Comparative Preclinical Drug Metabolism and
Pharmacokinetic Evaluation of Novel
4-Aminoquinoline Anti-Malarials

CHARLES B. DAVIS,

1

RAMESH BAMBAL,

1

GANESH S. MOORTHY,

1

ERIN HUGGER,

1

HONG XIANG,

1

BRIAN KEVIN PARK,

2

ALLISON E. SHONE,

3

PAUL M. O’NEILL,

4

STEPHEN A. WARD

3

1

Drug Metabolism and Pharmacokinetics, GlaxoSmithKline Drug Discovery, 1250 South Collegeville Rd, Collegeville,

Pennsylvania 19426

2

Department of Pharmacology and Therapeutics, University of Liverpool, Liverpool L693GE, UK

3

Molecular and Biochemical Parasitology, Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L35QA, UK

4

Department of Chemistry, University of Liverpool, Liverpool L697ZD, UK

Received 27 March 2008; accepted 12 May 2008

Published online 18 June 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21469

ABSTRACT: The disposition of three 4-aminoquinoline leads, namely isoquine (ISO),
des-ethyl isoquine (DEI) and N-tert-butyl isoquine (NTBI), were studied in a range of
in vivo and in vitro assays to assist in selecting an appropriate candidate for further
development. Analogous to amodiaquine (ADQ), ISO undergoes oxidative N-dealkylation
to form DEI

in vivo. Blood clearance of DEI was as much as 10-fold lower than that of ISO

in animals and after oral administration, metabolite exposure exceeded that of parent by
as much as 14-fold. Replacement of the N-ethyl with an N-tert-butyl substituent
substantially reduced N-dealkylation as blood clearance of NTBI was

2 to 3-fold lower

than DEI in mouse, rat, dog and monkey. Mean NTBI oral bioavailability was generally
higher than the other leads (

68%). Blood cell association was substantial for NTBI,

particularly in dog and monkey, where blood to plasma concentration ratios >4 were
observed. Human plasma protein binding was similar for NTBI, DEI, and des-ethyl
amodiaquine (DEA). Allometric scaling predicted human blood clearance (CL) for NTBI
to be low (

12% liver blood flow). All the 4-aminoquinolines inhibited recombinant

human cytochrome P450 2D6 with similar potency; DEI also inhibited 1A2. On balance,
NTBI appeared the most promising lead to progress towards full development.

ß

2008

Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 98:362–377, 2009

Keywords:

preclinical pharmacokinetics; drug design; pro-drugs; ADME; cyto-

chrome P450; MDCK cells; protein binding; blood partitioning; anti-malarials;
bioactivation

INTRODUCTION

Malaria is endemic to the poorest countries in the
world, causing more than 1 million deaths each
year. More than 90% of the deaths occur in Sub-
Saharan Africa and nearly all the deaths are in
children under the age of 5 years. The morbidity
and mortality from malaria have been increasing
over the last several decades due to deterioration
in health systems, growing drug and insecticide

Abbreviations used: CL,

in vivo clearance; V

ss

, volume of

distribution at steady-state;

t

1/2

, apparent terminal half-life;

AUC, area under the concentration-time curve; F, oral bioa-
vailability; CLi, intrinsic clearance; P-gp, P-glycoprotein;
MDR1, multidrug resistance 1; MDCKII, Madin Darby Canine
Kidney Type II; NTBI, N-tert-butyl isoquine, ISO, isoquine;
DEI, des-ethyl isoquine; ADQ, amodiaquine; DEA, des-ethyl
amodiaquine.

Correspondence to: Charles B. Davis (Telephone: 610-917-

5601; Fax: [phone]; E-mail: [e-mail])

Journal of Pharmaceutical Sciences, Vol. 98, 362–377 (2009)
ß

2008 Wiley-Liss, Inc. and the American Pharmacists Association

362

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

resistance, climate change, civil and social
instability, and population displacement.

1

The hemoglobin degradation pathway in

Plas-

modium falciparum has been successfully exploit-
ed as the therapeutic target of chloroquine and
other 4-aminoquinolines. As the parasite catabo-
lizes human hemoglobin to obtain the nutrient
amino acids, the haem bi-product is detoxified by
polymerization to hemozoin.

2

Chloroquine inter-

acts with the porphyrin ring of the haem

via p–p

stacking of the quinoline ring and electrostatic
interactions between the charged ammonium
side-chain of the drug and one of the haem
carboxylates.

3

Accumulation of the haem:drug

complex within the erythrocyte ultimately poisons
the parasite. Resistance to this mechanism of
action has proven difficult to induce. Chloroquine
resistance, which developed relatively slowly and
only after considerable human exposure, is the
result of point mutations in a parasite transporter
pfcrt.

4

Amodiaquine (ADQ, Fig. 1D) is active against

chloroquine-resistant strains of

P. falciparum.

However, its use has been severely restricted by
hepatotoxicity and agranulocytosis in humans.

5,6

This appears to be the result of bioactivation
to form a reactive quinoneimine metabolite that

binds irreversibly to cellular macromolecules and
may lead to direct toxicity as well as immune-
mediated hypersensitivity reactions. IgG antibo-
dies have been detected in patients with adverse
reactions to ADQ.

7,8

ADQ is a pro-drug. Des-ethyl

amodiaquine (DEA, Fig. 1E) is a potent anti-
malarial and it has been shown to be responsible
for the majority of the efficacy attributed to ADQ
therapy.

9

Recently, ADQ analogues have been described

that reduce the potential for formation of reactive
metabolites

via

the

quinoneimine

pathway

while retaining potent antimalarial activity.

10

Isoquine (ISO, Fig. 1B) is an isomeric ADQ
analogue, with the position of the 4

0

-hydroxyl

and Mannich side-chain interchanged. Oxidation
to the quinoneimine is not possible in this con-
figuration. ISO, des-ethyl isoquine (DEI, Fig. 1C)
and N-tert-butyl isoquine (NTBI, Fig. 1A) retain
potent antimalarial activity

in vitro, activity

against strains carrying resistance determinants
to chloroquine, and similar activity in preclinical
models of malaria infection following oral admin-
istration.

As part of the late lead optimization of these

promising ADQ analogues, we have studied their
pharmacokinetics following single intravenous
and oral administration to the mouse, rat, dog
and monkey. Protein binding and blood cell
association were investigated in the plasma or
whole blood from preclinical species and human,
in vitro. The routes and rates of metabolism were
studied in animal and human liver microsomes and
hepatocytes. Concentration- and time-dependent
human cytochrome P450 inhibition, permeability
and active transport were investigated

in vitro.

These studies have permitted detailed comparison
of the developability of these leads and ultimately
the selection of NTBI as the most promising
analogue for progression into definitive safety
assessment studies and clinical investigation.

METHODS

Chemicals

NTBI, ISO, DEI, and DEA were synthesized at
Liverpool University Department of Chemistry.
Selected studies with NTBI and ISO used
materials supplied by GlaxoSmithKline Chemical
Development. ADQ was purchased from Sigma
(St. Louis, MO). Mixed gender human liver
microsomes were obtained from XenoTech LLC

Figure 1. Structure of N-tert-butyl isoquine (NTBI,
Panel A), isoquine (ISO, Panel B), des-ethyl isoquine
(DEI, Panel C), amodiaquine (ADQ, Panel D), and des-
ethyl amodiaquine (DEA, Panel E).

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

363

(Kansas City, KS). Hepatocytes from male CD1
mouse, male Sprague–Dawley rat, male Beagle
dog, male Cynomolgus monkey and human were
from Cellzdirect (Dallas, TX). Basic William’s
medium E and Hank’s Balanced Salt Solution
(HBSS) were from JRH Biosciences (Lenexa, KS).
Microsomes prepared from human lymphoblast
cells expressing individual recombinant human
CYP450 isozymes CYP1A2, CYP2C8, CYP2C9,
CYP2C19, CYP2D6, and CYP3A4 were obtained
from BD Gentest (Woburn, MA). Twelve-multiwell
Transwell

1

systems (Catalog No. 3401; 0.4 mm

pore size, 1.134 cm

2

surface area) were obtained

from Corning, Inc. (Corning, NY). GF120918A
(HCl salt) and amprenavir methanesulfonate
were prepared at GSK. Dulbecco’s Modified Eagle
Medium (DMEM) with glutamine, 10% fetal
bovine serum, 50 U/mL penicillin and 50 mg/mL
streptomycin were from Invitrogen (Grand Island,
NY). Lucifer yellow was from Sigma. Encapsin

1

was purchased from Cerestar USA, Inc. (Ham-
mond, IN). All other chemicals and reagents were
of standard laboratory reagent grade or better.

Animal Husbandry and Handling

In vivo studies were approved by institutional
animal care and use committee. Male Sprague–
Dawley rats were obtained from Charles River
Laboratories (Kingston, NY) and housed indivi-
dually in polycarbonate cages in unidirectional air
flow rooms (25

28C, relative humidity 50 10%,

12-h light/dark cycle). Extruded diet #5L35 from
Purina Mills (St. Louis, MO) was available
ad libitum. Male CD-1 mice were obtained from
Charles River Laboratory (Raleigh, NC) and
housed as were the rats. Certified Rodent Chow
#5001 from Purina Mills was available

ad libitum.

Male Beagle dogs were obtained from Marshall
Farms (North Rose, NY) or Covance Research
Products (Cumberland, VA), group-housed until
placed on study, then individually housed in
stainless-steel cages in an environmentally con-
trolled room (18–298C; relative humidity 30–70%,
12-h light/dark cycle). Dogs received 2.5–3 cups
Certified Canine Diet #5007 from Purina Mills
once daily. Male Cynomolgus monkeys were
obtained from Primate Products Incorporated
(Miami, FL) or Covance Research Products, Inc.
(Alice, TX) and individually housed in stainless-
steel cages in a controlled environment (18–298C,
relative humidity 30–70%, 12-h light/dark cycle).
Each monkey received

6 to 8 biscuits Certified

Primate Chow #5048 from Purina Mills twice
daily plus two pieces of produce. In all cases,
filtered tap water was available

ad libitum. In all

studies, and for each study leg where appropriate,
animals were fasted overnight (12–14 h) and
food was provided

4 h postdose. Animals were

appropriately acclimated to handling procedures
and restraint devices prior to study.

Pharmacokinetics of NTBI

Thirty male mice (

30 g) received NTBI (6 mg/kg,

5 mL/kg) by bolus injection in the tail vein.
Twenty-seven (fasted) received NTBI (12.6 mg/kg,
10 mL/kg) by oral gavage. Blood samples (9

0.4 mL) were collected terminally by cardiac
puncture following anesthetization in a CO

2

chamber. Three male rats (300–360 g) had
cannulae surgically implanted in the femoral vein
and artery as described previously.

11

The animals

received NTBI (14 mg/kg, 10 mL/kg) by 60 min
iv infusion and 48 h later received NTBI (27 mg/
kg; 20 mL/kg) by oral gavage (fasted) in a cross-
over study design. Blood samples (14

0.15 mL)

were collected for up to 24 h after dosing. The
dog (

n

¼ 3 males, 9–14 kg) and monkey (n ¼ 3, 2.5–

5 kg) studies employed a crossover design with a
13 day washout period between dose sessions.
Dogs received NTBI as a 60 min infusion into a
cephalic vein (7 mg/kg, 4 mL/kg) or by oral gavage
(30 mg/kg, 6 mL/kg). Similarly, monkeys received
NTBI doses of 6 mg/kg iv (4 mL/kg) or 26 mg/kg
orally (6 mL/kg). Accept as noted, all formulations
for iv and oral administration were 2% (v/v)
DMSO and 20% (v/v) Encapsin

1

in isotonic saline

(iv) or water (oral). Blood samples were collected
into tubes containing heparin and placed on
crushed ice promptly after collection. Plasma
was collected by centrifugation, snap frozen and
stored at about

808C prior to analysis.

For the studies performed to assess pharmaco-

kinetic linearity of oral NTBI, male mice received
doses of 50, 100, or 300 mg/kg, male and female
rats received doses of 10, 50, 150, or 300 mg/kg and
1 male and 1 female monkey received escalating
doses of 60, 90, and 180 mg/kg by oral gavage
(day 1, 4, and 8). These formulations were 1%
(w/v) aqueous methyl cellulose (10 mL/kg) and
employed the di-HCl salt. Blood samples were
taken for up to 24 h (composite study design,

n

¼ 3

per time-point in the mouse only) then processed
and stored as described above. Studies of ISO and
DEI in mouse, rat, dog or monkey, though not

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

364

DAVIS ET AL.

described in detail, were performed using animal
models, formulations, and study designs that were
largely the same as those described above for
NTBI.

Pharmacokinetic Analysis

Concentration-time data were analyzed by non-
compartmental methods using the computer
program WinNonlin Professional (version 4.1)
largely as described previously.

11

In the NTBI dog

study, the terminal elimination phase following iv
administration was not well-characterized due to
assay sensitivity. Therefore, the half-life from the
oral leg was used to estimate pharmacokinetic
parameters from the iv dose session. Blood clear-
ance and volume were estimated from plasma
parameters using the same animal’s blood to
plasma ratio for NTBI dog and monkey studies
(assuming concentration and time-independent
blood partitioning). To estimate blood to plasma
ratio

ex vivo, at 1 and 3 h postdose for both the iv

and oral dosing sessions, both blood and plasma
concentrations were determined using calibration
curves for the appropriate matrix. Blood to plasma
ratios from independent

in vitro studies were used

to estimate blood PK parameters from the rodent
studies. In order to predict human blood CL,

V

ss

and

t

1/2

, allometric methods analogous to those

described previously were employed.

11

Blood Partitioning

The

in vitro blood partitioning of NTBI was

determined in fresh blood from mouse, rat, dog,
monkey, and human (single donor) at target
concentration of 1 mg/mL (

1% (v/v) methanol

final). Spiked blood samples were mixed gently
then incubated at

378C for 30 min. Aliquots of

blood were taken, mixed with an equal volume of
water, snap frozen and stored at approximately
808C prior to analysis. The remaining blood was
centrifuged to collect plasma. Protein was pre-
cipitated with acetonitrile containing internal
standard, supernatant was collected by centrifu-
gation, frozen on dry ice and stored at approxi-
mately

308C prior to analysis.

Protein Binding

The

in vitro plasma protein binding for NTBI,

DEI, and DEA were investigated using Rapid
Equilibrium Dialysis (RED) devices (8000 MW

cut-off, Linden Bioscience, Woburn, MA). Frozen
mouse and human plasma (Biological Specialty
Corporation, Colmar, PA) were thawed and pro-
tein aggregates were removed by vacuum filtra-
tion (Corning

1

Filter-top centrifuge tubes, pore

size of 0.22 mm). The plasma was then spiked at a
target concentration of 1 mg/mL (0.2% (v/v) DMSO
final). Duplicate aliquots were collected to verify
initial concentrations. Spiked plasma (300 mL)
was placed in the sample chamber of the RED
insert and phosphate-buffered saline (500 mL),
pH 7.4, in the buffer chamber (

n

¼ 6 inserts/species).

The incubation was carried out in a 5% (v/v)
carbon dioxide incubator (IR AutoFlow, NuAire,
Plymouth, MN) at 378C on an orbital titer plate
shaker (Lab-Line Instrument, Inc., Melrose Park,
IL.) at 300 rpm for 4 h. Single aliquots (50 mL)
from sample and buffer chambers of each insert
were removed and protein was precipitated with
acetonitrile containing a common internal stan-
dard (tolbutamide, 40 ng) prior to analysis by LC/
MS/MS. Peak area ratio of compound, relative to
internal standard was used to estimate concen-
trations following an appropriate correction to
account for difference in MS analyte response in
buffer and plasma. Percent binding was estimated
using standard equations.

Intrinsic Clearance in Animal and
Human Hepatocytes

Incubations were performed with 0.5 mM NTBI
in a 0.2 million cell/mL suspension in William’s
Medium E, pH 7.4. Culture plates were placed
into a 378C, 5% (v/v) carbon dioxide incubator on a
shaker at

40 rpm. At various time-points up to

240 min, 50 mL aliquots were removed and added
to 200 mL stop solution (80:20:1 (v/v/v) acetoni-
trile/ethanol/acetic acid) containing internal stan-
dard. A positive control was performed for activity
and inter-assay variability (ethoxycoumarin).
Samples were snap frozen and stored at

808C

until analyzed by LC/MS/MS. Prior to analysis,
samples were thawed at room temperature,
mixed, then centrifuged, and the supernatant
taken for analysis. Concentrations of ethoxycou-
marin were determined by HPLC.

The first-order elimination rate constant for

disappearance of parent compound was calculated
from the slope of the log-transformed concentration-
time curve using Grafit Version 5.0.8 (Erithacus,
Middlesex, UK). CLi was estimated using the
actual volume of the incubation and assuming

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

365

1.2

10

8

cells/g liver for hepatocytes (2.4

10

8

cells/g for dog hepatocytes

12

); clearances were

expressed in units of mL/min/g liver. The lower
limit of quantification was 0.5 mL/min/g liver and
this corresponded to <15% decrease in parent in
240 min. The highest CLi that can be reliably
measured is 50 mL/min/g liver; in this situation,
parent can be quantified for the first three time-
points (up to 45 min). We consider moderate
clearance to be in the range of 3–7 mL/min/g liver
in this assay.

Permeability and Active Transport

Apical-to-basolateral (A-to-B) and basolateral-to-
apical (B-to-A) transport of ISO, ADQ, DEI, NTBI,
and amprenavir (5 mM) were studied across
MDR1-MDCK cell monolayers in the absence and
presence of GF120918 (2 mM) in HBSS. Transport
assays were conducted with MDR1-MDCK cells
4 days postseeding onto 12-well Transwell

1

membrane inserts. Cells were pre-incubated in
HBSS with and without GF120918 at 378C for
30 min. Buffer was removed and test compound
(

GF120918) was added to each donor compart-

ment. HBSS transport buffer or GF120918 in
HBSS was added to the receiver compartment of
each well. The cells were incubated at 378C for
90 min under stirring (80 rpm) conditions, and
samples were withdrawn from both the donor and
receiver compartments. Samples of the donor
solutions were also taken at the beginning and the
end of the experiment to assess recovery. Intrinsic
apparent permeability (

P

app

) and apical efflux

ratio [or

P

app

(B-to-A)/

P

app

(A-to-B)] were estimat-

ed as described previously.

13,14

P

app

was estimated

by averaging the permeabilities in A-to-B and
B-to-A directions in the presence of GF120918.
In general, a compound is classified as a P-gp
substrate when the apical efflux ratio is >2.

P

app

values >50 nm/s are considered high in this model.

Human Cytochrome P450 Inhibition

The inhibition of P450 activity was assessed using
fluorescent probe substrates in a 96-well plate-
based assay format.

15

The following CYP450

activities were monitored in the presence and
absence of 4-aminoquinoline derivatives; 7-ethoxy-
resorufin

O-dealkylation (1A2), 7-methoxy-4-tri-

fluoromethylcoumarin-3-acetic acid

O-dealkylation

(2C9), 3-butyryl-7-methoxycourmarin

O-dealkyla-

tion (2C19), 4-methylaminomethyl-7-methoxycou-

marin

O-dealkylation (2D6), diethoxyfluorescein

O-dealkylation (2C8 and 3A4), 7-[3-(4-phenylpiper-
azin-1-ylmethyl) benzyl] resorufin

O-dealkylation

(3A4).

Incubation mixtures contained drug concentra-

tions ranging from 0 to 100 mM (seven concentra-
tions), probe substrate at a concentration near the
Michaelis-Menten constant, protein concentra-
tion of 0.1–0.4 mg/mL, and an NADPH-generating
system (as in the CLi methods, above, without
UDGPA) in 50 mM potassium phosphate buffer,
pH 7.4. Miconazole was included as a positive
control (all isozymes). Reactions were monitored
with fluorescence measurements at 1-min inter-
vals for 10 min in a fluorescence plate reader
(Cytofluor Series 4000, PE Biosystems Foster
City, CA) at 408C. Percentage activity versus drug
concentration data was fitted using Grafit Version
5.0.8 (Erithacus) to estimate IC50s. Generally, we
consider IC50s <10 mM to be of potential clinical
concern.

Time-dependent inhibition studies were per-

formed similarly except that fluorescence mea-
surements were made at 1-min intervals for
30 min and percentage control activity versus
NTBI concentration plots for each consecutive
5 min (1–5; 6–10; 11–15; 16–20; 21–25; and 26–
30 min) were generated to estimate potential
changes in the IC50 over time. In parallel
incubations, furafylline (CYP1A2), tienilic acid
(CYP2C9), ticlopidine (CYP2C19), metaclopro-
mide (CYP2D6), and troleandomycin (CYP3A4)
were included as positive controls.

Other

in vitro studies of ISO, DEI, ADQ, and

DEA, though not described in detail, were per-
formed largely as described above for NTBI.
Where comparisons between analogues have been
presented, compounds were studied head-to-head
in the same assay run on the same day.

Routes of Biotransformation

Microsomal incubations were performed with
100 mM compound with 1 mg/mL microsomal
protein, 0.34 mg/mL NADP, 1.56 mg/mL glucose-
6-phosphate and 1.2 U/mL glucose-6-phosphate
dehydrogenase, 4 mM UDPGA, 0.5% (v/v) metha-
nol in 50 mM potassium phosphate buffer, pH 7.4,
at 378C. Reactions proceeded for up to 1 h at 378C
and were quenched by adding an equal volume
of 80:20:1 (v/v/v) acetonitrile/ethanol/acetic acid.
Samples were snap frozen at

808C until ana-

lyzed. Prior to analysis, samples were thawed,
mixed, then centrifuged and 10 mL supernatant

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

366

DAVIS ET AL.

was injected onto the LC/MS for metabolite
identification. Hepatocyte studies were performed
essentially as described above (CLi) with 100 mM
compound and incubations up to 24 h at 378C
using adherent cells (Cedra, Austin, TX) and up to
4 h at 378C using suspension (0.7

10

6

cells/mL)

cell cultures. For studies with recombinant
human CYP450 isozymes, incubations contained
100 mM drug, 2–7 mg/mL isozyme and appro-
priate cofactor in 50 mM potassium phosphate, pH
7.4 and reactions proceeded for up to 1 h at 378C.
Reactions were quenched with stop solution and
samples stored and processed as described for the
microsomal incubations.

Drug Analysis

Specific and sensitive HPLC/MS/MS methods
were developed to quantify 4-aminoquinoline
analogues in rat, mouse, dog, monkey and human
blood (diluted 1:1 v/v with water) and plasma, as
well as in phosphate-buffered saline, and dialy-
sate from protein binding studies. Samples were
assayed using protein precipitation with acetoni-
trile followed by HPLC/MS/MS analysis employ-
ing positive-ion atmospheric pressure chemical
ionization or Turbo Ionspray ionization (API 4000
or API4000 QTrap, Applied Biosystems, Foster
City, CA).

The following two LC methods were used for

analysis. In the first method, the LC mobile phase
was 63:37 (v/v) acetonitrile: 10 mM ammonium
formate, pH 3.0. A quaternary pumping system
with degasser (Flux Instruments Rheos 2000,
Leap Technologies, Carrboro, NC) was used to
deliver mobile phase to a 2.1 mm

50 mm, 5 mm,

Aquasil C18 analytical column (Thermo Electron
Corporation, San Jose, CA) preceded by a 0.5 mm
filter. Flow rate was 550 mL/min. In the second LC
method, a TX2 turbulent flow system (Cohesive
Technologies, Franklin, MA) was utilized with
Cyclone 0.5

50 mm, 60 mm turbulent column

(Cohesive Technologies) and 2.1 mm

50 mm,

5 mm, Aquasil C18 analytical column (Thermo
Electron Corporation). The mobile phase con-
tained mixtures of aqueous 10 mM ammonium
formate, pH 3.0 and acetonitrile. The gradient
method employed was similar to one described
previously.

16

Flow rate was 350 mL/min. The

autosampler employed was a CTC Analytics HTS
PAL from Leap Technologies.

The lower limit of quantification was typically

10 ng/mL and the assays were typically linear over

a 100- to a 1000-fold drug concentration range
depending on the matrix and the expected
concentration range of the analyte. For CLi and
permeability studies, the ratio of analyte to
internal standard peak area was used to estimate
relative drug concentration.

For biotransformation studies, the analytical

HPLC system consisted of an HP-1100 solvent
delivery system, an HP-1100 degasser, an HP-
1100 diode array detector (Agilent, Piscataway,
NJ), and a CTC Analytics HTS PAL autosampler
(Leap Technologies). Chromatography was per-
formed on a 2 mm

150 mm, 3 mm, Aqua C18

column (Phenomenex, Torrance, CA) with a
mobile phase containing a mixture of aqueous
0.1% (v/v) formic acid (solvent A) and acetonitrile
(solvent B). The initial mobile phase composition
was 100% solvent A (first 5 min) then the gradient
progressed linearly to 30:70 (v/v) solvent A/solvent
B in 15 min. The mobile phase composition was
returned to the starting solvent mixture in 0.1 min
and the system equilibrated for approximately
10 min between runs. A flow rate of 0.2 mL/min
was employed.

LC/MS/MS was conducted with an LCQ Deca

XP (Thermo Electron Corporation) equipped with
an electrospray ion source. The effluent from the
HPLC column was introduced into a diode array
detector followed by the mass spectrometer atmo-
spheric ionization source. The diode array detec-
tor response was recorded in real time by the mass
spectrometer data system, which provided simul-
taneous detection of absorbance and MS data.
The electrospray interface was operated at
5000 V, and the mass spectrometer was operated
in the positive ion mode. Data were processed with
a VX1120 Gateway computer operating Xcalibur
1.2 software (Thermo Electron Corporation).

RESULTS

Pharmacokinetics and Blood Partitioning of NTBI

Following iv administration of NTBI to the mouse,
mean plasma concentrations declined in a bi-
phasic fashion with a terminal half-life of

3 h

(Fig. 2A and Tab. 1). Mean residence time was
somewhat longer (

7 h). Plasma clearance was

28 mL/min/kg, and the volume of distribution at
steady-state was 5 L/kg. Using a blood to plasma
concentration ratio of 1.6 (Tab. 4), the blood CL
was estimated to be 17 mL/min/kg or about 15% of
liver blood flow in the mouse.

17

The blood

V

ss

,

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

367

Figure 2. Mean (

SD) plasma concentration versus time profiles after iv (*) or oral

(*) administration of NTBI to animals. Panel A: Male CD-1 mice; Panel B: male
Sprague–Dawley rats; Panel C: male beagle dogs; Panel D: male Cynomolgus monkeys.
Data from the oral legs were dose-normalized for comparison with the iv data.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

368

DAVIS ET AL.

estimated from the plasma

V

ss

and the blood to

plasma ratio, was 3.1 L/kg. This value is

4 times

total body water in the mouse, and suggests there
was substantial penetration of NTBI into tissues.
Oral bioavailability of a solution formulation in
the mouse was high (

100%, Tab. 1). Following

oral administration, maximum plasma concentra-
tion was attained at

20 min; thereafter, mean

plasma concentrations declined bi-exponentially
with a terminal half-life similar to that observed
after iv administration (3.3 vs. 5.3 h). The shape of
the plasma concentration-time curve was gene-
rally similar for iv and oral administration as
absorption was rapid and complete (Fig. 2A).
There was evidence of a secondary increase in
plasma concentration, particularly in the oral
profile, at about 8 h after dosing.

The pharmacokinetics in the rat similarly were

multi-phasic after iv administration, and as in
the mouse after oral administration, there was a
secondary increase in plasma concentration about
8 h after dosing (Fig. 2B). In the rat, the terminal
half-life was similar following iv and oral admin-
istration with mean values of 7.9 and 9.0 h,
respectively. Mean residence time was somewhat
longer (

14 h). Blood CL and V

ss

were derived

from the plasma parameters (Tab. 1) and

in vitro

blood partitioning data as described for the mouse.
Blood partitioning in mouse and rat were indis-
tinguishable (B/P of

1.6, Tab. 4). Blood clearance

in the rat was estimated to be 26 mL/min/kg or

about 50% of liver blood flow.

17

Blood

V

ss

was 21 L/

kg, >20 times total body water in this specie. Oral
bioavailability was high (mean of 89

12%,

Tab. 1).

In the dog, following iv administration, plasma

concentrations declined in a multi-phasic manner
(Fig. 2C). Plasma clearance was high and variable
between animals (45.6

28.9 mL/min/kg, Tab. 1).

In the dog studies, blood and plasma samples were
collected to estimate blood partitioning for each
animal

ex vivo. Then for each individual animal,

plasma pharmacokinetic parameters and indivi-
dual animal blood partitioning data were used
to estimate blood CL and

V

ss

. Blood partition-

ing for these animals was high and ranged from
4.5 to 10.4. The animal with the highest plasma
clearance had the highest blood to plasma ratio.
Mean blood clearance was 6.3 mL/min/kg (Tab. 1)
or about 20% of liver blood flow in the dog.

17

Mean

blood

V

ss

was 30 L/kg (Tab. 1),

50 times total

body water in this specie. Oral bioavailability in
the dog again was high (68

18%, Tab. 1). With

the higher dose employed in the oral leg, plasma
concentrations were quantifiable for a substan-
tially longer period of time compared to the iv leg
and the long terminal phase was more completely
characterized. The mean terminal

t

1/2

after oral

administration was

49 h.

Following iv administration to the monkey,

plasma concentrations declined in a multi-phasic
manner with a terminal half-life of

11 h

Table 1. Pharmacokinetics of 4-Aminoquinolines after Single iv or Oral Administration to Animals

Lead

Species

CL

p

(mL/min/kg)

b

CL

B

(mL/min/kg)

c

V

p

ss

(L/kg)

b

V

B

ss

(L/kg)

c

t

1/2

(h)

F (%)

NTBI

Mouse

a

28

17

5.1

3.1

3.3

100

Rat

42

8

26

5

34

6

21

4

8

4

89

12

Dog

46

29

6.3

1.6

210

125

30

12

49

3

d

68

18

Monkey

87

30

14.5

0.7

59

26

11

5

11

3

e

100

DEI

Mouse

40

44

11.8

12.8

6.1

100

Rat

180

63

62

22

35

8

12

3

3

2

60

26

Dog

39

9

12

3

27

13

8

4

12

5

NE

f

Monkey

92

5

49

3

25

12

13

6

5

1

40

14

ISO

h

Mouse

224

219

6.3

6.2

0.4

21

Rat

178

9

89

4

11

1

5

1

1.4

0.1

17

4

Dog

448

70

151

24

31

5

10

2

1.0

0.2

NQ

g

Mean and standard deviation (

n

¼ 3) of parameter reported where appropriate.

a

For composite mouse studies, variability of parameters has not been estimated.

b

CL

p

and

V

p

ss

were derived from analysis of the plasma concentration-time data.

c

CL

B

and

V

B

ss

were calculated using the individual plasma parameter and either the individual mean dog or monkey

B/P values or

the mean mouse or rat

B/P values from in vitro studies.

d

Half-life estimated from oral leg.

e

N

¼ 2.

f

NE indicates not estimated (emesis observed in the oral leg).

g

NQ indicates nonquantifiable (solution dose of 3 mg/kg).

h

No studies were performed for ISO in the monkey given results in other species.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

369

(Fig. 2D and Tab. 1). As observed in other species,
there was a secondary increase in plasma
concentration about 8 h after dosing. As noted
in the dog, plasma CL and

V

ss

were high (Tab. 1)

and blood cell association was substantial (B/P
ranging from 4.2 to 7.9). As observed in the dog,
the monkey with the highest plasma CL had the
greatest extent of blood cell association. Mean
blood clearance of

15 mL/min/kg was low (about

30% of liver blood flow) and very similar between
animals (<5% coefficient of variation compared to
plasma clearance, where the coefficient of varia-
tion was 35%). Oral bioavailability in the monkey
was high (

100%) and absorption was relatively

rapid (maximum concentration observed

3 h

after dosing). The terminal half-life was a bit
longer than that observed after iv administration
(24 vs. 11 h).

Partitioning of NTBI into dog and monkey

bloods cells was also studied

in vitro and here

too, association with blood cells was substantial.
Because of differences observed between animals
ex vivo, in vitro studies were performed with two
individual animal donors. For the dog, mean blood
to plasma ratios of

16 and 4.5 were observed for

the two animals at a concentration of 1 mg/mL
(Tab. 4). For the monkey, mean blood to plasma
ratios of 3.9 and 3.7 were observed for the two
animals (Tab. 4).

In vitro, partitioning of NTBI in

normal human blood was much lower than in all
other species studied and indistinguishable from
the blood partitioning of DEA (

1.3, Tab. 4). DEI

had the lowest human blood to plasma ratio and a
fourfold higher blood cell association in the dog
compared to human (Tab. 4).

Pharmacokinetic Linearity of NTBI

Studies were performed in the mouse, rat and
monkey after oral administration of doses up to
300 mg/kg. Figure 3 compares the dose-normal-
ized plasma AUC versus dose for these studies.
There was no more than a twofold change in the
dose-normalized AUC across the dose range for
each specie and sex. Given that separate groups of
animals were studied in each case, we conclude
that the pharmacokinetics are largely dose-
proportional over the 25- to 30-fold dose range
studied in the rodents and the 7-fold dose range
studied in the monkey. Generally there were no
substantial differences between male and female
animals studied. On a dose for dose basis, mouse
exposure (plasma AUC) exceeded that observed in

all other species by about threefold. Studies of
pharmacokinetic linearity in the dog were con-
founded by lack of tolerability in this species.
Acute effects on the central nervous system were
apparent with varied frequency after oral doses as
low as 30 mg/kg in the dog. An oral dose of 30 mg/
kg chloroquine produced similar clinical signs
suggesting this species is particularly sensitive to
the 4-aminoquinolines (data not shown).

Prediction of Human Pharmacokinetics
by Allometry

Four different methods were used for allometric
scaling of the NTBI animal pharmacokinetic data

Figure 3. Linearity of the oral pharmacokinetics of
NTBI in mouse (squares), rat (circles) and monkey
(triangles). Mean dose-normalized oral AUCs are com-
pared by dose and specie. Filled symbols represent
female animals. Rodent exposures represent mean data
(

n

¼ 3) or were derived from composite study designs.

Monkey exposures are from individual animals (1M,
1F/group).

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

370

DAVIS ET AL.

to humans as described previously.

11

Table 3

summarizes the results of these analyses. The
highest

R

2

(0.991) and lowest

p-value (0.008,

representing significance of the slope from zero)
were obtained using the method in which CL was
scaled using a factor accounting for species
differences in mean life span and bile flow.

18

Blood CL was predicted to be low (2.5 mL/min/kg,
12% liver blood flow

17

) using this method although

all four methods predicted human CL to be <35%
liver blood flow. Human blood

V

ss

was predicted to

be 46 L/kg (

p < 0.05) for NTBI (9 L/kg for DEI,

p < 0.05) using a standard body weight scaling
approach. Human blood effective

t

1/2

for NTBI was

predicted to be 21 h (0.693/CL/

V

ss

). Predicted

blood CL for DEI was about two times higher than
NTBI for each for the four methods employed
(data not shown). This was similar to the differ-
ences in blood CL between NTBI and DEI
observed in the preclinical species (Tab. 1).

Comparative Pharmacokinetics of
4-Aminoquinoline Analogues

The pharmacokinetics of both ISO and DEI were
studied following iv and oral administration to
mouse, rat, dog, and monkey. By analogy with
ADQ, which undergoes oxidative N-dealkylation
to form an active metabolite in humans, we
postulated that ISO might form the analogous
metabolite

in vivo and thereby act as a pro-drug.

We also considered the possibility that DEI might
have suitable properties for development in its
own right. Table 1 includes a summary of DEI and
ISO pharmacokinetics along with those of NTBI.
Blood CL of ISO was very high, exceeding the rate
of blood flow to the liver in mouse, rat and dog.
Mean oral bioavailability of ISO in rodents was
20%. In the dog, concentrations of ISO were
nonquantifiable following an oral dose of 3 mg/kg
(LLQ of 10 ng/mL). DEI had lower blood CL
compared to ISO in mouse, rat, and particularly
dog (

p < 0.001, t-test). In the dog, blood clearance

of DEI was

2-fold higher than that of NTBI

(

p < 0.05, t-test). In the monkey, blood CL of

DEI was >3-fold higher than blood CL of NTBI
(

p < 0.001, t-test). Oral bioavailability of DEI was

40% in the monkey and higher in rodents. Emesis
in the dog study precluded an estimate of the oral
bioavailability in this specie.

Pharmacokinetic studies were performed fol-

lowing oral administration of ISO in mouse, rat
and dog to confirm that in fact ISO behaves as
a pro-drug like ADQ. Both ISO and DEI were
quantified in these studies and the metabolic ratio
was estimated by comparing AUC(0-t) for parent
and metabolite. In the three species, the exposure
of DEI exceeded the exposure of ISO by

6- to

14-fold (Tab. 2).

Plasma Protein Binding

Protein binding was similar for NTBI, DEI, and
DEA in human plasma (86–92%, Tab. 4). Mouse
protein binding was lower for DEA (74%) com-
pared to NTBI (93%) and DEI (93%). NTBI and
ADQ have similar ED50s against

Plasmodium

yoelii in a mouse model of infection and similar
total blood AUCs (DEA following administration
of ADQ) when administered the same dose in
this model (data not shown). This suggests the
differences in mouse protein binding are not
important pharmacologically in this rodent model.

Intrinsic Clearance

CLi of NTBI and related 4-aminoquinoline deri-
vatives was compared in animal and human
hepatocytes

in vitro. The results are summarized

in Table 5. As expected of this pro-drug, ADQ had
high CLi in mouse, rat, dog, monkey, and human
hepatocytes. ISO had high CLi in mouse, rat, dog,
and monkey hepatocytes but moderate CLi in
human hepatocytes. The major metabolites of
ADQ and ISO, the result of N-dealkylation, had
lower CLi compared to the parent drugs in all
species. NTBI had lower CLi in mouse, rat and dog

Table 2. Systemic Exposure of Isoquine and Des-Ethyl Isoquine Following Oral Administration of Isoquine to
Animals

Specie

Dose (mg/kg)

ISO, AUC

a

(mg h/mL)

DEI, AUC (mg h/mL)

DEI:ISO, AUC Ratio

Mouse

10

0.14

1.42

10.1

Rat

12

0.17

0.03

2.5

0.4

14.4

Dog

100

1.3

0.8

8.0

1.7

6.0

a

Plasma AUC from time 0 to the last quantifiable drug concentration.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

371

hepatocytes relative to DEI, DEA, ISO, and ADQ.
In monkey or human hepatocytes, CLi of NTBI,
DEI, and DEA was similar and low.

Biotransformation

A preliminary search for metabolites of NTBI and
DEI was conducted following incubation of the
compounds with liver microsomes and hepato-
cytes from mouse, rat, dog and human. For both
NTBI and DEI, there was no evidence for gluta-
thione conjugation or reactive metabolism

via the

quinone imine pathway associated with ADQ.

NTBI and DEI were metabolized by multiple
metabolic pathways

in vitro. In liver microsomes

of animal and human origin, mono-oxygenation
metabolites (M2, M3, M4), aldehyde (M5), car-
boxylic acid (M6) and glucuronides of parent, M5
and M6 were detected for both compounds (Fig. 4).
Although N-dealkylation metabolite (M1) was
detected for DEI, it was not detected for NTBI
in vitro. Ketone metabolite (M7) was detected for
NTBI in all species. However, DEI did not form
M7.

NTBI exhibited a molecular ion peak (MH

þ

)

with a mass to charge ratio (m/z) of 356. Collision
induced dissociation of NTBI produced a frag-
ment with

m/z 283 by loss of N-tert-butyl amine.

Mono-oxygenation metabolites M2, M3, and M4
had MH

þ

at

m/z 372. Metabolite M2 had a

fragment ion at

m/z 283 due to the loss of mono-

oxygenated N-tert-butyl amine, while metabolites
M3 and M4 had a fragment ion at

m/z 299 from

the loss of the N-tert-butyl amine moiety. This
suggested that the oxygenation in M3 and M4 did
not occur on this moiety. Aldehyde metabolite M5
and carboxylic acid metabolite M6 had MH

þ

at

m/z 299 and 315, respectively. Both metabolite
M5 and M6 lost a water molecule to produce a
fragment ion at

m/z 281 and 297, respectively.

Ketone metabolite M7 had MH

þ

at

m/z 370 and an

abundant fragment ion at

m/z 316 from the loss of

the N-tert-butyl moiety. M7 also had a fragment
at

m/z 272 due to the loss of carbonyl and N-tert-

butyl amine moieties. The fragmentation pattern
for glucuronides of parent, M5 and M6 were
characterized by neutral loss of glucuronide
moiety (176 Da) and the characteristic fragmen-
tation data described above for parent, M5 and
M6. The N-dealkylation metabolite M1 had MH

þ

at

m/z 300 which in turn produced a fragment ion

at

m/z 283 from loss of ammonia.

Table 4.

In Vitro Blood Cell Association and Plasma

Protein Binding of 4-Aminoquinolines

Compound

Specie

Blood to

Plasma Ratio

% Bound

NTBI

Mouse

1.63

0.22

92.8

1.2

Rat

1.63

0.12

ND

Dog (1)

15.65

5.04

ND

Dog (2)

4.48

0.24

ND

Monkey (1)

3.92

0.20

ND

Monkey (2)

3.73

0.39

ND

Human

1.27

0.02

88.3

3.1

DEI

Mouse

0.92

0.08

93.0

1.0

Rat

2.93

0.40

ND

Dog

3.37

0.39

ND

Monkey

1.87

0.18

ND

Human

0.79

0.04

91.7

1.4

DEA

Mouse

ND

74.4

2.7

Human

1.37

0.24

85.9

2.7

ND, no data.

Data are expressed as the mean

SD (n ¼ 3 for partition-

ing,

n

¼ 6 for protein binding, 1 mg/mL nominal blood or plasma

concentration). Dog and monkey blood was obtained from
different animals than those studied

in vivo.

Table 3. Prediction of Human Pharmacokinetics of N-Tert-Butyl Isoquine by
Allometry

Exponent b

a

R

2

Human CL (mL/min/kg)

p-value

b

Method 1

0.840

0.978

7.00

0.0195

Method 2

1.265

0.985

4.94

0.0138

Method 3

1.814

0.987

3.58

0.0119

Method 4

1.466

0.991

2.48

0.0082

a

b: slope from linear regression analysis of the log transformed data;

R

2

: square of correlation

coefficient; Method 1: CL

¼ a(W)

b

, where

W

¼ body weight; Method 2: CL MLP ¼ a(W)

b

and mean

life span (MLP in years)

¼ 185.4 (BW)

0.636

W

0.225

, where BW

¼ brain weight.; Method 3:

BW

CL ¼ a(W)

b

; Method 4: CL/BF

MLP ¼ a(W)

b

where BF is a species-dependent bile flow

correction factor based on liver weight.

18

b

Significance of slope from zero.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

372

DAVIS ET AL.

The exact position of oxygenation for M2, M3,

and M4 could not be ascertained due to the low
sensitivity in the MS

3

spectrum. Also for M6, due

to the very low extent of formation, the nature of
glucuronide, -acyl versus -hydroxyl, could not be
determined. The mass spectral fragmentation of
DEI and its metabolites was similar to that
described above for NTBI and NTBI metabolites.
The

in vitro metabolism of DEI was consistent

with urinary and biliary metabolites described
previously following administration of tritiated
ISO to rats.

10

Generally, hepatocyte metabolite

profiles were similar to those observed for liver
microsomes.

All metabolites observed

in vitro were similarly

detected in the plasma and/or urine of mice
administered NTBI or DEI orally. In contrast to
the

in vitro studies, M1 was detected for NTBI

in vivo and the M7 analog of DEI was detected
in vivo following oral administration of DEI. To
assess the relative importance of N-dealkylation
in the

in vivo biotransformation of these leads,

following oral administration of NTBI or DEI to
the mouse, systemic exposure to the N-dealky-
lation metabolite (M1) was compared. Dose-
normalized AUC for M1 following administration
of DEI was >1000 times the dose-normalized AUC
for M1 following administration of NTBI. Total

Table 5. Intrinsic Clearance of 4-Aminoquinolines in Animal and Human
Hepatocytes

Compound

Mouse

Rat

Dog

Monkey

Human

NTBI

6.4

1.0

<0.5

2.9

0.6

2.1

0.5

<0.5

ISO

41

7

11

4

24

1

21

1

3.3

0.1

DEI

17

1

1.0

0.2

5.4

0.4

1.2

0.1

<0.5

ADQ

>50

9.9

5.3

12.4

0.5

29

1

9.8

0.2

DEA

11

2

1.9

0.4

4.4

0.3

1.6

0.2

<0.5

CLi, mL/min/g liver.

Figure 4.

Metabolites of NTBI and DEI following incubation with animal and human

liver microsomes and hepatocytes

in vitro. Solid arrows designate pathways common to

both compounds and dashed arrows designate divergent pathways. For both compounds,
glucuronidation metabolites of parent, M5 and M6 were detected in liver microsomes
and hepatocytes. M1 was detected only with DEI; M7 was detected only with NTBI.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

373

urinary excretion of M1 was >60-fold higher for
DEI compared to NTBI in the mouse.

Human P450 Inhibition

The potential for inhibition of the major human
cytochrome P450 isozymes was compared, for
NTBI and related 4-aminoquinoline derivatives,
using

sensitive

screening

assays

employing

purified recombinant enzymes and fluorescent
probe substrates. The results of these studies are
summarized in Table 6. Inhibition of 2D6-
mediated

4-methylaminomethyl-7-methoxycou-

marin

O-dealkylation was observed for NTBI,

ISO, DEI, ADQ, and DEA. For this isozyme, IC50
values ranged from 3 to 7 mM across this series.
For NTBI, mean IC50 values were

23 mM for all

other isozymes studied, suggesting limited poten-
tial for inhibition of 1A2, 2C8, 2C9, 2C19, and 3A4.
NTBI did not show time-dependent inhibition of
1A2, 2C9, 2C19, 2D6, or 3A4 as there was no
consistent decrease in IC50 or in percent control
activity at 100 mM over a 30-min incubation
period.

Generally, the potential for inhibition of 1A2,

2C9, 2C19, and 3A4 was modest across this series.
However, DEI had a substantially lower IC50
against CYP1A2 (8 mM) compared to NTBI (29
m

M, Tab. 6,

p < 0.005, t-test). Activation of human

CYP3A4 was observed for NTBI and ISO (three-
fold stimulation of activity for both

19

). ADQ was a

potent 2C8 inhibitor (mean IC50

3 mM). DEA

had a much higher 2C8 IC50 (mean of 20 mM,
p < 0.005, t-test) and this value was similar to both
DEI (23 mM) and NTBI (23 mM).

Permeability and Active Transport

Consistent with the high solution oral bioavail-
ability of NTBI in animals, NTBI exhibited high
passive permeability across MDR1-MDCKII cell
monolayer (

324 nm/s, Tab. 7). NTBI had an

apical efflux ratio of

1.3, which was also

unaffected by the presence of the P-gp inhibitor
GF120918; therefore, NTBI was not a substrate
for this transporter. Amprenavir performed as
expected in this assay (efflux ratio of

16, which

collapsed to

1 in the presence of GF120918; high

intrinsic

P

app

). The high permeability appeared

common among the 4-aminoquinoline analogues
examined with little differentiation in this respect
between NTBI, ISO, DEI, and ADQ. In addition,
the 4-aminoquinolines, with the possible excep-
tion of DEI (efflux ratio of 2.1), appeared not to be
P-gp substrates.

DISCUSSION

A detailed comparison of the developability of
promising new 4-aminoquinoline antimalarials
has been undertaken in an effort to guide selection

Table 6. Comparative P450 Inhibition Potential of 4-Aminoquinolines

Compound

3A4 DEF

3A4 PPR

1A2

2D6

2C8

2C9

2C19

NTBI

A

a

>100

29

6

3.0

0.5

23

2

90

24

10

ISO

A

a

>100

23

8

7.0

0.2

13

2

75

8

39

11

DEI

>100

55

8

3

3.4

0.1

23

7

>100

62

16

ADQ

28

6

>100

15

5

6.1

0.5

2.7

1.4

62

9

85

DEA

24

3

>100

28

11

6.4

0.1

20

4

58

2

54

11

Mean and standard deviation of three independent IC50 (mM) determinations reported.

a

Activation observed.

Table 7. Comparative Permeability and Active Transport of 4-Aminoquinolines

Compound

P

app

(B-to-A)/

P

app

(A-to-B)

Average

P

app

(nm/s)

with GF120918

Without GF120918

With GF120918

NTBI

1.3

1.7

324

61

ISO

1.3

1.4

455

171

DEI

2.1

1.1

425

170

ADQ

1.5

1.6

457

87

Amprenavir

16

1.2

386

56

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

374

DAVIS ET AL.

of an appropriate candidate for further develop-
ment. The lead molecules investigated all possess
a distinct advantage over ADQ: elimination of the
potential for formation of reactive metabolites

via

the quinoneimine pathway. NTBI, ISO and DEI
retain potent antimalarial activity

in vitro, activ-

ity against strains carrying resistance deter-
minants to chloroquine, and similar activity in
preclinical models of malaria infection following
oral administration.

Pro-drug

conversion

to

active

metabolite

appears to be a characteristic of ISO as it is for
ADQ. DEI itself has good oral bioavailability in
animals but was a more potent inhibitor of
CYP1A2 compared to the other 4-aminoquino-
lines. Thus, there may be increased liability for
clinical drug–drug interactions for ISO and DEI
compared to NTBI. DEI had high blood clearance
in the monkey. This result was somewhat
surprising given the low CLi in monkey (and
human) hepatocytes. Nonetheless, it may suggest
an increased likelihood of unfavorable pharmaco-
kinetics for ISO and DEI in humans.

Replacement of the N-ethyl with an N-tert-

butyl substituent substantially reduced the rate of
N-dealkylation of NTBI compared to ISO and DEI.
This conclusion is supported by the lower blood
clearance of NTBI in mouse, rat, dog and monkey
and the very substantial differences in systemic
exposure and urinary excretion of the N-dealky-
lation metabolite (M1) following oral administra-
tion of NTBI or DEI to the mouse.

NTBI and the other 4-aminoquinoline anti-

malarials inhibited P450 2D6 in our assay system.
There is limited data in the literature suggesting
clinically relevant drug–drug interactions for
ADQ.

20

P450 2D6, 2C8, and 3A4 have been shown

to be involved in the biotransformation of
chloroquine

in vitro.

21

However, drug–drug inter-

actions involving chloroquine appear to be rare.

20

ADQ and to a lesser extent ISO inhibited
recombinant 2C8-mediated diethoxy fluorescein
O-dealkylation. NTBI was not a potent 2C8
inhibitor in the assay perhaps because it is not
as good a substrate. It has been shown that ADQ is
a relatively selective 2C8 substrate and that 2C8
is responsible for ADQ N-dealkylation.

22

2C8 also

catalyzes the N-dealkylation of ISO; the isozyme
selectivity of this biotransformation warrants
further investigation. As any new antimalarial
is likely to be used in combination therapy to
reduce the development of resistance, further
P450 enzymology studies of NTBI will be essential
to determine the most appropriate combinant and

to fully evaluate the potential for clinically
relevant drug–drug interactions.

Blood clearance of NTBI was lower in mouse

(15% liver blood flow) compared to the rat (50%
liver blood flow) despite higher apparent CLi in
hepatocytes. Interestingly, DEI also had higher
blood clearance in the rat compared to the mouse.
DEI had lower oral bioavailability in rat and
monkey compared to NTBI, but oral bioavail-
ability was nearly complete in mouse for both
leads. Given the high passive permeability of
these molecules, the lower bioavailability in rat
and monkey may reflect the higher blood clear-
ance in these species (and the possibility of a more
substantial first-pass effect).

We observed subtle secondary increases in

the plasma concentration-time profile of NTBI
in the mouse, rat, and monkey. Given that
glucuronide conjugates of parent were observed
in animal and human hepatocytes

in vitro, it is

plausible that entero-hepatic recirculation was
the cause of these minor secondary increases in
systemic drug concentration. In fact, in the mouse,
parent glucuronide was detected in the urine after
administration of NTBI.

Partitioning of drug into red blood cells is

potentially advantageous for an anti-malarial as
the site of action is the infected erythrocyte. We
describe here substantial partitioning of NTBI
into blood cells of the normal dog and monkey
(

in vitro and in vivo studies) and to less extent in

normal rodent and human blood

in vitro. Sub-

stantial accumulation of 4-aminoquinoline anti-
malarials has been described in blood cells of
infected animals as well as humans and this may
be the result of direct interaction of the drugs with
the target haem.

23

Consistent with this hypo-

thesis, we have observed, in our rodent model
of malaria, more extensive partitioning of NTBI
into blood cells of infected animals compared to
uninfected animals (data not shown).

Using allometric scaling, human blood CL was

estimated to be low (12% liver blood flow), blood
V

ss

was estimated to be very large (46 L/kg) and

t

1/2

was predicted to be

21 h. Chloroquine volume

of distribution is reportedly extremely high as well
(100–1000 L/kg

20

). DEA has an extremely long

terminal half-life life of 9–18 days; chloroquine
half-life is reportedly 8–56 days.

20

However,

the low drug concentrations associated with this
prolonged terminal concentration-time phase are
thought to be sub-therapeutic and may in fact
contribute to the selection of resistant strains of
the parasite.

20

Human NTBI pharmacokinetics

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

375

estimated here are admittedly based on data from
healthy animals and the impact of infection on
human blood partitioning and pharmacokinetics
are not known at this time. Therefore, these
estimates should be interpreted cautiously.

NTBI has many of the properties of an ideal

anti-malarial agent: half-life appropriate for short
duration of therapy by the oral route of admin-
istration, ease of chemical synthesis from readily
available starting materials, activity against
clinical isolates carrying resistance determinants
to marketed drugs and no cross-resistance with
agents currently in use. Clearly, this promising
profile warrants further investigation.

ACKNOWLEDGMENTS

The authors thank Richard Grater, Xinhe Jiang,
Lauren Kaskiel, Yan Liu, Thomas Mencken,
Dung Nguyen, Diane Talaber Kepner and Lisa
Woods for expert technical assistance, Young
Shin for his contribution to the biotransformation
studies, Chao Han for critical review of the manu-
script and our colleagues within GlaxoSmith-
Kline’s center for Diseases of the Developing
World (Tres Cantos, Spain) for many interesting
discussions. This work was supported by Glaxo-
SmithKline and the Medicines for Malaria
Venture.

REFERENCES

1. World Health Organization. 2000. Severe falci-

parum malaria. Trans R Soc Trop Med Hyg 94:1–10.

2. Francis SE, Sullivan DJ Jr, Goldberg DE. 1997.

Hemoglobin metabolism in the malaria parasite
Plasmodium falciparum. Annu Rev Microbiol 51:
97–123.

3. Bray PG, Janneh O, Raynes KJ, Mungthin M,

Ginsburg H, Ward SA. 1999. Cellular uptake of
chloroquine is dependent on binding to ferriproto-
porphyrin IX and is independent of NHE activity in
Plasmodium falciparum. J Cell Biol 145:363–376.

4. Bray PG, Martin RE, Tilley L, Ward SA, Kirk K,

Fidock DA. 2005. Defining the role of PfCRT in
Plasmodium falciparum chloroquine resistance.
Mol Microbiol 56:323–333.

5. Neftel KA, Woodtly W, Schmid M, Frick PG, Fehr J.

1986. Amodiaquine induced agranulocytosis and
liver damage. Br Med J 292:721–723.

6. Hatton CSR, Peto TEA, Bunch C. 1986. Frequency

of severe neutropenia associated with amodiaquine
prophylaxis against malaria. Lancet I:411–414.

7. Rouveix B, Coulombel L, Aymard JP, Chau F, Abel

L. 1989. Amodiaquine induced immune agranulo-
cytosis. Br J Haematol 71:7–11.

8. Christie G, Breckenridge AM, Park BK. 1989. Pro-

duction of antibodies towards amodiaquine and
amodiaquine quinoneimine in man and the rat.
Biochem Pharmacol 38:1451–1458.

9. Mount DL, Patchen LC, Nguyen-dinh P, Barber

AM, Schwartz IK, Churchill FC. 1986. Sensitive
analysis of blood for amodiaquine and three meta-
bolites by high-performance liquid chromatography
with electrochemical detection. J Chromatogr 383:
375–386.

10. O’Neill PM, Mukhtar A, Stocks PA, Randle LE,

Hindley S, Ward SA, Storr RC, Bickley JF, O’Neil
IA, Maggs JL, Hughes RH, Winstanley PA, Bray
PG, Park BK. 2003. Isoquine and related amodia-
quine analogues: A new generation of improved
4-aminoquinoline antimalarials. J Med Chem 46:
4933–4945.

11. Xiang H, McSurdy-Freed J, Subbanagounder G,

Hugger E, Bambal R, Han C, Ferrar S, Gargallo
D, Davis CB. 2006. Preclinical drug metabolism
and pharmacokinetic evaluation of GW844520, a
novel mitochondrial electron transport inhibitor.
J Pharm Sci 95:2657–2672.

12. Bayliss MK, Bell JA, Jenner WN, Park GR, Wilson

K. 1999. Utility of hepatocytes to model species
differences in the metabolism of loxitidine and to
predict pharmacokinetic parameters in rat, dog and
man. Xenobiotica 29:253–268.

13. Hugger ED, Novak BL, Burton PS, Audus KL,

Borchardt RT. 2002. A comparison of commonly
used polyethoxylated pharmaceutical excipients
on their ability to inhibit P-glycoprotein activity
in vitro. Pharm Res 91:1991–2002.

14. Polli JW, Wring SA, Humphreys JE, Huang L,

Morgan JB, Webster LO, Serabjit-Singh CS.
2001. Rational use of in vitro P-glycoprotein assays
in drug discovery. J Pharmacol Exp Ther 299:620–
628.

15. Crespi CL, Miller VP, Penman BW. 1997. Microtiter

plate assays for inhibition of human, drug-metabo-
lizing cytochromes P450. Anal Biochem 248:188–
190.

16. Xu X, Yan KX, Song H, Lo M-W. 2005. Quantitative

determination of a novel dual PPAR a/g agonist
using on-line turbulent flow extraction with
liquid chromatography-tandem mass spectrometry.
J Chromatogr B 814:29–36.

17. Davies B, Morris T. 1993. Physiological parameters

in laboratory animals and humans. Pharm Res 10:
1093–1095.

18. Mahmood I, Sahajwalla C. 2002. Interspecies scal-

ing of biliary excreted drugs. J Pharm Sci 91:1908–
1914.

19. Tang W, Stearns RA. 2001. Heterotropic co-

operativity of cytochrome P450 3A4 and potential

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

DOI 10.1002/jps

376

DAVIS ET AL.

drug-drug interactions. Curr Drug Metab 2:185–
198.

20. Giao PT, de Vries PJ. 2001. Pharmacokinetic inter-

actions of antimalarial agents. Clin Pharmacokinet
40:343–373.

21. Projean D, Baune B, Farinotti R, Flinois J-P,

Beaune P, Taburet A-M, Ducharme J. 2003.
In vitro metabolism of chloroquine: Identification
of CYP2C8, CYP3A4, and CYP2D6 as the main
isoforms catalyzing N-des-ethylchloroquine forma-
tion. Drug Metab Dispos 31:748–754.

22. Li XQ, Bjorkman A, Andersson TB, Ridderstrom M,

Masimirembwa CM. 2002. Amodiaquine clearance
and its metabolism to N-des-ethylamodiaquine is
mediated by CYP2C8: A new high affinity and
turnover enzyme-specific probe substrate. J Phar-
macol Exp Ther 300:399–407.

23. Fitch CD, Yunis NG, Chevli R, Gonzalez Y. 1974.

High-affinity accumulation of chloroquine by mouse
erythrocytes infected with Plasmodium berghei.
J Clin Investig 54:24–33.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 1, JANUARY 2009

COMPARATIVE DMPK OF NOVEL 4-AMINOQUINOLINE ANTI-MALARIALS

377