Applied Genomics: Of what use is genomics?
Most applications are directed towards improving human health
• Drug discovery; new drug targets • Determining mechanisms of drug actions • Diagnosis and classification of disease type for specific treatments • Identifying and mapping disease genes • Virulence properties of pathogens • Targets for vaccine development • Improvement of agricultural products • Pharmacogenomics
Pharmacogenomics: Inherited differences in drug response Pharmacogenomics:
The ultimate goal of pharmacogenomics is to profile the variation between individuals’ genotypes to predict particular responses to a drug and to prescribe proper amounts and types of medicines. Pharmacogenomic catch-phrases
“The right drug for the right person” “The right dose for the right patient” “Predictive medicine” “Personalized medicine” “Individualized medicine” “Designer medicine” Pharmacogenomics emphasizes correlations between of patient genotype and drug response in order to increase the chance that a drug will work without side effects
1
Pharmacogenomics: Inherited differences in drug response Pharmacogenomics:
To date, there are about 200 known pharmacogenomic differences
• Most disorders are caused by defective enzyme or receptor • Differences may be more than 10-fold among individuals • Differences can be 2 to 3-fold between ethnic / geographic groups
Uses forward genetics to discover the genetic basis of the phenotype
Association of known SNPs with certain drug response
SNPs may be linked to, or the actual mutation causing, the disease
How much variation exists between individuals?
Humans, on average, differ at 1-2 million of 3.3 billion basepairs
Differences are attributable to: (1) SNPs (which you have heard about) (2) Short Indels (causing frameshifts); (3) Repeat Number (microsatellites, trinucleotides)
dbSNP started in 1998, and by 2001, there were already 1.5 million SNPs based on five million sequence reads from 24 individuals of diverse ethnicity and this was predicted to increase ten-fold in 5 years.
Unique in database
To be considered a polymorphism (vs. a rare variant), a SNP must have a frequency of at least 1%.
There are 10-20 SNPs per CDS & perhaps half alter gene function
2
SNPs and other polymorphisms may be linked to, or may be the cause of, a particular phenotype
Relative frequencies of types of mutations underlying disease phenotypes _________________________________________________________ Change
Deletion Insertion/duplication Complex rearrangement Repeat variation Missense/nonsense Splicing Regulatory
Number
6085 1911 512 38 16441 2727 213
% of Total
21.8 6.8 1.8 0.1 58.9 9.8 0.8
Total
27027
_________________________________________________________
SNPs give rise to many (~100) haplotypes per gene and knowledge of precise haplotypes may allow prediction of phenotype
3
Most pharmacogenomic differences are covert i.e., not apparent from phenotype
The response to a drug is usually not known prior to being administered, and on average, 10% of population has adverse response. Therefore, the goal of pharmacogenomics to monitor individual genotypes, which will aid physicians in selecting the optimal dose and the best drug.
The first recognition of pharmacogenomic variation was observed during WWII African-American soldiers given anti-malarial drugs (primaquine) were more likely to develop anemia & haemolytic leukemia. In 1956, discovered that mutations in G6PDH (essential to maintain integrity of red blood cells) occurred in about 10% of Americans of African descent.
Pharmacogenomic polymorphisms in P450 cytochromes
P450 cytochromes are enzymatic heme proteins (found in liver tissue) that intervene in endogenous biosynthetic pathways, in oxidative drug metabolism, and in the inactivation of carcinogens.
CYP, family number, subfamily letter, number for isoenzyme, number after * is the allele
4
Allelic variation in human cytochromes
CYP2D6 variation discovered serendipitously when a researcher took debrisoquine for hypertension and could not oxidize drug (frequency of 3-10% Caucasians; 1% Japanese)
The ‘poor metabolizer’ phenotype also affects the metabolism (hydroxylation) of 40+ drugs (codeine, ß-blockers, tricyclics) and environmental chemicals. In contrast, enhanced metabolizer (‘UM’ phenotype is at 10% in Spain but at 1% in Sweden) is associated with risk of lung, liver, and bladder cancers
Efficacy pharmacogenomics of an obesity drug
(from A.D. Roses, Nat. Rev. Gen. 2004)
Heterozygous
Homozygous
(for 3 SNP alleles)
Homozygous
(for alternate SNP alleles)
Hyper-responders
(2-month trial)
In this study, 21 candidate genes (112 SNPs) that represent potential polymorphic sites in the mechanism of drug action were tested. A total of 3 SNPs segregated with the variation in drug response (2 in candidate genes related to proposed mechanism of drug action, 1 in an ‘obesity’ gene).
5
• ADH2 (converts alcohol to acetaldehyde) and ALDH2 (converts acetaldehyde to acetate) lead to differences in alcohol metabolism (and possibly addiction). • PON1 (paraoxonase) polymorphisms show differences in protection against cardiovascular diseases and toxicity to pesticides. • EPHX1 (epoxide hydrolase) polymorphisms differ in risks to ovarian carcinoma. • TPMT (thiopurine methyl transferase) converts active anti-leukemia drug 6mercaptopurine to 6-methylMP. (High amounts of drug are cytotoxic, so prescribed drug dosage relies on TPMT activity).
Differences in rate of drug metabolism by TPMT is trimodal, with ~10 allelic variants associated with weak enzymatic activities. Requires dose adjustment sufficient to treat leukemia (since higher dose is associated with higher survival) but to prevent drug toxicity.
1. Pharmacogenomic variation in drug metabolism
CYP2D6 (response to codeine, debrisoquine, et al.)
2. Pharmacogenomic variation due to polymorphism in drug targets
Differences in responses in patients taking drugs to: (1) reduce blood pressure (due to a mutation in angiotensin-converting enzyme) (2) treat asthma (due to mutation in 5-lipoxygenase pathway, or to adreno-receptor gene, depending on the particular drug target) (3) treat osteoporosis (point mutation in calcitonin receptor, which normally inhibits bone resorption)
3. Pharmacogenomic variation in drug response due to disease genes, rather than in the absorption or metabolism of the drug
Alzheimer patients with ApoE4 allele have higher susceptibility and lower prognosis for disease; they also have poorer response to tacrine treatment (80% of non-ApoE4 typically improve with tacrine, whereas 60% with ApoE4 deteriorate after treatment).
6
Active (& potentially profitable) research on drug therapy used for curable mood disorders
5-HTT (5-hydroxy-tryptamine; a serotonin transporter): • recaptures serotonin released at nerve ends and is among the principal mechanisms for eliminating this neurotransmitter from the synaptic cleft. (This transporter is the target of SSRI antidepressants.) • length variants of 5-HTT based on the number of copies of a 20-23 bp repeat: long allele is associated with a favorable response to SSRIs
7
8
Are there differences in the type of genetic testing for disease prognosis/diagnosis & for pharmacogenomics? pharmacogenomics?
One is predictive of onset (before symptoms begin), & the other for treatment (if symptoms begin)
Pharmacogenomic differences detected by transcriptional profiling
Golub et al. (Science 286: 531) compared transcription profiles of blood samples from patients with acute lymphoblastic leukemia (ALL) & acute myeloid leukemia (AML). Gene expression profiles were sufficient to classify patients into categories that are known to respond differently to chemotherapy.
9
There are now expression-based classification systems for many types of cancers and tumors, (including lung, breast, brain, kidney, prostate lymphoma, leukemia) for use in prognosis of disease progression and possible therapies.
10
Gene expression profiles can show effects or targets of therapeutic agents
A-matrix: relationship between tested compounds on cell lines
T-matrix: relates cells to gene expression levels or targets
Gene expression
Sensitivity to 1400 compounds
Product of A x T yields relationships between tested products and gene expression levels
11
Relationship between drug activity and gene expression (AT-matrix)
A block of red represents a positive correlation between a cluster of genes and a cluster of drugs (i.e., the drug is more active in those cell lines that express more of the gene)
Models of Human Disease
(from Gibson and Muse)
12
102 heritable disorders in humans are associated with defects in nuclearencoded mitochondrial proteins. (1000 proteins involved in mitochondrial function.) Screened the entire pool of 4700 homozygous diploid yeast deletion mutants to identify mt-associated proteins by growth on fermentable (glucose) and non-fermentable (glycerol, lactate, ethanol) substrates. (Mutants with mt respiratory defects - “petites” - have impaired growth on non-ferm substrates.)
M: Strains with deletions in genes of known mitochondrial function/biogenesis (n= 353) Red/Blue: fitness (growth rate) above/below strain median Tc1/2: replicate experiments
Identified 466 yeast nuclear genes whose deletions impaired mitochondrial respiration (265 not previously known to display this phenotype), Many have human homologs and were linked to heritable diseases based on map positions.
Using a yeast deletion library, Hughes et al. (Cell, 2000) created a reference database of 300 full-genome expression profiles in yeast carrying mutations in characterized or uncharacterized ORF, or treated with a known inhibitory compound. This compendium was compared to expression profiles of yeast (without mutations) grown in the presence of drugs with unknown targets.
13