New approaches to traditional anti-mitotic chemotherapy:
The structure based drug design of the Eg5 inhibitor
NVP-BQS481, from computer to clinic
Paul Barsanti
Novartis Institutes for Biomedical Research, Emeryville, CA
[email protected]
17th SCI/RSC Medicinal Chemistry Symposium
Churchill College, Cambridge UK Sept 2013
Presentation outline
 Pyridone and Carboline hit-to-lead campaigns
• use of crystal structure information
 Evolution of the BQS481 series
• structure based drug design
• SAR/ medchem issues
• efficacy, pre-clinical profiling
2
Why is Eg5 an attractive target ?
Normal Mitosis
Cdc2/cyclinB
(ATP)
G2/M
Eg5-P
Eg5 Inhibition
3
a-tubulin (green)
Chromosome (blue)
Eg5
• Eg5 is a motor protein required
for chromatid separation
• Inhibition of Eg5 causes
monopolar spindle formation
• Unlike other anti-mitotics
(taxanes, epothilones & Vinca
alkaloids), inhibition of Eg5 does
Normal bipolar spindle
not affect microtubule stability
• Evidence that Eg5 inhibition only
targets dividing cells & thus
should not cause neurotoxicities,
unlike microtubule disruptors
Monopolar spindles
Figures modified from Mayer et al; Science 1999, 286, 971
Cleveland et al; Cancer cell 2005, 8, 7-12
Eg5 offers the possibility for allosteric inhibition
“The concerted movement of switch-2 and of the necklinker region is
the most significant conformational change of the KSP–ADP (Eg5-ADP)
complex when monastrol binds”
“This comparison reveals that the KSP (Eg5) protein assumes the locked
conformation when it is bound with ADP and monastrol”
4
OH
O
HN
S
O
N
H
Kuo et al; J. Mol. Biol. 2004, 335, 547-554
The Pyridones
O
O
N
Cl
N
N
O
O
N
N
N
NH2
O
N
N
NH2
O
N
SB-715992
IC50 = 0.003 M
GI50 = 0.003 M (Colo205)
= 0.026 M (HCT-15)
= 0.006 M (KB3.1)
= 0.83 M (KBV1)
clogP=5.4
IC50 = 0.008 M
GI50 = 0.002 M (Colo205)
= 0.59 M (HCT-15)
= 0.003 M (KB3.1)
= 6.7 M (KBV1)
IC50 = 0.008 M
GI50 = 0.005 M (Colo205)
= 0.57 M (HCT-15)
= 0.011 M (KB3.1)
= 49
M (KBV1)
clogP=4.0
clogP=4.4
Due to close structural similarities, pyridone series terminated for
IP concerns and potential for similar liabilities as SB-715992
5
NH2
The Tetra-hydro--carbolines
H3C
N
H
N
N
H
O
OH
N
O
OH
HTS Hit
IC50 = 2.5 M
6
IC50 =
GI50 =
=
=
=
=
0.18 M
0.34 M (Colo205)
0.28 M (MDA435)
0.92 M (HCT-15)
0.64 M (KB3.1)
1.0 M (KBV1)
Binding induced rearrangement of Eg5
Bound
Apo
L5 Loop
IC50 = 0.183 M
7
Superposition of THB-Carboline with Quinazolinone
Large pocket
opens acommodating
Benzyl when
L214 Shifts
L214
Each class
explores
different
interactions
with protein
Pocket changes
due to backbone
movement of L214
Quinazolinone
8
Superposition of THB-Carboline with Quinazolinone
Each class
explores
different
interactions
with protein
Pocket changes
due to backbone
movement of L214
Carboline
9
Tetra-hydro--carbolines: from hit to lead
IC50 =
GI50 =
=
=
=
=
10
0.18 M
0.34 M (Colo205)
0.28 M (MDA435)
0.92 M (HCT-15)
0.64 M (KB3.1)
1.0 M (KBV1)
IC50 =
GI50 =
=
=
=
=
0.02 M
0.13 M (Colo205)
0.08 M (MDA435)
>2 M (HCT-15)
0.27 M (KB3.1)
>20 M (KBV1)
IC50 =
GI50 =
=
=
=
=
2.1 M
2.2 M (Colo205)
2.2 M (MDA435)
2.8 M (HCT-15)
5.1 M (KB3.1)
6.8 M (KBV1)
Tetrahydro--carboline series terminated
The series was discontinued due to:
• Steep SAR
• Modeling based on crystal structure proved unpredictive
• High interdependence of phenolic OH/activity
• Major pGP issues
• Constrained IP
IC50 = 0.007 M
GI50 = 0.10 M (Colo205)
= 0.06 M (MDA435)
= >2
M (HCT-15)
= 0.24 M (KB3.1)
= 85
M (KBV1)
Barsanti et. al., Bioorg. Med Chem Lett, 2010, 20(1), 157-160
11
Open-chain Carboline analogs: opening the C-ring
IC50 = 0.025 M
(scored -35 kJ/mol)
(scored -39.9kJ/mol)
IC50 > 25 M
12
Morphing the THB-carboline series into a new phenyl
imidazole core
Open A
ring
N
N
H
Open both A
and C rings
NH2
O
N
N
H
H
N
NH2
O
OH
Modeling score -30.2 kJ/mol
IC50 Eg5 >25 M
IC50 Eg5 23 M
Transpose Ph
Cl
N
O
N
O
N
Pocket opens up
when L214 shifts
13
NH2
IC50 Eg5 0.106 M
Phenylimidazoles: crystal structure
P137
W127
N
N
O
N
NH2
R119
Y211
IC50= 0.225 M
E116
Resolution: 2.6 A
• benzyl and toluamide share
large hydrophobic pocket
• propyl amine extends into
solvent channel to make
the only H-bond, with Glu116
Modification of the toluamide region
Quite the
“ magic methyl”
Eg5 IC50= 0.009 M
Eg5 IC50= 0.063 M
15
Eg5 IC50= 1.1 M
Eg5 IC50= 0.056 M
Phenyl imidazole series not locked into toluamide
Unanticipated broad amide SAR observed
Eg5 IC50= 0.051 M
16
Eg5 IC50= 0.082 M
Eg5 IC50= 0.013 M
Why do the t-butyl and chloride modifications
increase potency?
N
N
O
N
NH2
IC50 = 0.238 M
IC50 = 0.014 M
IC50 = 0.063 M
why is such diversity now tolerated in the amide region?
17
Meta-chloride pushes amide moiety into
a new binding pocket
18
Green
IC50 = 0.238 M
Purple
IC50 = 0.072 M
t-Butyl has same effect as the meta-chloro phenyl
N
N
O
Cl
N
N
NH2
N
N
Green
IC50 = 0.072 M
19
N
O
R119
E116
N
N
N
Purple
IC50 = 0.016 M
NH2
Exploiting new binding mode
ARG119
R119
E116
20
IC50 = 0.003 M
GI50 = 0.026 M (HCT-116)
0.028 M (KB3.1)
0.423 M (KB8.5)
RF = 15
Cyclic constraints based on modelling &
crystallography addresses pGP
3.3
2.9
E116
21
R119
IC50 = 0.042 M
GI50 = 0.250 M (HCT-116)
0.217 M (KB3.1)
0.406 M (KB8.5)
RF = 1.9
New binding mode allows increased diversification
SB-715992
IC50 = 0.007 M
22
IC50 = 0.02 M
IC50 = 0.007 M
Addressing pGP through amine basicity
R
NH2
KB3.1
GI50
KB8.5
GI50
RF
KB3.1
GI50
KB8.5
GI50
RF
0.03
0.60
22
0.34
0.43
1.3
0.50
6.6
13
0.13
0.15
1.1
0.004
0.88
221
0.04
0.09
2.2
0.02
0.43
22
0.23
0.35
1.5
0.184
1.0
5.4
0.25
1.4
5.6
IC50 values in M
23
R
Attenuation of pKa modulates hERG inhibition
pKa 7.6
GI50 KB8.5 0.0005 M
hERG (PC) 27% inhib @ 4.9M
pKa 4.4
KB8.5 > 0.20 M
hERG (PC) 17% inhib@10M
pKa 6.0
GI50 KB8.5 0.008 M
hERG (PC) 25% inhib@10M
pKa 6.2
GI50 KB8.5 0.008 M
hERG (binding) >30M
Di-F Methyl and oxetane successfully attenuates hERG channel activity
ONC8Eg5, Paul Barsanti, May 6th 2008
24
Potency / Cyp3A4 inhibition dichotomy
ARG119
X-ray shows H-bond
requires eclipsed
F
N
methyl group stabilizes
eclipsed conformation
F
N
F
ARG119
Of Eg5
F
N
O
O
H
N
NH2
F
GI50 KB8.5 0.0005 M
25
N
predicts dimethyl
might be even better
N
N
F
NH 2
N
O
HO
N
F
GI50 KB8.5 0.0002 M
F
NH 2
O
HO
F
GI50 KB8.5 0.014 M
Potency / Cyp3A4 inhibition dichotomy
Fe
Fe
Fe
CYP3A4
CYP3A4
CYP3A4
X-ray shows H-bond
requires eclipsed
F
N
methyl group stabilizes
eclipsed conformation
F
N
F
ARG119 and Fe
Of Eg5
3A4
F
N
O
O
H
N
NH2
F
GI50 KB8.5 0.0005 M
CYP 3A4 (MDZ) 1 M
N
N
F
NH 2
N
O
HO
N
F
GI50 KB8.5 0.0002 M
CYP 3A4 (MDZ) 0.6 M
F
NH 2
O
HO
F
GI50 KB8.5 0.014 M
CYP 3A4 (MDZ) 0.6 M
increased potency from bis-H bond to ARG119 also occurs with CYP3A4
ONC8Eg5, Paul Barsanti, May 6th 2008
26
N
predicts dimethyl
might be even better
Steric bulk around carbonyl and ether unable to
resolve Cyp3A4 inhibition issue
IC50 Eg5
F
N
O
F
MeO
0.7 nM
GI50 KB 8.5
Rf
N
N
NH 2
100 nM
1.5
CYP3A4 (MDZ)
F
N
O
1 M
F
F
O
O
O
O
O
NH
N
N
O
all < 3M CYP3A4
27
N
NH 2
IC50 Eg5
0.7 nM
GI50 KB 8.5
Rf
1.7 nM
1.4
CYP3A4 (MDZ) 0.3 M
F
O
O
N
O
N
O
O
O
S
O
O
5-6 M CYP3A4
NVP-BQS481 selected as most balanced molecule
to move forward
NVP-BQS481
F
N
N
O
F
HO
N
NH 2
HCT-116 = 0.1 nM
KB 8.5 = 0.6 nM
KB 3.1 = 0.6 nM
RF
=1
Rat PK (i.v, 1.5 mg/kg)
T½
= 4.9 h
Vss
= 20 L/kg
Cl
= 73 mL/min/kg
AUC
= 0.7 M*h
F
 Exceptionally potent in cellular assays including those that overexpress p-GP
 Well established PK/PD relationship; strong dose response (mitotic arrest and apoptosis)
 Monopolar spindles observed in tumors establishing in vivo MOA
 Consistent iv PK across species
 No red flags from Safety
• >5 M in pharmacology safety panel
• hERG patch clamp;, 27% @ 5M ; no findings during dog telemetry
• CYP3A4 ~0.5M
• GLP rat and dog toxicology findings consistent with MOA (anti-mitotic)
28
Key Eg5 protein interactions for BQS481 series
Key Interactions:
PRO137
5 Hydrogen Bonding Interactions:
TYR211
• 5-fluoro with H- N of ALA218
• Carbonyl & methoxy to back
bone H-N of ARG119
• Fluorine of amine arm with N
of ARG221
• N of amine arm with O of GLU116
LEU214
ARG119
GLU116
ALA218
ARG221
Other Interactions:
• Benzyl moiety extends into
hydrophobic pocket formed by
PRO137, TRP127, TYR211
• 2-Fluoro in hydrophobic pocket
formed by LEU214, ILE136,
PRO137
• t-Butyl, amide & amine moieties
extend into a solvent accessible
channel
NVP-BQS481 efficacy in KB8.5 (pGP positive)
mouse xenograft model (q4dx3, i.v)
2500
1: Vehicle
3
3)
Mean
Tumor
Volume(mm
(mm
Mean
Tumor
Volume
)(+/-SE)
06P-298
Tumor
Volumes
Mean Mean
KB8.5KB8.5
Tumor
Volumes
2000
2: NVP-BQS481
CHIR583780 5 mg/kg
1500
3: NVP-BQS481
CHIR583780 7.5 mg/kg
1000
5: SB-715992
CHIR370625 15 mg/kg
500
7: Taxol 30 mg/kg SD
0
0
5
10
15
Days Post Staging
Dosing: q4d for 3 cycles (d0, d4, d8)
Days Post Dosing Initiation
NVP-BQS481 differentiates itself from SB-715992 in p-GP over expressing model
30
Hematological malignancies are particularly
sensitive to BQS481
• Cell lines derived from hematological malignancies are amongst the
most sensitive to BQS481
AML
Mean
GI50=
2.2 nM
*
CML
ALL
Multiple Myeloma
1.50
1.00
0.50
0.00
-0.50
-1.00
-1.50
* Log of GI50 value relative to the log of the mean GI50 value of the entire
NCI-60 panel plus the GI50 values of the hematological cell lines panel
31
NHL
HL
Sustained tumor regression by NVP-BQS481 i.v. q4d x 3 in
MV4;11 AML subcutaneous mouse xenograft model
2000
Vehicle
NVP-BQS481
NVP-BQS481
NVP BQS481
NVP-BQS481
1500
1000
500
0
*p < 0.05 - ANOVA,
Dunn's post-hoc test
*
0 10 20 30 40 50 60 70 80
Days Post Dosing Initiation
32
0.25 mg/kg
0.5
1 mg/kg
/k
1 mg/kg
6 CR, 3 PR
Body Weights
Body Weight (Mean % Initial)
Mean Tumor Volume (mm3)
Tumor Volumes
20%
15%
10%
5%
0%
-5%
-10%
-15%
-20%
0
10
20
30
40
50
60
70
80
Days Post Dosing Initiation
NVP-BQS481 is efficacious against disseminated
KMS-11-luc multiple myeloma in SCID-Bg mice
Vehicle
Day 10 post-implant
ventral view
0.25 mg/kg IV Q4Dx3
10
8
6
6
x10
4
2
1 mg/kg IV Q4Dx3
0.5 mg/kg IV Q4Dx3
0
Color Bar
Min = 30000
Max = 1e+07
33
Available data supported clinical development of
NVP-BQS481 in AML and Multiple Myeloma
 AML
• In vitro
- Leukemia cell lines (n=28) are the most sensitive; majority have GI50’s below 350 pM
- Circulating blasts from AML patients were tested in tissue culture with BQS481.
• Survival /colony formation assays demonstrated cytotoxic effects of BQS481 with GI50‘s = 120pM
to 400 pM
• In vivo
- BQS481 is superior in efficacy compared to AraC or daunorubicin in 3 AML Xenograft
Models (1 subcutaneous and 2 disseminated)
 MM
• In vitro
- MM cell lines are among the most sensitive cell lines to BQS481 with subnanomolar GI50
values as observed in AML
• In vivo
- BQS481 is superior in efficacy compared to Velcade in 2/3 MM Xenograft Models (1
subcutaneous and 1 disseminated)
34
Human PK well predicted by Allometric Scaling
Allometric scaling for Vss
Allometric scaling for CL
3
dog
Log Cl
2
rat
1.5
monkey
1
1.5
monkey
rat
1
Log Vss
2.5
dog
2
0.5
0
0.5
-0.5
mouse
0
-0.5
mouse
-1
-2
-1.5
-1
-0.5
0
0.5
Log Body Weight (kg)
1
-2
-1.5
-1
-0.5
0
0.5
1
Log We ight Body (kg)
• Predicted human PK parameters
• CL: 799 – 833 mL/min
• Vss: 908 mL
• T1/2: 10 - 13 hours
Human Data:
 Half-life of BQS481 ~ 11-12h at the 1.2 mg/m2 cohort
 Highest plasma levels of BQS481 always reached at end of the 30-min infusion
No drug accumulation between Days 1, 8 and 15 were observed
35
Acknowledgements
 Med Chem













Paul Barsanti
Weibo Wang
Cheng Hu
Savithri Ramurthy
Sharadha Subramanian
Rama Jain
Yue Pan
Wooseok Han
Yu Ding
Paul Renhowe
Dan Poon
Rustum Boyce
Annie Xia
 Comp Chem
 Eric Martin
 Structural Chem




Mark Knapp
Robert Elling
Vincent Le
Dirksen Bussiere
 Chem Prep Lab
 Markus Furegati
 Sandro Nocito
36
 Biology








David Duhl
Li Zhang
Hanne Merritt
Annette Walter
Laura Doyle
Michel Faure
Joseph Castillo
Wei Zhang
 Analytical Chem
 Weiping Jia
 Pharmacology




Tinya Abrams
Millicent Embry
Jose Lapitan
Sok Chey
 Protein Sciences
 Laura Tandeske
 Eric Fang
 Formulations/CPP
 Andrew Phimister
 Vijay Sethuraman
 Sudhakar Garad
 Safety







Nancy Turner
Tao Wang
Andrea Mason
Werner Classen
Mary Arnella
Daniel Bauer
Berengere Dumotier
 Comparative Med





Charles Vitt
Jason Hulse
Pedro Benitez
Mateo Guzman
Ivan Barajas
 MAP/ DMPK








Richard Zang
Hong Lu
Helen Gu
Song Lin
Keshi Wang
Elaine Ginn
Hua Huang
David Pearlstein
 Clinical/Development





Angelo Kaplan
Jeannie Shen
Jerry Huang
Ashwin Gollerkeri
Erika Zannou
 Bioinformatics
 Vivien Chan
 Karen Yu
 Patents
 Gerald Suh
 Mike Smith
 Project Management
 Jason Ravenel