Oncology Drug Discovery in an Academic Setting : Pipeline or Pipedream ? Ian Waddell Head of Biology Drug Discovery Unit Paterson Institute for Cancer Research
© Paterson Institute for Cancer Research
Confessions • I do not consider myself to be an Academic • I have only worked in Oncology for 8 years
Confidential
© Paterson Institute for Cancer Research
MCRC Drug Discovery Unit Background
•Drug • •Manchester Discovery Cancer Unit for Research Paterson Institute Cancer Centre Research •Focus – •University on niche ofcancer Manchester cancer targets institute with clear route to clinical Specialist research core funded by evaluation – Paterson Cancer Research Institute for UK Cancer Research •Targets includeCancer cell Christie metabolic pathways, DNA repair Co-located with Hospital – •The Christie Hospital mechanisms and epigenetics • Fundamental molecular & cell biology through to •Access to PICR research investigators translational •Access to the clinic and translational medicine
Confidential
3
MCRC DDU Group
• Target Identification • Hit Identification
Confidential
© Paterson Institute for Cancer Research
Talk Structure • Context – Drug discovery in Oncology setting – Target selection
• Screening – Subset screening – Fragment screening
• Collaborations – Pharma – Hit finding
• Bioinformatics – Collateral vulnerability Confidential
© Paterson Institute for Cancer Research
Challenges of Drug Discovery in Oncology Neo adjuvant
Only primary tumour detected
“Cured”
Operable
Surgery
Adjuvant
Watch & wait
Initial cancer diagnosis & staging
Relapse
Metastatic
1st line
2nd line
Increasingly refractory
3rd etc.
Palliative
Early Drug Development
New drugs are tested first on the most refractory/resistant population
Confidential
© Paterson Institute for Cancer Research
Target identification: Strategic Considerations CRUK strategy: Novel/higher risk Leverage locality
Target Target
Function
Target validation Clinical
Preclinical
Mutant IDH 1 & 2
2 HO-glutarate onco-metabolite?
MonoTx or combo
Driver or initiator?
G6PDH Ret selective TTRAP /Tdp2 POL Q IDH1, wild type Malic enzyme
PPP glucose metab, NADPH regulation Receptor tyrosine kinase oncogene DNA repair DNA repair NADPH regulation? pyruvate metab, NADPH regulation
RTX combo Thyroid cancer Topo2 -resistant RTX combo RTX combo RTX combo
NADPH hypothesis
PDK1
growth /proliferation (PI3K /Akt p'way)
PTEN null
Taspase I MYD88 L265P UROD PSAT1 PYCR PHD3 BBOX1 Enolase II FEN 1 rad54 rad51 EGFRvIII SCL7A11 6PGDH HGDH PHGDH DNMT1 NLK MLK1 MLK2 BMI
suggested downregulator of Usp9x Adaptor protein CLL
Feasibility Chemistry
Biology
Competitive Biotech
Selectivity vs KDR
Academic
Margin, hypothesis Margin, hypothesis Hypothesis? Hypothesis? Allosteric inhibitor? Nothing to go at
Big Pharma Other part of pathway
Phosphoserine aminotranserase
review in progress DNA repair DNA repair DNA repair
Metabolism GBM / AML Metabolism Methyltransferase Kinase Kinase Kinase
RTX combo
JARID1B
H3K4 demethylase
Melanoma
IDHrev
reductive carboxylation
NADPH hypothesis NADPH hypothesis
MCRC strategy: Lung, melanoma, haems, RTx, women. Local PIs
RAC1 (P29S mutation)
C-FLIP
ABC transporters RNF2
PDPK1
DNA Repair
TET2
Areas of biology: Addiction, RTx, stem cells, DDR, hypoxia, metabolism but not cell cycle, inv/met, angiogenesis, immunology Confidential
Allosteric site best way forward
Lipid Kinase
ROS1
selectivity
selectivity
AML
RING1B
E3 ubiquitin-protein ligase
HIF1alpha/p300
protein/protein interaction
PARG LRP16
poly ADP ribose degradation
SNM1a CHFR Aprataxin ALC-1 APLF AAG APG? ANG? MGMT APE
nuclear receptor transactivation, poly ADP-ribose polymerisation DNA repair ubiquitin ligase DNA repair helicase - chromatin stabilisation poly ADP ribose stabilisation base excision repair base excision repair? base excision repair? base excision repair? AP base excision repair
PERK HIF-1
CTX combo / BRCA
Hypothesis
Awaiting structure
Emerging Emerging Emerging CTX/RTX combo
CTX combo CTX combo
Margin? Emerging Margin? Emerging Emerging Margin? Margin?
UPR stress response to hypoxia/anoxia
RTX combo
Hypothesis?
CRT hits
hypoxic response
RTX combo
What is mol target?
Awaiting hits
Hypox bioreductives, nitromidazole NQO1,2 NQO1 NQO2 Hypoxic gene profiling
delivery of warhead to hypoxic tumour active site inhibitor p53 stabilisation, ROS protection p53 stabilisation, ROS protection West/Harris studies
RTX combo RTX combo RTX combo RTX combo
Which warhead?
Stromal infiltration, post therapy mct-4
Awaiting details carboxylate transporter (eg lactate)
Warburg
Sarcosine
amino acid methylation
Inv /met target
γ-secretase FAK CD44 - intracellular signal CXCR4 LIF BMP6 Notch - selective downstream Notch, cbf-1 interaction Integrins /p-Erk
pan-Notch cleavage (GI toxicity) adhesion complex stem cells stem cells & stroma stem cells stem cells cell survival /proliferation nuclear medaitor of Notch sig. growth factor signalling
CTX combo
Academic
CRUK portfolio POM target
Comp chem hits? Hits? Hits? None tractable yet
Big Pharma
Selectivity vs. notch Hypothesis i-c Target?
Merck in clinic Pfizer in clinic Big Pharma
Validation? Validation? i-c Target? i-c Target?
Feasibility? Feasibility? Feasibility?
Wee-1 HSET No opportunities identified yet No opportunities identified yet No opportunities identified yet
Phosphatases No opportunities No opportunities No opportunities No opportunities No opportunities No opportunities
Big Pharma CRUK portfolio
Signalling identified identified identified identified identified identified
Selectivity
yet yet yet yet yet yet
TG2 -
AML oncogene signalling
CD45 phosphatase AEP LSD1 No opportunities identified yet No opportunities identified yet
AML oncogene signalling asparagine supply Leukemia
mcm2-7 helicase CDK RNA Primase cdc7 No opportunities identified yet No opportunities identified yet No opportunities identified yet MLK4 LZK
Helicase
AML ALL (paeds)
Selectivity vs normal AEP as target?
Selectivity watch for GSK
Target complexity Selectivity?
Primase
Target complexity Selectivity?
Big Pharma
Target complexity CRUK portfolio
ser thr Kinase ser thr Kinase
© Paterson Institute for Cancer Research
Competition: Avoid mainstream unless advantage Consider niches
Will it work?
Prioritising Targets :- Clinical Line of Sight Clinical Hypothesis
How clear is the path to clinical testing?
Preclinical Hypothesis
How strong is the preclinical target concept?
Can we do it?
Ratings: Chemical Feasibility
Can we deliver the required drug attributes in a molecule?
Biological Feasibility
Can we measure the desired biological profile?
Competitive Position
Normal risk profile for a drug discovery project Significant risk to be addressed in project plan Major risk to be addressed before starting project
Can we compete in this area (or not)?
Emphasis on clinical alignment Confidential
© Paterson Institute for Cancer Research
8
Talk Structure • Context – Drug discovery in Oncology setting – Target selection
• Screening – Subset screening – Fragment screening
• Collaborations – Pharma – Hit finding
• Bioinformatics – Collateral vulnerability Confidential
© Paterson Institute for Cancer Research
Medium Throughput Screening
89 Confidential
© Paterson Institute for Cancer Research
Subset Screening : Kinase Inhibitor Design •
Targeting a tyrosine kinase associated with a niche oncology setting – Aim to improve selectivity over related kinases implicated in clinical toxicity
Structure-Guided Optimisation
Library Screening
•
• • •
•
Exploit X-ray structures and literature scaffolds Focused med chem optimisation
Early insight into SAR relating to affinity/selectivity Identification of diverse scaffolds for optimisation
Confidential
Virtual screening Fragment library screen Kinase subset library
Subset screening : Kinase Library
•
Screening data typically presents challenges in analysis – Few actives among many inactives – Single-point data noisier than full IC50 – Chemical integrity/interference
•
Does combinatorial library offer advantages for understanding SAR? – i.e. 10’s of scaffolds represented by 100’s of compounds
% inhibition for Kinase B
• Kinase library profiled against 2 kinases – 9000-cpd combinatorial library based on diverse kinase inhibitor scaffolds – Single-point % inhibition at both 30 & 100mM – Confirmation and expansion around selected hits by re-synthesis and IC50 determination
% inhibition for Kinase A
Classical drug discovery Confidential
Subset screening : Library Analysis by Core • Explore activity profile for individual sub-libraries
Simple activity classification
As pie chart – full library Pie charts per sub-library
Docking of cores E & H examples
Confidential
R1
R2
Subset Screening : Decision Tree Analysis
Core H: low MW + low LogP high affinity for Kinase A But similar results found for Kinase B
xLogP
Which descriptors are related to observed activity/inactivity? - Two activity classes: HIGH (>50%) or LOW (50% inhib @ 500uM, light green 25-50%. Horizontal axis: specific compounds.
Dehydrogenase 1 (reverse)
DNA repair 1
DNA repair 2
DNA repair 3
Epigenetics 1
Tyrosine kinase
Primase 1
Confidential
© Paterson Institute for Cancer Research
Fragment Screening : Ligandability Assessment •
Fragment screening data broadly in accord with structural features – Protein kinase – typical kinase binding site, known ligands in literature – Epigenetics target – large polar binding site – DNA repair target – open, solvent exposed site, phosphate-binding subsite – Dehydrogenases – small polar substrate site adjacent to large co-factor site
Dehydrogenase 1
Dehydrogenase 2
Dehydrogenase 3
Heat-map of fragment library hits from biochemical screening
Dehydrogenase 4
Dehydrogenase 5
Dark green >50% inhib @ 500uM, light green 25-50%. Horizontal axis: specific compounds.
Dehydrogenase 1 (reverse)
DNA repair 1
DNA repair 2
DNA repair 3
Epigenetics 1
Tyrosine kinase
Primase 1
Confidential
© Paterson Institute for Cancer Research
Talk Structure • Context – Drug discovery in Oncology setting – Target selection
• Screening – Subset screening – Fragment screening
• Collaborations – Pharma – Hit finding
• Bioinformatics – Collateral vulnerability Confidential
© Paterson Institute for Cancer Research
Progressing Projects: Partnership Models
In vitro POP
Target Selection
Hit Identification
Lead Identification
In vivo POP
Lead Optimisation
DDU
Clinical POP
Preclinical Development
Clinical Development
Partner DDU
Partner
• DDU role is to “de-risk” target for a commercial partner • Timing of handover will be when POP* data reduces risk to acceptable level for partner DDU/Partner
DDU
Partner
• Earlier engagement of partners during HI • First refusal at agreed milestone * POP = Proof of Principle (i.e. compound produces desired biological effect in test system)
Confidential
AstraZeneca DDU Collaborations
http://www.cancerresearchuk.org/cancer-info/news/archive/pressrelease/2013-06-13-CRT-University-of-Manchester-and-AstraZeneca-work-together-to-seek-new-cancer-drugs
Confidential
© Paterson Institute for Cancer Research
AstraZeneca DDU Collaborations Project 1
Project 2
Joint target of interest DNA Repair AZ HTS AZ Crystal Structure AZ Hits
Joint target of interest Screening agreements DDU access AZ HTS and other screening facilities
Poster 30
MCRC DDU Progress to agreed criteria AZ have first right of refusal
Confidential
© Paterson Institute for Cancer Research
MCRC DDU Progress to agreed criteria AZ have first right of refusal
Talk Structure • Context – Drug discovery in Oncology setting – Target selection
• Screening – Subset screening – Fragment screening
• Collaborations – Pharma – Hit Finding
• Bioinformatics – Collateral vulnerability Confidential
© Paterson Institute for Cancer Research
What is Collateral Vulnerability ? – The concept was presented originally in a paper in Nature - Muller at al 2012.
ENO2 proposed as a target for a subpopulation of glioblastoma in which ENO1 is deleted (1p36 locus). Demonstrated experimentally that ENO2 inhibition by small molecules or RNA knockdown was selectively toxic to ENO1-null cells. Confidential
© Paterson Institute for Cancer Research
External Datasets
Complete Sequenced No data
Confidential
© Paterson Institute for Cancer Research
Identification of novel drug targets – We developed a bioinformatics pipeline to identify genes where the concept could be applied. – Simple workflow: • Identify genes near to known deletion loci (e.g. PTEN locus). • Use the Cancer Genome Atlas dataset to calculate percentage of cases in which the nearby genes were deleted. • Select deleted genes having a small number of paralogs. • Select genes that have an essential function. • Select genes that are chemically tractable (i.e. are suitable for a small molecule drug hunting project)
Confidential
© Paterson Institute for Cancer Research
Strategic Importance of Lung Cancer to MCRC
Cancer Research UK Lung Cancer Centre Of Excellence
TRACERx Programme
Confidential
© Paterson Institute for Cancer Research
Application of Workflow to Lung Cancer • Chose lung adenocarcinoma as the first cancer type: 230 samples in TCGA with RNA-seq, copy number, and sequencing Location Frequency Lethality Family sizedata. One of the most common cancer types with high unmet need: even a treatment for 2-3% of cases would be of value.
957 genes
535 genes
296 genes
957 genes identified in 20 deleted regions from Tumourscape. 535/957 genes > 2% 296/535 genes 1-4 paralogs 58/296 many:1 fly lethal 21/296 many:1 worm lethal 67/296 fly or worm lethal
Confidential
© Paterson Institute for Cancer Research
10 ‘follow up 21 ‘reserve’ 37 ‘rejected’.
67 genes
NSCLC Output Top 4 genes – – – –
Jak2 LTK PP2R2A Target X
Paralog siRNA – – – –
Yes Yes Yes Yes
Paralog Inhibitors – – – –
Yes Yes Yes ?
Work with a world leading lung cancer PI to validate the hypothesis with known targets and known inhibitors.
Confidential
© Paterson Institute for Cancer Research
Acknowledgements Drug Discovery Unit: Donald Ogilvie Ian Waddell Allan Jordan Niall Hamilton James Hitchin Stuart Jones Amanda Lyons Alison McGonagle Daniel Mould * Rebecca Newton Ali Raoof Kate Smith Kristin Goldberg Eleanor French Colin Hutton
Confidential
Ben Acton Phil Chapman Mark Cockerill Alex Boakes Emma Fairweather Samantha Fritzl Nicola Hamilton Sarah Holt Gemma Hopkins Dominic James Nikki March Helen Small Graeme Thomson Mandy Watson
Colleagues at: Paterson Institute for Cancer Research Beatson Institute for Cancer Research Cancer Research Technology Discovery Labs AstraZeneca Oncology iMED AstraZeneca Discovery Science
© Paterson Institute for Cancer Research