EXPLORING THE NEUROBLASTOMA EPIGENOME: PERSPECTIVES FOR THE DISCOVERY OF PROGNOSISTIC BIOMARKERS M. Ongenaert, A. Decock, J. Vandesompele, F. Speleman Center for Medical Genetics, Ghent University, Ghent, Belgium (mate.ongenaert@ugent.be) Neuroblastoma (NB) is a childhood tumor originating from sympathetic nervous system cells. Although recently new insights into genes involved in NB have emerged, the molecular basis of neuroblastoma development and progression still remains poorly understood. The best-characterized genetic alterations include amplification of the proto-oncogene MYCN, ALK activating mutations or amplification, gain of chromosome arm 17q and losses of 1p, 3p, and 11q. Epigenetic alterations have been described as well: caspase-8 (CASP8) and RAS-association domain family 1 isoform A (RASSF1A) DNA-methylation are important events for the development and progression of neuroblastoma. In total, there are about 75 genes described as epigenetically affected in NB cell lines and/or NB primary samples. Most of these methylation markers are found using ‘candidate gene’ approaches and the methylation frequencies are usually very low. In order to find novel methylation markers that can be used for improved prognosis, we applied a whole-genome methylation screen. This technique relies on capturing with the MBD2 protein, containing a methyl-binding domain (MBD), with a very high affinity towards methylated genomic regions. In an initial phase, MBD2-seq was performed on 8 NB cell lines (where we also had micro-array data of, before and after treatment with DAC). As these results are promising, we will explore the complete methylomes of 45 primary NB tumors. Based on an integrative analysis (re-expression results, expression micro-arrays, MBD2-sequencing on cell lines), 48 MSP (Methylation Specific PCR) assays were tested on 89 primary neuroblastoma patients of different risk categories. The results of this validation study demonstrate the power of epigenetic biomarkers as several assays are informative for prognosis and survival.

1. Exploring the neuroblastoma epigenome:
perspectives for the discovery of prognostic biomarkers
Cytometry 2011, Paris
26/10/2011
Maté Ongenaert
Center for Medical Genetics
Ghent University Hospital, Belgium

2. Overview
 Epigenetics - introduction
 Introduction
 DNA-methylation and histone modifications
 The interplay between epigenetics
 Applications of epigentics
 Mapping the neuroblastoma epigenome
 Sequencing the neuroblastoma epigenome
 Integrated data analysis
 Real-time methylation-assays for improved
prognosis

3. Epigenetics > Introduction
 -genetics
 Heritable changes to the DNA or histones without
affecting the DNA sequence
 A whole range of changes are described
• DNA-methylation
• Histone tail modifications
– Methylation
– Acetylation
– Phosphorylation
– ….
 Epigenetic changes are interconnected

4. Epigenetics > Introduction

5. Epigenetics > Introduction
DNA-methylation
Histone tail modifications

6. Epigenetics > DNA-methylation
 Isolated CG dinucleotides are in most
cases methylated
 Some regions are CG-rich: “CpG islands”
 More than half of the promoter regions have a
CpG island
 Are not methylated in most cases

7. Epigenetics > DNA-methylation
 Is a normal phenomenon
 Development, differentiation
• Genes, active only during specific stages of the embryonic development
 Genomic imprinting
• Only one of the parental copies is active
 Silencing large chromosomal domains, e.g. X-
chromosome
• Mosaic X-chromosome in females
 Protection against intra-genomic parasites:
retrotransposons and other junk in the genome

8. Epigenetics > DNA-methylation
 Is a normal phenomenon
 Development, differentiation
• Genes, active only during specific stages of the embryonic development

9. Epigenetics > DNA-methylation
 DNA-methylation and cancer
Local Global
hypermethylation hypomethylation

10. Epigenetics > DNA-methylation
 Dense methylation in promoter regions
causes transcriptional silencing
 Blocked binding of transcription machinery
(physical blockage)
 In reality, shows to be more complex
 Link between DNA methylation and histone
modifications

11. Epigenetics > Histone modifications
 Histone modifications
 Acetylation
• Activating (e.g. H3K9, H3K14, H3K18, H3K56)
 Methylation
• Repressing (H3K9 – H3K27 – H4K20)
• Activating (H3K4 – H3K36 – H3K79)
 Phosphorylation
• Activating (H3S10)
 Ubiquitinilation
• Repressing (H2A K119)
• Activating (H2B K123)
 Sumoylation

12. Epigenetics > Interplay
 Interplay between DNA-methylation and
histone modifications

13. Epigenetics > Detection
 Detection of DNA-methylation

14. Epigenetics > Detection
 Bisulfite conversion – MSP
 Primerdesign in CG regions

15. Epigenetics > Detection
 Detection of histone modifications
 Affinity-based
• Antibodies against modifications
• Enrichment of fragments, bound by antibody
 Platforms:
• Chip (ChIP-chip)
• Seq (ChIP-seq)

16. Epigenetics > Detection / Prognosis / Prediction
 Detection / Prognosis / Prediction
 Require ‘biomarkers’
• Easy to detect using molecular techniques
• Often an ‘early event’
• Suitable biomarker for detection / screening
• Can be detected in blood, urine, sputum (non-invasive sampling)
• Biomarkers in various cancer types
 Beyond tumor detection
• Stratification of patient groups
• Stage/grading classification
• Prognosis (survival, disease-free survival >)
• Chemotherapy respons (MGMT in brain cancer – temozolomide >)
• Personalized medicine

17. Epigenetics > Detection / Prognosis / Prediction
 Detection / Prognosis / Prediction
 DNA-methylation is not random
 Can be more frequent than mutations

18. Epigenetics > Detection / Prognosis / Prediction
 Detection / Prognosis / Prediction
 Dependent on environmental factors
 Methylation profile of twins (genetically identical)

19. Epigenetics > Detection / Prognosis / Prediction
 (Early) detection – diagnostic
 Diagnostic: who
 Screening programs

20. Epigenetics > Detection / Prognosis / Prediction
 Prognosis
 Prognostic: who
 Follow-up

21. Epigenetics > Detection / Prognosis / Prediction
 Prognosis Biomarker – bad progn.
 Prognostic: who
 Follow-up Biomarker – good progn.

22. Epigenetics > Detection / Prognosis / Prediction
 Prognosis
 Survival in colorectal cancer

23. Epigenetics > Detection / Prognosis / Prediction
 Prediction
 Predictive: what
 Treatment

24. Epigenetics > Detection / Prognosis / Prediction
 Prediction
 Predictive: what
 Treatment

25. Epigenetics > Detection / Prognosis / Prediction
 Prediction Biomarker
 Predictive: what
 Treatment

26. Epigenetics > Detection / Prognosis / Prediction
 Prediction
 Predictive: what
 Treatment

27. Epigenetics > Detection / Prognosis / Prediction
 Prediction
 Chemotherapy respons (MGMT in brain cancer -
temozolomide)

28. Overview
 Epigenetics - introduction
 Introduction
 DNA-methylation and histone modifications
 The interplay between epigenetics
 Applications of epigentics
 Mapping the neuroblastoma epigenome
 Sequencing the neuroblastoma epigenome
 Integrated data analysis
 Real-time methylation-assays for improved
prognosis

29. Mapping the neuroblastoma epigenome
 Neuroblastoma

30. Mapping the neuroblastoma epigenome
 Neuroblastoma

31. Mapping the neuroblastoma epigenome
 Neuroblastoma
Risk factors:
- Age at diagnosis
- MYCN amplification
- INSS Stage
Under- and overtreatment
- Molecular biomarkers - 59 gene signature (qPCR)
- Methylation biomarkers for improved prognosis

32. Mapping the neuroblastoma epigenome
 Neuroblastoma
 8 neuroblastoma cell lines
• CHP902R, CLBGA, IMR32, LAN2, N206, SHSY5Y, SJNB1, SKNAS
 Re-activation after DAC-treatment
• Expression: micro-array Affymetrix HGU-133plus2.0
 Sequencing
• Capture with MBD2 antibody – MBD2 has the highest affinity towards
methylated DNA
• Multiplex library preparation (MID tags to identify sample) and
sequencing
• Illumina GAIIx, paired-end sequencing (2x45bp)

33. Mapping the neuroblastoma epigenome
 Sequencing
Control of fragment sizes with high sensitivity DNA chips
Concentration determination of the fragmented DNA with Fluostar Optima plate reader
MBD2 immunoprecipitation reaction (MethylCollector Kit)

34. Mapping the neuroblastoma epigenome
 Sequencing data analysis
 Mapping on the human reference genome
• Input: 45 bp ‘sequence tags’
• Output: mapped sequence reads: chromosome-location
• Coverage: number of tags at a certain genomic location
• In this case: coverage ~ captured DNA fragments ~ MBD2 binding ~
methylation
 Peak detection
• Compared to the ‘background’, how unusual is the signal I see in a specific
region
 Peak annotation
• Genomic location > Genes / functions / …
 Visualisation

35. Mapping the neuroblastoma epigenome
 Sequencing data analysis

36. Mapping the neuroblastoma epigenome
 Sequencing data analysis

37. Mapping the neuroblastoma epigenome
 Sequencing data analysis

38. Mapping the neuroblastoma epigenome
 Neuroblastoma methylation
 Combine several data sources to filter the most
relevant biomarkers out: integrated data analysis
 Very specific for biological question
 Purpose: prognostic DNA-methylation biomarker
(risk groups)
 Expression results
• High stage vs. low stage
• MYCN amplified vs. MYCN non-amplified
• High risk vs. low risk
 Re-expression results
 Methylation capture results

39. Mapping the neuroblastoma epigenome
 Integrated data analysis
 Expression results
• Public expression data from a total of 380 primary NB samples
• Three different arrays, two different platforms
• Uniform scoring scheme (RankProd analysis - ranking statistics)
 Re-expression results
• Expression array before and after DAC treatment, 8 NB cell lines
• Score assigned (RankProd)
 Methylation capture results
• MBD2 sequencing in 8 NB cell lines
• Score assigned (TSS, p-value peaks, tags/length, FE)

40. Mapping the neuroblastoma epigenome
 Integrated data analysis (PCDHB-cluster)

41. Mapping the neuroblastoma epigenome
 Integrated data analysis

42. Mapping the neuroblastoma epigenome

43. Mapping the neuroblastoma epigenome
 Samples
 89 primary NB samples, 3 prognostic groups
Characteristic Classes Count (percentage)
HR_DOD 28/89 (31%)
Risk Classification HR_Surv 30/89 (34%)
LR 31/89 (35%)
1 21/89 (24%)
2 12/89 (14%)
INSS Stage
3 17/89 (19%)
4 39/89 (44%)
Not amplified (0) 50/89 (56%)
MYCN
Amplified (1) 39/89 (44%)
Age at diagnosis > 12 months
53/89 (60%)
(0)
Age
Age at diagnosis < 12 months
36/89 (40%)
(1)

44. Mapping the neuroblastoma epigenome
 MSP detection
 Roche LightCycler 480 Instrument: Cq
 LC480 melting curve analysis: Tm
 Caliper LabChip GX: Size
 Cq, Tm and size  methylation call (compared to
HCT-116 – SssI– methylated control)
 Controls
 HCT-116 / SssI: methylated control
 HCT-116 / DKO: unmethylated control
 Control primers: ACTB

45. Mapping the neuroblastoma epigenome
 MSP detection
 48 assays
 96 samples
 Total: 4608 MSP reactions
 12 384-well plates
 Pipetting: Tecan robot