Investigations into Etiology of Breast, Esophageal, and Gastric Cancers: Allele-specific Gene Expression and DNA Methylation Signature

Maxwell Lee, Ph.D.

National Cancer InstituteCenter for Cancer Research

Laboratory of Population Genetics andProgram in Bioinformatics and Computational Biology

May 14, 2013

Part 1

Part 2

Large-scale analyses of allele-specific gene expression and chromatin modifications

DNA methylation signatures for tumor classification and tumor progression

Part 3

Functional characterization of a novel oncogene identified through our genomic copy number analyses

Analyzing Allele-specific Gene Expression in Large-scaleUsing Affymetrix SNP Arrays

Genomic imprinting

X chromosome inactivation

cDNA

Affymetrix SNP array

normal human fetal tissues

Allele-specific Gene Expression and Implication for Genome Wide Association Studies

Allele-specific gene expression versus genomic imprinting and X-chromosome inactivation

•quantitative difference (2-4 fold)•20%~50% of the human genes•no parental origin preference

Implication of allele-specific gene expression for genome wide association studies

•SNPs that don’t change amino acid sequence•regulatory SNPs

277 genes (46%)equal expression

326 genes (54%)> 2-fold difference

Lo et al. Genome Res. 2003

Allele-specific ChIP-on-chip Experimental Design1347 1362

DNA

input

Pol II

H3Ac

H3K4H3K9

H3K27di

H3K27tri

active

inactive

control

96 microarray data

Genetic background influences the global epigenetic state

Samples cluster by family using allele-specific chromatin-binding activity

Family 1 Family 2

Kadota et al. PLoS Genet. 2007

1347

1362

active chromatin marks

inactive chromatin

marks

Genetic background influences the global epigenetic state

Somatic Mutations Identified through RNA-seq

4 pairs of breast tumor and normal140 millions reads

reads map to genome and transcriptome

342 somatic mutations

X

Elevated Expression of Mutant Alleles in Breast Tumors

cDNA

Genomic DNAINO80B

cDNA

Genomic DNA ARID1B

Implication for identifying driver mutationsre

lati

ve m

utan

t all

ele

inte

nsit

y in

cD

NA

no

rmal

ized

to g

enom

ic D

NA

Mean = 2.2, p-value = 0.05

Elevated Expression of Mutant Alleles in Breast Tumors

Summary of Functional data for Genes That

Displayed Elevated Expression of Somatic Mutations

gene mutation description

G3BP2 S48T GTPase activating protein (SH3 domain) binding protein 2;

oncogene, sequesting TP53

INO80B P306L INO80 complex subunit B; involved in prostate cancer

ARID1B Y1345N AT rich interactive domain 1B (SWI1-like). Its homolog, ARID1A, is frequently mutated in ovarian clear cell carcinoma

OSTF1 L20P osteoclast stimulating factor 1

GPRC5A S59C G protein-coupled receptor, family C, group 5, member A;

involved in lung cancer

RYBP K217Q RING1 and YY1 binding protein; stabilizing TP53

Part 1

Part 2

Large-scale analyses of allele-specific gene expression and chromatin modifications

DNA methylation signatures for tumor classification and tumor progression

Part 3

Functional characterization of a novel oncogene identified through our genomic copy number analyses

Identification of Novel Oncogenes through Focal Amplification Analysis

161 tumors

chro

mos

ome

161 breast tumors

putative novel oncogenes

Affymetrix SNP5 array

1q

8q

traditional approach

my approach

size of focal amplification

multiple genes

1 gene

frequency of tumors with

amplificationcommon

high frequency not required but must occur in ≥

1 tumor

Focal Amplification of TBL1XR1 in Breast Tumors

((()

Tumor 1

Tumor 2

TBL1XR1-shRNA Knockdown Suppresses In Vivo Tumor Growth

tum

or v

olum

e (m

m3 )

N=10 N=10 N=14implants

Kadota et al. Cancer Res. 2009

Western Blot

In collaboration with Lalage Wakefield

9 of 10 7 of 10 1 of 14tumor incidence

Day 39

p-value = 0.013

Part 1

Part 2

Large-scale analyses of allele-specific gene expression and chromatin modifications

DNA methylation signatures for tumor classification and tumor progression

Part 3

Functional characterization of a novel oncogene identified through our genomic copy number analyses

An algorithm for methylation and expression index (MEI)

Illumina Infinium HumanMethylation27 BeadChip

Illumina HumanRef-8 v2 Expression BeadChip

Differential methylation based on IHC (positive vs. negative for ER, PR, Her2, EGFR, or CK5)

2227 methylation markers in 1162 genes

Top 3% most variable gene expression

541 genes

128 methylation markers in 65 genes

MEI: the weighted sum of the gene expression where the weights are the negative numbers of the Spearman correlations.

Polish dataset: K-M survival based on MEI

p = 0.002

Sur

viva

l Pro

babi

lity

Year

Polish dataset: K-M survival using MEI for ER+ and ER- samples

Sur

viva

l Pro

babi

lity

p = 0.009 p = 0.360

Sur

viva

l Pro

babi

lity

ER+ cases ER- cases

Year Year

Validation: K-M survival using MEI for ER+ samples

TCGA ER+ GSE6532 ER+

NKI ER+ BT2000 ER+

Year Year

Year Year

Sur

viva

l Pro

bab

ility

OS

OS

DM

FS

p = 0.004

p = 0.001 p = 0.001

p = 0.00002

Collaborators

Lee Lab Mitsutaka Kadota Howard Yang Hailong Wu Beverly Duncan Sheryl Gere Guohong Song

Wakefield Lab Misako Sato Lalage Wakefield

Buetow Lab Chunhua Yan Michael Edmonson Rich Finney Daoud Meerzaman Ken Buetow

Nan Hu Phil Taylor Alisa GoldsteinChristian Abnet Neal Freedman Sandy DawseyJonine Figueroa Mark Sherman

Junya Fukuoka

NCI/CCR

Hunter Lab Kent Hunter

NCI/DCEG

Barbara Dunn Ronald Lubet Asad UmarNCI/DCP

Toyama University

Jiuping Ji James DoroshowNCI/DCTD

Chris Obiora Charles AdisaAbia State University

Jun RenBeijing Cancer Hospital

Purdue University Sulma Mohammed

Singer Lab Dinah Singer

Hewitt Lab Stephen Hewitt