Transcript What are Microarrays?

cDNA Microarrays and some of
their applications to Clinical
Medicine
Pascale F. Macgregor, Ph. D.
LMP 1019S
April 12, 2002
Outline of the lecture
• What are microarrays?
**Manufacturing
**Different types of microarrays
**Possible applications
**Data analysis
• Application of the microarray technology to the
study of ovarian cancer
• Combination with cytogenetics
• Laser Capture Microdissection
• Validation: tissue arrays, RT-PCR, ISH, IHC
©2002 University Health
Network/Ontario Microarray Network
What are Microarrays?
• Microarrays are simply small glass or silicon
slides upon the surface of which are arrayed
thousands of genes (usually between 50020,000)
• Via a conventional DNA hybridization process,
the level of expression/activity of those genes is
measured
• Data are read using laser-activated fluorescence
readers
• The process is “ultra-high throughput”
©2002 University Health
Network/Ontario Microarray Network
Glass slide
microarray of
thousands of
genes for
evaluation as new
biomarkers for
cancer prognostic
assessment
©2002 University Health
Network/Ontario Microarray Network
Yeast genome: 12,800 points
Diameter: 120 microns
Slide size: 170 mm
x 340 mm
©2002 University Health
Network/Ontario Microarray Network
Why analyze so many genes?
• <10% of the human genome has been studied
at the level of gene function. 40,000 odd genes
represent the pool of remaining drug targets.
• Patterns/clusters of expression are more
predictive than looking at one or two
prognostic markers.
• Increased accuracy/confidence.
• Unbiased. Empiric. Holistic. Independent of
“flawed” hypotheses
©2002 University Health
Network/Ontario Microarray Network
Microarray technologies
There are two primary microarray technologies
which are somewhat complimentary:
1. On-chip oligonucleotide synthesis: e.g.
Affymetrix® Gene Chips
2. Direct DNA deposition: “Pat Brown” approach
(oligos or gene fragments - Expressed Sequence
Tags, “ESTs”)
©2002 University Health
Network/Ontario Microarray Network
Affymetrix Gene Chip TM
• Produced by synthesizing tens of thousands
of short oligonucleotides in situ onto glass
wafers
• 16-20 nucleotides representing each gene on
the array
• Each oligo on the chip is matched with an
almost identical one, differing only by one
single base mismatch
• Comparison of target intensity between the 2
partners oligonucleotides
• Measure of the absolute level of expression
of genes
©2002 University Health
Network/Ontario Microarray Network
Affymetrix Gene Chip TM
Nature Genetics 21:2-24 1999
©2002 University Health
Network/Ontario Microarray Network
On-chip oligonucleotide synthesis
Nature Genetics 21:5-9 1999
©2002 University Health
Network/Ontario Microarray Network
Spotted cDNA microarrays
Advantages
• Lower price and flexibility
• Simultaneous comparison of two related
biological samples (tumor versus normal,
treated cells versus untreated cells)
• ESTs allow discovery of new genes
Disadvantages
• Needs sequence verification
• Measures the relative level of expression
between 2 samples
©2002 University Health
Network/Ontario Microarray Network
steel
Spotted arrays
spotting pin
chemically modified slides
384 well source
plate
1 nanolitre spots
90-120 um diameter
©2002 University Health
Network/Ontario Microarray Network
©2002 University Health
Network/Ontario Microarray Network
The 6 steps of a microarray experiment (1-3)
• 1- Manufacturing of the microarray: clone
collection acquisition (+ sequencing), PCR
amplification and insert verification, spotting,
QC.
• 2- Experimental design and choice of
reference: what to compare to what?
• 3- Target preparation (labeling) and
hybridization
©2002 University Health
Network/Ontario Microarray Network
The 6 steps of a microarray experiment (4-6)
• 4- Image acquisition (scanning) and
quantification (signal intensity to
numbers)
• 5- Database building, filtering and
normalization
• 6- Statistical analysis and data mining
©2002 University Health
Network/Ontario Microarray Network
From Macgregor and Squire, Clinical
Chemistry, 2002
©2002 University Health
Network/Ontario Microarray Network
©2002 University Health
Network/Ontario Microarray Network
©2002 University Health
Network/Ontario Microarray Network
Cy3
Cy5
©2002 University Health
Network/Ontario Microarray Network
Cy-3
Cy-5
©2002 University Health
Network/Ontario Microarray Network
Gene D
Overexpressed
in normal
tissue
Gene E
Overexpressed
in tumour
• Biomarkers
of prognosis
• Genes
affecting
Treatment
Response
©2002 University Health
Network/Ontario Microarray Network
Microarray Applications (some)
• Can identify new genes implicated in disease progression and
treatment response (90% of our genes have yet to be ascribed a
function)
• Can assess side-effects or drug reaction profiles
• Can extract prognostic information, e.g. classify tumours based on
hundreds of parameters rather than 2 or 3.
• Can detect gene copy number changes in cancer (array CGH)
• Can identify new drug targets and accelerate drug discovery and
testing
• ???
©2002 University Health
Network/Ontario Microarray Network
The challenges of microarrays
• Acquisition of high quality clinical samples,
tumor and normal tissues
• High Quality RNA
• Experimental design: what to compare to
what?
• Data analysis -1: what to do with the data?
• Data analysis -2: How do to it?
– Very large number of data points
– Size of data files
– Choice of data analysis strategy/algorithm/software
©2002 University Health
Network/Ontario Microarray Network
Experimental Design
• Choice of reference: Common (nonbiologically relevant) reference, or paired
samples?
• Number of replicates: how many are needed?
(How many are affordable...?). Are the
replicate results going to be averaged or
treated independently?
• Dye switches?
• Choice of data base: where and how to store
the data?
©2002 University Health
Network/Ontario Microarray Network
What is a “dye switch”:
– One slide with experimental sample
labeled with Cy5, and reference sample
labeled with Cy3 (“straight”).
– Replicate slide with experimental
sample labeled with Cy3, and reference
sample labeled with Cy5 (“switch”)
©2002 University Health
Network/Ontario Microarray Network
Data Pre-processing
• Filtering: background subtraction? Low
intensity spots? Saturated spots? Low
quality spots (ghosts spots, dust spots
etc).
• Filtering or flagging?
• Outliers?
©2002 University Health
Network/Ontario Microarray Network
• Data Pre-processing : Normalization
– Housekeeping genes/ control genes
– Intensity dependent (most commonly used): global
intensity or global ratio, calculates a single
normalization factor
– Intensity independent (LOWESS – Locally
Weighted Scatter plot Smoother) calculates a
function
– Global array or Sub-array
http://stat-www.berkeley.edu/users/terry
©2002 University Health
Network/Ontario Microarray Network
Microarray data analysis 101
• Scatter plots of intensities of tumor samples
versus normal samples: quick look at the
changers and overall quality of microarray
• Supervised versus unsupervised analysis
– Clustering: organization of genes that are
similar to each other and samples that are
similar to each other using clustering
algorithms
– Statistical analysis: how significant are the
results?
©2002 University Health
Network/Ontario Microarray Network
UP
log/log
scatter plot
DOWN
©2002 University Health
Network/Ontario Microarray Network
2 dimensional hierarchical (“Eisen”)
clustering (Eisen et al, PNAS (1998), 95, p. 14863)
• Unsupervised: no assumption on samples
• The algorithm successively joins gene
expression profiles to form a dendrogram
based on their pair-wise similarities.
• Two-dimensional hierarchical clustering first
reorders genes and then reorders tumors
based on similarities of gene expression
between samples.
©2002 University Health
Network/Ontario Microarray Network
Two dimensional hierarchical
(“Eisen”) Clustering
From:
Dhanasekaran et
al.
Nature, 421,
p.822.
©2002 University Health
Network/Ontario Microarray Network
Significance Analysis of Microarrays
(SAM)
• Supervised learning software
for genomic expression data mining
• Developed at Stanford University, based
on the paper of Tusher et al PNAS
(2001) 98, p. 5116.
©2002 University Health
Network/Ontario Microarray Network
What SAM does:
• SAM assigns a score to each genes on
the basis of the change in gene
expression relative to the standard
deviation of repeated measurements.
• For genes with scores above a certain
threshold (set by the user), SAM uses
permutations of the repeated
measurements to estimate the % of
genes identified by chance = the false
discovery rate (FDR).
©2002 University Health
Network/Ontario Microarray Network
SAM example: ovarian tumor versus
normal ovary
Significant: 210
Delta 0.80000
SAM Plot
Median # false significant: 6.15764
Fold Change 1.50000
12
UP
10
8
6
4
2
0
-4
-3
-2
-1
0
1
2
-2
-4
-6
DOWN
-8
-10
©2002 University Health
Network/Ontario Microarray Network
3
4
Every cell in the body has the same genetic constitution.
In cancer cells there is usually acquired aberrant DNA
Breast
Ovary
Prostate
Development of
“cancer genomics”
to better
understand the
molecular basis of
acquired genetic
change
©2002 University Health
Network/Ontario Microarray Network
TEAM WORK in ovarian cancer at the
UHN
• Clinicians / Pathologists / Basic Scientists /
Computer Scientists
• Tissue bank (tumor tissue and normal ovary
tissue) of ~ 500 samples
• New patients treated at Princess Margaret’s
Hospital for ovarian cancer (~100/ year)
• Microarray Centre / Basic Research labs
©2002 University Health
Network/Ontario Microarray Network
Epithelial Ovarian Cancer
• First cause of death by gynecological cancer
• 5th cause of death by cancer for women
• Early Diagnostic (20%): 5 year survival rate is
80%
• Late diagnostic (80%): 5 year survival rate is
10-15%
• Treatment: Carboplatin + Taxol
• Recurrence after treatment in 80% of cases
©2002 University Health
Network/Ontario Microarray Network
Urgent need for:
• Improved detection
• Better tumor classification
• Better evaluation of response to currently
used and experimental chemotherapy
• New therapeutic targets
• New treatments
©2002 University Health
Network/Ontario Microarray Network
My lab’s approach:
• Combine the powers of chromosome/DNA
analysis (cytogenetics/genomic) and RNA
expression (cDNA microarrays) to facilitate the
identification of smaller subsets of genes
pertinent to epithelial ovarian cancer.
©2002 University Health
Network/Ontario Microarray Network
Microarray study design
• All microarrays experiments were done on 19.2 K
arrays, in duplicate, with “dye switches”.
• 15 serous epithelial ovarian tumors, 7 normal ovaries
• Stratagene Universal Reference RNA was used as
an internal standard in all experiments.
• 52 successful microarrays were hybridized and
analyzed, for a total of 2 million data points.
©2002 University Health
Network/Ontario Microarray Network
Intensities scatter plot for normal sample Stratagene
Reference: Ambion normal ovary
Normalized
Normal ovary #1
versus normal ovary #2
Cy5 intensity (Ambion normal)
100000
10000
1000
100
Tumor ovary versus
normal ovary #1
10
10
100
1000
10000
100000
Cy3 intensity (Stratagene normal)
Intesities scatter plot for tumor sample OCA21B
Reference: Ambion normal ovary
Normalized
Cancer Research 2002
100000
Cy5 intensity (Ambion normal)
From Bayani et al,
10000
1000
100
10
10
100
1000
©2002 University Health
Cy3 intensity (OCA21B)
Network/Ontario Microarray Network
10000
100000
Combining supervised with unsupervised
strategies
From Macgregor and Squire, Clinical
Chemistry, 2002
©2002 University Health
Network/Ontario Microarray Network
Conclusions of microarray analysis of
ovarian samples (1)
• Normal ovary samples clustered separately
from tumor samples and early EOC
clustered as a group.
• Separation between early EOC and late
EOC less evident, but a group of late EOCs
clustered together.
©2002 University Health
Network/Ontario Microarray Network
Conclusions (2)
• Several new up-regulated genes identified in regions of
chromosomal gains
– Cis-platinum resistance related protein: Is differential
gene expression between patients a predictor of
treatment response?
©2002 University Health
Network/Ontario Microarray Network
Going further…
• Narrowing down the number of
candidate genes by combining
approaches: RNA and DNA /
chromosome analysis
• Are the genes that are overexpressed in regions of gene copy
number gains?
©2002 University Health
Network/Ontario Microarray Network
Chromosome CGH
©2002 University Health
Network/Ontario Microarray Network
SKY Method
• Uses chromosome ‘paints’ labelled with
combinations of fluorochromes so that unique
spectral signal will identify chromosome
region involved in aberration
©2002 University Health
Network/Ontario Microarray Network
CGH and SKY Analysis of Ovarian Patient 13A
(OCA13A)
From Bayani et al, Cancer Research 2002
©2002 University Health
Network/Ontario Microarray Network
Tumor heterogeneity challenge
**Some cancers, such as prostate cancer,
are multifocal / heterogeneous.
**For these tumors, “bulk” extraction of
genetic material from tumor tissue will
produce microarray results that are
“contaminated” by normal or premalignant tissue.
©2002 University Health
Network/Ontario Microarray Network
Laser Capture Microdissection (LCM)
©2002 University Health
Network/Ontario Microarray Network
Before LCM
After LCM
Microdissected PIN
(prostatic intraepithelial
neoplasia)
©2002 University Health
Network/Ontario Microarray Network
Microarray data validation
•
•
•
•
•
cDNA microarrays: one patient - 20,000 genes
Tissue arrays: one gene -1000 patients
RT-PCR
Immunohistochemistry (IHC)
In situ Hybridization (ISH)
• Cancer profiling arrays: one gene - 10
tumor/normal sample pairs for different tumors
©2002 University Health
Network/Ontario Microarray Network
Tissue Arrays
Monni et al, Seminars in Cancer Biology, (2001), 11,
p.395
©2002 University Health
Network/Ontario Microarray Network
Northern Blot, Tissue Arrays
a. Northern blot
b. Tissue array
c. IHC, anti-hepsin
antibody (1:benign2:cancer) X 100
d. X 400
Dhanasekaran et al, Nature, (2001)
421, p.822.
©2002 University Health
Network/Ontario Microarray Network
cDNA array, Northern, ISH, IHC
Mousses et al,
Cancer Research
(2002), 62, p. 1256
©2002 University Health
Network/Ontario Microarray Network
Cancer Profiling
Array (Clontech)
Wiechen et al,
American Journal
of Pathology,
(2001), 159,
p.1635
©2002 University Health
Network/Ontario Microarray Network
Acknowledgements
• Dr. Jim Woodgett, Director,UHN microarray Centre
and Head of the Division of Experimental
Therapeutics, OCI
• Dr. Jeremy Squire, Senior Scientist, OCI
• Drs J. Murphy, B. Rosen and P. Shaw, PMH/TGH
• Ms. Monique Albert, M. Sc., Dan Grisaru, M.D., Ph.
D., and Ms. Sabrina Allegro
• Mr. Jason Gonçalves, OCI
©2002 University Health
Network/Ontario Microarray Network