Integration of chemical-genetic and genetic interaction data links bioactive compounds to cellular target pathways

Parsons et al. 2003. Nature Biotechnology. 22(1):62-69

Presented by Obi Griffith

Outline

• Background• The problem• Approach• Methods• Results• Conclusions• Criticisms• Topics for

discussion

Background

• Yeast as a model organism

• Yeast genomics • Tools of yeast

genomics

Yeast as a model organism

• Studied for 100 years

• Convenient lab organism

• Stable haploid or diploid

• Unicellular but can display group characteristics

• Highly versatile transformation system

• Homologous recombination efficient

Yeast Genomics

• First eukaryotic genome to be sequenced• ~6000 annotated genes• 182 genes with significant similarity to human

disease genes. • No complete comparison between humans and

yeast yet completed but likely many more orthologous genes than this (Carroll et al, 2003).

• Many metabolic and signal transduction pathways are conserved

Tools of Yeast Genomics

• Expression profiling (microarrays, SAGE)

• Overexpression of yeast genes

• Two-hyrid analysis of yeast protein interactions

• Mass specroscopy analysis of protein complexes

• protein microarrays

• protein localization

Tools of Yeast Genomics (cont’d)

• Whole genome deletion collections Phenotypic screens Synthetic lethality

screens Haploinsufficiency

analysis Mutant gene mapping

The problem

• Determining how small organic chemicals interact with living systems

• Traditionally a very laborious process Eg biochemical or affinity purification

strategies Depend on ability to modify a test compound Affinity not always sufficient to allow

purification

The approach

• A global fitness test that reveals genes involved in mediating the response of yeast cells to a test compound

• A way to identify molecular targets without altering test compound

• Use synthetic lethal tests on a genomic scale.• Remember, synthetic lethal = lethal event

arising from ‘synthesis’ of two gene deletions or disruptions (eg. chemical inhibition)

Method

• Conduct 2 kinds of synthetic lethal tests:

• deletion collection + chemical

= chemical-genetic profile

• deletion collection + 2nd deletion

= genetic interaction profile

• Where profiles are the same the 2nd deletion is likely target of chemical

Chemical-genetic profiles

• Screened ~4700 viable yeast deletion mutants for sensitivity to 12 different chemical compounds. Eg. benomyl, a microtubule depolymerizing agent,

FK506, a calcineurin inhibitor, fluconazole, an antifungal agent that inhibits Erg11, etc…

• Confirmed interactions by serial-dilution spot assays to minimize false positives

• Assessed false-negatives by comparing results for rapamycin screen to previously published results

Genetic Interaction profiles

• First tested system with Erg11, which encodes the target of the antifungal drug fluconazole. Crossed the Erg11 mutation into the viable

deletion set. Screened double-mutant set for lethal or sick. Compared fluconazole chemical-genetic

interactions to Erg11 genetic interactions.

• Performed similar analysis with calcineurin (CNB1).

Clustering of chemical-genetic and genetic interaction profiles

• Used 2-d hierarchical clustering of a combined data set: Chemical-genetic profiles for FK506, CsA,

fluconazole, benomyl, hydroxyurea, and camptothecin

Genetic profiles for genes encoding for the target genes or their functionally related genes (57 total).

• Filtered out multidrug-resistance

Conclusions

• a powerful method of understanding pathways and targets for bioactive compounds

• A convincing proof of principle.

• Can identify target pathways for drugs that don’t interact with one specific target only.

• Adaptable to other organisms including mammals using methods like RNAi

Criticisms

• Reliance on GO annotations.• Convincing examples but no overall measure of

agreement between profile clustering and what we expect.

• false-negatives • Only detects more sensitive reactions to

compounds.• What about important interactions that do not

result in synthetic lethality?• In many cases, their method will identify target

pathway but not actual target

References

• Carroll PM, Dougherty B, Ross-Macdonald P, Browman K, FitzGerald K. 2003. Model systems in drug discovery: chemical genetics meets genomics. Pharmacol Ther. 99(2):183-220.

• Parsons AB, Brost RL, Ding H, Li Z, Zhang C, Sheikh B, Brown GW, Kane PM, Hughes TR, Boone C. 2003. Integration of chemical-genetic and genetic interaction data links bioactive compounds to cellular target pathways. Nat Biotechnol. 22(1):62-9

• Stockwell. 2003. The biological magic behind the bullet. Nat Biotechnol. 22(1):37-8

Topics for discussion

• Why don’t the two kinds of profiles match perfectly?

• Other possible applications of this approach• How could their method be incorporated or

supplemented with data from other methodologies (eg. microarray, haploinsufficiency)

• RNAi knockouts for each mouse gene to extend approach to mammals

• Others?

The First Eukaryotic Proteome Chip

• Zhu et al. (2001) demonstrate first Proteome chip.

• 6566 protein samples• Representing 5800 unique proteins

(80%)• Spotted in duplicate on nickel coated

microscope slide• GST fusion and probing with anti-GST• Tested with biotinylated Calmodulin• A highly conserved calcium binding

protein involved with many other proteins

• Detected by binding of Cy3-labelled streptavidin

• Found 39 proteins that bind to calmodulin

– 6 previously known– 6 missed because not in collection or

not successfully attached to chip• Found putative calmodulin binding

motif shared by 14 of 39 proteins

GO – Gene Ontology

• The goal of the Gene Ontology TM (GO) Consortium is to produce a controlled vocabulary that can be applied to all organisms even as knowledge of gene and protein roles in cells is accumulating and changing. GO provides three structured networks of defined terms to describe gene product attributes.

Why do the genetic interaction and chemical-genetic interaction profiles not match exactly?

• Incomplete inactivation by the chemical• Multiple gene targets for the gene• May reflect inherent differences in genetic versus

chemical mechanisms of target inhibition.• Gene deletion completely removes the target

protein from the system whereas chemical inhibition leaves a protein-chemical complex in the system that still may play some role in the cell or have unexpected effects.