Lecture 14: Use of genomic information for drug repurposing- Dr Williams Flashcards
How does APP lead to Alzheimers disease?
APP is a membrane spanning peptide chain, cleaved by two classes of secretases- Gamma and beta, resulting in soluble amyloid peptide. The mutation A to G. Glycine unlike other amino acids does not have a side chain, only has two hydrogens, consequence of this being that glycine impart higher flexibility to the peptide chain to protein chain, in this case the consequence of this being that these peptides can bind to others which leads to aggregates and form plagues leading to AD
How can gene association with disease be discovered?
GWAS
Separate population into two groups and identify what is common between groups
Compute a simple statistical test, whereby prevalence of a given allele variant between two groups are compared - Qualitative trait loci
Then assign an association level to the SNPs.
What constitutes a significant Qualitative trait loci?
Cohort size
What are examples of a GWAS elucidating disease mechanisms?
- Age-related Macular degeneration: A common SNP that causes the mutation Y402H in complement factor H (CHF) identified. Subsequently found that CHF
involved in retinal development - Crohn’s disease:
rs2241880 T300A in ATG16L1 compromises autophagy
rs1000113 intron variant, but lowers expression of IRGM
and compromises autophagy
Autophagy is critical defence against diverse stress
or stimuli, including infection and micro-organisms
- Genetic variants and drug response:
Cohorts based on differential response to hepatitis C
medication revealed multiple SNPs near IL28B gene.
Thus GWAS can inform personalised medicine.
How is the qualitative trait loci measured?
Association measured using odds ratio (ratio of the percentage occurrence of allele in two cohorts).
GWAS is usually performed with SNP arrays probing a limited set of variants. In many cases the causal SNP may lie in the vicinity of a non-causal but significant GWAS SNP (a synthetic association).
What are two ways this issue can be overcome?
Large scale genomic data, such as the 1000 genomes database, can be used to impute (predict) neighbouring SNPs and thereby discover a causal variant.
Resequencing can be performed in the neighbourhood of a QTL.
This way GWAS has identified genes implicated in T1&2DM, obesity and inflammatory bowel disease.
What are two ways to power the statistics when individual associations are weak?
- Grouping SNPs according to predefined criteria. For example, groups of weakly associated SNPs can be mapped to genes and these scored against gene interaction maps or pathways.
- Can build a multi-SNP model on GWAS data directly. This results in a polygenic risk score (PRS). Here, one trains on a set of data and tests on an independent cohort. - Most diseases are not strongly linked to single variants. But, there is a genetic susceptibility. Maybe this is spread over multiple Single Nuclear Polymorphisms SNPs
SNP candidates can be prioritised by associating them with biological effects or functional regions of DNA
What is one way of examinining
biological consequence or the phenotypic consequences of genetic variation?
GTEx portal.
How can SNPS be prioritised in GWAS?
On the basis of SNPS that reside in functional regions
Variants in functional regions of DNA more likely to be causal
Common SNPs can be markers of inheritance not disease susceptibility, therefore what should be considered?
Need to normalise for relatedness – eliminate Population Stratification
Why do many false positives arise from GWAS?
Due to multiple testing errors because the number of parameters measured is much bigger than the number of samples M»_space; N.
Can overcome this by:
expanding the cohort – bigger N
predefining SNP sets – smaller M
Elastic net fitting may be an option:
- keep weight small
- limit number of weights
GWAS has limited utility as a screen for disease susceptibility
Why?
Many more false positive attained
E.g Type 1 diabetes is rare and thereby more positives found in those who are not likely to develop disease
What are two ways in which rational drug discovery can be pursued?
- Peptide introduced to bind to protein and interfere with binding
- Screen compound library’s to bind into various regions- druggable pockets of protein
What is an example of structure based drug discovery?
The Nogo receptor (NgR) inhibitory complex
Develop small molecules to enable CNS neurons to overcome myelin inhibition – issue in CNS and nerve injury
Myelin scar generated and this prevents neurons from regrowth
NgR bound to GT1b and this blocks MAG, OMGP and NOGO66 effectively allowed neurite to grow over myelin scar
What steps facilitated the structure based drug discovery?
First peptide scan computed whereby systematically synthesised peptides from various putative sites of the protein protein interaction
Next located site of interaction
Use a ‘water’ probe to search for ‘druggable’ cavities
NgR has ‘druggable’ cavity proximal to active peptide
Prior to embarking on compound library screen, should try to reduce number of compounds one will test in lab
One way is virtual docking go compounds into cavity, this reduces time taken to screen as fewer compounds will be chosen to screen
What is a docking screen approach?
Flexible ligand library ~100k is docked into rigid cavity
e.g. GOLD – genetic algorithm to find best pose
then scores this with statistical potential
What is an issue with structure based drug discovery?
Some complex diseases have no single well-defined target for intervention.
Novel entity development takes time with massive attrition - 1 in ~10,000 get through the pipeline
What are alternative routes to drug discovery?
Use FDA drug for condition for which it wasn’t developed
There are 1000’s approved drugs and many more that have passed stringent criteria but failed in trials for the specific disease i.e. most of the hard work has been done
What is an epidemiological approach to drug discovery?
Direct correlation with disease incident and drug use. This can be carried out because Norwegian government data released of medical prescriptions and disease state. Lack of incidence of Parkinsons disease and the use of Salbutamol (β2-Adrenoreceptor agonist.
Found salbutamol inversely correlated with incidence of Parkinsons disease
If Propranolol taken (antagonist) opposite found
What does Transcription-based Repurposing involve?
What data typed are used for each?
Disease characterised by robust gene expression changes
Drugs with reverse effects may be candidate therapeutics
Disease data:
Human samples
Experiment in vitro & in vivo
Hypothesis
Drug data:
LINCS
~5000 drugs & multiple cell linesonly over 1000 genes
Connectivity map
1300 drugs on cancer cell lines
NMAP SMART-AD
~200 on iPSC neurons
Cells have more or less the same genome. But there are different cell types.
What are differences explained by?
Different expression of genes
Epigenetics
Transcription factor activation
What is an example of repurposing on the basis of cell culture transcriptional profiles?
EGF and proliferation/survival of NSCs
Compounds that correlated with gene expression profile of antagonising EGFR receptors were all cancer drugs- blocked proliferation
Drugs that anti correlate with blockages of antagonising EGFR were PPPAR agonists
Hypothesis that PPAR agonists will boost proliferation
Found that PPAR agonists reverse Age associated declineof NSC proliferation
What is an example of repurposing on the basis of cell culture transcriptional profiles for CNS regeneration?
MAG - myelin associated glycoprotein
MAG introduced to neuronal culture = comprise neurite length
Screened for profiles which are reverse of MAG induced transcriptional profile - transcriptional profile generated in cerebellar granule cells which are differentiated to highly specialised cells
GW- 8510 (CDK2 inhibitor)
CDK2 inhibits myelin protein, neurons regrow when introduce this compound