Bio-Layer Interferometry Technology Applied to Fragment ... · • Targeting Low-Druggability Bromodomains: Fragment Based Screening and Inhibitor Design Against the BAZ2B Bromodomain.

Alessio Bortoluzzi

Alessio Ciulli Group

University of Dundee

22nd April 2015

Targeting Protein-Protein Interactions (PPIs)

Ubiquitin Field Chromatin Field

• A bump-and-hole approach to engineer controlled selectivity of

BET bromodomain chemical probes.

Science 2014, 346 (6209), 638

• Targeting Low-Druggability Bromodomains: Fragment Based

Screening and Inhibitor Design Against the BAZ2B Bromodomain.

J. Med. Chem. 2013, 56 (24), 10183

• ………………

• Structure-Guided Design and Optimization of Small Molecules

Targeting the PPI between the E3 Ubiquitin Ligase VHL and the

HIF Alpha Subunit with In Vitro Nanomolar Affinities.

J. Med Chem. 2014, 57 (20), 8657.

• Biophysical studies on interactions and assembly of full-size E3

ubiquitin ligase: SOCS2:ElonginBC:Cullin5:Rbx2.

J.Biol.Chem.2015, 209(7), 4178.

• ………………

Why Target Protein-Protein Interactions?

Most of human proteins do not work on their own but in complex with other

protein(s), PPIs are central to all the biological processes and often dysregulated

in disease – the possibility to modulate PPIs holds enormous therapeutic

potential.

Nat. Rev. Drug Discov. 2002 1(9), 727

Targeting Protein-Protein Interactions by Fragment-Based Drug Discovery (FBDD)

PPI interfaces are generally flat and large (~1,000-2,000 A2) compared to the deep pockets that typically

bind small molecules (~300-500 A2) and unlike enzymes or GPCRs do not offer starting small molecules

for the drug discovery process. (Chem. Biol. 2014 21(9), 1102)

PPIs are notoriously challenging targets and traditional approaches such High Throughput Screening

(HTS) often fail.

HTS

FBDDFragments are low-molecular-

weight molecules (~<300 Da)

with low chemical complexity.(Nat. Protoc. 2013 8(11), 2309)

Adapted from: Top. Curr. Chem. 2012 317, 145

Targeting Protein-Protein Interactions by Fragment-Based Drug Discovery (FBDD)

Nat. Protoc. 2013 8(11), 2309

Selected clinical-stage compounds originating

from fragment-based lead discovery

Nat. Rev. Drug Discov. 2013 12(1), 5

Advantages and limitations of FBDD and HTS

Adapted from: Top. Curr. Chem. 2012 317, 145

How to Screen for Fragments?

Fragment screening methods – Pool result:

Erlanson, D. (Ed.) Practical Fragments 2011

Why not BLI?

• Label free technique

• Provides information on

the binding kinetics

• Versatile

• Good throughput

• ……

• Low MW limit of

detection (~150 Da)

• Run a screen on a 140-fragments library in triplicate on

eIF4E: 8/9 hits found in all 3

repetitions.

JNK-1 28/34: hits found in all 3

repetitions.

• eIF4E screened against larger library – 6500 fragments

– and of the selected hits 50% were reconfirmed in a

second screen.

• eIF4E was screened in a biochemical assay and a 52%

hit overlap was observed with the hits from BLI.

Conclusion: BLI is suitable for

small molecules characterization

and fragment screening

Bio-Layer Interferometry (BLI) Technology for Fragment Screening

Fragment Screening Approach

PPI Target Fragment library

Differential Scanning

Fluorimetry (DSF)

Bio-Layer

Interferometry (BLI)

1D Ligand Observed NMR

(CPMG, WaterLOGSY and STD)

1° Step

Minimize the number

of false negative hits

2° Step

Remove false positive

hits

Overview

- Description of the techniques used in the fragment screening cascade (DSF, BLI

and NMR).

- Analysis of the outcome of three fragment screenings.

- Conclusions from our experience in using BLI as fragment screening tool.

Differential Scanning Fluorimetry (DSF) Screen

Binding of a compound to a protein will result in an increase in the melting temperature (Tm).

Fluorescent dye – typically SYPRO

Orange – that is quenched in

aqueous solution but highly

fluorescent in no-polar environment

(e.g. hydrophobic patches of

unfolded proteins)

Adapted from: Nat. Protoc. 2007 2(9), 2212

Target protein

Typical Tm shift induced by fragments is of 0.5-2 °C (Nat. Protoc. 2013 8(11), 2309)

A fragment is considered a hit if induces a ΔTm > 2 x s.d. sample control.

Bio-Layer Interferometry (BLI) Screen at Single Point Concentration

Target protein is biotinylated in a ~1:1 molar ratio using commercially available kits.

Target

Protein

B= Linker

B = Biotin

Super Streptavidin

(SSA) Biosensor

Loading

10-50 μg/mL

Sensor Tray Sample Plate (384 wells)

16 SSA Biosensor loaded

with B-Protein and

quenched with Biocytin

16 Reference SSA Biosensors

– blocked with 1 μg/mL of

Biocytin

200 = Fragment at 200 μM

B = Buffer

= Positive control – if available

• 154 fragments per plate

• 90 minutes per plate

• 1200 fragments screened in 12 hours

Reference

wells

Bio-Layer Interferometry (BLI) Screen at Single Point Concentration

Instrument: Octet RED 384

Typical cycles are: Baseline – 60 seconds; Association – 60 seconds and Dissociation – 60 seconds

Experiment performed at 25 °CRaw Data

Reference

Sensors

Protein-Loaded

Sensors

Data are processed by double reference subtraction to remove drift and well-to-well artifacts:

Threshold: median

plus 3 X robust s.d.

Hit rate is target

dependent – typically

20-100 fragments (out

of ~1200) are selected

for follow up

Bio-Layer Interferometry (BLI) Screen – Hit Confirmation

Selected fragments are rescreened at 6 points concentration – 3 fold dilution from 500 μM.

Sensor Tray Sample plate (384 wells)

Reference

wells

Raw Data

• 28 fragment per plate

• Typical cycles are:

Baseline – 60 seconds;

Association – 60 seconds;

Dissociation – 90 seconds

• 3 hours and 15 minutes per

plate.

Bio-Layer Interferometry (BLI) Screen – Hit Confirmation

Data are processed by double reference subtraction to remove drift and well-to-well

artifacts and visually inspected to select fragments that give a concentration dependent

response and to discard fragments with not ideal behavior :

500 μM

167 μM

55 μM

18 μM

6&2 μM

Concentration dependent response

Association step reach saturation

“Reasonable” dissociation rate

✕

✕

✓

~ 50% of the fragments selected during the single

point concentration screening are discarded

Hit Validation by 1D-Ligand Observed NMR Experiments

CPMG WaterLOGSY STD

Reference Spectrum – Fragment 1.5 mM

Fragment 1.5 mM + Target Protein 30 μM

Small molecules have a relatively long

relaxation time (T2: ~1 s).

Protein have a shorter relaxation time

(T2: ~1-50 ms).

100-400 ms delay introduced before

accquisition

Bulk water’s magnetization is

disturbed and transferred to:

- Free ligand in solution

- Ligand bound to the protein

NOE of opposite sign

Spectrum A: saturation pulse ( 0.5-1

ppm, methyl region). Protein-selective

resonances saturation and transfer to

bound ligand.

Spectrum B: saturation pulse off-

resonance (80 ppm).

In A only bound ligand is saturated – the

difference spectrum (B-A) will show

only the resonances of the bound ligandFragments that show binding in at least one

of these experiment is considered a hit.

Outcome of Three Fragment Screenings

Target protein Fragment library

Differential Scanning

Fluorimetry (DSF)

Bio-Layer

Interferometry (BLI)

1D Ligand Observed NMR

(CPMG, WaterLOGSY and STD)

Target :

A, B and C

Fragment library

(~1200)

DSF hit rate:

1.4%, 1.6% and 2%

BLI hit rate:

0.8%, 1.6% and 3.5%

Confirmed hits by NMR

62.5%, 69.2% and 33.3%

Analysis of BLI and DSF Performances as Screening Techniques

DSF17

DSF17

Target A

BLI10

BLI10

3

DSF19

DSF19

Target B

BLI20

BLI20

0

DSF24

DSF24

Target C

BLI42

BLI42

3

DSF and BLI show very little hit overlap

Running the screening in parallel with two orthogonal

techniques reduces the number of false negative hits

Analysis of BLI and DSF Performances as Screening Techniques

NMR Validate Hits for DSF

11 (64.7%)

16 (84.2%)

7 (29.2%)

11.3 (59.3%)

NMR Validate Hits for BLI

6 (60.0%)

11 (55.0%)

15 (35.7%)

10.6 (50.2%)

NMR validated overlapping hits: 2 (66.7%)

NMR validated overlapping hits: 1 (33.3%)

NMR validated overlapping hits: 1.5 (50.0%)

Conclusions

BLI is an effective tool to identify genuine fragment hits.

Current experimental setting allow to perform a single-point concentration

step on ~1200 fragments plus a follow up 6-point concentration step using 15

28h of time machine and < 0.5 mg of protein

NMR-Validated hits

10.6 (50.2%)

BLI and DSF have similar hit rates but BLI has a higher propensity to select false positives

Average

NMR-Validated hits

10.6 (50.2%)NMR-Validated hits

11.3 (59.3%)

To reduce false negative hits is critical to

perform the primary screen with two

orthogonal techniques

Acknowledgments

Compound

Management Team

Manuel Blank

David Robinson

(Drug Discovery Unit)

Emil Bulatov

Federica Cettorino

Martina Casale

Alessio Bortoluzzi

Alessio Ciulli Group

University of Dundee

22nd April 2015