Patterns in expression profiles point to mode of action in drug discovery © 2001,Pharmacia, Inc. - All Rights Reserved.
Antifungal therapy: Opportunistic systemic infections:candidiasis and aspergillosis Candida albicans is 4th most-frequently infectious hospital isolate Nosocomial fungal infections affect > 2 million patient / year Available therapies: Polyenes (Amphotericin B): effective but with side effects Azoles (Fluconazole, Itraconazole): safe but less effective Candins (Cancidas): just approved Model organism: Saccharomyces cerevisiae, aka baker’s yeast
Characterization of novel agents Have we seen this type of agent before? What biological processes does it impact? Are improvements making it better or different? How can we measure activity? Drug discovery objectives:
Transcript profiles with Affymetrix microarrays Microarray profiles within an experimental class Relationships between experiments Identification of functional patterns Relationships among responsive genes Topics:
Transcript profiles: Transcript profile = snapshot of all mRNA species in sample Yeast:Profile =>unstressed, normal growth Profile => response to agent ? target pathway ? “secondary” response ? surrogate expression marker Stress:
Affymetrix Gene Chip Hybridization: Result: intensity value for each mRNA represented on chip
Measures of [mRNA] agree: log(Chip Intensity) ketoconazole itraconazole clotrimazole PNU E amorolfine fluconazole voriconazole terbinafine untreated ketoconazole itraconazole clotrimazole PNU E amorolfine fluconazole voriconazole terbinafine untreated PCR cycle number GeneChip ~ [mRNA] log (Chip Intensity) Taqman ~ 1 / [mRNA] (PCR cycle number)
Experimental design: Agents XExposure= Treatment
Aaaa aaaa genes TreatmentsTreatment profiles: Measure of similarity between biological response to different treatments Treatment profiles reveal similarity in response: Signal from chip
Identify common biological response:
Distinguish a compound with distinct effect:
Correlation: Pairwise Pearson correlation coefficient between each pair of treatments
Correlations between experiments:
Biological effects, proof of concept: 8 chemical agents: ergosterol ERG9 ERG1 ERG7 ERG11 ERG24 ERG25 ERG26 ERG6 ERG2 ERG3 ERG5 ERG24 allylamine morpholine 5 azoles Novel imidizole PNU E 3 genetic changes: ERG6 ERG2 ERG5 Define method to identify responsive transcripts farnesyl pyrophosphate
Expressed above background Significantly changed from untreated Changed in multiple related treatments Responsive genes: X X X
Responsive genes in blue 112 transcripts related to ergosterol 59 genes of unknown function 52 “other” changed transcripts AcylCoA-> -> ERG19 -> farnesyl pyrophosphate ERG9 ERG1 ERG7 ERG11, NCP1 ERG24 ERG25/ERG26 ERG6 ERG2 ERG3 ERG5 ERG4 ergosterol Response to ergosterol perturbation:
5 20 stress- response 36 mito 16 membrane -assoc. 5 vesicular transport 13 heme- responsive 29 lipid, fatty-acid sterol associated ergosterol plasma membrane plasma membrane inner mito membrane inner mito membrane Erg11p contains heme 12 hypoxic 5 cell wall Facets of response:
Signal transduction Major Facilitator Superfamily Transporters Protein processing Translation Lipid, sterol and fatty acid, biosynthesis Cell wall biosynthesis Stress responses Amino acid metabolism Carbohydrate metabolism Mitochondrial:metabolism energy production translation Nucleoside, etc. metabolism RNA synthesis and processing Structure DNA synthesis / repair / recombination
Global patterns of responsive genes: Each row is histogram of responsive genes in given treatment
Second dimension -- gene profiles:
Aaaa aaaa genes treatments => biological similarity => gene families co-regulated in response to treatment Gene profiles: Treatment profiles: Treatment profiles / gene profiles
profiles Treatment profiles / gene profiles Treatments Genes
Gene correlations Pairwise Pearson correlation coefficients for gene profiles
Conclusions: Expression profiles: identify responsive genes find significant pathway(s) in response find unanticipated responsive pathways identify surrogate expression markers identify agents eliciting similar responses distinguish biological response to apparently similar agents
Acknowledgements: Pharmacia: Gary Bammert, ID Genomics Chad Storer, ID Genomics Mark Johnson, Computer-Aided Drug Discovery Tom Vidmar, Biostatistics Affymetrix:Mike Lelivelt Proteome:Everyone supporting YPD Spotfire:Bill Ladd Shawn Kenner