Structure-based design and synthesis of lipophilic 2,4-diamino-6- substituted quinazolines and their evaluation as inhibitors of dihydrofolate reductases and potential antitumor agents

Article abstract of DOI:10.1021/jm980081y

The synthesis and biological activities of 14 6-substituted 2,4- diaminoquinazolines are reported. These compounds were designed to improve the cell penetration of a previously reported series of 2,4-diamino-6- substituted-pyrido[2,3-d]pyrimidines which h

Full text of DOI:10.1021/jm980081y

3426
J . Med. Chem. 1998, 41, 3426-3434
Str u ctu r e-Ba sed Design a n d Syn th esis of Lip op h ilic 2,4-Dia m in o-6-Su bstitu ted
Qu in a zolin es a n d Th eir Eva lu a tion a s In h ibitor s of Dih yd r ofola te Red u cta ses
a n d P oten tia l An titu m or Agen ts1
Aleem Gangjee,*,† Anup P. Vidwans,† Anil Vasudevan,†, Sherry F. Queener,‡ Roy L. Kisliuk,| Vivian Cody,§
Ruming Li,§ Nikolai Galitsky,§ J oe R. Luft,§ and Walter Pangborn§
Division of Medicinal Chemistry, Graduate School of Pharmaceutical Sciences, Duquesne University,
Pittsburgh, Pennsylvania 15282, Department of Pharmacology and Toxicology, School of Medicine, Indiana University,
Indianapolis, Indiana 46202, Department of Biochemistry, School of Medicine, Tufts University, Boston, Massachusetts 02111,
and Hauptman-Woodward Medical Research Institute Inc., 73 High Street, Buffalo, New York 14203
Received February 5, 1998
The synthesis and biological activities of 14 6-substituted 2,4-diaminoquinazolines are reported.
These compounds were designed to improve the cell penetration of a previously reported series
of 2,4-diamino-6-substituted-pyrido[2,3-d]pyrimidines which had shown significant potency and
remarkable selectivity for Toxoplasma gondii dihydrofolate reductase (DHFR), but had much
lower inhibitory effects on the growth of T. gondii cells in culture. The target N9-H analogues
were obtained via regiospecific reductive amination of the appropriate benzaldehydes with 2,4,6-
triaminoquinazoline, which, in turn, was synthesized from 2,4-diamino-6-nitroquinazoline. The
N9-CH3 analogues were synthesized via a regiospecific reductive methylation of the corre-
sponding N9-H precursors. The compounds were evaluated as inhibitors of DHFR from
human, Pneumocystis carinii, T. gondii, rat liver, Lactobacillus casei, and Escherichia coli,
and selected analogues were evaluated as inhibitors of the growth of tumor cells in culture.
These analogues displayed potent T. gondii DHFR inhibition as well as inhibition of the growth
of T. gondii cells in culture. Further, selected analogues were potent inhibitors of the growth
of tumor cells in culture in the in vitro screening program of the National Cancer Institute
with GI50s in the nanomolar and subnanomolar range. Crystallographic data for the ternary
complex of hDHFR-NADPH and 2,4-diamino-6-[N-(2′,5′-dimethoxybenzyl)-N-methylamino]-
pyrido[2,3-d]pyrimidine, 1c, reveal the first structural details for a reversed N9-C10 folate
bridge geometry as well as the first conformational details of a hybrid piritrexim-trimetrexate
analogue.
Infections with Pneumocystis carinii and Toxoplasma
gondii are a major cause of morbidity and mortality in
patients with the acquired immunodeficiency syndrome
(AIDS).2 P. carinii infections are treated with pentami-
dine, trimethoprim-dapsone, trimethoprim-sulfameth-
oxazole, and trimetrexate, while pyrimethamine plus a
sulfonamide is used for the treatment of T. gondii
infections.3,4 The treatment of P. carinii and T. gondii
infections with nonclassical antifolates such as trimeth-
oprim (TMP), trimetrexate (TMQ), and pyrimethamine
active transport system(s) required for the uptake of
classical antifolates with polar glutamate side chains.5
In addition, host tissues can selectively be protected by
the coadministration of leucovorin (5-formyltetrahydro-
folate), which is taken up by mammalian cells and
reverses toxicity associated with dihydrofolate reductase
(DHFR) inhibitors.6 Further, these lipophilic agents can
penetrate into the central nervous system (CNS) where
T. gondii infections usually occur.
TMP and pyrimethamine are weak inhibitors of
DHFR from P. carinii and T. gondii and hence have to
be used along with sulfonamides to augment their
potency.5 TMQ and piritrexim (PTX), which are 100-
takes advantage of the fact that these organisms are
permeable to lipophilic, nonclassical antifolates and,
unlike mammalian cells, lack the carrier-mediated
10 000 times more potent than TMP or pyrimethamine
against DHFR from P. carinii and T. gondii, are also
potent inhibitors of mammalian DHFR,7 and hence their
use is associated with significant toxicity. As a result,
TMQ has to be coadministered with leucovorin. How-
ever, side effects associated with the use of sulfa drugs
† Duquesne University.
‡ Indiana University.
| Tufts University.
§ Hauptman-Woodward Medical Research Institute Inc.
Current address: Abbott Laboratories, Abbott Park, Illinois 60064.
S0022-2623(98)00081-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/12/1998

Synthesis of 2,4-Diamino-6-Substituted Quinazolines
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18 3427
in combination with TMP or pyrimethamine, as well as
the high cost of leucovorin, often necessitate discontinu-
ation of therapy.8 Thus, there is a dire need for new
single agents with high potency and selectivity against
these organisms. This problem is particularly acute for
patients who cannot tolerate existing regimens since,
in the absence of a cure for AIDS-infected patients, the
prophylactic use of antiopportunistic agents needs to be
continued throughout the life of the patient.
As part of a program aimed at identifying potent and
selective inhibitors of P. carinii and/or T. gondii DHFR
of diverse structures, we have reported the design and
synthesis of a variety of 6-6 ring-fused analogues,9-11
as well as 6-5 ring-fused analogues,12,13 with the goal
of increasing the selectivity of TMQ and PTX for P.
carinii (pc)DHFR and T. gondii (tg)DHFR. One par-
ticularly promising series of compounds recently syn-
thesized by Gangjee et al.11 was the 2,4-diamino-6-
substituted-pyrido[2,3-d]pyrimidines of general structure
1. In these analogues, the normal C9-N10 bridge found
Ile123 of pcDHFR. That 5-methyl analogues are in
general more potent than the corresponding 5-des-
methyl analogues against DHFR is well documented for
both classical16,17 and nonclassical9,18 antifolates. The
fact that the 5-methyl of PTX makes hydrophobic
contact with both pcDHFR and hDHFR is part of the
reason that PTX lacks selectivity and is a potent
inhibitor of both pcDHFR and hDHFR.
Since 2,4-diamino-6-[N-(2′,5′-dimethoxybenzyl)-N-
methylamino]pyrido[2,3-d]pyrimidine, 1c, was the most
selective analogue for pcDHFR from compounds of
general structure 111 and was much more potent against
pcDHFR (IC50 ) 84 nM) and tgDHFR (IC50 ) 28 nM)
than hDHFR (IC50 ) 8500 nM), it was of interest to
crystallize this analogue with hDHFR to determine the
molecular reason for its low potency against hDHFR and
with pcDHFR to determine the reason for its high
selectivity. The crystal structure of compound 1c com-
plexed with hDHFR was determined from cocrystalli-
zation experiments, and the results are described in this
report for data to 2.1 Å resolution (Figure 1). The
crystal structure determination of 1c with pcDHFR is
currently underway.
In the crystal structure of 1c with hDHFR it is quite
evident that the N9-methyl moiety is not in van der
Waals contact (e4.0 Å) with any hydrophobic side chain
of hDHFR. This lack of hydrophobic interaction of the
N9-methyl moiety of 1c with the hydrophobic residues
of hDHFR in addition to the absence of the 5-methyl
moiety and its interaction with Val115 in part accounts
for the significantly lower inhibitory potency of 1c
against hDHFR compared to PTX or TMQ which
contain a 5-methyl moiety which is capable of hydro-
phobic contact with hydrophobic residues of hDHFR.14
in most classical as well as nonclassical DHFR inhibi-
tors was replaced with a reversed N9-C10 bridge.
Further, the methyl group at the 5-position which is
responsible for increased potency in nonclassical 5-deaza
antifolates9 was transposed to the N9 not only to induce
conformational restriction around the bridge geometry
τ1 and τ2 but also to evaluate hydrophobic interactions
of the N9-methyl moiety with DHFR, a feature not
explored previously with nonclassical antifolates. Some
analogues in the pyrido[2,3-d]pyrimidine series were
found to possess high potency against pcDHFR and
excellent selectivity and potency as inhibitors of tgDH-
FR, compared to human (h)DHFR, and represent some
of the most promising nonclassical antifolates reported
to date as candidates against opportunistic infections
with P. carinii and T. gondii.11 The premise for the
synthesis of the 5-desmethyl pyrido[2,3-d]pyrimidine
analogues 1 was to decrease the activity of these
analogues against all DHFR with the contention that
perhaps the decrease in inhibitory activity against
hDHFR would be greater than the decrease against
pcDHFR and/or tgDHFR which would translate into
improved selectivity against these organisms versus
mammalian cells while maintaining reasonable potency.
This hypothesis proved to be correct, and some com-
pounds in the pyrido[2,3-d]pyrimidine series were in-
deed potent and selective against pcDHFR and tgDH-
FR.11
Compound 1c was modeled on the crystal structure
of PTX complexed with pcDHFR,15 using SYBYL 6.03,19
in an attempt to determine the structural reason for the
increase in potency and for the selectivity of 1c com-
pared to PTX. Since the X-ray crystal structure coor-
dinates of PTX with pcDHFR are not available, a visual
model of the bound conformation of PTX was obtained
using the known TMP-pcDHFR coordinates15 and the
superimposition provided in the literature of TMP-
pcDHFR and PTX-pcDHFR.15 The pyrimidine ring of
1c in the bound conformation, obtained from the crystal
structure of 1c and hDHFR, was superimposed on the
pyrimidine ring of PTX bound to pcDHFR.15 In this
model the N9-methyl moiety of 1c is in van der Waals
contact (e4.0 Å) with both Ile123 and Ile65 of pcDHFR.
These additional hydrophobic interactions of 1c with
pcDHFR compared to the lack of hydrophobic interac-
tions of the N9-methyl moiety of 1c with hDHFR as
observed in the crystal structure could account, in part,
for both the increased potency and selectivity of 1c for
pcDHFR compared with hDHFR.
In light of its excellent potency and selectivity,
compound 1c was evaluated as an inhibitor of the
growth of T. gondii cells in culture. The cell culture/
enzyme IC50 ratio (447) for T. gondii indicated a greater
than 400-fold decrease in inhibition of the growth of cells
in culture compared with inhibition of the enzyme. This
suggests a lack of adequate cell penetration in T. gondii
cells in culture. Despite its poor cell culture/enzyme
IC50 ratio, compound 1c did show oral activity when
It is well documented from X-ray crystal structures
that the 5-methyl moiety of the 5-deaza-5-methyl clas-
sical and nonclassical antifolates such as PTX and TMQ
make hydrophobic contact with Val115 in hDHFR.14
Recently the crystal structure of pcDHFR complexed
with PTX was published15 which indicates that the
5-methyl group of PTX is in van der Waals contact with

3428 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18
Gangjee et al.
F igu r e 1. Stereodiagram showing the X-ray crystal structure of 1c (dark) in hDHFR F31G variant with NADPH (COE187). The
water molecule HOH1 is shown hydrogen bonded to Glu30 (3.11 Å), Trp24 (3.29 Å), and the N8 of 1c (3.49 Å). The distance
between the N9-CH3 and Val 115 is also designated (4.66 Å).
tested in the mouse model for T. gondii infection.11
Ta ble 1
However, improvement in the cell culture/enzyme IC50
ratio could provide significant improvement in in vivo
activity. It was therefore of interest to maintain the
structural features of compound 1c for potency and
selectivity and to concomitantly increase the lipid
solubility to provide for better inhibition of the growth
of cells in culture. We elected to achieve this goal by
isosterically replacing the N8 of pyrido[2,3-d]pyrim-
idines 1 with a carbon atom to afford quinazolines. We
have demonstrated that calculated log P values can be
used to predict enhanced inhibition of the growth of cells
in culture of a number of similar antifolates.20 Log P
calculations indicated that analogues in the reversed
bridge quinazoline series had significantly improved log
P values compared to those of the corresponding com-
pounds in the pyrido[2,3-d]pyrimidine series as shown
in Table 4 which were anticipated to provide better P.
carinii and T. gondii cell inhibition. Thus we designed
and synthesized the reverse bridge quinazoline series
2-15 (Table 1) where the N8-nitrogen of compounds
R
R′
2
3
4
5
6
7
8
9
10
11
12
13
14
15
H
H
H
H
H
H
H
H
H
2,5-OCH3
3,5-OCH3
2,4-OCH3
3,4,5-OCH3
2,3,4-OCH3
2,4,5-OCH3
2,4,6-OCH3
2,3-C4H4
2,5-OCH3
3,5-OCH3
2,4-OCH3
2,3,4-OCH3
2,3-C4H4
H
CH3
CH3
CH3
CH3
CH3
Molecular modeling using SYBYL 6.0319 and super-
imposing the low-energy conformations of 2,4-diamino-
6-substituted quinazolines on the crystal structure of
1c with hDHFR and on the crystal structure of PTX
with pcDHFR lent further credence to the rationale that
these quinazolines with the N9-methyl substitution
should bind much like compound 1c and hence should
provide the selectivity and potency demonstrated by 1c.
Thus analogues 2-15 and in particular 11-15 were
1 was replaced by a CH.

Synthesis of 2,4-Diamino-6-Substituted Quinazolines
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18 3429
Sch em e 1a
a
Reagents: (a) Raney nickel/DMF/H2 30 psi; (b) Raney nickel/DMF/AcOH/ArCHO/H2 30 psi; (c) HCHO/NaCNBH3/HCl.
anticipated to combine the potency and selectivity of 1
with significantly improved inhibition of the growth of
cells in culture. Such analogues could be used as single
agents or with much lower doses of sulfa drugs which
would overcome some of the drawbacks of currently used
DHFR inhibitors in the treatment of opportunistic
infections caused by P. carinii and T. gondii.
In addition, since 2,4-diaminoquinazolines such as
TMQ have been involved in phase II clinical trials as
antitumor agents, it was also of interest to evaluate
analogues 2-15 as potential antitumor agents. Clearly
those analogues which were significantly selective for
P. carinii and/or T. gondii DHFR were not anticipated
to function as potent inhibitors of the growth of tumor
cells in culture.
and also the first structural details for a hybrid molecule
containing key features of PTX and TMQ. As illustrated
in Figure 1, the reversal of this bridge places the N9-
methyl in a pocket of the hDHFR active site not
occupied by any other class of antifolate analogues. In
this orientation, the N9-methyl group does not make
any contacts within the range of <4 Å of the hydropho-
bic residues Thr56, Ile60, Val115, or the nicotinamide
ring carbon. This lack of interaction of the N9-methyl
moiety of 1c with the hydrophobic residues of hDHFR,
in addition to the absence of the 5-methyl moiety, in
part, accounts for the significantly lower potency of 1c
against hDHFR compared to TMQ or PTX which
contain a 5-methyl moiety and are capable of significant
contact with hydrophobic residues of hDHFR.
Modeling studies of this compound in the active site
of pcDHFR made by a least-squares fit of the backbone
structure pcDHFR from the furopyrimidine ternary
complex24 to this structure reveals that there are similar
hydrophobic contacts present; however, they are more
compact than in this human structure. In addition, the
substitution of Val115 in hDHFR by Ile123 in pcDHFR
permits closer intermolecular hydrophobic contacts to
be made via the N9-methyl moiety of 1c and pcDHFR
and could account for part of the enhanced potency
observed for this analogue against pcDHFR.
Ch em istr y
The target compounds 2-15 were synthesized (Scheme
1) from commercially available 2,4-diamino-6-nitro-
quinazoline 16 which was reduced with Raney nickel
and hydrogen at 30 psi, to afford the key intermediate
2,4,6-triaminoquinazoline 17. The appropriate benzal-
dehyde or naphthaldehyde was then reductively ami-
nated with 17 under hydrogen at 30 psi, to afford target
compounds 2-10 in yields ranging from 26 to 61%.
Reductive methylation of 3, 4, 5, 7, and 10, using
formaldehyde and sodium cyanoborohydride at a pH of
2-3, regiospecifically afforded the corresponding N9-
methyl analogues 11-15 in yields ranging from 89 to
95%.
Biologica l Activity a n d Discu ssion
Analogues 2-15 were evaluated as inhibitors of
pcDHFR,25 tgDHFR,26 and rat liver (rl)DHFR, and the
results (IC50) are reported in Table 2. Selectivity ratios
(IC50 rlDHFR/IC50 pcDHFR and IC50 rlDHFR/IC50 tgDH-
FR) were determined using rlDHFR as the mammalian
source and are also reported in Table 2. The choice of
side chain substituents for these quinazoline analogues
was based on the potent and selective pyrido[2,3-d]-
pyrimidine series reported by Gangjee et al.11 Com-
parison of the inhibitory activity of analogues 3-9 with
2 (Table 2) indicates that introduction of methoxy
groups in the side chain phenyl ring slightly improves
inhibition of pcDHFR and tgDHFR except for 9, the only
compound in the group with diortho substitution. The
rlDHFR inhibitory potency decreases slightly on addi-
tion of methoxy groups on the side chain phenyl ring;
thus, analogues 2-9 were selective inhibitors of tgDH-
FR versus rlDHFR. Despite the differential change in
the inhibition of pcDHFR and rlDHFR on introduction
of methoxy groups, selective inhibitors of pcDHFR were
not obtained.
X-r a y Cr ysta l Str u ctu r e
The overall characteristics of the hDHFR ternary
complex with NADPH and 1c are similar to those
reported for other hDHFR complexes, and the binding
orientation of the analogue 1c is similar to that observed
for PTX and TMQ.21 The carboxylate oxygens of Glu30
interact with the pyridopyrimidine ring forming contacts
with the N1 nitrogen by OE2 and OE1 to the 2-NH2
nitrogen. In addition, the carboxylate oxygen OE1 of
Glu30 forms a hydrogen bond to the hydroxyl of Thr136,
which in turn interacts with a conserved water and also
with the 2-NH2 group of 1c. There is a further hydrogen-
bonding network formed by another conserved water
molecule which forms contacts with OE2 of Glu30, the
N8 nitrogen of 1c, and NE1 of Trp24. These patterns
have been observed in most DHFR crystal structures.21-23
These data provide the first structural observation of
a reversed N9-C10 bridge in normal folate analogues

3430 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18
Gangjee et al.
Ta ble 2. Inhibitory Concentrations (IC50) in µM and
Selectivity Ratios of 2-15 against P. carinii (pc), T. gondii (tg),
and Rat Liver (rl)DHFRa
N9-methylation of compounds 3-5, 7, and 10 to
afford 11-15, respectively, results in a significant
increase in inhibition of all three DHFR. Thus, N9-
methylation of 3 to afford 11 results in a 53-, 5-, and
42-fold increase in inhibition of pcDHFR, tgDHFR, and
rlDHFR, respectively. However, as a result of the
greater increase in the inhibition of rlDHFR, its selec-
tivity against tgDHFR was decreased. N9-methylation
of 4 to afford 12 resulted in a 92-, 13-, and 102-fold
increase in the inhibition of pcDHFR, tgDHFR, and
rlDHFR, respectively, and provided the most potent
inhibitor of tgDHFR and rlDHFR of the entire series of
compounds in this study. Compound 12 is as potent as
TMQ against pcDHFR and tgDHFR. N9-methylation
of 5 affords 13 which shows a 44-, 4-, and 28-fold
increase in inhibition of pcDHFR, tgDHFR, and rlDH-
FR, respectively. The differential increase in DHFR
inhibition indicates that the active site of pcDHFR and
tgDHFR are quite different, in particular with respect
to the interaction of the N9-methyl moiety with the
enzymes and/or the effect of the N9-methyl group on
the conformation of the bridge and the side chain phenyl
ring. Comparison of the inhibition profiles of com-
pounds 5 and 7, and 13 and 14, suggests that the
addition of the third methoxy group does not afford a
beneficial effect for the inhibition of pcDHFR or tgDH-
FR, implying that it does not interact with any residue
in the active site. This information could be used to
impart beneficial properties to these molecules such as
improved solubility using substituents at this site for
the inhibition of P. carinii and/or T. gondii cells in
culture.
pc
8.7
4.6
2.2
4.4
6.8
4.9
5.4
rl
rl/pc
tg
rl/tg
2
3
4
5
6
7
8
9
0.66
1.1
0.84
1.2
0.9
1.3
0.08
0.24
0.38
0.27
0.13
0.27
0.29
0.19
0.26
0.29
0.34
0.43
0.36
1
0.32
0.16
0.12
0.17
0.084
0.19
0.124
0.95
0.099
0.03
0.0093
0.039
0.017
0.021
0.01
2.1
6.88
7
7.1
10.7
6.8
12.9
23.9
1.9
0.86
0.88
1.1
1.1
0.80
0.3
1.6
116
22.7
0.19
0.026
0.0082
0.043
0.019
0.017
0.003
133
10
11
12
13
14
15
TMQb
TMPb
0.72
0.087
0.0238
0.10
0.052
0.017
0.042
12
0.07
11.1
2.7
49
a
These assays were carrried out at 37 °C under saturating
conditions of substrate (90 µM dihydrofolic acid) and cofactor (119
µM NADPH); tgDHFR and rlDHFR were assayed in the presence
b
of 150 mM KCl.25,26 Data from ref 11.
Introduction of a 2,4,6-trimethoxy substituent in the
side chain phenyl (compound 9) results in a 13-, 3-, and
26-fold decrease in the inhibition of pcDHFR, tgDHFR,
and rlDHFR, respectively, compared with 2. A com-
parison of the inhibitory activity of 5 (2,4-diOCH3 Ph)
and 9 (2,4,6-triOCH3Ph) illustrates the detrimental
effect on DHFR inhibition with the introduction of an
additional ortho methoxy moiety on the phenyl ring.
This decreased activity of 9 could be attributed to an
intolerance of the enzyme for diortho substituents and/
or a change in the conformation of the side chain phenyl
ring, resulting from the bulk of a diortho substitution
which is not conducive to binding. As a result of its
significantly diminished rlDHFR potency compared to
tgDHFR, analogue 9 displayed a 23-fold selectivity for
tgDHFR and was the most selective inhibitor of tgDHFR
in this entire series versus rlDHFR. In addition the
decrease in DHFR inhibitory activity with a 2,4,6-
trimethoxy-substituted phenyl was the least for tgDH-
FR, indicating a greater tolerance for di-orthomethoxy
substituents.
On the basis of their superior selectivity toward
tgDHFR, compounds 6, 8, and 9 were tested as inhibi-
tors of the growth of T. gondii cells in culture; the IC50
values were 12, 11, and 3 µM, respectively. Thus the
IC50 T. gondii cell culture/IC50 tgDHFR ratio for 6, 8,
and 9 was 142, 89, and 3, respectively, which was
significantly better than the ratio of 447 for the pyrido-
[2,3-d]pyrimidine 1c. For comparison, the IC50 value
for pyrimethamine in the same assay was 0.64 µM.26
Replacement of the phenyl ring of compound 2 with
a 1-naphthyl moiety (10) provided interesting results.
The activity against all three DHFR was increased. The
active site of hDHFR has been shown to be large enough
to accommodate a naphthyl ring in the side chain;27
hence the increased inhibition of rlDHFR with a naph-
thalene side chain was not unexpected. The increased
inhibition of pcDHFR and tgDHFR as well by analogue
10 suggests that the size of the active sites of pcDHFR
and tgDHFR are also large enough to accommodate a
naphthyl moiety. However the selectivity for pcDHFR
and tgDHFR was not improved compared to the phenyl
unsubstituted analogue 2.
A comparison of the activities and selectivities (versus
rlDHFR) of these quinazoline analogues with the pyrido-
[2,3-d]pyrimidines11 with identical side chain substit-
uents reveals some differences. In the latter series, the
N9-methyl-2′,5′-diOCH3Ph analogue 1c was among the
most potent and selective against tgDHFR, and for the
other analogues, N9-methylation did not change the
selectivity. In contrast, the most selective analogues in
the quinazoline series were the N9-H analogues, and
contrary to the pyrido[2,3-d]pyrimidines, N9-methyla-
tion of the quinazolines abolished selectivity. These
N9-H quinazoline analogues were, in general, more
selective for tgDHFR than the corresponding pyrido-
[2,3-d]pyrimidines. With a few exceptions, the quinazo-
lines were also more potent than the pyrido[2,3-d]-
pyrimidines.
Analogues 2-15 were also evaluated against human
(h)DHFR, Escherichia coli (ec)DHFR, Lactobacillus
casei (lc)DHFR, as well as recombinant tgDHFR, under
slightly different assay conditions,28 and the results are
reported in Table 3, along with selectivity ratios versus
hDHFR. TMP and TMQ are included for comparison.
N9-methylation resulted in a dramatic increase in the
inhibition of hDHFR (as was observed for rlDHFR).
Thus the N9-methyl analogues 11 and 12 were 1077-
and 2545-fold more potent against hDHFR than the
corresponding N9-H compounds 3 and 4, respectively.
The inhibition of tgDHFR was also significantly in-
creased on N9-methylation, with 11 and 12 being 282-
and 179-fold more potent than 3 and 4, respectively.
Several analogues were identified as selective inhibitors
of tgDHFR with selectivity ratios of 1.7-33. Further,

Synthesis of 2,4-Diamino-6-Substituted Quinazolines
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18 3431
Ta ble 3. Inhibitory Concentrations (IC50) in µM and
augmented on N9-methylation where the quinazoline
analogues 11 and 12 are 327- and 290-fold more potent
against hDHFR than the corresponding pyrido[2,3-d]-
pyrimidines 1c and 1d , respectively. However, com-
pounds 11 and 12 were only 6-fold more potent as
inhibitors of tgDHFR than the corresponding pyrido-
[2,3-d]pyrimidines. These results indicate that the
presence of a quinazoline moiety is much more condu-
cive to potent inhibition of hDHFR than to tgDHFR
when compared with the corresponding pyrido[2,3-d]-
pyrimidines. It is this increased inhibition of hDHFR
that is responsible for the considerable loss of selectivity
against tgDHFR for the quinazolines.
Selectivity Ratios of Analogues 2-15 against Human (h)DHFR,
T. gondii (tg)DHFR, E. coli (ec)DHFR, and L. casei (lc)DHFRa
hDHFR tgDHFR h/tg ecDHFR h/ec lcDHFR h/lc
2
3
4
5
6
7
8
9
2.4
28
28
26
23
27
1.4
25
1.4
1.1
1.7
25
33
33
25
33
2.6
33
6.6
0.51
0.42
0.42
0.39
0.23
0.41
0.14
1.0
4.7
67
67
67
100
66
10
25
4
2.7
3.0
2.3
2.1
2.8
2.0
1.6
37
1.8
0.13
0.037 0.3
0.12 0.4
0.9
9.3
0.84
0.78
0.92
0.81
0.54
0.75
0.09
12.2
12.4
8.2
13.5
0.9
0.7
0.3
0.2
10
11
12
13
14
15
TMP
0.6
0.15
0.0078 3.3
0.0028 3.9
0.026
0.011
0.05
0.14
0.027
340
0.0039 6.6
0.0047 2.3
0.0039 12.8 0.013
0.0054 26
0.0041 6.6
6.8
3.8
The significant and selective augmentation of hDHFR
inhibition, attributed to the N9-methyl moiety of the
quinazolines as compared to the pyrido[2,3-d]pyrim-
idines (shown in Table 4), could result from a confor-
mational effect that positions the N9-methyl moiety in
the quinazolines to interact with residues on hDHFR
as well as other DHFR. This conformational positioning
of the N9-methyl moiety is probably not possible to the
same extent in the pyrido[2,3-d]pyrimidines, possibly
due to the N8-nitrogen. The N8-nitrogen in the
pyrido[2,3-d]pyrimidines has been shown to interact
through a hydrogen bridge via a conserved water
molecule to Glu30 and Trp24 in hDHFR, thus restrict-
ing the ability to position the N9-methyl in a conducive
orientation in hDHFR. Similar hydrogen bonding with
a water molecule in hDHFR has been observed for the
classical antitumor agent methotrexate and PTX.14,23
This could also explain, in part, the different structure-
activity and selectivity profiles observed with quinazo-
line analogues 2-15 as compared with the correspond-
ing pyrido[2,3-d]pyrimidines, since compounds with
identical side chains differ only in the substitution at
the 8-position in the two series. This hypothesis is
partially validated from the crystal structure of the
pyrido[2,3-d]pyrimidine analogue 1c with hDHFR as
shown in Figure 1, which indicates no van der Waals
or hydrophobic contact of the N9-methyl moiety with
Ile114 or Val115, thus accounting in part for its lower
inhibitory activity against hDHFR compared to other
DHFRs. In addition, the presence of a water molecule
which interacts with the N8-nitrogen (Figure 1) and
provides a bridge to Glu30 and Trp24 in hDHFR also
restricts the positioning of the N-methyl moiety as
suggested previously. Thus, the lack of high selectivity
in the quinazolines synthesized in this study as com-
pared to that of the pyrido[2,3-d]pyrimidines could arise,
in part, from a productive hydrophobic interaction of the
N9-methyl moiety of the quinazolines with hDHFR as
well as other DHFR. In addition, the conformational
effects of N9-methylation on the relative positioning
of the heterocyclic, as well as the phenyl ring and the
bridge, could play an important role in the lack of high
selectivity of these quinazoline analogues. Confirmation
of these hypotheses must await the crystal structures
of 11 with pcDHFR and hDHFR which is currently
underway.
0.0081 17
0.0054
0.01
0.059 2.4
0.021 1.3
5
50
34000 0.26
1308
TMQ 0.054
0.0072 7.5
0.018
3.0 0.054 1.0
a
Recombinant hDHFR was provided by Dr. J . H. Freisheim.
Recombinant tgDHFR was provided by Dr. D. V. Santi. Recom-
binant ecDHFR was provided by Dr. R. L. Blakley. DHFR assay
conditions:28 all enzymes were assayed spectroscopically in a
solution containing 50 µM dihydrofolate, 80 µM NADPH, 0.05 µM
Tris HCl, 0.001 M 2-mercaptoethanol, and 0.001 M EDTA at pH
7.4 and 30 °C. The reaction was initiated with an amount of
enzyme yielding a change in o.d. at 340 nm of 0.015/min.
Ta ble 4. Comparison of the HDHFR and TgDHFR Inhibition
(IC50) in µM and Calculated log P Values of 3, 6, 11, and 12
with the Corresponding Pyrido[2,3-d]pyrimidines 1a -1d 11
hDHFR tgDHFR h/tg log Pa
3
28
13
23
1.1
0.5
0.92
25
26
25
2.35
1.32
1.91
0.89
1a (R d H; R′ ) 2,5-OCH3)
6
1b (R ) H; R′ ) 3,4,5-OCH3) 76
11 0.026
1c (R ) CH3; R′ ) 2,5-OCH3)
12
0.395
0.0039
0.028
0.0047
0.029
192
6.6 2.90
1.87
2.3 2.90
8.5
0.01
2.9
304
1d (R ) CH3; R′ ) 3,5-OCH3)
100
1.87
a
Calculated using log KOWsestimation of log octanol/water
partition coefficient version 1.03sSyracuse Research Corporation.
analogues 11-15 were as potent as TMQ against
tgDHFR with equivalent or better selectivities. How-
ever, the remarkable selectivities observed against
tgDHFR with the pyrido[2,3-d]-pyrimidines, some as
high as 300-fold,11 were absent in the quinazoline series.
The selectivities and potencies against ecDHFR were
in general slightly greater for the N9-H analogues and
slightly lower for the N9-methyl analogues compared
to their effect on tgDHFR. Against lcDHFR the ana-
logues were less potent and selective compared to their
activities against tgDHFR.
Since compounds 2-15 were designed on the basis of
the previously synthesized pyrido[2,3-d]pyrimidines11
with the objective of increasing cell penetration, it was
of interest to determine the effects of the replacement
of the nitrogen (N8 of a pyrido[2,3-d]pyrimidine) with
a carbon (C8 of a quinazoline) on DHFR inhibition. A
comparison of the hDHFR and tgDHFR inhibitory
activities along with selectivity ratios of four quinazo-
lines 3, 6, 11, and 12 synthesized in this study, with
the previously synthesized pyrido[2,3-d]pyrimidines
1a -d ,11 respectively, and with identical substituents on
the side chain, are listed in Table 4. With the exception
of 3 and 1a , the hDHFR inhibition results indicate that
the quinazoline analogues are more potent than the
corresponding pyrido[2,3-d]pyrimidines. This effect is
The significant hDHFR inhibition of these quinazoline
analogues prompted their selection by the National
Cancer Institute in the antitumor in vitro preclinical
screening program.29 The remarkable increment in
hDHFR inhibition on N9-methylation was reflected in

3432 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18
Gangjee et al.
addition of 10-15 mL of glacial acetic acid, and the mixture
was further hydrogenated for 6-7 h at 35 psi and room
temperature. TLC analysis (CHCl3:CH3OH:NH4OH 5:2:1
drop) indicated the presence of a new product along with some
unreacted 2,4,6-triaminoquinazoline. The reaction was filtered
through a Celite pad, and the pad was washed repeatedly with
DMF until the washings were colorless. Silica gel (1.0 g) was
added to the filtrate, and the solvents evaporated (40 °C, 0.1
mmHg) to afford a dry plug. This plug was applied on the
surface of a 1.05 in. × 23 in. silica gel column and eluted with
CHC13:CH3OH using a gradient elution (95:5 to 80:20). Frac-
tions containing pure product (TLC) were pooled and evapo-
rated to afford analytically pure compound.
P r oced u r e B: Gen er a l P r oced u r e for th e P r ep a r a tion
of Com p ou n d s 11-15. To 2,4-diamino-6-[(N-substituted-
benzyl)amino]quinazoline (0.l g, 0.48 mmol) in 25 mL of
acetonitrile was added formaldehyde (0.04 g, 1.l mmol) with
constant stirring. To this suspension was added sodium
cyanoborohydride (0.05 g, 0.84 mmol), and the pH of the
mixture was adjusted to 2-3 using concentrated HCl. The
starting material initially went into solution at pH 2-3, and
after 5 min a bright yellow precipitate was obtained. TLC
(CHCl3:CH3OH:NH4OH-5:2:0.5) showed the presence of a new
spot with the absence of the starting material spot. The
acetonitrile was evaporated under reduced pressure and the
residue obtained was suspended in 10 mL of water. This
suspension was neutralized using NH4OH to afford the N9-
methyl derivative of the appropriate 2,4-diamino-6-[(N-
substituted-benzyl)amino]quinazoline.
the superior cytotoxicity of the N9-methyl analogues
as compared to the corresponding N9-H analogues.
Compounds 11, 12, and 14 displayed potent inhibitory
activity against the growth of various tumor cell lines
in culture with GI50 < 10-8 M. The cytotoxic properties
of these compounds can be attributed to their excellent
hDHFR inhibition.
In summary, several potent and selective inhibitors
of tgDHFR have been synthesized in this study. The
goal of this work was to improve the cell penetration of
highly selective pyrido[2,3-d]pyrimidines synthesized
previously, while maintaining their potency and selec-
tivity. This goal was partially achieved, as the synthe-
sized quinazolines had considerably improved calculated
lipophilicity parameters and a decrease in the IC50 T.
gondii cell culture/IC50 tgDHFR inhibitory ratio com-
pared to 1c as well as high activity against tumor cells
in culture. However, the replacement of the N8-
nitrogen in pyrido[2,3-d]pyrimidines with a carbon in
the quinazolines resulted in an unanticipated loss of the
excellent selectivity achieved with the pyrido[2,3-d]-
pyrimidines. As a result, alternate strategies to im-
prove the cell penetration of the pyrido[2,3-d]pyrimidine
analogues are currently underway and will be the
subject of future communications. Several of the ana-
logues synthesized in this study displayed exceptional
cytotoxic properties in the preclinical in vitro antitumor
screening program of the National Cancer Institute. The
potencies of these compounds against the growth of a
wide range of tumor cell lines in culture warrants their
further evaluation as antitumor agents. The goal of
developing antifolates as potent and selective inhibitors
of pcDHFR and tgDHFR with good cell penetration,
however, remains a challenge and in light of the AIDS
pandemic continues to be a worthwhile goal.
2,4-Dia m in o-6-[N-(ben zyl)a m in o]qu in a zolin e, 2. Com-
pound 2 was synthesized from 2,4-diamino-6-nitroquinazoline
(0.50 g, 2.43 mmol) and benzaldehyde (0.57 g, 2.91 mmol) to
afford after purification 0.43 g (55%) of 2 as a bright yellow
powder: mp 177.3 °C; TLC Rf 0.53 (CHCl3:CH3OH:NH4OH-
1
5:2:0.5); H NMR (DMSO-d6) δ 4.29 (d, 2 H, 10-CH2), 5.86 (s,
2 H, 4-NH2), 5.99 (d, 1 H, 9-NH), 6.98-7.43 (m, 10 H, Ar,
2-NH2). Anal. calcd for (Cl5Hl5N5‚0.5CH3COOH) C, H, N.
2,4-Dia m in o-6-[N-(2′,5′-d im eth oxyben zyl)a m in o]qu in -
a zolin e, 3. Compound 3 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 2,5-dimethoxy-
benzaldehyde (0.60 g, 3.65 mmol) using procedure A to afford
after purification 0.18 g (23.29%) of 3 as a yellow solid: mp
171-172 °C; TLC Rf 0.57 (CHCl3:CH3OH:NH4OH-5:2:0.5); 1H
NMR (DMSO-d6) δ 3.64 (s, 3 H, OCH3), 3.77 (s, 3 H, OCH3),
4.23 (d, 2 H, 10-CH2), 5.90 (t, 1 H, 9-NH), 6.34 (s, 2 H, 4-NH2),
6.78 (t, 1 H, Ar), 6.92 (t, 2 H, Ar), 7.09 (m, 3 H, 5,7,8-CH),
7.66 (s, 2 H, 2′-NH2). Anal. calcd for (C17H19N5O2‚0.5H2O‚
0.4CH3COOH) C, H, N.
2,4-Dia m in o-6-[N-(3′,5′-d im eth oxyben zyl)a m in o]qu in -
a zolin e, 4. Compound 4 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol), and 3,5-dimethoxy-
benzaldehyde (0.45 g, 2.68 mmol) using procedure A to afford
after purification 0.50 g (63%) of 4 as a light yellow solid: mp
182-183 °C; TLC Rf 0.61 (CHCl3:CH3OH:NH4OH-5:2:0.5); 1H
NMR (DMSO-d6) δ 3.70 (s, 6 H, 3′,5′-OCH3), 4.20 (d, 2 H, 10-
CH2), 5.85 (s, 2 H, 4-NH2), 6.00 (t, 1 H, 9-NH2), 6.34 (m, 1 H,
-Ar), 6.59 (s, 2 H, Ar), 7.04 (m, 3 H, 5,7,8-CH), 7.14 (s, 2 H,
2-NH2). Anal. calcd for (C17H19N5O2‚0.5CH3COOH) C, H, N.
2,4-Dia m in o-6-[N-(2′,4′-d im eth oxyben zyl)a m in o]qu in -
a zolin e, 5. Compound 5 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 2,4-dimethoxy-
benzaldehyde (0.45 g, 2.68 mmol) using procedure A to afford
after purification 0.25 g (32%) of 5 as a light brown solid: mp
204-205 °C; TLC Rf 0.59 (CHCl3:CH3OH:NH4OH-5:2:0.5); 1H
NMR (DMSO-d6) δ 3.70 (s, 3H, OCH3), 3.79 (s, 3 H, OCH3),
4.16 (d, 2 H, 10-CH2), 5.67 (t, 1 H, 9-NH), 5.92 (s, 2 H, 4-NH2),
6.70 (d, 1 H, Ar), 7.05 (m, 4 H, Ar), 7.28 (s, 2 H, 2NH2). Anal.
calcd for (C17H19N5O2‚0.5H2O‚0.7CH3COOH) C, H, N.
2,4-Diam in o-6-[N-(3′,4′,5′-tr im eth oxyben zyl)am in o]qu in -
a zolin e, 6. Compound 6 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mol), and 3,4,5-trimethoxy-
benzaldehyde (0.70 g, 3.65 mmol) using procedure A to afford
after purification 0.38 g (44%) of 6 as a bright yellow solid:
mp 193-194 °C; TLC Rf 0.52 (CHCl3:CH3OH:NH4OH-5:2:0.5);
Exp er im en ta l Section
Melting points were determined on a Fisher-J ohns melting
point apparatus or a Mel Temp apparatus and are uncorrected.
Nuclear magnetic resonance spectra for proton (1H NMR) were
recorded on a Brucker WH-300 (300 MHz). The data were
accumulated by 16K size with 0.5 s delay time with internal
standard tetramethylsilane; s ) singlet, br s ) broad singlet,
d ) doublet, t ) triplet, q ) quartet, m ) multiplet. Thin
layer chromatography (TLC) was performed on silica gel plates
with fluorescent indicator and were visualized with light at
254 and 366 nm, unless indicated otherwise. Column chro-
matography was performed with 230-400 mesh silica gel
purchased from Aldrich Chemical Co., Milwaukee, Wisconsin.
Elution was performed using a gradient, and 10 mL fractions
were collected, unless mentioned otherwise. All anhydrous
solvents were purchased from Aldrich Chemical Co. and were
used without further purification. Samples for micro analysis
were dried in vacuo over phosphorus pentoxide at 70 or 110
°C. Fractional moles of water or organic solvents, frequently
found in some analytical samples of the antifolates, could not
be prevented despite drying in vacuo and were confirmed by
1
their presence in the H NMR spectrum. Microanalyses were
performed by Atlantic Microlabs, Norcoss, Georgia.
P r oced u r e A: Gen er a l P r oced u r e for th e Syn th esis of
Com p ou n d s 2-10. 2,4-Diamino-6-nitroquinazoline (1 equiv)
was dissolved in 100 mL of N,N-dimethylformamide with
warming (75 °C). To this solution was added Raney nickel
(1.0 g), and the mixture was hydrogenated in a Parr hydro-
genation apparatus at 35 psi and room temperature for 3 h.
TLC analysis (CHCl3:CH3OH:NH4OH 5:2:1 drop) indicated
absence of starting material and the appearance of a new spot,
corresponding to 2,4,6-triaminoquinazoline. The appropriate
benzaldehyde (1.2 equiv) was then added, followed by the

Synthesis of 2,4-Diamino-6-Substituted Quinazolines
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18 3433
1H NMR (DMSO-d6) δ 3.61 (s, 3 H, 4′-OCH3), 3.74 (s, 6 H, 3′,5′-
OCH3), 4.20 (d, 2 H. 10-CH2), 6.08 (t, 1 H, 9-NH), 6.17 (s, 2
H, 4-NH2), 6.77 (s, 2 H, 2′, 6′-CH), 7.08 (d, 3 H, 5,7,8-CH), 7.45
(s, 2 H, 2NH2). Anal. calcd for (C18H21N5O3‚0.7H2O‚1.1CH3-
COOH) C, H, N.
2,4-Diam in o-6-[N-(2′,3′,4′-tr im eth oxyben zyl)am in o]qu in -
a zolin e, 7. Compound 7 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 2,3,4-trimethoxy-
benzaldehyde (0.57 g, 2.91 mmol) using procedure A to afford
after purification 0.23 g (27%) of 7 as a yellow solid: mp 213-
214 °C; TLC Rf 0.49 (CHCl3:CH3OH:NH4OH-5: 2:0.5); 1H NMR
(DMSO-d6) δ 3.75 (s, 6 H, OCH3), 3.76 (s, 3 H, OCH3), 4.16 (d,
2 H, 10-CH2), 5.59 (t, 1 H, 9-NH), 5.70 (s, 2 H, 4-NH2), 6.76 (d,
1 H, Ar), 7.04-7.11 (m, 6 H, Ar, 2-NH2). Anal. calcd for
(C18H2lN5O3‚0.2CH3COOH) C, H, N.
2,4-Diam in o-6-[N-(2′,4′,5′-tr im eth oxyben zyl)am in o]qu in -
a zolin e, 8. Compound 8 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 2′,4′,5′-trimethoxy-
benzaldehyde (0.57 g, 2.92 mmol) using procedure A to afford
after purification 0.28 g (32%) of 8 as a yellowish white solid:
mp 195-196 °C; TLC Rf 0.52 (CHCl3:CH3OH:NH4OH-5:2:0.5);
1H NMR (DMSO-d6) δ 3.61 (s, 3 H, OCH3), 3.74 (s, 3H, OCH3),
3.77 (s, 3 H, OCH3), 4.10 (d, 2 H, 10-CH2), 5.46 (t, 1 H, 9-NH),
5.57 (s, 2 H, 4-NH2), 6.66 (s, 1 H, Ar), 7.02 (m, 6 H, Ar, 2-NH2).
Anal. calcd for (C18H21N5O3‚0.25CH3COOH) C, H, N.
2,4-Diam in o-6-[N-(2′,4′,6′-tr im eth oxyben zyl)am in o]qu in -
a zolin e, 9. Compound 9 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 2,4,6-trimethoxy-
benzaldehyde (0.57 g, 2.91 mmol) using procedure A to afford
after purification 0.34 g (39%) of 9 as a brownish green solid:
mp 193-194 °C; TLC Rf 0.55 (CHCl3:CH3OH:NH4OH-5:2:0.5);
1H NMR (DMSO-d6) δ 3.77(m, 9 H, 2′,4′,6′-OCH3), 4.05 (s, 2
H, 10-CH2), 5.04 (s, 1 H, 9-NH), 5.99 (s, 2 H, 4-NH2), 6.26 (s,
2 H, 3′,5′-CH), 7.05 (m, 3 H, 5,7,8-CH). Anal. calcd for
(C18H2lN5O3‚0.4H2O‚0.55CH3COOH) C, H, N.
2 H, 4-NH2), 6.76 (m, 2 H, Ar), 7.06-7.23 (m, 4 H, Ar), 7.76 (s,
2 H, 2-NH2). Anal. calcd for (C18H2lN5O2‚0.5H2O‚1.5 HCl) C,
H, N.
2,4-Dia m in o-6-[N-(2′,3′,4′-tr im eth oxyben zyl)(m eth yl)-
a m in o]qu in a zolin e, 14. Compound 14 was synthesized from
7 (0.10 g 0.28 mmol), formaldehyde (0.04 g, l.l mmol), and (0.05
g, 0.84 mmol) using procedure B to afford 0.09 g (91%) of 14
as a yellow solid: mp 180-181 °C; TLC Rf 0.49 (CHCl3:CH3-
OH:NH4OH 5:2:0.5); 1H NMR (DMSO-d6) 2.91 (s, 3 H, N-CH3),
3.73 (s, 6 H, OCH3), 3.78 (s, 3 H, OCH3), 4.47 (s, 2 H, 10-CH2),
6.31 (s, 2 H, 4-NH2), 6.70 (s, 2 H, Ar), 7.25 (m, 3 H, 5,7,8-CH),
7.70 (s, 2 H, 2-NH2). Anal. calcd for (C19H23N5O3‚0.2CH3-
COOH‚0.3H2O) C, H, N.
2,4-Dia m in o-6-[N-(2′,3′-n a p h th yl)(m eth yl)a m in o]qu in -
a zolin e, 15. Compound 15 was synthesized from 10 (0.10 g,
0.28 mmol), formaldehyde (0.04 g, l.l mmol), and (0.05 g, 0.84
mmol) using procedure B to afford 0.09 g (87%) of 15 as a
yellow solid: mp 262-263 °C; TLC Rf 0.47 (CHCl3:CH3OH:
NH4OH-5:2:0.5); 1H NMR (DMSO-d6) δ 3.06 (s, 3 H, N-CH3),
5.09 (s, 2 H, 10-CH2), 6.93 (s, 2 H, 4-NH2), 7.14-8.09 (m, 10
H, Ar), 8.24 (s, 2 H, 2-NH2). Anal. calcd for (C20H19N5‚0.5H2O‚
0.5HCl) C, H, N.
Str u ctu r e Deter m in a tion a n d Refin em en t. hDHFR
was isolated and purified by Blakley as described30 and
crystals were grown by the hanging drop vapor diffusion
method. The protein solution of hDHFR was incubated
overnight at 4 °C with NADPH followed by the inhibitor, 1c.
Protein droplets contained 62% ammonium sulfate in 0.1 M
phosphate buffer, pH 7.0. Crystals grew in three weeks time
and are rhombohedral, space group R3 with hexagonal index-
ing, and diffract to 2.1 Å resolution. The lattice constants for
the ternary complex of hDHFR-NADPH-1c are 86.259 and
77.637 Å. Data were collected at room temperature on an
Rigaku RaxisIIc area detector and the data processed with the
Rigakau software. The Rmerge for 2σ data was 4.96% with a
3-fold redundancy. The overall completeness of the data was
86.4% and 63.6% for data in the shell between 2.0 and 2.1 Å
The data for the ternary complex refined to 20.2% for data to
2.1 Å resolution are included in the Supporting Information.
2,4-Dia m in o-6-[N-(1′-n a p h th ylm eth ylen e)a m in o]qu in -
a zolin e, 10. Compound 10 was synthesized from 2,4-diamino-
6-nitroquinazoline (0.50 g, 2.43 mmol) and 1-naphthaldehyde
(0.46 g, 2.92 mmol) using procedure A to afford after purifica-
tion 0.40 g (53%) of 10 as a yellow solid: mp 214-215 °C; TLC
The structure was solved by molecular replacement methods
using the coordinates of hDHFR-NADPH-MTX.22 Inspection
of the resulting difference electron density map using the
program CHAIN31 running on a Silicon Graphics Elan Work-
station revealed density for a ternary complex. The final cycles
of refinement were carried out with the inhibitor 1c and
cofactor NADPH using the program PROLSQ.32a,b The Ram-
achandran conformational paramaters for the last cycle of
refinement generated by PROCHECK33 shows that 94.3% of
the residues have the most favored conformation and none are
in the disallowed regions.
1
Rf 0.47 (CHCl3:CH3OH:NH4OH-5:2:0.5); H NMR (DMSO-d6)
δ 4.73 (d, 2 H, 10-CH2), 5.83 (s, 2 H, 4-NH2), 6.03 (t, 1 H, 9-NH),
7.06-7.14 (m, 3 H, 5,7,8-CH), 7.24 (s, 2 H, 2-NH2), 7.44-7.60
(m, 4 H, Ar), 7.84 (d, 1 H, Ar), 7.94 (d, 1 H, Ar), 8.15 (d, 1H,
Ar). Anal. calcd for (C19H17N5‚0.1H2O‚0.2CH3COOH) C, H,
N.
2,4-Diam in o-6-[N-(2′,5′-dim eth oxyben zyl)-N-m eth ylam i-
n o]qu in a zolin e, 11. Compound 11 was synthesized from 3
(0.10 g, 0.28 mmol), formaldehyde (0.04 g, l.l mmol), and
sodium cyanoborohydride (0.05 g, 0.84 mmol) using procedure
B to afford 0.09 g (81.4%) of 11 as a bright yellow solid: mp
143-144 °C; TLC Rf 0.57 (CHCl3:CH3OH:NH4OH-5:2:0.5); 1H
NMR (DMSO-d6) δ 2.93 (s, 3 H, N-CH3), 3.55 (s, 3 H, OCH3),
3.73 (s, 3 H, OCH3), 4.46 (s, 2 H, 10-CH2), 5.98 (s, 2 H, 4-NH2),
6.48 (d, 1 H, Ar), 6.72 (q, 1 H, Ar), 6.88 (s, 2 H, Ar), 7.11 (s, 2
H, Ar), 7.15 (s, 1 H, Ar), 7.40 (s, 2 H, 2-NH2). Anal. calcd for
(C18H21N5O2‚0.25H2O‚1.0 HCl) C, H, N.
Further refinement was continued for the ternary complex
and the model was manually adjusted to fit its electron density
values and verified by a series of OMIT maps calculated from
the current model with deleted fragments. The highest
thermal parameters (30-50 Å2) of the protein residues are
near the binding entrance and the flexible loop regions. The
remaining side chain thermal parameters range from 10 to
25 Å2. In the final stages of refinement, 102 water molecules
were added in accord with the criteria of good electron density
and acceptable hydrogen bonds to other atoms. The final
refinement statistics are available in Supporting Information
and available upon request. Coordinates have been deposited
with the Protein Data Bank (1BOZ).
2,4-Diam in o-6-[N-(3′,4′-dim eth oxyben zyl)-N-m eth ylam i-
n o]qu in a zolin e, 12. Compound 12 was synthesized from 4
(0.1 g, 0.28 mmol), formaldehyde (0.04 g, l.l mmol), and (0.05
g, 0.84 mmol) using procedure B to afford 0.09 g (86%) of 12
as a bright yellow solid: mp 230-231 °C; TLC Rf 0.61 (CHCl3:
1
CH3OH:NH4OH-5:2:0.5); H NMR (DMSO-d6) δ 3.26 (s, 3 H,
N-CH3), 3.64 (s, 6 H, 3′,5′-OCH3), 4.47 (s, 2 H, 10-CH2), 6.14
(s, 2 H, 4-NH2), 6.32 (s, 3-H, Ar), 7.21 (m, 3 H, 5,7,8-CH), 7.55
(s, 2 H, 2-NH2). Anal. calcd for (C18H21N5O2‚0.2H2O‚0.2CH3-
COOH) C, H, N.
Ack n ow led gm en t. This work was supported in part
by NIH Grants GM52811 (A.G.), GM40998 (A.G.), AI
41743 (A..G.), CA10914 (R.L.K.), and GM51670 (V.C.)
and NIH Contracts NO1-AI-87240 (S.F.Q.) and NO1-
AI-35171 (S.F.Q.), Division of AIDS. Human DHFR
(F31G) was kindly provided by Dr. Raymond L. Blakley,
St. J ude Children’s Research Hospital, Memphis, TN
38101.
2,4-Dia m in o-6-[N-(2′,4′-d im eth oxyben zyl)(m eth yl)a m i-
n o]-qu in a zolin e 13. Compound 13 was synthesized from 5
(0.1 g, 0.28 mmol), formaldehyde (0.04 g, l.l mmol), and (0.05
g, 0.84 mmol) using procedure B to afford 0.09 g (86%) of 13
as a yellow solid: mp 123-124 °C; TLC Rf 0.59 (CHCl3:CH3-
OH:NH4OH-5:2:0.5); 1H NMR (DMSO-d6) 2.91 (s, 3-H, NCH3),
3.73-3.82 (m, 6 H, 2′,4′-OCH3), 4.47 (s, 2 H, 10-CH2), 6.40 (s,

3434 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 18
Gangjee et al.
(17) Hynes, J . B.; Harmon, S. J .; Floyd, G. G.; Farrington, M.; Hart,
L. D.; Gale, G. R.; Washtein, W. L.; Susten, S. S.; Freisheim, J .
H. Chemistry and Antitumor Evaluation of Selected Classical
2,4-Diaminoquinazoline Analogues of Folic Acid. J . Med. Chem.
1985, 28, 209-215.
Su p p or tin g In for m a tion Ava ila ble: Refinement statis-
tics for the 1c-NADPH-hDHRF complex (1 page). Ordering
information is given on any current masthead page.
(18) Schornagel, J . H.; Chang, P. K.; Sciarini, L. J .; Moroson, B. A.;
Mini, E.; Cashmore, A. R.; Bertino, J . Synthesis and Evaluation
of 2,4-Diamino Quinazoline Antifolates with Activity Against
Methotrexate Resistant Human Tumor Cells. Biochem. Phar-
macol. 1984, 33, 3251-3255.
Refer en ces
(1) Presented in part at the 211th American Chemical Society
National Meeting, New Orleans, LA, March 24-28, 1996;
Abstr: MEDI 182.
(2) Mills, J .; Masur, H. AIDS-Related Infections. Sci. Am. 1991,
August, 50-57.
(19) Tripos Associates, Inc., 1699 S. Hanely Road, Suite 303, St.
Louis, MO 63144.
(3) Klepser, M. E.; Klepser, T. B. Drug Treatment of HIV-related
Opportunistic Infections. Drugs 1997, 53 (1), 40-73.
(4) Levine, S. J . Pneumocystis carinii. Clinics in Chest Med. 1996,
17 (4), 665-695.
(5) Allegra, C. J .; Kovacs, J . A.; Drake, J . C.; Swan, J . C.; Chabner,
B. A.; Masur, H. Activity of Antifolates against Pneumocystis
carinii Dihydrofolate Reductase and Identification of a Potent
New Agent. J . Exp. Med. 1987, 165, 926-931.
(6) Masur, H.; Polis, M. A.; Tuazon, C. V.; Ogota, A. D.; Kovacs, J .
A.; Katz, D.; Hilt, D.; Simmons, T.; Feuerstein, I.; Lundgren,
B.; Lane, H. C.; Chabner, B. A.; Allegra, C. J . Salvage Trial of
Trimetrexate-Leucovorin for the Treatment of Cerebral Toxo-
plasmosis. J . Infect. Dis. 1993, 167, 1422-1426.
(7) Kovacs, J . A.; Allegra, C. A.; Swan, J . C.; Drake, J . C.; Parillo,
J . E.; Chabner, B. A.; Masur, H. Potent Antipneumocystis and
Antitoxoplasma Activities of Piritrexim, a Lipid-soluble Anti-
folate. Antimicrob. Agents Chemother. 1988, 32, 430-433.
(8) Gordon, F. M.; Simin, G. L.; Wosby, C. B.; Mills, J . Adverse
Reactions to Trimethoprim-Sulfamethoxazole in Patients with
the Acquired Immunodeficiency Syndrome. Ann. Intern. Med.
1984, 100, 495-499.
(9) Gangjee, A.; Shi, J .; Queener, S. F.; Barrows, R. L.; Kisliuk, R.
L. Synthesis of 5-Methyl-5-Deaza Nonclassical Antifolate as
Inhibitors of Dihydrofolate Reductases and as Potential Anti-
Pneumocystis, Anti-Toxoplasma and Antitumor Agents. J . Med.
Chem. 1993, 36, 3437-3443.
(10) Gangjee, A.; Vasudevan, A.; Queener, S. F.; Kisliuk, R. L.
6-Substituted 2, 4-Diamino-5-Methyl-Pyrido[2,3-d]pyrimidines
as Inhibitors of Dihydrofolate Reductases from Pneumocystis
carinii and Toxoplasma gondii and as Antitumor Agents. J . Med.
Chem. 1995, 38, 1778-1785.
(11) Gangjee, A.; Vasudevan, A.; Queener, S. F.; Kisliuk, R. L. 2,4-
Diamino-5-Deaza-6-Substituted Pyrido[2,3-d]pyrimidines Anti-
folates as Potent and Selective Nonclassical Inhibitors of Dihy-
drofolate Reductases. J . Med. Chem. 1996, 39, 1438-1446.
(12) Gangjee, A.; Devraj, R.; McGuire, J . J .; Kisliuk, R. L.; Queener,
S. F.; Barrows, L. R. Classical and Nonclassical Furo[2,3-d]-
pyrimidines as Novel Antifolates: Synthesis and Biological
Activities. J . Med. Chem. 1994, 37, 1169-1176.
(13) Gangjee, A.; Mavandadi, F.; Queener, S. F.; McGuire, J . J . Novel
2,4-Diamino-5-Substituted Pyrrolo[2,3-d]pyrimidines as Classi-
cal and Nonclassical Antifolate Inhibitors of Dihydrofolate
Reductases. J . Med. Chem. 1995, 38, 2158-2165.
(20) Gangjee, A.; Vasudevan, A.; Queener, S. F. Synthesis and
Biological Evaluation of Nonclassical 2,4-Diamino-5-methyl-
pyrido[2,3-d]pyrimidines with Novel Side Chain Substituents as
Potential Inhibitors of Dihydrofolate Reductases. J . Med. Chem.
1997, 40, 479-485.
(21) Cody, V.; Wojtczak, A.; Kalman, T. I.; Freisheim, J . H.; Blakley,
R. L. Conformational Analysis of Human Dihydrofolate Reduc-
tase Inhibitor Complexes: Crystal Structure Determination of
Wild-Type and F31 Mutant Binary and Ternary Inhibitor
Complexes. Adv. Exp. Biol. Med. 1993, 481-486.
(22) Cody, V.; Luft, J . R.; Ciszak, E.; Kalman, T. I.; Freisheim, J . H.
Crystal Structure Determination at 2.3 Å of Recombinant
Human Dihydrofolate Reductase Ternary Complex with NADPH
and Methotrexate-γ-tetrazole. Anti-Cancer Drug Des. 1992, 7,
483-491.
(23) Blakley, R. L. Eukaryotic Dihydrofolate Reductase, Adv. Enzy-
mol. 1995, 70, 23-102.
(24) Cody, V.; Galitsky, N.; Luft, J . R.; Pangborn, W.; Gangjee, A.;
Devraj, R.; Queener, S. F.; Blakley, R. L. Comparison of the
Ternary Complexes of Pneumocystis carinii and Wild-Type
Human Dihydrofolate Reductase with a Novel Classical Anti-
tumor Furo[2,3-d]pyrimidine Antifolate. Acta Cryst. (D) 1997,
D53, 638-649.
(25) Broughton, M. C.; Queener S. F. Pneumocystis carinii Dihydro-
folate Reductase Used to Screen Potential Antipneumocystis
Drugs. Antimicrob. Agents Chemother. 1991, 35, 1348-1355.
(26) Chio, L.; Queener, S. F. Identification of Highly Potent and
Selective Inhibitors of Toxoplasma gondii Dihydrofolate Reduc-
tase. Antimicrob. Agents Chemother. 1993, 37, 1914-1923.
(27) Piper, J . R.; J ohnson, C. A.; Maddry, J . A.; Malik, D. N.; McGuire,
J . J .; Otter, G. M.; Sirotnak, F. M. Studies on Analogues of
Classical Antifolates Bearing the Naphthoyl Group in Place of
the Benzoyl in the Side Chain. J . Med. Chem. 1993, 36, 4161-
4171.
(28) Kisliuk, R. L.; Strumpf, D.; Gaumont, Y.; Leary, R. P.; Plante,
L. Diastereisomers of 5,10-Methylene-5,6,7,8-Tetrahydropteroyl-
D-Glutamic Acid. J . Med. Chem. 1977, 20, 1531-1533.
(29) We thank the Developmental Therapeutics Program of the
National Cancer Institute for performing the in vitro anticancer
evaluation.
(30) Chunduru, S. K.; Cody, V.; Luft, J . R.; Pangborn, W.; Appleman,
J . R.; Blakley, R. L. Methotrexate-resistant Variants of Human
Dihydrofolate Reductase: Effects of Phe31 Substitutitons. J .
Biol. Chem. 1994, 269, 9547-9555.
(31) Sack, J . S. CHAIN - A Crystallographic Modeling Program. J .
Mol. Graphics 1988, 6, 224-225.
(32) (a) Hendrickson, W. A.; Konnert, J . H. In Computing in Crystal-
lography; Diamond, R., Rameshan, S., Venkatesan, K., Eds.;
Indian Academy of Sciences: Bangalore, India, 1980; p 13.0. (b)
Finzel, B. C. Incorporation of Fast Fourier Transform to Speed
Restrained Least-Squares Refinement of Protein Structures. J .
Appl. Crystallogr. 1987, 20, 53-57.
(33) Laskowski, R. A.; MacArthur, M. W.; Doss, D. S.; Thornton, J .
M. PROCHECK: A Program to Check the Stereochemical
Quality of Protein Structures. J . Appl. Crystallogr. 1993, 26,
2891.
(14) Cody V.; Wojtczak, A.; Kalman, T. I.; Friesheim, J . H.; Blakley,
R. L. Conformational Analysis Of Human Dihydrofolate Reduc-
tase Inhibitor Complexes: Crystal Structure Determination Of
Wild-Type and F31 Mutant Binary and Ternary Inhibitor
Complexes. Adv. Exp. Med. Biol. 1993, 338, 481-486.
(15) Champness, J . N.; Achari, A.; Ballantine, S. P.; Bryant, P. K.;
Delves, C. J .; Stammers, D. K. The Structure of Pneumocystis
carinii Dihydrofolate Reductase to 1.9 Å Resolution. Structure
1994, 2, 915-924.
(16) Piper, J . R.; McCaleb, G. S.; Montgomery, J . A.; Kisliuk, R. L.;
Gaumont, Y.; Sirotnak, F. M. Synthesis and Antifolate Activity
of 5-Methyl-5-Deaza-Analogues of Aminopterin, Methotrexate,
Folic Acid and N-10 Methyl Folic Acid. J . Med. Chem. 1986, 29,
1080-1086.
J M980081Y