Transcriptomics:A general overview

By Todd, Mark, and Tom

Intro

• Transcriptomics => RNA in a cell• Either coding or non-coding (ncRNA).

mRNA vs microRNA, siRNAAlso non-functional RNA (pseudo-genes)

Transcriptomics focuses sets on:how

wherewhenwhy

Also diagnosing :developmental stages

tissue differentialsviruses

response to stimuli

Microarray

• Used for Biological Assays– DNA Microarrays– MMChips– Protein Microarrays– Tissue Microarrays– Antibody Microarrays

DNA Microarray

• Can be used to measure– Expression levels– SNPs– Genotyping– Comparative Genome Hybridization

Basic DNA Microarray Experiment

http://en.wikipedia.org/

Labeling

Lockhart and Winzeler 2000

Probes and Targets

• Probes– Known sequence bonded to substrate

• Target– Sample obtained to wash over chip• See what and how much is hybridized

Hybridization and Wash

Hybridization and Wash

Basic DNA Microarray Experiment

http://en.wikipedia.org/

Results

Lockhart and Winzeler 2000

Tiling Array

• Genome array consisting of overlapping probes

• Finer Resolution• Better at finding RNA in the cell– mRNA• Alternative splicing• Not Polyadenylated

– miRNA

Tiling Arrays

http://en.wikipedia.org/

Tiling Array

http://en.wikipedia.org/

Microarray

Wheelan et al. 2008

Gene expression profiling predicts clinical outcome of breast cancerLaura J. van 't Veer1,2, Hongyue Dai2,3, Marc J. van de Vijver1,2, Yudong D. He3, Augustinus A. M. Hart1, Mao Mao3, Hans L. Peterse1, Karin van der Kooy1, Matthew J. Marton3, Anke T. Witteveen1, George J. Schreiber3, Ron M. Kerkhoven1, Chris Roberts3, Peter S. Linsley3, René Bernards1 and Stephen H. Friend3

Divisions of Diagnostic Oncology, Radiotherapy and Molecular Carcinogenesis and Center for Biomedical Genetics, The Netherlands Cancer Institute, 121 Plesmanlaan, 1066 CX Amsterdam, The Netherlands Rosetta Inpharmatics, 12040 115th Avenue NE, Kirkland, Washington 98034, USA These authors contributed equally to this work

Nature, January 2002

• Use DNA microarray analysis and applied supervised classification to identify a gene expression signature predictive of metastases and BRCA1 carriers.

• Authors predicted that the expression profile would outperform all currently used clinical parameters in predicting disease outcome.

• Strategy to select patients who would benefit from adjuvant therapy (chemotherapy).

Metastases – spread of cancer from one area to another; characteristic of malignant tumor cells.

Angiogenesis – process of growing new blood vessels from pre-existing vessels. A normal process in growth and development, however also a fundamental step in the transition of tumors from a dormant state to a malignant state.

Estrogen Receptor alpha (ERα) – activated by sex hormone estrogen; DNA binding transcription factor which regulates gene expression; association with cancer known from immunohistochemical data (IHC).

BRCA1 – Human gene, Breast Cancer 1; Mutations associated with significant increase in risk of breast cancer.

• Belongs to a class of genes known as tumor suppressors (DNA damage repair, transcriptional regulation).

• BRCA1 represses ERα-mediated transcription, with a reduction of BRCA1 activity results in elevated ERα-mediated transcription and enhanced cell proliferation.

98 primary breast cancers:

34 from patients who developed metastases within 5 years

44 from patients who continued to be disease-free after 5 years

18 from patients with BRCA1 germline mutations

2 from BRCA2 carriers

• Hybridizations carried out on micoarrays (synthesized by inkjet technology) containing ~ 25,000 human genes

• ~ 5,000 genes found to be significantly regulated across the group of samples

• Total RNA isolated from patients and used to derive complementary RNA (cRNA)

• A reference cRNA pool was made by pooling equal amounts of cRNA from each cancer, for use in quantification of transcript abundance (fluorescence intensity in relation to reference pool).

Two distinct groups of tumours apparent on the basis of the set of ~5,000 significant genes.

• In upper group only 34% of patients were from group developing metastases within 5 years.

• In lower group 70% of patients had progressive disease.

• Clustering detects two subgroups of cancer which differ in ER status and lymphocytic infiltration

1) The correlation coefficient of the expression of ~ 5,000 significant genes was calculated, with 231 genes determined to be significantly associated with disease outcome.

2) These 231 genes were ranked on basis of magnitude.

3) Number of genes in ‘prognosis classifier’ optimized with the optimal number of marker genes reached at 70 genes.

To identify tumours that could reliably represent either a good or poor prognosis a three-step supervised classification method was applied:

Prognosis signature with prognostic reporter genes identifying two types of disease outcome:

• above dashed line good prognosis

• below dashed line poor prognosis

Predicted correctly the actual outcome of disease for 65 out of 78 patients (83%).

To validate prognosis classifier additional set analyzed (Fig. 2C).

The functional annotation of genes provided insight into the underlying mechanisms leading to rapid metastases with the following genes significantly upregulated in the poor prognosis signiture:

• genes involved in cell cycle

• invasion and metastasis

• angiogenesis

• signal transduction

A third classification was performed to look at the expression patterns associated with ER-positive and ER-negative tumours.

ER clustering has predictive power for prognosis although it does not reach the level of significance of the prognosis classifier.

Consensus conference developed guidelines for eligibility of adjuvant chemotherapy based on histological and clinical characteristics.

Prognosis classifier selects as effectively high-risk patients who would benefit from therapy, but reduces number to receive unnecessary treatment.

• Results indicate that breast cancer prognosis can be derived from gene expression profile of primary tumor.

Conclusions:

• ER signature - can be used to decide on hormonal therapy

• BRCA1 - knowing status of can improve diagnosis of hereditary breast cancer.

• Genes overexpressed in tumors with poor prognosis profile are targets for development of new cancer drugs

Recogmendations:

MicroRNA expression profiles classify human cancers

Jun Lu1,4*, Gad Getz1*, Eric A. Miska2*†, Ezequiel Alvarez-Saavedra2, Justin Lamb1, David Peck1,Alejandro Sweet-Cordero3,4, Benjamin L. Ebert1,4, Raymond H. Mak1,4, Adolfo A. Ferrando4, James R. Downing5,Tyler Jacks2,3, H. Robert Horvitz2 & Todd R. Golub1,4,6

Nature, June 2005

Short size of microRNAs (miRNAs) and sequence similarity between miRNA family members has resulted in cross-hybridization of related miRNAs on glass-slide microarrays.

Development of bead-based flow cytometric expression profiling of miRNAs.

miRNA profiles are informative with a general down regulation of miRNA in tumors compared with normal tissue

Expression profiles of miRNA are also able to classify poorly differentiated tumors, highlighting the potential for miRNA profiling in cancer diagnosis

RNA-Seq

Lockhart and Winzeler 2000

Wang et al. 2009

RNA-Seq

• Whole Transcriptome Shotgun Sequencing– Sequencing cDNA– Using NexGen technology

• Revolutionary Tool for Transcriptomics– More precise measurements– Ability to do large scale experiments with little

starting material

RNA-Seq Experiment

Wang et al. 2009

Mapping

• Place reads onto a known genomic scaffold– Requires known genome and depends on

accuracy of the reference

http://en.wikipedia.org/

Mapping

• Create unique scaffolds– Harder algorithms with such short reads

Comparisons

Wang et al. 2009

Comparisons

Wang et al. 2009

Biases

Wang et al. 2009

Directionality

Wang et al. 2009

Coverage Versus Depth

Wang et al. 2009

New Insights

• Mapping Genes and Exon Boundries– Single Base Resolution

• Transcript Complexity– Exon Skipping

• Novel Transcription– More accurate– No cross hybridization

Transcription Levels

• Can measure Transcript levels more accurately– Confirmed with qPCR and RNA spike-in

• Can compare measurements with different cellular states and environmental conditions – Without sophistication of normalization of data

What does mRNA tell you?

Gene expression not the same as phenotypic expression

Why no line?

Reasons?1)Noise and bias of sample

2)Lag time of translation3)Post-translational control

4)RNA/Protein half life5)?

Where is the genetics?

• How do you study the transcriptome?• What are the patterns of expression telling

you?• Differences between gene expression vs gene

function (i. e. protein code vs concentration)?

1) ‘guilt by association’2) Change environment, look for patterns;

compare known phenotypic mutants (cancer)3) Add ‘controlled’ knockout (specific locations/

times/ concentrations)4)Evolution; diversity of expression across intra

and inter speices5) Add entire chromosome

Evolution model: neutral vs selection

Mouse with Down syndromeWhat happens?

What would Mendel do?

Different environments

Rhythm