Structure Based Prediction of Influenza Drug-Resistant Mutations
BRYAN D. COX, PH.D.
EMORY UNIVERSITY, DEPARTMENT OF PEDIATRICS
LABORATORY OF BIOCHEMICAL PHARMACOLOGY
EMORY CENTER FOR AIDS RESEARCH
1
Baloxavir Marboxil (Xofluza)
2
Baloxavir Marboxil - Xofluza
(Roche, Shionogi)
Baloxavir Marboxil (Xofluza) Resistance
3
ILE38
Baloxavir Marboxil (Xofluza) Resistance
4
ILE38THR
30-50X
Decreased
Activity
Omoto, et. al, Sci Rep (2018) 8: 9633.
Baloxavir Marboxil (Xofluza) Resistance
5
ILE38THR
IF WE ONLY HAD THE BOUND STRUCTURES,
COULD WE PREDICT BALOXAVIR RESISTANCE MUTATIONS?
30-50X
Decreased
Activity
Omoto, et. al, Sci Rep (2018) 8: 9633.
Predicting Resistance: Central Hypothesis
6
Inhibitor InhibitorTight
Binding
WEAK
Binding
Wild-Type Enzyme Mutant Enzyme
I. Decrease Binding of Inhibitor
Hao, et. al, Drug Discovery Today (2012) 17: 1121-1126.
Predicting Resistance: Central Hypothesis
7
Inhibitor InhibitorTight
Binding
WEAK
Binding
Wild-Type Enzyme Mutant Enzyme
Tight
Binding
Wild-Type Enzyme
Substrate
Mutant Enzyme
SubstrateTight
Binding
I. Decrease Binding of Inhibitor
II. Maintain Affinity of Substrate
Hao, et. al, Drug Discovery Today (2012) 17: 1121-1126.
Predicting Resistance: Central HypothesisI. Decrease Binding of Inhibitor
II. Maintain Affinity of Substrate
III.Low Genetic Barrier
8
Inhibitor InhibitorTight
Binding
WEAK
Binding
Wild-Type Enzyme Mutant Enzyme
Wild-Type Enzyme
Substrate
Mutant Enzyme
Substrate
AUU → ACU
AUU → ACUTight
Binding
Tight
Binding
Single Nucleotide Substitution
Hao, et. al, Drug Discovery Today (2012) 17: 1121-1126.
Simulation of All I38 Mutations:OSPREY / K* Protein Design Algorithm
9
Sample Mutant
Low Energy
Rotamers
Prune by Sterics
and
Minimize Energy
Reeve, SM, et. al. Proc Nat Acad Sci (2015) 112(3): 749-754.
Gainza, P, et. al. Methods Enzymol (2013) 523: 87-107
Simulation of All I38 Mutations:OSPREY / K* Protein Design Algorithm
10
Sample Mutant
Low Energy
Rotamers
Prune by Sterics
and
Minimize Energy0
2
4
6
8
10
12
14
LO
G[K
* S
CO
RE
]More Favorable
Complex Energy
Reeve, SM, et. al. Proc Nat Acad Sci (2015) 112(3): 749-754.
Gainza, P, et. al. Methods Enzymol (2013) 523: 87-107
Simulation of All I38 Mutations:Energy Relative to Wild-Type
11
-14
-12
-10
-8
-6
-4
-2
0
2
4
ΔL
OG
[K*
SC
OR
E]
Improved Binding of BAL
USEFUL TO PREDICT ACTIVITY
AGAINST VARIANT STRAINS
Decreased Binding of BAL
Model of BAL Active Species
Bound to Flu A H1N1 PA
-14
-12
-10
-8
-6
-4
-2
0
2
4
0 10 20 30 40 50
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Simulation of All I38 Mutations:Considering Native Substrate Affinity
12
Improved Binding of
Native Substrate
Improved Binding of BAL
Decreased Binding of BAL
Model of Di-Nucleotide Substrate
Bound to Flu A H1N1 PA
Simulation of All I38 Mutations:Accessible by Low Genetic Barrier
13
-14
-12
-10
-8
-6
-4
-2
0
2
4
0 10 20 30 40 50
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Improved Binding of BAL
SINGLE NUCLEOTIDE MUTATIONS
Improved Binding of
Native Substrate
Decreased Binding of BAL
Model of Di-Nucleotide Substrate
Bound to Flu A H1N1 PA
Model of Di-Nucleotide Substrate
Bound to Flu A H1N1 PA
-7
-6
-5
-4
-3
-2
-1
0
1
38 38,5 39 39,5 40 40,5
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Simulation of All I38 Mutations:Single-Nucleotide Accessible Mutations
14
-14
-12
-10
-8
-6
-4
-2
0
2
4
0 10 20 30 40 50
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Improved Binding of BAL
I38T-Decrease BAL Binding
-High Substrate Binding
-Accessible by Single
Nucleotide Substitution
I38S
I38FImproved Binding of
Native Substrate
Decreased Binding of BAL
-7
-6
-5
-4
-3
-2
-1
0
1
38 38,5 39 39,5 40 40,5
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Simulation of All I38 Mutations:Single-Nucleotide Accessible Mutations
15
-14
-12
-10
-8
-6
-4
-2
0
2
4
0 10 20 30 40 50
ΔL
OG
[K*
SC
OR
E]
LOG[K* SCORE] for Nucleotide Substrate
Improved Binding of
Native Substrate
Improved Binding of BAL
Decreased Binding of BAL
I38TI38S
I38F
Correctly Identified I38T As a Likely
Resistance Mutation for Baloxavir Marboxil
-Decrease BAL Binding
-High Substrate Binding
-Accessible by Single
Nucleotide Substitution
Predicting Resistance Mutations for Pimodivir - Influenza A PB2 Inhibitor
16
Model of Pimodivir Bound
To Avian H5N1 FluA PB2
VX-787 (Pimodivir)
Predicting Resistance Mutations for Pimodivir - Influenza A PB2 Inhibitor
17
46
Mu
tati
on
s
Reduced Inhibitor
Binding from WT Maintains Fitness
Binds Substrate Well
22
Low Genetic Barrier
Single Nucleotide
Substitution?
Emerged in
Clinical Trial71
0 A
cti
ve
-Sit
e M
uta
tio
ns
Res WT Resistant
323 PHE L M V
325 PHE V
339 LYS D E
357 HIS D E
404 PHE M Y
431 MET I
510 ASN I T Y
Incorporating Resistance Prediction into Drug Design Platform
18
Res WT Resistant
323 PHE L M V
325 PHE V
339 LYS D E
357 HIS D E
404 PHE M Y
431 MET I
510 ASN I T Y
COMPUTATIONAL DESIGN
In Silico Library Enumeration
Targeted Modifications
Virtual Screening/Docking
WT Potency Prediction (FEP)
Desired Resistance Profile?
NO
YES
SYNTHETIC CANDIDATE
RESISTANCE MUTATION PROFILE
Summary and ConclusionsDeveloped approach to de novo predict resistance mutations given only an inhibitor-bound structure◦ Successfully applied to other viruses (HIV-1, HBV, HCV and ZIKV)◦ Recapitulated resistance mutations for Baloxavir Marboxil and Pimodivir
Useful tool in antiviral drug design◦ Rational design of broadly active agents against all strain variations◦ Rational design of agents with improved resistance profiles◦ Rational design of drug combinations that ensure no single mutation
delivers cross-resistance to all agents
Future Directions and Other Applications◦ Allosteric mutations not in direct contact with inhibitor◦ Compensatory mutations that improve substrate binding◦ Antibody escape mutations
19
Acknowledgements
Emory LOBP◦ Raymond F. Schinazi, DSc, PhD
◦ Baek Kim, PhD
◦ Franck Amblard, PhD
◦ Keivan Zandi, PhD
Funding Sources
◦ Emory CFAR 2P30AI050409
20