Selective tight binding inhibitors of trypanosomal glyceraldehyde-3- phosphate dehydrogenase via structure-based drug design

Article abstract of DOI:10.1021/jm9802620

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from the sleeping sickness parasite Trypanosoma brucei is a rational target for anti- trypanosomatid drug design because glycolysis provides virtually all of the energy for the bloodstream form of this parasite. Glycolysis is also an important source of energy for other pathogenic parasites including Trypanosoma cruzi and Leishmania mexicana. The current study is a continuation of our efforts to use the X-ray structures of T. brucei and L. mexicana GAPDHs containing bound NAD⁺ to design adenosine analogues that bind tightly to the enzyme pocket that accommodates the adenosyl moiety of NAD⁺. The goal was to improve the affinity, selectivity, and solubility of previously reported 2'-deoxy-2'-(3-methoxybenzamido)adenosine (1). It was found that introduction of hydroxyl functions on the benzamido ring increases solubility without significantly affecting enzyme inhibition. Modifications at the previously unexploited N⁶-position of the purine not only lead to a substantial increase in inhibitor potency but are also compatible with the 2'-benzamido moiety of the sugar. For N⁶-substituted adenosines, two successive rounds of modeling and screening provided a 330-fold gain in affinity versus that of adenosine. The combination of N⁶- and 2'- substitutions produced significantly improved inhibitors. N⁶-Benzyl (9a) and N⁶-2-methylbenzyl (9b) derivatives of 1 display IC₅₀ values against L. mexicana GAPDH of 16 and 4 μM, respectively (3100- and 12500-fold more potent than adenosine). The adenosine analogues did not inhibit human GAPDH. These studies underscore the usefulness of structure-based drug design for generating potent and species-selective enzyme inhibitors of medicinal importance starting from a weakly binding lead compound.

Full text of DOI:10.1021/jm9802620

4790
J . Med. Chem. 1998, 41, 4790-4799
Selective Tigh t Bin d in g In h ibitor s of Tr yp a n osom a l
Glycer a ld eh yd e-3-p h osp h a te Deh yd r ogen a se via Str u ctu r e-Ba sed Dr u g Design
Alex M. Aronov,† Christophe L. M. J . Verlinde,§ Wim G. J . Hol,§,‡ and Michael H. Gelb*,†
Departments of Chemistry, Biochemistry, and Biological Structure, University of Washington,
Seattle, Washington 98195, and Howard Hughes Medical Institute, Seattle,Washington 98195
Received April 30, 1998
Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from the sleeping sickness parasite
Trypanosoma brucei is a rational target for anti-trypanosomatid drug design because glycolysis
provides virtually all of the energy for the bloodstream form of this parasite. Glycolysis is
also an important source of energy for other pathogenic parasites including Trypanosoma cruzi
and Leishmania mexicana. The current study is a continuation of our efforts to use the X-ray
structures of T. brucei and L. mexicana GAPDHs containing bound NAD+ to design adenosine
analogues that bind tightly to the enzyme pocket that accommodates the adenosyl moiety of
NAD+. The goal was to improve the affinity, selectivity, and solubility of previously reported
2′-deoxy-2′-(3-methoxybenzamido)adenosine (1). It was found that introduction of hydroxyl
functions on the benzamido ring increases solubility without significantly affecting enzyme
inhibition. Modifications at the previously unexploited N6-position of the purine not only lead
to a substantial increase in inhibitor potency but are also compatible with the 2′-benzamido
moiety of the sugar. For N6-substituted adenosines, two successive rounds of modeling and
screening provided a 330-fold gain in affinity versus that of adenosine. The combination of
N6- and 2′-substitutions produced significantly improved inhibitors. N6-Benzyl (9a ) and N6-
2-methylbenzyl (9b) derivatives of 1 display IC50 values against L. mexicana GAPDH of 16
and 4 µM, respectively (3100- and 12500-fold more potent than adenosine). The adenosine
analogues did not inhibit human GAPDH. These studies underscore the usefulness of structure-
based drug design for generating potent and species-selective enzyme inhibitors of medicinal
importance starting from a weakly binding lead compound.
In tr od u ction
tational result that the presence of a competitive inhibi-
tor of hexokinase at a concentration in the range of 0-KI
(KI is the dissociation equilibrium constant for the E‚I
complex) has little effect on glycolytic flux, it has been
argued that competitive inhibitors of glycolytic enzymes
will not be useful for blocking energy production.18
However, there is no basis for this general statement,
as additional simulation studies have shown that gly-
colytic flux is reduced to 0 in the presence of competitive
inhibitors of glyceraldehyde-3-phosphate dehydrogenase
(GAPDH) and phosphoglycerate kinase present at [I]/KI
) 10-100.19
A comparison of the crystal structures of parasite and
mammalian glycolytic enzymes forms the basis for
rational design of selective inhibitors. Analysis revealed
distinct structural differences for GAPDH.20-23 Glyco-
somal GAPDH isolated from trypanosomes has been
studied previously,16 and irreversible nonselective in-
hibitors have been developed.24
The adenosyl moiety of NAD+ does not directly
contribute to catalysis and is sufficiently removed from
the essential active site cysteine residue, and the amino
acid sequence and the position of the protein backbone
around it are not completely conserved between human
and trypanosomal GAPDH (Figure 1). Significant dif-
ferences are found next to the conserved T. brucei Asp
37 (Asp 34 in human GAPDH) which anchors the
adenosine moiety by forming hydrogen bonds to its 2′-
and 3′-hydroxyls. In trypanosomal enzymes the back-
bone is further away from the nucleoside, resulting in
Sleeping sickness, caused by the protozoon Trypano-
soma brucei, is considered by the World Health Orga-
nization to be one of the main tropical parasitic dis-
eases,1 in which parasites invade and then freely live
in the bloodstream of the mammalian host.2,3 The
disease is fatal if left untreated.3,4 Existing chemother-
apy3-12 has serious drawbacks, including severe side
effects, low efficacy, increasing resistance, and no effect
on the most virulent form of sleeping sickness caused
by T. brucei rhodesiense. Other diseases caused by
Trypanosomatidae include leishmaniasis (Leishmania
genus) and Chagas disease (Trypanosoma cruzi).2,4
The work presented here focuses on glycolysis in
trypanosomes as a target for structure-based drug
design. It has been shown that upon going from insect
to bloodstream form the parasites become fully depend-
ent on glycolysis for energy production.2,13,14 Inhibiting
glycolysis in parasites would be expected to slow their
proliferation, as has been shown by Clarkson and Brohn
for the combination of salicylhydroxamic acid and
glycerol.15 Since the kinetic parameters for all of the
enzymes in the glycolytic pathway in T. brucei are
known,16 it has been possible to simulate parasite
glycolytic flux in computro.17,18 Based on the compu-
* To whom correspondence should be addressed.
† Departments of Chemistry and Biochemistry.
§ Department of Biological Structure.
‡ Howard Hughes Medical Institute.
10.1021/jm9802620 CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/31/1998

Selective Tight Binding Inhibitors of GAPDH
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24 4791
F igu r e 1. Stereofigure showing the adenosyl moiety of NAD+ bound to glycosomal T. brucei GAPDH. Hydrogen bonds are shown
as dashed lines.
Ch a r t 1
carried out. Inhibitors with significantly higher activity
have been obtained. Compatibility of 2′,N6-disubstitu-
tion has been established and should allow for further
improvement of the inhibitors currently under study.
Molecu la r Mod elin g
Initially, we decided to test the disposition of atoms
in the modeled structure of T. brucei GAPDH‚1 by
adding substituents on the benzamido group at positions
that were predicted from the trypanosomal GAPDH
X-ray structures not to sterically interfere with binding.
Because of the ease of synthesis and possible improve-
ment in solubility, introduction of hydroxyl groups was
considered. The proposed binding mode of 120 involves
the occupation by a 3-methoxybenzamido group of a
mainly hydrophobic and narrow enzyme cleft that starts
near the ribosyl 2′-position and widens near Asn 39
(Figure 1). The main recognition features of this
substituent are (i) a hydrogen bond between the amide
nitrogen and the carboxylate of Asp 37, (ii) sandwiching
of the aromatic ring by Met 38 and Val 205* (the latter
residue is provided by a neighboring subunit of the
GAPDH tetramer), and (iii) a hydrogen bond between
the methoxy group and ND2 of Asn 39 (predicted
hydrogen bond length 3.3 Å). From inspection of this
model it appeared that positions 4 and 5 of the 3-meth-
oxybenzamido were prime candidates for substitution
as they were pointing mainly into solvent. Position 2
was also considered, with the possibility of forming an
intramolecular hydrogen bond to the 2′-amide nitrogen;
such a conformation can be observed in 15 out of 51
2-hydroxybenzamido groups in the Cambridge Crystal-
lographic database with a hydrogen bond length of 2.63
( 0.04 Å. However, intermolecular contacts did not look
favorable since a hydroxyl at position 2 would bump into
the OD2 of Asp 37 (predicted distance 2.35 Å). In
conclusion, modeling suggested that the introduction of
a 4-hydroxyl and a 5-hydroxyl would leave the affinity
of 1 unchanged, whereas a 2-hydroxyl would disrupt its
binding mode and decrease the GAPDH binding affinity.
The next step was to increase the affinity of our best
parasite-selective lead compound 1 by adding substit-
uents to the adenine ring. Because the 8- and 2-sub-
stituents discovered during our previous design stud-
ies20 are incompatible with the 2′-(3-methoxybenzamido)
a cleft near the 2′-position of the sugar. Flanked by Asn
39 in the T. brucei enzyme (Ser 40 in the L. mexicana
enzyme), this cleft does not exist in the human enzyme
because it is essentially filled by Ile 37.23 This fact was
exploited by Verlinde et al.20,21 who designed, synthe-
sized, and tested a number of 2′-substituted adenosines
with adenosine as a lead (IC50 value around 50 mM).
The best result was achieved with 2′-deoxy-2′-(3-meth-
oxybenzamido)adenosine (1) (Chart 1) (IC50 values of 0.3
mM for L. mexicana GAPDH, 2.2 mM for T. brucei
GAPDH, and >10 mM for human GAPDH). While
using a 2′-amido functional group to preserve the
hydrogen bond to the Asp anchor, the design exploited
the differences in cleft size and achieved remarkable
specificity compared to adenosine, possibly involving a
hydrogen bond from the amido function of Asn 39 in T.
brucei GAPDH to the methoxy oxygen of 1.
Despite the success in achieving substantial selectiv-
ity, the affinity of the best trypanosomal GAPDH
inhibitor is still far too low for it to become a valuable
drug in treating trypanosomal infections. Attempts to
simplify the structure by using cyclopentyl and open-
chain adenine derivatives were only marginally suc-
cessful.22 The 2′-substituent proved to be incompatible
with substituents on the adenine ring such as C2-
methyl22 or C8-thienyl,20 which as monosubstitutions led
to a 10- and 180-fold increase in GAPDH affinity,
respectively, over adenosine.
In the present study mono- and dimodifications of
adenosine have been introduced at positions 2′ and N6
based on modeling studies. Their effect on GAPDH
inhibition has been measured, and optimization was

4792 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24
Aronov et al.
F igu r e 2. Superposition of the 15 substituted N6-methyleneadenines present in the Cambridge Crystallographic Database
(stereoview). All N6-substituents are oriented distal with respect to the imidazole of the adenine ring. In contrast, our proposed
binding mode in GAPDH has the substituent in the proximal orientation.
Sch em e 1a
moiety due to steric clashes between the introduced
substituents,22 we tried to explore the possibility of
introducing substituents at the N6 position. In proxim-
ity to that atom, two hydrophobic clefts exist (Figure
1). One is formed by the side chains of Leu 112 and
Phe 113 and the backbone and first two side-chain
atoms of Arg 91 (narrow cleft). The second site (wide
cleft) points in the opposite direction and is formed by
the side chains of Met 38 and Ala 89 and the backbone
of Ala 89. We assayed the L. mexicana GAPDH inhibi-
tion potency of 11 commercially available N6-substituted
adenosine derivatives that might project a hydrophobic
group into the aforementioned clefts. One of the best
inhibitors was N6-benzyladenosine with an affinity for
L. mexicana GAPDH which was nearly 10-fold higher
than that of adenosine. Because a semirandom syn-
thesis of nine more N6-adenosine derivatives (7a -i) did
not significantly improve on the lead (see Results and
Discussion), we decided to pursue derivatization by
structure-based inhibitor design methods.
Design was carried out as follows. First, we decided
on a synthesis concept of reacting commercially avail-
able primary amines with 6-chloropurine riboside.
From the Available Chemicals Directory (ACD 95.2),
1124 such compounds were retrieved. After filtering out
large compounds (Mr > 250) and compounds with
additional reactive functionalities, 88 compounds re-
mained. These were clustered based on their frame-
works: 19 mono-, 13 di-, and 3 trisubstituted benzy-
a
RCOCl, 10:1 CH2Cl/pyridine; (b) NH4F, methanol.
(Figure 2). However, steric interference is not so severe
because in none of the doubly N6-substituted adenosine
derivatives in the CSD do the R-methyl(ene) groups at
the N6-position rotate out of the plane of the purine ring.
We carried out a semiempirical quantum chemical
lamines, 22 R-substituted benzylamines, 8 bicyclic
compounds, and 23 structures with no apparent simi-
calculation with the AM1 Hamiltonian on N6-methyl-
adenine. It indicated that the distal conformation is
larity to each other. All of the substituents were
preferred by only 2 kcal/mol over the proximal one.
graphically attached at the N6-position of the adenosine
Hence, we concluded that our proposed binding confor-
framework and docked individually into the protein
mation about C6-N6 for the adenosine analogue bound
environment. Adenosyl moiety coordinates from X-ray
to GAPDH is plausible.
data for NAD+ were used as a starting point. Each time
both the narrow and the wide cleft binding modes were
Ch em istr y
considered for the N6-substituent. On the basis of steric
fit, amount of hydrophobic surface buried, and ease of
synthesis, six compounds were selected for synthesis
(7j-o).
Compounds 5b-f were prepared by direct acylation
of the Markiewicz-protected 2′-amino nucleoside 3 fol-
lowed by fluoride-promoted desilylation in MeOH21
(Scheme 1). Precursor 3 was made from adenine 9-â-
D-arabinofuranoside (Ara-A) according to the previously
published procedures.21,26 Hydroxyl functionalities on
the benzoyl fragment were protected via AlCl3-catalyzed
acetylation27 prior to activation of the substituted ben-
zoic acids with PCl5; the acetates were subsequently
removed in NH3/MeOH.
It should be noted that in all cases we opted for a C6-
N6 dihedral angle such that the N6-substituent was
oriented proximal to the imidazole fragment of adenos-
ine in order to avoid severe clashes with the protein
backbone of Arg 91. This is in contrast with the distal
orientation seen in 15 out of 15 structures in the
Cambridge Crystallographic Database, which is usually
attributed to avoidance of a steric interference between
N7 of adenosine and the R-methyl(ene) at position N6 25
Monosubstituted N6-alkyladenosines were prepared
in a one-step conversion from 6-chloropurine riboside

Selective Tight Binding Inhibitors of GAPDH
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24 4793
Sch em e 2
fold, respectively. Identical potency of compounds 1 and
5d , with a 3-OMe and a 3-OH substitution, respectively,
most likely argues against the earlier suggestion that
1 benefits from the interaction of its methoxy substitu-
ent with Val 207* of the neighboring subunit.22 How-
ever, both groups could possibly form a hydrogen bond
to Asn 39 of T. brucei GAPDH. The most soluble
compound of the series, 5e, with three hydroxyls on the
benzamide, is 3-fold more potent than 1. Unfortunately,
ortho-oxidation is common for gallic acid derivatives,34
and the resulting o-quinone metabolites are highly
reactive toward nucleophiles present in the cell.34,35 To
take advantage of both affinity and solubility of 5e,
compound 5f was synthesized. Its affinity was lower,
but the stability was improved. In summary, the effect
that different modifications at position 2′ had on inhibi-
tor potency is consistent with our model of the 2′-
benzamido moiety occupying the hydrophobic cleft in
trypanosomal GAPDHs.
The N6-position was initially investigated via the
testing of commercially available monosubstituted nu-
cleosides. GAPDH binding was selective for secondary
amines over amides (Table 2). Compound 6a , incorpo-
rating the isobutyryl moiety, exhibited both decreased
affinity and drastically lower solubility as compared to
1. The screening afforded hits 6b-d , which showed 12-
16-fold improvement in binding compared to adenosine
(Table 2). They all shared a secondary amine substitu-
tion at the N6-position.
Subsequently we synthesized a series of N6-substi-
tuted adenosine derivatives to explore the use of small
aliphatic chains in a semirandom manner. This re-
vealed a clear trend that substituents with more hy-
drophobic bulk increased binding to GAPDH (see Table
3). Isopropyl (7a ) and tert-butyl (7b) chains were too
short and thus unable to reach either of the two
hydrophobic clefts. Longer chains, both straight and
branched close to the nitrogen, had no effect on affinity.
As branching moved further away along the chain,
binding improved, and the N6-isoamyl derivative 7f was
the most active noncyclic amine adduct in the series
with an IC50 of 3.7 mM (10-fold improvement over
adenosine). This correlates with the activity of dihy-
drozeatin riboside (6d ), as the structures for these two
compounds are identical except for the terminal δ-al-
cohol present in 6d . Straightening the isoamyl chain
by introducing a double bond (7g) abbrogated nearly
half of the N6-isoamyl binding contribution. The rela-
tive success of adenosines containing N6-cyclic amines
(7h ,i) indicates that flexible cyclic substituents with
considerable hydrophobic surface complement the en-
zyme surface as well as flat aromatic systems.
At this point the N6-substituent was optimized by
using a benzyl substituent as the basis for the second
round of drug design. The benzyl moiety provided both
the rigid framework and the wealth of structural
analogues for further optimization of the N6-substituent.
In an effort to better fill the hydrophobic clefts, com-
pounds 7j-o were synthesized from the amines chosen
from the Available Chemicals Directory (see Molecular
Modeling) and tested against L. mexicana GAPDH. The
results are summarized in Table 3. As predicted from
modeling, a methyl group introduced at either the ortho-
or meta-position on the benzyl ring (7j,k ) improved
by nucleophilic substitution with an amine28,29 (Scheme
2). The only exception was N6-(3-methyl-2-butenyl)-
adenosine. In this case, prenyl bromide was used to
alkylate adenosine, and the N6-alkylated intermediate
was rearranged under Dimroth conditions to yield 7g.30
To obtain the N6-isobutyryl derivative 6a , the protected
analogue 4a was first acylated with isobutyryl chloride
and then subjected to NH4F/MeOH deprotection.
The quickest pathway to 2′,N6-disubstituted series
9a -e (see Scheme 3) involved subjecting 4a to alkyla-
tion by substituted benzyl bromides with subsequent
Dimroth rearrangement, the sequence explored by
Robins and co-workers.30 2-Methoxy-, 2,5-dimethyl-, and
2,3-dimethylbenzyl bromides were prepared from the
corresponding alcohols using PBr3;31 2,3-dimethylbenzyl
alcohol was made via LiAlH4 reduction of the respective
benzoic acid.32
Resu lts a n d Discu ssion
P a r a site GAP DH a n d Dr u g Design . Despite the
fact that modeling studies were carried out on GAPDH
from T. brucei, most of the structure-activity relation-
ship (SAR) work was done with the L. mexicana
GAPDH, which overexpresses in Escherichia coli sig-
nificantly better than the T. brucei enzyme. The amino
acid sequences of the two parasite GAPDHs are 81%
identical, and the structures of the binding pockets for
the adenosyl moiety of NAD+ are virtually identical
except at one position (Asn 39 for T. brucei GAPDH and
Ser 40 for L. mexicana GAPDH). The rms deviation in
backbone atoms for the residues in this pocket is 0.2 Å;
the rms deviation for the side-chain atoms is 0.5 Å.33
The conformations of NAD+ in the two enzymes are also
essentially superimposable. In previous studies, it was
found that 2′-(3-methoxybenzamido)adenosine ana-
logues typically bind ∼8-fold tighter to L. mexicana
GAPDH versus T. brucei GAPDH.20 This may be due
in part to the Asn 39/Ser 40 difference. In the present
study, our better inhibitors of L. mexicana GAPDH were
also tested on T. brucei and T. cruzi GAPDHs. The
residues comprising the adenosyl binding pockets of the
latter two enzymes are identical.
Str u ctu r e-Activity Rela tion sh ip s. Introduction
of hydroxyl groups on the benzamide ring of 1 moder-
ated affinity of the inhibitors for L. mexicana GAPDH
by less than 4-fold (Table 1). Hydroxyls were tolerated
at positions 2-5. The addition of the 2- and 4-hydroxyls
led to a minimal improvement in binding: 1.3- and 1.6-

4794 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24
Aronov et al.
Sch em e 3a
a
(a) RBr, DMF, 45 °C, then iPr-NH2/methanol (1:3), reflux; (b) NH4F, methanol.
The combination of N6-benzyl and 2′-(3-methoxyben-
zamide) groups on the adenosine scaffold (9a ) led to a
significant improvement in affinity for GAPDH. The
IC50 of 16 µM is 50-fold lower than that of 1 and 3100-
fold lower than that of adenosine, the original lead
compound. The modeled binding modes for 9a are
shown in Figure 3. The 2′-substituent fills the selectiv-
ity cleft, while the benzyl moiety can occupy either of
the two hydrophobic clefts proximal to N6 of adenosine.
There is a synergistic effect with respect to the indi-
vidual contributions of the substituents (Table 4).
Compound 9a did not detectably inhibit human GAPDH
when tested up to its solubility limit of ∼0.5 mM. It
was shown to be a competitive inhibitor of trypanosomal
GAPDH with respect to NAD+ (Figure 4). Based on the
results with 7j, a 2-methyl group was introduced onto
9a to give 9b. This compound is 4-fold more active than
9a , while retaining its high selectivity for the parasite
enzyme; compound 9b did not inhibit human GAPDH
up to 0.27 mM. Despite the fact that human GAPDH
also possesses the N6 pocket, the selectivity for 9a ,b is
maintained. This selectivity is due to clash of the 2′-
benzamido substituent with Ile 37 in the human en-
zyme, and this will occur when the disubstituted
adenosyl moiety sits in the active site of human GAP-
DH. Other analogues did not show improved potency
compared to 9b. The 2-methoxybenzyl derivative 9c
likely suffered from the putative desolvation of the
oxygen upon binding into either of the hydrophobic
enzyme clefts. Only an orientation into the narrow cleft
appeared possible for 9d ; in the alternative orientation,
the methyl in position 5 appeared to bump into the
protein backbone of Gln 90. Addition of the 2,3-
dimethylbenzyl substitution (9e) did not improve bind-
ing compared to 9b, despite the fact that the methyls
increase the lipophilicity of the N6-substituent.
Ta ble 1. Inhibition of L. mexicana GAPDH by
2′-Deoxy-2′-benzamidoadenosine Analogues
2′-benzamido substituent
compd
R2
R3
R4
R5
IC50 (µM)
1
H
H
OH
H
H
OMe
OMe
OMe
OH
OH
OH
H
OH
H
H
OH
H
H
H
H
H
OH
OH
800
500
600
850
250
650
5b
5c
5d
5e
5f
H
Ta ble 2. Initial Screening of N6-Substituted Adenosine
Analogues as Inhibitors of L. mexicana GAPDHa
compd
IC50 (µM)
50,00022
1,200
adenosine
N6-isobutyryl-2′-deoxy-2′-(3-methoxybenzamido)-
adenosine (6a )
N6-benzyladenosine (6b)
4,200
3,100
3,400
N6-(4-aminobenzyl)adenosine (6c)
N6-(4-hydroxy-3-methylbutyl)adenosine
(dihydrozeatin riboside) (6d )
a
KI values can be calculated from IC50 values using the
standard formula for competitive inhibition, [NAD+] ) 0.19 mM,
and KM(NAD+) ) 0.4 mM for L. mexicana GAPDH, 0.47 for T.
cruzi GAPDH, and 0.54 for T. brucei GAPDH.36
inhibition. This can be explained by a further insertion
of the substituent into the narrow cleft; molecular
mechanics analysis suggests a slight rotation of the side
chain of Leu 112 away from Phe 113 as a way to widen
the narrow cleft and make favorable interactions pos-
sible. Alternatively, partial interaction of benzyl with
the narrow cleft and of the methyl with the wide cleft
is also possible. Tetraline derivative 7l, modeled as
either S or R at the new asymmetric center, seemed to
cap the narrow hydrophobic cleft rather than enter it.
Compound 7m was the most potent inhibitor in the
series (330-fold better than adenosine), most likely by
making good van der Waals contacts with the narrow
or the wide cleft, but the presence of the naphthalene
moiety severely compromised its solubility, which is only
10% higher than its IC50 in 5% DMSO/buffer. The
aromatic thioether substituent in 7n seemed, according
to our model, to partially expose the second benzene ring
to solvent, while 7o with a diphenylmethyl substituent
may take advantage of both hydrophobic clefts.
Con clu sion s
The X-ray structures of L. mexicana, T. brucei, and
human GAPDH‚NAD+ complexes have been used to
identify clefts near the adenosyl moiety of NAD+ that
provide opportunity for structure-based inhibitor design.
The most potent compound 9b represents for L. mexi-
cana GAPDH a 200-fold improvement over 1 and an

Selective Tight Binding Inhibitors of GAPDH
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24 4795
Ta ble 3. Inhibition of L. mexicana GAPDH by N6-Substituted
Adenosine Analogues
F igu r e 3. Two possible orientations of the N6-benzyl sub-
stituent of 9a obtained by molecular modeling. Orientation A
fits in the narrow cleft, while B sits in the wide cleft.
selective for GAPDH. Compounds such as 9b are
promising antiparasite drug candidates since they may
be sufficiently lipophilic to cross parasite membranes
in a transporter-independent fashion.
Exp er im en ta l Section
Gen er a l. Anhydrous solvents were obtained as follows:
pyridine was distilled after refluxing overnight with CaH2 with
a CaSO4 dry tube; CH2Cl2 was refluxed overnight with P2O5
and distilled under Ar; DMF was dried by storing over 4-Å
molecular sieves, followed by distillation under reduced pres-
sure. TLC was carried out with precoated silica gel F254 plates
(EM Science), and spots were detected with UV light (254 nm).
Ninhydrin spray was used for detection of amines. Flash
chromatography was carried out with silica gel (0.040-0.063
mm; EM Science). NMR spectra were taken on Bruker AF-
300 and Bruker WM-500 spectrometers of solutions of ∼2-5
mg of compound in 0.4 mL of CDCl3 unless otherwise men-
tioned. Electrospray ionization (ESI) mass spectra were
obtained on a Kratos Profile HV4 mass spectrometer. All
target compounds were shown to be pure by reverse-phase
HPLC on C4 and C18 Vydac columns with MeOH/H2O elution
gradient.
In h ibition Stu d ies. (1) En zym es, Su bstr a tes, Cofa c-
tor s. Recombinant L. mexicana, T. cruzi, and T. brucei
glycosomal GAPDHs were obtained as a gift from Drs. H. Kim
and P. A. M. Michels.33,36 NAD+ and glyceraldehyde-3-
phosphate diethyl acetal were purchased from Sigma. Fresh
glyceraldehyde-3-phosphate was prepared from its diethyl
acetal according to the instructions provided by the manufac-
turer.
(2) GAP DH In h ibition Assa ys. The compounds tested
were dissolved in DMSO-d6, and concentrations were deter-
mined by integration of NMR peaks with methylene chloride
and/or chloroform as internal standards. NMR spectra were
collected with a pulse recycle delay of 8 s. The activity of
GAPDH was measured in the direction of NADH formation
by monitoring absorption at 340 nm at 21 °C. The 0.5-mL
reaction mixture contained 0.1 M triethanolamine-HCl buffer,
pH 7.6, 1 mM dithiothreitol, 1 mM NaN3, 5 mM MgSO4, 1 mM
EDTA, 10 mM K2HPO4, 0.8 mM glyceraldehyde-3-phosphate,
and 0.19 mM NAD+. The concentration of DMSO in the
reaction cuvette was kept at 5%. The reaction was started by
addition of enzyme. Control reactions were run in the absence
of inhibitors but with 5% DMSO. Remaining activity was
calculated as percent of control using the initial velocities
measured from 0 to 1 min. Inhibitor concentration in the
reaction cuvette was varied, with at least five different
a
b
Inactive up to 10 mM. Remaining enzyme activity at stated
inhibitor concentration.
overall greater than 104-fold affinity increase over
adenosine. Even though a large number of enzymes use
adenosine analogues as cofactors, one could anticipate
that highly substituted analogues 9a -e would be rather

4796 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24
Aronov et al.
Ta ble 4. Selective Parasite GAPDH Inhibition by 2′,N6-Disubstituted Adenosine Analogues (IC50, µM)
compd
N6-R
L. mexicana
T. cruzi
T. brucei
human
9a
9b
9c
9d
9e
benzyl
2-Me-benzyl
2-MeO-benzyl
2,5-diMe-benzyl
2,3-diMe-benzyl
16
4
25
25
5.5
159
30
220
30
ND
ND
ND
>530 (100%)b
>270 (100%)b
NDc
ND
ND
ND
ND
ND
a
b
All parasite GAPDHs are the glycosomal isoform, and the human enzyme is from erythrocytes. Insoluble above stated concentration.
c ND, not determined.
two residues are the ones in contact with the N6-purine
substituents.
To check conformational preferences of several of our
substituents, searches were carried out using the Cambridge
Crystallographic Database,41 followed by statistical analysis
with the accompanying VISTA software.42
To examine the conformational energy difference between
proximal and distal orientations of methyl with respect to the
imidazole of N6-methyladenine, their energies were calculated
with the AM1 Hamiltoninan using the AMSOL 5.0 program.43
Full geometry optimization was carried out.
Buried hydrophobic surface calculations were performed
with the program NACCESS.44
Syn th eses. 2′-Amino-2′-deoxy-3′,5′-O-(1,1,3,3-tetraisopro-
pylsiloxane-1,3-diyl)adenosine (3) and N6-(3-methyl-2-butenyl)-
adenosine were synthesized as described.30,45
P r otection of Hyd r oxy Acid s. Hydroxybenzoic acids
were protected via acetylation as described27 in 80-95% yields,
1
and the structures were verified by H NMR.
2′-N-Acyla tion . Each of the acetoxybenzoic acids (150 mg)
was suspended in 20 mL of dry CH2Cl2, and 100 mg of PCl5
was added with stirring under Ar. The solution was stirred
at room temperature for 5 h, then dried in vacuo overnight.
To a solution of 30 mg of 3 in 10 mL of CH2Cl2/1 mL of
pyridine, was added 3 equiv of acyl chloride, and the mixture
was refluxed for 3 h under Ar. Reaction progress was
monitored by TLC (10% MeOH/EtOAc: Rf ) 1 for acyl chloride,
0.5 for product, and 0.05 for 3). The solvent was evaporated;
the solid was dried in vacuo and redissolved in EtOAc. The
solution was extracted twice with saturated NaHCO3, then
dried, and chromatographed on silica. Nonpolar impurities
were removed with EtOAc, and the product was eluted with
10% MeOH/AcOEt. Rf values for the derivatives are ∼0.4
using 100% EtOAc as solvent.
2′-Deoxy-2′-(3-m eth oxyben za m id o)-3′,5′-O-(1,1,3,3-tet-
r a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4a ). The de-
scribed procedure with 10 mg (19.7 µmol) of 3 yielded 9.8 mg
(77%) of the title compound: 1H NMR δ 1-1.4 (m, 28, 4 iPr),
3.85 (s, 3, OMe), 4.05-4.20 (m, 3, H4′,5′,5′′), 4.95 (m, 1, H2′),
5.55 (m, 1, H3′), 6.16 (br s, 2, NH2), 6.27 (d, 1, H1′), 7.08 (m,
1, H4), 7.26-7.40 (m, 3, H6′′, 5′′, 2′′), 8.07 (s, 1, H2′′), 8.26 (s,
1, H8); identical by NMR and TLC to the reported compound.21
2′-Deoxy-2′-(3-m et h oxy-4-a cet oxyb en za m id o)-3′,5′-O-
(1,1,3,3-tetr a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4b).
The described procedure with 8 mg (15.7 µmol) of 3 yielded
11.4 mg (85%) of title compound: 1H NMR δ 0.9-1.3 (m, 28,
4 iPr), 2.33 (s, 3, acetate), 3.85 (s, 3, OMe), 4.05-4.15 (m, 3,
H4′,5′,5′′), 4.86 (m, 1, H2′), 5.57 (br s, 3, NH2, H3′), 6.09 (d, 1,
H1′), 7.11 (d, 1, H5′′), 7.28 (dd, 1, H6′′), 7.46 (d, 1, H2′′), 8.01
(s, 1, H2), 8.26 (s, 1, H8).
2′-Deoxy-2′-(3-m et h oxy-2-a cet oxyb en za m id o)-3′,5′-O-
(1,1,3,3-tetr a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4c).
The described procedure with 13.8 mg (27.2 µmol) of 3 yielded
6.6 mg (49%) product: 1H NMR δ 0.9-1.4 (m, 28, 4 iPr), 2.33
(s, 3, acetate), 3.82 (s, 3, OMe), 4.0-4.12 (m, 3, H4′,5′,5′′), 4.86
(m, 1, H2′), 5.50 (m, 1, H3′), 6.00 (br s, 3, NH2, H1′), 7.05-7.4
(m, 3, H4′′, 5′′, 6′′), 7.96 (s, 1, H2), 8.2 (s, 1, H8).
F igu r e 4. Competitive inhibition of L. mexicana GAPDH with
16 µM (triangles) and 32 µM (squares) 9a versus the control
(circles).
concentrations used to determine IC50 values. Statistical error
limits on the IC50 values have been calculated and amount to
10%.
Molecu la r Mod elin g. Prior to synthesis, qualitative dock-
ing experiments were carried out with the program BIOGRAF37
in conjunction with the Dreiding force field.38 The adenosine
moiety was fixed in the position observed in the crystal
structure of T. brucei‚NAD+ 23 for the adenosyl portion of
NAD+. Substituents were then added to adenosine followed
by conjugate gradient energy minimization (with the conver-
gence criterion set to 0.1 kcal mol-1 Å-1) in the presence of all
enzyme residues within 12 Å of the modeled inhibitor. In all
simulations the protein environment was kept rigid except
where stated otherwise. Also, a water molecule hydrogen
bonding to N1 of adenosine, the main-chain N of Glu 90, and
the ND2 of Asn 7 was included. The presence of this water
molecule can be inferred by analogy with the crystal structure
of Bacillus stearothermophilus GAPDH;39 it is not visible in
the T. brucei crystal structure due to the 3.2-Å resolution limit.
Because energy calculations were performed in vacuo, the
electrostatic potential energy function was turned off. The
most important effect of electrostatics, namely, the formation
of hydrogen bonds, was taken care of by explicit geometrical
hydrogen bond potentials of the Lennard-J ones 12-10 type
with angular dependence. This approach avoids the difficulties
of calculating and calibrating charges for each new ligand and
of treating dielectric effects in detail.40
For docking compounds 5b-f the side-chain conformation
of residue Met 38 was allowed to vary in order to alleviate
short contacts between the CG atom and the aromatic ring of
the 2′-ribosyl substituents. For docking compounds 9a -e the
side-chain conformations of residues Met 38, Leu 122, and Phe
113 were allowed to change to avoid short contacts. The latter
2′-Deoxy-2′-(3-a cet oxyb en za m id o)-3′,5′-O-(1,1,3,3-t et -
r a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4d ). The de-
scribed procedure with 10 mg (19.7 µmol) of 3 yielded 11 mg
(85%) of title compound: 1H NMR δ 0.85-1.3 (m, 28, 4 iPr),
2.35 (s, 3, acetate), 4.04-4.15 (m, 3, H4′,5′,5′′), 4.90 (m, 1, H2′),

Selective Tight Binding Inhibitors of GAPDH
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24 4797
5.55 (t, 1, H3′) 6.12 (d, 1, H1′), 6.40 (br s, 2, NH2), 7.4-7.7 (m,
4, H2′′, 4′′, 5′′, 6′′), 8.05 (s, 1, H2), 8.27 (s, 1, H8).
N6-Cou p lin g. All N6-alkyladenosines were synthesized
from 6-chloropurine riboside and the corresponding amines
under the conditions described by Fleysher in 90-100%
yield.28,29
2′-Deoxy-2′-(3,4,5-tr ia cetoxyben za m id o)-3′,5′-O-(1,1,3,3-
tetr a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4e). The
described procedure with 23 mg (45.3 µmol) of 3 yielded 15
mg (45%) of title compound: 1H NMR δ 0.8-1.3 (m, 28, 4 iPr),
2.25-2.31 (m, 9, acetates), 4.0-4.1 (m, 3, H4′,5′,5′′), 4.84 (m,
1, H2′), 5.53 (t, 1, H3′), 5.59 (br s, 2, NH2), 6.04 (d, 1, H1′),
7.52(s, 2, H2′′, 6′′), 7.96 (s, 1, H2), 8.24 (s, 1, H8).
N6-Isop r op yla d en osin e (7a ): 1H NMR (MeOH-d4) δ 1.27-
1.32 (d of d, 6, 2 CH3), 3.40 (m, 1, CH), 3.68-3.90 (m, 2, H5′,
5′′), 4.18 (m, 1, H4′), 4.30 (dd, 1, H3′), 4.71 (dd, 1, H2′), 5.91
(d, 1, H1′), 8.18-8.24 (2 s, 2, H8, 2); ESI-MS (methanol) 310.2
(M + H)+.
N6-ter t-Bu tyla d en osin e (7b): 1H NMR (MeOH-d4) δ 1.38
(s, 6, 2 CH3), 1.62 (s, 3, CH3), 3.70-3.95 (m, 2, H5′,5′′), 4.20
(m, 1, H4′), 4.32 (m, 1, H3′), 4.77 (dd, 1, H2′), 5.94 (d, 1, H1′),
8.20-8.24 (s, 2, H8, 2); ESI-MS (methanol) 324.2 (M + H)+.
2′-Deoxy-2′-(3,5-d ia cet oxyb en za m id o)-3′,5′-O-(1,1,3,3-
tetr a isop r op yld isiloxa n e-1,3-d iyl)a d en osin e (4f). The de-
scribed procedure with 40 mg (78.7 µmol) of 3 yielded 33 mg
(60%) of title compound: 1H NMR δ 0.90-1.3 (m, 28, 4 iPr),
2.28 (s, 6, acetates), 4.01-4.13 (m, 3, H4′,5′,5′′), 4.85 (m, 1,
H2′), 5.51 (t, 1, H3′) 6.05 (d, 1, H1′), 6.18 (br s, 2, NH2), 7.09
(t, 1, H4′′), 7.37 and 7.39 (s, 1 and 1, H2′′, 6′′), 7.99 (s, 1, H2),
8.21 (s, 1, H8).
1
N6-Am yla d en osin e (7c): H NMR (MeOH-d4) δ 0.9 (m, 3,
CH3), 1.3 (m, 4, γ,δ CH2), 1.65 (m, 2, â-CH2), 3.57 (m, 2, R-CH2),
3.65-3.9 (m, 2, H5′, 5′′), 4.15 (m, 1, H4′), 4.28 (dd, 1, H3′),
4.75 (dd, 1, H2′), 5.90 (d, 1, H1′), 8.1-8.2 (2 s, 2, H8, 2); ESI-
MS (methanol) 338.2 (M + H)+.
N6-(2-Am yl)a d en osin e (7d ): 1H NMR (MeOH-d4) δ 0.97
(dd, 3, CH3), 1.27 (dd, 3, CH3), 1.43 (m, 2, CH2), 1.59 (m, 2,
CH2), 3.29 (m, 1, CH), 3.70-3.94 (m, 2, H5′, 5′′), 4.18 (dd, 1,
H4′), 4.32 (dd, 1, H3′), 4.76 (dd, 1, H2′), 5.95 (d, 1, H1′), 8.18-
8.23 (2 s, 2, H8, 2); ESI-MS (methanol) 338.2 (M + H)+.
2′-Deoxy-2′-(3-m eth oxy-4-h yd r oxyben za m id o)a d en os-
in e (5b). Compound 4b (8.8 mg, 13.4 µmol) was dissolved in
8 mL of 10% NH3/MeOH and stirred for 15 min. The solution
was dried and subjected to the NH4F deprotection as de-
scribed,21 yielding 5.4 mg of 5b (97%): 1H NMR (DMSO-d6) δ
3.58-3.72 (m, 2, H5′,5′′), 3.76 (s, 3, OMe), 4.09 (m, 1, H4′),
4.29 (m, 1, H3′), 5.29 (m, 1, H2′), 5.61 (m, 1, 5′-OH), 5.76 (d, 1,
3′-OH), 6.17 (d, 1, H1′), 6.76 (d, 1, H5′′), 7.28-7.35 (m, 2, H2′′,
6′′), 8.23 (br s, 2, NH2), 8.09 (d, 1, NH), 8.11 (s, 1, H2), 8.23 (s,
1, H8).
1
N6-(2-Meth ylbu tyl)a d en osin e (7e): H NMR (MeOH-d4)
δ 0.92-1.07 (m, 6, 2 CH3), 1.28 (m, 1, γ-CH2), 1.51 (m, 1,
γ-CH2), 1.78 (m, 1, CH), 2.72 (dd, 1, R-CH2), 2.90 (dd, 1, R-CH2),
3.70-3.93 (m, 2, H5′, 5′′), 4.20 (dd, 1, H4′), 4.34 (dd, 1, H3′),
4.76 (dd, 1, H2′), 5.97 (d, 1, H1′), 8.20-8.26 (2 s, 2, H8, 2);
ESI-MS (methanol) 338.2 (M + H)+.
1
N6-Isoa m yla d en osin e (7f): H NMR (MeOH-d4) δ 1.0 (d,
2′-Deoxy-2′-(3-m eth oxy-2-h yd r oxyben za m id o)a d en os-
in e (5c). The procedure for the synthesis of 5b was followed
for 8.8 mg of 4c (13.4 µmol) and yielded 2.2 mg (40%) of 5c:
1H NMR (DMSO-d6) δ 3.6-3.7 (m, 2, H5′,5′′), 3.80 (s, 3, OMe),
4.10 (m, 1, H4′), 4.35 (m, 1, H3′), 5.32 (m, 1, H2′), 5.55-5.75
(m, 2, 3′,5′-OH), 6.12 (d, 1, H1′), 6.7-7.4 (3 m, 3, H4′′, 5′′, 6′′),
7.30 (br s, 2, NH2), 8.10 (s, 1, H2), 8.30 (s, 1, H8), 8.92 (d, 1,
NH).
6, 2 CH3), 1.62 (m, 2, â-CH2), 1.72 (m, 1, CH), 2.88 (m, 2,
R-CH2), 3.7-3.95 (m, 2, H5′, 5′′), 4.20 (m, 1, H4′), 4.32 (dd, 1,
H3′), 4.74 (dd, 1, H2′), 5.94 (d, 1, H1′), 8.20-8.25 (2 s, 2, H8,
2); ESI-MS (methanol) 338.3 (M + H)+.
N6-Cyclop en tyla d en osin e (7h ): 1H NMR (MeOH-d4) δ
1.70 (m, 4, 2 â-CH2), 2.10 (m, 4, 2 R-CH2), 3.58 (m, 1, CH),
3.70-3.93 (m, 2, H5′, 5′′), 4.19 (dd, 1, H4′), 4.33 (dd, 1, H3′),
4.75 (dd, 1, H2′), 5.94 (d, 1, H1′), 8.20-8.25 (2 s, 2, H8, 2);
ESI-MS (methanol) 336.2 (M + H)+.
2′-Deoxy-2′-(3-h yd r oxyben za m id o)a d en osin e (5d ). The
procedure for the synthesis of 5b was followed for 11 mg of
4d (16.8 µmol) and yielded 4.5 mg (70%) of 5d : 1H NMR
(DMSO-d6) δ 3.58-3.75 (m, 2, H5′,5′′), 4.07 (br s, 1, H4′), 4.30
(t, 1, H3′), 5.25 (dd, 1, H2′), 5.63 (br s, 1, 5′-OH), 5.71 (d, 1,
3′-OH), 6.17 (d, 1, H1′), 6.87 (m, 1, H4′′), 7.15-7.22 (m, 3, H2′′,
5′′, 6′′), 7.35 (br s, 2, NH2), 8.10 (s, 1, H2), 8.16 (d, 1, NH), 8.23
(s, 1, H8).
N6-Cycloh ep tyla d en osin e (7i): 1H NMR δ 1.3-2.0 (m, 12,
aliphatic protons), 3.0 (m, 1, R-CH), 3.7-3.9 (m, 2, H5′, 5′′),
4.25 (m, 1, H4′), 4.35 (dd, 1, H3′), 4.90 (dd, 1, H2′), 5.77 (d, 1,
H1′), 5.94 (br s, 1, NH), 7.80 (s, 1, H2), 8.20 (s, 1, H8); ESI-MS
(methanol) 364.3 (M + H)+.
N6-(2-Meth ylben zyl)a d en osin e (7j): 1H NMR (MeOH-d4)
δ 2.4 (s, 3, CH3), 3.7-3.9 (m, 2, H5′, 5′′), 4.15 (m, 1, H4′), 4.31
(dd, 1, H3′), 4.77 (dd, 1, H2′), 5.94 (d, 1, H1′), 7.1-7.4 (m, 4,
aromatic protons), 8.22 (2 s, 2, H8, 2); ESI-MS (methanol)
372.3 (M + H)+.
2′-Deoxy-2′-(3,4,5-tr ih ydr oxyben zam ido)aden osin e (5e).
The procedure for the synthesis of 5b was followed for 6 mg
of 4e (9.5 µmol) and yielded 3.8 mg (96%) of 5e: 1H NMR
(MeOH-d4) δ 3.75-3.92 (m, 2, H5′, 5′′), 4.23 (br s, 1, H4′), 4.47
(m, 1, H3′), 5.80 (dd, 1, H2′), 6.11 (d, 1, H1′), 6.70 (s, 2, H2′′,
6′′), 8.17 (s, 1, H2), 8.27 (s, 1, H8).
N6-(3-Meth ylben zyl)a d en osin e (7k ): 1H NMR (MeOH-d4)
δ 2.32 (s, 3, CH3), 3.7-3.9 (m, 2, H5′, 5′′), 3.87 (s, 2, CH2), 4.18
(m, 1, H4′), 4.31 (dd, 1, H3′), 4.75 (dd, 1, H2′), 5.93 (d, 1, H1′),
6.95-7.2 (m, 4, aromatic protons), 8.2 (2 s, 2, H8, 2); ESI-MS
(methanol) 372.3 (M + H)+.
2′-Deoxy-2′-(3,5-d ih yd r oxyben za m id o)a d en osin e (5f).
The procedure for the synthesis of 5b was followed for 5 mg
of 4f (7.2 µmol) and yielded 2.5 mg (86%) of 5f: 1H NMR
(DMSO-d6) δ 3.60-3.75 (m, 2, H5′,5′′), 4.08 (br s, 1, H4′), 4.30
(t, 1, H3′), 5.21 (dd, 1, H2′), 5.63 (br s, 1, 5′-OH), 5.69 (d, 1,
3′-OH), 6.16 (d, 1, H1′), 6.32 (br s, 1, H4′′), 6.61 (br s, 2, H2′′,
6′′), 7.33 (br s, 2, NH2), 8.00 (d, 1, NH), 8.13 (s, 1, H2), 8.23 (s,
1, H8).
N6-Acyla t ion . N6-Isob u t yr yl-2′-d eoxy-2′-(3-m et h oxy-
ben za m id o)a d en osin e (6a ). Compound 4a (9 mg, 14 µmol)
was dissolved in 10 mL of 50% CH2Cl2/50% pyridine, and 30
µL (0.29 mmol) of isobutyryl chloride was added. The mixture
was kept at room temperature for 14 h, and solvent was
removed in vacuo. The resulting oil was pure by TLC (Rf 0.5
in EtOAc) and was subjected to NH4F deprotection procedure
in MeOH at 45 °C overnight. The product was chromato-
graphed on silica (10-15% MeOH/EtOAc): yield 4.5 mg (45%);
1H NMR (MeOH-d4) δ 1.1-1.2 (d, 6, 2 Me), 2.9 (m, 1, CH),
3.80 (s, 3, OMe), 4.8-4.95 (m, 2, H5′,5′′), 4.27 (m, 1, H4′), 4.51
(dd, 1, H3′), 5.35 (dd, 1, H2′), 6.30 (d, 1, H1′), 7.0-7.5 (m, 4,
aromatic protons), 8.56-8.62 (s, 1, H8 and s, 1, H2).
N6-(1,2,3,4-Tet r a h yd r o-1-n a p h t h yl)a d en osin e (7l): 1H
NMR δ 1.8-2.4 (m, 6, H2′′, H3′′, H4′′), 3.65-3.95 (m, 2, H5′,
5′′), 4.14 (m, 1, H4′), 4.29 (m, 1, H3′), 4.4 (m, 1, H1′′), 4.98 (m,
1, H2′), 5.75 (d, 1, H1′), 6.20 (br s, 1, NH), 7.0-7.2 (m, 4, H5′′,
6′′, 7′′, 8′′), 7.70 (s, 1, H2), 8.28 (s, 1, H8); ESI-MS (methanol)
398.2 (M + H)+.
1
N6-(1-Na p h th a len em eth yl)a d en osin e (7m ): H NMR δ
3.7-3.95 (m, 2, H5′, 5′′), 4.20 (m, 1, H4′), 4.33 (m, 1, H3′), 4.40
(s, 2, CH2), 4.78 (m, 1, H2′), 5.93 (d, 1, H1′), 7.4-8.1 (m, 7,
aromatic protons), 8.19-8.23 (2 s, 2, H8, 2); ESI-MS (methanol)
408.1 (M + H)+.
N6-[2-(2-(H yd r oxym et h yl)p h en ylt h io)b en zyl]a d en o-
1
sin e (7n ): H NMR δ 3.71-3.91 (m, 2, H5′, H5′′), 4.18 (m, 1,
H4′), 4.31 (dd, 1, H3′), 4.70 (s, 2, CH2O), 4.73 (dd, 1, H2′), 4.90
(br s, 2, CH2N), 5.91 (d, 1, H1′), 7.05-7.50 (m, 8, aromatic
protons), 8.15-8.25 (2 s, 2, H8, 2); ESI-MS (methanol) 496.2
(M + H)+.
1
N6-(Dip h en ylm eth yl)a d en osin e (7o): H NMR (MeOH-
d4) δ 3.7-3.9 (m, 2, H5′, 5′′), 4.17 (m, 1, H4′), 4.29 (dd, 1, H3′),
4.72 (dd, 1, H2′), 5.34 (s, 1, CH), 5.92 (d, 1, H1′), 7.2-7.35 (m,

4798 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24
Aronov et al.
10, aromatic protons), 8.17 (s, 1, H2), 8.26 (s, 1, H8); ESI-MS
434.0 (M + H)+.
to M.H.G. A.M.A. is the recipient of a Dow Chemical
Fellowship Award. We wish to thank Drs. Paul A.
Michels and Hidong Kim for providing a source of
enzyme and Murdock Charitable Trust for a major
equipment grant to the University of Washington Bio-
molecular Structure Center.
N6-Alk yla tion . N6-Ben zyl-2′-d eoxy-2′-(3-m eth oxyben -
za m id o)-3′,5′-O-(1,1,3,3-t et r a isop r op yld isiloxa n e-1,3-d i-
yl)a d en osin e (8a ). Benzyl bromide (0.12 mL, 1 mmol) was
added to a solution of 24 mg (37.4 µmol) of 4a in 5 mL of DMF,
and the mixture was kept at 45 °C for 14 h until no starting
material could be detected by TLC (EtOAc). DMF was
evaporated in vacuo, and the remaining oil was dried by
coevaporation with MeOH. It was then dissolved in 25%
iPrNH2/75% MeOH and slowly refluxed for 24 h. The solid
was dried and chromatographed on silica with 1:1 hexane/
EtOAc (Rf 0.7 in EtOAc): yield 14.3 mg (52%); 1H NMR δ
0.85-1.2 (m, 28, 4 iPr), 3.81 (s, 3, OMe), 3.89-4.15 (m, 2, H5′,
5′′), 4.25 (d, 1, H4′), 4.80 (m, 3, H2′, CH2), 5.26 (t, 1, H3′), 6.12
(d, 1, H1′), 6.43 (br s, 1, NH amine), 7.2-7.4 (m, 9, aromatic
protons), 8.02 (s, 1, H2), 8.34 (s, 1, H8).
Refer en ces
(1) WHO Consultation Program. Bull. W. H. O. 1982, 60, 821-825.
(2) Clayton, C.; Haeusler, T.; Blattner, J . Protein Trafficking in
Kinetoplastid Protozoa. Microbiol. Rev. 1995, 59, 325-344.
(3) Despommier, D. D.; Gwadz, R. W.; Hotez, P. J . Parasitic
Diseases; Springer-Verlag: New York, 1995; pp 196-203.
(4) Meshnick, S. R. In Parasitic Diseases; Mansfield, J . M., Ed.;
Marcel Dekker: New York, 1984; Vol. 2, pp 165-199.
(5) Rew, R. S. Protozoan Infections of Man: African Trypanosomia-
sis. In Chemotherapy of Parasitic Diseases; Campbell, W. C.,
Rew, R. S., Eds.; Plenum: New York, 1986; pp 129-138.
(6) Bitonti, A. J .; Byers, T. L.; Bush, T. L.; Casara, P. J .; Bacchi, C.
J .; Clarkson, J .; McCann, P. P.; Sjoerdsma, A. Cure of Trypa-
nosoma brucei brucei and Trypanosoma brucei rhodesiense
Infections in Mice with an Irreversible Inhibitor of S-Adenos-
ylmethionine Decarboxylase. Antimicrob. Agents Chemother.
1990, 34, 1485-1490.
N6-(3-Met h yl-2-b u t en yl)a d en osin e (7g). 4-Bromo-2-
methyl-2-butene (0.05 mL, 0.43 mmol) was added to a solution
of 30 mg (112 µmol) of adenosine and 35 mg of BaCO3 in 5 mL
of DMF, and the mixture stirred at room temperature in the
dark for 48 h. DMF was evaporated in vacuo, and the
rearrangement was performed in a similar manner as for 10a
yielding 11.8 mg (31%) of title compound: 1H NMR δ 1.7 (br
s, 6, 2 δ-CH3), 3.65-3.95 (m, 2, H5′, 5′′), 4.15 (m, 1, H4′), 4.30
(m, 1, H3′), 4.15 and 4.40 (m, 2, R-CH2), 5.0 (m, 1, H2′), 5.30
(t, 1, â-CH) 5.78 (d, 1, H1′), 6.0 (br s, 1, NH), 7.75 (s, 1, H2),
8.05 (s, 1, H8); ESI-MS (methanol) 336.3 (M + H)+.
(7) Bacchi, C. J .; Nathan, H. C.; Yarlett, N.; Goldberg, B.; McCann,
P. P.; Sjoerdsma, A.; Saric, M.; Clarkson, A. B., J r. Combination
Chemotherapy of Drug-Resistant Trypanosoma brucei rhod-
esiense Infections in Mice Using DL-R-Difluoromethylornithine
and Standard Trypanocides. Antimicrob. Agents Chemother.
1994, 34, 563-569.
N6-Ben zyl-2′-d eoxy-2′-(3-m et h oxyb en za m id o)a d en os-
in e (9a ). Deprotection procedure described for 121 starting
with 2.1 mg (4.3 µmol) of 8a gave quantitative yield of product,
which was purified on silica (Rf 0.25 in EtOAc) and on a C18
Sep-Pak cartridge (loaded in MeOH/H2O and eluted with
MeOH): 1H NMR (DMSO-d6) δ 3.58-3.72 (m, 2, H5′, 5′′), 3.76
(s, 3, OMe), 4.08 (m, 1, H4′), 4.32 (m, 1, H3′), 4.68 (br s, 2,
CH2), 5.30 (m, 1, H2′), 5.57 (dd, 1, 5′-OH), 5.72 (d, 1, 3′-OH),
6.21 (d, 1, H1′), 7.1-7.4 (m, 9, aromatic protons), 8.17 (s, 1,
H2), 8.27 (s, 1, H8), 8.33 (d, 1, NH amide); ESI-MS (methanol)
491.2 (M + H)+.
(8) Gutteridge, W. E. Existing Chemotherapy and its Limitations.
Br. Med. Bull. 1985, 41, 162-168.
(9) Metcalf, B. W.; Bey, P.; Danzin, C.; J ung, M. J .; Casara, P.;
Vevert, J . P. Catalytic Irreversible Inhibition of Mammalian
Ornithine Decarboxylase (4.1.1.17) by Substrate and Product
Analogues. J . Am. Chem. Soc. 1978, 100, 2551-2553.
(10) Schechter, P. J .; Barlow, J . R. L.; Sjoerdsma, A. Clinical Aspects
of Inhibition of Ornithine Decarboxylase with Emphasis on
Therapeutic Trials of Eflornithine (DFMO) in Cancer and
Protozoan Disease. In Inhibition of Polyamine Metabolism;
McCann, P. P., Pegg, A. E., Sjoerdsma, A., Eds.; Academic: New
York, 1987; pp 345-364.
(11) Trop. Drug Res. News 1990, 34, 1.
N6-(2-Meth ylben zyl)-2′-deoxy-2′-(3-m eth oxyben zam ido)-
a d en osin e (9b). The procedures for 8a and 9a were applied
in sequence using 15.6 mg (24.3 µmol) of 4a and 0.12 mL (166
mg, 0.9 mmol) of 2-methylbenzyl bromide. The overall yield
for the two steps was 0.7 mg (6%): 1H NMR (DMSO-d6) δ 2.32
(s, 3, CH3), 3.6-3.75 (m, 2, H5′, 5′′), 3.79 (s, 3, OMe), 4.10 (m,
1, H4′), 4.32 (m, 1, H3′), 4.65 (br s, 2, CH2), 5.30 (m, 1, H2′),
5.59 (dd, 1, 5′-OH), 5.72 (d, 1, 3′-OH), 6.21 (d, 1, H1′), 7.05-
7.45 (m, 8, aromatic protons), 8.18 (s, 1, H2), 8.29 (s, 1, H8),
8.34 (d, 1, NH amide); ESI-MS (methanol) 505.3 (M + H)+.
N6-(2-Meth oxyben zyl)-2′-d eoxy-2′-(3-m eth oxyben za m i-
d o)a d en osin e (9c). The overall yield for the two steps was
38%: 1H NMR (MeOH-d4) δ 3.81 (m, 1, H5′), 3.92 (m, 1, H5′′),
3.79, 3.86 (2 s, 6, 2 OMe), 4.28 (m, 1, H4′), 4.50 (m, 1, H3′),
4.77 (br s, 2, CH2), 5.38 (m, 1, H2′), 6.19 (d, 1, H1′), 6.85-7.30
(m, 8, aromatic protons), 8.22 (2 s, 2, H8, 2); ESI-MS (metha-
nol) 521.4 (M + H)+.
(12) Bellofatto, V.; Fairlamb, A.; Henderson, G. B.; Cross, G. A. M.
Biochemical changes associated with R-difluoromethylornithine
uptake and resistance in Trypanosoma brucei. Mol. Biochem.
Parasitol. 1987, 25, 227-238.
(13) Opperdoes, F. R. Biochemical Peculiarities of Trypanosomes,
African and South American. Br. Med. Bull. 1985, 41, 130-136.
(14) Opperdoes, F. R. Compartmentation of Carbohydrate Metabo-
lism in Trypanosomes. Annu. Rev. Microbiol. 1987, 41, 127-
151.
(15) Clarkson, A. B., J r.; Brohn, F. H. Trypanosomiasis: An Approach
to Chemotherapy by the Inhibition of Carbohydrate Catabolism.
Science 1976, 194, 204-206.
(16) Lambeir, A.-M.; Loiseau, A. M.; Kuntz, D. A.; Vellieux, F. M.;
Michels, P. A. M.; Opperdoes, F. R. The cytosolic and glycosomal
glyceraldehyde-3-phosphate dehydrogenase from Trypanosoma
brucei. Kinetic properties and comparison with homologous
enzymes. Eur. J . Biochem. 1991, 198, 429-435.
(17) Bakker, B. M.; Michels, P. A. M.; Opperdoes, F. R.; Westerhoff,
H. V. Glycolysis in Bloodstream Form Trypanosoma brucei Can
Be Understood in Terms of the Kinetics of the Glycolytic
Enzymes. J . Biol. Chem. 1997, 272, 3207-3215.
(18) Eisenthal, R.; Cornish-Bowden, A. Prospects for Antiparasitic
Drugs. J . Biol. Chem. 1998, 273, 5500-5505.
(19) Bakker, B. M. Control and Regulation of Glycolysis in Trypa-
nosoma brucei. Ph.D. Thesis, Free University of Amsterdam,
1998.
(20) Verlinde, C. L. M. J .; Callens, M.; Van Calenbergh, S.; Van
Aerschot, A.; Herdewijn, P.; Hannaert, V.; Michels, P. A. M.;
Opperdoes, F. R.; Hol, W. G. J . Selective Inhibition of Trypano-
somal Glyceraldehyde-3-Phosphate Dehydrogenase by Protein
Structure-Based Design: Toward New Drugs for the Treatment
of Sleeping Sickness. J . Med. Chem. 1994, 37, 3605-3613.
(21) Van Calenbergh, S.; Van Den Eeckhout, E.; Herdewijn, P.; De
Bruyn, A.; Verlinde, C.; Hol, W.; Callens, M.; Van Aerschot, A.;
Rozenski, J . Synthesis and Conformational Analysis of 2′-Deoxy-
2′-(3-methoxybenzamido)adenosine, a Rational-Designed Inhibi-
tor of Trypanosomal Glyceraldehyde Phosphate Dehydrogenase
(GAPDH). Helv. Chim. Acta 1994, 77, 631-644.
N6-(2,5-Dim eth ylben zyl)-2′-d eoxy-2′-(3-m eth oxyben za -
m id o)a d en osin e (9d ). The overall yield for the two steps
was 25%: 1H NMR (DMSO-d6) δ 2.13, 2.27 (2 s, 6, 2 CH3),
3.58-3.75 (m, 2, H5′, 5′′), 3.76 (s, 3, OMe), 4.09 (m, 1, H4′),
4.32 (m, 1, H3′), 4.61 (br s, 2, CH2), 5.30 (m, 1, H2′), 5.59 (dd,
1, 5′-OH), 5.73 (d, 1, 3′-OH), 6.21 (d, 1, H1′), 6.9-7.45 (m, 7,
aromatic protons), 8.19 (s, 1, H2), 8.29 (s, 1, H8), 8.36 (d, 1,
NH amide); ESI-MS (methanol) 519.3 (M + H)+.
N6-(2,3-Dim eth ylben zyl)-2′-d eoxy-2′-(3-m eth oxyben za -
m id o)a d en osin e (9e). The overall yield for the two steps was
11%: 1H NMR (MeOH-d4) δ 2.26, 2.29 (2 s, 6, 2 CH3), 3.79 (s,
3, OMe), 3.8-3.93 (m, 2, H5′, 5′′), 4.27 (m, 1, H4′), 4.5 (m, 1,
H3′), 4.75 (br s, 2, CH2), 5.38 (m, 1, H2′), 6.2 (d, 1, H1′), 7.0-
7.35 (m, 7, aromatic protons), 8.22-8.27 (2 s, 2, H8,2); ESI-
MS (methanol) 519.3 (M + H)+.
(22) Van Calenbergh, S.; Verlinde, C. L. M. J .; Soenens, J .; De Bruyn,
A.; Callens, M.; Blaton, N. M.; Peeters, O. M.; Rozenski, J .; Hol,
W. G. J .; Herdewijn, P. Synthesis and Structure-Activity
Ack n ow led gm en t. This work was supported by
Grant HL36235 from the National Institutes of Health

Selective Tight Binding Inhibitors of GAPDH
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 24 4799
Relationships of Analogues of 2′-Deoxy-2′-(3-methoxybenzami-
do)adenosine, a Selective Inhibitor of Trypanosomal Glycosomal
Glyceraldehyde-3-phosphate Dehydrogenase. J . Med. Chem.
1995, 38, 3838-3849.
(34) Quideau, S.; Feldman, K. S.; Appel, H. M. Chemistry of Galloyl-
Derived o-Quinones: Reactivity Toward Nucleophiles. J . Org.
Chem. 1995, 60, 4982-4983.
(35) Harvey, R. G. Polycyclic Hydrocarbons and Carcinogenesis. ACS
Monograph No. 283; American Chemical Society: Washington,
DC, 1985.
(36) Hannaert, V.; Opperdoes, F. R.; Michels, P. A. M. Glycosomal
Glyceraldehyde-3-Phosphate Dehydrogenase of Trypanosoma
brucei and Trypanosoma cruzi: Expression in Escherichia coli,
Purification and Characterization of the Enzymes. Protein Expr.
Purif. 1995, 6, 244-250.
(37) Molecular Simulations Inc., Biograf 3.10; San Diego, CA.
(38) Mayo, S. L.; Olafson, B. D.; Goddard, W. A., III. DREIDING: A
Generic Force Field for Molecular Simulations. J . Phys. Chem.
1990, 94, 8897-8909.
(39) Skarzynski, T.; Moody, P. C. E.; Wonacott, A. J . Structure of
the holo-glyceraldehyde-3-phosphate dehydrogenase from Bacil-
lus stearothermophilus at 1.8 A resolution. J . Mol. Biol. 1987,
193, 171-187.
(40) Mueller, K.; Amman, H.-J .; Doran, D.; Gerber, P. R.; Gubernator,
K.; Schrepfer, G. The Use of Computer Modeling and Structural
Databases in Pharmaceutical Research. In Trends in Medicinal
Chemistry ‘88; van der Goot, H., Domany, G., Pallos, L.,
Timmerman, H., Eds.; Elsevier Science Publishers: Amsterdam,
1989; pp 1-12.
(41) Allen, F. H.; Davies, J . E.; Galloy, J . J .; J ohnson, O.; Kennard,
O.; Macrea, C. F.; Mitchell, E. M.; Mitchell, G. F.; Smith, M. J .;
Watson, D. G. The Development of Versions 3 and 4 of the
Cambridge Crystallographic Database System. J . Chem. Inf.
Comput. Sci. 1993, 31, 187-204.
(42) Watson, D. G. The Cambridge Structural Database (CSD):
Current Activities and Future Plans. J . Res. Natl. Inst. Stand.
Technol. 1996, 101, 227-229.
(23) Vellieux, F. M. D.; Hajdu, J .; Verlinde, C. L. M. J .; Groendijk,
H.; Read, R. J .; Greenhough, T. J .; Campbell, J . W.; Kalk, K.
H.; Littlechild, J . A.; Watson, H. C.; Hol, W. G. J . Structure of
glycosomal glyceraldehyde-3-phosphate dehydrogenase from
Trypanosoma brucei determined from Laue data. Proc. Natl.
Acad. Sci. U.S.A. 1993, 90, 2355-2359.
(24) Willson, M.; Lauth, N.; Perie, J .; Callens, M.; Opperdoes, F. R.
Inhibition of Glyceraldehyde-3-Phosphate Dehydrogenase by
Phosphorylated Epoxides and R-Enones. Biochemistry 1994, 33,
214-220.
(25) Saenger, W. Principles of Nucleic Acid Structure; Springer-
Verlag: New York, 1984.
(26) Robins, M. J .; Wilson, J . S.; Hansske, F. Nucleic Acid Related
Compounds. 42. A General Procedure for the Efficient Deoxy-
genation of Secondary Alcohols. Regiospecific and Stereoselective
Conversion of Ribonucleosides to 2′-Deoxynucleosides. J . Am.
Chem. Soc. 1983, 105, 4059-4065.
(27) White, B. B.; Barabash, E. U.S. Patent 2 581 565, 1952; Chem.
Abstr. 1952, 46, 5842.
(28) Fleysher, M. H.; Hakala, M. T.; Bloch, A.; Hall, R. H. Synthesis
and Biological Activity of Some N6-Alkyladenosines. J . Med.
Chem. 1968, 11, 717-720.
(29) Fleysher, M. H. N6-Substituted Adenosines: Synthesis, Biologi-
cal Activity, and Some Structure-Activity Relationships. J . Med.
Chem. 1972, 15, 187-191.
(30) Robins, M. J .; Trip, E. M. Sugar-Modified N6-(3-Methyl-2-
butenyl)adenosine Derivatives, N6-Benzyl analogues, and Cy-
tokinin-Related Nucleosides Containing Sulfur or Formycin.
Biochemistry 1973, 12, 2179-2187.
(31) Drew, J .; Letellier, M.; Morand, P.; Szabo, A. G. Synthesis from
Pregnenolone of Fluorescent Cholesterol Analogue Probes with
Conjugated Unsaturation in the Side Chain. J . Org. Chem. 1987,
52, 4047-4052.
(32) Raederstorff, D.; Shu, A. Y. L.; Thompson, J . E.; Djerassi, C.
Biosynthetic Studies of Marine Lipids. 11. Synthesis, Biosyn-
thesis, and Absolute Configuration of the Internally Branched
Demospongic Acid, 22-Methyl-5,9-Octacosadienoic acid. J . Org.
Chem. 1987, 52, 2337-2346.
(43) Liotard, D. A.; Hawkins, G. D.; Lynch, G. C.; Cramer, C. J .;
Truhlar, D. G. Improved Methods for Semiempirical Solvation
Models. J . Comput. Chem. 1995, 16, 422-440.
(44) Hubbard, S. J .; Thornton, J . M. NACCESS, London.
(45) Robins, M. J .; Hawrelak, S. D.; Hernandez, A. E.; Wnuk, S. F.
Nucleic Acid Related Compounds. 71. Efficient General Synthe-
sis of Purine (Amino, Azido, and Triflate)-Sugar Nucleosides.
Nucleosides Nucleotides 1992, 11, 821-834.
(33) Kim, H.; Feil, I.; Verlinde, C. L. M. J .; Petra, P. H.; Hol, W. G.
J . Crystal Structure of Glycosomal Glyceraldehyde-3-Phosphate
Dehydrogenase from Leishmania Mexicana: Implications for
Structure-Based Drug Design and
a New Position for the
Inorganic Phosphate Binding Site. Biochemistry 1995, 34,
14975-14986.
J M9802620