deposited in the database of the database of the
Developmental Therapeutics Program of the NCI (http://dtp.nci.nih.gov).

2.4. Growth inhibition assay

The in vitro response to drugs was evaluated by means of a growth inhibition assay as
described [10]. Aliquots of 5×104 cells/ml were seeded in 24-well plates and compounds were
immediately added at different drug concentrations to allow calculation of 50% inhibition
concentration (IC50) values. Cells were counted seven days after drug treatment. The resulting
growth data represent the net outcome of cell proliferation and cell death.

2.5. Preparation of pure α/β-tubulin (>95%)

Tubulin was isolated as pure α/β-tubulin from fresh pig brain according to a previously
described method [11]. Fresh brains were obtained from the local slaughterhouse and
processed immediately without prior cooling. In brief, 150-200 g cleaned pig brain was put
into ice-cooled depolymerization buffer (50 mM MES, 1 mM CaCl2, adjusted to pH 6.9 with
KOH) and homogenized in a Polytron mixer. The homogenate was centrifuged in a Sorvall
SLA-1500 rotor at 14,500 rpm for 60 min. The supernatant was transferred into an
Erlenmeyer flask in high-molar PIPES-buffer (1 M PIPES, 10 mM MgCl2, 20 mM EGTA
300 mL. The resulting suspension was mixed and incubated at 37°C for 1 h. Aliquots were
transferred into ultracentrifuge tubes and centrifuged in a Beckman Ti50.2 rotor at 32,500
rpm (96,000×g) for 75 min at 30°C. The microtubule protein pellets were suspended in
depolymerization buffer and put on ice prior to ultracentrifugation at 4°C. The procedure was
repeated for two polymerization cycles (total of three cycles) and the final α/β-tubulin pellets
were suspended in ice-cold Brinkley Buffer (BRB80; 80 mM PIPES, 1 mM MgCl2, 1 mM
peer-00672298, version 1 - 21 Feb 2012

Page 8

Page 7 of 42
Accepted Manuscript
curves. All experiments, along with both negative (untreated α/β-tubulin) and vehicle
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

7
EGTA adjusted to pH 6.8 with KOH) prior to shock-freezing in liquid nitrogen and
subsequent storage at -80°C. Purity and concentration of α/β-tubulin were determined by
SDS-PAGE gel electrophoresis and spectrophotometrically (A=ε . c . d with a given
extinction coefficient of 115,000 M-1 cm-1) at 280 nm. This procedure typically yielded 60-
100 mg of α/β-tubulin per 100 g of brain.

2.6. Radioligand displacement α/β-tubulin

Tubulin in BRB80 buffer (0.5 mL, 5-10 µM) was incubated with two different concentrations
of [G-3H]-vinblastine (American Radiolabeled Chemicals, Inc.) (5 and 10 μM) on ice for 30
min. Different concentrations of unlabeled vindoline, catharanthine, vincristine, vindesine, or
vinorelbine were added and the mixture was incubated at 37°C for 30 min. DMSO was used
as vehicle control. Bound radioactivity was separated by gel filtration on a Sephadex G50
column on micro columns and measured in a scintillation counter.

2.7. Turbidimetry analysis of polymer formation

Freshly thawed α/β-tubulin was centrifuged at 5,000×g for 5 min at 5°C and then incubated
with additional BRB80 buffer, drugs, DMSO vehicle, GTP/glutamate (25 mM/2.7 M) and/or
MAPs, respectively, in a 96-well plate in either 50 or 100 μL volumes. Experiments were
either carried out in a 96-well quartz plate or in normal 96-well polystyrene plates (Falcon).
The polymerization was monitored real-time at 340 nm in a temperature-controlled TECAN
GeniosPro spectrophotometer. Depending on the specific experiment, the temperature was set
either to RT (actual measuring temperature was 24-27°C) or 37°C. The concentration of
DMSO was found to be highly critical, because concentrations >2% DMSO induced
considerable MT formation. UV absorption of compounds was subtracted from absorption
controls, were carried out in triplicate. Experiments were performed at least with two different
α/β-tubulin batches.


peer-00672298, version 1 - 21 Feb 2012

Page 9

Page 8 of 42
Accepted Manuscript
Prior to start of the docking operation, essential hydrogens and Gasteiger chargers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

8
2.8. Immunofluorescence

Cells grown on cover slips were fixed in -20°C methanol/acetone (1:1) for 7 min and then
blocked in 10% goat serum/PBS. The primary mouse monoclonal antibody to Eg5
(Transduction Laboratories, Lexington, KY) was incubated for 1 h following 30 min
incubation with a fluorochrome-conjugated secondary antibody anti-mouse Cy3 (Jackson
ImmunoResearch Laboratories, West Grove, PN). Immunostained cells were examined using
a Zeiss Axiovert 200 M fluorescence microscope (Göttingen, Germany) and images were
processed with Photoshop software (Adobe, Munich, Germany).
Kinesin-related motor protein Eg5 (also known as kinesin-like protein KIF11, kinesin-like
spindle protein HKSP) is localized at the centrosomes, spindle microtubules, and intracellular
bridge. Eg5 is required for establishing the bipolar spindle.

2.9. Molecular docking of Vinca alkaloids on tubulin

The x-ray structure of tubulin–colchicines–vinblastine: stahmnin-like domain complex (SLD)
was used as docking template throughout the docking calculations. The coordinates of
vinblastine were extracted from the PDB file, obtained from the Brookhaven Protein Data
Bank (PDB 1Z2B) [12]. Docking calculations were performed using AutoDock program
(AutoDock 4.2, The Scripps Research Institute, La Jolla, CA, USA) installed on a Linux PC
under the SuSe operating system. AutoDock is an automated and robust docking algorithm
based on the Lamarckian genetic algorithm (GA) with several success rates in the virtual
screening of ligands.
The 2D structures of vincristine, vinblastine, vinorelbine, videsine, cantharanthine and
vindoline were energy-minimised and converted to 3D structures compatible for docking
operation using an open source program PRODRG [13].
were added to the macromolecule. In order to sample the binding site, a grid of 60 Å×60 Å×
60 Å with a spacing of 0.375 Å was first computed. In total, 100 cycles of flexible ligand
docking were performed in the grid representation of the receptor binding site, followed by
scoring the ligand-receptor interaction. AutoDock clustering was performed based on
similarities in binding modes and affinities in these cycles. The optimized orientations
represent possible binding modes of the ligand within the site. The AutoDock docking output
peer-00672298, version 1 - 21 Feb 2012

Page 10

Page 9 of 42
Accepted Manuscript
cluster tree (dendrogram). The merging of objects with similar features leads to the formation
of a cluster, where the length of the branch indicates the degree of relation. The distance of a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

9
contains solutions ranked according to the scoring functions with information about the
frequency of occurrence, mean energies, inhibition constant and low root mean square
deviations (RMSD) within the cluster each defined by the 3D coordinates. PyMOL
(Schrödinger, Portland, OR, USA) was used as visualization tool to further get a deeper
insight on the binding modes obtained from docking [14].

2.10. Statistical analyses

The mRNA expression values of 60 cell lines of the genes of interest were selected from the
NCI database (http://dtp.nci.nih.gov). The mRNA expression has been determined by
microarray analyses [15,16]. COMPARE analyses were performed to produce rank-ordered
lists of genes expressed in the NCI cell line panel. The methodology has been described
previously in detail [17]. Briefly, every gene of the NCI microarray database is ranked for
similarity of its mRNA expression to the IC50 values for Vinca alkaloids. To derive
COMPARE rankings, a scale index of correlations coefficients (R-values) is created. In the
standard COMPARE approach, greater mRNA expression in cell lines correlates with
enhanced drug resistance, whereas in reverse COMPARE analyses, greater mRNA expression
in cell lines indicates drug sensitivity.
The Ingenuity Pathway Analysis software (IPA) (Ingenuity Systems, Mountain View,
CA, USA; http://www.ingenuity.com) was utilized to identify networks and pathways of
interacting genes and other functional groups in genomic data. Using the IPA Functional
Analysis tool, we were able to associate biological functions and diseases to the experimental
results. Moreover, we used a biomarker filter tool and the Network Explorer for visualizing
molecular relationships.
For hierarchical cluster analysis, objects were classified by calculation of distances
according to the closeness of between-individual distances. All objects were assembled into a
subordinate cluster to a superior cluster represents a criterion for the closeness of clusters as
well as for the affiliation of single objects to clusters. Thus, objects with tightly related
features appear together, while separation in the cluster tree increases with progressive
dissimilarity. Recently, cluster models have been validated for gene expression profiling and
for approaching molecular pharmacology of cancer [15,18]. Cluster analyses applying the
WARD method were done with the WinSTAT program (Kalmia, Cambridge, MA, USA)).
peer-00672298, version 1 - 21 Feb 2012

Page 11

Page 10 of 42
Accepted Manuscript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

10
Missing values were automatically omitted by the program, and the closeness of two joined
objects was calculated by the number of data points they contained. In order to calculate
distances between all variables included in the analysis, the program automatically
standardizes the variables by transforming the data with a mean = 0 and a variance = 1.

Pearson’s correlation test was used to calculate significance values and rank
correlation coefficients as a relative measure for the linear dependency of two variables. This
test was implemented into the WinSTAT Program.. Pearson’s correlation test determines the
correlation of rank position of values. Ordinal or metric scaling of data is suited for the test
and data are transformed into rank positions. There is no condition regarding normal
distribution of the data set for the performance of Pearson’s correlation test.

peer-00672298, version 1 - 21 Feb 2012

Page 12

Page 11 of 42
Accepted Manuscript
takes place, when a tubulin ligand (here catharanthine) binds only to the complex formed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

11
3. Results

3.1. Cytotoxicity

The NCI cell line panel has been tested for sensitivity towards vinblastine, vincristine,
catharanthine, or vindoline. The mean log10IC50 values of eight different tumor types for these
compounds are shown in Figure 2. Cell lines of all tumor types revealed high cytotoxicity
with mean log10IC50 values for vinblastine in a range of -9.26 ± 0.04 M (leukemia) to -8.15 ±
0.41 M (ovarian carcinoma) and for vincristine in a range of -6.95 ± 0.01 M (leukemia) to -
6.14 ± 0.49 M (breast cancer). Catharanthine and vindoline displayed only weak activity.

3.2. Inhibition of tubulin polymerization by Vinca alkaloids

Vincristine potently inhibited microtubule assembly and catharanthine showed a less potent
effect on α/β-tubulin polymerization (Figure 3). Vindoline was not able to reach considerable
inhibition of microtubule assembly, suggesting that this compound does not or only very
weakly interact with α/β-tubulin. In addition, we analyzed the semi-synthetic derivatives,
vindesine and vinorelbine. Both drugs inhibited microtubule assembly in a comparable
manner as vincristine (Figure 3).

3.3. Uncompetitive inhibition of the binding [3H]vinblastine to tubulin by catharanthine

In competition experiments, catharanthine clearly interfered with [3H]-vinblastine for α/β-
tubulin binding in an apparently uncompetitive manner (Figure 4). Uncompetitive inhibition
between tubulin and vinblastine (the T-V complex).

3.4. Molecular docking

The x-ray structure of vinblastine bound to tubulin in a complex with the RB3 protein
stathmin-like domain was selected as a docking template (PDB 1Z2B) [19]. The template and
peer-00672298, version 1 - 21 Feb 2012

Page 13

Page 12 of 42
Accepted Manuscript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

12
the docking parameters, using AutoDock, were validated by docking the crystallographic
structure of vinblastine into the binding site of the protein template. Docking results showed
low RMSD values between the experimental and the calculated docked structure and set a
good platform to reliably dock other structurally similar chemical molecules. Vinca alkaloids
were individually docked into the validated grid defined in the crystal structure of α/β-tubulin
for appropriate conformational search. Next, 100 cycles of docking with approximately
250,000 energy evaluations in each cycle were carried out without any flexibility constraints
on the ligand. The results were finally analyzed upon setting identical conditions for docking
the tubulin structure template with vincristine, vinblastine, vinorelbine, vindesine,
catharanthine and vindoline chemical molecules throughout the operation (Figure 5).
The final docking results show that vindesine, vincristine, vinorelbine and vinblastine
posses a high-order in the frequency of occurrence in a cluster and closeness in parameters in
terms of mean docking energy (-8.52, -8.50, -8.47 and -8.37 kcal/mol respectively) and
predicted inhibition constant (1.49, 2.05, 1.54, 1.67 µM) as shown in Table I. In contrast,
vindoline and catharanthine showed relatively low mean docking energies (-7.28 and -7.0
kcal/mol, respectively) and inhibition constants (4.72 and 7.34 µM, respectively) (Table I),
which implies a low affinity to tubulin binding.
The ligand-protein interaction was analyzed using the LPC tool [20]. Binding site
analysis showed that interaction of Vinca alkaloids (except of catharanthine) with Asn249C
was a conserved feature (Table II). Interaction of Val177B was observed for four of the six
Vinca alkaloids analyzed and interaction of Pro175B with three compounds (Table II). No H-
bond interaction of catharanthine to any amino acid in the pharmacophore was found. The
four clinically established Vinca alkaloids revealed 12-15 residues involved in hydrophobic
interaction with the ligand, whereas vindoline and catharanthine exerted hydrophobic
interaction with only 9 residues each.

peer-00672298, version 1 - 21 Feb 2012

Page 14

Page 13 of 42
Accepted Manuscript
primary target, but also by mechanisms upstream of this target, we investigated the role of the
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

13
3.5. Induction of mono- and multipolar mitoses

To analyze whether vinblastine and its precursors, vindoline and catharanthine, exert a part of
their poisonous effects by induction of multipolar mitotic spindles, SCC114 cells were treated
for 24 h with each of the three compounds and immunostained with an anti-Eg5 monoclonal
antibody to visualize mitotic figures. Counterstaining of DNA was done using DAPI. As
shown in Figure 6A, untreated SCC114 cells contained 6.6% monopolar, 85% bipolar, and
8.3% multipolar mitoses. Treatment with 100-300 nM vindoline led to a dose-dependent
increase of monopolar mitoses up to 23.5%. At higher concentrations, the fraction of
monopolar mitoses decreased. At doses of 250-500 nM vindoline, a weak increase of
multipolar mitoses was recorded (12-15%). Above 500 nM vindoline, general cytotoxicity
precluded further analysis of mitotic spindles in SCC114 cells. According to the increase of
mono- or multipolar mitoses, the fraction of bipolar mitoses dose-dependently decreased in a
range between 100 and 500 nM.
Catharanthine showed a different effect on spindle polarity than vindoline. As shown
in Figure 6B, there was a strong increase of monopolar mitoses up to 37.5% and no increase
of multipolar mitoses upon treatment with 150-210 nM catharanthine. Accordingly, the
percentages of bipolar mitoses dose-dependently decreased from 84% (control) to 50% (210
nM). At higher concentrations, catharanthine was cytotoxic. Vinblastine induced a
considerable percentage particularly of multipolar mitoses (Figure 6C). Representative
examples of multipolar mitoses in SCC114 cells treated with vindoline, catharanthine, or
vinblastine and bipolar mitoses in untreated control cells are depicted in Figure 7.

3.6. Multidrug resistance

Since the activity of Vinca alkaloids may not only be determined by binding to tubulin as
multidrug resistance-conferring drug transporter, P-glycoprotein. We treated drug-sensitive,
parental CCRF-CEM leukemia cells and vincristine-resistant CEM/VCR1000 cells with
catharanthine, vindoline, and four clinically established Vinca alkaloids. Calculation of the
degrees of resistance from the IC50 values revealed that vincristine was one order and
vinblastine two orders of magnitude more cytotoxic towards CCRF-CEM cells than
catharanthine or vindoline (Table III), while vindesine was about two-fold less active than
peer-00672298, version 1 - 21 Feb 2012

Page 15

Page 14 of 42
Accepted Manuscript
criterion, only the top 10 genes for each compound are listed. If more than one clone of the
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65

14
vincristine, vinorelbine was two orders of magnitude more active than vincristine.
CEM/VCR1000 cells were 41.8-fold resistant towards vincristine as compared to the parental
CCRF-CEM cell line. Interestingly, CEM/VCR1000 cells showed higher degrees of
resistance (56.2- to 341.9-fold, respectively) towards vinblastine, vindesine, or vinorelbine
than the resistance-selecting agent, vincristine itself. Degrees of resistance towards
catharanthine or vindoline were only low (~2-fold) (Table III).

3.7. Cross-resistance

To further analyze determinants of resistance and sensitivity towards Vinca alkaloids, we took
advantage of the database of the Developmental Therapeutics Program of NCI, USA. We first
analyzed the cross-resistance profile of the NCI tumor cell panel to Vinca alkaloids. There
was a highly significant correlation between the log10IC50 values for vinblastine and
vincristine (R=0.789; P=6.5×10-16), whereas vincristine cytotoxicity did not correlate with
catharanthine or vindoline activity (Table IV). The log10IC50 values for catharanthine were
significantly associated with those for vindoline (P=0.01), but at a low R-value (0.310). Other
relationships were not found.

3.8. COMPARE analyses of microarray data

Next, we performed COMPARE analyses of the IC50 values for vinblastine, vincristine,
vindoline, and catharanthine and the transcriptome-wide mRNA expression of the NCI cell
lines to produce scale indices of correlation coefficients. We first performed a standard
COMPARE analysis in which cell lines that were most inhibited by Vinca alkaloids (lowest
IC50 values) were correlated with the highest mRNA expression levels of genes. Genes with
correlation coefficients of R>0.5 are shown in Table V. If more than 10 genes met this
same gene appeared in the COMPARE ranking lists, only the clone with the best R-value was
displayed. These genes may be considered as possible candidate genes, which determine
cellular resistance to Vinca alkaloids. Afterwards, reverse COMPARE analyses were done
(R<-0.5), which correlated the most inhibited cell lines with the lowest gene expression levels
(Table V). This approach provided genes that are associated with cellular sensitivity to Vinca
alkaloids. A comparison showed that no genes appeared in association with more than one of
peer-00672298, version 1 - 21 Feb 2012