Design, synthesis, and antifolate activity of new analogues of piritrexim and other diaminopyrimidine dihydrofolate reductase inhibitors with ω-carboxyalkoxy or ω-carboxy-1-alkynyl substitution in the side chain

Article abstract of DOI:10.1021/jm0581718

As part of a search for dihydrofolate reductase (DHFR) inhibitors combining the high potency of piritrexim (PTX) with the high antiparasitic vs mammalian selectivity of trimethoprim (TMP), the heretofore undescribed 2,4-diamino-6-(2′,5′-disubstituted benzyl)pyrido[2,3-d]pyrimidines 6-14 with O-(ω-carboxyalkyl) or ω-carboxy-1-alkynyl groups on the benzyl moiety were synthesized and tested against Pneumocystis carinii, Toxoplasma gondii, and Mycobacterium avium DHFR vs rat DHFR. Three N-(2,4-diaminopteridin-6-yl)methyl)-2′-(ω-carboxy-1-alkynyl) -dibenz[b,f]azepines (19-21) were also synthesized and tested. The pyridopyrimidine with the best combination of potency and selectivity was 2,4-diamino-5-methyl-6-[2′-(5-carboxy-1-butynyl)-5′-methoxy]benzyl] pyrimidine (13), with an IC₅₀ value of 0.65 nM against P. carinii DHFR, 0.57 nM against M. avium DHFR, and 55 nM against rat DHFR. The potency of 13 against P. carinii DHFR was 20-fold greater than that of PTX (IC₅₀ = 13 nM), and its selectivity index (SI) relative to rat DHFR was 85, whereas PTX was nonselective. The activity of 13 against P. carinii DHFR was 20 000 times greater than that of TMP, with an SI of 96, whereas that of TMP was only 14. However 13 was no more potent than PTX against M. avium DHFR, and its SI was no better than that of TMP. Molecular modeling dynamics studies using compounds 10 and 13 indicated a slight binding preference for the latter, in qualitative agreement with the IC₅₀ data. Among the pteridines, the most potent against P. carinii DHFR and M. avium DHFR was the 2′-(5-carboxy-1-butynyl) dibenz[b,f]azepinyl derivative 20 (IC₅₀ = 2.9 nM), whereas the most selective was the 2′-(5-carboxy-1-pentynyl) analogue 21, with SI values of > 100 against both P. carinii and M. avium DHFR relative to rat DHFR. The final compound, 2,4-diamino-5-[3′-(4-carboxy-1-butynyl)-4′-bromo- 5′-methoxybenzyl]pyrimidine (22), was both potent and selective against M. avium DHFR (IC₅₀ = 0.47 nM, SI = 1300) but was not potent or selective against either P. carinii or T. gondii DHFR.

Full text of DOI:10.1021/jm0581718

4420
J. Med. Chem. 2005, 48, 4420-4431
Design, Synthesis, and Antifolate Activity of New Analogues of Piritrexim and
Other Diaminopyrimidine Dihydrofolate Reductase Inhibitors with
ω-Carboxyalkoxy or ω-Carboxy-1-alkynyl Substitution in the Side Chain
David C. M. Chan,† Hongning Fu,† Ronald A. Forsch,† Sherry F. Queener,‡ and Andre Rosowsky*,†
Dana-Farber Cancer Institute and Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,
Boston, Massachusetts 02115, and Department of Pharmacology and Toxicology, Indiana University School of Medicine,
Indianapolis, Indiana 46202
Received January 19, 2005
As part of a search for dihydrofolate reductase (DHFR) inhibitors combining the high potency
of piritrexim (PTX) with the high antiparasitic vs mammalian selectivity of trimethoprim (TMP),
the heretofore undescribed 2,4-diamino-6-(2′,5′-disubstituted benzyl)pyrido[2,3-d]pyrimidines
6-14 with O-(ω-carboxyalkyl) or ω-carboxy-1-alkynyl groups on the benzyl moiety were
synthesized and tested against Pneumocystis carinii, Toxoplasma gondii, and Mycobacterium
avium DHFR vs rat DHFR. Three N-(2,4-diaminopteridin-6-yl)methyl)-2′-(ω-carboxy-1-alkynyl)-
dibenz[b,f]azepines (19-21) were also synthesized and tested. The pyridopyrimidine with the
best combination of potency and selectivity was 2,4-diamino-5-methyl-6-[2′-(5-carboxy-1-
butynyl)-5′-methoxy]benzyl]pyrimidine (13), with an IC50 value of 0.65 nM against P. carinii
DHFR, 0.57 nM against M. avium DHFR, and 55 nM against rat DHFR. The potency of 13
against P. carinii DHFR was 20-fold greater than that of PTX (IC50 ) 13 nM), and its selectivity
index (SI) relative to rat DHFR was 85, whereas PTX was nonselective. The activity of 13
against P. carinii DHFR was 20 000 times greater than that of TMP, with an SI of 96, whereas
that of TMP was only 14. However 13 was no more potent than PTX against M. avium DHFR,
and its SI was no better than that of TMP. Molecular modeling dynamics studies using
compounds 10 and 13 indicated a slight binding preference for the latter, in qualitative
agreement with the IC50 data. Among the pteridines, the most potent against P. carinii DHFR
and M. avium DHFR was the 2′-(5-carboxy-1-butynyl)dibenz[b,f]azepinyl derivative 20 (IC50
) 2.9 nM), whereas the most selective was the 2′-(5-carboxy-1-pentynyl) analogue 21, with SI
values of >100 against both P. carinii and M. avium DHFR relative to rat DHFR. The final
compound, 2,4-diamino-5-[3′-(4-carboxy-1-butynyl)-4′-bromo-5′-methoxybenzyl]pyrimidine (22),
was both potent and selective against M. avium DHFR (IC50 ) 0.47 nM, SI ) 1300) but was
not potent or selective against either P. carinii or T. gondii DHFR.
Introduction
As part of a larger program aimed at the discovery of
dihydrofolate reductase (DHFR) inhibitors as poten-
tial drugs against the AIDS-associated opportunis-
tic pathogens Pneumocystis carinii,1 Toxoplasma
gondii, and Mycobacterium avium, we have previously
studied a small number of 2,4-diamino-5-[2′-methoxy-
5′-(ω-carboxyalkoxy)benzyl]pyrimidines (1)2 and 2,4-
diamino-5-[(2′-methoxy-ω-carboxy-1-alkynyl)benzyl]py-
rimidines (2).3 More recently, we also synthesized and
tested the 3′,4′-dimethoxy-5′-(4-carboxy-1-pentynyl) ana-
logue 3,4 in which the 3′,4′,5′-trisubstituted pattern of
trimethoprim (TMP) is retained. The left half of the
bination with a sulfa drug like sulfamethoxazole or
molecule in these compounds resembles TMP (4) while
the right half embodies the 2′,5′-disubstituted pattern
of piritrexim (5, PTX).
While TMP has excellent species selectivity in its
binding to nonmammalian vs mammalian DHFR, its
potency as a DHFR inhibitor is relatively low. For this
reason, it is clinically useful only when given in com-
dapsone to block de novo reduced folate synthesis.5
However, a shortcoming of TMP/sulfa drug regimens is
that many patients develop allergic reactions to the
sulfa drug that can be severe enough to require discon-
tinuation of treatment.6 The binding of PTX to DHFR
is orders of magnitude tighter than that of TMP, but
unfortunately this increased potency is achieved at the
cost of a dramatic loss of species selectivity. Thus, when
PTX is administered to a patient it has to be combined
with leucovorin (5-formyltetrahydrofolate) to prevent
life-threatening myelosuppression.7 With these consid-
* To whom correspondence should be addressed. Tel: (617)-632-
3117. Fax: 617-632-2410. E-mail: andre_rosowsky@dfci.harvard.edu.
† Dana-Farber Cancer Institute, Harvard Medical School.
‡ Department of Pharmacology and Toxicology, Indiana University.
10.1021/jm0581718 CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/07/2005

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4421
erations in mind, a goal of our laboratory for several
years has been to design DHFR inhibitors that combine
the potency of PTX with the species selectivity of TMP.
A potentially attractive feature of such compounds is
that they might not require coadministration of either
a sulfa drug or leucovorin. Encouraged by the finding
that several examples of general structures 1 and 2,
seemed to have this desired property,2,3 we asked
ourselves whether similar modification of either the 2′-
or 5′-methoxy group in PTX would likewise yield DHFR
inhibitors retaining, or exceeding, both the potency of
PTX and the species selectivity of TMP. In the present
paper, we report the synthesis and DHFR-inhibitory
activity of the heretofore unknown PTX analogues
6-14. A novel feature of these compounds that sets
them sharply apart from PTX is the presence of an
ionizable COOH, which confers solubility in water at
physiological pH without the need to formulate the drug
as a salt.
ment with NaOMe.2 As shown in Scheme 1, O-alkyl-
ation of rigorously dried 24 with benzyl bromide af-
forded the known compound 25, which on Knoevenagel
condensation with ethyl acetoacetate followed by cata-
lytic hydrogenation was converted to the â-ketoester 26.
Then, in an adaptation of the method originally devel-
oped for the synthesis of PTX by Grivsky and co-
workers,11a and later improved by Rauckman and co-
workers,11b 26 was heated with 2,4,6-triaminopyrimidine
in N-methyl-2-pyrrolidinone (NMP), and the resulting
lactam 27 was converted to 28 with oxalyl chloride in
DMF. Debenzylation and dechlorination of 28 were then
achieved in a single step with 5% Pd-C and sodium
formate in 2-methoxyethanol.12 To complete the syn-
thesis, the thoroughly dry Na salt of the resulting
phenol 29 was treated overnight at room temperature
with ethyl 4-bromobutanoate, 5-bromopentanoate, or
6-bromohexanoate in dimethylsulfoxide (DMSO) solu-
tion to obtain, after purification by chromatography and
recrystallization, the esters 30-32. Saponification then
yielded the desired acids 6-8. Although the heterocyclic
intermediates 27-29 were used without extensive
purification, the structure and purity of each of the
In other work on compounds with a one-carbon bridge
between the heterocyclic moiety and the substituted
phenyl ring of the side chain, we recently also described
the novel pteridines 15-18, of which two, 16 (n ) 1)
1
esters 30-32 and acids 6-8 was established from H
NMR and mass spectra and by microchemical analysis.
For the preparation of the 2′-modified analogue 9, we
initially tried to adopt with minor modifications an
approach developed by Troschu¨tz and co-workers13 for
the synthesis of PTX. Thus, as shown in the upper half
of Scheme 2, the Wittig reaction between 2-benzyloxy-
5-methoxybenzaldehyde (33) and 1-triphenylphospho-
ranylidene-2-propan-one followed by catalytic hydroge-
nation afforded the ketone 34, which on Vilsmeier
reaction (POCl3/DMF) afforded the expected mixture
35a/35b. The geometrical isomers migrated with dif-
ferent Rf values on thin-layer chromatography (TLC)
and were separated by flash chromatography. The two
isomers could be distinguished easily by 1H NMR, with
the aliphatic Me group of the minor Z-isomer (18%
isolated yield) giving a singlet at δ 2.25 whereas the
corresponding peak for the major E-isomer (41% isolated
yield) was shifted downfield to δ 2.68. Condensation of
the E/Z mixture with cyanoacetamide proved more
difficult than initially anticipated, as the isolation of the
desired product 36 required extensive column chroma-
tography with monitoring of individual fractions by 1H
NMR. After considerable effort, pure 36 was obtained
in 45% yield and was converted first to the chloronitrile
37 with oxalyl chloride and then to the key intermediate
38 with guanidine free base in refluxing pyridine.
While we had intended originally to remove the
O-benzyl group at this stage and then the O-alkylate,
resulting in a product with a bromo ester as in the
and 17 (n ) 2), displayed the level of potency (IC50
<
10 nM) and selectivity (>100-fold against the DHFR
from at least two of the three nonmammalian species)
that we sought to achieve.8 Thus, in addition to the
pyridopyrimidines 6-14, we synthesized the pteridines
19-21 as analogues of 16-18 with a carbon-carbon
triple bond in place of the OCH2 group in the side chain.
As with 10-14, a potentially useful feature of 19-21
is aqueous solubility at physiologic pH. Last, as an
extension of our work on the TMP analogue 3, we
synthesized 2,4-diamino-5-[3′-methoxy-4′-bromo-5′-(4-
carboxy-1-butyl)benzyl]pyrimidine (22). Interest in this
previously unknown compound was prompted by a
desire to ascertain how removal of the putative out-of-
plane hydrophobic interaction of the 4′-methoxy group
in 3 might influence selectivity against P. carinii, T.
gondii, and M. avium DHFR vs rat DHFR. Compound
22 may be viewed as an analogue of brodimoprim (23),
which was initially reported to have some advantages
over TMP but was subsequently withdrawn from the
market.9,10
Chemistry
A reasonable approach to the synthesis of the 2′-
methoxy-5′-(ω-carboxyalkoxy) analogues 6-8 seemed to
be via the sodium salt 24, which was accessible in three
steps from 4-hydroxyanisole by sequential reaction with
methyl chloroformate, Lewis acid-catalyzed condensa-
tion with dichloromethoxymethane, and finally treat-
1
synthesis of 30-32, the H NMR spectra of the unpu-
rified product of the cyanoacetamide reaction of 35a/

4422 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Chan et al.
Scheme 1a
2-ol (62)8 with triflic anhydride in pyridine at ambient
temperature for 3 h, followed by flash chromatography
1
on silica gel, afforded a product whose H NMR spec-
trum confirmed it to be the desired triflate 63. Over-
night treatment of purified 63 at 90 °C with benzyl
3-butynoate, benzyl 4-pentynoate, or benzyl 5-hexynoate
in the presence of (Ph3P)2PdCl2 and Et3N yielded the
esters 64-66, which were saponified directly with
NaOH to obtain the acids 19-21. For purification, the
latter were subjected to a two-stage sequence consisting
of anion exchange chromatography on a Dowex 50W-
X2-100 resin followed by adsorption chromatography on
deactivated silica gel.
a Reagents and conditions: (a) PhCH2Br, DMF; (b) MeCOCH2-
COOEt, toluene, AcOH, piperidine, reflux; (c) H2/5% Pd-C; (d)
2,4,6-triaminopyrimidine, NMP, 190 °C; (e) (COCl)2, DMF; (f)
HCOONa, 5% Pd-C, MeOCH2CH2OH, reflux; (g) NaOEt, DMSO,
then Br(CH2)3-5COOEt; (h) NaOH, EtOH.
Initial attempts to prepare the bromide 22 were made
directly from the O-triflate ester of the known compound
2,4-diamino-5-(3′-hydroxy-4′-bromo-5′-methoxybenzyl)-
pyri-midine (67).18 However, a few test reactions of 67
with trifluoromethylsulfonyl chloride or triflic anhydride
in pyridine were found to give complicated mixtures of
O- and N-trifluoromethylsulfonyl derivatives, presum-
ably indicating the more reactive nature of the NH2
groups in 67 than those in 62. In addition, we were
concerned that an acidic NHTf group on the heterocyclic
moiety could create a problem during the alkynylation
step by coordinating with the Pd catalyst. While they
are not commonly used for N-protection in diaminopy-
rimidines, we were pleased to find that Boc groups
served our purpose quite well. Thus, as shown in
Scheme 5, treatment of 67 with excess Boc anhydride
in the presence of Et3N and DMAP afforded a 67% yield
35b indicated that it probably contained in addition to
lactam 36 a certain amount of its sterically less crowded
6-methyl isomer 36a whose complete removal was
exceedingly laborious. Thus, we abandoned further use
of 36 as an intermediate and turned instead to the
original approach using a â-ketoester rather than a
â-chlorovinyl ketone. As shown in the lower half of
Scheme 2, aldehyde 33 was reduced to alcohol 39 with
NaBH4, 39 was converted to the chloride 40 with thionyl
chloride, and 40 was condensed with the Na salt of ethyl
acetoacetate in refluxing THF to form the substituted
â-ketoester 41. Further elaboration of 41 via intermedi-
ates 42-45 then led to the acid 9.14
1
of a noncrystalline product whose H NMR spectrum
The method used to obtain 9 could presumably be
extended to other 2′-(ω-carboxyalkoxy)-5′-methoxy ho-
mologues. However, the interesting results we had
obtained earlier with the acetylenic diaminopyrimidine
derivatives 2 prompted us to focus instead on ω-carboxy-
1-alkynyl analogues of PTX. As shown in Scheme 3,
these compounds were prepared via the intermediates
47-54, which were obtained from 2- and 3-methoxy-
benzaldehyde, respectively, via the classical â-ketoester
route (cf. Scheme 2). Treatment of 53 and 54 with
N-iodosuccinimide in TFA yielded the iodides 55 and
56, respectively, and the latter were condensed with
benzyl ω-alkynoates in the presence of (Ph3P)2PdCl2,
CuI, and Et3N in a Sonogashira reaction.17 The resulting
esters 57-61 were then saponified to the acids 10-14
to complete the synthesis.
was consistent with the fully Boc-substituted structure
68. Selective cleavage of the O-Boc group in 68 with
morpholine in CH2Cl2 (1:3, v/v) at room temperature
then yielded the desired phenol 69 (53% yield) along
with two slightly more polar byproducts that were
1
separable from 69 by chromatography and whose H
NMR spectra showed only three N-Boc groups, indicat-
ing that partial N-deprotection had also taken place. To
complete the synthesis 69 was converted to the fully
N-protected triflate ester 70 (95% yield) with triflic
anhydride in pyridine, and 70 was warmed with benzyl
4-pentynoate in the presence of (Ph3P)2PdCl2, (Ph3P)3-
CuBr, and Et3N at 55 °C for 3 h to obtain ester 71, from
which the Boc groups and the benzyl ester group were
all removed at the same time by a reaction with NaOH
in DMSO. The final product 22 was purified by pre-
parative high-performance liquid chromatography
(HPLC) on C18 silica gel followed by ion-exchange
chromatography on (diethylamino)ethyl (DEAE)-cel-
lulose. The two-step yield of 22 from 71 was 20%.
Since it was important to be sure that the products
ultimately used in DHFR assays had the correct pattern
of substitution on the phenyl ring, the position of the
1
iodine atom in 55 and 56 was confirmed by H NMR.
The 3′-proton in the phenyl ring was found to give a
doublet at δ 6.8 in 55 and δ 7.8 in 56, the 4′-proton a
doublet of doublets at δ 7.5 in 55 and δ 6.6 in 56, and
the 6′-proton a doublet at δ 7.1 in 55 and δ 6.4 in 56.
Moreover, the coupling constants for each aromatic
proton were consistent with the assigned 1,2,5-trisub-
stituted structures, thereby excluding any possibility
that iodination had occurred next to the methoxy group.
For the synthesis of the third group of compounds
described in this paper, the dibenz[b,f]azepine deriva-
tives 19-21, we took advantage of the ability of alkynes
to react with aromatic triflate esters in the presence of
a Pd(0) catalyst. Thus, as shown in Scheme 4, treatment
of N-(2,4-diaminopteridin-6-yl)methyldibenz[b,f]azepin-
Enzyme Inhibition Assays. The ability of acids
6-14 and 19-22 to inhibit dihydrofolate reduction by
Pneumocystis carinii, Toxoplasma gondii, Mycobacte-
rium avium, and rat DFHR in the presence of NADPH
was determined spectrophotometrically at 340 nm as
previously described.19 The P. carinii DHFR used in
these assays was recombinant enzyme cloned from
organisms isolated from the lungs of infected rats.19 Also
tested were the esters 30-32, the benzyl ether 38, and
the iodides 55 and 56 as additional examples of PTX
analogues in which one or the other methoxy group is
replaced. The IC50 values, selectivity index (SI), and 95%
confidence limits of all of the compounds are shown in

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4423
Scheme 2a
a Reagents and conditions: (a) Ph3PdCHCOMe; (b) H2, PtO2; (c) POCl3, DMF; (d) NtCCH2CONH2, NaH, THF, reflux; (e) (COCl)2,
DMF, 0 °C; (f) guanidine, pyridine, reflux; (g) NaBH4, MeOH; (h) SOCl2, benzene; (i) MeCOCH2COOEt, NaH,THF, reflux; (j) 2,4,6-
triaminopyrimidine, NMP, 190 °C; (k) (COCl)2, DMF, 0 °C; (l) HCOONa, 5% Pd-C, MeOCH2CH2OH, reflux; (m) NaOEt, DMSO, then
Br(CH2)3COOEt; (n) NaOH, EtOH.
Scheme 3a
a Reagents and conditions: (a) MeCOCH2COOEt, piperidine, AcOH, toluene, reflux; (b) H2/5% Pd-C; (c) 2,4,6-triaminopyrimidine,
NMP, 190 °C; (d) (COCl)2, DMF, CH2Cl2; (e) HCOONa, 5% Pd-C, MeOCH2CH2OH, reflux; (f) N-iodosuccinimide, TFA; (g) HCtC(CH2)nCOOBn
(n ) 2-4),(Ph3P)2PdCl2, CuI, Et3N; (h) NaOH, EtOH.
Table 1. Also presented for comparison are data we have
already reported for TMP (a weak but selective inhibi-
tor), PTX (a potent but nonselective inhibitor), and
compound 3, an example of a DHFR inhibitor which is
both potent and selective.
(a) Pneumocystis carinii DHFR. The 2′-methoxy-
5′-(ω-carboxyalkyl)oxy analogues 6-8 were all potent
inhibitors of recombinant DHFR from rat P. carinii,20
with IC50 values of approximately 0.1-1.0 nM as
compared with 13 nM for PTX. Moreover, while PTX
was completely nonselective for P. carinii vs rat DHFR
(SI ) 0.26), these compounds all displayed a small
degree of selectivity, and of these, the O-(carboxypropyl)
analogue 6 was 30 times more potent than PTX and was
mildly selective (SI ) 5.1). Interchanging the positions
of the O-methyl and O-(carboxypropyl) groups as in 9
led to a 2.5-fold increase in binding and a higher degree
of selectivity (SI ) 54) than that found in 6. Thus, it
appeared that binding was more favorable when the
O-carboxyalkyl side chain projected into the active site

4424 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Chan et al.
Scheme 4a
a Reagents and conditions: (a) Tf2O, pyridine, rt, 3 h; (b) HCtC(CH2)1-3COOBn, (Ph3P)2PdCl2, Et3N, DMF; (c) NaOH.
Scheme 5a
a Reagents and conditions: (a) Boc2O, Et3N, DMAP, rt, 20 h; (b) 25% morpholine/CH2Cl2 (25% v/v), rt, 3 h; (c) Tf2O, pyridine, 0 °C, 20
h; (d) HCtC(CH2)2COOBn, (Ph3P)2PdCl2, (PhP)3CuBr, Et3N, DMF, 55 °C, 20 h; (e) NaOH.
from the 2′-position of the benzyl group than from the
5′-position.
was a ca. 7.4-fold decrease in binding between the 2′-
O-benzyl ether 38 and PTX, and a 25-fold binding
between 38 and the 2′-iodide 56, presumably reflecting
the bulky nature of the O-benzyl group.
Replacing the OCH2 moiety in 6-8 by a carbon-
carbon triple bond, which has the effect of straightening
and stiffening the side chain, led to improved binding
to P. carinii DHFR in the case of 10 vs 6, but not in the
case of 11 vs 7 or 12 vs 8, suggesting that the effective-
ness of this molecular modification depends on the
overall length of the side chain. Compound 10 was ca.
130-fold more potent than PTX against P. carinii DHFR,
but because its binding to the rat enzyme was even
tighter than that of 9, there was a 16-fold loss of
selectivity (SI ) 3.3). In our previous work on analogues
of TMP, we had observed that replacement of OCH2 by
CtC was also somewhat favorable, though only when
compounds with a relatively short side chain (i.e., 1 and
2, n ) 2) were compared. However, given the much
greater potency of PTX vs TMP against P. carinii
DHFR, we were disappointed to find that our 2′,5′-
disubstituted PTX analogues were not more selective
than the 2′,5′-disubstituted TMP analogues with similar
side chains that we had studied earlier. Turning next
to the analogues with ω-carboxy-1-alkynyl groups of
approximately the same length at the 2′-position instead
of at the 5′-position, 13 was found to be less potent than
10, whereas in the longer homologues 14 and 11 a
difference was no longer evident.
With regard to the esters 30-32 and 45, it was of
interest to note that, with the exception of the shortest-
chained analogue 30, a substantial loss of binding
occurred v´ıs a v´ıs the corresponding acids, supporting
the idea that a COOH group is favorable for binding to
the P. carinii enzyme regardless of whether it projects
into the active site from the 5′- or 2′-position of the
benzyl moiety. Not surprisingly, there was only a minor
difference in potency and selectivity between the iodides
55 and 56 in comparison with PTX itself. However there
Turning to the dibenz[b,f]azepine analogues 19-21
with ω-carboxy-1-alkyl groups at the 2′-position, the
inhibition data against P. carinii DHFR in Table 1
revealed a striking degree of dependence on the number
of CH2 groups. Thus, while 19 and 21 had IC50 values
of 9.4 and 12 nM, respectively, that of 20 (230 nM) was
ca. 20-fold higher. Similarly, while 20 showed no
selectivity, the SI values of 19 and 21 were in the range
of 80-100. This was not entirely surprising, as we had
reported earlier a similar, albeit less steep, correlation
of the IC50 value with the number of side chain CH2
groups among the O-ω-carboxyalkyl analogues 16-18,
with 17 (IC50 ) 1.1 nM, SI ) 1300) showing a 1-log
improvement in potency relative to the other compounds
in the series.
The brodimoprim analogue 22 was a more potent
inhibitor than TMP, but was also less selective, reflect-
ing a greater increase in binding to rat DHFR than to
P. carinii DHFR. It is also worth nothing that 22 was a
poorer inhibitor of P. carinii DHFR than the corre-
sponding TMP analogue 3 but was a better inhibitor of
the rat enzyme. Because the congener of 22 with one
additional CH2 group in the side chain was not available
for comparison, it cannot be said at this time whether
the difference between 3 and 22 is due to the change of
the 4′-substituent from OMe to Br or to the fact that
the carboxyalkynyl chain is shorter by one carbon.
Overall, the results in Table 1 indicate that all of the
new compounds studied fall well short of the potency
and selectivity of 3, which remains the benchmark
among all of the compounds we have tested to date.
(b) Toxoplasma gondii DHFR. The 5′-O-carboxy-
alkyl analogues 6-8 and the 2′-O-carboxyalkoxy ana-

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4425
Table 1. Inhibition of Dihydrofolate Reductases by 2,4-Diaminopyrido[2,3-d]pyrimidine and 2,4-Diaminopteridine Derivatives
(IC50, nM)
M. avium
selectivity indexa
compd
3c
P. carinii
T. gondii
rat
P. carinii
T. gondii
M. avium
1.0
34
(30-39)
2800
2.4
5000
5000
150
2100
(0.82-1.2)
13000
(2.1-2.7)
300
(4300-5800)
180000
(160000-210000)
3.3
(3600-7100)
14
(110-190)
65
(2000-2800)
610
4c (TMP)
(10000-16000)b
13
(2400-3300)
4.3
(260-350)
0.61
(10-20)
0.26
(48-87)
0.76
(460-810)
5.4
5c (PTX)
6
(9.0-17)
0.43
(4.0-4.6)
1.9
(0.53-0.70)
0.26
(2.9-3.9)
2.2
(0.17-0.42)
5.1
(0.63-0.97)
1.2
(4.1-4.7)
8.5
(0.37-0.50)
0.51
(1.8-2.0)
2.3
(0.23-0.30)
0.29
(2.0-2.4)
2.2
(4.0-6.5)
4.2
(1.0-1.3)
0.96
(6.7-10)
7.5
7
(0.35-0.75)
1.3
(2.1-2.4)
2.0
(0.23-0.36)
0.35
(1.9-2.8)
2.0
(2.5-8.0)
1.5
(0.79-1.3)
1.0
(5.3-12)
5.9
8
(0.92-1.8)
0.17
(1.8-2.2)
2.3
(0.30-0.39)
0.32
(1.8-2.2)
9.1
(1.0-2.0)
54
(0.82-1.2)
4.0
(4.6-7.3)
28
9
(0.12-0.23)
0.097
(2.0-2.7)
0.46
(0.29-0.35)
0.057
(6.7-12.5)
0.32
(29-100)
3.3
(2.5-6.3)
0.70
(19-43)
5.6
10
11
12
13
14
19
20
21
22
30
31
32
38
45
55
56
(0.086-0.11)
1.4
(0.42-0.51)
4.3
(0.050-0.065)
1.0
(0.28-0.37)
5.7
(2.6-4.3)
4.2
(0.55-0.88)
1.3
(4.3-7.4)
5.6
(1.2-1.5)
1.8
(3.8-4.9)
3.2
0.94-1.1)
0.69
(5.1-6.4)
4.0
(3.4-5.3)
2.2
(1.0-1.7)
1.3
(4.6-6.8)
5.8
(1.6-2.0)
0.65
(2.9-3.6)
4.1
(0.63-0.74)
0.57
(3.6-4.5)
55
(1.8-2.8)
85
(1.0-1.6)
13
(4.9-7.1)
96
(0.58-0.73)
1.2
(3.5-4.8)
5.8
(0.52-0.63)
1.3
(51-60)
24
(70-100)
20
(11-17)
4.1
(81-120)
18
(1.0-1.3)
9.4
(5.0-6.7)
100
(1.1-1.4)
17
(19-29)
740
(15-29)
79
(3.8-5.8)
7.4
(14-21)
43
(7.9-11)
230
(81-130)
77
(13-22)
2.9
(640-860)
560
(58-110)
2.4
(4.9-11)
7.3
(29-66)
190
(200-260)
12
(71-83)
21
(2.4-3.4)
6.3
(480-650)
1300
(1.9-3.3)
110
(5.8-9.2)
62
(140-270)
210
(8.4-17)
71
(19-24)
15
(4.8-8.3)
0.47
(1100-1500)
630
(65-180)
8.9
(46-79)
42
(130-310)
1300
(58-87)
6.5
(14-16)
3.4
(0.37-0.62)
1.8
(530-750)
4.2
(6.1-12.9)
0.65
(39-46)
1.2
(850-2100)
2.3
(5.7-7.3)
28
(3.0-3.8)
11
(1.3-2.5)
9.4
(3.6-5.0)
14
(0.49-0.68)
0.50
(0.95-1.7)
1.3
(1.4-3.9)
1.5
(24-33)
35
(4.9-6.5)
13
(8.4-11)
12
(12-16)
21
(0.36-0.67)
0.60
(1.9-3.3)
1.6
(1.1-1.9)
1.8
(25-48)
96
(12-13)
16
(10-13)
8.6
(18-25)
170
(0.38-1.0)
1.8
(1.4-2.1)
11
(1.4-2.5)
20
(84-110)
83
(9.6-14)
10
(7.2-10)
5.6
(130-220)
93
(1.2-2.6)
1.1
(9.3-23)
9.1
(13-31)
17
(77-90)
5.8
(7.7-14)
4.1
(4.7-6.6)
0.92
(87-101)
5.0
(0.97-1.3)
0.86
(6.2-13)
1.2
(13-21)
5.4
(5.0-6.8)
3.8
(3.1-5.4)
5.8
(0.84-1.0)
1.8
(4.2-5.8)
4.0
(0.62-1.2)
1.1
(0.78-1.9)
0.69
(4.2-6.9)
2.2
(3.6-4.1)
(5.6-5.9)
(1.6-2.0)
(3.9-4.2)
(0.95-1.0)
(0.66-0.75)
(2.0-2.6)
a Selectivity Index (SI) ) IC50 (rat)/IC50 (P. carinii, T. gondii, or M. avium). All of the values are rounded off to two significant figures.
b Numbers in parentheses are 95% confidence limits. c Data for 3-5 are taken from refs 1-3.
logue 9 all gave IC50 values of ca. 2 nM and were only
slightly more potent than PTX, and their selectivity was
quite modest. The esters 30-32 and 45 showed es-
sentially the same binding activity as the corresponding
acids, suggesting that in the case of the binding of these
PTX analogues to the T. gondii enzyme, a carboxyalkoxy
group at either the 5′-position or the 2′-position hardly
mattered. The same was also the case generally speak-
ing for the carboxyalkynyl analogues 10-14. It was
clear from the results that the addition of a COOH
group was much less effective for T. gondii DHFR
inhibition than for P. carinii DHFR inhibition, suggest-
ing that there is little likelihood of an interaction with
a basic Lys or Arg residue.
The dibenz[b,f]azepine analogues 19-21 inhibited T.
gondii DHFR with IC50 values in the 20-100 nM range
and thus were more potent than TMP but less potent
than PTX. Although 21 did have a respectable SI value
of 62, neither its potency nor its selectivity equaled that
of 17 (IC50 ) 9.9 nM, SI ) 120).8 As in the case of the
P. carinii enzyme, the potency and selectivity of 22
exceeded that of TMP but was unremarkable in com-
parison with 3. Again, however, it would be of interest
to test the congener of 22 with one more carbon in the
side chain, but this compound was not made.
(c) Mycobacterium avium DHFR. The 5′-O-car-
boxyalkyl analogues 6-8 and the 2′-O-carboxyalkyl
analogue 9 inhibited M. avium DHFR with IC50 values
in the 0.2-0.4 nM range, but 9 (SI ) 28) was consider-
ably more selective than the others. The corresponding
esters were 5-10 times less potent, suggesting that a
COOH group in the side chain of the inhibitor may be
able to interact with a basic Arg or Lys residue in the
active site of the enzyme, as seems to be the case for
the P. carinii enzyme. The 5′-(ω-carboxy-1-alkynyl)
analogues 10-12 and the 2′-(ω-carboxy-1-alkynyl) ana-
logues 13 and 14 inhibited the enzyme with IC50 values
several times higher than those of the ethers 6-9.
However, despite the fact that it was less potent than
some of the other acids, the 2′-(4-carboxy-1-butynyl)
analogue 13 appeared to be by far the most selective
for M. avium vs rat DHFR (SI ) 96) just as it was for
the P. carinii enzyme (SI ) 85).
Turning to the dibenz[b,f]azepines as inhibitors of M.
avium DHFR, the most potent was the 4-carboxy-1-
butynyl derivative 20, with IC50 and SI values of 2.9

4426 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Chan et al.
Table 2. In Vitro Growth Inhibition of CCRF-CEM Leukemic
intact cells, suggesting that uptake rather than DHFR
binding may be the limiting factor in the ability of the
analogues to inhibit growth. At this point, it is unknown
how these lipophilic, yet negatively charged, molecules
are transported into cells. Studies addressing this issue
could be of interest in view of the fact that uptake of
these compounds across cell membranes may be more
complex than simply by partition in and out of the lipid
bilayer. Given the nearly 2-log selectivity of 13 at the
level of both P. carinii and M. avium DHFR inhibition,
the design of prodrugs of this compound might be
worthwhile to consider.
It should be noted that in vivo studies of the com-
pounds in this paper were not done, and we therefore
cannot say anything about the metabolic and/or phar-
macokinetic differences that may exist between 6-14
and PTX on one hand and between 22 and brodimoprim
on the other. While such studies could be of interest in
the future, they were not considered to be the scope of
the present work.
Lymphoblasts by Selected Piritrexim Analoguesa
compd
IC50 (µM)b
compd
IC50 (µM)b
5 (PTX)
6
9
0.013 ( 0.0023c
3.7 ( 1.1
5.3 ( 1.1
11
14
6.1 ( 0.33
2.8 ( 0.56
a Assays were carried out as described previously. b Each IC50
is the mean ( SD of three separate experiments on different days.
c Numbers are rounded off to two significant figures.
nM and 190, respectively. However, the activity of the
5-carboxy-1-pentynyl analogue 21 was essentially the
same. Activity and potency both decreased when the
carboxyalkynyl arm was shortened by one CH2 group
as in 19. In comparison with the previously reported
data for the ethers 15-18,8 compound 20 once again
most closely resembled 17 (IC50 ) 2.0 nM, SI ) 600) in
potency but was 3 times less selective. All in all, there
was a gratifying degree of qualitative consistency in
structure-activity trends against all three of the para-
site enzymes between the 2′-(ω-carboxy-1-alkynyl)-
dibenz[b,f]azepines in this paper and the 2′-O-(ω-
carboxyalkyl) analogues we had studied earlier. However,
as noted above, none of the new compounds matched
the remarkable potency and selectivity of 3,4 indicating
that a carboxyalkoxy or carboxyalkynyl arm is more
effective on a diaminopyrimidine scaffold than on a
diaminopyrido[2,3-d]pyrimidine or diaminopteridine scaf-
fold.
The brodimoprim analogue 22 was notable for the fact
that it combined high potency (IC50 ) 0.47 nM) with
>1000-fold selectivity. Because its potency against M.
avium DHFR was several times greater than that of 3,
this compound could represent an exploitable lead (e.g.,
by substituting other alkoxy groups for OMe at the 3′-
position or changing the length of the 5′-carboxyalkynyl
chain).
In Vitro Growth Inhibition Assays. Because our
therapeutic intent in synthesizing the PTX analogues
in this paper was to assess their therapeutic potential
in the treatment of AIDS-associated opportunistic infec-
tions in humans, we compared the activity of four
selected examples (6, 9, 11, and 14) as inhibitors of the
growth of cultured CCRF-CEM human leukemic lym-
phoblasts in a 72 h assay routinely used in our labora-
tory.21 Since there are no data as of yet on the ability of
these compounds to prevent growth of P. carinii, T.
gondii, or M. avium parasites in culture, the purpose
of this experiment was not to show that the compounds
could selectively block the growth of the parasites in
the presence of mammalian host cells but merely to
ascertain whether they could get into mammalian cells
at a concentration likely to be physiologically achievable
(<10 µM), as well as determine how their potency as
cell growth inhibitors would compare with their activity
as DHFR inhibitors and how these would compare with
the activity of PTX. As shown in Table 2, the four
compounds we tested all had similar IC50 values in the
1-10 µM range. Not surprisingly, in light of the pres-
ence of a COOH group in the side chain, they were
considerably less (>200-fold) potent than PTX despite
their much smaller differences in potency relative to
PTX in terms of binding to isolated DHFR enzymes (cf.
for example, the data against rat DHFR). The IC50 value
varied over a broader range than the IC50 value against
Molecular Modeling
A crystal structure of the ternary complex of PTX and
TMP with NADPH and P. carinii DHFR had already
been reported by Champness and co-workers22,23 when
this work began, but unfortunately the three-dimen-
sional (3D) coordinates of the ternary complex with PTX
were not in the public domain. PTX and TMP, according
to the published information, bind similarly to the active
site, the most notable difference being a conformational
change of the Phe69 side chain in the case of the PTX
complex, resulting in a face-edge interaction with the
dimethoxybenzyl group that was absent in the case of
the TMP complex. Interestingly, the 2′-OMe group of
PTX occupied a position close to the 3′-OMe group of
TMP. Unlike the trimethoxybenzyl group in TMP, the
dimethoxybenzyl group in PTX did not make van der
Waals contact with the nicotinamide group of NADPH.
There were 15 pairwise van der Waals contacts with
active site residues for TMP as compared with 20 for
PTX. The larger pyridopyrimidine moiety positions the
benzyl side chain nearer to the region encompassing
residues 65-69 than was the case with TMP. However
neither inhibitor had a side chain that was long enough
to reach the conserved Arg75 residue, a site with which
classical antifolates such as MTX can interact strongly
via the R-COOH group of the glutamate moiety. In the
crystal structure of the TMP ternary complex, electron
density was observed in this otherwise empty pocket,
and the density was believed to arise from two bound
water molecules.
There are two critical differences between the P.
carinii and human DHFR active sites; Lys37 and Phe69
in the P. carinii enzyme are replaced by Gln35 and
Asn64, respectively, in the human enzyme. In the case
of our recently described 2′-O-(ω-carboxyalkyl)]dibenz-
[b,f]azepine derivatives 15-18, improved binding se-
lectivity for the P. carinii enzyme was attributed to a
combination of hydrogen bonding to Lys37 and van der
Waals interaction with Phe69.8 Furthermore, previous
structure determinations indicated that Asn64 in hu-
man DHFR interacts with the COOH group of MTX.24
Guided by this background information, we performed
manual docking studies with two of the compounds

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4427
Figure 1. Stereoview of the active site ternary complexes of 10 (panel A) and 13 (panel B) with Pneumocystis carinii DHFR and
NADPH after an unrestrained 1-ns molecular dynamics simulation using the program AMBER-726 as described.8 Model generated
using the PYMOL molecular graphics system.
synthesized in the present work, namely, the acids 10
and 13, in which the alkyne side chain is more linear
and rigid than it would be in an ether side chain.
Molecular dynamic simulations were performed as
described in a previous paper, and a pictorial represen-
tation is given in Figure 1.
Notable interactions in 10 appeared to include H
bonds of the COOH group to the distal nitrogens of
Arg75 (3.3 and 2.9 Å) or the ꢀ-amino group of Lys37 (3.1
Å). There also appeared to be a conformational change
in the Phe69 side chain that allowed the formation of a
number of van der Waals contacts (3.7-4.0 Å) with the
alkynyl chain. In addition, Ile65 interacted extensively
with the benzyl moiety and Ile33 interacted with the A
ring of the heterocyclic moiety (ca. 4 Å). As in the case
of the TMP ternary complex, the CH2 bridge made van
der Waals contacts with the nicotinamide moiety of
NADPH.
For compound 13, the orientation of the benzyl ring
was similar to that of PTX.22 The COOH group appeared
to be able, once again, to form H bonds with the distal
nitrogens of Arg75 (3.1 and 3.0 Å) or the ꢀ-amino group
of Lys37. In addition, there was van der Waals contact
between the alkynyl side chain and Leu72 (4.3 Å), Ile33
(4.1 Å), and the phenyl ring of Ph69 (3.6-3.8 Å). Unlike
the 2′-OMe group in 10, which H bonds to the OH group
of Ser64, the 5′-OMe group in 13 was located 3.9 Å from
the carbonyl backbone of this residue and does not
interact with the OH. As in the complex of 10, The CH2
bridge of 13 again came within the van der Waals
distance of the nicotinamide group of NADPH (4.1 Å),
but rather distinctively, the CH2 bridge was tilted 2.2
Å further into the active site, allowing the carboxy-
alkynyl arm of the inhibitor to project into the region
of Arg75 and Lys37 despite being on the 2′- as opposed
to the 5′-carbon of the phenyl ring. All in all, the degree
of similarity of the interaction of 10 and 13 with the
enzyme was quite striking and was consistent with the
absence of a dramatic difference in binding as measured
by spectrophotometric assay.
The outcome of the dynamics simulations suggested
that both Arg75 and Lys37 could interact with the
bound ligands (Figure 1), and their moderate SI value
seemed to reinforce the hypothesis that, while the
presence of Lys37 in the P. carinii enzyme as opposed
to Gln35 in the human enzyme might confer selectivity
in the case of TMP analogues despite the fact that Arg75
is conserved, the effect of the latter takes precedence
where PTX analogues are concerned. A summary of the
individual active residues that contributed to the simu-
lated binding of 10 and 13 to P. carinii DHFR is given
in Table 3 of Supporting Information. When the inter-
and intramolecular interactions of the inhibitors with
all of the residues in the active site were summed up,

4428 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Chan et al.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(4-carboxybu-
tyl)oxybenzyl]pyrido[2,3-d]-pyrimidine (7). This com-
pound was obtained in 88% yield from 31 by the same
procedure as acid 6: mp 265-267 °C; MS calcd m/e 412.20
(MH+), found 412.28. Anal. (C21H25N5O4‚1AcOH‚0.5H2O) C, H,
N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(5-carboxypen-
tyl)oxybenzyl]pyrido[2,3-d]pyrimidine (8). This compound
was obtained in 87% yield from 32 by the same method as
acid 6: mp 267-268 °C; MS calcd m/e 426.21 (MH+), found
426.33. Anal. (C22H27N5O4‚0.6AcOH‚0.4H2O) C, H, N.
2,4-Diamino-5-methyl-6-[2′-(3-carboxypropyloxy)-5′-
methoxybenzyl]pyrido[2,3-d]pyrimidine (9). This com-
pound was prepared from 45 in 80% yield by the same method
as acid 6: mp 268-270 °C; MS calcd m/e 398.18 (MH+), found
398.06. Anal. (C20H23N5O4‚1.1H2O) C, H, N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(4-carboxy-1-
butynyl)benzyl]pyrido[2,3-d]pyrimidine (10). Iodide 55
(1.3 g, 3.1 mmol) was added to a mixture of benzyl 4-pen-
tynoate (875 g, 4.65 mmol), (Ph3P)2PdCl2 (20 mg), CuBr (2 mg),
Ph3P (5 mg), and Et3N (5 mL) in dry DMF (80 mL). The
suspension was kept at 80 °C for 2 days, after which point
additionalportionsofbenzyl4-pentynoate(850mg),(Ph3P)2PdCl2
(10 mg), and CuBr (1 mg) were added and the orange-colored
reaction mixture was kept at 80 °C for 3 more days. The
solvent was removed by rotary evaporation, and the residue
was purified by silica gel flash chromatography (9:1 CHCl3/
MeOH, Rf value of 0.26) to obtain ester 57 (0.32 g, 22%). The
ester was dissolved directly in EtOH (45 mL) and saponified
with 1 M NaOH (5 mL). The EtOH was removed on the rotary
evaporator, the residue was taken in H2O (50 mL), the solution
was adjusted to pH 5 with 20% AcOH, and after overnight
storage at 4 °C the precipitated solid was collected by
centrifugation, washed with cold H2O, and dried in a lyo-
philization chamber to obtain 10 as a beige solid (48 mg,
40%): mp >195 °C (dec), gradually turning black above 195
°C;MScalcdm/e392.17(MH+),found392.21.Anal.(C21H21N5O3‚
0.5AcOH‚1.5H2O) C, N. H: calcd, 5.87; found, 5.37.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(5-carboxy-1-
pentynyl)benzyl]pyrido[2,3-d]pyrimidine (11). This com-
pound was obtained from iodide 55 and benzyl 5-hexynoate
by the same method as 10: beige solid, mp 255-257 °C,
gradually turning black above 195 °C; MS calcd m/e 406.19
(MH+), found 406.06. Anal. (C22H23N5O3‚AcOH‚0.05H2O) C, H,
N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(6-carboxy-1-
hexynyl)benzyl]pyrido[2,3-d]pyrimidine (12). This com-
pound was obtained from iodide 55 and benzyl 6-heptynoate
by the same method as 10: beige solid, mp 173-175 °C (dec)
turning to a black oil at 192-194 °C; MS calcd m/e 420.20
(MH+), found 420.04. Anal. (C23H25N5O3‚1.75AcOH‚0.1H2O) C,
H, N.
2,4-Diamino-5-methyl-6-[2′-(4-carboxy-1-butynyl)-5′-
methoxybenzyl]pyrido[2,3-d]pyrimidine (13). This com-
pound was obtained from iodide 56 and benzyl 4-pentynoate
by the same method as 10: beige solid, mp 205-207 °C (dec),
becoming a black oil at 260-264 °C; MS calcd m/e 392.17
(MH+), found 392.14. Anal. (C21H21N5O3‚1.2AcOH‚0.8H2O) C,
N. H: calcd, 5.79; found, 5.10.
2,4-Diamino-5-methyl-6-[2′-(4-carboxy-1-pentynyl)-5′-
methoxybenzyl]pyrido[2,3-d]pyrimidine (14). This com-
pound was obtained from iodide 56 and benzyl 5-hexynoate
by the same method as 10: beige solid, mp 182-184 °C (dec),
becoming a black oil at 198-200 °C; MS calcd m/e 406.19
(MH+), found 406.22. Anal. (C22H23N5O3‚0.45AcOH‚H2O) C, H,
N.
2,4-Diamino-6-[2′-(3-carboxy-1-propynyl)-5H-dibenz-
[b,f]azepinyl]methylpteridine (19). Step 1. Triflic anhy-
dride (314 mg, 1.1 mmol) was added an ice-cold solution of 62
(383 mg, 1.0 mmol) in anhydrous pyridine (10 mL), and the
mixture was stirred at ambient temperature for 3 h. The
solvent was evaporated under reduced pressure, and the
residue was treated with 5% NaHCO3, causing a brownish-
yellow solid to form. The solid was collected, washed with H2O,
the calculated energy of interaction for 10 (-214.48 kcal/
mol) proved slightly lower than that of 13 (-196.87 kcal/
mol), in qualitative agreement with the finding that the
former was slightly more potent. The simulation clearly
suggests that, with some relatively minor spatial re-
alignments of a few key amino acid residues and a small
change in the relative conformation of the two halves
of the ligand, binding is possible for either 2′- or
5′-modified series of the PTX analogues. On the basis
of these results, a reasonable approach to the design of
better inhibitors of rat P. carinii DHFR based on the
PTX scaffold might be to modify the side chain of 10 in
such a way that the COOH group could project further
toward Lys37 while further improving van der Waals
contacts to Phe69. However, it must be stressed that
the modeling studies described here were done only for
the interaction of 10 and 13 with DHFR from rat P.
carinii and would be of limited value in predicting the
interactions of these compounds, or analogues thereof,
with the human pneumocystis strain P. jirovecii, in
which both the basic Lys37 residue and the Phe69
residue are replaced by serine.25 Selective inhibitors of
P. jirovecii vs human DHFR would obviously be more
important clinically than selective inhibitors of P. carinii
vs rat DHFR. At present, our ability to address this
problem via rational structure-based drug design is
limited by the lack of published 3D structures for P.
jirovecii DHFR ternary complexes with NADPH and
lipophilic antifolates. Once the structural coordinates
for at least one P. jirovecii ternary complex become
available, rapid progress will be likely.
Experimental Section
Melting points were determined on an electrothermal ap-
paratus and are uncorrected. Infrared (IR) spectra (cf. Sup-
porting Information) were measured using KBr disks or thin
films pressed between NaCl plates on a Perkin-Elmer double-
beam recording spectrophotometer. Only wavelengths above
1200 cm-1 are reported. 1H NMR spectra were recorded at 200
MHz on a Varian instrument. Each resonance is denoted as a
singlet (s), doublet (d), doublet of doublets (dd), triplet (t),
quartet (q), or multiplet (m). Low-resolution mass spectra (MS)
were provided by the Molecular Biology Core Facility of the
Dana-Farber Cancer Institute. Precoated Whatman MK6F
silica gel plates were used for TLC and were visualized under
UV light. Flash column chromatography was performed using
“Flash” grade silica gel (Baker, 20 µm particle size). Starting
materials and reagents were purchased from Sigma-Aldrich,
Fisher Scientific, or Alfa Aesar and were used without further
purification unless otherwise stated. Pooled eluates from flash
silica gel columns were typically dried over anhydrous Na2SO4
prior to rotary evaporation. Elemental analyses were per-
formed by the Robertson Microlit Laboratories, Madison, NJ,
and were within (0.4% of theoretical values except for the
found hydrogen values for 10, 13, 19, and 20, which were
uniformly on the low side of this range.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(3-carboxypro-
pyl)oxybenzyl]pyrido[2,3-d]pyrimidine (6). Ester 30 (60
mg, 0.14 mmol) was dissolved in EtOH (5 mL) with gentle
warming, and 2 M NaOH (2 mL) was added; when saponifica-
tion was complete, as indicated by disappearance of the
starting material according to TLC, the yellow solution was
adjusted to pH 5 with 10% AcOH, and the precipitate was
collected by centrifugation. The pellet was resuspended in a
small volume of distilled H2O and spun down, and the process
was repeated two more times. The final pellet was dried in a
lyophilization chamber, giving 6 as a white solid (50 mg,
83%): mp 272-273 °C; MS calcd m/e 398.18 (MH+), found
398.27. Anal. (C20H23N5O4‚0.9AcOH‚H2O) C, H, N.

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4429
dried in vacuo, and purified by silica gel flash chromatography
(95:5 CHCl3/MeOH) to obtain the triflate ester 63 as a yellow
solid (216 mg, 42%) which was sufficiently pure to be used
directly in the next step, mp 197 °C.
cm × 13 cm) afforded the triflate 70 as a soft solid (812 mg,
95%); 1H NMR (CDCl3) δ 1.32 (s, 18H, N(Boc)2), 1.37 (s, 18H,
N(Boc)2), 3.82 (s, 5H, bridge CH2 and OMe), 6.67 (s, 1H, aryl
H), 6.72 (s, 1H, aryl H), 8.50 (s, 1H, pyridine H5). The entire
batch of 70 (812 mg, 0.95 mmol) was combined with benzyl
4-butynoate (338 mg, 1.8 mmol) in a mixture of DMF (3 mL)
and Et3N (3 mL), and the solution was stirred under N2,
treated with 100 mg each of (Ph3P)3CuBr and of (Ph3P)2PdCl2,
and heated at 55 °C for 20 h. The volatiles were removed by
rotary evaporation, and the residue was taken up in 30 mL of
2:1 CH2Cl2/TFA to remove the Boc groups. After 1 h at room
temperature, the solution was concentrated to dryness, the
residue was taken up in DMSO (5 mL), and 2 N NaOH (15
mL) was added with swirling. A precipitate formed, suggesting
incomplete saponification. Additional H2O (60 mL) was added,
the supernatant was decanted, and the undissolved solid was
treated again with NaOH in DMSO as above. A small amount
of solid still failed to dissolve and was filtered off. The filtrate
was frozen and thawed, a final trace of undissolved material
was removed, and all of the solutions containing the product
were adjusted to pH 9 with 10% AcOH. Preparative HPLC
(C18 silica gel, 15% MeCN in 0.1 M NH4OAc, pH 7.4) yielded
a major peak, and a number of minor peaks which were
discarded. The major peak was collected and freeze dried, and
the residue was purified further on a column of DEAE-cellulose
(20 g, 1.5 cm × 22 cm, HCO3- form). The column was initially
eluted with H2O, then 0.4 M NH4HCO3, and finally 0.4 M
NH4HCO3 adjusted to pH 10 with 28% NH4OH. Eluates were
monitored by HPLC, pooled appropriately, and freeze dried
to obtain 22 as a white solid (83 mg, 20% based on 70): mp
>250 °C (dec); MS calcd m/e 405.06, 407.06 (MH+) for the two
stable Br isotopes; found 405.11, 407.11. Anal. (C17H17BrN4O3‚
2H2O) C, H, N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(3-carbethoxy-
propyl)oxybenzyl]pyrido-[2,3-d]pyrimidine (30). Metallic
Na (29.9 mg, 1.3 mmol) was dissolved in absolute EtOH (2
mL), the solvent was evaporated under reduced pressure, the
residue of NaOEt was taken up in anhydrous DMSO (2 mL),
and the mixture was stirred at room temperature for 30 min.
Crude 29 (4.36 g, 10 mmol; cf. Supporting Information) and
ethyl 4-bromobutanoate (254 mg, 1.3 mmol) were then added,
the reaction mixture was kept at room temperature overnight,
and H2O was added until a copious precipitate formed. The
product was extracted several times with CHCl3, the combined
extracts were evaporated, and the solid was purified by flash
chromatography (silica gel, 9:1 CHCl3/MeOH) followed by
recrystallization from EtOH to obtain ester 30 as a beige solid
(90 mg, 17%): mp 226-227 °C; MS calcd m/e 426.21 (MH+),
found 426.28. Anal. (C22H27N5O4) C, H, N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(4-carbethoxy-
butyl)oxybenzyl]pyrido-[2,3-d]pyrimidine (31). This com-
pound was obtained in 17% yield from crude 29 by the same
method as ester 30: mp 215-216 °C; MS calcd m/e 440.23
(MH+), found 440.37. Anal. (C23H29N5O4‚1.4H2O) C, H, N.
2,4-Diamino-5-methyl-6-[2′-methoxy-5′-(5-carbethoxy-
pentyl)oxybenzyl]pyrido-[2,3-d]pyrimidine (32). This com-
pound was obtained in 15% yield from crude 29 by the same
method as ester 30: mp 226-227 °C; MS calcd m/e 454.25
(MH+), found 454.23. Anal. (C24H31N5O4‚0.35H2O) C, H, N.
(Z/E)-4-Chloro-2-[(2-benzyloxy-5-methoxy)benzyl]-3-
propen-2-one (35a/35b). Phosphorus oxychloride (18.6 mL,
31 g, 0.2 mol) was added dropwise with stirring into DMF (20
mL) at 0 °C. An off-white suspension formed initially and
redissolved upon warming to room temperature. The resulting
amber-colored solution of Vilsmeier reagent was treated with
a solution of 34 (28.4 g, 0.1 mol) in DMF (20 mL), and the
mixture was stirred at room temperature for 1 h and treated
first with crushed ice and then carefully with 20% NaHCO3
until no more effervescence occurred. The resulting solution
was extracted several times with Et2O, and the pooled organic
layers were dried and evaporated to a greenish oil whose TLC
showed the presence of two spots. Although these two products
could be separated by silica gel chromatography and were able
to be identified as 35a and 35b, the reaction with cyanoacet-
Step 2. A mixture of freshly prepared 63 (260 mg, 0.50
mmol), benzyl 3-butynoate (87 mg, 0.50 mmol), (Ph3P)2PdCl2
(17 mg), and Et3N (0.75 mL) in dry DMF (40 mL) was stirred
at 90 °C overnight under argon. The solvent was removed
under reduced pressure, and the residue, consisting of the
benzyl ester 64, was treated with 2 N NaOH (1 mL) followed
immediately by the addition of H2O (60 mL). A small amount
of undissolved material was removed by suction filtration, and
the clear brownish-yellow filtrate was applied onto a column
of Dowex 50W-X2 resin (H+ form, 2 cm × 20 cm). The column
was eluted with H2O until the eluate was neutral and UV
transparent, after which the product was eluted with 1.5%
NH4OH. Appropriately pooled fractions of the eluate were
reduced to a small volume by rotary evaporation, followed by
freeze drying for 72 h. The residue was purified further by
silica gel chromatography using 85:15:5 CHCl3/MeOH/concen-
trated NH4OH as the eluent. Appropriate fractions were
combined and evaporated under reduced pressure to obtain
the hydrated partial ammonium salt of 19 as a dark yellowish-
brown solid (32 mg, 14%): mp 218 °C (dec) with preliminary
darkening; MS calcd m/e 449.16 (MH+), found 449.06. Anal.
(C25H19N7O2‚0.8NH3‚4.5H2O) C, N. H: calcd, 5.64; found, 4.86.
2,4-Diamino-6-[2′-(4-carboxy-1-butynyl)-5H-dibenz[b,f]-
azepinyl]methylpteridine (20). This compound was pre-
pared via ester 65 by the same method as 19 except that benzyl
5-pentynoate was used and the ion-exchange step was replaced
by silica gel chromatography using 85:15:1 (v/v/v) CHCl3/
MeOH/AcOH as the eluent. The product was a dark yellowish-
brown solid (36% yield); mp 221 °C (dec) with preliminary
darkening; MS calcd m/e 464.12 (MH+), found 464.18. Anal.
(C26H21N7O2‚2.6AcOH‚0.6H2O) C, N. H: calcd, 5.22; found,
4.68.
2,4-Diamino-6-[2′-(5-carboxy-1-pentynyl)-5H-dibenz[b-
,f]azepinyl]methylpteridine (21). This compound was pre-
pared via ester 66 by the same method as 19 except that benzyl
6-hexynoate was used and the product was purified in two
stages by ion-exchange chromatography using 1.5% NH4OH
followed by silica gel chromatography using 90:10:1.5 (v/v/v)
CHCl3/MeOH/AcOH. The product was a dark yellowish-brown
solid (16% yield): mp 205 °C (dec) with preliminary darkening;
MS m/e calcd 477.19 (MH+), found 477.15. Anal. (C27H23N7O2‚
1.9AcOH‚0.5H2O) C, H, N.
2,4-Diamino-5-[3′-(4-carboxy-1-butynyl)-4′-bromo-5′-
methoxybenzyl]pyrimidine (22). Step 1. A suspension of
67 (975 mg, 3 mmol)18 in dry THF (25 mL) was treated with
di-tert-butyl pyrocarbonate (5.45 g, 25 mmol), Et3N (840 µL,
606 mg, 6 mmol), and DMAP (60 mg, 0.6 mmol), and the
mixture was stirred at room temperature for 20 h. The
volatiles were removed by rotary evaporation, and the product
was chromatographed twice on silica gel (20-25 g, 2:1 hexanes/
EtOAc) to obtain 68 as a gum (1.67 g, 67%) which was then
taken up in 10 mL of a 4:1 (v/v) mixture of CH2Cl2 and
morpholine. The solution was left at room temperature for 3
h, and an additional amount of CH2Cl2 was added to allow all
of the morpholine to be removed by washing with dilute citric
acid until the pH of the aqueous phase was acidic. TLC (silica
gel, 1:1 EtOAc/hexanes) showed a major spot with a Rf value
of 0.5 and two minor spots with Rf values of 0.4 and 0.3.
Chromatography on flash-grade silica gel (50 g, 3 cm × 19 cm)
using 1:1 EtOAc/hexanes gave the phenol 69 as a soft solid
(772 mg, 53%). The minor impurities with Rf values of 0.4 and
0.3 eluted more slowly and appeared from 1H NMR spectra to
contain less than four N-Boc groups. The entire batch of 69
(772 mg, 1.06 mmol) was dissolved in pyridine (5 mL), and
the solution was cooled in an ice bath, treated with triflic
anhydride (673 µL, 1.13 mg, 4 mmol), and placed in the
refrigerator at 4 °C for 20 h. The reaction mixture was then
diluted with CH2Cl2 and washed with dilute citric acid until
all the pyridine was removed. Evaporation under reduced
pressure followed by flash chromatography (silica gel, 15 g, 2

4430 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Chan et al.
amide in the next step was carried out using the E/Z mixture.
35a (Z-isomer): white solid (2.91 g, 18%); mp 53-54 °C. Anal.
(C19H19ClO3) C, H, Cl. 35b (E-isomer): pale-green oil (6.80 g,
41%). Anal. (C19H19ClO3) C, H, Cl.
was collected by suction filtration and the filter cake washed
with boiling H2O to obtain 42 as a solid pure enough to be
used in the next step without additional purification. The solid
was initially colorless but turned yellow upon being dried in
vacuo at room temperature over P2O5; MS calcd m/e 418.19
(MH+), found 418.05.
5-[(2′-Benzyloxy-5′-methoxy)benzyl]-3-cyano-4-methyl-
2-pyridone (36). Sodium hydride (3.45 g, 60% dispersion in
mineral oil, 0.086 mol) was added in small portions to a
solution of cyanoacetamide (7.24 g, 0.086 mol) in dry THF (500
mL), and the white suspension was stirred at room tempera-
ture for 1 h. The mixture was then cooled to -15 °C and
treated dropwise with a solution of 35a/35b (9.50 g, 0.029 mol)
in THF (150 mL). The mixture was slowly warmed to room
temperature and refluxed for 1 h. The resulting yellow
suspension was then cooled and treated with H2O, and the
mixture was concentrated to dryness under vacuum. The
residue was purified extensively (3 times) by flash chroma-
tography on silica gel (97:3 CHCl3/MeOH), with individual
fractions of the eluent being monitored by TLC with the same
mixture of solvents. Fractions showing a single spot (Rf0.1)
were pooled and evaporated to obtain 36 as a white solid (4.62
g, 45%): mp 218-220 °C. Anal. (C22H20N2O3) C, H, N.
Step 2. Oxalyl chloride (21.5 mL, 31.7 g, 0.25 mol) was
added dropwise to a stirred solution of DMF (19.6 g, 20.8 mL,
0.27 mol) in CH2Cl2 (120 mL) at 0 °C. When addition was
complete, the mixture was brought back to room temperature
and stirred for 15 h. Compound 42 (10.4 g, 0.025 mol) was
then added, and the mixture was refluxed for 3 h, chilled to 0
°C, and treated slowly with n-BuNH2 (60 mL) during 2 h
followed by another 1.5 h of reflux. The reaction mixture was
cooled to room temperature, the volatiles were removed by
rotary evaporation, and the black residue was triturated with
a mixture of crushed ice and H2O until a beige solid formed,
which was dried in vacuo over P2O5 and partially purified by
silica gel flash chromatography (9:1 CHCl3/MeOH, Rf0.37). The
resulting chloride 43 (2.6 g, 25%) was used in the next step
without further purification; MS calcd m/e 436.15 (MH+), found
436.01.
Step 3. A mixture of partly purified 43 (7.76 g, 17.8 mmol)
and 5% Pd-C (3.6 g) in 2-methoxyethanol (200 mL) was
refluxed while slowly adding sodium formate (7.3 g, 107 mmol)
over a period of 4 h. The mixture was filtered through a Celite
pad, the filtrate was evaporated under reduced pressure, the
dark residue was taken up directly in DMSO (20 mL), and
the solution was added to ethanolic NaOEt prepared from 62.5
mg of metallic Na and EtOH (2 mL). The resulting dark-brown
solution, assumed to contain the Na salt of 44, was treated
dropwise with ethyl 4-bromobutanoate (0.389 mL, 530 mg, 2.72
mmol) and stirred at room temperature overnight. After the
addition of H2O (100 mL), the product was extracted with
CHCl3 (2 times), the pooled extracts were evaporated, and the
residue was purified by silica gel chromatography (9:1 CHCl3/
MeOH) and recrystallization from 95% EtOH to obtain 45 as
yellow needles (240 mg, 21%): mp 224-226 °C; MS calcd m/e
426.21 (MH+), found 426.15. Anal. (C22H27N5O4‚0.35H2O) C,
H, N.
A second faster-moving compound (2.60 g, 25% Rf 0.19) was
also isolated from the column and discarded. While the
1
elemental analysis on this material was not obtained, its H
NMR spectrum suggested that it was probably the unwanted
regioisomer 5-[(2′-benzyloxy-5′-methoxy)benzyl]-3-cyano-6-
methyl-2-pyridone (36a, cf. Scheme 2): 1H NMR (DMSO-d6)
δ 2.17 (s, 3H, 6-CH3), 3.60 (s, 3H, OMe), 3.78 (s, 2H, bridge
CH2), 5.05 (s, 2H, OCH2Ph, 6.34 (d, 1H, J ) 2.9 Hz, H3), 6.68
(dd, 1H, J ) 2.9 Hz, 9.2 Hz, H4′), 6.94 (d, 1H, J ) 9.2 Hz, H3′),
7.04-7.40 (m, 5H, aryl H), 7.69 (s, 1H, pyridone H4), 7.86 (br
s, lactam NH). Because of the extreme difficulty of eliminating
this impurity, further use of 36 as an intermediate to 2′-O-
substituted PTX analogues was abandoned except for its
conversion to 38 via 37 as detailed below.
2-Chloro-5-[(2′-benzyloxy-5′-methoxy)benzyl]-3-cyano-
4-methylpyridine (37). Oxalyl chloride (7.8 g, 61.5 mmol)
was added dropwise to CHCl3 (25 mL) and DMF (4.68 g, 64
mmol) at 0 °C, and the colorless solution was warmed to room
temperature and stirred for 3 h. Compound 36 (2.16 g, 6 mmol)
was then added, and the mixture was refluxed for 3 h before
being cooled back to room temperature, chilled to 0 °C, and
treated slowly with n-BuNH2 (30 mL) over a period of 1 h and
under reflux for another 30 min. The solvents were evaporated
under vacuum, and the dark-brown residue was chromato-
graphed twice on flash silica gel using 9:1 cyclohexanes/EtOAc
and then 99:1 CH2Cl2/EtOAc as the eluents. Evaporation of
appropriately pooled fractions yielded a white solid (0.73 g,
32%). Recrystallization of a portion of this solid from EtOH
gave 37 as fine white needles: mp 101-102 °C. Anal. (C22N19-
ClN2O2) C, H, N, Cl.
Molecular Modeling. Unrestrained dynamics simulations
using the program AMBER 726 were of 1-ns duration and were
performed with slight modifications as described earlier.8
Acknowledgment. This work was supported by
research Grant RO1-AI29904 (to A.R.) from the Na-
tional Institute of Allergy and Infectious Diseases,
National Institutes of Health, and Department of Health
and Human Services. We are grateful to the help of Dr.
Joel E. Wright and Dr. Ying-Nan Chen in performing
the cell growth inhibition assays.
2,4-Diamino-5-methyl-6-(2′-benzyloxy-5′-methoxyben-
zyl)pyrido[2,3-d]pyrimidine (38). Metallic Na (104 mg, 4.6
mmol) was dissolved in MeOH (10 mL), the solution was cooled
in an ice bath, and guanidine hydrochloride (500 mg, 5.2 mmol)
was added. The mixture was stirred for 15 min, the fine white
precipitate was suction filtered, and the filtrated was evapo-
rated to dryness. A solution of 37 (250 mg, 0.66 mmol) in
anhydrous pyridine (1.5 mL) was added. The solution was kept
at 70 °C for 1.5 h and heated to reflux for 4 h. After being
cooled to room temperature, the solution was diluted with H2O
and kept at 0 °C until a precipitate formed, which was filtered,
dried under vacuum, and purified by silica gel flash chroma-
tography (95:5 CHCl3/MeOH) to obtain 38 as a white solid (49
mg, 18%): mp 259-261 °C; MS calcd m/e 402.19 (MH+), found
402.08. Anal. (C23H23N5O2‚0.25H2O). C, H, N.
2,4-Diamino-5-methyl-6-[2′-(carbethoxypropyloxy)-5′-
methoxybenzyl]pyrido[2,3-d]pyrimidine (45). Step 1. A
mixture of 41 (41.8 g, 0.117 mol; cf. Supporting Information)
and 2,4,6-triaminopyrimidine (14.7 g, 0.117 mol) in NMP (100
mL) was heated at 190 °C in a two-neck round-bottom flask
fitted with a Dean-Stark trap for 3 h. The dark-amber
solution was cooled to room temperature until a beige-colored
solid formed. After the addition of MeOH (180 mL), the product
Supporting Information Available: Elemental analyses
for all of the new compounds. Detailed synthetic procedures
and physical constants for benzyl 4-pentynoate, benzyl 5-hex-
ynoate, benzyl 5-heptynoate, and intermediates 24-29, 34,
37-39, and 47-53, IR data for compounds 6-14, 19-22, 26,
1
30-32, 34, 36-41, 41, 45, 48, 53, 55, and 56. H NMR data
for compounds 6-14, 19-22, 25, 26, 30-32, 34-43, 45, 48,
51-57, and 63. This material is available free of charge via
the Internet at http://pubs.acs.org.
References
(1) Pneumocystis carinii is now also known as Pneumocystis jiroveci,
but this nomenclature change has been somewhat controversial.
For two divergent views on the subject, see: (a) Hughes, W. T.
Pneumocystis carinii vs Pneumocystis jiroveci: another misno-
mer (response to Stringer et al.). Emerging Infect. Dis. 2003, 9,
276-277. (b) Stringer, J. R.; Beard, C. B.; Miller, R. F.; Cushion,
M. T. A new name (Pneumocystis jiroveci) for Pneumocystis from
humans (response to Hughes), Emerging Infect. Dis. 2003, 9,
277-279. Because the name Pneumocystis carinii was used in
all of our previous papers reporting the inhibition of DHFR from
organisms isolated from immunosuppressed rats, we will retain
this name in the present paper.

Diaminopyrimidine Dihydrofolate Reductase Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4431
(2) Rosowsky, A.; Forsch, R. A.; Queener, S. F. Inhibition of
Pneumocystis carinii, Toxoplasma gondii, and Mycobacterium
avium dihydrofolate reductases by 2,4-diamino-5-[2-methoxy-5-
(ω-carboxyalkoxy)benzyl]pyrimidines marked improvement in
potency relative to trimethoprim and species selectivity relative
to piritrexim. J. Med. Chem. 2002, 45, 233-241.
(3) Rosowsky, A.; Forsch, R. A.; Queener, S. F. Further studies on
2,4-diamino-5- (2′,5′-disubstituted benzyl)pyrimidines as potent
and selective inhibitors of dihydrofolate reductases from three
major opportunistic pathogens of AIDS. J. Med. Chem. 2003,
46, 1726-1736.
(15) Hill, J. A.; Wisowaty, J. C.; Darnofall, M. E. Synthesis of carbon-
14 labeled piritrexim - a potential anticancer agent. J. Labelled
Compd. Radiopharm. 1993, 33, 1119-1130.
(16) Chan, D. C. M.; Rosowsky, A. Synthesis of the lipophilic
antifolate piritrexim via a palladium(0)-catalyzed cross-coupling
reaction. J. Org. Chem. 2005, 70 (4), 1364-1368.
(17) Sonogashira, K.; Tohda, Y.; Hagihara, B. A convenient synthesis
of acetylenes. Catalytic substitutions of acetylenic hydrogen with
bromo alkenes, iodo arenes, and bromopyridines. Tetrahedron
Lett. 1975, 16, 4467-4470.
(18) For the synthesis of brodimoprim analogues with ω-carboxy-
alkoxy side chains (as opposed to a ω-carboxy-1-alkynyl) side
chain from this key intermediate, see: Kompis, I.; Then, R. L.
Rationally designed brodimoprim analogues: synthesis and
biological activities. Eur. J. Med. Chem. 1984, 19, 529-534.
(19) Rosowsky, A.; Cody, V.; Galitsky, N.; Fu, H.; Papoulis, A. T.;
Queener, S. F. Structure-based design of selective inhibitors of
dihydrofolate reductase: synthesis and antiparasitic activity of
2,4-diaminopteridine analogues with a bridged diarylamine side
chain. J. Med. Chem. 1999, 42, 4853-4860.
(20) Gangjee, A.; Adair, O.; Queener, S. F. Pneumocystis carinii and
Toxoplasma gondii dihydrofolate reductase inhibitors and an-
titumor agents: synthesis and biological activities of 2,4-
diamino-5-methyl-6-[(monosubstituted anilino)methyl]pyrido-
[2,3-d]pyrimidines. J. Med. Chem. 1999, 42, 2447-2455.
(21) Wright, J. E.; Vaidya, C. M.; Chen, Y.-N.; Rosowsky, A. Efficient
utilization of the reduced folate carrier in CCRF-CEM human
leukemic lymphoblasts by the potent antifolate NR-(4-amino-4-
deoxypteroyl)-Nδ-hemiphthaloyl-L-ornithine (PT523) and its B-
ring analogues. Biochem. Pharmacol. 2000, 60, 41-46.
(22) 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.
(4) Forsch, R. A.; Queener, S. F.; Rosowsky, A. Preliminary in vitro
studies on two potent, water-soluble trimethoprim analogues
with exceptional species selectivity against dihydrofolate reduc-
tase from Pneumocystis carinii and Mycobacterium avium.
Bioorg. Med. Chem. Lett. 2004, 14, 1811-1815, 2693.
(5) Medina, I.; Mills, J.; Leoung, G.; Hopewell, P. C.; Lee, B.; Modin,
G.; Benowitz, N.; Wofsy, C. B. Oral therapy for Pneumocystis
carinii pneumonia in the acquired immunodeficiency syndrome.
A
controlled trial of trimethoprim-sulfamethoxazole versus
trimethoprim-dapsone. N. Engl. J. Med. 1990, 323, 776-782.
(6) Roudier, C.; Caumes, E.; Rogeaux, O.; Bricaire, F.; Gentilini, M.
Adverse cutaneous reactions to trimethoprim-sulfamethoxazole
in patients with acquired immunodeficiency syndrome and
Pneumocystis carinii pneumonia. Arch. Dermatol. 1994, 30,
1383-1386.
(7) Falloon, J.; Allegra, C. J.; Kovacs, J.; O’Neill, D.; Ogata-Arakaki,
D.; Feuerstein, I.; Polis, M.; Davey, R.; Lane, H. C.; LaFon, S.;
Rogers, M.; Zunich, K.; Turlo, J.; Tuazon, C.; Parenti, D.; Simon,
G.; Masur, H. Piritrexim with leucovorin for the treatment of
pneumocystis pneumonia (PCP) in AIDS patients. Clin. Res.
1990, 38, 361A.
(8) Rosowsky, A.; Fu, H.; Chan, D. C. M.; Queener, S. F. Synthesis
of 2,4-diamino-6- [2′-O-(ω-carboxyalkyl)oxydibenz[b,f]azepin-5-
yl]methylpteridines as potent and selective inhibitors of Pneu-
mocystis carinii, Toxoplasma gondii, and Mycobacterium avium
dihydrofolate reductase. J. Med. Chem. 2004, 47, 2475-2485.
(9) Periti, O. Brodimoprim, a new bacterial dihydrofolate reductase
inhibitor. J. Chemother. 1995, 7, 221-223.
(10) Then, R. L. Antimicrobial dihydrofolate reductase inhibitors - -
achievements and future options. J. Chemother. 2004, 16, 3-12.
(11) (a) Grivsky, E. M.; Lee, S.; Siegel, C. W.; Duch, D. S.; Nichol, C.
A. 2,4-Diamino-6-(2,5-dimethoxybenzyl)-5-methylpyrido[2,3-d]-
pyrimidine. J. Med. Chem. 1980, 23, 327-329. (b) Rauckman,
B. S.; Woolley, J. L., Jr.; Roth, B. Pyridopyrimidines methods
for their preparation and pharmaceutical formulations thereof.
EP 0278686; Wellcome Foundation, Ltd., 1988.
(12) It may be noted that numerous piritrexim analogues other than
those reported here were described in 1993, but unfortunately
the paper appeared in a Chinese language journal which is not
readily available in the U.S.; cf.: Zhang, Y.; Zhou, J.; Li, R.; Wu,
J.; Zhang, H. Synthesis and studies on the antitumor activities
of 2,4-diamino-6-substituted benzyl-5-methylpyrido[2,3-d]py-
rimidine derivatives. Chin. J. Med. Chem. 1993, 3, 85-96. From
the general synthetic scheme shown in this paper, it appears
that the â-keto ester route11a,b was the one that these authors
used.
(23) Stammers, D. K.; Delves, C.; Ballantine, S.; Jones, E. Y.; Stuart,
D. I., Achari, A. Bryant, P. K.; Champness, J. N. Preliminary
crystallographic data for Pneumocystis carinii dihydrofolate
reductase. J. Mol. Biol. 1993, 230, 679-680.
(24) Oefner, C.; D’Arcy, A.; Winkler, F. K. Crystal structure of human
dihydrofolate reductase complexes with folate. Eur. J. Biochem.
1988, 174, 377-385.
(25) Nahimana, A.; Rabodonirina, M.; Bille, J.; Francioli, P.; Hauser,
P. M. Mutations of Pneumocystis jirovecii dihydrofolate reductase
associated with failure of prophylaxis. Antimicrob. Agents
Chemother. 2004, 48, 4301-4305. As noted in this paper, a newly
identified problem associated with antifolate prophylaxis of
pneumocystis pneumonia in immunocompromised populations
is the emergence of drug-resistant DHFR in the face of selection
pressure with lipophilic inhibitors such as pyrimethamine. The
name Pneumocystis jirovecii has also been used instead of the
earlier version, Pneumocystis jiroveci; cf. Nahimana, A.; Rab-
odonirina, A.; Zanetti, G.; Meneau, I.; Francioli, P.; Bille, J.;
Hauser, P. M. Association between a specific Pneumocystis
jirovecii dihydropteroate synthase mutation and failure of py-
rimethamine prophylaxis in human immunodeficiency virus-
positive and -negative patients. J. Infect. Dis. 2003, 188, 1017-
1023.
(26) Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E.
III; Wang, J.; Ross, W. S.; Simmerling, C. L.; Darden, T. A.; Merz,
K. M.; Stanton, R. V.; Cheng, A. L.; Vincent, J. J.; Crowley, M.;
Tsui, V.; Gohlke, H.; Radmer, R. J.; Duan, Y.; Pitera, J.;
Massova, I.; Seibel, G. L.; Singh, U. C.; Weiner, P. K.; Kollman,
P. A. AMBER 7; University of California: San Francisco, CA,
2002.
(13) Troschu¨tz, R.; Zink, M.; Gnibl, R. An alternative synthesis of
piritrexim, a lipophilic inhibitor of human dihydrofolate reduc-
tase. J. Heterocycl. Chem. 1999, 36, 703-706.
(14) Two other approaches that could have been used to obtain
potential precursors to compounds 6-9 were not pursued. The
first, based on work by Hill and co-workers,15 would involve
consecutive reactions of an appropriately 2,5-disubstituted
4-arylbutan-2-one with malononitrile, diethoxymethyl acetate,
hydrogen bromide, and guanidine. The second, based on a new
regioselective synthesis of PTX recently developed in our own
laboratory,16 would involve Pd(0)-catalyzed coupling between
2-amino-5-bromo- or 2-amino-5-iodonicotinonitrile and a 2,5-
disubstituted benzylzinc halide, followed by reaction with SbBr3/
t-BuONO and then guanidine.
JM0581718