Electrophilic fluorination of pyroglutamic acid derivatives: Application of substrate-dependent reactivity and diastereoselectivity to the synthesis of optically active 4-fluoroglutamic acids

Article abstract of DOI:10.1021/jo0106804

Electrophilic fluorination of enantiomerically pure 2-pyrrolidinones (4) derived from (L)-glutamic acid has been investigated as a method for the synthesis of single stereoisomers of 4-fluorinated glutamic acids. Reaction of the lactam enolate derived from 9 with NFSi results in a completely diastereoselective monofluorination reaction to yield the monocyclic trans-substituted α-fluoro lactam product 21. Unfortunately, a decreased kinetic acidity in 21 and other structurally related monofluorinated products renders them resistant to a second fluorination. In contrast, the bicyclic lactam 12 is readily difluorinated under the standard conditions described to yield the α,α-difluoro lactam 24. The difference in reactivity between the two types of related lactams is attributed mainly to the presence or lack of a steric interaction between the base used for deprotonation and the protecting group present in the pyrrolidinone substrates. This conclusion was reached based on analysis of the X-ray crystal structure of 21, molecular modeling, and experimental evidence. The key intermediates 21 and 24 are converted to (2S,4R)-4-fluoroglutamic acid and (2S)-4,4-difluoroglutamic acid, respectively.

Full text of DOI:10.1021/jo0106804

J . Org. Chem. 2001, 66, 8831-8842
8831
Electr op h ilic F lu or in a tion of P yr oglu ta m ic Acid Der iva tives:
Ap p lica tion of Su bstr a te-Dep en d en t Rea ctivity a n d
Dia ster eoselectivity to th e Syn th esis of Op tica lly Active
4-F lu or oglu ta m ic Acid s
David W. Konas and J ames K. Coward*
Departments of Chemistry and Medicinal Chemistry, The University of Michigan,
Ann Arbor, Michigan 48109
jkcoward@umich.edu
Received J uly 6, 2001
Electrophilic fluorination of enantiomerically pure 2-pyrrolidinones (4) derived from (L)-glutamic
acid has been investigated as a method for the synthesis of single stereoisomers of 4-fluorinated
glutamic acids. Reaction of the lactam enolate derived from 9 with NFSi results in a completely
diastereoselective monofluorination reaction to yield the monocyclic trans-substituted R-fluoro lactam
product 21. Unfortunately, a decreased kinetic acidity in 21 and other structurally related
monofluorinated products renders them resistant to a second fluorination. In contrast, the bicyclic
lactam 12 is readily difluorinated under the standard conditions described to yield the R,R-difluoro
lactam 24. The difference in reactivity between the two types of related lactams is attributed mainly
to the presence or lack of a steric interaction between the base used for deprotonation and the
protecting group present in the pyrrolidinone substrates. This conclusion was reached based on
analysis of the X-ray crystal structure of 21, molecular modeling, and experimental evidence. The
key intermediates 21 and 24 are converted to (2S,4R)-4-fluoroglutamic acid and (2S)-4,4-
difluoroglutamic acid, respectively.
Sch em e 1
In tr od u ction
Fluorinated amino acids are powerful molecular tools
for bioorganic and medicinal chemists. The substitution
of fluorine for hydrogen within the natural amino acid
structure can impart dramatic, yet predictable, effects
on chemical reactivity and is sterically conservative
enough so that the resulting product is often still active
in biological systems. Because of this fact, fluoroamino
acids and larger molecules containing them have enjoyed
widespread popularity and use in applications such as
biochemical probes, alternate enzyme substrates, and
enzyme inhibitors.1,2 A consequence of this popularity is
the existence of a large body of literature describing
methods for the synthesis of these compounds.3,4 Despite
this progress, there remains a strong demand for im-
proved strategies and synthetic routes.
investigations, the most valuable data are obtained from
experiments utilizing synthetic substrates which are
stereochemically pure. Thus, efficient and stereoselective
routes to a variety of fluoroglutamate isomers are desired.
One previously effective synthetic method relies on the
general strategy shown in Scheme 1. Protected fluori-
nated prolines 1 are obtained by reaction of the corre-
sponding oxygenated heterocycles with DAST and then
converted to pyroglutamate derivatives 2 via oxidation
with RuO4. A final one-step hydrolysis of the protecting
groups and lactam ring gives the desired fluoroglutamate
products. Application of this strategy resulted in the
synthesis of (2S,4S)-4-fluoroglutamic acid7,10 and (2RS)-
3,3-difluoroglutamic acid11 for use as peptide building
blocks and substrates in biochemical experiments.
Unfortunately, the DAST/RuO4 method cannot be
applied to the synthesis of (2S)-4,4-difluoroglutamic acid.
Synthesis of the necessary enantiomerically pure di-
This laboratory is interested in studying the biochemi-
cal and physiological effects of incorporating fluorine-
containing glutamic acids into folates and antifolates in
terms of their ability to enhance or hinder the formation
and hydrolysis of poly-γ-glutamate conjugates.5-9 In such
(1) Welch, J . T.; Eswarakrishnan, S. Fluorine in Bioorganic Chem-
istry; Wiley: New York, 1991.
(2) Organofluorine Compounds in Medicinal Chemistry and Bio-
medical Applications; Filler, R., Kobayashi, Y., Yagupolskii, L. M., Eds.;
Elsevier: Amsterdam, 1993.
(3) Fluorine-Containing Amino Acids, Synthesis and Properties;
Kukhar, V. P.; Soloshonok, V. A., Eds.; Wiley: Chichester, West Sussex,
England, 1995.
(4) Sutherland, A.; Willis, C. L. Nat. Prod. Rep. 2000, 17, 621.
(5) Tsukamoto, T.; Coward, J . K.; McGuire, J . J . In Biomedical
Frontiers of Fluorine Chemistry; Ojima, I.; McCarthy, J . K., Welch, J .
T., Eds.; American Chemical Society: Washington, D.C., 1996; p 118.
(6) McGuire, J . J .; Hart, B. P.; Haile, W. H.; Rhee, M.; Galivan, J .;
Coward, J . K. Arch. Biochem. Biophys. 1995, 321, 319.
(7) Hart, B. P.; Haile, W. H.; Licato, N. J .; Bolanowska, W. E.;
McGuire, J . J .; Coward, J . K. J . Med. Chem. 1996, 39, 56.
(8) Tsukamoto, T.; Kitazume, T.; McGuire, J . J .; Coward, J . K. J .
Med. Chem. 1996, 39, 66.
(9) McGuire, J . J .; Hart, B. P.; Haile, W. H.; Magee, K. J .; Rhee,
M.; Bolanowska, W. E.; Russell, C.; Galivan, J .; Paul, B.; Coward, J .
K. Biochem. Pharmacol. 1996, 52, 1295.
(10) Hudlicky, M. J . Fluorine Chem. 1993, 60, 193.
(11) Hart, B. P.; Coward, J . K. Tetrahedron Lett. 1993, 34, 4917.
10.1021/jo0106804 CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/27/2001

8832 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
fluorinated proline12,13 and application of the strategy
shown in Scheme 1 was unsuccessful due to the complete
failure of the RuO4 ring oxidation.14 This result is
consistent with an oxidation mechanism proceeding
through a carbocation intermediate as suggested by
Lee,15 although evidence for alternate possibilities has
been reported.16 In any case, it is clear that the strongly
electron-withdrawing CF2 group produces an insurmount-
able destabilizing effect on this transformation. In addi-
tion, even in the cases of fluorinated proline derivatives
where RuO4 oxidation is successful, the transformation
suffers from the problem of extended reaction times and
inconsistent yields.
Electrophilic fluorination using so-called “N-F” re-
agents is a method emerging in recent years for the
selective introduction of fluorine at electron-rich centers
such as alkenes, aromatics, and carbanions.17,18 These
reagents are usually stable solids which can be used with
standard laboratory glassware and lack the high reactiv-
ity, corrosiveness, and toxicity of earlier sources of
positive fluorine. One such compound possessing par-
ticular popularity and commercial availability is N-
fluorobenzenesulfonimide (NFSi, 3).19 Electrophilic fluo-
rination of pyroglutamate-derived lactam enolates was
envisioned as a means to provide fluoroglutamates
substituted in the 4-position while completely avoiding
the problematic RuO4 ring oxidation. It was anticipated
that treatment of a compound such as 4 with a deproto-
nation/NFSi fluorination sequence would lead to an
intermediate 5 (eq 1) which could be converted to the
desired (2S)-4,4-difluoroglutamic acid as a single stereo-
isomer.
This report describes the investigation of electrophilic
fluorination of pyroglutamate-derived pyrrolidinone lac-
tams with NFSi and its application to the synthesis of
fluorinated glutamic acids.
Resu lts
Syn th esis of F lu or in a tion Su bstr a tes. The syn-
thetic routes utilized in preparing the protected lactams
for this work are shown in Scheme 2. These pathways
illustrate a general strategy for obtaining a wide variety
of N,O-protected pyroglutaminols. This class of com-
pounds was selected as substrates in preference to
protected pyroglutamates and acyclic glutamates based
on some unpromising preliminary results involving N-Boc
pyroglutamate ethyl ester and concerns over unwanted
side reactions such as R-carbon racemization which are
possible in the presence of an excess of various strong
bases.21 Path A allows access to trityl or silyl ether
protection at the primary alcohol (R) with any desired
amide protecting group (R′). Path B results in benzyl or
4-methoxybenzyl (PMB) protected amides (R′) and any
desired ether protecting group for the alcohol (R). The
starting material in every case is L-glutamic acid, which
makes these pathways attractive for use on larger scales.
Utilization of this starting material ultimately results in
the same natural L, or 2S, configuration in the final
products. While this configuration is of greatest interest
for most biological studies, the use of D-glutamic acid and
the same chemistry should provide access to the corre-
sponding D-fluoroglutamate stereoisomers. Various ex-
amples and combinations of each type of protection were
prepared and examined as fluorination substrates.
Mon oflu or in a tion of La cta m Su bstr a tes. To study
the reactivity and diastereoselectivity of the electrophilic
fluorination reaction, each substrate was subjected to a
set of standardized monofluorination conditions which
are described in the Experimental Section. The results,
derived from NMR analysis (1H, 19F) and material
recovered after chromatography, are summarized in
Table 1. With only one exception (10), successful diaster-
eoselective monofluorination was observed. In general,
substrates with benzyl protecting groups for the amide
led to higher yields and greater diastereoselectivity than
those protected with a Boc carbamate. Interestingly,
there was no particular correlation between the degree
of diastereoselectivity and the size of the oxygen protect-
ing group. There were also no predictable trends based
on whether this group was an alkyl or silyl ether.
Electrophilic fluorination of enolates is advantageous
in that it can utilize a large body of existing carbanion
chemistry. Reactions of electrophiles with enolates of
lactams related to the general structure 4 having various
combinations of protecting groups are numerous in the
literature and generally proceed with significant dia-
stereoselectivity. The major products of such monosub-
stitutions most commonly contain the two ring substit-
uents in a trans relationship to each other.20 Thus, it may
be possible to use electrophilic fluorinations or other
substitutions for the synthesis of monofluorinated 4-
fluoroglutamates while taking advantage of this stereo-
chemical bias to control the configuration at the site of
fluorination and avoiding the problematic RuO4 ring
oxidation.
Clear exceptions to these general trends were com-
pounds 9 and 10 which both contain a trityl ether.
Subjecting 10 to these conditions led mainly to un-
changed starting material while monofluorination of 9,
although low-yielding, showed complete diastereoselec-
tivity and produced only a single monofluorinated prod-
uct. An X-ray crystal structure of this product was
obtained, and the fluorine was found to be in a trans
relationship with respect to the protected hydroxymethyl
group across the lactam ring.22 By comparison of 1H and
(12) Sufrin, J . R.; Balasubramanian, T. M.; Vora, C. M.; Marshall,
G. R. Int. J . Pept. Protein Res. 1982, 20, 438.
(13) Demange, L.; Menez, A.; Dugave, C. Tetrahedron Lett. 1998,
39, 1169.
(14) Hart, B. P. Ph.D. Thesis, The University of Michigan, Ann
Arbor, 1995.
(15) Lee, D. G.; Van den Engh, M. Can. J . Chem. 1972, 50, 3129.
(16) Bakke, J . M.; Froehaug, A. E. J . Phys. Org. Chem. 1996, 9, 310.
(17) Lal, G. S.; Pez, G. P.; Syvret, R. G. Chem. Rev. 1996, 96, 1737.
(18) Taylor, S. D.; Kotoris, C. C.; Hum, G. Tetrahedron 1999, 55,
12431.
(19) Differding, E.; Ofner, H. Synlett 1991, 187.
(20) Brena-Valle, L. J .; Carreon Sanchez, R.; Cruz-Almanza, R.
Tetrahedron: Asymmetry 1996, 7, 1019.
(21) Del Bosco, M.; J ohnstone, A. N. C.; Bazza, G.; Lopatriello, S.;
North, M. Tetrahedron 1995, 51, 8545.
(22) Crystallographic data (excluding structure factors) for the
structure in this paper have been deposited with the Cambridge
Crystallographic Data Centre as supplementary publication no. CCDC
166557. Copies of the data can be obtained, free of charge, on appli-
cation to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: +44
1223 336033 or e-mail: deposit@ccdc.cam.ac.uk).

Fluorination of Pyroglutamic Acid Derivatives
J . Org. Chem., Vol. 66, No. 26, 2001 8833
Sch em e 2
difluorinations of the lactam substrates. A recent report23
described the electrophilic difluorination of the Corey
lactone using NFSi. Consistent with the results here,
difluorination of the lactone was not observed following
simple treatment of the substrate with base followed by
NFSi. Instead, it required a large excess of reagents and,
most importantly, the presence of ZnCl2 or MnBr2. As a
positive control, electrophilic difluorination of a com-
mercial sample of the Corey lactone was attempted, and
the reaction proceeded as described in the literature.
Unfortunately, these same conditions did not result in
difluorination of either unsubstituted (9, 15) or trans-
monofluorinated (21, 22) pyrrolidinone lactams. A sum-
mary of these efforts is shown in Table 2. A trace of the
desired product was periodically detected in the crude
reaction mixture by 19F NMR, but even this poor result
was difficult to reproduce, and it was never possible to
isolate, purify, and characterize any difluorinated mono-
cyclic lactams.
In dramatic contrast to these results was the reactivity
of the bicyclic substrate 12 toward difluorination. Under
the standard monofluorination conditions (Table 1) it
gave a good yield of monofluorinated product, but the
least diastereoselectivity of any substrate examined.
However, the most important observation regarding 12
was the appearance of a difluorinated product in the
reaction mixture when an excess of LDA and NFSi was
used, indicating a greater reactivity of 12 toward di-
fluorination. This greater reactivity was confirmed when
sequential treatment of 12 and then the monofluorinated
intermediate 23 (Scheme 3) with the standard mono-
fluorination conditions led to good yields of the difluori-
nated lactam 24. Compound 24 was then converted in
three steps to (2S)-4,4-difluoroglutamic acid providing a
practical and efficient route to this stereochemically pure
fluoroamino acid which was described in detail in a
previous publication from this laboratory.24
Ta ble 1. Su m m a r y of La cta m Mon oflu or in a tion s w ith
NF Sia
compound
R
R′
% yieldb isomer ratioc
15
17
16
19
10
18
11
20
9
Bn
Bn
Bn
Bn
Bn
TBDMS
TBDPS
TIPS
CH3
C(Ph)3
68
66
5.7:1.0d
3.3:1.0
4.9:1.0
5.7:1.0
N/A
4.9:1.0
1.6:1.0
1.7:1.0
single isomer
1.2:1.0
>50e
70
0
64
4-(MeO)-Bn TBDMS
Boc
Boc
Boc
N/A
TBDMS
CH3
C(Ph)3
40
<38f
20
12
N/A [bicylic]
59
a
b
See Experimental Section for details. Total yield of all
fluorinated products. c Determined by NMR analysis of crude
reaction mixtures. The trans isomer is the major product in each
case. Based on material recovered after chromatography. e Yield
d
based on incomplete recovery of products after chromatography.
f Yield based on complete recovery of products plus an inseparable
impurity after chromatography.
19F NMR coupling patterns and chemical shifts to this
known diastereomer, the major products in all of the
other standard monofluorination reactions were assigned
to have the trans relative stereochemistry as well.
Diflu or in a tion of La cta m Su bstr a tes. During the
course of examining the monofluorination of the mono-
cyclic lactam 9, it was noted that even in cases where an
excess of base and NFSi was used, difluorinated products
were not observed. Unfortunately, this observation con-
tinued when many subsequent deliberate attempts at
difluorination under the standard conditions (Table 1 and
Experimental Section) and closely related variants failed
to yield any of the desired difluorinated lactam corre-
sponding to 5 (eq 1).
(23) Nakano, T.; Makino, M.; Morizawa, Y.; Matsumura, Y. Angew.
Chem., Int. Ed. Engl. 1996, 35, 1019.
(24) Konas, D. W.; Coward, J . K. Org. Lett. 1999, 1, 2105.
In addition to these standard conditions, a number of
other attempts were made to carry out electrophilic

8834 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
Ta ble 2. Su m m a r y of Un su ccessfu l Electr op h ilic
Diflu or in a tion Exp er im en tsa
sufficient solubility in common organic solvents to allow
for easy product extraction, isolation, and purification.
Unfortunately, the best monofluorination substrate 9,
which yields a completely diastereoselective fluorination,
contains protecting groups with very similar reactivity.
A wide variety of attempts were made to selectively
remove the trityl group from 21 in good yield, although
none were successful. These experiments included the use
of protic acids (HCl, CSA, citric acid, formic acid), Lewis
acids (BCl3,25 Et2AlCl,26 Yb(OTf)3,27) hydrogenolysis (H2/
Pd-C, H2/Pd(OH)2, HCO2NH4/Pd-C), and others (CAN,
CAN-SiO2,28 I2/MeOH29). In some cases, the appearance
of only the desired product 30 was observed initially, but
it was soon followed by the undesired product in which
both protecting groups had been removed at a rate such
that a good yield of 30 could not be obtained in a single
reaction step. Attempts to increase the yield of 30 by
altering the reaction conditions were not successful.
An alternative to the use of 21 as a key intermediate
was compound 22 which was derived from 15 and easily
separated from its minor cis diastereomer in the crude
fluorination reaction mixture (Table 1) using silica gel
chromatography. Removal of the TBDMS ether (TBAF)
to yield 31 then oxidation (RuO4) and alkylation (CH2N2)
to the corresponding methyl ester proceeded cleanly in
good yields, but subsequent removal of the benzyl group
was problematic. The only method which proved to be
successful was a dissolving metal reduction (Na/NH3(liq)),
but these conditions caused undesired reduction of the
C-F bond.
a
Results were obtained by examination of the 1H and 19F NMR
spectra of the crude reaction mixtures. The syntheses of 9 and 15
are outlined in Scheme 2. The monfluorinated derivatives 21 and
22 were obtained as shown in Scheme 4 and Table 1 (15 f 22),
respectively. All reactions were run in THF unless noted other-
wise.
b
Sch em e 3
Conservative protecting group substitutions based on
compound 9 designed to increase the difference in
reactivity between the two different groups were envi-
sioned, and two more protected pyroglutaminols were
prepared. The first contained a 4,4′-dimethoxytrityl
(DMT) ether instead of the trityl ether (R ) DMT, R′ )
Boc)30 and the second contained a Cbz group instead of
Boc (R ) Tr, R′ ) Cbz).31 Unfortunately, attempted
monofluorination of these analogues resulted in a mix-
ture of multiple products and was not pursued further.
Ultimately, it became clear that conversion of 21 to
(2S,4R)-4-fluoroglutamic acid would have to take place
as shown in Scheme 5 via simultaneous removal of both
protecting groups followed by a single-phase oxidation
in a polar solvent. Thus, the Boc and trityl groups were
removed with either aqueous HCl in MeOH or anhydrous
trifluoroacetic acid (TFA) in the presence of triethylsilane
to give the deprotected pyroglutaminol 27. Oxidation of
Syn th esis of (2S,4R)-4-F lu or oglu ta m ic Acid . The
completely diastereoselective monofluorination of 9 re-
vealed great promise for the development of a synthetic
route to (2S,4R)-4-fluoroglutamic acid, but the initial low
yields required improvement. The reaction was examined
in detail with respect to the base, temperature, dilution,
reagent stoichiometry, and even reagent sources. The
final successfully optimized procedure (Scheme 4, path
A) used to obtain 21 (See Experimental Section) consis-
tently gives yields of nearly 60% while maintaining the
absolute trans diastereoselectivity.
Development of a method to convert 21 to (2S,4R)-4-
fluoroglutamic acid was not straightforward. It was
originally assumed that the best strategy to convert
protected monofluorinated pyroglutaminols to fluoro-
glutamates involved initial selective deprotection of the
oxygen protecting group as shown in eq 2. A partially
protected intermediate, 30 or 31, could then be oxidized
to the required carboxylic acid while still maintaining
(25) J ones, G. B.; Hynd, G.; Wright, J . M.; Sharma, A. J . Org. Chem.
2000, 65, 263.
(26) Koester, H.; Sinha, N. D. Tetrahedron Lett. 1982, 23, 2641.
(27) Lu, R. J .; Liu, D. P.; Giese, R. W. Tetrahedron Lett. 2000, 41,
2817.
(28) Hwu, J . R.; J ain, M. L.; Tsai, F. Y.; Tsay, S. C.; Balakumar, A.;
Hakimelahi, G. H. J . Org. Chem. 2000, 65, 5077.
(29) Wahlstrom, J . L.; Ronald, R. C. J . Org. Chem. 1998, 63, 6021.
(30) Prepared from 6 by reaction with DMT-Cl (1.1 equiv) and
pyridine (1.3 equiv) in CH2Cl2 followed by Boc protection according to
the method used for 9.
(31) Prepared from 7 by deprotonation with LiHMDS (1.1 equiv) in
THF at -78 °C followed by reaction with Cbz-Cl (1.2 equiv).

Fluorination of Pyroglutamic Acid Derivatives
J . Org. Chem., Vol. 66, No. 26, 2001 8835
Sch em e 4
Sch em e 5
BAIB combination were carried out, but unfortunately
this method also was found to be inconsistent and
problematic for the conversion of the alcohol 27 to the
corresponding carboxylic acid. Finally, success was
achieved by returning to the TEMPO/NaOCl system and
using a milder reaction medium. When this oxidation was
carried out in a mixture of acetone and 5% aqueous
NaHCO3, the desired acid product was obtained cleanly
without any of the previously observed undesired byprod-
ucts. Also successful was the application of a method
utilizing catalytic chromium trioxide (CrO3) and periodic
acid (H5IO6) in wet acetonitrile38 and modification of the
reported standard procedure to avoid an aqueous-
organic partition in the workup step. The crude carboxylic
acid product was esterified to give 28 which could be
isolated and purified by standard silica gel chromatog-
raphy. A final HCl hydrolysis followed by propylene oxide
neutralization of the crude amino acid hydrochloride gave
the desired product (2S,4R)-4-fluoroglutamic acid 29 as
a single stereoisomer.
small polar alcohols such as 27 to the corresponding
carboxylic acids in good yield with convenient product
isolation is a challenge. Initially, aqueous oxidation of
27 with O2 and a platinum catalyst32 was investigated,
but the reaction was found to be sluggish and inconsis-
tent. Both a single-phase RuO4 oxidation in aqueous
acetone33 and the J ones reagent under standard condi-
tions consumed the starting material, but the reaction
was not clean, and a mixture of undesired byproducts
was formed. Similar problems regarding the oxidation
of polar molecules have been encountered in the area of
carbohydrate chemistry. In some cases, these have been
overcome utilizing catalytic TEMPO and stoichiometric
hypochlorite as a primary oxidant.34 Treatment of 27 with
TEMPO/NaBr/NaOCl in deionized water kept basic (pH
g 10) with aqueous NaOH35 resulted in a complex
mixture of products. Spectroscopic analysis of the product
mixture gave evidence of NaOCl-mediated N-chlorination
of the amide in 27 followed by elimination and further
decomposition in the strongly basic medium. Thus, such
undesired chlorination reactions should be considered
when substrates with unprotected N-H bonds are in-
volved. An alternative to hypochlorite as a primary
oxidant when utilizing TEMPO is bis-acetoxyiodobenzene
(BAIB).36,37 A variety of experiments involving the TEMPO/
In d ir ect Cis F lu or in a tion of Ch ir a l P yr oglu ta m i-
n ols. Due to the trans directing effect observed for the
reaction of electrophiles with enolates derived from
lactams such as 9, it is not possible to synthesize (2S,4S)-
4-fluoroglutamic acid directly via electrophilic fluorina-
tion because the required intermediate contains a cis
relationship between the incoming fluorine and the
substituent at the existing chiral center.
The conversion of aliphatic alcohols to alkyl fluorides
with inversion of configuration is often successful as a
strategy for selective fluorination. Thus, it was proposed
that a diastereoselective electrophilic hydroxylation of 9
followed by fluorination with stereochemical inversion
could provide access to the cis-substituted intermediate
26 while still taking advantage of the strong trans
directing effect present in this substrate. A similar
strategy had been attempted previously for the synthesis
of intermediates which could be converted into fluoro-
glutamates.39 In the present case, the electrophilic hy-
droxylation of the lactam enolate of 9 with a good yield
and high stereoselectivity had already been reported.40
Thus, deprotonation of 9 and reaction with MoOPH41
according to the literature method as shown in Scheme
4 gave the trans-substituted alcohol 25. The diastereo-
(32) Saijo, S.; Wada, M.; Himizu, J .; Ishida, A. Chem. Pharm. Bull.
1980, 28, 1449.
(33) Garner, P.; Park, J . M. J . Org. Chem. 1990, 55, 3772.
(34) de Nooy, A. E. J .; Besemer, A. C.; van Bekkum, H. Synthesis
1996, 1153.
(38) Zhao, M.; Li, J .; Song, Z.; Desmond, R.; Tschaen, D. M.;
Grabowski, E. J . J .; Reider, P. J . Tetrahedron Lett. 1998, 39, 5323.
(39) Avent, A. G.; Bowler, A. N.; Doyle, P. M.; Marchand, C. M.;
Young, D. W. Tetrahedron Lett. 1992, 33, 1509.
(35) Li, K.; Helm, R. F. Carbohydr. Res. 1995, 273, 249.
(36) De Mico, A.; Margarita, R.; Parlanti, L.; Vescovi, A.; Piancatelli,
G. J . Org. Chem. 1997, 62, 6974.
(40) Shin, C. G.; Nakamura, Y.; Yamada, Y.; Yonezawa, Y.; Ume-
mura, K.; Yoshimura, J . Bull. Chem. Soc. J pn. 1995, 68, 3151.
(41) Vedejs, E.; Larsen, S. Organic Syntheses; Wiley & Sons: New
York, 1990; Collect. Vol. No. VII, pp 277-282.
(37) Epp, J . B.; Widlanski, T. S. J . Org. Chem. 1999, 64, 293.

8836 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
selectivity of this reaction was analogous to that of the
corresponding fluorination (9 f 21) and none of the cis-
substituted alcohol was observed within the limit of NMR
detection while analyzing the crude reaction mixture.
Reaction of 25 with either DAST or Deoxo-Fluor42
produced the desired product, but in low yields. Instead
of pursuing these reagents further, the alcohol was
converted to the corresponding triflate, and reaction with
a nucleophilic fluoride source was investigated. Treat-
ment of the triflate with an excess of TBAF solution at
-78 °C led to the formation of an initial reversible adduct
characterized by a bright yellow color and a new spot
observed by TLC. A quench of the reaction at this point
led to the recovery of the original triflate starting
material. However, when the reaction solution was
allowed to slowly warm and stir at room temperature,
the desired cis-substituted product 26 was observed by
NMR. During experiments designed to trap the initially
formed adduct, hypothesized to involve reversible addi-
tion of fluoride to the electron deficient lactam carbonyl,
it was discovered that addition of iodomethane to the
reaction mixture before warming to room-temperature
resulted in a cleaner substitution reaction with less
undesired side products.
Development of this route to 26 (Scheme 4, path B)
provides the key component required for an alternate
synthesis of (2S,4S)-4-fluoroglutamic acid utilizing a
diastereoselective electrophilic pyrrolidinone enolate sub-
stitution and avoiding the problematic RuO4 oxidation.
Utilization of the chemistry shown in Scheme 5 for the
synthesis of (2S,4R)-4-fluoroglutamic acid should allow
for conversion of 26 to the diastereomeric (2S,4S)-4-
fluoroglutamic acid. The most important use of the cis-
substituted compound 26, however, was as a tool for
examining the reactivity differences with respect to
difluorination between the monocyclic and bicyclic sub-
strates used in this work (see Discussion).
center. This finding provides further evidence that exist-
ing carbanion chemistry can be transferred successfully
to electrophilic fluorination reactions. The origin of the
facial selectivity in pyrrolidinone alkylations and sub-
stitutions has been investigated in particular detail in
recent years.43-45 While some debate still remains, it is
clear that the selectivity results from a combination of
steric, torsional, and electronic effects. It is not surprising
that compound 12, the fluorination substrate with the
least steric differentiation between the two faces of its
lactam ring, also exhibits the least diastereoselectivity
in monofluorination. In sharp contrast is the completely
diastereoselective monofluorination of 9, which might
initially be attributed strictly to steric hindrance from
the bulky triphenylmethyl protecting group. However,
Table 1 includes a variety of other ethers ranging in size
from the equally bulky TBDPS to the much smaller
methyl. These other compounds also show diastereose-
lectivity in fluorination, but the magnitude is much less
relative to 9, and this selectivity does not vary signifi-
cantly with ether size. This provides evidence that
contributions in addition to steric factors are also likely
to play a role in determining the degree of diastereo-
selectivity observed in 9. Thus, good to excellent levels
of diastereoselectivity can be obtained with the proper
choice of protecting groups, although it may not always
be possible to predict the absolute best combination a
priori. The most striking result of these experiments,
illustrated in Scheme 6, is the dramatic difference in
reactivity with respect to electrophilic difluorination
between protected monocyclic pyroglutaminols and the
bicyclic pyrrolidinone 12. To explain this difference, the
questions of why the monocyclic lactams such as 9 cannot
be difluorinated under these reaction conditions and what
factors in 12 make it more reactive toward difluorination
must be addressed.
It is now well established that the ease of electrophilic
fluorination using N-F reagents for a given substrate
depends strongly on the acidity of the hydrogen being
replaced. Accordingly, active methylene groups in mal-
onates are the most reactive and can be converted to the
corresponding difluoromethylene derivatives even by
reaction with only the N-F reagent and no added base.46
Protons positioned next to an amide carbonyl are much
less acidic, and therefore electrophilic fluorination at this
position is expected to be more difficult. In fact, prior to
the experiments reported here an electronic reaction
search utilizing CAS SciFinder and Beilstein Crossfire
yielded only a single example of the synthesis of an R,R-
difluoroamide using N-F reagents involving a substrate
which did not contain an additional activating group such
as a carbonyl in the â-position.47 Even in this single
example, the substrate was a small unfunctionalized
molecule and the yield was moderate at best. With
respect to esters and lactones, the case in the literature
was found to be largely the same, but a report of the
N-F-mediated electrophilic difluorination of the Corey
lactone in good yield23 proved that such a reaction is
Discu ssion
The development of convenient and readily available
sources of electrophilic fluorine over the past decade has
led to a rapidly expanding body of new literature describ-
ing their application in organic synthesis. In this work,
it was proposed that the use of electrophilic fluorination
of reduced pyroglutamic acid derivatives would provide
an efficient route to single stereoisomers of 4-fluoro-
glutamic acids and eliminate the need for a problematic
RuO4 oxidation associated with an earlier route to these
compounds.7,10 Of particular interest was (2S)-4,4-difluo-
roglutamic acid, a target which was readily available only
in racemic form.8 However, the ability to control the
stereochemistry of electrophilic monofluorinations lead-
ing to single stereoisomers of 4-fluoroglutamic acid was
also an attractive concept.
The electrophilic monofluorinations of the substrates
shown in Table 1 proceed, with the exception of the
bicyclic lactam 12, with significant diastereoselectivity.
Consistent with numerous other substitution reactions
of optically active 5-substituted-2-pyrrolidinone lactam
enolates appearing in the literature, the major products
of these fluorinations have a trans relationship between
the fluorine and the substituent at the adjacent chiral
(43) Meyers, A. I.; Seefeld, M. A.; Lefker, B. A.; Blake, J . F. J . Am.
Chem. Soc. 1997, 119, 4565.
(44) Ando, K.; Green, N. S.; Li, Y.; Houk, K. N. J . Am. Chem. Soc.
1999, 121, 5334.
(45) Groaning, M. D.; Meyers, A. I. Tetrahedron 2000, 56, 9843.
(46) Banks, R. E.; Lawrence, N. J .; Popplewell, A. L. J . Chem. Soc.,
Chem. Commun. 1994, 343.
(42) Lal, G. S.; Pez, G. P.; Pesaresi, R. J .; Prozonic, F. M.; Cheng,
H. S. J . Org. Chem. 1999, 64, 7048.
(47) Differding, E.; Ruegg, G. M.; Lang, R. W. Tetrahedron Lett.
1991, 32, 1779.

Fluorination of Pyroglutamic Acid Derivatives
J . Org. Chem., Vol. 66, No. 26, 2001 8837
Sch em e 6
indeed possible. Despite the fact that esters are more
acidic than amides,48 this result served as a significant
inspiration prior to beginning this study.
C(2)-O dihedral angle is 66.3°, a deviation from the ideal
90° for maximum orbital overlap in the transition state.
Figure 1B shows the position occupied by the base during
deprotonation of this intermediate. It is expected that the
bulky bases used in these experiments (Tables 1 and 2)
would encounter significant destabilizing steric repulsion
from the trityl or any of the other ether protecting groups
used in the monocyclic fluorination substrates. This is
verified by noting that the crystal structure of 21 shows
the trityl group is indeed positioned over the top face of
the lactam ring. This final factor is very significant
because if it is accepted that one fluorine will not
significantly affect the conformation of 9 vs 21 then the
deprotonation of the nonfluorinated substrate 9 will face
nonideal stereoelectronic factors, but it can take place
from the face opposite the trityl ether and not encounter
the unfavorable steric interactions created by this large
group. This is consistent with the experimental result
that 9 can be fluorinated but the trans-substituted
monofluorinated 21 cannot (Scheme 6). Thus, in addition
to the desirable role it plays in directing the diastereo-
selectivity of monofluorination, a steric effect due to the
ether protecting group in monocyclic pyroglutaminols
appears to be responsible for the failure of the attempted
electrophilic difluorinations (Table 2).
The success in monofluorination of 9 clearly indicates
that the acidity and reactivity of this compound, and
other related monocyclic lactams from Table 1, are
sufficient to allow deprotonation and reaction with NFSi.
Given the strong connection between electrophilic fluo-
rination and acidity, it is reasonable to assume that the
failure to fluorinate 21 and other related monofluorinated
monocyclic compounds (e.g., 22) a second time (Table 2)
is related to a decreased kinetic acidity of the fluoro-
methylene CHF proton in such intermediates relative to
the corresponding methylene CH2 protons in the original
unfluorinated starting material. Despite its high elec-
tronegativity, it has been demonstrated that fluorine
substitution can actually be a destabilizing factor for
resonance-stabilized sp2-hybridized carbanion formation
when it is directly bonded to the carbon developing the
negative charge.49,50 Because of this R-fluoro carbanion
effect, some fluorinated molecules are actually less acidic
than their corresponding protio analogues. This effect
should apply in compounds such as 21, rendering them
less acidic and more difficult to fluorinate a second time.
However, the same effect will also be present in the
bicyclic monofluorinated substrate 23 and therefore
cannot explain the difference in reactivity between these
two compounds.
The rate of deprotonation from a position next to a
carbonyl depends on stereoelectronic, torsional, and steric
effects.51 Since the X-ray crystal structure of 21 is
available, the structural and electronic features of this
monocyclic 2-pyrrolidinone can be qualitatively evaluated
and compared to the bicyclic lactam as shown in Figure
1. The transition state for CH deprotonation is stabilized
by the stereoelectronic “CH-π-overlap effect”. This sta-
bilization results from orbital alignment between the CH
σ bond to be broken, and therefore the p orbital of the
developing anion, and the carbonyl π system. Maximum
stabilization occurs when these orbitals are exactly
aligned, which is the case when the dihedral angle
between the proton being removed and the carbonyl
oxygen is 90°. Figure 1A shows the Newman projection
between C2 and C3 in 21. The important H-C(3)-
This explanation can be strengthened experimentally
by noting that the major factor proposed to limit diflu-
orination, an unfavorable steric interaction between the
base and ether protecting group on the top face of the
lactam ring, will only apply in the case of trans-
substituted monofluorinated lactams such as 21. There-
fore, the corresponding cis-substituted lactam 26 should
exhibit a greater reactivity than 21 under these condi-
tions and difluorination should be possible. Indeed, when
a sample of 26 obtained through the hydroxylation-
fluorination method (Scheme 4) was subjected to the
standard monofluorination conditions, the observed re-
sult was very different from that seen with 21. The 19F
NMR spectrum of the resulting crude reaction mixture
showed, in addition to a number of minor unidentified
signals, three prominent resonances corresponding to the
desired difluorinated lactam (δ -25 ppm), the trans-
substituted lactam 21 (δ -110 ppm) and the cis-
substituted starting material 26 (δ -108 ppm) in a ratio
of 2:2:1. A mass spectrum of this mixture also served to
verify the presence of these products which were not
separable by silica gel chromatography. Clearly, unlike
the monofluorinated substrates shown in Table 2, com-
(48) Bordwell, F. G.; Fried, H. E. J . Org. Chem. 1981, 46, 4327.
(49) Adolph, H. G.; Kamlet, M. J . J . Am. Chem. Soc. 1966, 88, 4761.
(50) Castejon, H. J .; Wiberg, K. B. J . Org. Chem. 1998, 63, 3937.
(51) Behnam, S. M.; Behnam, S. E.; Ando, K.; Green, N. S.; Houk,
K. N. J . Org. Chem. 2000, 65, 8970.

8838 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
F igu r e 1. Newman projections through the C2-C3 (A,C) and C3-C4 (B,D) bonds of the X-ray crystal structure of 21 and the
MM2 molecular mechanics model of trans-23.
pound 26 is deprotonated under these conditions, and the
resulting enolate can react with NFSi resulting in di-
fluorination. In addition, under the reaction conditions
used, it appears that not all of the enolate generated was
able to react with NFSi. Instead, some of it was repro-
tonated upon quenching to yield the epimerized product
21. The completely diastereoselective monofluorination
of 9 indicates that the top face of the O-trityl-protected
lactam enolate ring may be too hindered to react with
the bulky NFSi, but this must not be the case for the
smaller proton donors (NH4+, H3O+) used to quench the
reaction.
To complete the analysis, the factors outlined for the
failure of substrates such as 9 to form difluorinated
products must either be absent in the bicylic 12, or other
factors unique to this substrate which offset the decrease
in kinetic acidity upon monofluorination must be present.
Given the reactivity difference just described between the
cis- and trans-substituted monofluorinated pyrrolidino-
nes, some difluorination of bicyclic 23 might be expected
since it is obtained from 12 as a near equal mixture of
both cis and trans diastereomers (Table 1). On the basis
of the yield of the difluorination step to give the product
24, though, it is obvious that both stereoisomers of 23
have the ability to be difluorinated. By this same argu-
ment it might be suggested that difluorinated lactams
should have been observed in the experiments sum-
marized in Table 1 since cis-substituted lactams were
present for many of the substrates after addition of NFSi
to the reaction. It is likely, however, that difluorination
was not observed since only a small excess of base was
used and, in addition to the deprotonation/fluorination
reaction, there is also a decomposition pathway between
NFSi and the strong bases utilized in these experiments
which serves to consume the reagents.52
terization of distorted amides which possess either a
significantly pyramidalized nitrogen, a perpendicularly
twisted C(O)-N bond, or both. These distortions break
the resonance interaction between the nitrogen lone pair
and the amide carbonyl resulting in functional groups
which are better described based on their reactivity and
physical properties as amino-ketones than amides. As
such, protons adjacent to the carbonyl of a distorted
amide should be more acidic than those of a normal
resonance-stabilized amide. Therefore, it was proposed
that the constraints of the bicyclic system in 12 might
adversely affect the amide resonance and make the
R-protons more acidic than those in corresponding mono-
cyclic pyrrolidinones. The crystal structure of 21 indicates
that the amide nitrogen is almost perfectly planar with
the three groups deviating from planarity by only 0.017
Å. In contrast, MM2 molecular mechanics and semi-
empirical PM3 modeling of 12 and the corresponding
monofluorinated intermediates indicate a more distinct
pyramidalization of the amide nitrogen which supports
the hypothesis of a lack of amide resonance in 12.
However, the spectroscopic data contradict this. The 13C
NMR chemical shift is a reflection of the charge density
around the carbon atom. Accordingly, distorted amides
which lack electron donation from nitrogen are charac-
terized by larger downfield carbonyl 13C chemical shifts
relative to planar amides. The average of the amide
carbonyl 13C chemical shift for every substrate in Table
1 except for the bicyclic 12 is 175.58 ( 0.28 ppm. The
13C carbonyl chemical shift for 12 is 171.58 ppm. This
represents a significant difference between 12 and the
rest of the substrates, but the shift of 12 is upfield from
the rest and not consistent with decreased amide reso-
nance. The carbonyl IR absorption frequency is also
diagnostic for evaluating the extent of amide distortion.
The amide carbonyl of 12 absorbs at 1686 cm-1 which is
consistent with a normal nontwisted lactam.
The true nature of the concept of amide resonance has
been the subject of a debate in recent years.53 Ac-
companying this debate has been the study and charac-
(53) The Amide Linkage: Selected Structural Aspects in Chemistry,
Biochemistry, and Materials Science; Greenberg, A., Breneman, C. M.,
Liebman, J . F., Eds.; Wiley: New York, N.Y., 2000.
(52) Konas, D. W. Unpublished results.

Fluorination of Pyroglutamic Acid Derivatives
J . Org. Chem., Vol. 66, No. 26, 2001 8839
structure was obtained by the X-ray facility of the Department
of Chemistry, The University of Michigan. Elemental analyses
were obtained from the CHN/AA service of the Department of
Chemistry, The University of Michigan. Molecular modeling
was done using Spartan software from Wavefunction, Inc. and
Cambridge Software’s Chem3D Pro for Microsoft Windows.
(S)-5-(Hydroxymethyl)-2-pyrrolidinone 6 was synthesized by
the method of Pickering et al.54 (Scheme 2, path A) or
purchased from Sigma-Aldrich. (S)-5-(tert-Butyldimethylsiloxy-
methyl)-1-(tert-butyloxycarbonyl)-2-pyrrolidinone 11 was syn-
thesized from 6 via intermediate 8 using a minor modification
of the method of Ikota.55 (S)-5-(Hydroxymethyl)-1-(phenyl-
methyl)-2-pyrrolidinone 13 and (S)-5-(hydroxymethyl)-1-(4-
methoxyphenylmethyl)-2-pyrrolidinone 14 were synthesized by
modification of the procedure reported by Olsen et al.56
(Scheme 2, path B). (S)-5-(Methoxymethyl)-1-(phenylmethyl)-
2-pyrrolidinone 19 was synthesized from 13 as described by
Brena-Valle et al.20 Oxodiperoxymolybdenum(pyridine)-(hexa-
methylphosphorictriamide) (MoOPH) was prepared by the
method of Vedejs and Larsen.41 N-Fluorobenzenesulfonimide
was obtained from Honeywell International, Inc. Solutions of
n-butyllithium were obtained from Fisher-Acros. Tetrahydro-
furan (THF) was freshly distilled from sodium/benzophenone.
Dichloromethane was freshly distilled from calcium hydride.
Pyridine was distilled after heating 5 h at reflux over barium
oxide. Diisopropylamine was distilled from sodium hydroxide.
1,1,1,3,3,3-Hexamethyldisilazane was distilled from calcium
hydride. All other reagents and starting materials were
obtained from Sigma-Aldrich or Fisher-Acros and used without
further purification. Air- and moisture-sensitive reactions were
run in flame- or oven-dried (T > 100 °C overnight) glassware
under an atmosphere of dry nitrogen or argon.
(S)-5-(Tr ityloxym eth yl)-2-p yr r olid in on e (7). Compound
6 (1.0 g, 8.8 mmol) and triphenylmethyl chloride (3.64 g, 13.05
mmol, 1.5 equiv) were stirred in anhydrous pyridine (15 mL).
The reaction was heated to 100 °C under a condenser and
drying tube for 3 h. The mixture was cooled to rt and then
slowly poured into vigorously stirred ice-water. Stirring of this
mixture continued for 10 min until a yellow gum was deposited
in the cloudy white water. This gum was washed and decanted
with freshwater three times and then dissolved in CH2Cl2. This
organic layer was washed with water and brine, dried over
Na2SO4, and filtered. The solvent was removed in vacuo, and
the resulting orange oil was partially purified using silica gel
column chromatography (CH2Cl2 f CH2Cl2/MeOH, 19:1) to
give a light yellow solid which contained the desired tritylated
product plus a trace of residual pyridine. The product NMR
spectrum matched that of authentic material obtained from
Aldrich, and the sample was used for Boc protection without
further purification. 1H NMR (300 MHz, CDCl3) δ 1.65 (m, 1H),
2.14 (m, 1H), 2.30 (m, 2H), 2.99 (t, 1H, J ) 9.2 Hz), 3.20 (dd,
1H, J ) 9.2, 3.9 Hz), 3.86 (m, 1H), 6.07 (br, 1H), 7.24-7.42
(m, 15H).
To help evaluate stereoelectronic effects, an energy-
minimized MM2 molecular mechanics model of trans-23
was calculated for comparison with the crystal structure
of 21. Figures 1C and 1D illustrate the Newman projec-
tions of this model corresponding to those for the crystal
structure of 21 in Figures 1A and 1B. As shown in Figure
1C, the calculated H-C(3)-C(2)-O dihedral angle is
34.2° which would correspond to much less CH-π-overlap
stabilization for deprotonation from the top face relative
to 21. Therefore, what is probably the most important
factor in this analysis is shown in Figure 1D. In contrast
to 21, deprotonation of trans-23 will not face any steric
hindrance between the base and a bulky ether protecting
group because of the bicylic structure. Thus, the evidence
is strong that decreased unfavorable steric interactions
give trans-23 a kinetic acidity sufficient to allow electro-
philic difluorination.
In summary, it has been demonstrated that the 3,5-
cis-substituted monofluorinated pyrrolidinones used in
this work (26, cis-23) can be deprotonated, and the
resulting lactam enolates will react with NFSi leading
to difluorination. In the case of 3,5-trans-substituted
monofluorinated pyrrolidinones, the bicyclic trans-23 can
also be deprotonated and fluorinated a second time but
monocyclic substrates such as 21 cannot. Because mono-
fluorinated lactams with a trans relationship between
substituents are the major products produced from
electrophilic monofluorination of substrates such as 9,
these monocyclic compounds do not allow for successful
electrophilic difluorination. Only the bicylic 12 provides
access to the necessary difluorinated lactam intermediate
24 enroute to (2S)-4,4-difluoroglutamic acid.
This work has demonstrated that electrophilic di-
fluorination and diastereoselective electrophilic mono-
fluorination of pyroglutamic acid derivatives can be used
to synthesize (2S)-4,4-difluoroglutamic acid and (2S,4R)-
4-fluoroglutamic acid. These new synthetic routes utilize
only readily available reagents and avoid a problematic
RuO4 oxidation while providing products which are
stereochemically pure. Ironically, in the process of de-
veloping a solution to this long-standing oxidation prob-
lem, other oxidation difficulties were encountered in the
conversion of an alcohol to a carboxylic acid within a
small polar molecule. This problem was ultimately
solved, and the methods and process used should be of
interest to others synthesizing unnatural amino acids
and encountering similar challenges. Finally, given the
utility of pyroglutamic acid derivatives, this chemistry
and the various intermediates involved should be of
interest for many applications in organofluorine chem-
istry beyond the amino acid syntheses reported here.
(S)-1-(ter t-Bu t yloxyca r b on yl)-5-(t r it yloxym et h yl)-2-
p yr r olid in on e (9). Compound 7 (2.7 g, 7.6 mmol) was
dissolved in CH2Cl2 (35 mL), and to the stirred solution were
added Boc2O (3.3 g, 15.2 mmol, 2 equiv), DMAP (1.0 g, 8.4
mmol, 1.1 equiv), and Et3N (1.0 g, 9.9 mmol, 1.3 equiv). The
solution was stirred at rt for 3 d. The reaction was diluted
with CH2Cl2 (30 mL) and washed with water and then brine.
The resulting organic layer was dried over Na2SO4, filtered,
and concentrated in vacuo to give a dark orange oil which was
purified by silica gel column chromatography (hexanes/EtOAc,
4:1) to give the desired product as a pale yellow solid (3.79 g,
8.3 mmol, 94% yield (two steps from 6)). Rf ) 0.43 (hexanes/
EtOAc, 3:2); mp 113-115 °C (lit. 118-119 °C);40 [R]23D ) -32.3
(c 1.05, MeOH); 1H NMR (300 MHz, CDCl3) δ 1.44 (s, 9H),
1.94-2.11 (m, 2H), 2.44 (m, 1H), 2.78 (m, 1H), 3.10 (dd, 1H, J
Exp er im en ta l Section
Gen er a l Meth od s. Thin-layer chromatography (TLC) was
performed with standard Sigma-Aldrich brand silica gel 60
F-254 plates. Column chromatography was performed with
silica gel 60 (230-400 mesh). Melting points were obtained
on a Thomas-Hoover Mel-Temp apparatus and are uncor-
rected. 1H NMR, 13C NMR, and 19F NMR spectra were recorded
on Bruker AVANCE DRX300 and DRX500 spectrometers
using the X-WinNMR software. Chemical shifts are reported
in parts per million (ppm) upfield or downfield from tetra-
methylsilane (internal standard for 1H and 13C) or trifluoro-
acetic acid (external standard for 19F). Infrared spectra were
obtained on a Nicolet 5-DX spectrometer. Mass spectra were
obtained with a VG 70-250-S mass spectrometer made by
Micromass (UK) and an Opus data system. X-ray crystal
(54) Pickering, L.; Malhi, B. S.; Coe, P. L.; Walker, R. T. Nucleosides
Nucleotides 1994, 13, 1493.
(55) Ikota, N. Chem. Pharm. Bull. 1992, 40, 1925.
(56) Olson, G. L.; Cheung, H.-C.; Chiang, E.; Madison, V. S.;
Sepinwall, J .; Vincent, G. P.; Winokur, A.; Gary, K. A. J . Med. Chem.
1995, 38, 2866.

8840 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
) 9.4, 2.7 Hz), 3.48 (dd, 1H, J ) 9.4, 4.5 Hz), 4.26 (m, 1H),
7.29 (m, 15H); 13C NMR (75.5 MHz, CDCl3) δ 21.8, 28.4, 32.6,
58.0, 64.5, 83.1, 87.3, 127.6, 128.3, 129.0, 144.0, 150.1, 175.5;
FAB-MS (3-nba with Na+) m/e (rel intensity) 479.9 (MNa+,
100.0), 380.0 (100.0), 165.0 (42.8), 136.0 (38.0), 106.0 (35.7);
FAB-HRMS (3-nba with Na+) m/e calcd for C29H31NO4Na
(MNa+) 480.2151, found 480.2146.
(S)-1-(P h en ylm et h yl)-5-(t r iisop r op ylsiloxym et h yl)-2-
p yr r olid in on e (16). Synthesized via the general procedure
and purified by eluting with hexanes/EtOAc (3:2) to give the
desired product as a colorless oil (90% yield). Rf ) 0.33
1
(hexanes/EtOAc, 4:1); H NMR (CDCl3, 300 MHz) δ 1.05 (m,
21H), 2.02 (m, 2H), 2.30 (m, 1H), 2.54 (m, 1H), 3.53 (m, 1H),
3.66 (m, 1H), 3.77 (m, 1H), 4.04 (d, 1H, J ) 15.0 Hz), 5.06 (d,
1H, J ) 15.0 Hz), 7.30 (m, 5H); 13C NMR (CDCl3, 75.5 MHz)
δ 12.0, 18.1, 21.7, 30.6, 44.7, 58.5, 64.0, 127.6, 128.2, 128.8,
137.1, 175.8; CI-MS (NH3) m/e (rel intensity) 362.5 (MH+,
100.0), 170.2 (35.5), 154.2 (25.5); CI-HRMS (NH3) m/e calcd
for C21H36NO2Si (MH+) 362.2515, found 362.2517.
(S)-1-P h en ylm et h yl)-5-(t r it yloxym et h yl)-2-p yr r olid i-
n on e (10). Compound 7 (200 mg, 0.56 mmol) was dissolved
in THF and cooled to 0 °C. Dry NaH (18 mg, 0.73 mmol, 1.3
equiv) was slowly added. The mixture was stirred for 10 min
after complete addition of the NaH, and then benzyl bromide
(125 mg, 0.73 mmol, 1.3 equiv) was added. The flask was
warmed to rt and allowed to stir overnight. The reaction was
quenched with the slow addition of H2O, and the solvents were
removed in vacuo. The resulting residue was partitioned
between EtOAc and H2O. After separating layers, the aqueous
layer was extracted further with EtOAc. The combined organic
layers were dried (Na2SO4), filtered, and evaporated to give
an orange residue which was purified by silica gel column
chromatography (hexanes/EtOAc, 3:2) to give the desired
product as a colorless semisolid in quantitative yield. Rf ) 0.55
(hexanes/EtOAc, 3:2); 1H NMR (CDCl3, 500 MHz) δ 1.93-2.11
(m, 2H), 2.41 (m, 1H), 2.63 (m, 1H), 3.15 (dd, 2H, J ) 10.0,
3.7 Hz), 3.19 (dd, 1H, J ) 10.0, 4.2 Hz), 3.65 (d, 1H, J ) 15.0
Hz), 4.98 (d, 1H, J ) 15.0 Hz), 7.03-7.40 (m, 20H); 13C NMR
(CDCl3, 75.5 MHz) δ 22.0, 30.7, 44.5, 56.9, 63.4, 87.1, 127.4,
127.6, 128.1, 128.8, 136.8, 143.7, 175.7; CI-MS (NH3) m/e (rel
intensity) 448.3 (MH+, 100.0), 243.1 (22.4), 206.1 (34.7), 188.1
(12.8); CI-HRMS (NH3) m/e calcd for C31H30NO2 (MH+)
448.2276, found 448.2255.
(S)-5-(ter t-Bu tyld ip h en ylsiloxym eth yl)-1-(p h en ylm eth -
yl)-2-p yr r olid in on e (17). Synthesized via the general pro-
cedure and purified by eluting with hexanes/EtOAc (4:1) to
give the desired product as a colorless oil (93% yield). Rf ) 4.8
(hexanes/EtOAc, 4:1); 1H NMR (CDCl3, 300 MHz) δ 1.05 (s,
9H), 2.01 (m, 2H), 2.42 (m, 1H), 2.59 (m, 1H), 3.47-3.68 (m,
3H), 3.74 (d, 1H, J ) 15.0 Hz), 4.99 (d, 1H, J ) 15.0 Hz), 7.42
(m, 15H); 13C NMR (CDCl3, 75.5 MHz) δ 19.3, 21.6, 27.0, 30.7,
44.6, 58.2, 63.9, 127.6, 128.0, 128.8, 130.1, 130.2, 133.0, 135.8,
135.9, 136.9, 175.8; CI-MS (NH3) m/e (rel intensity) 444.6
(MH+, 100.0), 274.4 (16.3), 207.3 (22.2); CI-HRMS (NH3) m/e
calcd for C28H34NO2Si (MH+) 444.2359, found 444.2345.
(S)-5-(ter t-Bu tyld im eth ylsiloxym eth yl)-1-(4-m eth oxy-
p h en ylm eth yl)-2-p yr r olid in on e (18). Synthesized via the
general procedure and purified by eluting with hexanes/EtOAc
(1:1) to give the desired product as a colorless oil (93% yield).
Rf ) 0.38 (hexanes/EtOAc, 1:1); 1H NMR (CDCl3, 300 MHz) δ
0.03 (s, 6H), 0.88 (s, 9H), 1.90 (m, 2H), 2.36 (m, 1H), 2.50 (m,
1H), 3.49 (m, 2H), 3.67 (m, 1H), 3.79 (s, 3H), 3.96 (d, 1H, J )
14.8 Hz), 4.95 (d, 1H, J ) 14.8 Hz), 6.84 (d, 2H, J ) 8.7 Hz),
7.18 (d, 2H, J ) 8.7 Hz); 13C NMR (CDCl3, 75.5 MHz) δ -5.4,
18.4, 21.7, 26.0, 30.7, 44.1, 55.5, 58.3, 63.6, 114.1, 129.2, 129.5,
159.1, 175.7; CI-MS (NH3) m/e (rel intensity) 350.3 (MH+,
100.0), 230.2 (19.4), 136.1 (17.0), 121.1 (25.5); CI-HRMS (NH3)
m/e calcd for C19H32NO3Si (MH+) 350.2151, found 350.2140.
(S)-1-(ter t-Bu tyloxycar bon yl)-5-(m eth oxym eth yl)-2-pyr -
r olid in on e (20). THF (8 mL) was added to a flask containing
a glass stir bar which was then cooled to -78 °C in a dry ice-
iPrOH bath. NH3 (10 mL) was condensed into the flask and
the mixture continued to stir at -78 °C. A solution of
compound 19 (730 mg, 3.3 mmol) in THF (5 mL) was added
slowly in portions along with the periodic addition of sodium
so that a deep blue color persisted in the reaction mixture.
After complete addition of the starting material, the reaction
stirred for an additional 10 min before the remaining sodium
was slowly quenched with nBuOH. The flask was warmed to
rt and the NH3 evaporated. The remaining solvents were
removed in vacuo, and the resulting white residue was
dissolved in water. This aqueous solution was brought to pH
) 7.0 with 3.0 M HCl, and then it was repeatedly extracted
with CH2Cl2 until TLC analysis indicated that the extracts
no longer contained the product. The combined extracts were
dried (Na2SO4), filtered, and evaporated to give a light yellow
oil which was purified by silica gel column chromatography
(EtOAc/MeOH, 9:1) to give the desired deprotected amide as
a colorless oil (240 mg, 1.86 mmol, 56% yield). Rf ) 0.38
(EtOAc/MeOH, 9:1); 1H NMR (CDCl3, 300 MHz) δ 1.77 (m, 1H),
2.16-2.38 (m, 3H), 3.29 (dd, 1H, J ) 9.4, 7.4 Hz), 3.37 (s, 3H),
(5S )-2,2-Dim e t h yl-8-oxo-1-a za -3-oxa -b icyclo[3.3.0]-
octa n e (12). Compound 6 (10.4 g, 90.3 mmol) was dissolved
in 2,2-dimethoxypropane (DMP) (40 mL), and camphorsulfonic
acid (CSA) (500 mg, cat.) was added. The solution was refluxed
for 4 h, and then the volatile components (DMP, MeOH) were
removed in vacuo. Fresh DMP was added, and the mixture
was again refluxed for 4 h. The process of evaporation and
restarting was repeated a total of three times. After the final
evaporation, the remaining residue was dissolved in EtOAc
and washed with saturated aqueous NaHCO3, water, and then
brine. The organic layer was dried (Na2SO4), filtered, and
evaporated to give the desired product as a pale yellow oil (12.3
1
g, 79 mmol, 87% yield). Rf ) 0.52 (EtOAc); H NMR (CDCl3,
500 MHz) δ 1.47 (s, 3H), 1.67 (s, 3H), 1.76 (m, 1H), 2.17 (m,
1H), 2.53 (m, 1H), 2.81 (m, 1H), 3.45 (t, 1H, J ) 8.8 Hz), 4.08
(dd, 1H, J ) 8.1, 5.6 Hz), 4.26 (m, 1H); 13C NMR (CDCl3, 75.47
MHz) δ 23.9, 24.5, 27.0, 37.4, 61.8, 70.0, 91.4, 171.6; EI-MS
(70 ev) m/e (rel intensity) 155.1 (M+, 1.9), 140.1 (M+ - 15,
100.0), 98.1 (19.9), 84.0 (14.2), 70.1 (12.2); EI-HRMS (70 ev)
m/e calcd for C8H13NO2 (M+) 155.0946, found 155.0941.
Gen er a l P r oced u r e for t h e Sila t ion of N-Ben zyl-
P r otected 5-(Hyd r oxym eth yl)-2-p yr r olid in on es (15-18).
The free alcohol (13 or 14), imidazole (1.3 equiv), and DMAP
(0.1 equiv) were dissolved in CH2Cl2, and then TBDMS-Cl,
TIPS-Cl, or TBDPS-Cl (1.3 equiv) was added. The solutions
were stirred at rt for 30 min, and a white precipitate was
visible. The reaction was diluted with CH2Cl2 and H2O and
transferred to a separatory funnel. After separation of layers,
the remaining aqueous layer was extracted further with CH2-
Cl2. The combined organic layers were dried (Na2SO4), filtered,
and evaporated to give the crude product which was purified
by silica gel column chromatography.
3.40 (dd, 1H, J ) 9.4, 4.3 Hz), 3.83 (m, 1H), 7.10 (br, 1H); 13
C
NMR (CDCl3, 75.5 MHz) δ 23.1, 29.8, 53.8, 59.1, 76.2, 178.5;
EI-MS (70 eV) m/e (rel intensity) 129.1 (M+, 4.5), 84.1 (100.0),
(S)-5-(ter t-Bu tyld im eth ylsiloxym eth yl)-1-(p h en ylm eth -
yl)-2-p yr r olid in on e (15). Synthesized via the general pro-
cedure and purified by eluting with hexanes/EtOAc (1:1) to
give the desired product as a colorless oil in quantitative yield.
Rf ) 0.48 (hexanes/EtOAc, 1:1); 1H NMR (CDCl3, 300 MHz) δ
0.03 (s, 6H), 0.87 (s, 9H), 1.70-2.04 (m, 2H), 2.37-2.55 (m,
2H), 3.50 (m, 2H), 3.66 (m, 1H), 4.03 (d, 1H, J ) 15.0 Hz),
4.99 (d, 1H, J ) 15.0 Hz), 7.23-7.34 (m, 5H); 13C NMR (CDCl3,
75.5 MHz) δ -5.4, 18.4, 21.7, 26.0, 30.6, 44.7, 58.5, 63.6, 127.6,
128.2, 128.8, 137.1, 175.8; CI-MS (NH3) m/e (rel intensity)
320.3 (MH+, 100.0); CI-HRMS (NH3) m/e calcd for C18H30NO2-
Si (MH+) 320.2045, found 320.2040.
45.0 (10.5), 41.0 (31.05); EI-HRMS (70 ev) m/e calcd for C6H11-
NO2 (M+) 129.0789, found 129.0788. The intermediate second-
ary amide (200 mg, 1.55 mmol) was dissolved in CH2Cl2, and
DMAP (189 mg, 1.55 mmol, 1.0 equiv), Et3N (202 mg, 2.0
mmol, 1.3 equiv), and Boc2O (474 mg, 2.17 mmol, 1.4 equiv)
were added. The mixture stirred at rt for 24 h, and then the
solution was diluted with CH2Cl2 and washed with brine. The
organic layer was dried (Na2SO4), filtered, and evaporated. The
resulting crude residue was purified by silica gel column
chromatography (hexanes/EtOAc, 3.5:1.5), and the desired
product was obtained as a colorless oil (350 mg, 1.52 mmol,
98% yield, 55% yield from 19). Rf ) 0.38 (hexanes/EtOAc, 3:2);

Fluorination of Pyroglutamic Acid Derivatives
J . Org. Chem., Vol. 66, No. 26, 2001 8841
1H NMR (CDCl3, 300 MHz) δ 1.54 (s, 9H), 2.07 (m, 2H), 2.39
(m, 1H), 2.69 (m, 1H), 3.35 (s, 3H), 3.50 (dd, 1H, J ) 9.6, 3.1
Hz), 3.57 (dd, 1H, J ) 9.6, 5.0 Hz), 4.25 (m, 1H); 13C NMR
(CDCl3, 75.5 MHz) δ 21.5, 28.2, 32.2, 57.4, 59.5, 73.6, 83.1,
150.1, 175.1; CI-MS (NH3) m/e (rel intensity) 230.0 (M + H+,
1.7), 147.0 (65.9), 130.0 (100.0); CI-HRMS (NH3) m/e calcd
for C11H20NO4 (MH+) 230.1392, found 230.1390.
Gen er a l Meth od for Electr op h ilic Mon oflu or in a tion
of La cta m s (Ta ble 1). A solution of LDA (1.3 equiv) in THF
(4-5 mL) was prepared and cooled to -78 °C. A solution of
the lactam substrate in THF (4-5 mL) was then added slowly.
The resulting mixture was stirred for 90 min at -78 °C and
then a solution of NFSi (1.4 equiv) in THF (4-5 mL) was
added. After 30 min, the reaction was quenched at -78 °C with
saturated aqueous NH4Cl and then allowed to warm slowly
to rt. The solvents were removed in vacuo, and the resulting
residue was partitioned between EtOAc and H2O. After
separation of the EtOAc layer, the aqueous layer was further
extracted with EtOAc. The combined organic layers were dried
(Na2SO4), filtered, and evaporated to give a residue which was
analyzed by NMR and purified by silica gel column chroma-
tography.
which was purified by silica gel column chromatography
(hexanes/EtOAc, 3:2) to give the desired product as a white
solid (189 mg, 0.40 mmol, 61% yield). Rf ) 0.42 (hexanes/
EtOAc, 1:1); mp ) 92-95 °C; IR (thin film) 3450 cm-1 (b, OH);
1H NMR (CHCl3, 500 MHz) δ 1.51 (9H, s), 2.06 (1H, m), 2.36
(1H, dd, J ) 12.6, 8.6 Hz), 2.73 (1H, d, J ) 1.9 Hz), 3.06 (1H,
dd, J ) 9.8, 2.4 Hz), 3.57 (1H, dd, J ) 9.8, 3.5), 4.23 (1H, m),
4.77 (1H, m), 7.22-7.46 (15H, m); 13C NMR (CDCl3, 75.5 MHz)
δ 28.1, 31.7, 54.9, 63.8, 70.0, 83.7, 87.4, 127.4, 128.2, 128.7,
143.5, 149.5, 175.7; FAB-MS (na with Na+) m/e (rel intensity)
496.1 (MNa+, 24.7), 396.1 (34.6), 243.1 (100.0); FAB-HRMS
(na with Na+) m/e calcd for C29H31NO5Na (MNa+) 496.2100,
found 496.2104.
(3S,5S)-1-(ter t-Bu tyloxyca r bon yl)-3-flu or o-5-(tr ityloxy-
m eth yl)-2-p yr r olid in on e (26). A solution of alcohol 25 (200
mg, 0.42 mmol) and pyridine (99.7 mg, 1.26 mmol, 3.0 equiv)
in CH2Cl2 was cooled to 0 °C, and triflic anhydride (142 mg,
0.50 mmol, 1.2 equiv) was added dropwise. After 20 min the
reaction was diluted with CH2Cl2 and poured into half-
saturated aqueous NaCl. The organic layer was washed (2×)
with saturated aqueous CuSO4, dried (Na2SO4), and filtered.
The solvent was removed in vacuo to give the desired inter-
mediate triflate in quantitative yield as a pale yellow solid
which was used without further purification.57 1H NMR
(CHCl3, 300 MHz) δ 1.52 (s, 9H), 2.28-2.43 (m, 2H), 3.04 (dd,
1H, J ) 10.1, 2.0 Hz), 3.74 (dd, 1H, J ) 10.1, 2.1 Hz), 4.30
(dd, 1H, J ) 8.4, 1.7 Hz), 5.95 (t, 1H, J ) 9.4 Hz), 7.22-7.37
(m, 15H); 19F NMR (CDCl3, 282.38 MHz) δ 1.34 (s). The triflate
(100 mg, 0.16 mmol) was dissolved in THF and cooled to -78
°C. A commercial solution of TBAF (1.0 M in THF) (320 µL,
0.32 mmol, 2.0 equiv) was added dropwise. After 15 min,
methyl iodide (45.4 mg, 0.32 mmol, 2.0 equiv) was also added.
After 30 min, the solution was allowed to warm gradually to
rt and stir for 24 h. The resulting solution was diluted (2:1)
with EtOAc and half-saturated aqueous NaHCO3 was added
dropwise with vigorous stirring. After separation of layers, the
aqueous layer was extracted with EtOAc. The combined
organic layers were dried (Na2SO4), filtered, and condensed
in vacuo to give the crude product as an orange residue. The
desired product was purified by silica gel column chromatog-
raphy (hexanes/EtOAc, 3:2) and obtained as a white solid (55
mg, 0.12 mmol, 72% yield from the triflate). The sample for
elemental analysis was obtained by crystallization in EtOH.
(3R,5S)-1-(ter t-Bu tyloxyca r bon yl)-3-flu or o-5-(tr ityloxy-
m eth yl)-2-p yr r olid in on e (21). Diisopropylamine (1.59 g,
15.73 mmol, 1.8 equiv) was added to THF (100 mL), and the
mixture was cooled to -78 °C in a dry ice-iPrOH bath. A
solution of nBuLi (895 mg, 8.73 mL, 13.98 mmol, 1.6 equiv)
(1.6M) was slowly added, and the resulting solution was stirred
for 1 h. A solution of 9 (4.0 g, 8.74 mmol) in THF (7 mL) was
added slowly. The resulting light yellow solution was stirred
at -78 °C for 45 min and then at -55 °C for 5 min. A solution
of NFSi (4.13 g, 13.11 mmol, 1.5 equiv) in THF (15 mL) was
then added, and the reaction was allowed to stir at -55 °C for
35 min. The reaction was quenched with saturated aqueous
NH4Cl, and the flask was warmed to rt. The THF was removed
in vacuo, and the resulting residue was partitioned between
EtOAc and H2O. After separating layers, the aqueous layer
was extracted further with EtOAc. The combined organic
layers were dried (Na2SO4), filtered, and evaporated to leave
an orange residue which was purified by two sequential silica
gel columns: #1 (hexanes/EtOAc, 4:1), #2 (CHCl3/EtOAc, 19:
1) to give the desired product as a white solid (2.4 g, 5.0 mmol,
57% yield). The samples for elemental analysis and X-ray
crystal structure determination were obtained by crystalliza-
tion in EtOH. Rf ) 0.40 (hexanes/EtOAc, 4:1); mp 130-134
°C; [R]23D ) -1.63 (c 0.61, MeOH); 1H NMR (CDCl3, 300 MHz)
δ 1.50 (s, 9H), 2.32 (m, 2H), 2.97 (dd, 1H, J ) 9.9, 2.1 Hz),
3.66 (dt, 1H, J ) 9.9, 2.6, 2.2 Hz), 4.27 (d, 1H, J ) 9.1 Hz),
5.56 (dt, 1H, J ) 53.1, 8.9 Hz), 7.29 (m, 15H); 13C NMR (CDCl3,
Rf ) 0.46 (hexanes/EtOAc, 3:2); mp ) 104-106 °C; [R]23
)
D
1
-75.8 (c 0.55, MeOH); H NMR (CHCl3, 300 MHz) δ 1.42 (s,
9H), 2.30-2.48 (m, 1H), 3.34 (d, 2H, J ) 4.9 Hz), 4.24 (m, 1H),
4.98 (ddd, 1H, J ) 51.8, 7.8, 4.1 Hz), 7.21-7.43 (m, 15H); 13
C
NMR (CDCl3, 75.5 MHz) δ 27.8, 28.1, 55.4, 64.0, 84.1, 87.0,
88.5 (d, 1J C-F ) 185 Hz), 127.4, 128.1, 128.8, 143.7, 149.4, 168.8
2
(d, J C-F ) 20.1 Hz); 19F NMR (CDCl3, 282.38 MHz) δ -108.0
2
75.5 MHz) δ 28.2, 30.7 (d, J C-F ) 18.9 Hz), 54.5, 63.6, 84.1,
(m, J ) 52, 27, 3.0 Hz); FAB-MS (NBA with Na+) m/e (rel
intensity) 498.3 (MNa+, 28.21), 398.2 (29.0), 243.2 (100.0),
154.1 (19.5), 136.1 (12.7); FAB-HRMS (NBA with Na+) m/e
calcd for C29H30FNO4Na (MNa+) 498.2056, found 498.2054.
Anal. Calcd for C29H30FNO4: C, 73.24; H, 6.36; N, 2.95.
Found: C, 73.07; H, 6.28; N, 2.95.
86.9, 88.2 (d, 1J C-F ) 187 Hz), 127.6, 128.3, 128.6, 143.4, 149.5,
2
169.7 (d, J C-F ) 20.8 Hz); 19F NMR (CDCl3, 282.38 MHz) δ
-110.6 (dd, J ) 53.6, 25.4 Hz); FAB-MS (3-nba with Na+) m/e
(rel intensity) 498.0 (MNa+, 61.6), 398.0 (69.5), 243.0 (100.0),
136.0 (25.6); FAB-HRMS (3-nba with Na+) m/e calcd for
C
29H30FNO4Na (MNa+) 498.2056, found 498.2054. Anal. Calcd
(3R,5S)-3-Flu or o-5-h ydr oxym eth yl-2-pyr r olidin on e (27).
Dep r otection Meth od A. Compound 21 (2.73 g, 5.74 mmol)
was stirred at rt in 3% concentrated HCl in MeOH for 48 h.
The pH of the solution was adjusted to 7.0 with saturated
aqueous NaHCO3. The now cloudy solution was condensed in
vacuo to give a white residue. This residue was triturated with
CH2Cl2/MeOH (5:1), and the resulting suspension was filtered
to remove the insoluble solids. Evaporation of the filtrate in
vacuo left another white solid residue which was purified by
silica gel column chromatography (CH2Cl2/MeOH, 5:1) to give
the desired product as a white solid (554 mg, 4.16 mmol, 73%
for C29H30FNO4: C, 73.24; H, 6.36; N, 2.95. Found: C, 72.84;
H, 6.27; N, 2.84.
(3R,5S)-1-(ter t-Bu tyloxyca r bon yl)-3-h yd r oxy-5-(tr ityl-
oxym eth yl)-2-p yr r olid in on e (25). 1,1,1,3,3,3-Hexamethyl-
disilazane (314 mg, 1.95 mmol, 3.0 equiv) was added to THF
(4 mL), and the solution was cooled to -78 °C in a dry ice-
iPrOH bath. A solution of nBuLi (115 mg, 1.12 mL, 1.8 mmol,
2.8 equiv) (1.6 M) was added dropwise and the mixture stirred
for 35 min. A solution of compound 9 in THF (2 mL) was added
dropwise over 5 min. The mixture stirred for 30 min at -78
°C and then 5 min at -55 °C before solid MoOPH (425 mg,
0.98 mmol, 1.5 equiv) was added in one portion. The bright
yellow reaction mixture was stirred for 30 min before it was
quenched with saturated aqueous NH4Cl. The flask was
warmed to rt, and the THF was removed in vacuo. The
resulting residue was partitioned between EtOAc and H2O.
After separating layers, the aqueous layer was extracted
further with EtOAc. The combined organic layers were dried
(Na2SO4), filtered, and concentrated in vacuo to leave a residue
yield). Rf ) 0.51 (CH2Cl2/MeOH, 5:1); mp 111-113 °C; [R]23
D
) +94.1 (c 1.01, MeOH); 1H NMR (MeOH-d4, 300 MHz) δ 2.20
(m, 2H), 3.47 (m, 2H), 3.75 (m, 1H), 5.12 (dt, 1H, J ) 53.8, 6.5
Hz); 13C NMR (MeOH-d4, 75.5 MHz) δ 32.5 (d, J C-F ) 20.7
2
1
2
Hz), 54.9, 65.2, 90.1 (d, J C-F ) 180.7 Hz), 175.5 (d, J C-F
)
(57) The triflate is unstable at room temperature and should be used
immediately or stored in a freezer (<0 °C).

8842 J . Org. Chem., Vol. 66, No. 26, 2001
Konas and Coward
20.1 Hz); 19F NMR (MeOH-d4, 282.38 MHz) δ -113.22 (ddd, J
) 54, 29, 13 Hz); DCI-MS (NH3) m/e (rel intensity) 151.1 (82.4),
134.1 (MH+, 100.0) 114.1 (5.2); DCI-HRMS (NH3) m/e calcd
for C5H9FNO2 (MH+) 134.0617 found 134.0612. Anal. Calcd
for C5H8FNO2: C, 45.11; H, 6.06; N, 10.52. Found: C, 44.70;
H, 6.02; N, 10.36.
MeOH, 5:1) indicated the complete loss of starting material
and the formation of a new higher Rf product. Any remaining
CH2N2 was destroyed with a solution of acetic acid in Et2O.
The flask was warmed to rt and the solvents were removed in
vacuo. The crude product was obtained by triturating the
residue with EtOAc, filtering the insoluble salts, and condens-
ing the filtrate in vacuo. Finally, the desired ester product was
purified by silica gel column chromatography (EtOAc) to give
28 as a colorless oil with yields of approximately 70% in two
Dep r otection Meth od B. Compound 21 (1.0 g, 2.10 mmol)
was dissolved in CH2Cl2, and the solution was cooled to 0 °C.
Triethylsilane (1.1 g, 9.45 mmol, 4.5 equiv) was added followed
by the dropwise addition of anhydrous TFA (2.4 g, 21 mmol,
10 equiv). The resulting solution stirred at 0 °C for 2 h. The
solvent and volatile components were removed in vacuo to give
a white solid residue. Silica gel column chromatography (CH2-
Cl2/MeOH, 5:1) of this crude material and concentration of the
fractions containing the desired product (Rf ) 0.51) gave a
white solid which contained the desired product plus a minor
inseparable impurity with a total mass corresponding to
greater than 100% yield. This material exhibited the following
NMR signals in addition to those reported for pure 27 in
method A: 1H NMR (DMSO-d6, 300 MHz) δ 4.96 (m, D2O
exchange); 19F NMR (DMSO-d6, 282.38 MHz) δ -1.22 (s). This
material can be carried on for oxidation and esterification
without further purification.
(2S,4R)-4-F lu or o-5-oxo-p yr r olid in e-2-ca r boxylic Acid
Meth yl Ester (28). Oxid a tion Meth od A. Compound 27 (217
mg, 1.63 mmol) was dissolved in acetone (23 mL) and cooled
to 0 °C in an ice-water bath before 5% aqueous NaHCO3 (5
mL), NaBr (34 mg, 0.33 mmol, 0.2 equiv), and TEMPO (25
mg, 0.16 mmol, 0.1 equiv) were added with vigorous magnetic
stirring. A solution of household bleach (365 mg, 4.9 mmol,
3.0 equiv NaOCl) was slowly added dropwise over 2 h to the
white nonhomogeneous mixture. After complete addition, the
mixture stirred for an additional 1 h. The remaining excess
oxidant was quenched with the addition of iPrOH and stirring
for 15 min. The mixture was acidified to pH ) 3-4 using 1 M
aqueous KHSO4 and testing with pH indicator paper. The flask
was warmed to rt, and all solvents were removed in vacuo to
give a crude white solid residue.
Oxid a tion Meth od B. Compound 27 (30 mg, 0.23 mmol)
was dissolved in wet acetonitrile (1.4 mL, 0.75 vol % H2O) and
cooled to 0 °C in an ice-water bath. A stock solution containing
H5IO6 (128 mg, 0.56 mmol, 2.5 equiv) and CrO3 (0.24 mg, 2.42
µmol, 1.1 mol %) in wet acetonitrile (1.4 mL, 0.75 vol. % H2O)
was slowly added dropwise over 40 min. After complete
addition the mixture was stirred for an additional 45 min
before being quenched with aqueous phosphate buffer (pre-
pared by dissolving 600 mg of Na2HPO4 in 10 mL of H2O).
The solvents were removed in vacuo to give a crude solid
residue.
Ester ifica tion of th e Cr u d e Ca r boxylic Acid . Methanol
was added to the solid residues obtained from either oxidation
method which contained the crude carboxylic acid and inor-
ganic salts. After the mixture was stirred vigorously to dissolve
the desired carboxylic acid product, the insoluble solids were
filtered, and the resulting filtrate was stirred and cooled to 0
°C in an ice-water bath. This solution was treated dropwise
with portions of CH2N2 in Et2O58 until TLC analysis (CH2Cl2/
steps from 27 utilizing either oxidation method. Rf ) 0.5
1
(EtOAc); [R]23 ) +94.6 (c 1.0, MeOH); H NMR (CDCl3, 300
D
MHz) δ 2.54-2.68 (m, 2H), 3.79 (s, 3H), 4.39 (m, 1H), 5.14
(ddd, 1H, J ) 52.5, 7.3, 6.3 Hz), 7.62 (br, 1H); 13C NMR (CDCl3,
75.5 MHz) δ 32.2 (d, 2J C-F ) 21.6 Hz), 52.9, 53.1, 87.3 (d, 1J C-F
2
) 184.4 Hz), 171.7, 172.6 (d, J H-F ) 20.0 Hz); 19F NMR
(CDCl3, 282.38 MHz) δ -116.01 (m); CI-MS (NH3) m/e (rel
intensity) 179.1 ([M + NH4]+, 100.0), 162.1 (MH+, 30.2); CI-
HRMS (NH3) m/e calcd for C6H9FNO3 (MH+) 162.0566 found
162.0573. Anal. Calcd for C6H8FNO3‚0.25 H2O: C, 43.51; H,
5.17; N, 8.46. Found C, 43.69; H, 5.10; N, 8.41.
(2S,4R)-4-F lu or oglu ta m ic Acid (29). Compound 28 (171
mg, 1.06 mmol) was dissolved in 6 N aqueous HCl and refluxed
for 4 h. The solution was cooled and evaporated to dryness in
vacuo. The resulting solid was dissolved in distilled H2O and
again evaporated to dryness (2×) to give the crude fluoro-
glutamate hydrochloride salt. The salt was dissolved in a
minimum of 95% EtOH and cooled to 0 °C with magnetic
stirring in an ice-water bath. An excess of propylene oxide
was added dropwise to the solution, and a precipitate formed.
After 20 min all the volatile components were removed in
vacuo. The resulting tan solid was crystallized from H2O/
EtOH, washed with cold Et2O, and dried to give the neutral-
ized fluoroglutamate product (104 mg, 0.63 mmol, 60% yield).
mp ) 173-175 °C (lit. mp ) 182-183 °C dec);10 [R]23D ) +36.8
(c 0.90, 1N HCl) (lit. [R]25D ) +33.17 (c 1, 1N HCl));10 1H NMR
(D2O, 300 MHz) δ 2.31-2.42 (m, 1H), 2.50-2.65 (m, 1H), 4.13
(dd, 1H, J ) 6.8, 6.1 Hz), 5.17 (ddd, 1H, J ) 50.2, 10.3, 2.8
2
Hz); 13C NMR (D2O, 75.5 MHz) δ 33.1 (d, J C-F ) 20.6 Hz),
51.3, 87.93 (d, 1J C-F ) 180.5 Hz), 172.1, 174.3 (d, 2J C-F ) 21.5
Hz); 19F NMR (D2O, 282.38 MHz) δ -110.14 (ddd, J ) 51, 37,
15 Hz); DCI-MS (NH3) m/e (rel intensity) 165.1 ([M + NH4
-
H2O]+, 100.0); DCI-HRMS (NH3) m/e calcd for C5H10FN2O3
([M + NH4 - H2O]+) 165.0675 found 165.0678. Anal. Calcd
for C5H8FNO4: C, 36.37; H, 4.88; N, 8.48. Found: C, 35.99;
H, 5.12; N, 8.51.
Ack n ow led gm en t. This research was supported by
a grant from the National Cancer Institute (CA28097).
We would like to thank Dr. George Shia and Honeywell,
Inc. for a gift of N-fluorobenzenesulfonimide, Dr. Adam
Matzger for a helpful discussion regarding molecular
modeling, and Ms. J oyce Reese for assistance in the
preparation of this manuscript.
Su p p or tin g In for m a tion Ava ila ble: ORTEP plot of
compound 21 and X-ray crystallographic data. This informa-
tion is available free of charge on the Internet at http://
pubs.acs.org.
(58) Ethereal solutions of diazomethane were carefully prepared in
a fume hood using Diazald and a procedure based on those outlined
in Aldrich Technical Bulletin AL-180.
J O0106804