NMR Fragment Screening at UCB Richard J. Taylor CCPN Conference - - PowerPoint PPT Presentation

Developments in NMR Fragment Screening at UCB

CCPN Conference 13th.July.2017 Richard J. Taylor

High Quality, Soluble, Chemically Diverse Fragment Subset

NMR Hardware

AVIII HD 600 MHz. 5 mm QCI-F He Cryoprobe SampleJet with cooling

Assemble 19F NMR database Construct Screening Cocktails Automated Deconvolution

Collaboration with ACD/Labs to automate 19F NMR ligand screening

• 19F NMR screen is sufficiently rapid to

allow ranking in proteins for STD screening

• High hit rates may be indicative of

non-specific binding

Results of 19F CPMG Screens

Target A

• 131 hits (12 %)
• 94 / 131 confirmed by STD NMR (20 strong)
• Crystal structure of 2 of the hits

Target B isoform 1

• 30 hits (3 %)
• 18 / 30 confirmed by STD NMR (6 strong)

Target B isoform 2

• 24 hits (2 %)
• 18 / 24 confirmed by STD NMR (8 strong)

Target B isoform 3

• 15 hits (1.4 %)
• 8 / 15 confirmed by STD NMR (8 strong)

Target C (tetrameric ion channel)

• 18 hits (1.6 %)
• 11 / 18 confirmed by STD NMR (7 strong)
• HSQC interaction site mapping of 2 of the hits

UCB: Christine Prosser, Zara Sands UoL: Philip Renshaw, Sarah Strong, Gareth Hall, Lorna Waters, Vaclav Veverka, Mark Carr

“….uses too much protein” “…. is too slow to inform Medicinal Chemistry iteration” Protein Observe experiments (HSQC, HNCO) are information rich BUT are often dismissed ….

Ph.D. Studentship Goals:

• Increase sensitivity
• Decrease sample consumption
• Reduce data acquisition time

Fast Pulsing Problem: 2D Experiments are slow Shaped Pulses Fewer Scans Optimise Spectral Width ASCOM BEST-TROSY SOFAST-HMQC Non-Uniform Sampling of t1 Ernst-angle excitation cos(θ) = e -(d1+at) / T1

• Dr. Matthew Goodwin

t1 t2

band-Selective Optimised Flip-Angle Short-Transient (SOFAST) Band-selective Excitation Short Transient (BEST) Tranverse Relaxation Optimised SpectroscopY (TROSY) Automated Spectral COMpression (ASCOM)

294 μg 3.5 hours 294 μg 9.5 hours

100 μM 42 kDa protein

Standard TROSY No shaped pulses Slower scans – interscan delay 1s BEST-TROSY Shaped pulses Faster scans – interscan delay 0.25 s More scans in same time 1.6x more S/N Allows time reduction from 9.5 h to 3.5 h

• Dr. Matthew Goodwin

BEST-TROSY –Reducing D1

Buffers:

• Proteins require salt for

solubility and stability, however, salt reduces S/N

• Low salt buffers result in a

~20 % increase in peak intensity

Phosphate Buffered Saline Arg-Glu Buffer

Increase relaxation:

• Gd3+ chelate based contrast agents
• ~ 1mM Gadbutrol, available

commercially as Gadovist, used as an MRI contrast agent

Gd3+ - Sibille et al. Low concentrations of a Gd-chelate increases the signal to noise ratio in fast pulsing BEST experiments.

• J. Magn. Reson. 224, 32-37 (2012)
• Dr. Matthew Goodwin

Buffer Optimisation

700 MHz 1.7 mm Microcryoprobe

• Bruker 700 MHz 1.7 mm

microcryoprobe ~ 5x the mass sensitivity of the 600 MHz. 5mm cryoprobe

• Optimised for sample

volumes of 35 µL – rather than 70-600 µL in a 5mm probe

• Tube based operation
• SampleJet for automation
• Gilson 215 robot for

sample preparation

High Sensitivity - Low volume - Automated

• Prof. Matthew Crump, Dr. Matthew Goodwin

Schanda et al. SO-FAST HMQC experiments for recording two-dimensional heteronuclear correlation spectra of proteins in a few seconds. J. Biomol. NMR, 33, 199-211 (2005)

25 μM 15N IL6 (~18 μg) 1 mM ligand Arg Glu buffer 1 hour (cf 100 μM 15N IL6 (~147 μg) 1 mM ligand 30 minutes at 600 MHz in 70mL

• Dr. Matthew Goodwin

SOFAST 1H-15N HMQC 700 MHz 1.7 mm Microcryoprobe

Binders (and water) Non-Binders

WaterLOGSY 25 μM IL6 (~18 μg) 1mM binder 6 PBS, 10% D2O 5 minutes Saturation Transfer Difference 25 μM IL6 (~18 μg) 1mM binder 6 PBS, 63% D2O 15 °C 8 minutes

Ligand Observe - 700 MHz 1.7 mm Microcryoprobe

63% D2O, 15 C, 8 mins experiment time are optimised conditions for acquisition of STD and WaterLOGSY on the same sample with sufficient S/N.

Water Ligand Observe via Gradient SpectroscpY (WaterLOGSY)

• Dr. Matthew Goodwin

To screen 1000 fragments by STD and waterLOGSY on microcryoprobe:

• ~20 days, ~18 mg protein
• Compares to ~7 days ~62 mg on 5 mm

probe, running in 5 ligand mixtures, plus deconvolution experiments To screen 100 compounds by protein detect

15N-1H SOFAST on microcryoprobe:

• 4 days, ~180 μg protein
• Compares to ~1 days ~31 mg on 5 mm

probe

• Dr. Matthew Goodwin

Conclusion For Example

A series of options each offer incremental improvements in sensitivity and acquisition rate

• 1.7 mm Microcryoprobe
• 25% NUS with Compressed Sensing reconstruction
• Arg-Glu Low Conductivity Buffers
• Chelated PREs (e.g. Gadbutrol)
• Optimisation of SOFAST-HMQC ( <20k Da) and BEST TROSY (>20 kDa)

No ideal single solution – still need to compromise Likely to be protein and application specific

Where next?

GPRCs and lipodisqs (SMALPS) Selective labelling strategies IDPs by Solid State NMR

Ralph Adams, Rebecca J. Burnley, Chiara R. Valenzano, Omar Qureshi, Carl Doyle, Simon Lumb, Maria del Carmen Lopez, Robert Griffin, David McMillan, Richard D. Taylor, Chris Meier, Prashant Mori, Laura M. Griffin, Ulrich Wernery, Jörg Kinne, Stephen Rapecki, Terry S. Baker, Alastair D. G. Lawson, Michael Wright and Anna Ettorre

Antibody Assisted Drug Discovery Explore application of APSY for 3D NHCO fragment screening

Automated Projection SpectrocscopY (APSY)

Acknowledgements

Professor Matthew Crump

• Dr. Matthew Goodwin

Erik Landin Professor Mark Carr

• Dr. Lorna Waters
• Dr. Vaclav Veverka
• Dr. Frederick Muskett
• Dr. Christine Prosser

Harry Mackenzie

• Dr. Alistair Henry

Questions?

20