Efficient syntheses of pyrofolic acid and pteroyl azide, reagents for the production of carboxyl-differentiated derivatives of folic acid

Article abstract of DOI:10.1021/ja971568j

Reaction of folic acid (1) with excess trifluoroacetic anhydride provides access to both the previously unknown N¹⁰-(trifluoroacetyl)pyrrofolic acid (8) and pyrofolic acid (9). Reaction of either of these materials with hydrazine selectively affords pteroyl hydrazide (13), which may be oxidized to pteroyl azide (27) on a large scale (62% overall from I without the need for chromatography). Treatment of 27 with differentially protected glutamates provides a convenient and high-yielding synthesis of differentially protected, optically pure folates.

Full text of DOI:10.1021/ja971568j

10004
J. Am. Chem. Soc. 1997, 119, 10004-10013
Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide,
Reagents for the Production of Carboxyl-Differentiated
Derivatives of Folic Acid
Jin Luo, Michael D. Smith, Douglas A. Lantrip, Susan Wang, and P. L. Fuchs*
Contribution from the Department of Chemistry, Purdue UniVersity,
West Lafayette, Indiana 47907
ReceiVed May 14, 1997X
Abstract: Reaction of folic acid (1) with excess trifluoroacetic anhydride provides access to both the previously
unknown N10-(trifluoroacetyl)pyrofolic acid (8) and pyrofolic acid (9). Reaction of either of these materials with
hydrazine selectively affords pteroyl hydrazide (13), which may be oxidized to pteroyl azide (27) on a large scale
(62% overall from 1 without the need for chromatography). Treatment of 27 with differentially protected glutamates
provides a convenient and high-yielding synthesis of differentially protected, optically pure folates.
A recent trend in cancer chemotherapy is the highly aggres-
inactive R-conjugate 25c and the active γ-conjugate 3,5c often
accompanied with the bis-functionalized derivative 4 (not
shown) and/or recovered 1 (Scheme 1). Separation of these
mixtures is often difficult and some of the exciting biological
results have been obtained on mixtures.5
sive application of high-dose multiple drug treatment regimens
at the earliest point of diagnosis.1 These protocols are limited
by drug toxicity, and severe physiological effects and patient
fatalities are not uncommon. This situation has caused several
members of the medical community to question whether the
benefit/risk boundary has been exceeded with the agents
currently available.2 Enhancement of the differential specificity
of anticancer agents by selectiVe targeting mechanisms might
diminish such problems. The vitamin folic acid (1) has attracted
considerable attention as a potential means of delivery of
covalently bound drug conjugates.3 Many human cancer cell
lines have been found to have highly overexpressed levels of
the protein which binds folic acid,4 a finding which is beginning
to be exploited with the preparation of folate-drug conjugates.5
Regiospecific functionalization of the γ-carboxylate of glutam-
ic acid is routinely accomplished using urethane derivatives of
pyroglutamic acid which exploit the γ-lactam moiety as an
acylating agent.6 A search of the literature surprisingly revealed
that pyrofolic acid (9) and its N-acylated derivatives are
apparently unknown.7 Nevertheless, treatment of 1 (100 g) with
excess trifluoroacetic anhydride in THF for 10 h from 0 to 25
°C produces N2,10-bis(trifluoroacetyl)pyrofolic acid (7) as an
extremely water-labile material which is believed to be a mixture
of diacylated anhydride 6 and diacylated carboxylic acid 7, as
judged by 19F-NMR of the crude reaction mixture. In any event,
simply stirring the aforementioned material with neutral water
affords N10-(trifluoroacetyl)pyrofolic acid (8) in a quantitative
yield (Scheme 2). Compound 8 can be further transformed to
pyrofolic acid (9) via deacylation with cesium carbonate and
water (87%). Both 8 and 9 are essentially racemic, as
determined using an enzymatic assay based on carboxypeptidase
G (see Table 3).
Unfortunately, a major impediment in drug design centers
around the synthetic aspects of preparing the folate-drug
conjugates. Current practice simply involves treatment of the
substrate of choice with folic acid (1) and a given dehydrating
agent, such as DCC. This results in a mixture of both the
X Abstract published in AdVance ACS Abstracts, October 1, 1997.
(1) Gurney, H.; Dodwell, D.; Thatcher, N.; Tattersall, M. H. N. Ann.
Oncol. 1993, 4, 103.
Unfortunately, the N10-trifluoroacetyl derivative of azalactone
8′ (not the isomeric pyrofolate 8) has been alleged to result from
reaction of 1 with trifluoroacetic anhydride (Scheme 3).7a In
addition, the same authors have reported that the N10-acetyl
azalactone Me-8′ results from the analogous acetic anhydride
reaction.7b Basic hydrolysis of Me-8′ yields a mixture of 1 and
pteroic acid (10), entirely consistent with the pyrofolate structure
as per our observations in Table 1 (below), while the authors
had to invoke an unprecedented8 hydrolysis of the azalactone
imine moiety to explain the production of 10. This structural
misassignment has dire consequences Vis-a`-Vis regiospecific
functionalization of the two carboxylates of the glutamate
(2) (a) Dodwell, D. J. Lancet 1993, 341, 614. (b) Davidson, N. E. J.
Clin. Oncol. 1992, 10, 517. (c) Henderson, I. C.; Hayes, D. F.; Gelman, R.
J. Clin. Oncol. 1988, 6, 1501.
(3) Folic acid and its reduced counterparts enter cells via two unrelated
pathways (Dixon, K. H.; Lanpher, B. C.; Chiu, J.; Kelley, K.; Cowan, K.
H. J. Biol. Chem. 1994, 269, 17). The primary intake route is mediated by
the reduced folate carrier, a transport protein that exhibits a Km for folate
near 5 × 10-6 M and a strong preference for reduced folates over folic
acid. The second path of folate uptake is mediated by a family of membrane
receptors termed the folate binding protein (FBP) (Brigle, K. E.; Spinella,
M. J.; Westin, E. H.; Goldman, I. D. Biochem. Pharmacol. 1994, 47, 337.
Shen, F.; Ross, J. F.; Wang, X.; Ratnam, M. Biochemistry 1994, 33, 1209.
Leamon, C. P.; Low, P. S. Biochem. J. 1993 291, 855. Garin-Chesa, P.;
Campbell, I.; Saigo, P. E.; Lewis, J. L., Jr.; Old, L. J.; Rettig, W. J. Am. J.
Pathol. 1993, 142, 557). FBP has a KD for folate that varies from 10-9 to
10-11 and exhibits a strong preference for folic acid over its reduced
counterparts (Low, P. S. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 5572).
(4) (a) Franklin, W. A.; Waintrub, M.; Edwards, D.; Christensen, K.;
Prendegrast, P.; Woods, J.; Bunn, P. A.; Kolhouse, J. F. Int. J. Cancer
1994, 8, 89. (b) Mantovani, L. T.; Miotti, S.; Me´nard, S.; Canevari, S.;
Raspagliesi, F.; Bottini, C.; Bottero, F.; Colnaghi, M. I. Eur. J. Cancer 1994,
30A, 363. (c) Ross, J. F.; Chaudhuri, P. K.; Ratnam, M. Cancer 1994, 73,
2432.
(6) (a) Danishefsky, S.; Berman, E.; Clizbe, L. A.; Hirama, M. J. Am.
Chem. Soc. 1979, 101, 4385. (b) Flynn, D. L.; Zelle, R. E.; Grieco, P. A.
J. Org. Chem. 1983, 48, 2424. (c) Smith, A. B. 2, III; Salvatore, B. A.;
Hull, K. G.; Duan, J. J.-W. Tetrahedron Lett. 1991, 32, 4859. (d) Collado,
I.; Ezquerra, J.; Vaquero, J. J.; Pedregal, C. Tetrahedron Lett. 1994, 35,
8037. (e) Dixit, A. N.; Tandel, S. K.; Rajappa, S. Tetrahedron Lett. 1994,
35, 6133 and references therein.
(5) (a) Leamon, C. P. and Low, P. S. J. Biol. Chem. 1992, 267, 24966.
(b) Leamon, C. P.; Low, P. S. J. Drug Targeting 1994, 2, 101. (c) Leamon,
C. P.; Pastan, I.; Low, P. S. J. Biol. Chem. 1993 268, 24847.
(7) (a) Temple, T., Jr.; Montgomery, J. A. Folates and Pterins; Wiley:
New York, 1984; Chapter 2, p 106. (b) Temple, C., Jr.; Rose, J. D.;
Montgomery, J. A. J. Org. Chem. 1981, 46, 3666.
S0002-7863(97)01568-0 CCC: $14.00 © 1997 American Chemical Society

Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10005
Scheme 1
Scheme 2
Scheme 3
Scheme 4
moiety, since nucleophilic attack on the azalactone 8′ would
be expected to afford functionalization of the R-carboxyl moiety,
while activation of the carboxylic acid should enable regiospe-
cific γ-functionalization. This is exactly the opposite regio-
chemistry of that which actually results from functionalization
of the correct pyrofolate structure 8 and has resulted in further
incorrect structural assignments.9
hydroxide at 25 °C for several hours to afford pteroylamide
(11) along with pyroglutamic acid (12) in high yield (Scheme
4), a finding which would be exceptionally difficult to rationalize
with the alternative azalactone structures 8′ and 9′. As expected,
concomitant deacylation of the N10-trifluoroacetyl moiety oc-
curred during the reaction. The isolated 12 exhibited only e5%
optical activity, in accord with extensive racemization occurring
during the synthesis of 8. Monitoring a pair of similar reactions
of 8 and 9 with hydrazine (10 equiv) in DMSO-d6 at 25 °C for
9 h unambiguously produces a 1:1 mixture of pteroyl hydrazide
(13) and (12) in near-quantitative yield, as assayed by 1H-NMR
and HPLC (Scheme 4 and Table 1, entries 12 and 13).
In order to provide a definitive structural assignment, both 8
and 9 were treated with excess concentrated ammonium
(8) (a) Carter, H. E. Organic Reactions; Wiley: New York, 1946; Vol.
III, p 198. (b)Alkaabi, S. S.; Shawali, A. S. Can. J. Chem. 1992, 70, 2515.
(c) Homami, S. S.; Joseph, K. Can. J. Chem. 1994, 72, 1383. (d) Jones, J.
H.; Witty, M. J. J. Chem. Soc., Perkin Trans. 1 1980, 858. (e) King, S. W.;
Riordan, J. M.; Holt, E. M.; Stammer, C. H. J. Org. Chem. 1982, 47, 3270.
(9) Dunlap, R. B.; Silks, V. F. Chem. Biol. Pteridines 1989, 77.
Spectral evidence strongly supports the pyrofolate structure
assigned to 8 and 9. In particular, the 13C-NMR shift (DMSO-

10006 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Luo et al.
Figure 1.
Table 1. Reactions of Nucleophiles with N10-(Trifluoroacetyl)pyrofolic Acid (8) and Pyrofolic Acid (9)
nucleophile
and conditions
folic acid derivative 25
pteroic acid derivative 26
folate/pteroate
run
SM
(yielda and nucleophile)
(yielda and nucleophile)
(25/26) ratioa
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
8
9
8
9
9
9
8
9
9
9
9
8
9
8
9
8
H2O, NaOH
H2O, NaOH
H2O, HCl
25a ≡ 1 (52%, OH)
25a ≡ 1 (68%, OH)
25bb (33%, OH)
25c (49%, OMe)
25c (81%, OMe)
25c (58%, OMe)
25c (35%, OMe)
25c (69%, OMe)
25d (80%, O-iPr)
25e (nr,c Ot-Bu)
26a ≡ 10 (48%, OH)
26a ≡ 10 (32, OH)
26bb (56%, OH)
26c (44%, OMe)
26c (19%, OMe)
26c (42, OMe)
26c (58%, OMe)
26c (30%, OMe)
1.1
2.1
0.5
1.1
4.3
1.4
0.6
2.3
11.5
nrc
2.2
0.01
0.1
0.5
1.5
0.01
MeOH, NaOMe
MeOH, LiOMe
MeOH, DBU
MeOH, TlOMe
MeOH, TlOMe
i-PrOH, TlOi-Pr
t-BuOH, TlOt-Bu
DEG,d NaH
NH2NH2, DMSO
NH2NH2, DMSO
NH4OH, H2O
NH4OH, H2O
PMBAf
26d (7%, O-iPr)
26e (nr,c Ot-Bu)
25f (48%, DEG′e)
25g (0.9%, NHNH2)
25g (9.1%, NHNH2)
25h (33%, NH2)
25h (60%, NH2)
25I (0.9%, PMBA′g)
26f (22%,a DEG′e)
26g ≡ 13 (91%, NHNH2)
26g ≡ 13 (90%, NHNH2)
26h ≡ 11 (67%, NH2)
26h ≡ 11 (40%, NH2)
26i (75%, PMBA′g)
a Product ratio assayed by HPLC. b This is the only instance in which the N10-trifluoroacetyl moiety survived the reaction conditions. c nr ) no
reaction. d DEG ) HOCH2CH2OCH2CH2OH. e DEG′ ) OCH2CH2OCH2CH2OH. f PMBA ) p-methoxybenzylamine. g PMBA′ ) p-methoxyben-
zylamino.
d6) of the methine-bearing chiral carbon is highly diagnostic
(Figure 1). Folic acid (1) and the truncated p-aminobenzoyl
glutamates 14-1610 all resonate at ∼52δ. Pyrofolates 8 and 9
and model p-aminobenzoyl pyroglutamates 17-21 exhibit their
methine carbons between 59 and 60δ. This can be contrasted
to model azalactones 22-24,10,11 which have chemical shifts
around 64-66δ.
Having secured the pyrofolate structure of derivatives 8 and
9, we turned our attention to a brief survey of the regiospecificity
of reaction of these substrates with oxygen and nitrogen
nucleophiles. While the pyrofolate moiety ensures complete
regioselection between the R- and γ-carbonyl groups of folic
acid, the problem of differentiating between the two imide
carbonyl groups (γ-CO and Pte-CO) of 8 and 9 remained to be
determined. As can be seen in Table 1, thallium(I)-mediated
alcoholysis provides preferential generation of the folic acid
derivatives 25c,d; the selectivity is an important function of the
steric environment of the alcohol. Aminolysis is currently
unacceptable for direct drug conjugation, but, as will be seen
in Scheme 5, the acyl hydrazide 26g (≡13) is an ideal
intermediate for the indirect synthesis of folates via nitrogen
acylation of glutamates with pteroyl azide (27). As expected,
N10-trifluoroacetyl derivative 8 consistently exhibits a smaller
folate/pteroate ratio (25/26) relative to 9, since increased
nucleophilic attack at the benzoate carbonyl moiety is favored
when the p-amino lone pair is unavailable for resonance
deactivation. The full magnitude of this effect cannot be
accurately assessed simply by inspection of Table 1, since we
did not attempt to determine the rate of deacylation of the N10-
trifluoroacetyl group relative to nucleophilic attack at the two
competing carbonyl groups (Scheme 5).
Since we had already determined that pyrofolic acid deriva-
tives 8 and 9 were essentially racemic, we elected to pursue an
indirect synthesis of the desired γ-functionalized folic acid
derivatives 3 shown in Scheme 1 via exploitation of the
exceptionally selective (>99:1) reaction of hydrazine with
compound 8 (Table 1, entry 12). To this end, it has been
observed that conversion of 13 to 27 can be conveniently
effected on a reasonable (28 g) scale simply by treatment with
1.0 equiv of tert-butyl nitrite in trifluoroacetic acid containing
5 mol % potassium thiocyanate for 4 h at 10 °C. HPLC analysis
reveals that reactions run in the absence of potassium thiocy-
anate, an N-nitroso transfer catalyst,12 also produce 20-30%
of N10-nitrosopteroyl azide (28),13 along with 27.14 In the
KSCN-free condition, the remaining 13 slowly reacts with 28
(10) Synthesis and characterization information for this compound is
given in the Supporting Information.
(11) We thank Professor Sih for providing us samples of compounds 22
and 23. For their synthesis, see: Crich, J. Z.; Brieva, R.; Marquart, P.; Gu,
R.-L.; Flemming, S.; Sih, C. J. J. Org. Chem. 1993, 58, 3252.
(12) Singer, S. S.; Singer, G. M.; Cole, B. B. J. Org. Chem. 1980, 45,
4931.

Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10007
Scheme 5
Scheme 6
to afford additional amounts of 27, but the reaction produces
unacceptable levels of impurities relative to the optimized
procedure (Scheme 6).
apparently resulting from Curtius rearrangement of 27, ∼2%
of pteroylamide (PteCO-NH2, 11), and <1% of 1. Pteroyl azide
(27) may be further purified by dissolution/precipitation from
Assay of 27 by HPLC reveals a purity of 91%. The known
impurities include ∼4% of 10, ∼2% of the aniline (PteNH2),
(13) Nitrosation of pteroic acid derivatives like folic and tetrahydrofolic
acid is known to occur at N10: Reed, L. S.; Archer, M. C. J. Agric. Food
Chem. 1979, 27, 995.
(14) Large-scale reactions which use crude pteroyl hydrazide (13) may
produce small amounts of N-nitrosopteroyl azide (28) because of the
inadvertant use of an excess of tert-butyl nitrite. In this instance, treatment
of the 27/28 mixture with sodium azide (see ref 12) effects smooth reduction
of the N-nitroso moiety of 28 to 27.

10008 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Luo et al.
Table 2. Acylation of Glutamic Acid Derivatives 30a-d in DMSO with Acyl Azides 27, 29, 31, and 32
acyl azide
and glutamate
base (amt);
DMSO pKa
product, HPLC conversion,
and/or (isolated yield)
run
conditions
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
27 + L-30a
27 + L-30a
27 + L-30a
27 + L-30a
27 + L-30a
27 + L-30a
27 + L-30b
27 + L-30c
27 + L-30d
27 + D-30d
27 + DL-30d
29 + L-30d
31 + L-30d
31 + L-30a
32 + L-30a
Et3N (3 equiv); 18.5
Et3N (5 equiv) ;18.5
TMG (2 equiv); 20
MTBD (2 equiv); 20
BTMG (2 equiv) 21
DBU (2 equiv); 19
TMG (2 equiv); 20
TMG (2 equiv); 20
TMG (3 equiv); 20
TMG (3 equiv); 20
TMG (3 equiv); 20
TMG (3 equiv); 20
TMG (3 equiv); 20
TMG (2 equiv); 20
TMG (2 equiv); 20
48 h, 40 °C
48 h, 40 °C
6 h, 25 °C
12 h, 25 °C
9 h, 25 °C
8 h, 25 °C
10 h, 25 °C
12 h, 25 °C
9 h, 25 °C
4 h, 25 °C
3 h, 25 °C
1 h, 25 °C
1 h, 25 °C
1 h, 25 °C
1 h, 25 °C
L-33a, 23%
L-33a, 25%
L-33a, 100% (88%)
L-33a, 100%
L-33a, 100%
L-33a, 77%
L-33b, 100% (60%)
L-33c, 100% (82%)
L-25a ≡ L-1, 100% (67%)
D-25a ≡ D-1, 100% (40%)
DL-25a ≡ DL-1, 100% (38%)
L-34d (100%)
L-35d34 (73%)
L-36a ≡ 15 (67%)
L-37a ≡ 16 (72%)
trifluoroacetic acid/2-isopropanol, but the known impurities do
not interfere, and we routinely use the ∼90% pure material for
virtually all coupling operations. Unlike some acyl azides,15
27 is quite stable; samples of the lemon-yellow solid may be
stored in the dark in a freezer for at least 12 months without
appreciable decomposition. Room temperature samples of 27
protected from light appear to have shelf lives of at least a
month; while samples of 27 darken appreciably when exposed
to light, their HPLC profiles are not substantially degraded.
The traditional approach to the synthesis of differentially
functionalized folic acid derivatives relies on the acylation of
the readily available monoesters of glutamic acid16 with pteroic
acid (10). Unfortunately this highly attractive strategy17 is
compromised by the difficulty of accessing reasonable quantities
of the prohibitively expensive 1018 (∼$1000/g). Fortunately,
27 now both is easily available and serves as an excellent reagent
for nitrogen acylation of differentially protected glutamates. As
can be seen in Scheme 6 and Table 2, reaction of 27 in DMSO
with L-glutamic acid (30d) or glutamates 30a-c is strongly
influenced by the nature of the added base, tetramethylguanidine
(39a, TMG), tert-butyltetramethylguanidine (39b, BTMG),19 and
the expensive N-methyl-1,5,9-triazabicyclo[4.4.0]decene (38,
MTBD),20 all giving superb reactions. Presumably, the more
basic nature of the guanidine bases, in concert with their ability
to form soluble guanidinium carboxylates, is responsible for
their ability to foster the acylation reaction. We were, therefore,
surprised to note that both Rosowsky et al.21 and Chaykovsky
et al.22 have reported that the closely related acyl azide 29 is
insufficiently reactive (in DMF or DMAC) to acylate glutamates
or other R-amino esters.
Our initial hypothesis was that this dramatic reactivity
difference between 27 and 29 was a consequence of abstraction
of the N-H proton of the p-aminobenzoyl azide moiety by the
guanidine base, followed by 1,6-elimination of the azide anion
to generate the neutral p-quinoketene monoimine analog of 27′
(Scheme 6, Y ) lone pair; no charge on N). Presumably, such
a species would be an exceptionally reactive acylating agent.
Two facts mitigate against this intriguing possibility: (1) The
pKa of the N-H proton of 27 should be similar to that of
p-methylaminomethyl benzoate, having a pKa of 25.7 in
DMSO,23 which is fairly distant from the pKa of H-TMG+ at
13.6 in water24 (∼20 in DMSO); consequently, the concentration
of the deprotonated form of 27 would be relatively small, but
still attainable; (2) A second observation which argues against
27′ (Scheme 6, Y ) lone pair; no charge on N) being a requisite
intermediate is that azide 29,21,22 which bears no N-H proton,
is a perfectly fine acylating agent, providing a high yield of
synthetic L-methotrexate (34d) upon reaction with L-glutamic
acid (30d) and tetramethylguanidine (39a), provided that DMSO
is employed as the reaction solvent (Table 2, entry 12). The
fact that 29 reacts more rapidly with 30d than does 27 rules
out the possibility that N-H deprotonation is an integral feature
in these acylation reactions.25 Furthermore, there is no special
reactivity conferred to these acyl azides by the pterin moiety,
since p-(monomethylamino)benzoyl azide (31) and p-(dimeth-
ylamino)benzoyl azide (32) both react with γ-methylglutamate
(30a) in DMSO in the presence of tetramethylguanidine to afford
(15) (a) Hamada, Y.; Shioiri, T.; Yamada, S. I. Chem. Pharm. Bull. 1977,
25, 221. (b) Shioiri, T.; Ninomiya, K.; Yamada, S.-i. J. Am. Chem. Soc.
1972, 94, 6203. (c) L’Abbe, G. Chem. ReV. 1969, 69, 345. (d) Tsuji, J.;
Nagashima, T.; Qui, N. T.; Takayanagi, H. Tetrahedron 1980, 36, 1311.
(e) Pozsgay, V.; Jennings, H. J. Tetrahedron Lett. 1987, 28, 5091. (f) Kim,
Y. H.; Kim, K.; Shim, S. B. Tetrahedron Lett. 1986, 27, 4749-4752. (g)
van Reuendam, J. W.; Baardman, F. Synthesis 1973, 73, 413.
(16) (a) Scholtz, J. M.; Bartlett, P. A. Synthesis 1989, 542. (b) Van
Heeswuk, W. A. R.; Eenink, M. J. D.; Feuen, J. Synthesis 1982, 744. (c)
Pugniere, M.; Castro, B.; Domergue, N.; Previero, A. Tetrahedron Asym-
metry 1992, 3, 1015. (d) Chevallet, P.; Garrouste, P.; Malawska, B.;
Martinez, J. Tetrahedron Lett. 1993, 34, 7409.
(17) For an excellent recent review on folate-based anticancer agents,
see: (a) Rosowsky, A. Am. J. Pharm. Ed. 1992, 56, 453. For references to
folate syntheses via pteroic acid acylation chemistry, see: (b) Rosowsky,
A.; Forsch, R.; Uren, J.; Wick, M. J. Med. Chem. 1981, 24, 1450. (c)
Rosowsky, A.; Beardsley, G. P.; Ensminger, W. D.; Lazarus, H.; Yu, C.-S.
J. Med. Chem. 1978, 21, 380. (d) Godwin, H. A.; Rosenberg, I. H.; Ferenz,
C. R.; Jacobs, P. M.; Meienhofer, J. J. Biol. Chem. 1972, 247, 2266. (e)
Krumdieck, C. L.; Baugh, C. M. Biochemistry 1969, 8, 1568. (f) Mala-
chowski, W. P.; Coward, J. K. J. Org. Chem. 1994, 59, 7625. (g) Hart, B.
P.; Haile, W. H.; Licato, N. J.; Bolanowska, W. E.; McGuire, J. J.; Coward,
J. K. J. Med. Chem. 1996, 39, 56.
(21) (a) Rosowsky, A.; Forsch, R.; Uren, J.; Wick, M. J. Med. Chem.
1981, 24, 1450. (b) Rosowsky, A.; Wright, J. E.; Ginty, C.; Uren, J. J.
Med. Chem. 1982, 25, 960.
(22) (a) Chaykovsky, M.; Brown, B. L.; Modest, E. J. J. Med. Chem.
1975, 18, 909. (b) Martinelli, J. E.; Chaykovsky, M.; Kisliuk, R. L.;
Gaumont, Y.; Gittelman, M. C. J. Med. Chem. 1979, 22, 869. (c) Martinelli,
J. E.; Chaykovsky, M.; Kisliuk, R. L.; Gaumont, Y. J. Med. Chem. 1979,
22, 874.
(23) We wish to thank Dr. Liu and Professor Bordwell of Northwestern
University for determining this value.
(24) (a) Barton, D. H. R.; Elliott, J. D.; Gero, S. D. J. Chem. Soc., Perkin
Trans. 1 1982, 2085. (b) Angyal, S. J.; Warburton, W. K. J. Chem. Soc.
1951, 2492.
(18) Pteroic acid has been obtained on a small scale from folic acid by
both enzymatic (D’Ari, L.; Rabinowitz, J. C. Methods Enzymol. 1985, 113,
169 and references therein) and chemical7 methods.
(19) Barton, D. H. R.; Elliott, J. D.; Gero, S. D. J. Chem. Soc., Perkin
Trans. 1 1982, 2085-2090. BTMG (39b) is about 1 pKa unit more basic
than TMG (39a) in ethanol/water.
(20) (a) McKay, A. F.; Kreling, M.-E. Can. J. Chem. 1957, 35, 1438.
(b) Schmidtchen, F. P. Chem. Ber. 1980, 113, 2175. (c) Alder, R. W. Chem.
ReV. 1989, 89, 1215. (d) Schwesinger, R. Chimia 1985, 39, 269.
(25) Direct competition NMR studies using g-methylglutamate (30a, 1
equiv) and TMG (39, 3 equiv) in DMSO-d6 with a 1:1 mixture of model
acyl azides 31 and 32 reveal that the p-(dimethylamino)benzoyl azide (32)
reacts 3 times faster with 30a than does the (monomethylamino)benzoyl
azide (31).

Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10009
N-acyl glutamates 36a and 37a, respectively.25 The question
as to whether these acylations are simply proceeding via the
standard tetrahedral adduct or may progress via the intermediacy
of p-quinoketene monoiminium ions 27′, 29′, 31′, and 32′
(Scheme 6) still remains to be determined.
Table 3. Carboxypeptidase G Enantiospecific Hydrolysis of Folic
Acid Derivatives
run
compound (source)
L-1 (commerciala)
L-25a ≡ L-1 (from Table 2)
L-25a ≡ L-1 (from Table 3)
D-25a ≡ D-1 (from Table 2)
DL-25a ≡ DL-1 (from Table 2)
L-34d (NCI MTXb)
L-34d (MTX from Table 2)
DL-33a (from Table 1)
L-33a (from Table 2)
% hydrolysis
1
2
3
4
5
6
7
8
9
10a
10b
11
12
96.7
98.4
98.6
3.0
52.3
96.1c
99.2c
54.4
96.8
85.9
96.7
49.5
51.7
Control studies show that in the absence of an added substrate,
tetramethylguanidine (TMG, 39a) slowly reacts with 27 or 31
to afford acylated guanidines 40 or 4126 (Scheme 6), but HPLC
analyses of the reactions of 27 with glutamates 30a-d show
only traces of 40. By comparison, HPLC analysis of a DMSO
solution of 27 with pentaalkylguanidines BTMG (39b) and
MTBD (38) gives no evidence of guanidine acylation or any
other reaction after 18 h at 25 °C. This finding requires that,
if formed, p-quinoketene monoiminium ion 27′ must be in ready
equilibrium with the starting acyl azide 27. DBU (pKa in DMSO
∼19)20c was found to be a less effective base in these reactions
(Table 2, run 5), since a portion of the 27 was consumed via an
unprofitable acylation-fragmentation sequence,27 providing
byproduct 45 in 20% yield.
L-33c (from Table 2)
L-33c (from Table 2 Scheme 7)
8 (from Scheme 2)
9 (from Scheme 2)
a Commercial L-1 from Vitamins, Inc. b A sample of L-methotrexate
c
(L-34d) was provided by the National Cancer Institute. D-34d is known
not to be hydrolyzed by carboxypeptidase G.35
results in partial racemization of glutamate 51b; in this instance,
2.1 equiv of sodium generated a product which was only 85.9%
optically pure (Table 3, run 10a), while employment of the 1.9
equiv conditions afforded material which was 96.7% optically
pure as judged by the enzymatic assay (Table 3, run 10b). Since
the synergistic value of palladium(0)-mediated deprotection of
the allyl and dimethylallyl protecting groups has been convinc-
ingly demonstrated for N-alloc esters by the Genet group,30 we
adopted this superb strategy for cleavage of 51a,b. As can be
seen in Scheme 7, reaction of 51a with diethylamine and 10
mol % palladium tetrakistriphenylphosphine provides a quan-
titative yield of 30d, while similar reaction of dimethyl allyl
ester 51b affords a 70% yield of R-DMA-protected glutamate
30c. As indicated above, both these materials are shown to be
essentially optically pure, as assayed after their conjugation (69-
82%) with pteroyl azide (27), to afford 25a ()L-1) and L-33c,
respectively (Scheme 7 and Table 3, entries 3 and 10b).
The value of 27 for the synthesis of folic acid analogs is
exemplified in the preparation of DTPA folate (γ) (53), a metal-
binding ligand of an important new imaging agent with
outstanding tumor specificity (Scheme 7).31,32 The initial sample
of 52, the precursor of 53, was prepared via the direct DCC/
NHS coupling of ethylenediamine with 1, followed by separation
of the various reaction components by extensive HPLC, the final
yield of the requisite γ-conjugate 52 (Scheme 1, 3, X-Z )
NHCH2CH2NH2) being on the order of 10-15%, on small
scale.32 Our current synthesis of 52, which is far superior in
practical terms, now simply involves the reaction of synthetic
γ-methylfolate (L-33a, Table 2, run 3, 1.1 equiv) with neat
ethylenediamine (50 equiv) for 3 h at 25 °C to afford 52 as a
yellow solid in 87% yield, without the need of chromatography.
Conversion of 52 to the tumor-specific metal-binding ligand
DTPA folate (γ) (53) has been previously described (Scheme
8).31
Since it had been previously observed that guanidine bases
are capable of racemizing acylated glutamates (under more
forcing conditions than were employed in the above table),28
we needed to evaluate the optical purity of the folic acid ()25a)
and several of its derivatives which had been produced. For
this purpose we elected to employ the enzyme carboxypeptidase
G,29 which has been shown to only deacylate the natural
L-enantiomer of folic acid (1) to L-glutamic acid and pteroic
acid (10), the yield of the latter material being determined by
HPLC. In order to assay the pyrofolates 8 and 9 as well as the
R- and γ-monoesters of folic acid L-33a and L-33c, an initial
mild basic hydrolysis was required prior to the enzyme assay.
As can be seen in Table 3 (runs 1 and 2), the synthetic folic
acid L-1 ()25a) was produced in enantiomerically pure form
(within the limits of experimental error). Control reactions on
synthetic D-1 ()25a) and DL-1 ()25a) (Table 3, runs 3 and 4)
serve to confirm the error limits of the enzymatic method at
about (2-3%.
Synthesis of bis-allylated glutamates 51a,b was initiated via
reaction of L-glutamic acid (30d) with allyl chloroformate to
afford N-alloc-protected glutamic acid (47). This material was
not purified but directly reacted with paraformaldehyde under
the standard conditions16 to provide N-alloc oxazolidinone (48)
in 50% overall yield for the two steps. Reaction of this substrate
with slightly less than 2 equiv of sodium metal in the presence
of an excess of allyl (49) or dimethylallyl alcohol (50) provides
R-allylated and R-dimethylallylated N-alloc glutamates 51a,b,
in 79 and 71% yield, respectively. Use of an excess of sodium
(26) Although simple acetyl and benzoyl tetramethylguanidines show
equivalence of the four methyl groups in the 1H-NMR at 25 °C (see (a)
Matsumoto, K.; Rapoport, H. J. Org. Chem. 1968, 33, 552. (b) Kantlehner,
W.; Jaus, H.; Kienitz, L.; Bredereck, H. Liebigs. Ann. Chem. 1979, 2096.
(c) Kessler, H.; Leibfritz, D. Liebigs. Ann. Chem. 1970, 737, 53), a Hamett
study (see Kessler, H.; Leibfritz, D. Tetrahedron 1970, 26, 1805) on
p-substituted N-phenyltetramethylguanidine derivatives reveals that electron-
releasing substituents increase the NMR colalesence temperature to near 0
°C. This is consistent with our observations on the NMR of pteroyl-
substituted tetramethyl guanidine (40). The 1H-NMR shows two broad
N-methyl singlets, while the 13C-NMR reveals eight closely separated
N-methyl absorptions, suggesting the presence of both E and Z conforma-
tional diastereomers (see Supporting Information).
In conclusion, we have provided the first efficient syntheses
of racemic N10-(trifluoroacetyl)pyrofolic acid (8), racemic
pyrofolic acid (9), pteroyl hydrazide (13), and pteroyl azide (27).
The latter reagent is an effective and economical reagent for
the synthesis of differentially functionalized folic acid deriva-
(27) Similar functionaliation/fragmentation reactions of DBU have been
previously observed. See: (a) Juneja, T. R.; Garg, D. K.; Schafer, W.
Tetrahedron 1982, 38, 551. (b) Lammers, H.; Cohen-Fernandes, P.;
Habraken, C. L. Tetrahedron 1994, 50, 865.
(28) Mautner, H. G.; Kim, Y.-H. J. Org. Chem. 1975, 40, 3447.
(29) (a) McCullough, J. L.; Chabner, B. A.; Bertino, J. R. J. Biol. Chem.
1971, 246, 7207. (b) Rosowsky, A.; Forsch, R.; Uren, J.; Wick, M.; Kumar,
A. A.; Freisheim, J. H. J. Med. Chem. 1983, 26, 1719. (c) Rosowsky, A.;
Forsch, R. A.; Moran, R. G.; Freisheim, J. H. J. Med. Chem. 1991, 34,
227.
(30) (a) Lemaire-Audoire, S.; Savignac, M.; Blart, E.; Pourcelot, G.;
Genet, J. P. Tetrahedron Lett. 1994, 35, 8783. (b) Genet, J. P.; Blart, E.;
Savignac, M.; Lemeune, S.; Lemaire-Audoire, S.; Bernard, J. M. Synlett
1993, 680. (c) For synthesis of a N-alloc C-allyl L-aspartate, see: Shibata,
N.; Baldwin, J. E.; Jacobs, A.; Wood, M. E. Synlett 1996, 519.
(31) Wang, S.; Mathias, C. J.; Green, M. A.; Low, P. S.; Luo. J.; Lantrip,
D. A.; Fuchs, P. L. Bioconjugate Chem., submitted.
(32) (a) Wang, S.; Lee, R. J.; Mathias, C. J.; Green, M. A.; Low, P. S.
Bioconjugate Chem. 1996, 7, 56. (b) Mathias, C. J.; Wang, S.; Lee, R. J.;
Waters, D. J.; Low, P. S.; Green, M. A. J. Nucl. Med. 1996, 37, 1002.

10010 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Luo et al.
Scheme 7
Scheme 8
obtained on a Finnigan 4000 mass spectrometer (low resolution) and a
CEC 21 110 B high-resolution mass spectrometer, with the molecular
ion designated as M. All the chemicals were supplied by Aldrich
Chemical Co., Inc, Milwaukee, WI, unless otherwise indicated.
tives. Application of this technology for the synthesis of new
folate-anticancer drug conjugates is under active investigation.
Experimental Section
Typical Experimental Procedure for Coupling of Pteroyl Azide
(27) with Glutamates. To a stirred DMSO suspension of equal molar
amounts of pteroyl azide (27, 0.02-0.3 M) and a glutamate (Table 2)
was added 2 (3 for glutamic acid) equiv of tetramethylguanidine at
room temperature. The mixture soon became homogeneous, and the
reaction was normally complete within 12 h as indicated by analytical
HPLC. The solution was filtered through a pad of Celite to remove
any traces of solid residue, and acetonitrile was slowly added to the
stirred filtrate. The precipitated solid was centrifuged to give the crude
product after washing with diethyl ether and drying 18 h under vacuum.
When appropriate, preparative LC was used to provide a purer sample.
The procedure is exemplified below by the synthesis of folic acid (L-
1) and tetramethylguanidinium L-methyl folate (γ) (33a).
Synthetic L-1 ()25a). To a suspension of 27 (200 mg, 0.590 mmol)
and L-glutamic acid (131 mg, 0.890 mmol) in DMSO (30 mL) was
added neat tetramethylguanidine (0.222 mL, 1.77 mmol). The stirred
mixture soon became homogeneous, and the reaction was complete
after 9 h, as indicated by analytical HPLC. The solution was filtered
through a pad of Celite to remove any traces of solid residue, and
acetonitrile (50 mL) was slowly added to the stirred filtrate. The
precipitated solid was then centrifuged to give the crude folic acid (248
mg) after washing with aqueous 1% HCl solution (10 mL × 1),
acetonitrile (10 mL × 1), and diethyl ether (25 mL × 2) and drying 18
General Methods. Melting points were obtained on a MEL-TEMP
apparatus and are uncorrected. Unless otherwise stated, reactions were
carried out under argon in flame-dried glassware. Tetrahydrofuran
(THF) and diethyl ether (Et2O) were distilled from sodium benzophe-
none ketyl. Dichloromethane and benzene were distilled from calcium
hydride. Cyclohexane was stored over sodium metal. Deuterated NMR
solvents (CDCl3 and CD3CN) were stored over 4 Å molecular sieves
for several days prior to use. Flash chromatography on silica gel was
carried out as described by Still33 (230-400 mesh silica gel was used),
and reversed phase LC was used for preparative purposes (LiChroprep
C-18, 310 mm × 25 mm). 1H- and 13C-NMR spectra were obtained
using GE QE-300 NMR and Varian Gemini 200 NMR spectrometers
at 300 or 200 MHz and 75 or 50 MHz respectively. 1H-NMR chemical
shifts are reported in ppm relative to the residual protonated solvent
resonance: CHCl3, δ 7.26; C6D5H, δ 7.15. Splitting patterns are
designated as s, singlet; d, doublet; t, triplet; q, quartet; p, pentet; m,
multiplet; br, broadened. Coupling constants (δ) are reported in Hertz.
13C-NMR chemical shifts are reported in ppm relative to solvent
resonance: CDCl3, δ 77.00; C6D6, δ 128.00. Mass spectral data were
(33) Still, W. C.; Kahn, M.;Mitra, A. J. Org. Chem. 1978, 43, 923.
(34) Fu, S.-C. J.; Reiner, M. J. Org. Chem. 1965, 30, 1277.
(35) Cramer, S. M.; Schornagel, J. H.; Kalghatgi, K. K.; Bertino, J. R.
J. Cancer Res. 1984, 44, 1843.

Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10011
h under vacuum. An analytical sample (174 mg, 67%) was obtained
using preparative reversed-phase HPLC. 1H-NMR, UV, and analytical
HPLC were all the same as those of commercial folic acid. Decom-
position point, ∼238 °C. [R]25D ) +17.4° (c ) 0.5 in 0.1 N NaOH);
0.5 in DMSO) (cf. commercial folic acid from Vitamins, Inc. ) +14.4°
at the same concentration). Analytical HPLC: tR ) 11.9 min (flow
rate, 0.7 mL/min; eluent A, water and phosphate buffer 50 mmol/dm3,
pH 7; eluent B, acetonitrile; gradient, 0 min, 2% B; 25 min, 50% B;
column, Econosphere C18, 150 mm × 4.6 mm).
98.4% of L-folic acid by the enzyme assay. Analytical HPLC: tR
)
Pyrofolic Acid (9). To a stirred homogeneous solution of N10-
(trifluoroacetyl)pyrofolic acid (8, 14 g, 27 mmol) and DMF (250 mL)
was slowly added aqueous cesium carbonate (10 mL, 82 mmol) at 25
°C. Analytical HPLC indicated completion of the reaction by disap-
pearance of 8 in 5 h. The mixture was filtered through a pad of Celite,
and the filtrate was carefully acidified to pH 4 with 5% aqueous
hydrochloric acid. The resulting precipitate was thoroughly washed
with water (100 mL × 3) by centrifugation, acetonitrile (100 mL ×
1), and diethyl ether (100 mL × 2) by aspirator filtration. The yellowish
product 9 (10.2 g, 89%) was obtained after drying for 24 h at 60 °C
under vacuum. 1H-NMR (300 MHz, DMSO-d6): δ 8.64 (s, H, C7-H),
7.41 (d, J ) 8.3 Hz, 2H, Ar), 6.58 (d, J ) 8.3 Hz, 2H, Ar), 4.70 (m,
1H, C19-H), 4.47 (d, J ) 4.7 Hz, 2H, C9-H2), 2.57-1.80 (overlap,
4H). 13C-NMR (300 MHz, DMSO-d6): δ 174.0, 172.8, 169.1, 161.2,
153.9, 152.2, 148.7, 148.4, 132.4, 132.3, 128.1, 120.4, 110.6, 59.0,
45.8, 31.6, 21.6. Mass spectrum (FAB): m/z 424 (MH+). HRMS:
calcd for C19H17N7O5 424.1369; found, 424.1365. Mp ) ∼269 °C
dec. [R]25D ) -1.2° (c ) 0.5 in DMSO). Analytical HPLC: tR ) 9.0
min (flow rate, 0.7 mL/min; eluent A, water and phosphate buffer 50
mmol/dm3, pH 7; eluent B, acetonitrile; gradient, 0 min, 2% B; 25
min, 50% B; column, Econosphere C18, 150 mm × 4.6 mm).
7.2 min (flow rate, 0.7 mL/min; eluent A, water and phosphate buffer
50 mmol/dm3, pH 7; eluent B, acetonitrile; gradient, 0 min, 2% B; 25
min, 50% B; column, Econosphere C18, 150 mm × 4.6 mm).
Determination of Optical Purity of Folate Derivatives Listed in
Table 3 Using Enzymatic Hydrolysis with Carboxypeptidase G.29
The folic acid derivative (∼2.5 mg) was dissolved in 1 mL of TRIS
buffer containing 20 µg of ZnCl2, and 50 microunits of enzyme (Sigma
Chemical Co.) was added. The solution was incubated at 37 °C for 2
h. A 20 µL sample was tested by HPLC on a 4.6 mm × 250 mm
Microsorb C18 reversed-phase column (eluent A, 5 mM phosphate
buffer; eluent B, acetonitrile; flow rate, 1 mL/min; gradient, 0-5 min,
5% B, 10-15, 25% B). Folic acid was eluted at 4.1 min, pteroic acid
at 10.6 min, methotrexate at 10.8 min, and amino-N10-methylpteroic
acid at 11.9 min. Peak area was used as the standard for analysis.
For folate derivatives, ∼5 mg was first dissolved in 1 mL of 0.25
M NaOH and incubated at room temperature for 1 h. The solution
was then acidified by 1 M HCl until precipitation. After centrifugation,
washing, and drying, the ∼2.5 mg yellow pellet was dissolved in 1
mL of TRIS buffer and reacted with enzyme under the conditions
described above.
N2,10-Bis(trifluoroacetyl)pyrofolic Acid/Anhydride (6/7). To a
mechanically stirred suspension of 1 (100 g, 0.23 mol, U.S.P. grade,
supplied from Vitamins Inc., Chicago, IL) and anhydrous tetrahydro-
furan (1000 mL) in a three-neck flask was slowly added trifluoroacetic
anhydride (256 mL, 1.81 mol) at 0 °C over ∼0.5 h before warming
the solution to 25 °C. The mixture gradually turned into a dark brown
homogeneous phase as the reaction proceeded. After 10 h, analytical
HPLC showed that the reaction was complete by the appearance of a
number of peaks (presumably a mixture of 7 and other trifluoroacety-
lated pyrofolic acids) and confirmed the absence of 1. The solution
was filtered through a pad of Celite to remove a small amount of solid
residue. Using a rotary evaporator, the filtrate was concentrated to a
dark brown viscous liquid (∼300 mL), which was slowly transferred
with the aid of tetrahydrofuran (∼20 mL) to a flask of well-stirred
benzene (1500 mL). The precipitated yellowish solid was collected
by filtration and washed with diethyl ether (250 mL × 1) to yield crude
product 6/7 (151 g). 1H-NMR (300 MHz, DMSO-d6): δ 8.89 (s, 1H,
C7-H), 7.65 (s, 4H, Ar), 5.25 (s, 2H, C9-H2), 4.71 (dd, J ) 4.2, 8.9
Hz, C19-H), 2.59-1.98 (overlap, 4H). 13C-NMR (200 MHz, DMSO-
d6): δ 174.4, 172.6, 170.9, 169.1, 166.0, 165.3, 159.8, 159.1, 159.0,
158.3, 157.5, 156.4, 155.7, 155.1, 149.2, 147.4, 142.2, 135.1, 130.1,
129.4, 128.4, 124.9, 124.8, 119.2, 118.1, 113.5, 112.4, 107.7, 106.6,
58.7, 54.1, 31.4, 21.6. 19F NMR (300 MHz, DMSO-d6): δ -65.66,
-74.13, -80.47 (integration ratio 1.0:0.93:0.13). Analytical HPLC:
tR ) 12.7 min (flow rate, 0.7 mL/min; eluent A, water and phosphate
buffer 50 mmol/dm3, pH 7; eluent B, acetonitrile; gradient, 0 min, 2%
B; 25 min, 50% B; column, Econosphere C18, 150 mm × 4.6 mm).
N10-(Trifluoroacetyl)pyrofolic Acid (8). The crude N2,10-bis-
(trifluoroacetyl)pyrofolic acid/anhydride (6/7, 150 g) was dissolved in
tetrahydrofuran (500 mL), followed by addition of ice (∼100 g) with
stirring. Analytical HPLC indicated that all the original peaks
converged after ∼3 h at 25 °C into a single one (N10-(trifluoroacetyl)-
pyrofolic acid (8)). The mixture was then slowly transferred to
efficiently stirred diethyl ether (2000 mL). The precipitated yellowish
powder was collected by filtration, triturated with diethyl ether, washed
thoroughly with diethyl ether (200 mL × 3), and dried 18 h under
vacuum, giving 8 (123 g) in a quantitative yield from 1. 1H-NMR
(300 MHz, DMSO-d6): δ 8.64 (s, H, C7-H), 7.62 (s, 4H, Ar), 5.12 (s,
2H, C9-H2), 4.70 (dd, J ) 3.2, 4.9 Hz, 1H, C19-H), 2.54-2.43 (overlap,
4H). 13C-NMR (300 MHz, DMSO-d6): δ 176.8, 174.7, 172.9, 169.4,
161.2, 156.9, 156.4, 156.0, 155.8, 155.5, 154.5, 149.8, 149.7, 145.1,
142.5, 135.3, 130.3, 128.7, 118.5, 114.7, 110.9, 59.0, 54.4, 31.8, 21.9.
19F-NMR (300 MHz, DMSO-d6): δ -65.2. Mass spectrum (FAB):
m/z 519 (MH+). HRMS: calcd for C21H16F3N7O6, 519.1114; found,
519.1112. Softens and decomposes at 208-214 °C. Elemental analysis
calcd for C21H26F3N7O6‚0.5H2O: H, 3.24; C, 47.73; F, 10.79; N, 18.65.
Pteroylamide (11) and Pyroglutamic Acid (12). N10-(Trifluoro-
acetyl)pyrofolic acid (8, 540 mg, 1.04 mmol) was dissolved in aqueous
concentrated ammonium hydroxide solution (25 mL). As the reaction
proceeded, yellowish solid precipitated. After 14 h, the solid powder
11 (217 mg, 67%) was isolated by filtration and washed with water
(20 mL × 3), methanol (20 mL × 1), and ether (25 mL × 2) and dried
under vacuum. 1H-NMR (300 MHz, DMSO-d6): δ 8.63 (s, 1H, C7-
H), 7.61 (d, J ) 8.3 Hz, 2H, Ar), 6.68 (d, J ) 8.3 Hz, 2H, Ar), 4.44
(d, J ) 5.2 Hz, 2H, C9-H2), 13C-NMR (300 MHz, TFA/inserted DMSO-
d6 tube): δ 172.6, 158.6, 151.5, 149.6, 146.7, 145.9, 138.0, 132.6, 130.2,
125.8, 122.9, 53.2. Mass spectrum (FAB): m/z 311 (M+). HRMS:
calcd for C14H13N7O2, 312.1209 (MH+); found, 312.1205. Decomposi-
tion point, ∼293 °C. Analytical HPLC: tR ) 9.8 min (flow rate, 0.7
mL/min; eluent A, water and phosphate buffer 50 mmol/dm3, pH 7;
eluent B, acetonitrile; gradient, 0 min, 2% B; 25 min, 50% B; column,
Econosphere C18, 150 mm × 4.6 mm). To isolate pyroglutamic acid
12, the solvent of the above aqueous filtrate was removed by a rotary
aspirator, and the residue was then purified using preparative HPLC
to give 12 (112 mg, 84%). 1H-NMR was the same as that one of
commercial pyroglutamic acid. [R]25D ) -0.74° (c ) 1.84 in water).
Mp ) 154-158 °C. Pyrofolic acid (9, 100 mg, 0.236 mmol), in a
similar treatment with ammonium hydroxide, yielded 11 (11 mg, 15%)
and 12 (23 mg, 92%). For 11, [R]25 ) +0.09° (c ) 0.46 in water).
D
Pteroyl Hydrazide (13). N10-(Trifluoroacetyl)pyrofolic acid (8, 49
g, 94 mmol) was dissolved in DMSO (1000 mL) with mechanical
stirring. To this homogeneous solution was added hydrazine (30 mL,
0.94 mol) while maintaining the temperature at 25 °C. During the
process, the flask was immersed in a water bath at 25 °C, and the
hydrazine was added slowly in order to moderate a gentle exotherm.
The reaction was complete after 8 h, as indicated by analytical HPLC,
and the mixture was filtered through a pad of Celite to remove a trace
of solid residue. To the filtrate was then slowly added methanol (1000
mL), and the resulting precipitated solid was collected by aspirator
filtration (or centrifugation) and washed thoroughly with methanol (200
mL × 3), followed by diethyl ether (200 mL × 2), to yield crude
product 13 (28 g, 91%) after drying for 18 h under vacuum. 1H-NMR
(300 MHz, D2O/NaOD): δ 7.91 (s, H, C7-H), 6.98 (d, J ) 8.5 Hz,
2H, Ar), 6.47 (d, J ) 8.5 Hz, 2H, Ar), 4.04 (s, 2H, C9-H2). 13C-NMR
(300 MHz, DMSO-d6/concentrated HCl ∼10/1 v/v): δ 166.3, 158.6,
152.9, 152.8, 152.2, 148.1, 146.9, 130.1, 128.6, 117.9, 112.1, 46.0.
Mass spectrum (FAB): m/z 327 (MH+). HRMS: calcd for C14H14N8O2,
327.1318; found, 327.1307. Decomposition point, ∼291 °C. Analytical
HPLC: tR ) 14.2 min (flow rate, 0.7 mL/min; eluent A, water and
phosphate buffer 50 mmol/dm3, pH 7; eluent B, acetonitrile; isocratic,
5% B; column, Econosphere C18, 150 mm × 4.6 mm).
Found: H, 3.04; C, 47.63; F, 10.72; N, 18.64. [R]25 ) +2.9° (c )
D

10012 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Luo et al.
DL-Methyl Folate (γ) 25c ()33a)). To a stirred suspension of
pyrofolic acid (9, 1.1 g, 2.6 mmol) and methanol (150 mL) was added
lithium hexamethyldisilazide (1.0 M in THF, 7.8 mL, 7.8 mmol) at
-10 °C. The mixture soon became homogeneous, and the reaction
was complete in 9 h at this temperature, as indicated by analytical LC.
The reaction was quenched by adding acetic acid (5 mL) at -78 °C.
The yellow precipitate was isolated by centrifugation and then purified
by preparative reversed-phase HPLC to give the methyl folate 25c (804
mg, 70%). 1H-NMR (300 MHz, DMSO-d6): δ 8.60 (s, 1H, C7-H),
7.91 (d, J ) 7.3 Hz, 1H, N18-H), 7.58 (d, J ) 8.4 Hz, 2H, Ar), 7.25
(s, 2H, C2-NH2), 6.90 (t, J ) 5.3 Hz, 1H, N10-H), 6.60 (d, J ) 8.4
Hz, 2H, Ar), 4.45 (d, J ) 5.3 Hz, 2H, C9-H2), 4.22 (dd, J ) 7.6, 12.3
Hz, 1H, C19-H), 3.51 (s, 3H, OCH3), 2.47 (t, J ) 1.7 Hz, 2H, C21-
H2), 2.36-1.88 (m, 2H, C20-H2). 13C-NMR (300 MHz, DMSO-d6): δ
174.9, 173.7, 166.7, 162.1, 156.6, 154.8, 151.2, 149.0, 148.9, 129.4,
128.3, 122.0, 111.8, 52.9, 51.8, 46.4, 30.7, 27.3. LRMS (PDMS) for
C20H21N7O6 (MH+): calcd, 455, found, 455.5. Decomposition point
∼251 °C. Analytical HPLC: tR ) 10.4 min (flow rate, 0.7 mL/min;
eluent A, water and phosphate buffer 50 mmol/dm3, pH 7; eluent B,
acetonitrile, gradient, 0 min, 2% B; 25 min, 50% B; column,
Econosphere C18, 150 mm × 4.6 mm).
(γ) (2.63 g, 16.3 mmol) in DMSO (50 mL) was added neat tetrameth-
ylguanidine (3.7 mL, 29.7 mmol). The stirred mixture soon became
homogeneous, and the reaction was complete after 6 h, as indicated
by analytical HPLC. The solution was filtered through a pad of Celite
to remove a trace of solid residue, and acetone (400 mL) was slowly
added to the stirred filtrate. The precipitated solid was then filtered to
give the crude tetramethylguanidinium salt of methyl folate (γ) (33a,
R ) TMG-H+) (6.1 g, 88%) after washing with diethyl ether (50 mL
× 2) and drying for 18 h under vacuum. 1H-NMR (300 MHz, DMSO-
d6): δ 8.58 (s, 1H, C7-H), 7.55 (d, J ) 8.4 Hz, 2H, Ar), 6.87 (t, J )
5.3 Hz, 1H, N10-H), 6.62 (d, J ) 8.4 Hz, 2H, Ar), 4.41 (d, J ) 5.3
Hz, 2H, C9-H2), 3.98 (dd, J ) 7.6, 12.3 Hz, 1H, C19-H), 3.49 (s, 3H,
OCH3), 2.81 (s, 12H, teramethylguanidine), 2.40-2.18 (m, 2H, C21-
H2), 2.17-1.78 (m, 2H, C20-H2). Mass spectrum (FAB): m/z 454 (M
- H+). HRMS: calcd for C20H21N7O6, 454.1475; found, 454.1445.
Spiking experiments in 1H-NMR and analytical HPLC showed this
material to be identical to 25c, except for the added presence of TMG.
N-Alloc-glutamic Acid (47). To 50 g (340 mmol) of L-glutamic
acid in 340 mL of distilled water was added 72 g (857 mmol) of sodium
bicarbonate in several portions. After the bubbling ceased, 56 mL (510
mmol) of allyl chloroformate was added all at once and the solution
was left to stir for 18 h at 25 °C. The pH was adjusted to 1 using
concentrated hydrochloric acid and assayed using pH paper. The
solution was transferred to a separatory funnel and extracted 10 times
with ethyl acetate. The combined organics were rinsed with brine and
dried over magnesium sulfate. The solvent was removed in Vacuo to
give 57 g (73%) of crude 47 as a colorless syrup.
3-[3-((Allyloxy)carbonyl)-5-oxo-1,3-oxazolan-4-yl]propanoic Acid
(48). 47 (56.9 g, (246 mmol), paraformaldehyde (16 g 492 mmol),
p-toluenesulfonic acid (4.7 g (24.6 mmol), and benzene (1.2 L) were
combined in a round-bottom flask equipped with a stir bar and a Dean-
Stark trap and heated at reflux for 1 h. The solvent was removed in
Vacuo, and the residue was purified by column chromatography (SiO2
4:1 hexane/ethyl acetate) to afford 48 (40.6 g, 68%) as a white solid.
1H-NMR (300 MHz, CDCl3): δ 5.90 (m, 1H), 5.55 (m, 1H), 5.30 (m,
1H), 4.66 (d, J ) 5.85 Hz, 2H), 4.41 (t, J ) 6.01 Hz, 1H), 2.54 (t, J
) 6.95 Hz, 2H), 2.28 (m, 2H). 13C-NMR (75 MHz, CDCl3): δ 177.4,
171.6, 152.9, 131.4, 118.7, 77.6, 66.8, 53.7, 28.9, 25.4. Mass spectrum
(CI): 244 (M + H+, base peak). HRMS: calcd for C10H13NO6,
244.0821; found, 244.0823.
N-Alloc r-Allylglutamate (51a). Elemental sodium (0.43 g, 18.7
mmol) was added in small pieces to 197 mL of allyl alcohol cooled to
0 °C under argon in a flame-dried round-bottom flask. After all the
sodium had dissolved, 2.39 g (9.83 mmol) of oxazolidinone 48 in 20
mL of allyl alcohol was cannulated into the solution of alkoxide. After
the solution was stirred for 1 h at 0 °C, the pH was adjusted to 1 using
concentrated hydrochloric acid and pH paper. The solution was
transferred to a separatory funnel, diluted with an equal volume of water,
and extracted 6 times with ethyl acetate. The combined organics were
rinsed with brine and dried over magnesium sulfate. The solvent was
removed in Vacuo, and the residue was purified by column chroma-
tography (SiO2, 4:1 hexane/ethyl acetate) to afford 51a (2.1g, 79%) as
a pale yellow oil. 1H-NMR (300 MHz, CDCl3) δ 5.85 (m, 2H), 5.43
(d, J ) 7.98 Hz, N-H), 5.26 (m, 4H), 4.64 (d, J ) 5.75 Hz, 2H), 4.57
(d, J ) 5.33 Hz, 2H), 4.42 (m, 1H), 2.48 (m, 2H), 2.24 (m, 1H), 2.0
(m, 1H). 13C-NMR (75 MHz, CDCl3): δ 176.8, 171.4, 155.8, 132.1,
131.0, 118.2, 117.2, 65.5, 65.4, 52.8, 29.4, 26.4. Mass spectrum (CI):
m/z 272 (M + H+, base peak). HRMS: calcd for C12H17NO6, 272.1134;
found, 272.1142.
N-Alloc r-Dimethylallylglutamate (51b). Elemental sodium (0.54
g, 23.5 mmol) was added in small pieces to 250 mL of allyl alcohol
cooled to 0 °C under argon in a flame-dried round-bottom flask. After
all the sodium had dissolved, 3.0 g (12.3 mmol) of 48 in 25 mL of
allyl alcohol was cannulated into the solution of alkoxide. After the
solution was stirred for 1 h at 0 °C, the pH was adjusted to 1 using
concentrated hydrochloric acid as monitored using pH paper. The
solution was transferred to a separatory funnel, diluted with an equal
volume of water, and extracted 6 times with ethyl acetate. The
combined organics were rinsed with brine and dried over magnesium
sulfate. The solvent was removed in Vacuo, and the remaining liquid
was distilled under vacuum to remove the excess dimethylallyl alcohol
(0.10 mmHg, bp 39 °C). The residue was purified by column
Pteroyl Azide (27) and N10-Nitrosopteroyl Azide (28). To a stirred
suspension of pteroyl hydrazide (13, 28 g, 86 mmol) and potassium
thiocyanate (0.41 g, 4.2 mmol) was added ice-cold trifluoroacetic acid
(220 mL). After the solid dissolved, the reaction mixture was cooled
to -10 °C, followed by slow addition of neat tert-butyl nitrite (10 mL,
86 mmol). Monitoring by analytical HPLC indicated that the reaction
was complete in 4 h, after which time the reaction was warmed to 25
°C. In addition to the desired pteroyl azide (27), analytical HPLC
sometimes showed generation of a small amount (∼10%, due to 90%
purity of pteroyl azide (27)) of N10-nitrosopteroyl azide (28), which
could be instantly converted into 27 at 25 °C simply by addition of
sodium azide (0.5 equiv per equiv of 13 used in the reaction) to the
reaction mixture. The solution was then filtered through a pad of celite
to remove a trace of solid residue. Slow addition of 2-propanol (250
mL) to the stirred filtrate led to an orange powder, which was collected
by centrifugation, washed thoroughly with water (500 mL × 3),
acetonitrile (500 mL × 1), and diethyl ether (200 mL × 2), and finally
dried for 24 h under vacuum. Multiple cycles of the precipitation
process were used in the trifluoroacetic acid-2-propanol combination
to obtain material of even higher purity (known impurities include ∼4%
of pteroic acid (10), 2% of the aniline apparently resulting from Curtius
rearrangement of 27, ∼2% of pteroylamide (11), and <1% of folic
acid.). Normally, a purity of >90% (determined by HPLC) was
achieved after a single purification cycle. The 27 (21 g, 72%) thus
obtained was stored at -15 °C with protection from light.
27. 1H-NMR (300 MHz, DMSO-d6): δ 8.63 (s, 1H, C7-H), 7.65
(d, J ) 8.7 Hz, 2H, Ar), 7.57 (m, 1H, N10-H), 6.67 (d, J ) 8.7 Hz,
2H, Ar), 4.49 (d, J ) 5.8 Hz, 2H, C9-H2). 13C-NMR (300 MHz,
DMSO-d6): δ 171.0, 158.6, 153.9, 152.7, 152.5, 148.3, 146.8, 131.8,
128.3, 117.6, 112.2, 45.8. Mass spectrum (FAB): m/z 338 (MH+).
HRMS: calcd for C14H11N9O2, 338.1114; found, 338.1110. Decom-
position point, ∼180 °C. Analytical HPLC: tR ) 18.0 min (flow rate,
0.7 mL/min; eluent A, water and phosphate buffer 50 mmol/dm3, pH
7; eluent B, acetonitrile; gradient, 0 min, 2% B; 25 min, 50% B; column,
Econosphere C18, 150 mm × 4.6 mm).
28. 1H-NMR (300 MHz, DMSO-d6): δ 8.74 (s, 1H, C7-H), 8.08
(d, J ) 8.6 Hz, 2H, Ar), 7.91 (d, J ) 8.6 Hz, 2H, Ar), 5.52 (s, 2H,
C9-H2). 13C-NMR (300 MHz, DMSO-d6): δ 171.6, 160.3, 153.8, 153.3,
149.3, 146.7, 145.1, 131.1, 129.0, 128.5, 120.0, 46.9. Analytical
HPLC: tR ) 19.2 min (flow rate, 0.7 mL/min; eluent A, water and
phosphate buffer 50 mmol/dm3, pH 7; eluent B, acetonitrile; gradient,
0 min, 2% B; 25 min, 50% B; column, Econosphere C18, 150 mm ×
4.6 mm).
4-{[N-[(2,4-Diamino-6-pteridinyl)methyl]-N-methyl]amino}ben-
zoyl Azide (29).22b A solution of 56 (100 mg, 0.31 mmol), diphen-
ylphosphoryl azide (0.13 mL 0.46 mmol), and Et3N (0.09 mL, 0.62
mmol) in DMSO (5 mL) was stirred for 20 h. THF (10 mL) was then
added, and the precipitate was filtered to give 29 (92 mg, 83%). The
1H-NMR spectrum is identical to the one in the literature.22b
Tetramethylguanidinium L-Methyl Folate (γ) (33a, R ) TMG-
H+). To a suspension of 27 (5 g, 14.8 mmol) and methyl glutamate

Efficient Syntheses of Pyrofolic Acid and Pteroyl Azide
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10013
chromatography (SiO2, 4:1 hexane/ethyl acetate) to afford 51b (2.6 g,
71%) as a pale yellow oil. 1H-NMR (300 MHz, CDCl3): δ 5.90 (m,
1H), 5.42 (d, J ) 8.0 Hz, N-H), 5.30 (m, 3H), 4.64 (d, J ) 7.31 Hz,
2H), 4.57 (d, J ) 5.56 Hz, 2H), 4.40 (m, 1H), 2.47 (m, 2H), 2.22 (m,
1H), 1.98 (m, 1H), 1.76, (s, 3H), 1.71 (s, 3H). 13C-NMR (75 MHz,
CDCl3), δ 177.0, 171.8, 155.8, 139.5, 132.2, 117.6, 117.4, 65.5, 62.1,
52.9, 29.6, 27.0, 25.3, 17.6. Mass spectrum (CI): m/z 300 (M + H+,
base peak). HRMS: calcd for C14H21NO6, 300.1447; found, 300.1459.
2-Aminoethylfolic Acid, γ-Amide (52). To the crude tetrameth-
ylguanidinium l-methyl folate (γ) (33a, R ) TMG-H+) (2.8 g, 5.95
mmol) was added ethylenediamine (20 mL, 0.3 mol) with stirring at
25 °C. The solid gradually dissolved as it reacted with the diamine.
The reaction was complete in 3 h, as indicated by analytical HPLC,
and filtration gave a clear solution, which was then transferred to a
well-stirred mixture of acetonitrile and diethyl ether (1:1 v/v, 500 mL).
The precipitated solid was collected by centrifugation and redissolved
in water (500 mL), followed by addition of aqueous 5% hydrochloric
acid until pH 7.0, which effected precipitation of the product.
Centrifugation was used to collect the precipitate, which was triturated
with water and washed thoroughly with water (250 mL × 3) to remove
any trace of ethylenediamine (1H-NMR was used to assay for the
presence of the diamine). A yellow solid (2.1 g, 88%) was obtained
after washing with acetonitrile (100 mL × 1) and diethyl ether (50 mL
× 3) and drying for 24 h under vacuum. 1H-NMR (300 MHz, DMSO-
d6/CF3CO2D ∼10:1 v/v): δ 8.75 (s, 1H, C7-H), 7.66 (d, J ) 8.6 Hz,
2H, Ar), 6.63 (d, J ) 8.6 Hz, 2H, Ar), 4.57 (s, 2H, C9-H2), 4.34 (dd,
J ) 4.0, 9.6 Hz, 1H, C19-H), 3.28∼3.13 (m, 2H, C24-H2), 2.80 (m,
2H, C25-H2), 2.16 (m, 2H, C21-H2), 2.15-1.85 (m, 2H, C20-H2). 13C-
NMR (300 MHz, D2O/NaOD): δ 178.8, 176.0, 173.1, 169.5, 164.0,
155.5, 151.1, 147.3, 147.2, 129.0, 128.2, 121.4, 112.5, 55.3, 45.8, 42.0,
39.8, 32.8, 28.0. Mass spectrum (FAB): m/z 484 (MH+). HRMS:
calcd for C21H25N9O5, 484.2057; found, 484.2062. [R]25D ) +4.5° (c
) 0.4 in 1.0 N NaOH). Decomposition point, ∼278 °C. Analytical
HPLC: tR ) 15.4 min (flow rate, 0.7 mL/min; eluent A, water and
phosphate buffer 5 mmol/dm3, pH 7; eluent B, acetonitrile; gradient, 0
min, 1% B; 15 min, 10% B; column, Econosphere C18, 150 mm ×
4.6 mm).
4-{[N-[(2,4-Diamino-6-pteridinyl)methyl]-N-methyl]amino}ben-
zoic Acid 56.22b The sodium salt of 34d (equivalent to 200 mg, 0.44
mmol, from the National Cancer Institute) was dissolved in 5 mL of
0.1 M TRIS buffer. Carboxypeptidase G (∼5 units) and ZnCl2 (1 mg)
were added, and the pH of the solution was adujsted to 7.2 with
concentrated HCl. The solution was shaken in an incubator at 38 °C
for 2 days, and the mixture (pH 8.4) was then adjusted to pH 3.5 with
dilute HCl. The precipitated yellow solid was filtered, washed with
H2O, EtOH, and Et2O, and dried to give the known acid 56 (152 mg,
1
96%). The H-NMR spectrum is identical to the one from Aldrich.
Acknowledgment. We wish to thank Endocyte, Inc. for
partial support of this work. Arlene Rothwell provided the MS
data. We thank Dr. Jae Wook Lee for valuable intellectual
contributions to the folate-mediated multiple drug delivery
program. M.D.S. would like to thank Kris Krishnamurthy and
the Kodak Corp. for generous fellowship support.
Supporting Information Available: Experimental proce-
dures for the synthesis of compounds: 14-22, 25d,f,h,i,
26c,d,f,i, 30c, 31, 33b,c, 40, 41, and 45, as well as 1H and 13C-
NMR of all new compounds (96 pages). See any current
masthead page for ordering and Internet access instructions.
JA971568J