Integrating Microflow NMR into Fragment-based Drug Discovery Daniel S. Sem Chemical Proteomics Facility at Marquette (CPFM) Department of Chemistry Marquette University Protasis Webinar 7/15/08

Integrating Microflow NMR into Fragment- based Drug Discovery Research Focus: Drug discovery using NMR Background on Fragment-based Drug Discovery NMR Equipment in the CPFM Flow Probe Applications – Routine quality control of compound collections – Protein 1 H- 15 N HSQC screening of labeled proteins – Fragment-based screening using STD (saturation transfer difference) NMR

New Drug Design Paradigm: Fragment-Assembly San Diego, California Initially funded June, 1999 Raised $42.5 M in A and B rounds; $15 M in C round 50+ employees Companies using fragment assembly approaches:  Advanced Medicines / Theravance  Sunesis (Thiol tethering)  Structural GenomiX  Abbott (SAR by NMR)  Vertex (SHAPES) Triad Therapeutics And from a Chemical Proteomics slant: Triad Therapeutics

1.Reviewed by: Pellecchia, Sem & Wuthrich (2002) Nat. Rev. Drug Disc. 1, Shuker, Hajduk, Meadows & Fesik (1996) Science 274, Fejzo et al. (1999) Chem. Biol. 6, Modular Drug Design / Fragment Assembly 1 SAR by NMR 2 SHAPES 3

Reviewed by: Pellecchia, Sem & Wuthrich (2002) Nat. Rev. Drug Disc 1, Proof of concept: Sem et al. (2004) Chemistry and Biology 11, 185. Combinatorial Library for Dehydrogenases Chemical inhibitors discovered across a gene family (dehydrogenases) Oxidoreductase-1 Oxidoreductase-2 Oxidoreductase-N Common Variable scaffold

LDH DHPRDOXPR 55  M 26  M >50  M 42 nM> 50  M10  M 12  M> 25  M202 nM 620 nM100 nM 7.9  M proteomic leverage target specificity (IC 50 ) Screen NMR-designed Library for Inhibitors Specific for Target versus Antitarget: Assay results Sem et al. (2004) Chemistry and Biology 11, 185. TB TargetMalaria Target scaffold Sem et al. (2004) Chemistry and Biology 11, 185. TB targetMalaria target

Triad Technology Platform – Internal use only – Proprietary NMR-based drug design platform to generate drug leads – Chemicals, proteins, software, databases, methods Proof of Principle completed (Series B, $30M) Leads for infectious disease targets Dual Business Strategy (Drug Discovery) – Internal drug discovery & licensing early-stage leads – Evolved into late-stage licensing (IND candidates) Triad ceased operations on 3/19/04: Drug licensing business model not practical

Drug Discovery in Academics Can focus on developing enabling methods rather than drug leads. The longer view … Focus on diseases with smaller markets; third world diseases. The CPFM has resources to aid in drug discovery: compounds; databases; software; NMR screening capability – automation & labeled probes.

NMR Equipment in the CPFM 600 MHz Varian NMR System  Cryogenic probe ( 1 H/ 2 H/{ 13 C}/{ 15 N})  z-axis gradients  4 channels 300 MHz Varian NMR System  Broadband probe  z-axis gradients  60 sample change

NMR Equipment in the CPFM 400 MHz Varian NMR System  x,y,z-axis gradients (imaging capability)  2 channels  BB and inverse probes  Protasis CapNMR Microflowprobe:  TXI triple resonance, 1H/2H/{ 13 C}/{ 15 N} detection  z-gradient, variable temp., 10 uL flowcell  Automated sample introduction using LEAP Technologies (CTC Analytics) liquid handler  Automation managed via Protasis One-Minute NMR (OMNMR) software

NMR-based Drug Discovery at Marquette’s CPFM Focus on developing new methods Blending: chemistry, NMR screening, informatics Integrating use of microflow NMR Focus on infectious disease

Could we discover a new version of this CR-based biligand more easily? a)Avoiding extensive synthesis (for linking)? b)Using microflow NMR (speed; conserve samples)? Strategy: Combine thiol tethering and STD-based screening, using a Flow NMR platform CR = catechol rhodanine Privileged scaffold

CF-STD NMR 1,2 : Cofactor fingerprinting with saturation-transfer-difference NMR STD NMR 3 Cofactor structures 1.Stockman & Dalvit (2002) Prog. NMR Spectr. 41, Yao & Sem (2005) FEBS Lett., 579, Mayer & Meyer (1999) Angew. Chem. Int. Ed. 38, STD-based screening (STD = saturation transfer difference)

CF-STD NMR 1,2 : Cofactor fingerprinting with saturation-transfer-difference NMR STD NMR 3 1.Stockman & Dalvit (2002) Prog. NMR Spectr. 41, Yao & Sem (2005) FEBS Lett., 579, Mayer & Meyer (1999) Angew. Chem. Int. Ed. 38, MHz; 25  M protein; 1mM cofactors STD: PKA + cAMP, cCMP, cGMP STD: RSP2 + cAMP, cCMP, cGMP 1D: cAMP, cCMP, cGMP

Flow Probe Applications

Using Flow Probe for HSQC Experiments  To screen for folding conditions (ex. structural proteomics)  To screen for fragment binding (ex. SAR by NMR) Our model protein: GB1 (IgG binding domain from protein G) Well studied (ex. Frank et al. (2002) NSB 9, )

Using Flow Probe for HSQC Experiments  Our model protein: GB1 (IgG binding domain from protein G; 56 AA)  ~1 mM, pH 7  Spectra taken on the 400 MHz flow probe (10 uL sample volume)  Acquisition time varied 2 hrs.5 hrs.10 hrs. Main advantage: automation and low sample consumption

100 uM DHPR + ligand (NAD + ): 10 mM NAD + 20 mM NAD + 40 mM NAD + 80 mM NAD + STD based screening with our drug target: DHPR (Dihydrodipicolinate reductase) Optimizing concentrations for flow-based screening (generally, we need > 50 uM protein, and high ligand concentration)  This requires use of reporter ligands to detect binding!  Of course, [competitor] > [enzyme target] Acquisition time = 47 minutes (96-well plate in < 4 days; this might be a secondary assay) Samples in D 2 O, 20 mM K-phosphate, pH 7.6, 298K

100 uM DHPR + 10 mM PDC + ? : 10 mM PDC STD based screening with our drug target: DHPR (Dihydrodipicolinate reductase) Optimizing concentrations for flow-based screening – binding 2 ligands (generally, we need > 50 uM protein, and high ligand concentration) Acquisition time = 47 minutes Samples in D 2 O, 20 mM K-phosphate, pH 7.6, 298K 10 mM PDC + 10 mM NAD + 10 mM PDC + 10 mM NADH

What’s going on? Flow-probe-based STD screening sees NAD but not NADH  NADH binds too tightly NADH seemed to increase the PDC STD effect? They bind to different sites, so possible synergy? Follow-up titration (this one is at 600 MHz w/cryoprobe):

Erlanson et al. (2000) PNAS 15, Erlanson et al. (2003) Nat. Biot. 21, Erlanson et al. (2004) Curr. Opin. Chem. Biol. 11, Next: combine STD-based screening w/ thiol-tethering Search for fragments that bind in cofactor (NAD or CRAA) and substrate (PDC) sites {CRAA = catechol rhodanine acetic acid} What is thiol tethering?  Bring weak binding fragments together (link) using disulfide bonds  Pioneered by Erlanson, Wells and other at Sunesis

Next: combine STD-based screening w/ thiol-tethering Search for fragments that bind in cofactor (NAD or CRAA) and substrate (PDC) sites Our approach  Bring weak 2 weak binding thiol-containing fragments together in the cofactor and substrate sites of DHPR, then link them later  Start with a first fragment that binds (CRAA) and screen for the second  Detect binding based on competition STD and reporter ligands (NAD, PDC) CRAA

Relative STD for 99 thiols (screened in pools of 5) Flow probe, using 10 mM PDC as a reporter (1 mM thiol; 100 uM DHPR; 45 min acquisition) Thiol Fragment Database: (99 thiols) Use STD Screening to Identify Thiol Fragments that Fit in the PDC Site

Relative STD for 99 thiols (screened in pools of 5) Flow probe, using 10 mM PDC as a reporter (1 mM thiol; 100 uM DHPR; 45 min acquisition) Why do some thiol fragments cause an increase in the PDC STD signal? Perhaps binding at other sites in the tetramer. PDC

STD with 400 MHz microflowprobe, 1 hr. acquisition time => 25% decrease in TNB STD signal due to PDC 20 mM TNB, 200 uM DHPR 20 mM TNB, 200 uM DHPR + 4 mM PDC Discovery that TNB (5-thio-2-nitrobenzoic acid) binds in the PDC site

Proof that PDC and TNB (a thiol fragment) occupy the same site => 600 MHz Competition of PDC (varied) against 2 mM TNB (reporter) (100 uM DHPR) Competition of TNB (varied) against 2 mM PDC (reporter) (100 uM DHPR) Note: STD doesn’t go to zero - perhaps because there are 4 active sites that are not equivalent?

A New Strategy: In situ thiol tethering and competition STD screening a same time The Goal:  Screen various thiols (RS - ) to see which can form a higher affinity biligand, blocking both sites (thereby decreasing STD signals for NAD and PDC reporters)  In this way, we discovered a biligand with TNB (5-thio-2-nitrobenzoic acid)

Results of in situ thiol tethering / competitive STD

Synthesis of the Thiol-Tethered Biligand (to verify in situ hit)

Secondary Assay:  In-gel binding to the colored CRAA2-TNB biligand

Secondary Assay:  Steady-state inhibition: CRAA2-TNB biligand is competitive vs. NADH (K i < 7+2  M)

Acknowledgements Graduate Students: Aurora Costache Huili Yao Xia Ge NMR Facility Manager: Sheng Cai, Ph.D. Fragment database, created with SciTegic software Chemistry Dept.  American Heart Association  Biomedical Technology Alliance  NIH (600 MHz spectrometer)  Marquette University