Focused Virtual Screening Lead discovery in the Human Estrogen - - PowerPoint PPT Presentation

▶
focused virtual screening
SMART_READER_LITE
LIVE PREVIEW

Focused Virtual Screening Lead discovery in the Human Estrogen - - PowerPoint PPT Presentation

Mar 08, 2024 231 likes •546 views

Focused Virtual Screening Lead discovery in the Human Estrogen Receptor a presentation by Dr David Lloyd Trinity College Dublin Daylight EuroMUG 2004 1592 750 AD Biochemistry in TCD largest Department in Country Significant research

slide-1
SLIDE 1

Focused Virtual Screening

a presentation by Dr David Lloyd

Trinity College Dublin Daylight EuroMUG 2004

Lead discovery in the Human Estrogen Receptor

slide-2
SLIDE 2
slide-3
SLIDE 3

1592 750 AD Biochemistry in TCD – largest Department in Country Significant research output

slide-4
SLIDE 4

Centre for Structural Biology and Molecular Design building on PRTLI and SFI investment in structural biology building on PRTLI and SFI investment in structural biology

Molecular Design Group

Established 2004 Ireland’s first protein X-ray facility

slide-5
SLIDE 5

Integrated Drug Discovery

C h e m istry B io lo g y C

  • m

p u ta tio n

slide-6
SLIDE 6

Structure Based Design – looking in the ER

slide-7
SLIDE 7

Structure Based Design – looking in the ER

Curr Med Chem 2003, Frontiers Med Chem 2005 (in press)

slide-8
SLIDE 8

Significance of ER

  • Estrogens regulate cell growth, differentiation & development of reproductive tissues

in men and women.

  • Maintain bone density preventing osteoporosis.
  • Exerts anti-atherosclerotic effects which lowers Cholesterol levels.
  • Involved in many CNS effects (Parkinson's) and implicated in Alzheimer's.

ER as a target

  • 60% of primary breast cancers contain ER- alpha
  • Estrogens are mitogenic for ER-positive breast cancer cells.

Target: Target: Estrogen Estrogen Receptor Receptor

slide-9
SLIDE 9

Structure Based Design – Docking in Nuclear Receptors

Docking Algorithms Docking Algorithms Scoring Functions Scoring Functions

slide-10
SLIDE 10

Structure Based Design – Docking in Nuclear Receptors

FRED FRED FlexX FlexX Discover3 Discover3 eHits eHits In house In house PLP PLP ChemScore ChemScore CScore CScore PLC PLC

slide-11
SLIDE 11

Structure Based Design – looking in the ER

slide-12
SLIDE 12

Building on knowns : using receptor structural knowledge– semi-rational design

slide-13
SLIDE 13

Traditional scaffold hopping –human de novo rational design

J Med Chem 2001 Anti Cancer Drug Design 2001

slide-14
SLIDE 14
slide-15
SLIDE 15

Traditional scaffold hopping –human de novo rational design Computer-enhanced!! Benzoxepin antiestrogens

N aminoalkylation aminoalkylation BBr3 PhZnCl Pd(PPh3)4 boronic acids Pyr.HCl PyHBr3 H+ n-BuLi OMe Br 11 10 9 8 7 para CN (19) para Me (15) meta Me (16) para Cl (17) meta NO2 (18) para OMe (12)

  • rtho OMe (13)

meta OMe (14) R2 = 21-28 20 R1= N N N O N 6 7-11 5 Suzuki Route Heck Route 4 3 R2 O O N O R1O O HO Br O MeO O HO R2 2 O HO O MeO Br O MeO O HO OMe O O

slide-16
SLIDE 16

Computer-enhanced human de novo rational design

slide-17
SLIDE 17

Ortho-ring substitution is tolerated - meta is not - elcectic binding mode

Computer-enhanced human de novo rational design

J Med Chem 2004

slide-18
SLIDE 18

Haystack built from 880 ‘drug-like’ compounds from WDI 40 Cox-2 inhibitors 40 Estrogen Receptor Modulators 40 Histamine ‘modulators’

Active ‘needles’ introduced from a separate validated ligand set

Let the computer decide : Virtual Screening

slide-19
SLIDE 19

Virtual Screening

vHTS – Performance Measures – Validation

Enrichment =

hit rate observed in subset hit rate in database (random)

Enrichment Subset Size (%) 1 5 10 15 20 Ligands 10 50 100 150 200 Max Actives 10 40 40 40 40 Best Possible Value 25 20 10 6.7 5.0

e.g. 1% sampled = 10 compounds. Subset - 10 actives = hit rate of 10/10 = 1.0, Hitrate in database is 40/1000 = 0.04 : enrichment = 1 / 0.04 = 25

slide-20
SLIDE 20

Target Database Docking Protocol Rescoring Generation of Hits Active set of compounds for development

  • Remove waters & Calculate

centre of bound ligand.

  • Use multiple structures

PreProcessing

slide-21
SLIDE 21
  • Samples search space and generates a set of

binding poses for each ligand conformer.

  • Docked

positions have their respective hydrogen bond lengths optimized to allow for refinement of the final structure.

  • CF (Complementarity Function) evaluates fit
  • Ranks these modes/ligand positions
  • Provides a numerical score allowing for ‘hit’

identification

In-house docking protocol

slide-22
SLIDE 22

Actual Hits Retrieved

5 10 15 20 25 30 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 % Sample Database % Hits Retrieved F_Score G_Score PMF_Score D_Score ChemScore Xscore PLP_Score Fresno Screenscore Hammerhead Theoretical Maximum Hits Retreived

Chemscore performs best of scoring functions. Accounts for: Hydrogen bond contacts, Lipophilic contacts, entropic penalty. G-score focuses on hydrogen bonding interactions

  • nly for

example.

Getting it right – Scoring Functions

slide-23
SLIDE 23
  • 38.208

COX20080

  • 94.16

(Drug_401)

  • 39.083

ESTR0067

  • 94.24

(Drug_472)

  • 39.112

Drug_61

  • 94.43

(Drug_751)

  • 39.176

Drug_259

  • 94.5

(Drug_476)

  • 39.624

Drug_421

  • 94.53

(Drug_466)

  • 39.874

Drug_265

  • 95.72

(ESTR0085)

  • 40.123

ESTR0068

  • 96.07

(ESTR0024)

  • 40.154

Drug_315

  • 96.18

(ESTR0045)

  • 40.389

Drug_257

  • 96.33

(Drug_161)

  • 40.687

Drug_217

  • 96.64

(ESTR0043)

  • 40.991

Drug_249

  • 97.1

(Drug_163)

  • 41.031

Drug_416

  • 100.14

(ESTR0034)

  • 41.635

Drug_823

  • 101.03

(ESTR0046)

  • 41.647

Drug_219

  • 102.62

(Drug_474)

  • 42.169

Drug_440

  • 103.16

(Drug_160)

  • 42.209

Drug_353

  • 103.35

(Drug_633)

  • 42.485

Drug_344

  • 103.42

(Drug_159)

  • 44.427

ESTR0072

  • 104.85

(Drug_158)

  • 46.568

ESTR0079

  • 107.73

(ESTR0025) FRED CHEMSCORE Name_ID

Sybyl6.91

CHEMSCORE Name_ID

Getting it right – Scoring Functions

slide-24
SLIDE 24

Hit Retrieval

5 10 15 20 25 30 35 40 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 % Database Screened % Hit Retrieved In House Protocol FlexX FRED Best V alue

Getting it right – early method validation

slide-25
SLIDE 25

Getting it right – Ligand Pre Processing Virtual High-Throughput Screening <1 sec per compound – rigid/rigid system

Corina Corina Omega Omega Stergen Stergen Rubicon Rubicon QuacPac QuacPac

slide-26
SLIDE 26

0.47 40_RUBICON_LEVEL1 0.57 40_CATALYST_LEVEL1 0.48 40_OMEGA_LEVEL1 0.54 40_CORINA_LEVEL1 X 0.64 40_RUBICON_LEVEL2 0.62 40_CATALYST_LEVEL2 0.66 40_OMEGA_LEVEL2 0.64 40_CORINA_LEVEL2 Quac-X 0.69 40_RUBICON_LEVEL3 0.69 40_CATALYST_LEVEL3 0.71 40_OMEGA_LEVEL3 0.69 40_CORINA_LEVEL3 Quac-X-10 Confs 0.74 40_RUBICON_LEVEL4 0.63 40_CATALYST_LEVEL4 0.70 40_OMEGA_LEVEL4 0.64 40_CORINA_LEVEL4 Quac-Ster-X 0.74 40_RUBICON_LEVEL5 0.73 40_CATALYST_LEVEL5 0.69 40_OMEGA_LEVEL5 0.69 40_CORINA_LEVEL5 Quac-Ster-X-10 Confs

Getting it right – Ligand Pre Processing

Random screening – 40 actives in 1000 – each active returns a score – the bigger the difference between the active and inactive scores, the better the method Preprocessing can increase the cutoff value for ligand consideration – reducing the subset we must consider in order to find our active ligands.

slide-27
SLIDE 27

Getting it right – Ligand Pre Processing

6.5 8.75 12.5 16.25 20 LEVEL1_RUBICON 6 7.5 9.17 12.5 17.5 LEVEL1_CATALYST 7 8.75 10 13.75 22.5 LEVEL1_OMEGA 10 11.88 15 16.25 22.5 LEVEL1_CORINA

Avg enrichment

4 3 2 1 Subset % 11.5 13.13 14.2 13.75 10 LEVEL2_RUBICON 5 5.625 6.66 8.75 15 LEVEL2_CATALYST 9.5 11.25 14.2 18.75 22.5 LEVEL2_OMEGA 7.5 11.25 18.3 20 25 LEVEL2_CORINA 9 10.625 14.16 18.75 22.5 LEVEL3_RUBICON 9 10.625 13.33 20 25 LEVEL3_CATALYST 9.5 11.25 15 21.25 25 LEVEL3_OMEGA 9.5 11.875 15 21.25 25 LEVEL3_CORINA 11 13.75 14.17 18.75 22.5 LEVEL4_RUBICON 6.5 6.875 8.33 10 17.5 LEVEL4_CATALYST 9.5 10.625 13.33 16.25 20 LEVEL4_OMEGA 9.5 11.25 15 21.25 25 LEVEL4_CORINA 12 13.75 16.66 22.5 25 LEVEL5_RUBICON 9 11.25 13.33 18.75 25 LEVEL5_OMEGA 8 9.375 14.17 20 25 LEVEL5_OMEGA 9.5 11.25 15 21.25 25 LEVEL5_CORINA

slide-28
SLIDE 28

18 purchased and assayed

Does it really work ?– Validate, Validate, Validate

Compound Number IC50 in MCF-7 MTT

MDG-ER-001 8.23E-07 MDG-ER-002 8.00E-06 MDG-ER-003 2.02E-05 MDG-ER-004 5.59E-04 MDG-ER-005 6.06E-04 TAMOXIFEN 5.51E-06

Screen ligands, prepare ranked hitlist cluster hits – 20 clusters 5 Hits 5 Hits µ µ µ µ µ µ µ µm range m range 4 Chemical Classes 4 Chemical Classes 3 novel Chemotypes 3 novel Chemotypes MW 450 MW 450-

  • 550

550 LOGP 4.8 LOGP 4.8-

  • 6.5

6.5

slide-29
SLIDE 29

What else do we need? Familial scoring functions Validation Validation Validation Flexible systems – dynamics in docking Chemical intelligence in fragment assembly Tiered Discovery – integration of technologies System Simulation

slide-30
SLIDE 30

Acknowledgements

The ER collaborators Dr Mary Meegan – School of Pharmacy TCD Dr Vladimir Sobolev – Weismann Institute, Israel Prof James Sexton – Trinity Centre for High Performance Computing Prof Clive Williams – Biochemistry TCD Dr Daniela Zisterer – Biochemistry TCD Dr Amir Khan – Biochemistry TCD The workers The facilitators Andy Knox Dermot Frost Yidong Yang Giorgio Carta Valeria Onnis Georgia Golfis