Structure-activity relationships of synthetic cordycepin analogues as experimental therapeutics for African trypanosomiasis

Article abstract of DOI:10.1021/jm401530a

Novel methods for treatment of African trypanosomiasis, caused by infection with Trypanosoma brucei are needed. Cordycepin (3′-deoxyadenosine, 1a) is a powerful trypanocidal compound in vitro but is ineffective in vivo because of rapid metabolic degradation by adenosine deaminase (ADA). We elucidated the structural moieties of cordycepin required for trypanocidal activity and designed analogues that retained trypanotoxicity while gaining resistance to ADA-mediated metabolism. 2-Fluorocordycepin (2-fluoro-3′-deoxyadenosine, 1b) was identified as a selective, potent, and ADA-resistant trypanocidal compound that cured T. brucei infection in mice. Compound 1b is transported through the high affinity TbAT1/P2 adenosine transporter and is a substrate of T. b. brucei adenosine kinase. 1b has good preclinical properties suitable for an oral drug, albeit a relatively short plasma half-life. We present a rapid and efficient synthesis of 2-halogenated cordycepins, also useful synthons for the development of additional novel C2-substituted 3′-deoxyadenosine analogues to be evaluated in development of experimental therapeutics.

Full text of DOI:10.1021/jm401530a

Article
pubs.acs.org/jmc
Structure−Activity Relationships of Synthetic Cordycepin Analogues
as Experimental Therapeutics for African Trypanosomiasis
Suman K. Vodnala,† Thomas Lundback,‡,⊥ Esther Yeheskieli,† Birger Sjoberg,‡,⊥
̈
̈
Anna-Lena Gustavsson,‡,⊥ Richard Svensson,§,⊥ Gabriela C. Olivera,† Anthonius A. Eze,∥
Harry P. de Koning,∥ Lars G. J. Hammarstrom,‡,⊥,# and Martin E. Rottenberg*,†,#
†Department of Microbiology, Tumor and Cell Biology and ‡Division of Translational Medicine and Chemical Biology, Department
of Medical Biochemistry and Biophysics, Karolinska Institutet, 171 77 Stockholm, Sweden
§Uppsala University Drug Optimization and Pharmaceutical Profiling Platform (UDOPP), Department of Pharmacy, Uppsala
̈
University, 753 12 Uppsala, Sweden
∥Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences, University of Glasgow,
Glasgow, G12 8QQ, U.K.
⊥Chemical Biology Consortium Sweden, Sweden
ABSTRACT: Novel methods for treatment of African
trypanosomiasis, caused by infection with Trypanosoma brucei
are needed. Cordycepin (3′-deoxyadenosine, 1a) is a powerful
trypanocidal compound in vitro but is ineffective in vivo
because of rapid metabolic degradation by adenosine
deaminase (ADA). We elucidated the structural moieties of
cordycepin required for trypanocidal activity and designed
analogues that retained trypanotoxicity while gaining resistance
to ADA-mediated metabolism. 2-Fluorocordycepin (2-fluoro-
3′-deoxyadenosine, 1b) was identified as a selective, potent, and ADA-resistant trypanocidal compound that cured T. brucei
infection in mice. Compound 1b is transported through the high affinity TbAT1/P2 adenosine transporter and is a substrate of
T. b. brucei adenosine kinase. 1b has good preclinical properties suitable for an oral drug, albeit a relatively short plasma half-life.
We present a rapid and efficient synthesis of 2-halogenated cordycepins, also useful synthons for the development of additional
novel C2-substituted 3′-deoxyadenosine analogues to be evaluated in development of experimental therapeutics.
to 21 000 cases in 2012.3 Despite this positive trend, the limited
number of safe and efficacious drugs available for treatment and
the difficulties in administration of current therapies warrant
continued investigation of new methods to treat this
INTRODUCTION
■
Human African trypanosomiasis (HAT) is caused by infection
with the extracellular protozoan parasite Trypanosoma brucei.
Parasites are transmitted through the bite of the tsetse fly
(Glossina sp.) vector, which infests vast areas of sub-Saharan
Africa. Infection with T. brucei causes a debilitating wasting
disease in livestock, and sleeping sickness in humans. The
subspecies T. b. gambiense causes a protracted human infection
prevalent in Western and Central Africa, while human T. b.
devastating disease. It is worth remembering that sleeping
sickness was almost eradicated in the late 1950s/early 1960s
but returned to epidemic proportions in the 1990s.4
Management of HAT is critically dependent on an accurate
diagnosis of the stage of infection, since most current
trypanocidal drugs do not transverse the BBB and are thus
rhodesiense infection, with a more rapid disease progression,
unsuitable for treatment of late stage infections. The few drugs
dominates in Eastern and Southern Africa, where it also infects
wild and domestic animals.1
The human disease is characterized by two stages. In the
early stage the parasite is found in the blood and lymph;
infected individuals show fever, joint pain, headaches, and
itching as clinical symptoms. In the late stage, the parasite
invades the central nervous system (CNS) after crossing the
that are BBB penetrating and thereby efficacious for treatment
of late stage infection are associated with severe side effects,
require a complicated treatment regimen, and/or are subspecies
specific.
Melarsoprol, an arsenic derivative discovered in 1949, is used
for treatment of late stage infection with both T. brucei
subspecies. This drug, together with eflornithine,5 which targets
blood−brain barrier (BBB), causing severe neurological
T. b. gambiense polyamine synthesis, is currently first-line
symptoms, sensory alterations such as hyperalgesia and
allodynia, poor coordination, and sleep disturbances.2 The
therapy for treating late stage infection. Treatment with
disease eventually results in coma and is invariably lethal if left
untreated. The incidence of African trypanosomiasis has been
reduced from an estimated 300 000 cases per year 12 years ago
Received: July 15, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
melarsoprol has many undesirable side effects, and 10% of
treated patients develop reactive encephalopathy.6 Eflornithine
is less toxic than melarsoprol, albeit requiring a cumbersome
treatment regimen. Nifurtimox−eflornithine combinational
therapy (NECT) was introduced by WHO in 2009, and
phase III clinical studies indicate that NECT is equally
efficacious as standard eflornithine monotherapy and is easier
to administer.7,8 However, resistance toward eflornithine and
nifurtimox is easily achieved in vitro with loss of the AAT6
amino acid transporter involved in eflornithine uptake or with
the loss of the nitroreductase that activates nifurtimox.9,10
All protozoan parasites so far studied, including trypano-
somes, lack the ability to synthesize purines de novo and are
thus completely dependent on the host as a purine source.11 As
a result, trypanosomes have developed a family of high affinity
nucleoside and/or nucleobase transporters for purine scaveng-
ing. This host-dependent purine metabolism presents an
opportunity for pharmacological intervention, and several
purine analogues have shown potent and selective trypanocidal
properties.5,12−14 Sinefungin, formycin B, and 7-deazainosine
have the ability to cure acute (early) models of trypanosomiasis
but fail in curing late encephalitic stage infections.5 The
trypanocidal properties of cordycepin (3′-deoxyadenosine, 1a,
Figure 1), isolated from the parasitic fungus Cordyceps militaris,
Despite exhibiting a potent trypanocidal activity in vitro, in
vivo administration of 1a failed to cure T. brucei infections in
mice.17 However, when 1a was coadministered with the
adenosine deaminase (ADA) inhibitor deoxycoformycin
(pentostatin, Dcf), the combination led to a complete cure of
infected mice with both human pathogenic subspecies even
when administered during the late stage of infection,
implicating mammalian ADA as a rapid and limiting
metabolizer of 1a in vivo (Figure 1). Dcf protects 1a from
deaminase metabolism but displays no parasiticidal activity in
itself. The doublet 1a−Dcf was effective when administered
orally. Although resistant T. brucei parasites were generated
after prolonged incubation with low doses of 1a, these
displayed low virulence, suggesting that clinical resistance
would not develop easily.17
Clinical development of 1a−Dcf as a combination therapy
for treatment of African trypanosomiasis is not considered
feasible, since Dcf itself has been shown to be teratogenic.26
The goal of this study was thus to develop novel, ADA-resistant
analogues to 1a that have the promise of functioning as stand-
alone drugs and to provide new insights into the mechanisms
and metabolism of nucleoside-based therapeutics against HAT.
RESULTS AND DISCUSSION
■
We aimed to identify novel cordycepin analogues that conserve
trypanocidal activity mediated through retained transport by
TbAT1 and processing by AK but gain resistance to ADA-
mediated degradation, thus increasing compound efficacy in
vivo. Through chemical synthesis and evaluation of previously
described nucleoside derivatives, we elucidated the structural
moieties on 1a required for trypanocidal activity and selected
modifications that would not alter trypanotoxicity while gaining
resistance to ADA-mediated metabolism. This work resulted in
a series of analogues closely related to 1a, the activities of which
were first tested in vitro and concomitantly evaluated in early
stage in vivo models of the disease.
Figure 1. Cordycepin 1a is rapidly metabolized in vivo in host plasma
by mammalian adenosine deaminase to the corresponding inactive
inosine derivative.
Structure−Activity Relationships of Cordycepin Ana-
logues. In order to design adenosine deaminase resistant
analogues of 1a, we investigated which structural components
of the nucleoside were critical for trypanocidal activity and
which positions of the scaffold were amenable to synthetic
alteration. In combination with literature data, we designed
novel scaffolds that were tested for in vitro and in vivo efficacy.
A summary of all investigated compounds can be seen in Table
1.
A Free Amine Moiety at C6 Is Essential To Retain
Trypanocidal Activity. ADA catalyzes the deamination of the
C6-NH2 group of adenosine analogues to their corresponding
inosine nucleobases (Figure 1); hence, modification or
protection of the C6-NH2 was the first approach investigated
to develop ADA-resistant cordycepin analogues. Wei and co-
workers have described the increased metabolic stability of N-
acylated derivatives of 1a in vivo.27 Despite exhibiting potent in
vitro trypanocidal activity in our assays (IC50 = 28 nM), N-
octanoylcordycepin (compound 2, Table 1) was unable to cure
mice infected ip with 2000 T. brucei and, similarly to 1a,
appears to be deactivated by ADA. Specifically, mice were
treated with 4 mg/kg (10 doses, once daily) 2 starting at 2 h
after infection. Parasitemia was undetectable for 18 days after
infection. However, all mice relapsed to control levels between
day 19 and day 30. We suspect that this compound acts as a
prodrug that is actively converted to 1a and concomitantly
hydrolyzed by ADA. In line with this theory, three doses of
were recognized as early as the 1970s.15 We more recently
confirmed that cordycepin is a potent (in vitro IC50 = 32
nM)13,16 and selective trypanocidal compound after screening a
nucleoside library.17
Although the exact mode of action of 1a remains somewhat
elusive, it is well-known to affect RNA metabolism. Points of
interference include an impact on normal RNA chain
elongation resulting from the absence of the 3′-hydroxyl
group.18 Recent studies in yeast demonstrated an additional
role for 1a in the processing of the poly(A) tails added at the 3′
end during maturation of mRNA.19 The formation of the
triphosphorylated form of 1a is important for these effects, as
the impact was abolished in mutant yeast strains defective in
either adenosine kinase (AK) or adenylate kinase.20 The
dependency on adenosine kinase for the trypanocidal effects of
1a has also been established.21 Interestingly, the high affinity
and selectivity of trypanosomal AK for adenosine and its
analogues may make this pathogen particularly susceptible to
1a and its analogues.21,22
To elicit toxicity in T brucei, 1a must be actively transported
into the parasite. The transport of adenosine analogues is
mainly mediated by the trypanosomal adenosine transporter
P2, encoded by the TbAT1 gene,23,24 although a few
trypanocidal analogues, including formycin A and formycin B,
have much higher affinity for the P1 nucleoside transporter.25
B
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
either 2 mg/kg 1a or 4 mg/kg 2 resulted in complete cure
when administered in combination with 0.2 mg/kg Dcf (Figure
2B). Coadministration of Dcf also potentiated the trypanocidal
activity of 2 in vitro (from 28 to 1.9 nM, Figure 2A). ADA
activity was absent in parasite lysates but present in the serum
used for the trypanocidal test (data not shown).
Table 1. Compounds Investigated and Their Corresponding
in Vitro Activity on Trypanosoma brucei Subspecies and
Mammalian Murine Cell Line L929 Viability
As the C6-NH2 is the site of metabolic deactivation of 1a, we
further studied whether presence of this group was essential for
trypanocidal activity. Nebularine (purine-9β-D-ribofuranoside)
3 has previously been shown to be cytotoxic against both
eukaryotic and prokaryotic cells28 and to be an inhibitor of
ADA.29 In our hands, however, 3 or 3′-deoxynebularine 4
showed no substantial selective toxicity (Table 1). From these
data we deduced that N6-deaminocordycepin derivatives
display no intrinsic trypanocidal activity and that the presence
of a free exocyclic amine moiety at C6 is critical to elicit potent
trypanocidal effects.
Ribose/Arabinose Modifications. Possible modifications
to the ribose sugar were studied next. As the only structural
difference between adenosine and 1a is the presence or absence
of the 3′-OH group, we assumed that the lack of the 3′-OH
does not interfere with the ability of 1a or related analogues to
act as a substrate for AK, does not inhibit AK, and does not
prevent the compounds from acting as substrates for adenosine
transporters. Thus, 1a was used as a positive control to relate
the potency of new analogues. Cordycepin triphosphate,
biosynthesized intracellularly by adenosine kinase (AK) and
adenylate kinase, has been shown to be a critical intermediate
metabolite in limiting yeast cell growth by inhibiting RNA
synthesis and maturation.20 Thus, in order to retain the ability
to phosphorylate cordycepin analogues in situ, no modifications
were done on the ribose 5′-OH position. Although 9-β-D-
arabinosyladenine (Ara A, 5) has previously shown to display
low in vitro toxicity against T. brucei,22,30 neither 5 nor the
corresponding 2-fluoro-3′-deoxyadenine arabinoside 6 showed
detectable trypanocidal activity (IC50 > 100 μM, Table 1). The
furanose epoxide intermediate 7 was also inactive. Thus, we
show that the stereochemistry of the 2′-OH group is crucial for
trypanocidal activity of the adenosine analogues.
Design of C2-Substituted Cordycepin Analogues. The
ADA-resistance of 2-fluorocordycepin (1b) was noted as early
as 1967, although little was known of the underlying molecular
interactions responsible for resistance.31 On the basis of the
available protein crystal structure,32 Gillerman and Fischer
recently described a comprehensive assessment of structural
a
IC50 values reported as averages and standard error (SE) of at least
b
three individual measurements. Apparent IC50, due to hydrolysis of
amide in vitro yielding 1a.
Figure 2. Adenosine deaminase inhibitor deoxycoformycin (Dcf) potentiate the effects of 2, indicating hydrolytic conversion to 1a in vitro and in
■
●
▲
vivo. (A) Compound 2 is potentiated in the presence of Dcf in vitro ( , compound 2 alone, IC50 = 24 nM; , 2 with 1.5 μM Dcf, IC50 = 1.0 nM).
(B) In vivo assessment of 2 indicating ADA-dependent deactivation. Group treated with compound 2 ( , 4 mg/kg, OD, 10 days) relapse after
■
●
infection to 100% of untreated control ( ). Group treated with 2 ( , 4 mg/kg, OD 3 days, coadministration of 0.2 mg/kg Dcf) showed full
clearance of parasites for 40 days after infection. Positive control group treated with 1a (×, 2 mg/kg, OD, 3 days, coadministration of 0.2 mg/kg Dcf)
showed full clearance of parasites for 40 days after infection. At least six mice per group were used.
C
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
motifs affecting substrate and inhibitor binding to human ADA
through introduction of C2 or C8 substituents onto
adenosine.29 This work suggests that C2 substitutions on
adenosine shift the compound from acting as substrate
(adenosine) to nonsubstrate (small C2 substituents) toward
inhibitors (larger C2 substituents) of ADA.
The crystal structure of mammalian ADA-adenosine
complex32 reveals key interactions of Glu-217 and Gly-184
with N1 and N3, respectively, of the adenine scaffold (Figure
3). These interactions are critical both for binding and for the
Figure 3. Schematic representation of the molecular mechanism by
which ADA binds and activates adenosine toward hydrolysis. Gly-184
and Glu-217 bind to adenosine N1 and N3, steering the C6-NH2
toward the catalytic site where Zn, His-238, and Asp-295 activate a
water molecule for nucleophilic attack of C6. The figure is adapted
from Wilson et al.32
Figure 4. (A) Calculated properties of 2-fluorocordycepin (1b) and 2-
chlorocordycepin (10) according to ChemAxon Marvin and site-
specific pK values calculated using Schrodinger software confirm a
̈
a
decrease in overall hydrogen bond acceptor capabilities at N1, N3, and
N7. (B) Superimpositioning of 1a, 1b, and 10 with their surface areas,
illustrating the increase in steric bulk at C2 as a result of halogenation,
possibly affecting TbAT1 transport ability.
catalytic conversion of adenosine to inosine, holding the
adenine scaffold in a position so that the C6-NH2 is positioned
toward the Zn-dependent catalytic site of ADA defined by Asp-
295 and His-238. Glu-217 hydrogen-bonds to N1 and also acts
as a hydrogen bond acceptor of the transition-state generated
NH at N1 when the tetrahedral intermediate at C6 is formed
prior to NH3 elimination.32 Since the initial binding of the
adenine core to ADA residues Glu-217 and Gly-184 depends
on electrostatic interactions, we hypothesized that a decrease of
the nucleobase electron density through C2 modification was
responsible for the decreased affinity for the enzyme.
The calculation of the electron densities and site specific pKa
values at N1 and N3 of cordycepin using Chemaxon software
verified that introduction of a fluorine or chlorine at C2 reduces
the partial negative charge on both N1 and N3 and thereby
dramatically reduces the intrinsic hydrogen bond acceptor
capabilities of these positions, as well as overall pKa (Figure
4A). Analogous QSAR calculation of site-specific electron
be resistant to deamination by ADA.33 Thus, we initially
prepared 2-fluorocordycepin 1b to evaluate its effects.
Compounds 1b and 10 Can Be Readily Synthesized
̈
Using Standard Silyl Hilbert−Johnson (Vorbruggen)
Chemistry. The synthesis of 1b was originally performed
through a modified procedure of the synthetic preparation of
1a via the corresponding epoxide 7 as described by Robbins et
al. (Scheme 1).34−36 Briefly, polybenzoylation of 1a in the
presence of pyridine afforded the N,N,O,O-tetrabenzoylated
cordycepin derivative 12 in 47% yield. Subsequent nitration at
the C2 position of the purine ring with TBAN in trifluoroacetic
acid anhydride gave intermediate 13 in 45% yield. TBAF-
mediated substitution of the nitro group introduced fluorine at
the C2-position of 14, and subsequent deprotection with
methanolic ammonia gave the desired 1b in 2% overall final
yield starting from adenosine.
densities using the Jaguar module within Maestro (Schro-
̈
dinger) indicates that the pKa values for N1, N3, and N7 are all
decreased upon C2 halogenation, thereby rendering the
molecule less prone to association with acidic residues in the
protein. Introduction of fluorine and chlorine significantly
increases the steric bulk of the region surrounding C2 (Figure
4).
Although reproducible, this synthesis was low yielding and
required a considerable effort to generate sufficient quantities of
1b for in vivo studies. Thus, alternative strategies for its
preparation were investigated. Surprisingly, 1b and 10 could be
accessed readily through a modified silyl Hilbert−Johnson
Analogues with the relatively small substituents fluorine and
chlorine should neither be substrates for ADA nor act as
inhibitors of the enzyme, as this has previously been described
to be promoted by larger hydrophobic substituents at C2.29 In
agreement, both 2-fluoroadenosine and 9-β-D-arabinosyl-2-
fluoroadenine (fludarabine) have previously been shown to
(Vorbruggen) chemistry via direct ribosylation of commercially
̈
available nonprotected 2-fluoroadenine or 2-chloroadenine with
5-O-benzoyl-1,2-di-O-acetyl-3-deoxy-D-ribose in the presence of
N,O-bis(trimethylsilyl)acetamide (BSA) and trimethylsilyl-
trifluoromethane sulfonate (Scheme 2) as recently described
by Nicolau et al.37 The reaction proceeds with excellent stereo-
D
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Scheme 1. Synthesis of 1b, Method A: Classical Synthesis from Adenosine
a
Conditions and reagents: (a) 2-acetoxyisobutyryl bromide, Amberlite 400, −OH, CH3CN (62%); (b) (C2H5)3LiH, DMSO (53%); (c) Bz-Cl,
pyridine (47%); (d) TBAN, TFAA, (45%); (e) TBAF, CH3CN (59%); (f) NH3, CH3OH (59%).
a
Thus, it is reasonable to assume that 1b elicits its
pharmacological effects through a similar mechanism as 1a.
Similar to 1a, 1b also showed trypanocidal activity against other
human pathogenic subspecies and showed a 300-fold differ-
ential toxicity to L929 cells (Table 1).
Scheme 2. Synthesis of 1b and 10, Method B: Modified
Silyl Hilbert−Johnson Synthesis
b
Surprisingly, 2-chlorocordycepin 10 and 2-(2-
hydroxyethylamino)cordycepin 11 exhibited very poor trypa-
nocidal activity in vitro (see Table 1). Although compound 10
shows calculated physiochemical properties (electronic dis-
tribution and predicted pKa) very similar to 1b, its steric bulk is
significantly larger (Figure 4). It is likely that the bulkier
substituents prevent the activation of the prodrug to the active
triphosphate by reducing the uptake via the adenosine
transporters, or the affinity for adenosine kinase, adenylate
kinase, and nucleoside-diphosphate kinase.
a
Conditions and reagents: (a) BSA, 5-O-benzoyl-1,2-di-O-acetyl-3-
deoxy-D-ribose, TMSOTf (98%); (b) NH3, CH3OH, (36−57%); (c)
ethanolamine, dioxane, reflux (78%). Introduction of a fluorine at C2
allows for subsequent synthesis of 2-amino-functionalized analogue 11
from 1b.
b
Compounds 1a and 1b, but Not 10, Are Excellent
Substrates for TbAT1/P2 Transport. The TbAT1 gene
codes for the P2 aminopurine transporter in T. brucei.38 This
unique transporter has been implicated in the uptake of
numerous trypanocidal compounds including diamidines,
melaminophenyl arsenicals, and adenosine analogues;39,40 its
structure−activity relationships have been investigated in
detail.41 We recently developed an optimized system for the
study of P2 activity by expressing the TbAT1 gene in the B48
clone of T. b. brucei, which lacks both an endogenous copy of
TbAT1 and the other main drug transporter, the high affinity
pentamidine transporter (HAPT1).42 Using this highly
sensitive system, we performed transport assays with [3H]-
adenosine in competition with the 2-H, 2-F, and 2-Cl
derivatives of adenine, adenosine, and cordycepin to obtain
their inhibition constants (Ki) (Figure 5). From the Ki values,
the Gibbs free energy of the transporter−substrate interaction
was calculated (Table 2).30 We found that 1a is indeed, as
predicted by current models, an excellent substrate of TbAT1/
P2, in fact better than adenosine itself (p = 0.025; Student’s t
test), as the 3′-hydroxy group slightly hinders binding rather
than helps affinity to this transporter. This further confirms that
P2 is primarily an adenine transporter tolerating the ribose
group of adenosine, as postulated on the basis of the
translocation efficiency of the aminopurines.11 2-Fluoro
substitution reduced binding energy to P2 by 3−4 kJ/mol
relative to each of the nonsubstituted purines (Table 2). The
and regioselectivity, requires no intermediate purification, and
can be performed on a multigram scale. Subsequent
deprotection of the 2′-O-acetyl/5′-O-benzoylribose moiety of
the O,O-diprotected intermediates 15 and 16 with methanolic
ammonia directly on the crude coupling products yields the
desired 1b and 10, which were isolated in 100% purity by
crystallization from water in 36−56% overall yield (see section
“Synthesis of 1b: Method B”) from 2-fluoroadenine.
Compound 1b Is a Synthon for Additional 2-
Substituted Cordycepin Analogues. In addition to the
interesting trypanocidal properties of 1b, it was discovered that
1b itself can be used as a useful synthon for the synthesis of
additional 2-substituted cordycepin analogues. Conventional
heating of 1b with excess hydroxylamine in dioxane for 24 h
gave quantitative conversion to the corresponding 2-(2-
hydroxyethylamino)cordycepin 11 by nucleophilic aromatics
substitution of the C2-fluorine (Scheme 2). Compound 11
showed no trypanotoxic activity in our assays, presumably
because of lower affinity for TbAT1.
In Vitro Evaluation of C2-Substituted Cordycepin
Analogues. Compound 1b displayed an in vitro trypanocidal
IC50 of 174 nM, i.e., 3- to 6-fold lower activity than 1a. The
observed trypanotoxicity did not depend on the halogen
modification itself, since 2-fluoroadenosine 8 and 2-fluoroade-
nine 9 proved to be nontoxic for the parasites (see Table 1).
E
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
of concern for the dependence of cordycepin analogues on the
P2 transporter.43
Compound 1b Is Resistant to ADA Metabolism in
Vitro. The in vitro conversion of 3′-deoxyadenosine to 3′-
deoxyinosine can be monitored spectrophotometrically, as they
absorb at different wavelengths.44 The presence of a 2-fluoro
substituent does not dramatically affect these absorbances.45
Compounds 1a and 1b were thus incubated with recombinant
ADA, and deamination of the compounds was monitored.
Compound 1a alone was readily metabolized by ADA. This
effect could be fully inhibited by the addition of 10 μM Dcf. In
contrast, 1b showed significantly attenuated ADA-mediated
deamination in the absence or presence of Dcf (Figure 6).
MOLT4 cells express high levels of ADA; thus, 1a is heavily
metabolized by these cells. We next evaluated the cytotoxicity
of 1b in MOLT4 cells as an additional model to evaluate its
susceptibility to ADA-mediated degradation. Compound 1b
alone or in the presence of Dcf on MOLT4 cells displayed
Figure 5. Uptake of 0.05 μM [3H]adenosine by bloodstream form T.
b. brucei of the B48/TbAT1 strain in the presence of various
concentrations of inhibitor as indicated. Transport was expressed as a
percent of control incubations without inhibitor. Data shown are the
average and SEM of triplicate determinations within a representative
experiment. Each inhibitor was tested in triplicate on at least three
independent occasions.
similar in vitro IC50 of 2.44
0.69 and 2.13
0.87 μM,
Table 2. Relative Affinity of Evaluated Analogues toward the
T. brucei Adenosine Transporter TbAT1/P2 Relative to
Native Substrates
confirming resistance against deamination. In contrast,
compound 1a was highly toxic for MOLT4 cells only when
co-incubated with Dcf (IC50 = 0.10 0.03 μM) versus 1a alone
(IC50 ≥ 100 μM) (Figure 6E,F). The IC50 of 1a for T. brucei
was 42 nM in absence and 1.2 nM in presence of 1.5 mM Dcf
(Figure 6G). In contrast, the toxicity of 1b for trypanosomes
was unaffected by the presence of Dcf (IC50 of 170 25 nM in
presence of Dcf or 176 23 nM in absence of Dcf) (Figure 6
H).
The Toxicity of 1b Is Dependent on Adenosine Kinase
(AK). We next studied if the activity of cordycepin analogues is
dependent on intracellular phosphorylation through AK by
evaluating the effects of the AK inhibitor ABT702 on the
trypanocidal activity of our compounds. The toxicity of 1a and
1b for T. b. brucei was significantly attenuated in the presence of
the inhibitor (Figure 7A), suggesting that conversion of 1a and
1b to their corresponding monophosphate is critical to elicit
activity.
In Vitro Preclinical and in Vivo Pharmacokinetics
Evaluation of 1b Suggests Good Preclinical Properties.
Microsomal metabolic stability (human, mouse), Caco-2 cell
membrane permeability, human plasma protein binding, and
stability of 1b were measured in vitro to give initial estimates of
the oral drug suitability. The apparent permeability (Papp) in the
apical to basolateral direction was determined to 3.0 × 10−6
cm/s (mass recovery 102%), which is in the region of moderate
to high permeability according to our in-house classification.46
The Papp in the basolateral to apical direction was determined to
1.4 × 10−6 cm/s (mass recovery 104%) and an efflux ratio of
0.5 calculated after assessing the transport in both directions
indicated apparent active transport in the absorptive direction.
No metabolic degradation was detected with either human or
mouse liver microsomes (HLM, MLM), indicating a very high
stability toward primarily oxidative metabolism. The plasma
protein binding was nonexistent as measured by equilibrium
dialysis (fraction unbound, 100 4%) which indicates that the
drug is freely available to tissues in systemic circulation. The in
vitro plasma stability over 4 h was 98%.
Km or Ki (μM)
δ(ΔG0)
a
compd
mean
SE
n
(kJ/mol)
relative to
b
adenosine
0.92
2.71
9.65
0.28
0.9
0.06
0.35
3.42
0.02
0.2
3
3
3
4
3
3
3
3
3
0.00
2.70
adenosine
adenosine
adenosine
adenosine
adenine
2-F-adenosine
2-Cl-adenosine
adenine
5.85
−2.92
2.97
2-F-adenine
2-Cl-adenine
2.6
0.3
5.55
adenine
cordycepin (1a)
2-F-cordycepin (1b)
2-Cl-cordycepin (10)
0.18
0.96
3.97
0.02
0.16
0.71
−4.02
4.15
adenosine
cordycepin
cordycepin
7.67
a
The Gibbs free energy ΔG0 of adenosine was −34.5 kJ/mol,
calculated from ΔG0 = −RT ln(Km), in which R is the gas constant and
T is the absolute temperature. The energy gain or loss (δ(ΔG0))
relative to adenosine is indicated. For adenine analogues the δ(ΔG0)
relative to adenine and for cordycepin analogues relative to cordycepin
is listed in order to highlight the effect of the modification. SE =
b
standard error. Km value. All other values are Ki.
binding of 2-chloro substitution was 5.5−7.7 kJ/mol reduced in
relation to the unsubstituted compounds, i.e., unfavorable by
approximately 3 kJ/mol when compared to the 2-fluoro
substitution. The observation that for each of the aminopurine
substrates, cordycepin, adenosine, and adenine, a similar
amount of binding energy is lost on the substitution with
fluorine or chlorine at position 2 strongly supports the view
that all three compounds bind in the same orientation in the
binding pocket and that a chlorine at C2 results in steric
interference with transport.41
As shown in Table 1, the trypanocidal activity of compound
1a diminished in 1b and was absent in compound 10. It seems
unlikely that the relatively small loss of affinity for the P2
transporter between 1a and 1b (Ki values of 0.18 0.02 and
The in vivo pharmacokinetics of 1b were next evaluated.
Mice were administered intraperitonally or orally with 25 mg/
kg 1b. Plasma samples were collected over 24 h (Figure 7B).
Bioanalysis revealed that despite being resistant to ADA and
showing good metabolic stability, 1b has a relatively short half-
life (t1/2 ≈ 1 h) and only covers the in vitro EC50 for 3−4 h.
0.96
0.16 μM, respectively) is sufficient to explain the
difference in trypanocidal potency, as the Ki value for 1b still
represents a very high affinity for a nucleoside transporter and
is statistically identical to the Michaelis−Menten constant for
adenosine (Km = 0.92 0.06 μM). The emergence of clinical
strains with resistant TbAT1 alleles constitutes potential issue
F
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 6. Compound 1b is resistant to adenosine deaminase metabolism in vitro. Cordycepin (1a, UVmax = 264 nm, arrow down) in the presence of
ADA is gradually converted to the corresponding inosine (UVmax = 254 nm, arrow up) over 10 h (A). Compound 1b shows dramatically attenuated
ADA-mediated degradation (B). Addition of ADA inhibitor deoxycoformycin (Dcf) abrogates effect of ADA on 1a (C), while 1b is unaffected (D).
Curves illustrate UV absorbance from 230 to 300 nm measured at 2 nm intervals over 10 h. Each data point indicates average UV absorbance of
○
▲
experimental triplicates. Cytotoxicity of 1a (E, G) and 1b (F, H) with ( ) and without ( ) coadministration of 1.5 μM Dcf on ADA-rich MOLT4
cells (E, F) and T. brucei (G, H) shows high dependence on ADA activity (upper graph). The mean SEM cell viability was determined in triplicate
cultures by the WST-1 assays 72 h after incubation with the nucleosides. A representative of one of at least three independent experiments is shown.
Overall the PK parameters for 1b were similar to those
previously reported for cordycepin where the main clearance
mechanism was attributed to ADA.27 For 1b however, clearance
mechanisms other than ADA or oxidative liver metabolism are
determining its fate. The relative bioavailability (po/ip ≈ 0.6)
suggests extensive intestinal elimination of current unknown
origin. In addition, the low protein binding could suggest that
renal filtration could be a major pathway.
Treatment of Mice with 1b Cures Infection with T. b.
brucei. Groups of mice were next treated ip with 1 mg/kg (10
G
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
■
●
Figure 7. (A) Trypanocidal activity of compounds 1a (left) and 1b (right) in vitro in the absence ( ) or presence ( ) of the adenosine kinase (AK)
inhibitor ABT702. Inhibition of AK dramatically reduces the in vitro efficacy of both compounds, revealing a critical dependence on intracellular
kinase activity for compounds to elicit activity. (B) In vivo pharmacokinetics of 1b in mice as determined through oral (po, 25 mg/kg) or
intraperitoneal (ip, 25 mg/kg) show a relatively short in vivo half-life in plasma.
doses, OD), 5 mg/kg (5 doses, OD), 15 mg/kg (4 doses, OD),
or 30 mg/kg 1b (6 doses, OD) starting 2 h after ip infection
with T. b. brucei (a rodent infecting subspecies of T. brucei).
Treatment with 10 doses of 1 mg/kg 1b decreased parasitemia
levels but did not clear the infection, and mice quickly relapsed
after dosing ended. Mice treated with either the 5 or 15 mg/kg
doses of 1b showed no detectable parasitemia for 7 days after
treatment, but eventually, all mice in this group relapsed.
Coadministration of 0.2 mg/kg Dcf to the 5 mg/kg cohort did
not improve the results in these dosing regimens, indicating
that ADA metabolism is not a contributing factor to the effects
of 1b. However, a complete clearance of parasites as assessed by
lack of detectable parasitemia for more than 90 days was
achieved in mice treated with six doses of 30 mg/kg 1b (Table
3).
CONCLUSION
■
In summary, we have evaluated the in vitro and in vivo
pharmacological effects of 2-fluorocordycepin 1b and related
analogues. This work resulted in an improved understanding of
the exquisitely delicate SAR of synthetic nucleosides as
trypanocidal compounds. On the basis of the effects of
structural modification of 1a, rational design led to the
identification of 2-fluorocordycepin (1b, Figure 1A) as an
ADA-resistant cordycepin analogue that retains trypanocidal
activity and selectivity. A novel route toward the synthesis of 1b
and other 2-halogenated cordycepin analogues was developed,
and subsequent chemistry demonstrates that 1b is a useful
synthetic intermediate allowing for the preparation of addi-
tional 2-substituted cordycepin analogues. Predictive pharma-
cokinetic assays suggest that 1b has a good in vitro preclinical
profile, although in vivo pharmacokinetic evaluation shows a
relatively short half-life. Oral administration of 1b at 30 mg/kg
once daily cured an acute in vivo model of trypanosomiasis in
mice, thus implicating 2-substituted cordycepin analogues as
promising candidates for the development of novel therapies
for HAT.
Table 3. Oral Dosing of 30 mg/kg 1b Cures Mice in Acute
Murine Model of HAT
a
dose (mg/kg)
dosing
day of relapse (after treatment)
EXPERIMENTAL SECTION
■
1
10 doses b.i.d.
5 doses OD
5 doses OD
5 doses OD
6 doses OD
2
Chemistry: General. Cordycepin (catalog no. PR 3430) and 5-O-
benzoyl-1,2-di-O-acetyl-3-deoxy-D-ribose (catalog no. CR 2010) were
purchased from Berry & Associates, Inc., Dexter, MI, USA. 2-
Fluoroadenine (catalog no. UN2811-3) was purchased from SynQuest
Laboratories, Inc., Alachua, FL, USA. Nebularine (catalog no.
N388500) was purchased from Toronto Research Chemicals,
Toronto, Canada. N-Octanoyl cordycepin27 was a kindly donated by
5
7
5 (+0.2 Dcf)
5
15
30
7
no relapse
a
Number of mice per dosing group is two to four.
H
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Dr. X. G. Li , School of Pharmaceutical Science, Southwest University,
China. All other starting materials and reagents were purchased from
Sigma Aldrich and used without further purification unless otherwise
noted. All evaporations were carried out in vacuo with a rotary
evaporator at 10−30 mmHg. Analytical samples were dried under high
vacuum (1−5 mmHg) at room temperature. Thin layer chromatog-
raphy (TLC) was performed on silica gel plates (Merck) with
fluorescent indicator. Spots were visualized by UV light (214 and 254
nm). Flash chromatography was performed using silica gel 60 from
Merck. 1H and 13C NMR spectra were recorded using a Bruker
DPX400 spectrometer (400 MHz) using deuterated solvents. 1H
chemical shifts are reported as δ values in ppm with the deuterated
solvent as an internal standard. Data are reported as follows: chemical
shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, br =
broad, m = multiplet), coupling constant (Hz), integration. All
compounds evaluated in biological tests were purified to ≥95% as
determined by reverse-phase high-pressure liquid chromatography
coupled mass spectrometry (HPLC−MS) on an Agilent/HP 1200
systems 6110 mass spectrometer with electrospray ionization (ESI+).
HPLC−MS methods were the following. Method 1: Waters XBridge
C18 3.5 μm column (3.0 mm × 50 mm), 3.5 min gradient, mobile
phase [CH3CN]/[10 mM NH4HCO3/H2O]. Method 2: ACE C18 3.5
μm column (3.0 mm × 50 mm), mobile phase [0.1% TFA/CH3CN]/
[0.1% TFA/H2O]. Absorbance was monitored at 305 90 and 254
nm. All solvents used were HPLC grade.
2.96 (m, 1H), 4.62 (dd, 1H), 7.73 (dd, 1H), 4.80−4.93 (m, 1H), 5.95
(d, 1H), 6.20 (s, 1H), 7.35−7.70 (overlapping tripletts, 12H), 7.87 (d,
4H), 8.03 (d, 2H), 8.07 (d, 2H), and 8.19 (s, 1H). ESI-MS (m/z): 686
[M + H]+.
(2R, 3R, 5S)-2-(6-Amino-2-fluoro-9H-purin-9-yl)-5-
(hydroxymethyl)oxolan-3-ol (1b). 14 (29 mg, 0.042 mmol) was
suspended as a slurry in dry methanol (5 mL) and dry ammonia gas
was bubbled through the mixture, resulting in a clear solution within
15 min. The mixture was left to stir overnight and was then
concentrated in vacuo and purified through flash chromatography with
an ethyl acetate/methanol gradient 100/0 to 84/16 to give 6.5 mg
1
(59%) of 1b. H NMR (DMSO-d6) δ 1.90−1.95 (m, 1H), 2.20−2.27
(m, 1H), 3.50−3.58 (m, 1H), 3.67−3.75 (m, 1H), 4.30−4.40 (m, 1H),
4.54 (m or br s, 1H), 5.04 (t, 1H), 5.70 (br s, 1H), 5.80 (br s, 1H),
7.84 (br s, 2H), and 8.37 (s, 1H). ESI-MS (m/z): 270 [M + H]+.
Synthesis of 1b: Method B. [(2S,4R,5R)-4-(Acetyloxy)-5-(6-
amino-2-fluoro-9H-purin-9-yl)oxolan-2-yl]methyl Benzoate (15).
2-Fluoro-9H-purin-6-amine (300 mg, 1.96 mmol) was suspended in
10 mL of anhydrous acetonitrile, and N,O-bis(trimethylsilyl)acetamide
(632 mg, 2.36 mmol) was added in one portion. The mixture was
flushed with nitrogen, sealed, and heated to 80 °C for 1 h or until
homogeneous. If not homogeneous after 1 h, N,O-bis(trimethylsilyl)-
acetamide was added dropwise until the solution became clear. The
mixture was then cooled to 0 °C, and 5-O-benzoyl-1,2-di-O-acetyl-3-
deoxy-D-ribose (632 mg, 1.96 mmol), dissolved in 3 mL of acetonitrile,
was added in one portion, followed by the dropwise addition of
trimethylsilyl triflate (523 mg, 2.35 mmol) at 0 °C while stirring. After
addition, the reaction was sealed and the mixture was heated to 80 °C
for 1 h. The mixture was cooled to room temperature and quenched
by pouring into saturated aqueous sodium bicarbonate solution. The
organic crude was extracted with three 30 mL portions of ethyl acetate.
Combined organic fractions were washed with two 20 mL portions
water and one 20 mL portion of brine, dried over anhydrous
magnesium sulfate, filtered, and concentrated in vacuo to yield 15 (804
mg, 98% yield) as an off-white solid which was used without further
purification or characterization.
Synthesis of 1b: Method A. (2R,3R,5S)-2-[6-(N-Benzoylbenza-
mido)-9H-purin-9-yl]-5-[(benzoyloxy)methyl]oxolan-3-yl Benzoate
(12). Cordycepin 1a (70 mg, 0.28 mmol) was dissolved in pyridine
(5 mL). Benzoyl chloride (392 mg, 2.79 mmol) was added dropwise
under stirring, and the mixture was then heated in an oil bath at 125
°C for 3 h before concentrating the reaction mixture in vacuo. The
crude was dissolved in ethyl acetate and washed with aqueous sodium
bicarbonate, water, and brine. The organic fraction was dried over
anhydrous sodium sulfate and concentrated in vacuo. The crude
material was purified through flash chromatography on silica using a
gradient of pentane/ethyl acetate 95:5 to 0:100 to afford 97 mg of the
1
analytically pure product (yield 47%). H NMR (CDCl3) δ 2.51 (dd,
(2R, 3R, 5S)-2-(6-Amino-2-fluoro-9H-purin-9-yl)-5-
(hydroxymethyl)oxolan-3-ol (1b). [(2S,4R,5R)-4-(Acetyloxy)-5-(6-
amino-2-fluoro-9H-purin-9-yl)oxolan-2-yl]methyl benzoate (15, 300
mg, 0.72 mmol) was dissolved in 25 mL of fresh 7 N methanolic
ammonia. The reaction was sealed carefully and the mixture was
allowed to stir for 24 h at room temperature. The resulting clear
solution was evaporated in vacuo, and the resulting solid material was
recrystallized from boiling water over 48 h (100 °C → room
temperature, 24 h; room temperature → 5 °C, 24 h). The resulting
small off-white crystals were collected by vacuum filtration and washed
with two portions of ice-cold methanol and two portions of diethyl
ether followed by drying under vacuum to yield analytically pure 1b
1H), 2.99 (ddd, 1H), 4.64 (dd, 1H), 4.74 (dd, 1H), 4.87−4.96 (m,
1H), 6.07 (d, 1H), 6.30 (s, 1H), 7.35−7.68 (overlapping phenyl peaks,
12H), 7.89 (d, 4H), 8.02 (d, 2H), 8.07−8.13 (distorted d, 2H), 8.28
(s, 1H) and 8.60 (s, 1H). ESI-MS (m/z): 668 [M + H]+.
(2R,3R,5S)-2-[6-(N-Benzoylbenzamido)-2-nitro-9H-purin-9-yl]-5-
[(benzoyloxy)methyl]oxolan-3-yl Benzoate (13). A solution of
tetrabutylammonium nitrate (TBAN) (66 mg, 0.22 mmol) in
dichloromethane (2 mL) was cooled in an ice/water bath for 15
min before dropwise addition of trifluoroacetic anhydride (TFAA) (45
mg, 0.21 mmol), and the mixture was stirred for 1 h. A solution of 12
(96 mg, 0.14 mmol) in dichloromethane (2 mL) was cooled before
dropwise addition of the TBAN/TFAA solution. The reaction mixture
was stirred cooled for 2 h before pouring it into an aqueous sodium
bicarbonate (3%, 8 mL) solution. The organic phase was separated and
the water phase extracted with two portions of dichloromethane.
Combined organic fractions were dried over anhydrous sodium sulfate
and concentrated to yield an oily residue that was purified via flash
chromatography on silica gel, eluting with dichloromethane followed
1
(96.5 mg, 50% yield). H NMR (400 MHz, DMSO-d6) δ ppm 1.85−
1.93 (m, 1H), 2.17−2.26 (m, 1H), 3.49−3.56 (m, 1H), 3.65−3.72 (m,
1H), 4.31−4.38 (m, 1H), 4.49−4.55 (m, 1H), 5.01 (t, J = 5.52 Hz,
1H), 5.67 (d, J = 4.02 Hz, 1H), 5.77 (d, J = 2.01 Hz, 1H), 7.82 (br s,
1H), 8.34 (s, 1H). 13C NMR (101 MHz, DMSO-d6) δ 159.5, 157.6,
150.0, 139.4, 117.4, 90.7, 80.9, 74.8, 62.3, 33.9. ESI-MS (m/z): 270 [M
+ H]+.
1
by dichloromethane with 1% of acetone to get 49 mg of product. H
Synthesis of 3′-Deoxynebularine (4). ((2S,4R,5R)-4-Acetoxy-5-
(6-amino-9H-purin-9-yl)tetrahydrofuran-2-yl)methyl Acetate. 1a
(200 mg, 0.80 mmol) was dissolved in pyridine (5 mL), and acetic
anhydride (1.02 g, 1 mmol) was added. The mixture was stirred at
room temperature for 30 min. The mixture was concentrated in vacuo
and the crude purified by flash column chromatography on silica gel
using ethyl acetate/methanol (100:0 to 90:10) as mobile phase. Yield:
220 mg (82%). ESI-MS (m/z): 336 [M + H]+. The material was used
in the next step without further characterization.
(2R,3R,5S)-5-(Hydroxymethyl)-2-(9H-purin-9-yl)tetrahydrofuran-
3-ol (4). 4 was prepared according to the procedure by Francom et
al.47 ((2S,4R,5R)-4-Acetoxy-5-(6-amino-9H-purin-9-yl)-
tetrahydrofuran-2-yl)methyl acetate (220 mg, 0.66 mmol) was
dissolved in dry THF (10 mL), and tert-butyl nitrite (1.35 g, 13.1
mmol) was added in one portion. The mixture was heated to 60 °C for
NMR (CDCl3) δ 2.56 (dd, 1H), 3.03−3.14 (m, 1H), 4.70 (dd, 1H),
4.80 (dd, 1H), 4.90−4.99 (m, 1H), 5.90 (d, 1H), 6.30 (d, 1H), 7.37−
7.59 (overlapping d and t, 14 H), 7.64 (t, 1H), 7.88 (d, 4H), 8.00 (d,
2H), 8.07 (d, 2 H), and 8.44 (s, 1H). ESI-MS (m/z): 713 [M + H]+.
(2R,3R,5S)-2-[6-(N-Benzoylbenzamido)-2-fluoro-9H-purin-9-yl]-5-
[(benzoyloxy)methyl]oxolan-3-yl Benzoate (14). A solution of 13
(47.7 mg; 0.067 mmol) in acetonitrile was cooled on an ice/water
bath. Tetrabutylammonium fluoride (TBAF) (23.0 mg, 0.088 mmol; 1
M in THF) was added, and the reaction mixture was stirred at 0 °C for
40 min and then slowly raised to room temperature. Concentration of
the reaction mixture by rotary evaporation at room temperature
resulted in a crude residue which was purified by flash chromatography
with a dichloromethane/acetone gradient from 100/0 to 95/5 to yield
1
29 mg (59%) of product. H NMR (CDCl3) δ 2.50 (dd, 1H), 2.88−
I
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
30 min until complete by analytical HPLC−MS. The crude material
was concentrated in vacuo and directly treated with 7 N NH3 in
methanol at room temperature for 1 h. The material was purified by
preparative HPLC (method 1) to yield the product as a light solid.
Yield: 81.0 mg (52%). 1H NMR (400 MHz, CH3OD) δ ppm 2.06 (m,
J = 6.71, 6.71, 2.64 Hz, 1 H) 2.40 (ddd, J = 13.55, 9.29, 5.52 Hz, 1 H),
3.70 (dd, J = 12.42, 3.64 Hz, 1 H), 3.95 (dd, J = 12.30, 2.76 Hz, 1 H),
4.53−4.61 (m, 1 H), 4.74−4.81 (m, 1 H), 6.17 (d, J = 2.01 Hz, 1 H),
8.89 (s, 1 H), 8.95 (s, 1 H), 9.10 (s, 1 H). ESI-MS (m/z): 237 [M +
H]+.
methyl benzoate (235 mg, 0.54 mmol) was dissolved in 20 mL of fresh
7 N methanolic ammonia and the sealed reaction allowed to stir at
room temperature for 24 h. The resulting clear solution was
evaporated in vacuo, and the resulting solid material was recrystallized
from boiling water over 48 h (100 °C → room temperature, 24 h;
room temperature → 5 °C, 24 h). The resulting white crystals were
collected by vacuum filtration and washed with two portions of ice-
cold methanol and two portions of diethyl ether followed by drying
1
under vacuum to yield analytically pure 10 in 38% yield. H NMR
(400 MHz, DMSO-d6) δ ppm 1.90 (ddd, J = 13.07, 6.25, 2.91 Hz, 1
H), 2.22 (ddd, J = 13.20, 8.91, 5.68 Hz, 1 H), 3.52 (ddd, J = 11.94,
5.49, 4.04 Hz, 1 H), 3.69 (ddd, J = 12.13, 5.43, 3.41 Hz, 1 H), 4.31−
4.39 (m, 1 H), 4.49−4.55 (m, 1 H), 5.02 (t, J = 5.56 Hz, 1 H), 5.68 (d,
J = 4.29 Hz, 1 H), 5.80 (d, J = 2.27 Hz, 1 H), 7.81 (br s, 2 H), 8.38 (s,
1 H). 13C NMR (101 MHz, DMSO-d6) δ = 156.7, 152.9, 149.9, 139.4,
118.0, 90.6, 80.9, 74.8, 62.3, 33.8. ESI-MS (m/z): 286 [M + H]+.
(2R,3R,5S)-2-(6-Amino-2-((2-hydroxyethyl)amino)-9H-purin-9-yl)-
5-(hydroxymethyl)tetrahydrofuran-3-ol (11). (2R,3R,5S)-2-(6-
Amino-2-fluoro-9H-purin-9-yl)-5-(hydroxymethyl)oxolan-3-ol (1b)
(20.0 mg, 0.07 mmol) was dissolved in dioxane (2 mL), and
ethanolamine (23.0 mg, 0.37 mmol) was added. The mixture was
heated to 100 °C for 20 h in a sealed stem block tube. The crude
material was concentrated in vacuo and purified by preparative HPLC
(method 1) to yield the product as a clear waxy solid. Yield: 18.0 mg
Synthesis of 2-Fluoro-3′-deoxyadenine Arabinoside (6).48
((2R,3R,4R,5R)-5-(6-Amino-2-fluoro-9H-purin-9-yl)-3-
((methylsulfonyl)oxy)-4-(pivaloyloxy)tetrahydrofuran-2-yl)methyl
Pivalate. Fluoroadenosine (285 mg, 1.00 mmol) was dissolved in
pyridine (20 mL) and cooled to −15 °C. Pivaloyl chloride (376 mg,
3.12 mmol) was added in one portion and the mixture allowed to stir
for 2 h at 0 °C. Methanesulfonyl chloride (252 mg, 2.20 mmol) was
added. The mixture was allowed to rise to room temperature and was
stirred at room temperature for 3 h. The reaction was quenched with
ice−water and the product extracted twice with ethyl acetate.
Combined organic fractions were washed with saturated sodium
bicarbonate, brine and dried over anhydrous MgSO4. The solution was
filtered and concentrated in vacuo. Residual pyridine was removed by
coevaporation with toluene. The crude material was purified by
column chromatography on silica gel using dichloromethane/
methanol (100:0 to 95:5). Yield: 417 mg (78%). 1H NMR (400
MHz, CDCl3) δ ppm 1.18 (s, 9 H), 1.25 (s, 9 H), 1.26 (s, 9 H), 3.16
(s, 3 H), 4.34−4.48 (m, 2 H), 4.52−4.58 (m, 1 H), 5.69 (t, J = 5.52
Hz, 2 H), 5.92 (dd, J = 5.40, 4.39 Hz, 1 H), 6.05 (d, J = 4.52 Hz, 1 H),
7.90 (s, 1 H), 8.63 (br s, 2 H). ESI-MS (m/z): 532 [M + H]+.
(2R, 3S, 5S)-2-(6-Amino-2-fluoro-9H-purin-9-yl)-5-
(hydroxymethyl)tetrahydrofuran-3-ol (6). ((2R,3R,4R,5R)-5-(6-
Amino-2-fluoro-9H-purin-9-yl)-3-((methylsulfonyl)oxy)-4-
(pivaloyloxy)tetrahydrofuran-2-yl)methyl pivalate (417 mg; 0.78
mmol) was dissolved in anhydrous DMSO/THF (1:1, 5 mL) and
cooled to 0 °C. Lithium triethylborohydride (1 M in THF; 4 mL, 4.00
mmol) was added dropwise and the mixture stirred at room
temperature for 20 h. The reaction was quenched with 1 mL of 5%
acetic acid (aqueous) and treated with tetrabutylammonium hydroxide
(2.60 g, 10.0 mmol). The mixture was allowed to stir for 3 h, then
concentrated in vacuo and the product isolated by column
chromatography on silica gel using methanol (0−20%) in ethyl
acetate as gradient to yield the desired product. Yield: 83 mg (39%).
1H NMR (400 MHz, DMSO-d6) δ ppm 1.98 (dt, J = 12.80, 8.16 Hz, 1
1
(78%). H NMR (400 MHz, DMSO-d6) δ ppm 1.86−1.95 (m, 1 H),
2.21−2.35 (m, 1 H), 3.27−3.35 (m, 2 H), 3.49 (m, J = 4.55 Hz, 1 H),
3.44−3.54 (m, 3 H), 3.57−3.65 (m, 1 H), 4.24−4.32 (m, 1 H), 4.54−
4.60 (m, 1 H), 5.69 (d, J = 2.53 Hz, 1 H), 6.07 (t, J = 1.00 Hz, 1 H),
7.89 (s, 1 H). 13C NMR (101 MHz, DMSO-d6) δ ppm 159.4, 156.0,
151.1, 135.9, 113.4, 90.3, 80.2, 74.3, 63.1, 60.3, 43.9, 40.0. ESI-MS (m/
z): 311 [M + H]+.
Ab Initio Calculation of pKa Values. All calculations in Figure 5
were performed using ChemAxon software (version 6.0) and
Schrodinger Suite 2012 (Schrodinger Suite 2012, Maestro, version 9.3;
̈
̈
Schrodinger, LLC: New York, NY, 2012.) using the Maestro interface
̈
for the Microsoft Windows 7 platform. pKa calculations were
performed using Jaguar (version 7.9). This is a quantum chemical
method (density functional theory) with empirical correction. The
workflow in Jaguar was used starting with a conformational serach
using MacroModel, version 9.9, before pKa calculations. Jaguar
calculates microscopic (atomic) pKa values. The pKa values for the
three nitrogens N1, N3, and N7 in all three compounds, differing in
the C2 substituent H, F, or Cl, were calculated. Default settings with
water as solvent were used.
H), 2.25 (dt, J = 12.74, 6.56 Hz, 1 H), 3.53−3.70 (m, 2 H), 4.07 (dd, J
= 4.14, 1.88 Hz, 1 H), 4.46−4.56 (m, 1 H), 5.11 (t, J = 4.77 Hz, 1 H),
5.41 (d, J = 4.77 Hz, 1 H), 6.00 (d, J = 5.52 Hz, 1 H), 7.79 (br s, 2 H),
8.28 (s, 1 H). ESI-MS (m/z): 270 [M + H]+.
Parasite and Mammalian Cell Cultures. T. b. brucei
(AnTat1.1E) (an animal-infecting subspecies of T. brucei), T. b.
rhodesiense (STIB 851), and T. b. gambiense (MBA)29,32 were kindly
provided by P. Buscher (Institute of Tropical Medicine, Antwerp,
̈
Synthesis of 2-Chlorocordycepin (10) and 2-(2-
Hydroxyethylamino)cordycepin (11). ((2S,4R,5R)-4-Acetoxy-5-
(6-amino-2-chloro-9H-purin-9-yl)tetrahydrofuran-2-yl)methyl Ben-
zoate (16). 2-Chloroadenine (100 mg; 0.59 mmol) was suspended in
3 mL of anhydrous CH3CN in a small stem block tube, and N,O-
bis(trimethylsilyl)acetamide (180 mg, 0.89 mmol) was added. The
reaction tube was flushed with nitrogen and sealed. The mixture was
heated at 80 °C while stirring vigorously until homogeneous. The
mixture was then cooled to 0 °C, and [(2S,4R,5S)-4,5-bis(acetyloxy)-
oxolan-2-yl]methyl benzoate (190 mg, 0.59 mmol) dissolved in 2 mL
of CH3CN was added in one portion. The mixture was stirred for 5
min, followed by the dropwise addition of trimethylsilyl triflate (157
mg, 0.71 mmol, 128 μL). The mixture was heated to 80 °C for 1 h.
The mixture was cooled to room temperature, quenched by pouring
into saturated NaHCO3, and extracted with 3 × 30 mL of ethyl
acetate. The combined organic fractions were washed with water, brine
and dried over MgSO4. The crude was evaporated in vacuo to yield
235 mg (93%) of the desired product as an off-white solid. The
material was used directly without further characterization or
purification.
Belgium), were cultured at 37 °C and 5% CO2 in HMI-9 medium
Iscove’s modified Dulbecco’s medium (IMDM, Hyclone) containing
10% heat-inactivated calf serum (Invitrogen, Carlsbad, CA), 28 mM 4-
(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.14%
glucose, 1.5% sodium bicarbonate, 2 mM L-glutamine, 0.14 mg/mL
gentamicin, 0.3 mM dithiothreitol (DTT), 1.4 mM sodium pyruvate,
0.7 mM L-cysteine, 28 μM adenosine, 14 μM guanosine, 100 units/mL
penicillin, and 0.1 mg/mL streptomycin (Hyclone, catalog no.
SV30010). T. brucei forms were harvested by centrifugation at 2500
rpm for 6 min followed by resuspension in fresh medium. Parasites
were utilized when in the log phase of growth.
Human MOLT-4 (ATCC-CRL-1582) and mouse fibroblast L-929
(ATCC-CCL-1) cells were cultured at 37 °C and 5% CO2 in Roswell
Park Memorial Institute medium (RPMI, Hyclone) supplemented
with 10% fetal calf serum (FCS), 100 units/mL penicillin, and 0.1 mg/
mL streptomycin. Cells were harvested by centrifugation at 1500 rpm
for 6 min followed by resuspension in fresh medium.
WST-1 in Vitro Cell Viability Assay. Compounds evaluated were
prepared as 10 mM stock solutions in dimethylsulfoxide (DMSO) and
diluted to the appropriate concentrations with IMDM and 10% FCS
(Hyclone) as previously described.49 For dose−response evaluation,
compound stocks were initially serial-diluted with 100% DMSO in 11
(2R,3R,5S)-2-(6-Amino-2-chloro-9H-purin-9-yl)-5-
(hydroxymethyl)tetrahydrofuran-3-ol (10). Crude ((2S,4R,5R)-4-
acetoxy-5-(6-amino-2-chloro-9H-purin-9-yl)tetrahydrofuran-2-yl)-
J
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
steps from 10 mM to 0.17 μM final concentrations prior to dilution.
Diluted compounds were then transferred to 96-well assay ready
plates. Maximum final DMSO concentration in each assay was 0.25%.
Suspensions of trypanosomes (for T. b. brucei, T. b. gambiense, and
T. b. rhodesiense: 6000 bloodstream T. b. forms/well) in IMDM and
10% FCS (Hyclone) were next added to each well using a multipipette
and plates were incubated at 37 °C and 5% CO2 for 72 h. DMSO
(0.25% final concentration) and 3′-deoxyadenosine (10 μM final
concentration) were used as negative and positive controls,
respectively.
using the spiked frozen plasma sample. Caco-2 samples were
quantified using an external standard curve between 10 and 1000
nM. Analysis of metabolic stability does not require a standard curve,
since the individual time points are analyzed on a relative basis.
Human plasma protein binding and stability of 1b was performed
using rapid equilibrium dialysis (rapid equilibrium dialysis (RED)
device inserts, Thermo Scientific), allowing for short dialysis times (2−
4 h) compared to traditional methods (>8 h) and minimal drug
consumption in a 96-well format. Pooled human plasma was provided
by Uppsala Academic Hospital and was collected from two male and
two female donors (nonsmoking). In brief, 0.2 mL of the plasma test
solution, spiked with 10 μM 1b (final compound concentration), was
dialyzed against 0.35 mL of isotonic phosphate buffer, pH 7.4, for 4 h
on a Kisker rotational incubator at 37 °C (900 rpm) to achieve
equilibrium.54 A stability test of the test solution was also prepared (to
allow detection of drug degradation), which was incubated separately
on a sealed plate at 37 °C for 4 h. After the incubation, the contents of
the dialyzed plasma, plasma stability test, and the dialyzed buffer
compartment were removed and immediately frozen until analysis.
The samples were analyzed by LC−MS/MS as described above.
Microsomal metabolic stability of 1b used pooled human or male
CD-1 mouse liver microsomes (Xenotech LLC, USA) with
supplemented cofactor (NADPH) to primarily facilitate CYP450
oxidative activity against the target compound. 1b (1 μM incubation
concentration) and microsomes (0.5 mg/mL incubation concen-
tration) were diluted in 0.1 M phosphate buffer, pH 7.4, and the
reaction was initiated with addition of NADPH (1 mM incubation
concentration). Then 0 min control samples were precipitated prior to
addition of NADPH. Midazolam and dextromethorphan were used as
positive controls of cytochrome P450 metabolism (CYP3A4,
CYP2D6) in separate wells on the same plate. The incubation times
were 0, 5, 15, 40 min, and the reaction was quenched by addition of
ice-cold acetonitrile (40% v/v final concentration) containing warfarin
as analytical internal standard. The plate was then sealed, centrifuged,
and frozen at −20 °C until LC−MS/MS analysis. The intrinsic
clearance was calculated using the well-stirred model.55
For mammalian cell cytotoxicity, 90 μL of a suspension of 104
MOLT4 cells or 5 × 103 L929 cells were seeded in 96-well transparent
plates containing compound solutions prepared as indicated above.
Plates were incubated at 37 °C and 5% CO2 for 72 h.
For detection of cell viability in all of the above-described assays, 10
μL of WST-1 reagent (Roche) was then added to each well and the
plates were incubated for 2 h at 37 °C and 5% CO2. Absorbance was
recorded from the plates using a multiwell spectrophotometer Victor2
(PerkinElmer) at a wavelength of 450 nm. For dose−response
measurements, EC50 values and standard error were calculated by
fitting resulting data as the log of inhibitor concentration versus
normalized response with nonlinear regression analysis (variable
slope) using GraphPad Prism software.
Transporter Studies. Affinity for the TbAT1/P2 transporter was
determined using culture-derived bloodstream trypanosomes of the
B48 clone of T. b. brucei,50 manipulated to overexpress the TbAT1
gene. The endogenous TbAT1 gene has been deleted from these
trypanosomes by targeted homologous recombination,51 resulting in a
higher and more stable expression level of the transporter and as such
a more sensitive assay.38 The transport assays were performed
essentially as described,52,53 incubating B48/TbAT1 trypanosomes
with 0.05 μM [2,8-3H]adenosine (American Radiolabeled Chemicals,
Inc.; 40 Ci/mmol) for 20 s and terminating the incubation by the
addition of ice-cold 1 mM adenosine and rapid centrifugation through
an oil layer. Ki values and Gibbs free energy were determined from
competition experiments in the presence of various concentrations of
inhibitor (typically 10 nM to 100 μM) as described.30
Caco-2 membrane permeability of 1b was performed in accordance
with published protocols.56 Caco-2 cell monolayers (passage 94−105)
were grown on permeable filter support and used for transport study
on day 21 after seeding. Prior to the experiment a drug solution of 10
μM was prepared and warmed to 37 °C. The Caco-2 filters were
washed with prewarmed HBSS prior to the experiment, and thereafter
the experiment was started by applying the donor solution on the
apical side or basolateral side, depending on which direction that was
monitored. The transport experiments were carried out at pH 7.4 in
both the apical and basolateral chambers. The experiments were
performed at 37 °C and with a stirring rate of 500 rpm. The receiver
compartment was sampled at 15, 30, and 60 min, and at 60 min also a
final sample from the donor chamber was taken in order to calculate
the mass balance of the compound. Directly after the termination of
the experiment the filter inserts were washed with prewarmed HBSS
and the membrane integrity was checked. This was performed by
trans-epithelial electrical resistance (TEER) measurement. The
experiment was validated by inclusion of the para-cellular marker
14C-mannitol and monitoring its permeability during the experiment
with 1b. Mannitol is a para-cellular marker used for cell-integrity
measurements. All samples were analyzed by LC−MS/MS.
In Vivo Pharmacokinetics. Male C57Bl/6 mice weighing
approximately 25 g were administered intraperitoneally or orally by
gavage with 1b to study the pharmacokinetic properties. Stock
concentrations of 1b were dissolved in 100% DMSO, and further
dilutions were made in PBS. Mice received a single dose of 25 mg/kg
intraperitoneally or orally. Blood samples were collected from tail vein
and by cardiac puncture under terminal anesthesia in EDTA-coated
capillary tubes. Samples were centrifuged for the preparation of plasma
and stored at −80 °C. The plasma samples were analyzed essentially as
described above for plasma protein binding. Standard curves were
constructed by spiking 1b in naive mouse plasma between 10 and
5000 nM. The limit of detection was estimated to ∼30 nM (3 × S/N
ratio) and limit of quantitation to ∼100 nM (10 × S/N ratio).
Adenosine Deamination Assay. Human recombinant adenosine
deaminase (ADA) was purchased from Sigma (Sigma-Aldrich, product
number 93985). Before use the ADA suspension at 1.9 U/mL was
thawed and then centrifuged for 2 min in a benchtop centrifuge to
remove precipitates that could interfere with the absorbance
measurements; i.e., the supernatant was used for the degradation
experiments. The experiments were conducted in UV transparent 96-
well plates (Nunc, product number 8404) using an Envision
microplate reader (PerkinElmer). Scans of the absorption spectrum
were taken between 230 and 300 nm with 2 nm resolution every 15
min for three independent samples containing 100 μM cordycepin or
2-fluorocordycepin in the presence of 3.5 mU/mL of ADA for 10 h.
Experiments were conducted in the absence or presence of 10 μM
ADA inhibitor deoxycoformycin (Dcf). Triplicate samples containing
only ADA or DCF were also obtained, and these were subtracted to
obtain the spectrum over time for cordycepin 1a and 2-
fluorocordycepin 1b. The raw data were exported as a comma
separated value (csv) file from the Envision instrument and then
imported into GraphPad Prism (version 6.02) for analysis and
visualization.
In Vitro Preclinical Profiling of 1b: General. All assay samples
were analyzed by LC−MS/MS using a Waters XEVO TQ triple-
quadrupole mass spectrometer (electrospray ionization, ESI) coupled
to a Waters Acquity UPLC (Waters Corp.). For chromatographic
separation a general gradient was used (1% mobile phase B to 90%
over 2 min total run) on a C18 HSS T3 1.8 μm column, 2 mm × 50
mm (Waters Corp.). Mobile phase A consisted of 5% acetonitrile and
0.1% formic acid, and mobile phase B consisted of 100% acetonitrile
and 0.1% formic acid. The flow rate was 0.5 mL/min. An amount of 5
μL of the sample was injected and run in negative MRM mode
detection (compound 1b parent/daughter ion m/z 268.1/152).
Plasma protein binding samples were precipitated (methanol 4:1)
and centrifuged prior to analysis. A standard curve was constructed
K
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
In Vivo Efficacy Studies. All experiments were conducted with
protocols that received institutional authorization by the Stockholm
Region’s animal protection committee. Adult C57BL6J (B6) mice of
8−10 weeks old were kept with food and water ad libitum under
specific pathogen-free conditions prior to infection. For postexposure
prophylaxis (PEP) treatment, to check whether the compounds have
the ability to prevent infection, mice (four in each group) were
infected intraperitoneally (ip) with 2000 T. b. brucei (Antat1.1E)
parasites and treated 2 h later with 200 μL of 1a or 1b ip. Compound
2 was administered ip once daily (OD) in presence or absence of Dcf.
Following infection, parasitemia and weight were recorded every
alternate day and twice a week after 30 dpi. All investigated
compounds were dissolved in 100% DMSO and further dissolved in
phosphate-buffered saline for treatment so that the final concentration
of DMSO was always 1.25%.
(6) Simarro, P. P.; Franco, J.; Diarra, A.; Postigo, J. a R.; Jannin, J.
Update on field use of the available drugs for the chemotherapy of
human African trypanosomiasis. Parasitology 2012, 139, 842−846.
(7) Priotto, G.; Kasparian, S.; Mutombo, W.; Ngouama, D.;
Ghorashian, S.; Arnold, U.; Ghabri, S.; Baudin, E.; Buard, V.;
Kazadi-Kyanza, S.; Ilunga, M.; Mutangala, W.; Pohlig, G.; Schmid,
C.; Karunakara, U.; Torreele, E.; Kande, V. Nifurtimox−eflornithine
combination therapy for second-stage African Trypanosoma brucei
gambiense trypanosomiasis: a multicentre, randomised, phase III, non-
inferiority trial. Lancet 2009, 374, 56−64.
(8) Alirol, E.; Schrumpf, D.; Amici Heradi, J.; Riedel, A.; de Patoul,
C.; Quere, M.; Chappuis, F. Nifurtimox−eflornithine combination
therapy for second-stage gambiense human African trypanosomiasis:
́ ̀
Medecins Sans Frontieres experience in the Democratic Republic of
the Congo. Clin. Infect. Dis. 2013, 56, 195−203.
(9) Baker, N.; Alsford, S.; Horn, D. Genome-wide RNAi screens in
African trypanosomes identify the nifurtimox activator NTR and the
eflornithine transporter AAT6. Mol. Biochem. Parasitol. 2011, 176, 55−
57.
AUTHOR INFORMATION
Corresponding Author
*Phone: ++468 5248 6711. E-mail: Martin.Rottenberg@ki.se.
■
(10) Wilkinson, S. R.; Taylor, M. C.; Horn, D.; Kelly, J. M.;
Cheeseman, I. A mechanism for cross-resistance to nifurtimox and
benznidazole in trypanosomes. Proc. Natl. Acad. Sci. U.S.A. 2008, 105,
5022−5027.
Author Contributions
#L.G.J.H. and M.E.R. equally to this work
Notes
The authors declare no competing financial interest.
(11) De Koning, H. P.; Bridges, D. J.; Burchmore, R. J. S. Purine and
pyrimidine transport in pathogenic protozoa: from biology to therapy.
FEMS Microbiol. Rev. 2005, 29, 987−1020.
ACKNOWLEDGMENTS
■
(12) Sufrin, J. R.; Spiess, A. J.; Marasco, C. J.; Rattendi, D.; Bacchi, C.
J. Novel trypanocidal analogs of 5′-(methylthio)-adenosine. Anti-
microb. Agents Chemother. 2008, 52, 211−219.
T.L., B.S., A.-L.G., R.S., and L.G.J.H. were funded by the
Swedish National Research Council as part of Chemical
Biology Consortium Sweden (www.cbcs.se). A.A.E. is funded
by a British Commonwealth studentship. The work was
supported by grants from Swedish International Developing
Agency (SIDA) and the Karolinska Institutet.
́
(13) Vodnala, S. K.; Ferella, M.; Lunden-Miguel, H.; Betha, E.; van
Reet, N.; Amin, D. N.; Oberg, B.; Andersson, B.; Kristensson, K.;
Wigzell, H.; Rottenberg, M. E. Preclinical assessment of the treatment
of second-stage African trypanosomiasis with cordycepin and
deoxycoformycin. PLoS Neglected Trop. Dis. 2009, 3, e495.
We thank Dr. H. P. Wei, Southwest University, China, for
kindly providing C6-N-acylated cordycepin.
(14) Rodenko, B.; van der Burg, A. M.; Wanner, M. J.; Kaiser, M.;
Brun, R.; Gould, M.; de Koning, H. P.; Koomen, G.-J. 2,N6-
disubstituted adenosine analogs with antitrypanosomal and antima-
larial activities. Antimicrob. Agents Chemother. 2007, 51, 3796−3802.
(15) Williamson, J.; Scott-Finnigan, T. Trypanocidal activity of
antitumor antibiotics and other metabolic inhibitors. Antimicrob. Agents
Chemother. 1978, 13, 735−744.
ABBREVIATIONS USED
■
ADA, adenosine deaminase; AK, adenosine kinase; APRT,
adenine phosphoribosyltransferase; BBB, blood−brain barrier;
BSA, N,O-bis(trimethylsilyl)acetamide; CNS, central nervous
system; Dcf, deoxycoformycin; DMSO, dimethylsulfoxide;
EDTA, ethylenediaminetetraacetic acid; FCS, fetal calf serum;
HAT, human African trypanosomiasis; NECT, nifurtimox−
eflornithine combinational therapy; QSAR, quantitative struc-
ture−activity relationship; RED, rapid equilibrium dialysis;
TBAF, tetrabutylammonium fluoride; TBAN, tetrabutylammo-
nium nitrate; TbAT1, Trypanosoma brucei P2 adenosine
transporter gene; TEER, trans-epithelial electrical resistance;
TFAA, trifluoroacetic acid; THF, tetrahydrofuran; TMSOTf,
trimethylsilyl triflate; WHO, World Health Organization
(16) Maser, P.; Vogel, D.; Schmid, C.; Raz, B.; Kaminsky, R.
̈
̈
Identification and characterization of trypanocides by functional
expression of an adenosine transporter from Trypanosoma brucei in
yeast. J. Mol. Med. 2001, 79, 121−127.
(17) Rottenberg, M. E.; Masocha, W.; Ferella, M.; Petitto-Assis, F.;
Goto, H.; Kristensson, K.; McCaffrey, R.; Wigzell, H. Treatment of
African trypanosomiasis with cordycepin and adenosine deaminase
inhibitors in a mouse model. J. Infect. Dis. 2005, 192, 1658−1665.
(18) Siev, M.; Weinberg, R.; Penman, S. The selective interruption of
nucleolar RNA synthesis in HeLa cells by cordycepin. J. Cell Biol.
1969, 41, 510−520.
(19) Muller, W.; Seibert, G.; Beyer, R.; Breter, H. Effect of
REFERENCES
̈
■
cordycepin on nucleic acid metabolism in L5178Y cells and on nucleic
acid-synthesizing enzyme systems. Cancer Res. 1977, 37, 3824−3833.
(20) Holbein, S.; Wengi, A.; Decourty, L.; Freimoser, F. M.; Jacquier,
A.; Dichtl, B. Cordycepin interferes with 3′ end formation in yeast
independently of its potential to terminate RNA chain elongation.
RNA 2009, 15, 837−849.
(1) Bacchi, C. J. Chemotherapy of human african trypanosomiasis.
Interdiscip. Perspect. Infect. Dis. 2009, 2009, 1−5.
(2) Kristensson, K.; Nygard, M.; Bertini, G.; Bentivoglio, M. African
̊
trypanosome infections of the nervous system: parasite entry and
effects on sleep and synaptic functions. Prog. Neurobiol. 2010, 91,
152−171.
(3) Maurice, J. New WHO plan targets the demise of sleeping
sickness. Lancet 2013, 381, 13−14.
(4) Stich, A.; Barrett, M. P.; Krishna, S. Waking up to sleeping
sickness. Trends Parasitol. 2003, 19, 195−197.
(5) Bacchi, C. J.; Berens, R. L.; Nathan, H. C.; Klein, R. S.; Elegbe, I.
A; Rao, K. V; McCann, P. P.; Marr, J. J. Synergism between 9-
deazainosine and DL-alpha-difluoromethylornithine in treatment of
experimental African trypanosomiasis. Antimicrob. Agents Chemother.
1987, 31, 1406−1413.
(21) Luscher, A.; Onal, P.; Schweingruber, A.-M.; Maser, P.
̈
̈
Adenosine kinase of Trypanosoma brucei and its role in susceptibility
to adenosine antimetabolites. Antimicrob. Agents Chemother. 2007, 51,
3895−3901.
(22) Vodnala, M.; Fijolek, A.; Rofougaran, R.; Mosimann, M.; Maser,
̈
P.; Hofer, A. Adenosine kinase mediates high affinity adenosine salvage
in Trypanosoma brucei. J. Biol. Chem. 2008, 283, 5380−5388.
(23) Barrett, M. P.; Gilbert, I. H. Targeting of toxic compounds to
the trypanosome’s interior. Adv. Parasitol. 2006, 63, 125−183.
L
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(24) Geiser, F.; Luscher, A.; de Koning, H. P.; Seebeck, T.; Maser, P.
(43) Matovu, E.; Geiser, F.; Schneider, V.; Maser, P.; Enyaru, J. C.;
̈
̈
̈
Molecular pharmacology of adenosine transport in Trypanosoma
brucei: P1/P2 revisited. Mol. Pharmacol. 2005, 68, 589−595.
(25) De Koning, H. P.; Jarvis, S. M. Adenosine transporters in
bloodstream forms of Trypanosoma brucei brucei: substrate recognition
motifs and affinity for trypanocidal drugs. Mol. Pharmacol. 1999, 56,
1162−1170.
(26) Airhart, M. J.; Robbins, C. M.; Knudsen, T. B.; Church, J. K.;
Skalko, R. G. Developing allantois is a primary site of 2′-
deoxycoformycin toxicity. Teratology 1996, 53, 361−373.
(27) Wei, H.-P.; Ye, X.-L.; Chen, Z.; Zhong, Y.-J.; Li, P.-M.; Pu, S.-C.;
Li, X.-G. Synthesis and pharmacokinetic evaluation of novel N-acyl-
cordycepin derivatives with a normal alkyl chain. Eur. J. Med. Chem.
2009, 44, 665−669.
(28) Lynch, T. P.; Paran, J. H.; Paterson, A. R. P. Therapy of mouse
leukemia L1210 with combinations of nebularine and nitrobenzylth-
ioinosine-5′-monophosphate. Cancer Res. 1981, 41, 560−565.
(29) Gillerman, I.; Fischer, B. Investigations into the origin of the
molecular recognition of several adenosine deaminase inhibitors. J.
Med. Chem. 2011, 54, 107−121.
(30) Wallace, L. J. M.; Candlish, D.; De Koning, H. P. Different
substrate recognition motifs of human and trypanosome nucleobase
transporters. Selective uptake of purine antimetabolites. J. Biol. Chem.
2002, 277, 26149−26156.
(31) Dickinson, M.; Holly, F. 3′-Deoxynucleosides. V. 3′-Deoxy-2-
fluoroadenosine. J. Med. Chem. 1967, 275, 1165−1167.
(32) Wilson, D.; Rudolph, F.; Quiocho, F. Atomic structure of
adenosine deaminase complexed with a transition-state analog:
understanding catalysis and immunodeficiency mutations. Science
1991, 252, 1278−1284.
(33) Brockman, R. W.; Schabel, F. M.; Montgomery, J. A. Biological
activity of 9-beta-D-arabinofuranosyl-2-fluoroadenine, a metabolically
stable analog of 9-beta-D-arabinofuranosyladenine. Biochem. Pharmacol.
1977, 26, 2193−2196.
(34) Hansske, F.; Robins, M. Regiospecific and stereoselective
conversion of ribonucleosides to 3′-deoxynucleosides. A high yield
three-stage synthesis of cordycepin from adenosine. Tetrahedron Lett.
1985, 4295−4298.
(35) Robins, M.; Hansske, F.; Low, N.; Park, J. A mild conversion of
vicinal diols to alkenes. Efficient transformation of ribonucleosides into
2′-ene and 2′,3′-dideoxynucleosides. Tetrahedron Lett. 1984, 25, 367−
370.
Kaminsky, R.; Gallati, S.; Seebeck, T. Genetic variants of the TbAT1
adenosine transporter from African trypanosomes in relapse infections
following melarsoprol therapy. Mol. Biochem. Parasitol. 2001, 117, 73−
81.
(44) Kalckar, H. Differential spectrophotometry of purine com-
pounds by means of specific enzymes II. Determination of adenine
compounds. J. Biol. Chem. 1947, 445−459.
(45) Antonino, L. C.; Wu, J. C. Human IMP dehydrogenase catalyzes
the dehalogenation of 2-fluoro- and 2-chloroinosine 5′-mono-
phosphate in the absence of NAD. Biochemistry 1994, 33, 1753−1759.
(46) Bergstrom, C. A.; Strafford, M.; Lazorova, L.; Avdeef, A.;
̈
Luthman, K.; Artursson, P. Absorption classification of oral drugs
based on molecular surface properties. J. Med. Chem. 2003, 46, 558−
570.
(47) Francom, P.; Janeba, Z. Nucleic acid related compounds. 116.
Nonaqueous diazotization of aminopurine nucleosides. Mechanistic
considerations and efficient procedures with tert-butyl. J. Org. Chem.
2002, 67, 6788−6796.
(48) Kawana, M.; Nishikawa, M.; Yamasaki, N.; Kuzuhara, H. Facile
transformation of beta-D-ribofuranosyl purines and pyrimidines into
their respective 3′-deoxy-threo-pentofuranosyl nucleosides. J. Chem.
Soc., Perkin Trans. 1 1989, 1593−1596.
(49) Vodnala, S. K.; Lundback, T.; Sjoberg, B.; Svensson, R.;
̈
̈
Rottenberg, M. E.; Hammarstrom, L. G. J. In vitro and in vivo activity
̈
of 2-aminopyrazines/2-aminopyridines in experimental models of
human African trypanosomiasis. Antimicrob. Agents Chemother. 2013,
57, 1012−1018.
(50) Bridges, D.; Gould, M.; Nerima, B. Loss of the high-affinity
pentamidine transporter is responsible for high levels of cross-
resistance between arsenical and diamidine drugs in African
trypanosomes. Mol. Pharmacol. 2007, 71, 1098−1108.
(51) Matovu, E.; Stewart, M.; Geiser, F.; Brun, R. Mechanisms of
arsenical and diamidine uptake and resistance in Trypanosoma brucei.
Eukaryotic Cell 2003, 2, 1003−1008.
(52) Natto, M. J.; Wallace, L. J. M.; Candlish, D.; Al-Salabi, M. I.;
Coutts, S. E.; de Koning, H. P. Trypanosoma brucei: expression of
multiple purine transporters prevents the development of allopurinol
resistance. Exp. Parasitol. 2005, 109, 80−86.
(53) Kelly, M. A.; Vestling, M. M.; Murphy, C. M.; Hua, S.; Sumpter,
T.; Fenselau, C. Primary structure of bovine adenosine deaminase. J.
Pharm. Biomed. Anal. 1996, 14, 1513−1519.
(54) Waters, N. J.; Jones, R.; Williams, G.; Sohal, B. Validation of a
rapid equilibrium dialysis approach for the measurement of plasma
protein binding. J. Pharm. Sci. 2008, 97, 4586−4595.
(55) Houston, J. Utility of in vitro drug metabolism data in predicting
in vivo metabolic clearance. Biochem. Pharmacol. 1994, 47, 1469−
1479.
(36) Robins, M. J.; Wilson, J. S.; Madej, D.; Low, N. H.; Hansske, F.;
Wnuk, S. F. Nucleic acid related compounds. 88. Efficient conversions
of ribonucleotides into the 2′,3′-anhydro, 2′(and 3′)-deoxy, 2′,3′-
didehydro-2′,3′-dideoxy, and 2′,3′-dideoxynucleoside analogues. J. Org.
Chem. 1995, 60, 7902−7908.
(37) Nicolaou, K. C.; Ellery, S. P.; Rivas, F.; Saye, K.; Rogers, E.;
Workinger, T. J.; Schallenberger, M.; Tawatao, R.; Montero, A.;
Hessell, A.; Romesberg, F.; Carson, D.; Burton, D. Synthesis and
biological evaluation of 2′,4′- and 3′,4′-bridged nucleoside analogues.
Bioorg. Med. Chem. 2011, 19, 5648−5669.
(56) Hubatsch, I.; Ragnarsson, E. G. E.; Artursson, P. Determination
of drug permeability and prediction of drug absorption in Caco-2
monolayers. Nat. Protoc. 2007, 2, 2111−2119.
(38) Maser, P.; Sutterlin, C.; Kralli, A.; Kaminsky, R. A nucleoside
̈
̈
transporter from Trypanosoma brucei involved in drug resistance.
Science 1999, 285, 242−244.
(39) Carter, N.; Barrett, M.; Koning, H. A drug resistance
determinant in Trypanosoma brucei. Trends Microbiol. 1999, 1997−
1999.
(40) Luscher, A.; De Koning, H. P.; Maser, P. Chemotherapeutic
̈
̈
strategies against Trypanosoma brucei: drug targets vs. drug targeting.
Curr. Pharm. Des. 2007, 13, 555−567.
(41) Collar, C.; Al-Salabi, M.; Stewart, M. Predictive computational
models of substrate binding by a nucleoside transporter. J. Biol. Chem.
2009, 284, 34028−34035.
́
(42) Munday, J. C.; Rojas Lopez, K. E.; Eze, A. A.; Delespaux, V.;
Van Den Abbeele, J.; Rowan, T.; Barrett, M. P.; Morrison, L. J.; de
Koning, H. P. Functional expression of TcoAT1 reveals it to be a P1-
type nucleoside transporter with no capacity for diminazene uptake.
Int. J. Parasitol. 2013, 3, 69−76.
M
dx.doi.org/10.1021/jm401530a | J. Med. Chem. XXXX, XXX, XXX−XXX