calcula&ng)mul&scale)rela&onships)...

Post on 27-Jul-2019

221 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Calcula&ng)mul&scale)rela&onships)among)drugs,)targets,)and)side)effects)

Michael(Keiser,(PhD(Assistant(Professor(of(Pharmaceu6cal(Chemistry(NS416A(Sandler(Neurosciences(Center,(Mission(Bay(Ins6tute(for(Neurodegenera6ve(Diseases((keiser@keiserlab.org(

Pharmacogenomics(245B(26(Jan(2015(

How(do(we(know(if(we’ve(found(all(of(a(drug’s(targets?(

nausea vomiting fatigue dizziness dry mouth headache itching muscle pains abdominal pain blurred vision constipation diarrhea loss of appetite memory loss failure to thrive palpitations problems with coordination ringing in the ears skin rashes or hives swelling of hands or feet heart attack syncope cafe-au-lait spots ecchymosis jaundice livedo reticularis necrolytic migratory erythema

heart attack

This(is(the(sort(of(thing(we’d(like(to(avoid(

$24$

$26$

$28$

$30$

$32$

$34$

$36$

$38$

$40$

$42$

$44$

$46$

16*Sep*04$ 23*Sep*04$ 30*Sep*04$ 7*Oct*04$ 14*Oct*04$ 21*Oct*04$ 28*Oct*04$ 4*Nov*04$

*41.5%$

9/30/04:(Merck$recalls$blockbuster$Vioxx$$

Heart(aTack(and(stroke(

$$$$ $$ $$

$$

$$

$$

$$$$

$$

$$

$6(billion(in(claims$

culprit:(5VHT2b(off6target(

1997( heart(valve(disorder(

already(cost(Wyeth(

$21$billion$by(2004$

But(which(sorts(of(toxicity(could(we(predict?(

OffVtarget( FenPhen(

1996(

Liver((metabolic)(

Rezulin((2000)(–(Parke(Davis(–(430+(pa6ents(suffered(liver(failures(

OnVtarget( Vioxx((2004)(–(Merck(–($6(billion(

TODAY$1.  known(pharmacology(2.  rela6ng(targets(to(other(targets(3.  predic6ng(drug(offVtargets(4.  wholeVorganism(effects(

Ligands(link(disparate(proteins(

An6folate(targets(lack(sequence(or(structural(similarity.((

✗$✗$

Cysteine(proteases(

Metalloproteases(

Serine(proteases(

Aspartyl(proteases(

Kinases(

Nuclear(hormone(receptors(

Ion(Channels(

Phosphodiesterases(

Aminergic(GPCRs(

Pep6de(GPCRs(

GPCRs((others)(

Enzymes(((others)(

Miscellaneous((

Paolini, Nat. Biotech, 2006

Polypharmacology(recapitulates(biology(

Aminergic(GPCRs((

Aspartyl(Proteases((

Cysteine(Proteases((

Enzymes((others)((

GPCRs(Class(AV(others((

GPCRs(Class(B((

GPCRs(Class(C((

Hydrolases((

Ion(ChannelsV(Ligand(Gated((

Ion(ChannelsV(others((

KinasesV(others((

Metalloproteases((

Nuclear(hormone(receptors((

Miscellaneous((

Oxidoreductases((

Phosphodiesterases((

Pep6de(GPCRs((

Protein(Kinases((

Serine(Proteases((

Transferases(

Aminergic(GP

CRs(

Aspartyl(Proteases(

Cysteine

(Proteases(

Enzymes((o

thers)(

GPCR

s(others(

GPCR

s(Class(B(

GPCR

s(Class(C(

Hydrolases(

Ion(Ch

anne

lsV(LG(

Ion(Ch

anne

lsV(

KinasesV(others(

Metalloproteases(

Nuclear(re

ceptors(

Misc

ellane

ous(

Oxido

redu

ctases(

Phosph

odiesterases(

Pep6

de(GPC

Rs(

Protein(Kinases(

Serin

e(Proteases(

Transferases(

Promiscuity((Gkl)(value(0((((((((((((((((((((((((((((60(

No(data(available(

Paolini,(Nat.(Biotech,(2006(

Promiscuity$in$&$across$gene$families$

How(can(we(predict(drug(&(target(networks?(

•  Hypothesis:(Proteins(can(be(interVrelated(through(their(ligands.(

–  We(represent(proteins(solely(by(their(ligands.(

–  This(works(for(protein(targets(that(are(unrelated(by(biological(criteria((sequence(&(structure).(

–  We(can(predict(new(links(known(between(targets(and(known(drugs.(

•  Lounkine(&(Keiser(et(al.(LargeVscale(predic6on(and(tes6ng(of(drug(ac6vity(on(sideVeffect(targets.(Nature.(2012.(

•  Keiser(&(Setola(et(al.(Predic6ng(new(molecular(targets(for(known(drugs.(Nature.(2009.(•  Keiser(et(al.(Rela6ng(protein(pharmacology(by(ligand(chemistry.(Nat)Biotech.(2007.(

1.  known(pharmacology(2.  rela6ng(targets(to(other(targets(3.  predic6ng(drug(offVtargets(4.  wholeVorganism(effects(

Core(idea:(Link(small(molecules(to(protein(targets(by(ligands,(not(by(sequence(or(structure(

•  Many(drugs(&(ligands(simultaneously(act(through(dissimilar(proteins.(

•  Methadone:(–  μ(opioid(receptor((GPCR)(–  NMDA(receptor((Ion(channel)(

•  Ligands(are(a(func6onal(link(between(molecular(targets.(

G*protein$coupled$receptor$(GPCR)$

Ion$channel$

Where(do(we(get(the(ligands?(

•  From(compound(ac6vity(databases(

DHFR(

Database$ Ligands$ Targets$ Data$points$MDDR((2005)( 201,761( 631( 391,406(

WOMBAT((2008)( 254,679( 2,100( 760,605(ChEMBLV14VB((2013)( 241,749( 2,657$ 1,165,336(

How(can(we(compare(two(proteins?(

Calculate(all(pairVwise(similari6es(among(ligands(across(the(sets(

?(DHFR( TS(

DHFR( TS(

How(do(we(encode(a(molecule?(

Fingerprint$

Bond(paTern(1(

Bond(paTern(2(

…(

F$T$

.(

.(

.(

How(similar(are(two(molecules?(

•  Use(Tanimoto(coefficients(to(compare(their(fingerprints(

•  TC ( (=(bC(((/ ((b1 (+(b2 (–(bC)(•  0.29( (=(2( (/ ((4( (+(5( (–(2)(

1

2

Comparative performance metrics: Chen et al. (2002) J Chem Inf Comput Sci. 42(6), 1407-14.

How(can(we(compare(two(ligand(sets?(

TC(a1,b1)(=(0.91(TC(a1,b2)(=(0.34(TC(a2,b1)(=(0.28(

TC(a2,b2)(=(0.31(TC(a3,b1)(=(0.43(TC(a3,b2)(=(0.39(

TC(=(Tanimoto(coefficient,(using(ECFP4(fingerprints(

Sa( Sb(

Similarity(distribu6on(for(Sa(vs.(Sb(

How(similar(are(ligands(across(targets?(

TS$

�Raw$Score�$

Similarity(to(Thymidylate(synthetase((TS)(ligands(

Num

ber(o

f(ligand(pairs(

Pairwise(ligand(similarity((%)(

Thrombin$

DHFR$

What(similarity(is(significant?(

Product(of(ligand(set(sizes(

Random)raw)scores)

Raw)score)standard)dev)

Raw(sc

ore(means(

Raw(sc

ore(std(de

vs(

Extreme$Value$Distribu\on$(EVD)$

ZVscore(Freq

uency(

Probability(of(any(given(raw(score(happening(by(random(chance(alone:(

Product(of(ligand(set(sizes(

But(wait;(how(should(we(choose(a(threshold?(

0.50(Tc(Threshold(chosen(by(fit(to(EVD(

0.48(Tc(Threshold(chosen(by(best(singleVmolecule(crossVfold(valida6on(performance((

Note:(Data(shown(uses(Daylight,(but(this(also(applies(to(ECFP4,(etc(

Similarity(ensemble(approach((for(targets)(

A( A(

1(Choose(target( 2(Build(ligand(set(

4(Rank(by(significance(

ZVscore(

Freq

uency(

A(vs.(B(

A(vs.(C(

Sta6s6cal(model(

…$

B(

3(Compare(against(database(sets(

?(?(

C(

,$

Similar(drug(targets(group(together(

?$Ligand(set(

Ligand(sets(

Similarity(paTerns(among(ligand(sets(

Similar MDDR Activity Classes E-value

Dihydrofolate Reductase Inhibitor 1.8e-247

1 Glycinamide Ribonucleotide Formyltransferase Inhibitor 8.5e-157

2 Thymidylate Synthetase Inhibitor 1.7e-94

3 Folylpolyglutamate Synthetase Inhibitor 7.1e-88

4 Uridine Phosphorylase Inhibitor 4.3e+3

1$

2$

3$

4$

Ligand(sets(form(archipelagos.(((

Does(this(feature(lead(to(an(informa6ve(global(map?(

Query:$Dihydrofolate(Reductase(Inhibitor(

MDDR$MDL(Drug(Data(Report((100k(compounds(annotated(by(patent(literature)(

Biological(clusters(emerge(from(chemistry(

Keiser(MJ,(et(al.(Rela6ng(protein(pharmacology(by(ligand(chemistry.(Nat)Biotech((2007).(

Node(A(set(of(ligands(that(represents(a(single(target((Edge(A(SEA(score(between(two(targets(

Sequence(&(pharmacology(are(complementary(

Sequence$similarity:$Log((PSIVBLAST(EVvalues)(

Pharmacological$similarity:$Log((SEA(EVvalues)(

1.  known(pharmacology(2.  rela6ng(targets(to(other(targets(3.  predic6ng(drug(offVtargets(4.  wholeVorganism(effects(

•  Although(drugs(are(intended(to(be(selec6ve,(at(least(some(bind(to(several(physiological(targets,(explaining(side(effects(and(efficacy.(

•  But(how(common(is(this?(

Known(network(pharmacology.(

(

Is(it(comprehensive?(

Drug((

Protein(target((

≤(1(uM(IC50(

Keiser(&(Setola,(et(al.(Nature,)2009(

3(Rank(by(significance((SEA(EVvalue)(

ZVscore(

Freq

uency(

vs.(DHFR(

vs.(thrombin(

Sta6s6cal(model(

DHFR(

Similarity(Ensemble(Approach((SEA)(predicts(targets((2000(targets,(300,000+(ligands)(

1(For(a(small(molecule,(

Keiser(et(al.,(Nature)Biotech.(2007;(Keiser(&(Setola(et(al.,(Nature,(2009.(

DHFR(

2(Compare(it(against(each(target’s(ligands(

…$,$thrombin(

First(tests:(Methadone(is(an(an6muscarinic(

Keiser(et(al,(Nat)Biotechnol,(2007(

$( Ac\vity( SEA$E*value(

1( An6muscarinic( 4.5(×(10V50(

2( Muscarinic(M3(Antagonist( 1.2(×(10V11(

3( Opioid(Agonist( 1.8(×(100(

4( NMDA(Receptor(Antagonist( 9.0(×(100(

%(of(kno

wn(ligand(bo

und(

Drug(concentra6on( Time((seconds)(

M3(ac6vity

((RFU

)(

Ac6vity(blocked(

((SEA(predic6ons(above(used(MDDR(v2005()(

Methadone(

Drugs(aren’t(always(“magic(bullets”(

Revanil((lysuride)(An6parkinsonian(that(inhibits(D1/2,(and(prolac6n(secre6on.(

α2(Antagonist(SEA:(8.9x10V82(

0.055(nM(

Permax((pergolide)(An6parkinsonian(withdrawn(due(to(heart(valve(disease.(

5VHT1D(Antagonist(SEA:(8.7x10V29(

13.1(nM(

Zyvox((linezolid)(An6Vinfec6ve(that(inhibits(bacterial(protein(synthesis(by(binding(to(23S(rRNA.(Also(affects(blood(pressure.(

MAOVB(Inhibitor(SEA:(1.3x10V27(

17(uM(

…We(found(>40(more(manual(and(182(more(automa6callyVvalidated(predic6ons(with(SEA(EVvalues(<(10V10(

…but(some(may(be(“magic(shotguns”(Dapsone(Old(use:((Leprosy((New(use:(Malaria(Why?(�CA(may(have(essen6al(roles(in(the(parasite(metabolism.(Thus,(the(aroma6c(sulfonamide(CA(inhibitors(may(have(poten6al(for(development(of(novel(an6malarial(drugs."(

Carbonic(anhydrase(

(

SEA:(2.9x10V12(

Miltefosine(Old(use:(Cancer(New(use:(Visceral(leishmania(Why?(�[analogues(of(sphingosine](have(been(assayed(in(vitro(against(Leishmania(spp.(and(Trypanosoma(cruzi.(Most(of(the(compounds(were(potent(parasi6cides..."(

SphingosineV1Vphosphate(

(

SEA:(1.9x10V6(

These(drugs(have(both(been(successfully(�repurposed�(

23+(novel((

offVtargets,(from(14(drugs(

…of$20$assays,$15$at$<$1$uM$5$at$<$100$nM$( )

Keiser,(et(al.(Nature)2009(

New(offVtargets(cross(tradi6onal(boundaries(

Rescriptor((HIVV1(NNRTI)(

Vadilex((NMDA)(

ROV25V6981((NMDA)(

Xenazine((VMAT2)(

Enzyme$

GPCR$

Ion$channel$

Transporter$α2$

H4$μ$opioid$

SERT$D4$

NET$κ$opioid$

E:(2.4x10V7(

E:(10V30(

E:(10V13V10V5(

Keiser(&(Setola,(et(al.(Nature,)2009(

Sequence(similarity(may(disagree(with(pharmacology(

ATempt(to(recapitulate(SEA(offVtargets(by(seq.(similarity(to(drug’s(known(targets(

Average(PSIVBLAST(rank:(43$

(of(244)(

(of(23)(

Case(study:(“Off”Vtargets(can(be(potent(

Keiser(&(Setola,(et(al.(Nature,)2009(

SEA(EVvalue:(2.8×10V27(

But(it(has(an$18$nM$Ki(for(D4R(

Baratol((an(an6hypertensive):(

(

200*600$nM$Ki$(

for(α1$

adrenergic((receptors.(

Target$$ SEA$E*value$$

1$ 5VHT1B(( 3.1(×(10V21(

2$ 5VHT2A( 1.2(×(10V13(

3$ 5VHT5A( 1.1(×(10V7(

4$ 5VHT7( 5.0(×(10V6(

Keiser(&(Setola,(et(al.(Nature,)2009(

Case(study:(Mode(of(ac6on(for(a(hallucinogen(

DMT$(dimethyltryptamine)(

What(is(the(target(behind(DMT’s(hallucinatory(effects?(

(A(2009(paper(implicated(the(σ1(receptor((14,750(nM(Kd)(

DMT:(129(nM(

Ergotamine((Ki(=(3.5(nM)((

DMT$

5*HT1B$receptor$

%([3H]5VCT

(binding(

log([drug](

DMT:(206(nM(%([3H]LSD(bind

ing(

log([drug](

Chlorpromazine((Ki(=(42(nM)((

DMT$

5*HT7A$receptor$

DMT:(127(nM(

%([3H]LSD(bind

ing(

Chlorpromazine((Ki(=(2.4(nM)((

DMT$

5*HT2A$receptor$

log([drug](

DMT(no(longer(hallucinogenic(in(5VHT2A(KO(mice(

5VHT2A(may(be(primary(target(of(DMT�s(hallucinogenic(effects.(

Keiser(&(Setola,(et(al.(Nature)2009(

With(Bryan(Roth((UNC(PDSP)(

Case(study:(OffVtarget(to(side(effect(

Target$ SEA$E*value$

1$ Adrenergic((β)(blocker( 3.9(×(10V15(

2$ 5(HT1A(Antagonist( 2.0(×(10V9(

3$ Acetylcholinesterase(Inhibitor(( 1.1(×(10+1(

Keiser,(et(al.(Nature,)2009;(Thomas,(et(al.(Psychopharmacol)Bulle&n,(2010.(

SSRI(discon6nua6on(syndrome(for(pa6ents(with(ADRB1(muta6on((SNP)(

Fluoxe6ne(binds(β1((4(uM)((

Paroxe6ne(binds(β1((10(uM)(%([1

25I]P

indo

lol(bou

nd(

Log([drug](

β1(receptor(

Alprenolol(Ki(=(2.0(nM((

Fluoxe\ne$Ki(=(4,385(nM(

Fluoxe6ne((Prozac)(( ((Using(MDDR(v2005()(

$Binding$assays$

Test(predic6ons(

We(sought(to(predict(ADRs(on(a(broad(scale(

$SEA$

predic\ons$

650(drugs(vs.)70(targets(

Liability(targets(

explaining(ADRS(

Target(consistent(with(ADR?(

$ADR$

databases$

115$new$links(found(

Lounkine(&(Keiser(et(al.,(Nature,(2012.(

ADR:(Adverse(drug(reac6on(

With(Laszlo(Urban((Novar6s)(

No)human)interven&on)

CTX(inhibits(platelet(aggrega6on(in(human(blood.(

In(silico,(in(vitro,(ex(vivo:(From(structure(to(ADR(

Chlorotrianisene((CTX)(is(a(synthe6c(

estrogen(causing(stomach(pain$

COXV1: (130(nM((Estrogen(receptor: (1,000(nM(

Lounkine(&(Keiser(et(al.,(Nature,(2012.(

v.s.(Indomethacin((230(nM(COXV1)(

?(

26%(of(new(targets(cross(family(boundaries(

Confirmed, 1%

0%(

5%(

10%(

15%(

20%(

25%(

30%(

Frac

tion

of n

ovel

off-

targ

ets

-log10 (BLAST E-value)

Sequence similarity of new target to closest known

HRH2(–(SCN10A(Not)related)

NR3C1(–(PGR(

CHRM1(–(HRH1(Dis&nct)subclass)

Lounkine(&(Keiser(et(al.,(Nature,(2012.(

Physical(proper6es(only(loosely(predict(drug(promiscuity(

V4( 0(

Probucol(

Ibandronate(

Chlorhexidine(

Fluspirilene(

Zafirlukast(

Tegaserod(

Lounkine(&(Keiser(et(al.,(Nature,(2012.(

1.  known(pharmacology(2.  rela6ng(targets(to(other(targets(3.  predic6ng(drug(offVtargets(4.  wholeVorganism(effects(

Drugs(act(via(‘mul6scale(mechanisms’(

Zhao(&(Iyengar.(Annu.(Rev.(Pharmacol.(Toxicol.(2012.(

channels((myocyte)(intermed.(+(target(intermediates(((drug(targets(drugs(

How(can(we(make(

this(simpler?(

Zhao(&(Iyengar.(Annu.(Rev.(Pharmacol.(Toxicol.(2012.(

3(Test in vitro and genetically

ZVscore(

Freq

uency(

vs.(DHFR(

vs.(thrombin(

Sta6s6cal(model(

DHFR(

Predic6ng(phenotypic(mechanism(of(ac6on(with(SEA)

1(For each active molecule

DHFR(

2(Compare to ligands of 2500 targets

…$,$thrombin(

Mul6Vphenotype(screen((3,200(cpds):(Some(compounds(affect(feeding(in(C.)elegans((measured(by(pharyngeal(pumping(rate).(

Kaveh’s(screen:(Can(we(find(mechanisms(systema6cally(

Food(source(

Posterior(pharynx(

Lemieux(et(al,(PLoS(Biology,(2013)

AR(8.6(nM( FltV3(

169(nM(

NK1(1.9(uM(

OTR(1.8(uM(

mGluR8(9.8(uM(

Using(SEA,(hits(were(antagonists(at(human(targets:(

We(mapped(human(targets(back(to(C.)elegans)

V10(

V5(

0(

5(

10(

Interac6on:(•  Addi6ve(•  Epista6c(•  Antagonis6c(

Change(in(fe

eding(rate(

Oxytocin(receptor((

(OTR)(antagonists(

Control((4.6(nM)(

From(screen:(F15((1.8(uM)(

Gene6c(condi6ons((gene(inac6va6ons)(

Lemieux(et(al,(PLoS(Biology,(2013)

We(mapped(human(targets(back(to(C.)elegans)

V10(

V5(

0(

5(

10(

Interac6on:(•  Addi6ve(•  Epista6c(•  Antagonis6c(

Change(in(fe

eding(rate(

Oxytocin(receptor((

(OTR)(antagonists(

Phenocopy$(4.6(nM)(

From(screen:(F15((1.8(uM)(

Gene6c(condi6ons((gene(inac6va6ons)(

gnrr*1$$

(3x10V30(by(BLAST(to(human(OTR)((

Lemieux(et(al,(PLoS(Biology,(2013)

We(mapped(human(targets(back(to(C.)elegans)

V10(

V5(

0(

5(

10(

Gene6c(condi6ons(

Change(in(fe

eding(rate(

Chemical(condi6ons(Interac6on:(•  Addi6ve(•  Epista6c(•  Antagonis6c(

Let’s(switch(a(topVdown(view(

Lemieux(et(al,(PLoS(Biology,(2013)

mglV2(

verV3(

gnrrV1(

verV2(

tkrV1((RNAi)(

gnrrV3(

This(yielded(four(targets(newly(regula6ng(appe6te(

Interac6on:(•  Addi6ve(•  NonVaddi6ve(•  Antagonis6c(

Gene6c(perturba6ons(

Chemical(perturba6ons(

3(

1(

2(

4(

mGluR8( FltV3( OTR( NK1/3( Human)(chemical))mglV2( verV3( gnrrV1( tkrV1( C.)elegans)(gene&c))

Lemieux,(Keiser,(et(al.(PLoS(Biol,(2013.) [c](=(posi6ve(control(compound(

Winter(notes(on(summer(impressions(

•  1(People(usually(choose(targets(by(dint(of(domain(exper6se.(–  If(we(didn’t,(would(we(just(get(‘useless�(offVtargets?(Or(learn(something(new?(

•  2(We(can(usually(cross(organisms.(–  Zebrafish(<V>(human(–  Worm(<V>(human(

•  3(One(phenotype(≠(one(target.(–  How(many(of(these(are(stackable?(–  What(happens(“offVdiagonal”?(

Most(offVtargets(are(s6ll(unexplored(

You(can(try(SEA(yourself(

hTp://sea.bkslab.org(

top related