Transcript Slide 1

Gene expression studies of breast tumors
with different responses to immunotherapy
Elizabeth Chun
MSc. Candidate
Jones Lab, The Genome Sciences Centre
2009. 11. 26.
Adoptive T-cell Transfer Immunotherapy
1.
Isolation of antigen-specific Tlymphocytes from a cancer patient
Ex vivo expansion and activation of
T-lymphocytes
Transfer of anti-tumor T-lymphocytes
back to the patient
2.
3.
•
Several attractive tumor antigens
e.g. Her2/neu
Low efficacy of immunotherapy
•
–
Many factors limiting immune
response
Gattinoni L. et al. (2006) Nature Reviews in Immunology. 6:383-393.
Mouse model
ACT
Cysteinerich domain
Extracellular
CR
Tyrosinekinase
domain
NOP-21
PR
CD8+ epitope
Neu+/p53- mouse
CD4+ epitope
C57BL/6J
PD
NOP-12, 23
NOP-6,17,18
NOP cell lines generated
Affymetrix MoEx-1_0-st-v1
Neu+ mouse
Mouse image from http://www.taconic.com/userassets/Images/Producs-Services/em_mod_black.jpg
Mammary tumor image from
http://www.nature.com/onc/journal/v25/n54/images/1209707
f4.jpg
SOLiD sequencing – miRNA, transcriptome
Affymetrix chip image from
http://www.molecularstation.com/molecular-biologyimages/data/508/affymetrix-microarray.jpg
Class specific DE genes
• DE genes are detected by a bio-conductor tool, siggenes, using the
Significance Analysis of Microarray (SAM) at FDR 10% or 15%
• Detection of class-specific DE genes
– the variation of gene expression between classes is greater than within the
class
– E.g. CR-specific DE genes
E. g. PR-specific DE genes
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
CR
PR
PD
4
0
0
CR
PR
CR
PD
E.g. PD-specific DE genes
PR
CR
PD
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
CR
PR
PD
0
CR
PR
PD
??? But interesting still…
4
4
PR
PD
0
CR
PR
PD
CR
PR
PD
Overlap from pair-wise comparisons and
combined classes
• Overlap of the “class-specific” gene sets found by the two-way pair-wise
comparison and the comparison against the combined classes
CR-specific
PR-specific
229
42
CR vs (PR and PD) CR vs PD
(N= 293)
(N = 1242)
PR vs PD
(N = 1466)
PD-specific
PR vs (CR and PD) PR vs PD
(N= 47)
(N = 1466)
899
PD vs (CR and PR) CR vs PD
(N= 3601)
(N = 1242)
CR vs PR
(N = 31)
CR vs PD
(N = 1242)
Class-specific pathway analysis
• Class-specific DE genes in CR and PD
– CR: N = 229
– PD: N = 889
• DAVID (KEGG, BioCarta), Ingenuity tools used
– Top pathways overlap in all three pathway databases
• Common pathways found to be involved
– Complement system: CR / PD
– Pattern recognition: CR / PD
– Stroma-related pathways: CR / PD
• Class-specific pathways
– CR-specific: TREM1 signaling; LXR/RXR activation
– PD-specific: IL-3 signaling; FcyRIIB signaling; GM-CSF signaling; Leukocyte
extravasation
• 71 genes were selected for qRT-PCR by ranking by fold-change,
involvement of > 1 pathways, found as good classifier by Predictive
Analysis of Microarray (PAM)
Comparison with the human breast tumor data
Select genes with 1-to-1 orthologous
relationship with human (N = 15K)
1300 human intrinsic breast cancer
gene set by Hu et al. (2006) (Agilent)
Collapse data from probe to gene level
• Median for probes targeting a single gene
Merge human (HG-U133A from
Rouzier et al. (2005)) and mouse
(MoEx) breast tumor expression data
• Batch correction by DWD
Filter out genes probed in both MoEx
and HG-U133 arrays (N = 8852)
866 mouse intrinsic breast cancer
gene set by Herschkowitz et al.
(2007) (Agilent)
Human
106
(1300)
Mouse
(866)
Cross-species intrinsic breast cancer
gene sets (N = 106)
82 genes common to mouse-human breast cancer
intrinsic gene sets in the merged dataset
Herschkowitz
et al. (2007)
Cluster analysis of mouse and human tumors
• Hierarchical clustering on the subset of genes common to both species
breast cancer intrinsic gene list
PD PD
Basal-like
PD
Luminal A
CR PR PR
Her2-overexp
Lum B
Lum A
ER- = 17/17 (100%)
ER+ = 11/13 (85%)
ER- = 11/12 (92%)
ER+ = 7/8 (88%)
ER+ = 28/32 (88%)
Her2- = 15/17 (88%)
Her2- = 10/13 (77%)
Her2+ = 8/12 (67%)
Her2+ = 6/8 (75%)
Her2- = 26/32 (81%)
PR- = 13/17 (76%)
PR- = 7/12 (58%)
PR- = 11/12 (92%)
PR+ = 6/8 (75%)
PR+ = 19/30 (63%)
Ongoing research
• Improve cluster analysis of mouse and human breast cancer data
• Experimental validation of pathway-specific, class-specific DE genes by RTqPCR
• miRNA analysis from SOLiD data
– Better alignment tools to account for adapter sequence
– Identification of miRNA target genes and their functional enrichment
– Correlation of target gene expression changes
• WTSS data analysis from SOLiD data
– Somatic point mutation survey of CR, PR, PD tumors
– PCR validation of the putative mutations
– Possible novel targets for tumor vaccine development
Acknowledgement
Supervisor
SOLiD Library Construction & Sequencing
•
•
•
•
•
•
Dr. Steven Jones
Microarray Analysis
• Dr. Allen Delaney
Dr. Martin Hirst
Yongjun Zhao
Thomas Zeng
Kevin Ma
Angela Tam
The Deeley Research Centre
•
•
Dr. Brad Nelson
Dr. Michele Martin
SOLiD WTSS Analysis
•
•
•
Dr. Inanc Birol
Nina Thiessen
Timothee Cezard
ABI bioinformatics support
•
Dr. Yongming Sun
LIMS & Systems team at GSC