Systems Approaches in the Common Metabolomics in Acute Lymphoblastic Leukemia and Rhabdomyosarcoma Cells: A Computational Approach

Abstract

Acute lymphoblastic leukemia is the most common childhood malignancy. Rhabdomyosarcoma, on the other hand, is a rare type of malignancy which belongs to the primitive neuroectodermal family of tumors. The aim of the present study was to use computational methods in order to examine the similarities and differences of the two different tumors using two cell lines as a model, the T-cell acute lymphoblastic leukemia CCRF-CEM and rhabdomyosarcoma TE-671, and, in particular, similarities of the metabolic pathways utilized by two different cell types in vitro. Both cell lines were studied using microarray technology. Differential expression profile has revealed genes with similar expression, suggesting that there are common mechanisms between the two cell types, where some of these mechanisms are preserved from their ancestor embryonic cells. Expression of identified species was modeled using known functions, in order to find common patterns in metabolism-related mechanisms. Species expression manifested very interesting dynamics, and we were able to model the system with elliptical/helical functions. We discuss the results of our analysis in the context of the commonly occurring genes between the two cell lines and the respective participating pathways as far as extracellular signaling and cell cycle regulation/proliferation are concerned. In the present study, we have developed a methodology, which was able to unravel some of the underlying dynamics of the metabolism-related species of two different cell types. Such approaches could prove useful in understanding the mechanisms of tumor ontogenesis, progression, and proliferation.

Full text

Author Copy
1
Systems Approaches in the Common Metabolomics in Acute
Lymphoblastic Leukemia and Rhabdomyosarcoma Cells: A
Computational Approach
Christos Tselios, Apostolos Zaravinos, Athanasios N. Tsartsalis, Anna Tagka, Athanasios Kotoulas,
Styliani A. Geronikolou, Maria Braoudaki and George I. Lambrou
Abstract Acute lymphoblastic leukemia is the most common childhood malignancy. Rhabdomyosarcoma, on the
other hand, is a rare type of malignancy which belongs to the primitive neuroectodermal family of tumors. The aim
of the present study was to use computational methods in order to examine the similarities and differences of the two
different tumors using two cell lines as a model, the T-cell acute lymphoblastic leukemia CCRF-CEM and
rhabdomyosarcoma TE-671 and in particular, similarities of the metabolic pathways utilized by two different cell
types in vitro. Both cell lines were studied using microarray technology. Differential expression profile has revealed
genes with similar expression, suggesting that there are common mechanisms between the two cell types, where some
of these mechanisms are preserved from their ancestor embryonic cells. Expression of identified species, was modelled
using known functions, in order to find common patterns in metabolism-related mechanisms. Species expression
manifested very interesting dynamics and we were able to model the system with elliptical/helical functions. We
discuss the results of our analysis in the context of the commonly occurring genes between the two cell lines and the
respective participating pathways as far as extracellular signaling and cell cycle regulation/proliferation are concerned.
In the present study, we have developed a methodology, which was able to unravel some of the underlying dynamics
of the metabolism-related species of two different cell types. Such approaches could prove useful in understanding
the mechanisms of tumor ontogenesis, progression and proliferation.
Christos Tselios, MSc
National and Kapodistrian University of Athens
Laboratory for the Research of Musculoskeletal Disorders
e-mail: christselios68@gmail.com
Apostolos Zaravinos, PhD
European University Cyprus
Department of Life Sciences
e-mail: a.zaravinos@euc.ac.cy
Athanasios Tsartsalis, MD, PhD
Naval Hospital of Athens
Department of Endocrinology Diabetes and Metabolism
Dinokratous 70, 11521, Athens Greece
e-mail: ttsartsal@yahoo.gr
Anna Tagka, MD, PhD
First Department of Dermatology and Venereology, “Andreas Syggros” Hospital
National and Kapodistrian University of Athens, Medical School
Ionos Dragoumi 5, 11621, Athens, Greece
Email: annatagka3@gmail.com
Athanasios Kotoulas, PhD
National Technical University of Athens
School of Electrical and Computer Engineering
Biomedical Engineering Laboratory
e-mail: athkot@gmail.com
Styliani A. Geronikolou, PhD
Biomedical Research Foundation of Academy of Athens
Clinical, Translational, Experimental Surgery Research Centerment of Pediatrics
Choremeio Research Laboratory
e-mail: sgeronik@gmail.com
Maria Braoudaki, PhD
Department of Life and Environmental Sciences
School of Life and Health Sciences
University of Hertfordshire
e-mail: m.braoudaki@herts.ac.uk
George I. Lambrou, PhD (Corresponding Author)
National and Kapodistrian University of Athens
First Department of Pediatrics
Choremeio Research Laboratory
e-mail: glamprou@med.uoa.gr
ResConfP032

Author Copy
2
Key Words: CCRF-CEM, TE_671, leukemia, rhabdomyosarcoma, metabolomics, microarrays
Table of Contents
Table of Contents ............................................................................................................................................. 2
1 Introduction .................................................................................................................................................. 2
1.1 Bioinformatics and the New Advent of Systems Biology ...................................................................... 3
1.2 Systems Theory and Modelling .............................................................................................................. 3
1.3 Systems Theory and Systems Biology .................................................................................................... 3
2 Materials and Methods ................................................................................................................................ 4
2.1 Experimental Setup ................................................................................................................................ 4
2.2 Microarray Experimentation.................................................................................................................. 4
2.3 Bioinformatics and Systems Analyses .................................................................................................... 4
2.4 Simulation ............................................................................................................................................... 4
3 Results ............................................................................................................................................................ 6
3.1 The Ontology of the Transcriptional Profile ......................................................................................... 6
3.2 The Time-Dependent Evolution of Molecules ....................................................................................... 6
3.3 Polar Transformation of Time-Dependent Evolution ........................................................................... 7
3.4 Species-Dependent Evolution of the Total System for all time-points .................................................. 7
4 Discussion ...................................................................................................................................................... 8
4.1 Conclusions ............................................................................................................................................. 9
References ........................................................................................................................................................ 9
1 Introduction
Acute lymphoblastic leukemia (ALL) is a cancer of the lymphoid line of blood cells that emerges early in the childhood.
It originates from the undifferentiated lymphoblast which ceases its development at different stages during their
maturation to lymphocytes and gains many mutations to genes that affect blood cell development and proliferation, thus,
creating cancerous cells. Rhabdomyosarcoma (RMS) on the other hand, is a rare cancer that comprises 5-8% of all tumors
emerging during childhood. The cell of origin is the myoblast or cells responsible for the formation of the skeletal muscle.
Theoretically, rhabdomyosarcomas can emerge at any part of the body that has skeletal muscle but it originates mainly
at the head and neck. Thus, these two malignancies are of different origin. The common aspect between those
malignancies is that they both comprise of cells that are undifferentiated, immortal and potentially divide infinitely. Also,
looking back to their developmental history, both cell types originate from the embryonic mesoderm. From this point,
differentiation enables these two cell types with different functions and roles in the body, through differential gene
regulation. Interestingly it has been reported that rhabdomyosarcoma can be present in the bone marrow of patients
presenting a leukemic image, without the presence of a primary tumor [1, 2].
The metabolic mechanics of tumors is still largely unknown and in particular, the metabolic mechanics of ALL and
RMS. It is possible that an answer to the treatment of cancer lies within the delicate mechanisms of metabolic processes.
As aforementioned, the first observation on tumor cells metabolism was stated by Warburg et al. (1924) where it was
observed that a shift occurred in tumors from oxidative phosphorylation to aerobic glycolysis, known today as the
Warburg effect [3]. The main governing idea was that the Warburg effect is rather the consequence of the cancer process,
mainly due to hypoxic growth, and not a prerequisite for cancer progression and proliferation per se [4]. However,
previous reports have highlighted that the process of aerobic glycolysis is reversible and as a matter of fact from very
simple compounds such Dichloroacetate (DCA) [5, 6]. In fact, the aspect of interfering with cancer metabolic pathways
as a mean for tumor treatment and therapy it seems to be of great importance [7]. This became apparent from discoveries
made in the field of metabolomics involving molecules previously thought to be solely metabolic and without the slightest
suspicion for regulatory functions [8-10]. It is clear nowadays, that metabolites or metabolic molecules, not only
participate in metabolic processes which have to do solely with energy production and thermodynamical conservation of
the cell but also possess multiple functions especially on the level of signal transduction. Therefore, it would not be an
exaggeration to say that the same molecules used as nutrient could also be the targets for therapies.

Author Copy
3
1.1 Bioinformatics and the New Advent of Systems Biology
The rapid development of high throughput methodologies has brought about the need for new tools of analysis.
Massive data are not able to be processed with classical statistical approaches, but new algorithms should be developed.
Towards that purpose, systems theoretic approaches borrowed from engineering and natural sciences have been employed
in order to create a new system biological discipline. Basically, as a consequence of the –omics “explosion”, new
questions arose which are mainly concerned with the interactions between molecules, questions related to cellular
communication and how cells make decisions, as well as with the signaling and expression of genes, respectively.
Systems biology is often confused with bioinformatics, which, while important and complementary to each other,
differs significantly as it uses a different approach to understand biological systems and in particular different
methodology for extracting results. Bioinformatics could be summarized as follows: a) it recognizes patterns, b) connects,
correlates, c) groups and d) sorts variables, while the systematic approach to biology; a) studies space-time principles and
mechanisms of cellular function, i.e. the dynamics of cellular processes, b) draws experimental responses to induced
stimuli, c) models and simulates cellular functions [11]. At the same time, bioinformatics manages the uncertainty in
biological systems, with the help of statistical modeling. Systems biology, on the other hand, must manage the complexity
that results from dynamic, non-linear space-time interactions with mathematical modeling. The study of signaling
pathways through mathematical modeling is, mandatory as they are governed by dynamic principles of control, regulation
and coordination, where feedback mechanisms are most often involved [11].
1.2 Systems Theory and Modelling
A system is defined as a set of objects that have well-established relationships with each other. However, this definition
is not entirely appropriate from the mathematical point of view, as the emphasis should not be on the objects but on the
processes of the system. For the purposes of modeling a system, two types of systems can be distinguished [11]; a) a
physical system which is a reconstruction of the apparent world and b) a standard system, which constitutes the
mathematical framework describing the physical system. Both system types can be modelled with the help of
mathematical modelling, which can be described as the process of interpreting a physical system with mathematical
formulation. In a mathematical model, a series of variables is defined along with their interactions. On the other hand, a
system can be also described with the help of a simulation, which can be described as the process of executing or
experimenting with the mathematical model. In other words, simulation concerns the “feeding” of the model with data
and predicting the system’s behavior. The key to a successful modelling attempt is to ultimately find an etiological relation
between the investigated variables. In general, systems theory and mathematical modelling introduces concepts such as
a) the effect of variables on the system, b) the differences of the system under different conditions and c) the conclusions
about the system based on the procured results. To be consistent with formality, we could describe a general system S,
which represents the relation among a number of variables. Thus, such a system of one or more variables/objects is a
subset of the Cartesian product [12] of all variables such as S[12], where i=1,2,…,n or equivalently S
⊂
Ο1×Ο2×...,n.
Similarly, we can define a complex system, comprised of other subsystems where each one of them is also a subset of the
variables Cartesian product.
1.3 Systems Theory and Systems Biology
The eukaryotic cell can be considered, and it is, as a complex system. In order to understand the cell’s function (or
functions) it is important to be able to understand the relations between its constituting parts, that would be mainly the
genes and proteins. A very interesting approach to this problem is the understanding of signal transduction pathways,
which can be described with the use of mathematical formulations. In particular, such an approach would assist us to
address questions such as: a) how do molecules interact, create signals, which store and transmit information, b) how do
signals unify to become “cellular decisions”, c) what is the spatio-temporal evolution of the cellular system and d) how
do the cellular events interact to produce higher level functions such as those of a tissue or an organism. The key to
understanding all previous questions, is the finding of an etiological structure of the physical system identified through
the standard system. In more plain words, ideally we would like to find a mathematical equation that would be able to
predict the spatio-temporal evolution of a biological system with extreme precision.
A very important concept of systems biology is that it regards signal transduction pathways as a set of biochemical
reactions, not static,but dynamic changing with time. This can be stated in a more formal way as:
:







(1)
Where, σ is a function defined from a set of stimuli (Ω) to a set of responses (Γ) and ω is the variable describing a stimulus
to a response γ. In other words:
Ω={ω:Ι→U}, Γ={γ:Ι→Υ} (2)
Such systems are usually described with the help of ordinary differential equations of the form '
x
yt


, or dx
ydt
, where
x can be the concentration of a molecule with respect to time. Such a reaction could be described in a general form such
as:

Author Copy
4
12 12
12 12
: ... ...
nn
k
nn
RIX IX IX JYJY JYX

     

 
 (3)
Where, Iμ, Jμ are the stoichiometric factors of the reaction, X and Y are the reactants and products respectively and kμ is
the reaction rate. In the case of a series of such reactions, as in the case of a signal transduction pathway, this can be
solved by solving a system of equations of the form:
11
() ( ), 1, 2,...,
j
i
n
Mi
i
j
j
dx t vk X t i n
dt





 (4)
In the present work we have investigated gene expression data from two cell lines, the T-cell acute lymphoblastic
leukemia CCRF-CEM and the rhabdomyosarcoma TE-671 cell line [13]. In addition, in the present study we used systems
biology approaches in order to simulate signal transduction pathways common to both cell lines and in particular, those
identified genes that were related to metabolism and energy transformation.
2 Materials and Methods
2.1 Experimental Setup
The CCRF-CEM (ALL) [14] and the TE-671 (RMS) cell lines were used as the model, both obtained from the
European Collection of Cell Cultures (ECACC). Both cell lines as well as the process and conditions of cell culture have
been described previously in detail [13]. Also, microarray experimentation has been previously described in detail [13].
2.2 Microarray Experimentation
For the assay of mRNA levels two sets of microarray chips were used: cDNA microarray chips (4.8k genes) obtained
from TAKARA (IntelliGene™ II Human CHIP 1) [15] and microarray chips (9.6k genes) from the Institut fuer
Molekularbiologie und Tumorforschung, Microarray Core Facility of the Philipps-Universitaet, Marburg Germany
(IMT9.6k) as previously described [13]. The microarray data have been submitted to the GEO Database under the
Accession Number GSE34522. Microarray data pre-processing analysis was performed with ImaGene® v.6.0 Software
(BioDiscovery Inc, CA) and ARMADA software (National Hellenic Research Foundation, Athens Greece) [16]. Data
were collected from exported text file and data pre-processing was performed using the Microsoft Excel® environment.
Data were processed in two ways: The first included the separation of each channel (Cy3 and Cy5), and the second
included pre-processing of the ratio between the two samples.
A common preprocessing stage was applied to the raw data (the median intensity value in each channel) of both
platforms. Specifically, the well performing version of the robust loess-based background correction (rLsBC) approach,
as proposed by Sifakis et al. (2011) was applied [17]. The background corrected signal intensities were further normalized
in order to mitigate the effect of extraneous, non-biological variation in the measured gene expression levels.
The data were further analyzed in order to identify the differentially expressed genes (DEGs) and the groups of genes
that share common expression characteristics. Analysis steps were conducted in the Matlab® computing environment.
2.3 Bioinformatics and Systems Analyses
Commonly expressed genes were annotated for their biological functions using the eGOn online tool for Gene Ontology
(The Norwegian University of Science and Technology, Trondheim, Norway)9 [18], Genesis 1.7.2 software [19],
gprofiler10 [20] and WebGestalt web-tools [21-24]. Out of these genes 185 were found to be related to metabolism and
energy conversion. Those genes were further inserted in the SimBiology tool of the Matlab computational environment.
Gene definitions and functions were based on the National Institute of Health databases11. The annotated species are
presented in Fig. 1, exactly as they have been used for simulation. Interactions among molecules was investigated using
the Coremine12 web-tool, which identifies known interactions between molecules as well as known functions and diseases.
2.4 Simulation
For the present model, in order to describe interactions between molecules (the species) we have used the law of mass
action. In particular, in a reaction described as:
k
A
BCD




 (5)
Where the reaction constant k is given by:

9http://www.genetools.microarray.ntnu.no/egon/
10https://biit.cs.ut.ee/gprofiler/gost
11http://www.ncbi.nlm.nih.gov/sites/entrez/
12https://coremine.com/medical/

Author Copy
5
a
CD
a
A
B
CC
kCC


 (6)
Where CA,B,C,D are the concentrations of the respective reactants and products. Therefore, for the reaction of two or more
species we can formulate a differential equation describing the concentration change rate with respect to time such as:
,,,
1111
C
A
BD
ABCD
dC
dC dC dC
udt dt dt dt

 (7)
For solving such a massive system of differential equations we have used the explicit tau approximation method,
which involves the calculation of the stochastic system by leap τ, where the change in each step is described by:
()()('())
x
txtPxt


  (8)
Where, P(τx’(t)) is a Poisson distributed random variable with mean τx’(t).
Fig. 1. The SimBiology environment, where all species (i.e. molecules that are represented as colored spots) are entered and will be
further simulated. In the present figure the interactions between the species are not shown (Legend: GA: Golgi Apparatus, ER:
Endopasmatic Reticulum, DG: Dense Granule).
All simulations were performed for 21sec, which produced a matrix with dimensions 84×185, i.e. 84 time points and 185
for 185 species. This matrix was further used for data analysis. In addition, in order to graphically represent our data we
have applied a polar transformation of Cartesian data. Polar transformation for two coordinates x, y is estimated by the
reverse tan function of the angle θ and the radius (r) of the circle passing from the (x,y) point (Fig. 2).
Fig. 2. Transformation of two variables x, y on a Cartesian system to polar coordinates.

Author Copy
6
3 Results
3.1 The Ontology of the Transcriptional Profile
As a first step, we have investigated the ontological profile of the transcriptional profile of our gene dataset. Interestingly,
the majority of the identified genes was found to participate in metabolic processes as initially stated (Fig. 3). In particular,
identified genes participated in 230 biological processes, which were metabolism-related in their majority, also
participated in 14 KEGG pathways, 36 cellular components, 4 wiki pathways, 3 reactome pathways (REAC), 9
transcription factors (TF) and two predicted miRNAs (Fig. 3).
Fig. 3. Functional profiling of the identified genes. Genes were found to participate in 31 molecular functions, 230 biological processes,
36 cellular components (CC), 14 KEGG pathways, four wiki pathways (WP), three reactome pathways (REAC), nine transcription
factors (TF) and two predicted miRNAs.
3.2 The Time-Dependent Evolution of Molecules
Following the ontological investigation of the identified genes, the next step in our analysis was the investigation of the
time-dependent evolution of the simulated species. Most of them manifested a logistic-like graph along with an oscillatory
pattern. Indicatively, in Fig. 4, we present some examples of the evolution of ANG (Fig. 4a), DDR2 (Fig. 4b), EPHA3
(Fig. 4c), LPL (Fig. 4d), LRRK2 (Fig. 4e), MADD (Fig. 4f), MET (Fig. 4g), PLA2G6 (Fig. 4h), PLK1 (Fig. 4i),
PRKAR2B (Fig. 4j), STK25 (Fig. 4k), XRCC5 (Fig. 4l) with respect to time. All these molecules participate in the
positive regulation of phosphorus metabolic processes, AMP metabolic processes, the positive regulation of sequestering
of triglycerides, as well as in homeostatic processes. Since, each species was represented with respect to time, it would
be interesting to investigate the relation of species among them at all time points. An example is presented in Fig. 5,
where the expression of ANG, EPHA3, MADD, MET and PLA2G6 are presented. All species manifested a pattern of
expression, indicating a possible relational and etiological connection.
Fig. 4. Time-dependent evolution of simulated species. Examples include ANG (A), DDR2 (B), EPHA3 (C), LPL (D), LRRK2 (E),
MADD (F), MET (G), PLA2G6 (H), PLK1 (I), PRKAR2B (J), STK25 (K) and XRCC5 (L).

Author Copy
7
3.3 Polar Transformation of Time-Dependent Evolution
As we tried to further understand the dynamics of molecule evolution we have plotted the transformed polar coordinates
of each species with respect to time. This representation manifested some interesting dynamics, which are presented in
Fig. 6. Each species manifested quasi-turbulent dynamics, relative to its time-dependent behavior. Since, we attempted
to find specific patterns that could describe the transition of the simulated molecules with respect to time we have created
a diagram manifesting all trajectories simultaneously.
Fig. 5. Three-Dimensional representations of the species presented in Fig. 1 and Fig. 4.
Fig. 6. Polar transformations of Cartesian time-series data. Examples include ANG (A), DDR2 (B), EPHA3 (C), LPL (D), LRRK2
(E), MADD (F), MET (G), PLA2G6 (H), PLK1 (I), PRKAR2B (J), STK25 (K) and XRCC5 (L).
3.4 Species-Dependent Evolution of the Total System for all time-points
The aforementioned observations have led us to the question whether it would be possible to model these graphs in a
more formal way. For that reason, since the species’ trajectories resembled helices we have attempted to perform
regressions based on functions that produce helical structures. Our approximation gave very interesting results, since it
appeared that those trajectories were able to be modelled with functions of the form:
12
12
(, ) sin( ) cos( )
sin( ) cos( )
(, )
f
xy ax r ay r
ax ay
fxy rr
 
 (9)
The result of the simulation, of all species, at all time-points is presented in Fig. 7. This simulation models the
regressions of one species vs. the other at all time-points and in all possible combinations. The observed behavior of one
species vs. the other, in all possible combinations (Fig. 7b) manifested a quasi-helical structure, which resembled very
much the experimental result (Fig. 7a). Further on, we have performed the same approach, by modelling the species
presented in the previous figures vs. all other. The result is presented in Fig. 8, where it appeared that species expression
manifested interesting elliptic helical dynamics.

Author Copy
8
Fig. 7. All regressions of simulated species were represented in a 3D graph, where each species was regressed vs. the other (A) as well
as in a 3D graph of the same regressions, modelled as a spiral (B).
Fig. 8. Helix-modelled data with respect to all species. Each species was regressed with all others using the log2 expression of the
simulated species. In the present figure we present representative modelling attempts for the species ANG (A),DDR2 (B), EPHA3 (C),
NADD (D), MET (E), PLA2G6 (F), PLK1 (G), PRKAR2B (H) and STK25 (I).
4 Discussion
In a previous work we have reported that simulated species, as identified by microarray experimentation, manifested
quasi-chaotic behaviors, such as sinks and repellers [13]. In the present work we have expanded our concept of thinking
by attempting to examine a larger cohort of the identified species and more specifically those participating in metabolism
and energy conversion. We have simulated the identified genes/species in order to understand the underlying dynamics.
To the best of our knowledge there are no previous works on the topic.
Most research to date, concerning cancer therapy, is focused on adult tissue and derives conclusions from data gathered
from adult patients. Thus, very few studies are concerned with paediatric tumours and therefore they consist of a
promising field of study. Further on, increasing evidence from clinical trials of promising anticancer drugs points at the
need for an enhancement of our understanding of intracellular signal pathway networks; targeting exclusively cellular
receptors works well with one type of human cancer, and fails at another type. Cancer in a human patient has a far larger
temporal and spatial frame of complexity than any type of experimental model. A possible answer to this complexity

Author Copy
9
would be to target ancillary metabolic pathways for the intracellular signals that maintain the pool of malignant cells, in
addition to the main pathway that sustains aggressive cancer properties.
A major part of biomedical progress has been based until now on monitoring and interfering with hormone receptors
(nuclear and transmembrane). Advanced cancer stages, however, often fail to respond to receptor-targeted therapy. These
stages are often linked to aberrant activation of proteins that mediate intracellular survival and growth signals downstream
of hormone receptors, and are therefore not expected to be influenced by drugs aimed solely at the receptor molecule.
Effectiveness of interference aimed at signal mediators has been already demonstrated at the level of tissue culture [12,
25] at the level of animal xenograft [26, 27] and at the level of clinical trials on patients [28, 29]. The need to understand
basic mechanisms of intracellular signals is very large. In studies of cell culture, it has been shown that combining
anticancer drugs may yield effects that cannot be always beneficial to the patient. Essential to the utilization of advanced
knowledge in cancer cell physiology is basic research in the interactions of key mechanisms of response to drugs.
There is a need for study of major intracellular signals active in childhood leukaemia and RMS cells. To that direction,
we have attempted to report a methodology that would enable us understand the mechanics of metabolism-related
molecules in acute leukaemia and RMS. This understanding is critical for the mechanistic design of further preclinical
and clinical studies, aimed to improve effectiveness of treatment for childhood tumours by prevention of undesirable side
effects.
4.1 Conclusions
To conclude, the current study showed that metabolism-related genes, as identified by microarray experimentation,
manifested specific patterns of expression, which could be modelled using systems biology and mathematical tools. Such
approaches could be considered a valuable tool towards the understanding of therapeutic targets in childhood neoplasms.
References
1. Parham D.M. et al. 1998. Rhabdomyosarcoma mimicking acute leukemia. Arch Pathol Lab Med 122 (12):1047
2. Sandberg A.A. et al. 2001. Hematologic masquerade of rhabdomyosarcoma. Am J Hematol 68 (1):51-57
3. Warburg O. et al. 1924. Ueber den Stoffwechsel der Tumoren. Biochemische Zeitschrift 152:319-344
4. Arakaki A.K. et al. 2008. Identification of metabolites with anticancer properties by computational
metabolomics. Molecular cancer 7:57
5. Bonnet S. et al. 2007. A mitochondria-K+ channel axis is suppressed in cancer and its normalization promotes
apoptosis and inhibits cancer growth. Cancer cell 11 (1):37-51
6. Wong J.Y. et al. 2008. Dichloroacetate induces apoptosis in endometrial cancer cells. Gynecologic oncology 109
(3):394-402
7. Araujo E.P. et al. 2006. Disruption of metabolic pathways--perspectives for the treatment of cancer. Current
cancer drug targets 6 (1):77-87
8. Griner E.M., Kazanietz M.G. 2007. Protein kinase C and other diacylglycerol effectors in cancer. Nat Rev Cancer
7 (4):281-294
9. Ogretmen B. 2006. Sphingolipids in cancer: regulation of pathogenesis and therapy. FEBS letters 580 (23):5467-
5476
10. Pollak N. et al. 2007. The power to reduce: pyridine nucleotides--small molecules with a multitude of functions.
The Biochemical journal 402 (2):205-218
11. Vera J., Wolkenhauer O. 2008. A system biology approach to understand functional activity of cell
communication systems. Methods in cell biology 90:399-415. doi:10.1016/s0091-679x(08)00817-0
12. Kim J.Y. et al. 2010. Molecular mechanisms of cellular proliferation in acute myelogenous leukemia by leptin.
Oncol Rep 23 (5):1369-1374. doi:10.3892/or_00000773
13. Lambrou G.I. et al. 2012. Pathway simulations in common oncogenic drivers of leukemic and
rhabdomyosarcoma cells: a systems biology approach. International journal of oncology 40 (5):1365-1390.
doi:10.3892/ijo.2012.1361
14. Foley G.E. et al. 1965. Continuous Culture of Human Lymphoblasts from Peripheral Blood of a Child with
Acute Leukemia. Cancer 18:522-529
15. Chung H.W. et al. 2004. Differences in genetic expression profiles between young-age and old-age gastric
adenocarcinoma using cDNA microarray for endocrine disruptor study. Oncology reports 12 (1):33-39
16. Chatziioannou A. et al. 2009. Gene ARMADA: an integrated multi-analysis platform for microarray data
implemented in MATLAB. BMC bioinformatics 10:354
17. Sifakis E.G. et al. Evaluating the effect of various background correction methods regarding noise reduction, in
two-channel microarray data. Computers in biology and medicine
18. Beisvag V. et al. 2006. GeneTools--application for functional annotation and statistical hypothesis testing. BMC
bioinformatics 7:470
19. Sturn A. et al. 2002. Genesis: cluster analysis of microarray data. Bioinformatics (Oxford, England) 18 (1):207-
208
20. Raudvere U. et al. 2019. g:Profiler: a web server for functional enrichment analysis and conversions of gene lists
(2019 update). Nucleic acids research 47 (W1):W191-W198. doi:10.1093/nar/gkz369

Author Copy
10
21. Zhang B. et al. 2004. GOTree Machine (GOTM): a web-based platform for interpreting sets of interesting genes
using Gene Ontology hierarchies. BMC bioinformatics 5:16
22. Zhang B. et al. 2005. WebGestalt: an integrated system for exploring gene sets in various biological contexts.
Nucleic acids research 33 (Web Server issue):W741-748
23. Wang J. et al. 2013. WEB-based GEne SeT AnaLysis Toolkit (WebGestalt): update 2013. Nucleic acids research
41 (W1):W77-W83. doi:10.1093/nar/gkt439
24. Wang J. et al. 2017. WebGestalt 2017: a more comprehensive, powerful, flexible and interactive gene set
enrichment analysis toolkit. Nucleic acids research 45 (W1):W130-W137. doi:10.1093/nar/gkx356
25. Pillozzi S. et al. 2011. Chemotherapy resistance in acute lymphoblastic leukemia requires hERG1 channels and
is overcome by hERG1 blockers. Blood 117 (3):902-914. doi:10.1182/blood-2010-01-262691
26. Wang J. et al. 2010. Notch promotes radioresistance of glioma stem cells. Stem cells (Dayton, Ohio) 28 (1):17-
28. doi:10.1002/stem.261
27. Wysoczynski M. et al. 2010. Selective upregulation of interleukin-8 by human rhabdomyosarcomas in response
to hypoxia: therapeutic implications. International journal of cancer 126 (2):371-381. doi:10.1002/ijc.24732
28. Karp J.E. et al. 2001. Current status of clinical trials of farnesyltransferase inhibitors. Current opinion in
oncology 13 (6):470-476. doi:10.1097/00001622-200111000-00009
29. Fathi A.T. et al. 2010. Exploiting cellular pathways to develop new treatment strategies for AML. Cancer
treatment reviews 36 (2):142-150. doi:10.1016/j.ctrv.2009.12.004