Synthesis and antifolate activity of new pyrrolo[2,3-d]pyrimidine and thieno[2,3-d]pyrimidine inhibitors of dihydrofolate reductase

Article abstract of DOI:10.1002/jhet.5570410523

Three previously undescribed dihydrofolate reductase (DHFR) inhibitors, N^α-[4-[N-[(2,4-diaminopyrrolo[2,3-d]pyrimidin-5-yl)methyl] amino]benzoyl]-N^δ-hemiphthaloyl-L-ornithine (7), N ^α-[4-[N-[(2,4-diaminothieno[2,3-d]pyrimi

Full text of DOI:10.1002/jhet.5570410523

Sep-Oct 2004
787
Synthesis and Antifolate Activity of New Pyrrolo[2,3-d]pyrimidine
and Thieno[2,3-d]pyrimidine Inhibitors of Dihydrofolate Reductase
Chitra Vaidya, Joel E. Wright, and Andre Rosowsky*
Dana-Farber Cancer Institute and Department of Biological Chemistry and Molecular Pharmacology,
Harvard Medical School, Boston, MA 02115
Received March 17, 2004
α
Three previously undescribed dihydrofolate reductase (DHFR) inhibitors, N -[4-[N-[(2,4-diaminopy-
δ
α
rrolo[2,3- d]pyrimidin-5-yl)methyl]amino]benzoyl]-N -hemiphthaloyl-L-ornithine (7), N - [ 4 - [N-[(2,4-
δ
diaminothieno[2,3-d]pyrimidin-5-yl)methyl]amino]benzoyl]- N -hemiphthaloyl-L-ornithine (8), and N-[4-
[N-[(2,4-diaminothieno[2,3-d]pyrimidin-5-yl)methyl]amino]benzoyl]-L-glutamic acid (1 2), were synthe-
sized and their antifolate activity was assessed. The ability of 7 and 8 to bind to DHFR and inhibit the growth
of CCRF-CEM human lymphoblastic leukemia cells in culture were dramatically reduced in comparison
α
δ
with the corresponding pteridine analogue, N -(4-amino-4-deoxypteroyl)-N -hemiphthaloyl-L-ornithine (1,
PT523). In a similar manner, the antifolate activity of 12 was markedly reduced in comparison with that of
the corresponding glutamate analogue, aminopterin (5, AMT). In contrast, 7, 8, and 12 all displayed excel-
lent affinity for the reduced folate carrier (RFC) of CCRF-CEM cells as measured by a standard competitive
influx assay. Lack of a consistent correlation between the results of the growth inhibition assays and those of
the DHFR and RFC binding assays results suggest that additional factors also play a role in the antifolate
activity of these compounds.
J. Heterocyclic Chem., 41, 787 (2004).
Nα-(4-amino-4-deoxypteroyl)-Nδ-hemiphthaloyl-L-
ornithine (PT523, 1) was synthesized in our laboratory a
number of years ago [1a,b], and is currently in Phase 1
clinical trial. In contrast to classical antifolates with a glu-
tamic acid side chain, the potent growth inhibitory activity
of PT523 does not depend on the ability of a cell to form
γ-polyglutamyl conjugates. It has been suggested that this
property may confer a therapeutic advantage for PT523 in
the treatment of tumors with low γ-folylpolyglutamate
synthetase activity in comparison with dose-limiting tis-
sues of the host [2].
As part of an ongoing effort to optimize the biological
activity of PT523, we previously examined the effect of
lengthening or shortening the amino acid side chain, mov-
ing the carboxyl group from the ortho to the meta or para
position of the phthaloyl moiety, modifying the 9,10-
bridge and 4-aminobenzoyl moiety comprising the central
region of the molecule, and altering the electronic charac-
ter of the B-ring by replacing either or both nitrogen atoms
in the B-ring by carbon or by introducing Me or Cl substi-
tution at the 5-position [3-7]. The various analogues were
compared with respect to their ability to bind to dihydrofo-
late reductase (DHFR), serve as competitive inhibitors of
the reduced folate carrier (RFC), and inhibit the growth of
CCRF-CEM human leukemic lymphoblasts in culture
[8,9]. Three of the more interesting members of the series
thus far have been the 5,8-dideaza, 10-deaza-, and 5,8,10-
trideaza derivatives 2-4, all of which were superior to the
classical DHFR inhibitors aminopterin (AMT, 5) and
methotrexate (MTX, 6) in these assays.
In the present paper we report the synthesis of the first
two analogues of 1 in which the B-ring is five-membered,
namely the pyrrolo[2,3-d]pyrimidine derivative 7 and the
thieno[2, 3- d]pyrimidine derivative 8. Although the
furo[2,3-d]pyrimidine and pyrrolo[2,3-d]pyrimidine deriv-
atives 9 and 10 with a glutamate side chain and CH2NH
bridge were known when this work began [10,11], and a
synthesis of the thieno[2,3-d]pyrimidine derivative 11 with
a glutamate side chain and CH2CH2 bridge had also been
reported (albeit without biological data) [12], we were sur-
prised to discover that the thieno[2,3-d]pyrimidine ana-
logue 12 with a glutamate side chain and CH2NH bridge
was unknown. Thus, we also synthesized 12 and compared
its in vitro antifolate activity to that of 8.
As shown in Scheme 1, the preparation of 7 began with
the known compounds 2,4-diaminopyrrolo[2,3-d] p y r i m i-
dine-5-carbonitrile (1 3) [13] and methyl L-2-[4-aminoben-
zoyl)amino]-5-phthalimidopentanoate (1 4) [3], and was
analogous to that of 1 0 from 1 3 and diethyl N- ( 4 -
(aminobenzoyl)-L-glutamate (1 5) described by Gangjee

788
C. Vaidya, J. E. Wright, and A. Rosowsky
Vol. 41
and coworkers [11]. The product was purified by flash
chromatography on silica gel and characterized by its H
chloride in aqueous tetrahydrofuran at room temperature
as described earlier. After purification to homogeneity, the
product was treated directly with sodium hydroxide at
room temperature for 5 days to cleave the methyl ester,
remove the pivaloyl groups, and open the phthalimide
ring. The same sequence starting from 15 instead of 14
afforded 12 via the bromo diester 21 and the debrominated
diester 22.
As in the synthesis of the pyrrolo analogue 7, the iden-
tity of each compound in Scheme 2 was confirmed from
1H NMR spectra. Thus, reduction of 19 to 20 and of 21 to
22 was accompanied by the appearance of a distinctive
vinyl singlet, and deprotection of 20 and 22 with sodium
hydroxide was evident from the loss of the signals corre-
sponding to the pivaloyl and methyl ester groups.
Concomitant opening of the phthalimide ring was con-
firmed by the appearance of a new CONH peak along with
the same change in appearance of the signal for the δ-CH2
protons of the ornithine side chain as we had noted earlier
in other examples of this cleavage reaction [6]. A small
downfield shift for the C6 proton in 12 relative to 7 by ca.
was consistent with the lower electronegative character of
the sulfur atom.
1
NMR spectrum in DMSO-d6 solution, which featured the
expected singlets for the C6 and N7 protons of the pyrrole
moiety and the CH=N proton of the bridge. The presence of
a methyl ester group and closed phthalimide ring were also
evident from the spectrum, and were supported by mass
spectrometry. Treatment of 16 for 4 hours at room tempera-
ture with NaCNBH3 in MeOH at pH 2 led to the desired
reduction product 1 7, whose 1H NMR spectrum showed the
required loss of the bridge CH=N proton and confirmed that
the methyl ester and closed phthalimide ring had remained
intact. The identity of 17 was also supported by mass spec-
trometry and microchemical analysis. Deprotection of 17
was accomplished readily by treatment with barium hydrox-
1
ide. The H NMR spectrum of the deprotected product in
DMSO-d6 solution again confirmed the presence of the C6
and N7 protons on the pyrrolo[2,3-d]pyrimidine moiety.
For the synthesis of the thieno[2,3-d]pyrimidine 8
(Scheme 2), we began with 2,4-dipivaloylamino-5-bro-
momethyl-6-bromothieno[2,3-d]pyrimidine (1 8) [14a,b].
Heating a mixture of 18 and 14 with dried sodium bicar-
bonate in dimethylformamide solution at 55 °C for 48
hours, followed by purification on a silica gel column
a fforded the bromo ester 19, which was debrominated
directly to 20 with sodium borohydride and palladium
The pyrrolo[2,3-d]pyrimidine 7 and thieno[2,3-d]pyrim-
idines 8 and 12 were tested for their ability to bind to
recombinant human DHFR according to standard assays

Sep-Oct 2004
New Pyrrolo[2,3-d]pyrimidine and Thieno[2,3-d]pyrimidine Inhibitors
789
previously described [8,9]. As shown in Table 1, the
potency of the pyrrolo[2,3-d]pyrimidine 7 was decreased
7-fold relative to PT523, whereas that of the thieno[2,3-
d]pyrimidine 8 was decreased 60-fold, indicating that
these structural modifications of the B-ring were both
detrimental to binding. The weaker inhibition observed
with 7 in comparison with PT523 was consistent with the
results of Gangjee and coworkers [11], who found, using
rat DHFR and IC50 rather than Ki values, that compound
10, the analogue of 7 with a glutamate side chain, was 15-
fold less potent than MTX (6). Although a three-dimen-
sional structure for the ternary complex of 10 with
NADPH and rat DHFR is not available, possible reasons
for its relatively weak binding might be (i) that the change
of the B-ring nitrogen atom from an H-bond acceptor to an
H-bond donor is unfavorable and/or (ii) that the side chain
is less optimally positioned for binding when the B-ring is
five-membered than when it is six-membered. The weaker
binding of 7 to human DHFR relative to PT523 presum-
ably reflects steric and electronic differences analogous to
those between 10 and MTX. In the case of the thienopy-
rimidine analogue 8, we surmise that decreased affinity
relative to 7 reflects (i) the inability of the sulfur atom to
function as an H-bond acceptor and/or (ii) its larger size in
comparison with nitrogen. As a consequence of the larger
size of sulfur, this substitution in the five-membered ring
might produce a less favorable orientation of the bridge,
and hence the hemiphthaloylornithine side chain, than is
the case in the pyrrole analogue 7 [16]. It was of interest to
note, as well, that 8 was a 27-fold better inhibitor than 12,
in accord with our earlier findings with other types of B-
ring analogues containing a hemiphthaloylornithine side
chain as opposed to a glutamate side chain [8].
In addition to their interaction with DHFR, we also
examined the ability of 7, 8, and 12 to compete with
[3H]MTX for binding to the reduced folate carrier (RFC)
of CCRF-CEM human lymphoblastic leukemia cells using
the same standardized influx assay we had used earlier to
compare the RFC binding of PT523 and its 5-deaza, 8-
deaza, and 5,8-dideaza analogues [9]. As can be seen from
Table 1, compounds 7 and 8 were both better inhibitors of
RFC binding than either AMT or PT523, and in fact
proved to have the lowest Kis of any AMT analogues with
Table 1
Dihydrofolate Reductase (DHFR) Binding and CCRF-CEM Cell Growth
Inhibition by Pyrrolo[2,3-d]pyrimidine and
Thieno[2,3-d]pyrimidine Analogues of PT523 (1) and Aminopterin (5)
Compound DHFR Inhibition
RFC binding
Cell Growth
K (pM) [a]
i
K (µM) [a]
IC (_M, 72 h) [a]
50
i
1 [b]
5 [b]
7
0.30 ± 0.036 (n = 7) 0.71 ± 0.12 (n = 3) 1.5 ± 0.39 (n = 10)
3.7 ± 0.35 (n = 3)
2.2 ± 0.51 (n = 4)
18 ± 6.1 (n = 6)
5.4 ± 0.09 (n = 3)
0.14 ± 0.05 (n = 3) 68 ± 0.11 (n = 6)
0.075 ± 0.01 (n = 3) 120 ± 28 (n = 5)
4.4 ± 0.10 (n = 3)
8
12
480 ± 110 (n = 3) [c] 0.53 ± 0.12 (n = 3) 43 ± 7.0 (n = 4)
[a] Assays were carried out as described in reference 8 and 9. [b] Data for a hemiphthaloylornithine side we have tested to date.
PT523 (1) and AMT (5) included for comparison are taken from reference
9. All numbers are rounded off to two significant figures and are followed
membered B-ring appeared to be highly favorable for RFC
by the standard deviation for n experiments performed on different days. [c]
Thus, in contrast to DHFR binding (see above), the five-
binding [18]. Interestingly, while it was not quite as low as
In contrast to 7 and 8, the very weak inhibition of DHFR by 12 was typical
of Zone B equilibrium binding [15].
that of 8, the Ki of 12 was also lower than that of AMT,

790
C. Vaidya, J. E. Wright, and A. Rosowsky
Vol. 41
providing further evidence of the advantage of a five-
membered over a six-membered B-ring where binding to
the RFC protein is concerned.
With regard to cell growth inhibition, 7 was 45-fold less
potent than PT523, 8 was 80-fold less potent than PT523,
and 12 was 10-fold less potent than AMT during 72 hours of
continuous drug exposure, at which point the DHFR would
presumably be fully saturated with inhibitor. Thus there was
a greater decrease in growth inhibitory activity than could be
In summary, pyrrolo[2,3-d]pyrimidine and thieno[2,3-d]-
pyrimidine analogues of the potent nonpolyglutamatable
DHFR inhibitor PT523 were synthesized and tested in vitro
as antifolates. Also synthesized was a previously unknown
thieno[2,3-d]pyrimidine analogue of AMT. Replacement of
the pteridine scaffold by these 6/5 ring systems with reten-
tion of the CH2NH bridge and hemiphthaloylornithine side
chain was favorable for RFC binding but not for DHFR
binding or cell growth inhibition. These results add to our
understanding of the structure-activity correlations that
have to be to be considered during the development of eff i-
cacious second-generation PT523 analogues.
accounted for by the K differences in the DHFR and RFC
i
binding assays alone. In the case of 7 and 8, the fact that
improved RFC binding was insufficient to offset decreased
DHFR binding indicates that other factors are probably at
work. For example, the steady-state level of free drug (i.e.
not bound to DHFR) in the cell [19] might be lower with 7
and 8 than it is with PT523 because a difference in the ability
of the RFC to act as a bidirectional pump mediating efflux as
well as influx once the DHFR is saturated with drug [20]. In
the case of 1 2, a critical factor in addition to DHFR binding
and cellular uptake of the monoglutamates form of the drug,
is that polyglutamation (which we did not examine) might be
less efficient than in the case of AMT.
EXPERIMENTAL
2,4-Diaminopyrrolo[2,3-d]pyrimidine-5-carbonitrile (1 3) was
synthesized from 2-amino-5-bromo-3,4-dicyanopyrrole [13]
according to Gangjee and coworkers [11]. Methyl 2-L-[N- ( 4 -
aminobenzoyl)amino]-5-phthalimidopentanoate (1 4) was pre-
pared as described earlier [3]. 2,4-Bis(pivaloylamino)-5-bro-
momethyl-6-bromothieno[2,3-d]pyrimidine (1 8) was the same
minimally purified material we had used to prepare a series of
lipophilic 2,4-diamino-5-thieno[2,3-d]pyrimidines [14a,b], and
was obtained from 2,4-diamino-5-methylthieno[2,3-d]pyrimi-
dine [22,23] by acylation with pivalic anhydride in pyridine fol-
lowed by reaction with N-bromosuccinimide and dibenzoyl per-
oxide in chloroform. Diethyl N-(4-aminobenzoyl)-L-glutamate
(1 5 ) and other chemicals were purchased from Aldrich,
Milwaukee, WI. Other reagents were from Fisher, Boston, MA.
Infrared spectra were obtained in potassium bromide disks with a
Perkin-Elmer model 781 double-beam spectrophotometer. Only
It is of interest to note that, when our work on 7 began,
we were unaware that a Japanese group was independently
synthesizing DHFR inhibitors with a hemiphthaloylor-
nithine or other non-glutamate side chain linked to a
pyrrolo[2,3-d]pyrimidine scaffold [21]. Six compounds of
general structure 23 were prepared in which the length of
the (CH2)m and (CH2)n linkers was varied with a view to
optimizing DHFR binding and cell growth inhibition. The
analogue closest in structure to PT523, and the one that
seemed to give the best combination of DHFR binding and
cell growth inhibition, was 23 (m = 2, n = 3), with an IC50
of 6.9 nM against bovine DHFR, an IC50 of 1.5 nM against
a mouse fibrosarcoma cell line (MethA), and an IC50 of
7.5 nM against CCRF-CEM cells (72 h treatment) [21].
Thus, under incubation conditions essentially identical to
ours, 23 was a more potent inhibitor of the growth of
CCRF-CEM cells than 7 by almost one order of magni-
tude, showing that a CH2CH2 group is better than a
C H2NH bridge when the heterocyclic scaffold is a
pyrrolo[2,3-d]pyrimidine, just as we had observed, though
to a lesser extent, when the scaffold was a pteridine (i.e., in
the case of the 10-deaza analogue of PT523) [6].
-1
peaks with wave numbers greater than 1200 cm are reported.
1
H-nmr spectra were recorded at 200 MHz on a Varian VX200
instrument. Fast-ion bombardment mass spectra (FABMS) were
obtained by staff of the Dana-Farber Cancer Institute Molecular
Biology Core Facility. High-resolution mass spectra (HRMS)
were obtained by staff of the Department of Chemistry, Harvard
University, Cambridge, MA. TLC analyses were performed on
Whatman MK6F silica gel plates with UV illumination at 254
nm. Column chromatography was performed on Baker 7025
flash silica gel (40 µm particle size). HPLC separations were on
C
silica gel radial compression cartridges (Millipore, Milford,
18
MA; analytical, 5-µm particle size, 5 x 100 mm; preparative, 15-
µm particle size, 25 x 100 mm). Melting points were measured in
Pyrex capillary tubes in a Mel-Temp "Electrothermal" apparatus
(Fisher), or on a hot-stage microscope apparatus (Fisher), and are
not corrected. Elemental analyses (C, H, N) were performed by
Robertson Laboratories, Madison, NJ, and were within ± 0.4% of
theoretical values unless otherwise specified.
2,4-Diaminopyrrolo[2,3-d]pyrimidin-5-yl)methyl]amino]benzoyl]-
δ
amino]-N -hemiphthaloyl-L-ornithine (7).
Step 1. Raney nickel (3.3 g) was added to a solution of 13
(0.67 g, 3.88 mmole) and 14 (2.4 g, 5.82 mmole) in 80% acetic
acid (65 mL) in a Parr apparatus and the mixture was shaken
under hydrogen at 15 psi for 12 hours. The mixture was filtered
through Celite, the filter pad was washed with 10% acetic acid,
and the combined filtrate and wash solution were removed on the
rotary evaporator. A cold solution of saturated aqueous sodium

Sep-Oct 2004
New Pyrrolo[2,3-d]pyrimidine and Thieno[2,3-d]pyrimidine Inhibitors
791
bicarbonate was added to the residue, and the mixture was kept on
ice for 1 hour. The yellowish-brown solid was collected by filtra-
tion, washed with water, dried in a vacuum oven, taken up in a
minimal volume of 9:1 chloroform-methanol, and chro-
matographed on column of flash-grade silica gel packed with 95:5
chloroform-methanol and eluted sequentially with 95:5 and 93:7
α-CH), 5.44 (br s, NH ), 6.08 (br s, NH ) 6.49 (t, H, bridge
2
CH NH), 6.68 (s, H, 6-H), 6.71-6.75 (m, 2H, 3'- and 5'-H), 7.37-
2
2
7.47 (m, 3H, phthaloyl ring protons), 7.63 (m, 1H, phthaloyl ring
proton), 7.64-7.71 (m, 2H, 2'- and 6'-H), 8.03-8.06 (d, H,
CONH), 8.36 (t, H, phthaloyl CONH), 10.53 (br s, H, 7-NH).
.
Anal. Calcd. for C H N O H O; C, 56.05, H, 5.23, N,
27 28
19.37. Found; C, 55.72; H, 4.92; N, 19.49.
8
6
2
chloroform-methanol. Fractions containing a major tlc spot at R
f
0.31 (silica gel, 9:1 chloroform-methanol) along with a minor
impurity were pooled and evaporated to dryness. The residue was
dissolved in small volume of 9:1 chloroform-methanol, and the
solution was poured into a large volume of ether. The precipitate
was collected by filtration, washed well with ether, and dried in a
vacuum oven to obtain Schiff's base 16 as yellow powder (0.475
g, 22%); ir: ν 3350, 3250, 2960, 1750, 1710, 1650, 1610, 1550,
α
N -[4-[N-[(2,4-Diaminothieno[2,3-d]pyrimidin-5-yl)methyl] -
δ
amino]benzoyl]-N -hemiphthaloyl-L-ornithine (8).
Step 1. Compound 14 (2.4 g, 6.1 mmole) and sodium bicar-
bonate (7.7 g, 92 mmole) were added sequentially to a solution of
18 (4.37 g, 9.2 mmole) in dry dimethylformamide (20 mL) and
the reaction mixture was stirred at 55 °C for 48 hours. The sol-
vent was removed by rotary evaporation, and the residue was
taken up in ethyl acetate (200 mL). Some dark-brown insoluble
material did not dissolve and was discarded. The solution was
washed with water (100 mL), dried over anhydrous sodium sul-
fate, and evaporated. The product was chromatographed on a
flash-grade silica gel column packed and eluted with 99:1 chloro-
-1
1
1510, 1400, 1380, 1270, 1200 cm ; H-nmr: δ 1.70-1.78 (m, 4H,
β- and γ-CH ), 3.57-3.60 (m, 5H, COOCH and δ-CH ), 4.44 (m,
2
H, α-CH), 5.67 (br s, NH ), 6.70 (br s, NH ), 7.22-7.26 (m, 2H, 3'-
3
2
2
2
and 5'-H), 7.52 (s, H, 6-H), 7.80-7.89 (m, 6H, 2'-H, 6'-H, and
phthalimide ring protons), 8.47 (s, H, CH=N), 8.63-8.67 (d, H,
CONH), 11.51 (br s, H, 7-NH); FABMS: m/e 555 (M+H). This
material was used directly in the next step.
Step 2. Solid sodium cyanoborohydride (70 mg, 1.12 mmole)
was added to a stirred solution of 16 (450 mg, 0.81 mmole) in
methanol (50 mL), and the pH of the solution was adjusted to 2
using 6 N hydrochloric acid. Stirring was continued for 4 hours
at room temperature, and the precipitated solid was collected by
filtration, washed with methanol, dried in a vacuum oven, and
flash chromatographed on a silica gel column packed and eluted
with 95:5 chloroform-methanol. Fractions containing a single tlc
form-methanol. Fractions whose tlc contained a spot with R 0.50
f
(silica gel, 95:5 chloroform-methanol) were pooled and concen-
trated to dryness to obtain the bispivaloyated bromo ester 19 (690
1
mg, 10%); H-nmr (deuteriochloroform): δ 1.3 (m, 18H, 2 x t-
Bu), 1.65-1.80 (m, 4H, β- and γ-CH ), 3.65-3.72 (m, 5H,
2
COOMe and δ-CH ), 4.4-4.6 (m, 3H, α-CH and bridge CH NH),
2
6.3 (m, 1H, bridge CH NH), 6.65-6.9 (m, 2H, 3'- and 5'-H), 7.56-
2
2
7.60 (m, 2H, 2'- and 6'-H), 7.68-7.82 (m, 4H, phthalimide ring
protons), 8.0 (s, 1H, t-BuCONH), 8.2 (d, 1H, CONH), 8.65 (s,
1H, t-BuCONH); ms (FAB): m/e 844 (M+Na). This material was
used directly in the next step.
spot with R 0.23 (silica gel, 9:1 chloroform-methanol) were
f
pooled and evaporated to dryness. The residue was dissolved in
a small volume of 9:1 chloroform-methanol and the solution was
added to a large volume of ether. The precipitate was collected
by filtration, washed with ether and dried in a vacuum oven to
obtain 17 as pale-yellow powder (213 mg, 47%); mp 190 °C dec;
ir: ν 3350, 3200, 1710, 1640, 1590, 1530, 1500, 1400, 1350,
Step 2. A stirred solution of 19 (650 mg, 0.79 mmole) in
tetrahydrofuran (100 mL) was cooled to 0 °C, and treated sequen-
tially with water (100 mL), palladium chloride (280 mg, 1.58
mmole), and sodium borohydride (299 mg, 7.9 mmole). After 20
minutes at 0 °C, the reaction mixture was left to stir at room tem-
perature for 4 hours. Thin-layer chromatography showed two new
spots with R 0.40 and R 0.23 (silica gel, 95:5 chloroform-
-1
1
1210 cm ; H-nmr (d -dimethylsulfoxide): δ 1.67-1.74 (m, 4H,
6
β- and γ-CH ), 3.57-3.61 (m, 5H, COOCH and δ-CH ), 4.31-
2
4.38 (m, 3H, bridge CH NH and α-CH), 6.6 (m, H, bridge
3
2
f
f
methanol), and no starting material (R 0.50). The reaction mixture
f
2
CH NH), 6.68-6.72 (m, 2H, 3'- and 5'-H), 6.95 (s, H, 6-H), 7.21
was filtered through Celite, the filter pad was washed with 1:1
methanol-water, and the tetrahydrofuran and methanol were
removed from the combined filtrate and wash by rotary evapora-
tion. The remaining aqueous solution was diluted with water (100
mL) and extracted with chloroform (2 x 200 mL). The combined
organic layers were washed with water (2 x 25 mL), dried over
anhydrous sodium sulfate, and evaporated under reduced pressure.
The residue was chromatographed on a column of flash-grade sil-
ica gel packed and eluted with 98:2 chloroform-methanol.
Evaporation of appropriately pooled fractions containing a single
2
(br s, NH ), 7.61-7.65 (m, 2H, 2'- and 5'-H), 7.80-7.85 (m, 4H,
2
phthalimide ring protons), 7.9 (br s, NH ), 8.23-8.27 (d, H,
2
CONH), 11.62 (br s, H, 7-NH); ms: m/e 579 (M+Na).
Step 3. A solution of 17 (50 mg, 0.09 mmole) in methanol (25
mL) containing a few drops of water was diluted with additional
water (25 mL), treated with solid barium hydroxide (80 mg, 0.25
mmole), and stirred at room temperature for 48 hours. Solid
ammonium bicarbonate (100 mg) was then added, and stirring
was continued for 30 minutes. The precipitated barium carbonate
was filtered, the solvents were evaporated under reduced pres-
sure, and the residue was dissolved in 10% ammonium hydrox-
ide. The solution was cooled in ice and adjusted to pH 4.5 with
10% acetic acid. The precipitate was collected by filtration
immediately, washed with water, and dried in a lyophilizer and
then in vacuo at 50 °C over phosphorous pentoxide to obtain 7 as
a white solid (30 mg, 70%); mp >200 °C, darkening above 190
tlc spot with R 0.40 afforded the debrominated bispivaloyl ester 20
1
f
as a pale-yellow powder (98 mg, 17%); H-nmr (deuteriochloro-
form): δ 1.3 (s, 18H, 2 x t-Bu), 1.65-1.80 (m, 4H, β- and γ-CH ),
2
3.65-3.70 (m, 5H, COOMe and δ-CH ), 4.6-4.8 (m, 3H, α-CH and
2
bridge CH NH), 5.8 (t, 1H, bridge CH NH), 6.56-6.62 (m, 2H, 3'-
2
2
and 5'-H), 7.01 (s, 1H, 6-H), 7.59-7.63 (m, 2H, 2'- and 6'-H), 7.67-
7.84 (m, 5H, t-BuCONH and phthalimide ring protons), 8.15 (d,
1H, CONH), 8.55 (s, 1H, t-BuCONH); FABMS: m/e 764 (M+Na).
This material was used directly in the next step.
Step 3. A solution of 20 (98 mg, 0.13 mmole) in a mixture of
methanol (12 mL) and 1 N sodium hydroxide (6 mL) was stirred at
room temperature for 5 days, then concentrated to dryness on the
°C; hplc: 21 min (C silica gel, 1 to 25% acetonitrile gradient
18
over 20 min in 0.1 M ammonium acetate, pH 7.5, 1 mL/min); ir:
-1
1
ν 3350. 2950, 1700, 1600, 1570, 1520, 1430, 1390 cm ; H-nmr
(d -dimethylsulfoxide): δ 1.54-1.95 (m, 4H, β- and γ-CH ), 3.16
(m, 2H, δ-CH ), 4.24-4.26 (m, 2H, bridge CH NH), 4.35 (m, H,
6
2
2
2

792
C. Vaidya, J. E. Wright, and A. Rosowsky
Vol. 41
1
rotary evaporator. The residue was taken up in water (5 mL), and
the solution was cooled on ice and adjusted to pH 4.5 with 1 N
hydrochloric acid. The precipitate was filtered, washed with water,
(301 mg, 77.5%); H-nmr (deuteriochloroform): δ 1.2 (m, 6H, 2
x CH CH ), 1.3 (s, 18H, 2 x t-Bu), 1.8 (m, 2H, β-CH ), 2.4 (t,
3
2
2
2H, γ-CH ), 4.0-4.2 (m, 4H, 2 x CH CH ), 4.6-4.8 (m, 3H, α-CH
2
3
and bridge CH NH), 5.8 (t, 1H, bridge CH NH), 6.7 (m, 2H, 3'-
2
dried in a vacuum oven, and purified by preparative hplc (C sil-
18
2
2
ica gel, 18% acetonitrile in 0.1 M ammonium acetate, pH 7.4, 10
mL/min). Appropriate fractions were pooled and lyophilized to
obtain 8 as a pale-yellow solid (25 mg, 33%); mp >200 °C, darken-
and 5'-H), 6.95 (s, 1H, 6-H), 7.4 (m, 2H, 2'- and 6'-H), 8.0-8.2 (m,
1H, t-BuCONH); FABMS: m/e 669 (M+H). This material was
used directly in the next step.
ing above 170 °C; analytical hplc: 8 min (C silica gel, 18% ace-
18
Step 3. A solution of 22 (300 mg, 0.4 mmole) in a mixture of
95% ethanol (40 mL) and 1 N sodium hydroxide (20 mL) was
stirred at room temperature for 4 days, then concentrated to dry-
ness on the rotary evaporator. The residue was taken up in water
(10 mL), and the solution was cooled on ice and adjusted to pH
4.5 with 1 N hydrochloric acid. The precipitate was collected by
filtration, washed with water, dried in a vacuum oven, and puri-
tonitrile in 0.1 M ammonium acetate, pH 7.4, 1 mL/min); ir: ν
-1
1
3300, 2940, 1620, 1550, 1500, 1400, 1300, 1250 cm ; H-nmr
(d -dimethylsulfoxide): δ 1.51-1.75 (m, 4H, β- and γ-CH ), 3.13
6
2
(m, 2H, δ-CH ), 4.31 (m, 3H, α-CH and bridge CH NH), 5.99 (br
2
s, NH ), 6.48 (br s, NH ), 6.69-6.73 (m, 2H, 3'- and 5'-H), 6.77 (s,
2
2
2
1H, 6-H), 7.31-7.45 (m, 3H, phthaloyl ring protons), 7.65-7.69 (m,
3H, 2'-, 6'-, and phthaloyl ring proton), 8.15 (d, 1H, CONH), 8.22
(m, 1H, phthaloyl CONH); HRMS: m/e found 578.1823 (M+H),
calcd 578.1822. Because it was purified to homogeneity by prepar-
fied by preparative hplc (C silica gel, 12% acetonitrile in 0.1 M
18
ammonium acetate, pH 7.4, 10 mL/min). Appropriate fractions
were pooled and lyophilized to obtain 12 as a pale-yellow solid
(33 mg, 17%); mp >200 °C, darkening above 175 °C; ir: ν 3300,
1
ative hplc and its HRMS and H-nmr data were fully consistent
-1
1
with the assigned molecular structure, elemental microanalysis
was omitted and the product was used for DHFR binding, trans-
port, and cell growth assays directly.
1650, 1600, 1560, 1500, 1450, 1450, 1400, 1300, 1250 cm ; H-
nmr (d -dimethylsulfoxide): δ 1.8-2.0 (m, 2H, β-CH ), 2.3 (t,
6
2
2H, γ-CH ), 4.0-4.2 (m, 3H, α-CH and bridge CH NH), 6.0 (br s,
2
2
NH ), 6.5 (br s, NH ), 6.65 (t, 1H, bridge CH NH), 6.73-6.77 (m,
2
2H, 3'- and 5'-H), 6.82 (s, 1H, 6-H), 7.66-7.70 (m, 2H, 2'- and 6'-
2
2
N-[4-[[N-(2, 4-Diaminothieno[2, 3-d]pyrimidin-5-yl)methyl]-
amino]benzoyl]-L-glutamic Acid (12).
H); 8.1 (m, 1H CONH); FABMS: m/e 445 (M+H).
.
Step 1. Diethyl N-(4-aminobenzoyl)-L-glutamate (15) (1.36 g,
4.22 mmole) and sodium bicarbonate (5.2 g, 63 mmole) were
added sequentially to a solution of 18 (3 g, 6.3 mmole) in dry
dimethylformamide (15 mL) and the reaction mixture was stirred
at 55 °C for 24 hours. The solvent was removed by rotary evap-
oration, and the residue was taken up in ethyl acetate (200 mL).
The solution was washed with water (100 mL), dried over anhy-
drous sodium sulfate, and evaporated. The product was taken up
in a minimal volume of 99:1 chloroform-methanol and chro-
matographed on a flash-grade silica gel column packed and
eluted with the same solvent mixture. Fractions containing a tlc
Anal. Calcd for C H N O S 0.7H O: C, 49.92; H, 4.71; N,
19 20
6
5
2
18.38. Found: C, 49.60; H, 4.53; N, 18.38.
Acknowledgments.
This work was supported by grant RO1-CA25394 from the
National Cancer Institute, NIH, DHHS. Technical assistance in
the cell culture and transport experiments was provided by Ying-
Nan Chen.
REFERENCES AND NOTES
spot with R 0.48 (silica gel, 95:5 chloroform-methanol) and a
f
faint slower-moving impurity were pooled and concentrated to
[1a] A. Rosowsky, H. Bader, C. A. Cucchi, R. G. Moran, W.
Kohler, and J. H. Freisheim, J. Med. Chem., 31, 1332 (1988); [b] A.
Rosowsky, H. Bader, and R. A. Forsch, . Pteridines, 1, 91 (1989).
[2] A. Rosowsky, Current Med. Chem., 6, 329 (1999).
[3] A. Rosowsky, H. Bader, J. E. Wright, K. Keyomarsi, and L. H.
Matherly, J. Med. Chem., 37, 2167 (1994)
[4] A. Rosowsky, C. M. Vaidya, H. Bader, J. E. Wright, and B. A.
Teicher, J. Med. Chem., 40, 286 (1997).
[5] A. Rosowsky, J. E. Wright, C. M. Vaidya, H. Bader, R. A.
Forsch, C. Mota, J. Pardo, C. S. Chen, and Y. N. Chen, J. Med. Chem., 41,
5310 (1988)
[6] A. Rosowsky, J. E. Wright, C. M. Vaidya, R. A. Forsch, and
H. Bader, J. Med. Chem., 43, 1620 (2000)
[7] C. Vaidya, J. E. Wright, and A. Rosowsky. J. Med. Chem., 45,
1690 (2002)
[8] J. E. Wright, C. M. Vaidya, Y. N. Chen, and A. Rosowsky,
Biochem. Pharmacol., 60, 41 (2000).
[9] J. E. Wright, G. K. Yurasek, Y. N. Chen, and A. Rosowsky,
Biochem. Pharmacol. 65, 1427 (2003).
dryness to obtain the bispivaloyl bromo diester 21 (327 mg,
1
10%); H-nmr (deuteriochloroform): δ 1.1 (m, 6H, 2 x CH CH ),
3
2
1.3 (s, 18H, 2 x t-Bu), 1.8 (m, 2H, β-CH ), 2.4 (t, 2H, γ-CH ),
2
2
4.0-4.2 (m, 4H, 2 x CH CH ), 4.4-4.6 (m, 3H, α-CH and bridge
3
CH NH), 6.25 (m, 1H, bridge CH NH), 6.6 (m, 2H, 3'- and 5'-H),
2
2
2
7.4 (m, 2H, 2'- and 6'-H), 8.0 (s, 1H, t-BuCONH), 8.15 (d, 1H,
CONH), 8.8 (s, 1H, t-BuCONH); ms (FAB); m/e 748 (M+H).
This material was used directly in the next step.
Step 2. A stirred solution of 21 (433 mg, 0.58 mmole) in
tetrahydrofuran (65 mL) was cooled to 0 °C, and treated sequen-
tially with water (65 mL), palladium chloride (205 mg, 1.16
mmole), and sodium borohydride (219 mg, 5.8 mmole). After 20
minutes at 0 °C, the reaction mixture was left to stir at room tem-
perature for 4 hours. Thin-layer chromatography showed a new
spot with R 0.40 (silica gel, 95:5 chloroform-methanol), along
f
with disappearance of the starting material (R 0.48). The reac-
f
tion mixture was filtered through Celite, the filter pad was
washed with 1:1 methanol-water, and the tetrahydrofuran and
methanol were removed from the combined filtrate and wash by
rotary evaporation. The remaining aqueous solution was diluted
with water (50 mL) and extracted with chloroform (2 x 100 mL).
The combined organic layers were washed with water (2 x 25
mL), dried over anhydrous sodium sulfate, and evaporated under
reduced pressure to obtain the debrominated bispivaloyl ester 22
[10] A. Gangjee, R. Devraj, J. J. McGuire, R. L. Kisliuk, S. F.
Queener, and L. R. Barrows, J. Med. Chem., 37, 1169 (1994).
[11] A. Gangjee, F. Mavandadi, S. F. Queener, and J. J McGuire, J.
Med. Chem., 38, 2158 (1995).
[12] E. C. Taylor, H. Patel, G. Sabitha, and R. Chaudhari, R.
Heterocycles, 4, 349 (1996).
[13] E. E. Swayze, J. M. Hinkle, and L. B. Townsend, in Nucleic
Acid Chemistry, Part 4; L. B. Townsend and R. S. Tipson, R., Eds.,

Sep-Oct 2004
New Pyrrolo[2,3-d]pyrimidine and Thieno[2,3-d]pyrimidine Inhibitors
793
Wiley-Interscience, 1991, pp 16-17.
obtained using a radioactive substrate, and radiolabeled 7 and 8 were not
[14a] A. Rosowsky, A. T. Papoulis, and S. F. Queener, J. Med.
Chem. , 4 0, 3694 (1997); [b] A. Rosowsky, A. T. Papoulis, and S. F.
Queener, J. Med. Chem., 46. 3424 (2003).
[15] O. H. Straus and A. Goldstein, J. Gen. Physiol., 26, 559
(1943).
[16] The spatial orientation of the side chain in the furan analogue
9 in its ternary complex with NADPH and human DHFR has been found
by X-ray crystallographic analysis to be different from that of the side
chain in MTX [17], and this may also be true in the case of the thiophene
analogue 12.
[17] V. Cody, N. Galitsky, J. R. Luft, W. Pangborn, A. Gangjee, R.
Devraj, S. F. Queener, and R. L. Blakley, Acta Crystal., D53, 638 (1997).
[18] Although this unanticipated result was gratifying, it does not
necessarily signify that 7 and 8 accumulate to higher steady-state levels in
the cell than PT523. High levels of cellular uptake of an antifolate via the
available, measurement of the V /K for these compounds was outside
max m
the scope of this work.
[19] Because the binding of diaminoheterocyclic antifolates to the
active site of DHFR is thermodynamically reversible except in those
instances where a covalent bond is formed, it is generally considered that
an excess of free drug is important as a way to minimize dissociation of
the antifolate from the enzyme, thereby maximally prolonging DNA syn-
thesis inhibition.
[20] F. M. Sirotnak, Pharm. Ther., 8, 71 (1980).
[21] F. Itoh, Y. Yoshioka, K. Yukishige, S. Yoshida, K. Ootsu, and
H. Akimoto, Chem. Pharm. Bull., 49, 1270 (2000).
[22] B. Roth, R. Laube, M. Y. Tidwell, and B. S. Rauckman, J.
Org. Chem., 45, 3651 (1980).
[23] Although the preparation of 2,4-diamino-5-methylthieno[2,3-
d]pyrimidine from 2,4-diamino-6-mercaptopyrimidine and chloroacetone
RFC requires not only a high affinity (low K ) for the carrier protein but
described in ref. 14a was originally said to work with KHCO as the base,
m
3
also a high V , the latter of which is then used to calculate the more rel-
this statement was an error [14b] and we subsequently returned to the
original procedure using NaOMe [22].
max
evant first-order rate constant (V /K ). Because V is typically
max
max
m