PROJECT PERIODIC REPORT - University College Dublin Annual Report 2012.pdf · Project title: Analysing and Striking the Sensitivities of Embryonal Tumours ... 3.1 Publishable summary

PROJECT PERIODIC REPORT

Grant Agreement number: 259348-2

Project acronym: ASSET

Project title: Analysing and Striking the Sensitivities of Embryonal Tumours

Funding Scheme: Two-Stage Collaboration project

Large-scale integration project

Date of latest version of Annex I against which the assessment will be made:

Periodic report: 1st Completed 2nd x 3rd □ 4th □

Period covered: from 1 Nov 2011 to 31 Oct 2012

Name, title and organisation of the scientific representative of the project's coordinator1:

Prof Walter Kolch, MD, FRSE Systems Biology Ireland University College Dublin Conway Institute, Belfield Dublin 4 Ireland

Tel: ++353-1-716 6931

Fax: ++353-1-7166856

E-mail: [email protected]

Project website2 address:

http://www.asset-fp7.eu

1 Usually the contact person of the coordinator as specified in Art. 8.1. of the Grant Agreement . 2 The home page of the website should contain the generic European flag and the FP7 logo which are available in electronic

format at the Europa website (logo of the European flag: http://europa.eu/abc/symbols/emblem/index_en.htm logo of the 7th

FP: http://ec.europa.eu/research/fp7/index_en.cfm?pg=logos). The area of activity of the project should also be mentioned.

http://www.asset-fp7.eu/

http://europa.eu/abc/symbols/emblem/index_en.htm

http://ec.europa.eu/research/fp7/index_en.cfm?pg=logos

Declaration by the scientific representative of the project coordinator

I, as scientific representative of the coordinator of this project and in line with the obligations as

stated in Article II.2.3 of the Grant Agreement declare that:

The attached periodic report represents an accurate description of the work carried out in this project for this reporting period;

The project (tick as appropriate) 3:

x has fully achieved its objectives and technical goals for the period;

□ has achieved most of its objectives and technical goals for the period with relatively minor deviations.

□ has failed to achieve critical objectives and/or is not at all on schedule. The public website, if applicable

X is up to date

□ is not up to date To my best knowledge, the financial statements which are being submitted as part of this

report are in line with the actual work carried out and are consistent with the report on the resources used for the project (section 3.4) and if applicable with the certificate on financial statement.

All beneficiaries, in particular non-profit public bodies, secondary and higher education establishments, research organisations and SMEs, have declared to have verified their legal status. Any changes have been reported under section 3.2.3 (Project Management) in accordance with Article II.3.f of the Grant Agreement.

3 If either of these boxes below is ticked, the report should reflect these and any remedial actions taken.

Name of scientific representative of the Coordinator: Prof. Walter Kolch

Date: 12/ 02/ 2013

For most of the projects, the signature of this declaration could be done directly via the IT

reporting tool through an adapted IT mechanism.

3.1 Publishable summary Summary description of the project context and objectives

ASSET applies a systems level approach to study embryonal tumours (ET), which are dysontogenetic

tumours whose pathological features resemble those of the developing organ or tissue of origin and include

the entities neuroblastoma (NB), medulloblastoma (MB) and Ewing sarcoma family tumours (ESFT). These

tumours arise in babies and children. Therefore, they tend to be especially devastating for the patients and

their families. ETs pose significant clinical challenges in terms of disease stratification for prognosis and

treatment as well as in the paucity of drugs available for treatment. ETs seem to share common aberrations

in core signalling networks with “modulator” pathways determining disease-specific manifestations.

Combining state-of-the-art genomics, proteomics and mathematical modelling, ASSET will analyse ETs with

the aim to deconvolute the plethora of molecular pathogenetic cancer aetiologies to the common core

principles.

ASSET concentrates on elucidating the aberrations in signal transduction networks that are brought about by

genetic alterations and drive the pathogenesis of ETs. Through predictive mathematical and computational

modelling we will harvest this knowledge and use it to develop functional, network based indicators for

patient stratification and new treatments. Our core hypothesis is that we can make phenotypically diverse

cancers accessible to therapeutic approaches by targeting the shared core networks. ASSET will test this

hypothesis by (i) predictive models that identify these network vulnerabilities; and by (ii) rationally designed

screens for inhibitors (drugs, siRNAs) that target these vulnerabilities. The resulting information on

drugs/siRNAs and efficacious combinations in ETs will be the main practical output of ASSET. It should be

noted that while ASSET focuses on examining this hypothesis in respect to ETs, a positive validation could

also open completely new concepts for the treatment of adulthood cancers by shifting the focus from trying

to address the diversity of cancers towards a focus on finding and exploiting the commonalities.

ASSET builds on a wealth of high quality, high-throughput "omics" datasets generated in the recently

finished FP6 project, the European Embryonal Tumour Pipeline (EETP; http://www.eet-pipeline.eu/) project,

and also will benefit from access to the largest collection of clinical ET samples in Europe. Existing EETP

data comprise >1000 molecularly characterised clinical samples, and mRNA and miRNA transcriptome

profiles derived from ETs and cell lines. Importantly, ASSET will advance from an observatory domain into

the functional, mechanistic domain informed and driven by mathematical models. This will be achieved by

generating and integrating quantitative, large-scale datasets to produce predictive models that will be

rigorously validated in disease models and clinical samples.

The ASSET objectives

Using a systems biology-driven discovery and validation engine our aims are:

the combined analysis of genomic mutations, transcriptome, miRNA expression and dynamic proteome changes in ET model cell lines

http://www.eet-pipeline.eu/

mathematical modelling to elucidate molecular pathogenetic networks and their emergent properties

systematic perturbations to probe and refine these networks

implementation of a virtuous cycle of model making and validation in relevant biological model systems (cell culture models and preclinical mouse models) and clinical samples

The ASSET workprogramme

The general conceptual flow of the project is to start from the reconstruction of global, but static networks

(GRN networks based on existing EETP data and protein signalling networks based on the quantitative

proteomics experiments performed in ASSET), then use large-scale and targeted perturbation analysis to

validate the network topologies, and in combination with model predictions, extract functionally important

subnetworks that subsequently will be modelled using dynamic, kinetic methods. The latter models will allow

us to analyse emergent network properties and simulate the behaviour of networks to perturbances. This

analysis will be deployed to identify nodes that are individually, or in combination, vulnerable to interference

by drugs or siRNAs. These predictions will be tested in ET cell models with conditional oncogene

expression, preclinical xenograft models and clinical specimens. The data throughout the project are

collected in a data warehouse in a semantically linked form so that both experimental data and models are

tractable and amenable to meta-analysis. As a result, we expect to achieve a rational prediction and

validation of drug sensitivities in pathogenetic ET signalling networks.

Work performed since the beginning of the project and the main results achieved so far The initial setup phase of the project has been successfully concluded in the first year. The work is now well underway progressing largely as scheduled in the work plan. There were some technical problems with instability of protein expression of inducible TrkA cell lines, which incurred a small delay in some subprojects,

Transcriptomics

Proteomics

ChIP

Reconstruction of

static global

networks

Dynamic

mechanistic

models of

subnetworks

Identification

of vulnerable

nodes

Mathematical

models Validation

ET cell lines

Xenografts

Clinical

samples

Perturbations

Predictions

Experimental Data

DNA

sequencing

New drug

combinations

Outline of the ASSET Workflow

TranscriptomicsTranscriptomics

ProteomicsProteomics

ChIPChIP

Reconstruction of

static global

networks

Dynamic

mechanistic

models of

subnetworks

Identification

of vulnerable

nodes

Mathematical

models Validation

ET cell linesET cell lines

XenograftsXenografts

Clinical

samples

Clinical

samples

Perturbations

Predictions

Experimental Data

DNA

sequencing

DNA

sequencing

New drug

combinations

Outline of the ASSET Workflow

in particular the proteomics profiling of TrkA signal transduction and the modelling of TrkA signalling. However, these delays will not affect overall progress, and the problems have recently been solved through the construction of new cell lines. Work in some areas, e.g. the drug screens and establishment of mouse models for drug testing and model validation, is ahead of schedule. Exceptional progress was made in the work on (i) new drug targets being discovered from the screens and computational network analysis; and (ii) screening of drugs, miRNAs and new drug combinations. The work on the data warehouse for the sharing and integration of data has made further progress and keeps contributing to projects, mainly the drug screening experiments. The exome sequencing of the agreed cell lines panel has been concluded. Facilities for data sharing and mining are being populated leading to an increase in functionality. The mathematical and computational modelling efforts were focusing on developing the statistical inference methods necessary to optimally exploit the high throughput transcriptomics and proteomics data. The application of modelling to data has commenced. There were 4 deliverables were due during this reporting period, which are reported on separately. Overall, we have made excellent progress and achieved promising results in many areas. They are described in detail below, but briefly the main results are: WP1 1. Potential network structures have been identified and detailed mechanistic models based on ODEs have

been developed to study the regulation of E2F target genes, induction of a p53 via the NOTCH pathway, and contribution of the kinases CDK2 and ATK to the regulation of FOXO1 target genes in Ewing sarcoma.

2. Additional time resolved RT-qPCR data have been generated for the validation of these models. 3. We have established a statistical method for the inference of GRNs based on time-resolved

transcriptome data and applied it to the neuroblastoma SH-EP cell lines inducing MYCN under the control of tetracycline.

4. Markov Chain Monte Carlo algorithms for efficient inference of ODE models have been developed. 5. Multiple patterns of dynamic gene expression in response to MYCN induction were deduced, revealing

targets of MYCN, both activated and repressed. This enlarged our knowledge about mechanisms of regulation by MYCN, in particular, on its ability to overcome proliferation arrest under doxorubicin treatment. Activating and repressive effects of MYCN on its targets were confirmed in clinical data on neuroblastoma patients bearing MYCN amplification.

6. Promoter analysis of genes in each pattern indicated transcription factors which possibly interfere with MYCN. The hypotheses will be verified with ChIP, towards the milestone M.1.5. “GRN and motif structures for MYCN and MYC” (Mo48)

WP2 1. MYCN and EWS-FLI1 regulated genes were identified.

2. Combinatorial assessment has revealed synthetic lethalities, e.g. with amplified MYCN and spindle

checkpoint genes.

3. The screening platforms for drugs, siRNAs and shRNAs have been successfully established and

adapted for use with all three ET entities.

4. First results already have identified genes involved in chemoresistance in NB (FGFR2), MB (ATR, LYK5,

MPP2, PIK3CG, PIK4CA, and WNK4).

WP3 1. ET cell lines with regulatable expression of ET oncogenes have been established and characterised.

They have been validated for signalling and screening studies, although the TrkA inducible NB cells may

not be suitable for long term biological studies due to an unexpected over-induction of the TrkA construct

after prolonged periods of NGF stimulation.

2. EWS-FLI1 regulated miRNA profiles have been identified and are currently analysed and validated.

3. HTS with pre-mir and anti-mir libraries were performed in MB, NB and Ewing sarcoma cells under

condition of oncogene induction switched on or off. Data has been reported to the partners and a

deliverable (D3.1: miRNAs regulating ET cell viability (Month 24)) was reported.

4. We show that the MYCN regulated miR-17-92 directly controls expression and protein levels of DKK3

through binding to seed sequences in the 3'UTR of the gene.

5. We characterised MYCN regulation by Let-7 miRNAs in cells and animal models, identifying that MYCN

protein expression is negatively regulated by the Let7 miRNA, which is repressed by the LIN28B

transcription factor.

WP4

1. Comprehensive data sets of the response to MYCN overexpression at multiple molecular levels in SH-

SY5Y have been generated. These data sets can be summarised as follows:

a. of approximately 15,000 genes expressed the cell line over 600 are differentially expressed

upon MYCN overexpression.

b. of 1,073 miRNAs expressed 30 were found to be differentially expressed upon MYCN

overexpression.

c. of approximately 4,000 detectable proteins 278 were differentially expressed upon MYCN

overexpression.

2. A method to specifically identify newly transcribed genes in yeast (Dynamic Transcriptome Analysis) has

been successfully established in mammalian cells and used to map MYCN target genes.

3. The MYCN TF transcriptional network has been reconstructed and can now be used for identification of

NB outcome relevant genes which will be functionally validated. Such genes will be identified by cross

comparison of the MYCN network with non-amplified vs. amplified MYCN NB data sets, and NB early

stage differentiation data.

4. Genome wide MYCN binding sites have been mapped, and results suggest that a repressor complex

abrogates binding of MYCN near transcriptional start sites during differentiation of NB cells.

5. TF complex components in ESFT were characterised by ChIP-PCR and reporter gene assays identifying

a EWS-FLI1/E2F transcriptional module in Ewing sarcoma.

WP5

1. Data for quantitative profiling of protein phosphorylation in SY5Y-TR-TrkA cells is being re-analyzed.

Proteome data within the short term NGF stimulation up to 2 hours has been included to take changes in

protein abundance into account.

2. Data for quantitative profiling of protein expression in SY5Y-TR-TrkA cells (long term NGF stimulation, 0,

24, 48 hours) are being evaluated together with the generated phosphoproteomics data to create list of

hits for validation.

3. Quantitative profiling of protein expression in UW228 Myc-ER cell line upon induction of MYC for 48 and

72 hours has been performed.

4. Quantitative profiling of protein expression in the Asp14 shEWS-FLI1 cell line upon downregulation of

EWS-FLI1 for 18 and 48 hours has been performed.

5. Quantitative profiling of newly synthesized protein (pulsed SILAC) upon EWS-FLI1 downregulation (48

hours of doxycycline treatment) in the Asp14 shEWS-FLI1 cell line has been performed. MS and data

analysis on Asp14 shEWS-FLI1 work has been completed.

6. Protein array analysis has pinpointed PLCγ as potential critical TrkA downstream transducer.

WP6

1. The SY5Y-TR-Trka cell line shows no significant changes in MYC and MYCN levels within the early

phase of TrkA signalling.

2. As alternative to the TrkA model, an integrated, dynamic model of ERK, p38, JNK and AKT signalling in

response to growth factors and stress has been developed and validated experimentally using the SY5Y

cell line.

3. An MYCN interaction model relating to apoptosis and chemo-resistance in neuroblastoma has been

developed, partially validated using the SY5Y-MYCN cell line and is currently being refined and

parameterised. The model predicts that inhibition of the HMGA1-HIPK2 interaction facilitates the

induction of apoptosis in response to DNA damaging agents, which needs experimental validation.

4. The establishment and characterisation of a TrkA inducible cell line on a MYCN amplified background

suggests a late block of TrkA-mediated biological effects in MYCN-amplified cells.

WP7

1. A method for analysing quantitative data with influence networks (PIQuant for Path Influence

Quantification) was developed.

2. A model relating HMGA1, a molecule involved in tumour chemo-resistance, to the p53-Mdm2 module

was developed and partially validated experimentally.

3. We performed genetic perturbation experiments of players identified in CDK-Rb-E2F-Skp2 and p53-

MDM2/MDMX networks in ESFT cell lines.

WP8

1. An ET cell line panel representing NB, MB and ESFT with inducible TrkA, MYC and repressible EWS-

FLI1, respectively, was screened with the CEMM panel of inhibitors under induced and uninduced

conditions, and single dose response curves were determined.

2. Panobinostat , the most potent drug emerging from this screen was modified for immobilisation on a

solid matrix and successfully used for drug pull-down experiments.

3. A disease protein network based on compounds and their mapped target proteins in each NB, MB and

ESFT has been developed demonstrating that compounds indirectly affect larger target networks.

WP9

1. Regulation of E2F targets by EWS-FLI1 established.

2. Distinct growth regulatory roles of mir-631 and hsa-mir-552 in the Ewing sarcoma cell line model.

3. MYCN/MYC-mediated overactivation of the metaphase-anaphase checkpoint synergizes with loss of

p53-p21 function to prevent arrest or apoptosis of tetraploid neuroblastoma cells.

4. 150-gene signature representative for high ALK activity in NB established.

5. Genetically engineered mouse models (ALK mutation +/- MYCN overexpression) for neuroblastoma

established.

6. ALK inhibitors tested in above mouse models.

7. MAPK driven ETV5 oncogene was identified as a robustly regulated ALK target in NB and other ALK

activated cancers.

8. PCTK1, EPHA7 and AKAP12 identified as MYC dependent growth regulatory kinases in

medulloblastoma.

9. PRKCB identified as essential kinase target of EWS-FLI1 affecting histone code and Ewing sarcoma cell

proliferation.

10. Regulatable TrkA expression model established and tested in mouse models.

11. CUL1 was identified as a new potential target of EWS-FLI1 using an influence network linking EWS-

FLI1 to cell cycle and apoptosis phenotypes.

12. SIRT1 identified as a metastasis associated key regulator downstream of EWS-FLI1 in Ewing sarcoma.

WP 10

1. System Level Data Integration and Storage

Through the chemoinformatics work on proposing compounds for synergy screens, 3 additional

compounds were suggested, and one of these, a PI3K inhibitor, has been identified as a top candidate

from the synergy screen data for SY5Y.

Data generated on different parts of the project are being catalogued and transfers to the central data

warehouse have begun.

The CNIO and UCPH partners have identified databases containing information about pathways and

mutation phenotype data and also the algorithms required to integrate this data. Focus has been on

pipelines for cancer genome analysis and key databases and software helpful to integrate genomic data

with phenotypic information and drug response.

CURIE Systems Biology team (AZ) has developed computational tools that will be useful to 1) model

biochemical pathways with respect to the effect of perturbations on cell fate (MaBoSS software), 2)

analyse the structure of biological pathways involved in pediatric tumours (BiNoM software), 3) predict

the effect of combinatorial perturbations on particular cell fates (cell death, proliferation, survival, etc.)

and the side effects of such perturbations (for example, cell toxicity).

BiNoM (http://binom.curie.fr ), made freely available to the community, includes multiple features

connected with analysis and visualization of biological pathways (Bonnet et al, 2012). MaBoSS

(https://maboss.curie.fr/ ) implements a modeling framework based on qualitative approaches that is

http://binom.curie.fr/

https://maboss.curie.fr/

intrinsically continuous in time1. OCSANA (http://bioinfo.curie.fr/projects/ocsana/) allows identifying and

ranking optimal combinations of intervention points in a network to block signals from specified source

nodes to specified targets (Vera-Licona, 2012).

2. Exome Sequencing and Comparative Analysis of ETs

Next generation sequencing technology was used to fully sequence the exomes of the core cell lines of the

three childhood cancers in order to survey their exact mutual status. The cell lines representing the three ET

entities, which were analyzed so far, are:

ASP14 (Ewing Sarcoma cell line) with tet-regulatable EWS-FLI1.

SH-SY5Y (Neuroblastoma cell line)

IMR-5/75 (Neuroblastoma cell line)

3. The analysis framework of the data generated from the exome sequencing of the selected cell lines

involved identifying the differences between the ET cell lines sequenced and the human reference genome.

A number of sequence variants per cell line could be detected.

WP11 1. Systematic detection of gene fusion products in human cancers

The expected final results and their potential impact and use ASSET’s major goal is to identify mechanistically understood network vulnerabilities that can be exploited for new approaches to the diagnosis and treatment of major paediatric tumours. Elucidating such core mechanisms will (i) improve the understanding of and therapeutic options for these devastating childhood malignancies, and (ii) inform a rational approach to deal with the complexity of the pathogenesis of adulthood cancers. Several single targeted drugs with promising clinical activity have already been approved for the treatment of advanced cancer types. However, most single agents fail to induce complete responses, and the treated patients often develop resistance during therapy. Here, we will go beyond these initial targeted approaches to identify intelligent and complementary combinations of targeted agents based on mechanistic insights into ET-specific signalling networks. This approach matches therapy to genetic and functional aberrations, and represents the personalised medicine needed to increase treatment responses and to overcome therapy resistance induced by single-agents. As a result, we expect to obtain comprehensive insight into pathogenetic mechanisms of ETs based on validated computational models that are useful for (i) identifying fragile nodes where pharmacological interference will have maximal disease-specific effects while minimising side effects, (ii) improving therapeutic stratification of patients by molecular functional features and (iii) guiding the search for similar “core pathogenetic networks in adulthood cancers. In the current reporting period we have made excellent progress towards these goals as detailed below.

The address of the project public website http://www.asset-fp7.eu

http://bioinfo.curie.fr/projects/ocsana/


3.2 Core of the report for the period: Project objectives, work progress and achievements, project management 3.2.1 Project objectives for the period

Four deliverables were due during the reporting period, all at month 24:

D3.1 miRNAs regulating ET Cell viability

D5.1 Quantitative phosphoproteome analyses of the activated ALK- and TrkA-expressing NB cell lines

D6.1 Core model of Myc protein regulation by TrkA in NB Cell lines

D10.1 A description of sources of pathways, protein-protein complexes and mutation-phenotype data

They are described below in more detail. The specific objective for this reporting period included: D.3.1 List of miRNAs regulating ET cell viability (Mo 24) – VTT Task 3.1. Profiling miRNA expression in ETs. Large data sets consisting of miRNA expression profiles in

clinical NB, MB and ESFT samples, and cell lines (from E.E.T.-Pipeline and OD) will be provided and

expanded to use as a base to analyse the in vivo relevance of functional miRNA results collected in this

workpackage. For instance, an NB dataset for at least 100 cases assessed for 450 miRNAs exists, which we

plan to expand to apprx. 650 miRNAs combined with sequencing of miRNAs from fetal neuroblasts.

D5.1 Quantitative phosphoproteome analyses of the activated ALK- and TrkA-expressing NB cell lines Task 5.1. Global profiling of protein expression of the core ET cell lines by quantitative mass spectrometry. We will establish stable isotope labelling via amino acids in cell culture (SILAC) of the different ET cell lines. The proteome of induced and untreated SILAC lysates from the core ET cell models for MB, NB and ESFT will be analysed by high-resolution MS. Whole-cell lysates will be mixed and digested with trypsin. The resulting peptides separated off-line by iso-electric focusing into 12 fractions before online nanoflow RP-HPLC-MS/MS analysis. This standardised and largely automated workflow uses the latest generation of state-of-the-art high-resolution MS instrumentation, i.e. the LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) coupled to an online nano HPLC system (Proxeon Biosystems) available at UCPH. From replicated experiments, we expect to be able to quantify more than six thousand proteins in each of the cell lines. D6.1 Core model of Myc protein regulation by TrkA in NB Cell lines Task 6.1. Modelling of the effects of TrkA on MYCN and MYC expression and protein stability. This will involve measurements of activities of TrkA and downstream pathways1,2, in particular Ras/ERK and Akt/GSK3, which are known to regulate Myc protein stability by direct phosphorylation3. These measurements will be taken in part from the quantitative phospho-proteomics experiments conducted in WP5, and especially for the iterative phase of model refinement, supplemented by additional wet-lab experiments to obtain detailed kinetic data on selected network components. These measurements will be

correlated with myc gene promoter activity and Myc protein stability to develop a core kinetic model considering regulation of MYCN gene activity and protein stability by TrkA. This core model will be used to test how functionally induced elevation of MYCN and MYC expression or activity compare with genetically induced (by gene amplification) Myc deregulation. D10.1 A description of sources of pathways, protein-protein complexes and mutation-phenotype data Task 10.1. Construction of a state-of-the-art and data warehouse for ASSET. An analysis of data already generated by the consortium will include working with existing databases and data formats, understanding their semantic relationships and assessing gaps within the data as well as gaps in data connectivity and usability by the consortium for systems biology mining efforts. Similarly relevant publicly available data will be assessed. A datawarehouse will be constructed to host new data generated by the consortium as well as relevant publicly available data. Attention will be paid to facilitating queries across data types and linking to disease complexes and pathway reactions. Use of semantic web concepts will be evaluated if appropriate to enhance query possibilities. The developed database will be a shared resource of the ASSET consortium. Task 10.2: Data mining of publicly available mutation-phenotype data, protein-protein complexes and pathways and incorporation of these data within the relational database framework. The task will include building a database of childhood cancer pathway knowledge available from publicly accessible databases with an emphasis on pathways active in the tumour types in the study. The task will involve additional annotation of selected pathway circuits and protein-protein complexes key to this project. Using clinical phenotypes based on public databases such as OMIM and electronic patient records when relevant, the WP bridges the molecular level and the macroscopic description of cancer phenotypes in novel ways, building on recent advances in text mining and data integration1.

3.2.2 Work progress and achievements during the period

Below is a concise overview of the progress of the work broken down in Work Packages in line with the structure of Annex I to the Grant Agreement.

WP1. Reconstruction of Gene Regulatory Networks (GRN) driven by the ET

transcription factor oncogenes, MYCN, MYC and EWS-FLI1.

Project Objectives for the Period

The emphasis was on inferring GRNs and relating them to the pathology of ETs. For this purpose new

computational tools were developed, several network models were generated, and a systematic

experimental validation approach was started.

Summary of progress towards objectives and details for each task

Task 1.1. Generation of time resolved mRNA and miRNA data and ChIP-chip/seq data.

During this reporting period additional data for specific target structures involving EWS-FLI1 have been

generated extending previous data reported in period 1. Small scale Gene Regulatory Networks (GRNs)

involving EWS-FLI1 have been constructed using prior bibliographical and biological knowledge as well as

time resolved experimental data, and they have been used to construct detailed mechanistic models which

will provide the means to test alternative hypotheses about structure and network topologies. Validated

mechanistic models will be subsequently used for dynamic simulations and feed into other workpackages.

Moreover, a group of direct targets of MYCN with similar pattern of dynamic behaviour have been identified

which are involved in the regulation of the cell cycle (M phase), cell division, response to DNA damage etc.

More work is being conducted on characterizing the patterns in terms of known cellular pathways and re-

constructing GRNs from this group of genes. Finally, initial progress has been made towards the integration

of promoter occupancy data by an initial promoter analysis study, which revealed candidate MYCN

transcription factors.

We formulated three mathematical models in order to assess: (i) the regulation of E2F target genes, (ii) the

induction of a p53 response as a result of EWS-FLI1 knockdown, and (iii) the contribution of kinases CDK2

(directly EWS-FLI1 / E2F3 activated) and AKT (indirectly EWS-FLI1 regulated) to the regulation of FOXO1

target genes in Ewing sarcoma. In order to calibrate these models time course experiments were performed

in Asp14 cells. Early upon doxycycline induction of conditional EWS-FLI1 knockdown by specific shRNA,

RNA was extracted and mRNA expression of EWS-FLI1, E2F3, E2F4, JAG1, HEY1, SIRT1, CDK2, AKT and

FOXO1 was determined by real-time RT-qPCR. These data is being used to refine the models.

Additional time resolved data has been generated for the inducible 1c and 2 clones that consist of a

doxycycline inducible EWS-FLI1-specific shRNA. Time course of inhibition have been generated until day 17

of inhibition. Time course of expression profiles following EWS-FLI1 re-induction following withdrawal of

doxycycline in the medium has also been generated. These data have been compared with the time course

data generated by CCRI.

MYCN gene regulatory networks

Neuroblastoma is an embryonic tumour arising from immature sympathetic nervous system cells. Frequently

recurrent genomic alterations include MYCN and ALK amplification as well as recurrent patterns of gains and

losses of whole or large partial chromosomal segments. A recent whole genome sequencing effort yielded

no frequently recurring mutations in genes, other than those affecting ALK but further underscored the

importance of DNA copy number alterations in this disease, in particular for genes implicated in

neuritogenesis2. Here, we provide further evidence for the importance of focal DNA copy number gains and

losses in the pathogenesis of neuroblastoma through targeting MYCN regulated genes. A focal 5 kb gain

encompassing the MYCN regulated miR-17∼92 cluster as sole gene was detected in a neuroblastoma cell

line and further analyses of the array CGH data set demonstrated enrichment for other MYCN target genes

in focal gains and amplifications. Next we applied an integrated genomics analysis to prioritize MYCN down

regulated genes mediated by MYCN driven miRNAs within regions of focal or homozygous deletion. We

identified RGS5, a negative regulator of G-protein signalling, targeted by a focal homozygous deletion, as a

new bona fide indirect MYCN down regulated gene through MYCN driven miRNA binding. In addition, we

expand the miR-17∼92 regulatory network controlling TGFß signalling in neuroblastoma with the ring finger

protein 11 encoding gene RNF11. RNF11 is capable of antagonizing Smurf2-mediated inhibition of TGFß

signalling and is a critical component of the A20 ubiquitin-editing complex and NF-ß signalling and was

previously shown to be targeted by the miR-17∼92 member miR-19b. Taken together, our data indicate that

focal DNA copy number imbalances in neuroblastoma target genes that are implicated in MYCN signalling,

further underscoring the functional relevance of such alterations to the oncogenic phenotype of the tumour

cell and identifying new molecular targets for treatment.

MiRNA perturbation in gene regulatory networks in NB.

In previous work we already had discovered a major contribution for MYCN regulated miRNAs to the NB

oncogenic phenotype3. Amongst others, miR-17-92 was demonstrated as a central player in this respect.

Our initial findings have now been further extended and established by follow up studies also including

studies related to the prognostic relevance of miRNA signatures which are also intimately linked to the

underlying biology of the tumours in relation to their clinical behaviour. Interestingly, miR-17-92 was also

studied in another paediatric tumour (retinoblastoma) and shown to be of crucial importance in tumour

formation4,5.

Task 1.2. Inference of network structures of EWS-FLI1 and MYCN/MYC (dys)regulated GRNs

A novel method (SwitchFinder) for the analysis of time-resolved gene expression data based on inferring

time-points of switches between up-regulated and down-regulated behaviour of an individual gene has been

developed. The method is based on Bayesian statistics and Gibbs sampling. The method was applied to

kinetic data measured after induction of MYCN in SH-EP cell lines expressing a MYCN transgene under the

control of a tetracycline-repressible element developed by Partner FW (DKFZ). The statistical method

indicated genes with similar pattern of dynamic behaviour over 3 biological conditions. 10 such groups of

genes (patterns) were revealed comprising more than 1800 genes. In Table 1.1 we demonstrate 3 exemplar

groups with 700, 317 and 327 genes; the mean of expression values of the genes in each group is shown in

red. The most prominent pattern reveals direct targets of MYCN which are slightly repressed and then re-

induced upon remove of tetracycline. Gene Ontology analysis of this group of genes indicates their function

in the regulation of cell cycle (M phase), cell division, response to DNA damage etc. We are working on

characterizing the patterns in terms of known cellular pathways.

Dynamic patterns Gene Ontology terms Transcription Factors

cell cycle, M phase of mitotic cell cycle,

DNA replication, cell division,

chromosome segregation, response to

DNA damage stimulus, regulation of cell

cycle, DNA repair, cell cycle checkpoint,

cellular response to stress,

ribonucleoprotein complex biogenesis,

microtubule-based process, ribosome

biogenesis, spindle organization etc.

AP2

CREB

E2F

ELK

EGR

GATA

Klf4

NFY

NGFIC (EGR4)

NKX

NRF

STAT

skeletal system development, bone

development, negative regulation of

signal transduction, regulation of cell

motion, integrin-mediated signalling

pathway, regulation of cell migration,

blood vessel development, negative

regulation of cell communication,

extracellular matrix organization,

vasculature development, cell cycle

arrest, platelet-derived growth factor

receptor signalling pathway, negative

regulation of cell proliferation, cell

adhesion, developmental growth etc.

EWSR1-FLI1

FOXO

GFI

IRF

MAZR

MZF-1

NGFIC

NFAT

RREB1

enzyme linked receptor protein signalling

pathway, positive regulation of

macromolecule metabolic process,

regulation of transcription from RNA

polymerase II promoter, regulation of

apoptosis, chromatin modification,

negative regulation of cell proliferation,

cellular response to stress, localization of

cell, protein localization, response to

DNA damage stimulus, regulation of

hydrolase activity, neuron development

etc.

AP2, AP4

CREB

E2F

ELK

EGR

EWSR1-FLI1

GABP

MZF-1

NFkB

P53

Table 1.1 Three patterns of dynamic gene expression over three platforms reveal the differential

response of genes upon MYCN induction and doxorubicin treatment. Gene Ontology terms significantly

overrepresented in each group of genes as found by the program DAVID; TFs revealed by promoter analysis

of the genes with PScan based on both Jaspar and TRANSFAC PWM-libraries.

Three small scale network structures and corresponding mathematical models based on Ordinary Differential

Equation systems have been developed to study the following questions: (i) if EWS-FLI1 regulates E2F

target genes alone or in synergy with E2F3, (ii) if EWS-FLI1 knockdown induces a p53 response solely via

upstream activation of the NOTCH pathway (JAG1 induction) or also via downstream activation of the

NOTCH effector HEY1, (iii) the contribution of kinases CDK2 (directly EWS-FLI1 / E2F3 activated) and AKT

(indirectly EWS-FLI1 regulated) to the regulation of FOXO1 target genes in Ewing sarcoma. The models will

be parameterised with the data generated from Task 1.1 and alternative hypotheses will be tested by

evaluating the probabilities of competing models conditioned on time resolved qPCR data. Due to the

complex nature of such models a novel Markov Chain Monte Carlo algorithm has been developed6 which

allows for efficient exploration of the model parameter posterior distributions.

A meta-analysis using gene expression profiles from different paediatric tumours, including, Ewing's

sarcoma, rhabdoid sarcoma, neuroblastoma, medulloblastoma and desmoplastic round cell tumours using

Independent Component Analysis (ICA) has revealed common components reproducible in all independent

datasets. These have been associated to tumour related processes such as cell cycle, cell growth, chromatin

remodelling and invasion. A component common to all brain tumours is characterized by multiple ELAV-like

family genes, known to increase MYCN mRNA stability in neuroblastoma. The pathways involved in

neuroblastoma tumourigenesis have been further assessed through analysis of genomic alterations in

neuroblastoma, a CGH array for 255 neuroblastoma patients was analysed for segmental chromosome

alterations7. A list of genes was obtained which are frequently lost or gained in neuroblastoma and whose

gain or loss status has the highest association with a poor outcome in their regions. A clustering of over-

represented functional annotations of genes (GO Terms) with the David software was performed, which

indicated the role of DNA repair, neuron development, cell-cell junction, motor activity pathways in

neuroblastoma. More focused analysis identified rearrangements involving the ALK gene, associated with

aggressiveness. Detailed investigation of this association was published in Cancer Research8.

Task 1.3. Refinement of inferred GRNs by integration of promoter occupancy data.

Promoter analysis of genes with similar dynamic expression pattern revealed TFs whose relation to MYCN

(synergistic or antagonistic) should be verified with ChIP or ChIP-chip, see Table 1.1 (Please note that all

groups contain binding sites for MYCN). Some of the TFs are already described in the literature, e.g. as

competitors of MYC for the promoter occupancy on the targets (FOXO). This work is in progress and brings

hypotheses on the interactions in GRNs of ETs.

Significant results

1. Potential network structures have been identified and detailed mechanistic models based on ODEs

have been developed to study the regulation of E2F target genes, induction of a p53 via the NOTCH

pathway, and contribution of the kinases CDK2 and ATK to the regulation of FOXO1 target genes in

Ewing sarcoma.

2. Additional time resolved RT-qPCR data have been generated for the validation of these models.

3. We have established a statistical method for the inference of GRNs based on time-resolved

transcriptome data and applied it to the neuroblastoma SH-EP cell lines inducing MYCN under the

control of tetracycline.

4. Markov Chain Monte Carlo algorithms for efficient inference of ODE models have been developed.

5. Multiple patterns of dynamic gene expression in response to MYCN induction were deduced,

revealing targets of MYCN, both activated and repressed. This enlarged our knowledge about

mechanisms of regulation by MYCN, in particular, on its ability to overcome proliferation arrest under

doxorubicin treatment. Activating and repressive effects of MYCN on its targets were confirmed in

clinical data on neuroblastoma patients bearing MYCN amplification.

6. Promoter analysis of genes in each pattern indicated transcription factors which possibly interfere

with MYCN. The hypotheses will be verified with ChIP, towards the milestone M.1.5. “GRN and motif

structures for MYCN and MYC” (Mo48)

Statement on the use of resources

Resources were largely used as planned as detailed below.

Participant number

Participant short name Person-months per participant

Personnel & Resources used: November 1, 2011 to October 31, 2012

4 CURIE 1 1 Postdoc – 1PM

5 DFKZ 5 1 PhD Student – 5PMs

9 UCL 6.46 Postdoc: Vasileios Stathopoulos –

6.46PMs

10 UBERN 6.2 Salaries for 2 PhD students (50%) for 2.4 months and 1 PhD student (50%) for 1.4 months.

Total as of October 31, 2012

18.66

WP2. Targeted and global high content network perturbation screens with

siRNAs and drugs for network reconstruction.


This WP employs initial models of ET GRNs produced in WP1 to generate high-content perturbation data

that can be used to (i) test GRN structures predicted by the models; (ii) enrich and refine GRN structures by

functional data, thereby enabling the iterative cycle of modelling informing experimental work and

experimental work informing the modelling; (iii) functionally link GRN structures to biological outcomes such

as cell viability. In addition, we are performing (iv) hypothesis driven and mechanistically oriented drug

discovery work by unbiased compound screens as well as (v) predicted synthetic effects between siRNA –

siRNA and siRNA – drug combinations. The activities in this reporting period concentrated on the

transcription factor networks deregulated in Ewing sarcoma, medulloblastoma and neuroblastoma (EWS-

Fli1, MYC and MYCN), and siRNA and drug screens that uncover vulnerabilities arising from the

deregulation of these transcription factors.


As a general summary of WP2, the screening platforms for drugs, siRNAs and shRNAs have been

successfully established and adapted for use with all three ET entities. Several screens have already been

performed and are in the validation phase. The rest are on their way and should give results during the third

year. UBERN and CURIE, as well as UBERN and DKFZ have performed a kinome-wide siRNA screens in

established medulloblastoma (MB) and neuroblastoma (NB) cell lines, respectively, to identify genes

involved in cell survival and resistance to Cisplatin. This data has been already published9,10 or submitted.

Partner CEMM will start work on WP2 during year three of the project. Overall the work package is

progressing well.

Task 2.1. Functional role of EWS-FLI1 downstream genes to refine the ESFT network model.

A part of the influence network constructed by CURIE was validated using systematically collected

siRNA/QPCR data (11 genes) in order to test and refine the influence network. The influence network

appeared to be consistent with the data: 9 connections were confirmed, other influences can be explained by

indirect paths in the network (except one). A publication on the Influence network and its validation by the

siRNA/QCPR data is in preparation.

CURIE (AZ) worked on the organization of high-throughput imaging phenotypic data obtained as a result of

HTS siRNA screening of EWS/FLI-1 inducible cell line (273 different transfections, fluorescent cellular

phenotyping). The results were organized in the form of an interactive web-service, allowing to access to the

screening results including the results of the statistical data treatment and the raw imaging data. The service

will be soon made available to the members of the consortium. In addition, a statistical method for

quantification of High-Content Screening (HCS) data was developed. A probabilistic model allows

representing and comparing cell populations as well as fully exploiting the HCS-specific information:

“dependence structure of population descriptors” and “within-population variability”11. The identification of

new genes and pathways that modify cellular phenotypes is in progress; the updating of the network based

on this analysis has already started.

Task 2.2. Identification of synthetic lethal genes with amplified or deregulated MYCN in NB cells.

Based on our pilot siRNA screening project using a limited number of candidates (n=250, see also WP9 for

details) we are currently scaling up the siRNA screening procedure for a druggable genome siRNA screen at

the (DKFZ). This screen will be performed in the ASSET culture model system for MYCN amplified

neuroblastoma cells, IMR5/75 with regulatable MYCN. This means that we will screen for synthetic lethal

combinations using the druggable genome under two conditions, i.e (i) with MYCN on, and (ii) with MYCN

off. We have tested the feasibility of this approach and defined a set of positive controls, namely those which

reduce the fitness of the cells upon knock down only or predominantly in the MYCN on but not or less in the

MYCN off condition.

In NBs with a non-amplified MYCN gene MYCN still can be overexpressed due to enhanced signalling by

RTKs, especially TrkA. More recently, also ERBB family receptors have been associated with prognosis in

NB and MB. RTK stability and function is regulated by ubiquitination and de-ubiquitination. In order to identify

RTK-regulating enzymes, Weizmann performed siRNA screens of 95 of the ca. 100 known human

deubiquitinating enzymes (DUBs) in human KB cells assessing ligand induced degradation of ERBB1. This

screen identified Cezanne-1, a DUB previously implicated in NFkB signalling, and its family member,

Cezanne-2, as candidate DUBs of RTKs. The simultaneous knockdown of Cezanne-1/2 moderately

enhanced both basal and inducible receptor degradation. Comparative Genomic Hybridization (CGH)

indicated that Cezanne-1 is amplified in one third of the NCI-60 panel of tumour cell lines. Currently, we are

screening several NB and other ET lines for aberrant expression of Cezanne-1/2, and analyze potential

interactions of Cezanne with Trk and ALK family RTKs.

Task 2.3. Identification of biological consequences of global ET GRN perturbations in cultured ET

cells to refine the GRN models

Task 2.3 was expanded from a targeted approach to a broader, unbiased screen using VTT’s druggable

siRNA library. The screen will be performed once with and without induction of the oncogene or anti-

oncogene shRNA (in the case of EWS-Fli1) using pooled siRNAs and measuring cell viability as an endpoint.

This decision was made in the ASSET-project meeting in May 2011 taking account of the fact that the

reconstruction of gene regulatory networks (GRNs, see WP1) will take time (delivery month 48), and that

therefore there would be only one year left to design the siRNA library, order the siRNAs and perform the

screens with a variety of endpoints. Moreover, a more unbiased approach may highlight vulnerabilities

around GRNs that would not have been identified using a hypothesis driven model and the siRNA screening

results can be incorporated to GRNs at an earlier phase of the project to support generation of a systems

view.

So far, as prelude to the screen the functionality of ET cell models provided by other ASSET partners has

been tested and confirmed by quantitative RT-PCR:

Neuroblastoma (NB): MYCN induction (24 fold increase in NMYC mRNA in response to tetracycline

exposure for 24 hours) was confirmed in inducible SY5Y cells (SY5Y/6TR (EU)/pTrex-Dest-

30/MYCN cells from DKFZ)

Medulloblastoma (MB): MYC induction (4 fold) was confirmed in tamoxifen induced MB (UW-228)

cells (24 h induction, cells received from UBERN)

Ewing Sarcoma Family Tumours (ESFT): EWS-FLI1 knock-down (79 % decrease in EWS-FLI1

mRNA in response to 64 hours exposure to doxycycline) was confirmed in conditional ESFT (A673)

cells received from CCRI.

In addition, the optimal cell number and screening conditions for a 72 hour high-throughput siRNA screen

using cell viability as an end point has been defined. The high-throughput siRNA screens are scheduled for

the third year of the consortium.

Our working hypothesis assumes that MYCN regulates the balance between symmetric and asymmetric

divisions of neural stem cells, or their early progenitors. To test this model, partner Weizmann is performing

siRNA and other screens aimed at identifying markers of stemness, which are driven by MYCN

overexpression. Our screens are making use of two inducible cell lines (SH-SY5Y-TO-MYCN, IMR5/75-

MYCN-SH), previously established by member labs of ASSET. These cellular systems allow us to compare

the fine cellular states under high or low MYCN expression. We are currently establishing specific molecular

markers of pluripotency, which we test on the inducible cell lines. We will continue by screening for proteins

that together with MYCN co-regulate signalling pathways leading to enhanced stemness. By mutating or

inhibiting the signalling proteins we will test the hypothesis that altering the function of certain signalling

pathways can overcome the effects of aggressiveness in the engineered NB cell lines.

Task 2.4. Application of cell based HTS to identify growth inhibitory compounds and synthetic

lethalities in genetically engineered isogenic ET cell line pairs.

The cell number for the high throughput screen (HTS) has been optimized (1500/well for NB and MB cells

and 1700 for ESFT cells), and VTT has performed the HT compound screens. ESFT MB and NB cells (+/-

induction) have been screened using Biomol (80 known kinase and phosphatase inhibitors; 10, 1, 0.1 and

0.01 µmol/L), LOPAC (1280 compounds; 1 and 0.1 µmol/L), Microsource Cancer (80 compounds) and

Microsource Spectrum (2000 compounds including most of the known drugs and other bioactive compounds

and natural products; 1 and 0.1 µmol/L) libraries totalling 3500 compounds (see table 2.1 below). Results

from the screens were reported to the partners in the annual meeting in Majorca 2012 and shared with

partners electronically.. The HT compound screens will be continued with a newly acquired library of FDA

approved and 96 novel compounds and the results are analyzed and validated by the partners during the

coming years (delivery month 60).

Table 2.1 Compound libraries and concentrations of drugs used for screening ESFT MB and NB cells

(+/- induction).

Figure 2.1 Hits specifically reducing the proliferation of MYC expressing UW228 MYC_ER

medulloblastoma cells. Black bars, MYC-ER induced by 4–hydroxytamoxifen treatment (4OHT); grey bars,

uninduced control cells.

An example of compound screen yielding 22 hits specifically inhibiting MYC dependent medulloblastoma cell

growth is shown in figure 2.1. MYC has been previously shown to be linked to Resveratrol induced cell cycle

arrest and apoptosis in medulloblastoma12.

To search for synthetic lethality, 1c cells derived from A673 parental cell line are currently being infected with

lentivirus vectors carrying 20,000 different shRNAs at CURIE. Cell pools will be analysed under conditions

where EWS-FLI1 is upregulated or downregulated (via a doxycycline inducible shRNA). Each shRNA

transfectant can be uniquely identified by a barcode. Following infection with the shRNA library shRNAs that

are over- or underrepresented in growing cells (and hence regulate cell proliferation) will be identified by next

generation sequencing.

Task 2.5. Combination of data from chemical biology with signalling network model to refine critical

pathways for ET proliferation.

The work on this task has not started yet.

Task 2.6. Identification of siRNAs sensitizing NB and MB cells to selected drugs.

UBERN and CURIE have performed a kinome-wide siRNA screen in established medulloblastoma (MB) cell

lines to identify genes involved in cell survival and resistance to the chemotherapeutic agent Cisplatin. A set

of 6 genes comprising ATR, LYK5, MPP2, PIK3CG, PIK4CA, and WNK4 were identified as contributing to

both cell proliferation and resistance to Cisplatin treatment in MB cells. An analysis of the expression of the 6

target genes in primary MB tumour samples and cell lines revealed overexpression of LYK5 and PIK3CG.

The results of the siRNA screen were validated by target inhibition with specific pharmacological inhibitors. A

pharmacological inhibitor of p110gamma (encoded by PIK3CG) impaired cell proliferation in MB cell lines

and sensitized the cells to cisplatin treatment. Together, our data show that the p110gamma

phosphoinositide 3-kinase isoform is a novel target for combinatorial therapies in medulloblastoma. We have

published these data 9.

UBERN and DKFZ have performed a kinome-wide siRNA screen in established neuroblastoma (NB) cell

lines to identify genes involved in cell survival and resistance to Cisplatin. As a result we identified the

fibroblast growth factor receptor 2 (FGFR2) as an important determinant of Cisplatin resistance.

Pharmacological inhibition of FGFR2 confirmed the importance of this kinase in NB chemo-resistance.

Silencing of FGFR2 sensitized neuroblastoma cells to Cisplatin-induced apoptosis, which was regulated by

the downregulation of the anti-apoptotic protein BCL2. Mechanistically, FGFR2 was shown to activate

protein kinase C delta (PKC-δ) to induce BCL2 expression. FGFR2, as well as the ligand FGF-2, were

consistently expressed in primary NB and NB cell lines, indicating the presence of an autocrine loop.

Microarray data revealed that FGFR2 expression correlates MYCN amplification and with advanced stage

disease, demonstrating the clinical relevance of FGFR2 in neuroblastoma. These findings suggest a novel

role for FGFR2 in chemo-resistance and provide a rationale to combine pharmacological inhibitors against

FGFR2 with chemotherapeutic agents for the treatment of neuroblastoma. These findings are reported in the

following submitted manuscript: Salm, F., Cwiek, P., Ghosal, A., Bucarello, A.L., Largey, F., Wotzkow, C.,

Bodmer, N., Gross, N., Westermann, F., Schäfer, S., and Arcaro, A. RNA interference screening identifies a

novel role for autocrine fibroblast growth factor signalling in neuroblastoma chemoresistance..

Task 2.7. Use of siRNA screens to ascertain the identity of the kinases which mediate synergistic

response induced by kinase inhibitor combinations

The work on this task has not started yet.

Significant results

5. MYCN and EWS-FLI1 regulated genes were identified.

6. Combinatorial assessment has revealed synthetic lethalities, e.g. with amplified MYCN and spindle

checkpoint genes.

7. The screening platforms for drugs, siRNAs and shRNAs have been successfully established and

adapted for use with all three ET entities.

8. First results already have identified genes involved in chemoresistance in NB (FGFR2), MB (ATR, LYK5,

MPP2, PIK3CG, PIK4CA, and WNK4).

Deviations from Annex I and their impact

Validation work of DKFZ (partner 5) on candidates (e.g. MAD2L1) from the functional screens has been

moved to WP9 as it is better suited there. This has been communicated with and agreed by the scientific

coordinator.


Resources were largely used as planned as detailed below.

Participant number


Personnel & Resources used to date November 1, 2011 to October 31, 2012


6 VTT 4.6 1 Postdoc – 4.6PM

10 UBERN 6.2 2 PhD students (50%) for 2.4 months and 1 PhD student (50%) for 1.4 months

11 Weizmann 27.61 1 Postdoc - 11.96PMs, 1 Postdoc- 5.45PMs, 1 Postdoc - 9PM, PI -1.2PMs.


40.41

WP3. microRNA expression in ETs: regulation by ET oncogenes, identification

of targets & effects on proteomes.


MicroRNAs (miRNAs) are central connections between the genome, transcriptome and proteome. Their

crucial roles in cancer are increasingly recognised. We are assessing aberrations of miRNA expression in

ETs and their effects on the transcriptome and proteome to include miRNA effects into our signalling network

models. Global and focussed approaches are combined: high-throughput cell-based miRNA transfection

screens with a comprehensive library of >800 pre-mirs have generated data concerning ET cell proliferation.

This objective has been completed and the three ET model systems have been screened in cell-based high-

throughput miRNA transfection screen using Dharmacom miRIDIAN Mimics (N=810) and miRIDIAN

inhibitors (N=896). The corresponding deliverable (D3.1: miRNAs regulating ET cell viability (Month 24)) is

reported separately.

The analysis of the effects of selected miRNAs regulated by EWS-FLI1, MYCN/MYC (for instance miR-17-

92), ALK and TrkA on the transcriptome and proteomes of ET cells has started. The EWS-FLI-1-regulated

miRNAs were identified by microarray miRNA profiling of a Ewing’s sarcoma cell line model expressing an

inducible shRNA targeting EWS-FLI1. Validation of the ET cell specific, oncogene regulated miRNAs from

HT-screens are ongoing.

The generation of time-resolved miRNA expression data after EWS-FLI1, MYCN and MYC perturbation also

has commenced. A microRNA-seq time course experiment has been performed in the NB MYCN inducible

cell line SY5Y-MYCN, to identify MYCN target miRNAs (UCD).


Task 3.1. Profiling miRNA expression in ETs

CURIE has profiled miRNA expression in 40 ET tumours. Correlation analysis has been applied to

microRNA and transcriptome patient data. Pair-wise correlation coefficients have been computed for each

miRNA in the dataset versus each transcript by Pearson coefficient. The large-scale analysis has permitted

the identification of pairs of miRNAs and mRNAs with a negative and positive correlation. The study of

miRNA and its impact on target expression is now being tested by modulation of miRNA expression, either

inhibition or overexpression. Moreover, the effect of these miRNA/mRNA pairs on the phenotype of Ewing

sarcoma tumour growth, ET cell invasion and migration will be studied. UCD has performed a microRNA-seq

time course experiment in the NB MYCN inducible cell line SY5Y-MYCN, to identify MYCN target miRNAs

(see WP4).

Task 3.2. Global approaches to identify critical miRNAs for ET cell viability using high-throughput

screening (HTS).

CURIE also has correlated the expression of miRNAs in Ewing sarcoma patient samples with changes in

mRNA transcript expression for all genes showing a modulation of their cognate mRNA levels in the tumour

phenotype. The structure of the Ewing sarcoma miRNA-transcriptome correlation network highlighted groups

of miRNAs and target genes that are believed to participate cooperatively in post-transcriptional gene

regulation. Among them, one module is composed by miRNAs and target genes known to be specifically

expressed in muscle. Since the Ewing's sarcoma samples are surgically derived from patient bones, the

observed results suggest contamination of patient derived samples by tumour neighbouring tissues and non-

tumour cells. However, another relevant module is enriched by genes known to be modulated by EWS-FLI1,

accordingly to the published Ewing's sarcoma signature genes. This module is predicted to be regulated

cooperatively by hsa-miR-145 and hsa-miR-199b. A list of 20 genes inside this module has been selected for

validation by experiments. To enable this analysis CURIE has developed a new statistical method for

assessing connections between mRNA and microRNA expression changes, based on the antagonism

pattern13.

Recently, cancer cell lines have been shown to express substantial amounts of mRNA isoforms with

shortened 3' untranslated regions (UTRs). This is an enormously important observation from the point of

view of post-transcriptional regulation, because if the 3’UTR of a mRNA is shorter or missing, miRNAs and

other regulatory molecules are no longer able to bind. CURIE has investigated this mechanism in Ewing's

sarcoma data, as it could be possibly linked to observed positive correlations. We based the analysis on a

large dataset of 75 patient samples from Affymetrix U133A GeneChip microarrays, and the team developed

a method to characterize specific loss of 3’UTR of some mRNAs14. Initially, Affymetrix individual probes were

regrouped into customized probesets mapping specifically the CDS or the 3’UTR of the transcript, according

to RefSeq annotation. Then, candidate 3’UTR shortening events are identified by statistical differential

expression analysis of customized probesets in different patient samples. Potentially, these genes have

oncogenic role and try to avoid inhibition by microRNA. Their list is compiled and possible experimental

validations are currently discussed.

VTT has confirmed the functionality and suitability of the cell lines provided for screening, and the HTS

conditions have been established for NB, MB and ESFT models (see WP2 Task 2.3). The screens with

Dharmacom miRIDIAN Mimics (N=810) and miRIDIAN inhibitors (N=896) have been performed at least

twice in all selected cell models both with and without induction (Figure 3.1). In total, screens consisted of

84x 384-well plates. The normalization of all the results and the further bioinformatic analysis of the data

were performed during spring 2012 and results were presented in the ASSET annual meeting. Results were

also distributed to partnersfor further analyses by the partners. For continuation partners UBERN, CCRI and

DKFZ are validating the hits found in the miRNA high throughput functional screens. The deliverable D3.1:

miRNAs regulating ET cell viability (Month 24) is reported separately.

Figure 3.1 Cell viability results of pre-mir and anti-mir screens performed in MB (UW228 Myc/ER) cells

without (upper part, no MYC induction) and with Tamoxifen treatment (lower part, MYC induction). The pre-

mirs (Dharmacom miRIDIAN Mimics) were distributed in three 384 well plates (on left hand side) and the

anti-mirs (miRIDIAN inhibitors) in another three 384-well plates. The results from wells derived from different

plates are separated in the figure using dashes lines. The positive controls (AllStars death, PLK1 siRNA and

KIF11 siRNA) can be clearly identified due to their growth inhibitory effect whereas the majority of negative

controls (buffer, AllStars negative control, miRIDIAN) are located close to zero (no effect on cell viability),

indicating that the HTS has functioned as expected.

Figure 3.2 Example of the specific results from the high throughput miRNA screens performed in MB

(UW228 Myc/ER) cells without (light shades, no MYC induction) and with Tamoxifen treatment (darker

shades, MYC induction). The hsa-mir-106b and hsa-mir-892a seem most specific for the MYC induction.

CCRI has successfully established the PAR-CLIP method for genome wide identification of miRNAs and

target mRNAs engaged in the RISC complex. PAR-CLIP was performed on the Asp14 cell line model in

duplicate experiments of uninduced and 56 hour EWS-FLI1 suppressed cells to assess mRNAs involved in

AGO2 containing RISC complexes. Results were compared to mRNA patterns obtained by expression

profiling in the same cell line under identical conditions. Further, candidate targets (E2F3, RAS, YEATS4,

CGGBP1, USP37) identified by PAR-CLIP were validated by PCR and, in part, also by protein analysis.

Surprisingly, only very little overlap between EWS-FLI1 dependent mRNA expression levels and AGO2

association was observed. In fact, knockdown of EWS-FLI1 changed miRNA-association patterns of 3´-

UTRs independent of miRNA and mRNA expression. Testing mRNA expression of gene products known to

be involved in miRNA biogenesis or function, or being associated with AGO1 and/or AGO2 containing

complexes, we found 70-80% to be changed in expression, the majority being suppressed upon EWS-FLI1

knockdown, providing a potential explanation for the observed shift in general miRNA/mRNA association

patterns. Yet the lack of concordance between miRNA/mRNA abundancies and AGO2 association patterns

raises the question if AGO2 binding leads to either stabilization or destabilization of mRNA targets, and how

this translates into protein expression regulation. This question is currently addressed in AGO2 knockdown

experiments and together with UCPH by quantitative proteome profiling experiments using SILAC.

Task 3.3. Targeted approaches to identify targets of miR-17-92 (Myc) targets, and TrkA- and ALK-

regulated miRNAs in NB by transcriptomics and proteomics.

UKE has established and characterized a neuroblastoma cell line (SY5Y) for Tet-inducible expression of a

codon optimized TrkA cDNA. The system is not leaky, and no TrkA autophosphorylation was observed in the

absence of NGF (Figure 3.3). In turn, NGF induced strong TrkA phosphorylation, cell differentiation, cell

cycle arrest and apoptosis. This system was made available to all ASSET collaboration partners for further

analysis.

Figure 3.3. Induction of TrkA expression in the SY5Y-TR-TrkA#1 cell lines by tetracycline treatment,

and phosphorylation of the induced TrkA receptor by stimulation with the ligand NGF.

In the course of further experiments studying the regulation of differentiation and apoptosis by NGF it turned

out that NGF treatment led to a upregulation of TrkA expression from the transduced construct commencing

between 2-4 hours after NGF stimulation. This is due to a delayed enhanced Tet promoter activity following

NGF stimulation. Thus, while this cell line is ideal for short term experiments, we are concerned that the

upregulation of TrkA may compromise its usefulness for experiments involving long-term NGF stimulation.

To address this issue we are constructing two new SH-SY5Y model lines with stable TrkA expression or

inducible expression from an alternative promoter type (UKE and UCD). The current cell line can be used for

short term signalling experiments, such as the reverse protein arrays and the phospho-proteomics

experiments in WP5, which are proceeding as planned. In this respect the SILAC labelling for SH-SY5Y cells

(the TrkA model) has been optimised by UCPH, and conditions for the reverse protein array work have been

established.

To analyse interplay between TrkA and MYCN, the MYCN-amplified cell line, IMR-5, was stably transfected

with Tet-inducible TrkA expression vectors. This new cell model was phenotypically analysed (details

reported in WP6). In addition SY5Y-TR-TrkA has been transfected with a tamoxifen-regulatable dominant

negative MYCN (“Omomyc”, details also reported in WP6).

The focus on miR-17-92, which is upregulated by MYCN, yielded exciting results. Several miRNAs are

activated downstream of MYCN as established in different published studies15,16. The miR-17-92 is a well

known oncogenic miRNA cluster and directly up regulated by MYCN. Proteome, mRNA and miRNAome

profiling of a miR17-92 inducible system revealed broad but reproducible effects of this cluster on

transcriptome and proteome3. In particular this first study uncovered an important role for the miR-17-92

cluster in control of the TGFbeta signalling pathway as an important downstream axis controlled by MYCN.

Of further interest, this study illustrates that this miRNA cluster targets simultaneously multiple components

of the TGFB pathway thus illustrating a particular mode of action of miRNAs in complex regulation of

oncogenic signalling. In two follow up studies, we provided further evidence for a crucial role of miRNAs in

down regulating protein coding genes under control of MYCN. Until recently, this was mainly assumed to be

the result of the repressive direct interaction through MIZ1, but our own studies and other recent data are

show that multiple mechanisms, including miRNA controlled regulation, drive the downregulation of MYCN

target genes. In a first follow up study, we focussed on the DKK3 gene, a member of the DKK family of WNT

inhibitors. In previous studies, we identified DKK3 as one of the most robustly down regulated protein coding

genes under control of MYCN. We now show that miR-17-92 directly controls expression and protein levels

of DKK3 through binding to seed sequences in the 3'UTR of the gene. Presently, the mode of action of

DKK3 is still enigmatic and further studies are warranted to investigate whether this might serve as a

possible therapeutic target17. More recently, we finalized a second follow up study, illustrating that MYCN

downregulated genes are enriched in focal deletions in primary neuroblastomas. Using an integrated

genomics analysis of miRNA, mRNA and DNA copy number data sets, we were able to identify several

candidate protein coding genes that downregulated due to MYCN driven miRNA expression18. Two of these,

RGS5 and RNF11, were formally established as MYCN down regulated genes under control of miRNAs that

were MYCN directly up regulated.

MYCN regulation by miRNAs: Identification of the LIN28B/Let7/MYCN axis. The hypothesis driven

approach was expanded to Let-7 miRNAs. MYCN harbours binding sites for Let-7 miRNAs in its 3'UTR.

LIN28B regulates developmental processes by modulating miRNAs of the let-7 family. A role for LIN28B in

cancer has been proposed, but has not been established in vivo. We detected LIN28B amplifications or

overexpression in high-risk neuroblastomas. LIN28B repressed let-7 family miRNAs and induced MYCN

expression in neuroblastoma cells in vitro and in mouse tumours (Fig. 3.4). LIN28B-let-7-MYCN signalling

induced proliferation and blocked differentiation of normal neuroblasts and neuroblastoma cells. The effect of

LIN28B knock-down was rescued by overexpression of let-7-resistant MYCN.

Figure 3.4.The Let-7 miRNA regulates MYCN. (A) The Lin28b-let-7-MYCN regulatory axis is shown

diagrammatically. Lin28b down-regulates let-7 miRNAs, which target MYCN. Low let-7 miRNA levels lead to

high MYCN levels and subsequently to the development of neuroblastoma. (B) Western blot analysis

showing high Lin28b transgene expression in murine tumours and MYCN up-regulation. Neither Lin28b nor

MYCN expression was detectable in control tissues. (C) Stem-loop RT-qPCR analysis shows that

expression of all 8 let-7 miRNAs were absent in tumours driven by Lin28b. Expression in the adrenals or

brains of Lin28b litter-mates and brains or neuroblastomas derived from TH-MYCN transgenic mice are

shown as controls.

To analyse the tumour-initiating capacity and oncogenicity of Lin28b in vivo, mice were generated by knock-

in of the CAG-LSL-Lin28b-IRES-Luciferase vector (LSL-Lin28b) into the ROSA26 locus. Cross-breeding

these mice with DBH-iCre mice targeted Lin28b expression to the neural crest. Abdominal tumours

developed in 7 of 18 DBHiCre;LSL-Lin28b mice between 36 and 62 days of age. Autopsy revealed uni- or bi-

lateral adrenal tumours in all mice, reflecting the most frequent location of neuroblastomas in humans.

Tumours consisted of small round blue cells and expressed the neuroblastoma markers, DBH, TH and

Phox2b. Macroscopic tumour appearance, primary tumour sites, tumour histology and marker gene

expression confirmed these tumours as neuroblastomas. We could successfully serially transplant and grow

the original tumours from this mouse model, confirming them as fully transformed malignant tumours. Both

the Lin28b and MYCN proteins were strongly expressed in all tumours, and members of the let-7 miRNA

family were significantly down-regulated. These experiments demonstrate that over-expressing Lin28b in the

neural crest can drive neuroblastomagenesis in mice, supporting LIN28B as an important oncogene for

neuroblastoma and potentially other malignancies. The LSL-Lin28b mouse model will be used for further

perturbation experiments (WP9). Inhibiting MYCN with the targeted inhibitor, JQ1, in DBHiCre;LSL-Lin28b-

derived tumours in vivo resulted in widespread apoptosis and proliferation arrest of tumour cells (Fig. 3.5).

These results demonstrate the addiction of the tumour cells to the LIN28B-let-7-MYCN axis, and show that

MYCN is a central signalling element in these tumours. These experiments indicate that therapeutic

approaches aimed at inhibiting Lin28b or let-7 family re-expression may be useful to circumvent MYCN

addiction in high-risk neuroblastomas.

Figure 3.5. Pharmacological inhibition of MYCN function blocks tumour growth. (A) Tumours derived

from DBH-iCre;LSL-Lin28b transgenic mice were retransplated into immune-compromised nu/nu mice. We

proposed that the Lin28b-Let7-MYCN axis is relevant in these tumours, with MYCN driving neuroblastoma

formation. As LSL-Lin28b is ectopically expressed as a knock-in transgene driven by the CAG promotor,

transcriptional regulation of Lin28b by MYCN can be excluded in this system. JQ1 is a bromodomain

inhibitor, which inhibits both MYCN transcription and transcription initiated by MYCN. Tumour bearing mice

were treated twice daily with intraperitoneal injections of JQ1 or solvent (control) for three consecutive days.

(B) Western blot analysis of MYCN and Lin28b expression in two tumours from mice treated with JQ1 and

two tumours from untreated mice. Actin serves as a loading control. (C) Representative histology is also

shown of the two JQ1-treated and untreated tumours after HE staining or immunohistochemistry for Ki67

(indicating proliferating cells) or ClCasp3 (indicating cells undergoing apoptosis). Note the extensive necrosis

observable in JQ1-treated tumours.

miRNAs downstream of TrkA: miR-542-3p targets survivin. We have previously shown that miR-542 is

induced by TrkA19. Analysis of miRNA expression in a small cohort of primary neuroblastomas revealed that

miR-542 is down-regulated in tumours from patients with adverse outcome, and that miR-542 expression

inversely correlates with MYCN gene amplification. We addressed the function of miR-542 in neuroblastoma

tumour biology using cell and mouse models. miR-542-3p re-expression in neuroblastoma cell lines reduced

cell viability and proliferation, induced apoptosis and downregulated survivin expression. Survivin expression

was also inversely correlated with miR-542-3p expression in primary neuroblastomas. Reporter assays

confirmed that miR-542-3p directly targeted survivin in neuroblastoma cells. Downregulation of survivin using

siRNA phenocopied miR-542-3p re-expression in neuroblastoma cell lines, while the enforced expression of

survivin partially rescued the phenotype induced by miR-542-3p re-expression. Treating nude mice bearing

neuroblastoma xenografts with miR-542-3p-loaded nanoparticles repressed survivin expression, decreased

cell proliferation and induced apoptosis in the xenograft tumours. We conclude that miR-542-3p exerts its

tumour suppressive function in neuroblastoma, at least in part, by targeting survivin.

Figure 3.6. miR-542-3p regulates survivin and Aurora B kinase expression. Western blot showing

AURKB and survivin expression in neuroblastoma cell lines after control (NC), miR-542-5p or miR-542-3p re-

expression. Actin was used as a loading control.

Figure 3.7.: Re-expression of miR-542-3p reduces SHEP cell viability. Cell viability was measured using

MTT assays at the indicated time points.

In order to study ALK signalling a stable system for the Tet-inducible expression of wt ALK, ALKR1275Q

and ALKF1174L in the ALK-negative neuroblastoma cell line SKNAS was established by UKE in order to

analyse the effects of wild-type ALK and mutated ALK on neuroblastoma cells. Initial analysis of signalling

was performed using Western blot analysis20. Further high throughput Affymetrix gene array analysis to

identify ALK regulated genes is underway, and the cells are available to ASSET partners.

Figure 3.8. Western blot analysis of inducible wildtype and mutant (RQ, FL) ALK expression cell

clones. The system is not leaky and the expression of ALK is well regulated in all clones. The mutant ALK

proteins are phosphorylated, and high expression of the FL mutant induces the activation of ERK and STAT3

as assayed by phosphospecific-antibodies.

In a first exploratory analysis to identify ALK downstream miRNAs, we performed ALK inhibitor treatment of

NB cell lines using the TAE-684 compound and subsequently established the miRNA expression profile (see

also WP9). Analysis revealed no robust changes of miRNA expression levels after ALK inhibition. Currently,

we are setting up a miRNA expression profiling analysis using the ALK inducible cell lines for further analysis

of the impact of ALK signalling on miRNA expression. In parallel, we will establish additional miRNA profiles

from ALK driven mouse tumours.

Task 3.4. Characterization of the functional interrelation between TrkA and MYCN in regulating NB

cell fate.

Work on this task has not started yet, but UKE has undertaken preparatory work. To support robust

statistical analysis of miRNA variation in primary tumours, miRNA sequencing will be performed in a cohort

of 100 primary neuroblastomas. Sequencing will be carried out in 2013, but some preparation for this part of

task 3.4 has already taken place. The 100 primary neuroblastomas have been selected from the tissue bank

of the German Neuroblastoma Trial in Cologne. The tumours were selected for the completeness of

supporting molecular information. For the respective tumour samples, mRNA expression array data, mRNA

transcriptome sequencing data and, for a subset even exome sequencing data, is available or will be

available within 2013. For miRNA sequencing, tissue samples have been acquired from Cologne, and total

RNA has been isolated using Qiazol. Sequencing will be performed by the Beijing Genome Institute (BGI) on

a fee-for-service basis.

Task 3.5. Analysis of time-resolved EWS-FLI1 dependent miRNA expression profiles in order to study

molecular and phenotypic effects of selected miRNAs in ESFT cell lines.

CCRI has monitored miRNA expression profiles at four time points after doxycycline inducible knockdown of

EWS-FLI1 by stem-loop PCR in the ASSET model Ewing´s sarcoma cell line Asp14. In addition, mRNAs

engaged in the RISC complex in the presence of EWS-FLI1 and upon a 56 hour knockdown of EWS-FLI1

were analysed using the PAR-CLIP method (see also Task 3.2.).

CURIE has carried out an analysis of EWS-FLI1-regulated miRNAs by microarray miRNA profiling with

Ewing’s sarcoma cell line model expressing an inducible shRNA targeting EWS-FLI1. The genome-wide

analysis of miRNAs where EWS-FLI1 was downregulated showed that the expression of 92 miRNAs was

affected. The impact of each miRNA on the phenotype of ET cells is being analyzed by modulation of the

respective miRNA expression. The EWS-FLI-1-regulated miRNAs were identified by microarray miRNA

profiling. Downregulation of EWS-FLI-1 affected the expression of 92 miRNAs. The implication of each

miRNA on the phenotype of ET is being analyzed.

Task 3.6. Modelling to identify commonly influenced miRNAs in ETs and separate miRNA profiles for

NB, MB and ESFT

The data generation for this task has started. Using a high throughput 3’UTR luciferase screen for miRNA

binding sites we were able to establish the miRNA-ALK and miRNA-MYCN regulomes (see Figure 3.5).

These unique validated datasets offer insights into the complex upstream regulation of both crucial

oncogenes in neuroblastoma. This dataset is being made available for modelling the mRNA/miRNA

regulatory networks of the ET genes.

Figure 3.5: Cytoscape view of 3’UTR screen results of MYC, MYCN and ALK. Edge width is representative of the interaction scores between the miRNA and gene.

Significant results

6. ET cell lines with regulatable expression of ET oncogenes have been established and characterised.

They have been validated for signalling and screening studies, although the TrkA inducible NB cells may

not be suitable for long term biological studies due to an unexpected over-induction of the TrkA construct

after prolonged periods of NGF stimulation.

7. EWS-FLI1 regulated miRNA profiles have been identified and are currently analysed and validated.

8. HTS with pre-mir and anti-mir libraries were performed in MB, NB and Ewing sarcoma cells under

condition of oncogene induction switched on or off. Data has been reported to the partners and a

deliverable (D3.1: miRNAs regulating ET cell viability (Month 24)) was reported.

9. We show that the MYCN regulated miR-17-92 directly controls expression and protein levels of DKK3

through binding to seed sequences in the 3'UTR of the gene.

10. We characterised MYCN regulation by Let-7 miRNAs in cells and animal models, identifying that MYCN

protein expression is negatively regulated by the Let7 miRNA, which is repressed by the LIN28B

transcription factor.


None.

Corrective actions. The initial cell line model of TrkA signalling in neuroblastoma may be not suitable for

longer term studies due to the promoter instability discussed above, and will be remade using different

inducible promoter systems. This is not expected to adversely affect deliverables.


Resources were largely used as planned as detailed below:

WP4. ET transcription factor protein networks.

Objectives

This WP characterises TF networks on the functional protein level. Towards this we are mapping the

dynamic protein-protein interactions of TFs altered in ETs, such as MYCN, MYC, EWS-FLI1, p53 and Rb

using quantitative proteomics. These data will enrich mathematical models of GRNs by dynamic and

mechanistic data.


Task 4.1. Dynamic analysis of protein-protein interactions in ET TF complexes.

The MYCN inducible Neuroblastoma cell model (DKFZ, FW), TrkA-inducible neuroblastoma model (UKE)

and EWS-FLI1 inducible Ewings Sarcoma model (CCRI) have been generated. The necessary labelling

conditions for quantitative proteomics have been established (UCPH), and optimisation of immuno-

precipitation and sample preparation conditions are currently underway. SILAC based MS quantitative

proteomics has been performed do detect changes in global protein levels upon MYCN overexpression

(UCD). Mapping MYCN protein interaction partners by proteomics has commenced (UCD) and will be

expanded to generate dynamic quantitative data.

Task 4.2. Validation of results by perturbation studies with siRNA and drugs.

Participant number

Participant short name

Person-months per participant

Personnel & Resources used to date November 1, 2011 to October 31, 2012.

2 CCRI 7 1 Postdoc – 7 PM


6 VTT 11.5 1 Postdoc – 11.5 PM

7 UKE 5 1 PhD student – 3 PM, 1 Technician 2 PM

10 UBERN 6.2 2 PhD students (50%) for 2.4 months and 1 PhD student (50%) for 1.4 months

12 ZEPTO 1.82 1 staff scientist 1.8PM, 1 technician .02PM

8 UGENT 2 1 Postdoc: David Camacho-Trujillo - 2PMs


37.52

The experiments in this task will follow from data obtained in Task 4.1, and hence have not started yet. Pilot

experiments testing drug responses and optimmising siRNA efficacies in ET cell lines are underway in

several groups (UKE, DKFZ, UCD, UBERN, CCRI, UGENT).

Task 4.3. Generation of dynamic protein-DNA and protein-protein interaction maps for modelling.

As part of this global modelling approach the transcriptional network induced downstream of amplified MYCN

in the SH-SY5Y model is being mapped by a novel method called Dynamic Transcriptome Analysis (DTA).

The method was originally developed for yeast21, and adapted for use in mammalian cells by UCD. This

method facilitates the identification of direct transcriptional target genes. Briefly, the method involves

metabolic labelling of nascent mRNA with a uridine analogue that subsequently can be modified in vitro by

biotinylation. Pulldown of the biotinylated mRNA species with streptavidin beads permits the selective

purification of newly synthesised mRNA, and the quantitative identification of these by next-generation RNA

sequencing. We have successfully established DTA, and have obtained data from NB cell lines where

MYCN expression was conditionally induced. In conjunction with this, promoter occupancy by MYCN is

currently being analysed by CHIP-PCR. The data for MYCN protein-protein interactions will be merged with

this dynamic transcriptome and ChIP data to form this complete map of protein-DNA, protein-protein and

subsequent transcriptional network.

All primary and secondary MYCN TF responsive target genes, proteins and miRNAs have been

quantitatively measured (mRNA-seq, SILAC MS & miRNA-seq) and their dynamic response to MYCN

overexpression has been used to generate statistical clustering models. Preliminary data on all MYCN DNA

interacting sites in NB have also been generated (ChIP-seq). These -omic data sets provide dynamic

quantitative gene expression levels for every gene in the core NB cell line SH-SY5Y. This data will form the

basis for the establishment of mathematical models. Preliminary MS analysis of MYCN interacting proteins

has been performed and this will be expanded to generate dynamic quantitative datasets.

The global MYCN transcriptional network has been reconstructed from the mRNA-seq and miRNA-seq data.

Results from RNA-seq have been analysed and extended to include further time points and cell states

(UCD). miRNA-seq has been performed and analysed. Preliminary ChIP-seq experiments have been

optimised, performed and analysed. Data sets are currently being integrated to identify a short list of NB

outcome relevant candidate MYCN targets for subsequent validation. Transcriptomic data from MYCN

amplified cell lines is also being generated for cross comparison with the reconstructed MYCN transcriptional

network. Prior to the generation of mechanistic models time course analysis methodology for -omics data

sets and development of statistical clustering models of the time course data had to be established (UCD).

DKFZ and BONN have focused on the question whether MYCN oncoprotein acts as a general amplifier of

already active genes in neuroblastoma cells. Chromatin-Immunoprecipitation (ChIP) combined with oligo-

microarrays or next-generation sequencing was used. We used well-established NB cell lines as well as the

ASSET model systems and enriched for MYCN/MYC binding as well as for histone modifications for active

(H3K4me3), silent (H3K27me3), and elongated genes (H3K36me3) using ChIP. In neuroblastoma cell lines

derived from tumours with aggressive clinical behaviour, MYCN binding was mainly observed at the

transcriptional start site (TSS) of active and elongated genes, where the proximity to the TSS as well as the

number of MYCN binding events (sequence tags) per gene significantly correlated with mRNA expression

levels. Induction of neuronal differentiation in these cells dramatically reduced MYCN binding events at gene

promoters where MYCN was initially bound close to the TSS. Surprisingly, the genome-wide overall number

of MYCN binding events did not change during neuronal differentiation. This argues in favour of a repressor

complex that abrogates MYCN binding at the TSS in differentiated neuroblastoma cells. In line with a

prominent role of MYCN binding close to the TSS in tumour progression, we found that those genes bound

by MYCN in cell lines were also highly expressed in primary neuroblastoma tumours. Moreover, high

expression of MYCN-bound genes was significantly correlated with unfavourable biology of neuroblastomas.

In subsequent experiments we will explore the identity and the dynamics of putative repressor complexes at

the TSS of MYCN target genes during neuronal differentiation. Furthermore, we will adapt the ChIP-seq

protocol for primary tumour samples.

CCRI has investigated promoter occupancy by identified TF complex components in ESFT by ChIP-PCR

and reporter gene assays. By TF recognition motif analyses of EWS-FLI1 activated promoters and by ChIP-

seq in collaboration with Sven Bilke and Paul Meltzer (NCI, NIH) we identified a EWS-FLI1/E2F

transcriptional module in Ewing sarcoma. Transcriptional activation through this module was validated for 10

arbitrarily chosen genes, including CDK2, E2F3, SKP2, GEMIN4 and ATAD2. By ChIP-PCR, binding

elements were delineated and functional interdependencies were studied using ChIP-PCR on ETS-motif

mutated transfected constructs. We could demonstrate that a functional ETS binding site is required to allow

for E2F3 binding to the promoters of GEMIN4 and E2F3. Upon knockdown of EWS-FLI1, activating E2F3

(and the pRb1) was found rapidly exchanged by repressive E2F4 (and the corresponding pocket protein

p130) suggesting that binding of EWS-FLI1 leads to a E2F4/p130 to E2F3/pRb1 exchange. We were able to

extend our observations to another ETS-driven cancer, TMPRSS2-ERG expressing prostate cancer.

Significant results

Comprehensive data sets of the response to MYCN overexpression at multiple molecular levels in SH-SY5Y

have been generated. These data sets can be summarised as follows:

of approximately 15,000 genes expressed the cell line over 600 are differentially expressed upon

MYCN overexpression.

of 1,073 miRNAs expressed 30 were found to be differentially expressed upon MYCN

overexpression.

of approximately 4,000 detectable proteins 278 were differentially expressed upon MYCN

overexpression.

A method to specifically identify newly transcribed genes in yeast (Dynamic Transcriptome Analysis) has

been successfully established in mammalian cells and used to map MYCN target genes.

The MYCN TF transcriptional network has been reconstructed and can now be used for identification of NB

outcome relevant genes which will be functionally validated. Such genes will be identified by cross

comparison of the MYCN network with non-amplified vs. amplified MYCN NB data sets, and NB early stage

differentiation data.

Genome wide MYCN binding sites have been mapped, and results suggest that a repressor complex

abrogates binding of MYCN near transcriptional start sites during differentiation of NB cells.

TF complex components in ESFT were characterised by ChIP-PCR and reporter gene assays identifying a

EWS-FLI1/E2F transcriptional module in Ewing sarcoma.


None.


Resources were used as planned as detailed below:

WP5. Mapping cytosolic ET signalling protein networks by quantitative

proteomics.

Objectives

Protein functions, hence, network connectivities are critically regulated by phosphorylation. Subsets of NBs

feature aberrant receptor tyrosine kinase (RTK) signalling by TrkA and ALK. To identify the regulatory

circuits downstream of ALK and TrkA we are using quantitative proteomics to map (i) tyrosine

phosphorylation sites on TrkA and ALK, (ii) proteins which bind them and (iii) global phosphoproteome

changes downstream of these RTKs. ALK expression is a better prognostic predictor than ALK mutational

status in NB. We are using NB cell models with regulatable ALK or TrkA expression, and to further

parameterise our kinetic models we are quantitating protein network components and their phosphorylation

states using Zeptosens’ reverse protein arrays for highly parallel relative quantitation and MS-based

methods for absolute protein quantitation.

Participant number


Personnel & Resources used: November 1, 2011 to October 31, 2012

1 NUID UCD 13.45 2 Postdocs: David Duffy – 12PM, Melinda Halasz 1.45PM

7 UKE 1 1 PhD student – 1PM

8 UGENT 2 1 Postdoc: David Camacho-Trujillo – 2PM


16.45


Task 5.1. Global profiling of protein expression of the core ET cell lines by quantitative mass

spectrometry.

UCPH has generated proteomics data for two biological replicates of a triple SILAC experiment using 0, 24

and 48 hrs of NGF treatment as time points to identify global protein expression changes associated with

differentiation of TrkA activated SY5Y cells. Data analysis using the MaxQuant software suite has resulted in

a list of significantly regulated proteins. These results were obtained before the promoter instability issues

with the SY5Y-TR-TrkA cells were discovered. Thus, the validation of the results will be done in the new

TrkA cell lines, which are currently being established as well as in the original cell line.

In addition, UCPH has established SILAC conditions for the medulloblastoma cell line, UW228 Myc-ER and

the Ewing Sarcoma cell line, Asp14 shEWS-FLI1 (A673/TR/shEF) from UBERN and CCRI, respectively. For

the UW228 Myc-ER cell line, a triple SILAC experiment using 0, 48 and 72 hours of MYC induction has been

performed. For the Asp14 shEWS-FLI1 as triple SILAC experiment has been performed using 0, 18 and 48

hours of doxycycline-induced EWS-FLI1 downregulation. In addition, a pulsed SILAC experiment was

performed to analyse changes in newly synthesized protein upon 48 hours of doxycycline induction. All

experiments were performed in two biological replicates. Western analysis was performed to ensure proper

induction prior to sample preparation for MS.

Whole cell lysates from the three SILAC conditions have been mixed, in-gel digested with trypsin and

analyzed by online nanoscale-LC-MS/MS using a hybrid orbitrap mass spectrometer with high-resolution

tandem mass spectrometric peptide sequencing. For the Asp14 EWS-FLI1 and UW228 Myc-ER

experiments data analysis has been performed using the MaxQuant software package. Lists of significantly

regulated proteins have been handed over to CCRI and UBERN, who will be responsible for subsequent

validation of hits from the screens. The incorporated ET cell lines are also now available for

phospoproteomics on kinase inhibitor synergy screen studies (WP8, task 8.5).

Task 5.2. Quantifying the phosphoproteome of different NB cell lines with induced TrkA and ALK

expression/activation levels.

Phosphoproteomics data for the SY5Y-TR-TrkA cell line in terms of early time-point NGF stimulation have

been generated. Western blot analysis of NGF time course stimulation (within 0-4 hours) of TrkA induced

and activated cells showed biphasic signalling by phosphorylation of downstream adaptor proteins.

Therefore, we chose to setup a double triple SILAC experiment to cover these observed phases of TrkA

signalling. We chose to work with the following time points (min): 0, 10, 45, 120 and 120+cycloheximide.The

cycloheximide condition was included to distinguish the signalling contribution from newly synthesized TrkA

receptor and early gene induction. The phosphoproteomics setup and workflow is illustrated in Figure 5.1.

Two biological replicates have been performed. Correlation of data for the replicates ranges from 0.74-0.84.

See Figure 5.2.

Figure 5.1: Phosphoproteomics workflow

Figure 5.2: Correlation between replicates

Processing of the MS data by the MaxQuant software and evaluation of data quality is on-going. Data is

being re-analyzed including proteome data within the short term NGF stimulation up to 2 hours to take

changes in protein abundance into account. Preliminary phosphoproteomics data only is listed in Figure 5.3.

Figure 5.3: Preliminary evaluation of phosphoproteomics data.

The cell model, initially planned for the ALK studies (UKE), SKNAS-TR-ALK, was created and is functional

(details in WP3). Gene expression profiles induced upon ALK activation in SKNAS-TR-ALK were compared

to expression profiles correlating with ALK inhibition in neuroblastoma cell lines (induced by ALK inhibitors or

siRNA) that correlated with ALK expression in primary neuroblastomas or ALK-driven murine

neuroblastomas (together with UGENT). From these preliminary analyses, SKNAS-TR-TrkA appeared to be

a good model. However, some target genes, in particular ETV5, induced by ALK in other systems were

strongly expressed in SKNAS-TR-ALK cells, even in the absence of ALK activation, and did not significantly

increase upon ALK activation. This result was interpreted such that the absence of ALK in the parental SK-

N-AS cell line may have been solved by the tumour cells by activating downstream signalling pathways via

other aberrations converging on ALK downstream signalling and yielding a tumour advantage. For this

reason, we decided (see deviations to Annex I section below) to generate a second cell model from another

ALK-negative neuroblastoma cell line, CHP134, and compare which of the models, SKNAS-TR-TrkA or

CHP134-TR-TrkA, more closely resembles ALK-modulated gene expression in other models, and be best

suited for phospho-proteomics analyses. Since CHP134-ALK evaluation is not yet complete, cell lines have

not yet been distributed to ASSET partners for ALK analysis. Either SKNAS-TR-TrkA or CHP134-TR-TrK will

be distributed to ASSET partners in 2013.

Task 5.3. Identification of phospho-tyrosine dependent protein-protein interactions for ALK and TrkA

using antibody-based affinity-purification coupled to high-resolution MS

These studies have not been initiated yet.

Task 5.4. Absolute quantification of selected phosphorylation sites and proteins using AQUA.

This part is dependent on the output from 5.2, and will start after the analysis of the corresponding data.

Task 5.5. Routine MS-based measurements of selected proteins and PTMs.

This work is dependent on output from 5.1 and 5.2 and will start after the corresponding analysis is

completed.

Task 5.6. Investigation of the dynamic behaviour of protein signalling networks using reverse protein

microarrays

Protocols to prepare protein lysates for Zeptosens Phosphorylation Array Analysis were optimized for the tet-

inducible SY5Y-TR-TrkA cell line. Johannes Schulte and Sven Lindner (UKE) visited partner ZEPTO (at

Bayer Technology Services) to discuss operations for the Zeptosens array analyses, and these visits

continued throughout the cooperation. SY5Y-TR-TrkA cells were prepared for analyses by UKE. SY5Y-TR-

TrkA and SY5Y-TR-TrkB cells were seeded onto 10 cm cell culture dishes. After 24 hours, medium was

changed to medium containing tetracycline, and 21 hours later medium was changed again to medium

containing 1% FCS with tetracycline. This medium was left on for 3 hours to reduce background signalling

noise and keep cellular stress as low as possible. After this 3 hours of cellular model normalisation, new

medium containing either NGF (to SY5Y-TR-TrkA cells) or BDNF (to SY5Y-TR-TrkB cells) were added to

stimulate the respective TrkA/B receptor and initiate downstream signalling.

Figure 5.4: Validation of PLCγ as promising target of Trk signalling. (A) Read-out of reverse-phase protein array

(Zeptosens) showing the different regulation of phosphorylated PLCγ. Whereas phosphorylated PLCγ was strongly up-

regulated in SY5Y-TR-TrkA cells upon NGF treatment (green curve), there was only a modest increase in

phosphorylated PLCγ in SY5Y-TR-TrkB upon BDNF treatment (blue curve). (B) Western blotting analysis confirming

the read-out from the reverse-phase protein array (A3/B3 = 2 replicates after 2 min NGF treatment; D3/E3 = 2

replicates after 2 min BDNF treatment).

After 2 – 12 hours (detailed timepoints see Figure 5.4) the media was aspirated and the cells were washed

with ice cold PBS. Afterwards the cells were collected in CLB1 lysis buffer (Zeptosens). Before delivering the

cell lysates to ZEPTO for reverse-phase array analysis, concentrations were determined using Coomassie

Plus Protein Assay (Pierce). Reliable total protein concentration determination in samples is crucial for the

RPA process, as samples are normalized to the same total protein concentration to ensure comparability

among them. Protein determinations were done at UKE as well as at ZEPTO showed an excellent

correlation. Single droplets of about 400 pL sample were arrayed with a NanoPlotter NP2.1E (GeSim) on

freshly prepared Zeptosens Hydrophobic Chips (Zeptosens/BTS) under clean room conditions. After spotting

the arrays were blocked and incubated in a two-step assay with primary antibody and fluorescently labelled

secondary antibody. Arrays were imaged in a fluorescent reader and processed with the image analysis

software.

Among 122 analyzed proteins/phosphorylated proteins 16 showed regulation. However the significance of

these results has to be analyzed in detail. Among the regulated proteins are pAkt, CDK1, CyclinD1, pGSK3β,

pMEK1/2, pStat3, pPKCα, pPLCγ, pTyk2, S6 ribosomal protein and SHP2. The overlap of similar regulated

proteins between NGF (TrkA) and BDNF (TrkB) signalling was seen in 15 out of the 16 regulated proteins.

So far, the most promising target was Phospholipase C gamma (PLCγ) which was upregulated only in

SY5Y-TR-TrkA after NGF stimulation and not in SY5Y-TR-TrkB after BDNF stimulation. This hints to a role in

differentiation and cell death of PLCγ. In further experiments a knock down of PLCγ in SY5Y-TR-TrkA

stimulated with NGF could shed light on the role of PLCγ.

Figure 5.5: Validation of TrkA or TrkB induction and stimulation with NGF or BDNF in SY5Y-TR-TrkA and AY5Y-

TR-TrkB cells used for reverse-phase protein array analyses.

To validate the ZEPTO results on PLCγ phosphorylation in SY5Y-TR-TrkA and SY5Y-TR-TrkB cells

independently, UKE conducted Western blotting analysis on the protein lysates (Figure 5.5). Cell lysates

were prepared under highly denaturing conditions using the ZEPTO proprietary buffer CLB1.

Significant results

7. Data for quantitative profiling of protein phosphorylation in SY5Y-TR-TrkA cells is being re-analyzed.

Proteome data within the short term NGF stimulation up to 2 hours has been included to take changes in

protein abundance into account.

8. Data for quantitative profiling of protein expression in SY5Y-TR-TrkA cells (long term NGF stimulation, 0,

24, 48 hours) are being evaluated together with the generated phosphoproteomics data to create list of

hits for validation.

9. Quantitative profiling of protein expression in UW228 Myc-ER cell line upon induction of MYC for 48 and

72 hours has been performed.

10. Quantitative profiling of protein expression in the Asp14 shEWS-FLI1 cell line upon downregulation of

EWS-FLI1 for 18 and 48 hours has been performed.

11. Quantitative profiling of newly synthesized protein (pulsed SILAC) upon EWS-FLI1 downregulation (48

hours of doxycycline treatment) in the Asp14 shEWS-FLI1 cell line has been performed. MS and data

analysis on Asp14 shEWS-FLI1 work has been completed.

12. Protein array analysis has pinpointed PLCγ as potential critical TrkA downstream transducer.


The discovery of promoter instability issues with the SY5Y-TR-TrkA cell line has delayed the progress

towards objectives. Changes in experimental setup and how to analyse data was required to meet the issue

in the best possible way.

At a meeting in Ghent in October 2011, it was decided that the initial cell line for the ALK studies, SK-N-AS-

TR-ALK should be compared with 3 additional cell lines due to specific genetic background issues to decide

which cell line is best suitable particularly for further phospho-proteomics analyses. We are awaiting a final

conclusion on which cell line to use before distributing the final choice among partners to initiate the ALK-

related work in 2013. As a result the project is not progressing to according to schedule and a delay in

accomplishment of deliverables and milestone is to be expected especially for the ALK-related work.



WP6. Computational-kinetic models of critical Myc-dependent regulatory

networks that drive ETs.

Objectives

To make critical network parts accessible to mechanistic simulations and predictions, we are developing

predictive ODE models from the global network models developed in SP1 based on perturbation analyses

Participant number



1 NUID UCD 1.19 1 Postdoc: Melinda Halasz 1.19PM

7 UKE 7 1 PhD student 3PM, 1 Technician 4PM

12 ZEPTO 3.3 1 staff scientist 2.3PM, 1 technician 1PM

13 UCPH 5 Senior Scientist: Kristina Bennet Emdal – 5PMs


16.49

and populated by data on protein-protein and protein-DNA interactions obtained in SP2. Stochastic features

will be added using advanced statistical approaches and ensemble modelling which considers the

distribution of output functions of an “ensemble” of models rather than single deterministic models. The aim

is to develop experimentally validated mathematical models of the core regulatory networks, deregulation of

which drives Myc-dependent ETs. This WP is integrating input from global network models, biochemical

experimentation from previous WPs and apply dynamic kinetic modelling methods in order to elaborate

models of network submodules providing mechanistic insights. These results are being fed back into the

experimental WPs as well as into the validation WPs in order to provide concrete testable predictions of a

quantitative rather than qualitative nature.


Task 6.1. Modelling of the effects of TrkA on MYCN and MYC expression and protein stability.

The literature suggested that MYCN and MYC are regulated by TrkA22,23,24, but our own experiments in

SY5Y cells showed no significant changes in the protein levels of MYCN or MYC in response to NGF (Fig.

6.1). These experimental data rule out the necessity to construct a mathematical model of MYC regulation by

TrkA in SY5Y cell line, which is the common cell line for all experimental labs in the consortium. However,

our experimental observations showed that retinoic acid (RA) significantly regulated MYCN, MYC and TrkB

on the mRNA and protein level (Fig. 6.2 & 6.3). Therefore, we shifted our modelling studies to account for

MYC regulation by RA on the gene regulatory level. First, we established a reaction scheme of this model

informed by the literature and our experiments on the mRNA level (Fig. 6.4). Initial results on the protein

level suggest that the model needs further refinement to account for cell-type specific differences in different

neuroblastoma cell lines.

Figure 6.1. Responses of the SY5Y-TR-Trka cell line to different dosages of NGF treatment. Data were

obtained by Western blotting using the antibodies and time-points indicated.

Figure 6.2. TrkA, TrkB and MYC mRNA levels in response to retinoic acid treatment in the SY5Y-

MYCN cell line. Data were obtained by sequencing 24h after the RA treatment in two different MYCN

conditions: “RA –” indicates DMSO control conditions, “RA +” indicates RA treatment, “DOX –” indicates that

MYCN is not induced, “DOX +” indicates that MYCN is induced.

Figure 6.3. Time-courses of MYCN protein levels in response to RA treatment in the indicated cell lines.

Data were obtained by Western blotting.

Figure 6.4. Simplified scheme of the RA-Myc-Trk model. The model only includes interactions that were

consistent with our own data.

Additionally, and as an alternative to the TrkA model, we developed a model of how SY5Y cells respond to

growth factors and stress in general25. The literature suggests that both MYC mRNA and protein levels are

regulated by MAPK and AKT/GSK3 signalling in response to extracellular cues such as insulin like growth

factor or RA26,27,28,29. Hereby, the relative activation of different pathways in relation to each other seems

critical, as activation of ERK and AKT stabilised MYC protein (by phosphorylating MYC at S62 and inhibiting

GSK3 mediated phosphorylation of T58, respectively), whereas JNK activity targeted MYC for proteasomal

degradation27,30. Further, MYC can exhibit both proliferative and pro-apoptotic functions depending on its

phosphorylation status as regulated by ERK and JNK, respectively31. Thus, it is critical to understand how

these pathways process extracellular cues in order to decipher how MYC proteins are regulated. To that

end, we constructed a mathematical model of the three major MAPK signalling cascades (ERK, p38, JNK)

and PI3K/AKT signalling in response to growth factors and stress25 (Fig. 6.5). Our model demonstrates that

the one signalling pathway cannot be considered in isolation from the others, and crosstalk and feedback

structures are the determining elements of cell fate decisions. For example, JNK can switch from proliferative

to pro-apoptotic signalling due to a positive feedback loop, which is itself be regulated by crosstalk from ERK

and AKT25. Importantly, the key regulatory elements of the model (the JNK positive feedback loop and AKT–

JNK crosstalk) have been confirmed experimentally using the SY5Y cell line.

Figure 6.5. Simplified scheme of the MAPK and AKT interaction model.

Task 6.2. Modelling of the MYCN and MYC interaction network.

A preliminary model of direct MYCN and MYC interactions on their expression levels has been formulated.

Analysis targeted at the characterisation of possible behaviours in this system suggest that a positive

feedback loop consisting of MYCN and MYC mutual repression can exhibit bistable behaviour in the form of

a toggle switch. Over-expression of either MYC or MYCN can lock the system into a MYC or MYCN active

state, respectively. However, depending on the strength of MYCN and MYC auto-repressive feedback, the

parameter region for which bistable behaviour occurs is diminished. Whether the MYCN/MYC switch has

biological relevance is currently unknown. The model currently awaits parameterisation and validation.

A model of the MYCN regulatory network relating to apoptosis and chemo-resistance has been developed

and partially validated. This model is based on established links in the literature between HMGA1, a

molecule involved in tumour chemo-resistance, the DNA-damage response system surrounding p53 and

MYCN (Figure 6.6). The model accounts for several MYCN dependent regulatory loops, predicts an optimal

range of both MYCN and HMGA1 levels for apoptosis induction and offers hints for optimal therapeutic

intervention. Initial experiments using the SY5Y-MYCN validated the predicted MYCN and HMGA1 effects

on apoptosis. Experiments for further model validation and calibration are currently under way.

Figure 6.6. Illustration of the key elements in the MYCN regulatory model concerning apoptosis and

chemo-resistance.

In addition to the SY5Y-TR-TrkA cell model system described above, UKE established a second model

system in order to analyse the effects of TrkA expression on a background of MYCN amplification. For this

we used the MYCN-amplified cell line, IMR5, and transfected it with the tet-inducible TrkA expression vector,

generating IMR5-TR-TrkA. Interestingly, TrkA activation did not induce differentiation, apoptosis or

proliferation arrest in IMR5-TR-TrkA cells, as compared with SY5Y-TR-TrkA cells. As similar immediate early

genes are activated in both models, this suggests a late block of TrkA-mediated biological effects in MYCN-

amplified cells.

Figure 6.6 Characterisation and validation of the IMR5-TR-TrkA cell model. Induction of TrkA and

stimulation of TrkA with NGF do not have an effect on cell viability (A, MTT assay), cell death (B) or

proliferation (C, BrdU incorporation). (D) Western blot shows the TrkA expression before and after treatment

of cells with tetracycline.

Task 6.3. Fragility analysis.

This modelling step is reliant upon data from Task 6.1 and 6.2 and will therefore be conducted once these

models are fully validated and parameterised.

Significant results

The SY5Y-TR-Trka cell line shows no significant changes in MYC and MYCN levels within the early phase of

TrkA signalling.

As alternative to the TrkA model, an integrated, dynamic model of ERK, p38, JNK and AKT signalling in

response to growth factors and stress has been developed and validated experimentally using the SY5Y cell

line.

An MYCN interaction model relating to apoptosis and chemo-resistance in neuroblastoma has been

developed, partially validated using the SY5Y-MYCN cell line and is currently being refined and

parameterised. The model predicts that inhibition of the HMGA1-HIPK2 interaction facilitates the induction of

apoptosis in response to DNA damaging agents, which needs experimental validation.

The establishment and characterisation of a TrkA inducible cell line on a MYCN amplified background

suggests a late block of TrkA-mediated biological effects in MYCN-amplified cells.


Based on our experiments (see Task 6.1) the deliverable D6.1 has been modified. Instead of the NGF-TrkA-

Myc model, we developed and validated a model how SY5Y cells respond to growth factors and stress in

general. This model constitutes the modified deliverable D6.1a and is published25. Additionally, we are

currently developing a model of RA induced regulation of MYC proteins and neurotrophin receptors. We

expect to deliver this model in month 60, as we also want to consider results from the analysis of RA effects

on the transcriptome. Task 6.3. will now concern these models. Task 6.2 is not affected.



Participant number



1 NUID UCD 18.36 3 Postdocs: David Croucher 6PM. Dirk Fey 12 PM, Melinda Halasz .36PM

7 UKE 1 1 PhD student 1PM

9 UCL 5.54 1 Postdoc: Vasileios Stathopoulos –

5.54PM


23.9

WP7. Network fragility analysis of apoptosis and proliferation decisions

Objectives

In this WP we are developing mathematical models of the network that determines the balance between

decisions for proliferation or apoptosis in ETs, centring around the CDK-Rb-E2F-Skp2 and p53-

Mdm2/MDMX modules. Our initial focus is on MYCN-driven NB, where in collaboration with WP6 we aim to

rationalise on a mechanistic level how the different degrees of MYCN overexpression or amplification

observed in NBs give rise to the various clinically recognised stages of the tumour. NBs range from

moderately hyperproliferative through hyperproliferative yet apoptosis-sensitive (both of which respond to

current therapies) to therapy-resistant hyperproliferative and apoptosis-insensitive stages. This model tries to

dissect the regulatory effects of molecular aberrations in ESFT and MB. New molecular players and

regulatory links identified in our functional genomic and proteomic screens are incorporated into the models

as required. We will systematically characterise in silico fragile nodes of the network that can be targeted to

(i) sensitise tumour cells toward apoptosis or/and (ii) inhibit unchecked cell cycle progression. The results of

this workpackage and WP6 will inform the biological validation SP4 using cell lines and mouse models.


Work in this WP is in large parts dependent on results from other WPs and therefore has only started in a

limited fashion as warranted by existing results.

Task 7.1. Modelling of the CDK-Rb-E2F-Skp2 and p53-Mdm2/MDMX modules in NB.

UCD has developed an initial, partially validated model relating HMGA1, a molecule involved in tumour

chemo-resistance, to the p53-Mdm2 module and surrounding structures (also see WP6 Task 6.2 Modelling

the MYCN and MYC interaction network).

DKFZ further characterize both modules, namely p53-Mdm2/MDMX and CDK-Rb-E2F-Skp2 in

neuroblastoma cells we have generated model systems that allow conditional expression of p16 and

p14ARF at the DKFZ. We have extensively studied the consequences of p14ARF reexpression on the p53-

MDM2 axis in these cells. Furthermore, we analysed the regulation of p14ARF in neuroblastoma cells and

described epigenetic mechanisms that inhibit expression of p14 ARF expression.

CCRI has performed genetic perturbation experiments of players identified in CDK-Rb-E2F-Skp2 and p53-

MDM2/MDMX networks in ESFT cell lines. In WP4 we have identified an ETS/E2F transcriptional module

operating in EWS-FLI1 expressing Ewing sarcoma and TMPRSS2-ERG expressing prostate cancer. We

found E2Fs, CDK2 and SKP2 among genes activated by this module. Here, we have performed CDK2

knockdown and pharmacological inhibition experiments in the Asp14 model cell line and obtained evidence

that CDK2 inhibition is sufficient to drive the transcription factor FOXO1 into the nucleus allowing for re-

expression of about 10% of EWS-FLI1 repressed genes. We also performed E2F3 and E2F1 knockdown

experiments and monitored effects on genome-wide gene expression. Surprisingly, the transcriptional

consequences were relatively modest supporting our results described in WP4 that it is EWS-FLI1 which is

primarily responsible for the regulation of E2F target genes in Ewing sarcoma.

Using siRNAs to p53, MDM2, and MDMX, we studied the consequences of perturbation at each of these

nodes on EWS-FLI1 dependent p53 regulation. By a comparative genomics approach we identified the NAD-

dependent deacetylase SIRT1, which we found highly expressed in Ewing sarcoma cell lines and in about

30% of primary tumours (but more than 60% of metastases), to be involved in the inactivation of p53

downstream of EWS-FLI1. Upon knockdown of EWS-FLI1 we previously described activation of NOTCH

signalling and eventually of HEY1 transcription. We now identified SIRT1 as a HEY1 target. Knockdown or

pharmacological inhibition of SIRT1 leads to p53 acetylation and activation.

Task 7.2. Connection with signal transduction.

During this year the CURIE Systems Biology team (AZ) mainly contributed to methodological development of

software tools that will help to predict probabilities of cell fates in response to various perturbations, including

combinatorial perturbations. The tools developed (MaBOSS, OCSANA and BiNoM) are described in WP10

in more details. In particular, a method for confronting quantitative data with influence networks (PIQuant for

Path Influence Quantification) was developed, implemented and applied to the expression data for EWS/FLI-

1 inducible system. This method was applied to the influence network constructed for WP1; it confirmed the

expected effects of EWS-FLI1 on cellular phenotypes. It also allowed to extract enriched sub-networks for

further finer mathematical analysis.

Task 7.3. Adaptation of the model to ESFT and MB.

This work has not started yet.

Task 7.4. Predictive fragility analysis of single and combined nodes.

CURIE has developed a method for analysing quantitative data with influence networks (PIQuant for Path

Influence Quantification). This method was implemented and applied to the influence network constructed for

WP1 using the expression data for EWS/FLI-1 inducible system. The results confirmed the expected effects

of EWS-FLI1 on cellular phenotypes. It also allowed us to extract enriched sub-networks for further finer

mathematical analysis.

Significant results

A method for analysing quantitative data with influence networks (PIQuant for Path Influence Quantification)

was developed.

A model relating HMGA1, a molecule involved in tumour chemo-resistance, to the p53-Mdm2 module was

developed and partially validated experimentally.

We performed genetic perturbation experiments of players identified in CDK-Rb-E2F-Skp2 and p53-

MDM2/MDMX networks in ESFT cell lines.


None.


Participant number



2 CCRI 2 1 Postdoc – 2PMs

5 DFKZ 63.5 15 PMs Postdoc, 43.5 PMs PhD and 10 PMs Technician.

WP8. Probing network fragilities with kinase inhibitors

Objectives

Kinases are crucial in signal transduction. Ten selected kinase inhibitors that were shown to be safe in adult

humans, and that cumulatively cover the majority of the druggable kinome are being characterised by

combinatorial perturbation experiments in different ET cell lines. They include approved drugs: nilotinib,

dasatinib, erlotinib, lapatinib, sunitinib, sorafenib; and drugs in clinical development: bosutinib, bafetinib,

danusertib, midostaurin. We are using this panel (i) to identify synergistic drug combinations; and (ii) to

perform proteomic analyses to identify direct and distal protein targets of the most promising drug

combinations. The resulting datasets are being used to (i) correlate targets and biological effects using

transcriptomics, phospho- and chemical proteomic approaches; (ii) identify vulnerable nodes via mechanistic

modelling in order to predict effective inhibitors and inhibitor combination. In order to assure covering the

majority of attractive and druggable targets in selected ET cell lines, we selected a variety of non-kinase

inhibitors (HDAC inhibitors, IGFR inhibitors, Hsp90 inhibitors, etc.). These compounds were added to the

initial panel of kinase inhibitors and all further experiments were done with the entire panel (around 30 drugs

in total). Clinical applicability remained to be the main criterion for the selection.


Task 8.1. Establishing the screening cell line panel and screening conditions.

Screening cell line panel was supposed to be comprised of 4 neuroblastoma conditions (SH-SY5Y inducible

for TrkA or MYCN), 2 Ewing’s sarcoma family of tumours conditions (A673-derived cell line ASP14 that

allows for a doxycycline-inducible knockdown of EWS-FLI1) and 2 medulloblastoma conditions (UW-228 with

MYC on/off). The MYCN inducible SH-SY5Y was still unavailable, so we preformed the experiments with

TrkA on/off system only. Screening conditions for the manual and robotics-assisted combinatorial drug

screening platform were optimized (cell culture, screen conduction and data collection), which led to some

conclusions: (i) SH-SY5Y system is very robust and the results from on and off conditions can be easily

normalized and compared; (ii) knockdown of EWS-FL1 in ASP14 cell line arrests proliferation making it

difficult to use our screening conditions t for both non-induced and induced state. Therefore, we will use the

EWS-FLI1 off condition to confirm that effectiveness of identified inhibitors and inhibitor combinations is

dependent on EWS-FLI1 expression and thus specific for Ewing's sarcoma; (iii) the MYC induction in UW-

228 cell line was not working properly and hampered the initial design. However, the cells are expressing

endogenous MYC so that we would still provide convincing synergy screen results.

Task 8.2. Binary combinatorial drug screen.

Firstly, we determined the individual drug effects on each cell line for all the compounds in proliferation

assays using Cell Titer Glo (Promega). Single dose response curves were determined in a semi-automatic


65.5

way and then confirmed in manual validation experiments. The concentrations where all drugs are tested

range from 0.1nM to 20µM. Based on the respective IC50 values we divided the drugs from the panel in 3

groups: (i) potent (mid-nanomolar range); (ii) measurable / not potent and (iii) compounds with no activity

(these are excluded for the list that will serve for the binary combinatorial screens). (Overview in Figure 8.1)

Figure 8.1. Examples of the IC50 based groups of drugs in the panel. (a) potent; (b) measurable – not

very potent; (c) no activity

Starting with the most potent compound for each cell line, we screened the remaining panel of compounds

for synergistic antiproliferative effects through pair wise combinations. The InWeb in-house tool as well as

the Cytoscape visualization tool were used in collaboration with WP10 to depict the disease protein network

and the compounds targeting the proteins involved in each one of the cancer types of interest. The results

show that although the 20 selected compounds target directly only 32 disease related proteins, based on this

network approach they target indirectly a large number of other disease proteins (specifically 87 disease

proteins are indirectly targeted).

In order to assure covering the majority of attractive and druggable targets in selected ET cell lines, CEMM

selected a variety of non-kinase inhibitors (HDAC inhibitors, IGF1R inhibitors, Hsp90 inhibitors, etc.) and

added these compounds to the initial panel of kinase inhibitors. Clinical applicability remained to be the main

criterion for the selection. All further experiments were done with the entire panel (around 30 drugs in total).

Dose response curves for the enlarged panel of drugs were successfully determined in a semi-automatic

way and validated by hand in a cell line panel comprising 2 neuroblastoma conditions (SH-SY5Y inducible

for TrkA and upon activation with NGF), 1 Ewing Sarcoma condition (ASP14 without knockdown of the

oncogenic EWS-FLI1) and 1 medulloblastoma condition (UW228 without MYC induction). Potency of the

drugs was assayed and classified as described above.

Task 8.3. Synthesis and validation of coupleable drug analogues.

Based on kinase co-crystal and structure-activity relationship data from the literature we designed

coupleable analogues of the drugs that show the highest inhibitory potency in at least one cell line. Four

drugs (Nilotinib, Dasatinib, Bosutinib, Bafetinib) were already available in CEMM as coupleable analogues

and fully validated against a large number of kinase and non-kinase targets. In addition to these, we

designed coupleable analogues based on kinase co-crystal and structure-activity relationship data for all the

other kinase inhibitors from the initial panel (Erlotinib, Lapatinib, Sunitinib, Sorafenib, Danusertib,

Midostaurin) and some of the drugs from the additional compound list. Compound synthesis was either

performed in-house or outsourced to commercial providers. The drug analogues were tested in comparison

with the parent compounds against their cognate targets as well as in pilot drug pull-down experiments.

Since Panobinostat turned out to be the most potent single drug in all investigated ET cell line systems, it is

very promising that the pilot drug pull-down experiments were carried out successfully. Other compounds

from the supplementary set suitable for immobilization are Crizotinib, Tozasertib, Everolimus, Imatinib and

17-AAG (or more precisely, Geldanamycin – its structural analogue – that is converted to 17-AAG during the

coupling procedure).

Panobinostat (Figure 8.1, panel a), left) is the HDAC inhibitor that consistently showed the lowest IC50 value

in all three cell lines. We managed to couple it, and the pilot drug pull-down experiments are done in order to

validate the results (Figure 8.2). For the other promising candidates from the drug panel we will either

perform in-house compound synthesis or outsource it to commercial providers.

Figure 8.2: Coupling of Panobinostat.

Task 8.4. Generation of drug target profiles.

Using the validated immobilised kinase inhibitor analogues we are performing MS-supported drug affinity

chromatography experiments. We are focussing on those drugs that showed synergistic effects in the

combinatorial screen (task 8.2). Control experiments for target deconvolution will use unrelated drugs (e.g.

ampicillin), other kinase inhibitors from our drug panel which have no significant effect on the cell line of

interest, or competition with high concentrations of the parent drug.

Drugs from the above described group (i) and (ii) were screened for synergistic antiproliferative effects

through pair wise combinations using a robotics-assisted platform and validated by three-dimensional dose

response matrices. Mathematical model (Bliss independence) was used to predict the additive effect of the

two compounds and therefore help elucidate the most promising synergistic drug interactions. Selected hits

were investigated using 6x6 dose response matrices centred on the IC50 of the each drug. The resulting

three-dimensional response surfaces contain valuable information that can be used to validate vulnerabilities

predicted by the model.

CEMM and CNIO are collaborating to explore the application of statistical methods to identify genomic

features that correlate with the response to drug in different cell lines. This area is still in a very early phase

due to the complexity of the algorithms to be applied and the availability of the data. These activities are

expected to increase in 2013 because the drug response curves for the three tumour type cell lines (Ewing’s

sarcoma, medulloblastoma and neuroblastoma) have been generated by CEMM and are now available to

the consortium.

Task 8.5. Characterisation of drug-dependent phosphoproteome signatures.

In parallel to the described chemical proteomics approach (task 8.4), we are investigating the kinase

inhibitor-dependent global phosphoproteome changes for the selected drug pairs using the

phosphoproteomics MS platform. The drug profile and phosphoproteome signatures will be further used in

SP4 to control validation experiments and in SP5 to generate dynamic non-linear mathematical models.

SILAC conditions were established for the ET cell line panel representing NB, MB and ESFT with inducible

TrkA, MYC expression and downregulation of EWS-FLI1, respectively. The incorporated NB cell line

including SILAC growth medium and supplements were shipped to CEMM in order for them to test optimal

experimental conditions for the incorporated NB cell line e.g. concentrations of the selected drug synergy

combinations and incubation time. We are awaiting these data in order to continue the global

phosphoproteomics studies.

Significant results

4. An ET cell line panel representing NB, MB and ESFT with inducible TrkA, MYC and repressible

EWS-FLI1, respectively, was screened with the CEMM panel of inhibitors under induced and

uninduced conditions, and single dose response curves were determined.

5. Panobinostat , the most potent drug emerging from this screen was modified for immobilisation on a

solid matrix and successfully used for drug pull-down experiments.

6. A disease protein network based on compounds and their mapped target proteins in each NB, MB

and ESFT has been developed demonstrating that compounds indirectly affect larger target

networks.


None.



Participant number



3 CEMM 12 1 PhD Student: Branka Radic – 12PMs


12

WP9. Validation of mathematical models in biological models and clinical samples

Progress towards Objectives

The focus of this WP is on validating network nodes predicted to be effective drug targets in ET cell lines

engineered to permit tunable modulation of the involved genes (e.g. inducible regulation via tetracycline,

shRNA expression, etc.), mouse xenograft models and human clinical samples. For all three ET entities

(neuroblastoma, Ewing sarcoma and medulloblastoma), tunable cell lines models have been established and

already been used to experimentally identify key players of cell cycle and apoptosis regulation. Kinetic

expression patterns have been established and high throughput siRNA and miRNA screens have been

performed. First steps to use these experimental data in mathematical modelling have been undertaken. In

the reporting period, WP9 mainly focussed on re-analysis of existing and the production of de-novo data

sets, and the experimental identification and validation of candidate key players as potential drug targets in

the three ET entities. Further, for selected molecules, mouse xenograft models (i.e. TrkA) were established

and tissue micro arrays of patient samples were screened (i.e. SIRT1).


Task 9.1. Validation of predicted fragility nodes in cellular models for ET.

Ewing Sarcoma

Functional interaction of EWS-FLI1 and E2Fs. Whole genome expression, ChIP-seq, structural and

functional data obtained in Ewing sarcoma cell lines suggest that EWS-FLI1 is regulating at least 50% of

E2F target genes (Bilke, Schwentner et al. Genome Res. In revision). ChIP-PCR data further suggest that

binding of EWS-FLI1 to E2F target genes leads to a replacement of repressive E2F4 by activating E2F3.

However, due to the limitations of reporter gene assays, no direct evidence for functional EWS-FLI1 / E2F3

synergy has been obtained so far. During a one week stay of the computational scientist Vassilios

Stathopoulos from UCL at the CCRI in Vienna, a mathematical model was formulated according to the

methodology of Locke et al.32 that shall help to decide between three possibilities: (i) EWS-FLI1 binding is

sufficient to regulate E2F3 target genes; (ii) since E2F3 is a target of EWS-FLI1 itself, E2F3 binding is

sufficient to regulate E2F3 target genes; (iii) EWS-FLI1 and E2F3 regulate E2F3 target genes synergistically.

The model is developed using existing whole genome mRNA expression data (Affymetrix array) from a time

course experiment in the EWS-FLI-1 shRNA inducible cell line model Asp14. Additional early time points are

currently analysed by real-time RT-qPCR.

EWS-FLI1/ p53 interaction. Additionally, mathematical equations have been drafted to model regulation of

p53 by the NOTCH pathway in Ewing sarcoma. Specifically, the question if EWS-FLI1 knockdown induces

p53 solely via upregulation of the NOTCH ligand JAG1 or also via direct induction of the JAG1 downstream

transcriptional regulator HEY1 shall be assessed.

EWS-FLI1 regulated kinases in post-transcriptional regulation of FOXO1. As we have experimentally

observed transcriptional and post transcriptional regulation of FOXO1 by EWS-FLI1, a model is under

development that assesses the role of the directly EWS-FLI1/E2F co-regulated kinase CDK2 and the

indirectly EWS-FLI1 regulated kinase AKT in post-transcriptional FOXO1 target gene regulation.

Validation of growth inhibitory miRNAs. Validation of two miRNAs identified by VTT in WP3 to reduce

growth of the Ewing sarcoma cell line model Asp14 only in the presence but not in the absence of EWS-FLI1

(hsa-mir-631 and hsa-mir-552) is ongoing. So far, we demonstrated that miR-552 reproducibly induces a G1

arrest while miR-631 elicits a G2 cell cycle arrest in the model cell line. None of the two miRNAs was found

to be expressed in Ewing sarcoma cell lines, neither in the presence nor in the absence of EWS-FLI1, nor

were they found to be present in mesenchymal stem cells that serve as a relevant common control for Ewing

sarcoma.

Neuroblastoma

MYC / p53 interaction in regulation of mitotic checkpoint of neuroblastoma. High-risk neuroblastomas

often harbour structural chromosomal alterations, including amplified MYCN, and usually have a near-

di/tetraploid DNA content, but the mechanisms creating tetraploidy remain unclear. Gene-expression

analyses revealed that certain MYCN/MYC and p53/pRB-E2F target genes, especially regulating mitotic

processes, are strongly expressed in near-di/tetraploid neuroblastomas. Using a functional RNAi screening

approach and live-cell imaging, DKFZ identified a group of genes, including MAD2L1, which after knockdown

induced mitotic-linked cell death in MYCN-amplified and TP53-mutated neuroblastoma cells. We found that

MYCN/MYC-mediated overactivation of the metaphase-anaphase checkpoint synergizes with loss of p53-

p21 function to prevent arrest or apoptosis of tetraploid neuroblastoma cells. These results reveal novel

insights into how genetic aberrations of the p53-p21 axis contribute to tetraploidy in neuroblastoma cells.

These data enhance our understanding of how MYCN/MYC mediates aggressive behaviour in

neuroblastomas. Since overactivation of the metaphase-anaphase checkpoint supports the survival of

tetraploid cells lacking p53-p21 function, targeted inhibition of certain metaphase-anaphase checkpoint

members, such as MAD2L1, may provide a therapeutic option for neuroblastomas harboring genomic

alterations reducing p53-p21 function.

Figure 9.1. MAD2L1 silencing after vincristine treatment induces tetraploidization in neuroblastoma cells with functional p53-p21. (A) Western blot showing MAD2L1 and MYCN expression in whole-cell lysates from WAC2 and SH-EP cells stably transfected with shRNA targeting MAD2L1. (B) Flow cytometric analysis of cell cycle and ploidy in WAC2-shMAD2L1 cell cultures 36h after treatment. Curves are paired with bar-graph quantifications (below) for each treatment group. (C) 2-color FISH of WAC2-shMAD2L1 after 36h of vincristine treatment and MAD2L1 shRNA induction using centromeric probes for chromosome 6 and 8 (pink and green, respectively) and counterstained with DAPI (blue). Representative images from 250 interphases are shown. (D) Merged immunofluorescence images of WAC2-shMAD2L1 stained for centromers with CREST antibodies (green) and DNA (blue) to visualize altered nuclear size after combined MAD2L1 silencing and vincristine treatment.

An ALK transcriptional signature in neuroblastoma. As part of this work package UGENT has performed

extensive further data mining and downstream analyses following the delivery of high throughput gene

expression profiles for ALK pharmacological and shRNA knock down in a panel of selected NB cell lines.

Activating ALK mutations are present in almost 10% of primary neuroblastomas (NB) and serve as new

therapeutic targets for treatment. Clinical trials for small molecule ALK inhibitors have been initiated for NB

and other ALK driven tumour entities. However, in many instances, tumours acquire resistance to small

molecule inhibitors, illustrating the need for additional compounds directed against downstream target genes

or alternative survival pathways. To achieve this goal, we analyzed aberrant ALK signalling to identify such

vulnerable nodes for combined pharmacological targeting.

Transcriptome profiling was performed on 10 NB cell lines (ALK wild type, ALKF1174L, ALKR1275Q mutant

or amplified) following NVP-TAE684 treatment. Data mining analysis and functional validation experiments

were integrated to identify ALK driven functional cellular networks and aberrantly regulated downstream

pathway components. Differential gene expression analysis allowed the delineation of a 150-gene signature

representative for high ALK activity in NB. This signature was significantly enriched for genes implicated in

MAPK/ERK signalling, including several negative MAPK regulators, indicating strong ALK induced MAPK

activity. In addition, genes implicated in neuronal differentiation and growth control were identified. In

keeping with this observation, further analysis using Gene Ontology and Gene Set Enrichment Analysis, we

identified amongst others MAPK and AKT/mTOR pathway signalling as well as MYC/MYCN activation. The

latter is of importance in relation to the ongoing modelling for MYCN in NB as recent findings indeed have

indicated various levels of interaction between ALK and MYCN. These include transcriptional regulation and

MYCN protein stabilization. Of further relevance, this interconnection between ALK and MYCN has been

firmly established through both mouse and zebrafish models. A paper describing the cooperative effect

between ALK and MYCN in a mouse neuroblastoma tumour model has been recently published as a close

collaborative effort between the UGENT and UKE teams33.

Regulation of ETV5 by ALK. Further downstream analyses have mainly been focussed on ETV5 which was

shown to be one of the most robustly regulated ALK downstream genes. This evidence was obtained

through the analysis of several cell line models (in vitro) and mouse model data (in vivo by treating mice with

ALK driven NB tumours with ALK inhibitors) and through comparison of gene expression data of ALK

modulated cell lines and ALK, MYCN and ALK/MYCN driven mouse tumours. Of further notice, ETV5 is

known to be involved in neuronal fate decision and metastasis/invasion and is activated in a subset of

prostate cancers as a result of gene fusion events. We investigated the phenotypic effects of modulating

ETV5 in NB cells by RNAi-mediated ETV5 knock down. This showed drastic reduction in cellular growth

measured in NB cells with activated ALK but also showed effects in some cell lines with wild type ALK

indicating a broader relevance for ETV5 in NB thus also opening broader therapeutic opportunities. Elevated

ETV5 levels were apparent in human and mouse ALK positive NB. Remarkably, inhibition of ALK signalling

in NPM/ALK positive lymphoma and EML4/ALK positive lung cancer also strongly reduced ETV5 expression

which extends the relevance of ETV5 beyond NB to other ALK driven, so-called alkoma tumour entities.

We obtained for the first time a detailed picture of the transcriptional consequences of sustained ALK

signalling in human and mouse NB cells. These data further support the ALK-MYCN functional connection

which is of major importance to the other ongoing modeling efforts in ASSET. The MAPK driven ETV5

oncogene was identified as a robustly regulated ALK target in NB and other ALK activated cancers, thus

offering new therapeutic targets for molecular therapy.

Biological characterization of SY5Y neuroblastoma cells with regulatable TrkA expression. The

neuroblastoma cell model, SY5Y-TR-TrkA (details in WP3), is used for perturbation experiments. TrkA

expression can be induced by tetracycline treatment of these cells. Our first round of perturbation

experiments focuses on understanding the relationship between TrkA signalling and the following signalling

pathways or tumour physiological processes:

1. Signalling via PLCγ (details reported in WP5).

2. Interactions between TrKA-expressing neuroblastoma cells and cells of the tumour stroma by

paracrine signalling.

3. TrkA signalling preventing immune escape of neuroblastoma cells (Pajtler et al. IJC in Press): Upon

TrkA expression/activation, neuroblastoma cells express several proteins that stimulate activation of

or recognition by T cells and NK cells (Figure 9.2)

Figure 9.2. MHC class I molecules are up-regulated on TrkA-expressing neuroblastoma cells independently of

whether the cells exhibit qualities of differentiation or whether NGF is present. Flow cytometry analysis of MHC

class I expression on the surface of the indicated neuroblastoma cell lines. Bars indicate percentages of MHC class I-

positive cells, mean +/- SD of three or more experiments is presented. A. ** (P<0.01) indicates significantly higher

expression of MHC class I on SY5Y-TR-Trka cells compared to SY5Y-TR-TrkB and the empty vector control,

SY5Yvec, cells. B. Significantly higher expression of MHC class I on TrkA-expressing cells * (P<0.05) was confirmed

in a second stably transfected neuroblastoma cell line (NB69).

As a prerequisite to analysing the complex network of tumour-host signalling in the SY5Y-TR-TrkA cell

model in vivo, SY5Y-TR-TrkA cells were xenografted into nude mice, and mice were treated with

doxycycline. The doxycycline administration scheme was optimised to achieve maximum expression levels

of TrkA in the xenograft tumours (Figure 9.3). Notably, we observed no activation of TrkA in vivo in the

absence of NGF, and therefore, no effect of TrkA on tumour growth. These experiments provide the

necessary pilot data for the testing of compounds in this cell model grown as xenografts in mice in an

interventional setting.

Figure 9.3. Strong TrkA expression was observed in xenografted tumours after treatment of mice with

doxycycline. Tumours 1-6 arose from SY5Y-TR-TrkA cells xenografted subcutaneously into nude mice. Mice

harbouring xenografted tumours 4-6 were treated with doxycycline, while mice harbouring tumours 1-3 were treated

with control vehicle. Whole-cell lysates of SY5Y-TR-TrkA cells cultured in vitro are included as positive controls in the

right-hand lanes of the Western blot showing expression of TrkA receptor and the activated form of the TrkA receptor

(P-TrkA). Actin (β-actin) was used as a loading control.

Generation and validation of mouse models to test models / perturbations in vivo. Tumours

developing in genetically engineered mouse models (GEMMs) are well defined for their genetic background

and the oncogenic drivers involved. GEMMs are ideal models to perform perturbation experiments to assess

the functionality and fragility of specific pathways driving tumour formation. After tumour development, mice

can be treated with small molecule inhibitors. In addition, by introducing additional transgenes in the

respective models, also genetic modifications/perturbations of the networks can be induced. We generated

mice for tissue-specific, conditional ALKF1174L or MYCN overexpression by knock-in of a construct into the

Rosa26 locus that harbours the respective oncogenes downstream of a strong promoter and a stop cassette

flanked by loxP sites (Fig. 9.4). Cross-breeding these mice (LSL-ALKF1174L and LSL-MYCN) with mice that

express the Cre recombinase specifically in the neural crest (DBHiCre mice) resulted in tissue-specific

deletion of the stop cassette in double transgenic offspring, and thereby neural crest-specific expression of

ALKF1174L or MYCN. Both mouse models developed neuroblastomas at high incidence (Fig. 9.5).

Figure 9.4. Schematic of a of a construct for the tissue-specific, conditional expression of

oncogenes, in this case MYCN.

Figure 9.5. Double-transgenic LSL-MYCN;Dbh-iCre mice develop tumours derived from the neural crest. (A)

Kaplan-Meier analysis indicating when palpable tumours were detected in mice heterozygous for LSL-MYCN and mice

double-transgenic for LSL-MYCN and Dbh-iCre (statistically analysed by the log-rank test) (B) Bioluminescent imaging

of three representative mice carrying palpable tumours in the regions of the superior cervical ganglion (I), adrenal

glands (I, II, III) and celiac ganglion (III). Luciferase activity: low = blue, high = red.

Development of an ALK mouse model. To generate a valid GEMM for the most common ALK mutation

detected in human neuroblastomas and demonstrate oncogenecity of ALK in vivo, we targeted ALKF1174L

expression to the neural crest of transgenic mice, which indeed developed tumours at high incidence33.

These tumours resembled human neuroblastomas in morphology, metastasis pattern, gene expression and

the presence of neurosecretory vesicles as well as synaptic structures (Fig. 9.6). This ALK-driven

neuroblastoma mouse model precisely recapitulated the genetic spectrum of the human disease.

Chromosomal aberrations were syntenic to those in human neuroblastomas, including 17q gain and MYCN

oncogene amplification.

Figure 9.6. Autopsies of LSL-ALK,DBHiCre mice carrying palpable tumours. (I) Primary tumour arising from the

left adrenal displacing the left kidney caudally (tu, tumour; ki, kidney). (II) Liver from a mouse with a large

retroperitoneal tumour, showing multiple metastatic nodules. (III) Thoracic cavity of a mouse with a large

retroperitoneal tumour and metastatic lesion or second primary tumour in the left upper thorax (tu, second primary

tumour; he, heart; lu, lung). (IV) Primary tumour arising from cranial paravertebral ganglion.

Genetic perturbation experiment: Synergism of ALK an MYCN in vivo. The ALKF1174L mutation has been

associated with MYCN amplification in human neuroblastomas. We hypothesized that ALK and MYCN not

only are associated but also functionally synergize to induce neuroblastoma. To address this question, we

crossbred ALKF1174L;DBHiCre mice with the TH-NMYC transgenic mouse model, which expresses MYCN in

the neural crest driven by a rat Th promoter in the 129x1/SvJ strain. We observed significant acceleration of

tumour formation in mice transgenic for ALK and MYCN, confirming the synergism of both genes in vivo (Fig.

9.7).

Figure 9.7. Kaplan-Meier analysis showing that ALK and MYCN synergistically accelerate neuroblastoma

formation (red curve). 129B6F1: F1 background from 129x1/SvJ × C57Bl/6 mouse cross. Endpoint defined as

detection of palpable tumours.

Targeting ALK-driven tumours with ALK Inhibitors in vivo

To analyze the effect of ALK inhibition on tumour growth in vivo, GEMMs were treated with the ALK inhibitor,

TAE-684, after ALK/MYCN-driven tumour development (Fig. 9.8). Treatment caused tumour regression. We

then used a xenograft mouse model to analyse the effect of TAE-684 on human neuroblastoma cells

harbouring the ALKF1174L mutation grown as xenografts. Tumour regression was also observed in tumour

xenografts. Both GEMM and xenograft tumours relapsed after discontinuation of treatment. These results

most recently resulted in the initiation a Phase I trial to test the ALK inhibitor, LDK-378, in patients with

relapsed or progressive NB

(http://clinicaltrials.gov/ct2/show/NCT01742286?term=rhabdomyosarcoma&recr=Open&no_unk=Y&lup_s=1

1%2F07%2F2012&lup_d=30). TAE-684 showed higher efficacy than Crizotinib, an ALK inhibitor in clinical

use for non-small cell lung carcinoma34 (Fig. 9.9),

Figure 9.8. Images from TH-NMYC;ALKF1174L;DBHiCre triple-transgenic mice treated with either the ALK

inhibitor, TAE-684, or control. Bioluminescent luciferase images are shown before treatment and after 14 days of

treatment.

http://clinicaltrials.gov/ct2/show/NCT01742286?term=rhabdomyosarcoma&recr=Open&no_unk=Y&lup_s=11%2F07%2F2012&lup_d=30

http://clinicaltrials.gov/ct2/show/NCT01742286?term=rhabdomyosarcoma&recr=Open&no_unk=Y&lup_s=11%2F07%2F2012&lup_d=30

Figure 9.9. TAE-684 reduces viability of human neuroblastoma cells harbouring the ALKF1174L mutation grown

as xenografts in mice. Mice were subcutaneously inoculated with 2 × 107 SH-SY5Y neuroblastoma cells, and treated

after palpable tumours appeared. Mice were treated daily with TAE-684, crizotinib (an ALK inhibitor in clinical use for

non-small cell lung carcinoma), or vehicle (control), and tumour size was measured with a digital caliper. Mean tumour

size is displayed until time of euthanasia of the first mouse in each group (treatment day 12 for control and crizotinib

groups, treatment day 28 for TAE-684 group).

Medulloblastoma

Growth regulatory kinases downstream of MYC in medulloblastoma. MYC is known to promote tumour

formation and metastasis by regulating proliferation, differentiation, angiogenesis and metabolism. As a

transcription factor without evident druggable domain and regulating essential processes in proliferative

tissue, MYC may not be an appropriate drug candidate. An alternative approach is to identify genes required

for proliferation and survival of MYC over-expressing cells. With the goal to identify therapeutic targets in

MYC-overexpressing MB cells, UBERN performed a high-throughput siRNA screen targeting protein and

lipid kinases. 719 known kinases were silenced with 3 different siRNA sequences in vector-transfected or

MYC-transfected DAOY cells. The comparison of cell proliferation in both cell lines yielded 3 top candidates:

PCTK1, EPHA7 and AKAP12. Depletion of these genes selectively decreased cell proliferation of MYC-over-

expressing cells suggesting an essential role in the context of MYC. We have validated the three top

candidates (PCTK1, EPHA7 and AKAP12) in the UW228 MB cell line with inducible MYC expression, by

showing a comparable effect of the siRNAs on cell proliferation upon expression of MYC. We have also

validated the effects of the siRNAs by analyzing target knock-down by RT-PCR and Western blot analysis. In

addition, the expression of the top candidates in primary MB was analyzed in two available cDNA microarray

datasets from primary MB. Together our results have started to uncover new potential molecular targets in

medulloblastoma hallmarked by MYC over-expression.

Task 9.2. Transcriptomic expression profiles for selected combinations of kinase inhibitors in the

ETcell line panel.

Receptor tyrosine kinases in embryonal tumours. UBERN has assembled a 15 cell-line panel comprising

5 cell lines from each ET entity, namely Ewing’s sarcoma (ESFT), medulloblastoma (MB) and neuroblastoma

(NB). The panel consists of RDES, A673, TC71, TC-252, SK-N-MC (ESFT), DAOY, D341, D458, PFSK,

UW228 (MB), SK-N-BE, LAN5, IMR5, SH-SY5Y, SKNAS (NB). We have screened the cell line panel by

Western blot and RT-PCR analysis for the expression of selected receptor tyrosine kinases (RTKs) and

downstream signalling mediators, as well as for MYC, MYCN and EWS-Fli1.

UBERN further has investigated the potential of targeting the axis of the insulin-like growth factor-1 receptor

(IGF-1R) and PI3K signalling in two ET entities neuroblastoma and medulloblastoma. By treating

neuroblastoma and medulloblastoma cells with R1507, a specific humanized monoclonal antibody against

the IGF-1R (Roche), we could observe cell line-specific responses and in some cases a strong decrease in

cell proliferation. In contrast, targeting the class I PI3K p110α with the selective inhibitor PIK75 resulted in

broad anti-proliferative effects in a panel of neuro- and medulloblastoma cell lines. Additionally, sensitization

to commonly used chemotherapeutic agents (cisplatin and doxorubicin) occurred in medulloblastoma and

neuroblastoma cells upon treatment with R1507. Furthermore, by studying the expression and

phosphorylation state of IGF-1R/PI3K downstream signalling targets we confirmed that R1507 and PIK75

downregulated the activation of the IGF-1R/PI3K signalling pathway. In addition, apoptosis occurred in

embryonal tumour cells after treatment with PIK75 or R1507. Together, our studies demonstrate the potential

of targeting the IGF-1R/PI3K signalling axis in embryonal tumours.

Ewing sarcoma

PRKCB as essential EWS-FLI1 target in Ewing sarcoma. Using time series analysis of Ewing sarcoma

cell line inhibited for EWS-FLI1, CURIE identified Protein Kinase C Beta (PRKCB) as a gene strongly

activated by EWS-FLI135. Moreover, comparison with other paediatric or bone cancers showed that PRKCB

is highly and specifically overexpressed in Ewing sarcomas. Its transcriptional activation is directly regulated

by the EWS-FLI1 oncogene as demonstrated by luciferase reporter and ChIP experiments. Getting insights

in PRKCB activity we show that, together with PRKCA, it is responsible for the phosphorylation of histone

H3T6, allowing global maintenance of H3K4 trimethylation on a variety of gene promoters. In the long term,

PRKCB RNA interference induced apoptosis in vitro. More importantly, in xenograft mice models, complete

impairment of tumour engraftment and even tumour regression were observed upon PRKCB inhibition,

highlighting PRKCB as a most valuable therapeutic target. Actually, using enzastaurin, a specific inhibitor of

PRKCB we could inhibit tumour growth in vivo providing proof of concept for the use of such drugs in

preclinical studies. Deciphering PRKCB roles in Ewing sarcoma using expression profiling, we found a

strong overlap with genes modulated by EWS-FLI1 and an involvement of RPKCB in regulating crucial

signalling pathways, such as NFKB or TGFβ. Altogether, we show that PRKCB may have two important

independent functions and should be considered as highly valuable for understanding Ewing sarcoma

biology and as a promising target for new therapeutic approaches in Ewing sarcoma.

Task 9.3. Testing of top drug combinations for each ET entity in xenograft mouse models.

Work on this task has not yet started, and awaits identification of promising drug combinations to test. The

first combination is planned for 2013, and results from output of Task 8.2.

Task 9.4. Re-analysis of existing high-throughput data sets from primary human NB, MB and ESFT

for the validation of model predictions.

Influence network linking EWS-FLI1 to cell cycle and apoptosis phenotypes. Silencing EWS-FLI1

induces cell cycle alteration and ultimately leads to apoptosis, but the exact molecular mechanisms are still

unclear as many genes are implicated in this signalling process. In order to clarify these mechanisms,

CURIE has used a systems biology approach to establish an influence network linking EWS-FLI1 to cell

cycle and apoptosis phenotypes. Transcriptome time-series of the shEWS-FLI1 tetracycline inducible clones

(shA673-1C shA673-2C) were used to identify core modulated genes. For that, a scoring method based on

fitting curves was applied to select genes varying coherently along the time-series. Literature data mining

was then performed to connect these modulated genes into a network. The validity of the network was

assessed by siRNA/RT-QPCR experiments on a subpart of this network and confirmed most of the links.

Based on this network and the transcriptome data, CUL1 was identified as a new potential target of EWS-

FLI1 that was then experimentally confirmed using ChIP and silencing experiments. Altogether, using an

original methodology of data integration, we provide the first version of EWS-FLI1 network model around cell

cycle and apoptosis (Stoll and Surdez, submitted).

Task 9.5. Immunohistochemical validation of selected proteins on existing tissue microarrays.

Validation of SIRT1 as a key cell cycle regulator downstream of EWS-FLI1 in Ewing sarcoma. More

than 90% of Ewing sarcomas express wildtype p53. CCRI has previously demonstrated that knockdown of

EWS-FLI1 in wildtype p53 Ewing sarcoma cell lines results in p53 induction via activation of the NOTCH

signalling pathway36. We had found that the NOTCH transcriptional target HEY1 upregulates p53 protein

levels leading to induction of p21 and subsequent cell cycle arrest. Re-analysis of transcriptomic data from

EWS-FLI1 knockdown cell lines and from ectopic HEY1 overexpression cell lines identified the NAD

dependent type III deacetylase SIRT1 as being activated by EWS-FLI1 but repressed by HEY1. Knockdown

of EWS-FLI1 was found to induce HEY1 and suppress SIRT1 expression allowing for p53 acetylation and

consequently activation. These results identified SIRT1 as a key cell cycle promoting protein downstream of

EWS-FLI1.

Now, we analysed a total of 310 Ewing sarcoma tumours on two tissue micro arrays for

immunohistochemical detectability of SIRT1. 5% of primary tumours from patients with localized disease

expressed very high and 25% high nuclear levels of SIRT1, while 50% were completely negative, and 20%

almost completely negative. Strikingly, when analysing metastasis samples (n=30), 30% expressed very

high, and another 33% high levels of SIRT1, while only 20% were completely and 17% almost completely

negative. These results suggest that SIRT1 is primarily associated with metastasis.

Significant results

13. Regulation of E2F targets by EWS-FLI1 established.

14. Distinct growth regulatory roles of mir-631 and hsa-mir-552 in the Ewing sarcoma cell line model.

15. MYCN/MYC-mediated overactivation of the metaphase-anaphase checkpoint synergizes with loss of

p53-p21 function to prevent arrest or apoptosis of tetraploid neuroblastoma cells.

16. 150-gene signature representative for high ALK activity in NB established.

17. Genetically engineered mouse models (ALK mutation +/- MYCN overexpression) for neuroblastoma

established.

18. ALK inhibitors tested in above mouse models.

19. MAPK driven ETV5 oncogene was identified as a robustly regulated ALK target in NB and other ALK

activated cancers.

20. PCTK1, EPHA7 and AKAP12 identified as MYC dependent growth regulatory kinases in

medulloblastoma.

21. PRKCB identified as essential kinase target of EWS-FLI1 affecting histone code and Ewing sarcoma

cell proliferation.

22. Regulatable TrkA expression model established and tested in mouse models.

23. CUL1 was identified as a new potential target of EWS-FLI1 using an influence network linking

EWS-FLI1 to cell cycle and apoptosis phenotypes.

24. SIRT1 identified as a metastasis associated key regulator downstream of EWS-FLI1 in Ewing

sarcoma.


None


Participant number



2 CCRI 3 1 Postdoc salary – 3PMs

7 UKE 9 1 Technician 6PM, 1 PhD student 3PM

8 UGENT 2 1 Postdoc: David Camacho-Trujillo – 2PMs


14

WP10. Data management & Systems level knowledge integration

Objectives

This work package provides a data management framework for all experimental data and models generated

in ASSET. It establishes an interrogable project database for holding shared experimental data, and a

repository for mathematical models and data integration pipelines that also exploit publicly available data. At

its heart is a data warehouse to support and enable the combination of project proprietary data with publicly

available data sources. The use of semantic technologies will allow integrating and interrogating data at

many levels. Data frameworks will be built with future maintenance in mind, with provisions to integrate with

alternative systems, for example through the use of web services. This database will be made accessible to

the scientific community at the end of the project, and represents the lasting legacy of ASSET to the

scientific community.


Task 10.1. Construction of a state-of-the-art and data warehouse for ASSET.

The different data types on the ASSET project, and the cell line systems used by different partners are being

catalogued. This catalogue is a living document and is being updated as more data is being generated. An

upload of this data to the central warehouse has begun and interfaces to query the data are expected to

begin being visible in 2013.

Task 10.2: Data mining of publicly available mutation-phenotype data, protein-protein complexes and

pathways and incorporation of these data within the relational database framework.

Publicly available mutation data on the cell lines SH-SY5Y, IMR and ASP14 are being compiled internally for

comparison with exome sequencing data currently being generated. Data from the CCLE compendium made

available in 2012 has been downloaded and will be linked with ASSET generated data as appropriate in

2013.

Task 10.3: Develop web-services and data visualisation.

Not applicable at this stage of the project.

Task 10.4: Linking the model repository for ASSET to community-wide efforts (e.g. Reactome,

CellML, BioModels, DOQCS, ModelDB, etc.).

Not applicable at this stage of the project.

Significant results

1. System Level Data Integration and Storage

Through the chemoinformatics work on proposing compounds for synergy screens, 3 additional

compounds were suggested by the UCPH partner. One of these, a PI3K inhibitor, has been identified as

a top candidate from the synergy screen data for SY5Y. Follow-up work on the synergy screening data is

now in progress.

Data generated on different parts of the project are being catalogued and transfers to the central data

warehouse have begun. As this catalogue gets richer through the project, it is expected that appropriate

questions interrogating the data at a systems level will be addressed through manual and automated

analyses.

The CNIO and UCPH partners have identified databases containing information about pathways and

mutation phenotype data and also the algorithms required to integrate this data. Focus has been on

pipelines for cancer genome analysis and key databases and software helpful to integrate genomic data

with phenotypic information and drug response.

CURIE Systems Biology team (AZ) has developed computational tools that will be useful to 1) model

biochemical pathways with respect to the effect of perturbations on cell fate (MaBoSS software), 2)

analyse the structure of biological pathways involved in pediatric tumours (BiNoM software), 3) predict

the effect of combinatorial perturbations on particular cell fates (cell death, proliferation, survival, etc.)

and the side effects of such perturbations (for example, cell toxicity).

BiNoM (http://binom.curie.fr ), made freely available to the community, includes multiple features

connected with analysis and visualization of biological pathways (Bonnet et al, 2012). MaBoSS

(https://maboss.curie.fr/ ) implements a modeling framework based on qualitative approaches that is

intrinsically continuous in time1. OCSANA (http://bioinfo.curie.fr/projects/ocsana/) allows identifying and

ranking optimal combinations of intervention points in a network to block signals from specified source

nodes to specified targets (Vera-Licona, 2012).

2. Exome Sequencing and Comparative Analysis of ETs

Next generation sequencing technology was used to fully sequence the exomes of the core cell lines of the

three childhood cancers in order to survey their exact mutual status.The cell lines representing the three ET

entities, which were analyzed so far, are:

ASP14 (Ewing Sarcoma cell line) with tet-regulatable EWS-FLI1.

SH-SY5Y (Neuroblastoma cell line)

IMR-5/75 (Neuroblastoma cell line)

3. The analysis framework of the data generated from the exome sequencing of the selected cell lines

involved identifying the differences between the ET cell lines sequenced and the human reference genome.

A number of sequence variants per cell line can be detected including:

Restriction of the search to non-synonymous variants (nsSNVs). nsSNVs, lead to a single amino

acid change in the protein product so they are more likely to change the protein functions and impact

cellular phenotypes.

A number of computational biology tools were used to predict degree of deleteriousness of the

nsSNVs by various computational algorithms using genomic features like amino acid properties,

protein structure and cross-species conservation.

http://binom.curie.fr/

https://maboss.curie.fr/

http://bioinfo.curie.fr/projects/ocsana/

Since for rare severe diseases, underlying causal mutations are very unlikely to be common in

human populations, the next step was the filtering the “common variants” (variants with more than

1% Minor Allele Frequency (MAF)).

Variant quality filtering. Selection of variants with sufficient quality. (Quality is defined as the Phred

scaled probability of Probability that REF/ALT polymorphism exists at this site given sequencing

data).

Filtering according to read depth coverage of each variant. We have considered as confident the

variants with read depth coverage more than 10.

Knowledge level analysis. Functional and structural information was used in order to evaluate the

effectiveness of the final variant dataset. In addition exploration of the protein network of the

respective genes was performed to identify possible physical and functional interactions with

candidate genes for this specific type of cancer. Preliminary results in terms of the number of the

variants and the respective genes of interest in each cell line is illustrated on Table 10.1.

Table 10.1: The numbers of the prioritized variants as well as the respective proteins and genes for each one

of the cell lines analyzed – Preliminary Results.

As a starting point we focused on the variants prioritized as possibly functional at the neuroblastoma cell line

SH-SY5Y (more details about the analysis framework in figure 10.1). During the analysis, we identified with

high confidence variants in genes involved in fundamental biological processes like proliferation, cell cycle

regulation and apoptosis. Our results expand the list of the potential neuroblastoma related genes,

contributing not only to our knowledge of this specific type of cancer but also to our understanding of the

cancer genome in general. Future plans include the further exploration and comparative analysis of the cell

lines that are already provided from the consortium.

CELL LINE NAMES Variants Genes

SH-SY5Y 54 51

IMR-5/75 109 107

ASP14 (Wild type) 76 74

ASP14 (with knockdown of EWS-

FLI) 58 58


None.



Participant number



7 UKE 1 1 PhD Student 1PM

14 CNIO 12 1 Senior Staff: 12PMs

13 UCPH 11 Scientist Kalliopi Popi Tsafou (7 PM), PhD Kalliopi Popi Tsafou, (4PM).


24

Figure 10.1. Exome variants prioritization and filtration framework. At the right part of the figure, are provided the numbers of variants at each step on the SY5Y cell line.

WP11. Build a network/dynamic model as a reference framework to correlate genetic

alterations with clinical cancer phenotypes.

Objectives

This work package provides a systems level approach to integrating various data from the WP10 data

warehouse. The emphasis is on mechanistic interpretations based on pathways and protein-protein

interaction complexes linking the generated data with clinical phenotypes or endpoints. This workpackage

aims to combine qualitative and quantitative modelling, where qualitative data integration schemes will lead

to models that can be used to select and design quantitative simulation efforts targeting key pathways and

temporal cancer-linked mechanisms. WP11 will, thus, address the issue of integrating data and models at

two levels. Components can be linked using semantic annotation to allow viewing and component analysis in

different modelling contexts. This allows, for instance, the integration of data across regulatory and signalling

networks. Secondly, expanding the semantic annotations into rule-based frameworks will permit zooming in

and out of models from coarse- to fine-grained resolution. A systems level analysis of data and association

with phenotypes will go beyond current studies where single entity results or perturbations show limited

reproducibility due to the multiple routes of signalling in a cell leading to similar end effects.


Workpackage 11 tasks were scheduled to start at month 19. During this reporting period, however, we have

been working together with UCPH on setting the basis needed to achieve the tasks listed below. This work

has contributed to the highlights reported for WP10 above.

Task 11.1. Construction of cancer-specific protein-protein interaction networks (rewired cancer

interactomes) linked to expert annotation of selected disease-related protein complexes.

This work has started evaluating such networks from the view of druggable kinase pathways as described

under highlights in WP10.

Task 11.2. Interpretation and mapping variation data.


Task 11.3. Exome and targeted exon sequencing of ETs.


Task 11.4. Comparative analysis of ET and adult cancer genetic alterations as an approach to

delineate driver from passenger mutations

The goal of the ASSET project is to capture common pathogenetic principles shared by different Embryonal

tumours (ET) by combining high-throughput analysis and high content analysis of the genome, transcriptome

and proteome with mathematical modelling. One of the alterations that cause cells to become cancerous is

the formation of chimeric transcripts. For example most cases of Ewing's sarcoma (85%) are the result of a

translocation between chromosomes 11 and 22, which fuses the EWS gene of chromosome 22 to the FLI1

gene of chromosome 11. Other specific cellular phenotypes are characterized by expression of chimeric

transcripts, for example, the fused BCR/ABL, FUS/ERG, MLL/AF6 and MOZ/CBP genes are expressed in

acute myeloid leukemia (AML) (Panagopoulos et al. 2003; Nambiar et al. 2008), and the TMPRSS2/ETS

chimera is associated with overexpression of the oncogene in prostate cancer (Nambiar et al. 2008). In

principle, chimeric transcripts can augment the number of gene products available in a given genome and

are suspected to function not only in cancer (Thomson et al. 2000; The ENCODE Project Consortium 2007;

Gingeras 2009) but also in normal cells (Akiva et al. 2006; Parra et al. 2006). Chimeric RNAs comprise

exons from two or more different genes and have the potential to encode novel proteins that alter cellular

phenotypes. To date, numerous putative chimeric transcripts have been identified among the ESTs isolated

from several organisms and using high throughput RNA sequencing. The few corresponding protein

products that have been characterized mostly result from chromosomal translocations and are associated

with cancer.

Due to the importance these chimeric transcripts have in the cancer onset and progression partner CNIO

has dedicated its effort during this year to scan the human genome searching for putative chimeric

transcripts. The results obtained will swell the knowledge produced by WP11 about the possible causes of

cancer progression. We systematically established that some of the putative chimeric transcripts are

genuinely expressed in human cells. Using high throughput RNA sequencing, mass spectrometry

experimental data, and functional annotation, we studied 7424 putative human chimeric RNAs. We

confirmed the expression of 175 chimeric RNAs in 16 human tissues, with an abundance varying from 0.06

to 17 RPKM (Reads Per Kilobase per Million mapped reads). We show that these chimeric RNAs are

significantly more tissue-specific than non-chimeric transcripts. Moreover, we present evidence that

chimeras tend to incorporate highly expressed genes. Despite the low expression level of most chimeric

RNAs, we show that 12 novel chimeras are translated into proteins detectable in multiple shotgun mass

spectrometry experiments. Furthermore, we confirm the expression of three novel chimeric proteins using

targeted mass spectrometry. Finally, based on our functional annotation of exon organization and preserved

domains, we explored potential features of chimeric proteins. The results suggest that chimeras significantly

exploit signal peptides and transmembrane domains, which can alter the cellular localization of cognate

proteins. Taken together, these findings establish that some chimeric RNAs are translated into potentially

functional proteins in humans. These results have been published.37

Significant results

Systematic detection of gene fusion products in human cancers


None.


Participant number



4 CURIE 2 1 Postdoc 2PM

14 CNIO 3 1 Postdoc – 3PM


3.2.3 Project Management (WP12) & Training (WP13) Consortium Management Tasks and Achievements A. Coordination of ASSET Meetings

1) April 25th and 26th

2012, ASSET Annual Meeting, Majorca Spain

The ASSET 2nd annual General Assembly meeting was held in Spain, on April 25th and 26th 2012 and all partners attended. Walter Kolch opened the meeting and Lauren Montague updated the partners on all key financial and administrative details for the period. Philip Smyth provided an update on the education and outreach activities of the consortium and Ian Barwick gave an update on industry engagement. An update per WP was provided by the leader of each WP. The WP partners presented results from the previous 12 months and outlined future planned work for the following 12 months and provided an update on upcoming milestones and deliverables. The scientific discussions on day one and two continued into the evenings with the key next steps outlined for each WP as the key emphasis of the two day meeting was progress towards the workplan milestones.

The ASSET Steering Committee and the Industrial Liaison & Exploitation Committee (ILEC) held closed meetings on day one also. The Steering committee is made up of the project coordinator, the WP leaders and the chair of the Industrial Liaison and Exploitation Committee. During the ILEC meeting Ian Barwick Business Development Manager from Systems Biology Ireland was appointed as the chair of the ILEC.

2) September 9th 2012, ASSET Workshop, Dublin The ASSET workshop was held in conjunction with UCD System Biology Ireland’s (Partner 1), International Conference in Systems Medicine which was held in Ireland from the 9th to 13th September. This was held during Dublin’s tenure as the 2012 European City of Science. In keeping with the ASSET work plan, a half-day workshop on Embryonal Tumours, to enhance and highlight ASSET’s clinical/biological interface, was included as part of the program. This workshop was held on day one of the conference. This workshop was open to all conference participants. There was also a closed ASSET meeting for partners only, at the end of the workshop to discuss the upcoming reporting requirements and deadlines. It was agreed at this meeting that a number of smaller working group meetings would be held over the next 6 months. These included working groups on a) RNA sequencing linked to microRNA profiling; b) Signalling and Proteomics and c) Drug and Drug Screens as detailed in section 5: Future Meetings.

The ASSET Workshop speakers included the following:

Workshop : Cell Cycle Perturbations in Embryonal Tumours

Neuroblastoma and Cell Cycle Overview Frank Westermann, Department of Tumor Genetics, Cancer Research Center- DKFZ, Heidelberg

N-myc Dependent Tumour Models Johannes H. Schulte, Pediatric Oncology Research Lab, University Hospital of Essen, Germany

Copy number alterations in neuroblastoma preferentially target MYCN down-stream genes: implications for modeling and study of MYCN regulated signaling pathways Frank Speleman, Center Medical Genetics, Ghent University, Belgium

Application of Omics Technologies to Elucidate the MYCN Transcriptional Network in Neuroblastoma David Duffy, Systems Biology Ireland, University College Dublin, Ireland

Medulloblastoma Overview Alexandre Arcaro, Paediatric Oncology, Department of Clinical Research, University of Bern, Germany

Ewing Sarcoma Overview Heinrich Kovar, Children’s Cancer Research Institute, Vienna, Austria

Gene Networks in EWS Loredana Martignetti, Institut Curie, Paris, France

Data Integration Ramneek Gupta, University of Copenhagen, Denmark

Modelling approaches Florian Lamprecht, Cancer Research Center - DKFZ, Heidelberg, Germany

3) Mini-meetings were held throughout the year including.

UCL – CCRI Meeting - 20 – 24 August 2012 During the research visit UCL and CCRI established two models for small network structures involved in Ewing’s sarcoma tumors based on current time resolved expression data from CCRI. The first model will be used for studying the mechanism of E2F target genes regulation by EWS-FLI1 while the second model is intended to study the activation of the Notch pathway by EWS-FLI1 and the regulation of HEY1. The models still need parameterization and validation using the available experimental data. Initial simulations with the models will be used in order to test the experimental conditions under which the different hypothesis can be discriminated and also study the behaviour of the models. These simulations will inform the experimental conditions for lab experiments in order to produce new and more detailed time course data.

4) Teleconferences and Skype video conferences were organized and held throughout the year.

5) Future Meetings

Preparations for future meetings have begun as follows:

5a) Working Group/Round-table Meetings:

December 14th, 2012, RNA sequencing working group, Essen The first working group meeting of the RNA Sequencing group was held 14th December 2012 in Essen. This meeting focussed on RNA sequencing. The key aim of this RNA sequencing working group was to organize the current and future work in data production and analysis in order to maximize synergies across the partners involved. The meeting format brought together the partners who are working in that space. The partners that participated in this meeting include UCD, CCRI, UKE, UGENT, UBERN and DKFZ. At a later stage these will link into the other working groups. Presentations of their work were followed by round table discussions on prospective joint publications and outcomes for the RNA Sequencing based data. As result we have harmonised data generation and designed plans for several publications.

Signalling and Proteomics and Drug and Drug Screens working groups These working group meetings will be held in 2013.

5b) ASSET Annual Meeting (Workshop and Midterm Review), Vienna, June 10th and 11th, 2013. CCRI’s Heinrich Kovar (Partner 2) is organizing a conference in Vienna (celebrating the 25th birthday of the CCRI) with the title "Paediatric Cancer Research at the INTERFACE" from June 6th to 8th, 2013. This meeting will be followed by an ASSET workshop that will take place at the Hotel Schloss Wilhelminenberg. This will be a 1 day workshop and aims to cover the three major areas of ASSET; data generation, data integration and modeling, validation) individually for each tumor entity. The following day the ASSET midterm review meeting will be held at the same venue. During the midterm review the WP leaders will present their WP update and the researchers will present their work more informally at a poster session. Members of the Scientific Advisory Board will attend the workshop and the midterm review and they will subsequently provide a short report and recommendations about the future of ASSET for submission to the project officer.

B. Training and Dissemination Activities (WP13)

1. Website (www.asset-fp7.eu; also www.ucd.ie/sbi/asset).

The ASSET project website has been updated with news and events and training videos on the password secured extranet. Online seminars are also available in other areas of biomedical research to ensure the ASSET researchers learn about a broad range of research and techniques.


http://www.ucd.ie/sbi/asset).

2. ASSET Twitter (@ASSETFP7)

The ASSET Twitter social media site was set up to create further awareness of the ASSET project and to highlight events at which ASSET are involved in. During period 3 we will focus on increasing the number of followers on the ASSET twitter.

3. Training ASSET introduces a training program to break down barriers and encourage innovation. The training program aims to facilitate the communication and active engagement between the wet and dry disciplines. Therefore, while it mainly targets PhD students and postdoctoral scientists, it also engages PIs. During the past year meeting, a multi-faceted approach to the project training strategy described in above and detailed in WP13 was developed and included

• Direct exchange of wet and dry laboratory personnel and mini-meetings between partner groups to create an immersive environment to learn techniques, language and culture. To this end, the following laboratory exchanges occurred during the past year:

• Vienna - Joint meeting of CCRI (Partner 2) and DKFZ (Partner 5) (participants H. Kovar, F.

Westermann, T, Höfer, F. Lamprecht, Ban, Schwentner) (November 2011).

• Dublin – (Partner 1) Kolch/Kholodenko (NUID UCD) hosted Gur Pines from Weizmann (Partner

11) From June 17th to 22nd 2012.

• Svetlana Bulashevska (BONN) visited Weizmann Institute of Science from 18-23 September. Svetlana met with Yosef Yarden the Department of Biological Regulation (Partner 11

WEIZMANN ) and presented her work on the 19th of September, the talk was entitled “SwitchFinder: A novel method for the analysis of time- resolved biological data.”

• Future Exchange/Training Visits

• Annelies Fieuw (Partner 8 UGENT) a PhD student in the lab of Frank Speleman in University of Ghent will spend two weeks in the lab of the Barillot research group in Paris as an exchange under ASSET.

Moving forward researchers will be asked to submit short feedback forms on their exchange visits, this will ensure an accurate log of the potential benefits and techniques that researchers can learn when attending different institutes. Please see short Exchange visit log below which will be available via an online link, which will be circulated with an online news update, for researchers to complete:

Research Visitor Name

Objectives of visit (maximum 100 words)

Name of Institute visited

Visit Dates

Training/Techniques/Topics discussed

Seminars attended and given whilst at host institute. Please provide title of each seminar given or attended.

Did you find the visit beneficial

Any further comments

• Active Participation by Early Stage Researchers (ESRs) at all Project Meetings All post-doctoral researchers and graduate students who attended the ASSET annual meeting held in Majorca 2012 presented minimum 15 minute talks regarding their current and future research plans for the ASSET project. Each researcher’s presentation was recorded for training purposes so that they could review their presentations after the meeting. As part of researchers career development they are encouraged to take part in training schemes held at an institutional level such as Project Management in the Research Context, Managing Research Projects, Writing Effective Research Proposals, Getting Published in the Sciences and Research Protocol and Design.

• Training Seminars Seminars in the area of Biomedical Research Audio Class seminars held by Systems Biology Ireland, NUID UCD, (partner 1) from international experts in various areas of biomedical research are available on the ASSET Extranet to give the ASSET researchers access to a broad range of topics that may be of interest to them.

Other Dissemination Activities

• SYSMED Conference Dublin 9th to 13th September 2012 – ASSET had a one day workshop during the SYSMED conference and therefore the ASSET logos were included on all promotional material including the SYSMED abstract book, programme and website.

ASSET partners attended a number of conferences and workshops during the period presenting the research from the ASSET project, the details of these oral presentations and poster presentations are included in Annex 1 and are also included in the list of dissemination activities on the portal.

C. Changes in the consortium membership 1. As per the first annual report, Bayer Technology Services, Leverkusen merged the activities performed in Witterswil (Zeptosens) with their operations in Leverkusen. In this period the paperwork was finalized to confirm that the patent and trademark rights related to the ZEPTO contribution to the grant agreement was

transferred from Bayer Technology Services Gmbh to Bayer Intellectual Property Gmbh effective of 1st April 2012. Agreements remain unchanged and any reference in the grant agreement including annexes to Bayer Schweiz AG or ZEPTO shall be deemed to be a reference to Bayer Technology Services Gmbh. Bayer Technology Services Gmbh is provided with all necessary rights to fulfill all obligations under said agreements. The letters confirming this transfer of rights was submitted to the Commission for review.

2. The addition of Rheinische Friedrick-Wilhelms-Universitaet BONN (BONN) as a new beneficiary to the grant agreement. Svetlana Bulashevska, one of the DKFZ (Partner 5) PIs, has moved to the University of BONN. The budget and effort split have been agreed between DKFZ and BONN and agreed by the coordinator. The agreement of each of the affected partners is in place, along with the agreement of the rest of the consortium. The contract amendment was initiated in P2 and the final paperwork was submitted to the Commission for review. D. Financial Issues • Internal approval was sought from the project officer for CNIO (Partner 14) to use part of their budget for subcontracting. CNIO’s role within the project is to provide a systems level approach to integrating various data from the WP10 data warehouse. Their emphasis is on mechanistic interpretations based on pathways and protein – protein interaction complexes linking the generated data with clinical phenotypes or endpoints. The final workpackage aims to combine qualitative and quantitative modelling, where qualitative data integration schemes will lead to models that can be used to select and design quantitative simulation efforts targeting key pathways and temporal cancer-linked mechanisms. Recently, they illustrated their systems level approach using integrative RNA- sequencing and mass-spectrometry analysis to study three types of tumours (breast, prostate and ovary). However, in order to assess at a high level, the sensitivity and correct biological interpretations of their system modelling, they need to produce supportive experiments and to present concrete biological evidences for the uncovered cancer- linked mechanisms. The experiments include the verification of concrete cases in vitro using the RT-qPCR method as well as targeted proteomics analysis using the de-novo synthesized peptide standards. As it is not possible to conduct experiments identical or similar to those suggested at their institute or at any of the ASSET partners, they need to conduct them externally. These are in line with the goals of the ASSET project. None of the other tasks and milestones promised in the workplan will be affected by this. • Normal queries from partners regarding the appropriateness of expenses were routinely fielded by the Coordinator (UCD-Partner 1). E. Other Consortium issues None.

References

1 Stoll, G., Viara, E., Barillot, E. & Calzone, L. Continuous time boolean modeling for biological

signaling: application of Gillespie algorithm. BMC Syst Biol 6, 116 (2012). 2 Molenaar, J. J. et al. Sequencing of neuroblastoma identifies chromothripsis and defects in

neuritogenesis genes. Nature 483, 589-593 (2012). 3 Mestdagh, P. et al. The miR-17-92 microRNA cluster regulates multiple components of the TGF-

beta pathway in neuroblastoma. Molecular cell 40, 762-773 (2010). 4 Conkrite, K. et al. miR-17~92 cooperates with RB pathway mutations to promote retinoblastoma.

Genes & development 25, 1734-1745 (2011). 5 Nittner, D. et al. Synthetic lethality between Rb, p53 and Dicer or miR-17-92 in retinal progenitors

suppresses retinoblastoma formation. Nat Cell Biol 14, 958-965 (2012). 6 Lan, S., Stathopoulos, V., Shahbaba, B. & Girolami, M. Lagrangian dynamical Monte Carlo.

arXiv:1211.3759 [stat.CO] (2012). 7 Boeva, V. et al. Control-FREEC: a tool for assessing copy number and allelic content using next-

generation sequencing data. Bioinformatics (Oxford, England) 28, 423-425 (2012). 8 Cazes, A. et al. Characterization of Rearrangements Involving the ALK Gene Reveals a Novel

Truncated Form Associated with Tumor Aggressiveness in Neuroblastoma. Cancer Res 73, 195-204 (2013).

9 Guerreiro, A. S. et al. A sensitized RNA interference screen identifies a novel role for the PI3K p110gamma isoform in medulloblastoma cell proliferation and chemoresistance. Mol Cancer Res 9, 925-935 (2011).

10 Salm, F. et al. RNA interference screening identifies a novel role for autocrine fibroblast growth factor signaling in neuroblastoma chemoresistance. Oncogene (2012).

11 Pauwels, E. et al. A probabilistic model for cell population phenotyping using HCS data. PloS one 7, e42715 (2012).

12 Zhang, P. et al. c-Myc downregulation: a critical molecular event in resveratrol-induced cell cycle arrest and apoptosis of human medulloblastoma cells. Journal of neuro-oncology 80, 123-131 (2006).

13 Martignetti, L. et al. Antagonism pattern detection between microRNA and target expression in Ewing's sarcoma. PloS one 7, e41770 (2012).

14 Martignetti, L., Zinovyev, A. & Barillot, E. Identification of shortened 3' untranslated regions from expression arrays. Journal of bioinformatics and computational biology 10, 1241001 (2012).

15 Ma, L. et al. miR-9, a MYC/MYCN-activated microRNA, regulates E-cadherin and cancer metastasis. Nat Cell Biol 12, 247-256 (2010).

16 Mestdagh, P. et al. MYCN/c-MYC-induced microRNAs repress coding gene networks associated with poor outcome in MYCN/c-MYC-activated tumors. Oncogene 29, 1394-1404 (2010).

17 De Brouwer, S. et al. Dickkopf-3 is regulated by the MYCN-induced miR-17-92 cluster in neuroblastoma. International journal of cancer 130, 2591-2598 (2012).

18 Kumps, C. et al. Focal DNA Copy Number Changes in Neuroblastoma Target MYCN Regulated Genes. PloS one 8, e52321 (2013).

19 Schulte, J. H. et al. Accurate prediction of neuroblastoma outcome based on miRNA expression profiles. International journal of cancer 127, 2374-2385 (2010).

20 Schulte, J. H. et al. High ALK receptor tyrosine kinase expression supersedes ALK mutation as a determining factor of an unfavorable phenotype in primary neuroblastoma. Clin Cancer Res 17, 5082-5092 (2011).

21 Miller, C. et al. Dynamic transcriptome analysis measures rates of mRNA synthesis and decay in yeast. Molecular systems biology 7, 458 (2011).

22 Bogenmann, E., Peterson, S., Maekawa, K. & Matsushima, H. Regulation of NGF responsiveness in human neuroblastoma. Oncogene 17, 2367-2376 (1998).

23 Matsushima, H. & Bogenmann, E. Expression of trkA cDNA in neuroblastomas mediates differentiation in vitro and in vivo. Molecular and cellular biology 13, 7447-7456 (1993).

24 Woo, C. W., Lucarelli, E. & Thiele, C. J. NGF activation of TrkA decreases N-myc expression via MAPK path leading to a decrease in neuroblastoma cell number. Oncogene 23, 1522-1530 (2004).

25 Fey, D., Croucher, D. R., Kolch, W. & Kholodenko, B. N. Crosstalk and signaling switches in mitogen-activated protein kinase cascades. Frontiers in physiology 3, 355 (2012).

26 Kapeli, K. & Hurlin, P. J. Differential regulation of N-Myc and c-Myc synthesis, degradation, and transcriptional activity by the Ras/mitogen-activated protein kinase pathway. The Journal of biological chemistry 286, 38498-38508 (2011).

27 Sears, R. et al. Multiple Ras-dependent phosphorylation pathways regulate Myc protein stability. Genes & development 14, 2501-2514 (2000).

28 Strieder, V. & Lutz, W. Regulation of N-myc expression in development and disease. Cancer Lett 180, 107-119 (2002).

29 Vervoorts, J., Luscher-Firzlaff, J. & Luscher, B. The ins and outs of MYC regulation by posttranslational mechanisms. The Journal of biological chemistry 281, 34725-34729 (2006).

30 Alarcon-Vargas, D. & Ronai, Z. c-Jun-NH2 kinase (JNK) contributes to the regulation of c-Myc protein stability. The Journal of biological chemistry 279, 5008-5016 (2004).

31 Noguchi, K. et al. Regulation of c-Myc through phosphorylation at Ser-62 and Ser-71 by c-Jun N-terminal kinase. The Journal of biological chemistry 274, 32580-32587 (1999).

32 Locke, J. C. et al. Extension of a genetic network model by iterative experimentation and mathematical analysis. Molecular systems biology 1, 2005 0013 (2005).

33 Heukamp, L. C. et al. Targeted expression of mutated ALK induces neuroblastoma in transgenic mice. Science translational medicine 4, 141ra191 (2012).

34 Mologni, L. Inhibitors of the anaplastic lymphoma kinase. Expert opinion on investigational drugs 21, 985-994 (2012).

35 Surdez, D. et al. Targeting the EWSR1-FLI1 oncogene-induced protein kinase PKC-beta abolishes ewing sarcoma growth. Cancer Res 72, 4494-4503 (2012).

36 Ban, J. et al. EWS-FLI1 suppresses NOTCH-activated p53 in Ewing's sarcoma. Cancer Res 68, 7100-7109 (2008).

37 Frenkel-Morgenstern, M. et al. Chimeras taking shape: potential functions of proteins encoded by chimeric RNA transcripts. Genome research 22, 1231-1242 (2012).

Annex 1 - Other Dissemination Activities

David Croucher, (NUID UCD) - Poster presentation “A Systems Biology Model of Cell Fate Decisions in

TrkA Expressing Neuroblastoma” at ICSB, Mannheim, 28th August – September 1st, 2011.

Dirk Fey (UCD) - Poster presentation "Identification of adaptive signal processing for TrkA mediated cell fate

decisions in neuroblastoma" at the International Conference on Systems Biology in Mannheim, Germany 18

August - 1 September 2011

Dirk Fey (UCD) - Oral presentation "Understanding TrkA and Myc dysregulation in neuroblastoma using

dynamic modelling approaches" at Mathematical Oncology: New Challenges for Systems Biomedicine in

Erice, Italy 26 - 30 September 2011

Valeriya Dimitrova (UBERN) -Poster presentation”Novel c-MYC target genes in medulloblastoma” at the day

of Clinical Research in Bern 1 November- 2 November 2011

Paulina Cwiek (UBERN) - Poster presentation”Identification and validation of novel drug targets which can

be applied to the treatment of embryonal tumors ” at the Day of Clinical Research in Bern 1

November- 2 November 2011

David Duffy (NUID UCD) – Poster presentation “MYCN Transcriptional Regulation in an Embryonal Tumour”

at the Irish Network of Developmental Biology Meeting on 12th Dec 2011.

Olivier Delattre (CURIE) – Oral Presentation “Ewing sarcoma, molecular and cellular aspects” at the

NIH/NCI, Pediatric Oncology Branch Bethesda, MD USA – December 2011.

David Duffy (NUID UCD) – Poster presentation “Apoptosis at the Organismal Level: programmed cell death

in metamorphosis and regeneration” at GATSBY on 19th January 2012.

Valeriya Dimitrova (UBERN) - Oral presentation”Novel c-MYC target genes in medulloblastoma” at the

SPOG Scientific Meeting in Lugano 27 January-28 January 2012

Paulina Cwiek (UBERN) - Oral presentation”Identification and validation of novel drug targets which can be

applied to the treatment of embryonal tumors ” at the SPOG Scientific Meeting in Lugano 27 January-28

January 2012

Olivier Delattre (CURIE) – Oral Presentation “Ewing sarcoma, from somatic to germline genetics and

back” at the Pædiatric cancer translational genomics conference - Scottsdale, Arizona, USA –

February 2012

Mark Girolami (UCL) – Oral presentation “Statistical inference for markov jump process models

via differential geometric Monte Carlo methods and the linear noise approximation” at the Royal Society

Signal processing and inference for the physical sciences meeting, 26th to 27th March 2012.

Olivier Delattre (CURIE) – Oral Presentation “Gene fusion detection in sarcoma by next generation

sequencing” at the 1st European Symposium of Biopathology- Paris, France – June 2012.

David Duffy (NUID UCD) – Poster presentation “Wnt signalling at the base of the Metazoa: Revealing

Cryptic Wnt Functions and their Application in Cancer Research” at EMBO 30years of Wnt from the 27th

June to 1st July 2012.

Vassilios Stathopoulos (UCL) – Oral presentation “Riemann Manifold Hybrid Monte Carlo and alternative

metrics” at the 9th AIMS International Conference on Dynamical Systems, Differential Equations and

Applications July 1st to 5th 2012.

Valeriya Dimitrova (UBERN) - Poster presentation” The phosphoinositide 3-kinase p110α_ isoform

regulates leukemia inhibitory factor receptor α expression to promote cell proliferation and survival in

Medulloblastoma” at the 7th Swiss Apoptosis Meeting in Bern 30 August-31 August 2012

Dirk Fey, (NUID UCD) - Poster presentation “Opposing positive and negative feedback loops regulate JNK

signalling dynamics upon activation by multiple stimuli” at SYSMED conference 9th to 13th September

2012.

David Duffy (NUID UCD) - Oral and Poster presentation “Omics of Mycs: Application of Omics

Technologies to Elucidate the MYCN Transcriptional Network in Neuroblastoma” at the SYSMED

conference 9th to 13th September 2012.

Heinrich Kovar (CCRI) – Oral presentation “Multilayered gene regulation in Ewing sarcoma: from

genomics to biology” at the ESF- EMBO Symposium Molecular Biology and Innovative Therapies in

Sarcomas of Childhood and Adolescence 29 September - 4 October 2012.

Olivier Delattre (CURIE) – Oral Presentation “Ewing sarcoma, the right oncogene in the appropriate genetic

and cell backgrounds” at the ESF-EMBO symposium « Molecular biology and innovative therapies in

sarcomas of childhood and adolescence » - Pultusk, Poland – September 2012.

Olivier Delattre (CURIE) – Oral Presentation “Ewing sarcoma, the right oncogene in the appropriate genetic

and cell backgrounds” at the Annual EMBO members meeting – Heidelberg, Germany – October 2012

Miguel Vazquez (CNIO) – Oral presentation “Cancer Genome Analysis Pipeline” at 2nd Conference on

Systems Biology and New Sequencing Techniques" 2nd to 4th November 2012.

Dirk Fey (NUID UCD) - Oral presentation (Hosting) "Crosstalk and signalling switches in MAPK

systems" at D-BSSE ETH Zurich 12th Nov 2012

Vassilios Stathopoulos and Mark Girolami (UCL) – Journal article “Markov chain Monte Carlo inference

for Markov jump processes via the linear noise approximation” Philosophical Transactions of the Royal

Society A 13 February 2013 vol. 371 no. 1984 20110541, DOI: 10.1098/rsta.2011

PROJECT PERIODIC REPORT - University College Dublin Annual Report 2012.pdf · Project title: Analysing and Striking the Sensitivities of Embryonal Tumours ... 3.1 Publishable summary

Documents