Utility of Alkylaminoquinolinyl Methanols as New Antimalarial Drugs

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Dec. 2006, p. 4132–4143 Vol. 50, No. 120066-4804/06/$08.00�0 doi:10.1128/AAC.00631-06

Utility of Alkylaminoquinolinyl Methanols as New Antimalarial Drugs�

G. S. Dow,1* T. N. Heady,1 A. K. Bhattacharjee,1 D. Caridha,1 L. Gerena,1 M. Gettayacamin,3C. A. Lanteri,1 N. Obaldia III,4 N. Roncal,1 T. Shearer,1 P. L. Smith,1 A. Tungtaeng,3

L. Wolf,2 M. Cabezas,2 D. Yourick,2 and K. S. Smith1

Division of Experimental Therapeutics, Walter Reed Army Institute of Research, 503 Robert Grant Ave., Silver Spring,Maryland 209101; Division of Neuroscience, Walter Reed Army Institute of Research, 503 Robert Grant Ave.,

Silver Spring, Maryland 209102; Department of Veterinary Medicine, United States Army Medical Component,Armed Forces Research Institute of Medical Sciences, 315/6, Rajthevi, Bangkok, Thailand 104003;

and Tropical Medicine Research/Gorgas Memorial Research Institute,Ave. Justo Arosemena no. 3530, Panama City, Panama4

Received 23 May 2006/Returned for modification 8 July 2006/Accepted 30 August 2006

Mefloquine has been one of the more valuable antimalarial drugs but has never reached its full clinicalpotential due to concerns about its neurologic side effects, its greater expense than that of other antimalarials,and the emergence of resistance. The commercial development of mefloquine superseded that of anotherquinolinyl methanol, WR030090, which was used as an experimental antimalarial drug by the U.S. Army in the1970s. We evaluated a series of related 2-phenyl-substituted alkylaminoquinolinyl methanols (AAQMs) fortheir potential as mefloquine replacement drugs based on a series of appropriate in vitro and in vivo efficacyand toxicology screens and the theoretical cost of goods. Generally, the AAQMs were less neurotoxic andexhibited greater antimalarial potency, and they are potentially cheaper than mefloquine, but they showedpoorer metabolic stability and pharmacokinetics and the potential for phototoxicity. These differences inphysiochemical and biological properties are attributable to the “opening” of the piperidine ring of the4-position side chain. Modification of the most promising compound, WR069878, by substitution of anappropriate N functionality at the 4 position, optimization of quinoline ring substituents at the 6 and 7positions, and deconjugation of quinoline and phenyl ring systems is anticipated to yield a valuable newantimalarial drug.

In the late 1960s to early 1970s, Plasmodium falciparummalaria in Southeast Asia had begun to develop resistance toall of the available antimalarial drugs (6). Cure rates were 11 to20% and 26 to 50% for chloroquine and quinine, respectively,and had declined to only 90% for the triple combination ofquinine/pyrimethamine/dapsone (6). All of these regimenswere associated with adverse side effects (6). As a conse-quence, the U.S. Army began routinely employing two exper-imental antimalarial drugs, WR030090 and WR033063, for thetreatment of recrudescent malaria infections at the WalterReed Army Medical Center (6). Subsequent field trials dem-onstrated that WR030090, a quinolinyl methanol, exhibitedcure rates of at least 88% and was better tolerated than quinine(6, 21).

Shortly thereafter, mefloquine was discovered and was de-veloped commercially by Hoffman La Roche and the U.S.Army. Mefloquine exhibited a long half-life in humans, andthis desirable property facilitated its administration as a singledose for malaria treatment and as a once-weekly dosing forprophylaxis (50). In contrast, WR030090 was only partiallyeffective as a prophylactic agent, required a dosing regimensimilar to that of quinine to effect cures, and was subsequentlyabandoned (6, 9, 21, 30). However, it is important to recognize

that this occurred because of unfavorable pharmacokineticcharacteristics, not as a consequence of unacceptable toxicity.

Mefloquine combined with artesunate constitutes one of themost effective combination agents for treatment of malaria(41). Mefloquine is also the only once-weekly drug approvedfor malaria chemoprophylaxis in the United States that, bar-ring the Thai border regions, is effective in almost all areas ofthe world (7). However, mefloquine is relatively expensivecompared to other antimalarials (8), which limits its accessi-bility to developing countries. Also, mefloquine use is associ-ated with debilitating neurological effects in a small proportionof patients, and milder but nevertheless concerning effects arealso frequently observed (31, 40, 43, 47). These negative char-acteristics have limited the scope of the possible clinical utilityof the drug.

In the present study, we report the antimalarial activity andpharmacological properties of a number of 2�-substituted alkyl-aminoquinoline methanols (AAQMs) related to WR030090. Ourgoals were (i) to develop an understanding of structure-activityrelationships that might allow the synthesis of improved quino-linyl methanols and (ii) to determine whether compounds relatedto WR030090 might prove to be more efficacious antimalarials.

MATERIALS AND METHODS

In vitro antimalarial activities of AAQMs. The in vitro activities of AAQMsagainst P. falciparum strains W2, D6, TM91C235, and TM90C2A were evaluatedusing the method of Desjardins et al. (12), as modified by Milhous et al. (32) anddescribed in one of our earlier papers (16). W2 is chloroquine resistant andmefloquine sensitive, D6 is chloroquine sensitive but naturally less susceptible to

* Corresponding author. Mailing address: Division of ExperimentalTherapeutics, Walter Reed Army Institute of Research, 503 RobertGrant Ave., Silver Spring, MD 20910. Phone: (310) 319-9009. Fax:(301) 319-9954. E-mail: [email protected].

� Published ahead of print on 11 September 2006.

4132

mefloquine, and TM91C235 is resistant to mefloquine, chloroquine, andpyrimethamine, as is TM90C2A; however, this last parasite is a two-pfmdr1-copystrain. We routinely run mefloquine in this screen as a control to ensure assayvalidity. Mefloquine has a mean 50% inhibitory concentration (IC50) � standarddeviation (SD) against P. falciparum TM91C235 of 15.7 � 2.7 ng/ml (last 15assays).

Antimalarial activities of AAQMs with mice and monkeys. The Plasmodiumberghei mouse efficacy data were obtained from the WRAIR chemical informa-tion system (subcutaneous testing) or from recent tests conducted at AFRIMS,Thailand, or at the University of Miami using a modified version of the Thomp-son test. Basically, groups of five mice were inoculated through intraperitonealinjection on day 0 (usually with 1 � 106 P. berghei-parasitized erythrocytes).Drugs were administered either subcutaneously or orally for 3 days (usually ondays 3 to 5) at doses of 1.25 to 160 mg/kg of body weight in two- to fourfoldincrements. Cure was defined as survival until day 60 (subcutaneous dosing) orday 31 posttreatment (oral dosing). Nontreated control mice usually die on days6 to 10 postinfection. The minimum effective dose was the lowest dose level thatcured at least one of five mice. The Aotus studies were performed as previouslydescribed (35, 36). Briefly, groups of two Aotus lemurinus lemurinus male andfemale monkeys of the karyotype VIII or IX (27) with weights ranging from 742to 970 g were inoculated with 5 � 106 parasites of either the FVO strain of P.falciparum or the AMRU1 strain of Plasmodium vivax. Both of these strains arechloroquine resistant. The course of infection with P. falciparum is usually lethal,while the AMRU1 P. vivax strain induces a potentially lethal thrombocytopeniaif left untreated. Monkeys were examined, and thick Giemsa-stained bloodsmears were prepared and enumerated daily (17) to monitor the course ofinfection. When parasitemias increased to greater than 5,000 parasites per �l,monkeys were treated orally with the test drug. If drug treatment failed, i.e.,parasitemia did not decline or increased again to �5,000 parasites per �l, themonkeys were rescued with a single dose of orally administered mefloquine (20mg/kg). In some instances, a high dose of the test drug was used to retreatmonkeys instead of the mefloquine rescue. Monkeys were considered cured ifthey were parasite free 90 days posttreatment. For the P. falciparum studies, thedose rate selected was 10 mg/kg/day � 7, since earlier studies suggested thatWR030090, a compound that later proceeded to clinical studies, cleared but didnot cure infections at this dose (WRAIR archival data). Cure at this dose forrelated analogs would therefore be a good indicator of their superiority over acompound that had already proceeded into clinical development. Plasma sam-ples were taken from the monkeys in some studies for quantification of metab-olites. Additional blood samples were taken as appropriate for complete bloodcounts and serum chemistry. Research was conducted in compliance with theAnimal Welfare Act and other federal statutes and regulations relating to ani-mals and experiments involving animals, and it adheres to principles stated in theGuide for the care and use of laboratory animals, NRC publication, 1996 edition(34).

Metabolic stability analysis and metabolite determination. Mefloquine ana-logs (10 �M) were added to a mixture containing an NADPH-regeneratingbuffer (1.25 mM �-NADP�, 3.3 mM glucose-6-phosphate, and 3.3 mM MgCl2)and 0.5 mg/ml pooled human liver microsomes to a final volume of 125 �l. Themixtures were incubated for 5 min at 37°C, and the reactions were initiated byadding 25 �l glucose-6-phosphate dehydrogenase to a final concentration of1 unit/ml. The reactions were maintained at 37°C until they were terminated bythe addition of an equal volume of 100% ice-cold acetonitrile at 0, 10, 30, 60, and120 min. Samples were centrifuged to pellet the proteins, and the supernatantwas analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) in duplicate using a fast LC gradient or isocratic method. The parent drugwas quantified with external calibration, using plots of the parent drug responseversus the amount. Chromatograms were analyzed using the mass spectrometrysoftware Xcalibur QuanBrowser (for ThermoFinnigan instruments) or Mass-Lynx (for Waters instruments). The concentration of parent drug remaining ateach time point was calculated using the unknown peak areas and correspondingcalibration curves. In order to calculate the half-life, a first-order rate of decaywas assumed. A plot of the natural log of the drug concentration versus time wasgenerated, where the slope of that line was �k. The half-life was calculated as0.693/k. The positive control tested with each of these assays was nifedipine,which exhibited a mean half-life � SD of 31.7 � 5.3 min with human livermicrosomes and 27.6 � 2.6 min with mouse liver microsomes (based on fiveassays). Mefloquine, which has been run �5 times in this assay, consistentlyexhibits a half-life of �120 min in the presence of both human and mousemicrosomes. All reagents were purchased from Sigma except for the microsomes,which were obtained from BD Gentest.

For metabolite identification, samples were prepared as described above withhuman liver microsomes. Additional samples were prepared with each drug using

mouse, rat, and rhesus monkey liver microsomes. Samples were separated usingan LC gradient method and analyzed by full-scan LC-MS and LC-MS/MS. Theparent compound and putative metabolites were all fragmented, and theseMS/MS experiments were used in combination with the no-NADPH controlexperiments to confirm the assignment of peaks as metabolites. These MS/MSdata were also used to do preliminary structural elucidation. Although dealkyl-ated metabolites could be straightforwardly identified, the regioselectivity ofother modifications, specifically the position(s) of hydroxylation, could not bedefinitively ascertained. The relative percentage of formation of each metabolitewas determined in a semiquantitative manner, since standards of each metabolitewere not available. Peak areas of each metabolite and the internal standard weredetermined, and their ratios were calculated as metabolite area/internal standardarea. The percent formation of each metabolite was determined as the area ratiodivided by the sum of all the metabolite area ratios.

Cytochrome P450 inhibition assays. Two assays were used to evaluate thedrug-drug interaction potential of the compounds. An LC-MS-based assay wasfirst used as a prescreen of all the compounds against each isozyme, and thenconfirmation of significant inhibition was performed using the standard fluoro-metric assay. The LC-MS assay was conducted in a 96-well format using acocktail of four cytochrome P450 (CYP) substrates specifically metabolized byfour of the CYP isozymes. A mixture of tolbutamine (106 �M), omeprazole (10.6�M), bufuralol (10.6 �M), and midazolam (10.6 �M) was used to evaluate theactivities of CYP2C9, CYP2C19, CYP2D6, and CYP3A4, respectively. The testcompounds were incubated for 5 min at four different concentrations (singlereplicates) in the presence of the CYP substrate cocktail and 0.5 mg/ml humanliver microsomes in NADPH-regenerating buffer. The reaction was initiated bythe addition of 1 U/ml glucose-6-phosphate dehydrogenase. The reactions werestopped after 25 min by the addition of an equal volume of 100% ice-coldmethanol containing a dextromethorphan internal standard (5 �M). The sampleswere then centrifuged and transferred to a new plate for analysis.

Quantification of the production of specific metabolites from each CYP sub-strate was performed using LC-MS. Known inhibitors for each isoenzyme (quin-idine, sulfaphenazole, tranylcipromine, and ketoconazole) were used as positivecontrols for the inhibition of specific enzymatic activity. The inhibition obtainedfrom the test drug was expressed as a percentage of results for the controlswithout inhibitors. In general, inhibition of enzymatic activity seen at test drugconcentrations of �50 �M was not considered to be significant enough forfurther evaluation. Values between 10 and 50 �M were considered to be mod-erate inhibition, and values lower than 10 �M were significant.

Confirmation of CYP2D6 inhibition by phenylquinolinyl methanols was doneusing a modified version of the standard fluorometric assay recommended by thesupplier (BD Gentest). Briefly, the assay used the cDNA-expressed CYP2D6isoenzyme at 15 pmol/ml protein and 3-[2-(N,N-diethyl-N-methylamino)ethyl]-7-methoxy-4-methylcoumarin as the 2D6-specific fluorescent substrate. The re-action mixtures were placed in a 96-well plate with both a quinidine control anda range of concentrations of the test compound. The plate was incubated for 30min at 37°C, and the production of 3-[2-(N,N-diethylamino)ethyl]-7-hydroxy-4-methylcoumarin hydrochloride, the fluorescent metabolite, was quantified usinga fluorescence plate reader. Inhibition at each concentration was calculated asdescribed above. IC50s were then calculated using the two concentrations thatbracket 50% inhibition. IC50s for the control, quinidine, were in each case foundto be within the accepted range of 0.001 to 0.05 �M as specified by the manu-facturer.

Neurocytotoxicity, neuroprotection assays, and confocal microscopy. The neu-rocytotoxicity assay of AAQMs was conducted as previously described (16). Thisassay utilizes primary rat forebrain neurons and is a multi-end-point screen. Inthis system a large component of the neurotoxicity of mefloquine can be attrib-uted to the disruption of calcium homeostasis via discharge of the endoplasmicreticulum calcium store and activation of ill-defined plasma membrane calciumchannels (14, 15). Mefloquine may also induce other, uncharacterized effects inthese cells. Compounds were screened at 10, 100, and 1,000 �M in triplicate, andthe reduction in viability observed was determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as previously described (16). Ap-proximate IC50s were calculated on the basis of the level of inhibition observedat each concentration. Full dose-response assays were run for compounds ofinterest (e.g., WR069878) with two-fold dilutions (n 6 wells per dilution) ofeach drug. IC50s were calculated using Prism. Each time the assay was conducted,we routinely included 20 �M mefloquine as a control. This concentration ofmefloquine reduces cellular viability by a mean value � SD of 52% � 4.4%.Assays are rerun if the loss of viability induced by mefloquine is 40% or �60%.

Neuroprotection experiments with 6,7-dinitroquinoxaline-2,3-dione (DNQX)and magnesium were performed as described previously (14). Basically, neuronswere exposed to DNQX (100 �M) or magnesium (12 mM) for 5 min, followed

VOL. 50, 2006 AAQMs AND ANTIMALARIAL THERAPY 4133

TA

BL

E1.

Stru

ctur

esan

dbi

olog

ical

data

for

vari

ous

alky

lam

inoq

uino

lines

Com

poun

d

Subs

titut

ion

atpo

sitio

na:

Pred

icte

dso

lubi

lity

atpH

5.5

(mg/

ml)

g

Log Pg

Act

ivity

agai

nst

P.f

alci

paru

mT

M91

C23

5(I

C50

�ng/

ml�)

Neu

roto

xici

tyra

nkin

g(I

C50

��M

�)

Met

abol

icst

abili

ty(h

alf-

life

�min

�)

CY

P2D

6in

hibi

tion

(IC

50

[�M

])

TO

PKA

T(n

o.of

pass

edte

sts/

30)

SCac

tivity

inm

ice

(min

imum

effe

ctiv

edo

se�m

g/kg

/day

�)h

Ora

lact

ivity

inm

ice

(min

imum

effe

ctiv

edo

se�m

g/kg

/day

�)i

Ora

lact

ivity

inA

otus

k

2a2b

2c3

4a4b

4c6

78

Mefl

oqui

neC

F3b

HPi

peri

dine

dC

F3

HH

8.29

2.87

1520

�12

018

.610

205j

Cur

eW

R03

0090

HC

lC

lH

Bu

Bu

Cl

Cl

Cl

0.00

48.

1811

2,00

0�

120

NT

1020

5,5j

Fai

lure

WR

0698

78H

HC

lH

Bu

Bu

Me

Cl

H0.

019

8.04

1.5

200

3818

.215

202.

5,5j

Cur

eW

R17

6399

HH

Cl

HB

uH

Me

HM

e1.

115.

470.

4860

0�

120

3.5

1540

10R

-21

WR

0075

24H

HC

lH

Bu

HC

lH

Cl

0.28

5.46

0.63

60�

120

2.2

1480

1.

25,2

.5j

WR

0412

94H

HC

lH

Et

Et

Cl

HC

l1.

595.

041.

160

�12

00.

618

�64

010

WR

0810

49H

HC

lH

Bu

Bu

HF

H0.

088

7.01

1.6

200

240.

2417

160

2.5

WR

0350

58H

HC

lH

Bu

Bu

Cl

HH

0.03

77.

182.

760

038

0.3

1640

1.25

Cur

eW

R09

8656

HH

Cl

HB

uB

uO

eC

lH

Cl

19.0

4.98

3.3

60�

120

0.34

3010

1.

25W

R07

4086

HH

FH

Bu

Bu

Cl

HC

l0.

043

7.10

3.4

2,00

046

2.8

1620

2.5j

R-2

5W

R21

1925

HH

Cl

HB

uB

uH

HM

e0.

090

7.29

5.2

600

540.

6515

160

5.0

WR

0292

52H

HC

lH

Bu

Bu

Cl

HC

l0.

004

7.74

7.6

600

�12

014

52.

5jF

ailu

reW

R21

1679

HH

HH

Hex

Hex

HH

Cl

0.02

48.

637.

72,

000

649

ND

m

WR

1779

73T

Buc

HB

uB

uC

lH

H0.

436.

787.

920

015

1432

0�

40W

R10

6752

HH

IH

Bu

Bu

Cl

HC

l0.

029

8.17

8.0

2,00

0N

Tl

1540

WR

0531

88C

lH

Cl

HB

uB

uC

lH

Cl

0.00

38.

2318

2,00

089

1180

WR

0300

91H

HH

HH

exH

exC

lH

Cl

0.00

69.

2919

3,00

0�

120

916

0W

R03

0464

HH

Cl

HH

exH

exH

HC

l0.

007

9.20

193,

000

�12

09

160

WR

1067

51H

HO

HH

Bu

Bu

Cl

HC

l0.

26.

3922

608

16�

640

WR

1491

05H

HC

lF

Bu

Bu

Cl

HC

l0.

003

8.48

3660

0N

T12

10W

R02

9656

HH

Cl

HH

exH

exC

lH

Cl

0.00

29.

8611

73,

000

�12

07

160

WR

1317

39H

OH

HH

Bu

Bu

Cl

HC

l0.

196.

4218

120

034

16�

640

WR

0918

84H

HC

lH

Bu

Bu

Mef

Cl

HC

l47

1.48

277

60�

120

11�

640

WR

1400

89H

HC

lC

lB

uB

uC

lH

Cl

0.00

38.

4350

03,

000

NT

1140

aPo

sitio

nof

subs

titut

ion

onqu

inol

ine

ring

,as

inF

ig.1

.Gro

ups

2ato

-cde

note

eith

ersu

bstit

utio

nat

the

2po

sitio

nor

mod

ifica

tions

toth

eph

enyl

ring

atta

ched

toth

equ

inol

ine

ring

atth

e2

posi

tion.

Gro

ups

4ato

-cde

note

mod

ifica

tions

toth

eN

side

chai

nat

the

4po

sitio

n.B

u,bu

tyl;

Hex

,hex

yl.

bT

riflu

orom

ethy

lgro

upis

atta

ched

atth

e2

posi

tion

ofth

equ

inol

ine

ring

.c

t-B

utyl

grou

pis

atta

ched

toth

e2

posi

tion

ofth

equ

inol

ine

ring

.d

Pipe

ridi

neri

ngas

inF

ig.1

.e

N-o

xide

.fN

-met

hyl.

gC

alcu

late

dus

ing

AC

D/L

ogD

SolS

uite

.h

Min

imum

daily

dose

that

cure

dat

leas

ton

eof

five

P.b

ergh

ei-c

halle

nged

mic

e.D

rugs

wer

egi

ven

subc

utan

eous

lyfo

r3

days

.iM

inim

umda

ilydo

seth

atcu

red

atle

asto

neof

five

P.b

ergh

ei-c

halle

nged

mic

e.D

rugs

wer

egi

ven

oral

lyfo

r3da

ys.C

ompo

unds

wer

ete

sted

atth

eU

nive

rsity

ofM

iam

iunl

esso

ther

wis

ein

dica

ted.

Gre

ater

/less

than

sym

bols

indi

cate

that

the

min

imum

effe

ctiv

edo

sew

asou

tsid

eth

edo

sera

nge

test

edas

indi

cate

d.jT

his

com

poun

dw

aste

sted

atA

FR

IMS,

Tha

iland

.k

Clin

ical

outc

ome

ingr

oups

oftw

oP

.fal

cipa

rum

-cha

lleng

edA

otus

mon

keys

.Cur

em

eans

that

both

mon

keys

wer

epa

rasi

tefr

ee90

days

afte

rtr

eatm

ent.

Fai

lure

mea

nsth

atm

onke

ysw

ere

resc

ued

with

mefl

oqui

ne.

Rm

eans

that

infe

ctio

nsw

ere

clea

red,

and

the

num

ber

indi

cate

sth

eav

erag

eda

yof

recr

udes

cenc

efo

rbo

thm

onke

ys.

lN

T,n

otte

sted

due

toin

solu

bilit

yor

insu

ffici

ent

com

poun

d.m

ND

,not

dete

rmin

ed.

4134 DOW ET AL. ANTIMICROB. AGENTS CHEMOTHER.

by mefloquine (25 �M) or WR069878 (250 �M) for 20 min, after which thereduction in cell viability was determined. Neither DNQX nor magnesium al-tered cellular viability alone. Each combination of treatments was tested inquadruplicate on two occasions, and similar trends were observed each time.DNQX is an inhibitor of non-N-methyl-D-aspartate receptors, while magnesiuminhibits the functioning of the inositol 1,4,5-trisphosphate-mediated calcium sig-naling pathway at several points.

The effects of mefloquine and WR069878 on neuronal calcium homeostasiswere assessed utilizing confocal microscopy as described elsewhere (16). Briefly,neurons were loaded with the calcium-sensitive dye Fluo 3-A for 1 h and werewashed prior to image experiments. Neurons were “spiked” with 100 �Mmefloquine or WR069878. Subsequent changes in neuronal calcium homeostasiswere recorded as fluctuations in the emitted fluorescence of fluo-3-complexedcalcium at 530 nm (excitation was 488 nm). Sequential image scans of fieldscontaining 5 to 25 neurons were used to construct temporal profiles. Scans weremade at 10-s intervals. Fluorescence levels for each neuron were normalized totime-zero values. Data were then pooled from three and five independent ex-periments for mefloquine and WR069878, respectively.

Phototoxicity pharmacophore and 3T3 neutral red uptake (NRU) phototox-icity test. We attempted to develop a pharmacophore for phototoxicity to beused as an in silico screening tool. An earlier study reported the minimumphototoxic concentration for a range of different quinolinyl methanols in a yeastassay system (22). These data were used to generate a three-dimensional phar-macophore model using the HypoGen algorithm of the CATALYST (AccelrysSoftware, Inc.) methodology as previously described (2, 16). The structures ofclinically used quinoline methanols were mapped onto the pharmacophores, andestimated values for minimum phototoxic concentrations were generated.

The 3T3 NRU phototoxicity test, conducted by MB Research Laboratories(Spinnerstown, PA), was used to identify quinoline methanols that have thepotential to exert in vivo phototoxicity after systemic application. Briefly, thecentral 60 wells of 2 96-well plates per test compound were seeded with BALB/c3T3 mouse fibroblast cells and maintained in culture for 24 h. These plates werethen preincubated with a range of eight different concentrations of test com-pound (six wells per concentration) for 1 h. Next, one plate was treated with aUVA dose of 5 J/cm2 by irradiating for 50 min at 1.7 mW/cm2, whereas the otherplate remained nontreated and in the dark. Next, the treatment medium wasreplaced with culture medium, and cell viability was determined after 24 h bymeasuring neutral red uptake for 3 h. Finally, the 3T3 NRU Phototox Predictionsoftware (version 2.0; ZEBET) was used to calculate 50% effective concentra-tions and photo-irritant factors (PIF) for each compound. Compounds showingpotential for phototoxicity have a PIF of �5.0.

In silico toxicity screening. We utilized TOPKAT (see TOPKAT user guide6.1; Accelrys, Inc., San Diego, CA) for assessment of the 2-phenyl-substitutedAAQMs for their potential toxicity to human health and the surrounding eco-system by screening through 30 different models that are currently available inthe software. Data are presented as the total number of tests passed by eachcompound (see Table 1). This software package allows a rapid assessment ofpotential toxicity of chemicals solely from two-dimensional molecular structures.The qualitative models in TOPKAT provide dichotomous output (yes/no) for

rodent carcinogenicity, Ames mutagenicity, developmental toxicity, skin sensiti-zation, skin irritancy, ocular irritation, and aerobic biodegradability. The quan-titative models provide point estimates for the lowest observed adverse effectlevel, oral rat lethal dose, lethal concentration, maximum tolerated dose, andoctanol-water partition coefficient (V log P) along with 95% confidence limits foreach. Within a class of chemicals, a relatively high number of failed tests may beindicative of potential toxicity.

Physiochemical properties. Important physiochemical properties, includinglog P, log D, predicted solubility, polar surface area, pKa, the number of Hdonors and acceptors, etc., were calculated using the Advanced Chemistry De-velopment LogD Sol Suite.

Cost and ease of synthesis analysis. The bench-scale cost analysis for meflo-quine presented here is based on a 1971 synthesis (37). The syntheses ofWR069878 and WR030090 (see Fig. 5) are hypothetical. Each synthetic step inthese syntheses is strongly supported in the literature, and we have included theimmediately relevant literature citations for each step. The isatin starting mate-rial for WR069878 was not commercially available at the time of this publication.The cost analysis for WR069878 incorporates synthesis of the starting 3-chloro-4-methyl isatin. Yields for the hypothetical syntheses are averages from thepreviously reported values. The pricing of the bench-scale mefloquine synthesisand WR069878 and WR030090 was done using current reagent pricing of com-monly used laboratory quantities, not bulk reagent quantities (Aldrich catalog,2000-2001). As in the mefloquine synthesis, condensation of the appropriateisatin and ketone should afford the quinoline core (37). Conversion of thequinoline acid to the acid chloride could be accomplished with either thionylchloride or oxalyl chloride (25). Conversion of the acid halide to the -haloketone could be accomplished through the diazoketone (1, 19, 25, 26). Treat-ment of the -halo ketone with sodium borohydride should afford the epoxide (3,29). Lastly, the epoxide could be opened with dibutyl amine to give an enantio-meric secondary alcohol (4, 18, 33).

RESULTS

In vitro antimalarial activities of AAQMs. A substructuresearch of the WRAIR chemical library identified 21 com-pounds structurally related to WR030090. These compoundswere subjected to batteries of in vitro and in vivo tests to assesstheir potential utility as antimalarials. Many of the AAQMsexhibited much greater antimalarial activity and decreasedneurotoxicity in vitro relative to those of mefloquine (Table 1).These characteristics appear to be correlated with “opening”of the piperidine ring of mefloquine, since “open-chain” N,N-dialkylaminoquinolines display a therapeutic index superior tothat of 4-quinoline carbinolamines that contain phenyl sub-stituents at the 2 position (Fig. 1; Table 2). In general, those

FIG. 1. Structures of mefloquine and AAQMs. The AAQMS are substituted at the positions indicated and as described in Table 1.


alkylaminoquinoline compounds showing relatively potent invitro antimalarial activity and neurotoxicity either containedshort alkyl amino chains (e.g., WR041294 versus WR029252[Table 1]) or had one chain removed (e.g., WR007524 versusWR029252 [Table 1]). The in vitro efficacies of AAQMstracked with that of mefloquine, since in most cases IC50s werein the following rank order (highest to lowest IC50) for P.falciparum strains: TM90C2A, TM91C235, D6, and W2 (datanot shown). The most promising compound, WR069878,

exhibited 90% inhibitory concentrations (IC90s) againstTM90C2A, TM91C235, D6, and W2 of 16, 11.7, 5.3, and 0.49ng/ml, respectively. Comparable values for mefloquine were101, 89, 20, and 3.9 ng/ml.

Relative metabolic stabilities of AAQMs. Many of theAAQMs tested were less metabolically stable than mefloquinein vitro (Table 1). Singly alkylated analogs (secondary amines)showed metabolic stability similar to that associated with mef-loquine (WR176399 and WR007524) (Table 1). In contrast,many of the dialkylated analogs were less metabolically stable(Table 1). The N,N-dialkyl analogs were metabolized in allspecies primarily (�90%) by N dealkylation (26), yielding thecorresponding secondary amine metabolite, as with WR069878(Fig. 2). Hydroxylation was a minor but secondary route ofmetabolic transformation. Singly alkylated analogs were me-tabolized to a much lesser extent via hydroxylation (WR176399)and/or N dealkylation (WR041294) or not at all (WR007524).Metabolism was similar across all the species tested (rat, hu-man, mouse, and rhesus). In cytochrome P450 inhibition as-says, many of the analogs were more potent inhibitors of cy-

FIG. 2. Metabolic scheme for AAQMs. As indicated for WR069878, the N,N-dialkyl analogs were metabolized in all species primarily by Ndealkylation (�90%), yielding the corresponding secondary amine metabolite. Hydroxylation was a minor but secondary route of metabolictransformation. Singly alkylated analogs were metabolized to a much lesser extent via hydroxylation and/or N dealkylation.

TABLE 2. Alkylaminoquinoline methanols exhibit a greatertherapeutic index than 4-quinoline carbinolamines (4QCs)

Compound

Median IC50against

P. falciparumTM91C235

(nM)

Median IC50against rat

neurons(�M)

Therapeuticindex (103)

Therapeuticindex

relative tomefloquine

Mefloquine 35 20 0.57 1.0AAQMs 17 600 35 614QCs 26 33 1.3 2.3


tochrome CYP2D6 relative to inhibition caused by mefloquine(IC50 of 18.6 �M), and one compound (WR069878) showedinhibitory activity equivalent to that of mefloquine (Table 1).

In vivo antimalarial activity of AAQMs. In terms of in vivoactivity, almost all of the AAQMs exhibited some curativeeffects in the P. berghei-mouse model when administered sub-cutaneously (Table 1). With the exception of WR030090 andWR211679 (which had higher IC50s), all AAQMs with an IC50

of 10 ng/ml were screened orally with the mouse. With theexception of WR177973, all were substantially more effectiveby the oral route, with minimum curative doses ranging from1.25 to 10 mg/kg/day for 3 days for most analogs (Table 1),although one must consider the fact that for the oral dosingexperiment, survival was assessed at day 31 instead of day 60.The least neurotoxic of these were then tested for efficacyagainst P. falciparum in Aotus. Cures were observed withWR069878 and WR035058 (Table 1). WR074086 andWR176399 cleared parasitemias, but recrudescence was sub-sequently observed (Table 1). Clinical failure was associatedwith a high in vitro IC90 (�20 ng/ml) against P. falciparumTM91C235 and/or relatively low plasma concentrations (Table3). No monkeys became ill, died, or showed signs of toxicityafter treatment with the analogs. None of the observedchanges in complete blood counts, serum chemistry, or bodyweight were attributable to toxicity. Adverse changes in hemat-ocrit, hemoglobin, platelet, white blood cell, and reticulocytecounts were observed with malaria infection but resolved upontreatment with either mefloquine (rescue) or the AAQMs.

AAQM plasma concentrations in monkeys. Overall, plasmaconcentrations of the six AAQMs in Aotus monkeys were rel-atively low (mean peak concentrations of 28 to 413 ng/ml witha 10-mg/kg/day oral dose � 7) (Table 3) compared to what onewould usually observe after a single mefloquine treatment dose(peak concentrations of 800 to 8,000 ng/ml) (49). Lower thanexpected plasma concentrations for AAQMs may be due to acombination of several contributing factors. First, some of theAAQMs tested were less metabolically stable than mefloquine,which could potentially result in lower overall plasma concen-trations of the parent compound (Table 1). Second, most ofthe AAQMs were several orders of magnitude less soluble thanmefloquine (Table 1). It is likely that the higher solubility ofmefloquine is attributable to the more polar secondary amineof the piperidine ring, whereas most of the other AAQMscontained more aliphatic tertiary amines. Finally, the predictedlog P values for most of the AAQMs were substantially higherthan that for mefloquine and, furthermore, substantiallygreater than that normally observed for commercially available

drugs (Table 1) (24). Indeed, poor solubility and permeabilityare the two leading causes of poor drug absorption (24).

WR069878 selected for further evaluation with monkeys.WR069878 was chosen for further Aotus studies, since, of thetwo curative analogs, it exhibited the least-potent inhibitoryeffects against CYP2D6 (Table 1). Single doses at 20 or 40mg/kg of WR069878 were not uniformly curative against P.vivax (Table 4). In contrast, 40 mg/kg for 3 days cured bothmonkeys. At lower dose rates, a 7-day treatment regimen isapparently required, since 10 mg/kg/day for 3 days did not cureP. vivax infections but 10 mg/kg/day for 7 days did cure P.falciparum infections. No monkeys became ill, died, or showedsigns of toxicity after treatment with WR069878. Plasma con-centrations of all the analogs peaked within 2 h of dosing anddeclined to 80% of their peak value by 24 h posttreatment.Collectively, these observations suggest that AAQMs requiredaily administration to induce cures. The approximate dose ofWR069878 that would be required to treat a human Plasmo-dium infection would most likely be 10 mg/kg/day for 3 days(assuming a scaling factor of approximately 4 based on surfacearea) (48).

TABLE 3. Relationship between efficacies of selected AAQMs in Aotus, in vitro antimalarial activity, and plasma concentrations

Compound Outcome withAotus

TM91C235 IC90(ng/ml)

Concn in plasma (ng/ml) on day:Mean peak

concn (ng/ml)Mean peakconcn/IC90

1 7

2 h 24 h 2 h 24 h

WR035058 Cure 8.5 454 84 371 105 413 49WR069878 Cure 3.3 38 16 91 40 65 20WR074086 Clear (R-25) 7.0 179 4.0 78 7.8 129 18WR176399 Clear (R-21) 0.5 27 2.7 30 3.8 28 49WR029252 Failure 20 67 1.0 21 1.0 44 2.2WR030090 Failure 52 597 11 16 5.6 307 5.9

TABLE 4. Efficacies of WR069878 against Plasmodium spp.with Aotus

Dose(mg/kg/day)

No. ofdays ofdosing

Type oftreatmenta Speciesb Outcome

(nc)

20 1 Retreatment P. vivax AMRU1 Cure20 1 Retreatment P. vivax AMRU1 Clear (8)80 1 Retreatment P. vivax AMRU1 Clear (11)80 1 Retreatment P. vivax AMRU1 Cure0.625 3 Primary P. vivax AMRU1 Failure0.625 3 Primary P. vivax AMRU1 Failure2.5 3 Primary P. vivax AMRU1 Failure2.5 3 Primary P. vivax AMRU1 Failure10 3 Primary P. vivax AMRU1 Clear (7)10 3 Primary P. vivax AMRU1 Cure40 3 Retreatment P. vivax AMRU1 Cure40 3 Retreatment P. vivax AMRU1 Cure10 7 Primary P. falciparum FVO Cure10 7 Primary P. falciparum FVO Cure

a Primary treatment refers to the initial treatment given when parasitemiareached 5,000 parasites/�l. Retreatment refers to administration of an additionalcourse of treatment in the event of recrudescence after, or failure of, the primarytreatment.

b These strains/species are chloroquine resistant.c Where the outcome of treatment was clearance, the number in parentheses

indicates the number of days before parasites recrudesced.


WR069878 is less neurotoxic than mefloquine. We furtherinvestigated the neurological effects of WR069878 with ourtest system. WR069878 was clearly less neurotoxic than meflo-quine (Table 1; Fig. 3). This effect does not result from adifference in solubility, since both drugs dissolved withoutevidence of precipitation across the concentration rangetested. The mechanism of neurotoxicity of mefloquine alsoappears to be different, since WR069878 did not disrupt cal-cium homeostasis to the same extent as mefloquine (Fig. 3). Inaddition, the neurotoxicity of WR069878 was not inhibited byDNQX and supraphysiological magnesium, agents shown topartially protect neurons from mefloquine-induced neurotox-icity (Fig. 3).

Assessment of phototoxicity potential of AAQMs. Phototox-icity may be associated with quinolinyl methanols that arephenyl substituted at the 2 position (44). We evaluated twoapproaches for assessing the possible phototoxicity of our mostpromising compound, WR069878. First, we generated a phar-macophore for phototoxicity based on published data for ayeast model (22). Features required for phototoxicity include ahydrogen bond acceptor and an aliphatic hydrophobic and twoaromatic hydrophobic functionalities (Fig. 4). The structures ofthree quinoline methanols were mapped to that of the phar-macophore. WR030090 and WR007930 both mapped all thefeatures of the pharmacophore, with estimated minimum pho-totoxic concentrations of 110 and 170 mg/ml, respectively (Fig. 4;

FIG. 3. WR069878 is less neurotoxic and has a different mechanism of toxicity than mefloquine. A: Mefloquine, with an IC50 of 27 �M,is more neurotoxic than WR069878 with an IC50 of 242 �M. B: Mefloquine but not WR069878 at a concentration of 100 �M increasesintracellular calcium concentrations, as indicated by the relative increase in Fluo3 fluorescence, as measured by confocal microscopy. Thedisruption of neuronal calcium caused by mefloquine occurs as a consequence of discharge of the endoplasmic reticulum calcium store andan influx of extracellular calcium through unknown mechanisms. Drugs were added at the time indicated by the arrow. C: The neurotoxicityof mefloquine but not WR069878 is blocked by DNQX and supraphysiological magnesium, indicating that the compounds have differentmechanisms of toxicity. Bars represent standard errors in all cases.


Table 5). WR007930 is historically associated with phototoxic-ity in phase I clinical trials, whereas WR030090 progressed tophase II clinical trials without phototoxicity being deemed asignificant concern (30, 42). Mefloquine mapped all the fea-tures of the pharmacophore, with the exception of an aromatichydrophobic functionality associated with the phenyl ring atthe 2 position, with an estimated minimum phototoxic concen-tration of 5,600 mg/ml (Table 5). In addition to the pharma-cophore assessment, these quinoline methanols were alsotested in the 3T3 NRU phototoxicity test. In this test, com-pounds with a resultant PIF value greater than 5.0 are likely tobe phototoxic when administered systemically. The results ofthe 3T3 NRU assay indicate that WR007930 and WR030090have the potential to exert phototoxicity (PIF 29.0 and 105.7,respectively), whereas mefloquine scored a PIF of 1.76 andtherefore, as expected, is not considered phototoxic (Table 5).

Thus, the pharmacophore, the 3T3 NRU test, and other mod-els outlined in Table 5 are all able to predict the absence ofphototoxicity but cannot grade the degree of potential photo-toxicity. WR069878 exhibited a PIF of 107.7 in the 3T3 assayand a minimum phototoxic concentration of 1,200 mg/ml in thepharmacophore model, suggesting the potential for phototox-icity.

Synthesis scheme for WR069878 and WR030090. We devel-oped a synthetic scheme for WR069878 and WR030090 mod-eled after one published for mefloquine (Fig. 5). Based on abench-scale synthesis, the two dialkylaminoquinolinyl meth-anols were much less costly to produce than mefloquine (Table6). The cost of a treatment course of each drug can be extra-polated based on the known or estimated effective dose. Usingthis scheme, the cost of treatment for WR069878 or a similardrug would likely be substantially lower than that for meflo-

FIG. 4. Phototoxicity pharmacophore maps. The phototoxicity pharmacophore was generated based on published studies (22) reporting theminimum phototoxic concentrations of various quinoline methanols in an in vitro yeast growth inhibition assay. All the features of the pharma-cophore map to WR007930 and WR030090. However, the aromatic hydrophobic functionality associated with the 2-position phenyl group doesnot map to mefloquine.

TABLE 5. Clinical and preclinical phototoxicity data for three quinoline methanols

Compound Clinical end pointPhototoxicity of compound:

PIF f

For mice For yeast With pharmacophore model

WR007930 Irritation/erythema after sunexposure (12 mg/kg p.o. for14 days)a

Minimum phototoxic doseis 7 mg/kg by “injection”d

MPCc 31 mg/mld Estimated MPC 170 mg/ml 29.0

WR030090 Clinically insignificant phototoxicity(10 mg/kg/day for 6 days in 4/124people)b

Doses of 25 or 50 mg/kgorally are phototoxice

MPCs were 25 and500 mg/mld

Estimated MPC 110 mg/ml 105.7

Mefloquine Not considered phototoxic Not phototoxic at tolerateddoses

Not tested Estimated MPC 5,600 mg/ml 1.76

a From the work of Pullman et al. (42). p.o., per os.b From the work of Martin et al. (30).c MPC, minimum phototoxic concentration.d From the work of Ison and Davis (22).e WRAIR archival data, method based on principles similar to those of the work of Ison and Davis (22).f As calculated by the 3T3 Neutral Red Uptake Phototox Prediction software (version 2.0; developed by ZEBET); compounds with the potential to be phototoxic

have a PIF of �5.0.


quine. This analysis has two caveats. First, the difference in thecost of production between the bench and industrial scales wasassumed to be equivalent to that for mefloquine. Second, thesynthesis of WR069878 is complicated by the lack of a com-mercially available source of starting material (Fig. 5). Fur-thermore, “from-scratch” synthesis of this material wouldresult in the synthesis of two isomers, the resolution ofwhich could potentially be complicated for an industrial-scale synthesis.

DISCUSSION

We have identified WR069878 as the most promising moleculefrom a group of structurally similar compounds. WR069878 wasshown to be orally active, without toxicity, against P. berghei in

mice and chloroquine-resistant P. falciparum and P. vivax inmonkeys. The compound is also an order of magnitude morepotent than mefloquine against mefloquine-resistant strains ofP. falciparum in vitro. We have also shown that WR069878 isintrinsically less neurotoxic than mefloquine and appears toexhibit a different mechanism of neurotoxicity in vitro. Fur-thermore, the cost of treatment with WR069878 (and a relatedquinoline, WR030090) has been estimated to be much lowerthan that with mefloquine. These favorable characteristics ar-gue for further development of the compound. However, thepotential phototoxicity of the molecule, its relative lack ofmetabolic stability, and the lack of commercially availablestarting materials suggest a more cautious approach. In thediscussion that follows, we outline the structural basis for eachof these characteristics and suggest modification to the molec-ular scaffold that is expected to yield a new antimalarial agent.Finally, we discuss the likely clinical use of such a drug giventhe concern about potential cross-resistance to mefloquine.

Several lines of evidence suggest that the neurotoxicity ofmefloquine, as assessed in the present study, appears to beassociated with the piperidine ring. We showed in an earlierstudy (16) that 4-quinolinecarbinolamines, which possess piper-idine functionality at the 4 position, exhibit calcium-dependentneurotoxicity at a level of potency similar to that of meflo-quine. In the present study, we show that “opening” of thepiperidine ring correlates with a lower neurotoxic potency.Furthermore, the reduced neurotoxicities of both WR069878(this study) and chloroquine (14, 16) are associated with afailure to disrupt calcium homeostasis in the same manner asmefloquine. It is likely that this screen has in vivo relevance,since mefloquine induces neuronal degeneration consistentwith a necrotic effect in rats (13). Of course, it remains to bedetermined whether this physiological effect caused by meflo-quine is solely responsible for its neuropathological effects invivo. Indeed, mefloquine accumulates in the central nervoussystem and has been shown to interact with numerous cellular

FIG. 5. Hypothetical synthesis of WR069878 and starting materials. Scheme 1, proposed synthesis of WR069878 with possible yields. Scheme2, synthesis of isatin starting material.

TABLE 6. Cost-of-goods analysis

Drug

Benchcost per

100 g($)a

Benchcost permg (¢)b

Marketprice of228-mg

basepill (¢)

Marketprice

per mg(¢)

Treatmentdose (mg)

Treatmentcost ($)i

Mefloquine 13,969 14 571c 2.50d 1,250f 31WR030090 2,200 2.2 NA 0.39e 4,140g 16WR069878 1,360 1.4 NA 0.24e 2,100h 5

a Calculated based on hypothetical bench synthesis shown in Fig. 5 and pub-lished prices for starting materials from Sigma.

b Value in column 2 divided by 1,000.c Cost of a 250-mg Lariam tablet (228-mg mefloquine base, from the work of

Chambers [8]).d Calculated by dividing cost of Lariam tablet in column 4 by 228.e Estimated by assuming that scale-up costs for WR069878 and WR030090 are

equivalent to that for mefloquine.f Standard mefloquine treatment dose is 1,250 mg (Rendi-Wagner et al. [43]).g Total dose required for cure based on published treatment dose of 690

mg/kg/day for 6 days (Martin et al. [30]).h Dose in mg assuming a scaling factor of 3.8 between Aotus and humans based

on surface area and assuming that the average human weighs 70 kg.i Calculated by multiplying the cost per mg in dollars by treatment dose in mg.


targets within a concentration range that one could reasonablyconsider pharmacological (10, 15, 20, 23, 28, 39, 51). Further-more, it is interesting to note that others have reported that thesubstitution of the piperidine ring with a pyridine ring alsoreduces the potency of mefloquine against at least one of theseother potential neurological targets (28).

Replacement of the piperidine ring with two noncyclic alkylside chains to form a tertiary amine renders these compoundsmore metabolically labile and susceptible to N dealkylation.Moreover, the nitrogen side chain appears to be less vulnera-ble to N dealkylation if the chains are shortened or if only asingle chain is present (a secondary amine). This modification,interestingly, also appears to make the compounds more bio-logically active. Therefore, it seems possible to modify thestructure-activity relationships (SAR) of the N-alkyl side chain,such that the ideal balance between metabolic stability, biolog-ical activity, and cost of goods is obtained. One approach toachieve this goal might be to include an N functionality in anonpiperidine ring system, either in the form of a pyridine ring(shown by Maertens et al. [28] to be less toxic in vitro to anionchannels), a quinuclidine ring (quinine is not neurotoxic in ourtest system) (16), or a heterocyclic ring of a smaller, differentsize. Indeed, a recent study reported the synthesis of a non-neuroactive molecule in which the piperidine ring of meflo-quine was replaced with a quinuclidine ring (46), although theveracity of this claim is difficult to judge given that mefloquinewas not included as a positive control in these neurologicalstudies.

Phototoxicity is potentially a problem with this class of drugs(42). However, for various reasons there are no useful dataavailable to determine whether WR069878 or analogs wouldbe likely to induce clinically important phototoxicity. First, theone study that reported in vivo phototoxicity (in mice) withWR069878 did not specify the route of administration beyondthe fact that the compound was “injected” (22). Second, quin-olinyl methanols are not uniformly phototoxic in the clinicalsetting (Table 5). Third, among the various test systems forphototoxicity, there are no comparable data for all the quino-linyl methanols (Table 5). We therefore attempted to resolvethis issue by developing a phototoxicity pharmacophore andutilizing the 3T3 NRU assay to rank the three compounds withwhich there is clinical experience. If successful, this approachmay have utility as an in silico screen for phototoxicity.

Pharmacophore mapping correctly identified mefloquine asa nonphototoxic drug; however, WR030090 and WR007930were indistinguishable (Table 5). This mapping also indicatedthat the feature which did not map to mefloquine was thearomatic hydrophobic functionality associated with the phenylgroup at the 2 position. The 3T3 assay was also able to clearlydistinguish the lack of phototoxicity of mefloquine but did notrank WR007930 and WR030090 correctly. Thus, the pharma-cophore, 3T3 assay, and other screens listed in Table 5 appearto be capable of ruling out phototoxicity but not ranking re-lated molecules in terms of the degree to which they couldinduce phototoxicity. The results of both of these screens sug-gest the potential for phototoxicity with WR069878. However,it may be possible to engineer out the phototoxicity ofWR069878, since it is thought that the phototoxicity of quin-oline methanols is due to extensive �-conjugation of the quin-oline and phenyl ring systems (44). With mefloquine, this was

avoided by addition of a trifluoromethyl group at the 2 posi-tion. However, addition of a phenyl ring system improves in-trinsic antimalarial activity. Future studies should be directedtowards investigating the SAR balance between phototoxicityand antimalarial activity at the 2 position. The phototoxicitypharmacophore could then be used as an in silico screeningtool to rule out phototoxicity prior to synthesis of new analogs.

Cost of goods is an important factor to consider when evalu-ating new antimalarial drugs, given that the disease is prevalent indeveloping countries. We have estimated that WR069878 andrelated compounds are likely to be less expensive to produce thanmefloquine based on comparing the respective bench-scale syn-theses for these compounds. Using this method, we determinedthat the starting materials needed for WR069878 are generallyless expensive and also feature more-reliable and higher-yieldingchemical steps. Furthermore, we anticipate that commercial pro-cess chemistry using more cost-effective bulk reagents will furtherimprove the cost basis of the AAQMs versus mefloquine.

The hypothetical synthesis of WR069878 and related com-pounds, while intrinsically cheaper than mefloquine on a benchscale, is likely to be complicated by the lack of a commerciallyavailable starting material needed for large-scale synthesis.This is because the synthesis of the appropriate starting mate-rial as shown would involve separation of isomers on a largescale (Fig. 5). One approach to resolving this issue would bethe strategic placement of blocking groups that could be easilyremoved on a large scale. For example, installing a methoxyfunctional group that would block cyclization at the undesiredposition and could be removed, for example, with boron tri-bromide, should yield only the desired isomer. Alternatively,this issue might be avoided by introduction of different ringsubstituents at the 6, 7, and 8 positions. The present study didnot, however, attempt to address the effect that such substitu-tions may have on intrinsic antimalarial activity, neurotoxicity,or metabolic stability.

Clinical resistance to mefloquine is mediated for the mostpart by amplified copy number of the gene pfmdr1 (41). pfmdr1encodes PgH-1, a malaria drug efflux protein homologous tomammalian P glycoprotein. It has been suggested that thisprotein is linked mechanistically to mefloquine resistance,since it acts to detoxify the malaria parasite by extrusion of thedrug into the food vacuole (45). However, clinical resistance tomefloquine may also occur in the absence of pfmdr1 amplifi-cation in some cases (41), and some parasites of West Africanorigin exhibit a naturally lower susceptibility to mefloquine.The P. falciparum strains TM90C2A, TM91C235, and D6 usedhere are representative of these scenarios. In an earlier studywe observed that the susceptibilities of different Plasmodiumstrains to a series of 4-quinolinecarbinolamines track withthose of mefloquine (16). A similar observation was made inthe present study with AAQMs. It is worth briefly consideringthe clinical implications of this observation.

In vitro susceptibility assays do not in themselves predictclinical efficacy, since resistance occurs when the MIC is notexceeded by plasma drug concentrations for a sufficient periodof time. Therefore, AAQMs which are more potent thanmefloquine should exhibit greater in vivo efficacy, assumingsimilar pharmacokinetics, regardless of in vitro cross-suscepti-bilities of parasites to mefloquine and AAQMS. However, thein vitro data do imply that deployment of AAQMS as mono-


therapy for treatment of malaria may eventually result in re-duced clinical utility through mechanisms similar to those withmefloquine. In order to avoid this, it would be necessary toappropriately combine AAQMS with other drugs. In SoutheastAsia, where mefloquine resistance is most prevalent, meflo-quine remains effective in combination with artesunate (11, 38,52) despite 10 years of continuous use and a high backgroundof pfmdr1-mediated resistance to mefloquine monotherapy(41). Furthermore, there is some evidence that the backgroundof mefloquine resistance has declined (5). There is thereforeevery reason to suspect that a future WR069878 analog wouldbe effective as a malaria treatment agent if used in combinationwith other antimalarials.

WR030090, one of the alkylaminoquinoline methanols in-vestigated here, progressed into phase II trials and was used tosuccessfully treat emerging multi-drug-resistant malaria inSoutheast Asia during the Vietnam War (6). However, it waseventually abandoned in favor of mefloquine, not because oftoxicity issues but as a consequence of the apparently superiorpharmacokinetics of mefloquine. Many of the compounds in-vestigated in the present study are superior to WR030090 interms of oral activity and pharmacokinetics and are superior tomefloquine in terms of neurotoxicity, intrinsic potency, andcost of goods. Structural improvement of the lead candidate,WR069878, along the lines suggested is likely to result in avaluable new antimalarial agent.

ACKNOWLEDGMENTS

The manuscript was reviewed by WRAIR and MRMC, and there isno objection to its publication or dissemination. The opinions ex-pressed herein are those of the authors and do not reflect the views oropinions of the Department of the Army and the Department ofDefense.

This work was conducted under the auspices of the Military Infec-tious Diseases Research Program (proposal no. A50030_03_WR_OC).

We acknowledge the contribution made by the University of Miamiin the form of the screening of various AAQMs against P. berghei inmice. We thank William Otero and Jorge Aparicio for performing themicroscopic analysis of blood films generated during the Aotus studies;Camilo Marin and the animal caretakers for supervision and animalhusbandry at the Gorgas Memorial Institute Aotus facility in PanamaCity, Panama; and Maritza Brewer and Gladyz Calvino for secretarialand administrative support. We also acknowledge the support ofLaurie Brown, Sandra Simpson, and Denise Anderson at WRAIRfor secretarial and administrative support.

REFERENCES

1. Adams, R., E. W. Bachmann, L. F. Fieser, J. R. Johnson, and H. R. Snyder(ed.). 1942. Organic reactions, vol. 1. John Wiley and Sons, New York, N.Y.

2. Bhattacharjee, A. K., D. E. Kyle, J. L. Vennerstrom, and W. K. Milhous.2002. A 3D QSAR pharmacophore model and quantum chemical structure-activity analysis of chloroquine(CQ)-resistance reversal. J. Chem. Inf. Com-put. Sci. 42:1212–1220.

3. Blumbergs, P., M. P. LaMontagne, A. Markovac, J. G. Moehring, A. B. Ash,and C. L. Stevens. 1972. Antimalarials. 2. 2,6-Bis(aryl)-4-pyridinemethanols.J. Med. Chem. 15:808–812.

4. Boyd, D. R., R. J. H. Davies, L. Hamilton, and J. J. McCullough. 1992.Synthesis of arene oxide and trans-dihydro diol metabolites of isoquinoline.J. Chem. Soc. Perkin Trans. 1:31–35.

5. Brockman, A., R. N. Price, M. van Vugt, D. G. Heppner, D. Walsh, P. Sookto,T. Wimonwattrawatee, S. Looareesuwan, N. J. White, and F. Nosten. 2000.Plasmodium falciparum antimalarial drug susceptibility on the north-westernborder of Thailand during five years of extensive use of artesunate-meflo-quine. Trans. R. Soc. Trop. Med. Hyg. 94:537–544.

6. Canfield, C. J., A. P. Hall, B. S. MacDonald, D. A. Neuman, and J. A. Shaw.1973. Treatment of falciparum malaria from Vietnam with a phenanthrenemethanol (WR 33063) and a quinoline methanol (WR 30090). Antimicrob.Agents Chemother. 3:224–227.

7. Centers for Disease Control and Prevention. 2005–2006. Health informationfor international travel. Elsevier Mosby, Philadelphia, Pa.

8. Chambers, J. A. 2003. Military aviators, special operations forces, and causalmalaria prophylaxis. Mil. Med. 168:1001–1006.

9. Clyde, D. F., V. C. McCarthy, C. C. Rebert, and R. M. Miller. 1973. Pro-phylactic activity of a phenanthrene methanol (WR 33063) and a quinolinemethanol (WR 30090) in human malaria. Antimicrob. Agents Chemother.3:220–223.

10. Cruikshank, S. J., M. Hopperstad, M. Younger, B. W. Connors, D. C. Spray,and M. Srinivas. 2004. Potent block of Cx36 and Cx50 gap junction channelsby mefloquine. Proc. Natl. Acad. Sci. USA 101:12364–12369.

11. Davis, T. M., H. A. Karunajeewa, and K. F. Ilett. 2005. Artemisinin-basedcombination therapies for uncomplicated malaria. Med. J. Aust. 182:181–185.

12. Desjardins, R. E., C. J. Canfield, J. D. Haynes, and J. D. Chulay. 1979.Quantitative assessment of antimalarial activity in vitro by a semiautomatedmicrodilution technique. Antimicrob. Agents Chemother. 16:710–718.

13. Dow, G., R. Bauman, D. Caridha, M. Cabezas, F. Du, R. Gomez-Lobo, M.Park, K. Smith, and K. Cannard. 2006. Mefloquine induces dose-relatedneurological effects in a rat model. Antimicrob. Agents Chemother. 50:1045–1053.

14. Dow, G. S., D. Caridha, M. Goldberg, L. Wolf, M. L. Koenig, D. L. Yourick,and Z. Wang. 2005. Transcriptional profiling of mefloquine-induced disrup-tion of calcium homeostasis in neurons in vitro. Genomics 86:539–550.

15. Dow, G. S., T. H. Hudson, M. Vahey, and M. L. Koenig. 2003. The acuteneurotoxicity of mefloquine may be mediated through a disruption of cal-cium homeostasis and ER function in vitro. Malaria J. 2:14.

16. Dow, G. S., M. L. Koenig, L. Wolf, L. Gerena, M. Lopez-Sanchez, T. H.Hudson, and A. K. Bhattacharjee. 2004. The antimalarial potential of 4-quino-linecarbinolamines may be limited due to neurotoxicity and cross-resistance inmefloquine-resistant Plasmodium falciparum strains. Antimicrob. Agents Che-mother. 48:2624–2632.

17. Earle, W. C., and M. J. Perez. 1931. Laboratory clinical medicine. 17:1124–1130.

18. Epling, G. A., and K.-Y. Lin. 1987. Sulfur-containing 2-arylquinolinemeth-anols as potential antimalarials. J. Heterocycl. Chem. 24:853–857.

19. Faucher, A. M., M. D. Bailey, P. L. Beaulieu, C. Brochu, J. S. Duceppe, J. M.Ferland, E. Ghiro, V. Gorys, T. Halmos, S. H. Kawai, M. Poirier, B. Simoneau,Y. S. Tsantrizos, and M. Llinas-Brunet. 2004. Synthesis of BILN 2061, an HCVNS3 protease inhibitor with proven antiviral effect in humans. Org. Lett. 6:2901–2904.

20. Gribble, F. M., T. M. Davis, C. E. Higham, A. Clark, and F. M. Ashcroft.2000. The antimalarial agent mefloquine inhibits ATP-sensitive K-channels.Br. J. Pharmacol. 131:756–760.

21. Hall, A. P., H. E. Segal, E. J. Pearlman, and P. Phintuyothin. 1975. Com-parison of a 9-phenanthrene methanol (WR33063), a 4-quinoline methanol(WR30090), and quinine for falciparum malaria in Thailand. Trans. R. Soc.Trop. Med. Hyg. 69:342–349.

22. Ison, A. E., and C. M. Davis. 1969. Phototoxicity of quinoline methanols andother drugs in mice and yeast. J. Investig. Dermatol. 52:193–198.

23. Lim, L. Y., and M. L. Go. 1985. The anticholinesterase activity of mefloquine.Clin. Exp. Pharmacol. Physiol. 12:527–531.

24. Lipinski, C. A. 2000. Drug-like properties and the causes of poor solubilityand poor permeability. J. Pharmacol. Toxicol. Methods 44:235–249.

25. Liu, Y., Y. Feng, R. Wang, Y. Gao, and L. Lai. 2001. Quinoline-4-acetamidesas sPLA(2) inhibitors. Bioorg. Med. Chem. Lett. 11:1639–1641.

26. Lutz, R. E., J. F. Codington, R. J. Rowlett, A. J. Deinet, and P. S. Bailey.1946. Antimalarials. 7-Chloro-4-hydroxy- and 4,7-dichloro-1-methylcarbostyrils.J. Am. Chem. Soc. 68:1810–1812.

27. Ma, N. S., R. N. Rossan, S. T. Kelley, J. S. Harper, M. T. Bedard, and T. C.Jones. 1978. Banding patterns of the chromosomes of two new karyotypes ofthe owl monkey, Aotus, captured in Panama. J. Med. Primatol. 7:146–155.

28. Maertens, C., L. Wei, G. Droogmans, and B. Nilius. 2000. Inhibition ofvolume-regulated and calcium-activated chloride channels by the antimalarialmefloquine. J. Pharmacol. Exp. Ther. 295:29–36.

29. Markovac, A., M. P. LaMontagne, P. Blumbergs, A. B. Ash, and C. L.Stevens. 1972. Antimalarials. 3. 2,6-Bis(aryl)-4-pyridinemethanols with tri-fluoromethyl substituents. J. Med. Chem. 15:918–922.

30. Martin, D. C., J. D. Arnold, D. F. Clyde, M. al-Ibrahim, P. E. Carson, K. H.Rieckmann, and D. Willerson, Jr. 1973. A quinoline methanol (WR 30090)for treatment of acute malaria. Antimicrob. Agents Chemother. 3:214–219.

31. Meier, C. R., K. Wilcock, and S. S. Jick. 2004. The risk of severe depression,psychosis or panic attacks with prophylactic antimalarials. Drug Saf. 27:203–213.

32. Milhous, W. K., N. F. Weatherly, J. H. Bowdre, and R. E. Desjardins. 1985.In vitro activities of and mechanisms of resistance to antifol antimalarialdrugs. Antimicrob. Agents Chemother. 27:525–530.

33. Muth, C. W., H. L. Minigh, S. Chookruvong, L. A. Hall, and H. T. Chao.1981. Syntheses of 9-acridine- and 2-phenanthridinemethanols as potentialantimalarials. J. Med. Chem. 24:1016–1018.

34. National Research Council. 1996. Guide for the care and usage of laboratoryanimals. National Academy Press, Washington, D.C.


35. Obaldia, N., III. 1991. Detection of Klebsiella pneumoniae antibodies inAotus l. lemurinus (Panamanian owl monkey) using an enzyme linked im-munosorbent assay (ELISA) test. Lab. Anim. 25:133–141.

36. Obaldia, N., III, R. N. Rossan, R. D. Cooper, D. E. Kyle, E. O. Nuzum, K. H.Rieckmann, and G. D. Shanks. 1997. WR 238605, chloroquine, and theircombinations as blood schizonticides against a chloroquine-resistant strainof Plasmodium vivax in Aotus monkeys. Am. J. Trop. Med. Hyg. 56:508–510.

37. Ohnmacht, C. J., A. R. Patel, and R. E. Lutz. 1971. Antimalarials. 7. Bis(tri-fluoromethyl)–(2-piperidyl)-4-quinolinemethanols. J. Med. Chem. 14:926–928.

38. Olliaro, P. L., and W. R. Taylor. 2004. Developing artemisinin based drugcombinations for the treatment of drug resistant falciparum malaria: a re-view. J. Postgrad. Med. 50:40–44.

39. Pham, Y. T., A. Regina, R. Farinotti, P. Couraud, I. W. Wainer, F. Roux, andF. Gimenez. 2000. Interactions of racemic mefloquine and its enantiomerswith P-glycoprotein in an immortalized rat brain capillary endothelial cellline, GPNT. Biochim. Biophys. Acta 1524:212–219.

40. Phillips-Howard, P. A., and F. O. ter Kuile. 1995. CNS adverse eventsassociated with antimalarial agents. Fact or fiction? Drug Saf. 12:370–383.

41. Price, R. N., A. C. Uhlemann, A. Brockman, R. McGready, E. Ashley, L.Phaipun, R. Patel, K. Laing, S. Looareesuwan, N. J. White, F. Nosten, andS. Krishna. 2004. Mefloquine resistance in Plasmodium falciparum andincreased pfmdr1 gene copy number. Lancet 364:438–447.

42. Pullman, T. N., L. Eichelberger, A. S. Alving, R. Jones, B. Craige, andC. M. Whorton. 1948. The use of SN-10,275 in the prophylaxis andtreatment of sporozoite-induced vivax malaria (Chesson strain). J. Clin.Investig. 27:12–16.

43. Rendi-Wagner, P., H. Noedl, W. H. Wernsdorfer, G. Wiedermann, A. Miko-lasek, and H. Kollaritsch. 2002. Unexpected frequency, duration and spec-

trum of adverse events after therapeutic dose of mefloquine in healthyadults. Acta Trop. 81:167–173.

44. Rice, K. C. 1976. Synthesis, antimalarial activity, and phototoxicity of somebenzo(h)quinoline-4-methanols. J. Med. Chem. 19:887–892.

45. Rohrbach, P., C. P. Sanchez, K. Hayton, O. Friedrich, J. Patel, A. B. Sidhu,M. T. Ferdig, D. A. Fidock, and M. Lanzer. 2006. Genetic linkage of pfmdr1with food vacuolar solute import in Plasmodium falciparum. EMBO J. 25:3000–3011.

46. Satayavivad, J., P. Watcharasit, P. Khamkong, J. Tuntawiroon, C. Pavaro,and S. Ruchirawat. 2004. The pharmacodynamic study of a potent newantimalarial (MC1). Acta Trop. 89:343–349.

47. Schlagenhauf, P. 1999. Mefloquine for malaria chemoprophylaxis 1992–1998: a review. J. Travel Med. 6:122–133.

48. Schmidt, L. H. 1978. Plasmodium falciparum and Plasmodium vivax infec-tions in the owl monkey (Aotus trivirgatus). II. Responses to chloroquine,quinine, and pyrimethamine. Am. J. Trop. Med. Hyg. 27:703–717.

49. Simpson, J. A., E. R. Watkins, R. N. Price, L. Aarons, D. E. Kyle, and N. J.White. 2000. Mefloquine pharmacokinetic-pharmacodynamic models: impli-cations for dosing and resistance. Antimicrob. Agents Chemother. 44:3414–3424.

50. Trenholme, C. M., R. L. Williams, R. E. Desjardins, H. Frischer, P. E.Carson, K. H. Rieckmann, and C. J. Canfield. 1975. Mefloquine (WR142,490) in the treatment of human malaria. Science 190:792–794.

51. Weiss, S. M., K. Benwell, I. A. Cliffe, R. J. Gillespie, A. R. Knight, J.Lerpiniere, A. Misra, R. M. Pratt, D. Revell, R. Upton, and C. T. Dourish.2003. Discovery of nonxanthine adenosine A2A receptor antagonists for thetreatment of Parkinson’s disease. Neurology 61:S101–S106.

52. Wongsrichanalai, C., S. Prajakwong, S. R. Meshnick, G. D. Shanks, and K.Thimasarn. 2004. Mefloquine—its 20 years in the Thai Malaria ControlProgram. Southeast Asian J. Trop. Med. Public Health 35:300–308.


Utility of Alkylaminoquinolinyl Methanols as New Antimalarial Drugs

Documents