fragment screening by nmr in drug discovery – f methods · fragment screening by nmr in drug...
TRANSCRIPT
Fragment Screening by NMR in Drug Discovery – 19F Methods
Dr. Stefan Jehle, Bruker BioSpin Dr. Pavel Kessler, Bruker BioSpin
German Users Meeting, November 9th, 2016
The principle of Fragment Based Lead Discovery from efficient fragments to Drug candidates
Fragment high “ligand efficiency” (LE), medium IC50
Candidate high LE, low IC50
0.91µM
0.59
0.07µM
0.55
5.9nM
0.49
3.0nM
0.42
IC50: LE:
Grow, optimize
Advantages of FBLD – start small and grow big: Fragment chemical space (109) << drug like chemical space (1030) • Fragment space easier to explore • Smaller libraries with higher diversity (~ 1’000 – 5’000 compounds) • Higher hit rates (0.1% - 10%) • Cost efficient
NMR in Fragment based Screening
More than one answer possible
http://practicalfragments.blogspot.ch/2016/10/poll-results-affiliation-metrics-and.html
% respondents using NMR technique
November 10, 2016 4
Fragment Based Screening NMR vs. SPR
NMR Fragment Screening SPR Fragment Screening
100 samples, 500-1000 compounds per day (19F 3000 compounds/day) (5-10 compounds per sample)
500 compounds per day, single point measurement (depends on instrument)
Operational costs: ~45k per year (96 well format NMR tubes, cryogens and service contract)
Operational costs: 45-50k per year (chips for target immobilization, consumables, solvents and service)
QC of fragments possible as part of process, inherent concentration information aids hit validation
No QC of fragments possible during process, independent QC required
No issue, one tube per sample: bad sample does not stop the screen
Sticky compounds may dismantle the chip during screening
Clumsy, home-built acquisition automation, data organization, no workflow support
Easy to use interface for operation, data analysis and export
Throughput
Running costs
Data Quality
Bad samples
Data Analysis
New in Topspin: FBS Analysis Tool
NMR Screening Methods
Ligand observed methods identify binders from mixtures
Protein observed methods identifiy binders and binding site on target
- NO isotopic labeling of the protein - NO size limitation of the target - Little amounts of protein needed (4-10
mgs) - Little amounts of ligands needed (0.025
– 0.250 mM in each sample) No crystal structure available? Binding site in crystal contact (often for
PPI targets)?
- Binding site on target can be identified - Isotopic protein labeling required (13C or 15N) - Size limitation of target - Amount of protein required depends on
application
Affinity Ranges and NMR Detection Limits
KD = 𝑘𝑜𝑜𝑜𝑘𝑜𝑜
Competition Experiments – Spy Molecules
+ +
Reporter/Spy, eventually with known affinity and binding site
Ligand or cocktail; affinity can be determined
19F 19F
No protein labeling required Minimal amount of protein required (low µM) Binding site on protein is known from spy molecule Affinity of test ligand can be calculated if affinity of spy molecule is known Also high affinity ligands detected (replacement of spy)
Experimental Parameters
Test Samples Mixture of tryptophan and tyrosine (2 mM each) and 0.04 mM human serum albumin
(HSA) (65 kDa) in 100 mM phosphate buffer pH 7.4 Mixture of benzamidine and sucrose (2 mM each) and 0.04 mM trypsin (23.3 kDa) in 100
mM phosphate buffer pH 7.4
The protein concentration and fragment to protein ratio strongly depend on the molecular weight (MW) of the protein
For targets with low sample availability the ratio can be as high as ~1:400
Sample Concentration
Excess of ligand (i.e. 0.050 mM) small molecule, different NMR properties than protein
NMR Fragment Based Screening The concept
Low protein concentration (i.e. 0.005mM) Large molecule with distinct NMR properties
High ligand concentration, low protein concentration: ideal for ligand observed NMR!
Ligand in contact with protein (=binding) adopt it’s physical properties during residence time
NMR Fragment Based Screening The concept
Excess of ligand (i.e. 0.050 mM) small molecule, different NMR properties than protein
Low protein concentration (i.e. 0.005mM) Large molecule with distinct NMR properties
norm
aliz
ed in
tens
ity
time
T2
Binding ligand keeps properties when back in the pool, non-binding ligands are unchanged
norm
aliz
ed in
tens
ity T2
time
NMR Fragment Based Screening The concept
Fluorine Fragment Screening
5’-19F-Tryptophan, H2O/D2O, phosphate buffer, pH 7.4
5’-19F-Tryptophan + Human Serum Albumin, H2O/D2O, phosphate buffer, pH 7.4
20 ms relaxation delay
200 ms relaxation delay
20 ms relaxation delay
200 ms relaxation delay
Spin-Echo pulse program for relaxation measurement
Two point measurement, binding fragments show faster relaxation when compared to non-binders
Fluorine Fragment Screening Advantages and Opportunities
Usually one peak per fragment
No water suppression No buffer signals Up to 30 fragments
per mixture Increased throughput Low compound and
protein concentration
typical conditions:
protein = 8 µM
compound = 20 µM
sample volume = 170 µL
(3mm tube)
measuring time = 8 min
600 MHz QCI-F CryoProbe
19F-detected FBS
Courtesy of Dr. M. Blommers, Novartis Pharma, Switzerland
Fluorine Fragment Screening Problem: Large Spectral Width
Phasing of the spectra is difficult if hard 180°pulses or adiabatic pulses with insufficient bandwidth are used
Fluorine Fragment Screening
Spin-Echo pulse program for relaxation measurement
Broadband Adiabatic Refocusing Pulse
WHAT WORKS @ 700MHz: 1.5ms Crp90.comp4 Power of 13 us 90-square (19.5kHz) Excitation profile 70kHz = ca. 110ppm 19F @ 700MHz
Fluorine Fragment Screening Broadband Adiabatic Refocusing Pulse
WHAT WORKS @ 700MHz: 1.5ms Crp90.comp4 Power of 13 us 90-square (19.5kHz) Excitation profile 70kHz = ca. 110ppm 19F @ 700MHz
Spin-Echo pulse program for relaxation measurement
Fluorine Fragment Screening Benefits of 1H decoupling
1H decoupling significantly improves signal to noise in 19F detected screening spectra
With 1H decoupling
Without 1H decoupling
CryoProbe QCIF
19F-observe 1H-decouple 19F S/N 6000:1 @ 600MHz
Smart probe BBFO
19F-observe, 1H-decouple 19F S/N 565:1 @ 600MHz
N2 cooled CryoProbe Prodigy TCI
19F-observe without 1H-decoupling 19F S/N 2400:1 @ 600MHz
19F Probe Options for FBS
How about “druggability assessment”? Determine the “ligandability” of a target
Computational hot-spot mapping is used since > 15 for druggability determination
Hereby a small set of probes is mapped on the surface and favorable binding modes are detected
If a certain number of probes bind and sufficient large hot spots are found, a target is considered druggable
How about “druggability assessment”? Determine the ligandability of a target
Computational methods
depend on the protein structure that is used Experimental condition like buffer, pH, cofactors etc. are not considered
Experimental druggability assessment
19F library with a diverse set of 900 fragments Cocktails with 30 fragments each makes 30
samples Measurement time 8 min per sample = 4 h total Data analysis 2 hours
One day of 19F NMR fragment screening can save many months ($$$) of work on a target or in buffer conditions under which ligands cannot be found
Immediate ROI for a typical NMR instrument 600 + SampleJet + CP QCIF
Data analysis and interpretation – up to now
For each sample, you
• Open the STD spectrum
• Open the STD reference
• Open the Water-LOGSY
• Open the T2 experiment
• Open the T2 reference
• Scale all of them for amplitude
• Search for, and open all single compound reference spectra for that mixture
Before you can start analyzing
The actual screening campaign Data analysis and interpretation – TS3.5 next pl
With TopSpin now, you
• Point to mixture table
• Indicate parent directory for single compound reference
• Indicate parent blank screening data (if applicable)
The actual screening campaign Data analysis and interpretation – TS3.5 next pl
Customizable Spectra Types by unique identifier
The actual screening campaign Data analysis and interpretation – TS3.5 next pl
Customizable display layout
…. you start analyzing!
November 10, 2016 27
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 28
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 29
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 30
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 31
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 32
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
My comment
November 10, 2016 33
New FBS tool in Topspin
Data analysis and interpretation – TS3.5 next pl
November 10, 2016 34
New FBS tool in Topspin Data analysis and interpretation – TS3.5 next pl
The actual screening campaign Data analysis and interpretation – TS3.5 next pl
Customizable report layout
November 10, 2016 36
Alavar Gossert, Wolfgang Jahnke, Marcel Blommers, Cesar Fernandez, Paul Erbel
Daniel Wyss and Hugh Eaton
Stefan Jehle, Pavel Kessler, Fabrice Moriaud, Matteo Penestri, Anna Codina
Bettina Elshorst
Markus Schade
© Copyright Bruker Corporation. All rights reserved.
Innovation with Integrity
Copyright © 2011 Bruker Corporation. All rights reserved. www.bruker.com
Thank you for your attention … and come see us for demos and discussions