visualising the molecular drivers behind drug resistance · visualising the molecular drivers...

1
Visualising the molecular drivers behind drug resistance The huge expense of developing a new drug can be wasted if natural muta7ons of amino acid residues in the targeted protein lead to a loss of drug binding affinity. Ra5onal drug design is a con5nual struggle, with evolu5on driving muta5ons that develop emerging drug resistance into widespread drug inefficacy. We have developed a new free energy method callled ProteinSwap, based on WaterSwap, that can provide drug designers with the insight needed to understand how protein muta5ons affect drug binding. This method allows the change in binding free energy of the drug associated with the muta7on of the protein to be calculated. In addi5on, as for other Swap-Based methods, this free energy change can be decomposed into components that can be used for visualis7on. These components allow the change in binding free energy to be understood in terms of changes in specific ligand-protein interac7ons, or changes in the solva5ng water network in the ac5ve site. This method allows drug designers to pro-ac5vely screen a new drug computa7onally against a range of likely protein mutants, thereby enabling them to get one step ahead of nature. Alterna5vely, it allows chemists to inves5gate the molecular basis for reduced binding affinity in known drug-resistant mutants of a protein. Maturos Malaisree and Christopher J. Woods, School of Chemistry, University of Bristol, Bristol, BS8 1TS, United Kingdom ProteinSwap is a new free energy method that can be used to explore drug resistance. Residue-based decomposi5ons provide a useful graphical visualisa5on. These enable the effect of protein muta5on on drug binding to be ra5onalised at the molecular level. ProteinSwap will soon be available as an open source command-line tool as part of Sire (hLps://siremol.org). A graphical interface will be made available by Cresset via Flare (hLps://www.cresset-group.com/flare). References (1) Woods, CJ, Malaisree, M, Hannongbua, S & Mulholland, AJ, “A water-swap reac5on coordinate for the calcula5on of absolute protein-ligand binding free energies”, Journal of Chemical Physics, vol 134, 2011 (2) Woods, CJ, Malaisree, M, Michel, J, Long, B, McIntosh-Smith, S & Mulholland, AJ, “Rapid decomposi5on and visualisa5on of protein-ligand binding free energies by residue and by water”, Faraday Discussions, vol 169, 2014 (3) Baek, YH, et al, “Profiling and Characteriza5on of Influenza Virus N1 Strains Poten5ally Resistant to Mul5ple Neuraminidase Inhibitors” , Journal of Virology, vol 89, 2015 Acknowledgements We thank the EPSRC and Cresset for funding via an EPSRC Impact Accelera5on Award. CW is an EPSRC Research Socware Engineering Fellow (EP/N018591/1). We thank the Advanced Compu5ng Research Centre at UoB for providing early access to the new BlueCrystal Phase 4 cluster (hfp://www.acrc.bris.ac.uk) λ=0.0 λ=0.2 λ=0.5 λ=0.8 λ=1.0 Wildtype Mutant E (λ) = (1 - λ) E ligand:wildtype + E water:mutant + λ E water:wildtype + E ligand:mutant ProteinSwap, based on WaterSwap 1,2 , calculates the change in protein-ligand binding free energy caused by protein muta5on. A λ-coordinate swaps the ligand from the wildtype protein to the mutant. Simultaneously, a cluster of water molecules is swapped from the mutant to the wildtype. Monte Carlo simula5ons at each value of λ are used to evaluate the free energy change using thermodynamic integra5on. The result is the free energy preference of the ligand for either the wildtype or the mutant protein. Valida7on, was performed by calcula5ng the ProteinSwap free energies of oseltamivir and peramivir ligands in wildtype and mutant forms of influenza neuraminidase (H1N1-2009). Logical tests, such as swapping the ligand between two iden5cal proteins, showed the method was sound. Reverse tests, comparing wildtypeàmutant against mutantàwildtype, showed the ProteinSwap free energies were consistent. Closure tests showed free energies closed. Finally, comparison against known experimental binding affini7es showed that ProteinSwap could corerctly iden5fy known drug-resistant muta5ons. -8 -6 -4 -2 0 2 4 6 8 0 0.2 0.4 0.6 0.8 1 Free Energy / kcal mol -1 -0.1±0.6 kcal mol -1 Peramivir: WildtypeàWildtype Logical Test - should be zero -16 -12 -8 -4 0 4 8 12 16 0 0.2 0.4 0.6 0.8 1 Free Energy / kcal mol -1 -16 -12 -8 -4 0 4 8 12 16 0 0.2 0.4 0.6 0.8 1 Free Energy / kcal mol -1 8.3±0.8 kcal mol -1 -7.7±1.2 kcal mol -1 Peramivir: WildtypeàR292K R292KàWildtype Reverse Test - should be equal and opposite Closure Test paths equal Wildtype R292K R152K E119G 8.3±0.8 kcal mol -1 Peramivir: Wildtype, R292K, R152K, E119G Comparison to Expt. Green Shows Agreement Oseltamivir Peramivir Experimental Resistance 3 Wildtype ≈ E119G < R152K < R292K Wildtype < R152K < R292K << E119G ProteinSwap Free Energy / kcal mol -1 0.0 ≈ 0.6 < 5.1 < 8.4 0.0 < 3.0 < 8.3 !! 3.0 Residue-based decomposi7on, allows ProteinSwap free energies to be broken down to residue-based components. Each residue can be colored as preferring binding ligand (blue) or preferring binding water (red). The difference of these between the mutant and wildtype shows how muta5on weakens binding for some residues (red), while strengthening binding to others (blue). -30 -25 -20 -15 -10 -5 0 5 10 R118 E119 D151 R152 R156 S179 D198 I222 R274 E277 S246 E276 E277 R292 N294 G344 K346 R371 Free Energy / kcal mol -1 -30 -25 -20 -15 -10 -5 0 5 10 R118 E119 D151 R152 R156 S179 D198 I222 R274 E277 S246 E276 E277 K292 N294 G344 K346 R371 Free Energy / kcal mol -1 -30 -25 -20 -15 -10 -5 0 5 10 R118 E119 D151 R152 R156 S179 D198 I222 R274 E277 S246 E276 E277 R292K N294 G344 K346 R371 Free Energy / kcal mol -1 - = Oseltamivir:Wildtype Oseltamivir:R292K Difference R292 R371 R118 R118 R371 K292 R292K R371 R118 E residue wildtype (λ) = (1 - λ) E residue ligand:wildtype + λ E residue water:wildtype E residue mutant (λ) = (1 - λ) E residue water:mutant + λ E residue ligand:mutant Conclusion

Upload: others

Post on 22-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Visualising the molecular drivers behind drug resistance · Visualising the molecular drivers behind drug resistance The huge expense of developing a new drug can be wasted if natural

Visualising the molecular drivers behind drug resistance Thehugeexpenseofdevelopinganewdrugcanbewastedifnaturalmuta7onsofaminoacidresiduesinthetargetedproteinleadtoalossofdrugbindingaffinity.Ra5onaldrugdesignisacon5nualstruggle,withevolu5ondrivingmuta5onsthatdevelopemergingdrugresistanceintowidespreaddruginefficacy.WehavedevelopedanewfreeenergymethodcallledProteinSwap,basedonWaterSwap,thatcanprovidedrugdesignerswiththeinsightneededtounderstandhowproteinmuta5onsaffectdrugbinding.Thismethodallowsthechangeinbindingfreeenergyofthedrugassociatedwiththemuta7onoftheproteintobecalculated.Inaddi5on,asforotherSwap-Basedmethods,thisfreeenergychangecanbedecomposedintocomponentsthatcanbeusedforvisualis7on.Thesecomponentsallowthechangeinbindingfreeenergytobeunderstoodintermsofchangesinspecificligand-proteininterac7ons,orchangesinthesolva5ngwaternetworkintheac5vesite.Thismethodallowsdrugdesignerstopro-ac5velyscreenanewdrugcomputa7onallyagainstarangeoflikelyproteinmutants,therebyenablingthemtogetonestepaheadofnature.Alterna5vely,itallowschemiststoinves5gatethemolecularbasisforreducedbindingaffinityinknowndrug-resistantmutantsofaprotein.

MaturosMalaisreeandChristopherJ.Woods,SchoolofChemistry,UniversityofBristol,Bristol,BS81TS,UnitedKingdom

ProteinSwapisanewfreeenergymethodthatcanbeusedtoexploredrugresistance. Residue-based decomposi5ons provide a useful graphicalvisualisa5on.Theseenabletheeffectofproteinmuta5onondrugbindingtobera5onalisedatthemolecularlevel.

ProteinSwapwillsoonbeavailableasanopensourcecommand-linetoolas part of Sire (hLps://siremol.org). A graphical interfacewill bemadeavailablebyCressetviaFlare(hLps://www.cresset-group.com/flare).

References(1)  Woods,CJ,Malaisree,M,Hannongbua,S&Mulholland,AJ,“Awater-swapreac5oncoordinateforthecalcula5onof

absoluteprotein-ligandbindingfreeenergies”,JournalofChemicalPhysics,vol134,2011(2)  Woods,CJ,Malaisree,M,Michel,J,Long,B,McIntosh-Smith,S&Mulholland,AJ,“Rapiddecomposi5onand

visualisa5onofprotein-ligandbindingfreeenergiesbyresidueandbywater”,FaradayDiscussions,vol169,2014(3)  Baek,YH,etal,“ProfilingandCharacteriza5onofInfluenzaVirusN1StrainsPoten5allyResistanttoMul5ple

NeuraminidaseInhibitors”,JournalofVirology,vol89,2015

AcknowledgementsWethanktheEPSRCandCressetforfundingviaanEPSRCImpactAccelera5onAward.CWisanEPSRCResearchSocwareEngineeringFellow(EP/N018591/1).WethanktheAdvancedCompu5ngResearchCentreatUoB forprovidingearly access to thenewBlueCrystalPhase4cluster(hfp://www.acrc.bris.ac.uk)

λ=0.0 λ=0.2 λ=0.5 λ=0.8 λ=1.0

Wildtype

Mutant

E(�) = (1� �)

Eligand:wildtype + Ewater:mutant

�+ �

Ewater:wildtype + Eligand:mutant

ProteinSwap, based onWaterSwap1,2, calculates the change in protein-ligand binding free energycaused by protein muta5on. A λ-coordinate swaps the ligand from the wildtype protein to the mutant.Simultaneously, a cluster of water molecules is swapped from the mutant to the wildtype. Monte Carlosimula5onsateachvalueofλareusedtoevaluatethefreeenergychangeusingthermodynamicintegra5on.Theresultisthefreeenergypreferenceoftheligandforeitherthewildtypeorthemutantprotein.

Valida7on,wasperformedbycalcula5ngtheProteinSwapfreeenergiesofoseltamivirandperamivirligandsinwildtypeandmutant formsof influenzaneuraminidase (H1N1-2009). Logical tests, such as swapping the ligandbetween two iden5calproteins, showed themethodwas sound.Reverse tests, comparingwildtypeàmutant againstmutantàwildtype, showed theProteinSwap free energies were consistent. Closure tests showed free energies closed. Finally, comparison against knownexperimentalbindingaffini7esshowedthatProteinSwapcouldcorerctlyiden5fyknowndrug-resistantmuta5ons.

-8

-6

-4

-2

0

2

4

6

8

0 0.2 0.4 0.6 0.8 1

FreeEne

rgy/kcalm

ol-1

-0.1±0.6kcalmol-1Peramivir:WildtypeàWildtype

LogicalTest-shouldbezero

-16

-12

-8

-4

0

4

8

12

16

0 0.2 0.4 0.6 0.8 1

FreeEnergy/kcalm

ol-1

λ-16

-12

-8

-4

0

4

8

12

16

0 0.2 0.4 0.6 0.8 1

FreeEnergy/kcalm

ol-1

λ

8.3±0.8kcalmol-1 -7.7±1.2kcalmol-1Peramivir:WildtypeàR292K R292KàWildtype

ReverseTest-shouldbeequalandopposite ClosureTest–pathsequal

Wildtype R292K

R152K

E119G

8.3±0.8kcalmol-1

Peramivir:Wildtype,R292K,R152K,E119G

ComparisontoExpt.GreenShowsAgreement Oseltamivir PeramivirExperimentalResistance3 Wildtype≈E119G<R152K<R292K Wildtype<R152K<R292K<<E119G

ProteinSwapFreeEnergy/kcalmol-1

0.0≈0.6<5.1<8.4 0.0<3.0<8.3!!3.0

R e s i d u e - b a s e ddecomposi7on, allowsProteinSwap free energies to bebroken down to residue-basedcomponents. Each residue can becolored as preferring bindingligand (blue) or preferring bindingwater (red). The difference ofthese between the mutant andwildtype shows how muta5onweakens b ind ing for someresidues(red),whilestrengtheningbindingtoothers(blue).

-30

-25

-20

-15

-10

-5

0

5

10

R118

E119

D151

R152

R156

S179

D198

I222

R274

E277

S246

E276

E277

R292

N294

G344

K3

46

R371

FreeEne

rgy/kcalm

ol-1

-30

-25

-20

-15

-10

-5

0

5

10

R118

E119

D151

R152

R156

S179

D198

I222

R274

E277

S246

E276

E277

K292

N294

G344

K3

46

R371

FreeEne

rgy/kcalm

ol-1

-30

-25

-20

-15

-10

-5

0

5

10

R118

E119

D151

R152

R156

S179

D198

I222

R274

E277

S246

E276

E277

R292K

N294

G344

K3

46

R371

FreeEne

rgy/kcalm

ol-1

- =

Oseltamivir:Wildtype Oseltamivir:R292K Difference

R292

R371

R118 R118

R371

K292 R292K

R371

R118

Eresiduewildtype(�) = (1� �)

Eresidue

ligand:wildtype

�+ �

Eresidue

water:wildtype

Eresiduemutant(�) = (1� �)

Eresidue

water:mutant

�+ �

Eresidue

ligand:mutant

Conclusion