Repurposing and Reformulation of the Antiparasitic Agent Flubendazole for Treatment of ...

13 Pages·2017·1.98 MB·English

PHARMACOLOGY
crossm
Repurposing and Reformulation of the Antiparasitic Agent
Flubendazole for Treatment of Cryptococcal
Meningoencephalitis, a Neglected Fungal Disease
GemmaL.Nixon,a,bLauraMcEntee,bAdamJohnson,bNicolaFarrington,bSarahWhalley,bJoanneLivermore,bCristienNatal,b
GinaWashbourn,aJaclynBibby,aNeilBerry,aJodiLestner,bMeganTruong,cAndrewOwen,dDavidLalloo,eIanCharles,f
D
WilliamHopeb
o
w
n
aDepartmentofChemistry,UniversityofLiverpool,Liverpool,UnitedKingdom lo
bAntimicrobialPharmacodynamicsandTherapeutics,DepartmentofMolecularandClinicalPharmacology, a
d
InstituteofTranslationalMedicine,Liverpool,UnitedKingdom
e
cithreeInstitute,UniversityofTechnologySydney,Sydney,Australia d

f
dDepartmentofMolecularandClinicalPharmacology,Liverpool,UnitedKingdom ro
eLiverpoolSchoolofTropicalMedicine,Liverpool,UnitedKingdom m

fQuadramInstituteBioscience,NorwichResearchPark,Norwich,UnitedKingdom h
t
t
p
ABSTRACT Current therapeutic options for cryptococcal meningitis are limited by :/
/
a
toxicity, global supply, and emergence of resistance. There is an urgent need to de-
a
velop additional antifungal agents that are fungicidal within the central nervous sys- c
.
a
tem and preferably orally bioavailable. The benzimidazoles have broad-spectrum an- s
m
tiparasitic activity but also have in vitro antifungal activity that includes Cryptococcus
.
neoformans. Flubendazole (a benzimidazole) has been reformulated by Janssen Phar- o
r
g
maceutica as an amorphous solid drug nanodispersion to develop an orally bioavail- /

able medicine for the treatment of neglected tropical diseases such as onchocercia- o
n
sis. We investigated the in vitro activity, the structure-activity-relationships, and both M
in vitro and in vivo pharmacodynamics of flubendazole for cryptococcal meningitis. a
y
Flubendazole has potent in vitro activity against Cryptococcus neoformans, with a
4
modal MIC of 0.125 mg/liter using European Committee on Antimicrobial Suscepti- ,
2
bility Testing (EUCAST) methodology. Computer models provided an insight into the 0
1
residues responsible for the binding of flubendazole to cryptococcal (cid:2)-tubulin. Rapid 9

fungicidal activity was evident in a hollow-fiber infection model of cryptococcal b
y
meningitis. The solid drug nanodispersion was orally bioavailable in mice with
g
higher drug exposure in the cerebrum. The maximal dose of flubendazole (12 Received14September2017 Returnedfor u
modification3November2017 Accepted3 e
mg/kg of body weight/day) orally resulted in an (cid:2)2 log CFU/g reduction in fungal s
10 January2018 t
burden compared with that in vehicle-treated controls. Flubendazole was orally bio- Acceptedmanuscriptpostedonline8
available in rabbits, but there were no quantifiable drug concentrations in the cere- January2018
brospinal fluid (CSF) or cerebrum and no antifungal activity was demonstrated in ei- CitationNixonGL,McEnteeL,JohnsonA,
FarringtonN,WhalleyS,LivermoreJ,NatalC,
ther CSF or cerebrum. These studies provide evidence for the further study and
WashbournG,BibbyJ,BerryN,LestnerJ,
development of the benzimidazole scaffold for the treatment of cryptococcal menin- TruongM,OwenA,LallooD,CharlesI,HopeW.
gitis. 2018.Repurposingandreformulationofthe
antiparasiticagentflubendazolefortreatment
ofcryptococcalmeningoencephalitis,a
KEYWORDS Cryptococcusneoformans,cryptococcalmeningoencephalitis,
neglectedfungaldisease.AntimicrobAgents
benzimidazole,flubendazole,(cid:2)-tubulin,antifungalagents,cryptococcal,meningitis, Chemother62:e01909-17.https://doi.org/10
pharmacodynamics,pharmacokinetics,tubulin .1128/AAC.01909-17.
Copyright©2018Nixonetal.Thisisanopen-
accessarticledistributedunderthetermsof
Cryptococcalmeningoencephalitis(herereferredtoasmeningitis)isacommonand theCreativeCommonsAttribution4.0
Internationallicense.
lethal disease in immunosuppressed patients (1, 2). This disease is predominately
AddresscorrespondencetoWilliamHope,
associated with advanced HIV infection and has the highest incidence in low- to
[email protected].
middle-incomecountries(1).Thenumberofeffectiveagentsisdespairinglysmall(3).
April2018 Volume62 Issue4 e01909-17 AntimicrobialAgentsandChemotherapy aac.asm.org 1
Nixonetal. AntimicrobialAgentsandChemotherapy
TABLE1MICdistributionsofflubendazoleagainstC.neoformansisolatesusingCLSIand
EUCASTmethodologies
No.ofisolateswithMIC(mg/liter)of:
Methodology No.ofstrains 0.03 0.06 0.125 0.25 0.5
EUCAST 50 1 19 25 5 0
CLSI 50 2 40 8 0 0
Allavailableinductionandmaintenanceregimensareconstructedwiththreeantifun-
galagents:amphotericinB(AmB),flucytosine(5FC),andfluconazole(4).Eachofthese
compoundshassignificantadverseeffectsthatincludeinfusionaltoxicity(AmB),neph-
rotoxicity(AmB[5]),bonemarrowsuppression(AmBand5FC[5,6]),andhepatotoxicity
(fluconazole and 5FC [7]). Moreover, there are significant inherent limitations that
D
include fungistatic effects (fluconazole [8]) and the potential emergence of drug
o
resistance(fluconazoleand5FC[9–11]).Thus,thereisanurgentimperativetodevelop w
n
new agents. Orally bioavailable agents are particularly important given the predomi-
lo
nanceofthisdiseaseinresource-constrainedsettings. a
d
During the process of screening a compound library against fungal pathogens, it e
d
wasnotedbyus(MeganTruongandIanCharles)thatflubendazolehaspotentinvitro

f
activityagainstCryptococcusneoformans.Aliteraturesearchrevealedothermembersof ro
m
the benzimidazole class (e.g., albendazole and mebendazole) of antiparasitic agents

hadpreviouslybeendemonstratedtohavepotentinvitroactivityagainstCryptococcus h
t
t
neoformans,withMICsof0.16to0.45mg/liter(12,13).Thepharmacologicaltargetof p
:
the benzimidazoles against Cryptococcus neoformans is (cid:2)-tubulin (14). The antifungal //
a
activity of parenterally administered flubendazole in a murine model of cryptococcal a
c
meningitiswasconfirmedbyusinaseriesofpreliminaryexperiments.Concurrently,we .a
became aware of the efforts by Janssen Pharmaceutica to develop a new orally s
m
bioavailable formulation of flubendazole that may be active against filariasis and .
o
onchocerciasis.Thepotentialvalueofthisnewformulationasanoralmedicineforthe r
g
treatmentofcryptococcalmeningitisinresource-poorhealthcaresettingswasthere- /
o
foreevident. n

Herewedescribetheinvitroactivity,putativestructure-activityrelationships,andin M
a
vivopharmacokinetic-pharmacodynamic(PK-PD)relationshipsofflubendazoleagainst y

Cryptococcus neoformans. A hollow-fiber infection model of cryptococcal meningitis 4
,
was developed as a first step for exploring dose exposure-response relationships. 2
0
Subsequently,twoextensivelyusedandwell-characterizedlaboratoryanimalmodelsof 1
9
cryptococcal meningitis were used to provide the experimental foundation for the

b
potentialuseoforalformulationsofflubendazoleoritscongenersforthetreatmentof y

aneglectedinfectionofglobalsignificance. g
u
e
RESULTS s
t
Invitrostudies.Flubendazoledisplayedpotentinvitroactivity(MICsof0.06to0.25
mg/liter[Table1])againstC.neoformans.TheMICswerecomparablewhenEuropean
Committee on Antimicrobial Susceptibility Testing (EUCAST) and Clinical and Labora-
toryStandardsInstitute(CLSI)methodologieswereused.
The flubendazole 50% inhibitory concentration (IC ) against porcine tubulin was
50
2.38 (cid:3)M. Other known tubulin inhibitors display similar efficacies in this assay (e.g.,
colchicinehasanIC of1.15(cid:3)M[unpublisheddata],paclitaxelhasanIC of3.9(cid:3)M
50 50
[15], and vinblastine has an IC of 5.3 (cid:3)M [15]). These data are consistent with the
50
knownmechanismofactionofflubendazole.
Aninvitrodrugmetabolismandpharmacokinetics(DMPK)assessmentofcommer-
cially available flubendazole powder confirmed a favorable log of distribution coeffi-
cientatpH7.4(logD7.4)of2.9.Plasmaproteinbindingwas90.6%,andtherewaslow
metabolic turnover (human microsomal intrinsic clearance (cid:3) 44 (cid:3)l/min/mg and rat
hepatic intrinsic clearance (cid:3) 39 (cid:3)l/min/106 cells). However, aqueous solubility was
poor (0.8 (cid:3)M), which is characteristic of the benzimidazoles. This in vitro DMPK
April2018 Volume62 Issue4 e01909-17 aac.asm.org 2
PharmacodynamicsofFlubendazoleforCryptococcus AntimicrobialAgentsandChemotherapy
D
o
w
n
lo
a
d
e
d

f
r
o
m

h
t
t
p
FIG1(AandB)HomologymodelofflubendazoledockedwithbothC.neoformans(A)andhuman(B)(cid:2)-tubulin.Redsphere,hydrogen :/
/
bond donors; blue sphere, hydrogen bond acceptors; yellow sphere, hydrophobic interactions. (C and D) The docking pose is a
a
visualizedwithPyMOL.Proteinisshownasasurfacerepresentationcolored40%transparentlightblue.Flubendazoleisrepresented
c
as sticks composed of carbon (light blue), hydrogen (white), nitrogen (dark blue), oxygen (red), and fluorine (cyan). Binding site .
a
residuesselectedaround4Åarerepresentedasstickswithcarbon(green),nitrogen(blue),oxygen(red),andsulfur(yellow). s
m
.
o
r
g
assessment was consistent with subsequent in vivo observations (see below). Poor /

o
aqueous solubility limits absorption through the gut, but once in the bloodstream, n

flubendazolehasfavorablepharmacokineticproperties(e.g.,abilitytopassthroughcell M
membranes, low metabolism, and high concentrations of free drug) that enable it to a
y
reachtheeffectsite. 4
,
Docking studies. There were two principal noncovalent binding interactions be-
2
tween flubendazole and the homology model of C. neoformans (cid:2)-tubulin. First, the 0
1
hydroxyl group of serine 350 (Ser350) acts as a hydrogen bond donor and binds the 9

ketone oxygen of flubendazole (Fig. 1A). Second, asparagine 247 (Asn247) acted as a b
y
hydrogen bond donor via the primary amide with the ketone of the carbamate on g
u
flubendazole, but it also acted as a hydrogen bond acceptor through the primary e
carbonyl group of Asn247 and the N-H on the benzimidazole core. There were also s
t
several hydrophobic interactions deeper in the binding pocket that involved the
benzene ring and the fluorine of flubendazole.
Docking studies of flubendazole and human (cid:2)-tubulin (Fig. 1B) showed that both
theN-HofthebenzimidazolecoreandtheN-Hofthecarbamatearehydrogenbond
donors (Fig. 1B) to the primary amide of the side chain of Asn247. As for the C.
neoformans interaction, there were hydrophobic interactions present from the para-
substituted benzene and the hydrophobic binding pocket. There was a lack of a
hydrogenbondacceptorrolefromtheketoneoxygen.Thisisduetothereplacement
ofSer350fromtheC.neoformansactivesitewithlysine350(Lys350)inhumans.
Hollow-fiber infection model of cryptococcal meningoencephalitis. Rapid fun-
gicidalactivitywasobservedinthehollow-fiberinfectionmodel.Controlsgrewfroman
initialdensityofapproximately6log CFU/mlto8to9log CFU/ml(Fig.2).Following
10 10
theadministrationofflubendazoletherewasaprogressivedeclineinthefungaldensity
in the hollow fiber in all arms. There was an exposure-dependent decline in fungal
burden.
April2018 Volume62 Issue4 e01909-17 aac.asm.org 3
Nixonetal. AntimicrobialAgentsandChemotherapy
D
o
w
n
lo
a
d
e
d

f
r
o
m

h
t
t
p
:
/
/
a
a
c
.
a
s
m
.
FIG 2Hollow-fiber infection model of cryptococcal meningitis. (A) Pharmacokinetics of flubendazole o
withthethreearmswithintendedpeakconcentrationsof1.25,2.5,and10mg/liter;(B)pharmacody- rg
namics in response to flubendazole administered at various doses q24h. Therapy was initiated 24 h /
postinoculation,afterwhichtimeCryptococcushadgrownfrom(cid:2)6log CFU/mlto8log CFU/ml. o
10 10 n

M
a
y
Preliminarystudiestodemonstrateinvivoefficacyofflubendazole.Therewas 4
,
no demonstrable antifungal effect of orally administered flubendazole as a pure 2
compound when formulated with sterile distilled water, 0.05% polysorbate 80 in 0
1
phosphate-buffered saline (PBS), 5% dimethyl sulfoxide DMSO, 10% polyethylene 9

glycol 400 (PEG400), or 85% hydroxyl-propyl-(cid:2)-cyclodextrin (data not shown). Anti- b
y
fungalactivitycouldbeestablishedonlywhenpureflubendazolewasformulatedwith g
u
polysorbate80(Tween80)andinjectedsubcutaneously(s.c.)toformadepot.Presum- e
s
ably, formulation with polysorbate 80 solubilized flubendazole to an extent that t
enabledittobecomesystemicallybioavailable.However,thiswasobservedonlywhen
flubendazole was administered s.c. This parenteral regimen resulted in a modest
reductioninfungalburden,1to2log CFU/g,comparedwiththatinvehicle-treated
10
controls(datanotshown).AlimitedPKstudywithconcentrationsmeasuredatasingle
timepointtheendoftheexperimentalsoconfirmedthatflubendazoleconcentrations
werequantifiableinplasmaandthecerebrumofmice(datanotshown).
These preliminary pharmacokinetic and pharmacodynamic data provided the im-
petus for further detailed experiments examining the pharmacodynamics of a new
orally bioavailable solid drug nanodispersion against Cryptococcus neoformans devel-
opedbyJanssen.
Pharmacokineticandpharmacodynamicstudiesoftheflubendazolenanofor-
mulationinmice.Whenflubendazolewasformulatedasasoliddrugnanodispersion,
it was rapidly absorbed after oral dosing and plasma concentrations were readily
quantifiable at the first sampling point (i.e., 0.5 h postdose [Fig. 3]). The pharmacoki-
netics was linear, with biexponential clearance from the bloodstream with mean and
April2018 Volume62 Issue4 e01909-17 aac.asm.org 4
PharmacodynamicsofFlubendazoleforCryptococcus AntimicrobialAgentsandChemotherapy
D
o
w
n
lo
a
d
e
d

f
r
o
m

h
t
t
p
:
/
/
a
a
c
.
a
s
m
.
o
r
g
/

o
n

M
a
y

4
,

2
0
1
9

b
y

g
u
e
s
t
FIG3Flubendazolepharmacokineticsinmiceandrabbits.(A)Mouseplasmaconcentration-timeprofiles
followingtheadministrationofflubendazoleat2,4,6,8,and12mg/kg.(B)Mouseconcentration-time
profilesinthebrainfollowingtheadministrationofflubendazoleat2,4,6,8,and12mg/kg.Dataare
means(cid:4)SDsfor3mice.(C)Plasmapharmacokineticsintheserumforindividualrabbitsreceiving6
mg/kg/day(brokenlinesandsolidtriangles)and22.5mg/kg(solidlinesandsolidsquares).
median values of 0.039 and 0.026 liter/h, respectively (Fig. 3). The pharmacokinetic
parameters are summarized in Table 2. There was rapid and extensive distribution of
drug to the cerebrum in mice, and concentrations of flubendazole were consistently
higherthanthoseobservedinplasma.Theratiooftheareaunderthecurve(AUC)for
serumandtheAUCforthecerebrumwas1:4.44.
Flubendazolehadasignificantanddiscernibleantifungaleffectinmice.Useofthe
highestdoseinthisstudy(12mg/kg)resultedinapproximatelya2-to3-logreduction
April2018 Volume62 Issue4 e01909-17 aac.asm.org 5
Nixonetal. AntimicrobialAgentsandChemotherapy
TABLE2ParametervaluesfromthePK-PDmodelfittedtomice
Parametera Mean Median SD
K (h(cid:5)1) 11.312 14.895 6.594
a
SCL/F(liters/h) 0.039 0.026 0.031
V/F(liters) 0.051 0.069 0.033
c
K (h(cid:5)1) 15.741 15.404 6.806
cp
K (h(cid:5)1) 16.997 16.915 5.962
pc
K (h(cid:5)1) 3.446 0.594 4.709
cb
K (h(cid:5)1) 0.056 0.056 0.030
bc
K (log CFU/g/h) 0.107 0.098 0.025
gmax 10
Hg 10.338 5.096 9.782
C (liters/h) 2.036 1.681 1.517
50g
POPMAX(CFU/g) 982,934,669.178 427,055,621.187 2,281,967,059.602
IC(CFU/g) 102.255 116.462 60.966
V /F(liters) 0.277 0.146 0.335
b
aK isthefirst-orderrateconstantcollectingthegutandthecentralcompartment;SCL/Fistheapparent D
a
clearanceofflubendazolefromthecentralcompartment;V/FandV/Faretheapparentvolumesofthecentral o
c b
compartmentandbrain,respectively;K ,K ,K ,andK arethefirst-orderintercompartmentalrateconstants; w
andK isthemaximalrateofcryptoccpoccpaclgcrbowth.PObcPMAXisthemaximumtheoreticalfungaldensity.C n
isthegcmoanxcentrationsofflubendazolethatinducehalf-maximaleffectsongrowth.Hgistheslopefunctionfor50g lo
a
theeffectofflubendazoleongrowth.ICisthedensityofCryptococcusimmediatelypostinoculation. d
e
d

f
in fungal burden relative to that of controls (Fig. 4). This regimen was limited by r
o
maximum permissible volumes for oral administration for mice (i.e., 20 ml/kg). In a m

singleexperimentinwhichtheeffectof6mg/kgevery12h(q12h;i.e.,12mg/kg/day) h
t
was compared to that of 12 mg/kg/day, there was no difference in antifungal effect tp
:
/
/
a
a
c
.
a
s
m
.
o
r
g
/

o
n

M
a
y

4
,

2
0
1
9

b
y

g
u
e
s
t
FIG4 Pharmacodynamicsofflubendazoleinamurinemodelofcryptococcalmeningitis.Flubendazolewasadministeredorallyoncedaily.Data(opensquares)
aremeans(cid:4)SDsfor3mice.Thesolidlineisthefitofthepopulationpredictedpharmacokinetic-pharmacodynamicmodel.Themaximallyadministereddose
inthisstudy(12mg/kg/day)slowedbutdidnotpreventfungalgrowthinthebrain.
April2018 Volume62 Issue4 e01909-17 aac.asm.org 6
PharmacodynamicsofFlubendazoleforCryptococcus AntimicrobialAgentsandChemotherapy
D
o
w
n
lo
a
d
e
d

f
r
o
m

h
t
t
p
:
/
/
a
a
c
.
a
s
m
.
o
r
g
FIG5 Pharmacodynamicsofflubendazoleinarabbitmodelofcryptococcalmeningitis.(A)TimecourseoffungalburdenintheCSFofuntreatedcontrols;(B) /
timecourseoffungalburdenintheCSFrabbitstreatedwithflubendazoleat6mg/kgq24horally;(C)timecourseoffungalburdenintheCSFrabbitstreated o
withflubendazoleat22.5mg/kgq24horally;(D)fungalburdeninthecerebrumofrabbitsattheendoftheexperiment(288hpostinoculationandafter10 n
daysoftreatmentwithflubendazole).Therearenodifferencesinthethreegroups(P(cid:3)0.464,analysisofvariance). M
a
y

4
,
(data not shown). This is preliminary evidence that the AUC is likely to be the
2
dynamicallylinkedindexforflubendazoleagainstCryptococcusneoformans. 0
1
Pharmacokineticandpharmacodynamicstudieswithrabbits.ThePKinrabbits 9

b
was linear, with a concentration-time profile similar to that observed in mice. The
y
plasmaconcentration-timeprofilesinrabbitshadasimilarshapetothoseofmicebut g
u
were lower for the dosages used in this study. Despite readily quantifiable plasma e
s
concentrations,therewerenoquantifiabledrugconcentrationsineitherthecerebro- t
spinalfluid(CSF)orthecerebruminrabbitsatthetimeofsacrifice.
There was no demonstrable antifungal effect in rabbits receiving 6 mg/kg/day.
Theremayhavebeensomeeffectinrabbitsreceiving22.5mg/kgq24h,butifpresent,
the effect was small, and these assessments were limited by the use of few animals.
There were no statistically significant differences in the area under the concentration
(log CFU/g)-time curve for each regimen, even though this may be a relatively
10
insensitivetestofantifungaleffect.Furthermore,therewasnodifferenceinthefungal
burdeninthecerebrumattheendoftheexperimentforanyofthegroupsofrabbits
usedinthisstudy(Fig.5).
DISCUSSION
Whengivensubcutaneously,flubendazolehasstrikingactivityinlaboratoryanimal
modelsoffilarialdiseasessuchasonchocerciasisandlymphaticfilariasis(16).Janssen
developed a novel amorphous solid drug nanodispersion to provide a potential new
therapeutic option for patients with these neglected tropical diseases. The systemic
April2018 Volume62 Issue4 e01909-17 aac.asm.org 7
Nixonetal. AntimicrobialAgentsandChemotherapy
drugexposurethatwasenabledbythenewformulationmandatedGoodLaboratory
Practice (GLP) toxicology studies before progression to early-phase clinical studies. It
wasalreadyknownthatflubendazoleisclastogenic(i.e.,induceschromosomalbreak-
ages) and aneugenic (i.e., induces aneuploidy), as well as embryotoxic (17). GLP
toxicology studies were performed by Janssen with rats (5, 15, and 30 mg/kg/day in
male rats and 2.5, 5, and 10 mg/kg/day in female rats) and dogs (20, 40, and 100
mg/kg/day)for2weeks.Theseexperimentsshowedevidenceoftoxicityrelatedtothe
pharmacologicalactivityofflubendazoleinthegastrointestinaltract,lymphoidsystem,
and bone marrow, as well as testicular toxicity in both rats and dogs. In dogs, liver
toxicitywasalsoobserved.Asaresult,thedevelopmentprogramwasstoppedbased
on an unacceptable risk/benefit profile in humans. This also halted our own efforts
to develop flubendazole for cryptococcal meningitis.
FlubendazolehasstrikinginvitroactivityagainstCryptococcusneoformansthatwas
evident in the MIC testing and the pharmacodynamic studies in the hollow-fiber D
o
infectionmodel.Therewasmodestantifungalactivityinthemurinemodel,whichisnot w
as prominent as that previously described by us for fluconazole, amphotericin B n
lo
deoxycholate, or liposomal amphotericin B (8, 18, 19). There was no unequivocal a
antifungal activity in the rabbit model of cryptococcal meningoencephalitis, which is d
e
largely explained by the absence of detectable flubendazole concentrations in the d

f
cerebrum or CSF (despite readily quantifiable plasma concentrations). The in vitro r
o
susceptibilitytestinganddatafromthehollow-fibermodelsuggestthatflubendazole m

is highly potent and fungicidal if able to reach its fungal target in sufficient concen- h
t
trations.Thediminishedactivityinmice(relativetohistoricalcontrols)andabsenceof tp
:
effectinrabbits(withnonquantifiableconcentrationsinthecerebrumandCSF)further //
a
supportthisconclusion.Hence,successfulexploitationofthebenzimidazolebackbone a
c
requirescarefulattentiontophysiochemicalpropertiesthatpromoteabsorptionacross .
a
thegutandtheabilitytopartitionintosubcompartmentsofthecentralnervoussystem s
m
(CNS).
.
o
Flubendazole did not display a degree of in vivo activity comparable to those of r
g
otherfirst-lineagentsforcryptococcalmeningitis(i.e.,fluconazoleandamphotericinB /

o
formulations).Evenifthesafetyprofilewasnotproblematic,thereisinsufficientprima n

facieevidencefromeitherthemurineorrabbitmodeltofurtherstudyflubendazoleas M
monotherapyforinductiontherapyinphaseIIclinicalstudies.Nevertheless,additional a
y
approaches, such as combination with other antifungal agents for induction therapy 4
,
and/or use as longer-term consolidation and maintenance therapy, may have been 2
possible. 0
1
Thepotentialofderivativesofflubendazoletobeusefulhumanmedicinesdepends 9

b
onthedifferentialactivitybetweencryptococcalandhumanproteins.Characterization
y
ofthe(cid:2)-tubulingenesofC.neoformanshasbeenundertaken,andtwoC.neoformans g
(cid:2)-tubulin genes (TUB1 and TUB2) have been identified. TUB1 was identified as the ue
s
primarytargetofthebenzimidazoleclassofcompoundsthroughgenecharacterization t
and expression (14). There is 90% homology between fungal TUB1 and human
(cid:2)-tubulin, although the former has not been crystallized and this has prevented
definitive structure-activity-relationship docking studies. The ability to develop new
agentsbasedonabenzimidazolescaffoldortofurtherexploit(cid:2)-tubulinasapharma-
cological target will depend on the degree of differential activity of a benzimidazole
with these proteins. The differential binding identified through the docking and
homology modeling of both human (cid:2)-tubulin (20) and C. neoformans var. grubii
serotypeA(strainH99)(cid:2)-tubulin(14)totheBosTaurus1SA0(cid:2)-tubulincrystalstructure
impliesanincreasednumberofbindinginteractionswithC.neoformans(cid:2)-tubulin.This
may provide the potential to exploit this differential binding to establish a favorable
therapeutic index. It is also worth emphasizing that the benzimidazoles may have
additional targets beyond (cid:2)-tubulin that have the further potential to provide differ-
entialactivitybetweenhumanandfungalproteins,butthisrequiresfurtherinvestiga-
tion(21–24).
Thepotentialutilityofcongenersofflubendazolenowrestswithmedicinalchem-
April2018 Volume62 Issue4 e01909-17 aac.asm.org 8
PharmacodynamicsofFlubendazoleforCryptococcus AntimicrobialAgentsandChemotherapy
istry programs. Compounds must be synthesized that exhibit differential activity
against cryptococcal and human tubulin (if that is possible) so that there is an
acceptablesafetymarginandtoxicityprofile.Furthermore,thecompoundmustbeable
to traverse the gut (compounds that are not orally bioavailable will be less clinically
valuable) and then the blood-brain barrier to achieve concentrations that are ideally
fungicidal. The latter will be promoted by new low-molecular-weight, lipophilic com-
pounds that are not substrates for efflux pumps such as P-glycoprotein. This will
undoubtedly also require the use of novel formulation technologies to ensure that
compoundsthatarepoorlysolublecanbecomeusefulagentsfordisseminatedinfec-
tions.
MATERIALSANDMETHODS
Drug.FlubendazolethatwasusedfordeterminationofMICs,hollow-fiberexperiments,andprelim-
inary murine experiments was purchased from Sigma. Subsequently, definitive pharmacokinetic-
D
pharmacodynamicexperimentsthatwereperformedwithorallyadministeredflubendazoleusedasolid
o
drugnanodispersionformulationdevelopedbyJanssenPharmaceuticals(batchBREC-1113-070;Janssen w
Pharmaceuticals).Thestabilityofthisformulationinliquidandsolidphaseswasconfirmedfor1and6 n
months,respectively. lo
A10-mg/mlmethylcellulose(4,000-cps)stocksolution(100-mlbatch)waspreparedfromadispersion a
d
of1gofmethylcellulose(4,000cps)withstirringinto70mlofdemineralizedwaterheatedto70°Cto e
80°C.Thesolutionwasstirredforatleast15min,followedbytheadditionof20mlofdemineralized d

water.Themixturewasstirreduntilitreachedroomtemperatureandwasthenmadeupto100mlwith f
r
demineralizedwater.Atotalof50mlof6-mg/mlspray-driedpowdersuspensionwasthenprepared(this o
m
correspondsto0.6mg/mlofflubendazoleastheactivedosein0.5%methylcellulose).Atotalof24.70

gofdemineralizedwaterwasaddedtoa50-mlclearglassvial.Atotalof0.30gofspray-drieddrugwas h
t
addedtothevial,whichwasthenclosedwithastopper.Thevialwasvortexedandthenhomogenized tp
usingaPolytrondisperser.A25-mlstocksolutionofmethylcellulose(4,000cps)wasaddedandthevial :
/
wasvortexed.Thesuspensionwasrefrigeratedat5°Cuntildosingforamaximumof14days.Priorto /a
dosing,thesuspensionwasvortexed. a
c
Strains. The initial in vitro susceptibility testing was performed with H99 (ATCC 208821). An .
a
additional49clinicalisolateswereobtainedfromtheNationalCentreforMicrobiologyInstitutodeSalud s
CarlosIII,Madrid,Spain(courtesyofAnaAlastruey-IzquierdoandManualCuenca-Estrella).Theseisolates m
wereidentifiedtothespecieslevelusingstandardmicrobiologicaltechniques. .o
MICs.TheMICofflubendazoleagainstH99(ATCC208821)andthe49isolateswasestimatedusing rg
methodologyoftheEuropeanCommitteeonAntimicrobialSusceptibilityTesting(EUCAST[25])andthe /

ClinicalandLaboratoryStandardsInstitute(CLSI[26]).TheendpointforMICdeterminationusingwas o
n
50%forboththeEUCASTandCLSImethods.MICswereperformedintriplicate.
M
Porcinetubulinpolymerizationassay.Porcinetubulinisgenerallyusedasasurrogateforhuman
a
tubulinbecauseofitshighdegreeofhomology(95%)(27).Inthestudiesdescribedhere,thisassaywas y
usedtodeterminetheextentofinteractionbetweenflubendazoleanditsputativetarget;ithasalso 4
beenusedforassessmentofthebindingofantineoplasticagents(28,29).Thecommerciallyavailable ,
2
porcinetubulinassay(BK011P;Cytoskeleton,Inc.,Denver,CO)quantifiesthetime-dependentpolymer-
0
izationoftubulintomicrotubulesandthustheabilityoftubulininhibitorstodisruptthisprocess. 1
Theporcinetubulinassaywasperformedaccordingtothemanufacturer’sinstructions.Briefly,the 9

96-wellassayplatewasprewarmedto37°Cpriortouse.Fivemicrolitersoftestcompound(s)andcontrols b
y
at0,1.25,2.5,5,and10(cid:3)Mwerealiquotedintoeachwellandprewarmedfor1min.Colchicineand
g
DMSOwereusedaspositiveandnegativecontrols,respectively.Polymerizationwasinitiatedbymixing u
45(cid:3)lofreactionbufferthatcontained2mg/mlofpurifiedporcinebraintubulin,10(cid:3)Mfluorescent e
s
reporter,PEMbuffer(80mMPIPES,0.5mMEGTA,2mMMgCl2[pH6.9]),1mMGTP,and20.3%glycerol. t
Tubulinpolymerizationwasfollowedbyanincreaseinfluorescenceintensityduetotheincorporation
ofafluorescencereporterintomicrotubulesaspolymerizationoccurred.Thechangeinfluorescencewas
measuredusinganexcitationandemissionwavelengthof360nmand450nm,respectively,every1min
for1hat37°CusingaVarioskanmultimodeplatereader(ThermoscientificInc.).Alldatapointswere
acquiredintriplicate,andIC swerecalculatedwithGraphPadPrism.TheIC wasdefinedasthedrug
50 50
concentrationrequiredtoinhibittubulinpolymerizationby50%comparedwiththenegativecontrol.
Homology modeling and docking studies. While the amino acid sequence of cryptococcal
(cid:2)-tubulinisknown(74%homologywithhuman(cid:2)-tubulin),theproteinhasnotbeencrystallized.A
homology model was therefore developed to investigate differential binding modes of flubendazole
withinC.neoformansandhuman(cid:2)-tubulin.Molecularmodeling(Modelerversion9.14;https://salilab
.org/modeller/) of both human (cid:2)-tubulin (20) and C. neoformans var. grubii serotype A (strain H99)
(cid:2)-tubulin(14)wasundertakenusingtheBosTaurus1SA0(cid:2)-tubulincrystalstructure(identity,364/447
[81.4%];similarity,405/447[90.6%]).
VirtualflubendazolewasbuiltinthemolecularmodelingsoftwareSpartan(WavefunctionInc.,Irvine,
CA)andenergyminimized.Flubendazolewasthensubjectedtoapiecewiselinearpotential(ChemPLP)
docking protocol (a scoring function to provide confidence in the docking pose adopted by the
molecule), consisting of 10 genetic algorithm (GA) runs before visualization using the molecular
visualizationsystemPyMOLwiththetopscoringcompounddepictedinFig.1.Theactive-sitebinding
interactionswereselectedbyidentifyingthoseaminoacidresidueswithin4Åofflubendazolewhen
April2018 Volume62 Issue4 e01909-17 aac.asm.org 9
Nixonetal. AntimicrobialAgentsandChemotherapy
dockedintothe(cid:2)-tubulinbindingsite.Polarcontactsbetweenflubendazoleandthesurroundingamino
acidswereidentified,whichaidedintheidentificationofhydrogenbondinginteractionsthatarekeyin
determiningtheefficacyofadrugagainstitspharmacologicaltarget.
Finally,hydrogenbonddonorinteractionsaswellashydrophobicinteractionswereidentifiedusing
thepharmacophore(i.e.,anabstractdescriptionofmolecularfeaturesthatinthiscasearenecessaryfor
molecular recognition of flubendazole by (cid:2)-tubulin) search software ZINCPharmer (30) at 4 Å for
hydrogenbondinginteractionsand6Åforhydrophobicinteractions.
Hollow-fiber model of cryptococcal meningoencephalitis. A new hollow-fiber infection model
wasdevelopedtoinvestigatetheinvitropharmacodynamicsofflubendazoleagainstC.neoformans.The
samecartridges(FiberCellSystems,Frederick,MD)andconfigurationaspreviouslydescribedforbacterial
pathogens was used (see, for example, reference 31). The extracapillary space of each cartridge was
inoculatedwith40mlofasuspensioncontaining6log CFU/mlofC.neoformansvar.grubii(ATCC
10
208821;H99).Yeastextract-peptone-dextrose(YPD)mediumwaspumpedfromthecentralcompartment
throughthecartridgeandbackagainusingaperistalticpump(205U;Watson-Marlow,UnitedKingdom).
Thehollow-fiberinfectionmodelwasincubatedat37°Cinambientair.Thetimecourseoffungalgrowth
wasdeterminedbyremoving1mlfromtheextracapillaryspaceofthecartridgeandplatingserial10-fold
dilutionstoYPDagar. D
Therelationshipbetweenflubendazoleexposureanditseffectwasexploredusingarangeofdrug o
exposures. Since there is no information on the pharmacokinetics of flubendazole in humans, we w
n
attempted to produce AUCs that were comparable to those observed in mice. Various doses of lo
flubendazolewereadministeredq24hbyinfusionover1hfor8daystothecentralcompartmentusing a
aprogrammablesyringedriver(Aladdinpump;WorldPrecisionInstruments,UnitedKingdom).Therewas d
e
a24-hdelayintheinitiationofflubendazoletherapypostinoculation.Togeneratefirst-orderpharma-
d
cokinetics, fresh YPD medium was pumped into the central compartment and the same volume of
f
drug-containing medium was simultaneously removed and discarded. Positive controls of currently ro
licensedagentswerenotstudiedintheseexperiments. m
Murine model of cryptococcal meningoencephalitis. A previously described (8) and well- h
characterizedmurinemodelofcryptococcalmeningitiswasusedtoinvestigatethepharmacodynamics tt
p
offlubendazole.AlllaboratoryanimalexperimentswereperformedunderUKHomeOfficeprojectlicense
:
PPL40/3630andwereapprovedbytheUniversityofLiverpool’sAnimalWelfareEthicsReviewBoard. //
a
MaleCD1micewerepurchasedfromCharlesRiverandwere20to30gatthetimeofexperimentation. a
Aninoculumof3(cid:6)108CFUin0.25mlwasusedforeachmouse.Groupsofmice(n(cid:3)3)wereserially c
.
sacrificed throughout the experimental period. The brains were removed and homogenized. Serial a
s
10-folddilutionswereplatedtoYPDagarsupplementedwithchloramphenicoltoenumeratethetotal m
fungalburden.Plateswereincubatedinairat30°Cforatleast48h. .
o
Studies of pharmacokinetics and pharmacodynamics of flubendazole in mice. Preliminary r
g
evidencefortheefficacyofflubendazolewasobtainedbydissolvingpurecompoundinavarietyof /

excipients that included cyclodextrin (F2G; Eccles, UK), DMSO (5%), and polysorbate 80 (10%) and o
n
injectingitsubcutaneouslyq24h.Ultimately,onlys.c.injectionwithTween80showedanyeffect.This

experimentprovidedtheimpetustofurtherexaminetheorallybioavailableformulationdevelopedby M
Janssen(seeabove). a
y
The pharmacokinetics of oral flubendazole was determined with two independently conducted
4
experiments.Treatmentwasinitiated24hpostinoculation.Dosesof2to12mg/kgwereused.Onlythe ,

first dosing interval was studied. A serial-sacrifice design was used with groups of 3 mice that were 2
0
sacrificedat0.5,1,2,8,and24hpostinoculation. 1
The pharmacodynamics of oral flubendazole was estimated over the course of three separate 9

independentlyconductedexperiments.Groupsof3miceweresacrificedat2,24,48,96,144,and168 b
y
hpostinoculation.Dosefindingstudieswereperformedusingflubendazoleat2,4,6,8,and12mg/kg
g
q24horally.TheupperdosagewaslimitedbythevolumerestrictionsformiceimposedbyHomeOffice
u
projectlicensePPL40/3630.Afourthexperimentcompared12mg/kgq24with6mg/kgq12htoexamine e
whethermorefractionatedregimensprovidedanyadditionalantifungaleffect. s
t
Rabbit model of cryptococcal meningitis. A previously described and well-characterized rabbit
modelofcryptococcalmeningoencephalitis(32)wasusedtofurtherinvestigatethepharmacodynamics
offlubendazole.MaleNewZealandWhiterabbitswerepurchasedfromHarlan.Rabbitsweighed2.5to
3kgatthetimeofexperimentation.Rabbitswereimmunosuppressedintramuscularlywithhydrocorti-
soneat10mg/kgonday(cid:5)1relativetoinfectionandthendailythroughouttheexperiment.
Cryptococcalmeningoencephalitiswasinducedwiththeintracisternalinoculationof0.25mlofa
suspensioncontaining3.8(cid:6)108CFU/mlundergeneralanesthesia(inducedwithmedetomidineand
ketamine).Thisinoculumresultsinprogressiveinfectionthatmanifestsasanincreaseinfungalburden
in the CSF and reproducible encephalitis. There was minimal clinical disease with no demonstrable
neurological signs in the experimental period. Mortality always occurred in the context of cisternal
tappingandrepeatedanesthesiaratherthanfromprogressiveinfection.
Studiesofpharmacodynamicsandpharmacokineticsinrabbits.PK-PDrelationshipsintherabbit
wereestimatedintwoindependentlyconductedexperimentsincluding6rabbitsineach.Rabbitswere
placedundergeneralanesthesiaforremovalofCSFviaintracisternaltappingat48hintervals.Overthe
courseofthetwoexperimentstherewere3controls(1rabbitdiedafterbeingtapped);flubendazolewas
givenat6mg/kgq24h(n(cid:3)6)and22.5mg/kgq24h(n(cid:3)6).Themaximumdosagethatwasusedwas
limitedbytheformulationprovidedbyJanssenandthelimitsoforalgavageinrabbist(15ml/kg/day).
Treatmentwasinitiated48hpostinoculationandcontinuedfor10days,afterwhichtimeallrabbitswere
sacrificed.Thus,thetotaldurationoftheexperimentwas288h.
April2018 Volume62 Issue4 e01909-17 aac.asm.org 10