Overview of Current Tools and Approaches Used to Demonstrate Epigenetic Effects

Transcription

1 Use of Emerging Science and Technologies to Explore Epigenetic Mechanisms Underlying the Developmental Basis for Disease NAS, Washington DC, July 2009 Overview of Current Tools and Approaches Used to Demonstrate Epigenetic Effects John M. Greally MB PhD Center for Epigenomics Division of Computational Genetics, Department of Genetics Division of Hematology, Department of Medicine Albert Einstein College of Medicine, Bronx, NY

2 Tools and approaches to study epigenomic dysregulation Challenges in studying epigenomic dysregulation in human disease Many regulators of epigenomic (and transcriptional) organisation to study Current technologies available to study the epigenome What we find when we study human disease Study design issues, ideal technologies

3 Which regulator of epigenomic organisation to study Challenges in studying epigenomic dysregulation in human disease Many regulators of epigenomic (and transcriptional) organisation to study Current technologies available to study the epigenome What we find when we study human disease Study design issues, ideal technologies

4 Which regulator of epigenomic organisation to study Molecular mediators of epigenetic and transcriptional regulation Histones and variants

5 Which regulator of epigenomic organisation to study Molecular mediators of epigenetic and transcriptional regulation Nucleosomal positioning, nucleosome-free/dnase HS regions

6 Which regulator of epigenomic organisation to study Molecular mediators of epigenetic and transcriptional regulation Histone post-translational modifications

7 Which regulator of epigenomic organisation to study Molecular mediators of epigenetic and transcriptional regulation Regulatory or transcriptional enzyme complexes Chromatin looping

8 Which regulator of epigenomic organisation to study Molecular mediators of epigenetic and transcriptional regulation DNA properties Cytosine methylation (hydroxymethylation) Physical properties of DNA DNA sequence variability

9 Which regulator of epigenomic organisation to study Transcriptional relationships

10 Which assay to use Challenges in studying epigenomic dysregulation in human disease Many regulators of epigenomic (and transcriptional) organisation to study Current technologies available to study the epigenome What we find when we study human disease Study design issues, ideal technologies

11 Epigenomic assays Chromatin immunoprecipitation (ChIP) assays

12 Epigenomic assays Chromatin immunoprecipitation (ChIP) assays

13 Epigenomic assays Other nucleosome positioning assays: nucleosome-free regions

14 Epigenomic assays Other nucleosome positioning assays: nucleosome-free regions

15 Epigenomic assays Transcriptional assays Microarray or massively-parallel sequencing Choice of RNA preparations: Total RNA mrna nuclear RNA, primary transcript small RNAs Massively-parallel sequencing strategies Quantitative (relative expression levels) Qualitative (alternative splicing) Directional

16 Epigenomic assays Cytosine methylation assays

17 Epigenomic assays Epigenomic assay comparisons Assay Sample preparation issues Cell numbers required ChIP Nucleosome positioning Transcription Needs immediate sample processing Needs immediate sample processing Sample should be immediately flash-frozen or processed (10 5 -) 10 7 < (10 5 -) 10 7 < (10 4 -) 10 6 Cytosine methylation Sample should be frozen < (10 4 -) 10 6

18 Epigenomic assays Epigenomic assay comparisons Assay Quantitative capability of genome-scale assay Resolution of genomescale assay ChIP Poor Hundreds of basepairs Nucleosome positioning Poor Hundreds of basepairs Transcription Good Exon-specific Cytosine methylation Fair One to hundreds of basepairs

19 Epigenomic assays Epigenomic assay comparisons When studying human clinical samples, current genome-wide approaches almost exclusively focus on cytosine methylation and transcriptional

20 Epigenomic assays Signal/noise issues: massively-parallel sequencing beats microarrays ENCODE Technology Development NIH (NHGRI) R01 HG Co-investigators Brad Bernstein (MGH/Broad), Andi Gnirke (Broad)

21 Epigenomic assays Signal/noise issues: cytosine methylation assays CG dinucleotide density dependence

22 Epigenomic assays Cytosine methylation assays Distribution of CG dinucleotides in human genome 23 million CGs per haploid male genome CpG islands 7% Generally unmethylated Repetitive DNA 51% Generally methylated Other DNA 41%? Most current assays directed at CpG islands/promoters Readily interpretable Just looking for acquisition of methylation as abnormal outcome

23 Epigenomic assays Cytosine methylation assays No single assay does it all Genome-wide Nucleotide resolution Quantitative High sample throughput Cost efficient Three types of assays have emerged DISCOVERY Genome-wide, lower resolution COMPREHENSIVE Quantitative, nucleotide resolution POPULATION High sample throughput

24 Epigenomic assays A choice of assays available DISCOVERY Restriction enzyme-based (HELP, mcrbc) Affinity-based (medip, MIRA) Large proportion of genome, low resolution, small sample number Massively-parallel sequencing (MPS) COMPREHENSIVE Cloning and sequencing Pyrosequencing (Biotage) MassArray (Sequenom) Nucleotide resolution, very limited proportion of genome, moderate sample number Massively-parallel sequencing -- SeqCap bisulphite POPULATION MethylLight, Methylation-Specific Primer (MSP), Infinium (Illumina) Studies ~1 CG dinucleotide at a time, large sample numbers

25 Epigenomic assays A choice of assays available DISCOVERY Restriction enzyme-based (HELP, mcrbc) Affinity-based (medip, MIRA) Large proportion of genome, low resolution, small sample number Massively-parallel sequencing (MPS) COMPREHENSIVE Cloning and sequencing Pyrosequencing (Biotage) MassArray (Sequenom) Nucleotide resolution, very limited proportion of genome, moderate sample number Massively-parallel sequencing -- SeqCap bisulphite POPULATION MethylLight, Methylation-Specific Primer (MSP), Infinium (Illumina) Studies ~1 CG dinucleotide at a time, large sample numbers

26 Epigenomic assays Discovery platforms and methylation states Methylation at a CG dinucleotide predictive of nearby CGs The 1.32 million locus human microarray HELP platform flags two-thirds of CGs in genome Eckhardt F, Lewin J, Cortese R, Rakyan VK, Attwood J, Burger M, Burton J, Cox TV, Davies R, Down TA, Haefliger C, Horton R, Howe K, Jackson DK, Kunde J, Koenig C, Liddle J, Niblett D, Otto T, Pettett R, Seemann S, Thompson C, West T, Rogers J, Olek A, Berlin K, Beck S. DNA methylation profiling of human chromosomes 6, 20 and 22. Nat Genet Dec;38(12):

27 Epigenomic assays HELP assay (version 1) Khulan B., Thompson R.F., Ye K., Fazzari M.J., Suzuki M., Stasiek E., Figueroa M.E., Glass J.L., Chen Q., Montagna C., Hatchwell E., Selzer R.R., Richmond T.A., Green R.D., Melnick A., Greally J.M. Comparative isoschizomer profiling of cytosine methylation: The HELP assay. Genome Research 2006 Aug;16(8):

28 Epigenomic assays Supporting analytical tools essential Hypomethylated Methylated

29 Epigenomic assays High-resolution HELP (version 2) Improved representations Multiple species Human > 1.32 million loci Mouse > 1.01 million loci Rat > 1.10 million loci Cow > 1.10 million loci

30 Epigenomic assays New applications: nanohelp Amit Verma lab (Einstein) 5,000 cell genome equivalents

31 Epigenomic assays New assays: HELP-tagging (version 4) Used Illumina platform for these data

32 Epigenomic assays HELP-tagging

33 Epigenomic assays HELP-tagging

34 Epigenomic assays HELP-tagging Similar assay to MSCC Relatively quantitative

35 Which assay to use Challenges in studying epigenomic dysregulation in human disease Many regulators of epigenomic (and transcriptional) organisation to study Current technologies available to study the epigenome What we find when we study human disease Study design issues, ideal technologies

36 Genomic distribution of epigenomic dysregulation Disease studies Epigenome-wide association studies (EWAS) Diseases: Cancers (oesophageal, breast) [human] Intrauterine growth restriction [rat] High-resolution microarray-based HELP assays Bisulphite MassArray validation

37 Epigenomic dysregulation in human disease Epigenomic dysregulation in cancer Amit Verma, Einstein

38 Genomic distribution of epigenomic dysregulation Human breast adenocarcinoma Pilot study: selected 10 pairs of tumour and grossly-normal adjacent tissue HELP assays performed on all Focus on genomics of changes Where in the genome are the most informative changes? Current beliefs suggest CpG islands should acquire methylation, transposable elements should lose methylation.

39 Genomic distribution of epigenomic dysregulation Human breast adenocarcinoma Overall results Top 1% of loci Changes from methylated to hypomethylated Changes from hypomethylated to methylated Repetitive element 72% 38% Intergenic DNA 24% 33% CG density 2.8% CG cluster 1.6% CpG island 24% CG cluster 12% CpG island Gene structure 0.6% promoter 5.3% gene body 7% promoter 26% gene body

40 Genomic distribution of epigenomic dysregulation Human breast adenocarcinoma Relative to genes Most of the consistent changes seen in these tumours are not located within promoters.

41 Genomic distribution of epigenomic dysregulation Human breast adenocarcinoma CG density The CG depleted regions of the genome are more informative than the CG rich

42 Genomic distribution of epigenomic dysregulation Human breast adenocarcinoma While data confirm prior observations of dysregulation of the cancer epigenome... Loss of methylation of transposable elements Acquisition of methylation by CG-dense sequences...the genome-wide approach makes new observations Intergenic regions often dysregulated CG clusters a better annotation than CpG islands when looking for targets of dysregulation The CG-depleted majority of the genome contains most of the informative changes

43 Genomic distribution of epigenomic dysregulation Rat intrauterine growth restriction (IUGR) Risk of adult type 2 diabetes mellitus in later life Does the epigenome mediate the cellular memory of prenatal event?

44 Genomic distribution of epigenomic dysregulation Rat intrauterine growth restriction (IUGR)

45 Epigenomic dysregulation in human disease Rat intrauterine growth restriction (IUGR) HELP tracks at Gch1 show non-promoter changes

46 Epigenomic dysregulation in human disease Rat intrauterine growth restriction (IUGR) Validated a conserved intergenic locus 45 kb upstream of Gch1

47 Epigenomic dysregulation in human disease Rat intrauterine growth restriction (IUGR) Bisulphite MassArray analysis of conserved 45 kb upstream element Significant changes of magnitude 20-30%

48 Epigenomic dysregulation in human disease Rat intrauterine growth restriction (IUGR) Bisulphite MassArray analysis of conserved 45 kb upstream element Significant changes of magnitude 20-30% Non-promoter cytosine methylation changes associated with local gene expression changes

49 Epigenomic dysregulation in human disease How to interpret small differences in epigenetic regulatory patterns In non-cancer conditions, degrees of difference are small and at restricted numbers of loci

50 Epigenomic dysregulation in human disease How to interpret small differences in epigenetic regulatory patterns Tissue-specific differences of comparable magnitude

51 Epigenomic dysregulation in human disease How to interpret small differences in epigenetic regulatory patterns If fixed, must be occurring in cellular subpopulations

52 Epigenomic dysregulation in human disease How to interpret small differences in epigenetic regulatory patterns If not fixed, may be oscillating in all cells

53 Epigenomic dysregulation in human disease How to interpret small differences in epigenetic regulatory patterns May be affecting functionally-specialised subpopulations of cells

54 Study design in epigenome-wide association studies Challenges in studying epigenomic dysregulation in human disease Many regulators of epigenomic (and transcriptional) organisation to study Current technologies available to study the epigenome What we find when we study human disease Study design issues, ideal technologies

55 Study design in epigenomics projects Epigenomics and human disease Technologies have to have two characteristics Quantitative Genome-wide Cohorts studied Well-characterised Adequate sample size for power Cell samples Pure (>90%) Well-characterised Two stage design Genome-wide, identify candidates Single-locus, quantitative

56 Study design in epigenome-wide association studies ChIP-based and nucleosomal positioning assays Probably insufficiently quantitative to detect moderate changes Shahin Rafii, Weill-Cornell Aaron Goldberg, David Allis, Rockefeller

57 Study design in epigenome-wide association studies HELP-tagging Similar assay to MSCC Relatively quantitative 100/80%80/60%60/40%40/20%20/0%

58 Study design in epigenome-wide association studies SNPs Especially a problem in cytosine methylation studies Deamination of methylated CG dinucleotides Need to sequence native as well as bisulphite-converted sequence Confound MassArray and restriction enzyme-based assays ACATCCACGTmeCGAC TT Bisulphite conversion AUATUUAUGT mecgau TT PCR ATATTTATGTCGATT T

59 Study design in epigenome-wide association studies Power calculations Number of subjects needed to identify real differences in cytosine methylation Using Church's regression equation (methylation = *counts), we may simulate individual counts based on control samples having 60% methylation and cases having 80% methylation using a poisson distribution with a count of 3.6 and 1.76 respectively. Power is defined as the Pr(Reject the null of no difference in counts between cases and controls) under differential methylation (80 vs. 60 percent). A p value cutoff of 0.05 represents the conventional threshold, without adjustment for multiple testing. A p value cutoff of 5e 8 represents a conservative (Bonferroni like) adjustment assuming 1 million independent tests are performed. Based on the simulated data described above, sample sizes of per group are necessary to have Melissa Fazzari sufficient (Einstein) power given the subtle differences in methylation hypothesized as well as the large number of tests performed.

60 Studying epigenomic dysregulation in human disease Epigenomic dysregulation studies at Einstein Center for Epigenomics Cancer epigenomics Neuroepigenomics Epigenomics of infectious disease Epigenomics of ageing

61 Acknowledgements Lab personnel Masako Suzuki Niru Narayanan Reid Thompson Edyta Stasiek Marién Pascual Not pictured: Niki Athanasiadou Maria-Paz Ramos Kevin Lau Former lab personnel Mayumi Oda Khulan Batbayar Jacob Glass Priti Tewari Einstein Center for Epigenomics investigators R. Suzanne Zukin Kami Kim Eric E. Bouhassira Simon Spivack Amit Verma Nir Barzilai Francine Einstein Einstein Epigenomics Shared Facility Shahina Maqbool Raul Olea Gael Westby Einstein Computational Epigenomics group Andy McLellan A.J. Jing Rob Dubin Collaborators Cornell University Ari Melnick Grenoble Saadi Khochbin Roche NimbleGen Systems Inc. Jeff Jeddeloh National University of Ireland, Galway Aaron Golden University of Tokyo Kunio Shiota MGH/Broad Brad Bernstein Andi Gnirke Einstein Computational and Statistical Epigenomics Group Melissa J. Fazzari Deyou Zheng Grant support NIH (NCI) R21 CA NIH (NICHD) R01 HD NIH (NHGRI) R01 HG High Q Foundation