This file includes the SLAS2013 presentations of Paul A. Johnston of University of Pittsburgh; Douglas Auld of Novartis Institutes for Biomedical Research; and Lisa Minor of In Vitro Strategies, LLC.
2. University of Pittsburgh Drug Discovery Institute
• Established 2005
– School of Medicine
• John S. Lazo – Department of Pharmacology & Chemical Biology
– School of Pharmacy
• Barry I. Gold – Department of Pharmaceutical Sciences
– School of Arts and Sciences
• Peter Wipf – Chemistry Department
• Pittsburgh Molecular Library Screening Center, 2005
– Member of the NIH pilot phase MLSCN, U54MH074411 (Lazo, PI)
• Pittsburgh Specialized Application Center, 2010
– Member of the NCI Chemical Biology Consortium
– Lazo & Johnston (Co‐PI’s)
• University of Pittsburgh Cancer Center, 2010
– Chemical Biology Facility (ChBF)
– Cancer Center Support Grant (Davidson, PI)
UPCI Chemical Biology
Facility (ChBF) 2
3. HTS Facility Functions
Assay Assay HTS/HCS Active Hit Lead
Development Validation Campaign Confirmation Characterization Optimization
• HTS/HCS assay development collaboration/consultation
– Development & optimization primary, secondary & tertiary assays
• HTS/HCS Validation
– Automated process
– Z’‐factor, S:B ratio, DMSO validation, LOPAC & NIH Clinical Collection library screening
• HTS/HCS campaign
– HTS/HCS data processing & quality control review
– Active identification & confirmation
• Data Generation & Reporting
– HTS/HCS data analysis
– Compound classification, clustering & similarity searches; Cross target queries – biological promiscuity
• Hit Characterization
– Counter screens, secondary & tertiary assays
• Lead Optimization
– Iterative bioassay support of the SAR effort
• Grant Submissions & Contracts ‐ collaboration/consultation
– Preliminary data
– HTS/HCS specific aims & statements of work
• Publications & Teaching
3
5. 21 Primary HTS/HCS Campaign Collaborations
• Johnston research group 2006‐2012
• 4.62 million data points collected & 3.85 million compounds screened
– Assay development & HTS/HCS implementation
• Caleb Foster*, Jennifer Phillips*, Sunita Shinde*, Salony Maniar*, John Skoko*, Yun Hua , Daniel
Camarco, David Close, Stephanie Leimgruber*, Seia Comsa*, & Richard DeBiasio*
– HTS/HCS informatics & Chem‐informatics
• Tong Ying Shun & Harold Takyi
– 21 publications (2006‐2012) & several manuscripts in preparation
5
6. Establishing an Academic HTS/HCS Facility
Initial funds from Institution & Grants
• Capital investment HTS/HCS hardware $$$
• Capital investment informatics hardware $$$
• Capital investment informatics software $$$
• Purchase a compound &/or siRNA library $$$
• Equipment service contracts $$$
• Software licensing fees $$$
• Suitable institution space available – rent? $$$
• Salaries, reagents & supplies – Grant $$$
6
7. Funding and Maintaining
an Academic Screening Center
• Institutional investment
– Space, equipment, IT hardware & software
– Compound & siRNA libraries
• Core facility or independent institute model?
– Core facility – institutional support
• Grants, contracts, foundations & donations
– Personnel salaries
– Equipment service contracts (multi‐year)
– Software licensing fees (multi‐users)
– Reagents & supplies
• Grants ‐ current funding level ≤ 7‐8 %, 3‐5 yrs support
– RO1 grant modular budget $250K/yr, 3‐5 yrs
– RO1 grant budget > $250K/yr ‐ need to justify
– R21 grant modular budget $125K/yr, 1‐2 yrs
• Large equipment grants – multi‐user consortium
7
8. Sustaining a Funding Stream:
It’s all about the collaborations!
• NIH pilot phase MLSCN, U54MH074411 (Lazo, PI)
– 11 HTS campaigns funded
• NCI Chemical Biology Consortium (Lazo & Johnston Co‐PI’s)
– NeXT STAT3 pathway inhibitor project (Grandis, PI)
– NeXT cMyc inhibitor project (Prochownik, PI)
• NCI contract
– Drug combination screening in the NCI 60 cell line panel (Eiseman, PI)
• HTS/HCS Collaborations –co‐investigators
– STAT3‐GFP nuclear localization assay development (Reich, PI)
– AR‐GFP nuclear localization assay development & HTS (Zhou, PI)
– TLR 3 signaling assay development and HTS (Sarkar, PI)
– MCAD assay development (Moshen, PI)
– ATZ assay development and HTS (Silverman, PI)
• HTS/HCS assay development and screening ‐ PI
– AR‐TIF2 protein‐protein interaction biosensor NINDS R21, (Johnston, PI)
– AR‐TIF2 protein‐protein interaction biosensor NCI RO1, (Johnston, PI)
8
9. Leveraging Focus Libraries and Quantitative HTS
in Assay Pilot Testing
Screen Design and Assay Technologies Special Interest Group:
Screening in this economy….What makes ‘cents
Douglas Auld, Ph.D.
Novartis Institutes for Biomedical Research
Cambridge, Mass., USA
10. Testing multiple hypothesis early
Better starting points for drug discovery
Focus libraries can be used to characterize assays
and help choose the right set of assays for the project
11. What to screen?
Types of libraries
LMW libraries
• Probes, drugs, tools
• Natural products
• Previously found program compounds (small molecules of unknown
targets –”SMUTS”)
RNAi libraries
• Down-regulation of target only
• Time-scale of response very different than LMW treatments
• Specificity
Biochemicals
• Peptides
• Metabolites
• Nucleic acid mimics
14. Focus libraries in understanding phenotypic assays
Two opportunities
• Old view: Assay is used to characterize the compound library
• New view: Compound libraries are used to characterize
assays –choose the best assay(s) and compound subset for
the project
1. Build a 2. Generate a
hypothesis prior hypothesis post HTS
?
to HTS
6 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
15. Focus libraries
Size and types of libraries at NIBR
1,400 – one 1536w plate (Challenge) –test for frequent hitters (solubility data
Read-Out artifacts (e.g. fluorescence)
2,733 – two 1536w plates. Drugs, clinical candidates, tool compounds (MoA)
4~10K – eight 1536w plates Random set hit rate estimate
250K - Focus screen (~180 1536w plates)
-plate based diversity set
-Biodiverse set –target class annotation.
Epi
Challenge BioDiv.
MOA
7 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
16. Sources of chemical biology information
Annotation of compounds, pathways, mechanisms
Compound Databases Bioinformatics Resources
• ChEMBL • EntrezGene
• ChEBI • InterPRO
• DrugBank • GeneGo
• Thomson Reuters Integrity • SCOP
• World Drug Index • UniProt
• PubChem structures • Clinicaltrials.gov
• eMolecules (8M compounds) • Broad Connectivity
Map
• PDB
• Pubchem BioAssay
• Binding DB
Semantic Standardization enables interoperability
•Merges chemical and biological data
•Internal, historical data + external data
•Maps assay metadata to results data
•Provides chemical structure (InChIKeys) and target
normalization (Gene ID)
17. NDFI: Novartis Data Federation Initiative
Goal: Allow rapid mining of data to generate knowledge
Forming a searchable chemical biology database
Reporting
Literature
Databases
Data Analysis Results Publication Data Data Analysis
Warehouse
Assay Data
Assay Metadata Assay
Assay data
registration
Sample Information
Results Capture Assay Registration
Chemist
Sample
Biologist registration Assay Request
Assay Configuration
eLN eLN
Assay Request
9 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
18. Focused screen with qHTS format
Allows for both plate-based performance and validation data – 1 experiment
quantitative HTS
“qHTS”
First four concentrations represent typical single concentration screening scenarios Last four
provide robust curve fitting. One experiment yields:
Plate-based single concentration data for assay performance stats.
Validation data obtained – dose response data
A “truth matrix” is the output: True positives (TP), false positives (FP),
false negatives (FN), true negatives (TN), confirmation rate (CR) and Hit rate
per concentration can be considered in choosing the final screening concentration.
10 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
19. Two types of data from qHTS
Scatterplots and concentration-response curves
1
2
CRCs
Retrospective analysis
Annotate activity
based on hit threshold
and CRC information
TP, FP, FN, TN
11 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
20. qHTS Retrospective analysis
Analysis with Pipeline Pilot
• Top tier – get curve fit information from database file and tag data as “CRC” if active criteria are met
• Bottom tier – get hits from database file based on a cut-off threshold (e.g. <-30%), tag data as ”Hits”
• Annotate all data as either TP, FP, FN, or TN depending if the “Hits” are found as “CRC” actives.
12 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
22. Challenge Library Construction
Focus on biochemical assay interferences
Challenge Library
Solubility / Aggregation/ “PAINS”
• Solubility data
• Read-Out artifacts (e.g. fluorescence)
• Hits in counter-screen
Covalent Protein Modifiers
• LC/MS assay for covalent modification data
During construction, target unselective ligands (e.g. non-specific kinase
inhibitors) were not taken as “frequent hitters”, left out of Challenge library
∑ 1,408 compounds, fits in one 1536-well plate
Luciferase inhibitors are available as a separate subset (for reporter-gene
assay characterization)
23. Challenge Library Construction
Focus on biochemical assay interferences
Frequent hitter analysis
Many interfere with fluorescence and AlphaScreen
Freq. hitter = screened in at least 10 assays and hit >50% of these
(compare - same analysis with1,400 randomly picked compounds
yields only 1 freq hitter)
24. Firefly luciferase (FLuc) is a popular choice for RGAs
FLuc inhibitors can confound the interpretation of RGA results
Observations:
• FLuc inhibitors compose a ~4% of typical screening libraries (determined in an biochemical FLuc enzyme
assay)
• FLuc inhibitors are highly enriched (40-98%) in hit list derived from FLuc-based RGAs
• Can show apparent activation in cell-based FLuc-RGAs due to inhibitor-based enzyme stabilization
Tool set:
• Known luciferase inhibitors available to characterize primary and counter-screen assays
• Mechanistic understanding of luciferase inhibitors can be used to develop robust orthogonal assays
PubChem examples: Frequent Hitter analysis of confirmed
See Thorne et al. (2012) Chem Biol. 2012 Aug 24;19(8):1060-72.
FLuc inhibitors
16 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
25. Biochemical assay against the Challenge Library
Comparison of two buffer systems biochemical assay
Biochemical assay for an essential enzyme in bacterial cell
wall synthesis
Fluorescent read-out in 1536w plates
Use Challenge library to identify a buffer system that
reduces interference with the assay
• Lower hit rate against Challenge library
26. Biochemical against Challenge Library
Comparison of two buffer systems biochemical assay
• “Modified buffer” showed a 1% lower hit rate at any of the concentrations
while FPR was similar. FNR shows weakening of interference so this was
chosen for the screen
• Additional definitions:
• Diagnostic FNR reports on the fraction false negatives relative to the
total true positives
• Absolute is relative to total samples
• Relative rates is to total # of hits
27. Role of MoA library in assay pilot testing
A comprehensive set for understanding the MoAs underlying an assay
A comprehensive chemical probe set for understanding the MoAs underlying
an assay
Understand targets and pathways influencing a screen in the pilot stage
• Most useful for understanding mechanism underlying phenotypic responses
Enable decisions about wanted or unwanted molecular mechanisms to
facilitate design of counter-screens or secondary assays for compound
triage and prioritization for follow-up work.
Add to knowledge base - linking known compounds to new biology
19 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
28. MoA Library composition
Reflective of current pharmacopeia
Anti infective
Antiinflammatory
Apoptosis
Enzyme(other)
Epigenetics
Ion Channels
Lipid kinase/metabolism
Metabolism/antioxidants
Nuclear receptors
P450s
Phosphodiesterases/cyclases
PPI
Proteases
Protein Kinases
Receptors
Stress
Transcription/Translation
Transportors
~3K compound, 1,700 targets
20 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
29. Use of MoA library analysis
Assay flow chart development
Enriched target classes
Primary
√
HDAC Counterscreen
binning Orthogonal
assayy
Hit
prioritizaiton
Secondary
21 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
30. Chemical vs. biodiversity
Chemical diversity is necessary but not sufficient for biodiversity
P.M. Petrone, A. M. Wassermann, E. Lounkine, P.Kutchukian, B. Simms, J. Jenkins, P. Selzer and M.Glick
31. Plate-based biodiversity selection
Diverse Gene Selection (DiGS)
Novartis screening deck, annotated with Sort plates according to number of targets per
biological activity plate
...
Targets that have been covered on plates higher in
the list, are not counted on plates lower in the list
Eliminate redundant scaffold‐target
occurrences
50-80% coverage
Select the top n plates of bioactive
e.g. 710 plates (384) = 250k compounds
compounds
covered.
P.M. Petrone, A. M. Wassermann, E. Lounkine, P.Kutchukian, B. Simms, J. Jenkins, P. Selzer and M.Glick
23 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
32. Two views
Annotated libraries
Annotations only tells us what we already know. Compounds
will target what we have already found and will not get us into
new areas of biology.
Annotations reflect “bioactive” (“privileged”) structures which are
capable of interacting with biological components (protein
surfaces, binding pockets) and therefore should be useful to
widely probe biology.
33. Two views
Annotated libraries
Annotations only tells us what we already know. Compounds
will target what we have already found and will not get us into
new areas of biology.
Annotations reflect “bioactive” (“privileged”) structures which are
capable of interacting with biological targets
Chemistry & Biology 18, October 28, 2011, 1205.
34. Summary
Focus library design and testing is an increasing practice
during assay optimization
• Generate target hypothesis pre and post-HTS
Three pilot libraries are available in qHTS format at NIBR
• Challenge (artifacts sensitivity)- anticipate counter-
screens and orthogonal assays
• MoA (pathway analysis) – anticipate secondary assays
• Random set – hit rate, screening concentration estimation
Focus library testing benefits from full titration-based
analysis
26 D.S.Auld | Pilot Libraries and qHTS| Sept. 21, 2012| 8th Compound Management & Integrity| Business Use Only
35. Acknowledgements
SLAS workshop
Focus libraries (NIBR):
Meir Glick
qHTS (NIBR):
Jeremy Jenkins
Ji Hu Zhang
Ansgar Schuffenhauer
Hanspeter Gubler
Jutta Blank
Ophelia Ardayfio
Peter Fekkes
Zhao Kang
Marjo Goette
Adam Hill
Martin Klumpp
Shin Numao
SMG (NIBR):
Johannes Ottl
Scott Bowes
Günther Schee
Manori Turmel
Caroline Engeloch
Greg Wendel
Ben Cornett
Florian Nigsch
Christian Parker
27 | Reporters in cell-based assays: Understanding fact from fiction| Doug Auld| 1-14-2013 | SLAS 2013t | Business Use Only
36. Implementation of qHTS paradigm
Pilot testing of focus libraries at NIBR
Traditional pilot testing: Run assay against Pilot151 library
at one concentration in duplicate.
• Examine hit rate, order hits, dilute compounds, validate by determining
CRCs, and calculate confirmation rate
• Oftentimes if the hit rate/confirmation rate is unacceptable - repeat the
process at a different concentration
qHTS approach
• Develop a titration-based archive for specific pilot libraries
- MoA, Challenge, Random sets
• Two data sets are obtained from experiment:
- Scatterplots at each concentration - primary hit rate
- Concentration-response curve – pharmacological information on every
compound
- Retrospective analysis of the data can be use to calculate FNR, FPR, CR,
and hit rate as a function of concentration to determine optimal screening
concentration.
37. Assay and Screening Strategies to
Survive an Ever Changing World
Lisa Minor
In Vitro Strategies, LLC
38. Outline
• The changing world of HTS
• Assay challenges
• Assay survival
• The changing world of the Drug Discovery
Industry
• Personal survival
39. The Changing World of HTS
• Long long ago- “the great new world of HTS”
– HTS was the place to be
• Screen more compounds/faster/cheaper
• Screen in simple uni-dimensional platforms
• The assay balance was weighted toward biochemical assays and fewer cellular assays
• Was a need for data analysis and archiving databases
• Need for new robotic platforms/new dispensing platforms/higher density formats….
• Long ago-
– HTS still the place to be but no longer a great new world
• Screen more compounds with smaller volume
• More complicated assay platforms
• Robots were common
– Attempts for large robotic platforms
– Attempts for large cell culture platforms
• Cellular assays with very directed output were emerging
• Data archiving and analysis databases are emerging
• Now:
– Screen target directed compound libraries
– Workstation robotic platforms
– Screen smaller diversified compound libraries
– Screen in small volumes
– Increase in cellular assays so cellular assays outnumber biochemical assays
– Increase in phenotypic cellular assays
– Finally, an increase in high content assays
40. Drug Discovery Process
• First: Target Identification and Validation
– Can be molecular target or phenotypic result
• Identify a screen to interrogate the target
• Identify parallel assay to test the hits so don’t have assay bias
(example: luciferase inhibitors paradoxically cause luciferase
increases in cell based assays)
• Identify lead series and begin chemistry
• Identify secondary testing strategy or testing funnel
– Test activity for similar receptors/enzymes/species activity overlap etc.
– Test toxicity profile
– Test for solubility
– Identify biomarkers for compound activity and ideally for target
engagement
– Test for in vivo activity
41. Assay Challenge/Assay type
• Biochemical target/assay
– Good: great SAR potential/best for target engagement/may be easier
to develop the assay/assay rules are in place/you know all of the
players
– Poor: there may be fewer low hanging fruit here as targets
• Defined cellular single target:
– Good: good SAR potential/OK for target engagement
– Poor: many targets may require more than one readout so single
target may not cut it
• Multiple readout for a single cellular target;
– GPCRs (multiple signaling path); good in that final compound may be
more specific, have better toxicity profile but need to develop multiple
compounds with each profile to ID the right profile
• Phenotypic target readout
– Good: may have more physiological relevance
– Poor: more difficult to lead SAR/no real target engagement so
development of the compounds may be more difficult
42. Assay Challenges: Cell type
• Cell line
– Easy to run
– Can transfect with target if target is known
– Good SAR development
• Physiological relevant cell – primary or primary like
– Good: may yield results that are more realistic
– SAR may be challenging if target is not known
– Can run fewer compounds
– Expensive
– Stem cells?
• 2D vs. 3D
– Potential for 3D being more relevant but still new area
• how is drug delivered completely to the 3D structure
• How are 3D structures organized, self organized?
• How based in reality is the 3D structure?
• Does your assay used in 2D exactly transfer to 3D?
– Not likely
– Needs complete validation ex. Does lysis reagent completely lyse the cells or are you
getting artifacts?
43. What is the Best Assay?
• Criteria
– Target known if possible
• Find a way to deal with multiple signaling pathways up front
• Strategy is key
– Good reliability/low variability
– Assay format to suit your company’s screening paradigm
– Adequate throughput
– Adequate cost
– Strategy required to follow through on hits both to validate
target and to develop druggable leads
– Toxicity strategy
– In vivo follow-up strategy
44. What can you do to be successful?
• Do your best to vet the target
– Ask questions of your target validation team
– Work with them to devise the best strategy
• Make the best assay possible
• Don’t run an assay if assay is not reliable: data is only as good
as the assay
• Have a secondary assay in place with appropriate throughput
– assay should be with different but parallel platform to
eliminate assay bias
• Run neat compound to validate hits
• Run resynthesized compound to validate hits
• Keep some of the initial compound in solution to test
composition if necessary
• Identify must haves assays vs nice to have as your time is
valuable
45. Changing World of Industry
• Big pharma Consolidations
• Big pharma outsourcing projects/chemical
synthesis, screening
• Big pharma reducing jobs
• Big pharma partnering more with
academics/biotech in early drug discovery
• More academic drug discovery research
• All makes for uncertainty in the employee
46. What can you do in this changing world?
• Do your best job/make the best assays/make the best
decisions/be reliable
• Present your work internally/externally
• Become invaluable internally
• Be confident
• Keep your CV current
• Network
– Inside your company
– Outside of your company
– In social networks such as Linkedin.
• Be willing to move from pharma to biotech or academia and
to move locations
• Be an entrepreneur and start your own business or work as a
temp if necessary
• Keep a positive attitude (glass half full)
• Have fun