Transcript (figure 4). - University of Warwick

A Comparison of c-Myc Regulated Gene Networks
Involved in Tumourigenesis of Two Distinct Tissues
Sam Robson 1, 2, S. Pelengaris 2, D. Epstein 3 and M. Khan 2, 4
1 MOAC
Doctoral Training Centre, Senate House, University of Warwick, Coventry, CV4 7AL, UK; 2 Biomedical Research Institute, Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK;
3 Department of Mathematics, University of Warwick, Coventry, CV4 7AL, UK; 4 Clinical Sciences Research Institute, Warwick Medical School, Walsgrave Hospital, Coventry, UK.
1: c-Myc
4: Aims
Deregulated expression of the c-myc (cellular Myelocytomatosis) protooncogene is seen in a large number of human cancers [1]. It transcribes a
transcription factor that forms a heterodimeric complex with Max (figure
1) to promote a variety of tumour related biological functions, such as
cell-cycle progression, inhibition of terminal differentiation, induction of
angiogenesis, and (perhaps somewhat paradoxically) induction of
apoptosis.
To analyse and categorize changes in gene expression at the post-transcriptional level as a result of Myc activation in
two distinct tissues, to compare and contrast these changes to study how different genotypes lead to radically different
phenotypes, and to define possible target genes for future analysis in Myc-related cancer therapy.
5: Laser Capture Microdissection
The use of whole pancreas tissue in microarray time course experiments poses problems for data analysis due to the
non-uniform nature of the islet mass. The ratio of islet cells to exocrine cells does not remain constant after cMycERTAM activation, making comparison of RNA samples from different time-points problematic.
Laser capture microdissection (LCM) allows isolation of pure cell populations (i.e. pancreatic islets or suprabasal
epidermal cells) from surrounding tissue using a low-power infra-red laser beam (figure 4). This ensures that there is
no RNA contamination from cells not expressing the c-mycERTAM transgene.
A
B
C
Figure 1: c-Myc works as a transcriptional activator in a heterodimeric complex with the protein Max.
This Myc-Max complex has been implicated in the initiation of a number of tumour related biological
functions, such as cell-cycle progression, induction of angiogenesis, inhibition of terminal differentiation, and
induction of apoptosis.
x20
2: Transgenic Model
Several therapeutic techniques have been targeted to the control of
aberrant c-Myc expression [2], but the precise genetic pathways involved
in c-Myc induced tumourigenesis are not yet fully understood. It is known
that, during normal biological functions, c-Myc is involved in the
regulation of a large number of genes [3]. However, it is not known
precisely which of these are direct targets, and which are activated
downstream of c-Myc activity [4]. Mouse models have been created [5]
allowing controlled expression of Myc in vivo (figure 2). In vivo analysis of
changes in gene expression at the transcriptional level using high density
oligonucleotide arrays allows large scale analysis of genotypic differences.
x20
x20
Figure 4: Laser capture microscopy (LCM) of a pancreatic islet. LCM is a powerful microscopy technique allowing isolation of pure cell colonies, or even single cells, from
surrounding tissue using a low powered infra-red laser beam. A) Tissue sections are freshly cut, fixed in 100% ethanol, stained with toluidine blue, and dehydrated thoroughly
before being placed on the LCM platform. The infra-red laser is used to cut around selected areas of the tissue. C) The lasered section can be lifted from the tissue using special
‘sticky’ capped epindorf tubes. D) The isolated tissue can be collected for RNA extraction.
6: Data analysis
The multi-parametered nature of the experimental design offers an interesting challenge for the analysis of the gene
expression data. The experiment contains nested data covering time-course, treated vs. untreated, and tissue
comparison. While this data can be used with standard gene expression data analysis techniques such as differential
expression analysis, cluster analysis, ANOVA etc, it is also suited for more advanced analysis techniques such as
unsupervised network structure algorithms.
It is hoped that comparison of differences in gene expression levels, and of network structures created using a variety of
methods (Bayesian inference, fuzzy neural networks, graphical Gaussian methods etc.), between the two tissues will
offer an insight into the genotypic reasons for the markedly different outcomes of c-MycERTAM activation. This study
runs parallel to work conducted by Vicky Ifandi on the proteomic side of c-MycERTAM activation.
To take advantage of the multi-parameter aspect of the data, the generalized linear models (Nelder and Wederburn,
1972) have been utilized. GLM unites a number of statistical methods for linear regression into one model, with the
added bonus that, unlike many other modelling techniques, few assumptions need to be made of the data. Specifically,
we need not assume that the data follows a normal distribution (a bold assumption with only 3 replicates to infer from).
The native GLM functions in the statistical programming language R have been taken advantage of to create a plugin
script for use with the Genespring GX expression analysis suite from Agilent, with a JAVA script planned in the near
future.
Figure 2: Activation of the c-MycERTAM transgene by addition of the ligand 4-Hydroxytamoxifen.
Binding of 4-OHT to the estrogen receptor of the chimeric protein opens up the bHLHZ domain allowing
dimerization with the protein Max and subsequent transcriptional activation.
3: Skin Vs. Pancreas
7: Conclusions and Future Work
Activation of Myc leads to differential expression of a large number of genes, although it is not known exactly how
many of these are direct targets. Also, the precise pathways involved in normal c-Myc regulated biological functions
are not yet fully understood. A combination of a targeted, switchable, in vivo model, allowing direct control over
activation of c-Myc activity, LCM to isolate pure colonies of targeted cells to ensure RNA purity, and high density
oligonucleotide arrays for the analysis of subsequent c-Myc induced gene expression changes offers a powerful
method for analysing specific c-Myc induced changes during tumourigenesis.
The experiment was designed in such a way as to allow multiple comparisons, and consists of time course data and
treated vs. untreated data over two distinct tissue types (skin and pancreas). In this way, we hope to deduce candidate
genes and gene networks responsible for the dichotomy of c-Myc induced phenotypes seen in differing tissues. This
does however create certain problems for the analysis of the high throughput data. A GLM based analysis technique is
being prepared to take full advantage of the multi-parameter nature of the experiment, and will be used in
conjunction with more standard gene expression analysis techniques. Candidate genes of interest will then be analysed
using more robust methods, such as RT-PCR, immunohistology, RNA inteference, and chromatin
immunoprecipitation (ChIP) which will allow us to test if candidate genes are direct c-Myc targets or not.
The data will also be comparable to proteomic studies, conducted by Vicky Ifandi at the University of Warwick. This
will allow a comparison of the genotypic response of the model to the subsequent phenotypic response, providing a
comprehensive overview of the effects of activation of c-MycERTAM, and thus its role in tumourigenesis.
Acknowledgements
Figure 3: Activation of c-Myc in epidermal kerotinocytes results in a vastly different phenotype than
activation of Myc in pancreatic β-cells. In suprabasal skin epidermal cells, activation of c-Myc results in an
increase in proliferation and formation of tumour-like growths [6]. In islet β-cells, c-Myc activation results in a
massive apoptotic response [7]. By studying the genotypic basis for this dichotomy it is hoped that a better
understanding of the tumour-related functions of c-Myc can be reached.
References
Many thanks to Mike, David and Stella for taking me on for this project, and to Vicky Ifandi, Sylvie Abouna,
Göran Mattson and Linda Cheung for their help in the lab. Thanks also to Helen Bird, Sue Davis and Lesley
Ward for all their work on the microarray side of things. This project is funded by the Engineering and
physical Sciences Research Council and the Association of International Cancer Research through the MOAC
Doctoral Training Centre.
M
o
Ac
1.
C. Nesbit, J. Tersak et al (1999), Oncogene.
2.
S. Robson et al (2006), Recent Patents on Anti-Cancer Drug Discovery.
3.
C. Dang (1999), Mol Cell Biol.
4.
K. Zeller et al (2003), Genome Biol.
5.
T. Littlewood et al (1995), Nucleic Acids Res.
6.
S. Pelengaris et al (1999), Mol Cell.
7.
S. Pelengaris, M. Khan et al (2002), Cell.