Synthesis and in vitro antifolate activity of rotationally restricted aminopterin and methotrexate analogues

Article abstract of DOI:10.1021/jm040122s

Heretofore unknown analogues of aminopterin (AMT) and methotrexate (MTX) in which free rotation of the amide bond between the phenyl ring and amino acid side chain is prevented by a CH₂ bridge were synthesized and tested for in vitro antifolate activity. The K_i of the AMT analogue (9) against human dihydrofolate reductase (DHFR) was 34 pM, whereas that of the MTX analogue (10) was 2100 pM. Both compounds were less potent than the parent drugs. However, although the difference between AMT and MTX was <2-fold, the difference between 9 and 10 was 62-fold, suggesting that the effect of N ¹⁰-methyl substitution is amplified in the bridged compounds. The K_i values of 9 and 10 as inhibitors of [³H]MTX influx into CCRF-CEM human leukemia cells via the reduced folate carrier (RFC) were 0.28 and 1.1 μM, respectively. The corresponding K_i and K_t values determined earlier for AMT and MTX were 5.4 and 4.7 μM, respectively. Thus, in contrast to its unfavorable effect on DHFR binding, the CH₂ bridge increased RFC binding. In a 72 h growth assay with CCRF-CEM cells, the IC₅₀ valuess of 9 and 10 were 5.1 and 140 nM, respectively, a 27-fold difference that was qualitatively consistent with the observed combination of weaker DHFR binding and stronger RFC binding. Although rotationally restricted inhibitors of other enzymes of folate pathway enzymes have been described previously, 9 and 10 are the first reported examples of DHFR inhibitors of this type.

Full text of DOI:10.1021/jm040122s

6958
J. Med. Chem. 2004, 47, 6958-6963
Synthesis and in Vitro Antifolate Activity of Rotationally Restricted
Aminopterin and Methotrexate Analogues
Andre Rosowsky,* Ronald A. Forsch, and Joel E. Wright
Dana-Farber Cancer Institute and Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,
Boston, Massachusetts 02115
Received June 25, 2004
Heretofore unknown analogues of aminopterin (AMT) and methotrexate (MTX) in which free
rotation of the amide bond between the phenyl ring and amino acid side chain is prevented by
a CH2 bridge were synthesized and tested for in vitro antifolate activity. The Ki of the AMT
analogue (9) against human dihydrofolate reductase (DHFR) was 34 pM, whereas that of the
MTX analogue (10) was 2100 pM. Both compounds were less potent than the parent drugs.
However, although the difference between AMT and MTX was <2-fold, the difference between
9 and 10 was 62-fold, suggesting that the effect of N10-methyl substitution is amplified in the
bridged compounds. The Ki values of 9 and 10 as inhibitors of [3H]MTX influx into CCRF-
CEM human leukemia cells via the reduced folate carrier (RFC) were 0.28 and 1.1 µM,
respectively. The corresponding Ki and Kt values determined earlier for AMT and MTX were
5.4 and 4.7 µM, respectively. Thus, in contrast to its unfavorable effect on DHFR binding, the
CH2 bridge increased RFC binding. In a 72 h growth assay with CCRF-CEM cells, the IC50
valuess of 9 and 10 were 5.1 and 140 nM, respectively, a 27-fold difference that was qualitatively
consistent with the observed combination of weaker DHFR binding and stronger RFC binding.
Although rotationally restricted inhibitors of other enzymes of folate pathway enzymes have
been described previously, 9 and 10 are the first reported examples of DHFR inhibitors of this
type.
Introduction
in vitro antifolate activity of these compounds in com-
parison with AMT and MTX. To our knowledge, 9 and
10 are the first examples of 2,4-diamino antifolates with
a rotationally restricted glutamate side chain.
Several examples of analogues of folic acid (1) in
which the B-ring, 9,10-bridge, or pABA moiety is modi-
fied, and in addition the glutamyl nitrogen is tethered
to the ortho position of the phenyl ring via a CH2 bridge
to prevent free rotation of the amide bond, have been
described in the literature. Specifically they include the
2-methylquinazolin-4(3H)-one 2,1 the 3-methylbenzo[f]-
quinazolin-1(2H)-one 3,2 the 2-aminopyrido[2,3-d]pyri-
midin-4(3H)-ones 4,3 5,4 and 6,4 the 2-aminopyrrolo[2,3-
d]pyrimidin-4(3H)-one 7,3 and the 2-aminothieno[2,3-
d]pyrimidin-4(3H)-one 8.3 Compounds 2 and 3 were
designed as inhibitors of the enzyme thymidylate syn-
thase (TS), compounds 5 and 6 as conformationally
restricted analogues of (6R,6S)-2,4-diamino-5-deaza-
5,6,7,8-tetrahydrofolate (DDATHF, LY237147), an in-
hibitor of the key purine pathway enzyme glycinamide
ribotide formyltransferase (GARFT), and compound 7
as an analogue of the novel multitargeted antifolate
pemetrexed (MTA, LY231514), whose polyglutamates
are potent inhibitors not only of TS and GARFT but also
to a lesser extent dihydrofolate reductase (DHFR). The
structures of 1-8 are shown in Figure 1.
Chemistry
The synthesis of 9 began with methyl 2-(bromo-
methyl)-4-nitrobenzoate (13), which was converted to
the key lactam 151,4 as summarized in Scheme 1. This
was followed by condensation with 2,4-diamino-6-bro-
momethylpteridine (16)5 and saponification of the re-
sulting ester 17 with Ba(OH)2. Although it proved
expedient to run through the sequence from 13 to 17
with only partial purification of the intermediates 14
and 15, the end product 9 was obtained in very pure
state by a two-stage process consisting of preparative
HPLC on C18 silica gel followed by ion-exchange chro-
matography on DEAE-cellulose. Lyophilization of ap-
propriately pooled fractions afforded 9 as a yellow
powder whose microchemical analysis showed it to be
a hydrated partial ammonium salt and whose 200 MHz
1H NMR spectrum in d6-DMSO solution showed the
requisite features including a pair of doublets centered
at δ 4.25 and δ 4.49 for the diastereotopically non-
equivalent CH2 protons of the isoindolinone ring. Like-
wise consistent with structure of 9 were a doublet at δ
7.39 and a singlet at δ 8.39, which could be assigned to
the isoindolinone H-7 and the pteridine H-7 protons,
respectively. The isoindolinone H-4 and H-6 aromatic
protons gave rise to a poorly resolved A2B2 pattern that
was partially obscured by a broad singlet corresponding
to one of the NH2 groups. The overall yield of 9 from
the amino lactam 15 was 28%.
Rotationally restricted inhibitors of a third key en-
zyme of folate metabolism, dihydrofolate reductase (DH-
FR), have, until now, not been described in the litera-
ture. In the present paper, we report the synthesis of
compounds 9 and 10, which may be viewed as rotation-
ally restricted analogues of the classical DHFR inhibi-
tors aminopterin (AMT, 11) and methotrexate (MTX,
12), respectively. Also reported are initial data on the
For the synthesis of the N10-methyl analogue 10, we
initially considered making the previously unknown
* Corresponding author. Phone: 617-632-3117. Fax: 617-632-2410.
E-mail: andre_rosowsky@dfci.harvard.edu.
10.1021/jm040122s CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/02/2004

Aminopterin and Methotrexate Analogues
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 27 6959
Figure 1. Structures of folic acid (1) and its previously reported analogues 2-8 with a rotationally restricted side chain amide
bond.
low yield of monomethylated product likely to be
encountered in the reaction of 15 with methyl iodide,
we chose instead to start with commercially available
4-bromo-2-methylbenzoic acid (19) and take advantage
of the elegant Pd0 catalyzed amination procedure re-
cently developed by Hartwig and co-workers.6 As shown
in Scheme 1, 19 was esterified (SOCl2/MeOH), and the
resulting ester 20 was condensed with N-benzylmethyl-
amine in the presence of tris(dibenzylideneacetone)-
dipalladium(0) (Pd2dba3), tri-tert-butylphosphine, and
cetyltrimethylammonium hydroxide as a phase transfer
catalyst in a well-stirred mixture of aqueous NaOH and
toluene. This afforded 21, which was then converted
stepwise to 22 and 23 by catalytic hydrogenolysis (H2/
Pd-C) and acylation with trifluoroacetic anhydride in
pyridine, respectively. Bromination of the methyl group
with N-bromosuccinimide (NBS) and benzoyl peroxide,
followed by reaction of the crude bromide 24 with
diethyl L-glutamate and removal of the trifluoroacetyl
group with K2CO3 in aqueous EtOH, afforded the
methylamino lactam 18. Condensation of 18 and 16,
followed by saponification using Ba(OH)2, yielded the
diester 25 and diacid 10, respectively. The yield of 10
Figure 2. Structures of aminopterin (AMT, 9), methotrexate
based on 18 was 48%. In contrast to 9, purification of
10 was accomplished by preparative HPLC on C18 silica
gel without need of additional ion-exchange chromatog-
raphy. Microchemical analysis indicated that in this
case the product contained approximately two molecules
(MTX, 10), and their analogues 11 and 12 with a rotationally
restricted side chain amide bond.
N-methylamino lactam 18 by direct methylation of 15.
However, given the separation problem and resultant

6960 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 27
Rosowsky et al.
Scheme 1
of water but no ammonia. Ultraviolet spectra of 9 and
10 showed same type of UV shift as was reported
originally in MTX versus AMT7 and has since been
observed in other cases involving replacement of the
pteridine by a different 2,4-diaminoheterocyclic ring
system8,9 or the glutamate side chain by another amino
acid.10,11 The likely reason for this difference between
N10-methyl and N10-unsubstituted AMT and MTX ana-
logues is that the methyl group enhances the basicity
of the nitrogen, thereby favoring delocalization of the
lone pair of electrons into the benzoyl moiety via the
coplanar resonance structures 9a and 10a.
ronment in both the phenyl and isoindolinone ring in
comparison with 9.
Formation of 18 from 24 and diethyl l-glutamate was
found to yield two minor byproducts that had to be
carefully removed by chromatography in order to avoid
carrying them to the next step. From the 1H NMR
spectrum, it appears that these byproducts were the
lactam methyl and ethyl esters 26 and 27, arising from
24 by nucleophilic attack of the glutamate amino group
on the BrCH2 group, ring closure onto the γ-glutamyl
carbon, and partial transesterification of the aromatic
CO2Me group during removal of the N-trifluoroacetyl
group with EtOH as the solvent. Whereas the chemical
shift for the CH2 group in isoindoline 15 was part of a
complex multiplet at δ 4.1-4.64 and this feature was
replicated in the spectrum of 18, the 1H NMR spectrum
of the putative byproducts 26 and 27 displayed a clearly
separated CH2 singlet downfield at δ 5.13, in agreement
with the presence of an electron-withdrawing ester
group at the ortho position of the phenyl ring. Although
analogous byproducts could have formed also during the
earlier synthesis of 15 from 13,4 this possibility was not
investigated.
The 1H NMR spectrum of 10 was different from that
of 9 in two notable respects. Although the isoindolinone
CH2 group of 9 gave rise to a pair of well-resolved one-
proton doublets, the spectrum of 10 contained only a
two-proton singlet, suggesting a smaller degree of
nonequivalence. Consistent with the ring-closed struc-
ture of 10 was the finding that the isoindolinone CH2
protons gave a 1H NMR signal (δ 4.38) in the same
general region as in 9, whereas if ring opening had
occurred during saponification then the CH2 group next
to the nonacylated nitrogen would presumably be
further upfield. Furthermore, although the exchange-
able peak for one of the NH2 groups in 9 overlapped the
H-4 and H-6 protons, in the spectrum of 10 this NH2
group appeared to be shifted downfield so that it
An alternative route to 9 that we also explored began
with the condensation of 15 with 2-amino-5-bromo-
methylpyrazine-3-carbonitrile (28) to form the substi-
tuted pyrazine 29. This was followed by ring closure
with guanidine and saponification with NaOH (Scheme
2). However the bioassay data for this product (Table
1) indicated that racemization had probably taken place
at the glutamyl R-carbon, just as Mautner and co-
1
overlapped the H-7 proton. Overall, the H NMR data
suggested a subtle perturbation of the magnetic envi-

Aminopterin and Methotrexate Analogues
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 27 6961
Scheme 2
the observed DHFR binding difference between the
bridged and the nonbridged compounds reflects not only
the inability of the amide bond to rotate freely but also
the fact that the phenyl ring is 6′-substituted. A
comparative X-ray crystallographic analysis of the
orientation of the amide carbonyl oxygen atom relative
to the R-COOH group in the bridged and nonbridged
compounds in ternary complexes with DHFR and NAD-
PH would shed light on this complex issue and is
currently planned.
In addition to the effect on DHFR binding, it was
equally of interest to assess whether the introduction
of a CH2 bridge to restrict free rotation of the side chain
amide bond of AMT and MTX would have a deleterious
effect on transport via the reduced folate carrier (RFC),
just as it does on DHFR binding. Thus, we determined
the Ki of 9 and 10 inhibitors of [3H]MTX influx into
CCRF-CEM human leukemic lymphoblasts over a 60-s
period as previously described.14 As shown in Table 1,
the Ki of 9 was 0.28 ( 0.10 µM, whereas that of 10 was
1.1 ( 0.11 µM, a 4-fold difference again favoring the N10-
unsubstituted analogue. The Ki value that we had
obtained earlier for AMT was 5.4 ( 0.09 µM, and the
Kt for MTX (equivalent to the Ki in this case) was 4.7 (
1.3 µM (Table 1). Thus, 9 and 10 were both able to bind
better to the RFC than their nonbridged counterparts.
Moreover, rotational restriction of the side chain amide
bond appeared to have the opposite effect on RFC
binding than on DHFR binding. This suggests that, in
contrast to DHFR binding, a coplanar arrangement
between the side chain amide bond and the phenyl ring
may be favorable where RFC binding is concerned. The
low Ki of 9 was gratifying, because until now the only
DHFR inhibitors we had found to give a Ki of <0.4 µM
in this assay were analogues of NR-(4-amino-4-deoxy-
pteroyl)-Nδ-hemiphthaloyl-L-ornithine (PT523) in which
the B-ring (5- or 5,8-dideaza) or 9,10-bridge (10-deaza
or 5,8,10-trideaza) are modified but the amide bond is
not locked in place. However, whether restricted rota-
tion of the amide bond would also promote RFC binding
when the side chain is Nδ-hemiphthaloyl-L-ornithine
remains to be determined.
With the knowledge that 9 and 10 were weaker
DFHR inhibitors than AMT and MTX, but could bind
somewhat better to the RFC, it was of interest to assess
their ability to inhibit cell growth in culture under the
standardized conditions routinely used in our laboratory
in the past.15 As shown in Table 1, the IC50 values of 9
and 10 against CCRF-CEM cells during 72 h of
exposure were 5.1 ( 0.25 and 140 ( 5.0 nM, as
compared with previously determined values of 4.4 (
0.10 and 14 ( 2.6 nM for AMT and MTX, respectively.
It thus appeared that the increase in RFC binding of 9
relative to AMT was able to approximately offset its
decrease in DHFR binding, whereas in the case of 10
versus MTX the modest gain in RFC binding was not
enough to offset a much larger decrease in DHFR
binding.
Table 1. In Vitro Antifolate Activity of Rotationally Restricted
AMT and MTX Analoguesa
growth IC50
(nM, 72 h)
cmpd
DHFR Ki (pM)
influx Ki (µM)
9
34 ( 3.0
0.28 ( 0.10
0.53 ( 0.03
1.1 ( 0.11
5.1 ( 0.25
8.4 ( 0.7
140 ( 5.0
4.4 ( 0.10
14 ( 2.6
rac-9
77 ( 1.2
10
AMT (11)
MTX (12)
2100 ( 200
3.7 ( 0.35
4.8 ( 0.45
5.4 ( 0.09
4.7 ( 1.3 (Kt)
a All of the numbers are rounded off to two significant figures
and are the means ( SD (n g 3). Data for AMT and MTX are
taken from refs 13b and 14. The racemic nature of rac-9 obtained
via Scheme 2 is indicated by its roughly 2-fold lower potency in
all three assays relative to 9 obtained via Scheme 1.
workers12a,b had observed earlier in their synthesis of
aminopterin and 10-thiaaminopterin from 26. Although
we had hoped that the absence of an ionizable hydrogen
on the amide group of 27 might prevent such racem-
ization, this turned out not to be the case.
Antifolate Activity. Because there was no informa-
tion in the literature on the effect that a CH2 bridge
might have on the biological activity of AMT and MTX,
the ability of 9 and 10 to bind to human DHFR was
assessed spectrophotometrically as described previ-
ously,13 and the Ki values were compared with those
that we had obtained earlier for AMT and MTX. As
shown in Table 1, the Ki of 9 was 34 ( 3.0 pM and that
of 10 was 2100 ( 200 pM. Both compounds were less
potent than their nonbridged counterparts, but there
was an unexpected difference between them in that 10
was 440 times less potent than MTX (Ki ) 4.8 ( 0.45
pM), whereas the difference in Ki between 9 and AMT
(Ki ) 3.7 ( 0.35 pM) was only 9-fold. Thus. N10-
methylation appeared to produce a 62-fold decrease in
binding when rotation about the side chain amide bond
was blocked by a CH2 bridge to the ortho position of
the phenyl ring, but almost no decrease was observed
in the absence of a bridge. This suggests that, for
optimal binding to the enzyme, the amide bond of both
MTX and AMT may need to be able to point further out
of the plane of the phenyl ring than is possible when
the amide nitrogen is part of a fused five-membered ring
and that this requirement is more easily met when N10
is unsubstituted. However, it cannot be ruled out that
As indicated in Table 1, the Ki values for DHFR
inhibition and [3H]MTX influx as well as the IC50 value
for cell growth inhibition were all ca. 2-fold higher for
the product obtained via Scheme 2 than for the one
obtained via Scheme 1. We interpreted these results to
mean that significant racemization had taken place at

6962 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 27
Rosowsky et al.
isoindolinone H-7), 8.70 (s, 1H, pteridine H-7). Anal. (C20H20-
N8O5‚2.75H2O) C, H, N.
the R-carbon under the highly basic conditions of
guanidine cyclization and moreover that essentially all
of the the biochemical and biological activity was
provided by the diastereomer with the “natural” con-
figuration. In the classic work of Mautner and co-
workers12a,b on the synthesis of AMT analogues by this
method, racemization was inferred from the absence of
optical activity in the product. The present findings,
based on in vitro biochemical and biological evidence
rather than physical data, complement the earlier study
and confirm that heating with guanidine must be
avoided if a chiral carbon is already present.16
The lower degree of proportionality between DHFR
binding, RFC binding, and cytotoxicity in the case of
10 versus 9 presumably reflects (a) differences in
substrate activity of the two compounds for γ-folylpoly-
glutamate synthetase (FPGS),17 (b) the nature of the
product(s) of the FPGS reaction and their relative ability
to inhibit DHFR,18 (c) the ability of γ-folylpolyglutamate
hydrolase (FPGH) to cleave the γ-diglutamyl derivatives
back to the original drug,19 and (d) the possible sub-
strate activity of each species for one or more of the
multidrug resistance proteins (MRPs) that have recently
been shown to mediate MTX efflux.20 A detailed analysis
of the pharmacodynamics of 9 and 10 to address these
complex interdependent variables would be of interest
but was beyond the scope of this study.
2-R,S-[5-[(2,4-Diaminopteridin-6-yl)methylamino]-2,3-
dihydro-1-oxo-2(1H)-isoindolyl]glutaric Acid (rac- 9).
Metallic Na (37 mg, 1.61 mmol) was dissolved in EtOH (10
mL), solid guanidine hydrochloride (152 mg, 1.6 mmol) was
added, and after 5 min of being swirled, the mixture was added
to the ethanolic solution of 29 obtained after workup of the
reaction of 15 with 28 (cf. Supporting Information). The
reaction mixture was stirred under reflux for 20 h, then cooled
to room temperature, treated with 1 M NaOH (5 mL), stirred
for 5 min, and evaporated under reduced pressure. The residue
was subjected to preparative HPLC (C18 silica gel, 3.3% MeCN
in 0.1 M NH4OAc, pH 6.9, 1.0 mL/min), and appropriately
pooled fractions (Rf ) 14 min, dual detection at 280 and 370
nm) were concentrated by rotary evaporation followed by
freeze-drying. The residue was taken up in dilute NH4OH, a
small amount of insoluble material was filtered off, and 10%
AcOH was added to reprecipitate the product. Freeze-drying
of the entire acidified mixture afforded rac-9 as a yellow
powder (124 mg, 23%) whose IR spectrum, 1H NMR spectrum,
and HPLC elution time from the C18 silica gel matched those
of 9 obtained from 14 and 16‚HBr as described above. Anal.
(C20H20N8O5‚0.25NH3‚4.5H2O) C, H, N.
2-S-[5-[N-(2,4-Diaminopteridin-6-yl)methyl)-N-methyl-
amino]-2,3-dihydro-1-oxo-2(1H)-isoindolyl]glutaric Acid
(10). From purified 18, prepared from 22-24 (cf. Supporting
Information), this compound was obtained via 25 by essentially
the same procedure as 9 except that the reaction mixture was
heated at 50 °C for only 3.5 days. After removal of the BaCO3,
the EtOH was removed from the aqueous EtOH filtrate by
rotary evaporation, a small amount of fine white solid was
filtered off, the filtrate was allowed to stand at room temper-
ature overnight, and an additional small quantity of fine solid
was removed. Analytical HPLC (C18 silica gel, 8% MeCN in
0.1 M NH4OAc, pH 7.4) showed a major peak eluting at 13
min, along with two faster-moving impurities. For preparative
HPLC, the initial MeCN concentration was reduced to 4% and
was increased to 8% as the main peak began to elute.
Appropriately pooled eluates were concentrated and freeze-
dried to obtain 10 as yellow solid (75 mg, 48%). Mp: >250 °C.
IR (KBr): ν 3340, 3150br, 2920br, 1615, 1560, 1505, 1450,
1405, 1385, 1360, 1300, 1260, 1205 cm-1. UV λmax (0.05 M
phosphate buffer, pH 7.4): 226 nm (ꢀ 25 800), 258 (25 500),
308 (24 000), 370 (8100). 1H NMR (DMSO-d6): δ 2.27 (m, 4H,
â- and γ-CH2), 3.38 (s, NMe, overlapping the H2O peak), 4.38
(s, 2H, isoindolinone H-3), 4.86 (m, 3H, CH2NMe and R-CH),
6.69 (br s, NH2), 7.00 (m, 2H, 4-H and 6-H), 7.52 (br s, NH2,
overlapping 7-H, d, J ) 8 Hz), 8.65 (s, 1H, 7-H). Anal.
(C21H22N8O5‚2.2H2O) C, H, N.
Experimental Section
2-S-[5-[(2,4-Diaminopteridin-6-yl)methylamino]-2,3-di-
hydro-1-oxo-2(1H)-isoindolyl]glutaric Acid (9). Compound
15 (200 mg, 0.58 mmol, as the crude product of reduction of
14 with SnCl2)4 and 16‚HBr (212 mg, 0.58 mmol, assumed to
be solvated with i-PrOH as reported in the literature)5 were
dissolved in DMAC (5 mL), and the solution was heated at 50
°C for 7 days. The reaction mixture, containing diester 17, was
diluted with 95% EtOH (30 mL) and H2O (20 mL), and solid
Ba(OH)2‚8H2O (1.26 g, 4.0 mmol) was added. The mixture was
then stirred at room temperature for 3 days, and a solution of
(NH4)2CO3 (384 g, 4.0 mmol) in H2O (5 mL) was added. After
the solution was stirred for another 10 min, the insoluble
BaCO3 was filtered and the filter cake was washed with H2O.
The filtrate was concentrated by rotary evaporation until most
of the EtOH was removed, and a small amount of insoluble
material was filtered off. The final pH of the filtrate was ca.
8.0. Analytical HPLC (C18 silica gel, 5% MeCN in 0.1 M NH4-
OAc, pH 7.4, 1.0 mL/min) showed a major peak eluting at 10
min along with a slower peak at >20 min, with each peak
being detectable at both 280 and 370 nm. A fast peak (2 min)
was also seen at 280 nm but not 370 nm. The entire sample
was then purified on a preparative scale using the same eluent
buffer except that the MeCN concentration was initially
decreased to 3% and then raised to 5% as the main peak began
to elute. Appropriately pooled eluates were concentrated by
rotary evaporation followed by freeze-drying. The residue was
taken up in 10% NH4OH, the solution was filtered to remove
a small amount of adventitious C18 silica gel, and 10% AcOH
was added to reprecipitate the product. The resulting solid was
filtered and subjected to final purification by ion-exchange
chromatography on DEAE-cellulose (HCO3-1 form, 1.5 cm ×
19 cm, 0.4 M NH4HCO3). Appropriately pooled fractions were
again concentrated and freeze-dried to obtain 9 as a yellow
powder (81 mg, 28%). Mp: >250 °C. IR (KBr) ν: 3350, 3200,
1645, 1615, 1540, 1505, 1450, 1405, 1385, 1290, 1260, 1225,
1130-1100br, 995, 810, 770, 690, 620 cm-1. UV λmax (0.05 M
phosphate buffer, pH 7.4): 223 nm (ꢀ 23 700), 259 (24 000),
285 (23 300), 370 (8000). 1H NMR (DMSO-d6): δ 2.49 (m, 4H,
â- and γ-CH2), 4.25 (d, 1H, isoindolinone H-3R), 4.49 (m, 4H,
CH2NH, R-CH, isoindolinone H-3â), 6.57 (s, 2H, NH2), 6.80
(4H, NH2, isoindolinone H-4 and H-6), 7.39 (d, J ) 8 Hz, 1H,
Acknowledgment. This work was supported in part
by Grant RO1-CA25394 from the National Cancer
Institute, National Institutes of Health, Department of
Health and Human Services. The authors are indebted
to Ying-Nan Chen for her technical assistance in the
transport and cytotoxicity experiments.
Supporting Information Available: Additional experi-
mental details for the synthesis of compounds 20-23 from 19,
compound 18 from 23 via 24, and compound 29 from 15. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Marsham, P. R.; Jackman, A. L.; Hayter, A. J.; Daw, M. R.;
Snowden, J. L.; O’Connor, B. M.; Bishop, J. A. M.; Calvert, A.
H.; Hughes, L. R. Quinazoline Antifolate Thymidylate Synthase
Inhibitors: Bridge Modifications and Conformationally Re-
stricted Analogues in the C2-Methyl Series. J. Med. Chem. 1991,
34, 2209-2218.
(2) Pendergast, W.; Dickerson, S. H.; Dev, I. K.; Ferone, R.; Duch,
D. S.; Smith, G. K. Benzo[f]quinazoline Inhibitors of Thymidylate
Synthase: Methyleneamino-Linked Aroylglutamate Derivatives.
J. Med. Chem. 1994, 37, 838-844.

Aminopterin and Methotrexate Analogues
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 27 6963
(3) Taylor, E. C.; Jennings, l. D.; Mao, Z.; Hu, B.; Jun, J.-G.; Zhou,
P. Synthesis of Conformationally Constrained Glutamate Ana-
logues of the Antitumor Agents DDATHF, LY254155, and
LY231514. J. Org. Chem. 1997, 62, 5392-5403.
(4) Rosowsky, A.; Forsch, R. A.; Null, A.; Moran, R. G. 5-Deazafolate
Analogues with a Rotationally Restricted Glutamate or Orni-
thine Side Chain: Synthesis and Binding Interaction with
Folylpolyglutamate Synthetase. J. Med. Chem. 1999, 42, 3510-
3519.
(5) Piper, J. R.; Montgomery, J. A. A Convenient Synthesis of
Aminopterin and Homologues via 6-(Bromomethyl)-2,4-diami-
nopteridine Hydrobromide. J. Med. Chem. 1974, 11, 379-380.
(6) Kuwano, R.; Utsunomiya, M.; Hartwig, J. F. Aqueous Hydroxide
as a Base for Palladium-Catalyzed Amination of Aryl Chlorides
and Bromides. J. Org. Chem. 2002, 67, 6479-6486.
(7) Seeger, D. R.; Cosulich, D. B.; Smith, J. M., Jr.; Hultquist, M.
E. Analogs of Pteroyl-Glutamic Acid. III. 4-Amino Derivatives.
J. Am. Chem. Soc. 1949, 71, 1753-1758.
(8) Srinivasan, A.; Broom, A. D. Pyridopyrimidines. 12. Synthesis
of 8-Deaza Analogues of Aminopterin and Folic Acid. J. Org.
Chem. 1981, 46, 1777-1781.
(9) Temple, C., Jr.; Elliott, R. D.; Montgomery, J. A. Pyrido[2,3-d]-
pyrimidines. Synthesis of the 5-Deaza Analogues of Aminopterin,
Methotrexate, Folic Acid, and N1-Methylfolic Acid. J. Org. Chem.
1982, 47, 761-764.
(10) Rosowsky, A.; Forsch, R. A.; Freisheim, J. H.; Moran, R. G.; Wick,
M. Methotrexate Analogues. 19. Replacement of the Glutamate
Side-Chain in Classical Antifolates by l-Homo-cysteic and Cys-
teic Acids: Effect on Enzyme Inhibition and Antitumor Activity.
J. Med. Chem. 1984, 27, 600-604.
(11) Rosowsky, A.; Forsch, R. A.; Moran, R. G.; Freisheim, J. H.
Synthesis and In Vitro Biological Evaluation of â,γ-Methano
Analogues of Methotrexate and Aminopterin. Pteridines 1990,
2, 133-139.
(12) (a) Mautner, H. G.; Kim, Y.-H. The Circular Dichroism of Folic
Acid and 10-Thiafolic and the Problem of Racemization in the
Synthesis of Analogues of Folic Acid through the Cyclization of
Substituted 2-Amino-3-cyanopyrazine. J. Org. Chem. 1975, 40,
3447-3448. (b) Kim, Y.-H.; Gaumont, Y.; Kisliuk, R. L.; Maut-
ner, H. G. Synthesis and Biological Activity of 10-Thia-10-deaza
Analogues of Folic Acid, Pteroic Acid, and Related Compounds.
J. Med. Chem. 1975, 18, 776-780.
(13) (a) Rosowsky, A.; Wright, J. E.; Vaidya, C. M.; Bader, H.; Forsch,
R. A.; Mota, C. M.; Pardo, J.; Chen, C. S.; Chen, Y.-N. Synthesis
and Potent Antifolate Activity and Cytotoxicity of B-Ring Deaza
Analogues of the Nonpolyglutamatable Dihydrofolate Reductase
Inhibitor NR-(4-Amino-4-deoxypteroyl)-Nδ-hemiphthaloyl-L-or-
nithine (PT523). J. Med. Chem. 1998, 41, 5310-5319. (b) Wright,
J. E.; Yurasek, G. K.; Chen, Y.-N.; Rosowsky, A. Further Studies
on the Interaction on Nonpolyglutamatable Aminopterin Ana-
logues with Dihydrofolate Reductase and the Reduced Folate
Carrier as Determinants of in Vitro Antitumor Activity. Biochem.
Pharmacol. 2003, 65, 1417-1433.
(14) 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.
or 9, 10-Bridge: Synthesis and In Vitro Antitumor Activity. J.
Med. Chem. 2000, 43, 1620-1634.
(16) On the basis of the findings of Mautner and co-workers12a,b and
those reported here, there is good reason to believe, as implied
by these authors, that prior data reported for DHFR and
bacterial growth inhibition by MTX synthesized via this ap-
proach were similarly off by a factor of 2. Cf.: Chaykovsky, M.;
Rosowsky, A.; Papathanasopoulos, N.; Chen, K. K. N.; Modest,
E. J.; Kisliuk, R. L.; Gaumont, Y. Methotrexate Analogues. 3.
Synthesis and Biological Properties of Some Side-Chain Altered
Analogues. J. Med. Chem. 1974, 17, 1212-1216.
(17) McGuire, J. J. Antifolate Polyglutamylation in Preclinical and
Clinical Antifolate Resistance. In Antifolate Drugs in Cancer
Therapy; Jackman, A. L., Ed.; Humana: Totowa, NJ, 1999; pp
339-363.
(18) (a) Matherly, L. H.; Voss, M. L.; Anderson, L. A.; Fry, D. W.;
Goldman, I. D. Enhanced Polyglutamation of Aminopterin
Relative to Methotrexate in the Ehrlich Ascites Tumor Cell In
Vitro. Cancer Res. 1985, 45, 1073-1078. (b) Kumar, P.; Kisliuk,
R. L. Gaumont, Y.; Freisheim, J. H.; Nair, M. G. Inhibition of
Human Dihydrofolate Reductase by Antifolyl-Polyglutamates.
Biochem. Pharmacol. 1989, 38, 541-543. (c) Rosowsky, A.;
Galivan, J.; Beardsley, G. P.; Bader, H.; O’Connor, B. M.;
Russello, O.; Moroson, B. A.; De Yarman, M. T.; Kerwar, S. S.;
Freisheim, J. H. Biochemical and Biological Studies on 2-Desa-
mino-2-methylaminopterin (dmAMT), an Antifolate the Poly-
glutamates of Which Are More Potent Than the Monoglutamates
against Three Key Enzymes of Folate Metabolism. Cancer Res.
1992, 52, 2148-2155.
(19) (a) Rhee, M. S.; Wang, Y.; Nair, M. G.; Galivan, J. Acquisition
of Resistance to Antifolates Caused by Enhanced γ-Glutamyl
Hydrolase Activity. Cancer Res. 1993, 53, 2227-2230. (b) Yao,
R.; Rhee, M. S.; Galivan, J. Effects of γ-Glutamyl Hydrolase on
Folyl and Antifolyl Polyglutamates in Cultured Rat Hepatoma
Cells. Mol. Pharmacol. 1995, 48, 505-511. (c) Rots, M. G.;
Pieters, R.; Peters, G. J.; Noordhuis, P.; van Zantwijk, C. H.;
Kaspers, G. J. L.; Hahlen, K.; Creutzig, U.; Veerman, A. J. P.;
Jansen G. Role of Folylpolyglutamate Synthetase and Folylpoly-
glutamate Hydrolase in Methotrexate Accumulation and Poly-
glutamylation in Childhood Leukemia. Blood 1999, 93, 1677-
1683. (d) Cole, P. D.; Kamen, B. A.; Gorlick, R.; Banerjee, D.;
Smith, A. K.; Magill, E.; Bertino, J. R. Effects of Overexpression
of γ-Glutamyl Hydrolase on Methotrexate Metabolism and
Resistance. Cancer Res. 2001, 61, 4599-4604.
(20) (a) Hooijberg, J. H.; Broxterman, H. J.; Kool, M.; Assaraf, Y. G.;
Peters, G. J.; Noordhuis, P.; Scheper, R. J.; Borst, P.; Pinedo,
H. M.; Jansen, G. Antifolate Resistance Mediated by the Mul-
tidrug Resistance Proteins MRP1 and MRP2. Cancer Res. 1999,
59, 2532-2535. (b) Zeng, H.; Chen, Z. S.; Belinsky, M. G.; Rea,
P. A.; Kruh, G. D. Transport of Methotrexate (MTX) and Folates
by Multidrug Resistance Protein (MRP) 3 and MRP1: Effect of
Polyglutamylation on MTX Transport. Cancer Res. 2001, 61,
7225-7232. (c) Chen, Z. S.; Lee, K.; Walther, S.; Raftogianis, R.
B. Kuwano, M.; Zeng, H.; Kruh, G. D. Analysis of Methotrexate
and Folate Transport by Multidrug Resistance Protein 4 (AB-
CC4): MRP4 Is a Component of the Methotrexate Efflux System.
Cancer Res. 2002, 62, 3144-3150.
(15) Rosowsky, A.; Wright, J. E.; Vaidya, C. M.; Forsch, R. A.; Bader,
H. Analogues of the Potent Nonpolyglutamatable Antifolate NR-
(4-Amino-4-deoxypteroyl)-Nδ-hemiphthaloyl-L-ornithine (PT523)
with Modifications in the Side Chain, p-Aminobenzoyl Moiety,
JM040122S