Assay for the enantiomeric analysis of [2H1]-fluoroacetic acid: Insight into the stereochemical course of fluorination during fluorometabolite biosynthesis in Streptomyces cattleya

Article abstract of DOI:10.1021/ja026654k

A sensitive method for the configurational analysis of (R)- and (S)-[²H₁]-fluoroacetate has been developed using ²H{¹H}-NMR in a chiral liquid crystalline solvent. This has enabled biosynthetic experiments to be

Full text of DOI:10.1021/ja026654k

Published on Web 12/13/2002
Assay for the Enantiomeric Analysis of [2H1]-Fluoroacetic
Acid: Insight into the Stereochemical Course of Fluorination
during Fluorometabolite Biosynthesis in Streptomyces
cattleya
David O’Hagan,*,† Rebecca J. M. Goss,‡ Abdelkrim Meddour,§ and
Jacques Courtieu*,§
Contribution from the School of Chemistry, UniVersity of St Andrews, Centre of Biomolecular
Sciences, North Haugh, St Andrews, United Kingdom, KY15 9EA, Department of Chemistry,
UniVersity of Durham, Science Laboratories, South Road, Durham, United Kingdom, DH1 3LE,
and Laboratoire de Chimie Structurale Organique, I.C.M.O., URA-CNRS 1384,
Bat. 410 Orsay Cedex, France
Received April 24, 2002 ; E-mail: do1@st-andrews.ac.uk
Abstract: A sensitive method for the configurational analysis of (R)- and (S)-[2H1]-fluoroacetate has been
developed using 2H{1H}-NMR in a chiral liquid crystalline solvent. This has enabled biosynthetic experiments
to be conducted which reveal stereochemical details on biological fluorination occurring during the
biosynthesis of fluoroacetate and 4-fluorothreonine in the bacterium Streptomyces cattleya. In particular,
feeding experiments to S. cattleya with isotopically labeled (1R, 2R)- and (1S, 2R)-[1-2H1]-glycerol 3d and
3e and [2,3-2H4]-succinate 4a gave rise to samples of enantiomerically enriched [2-2H1]-fluoroacetates 1a.
The predominant enantiomer resulting from each experiment suggests that the stereochemical course of
biological fluorination takes place with an overall retention of configuration between a glycolytic intermediate
and fluoroacetate 1. Consequently, this outcome suggests that the stereochemical course of the recently
identified fluorinase enzyme which mediates a reaction between fluoride ion and S-adenosyl-L-methionine
(SAM), occurs with an inversion of configuration.
Introduction
amino acid 4-fluorothreonine 2. Some recent insights into the
biosynthesis of fluoroacetate 1 and 4-fluorothreonine 2, in S.
Fluorinated metabolites are an extremely rare group of
secondary metabolites and the details of biological fluorination
are unclear.1-3
cattleya have been uncovered.7-13 For example a biosynthetic
feeding experiment with [2-13C]-glycerol 3a10 resulted in
isotopic incorporation into C-1 of fluoroacetate 1. Also isotopic
labeling experiments with (2R) and (2S) [2H2]-glycerols 3b and
3c have shown11 that the fluoromethyl groups of fluoroacetate
1 and 4-fluorothreonine 2 are labeled exclusively from the pro-R
hydroxymethyl group of glycerol 3 as shown in Scheme 1.
The experiment revealed that both deuterium atoms from
(2R)-[2H2]-glycerol 3b had become incorporated into the fluo-
romethyl groups of 1 and 2. The pro-R hydroxymethyl group
of glycerol 3 is activated in vivo by glycerol kinase to generate
(2R)-glycerol-3-phosphate prior to entry into the glycolytic
pathway. Accordingly, the hydroxymethyl group of glycerol 3
that is phosphorylated during glycerol activation is also that
Fluoroacetate 1, is the most widely distributed fluorometabo-
lite found as a toxin in many tropical and sub-tropical plants.4,5
Fluoroacetate 1 is also a metabolite of the bacterium Strepto-
myces cattleya6 where it is produced as a co-metabolite of the
† School of Chemistry, University of St Andrews, Centre of Biomolec-
ular Sciences.
(7) Hamilton, J. T. G.; Murphy, C. D.; Amin, M. R.; O’Hagan, D.; Harper, D.
B. J. Chem. Soc., Perkin Trans. 1. 1998, 759-767.
(8) Hamilton, J. T. G.; Amin, M. R.; Harper, D. B.; O’Hagan, D. Chem.
Commun. 1997, 797-798.
‡ Department of Chemistry, University of Durham, Science Laboratories.
§ Laboratoire de Chimie Structurale Organique, I.C.M.O., URA-CNRS.
(1) O’Hagan, D.; Harper, D. B. J. Fluorine Chem. 1999, 100, 127-133.
(2) O’Hagan, D.; Harper, D. B. Ramachardran, Asymmetric Fluoroorganic
Chemistry, Ed P. V.; ACS Symposium Series 746; 2000; pp 210-224.
(3) Harper, D. B.; O’Hagan, D. Nat. Prod. Rep. 1994, 11, 123-133.
(4) Twigg, L. E.; Wright, G. R.; Potts, M. D. Aust. J. Botany 1999, 47, 877-
880.
(9) Schaffrath, C.; Murphy, C. D.; Hamilton, J. T. G.; O’Hagan, D. J. Chem.
Soc., Perkin Trans. 1. 2001, 3100-3105.
(10) Tamura, T.; Wada, M.; Esaki, N.; Soda, K. J. Bacteriol. 1995, 177, 2265-
2269.
(11) Nieschalk, J.; Hamilton, J. T. G.; Murphy, C. D.; Harper, D. B.; O’Hagan
Chem. Commun. 1997, 799-800.
(5) O’Hagan, D.; Perry, R.; Lock, J. M.; Meyer, J. J. M.; Dasaradhi, L.;
Hamilton, J. T. G.; Harper, D. B. Phytochemistry 1993, 33, 1043-1046.
(6) Sanada, M.; Miyano, T.; Iwadare, S.; Williamson, J. M.; Arison, B. H.;
Smith, J. L.; Douglas, A. W.; Liesch, J. M.; Inamine, E. J. Antibiotics 1986,
39, 259-265.
(12) Schaffrath, C.; Cobb, S. L.; O’Hagan, D. Angew. Chem., Int. Ed. Engl.
2002, 41, 3913-3915.
(13) O’Hagan, D.; Schaffrath, C.; Cobb, S. L.; Hamilton, J. T. G.; Murphy, C.
D. Nature 2002, 416, 279.
9
10.1021/ja026654k CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 379-387
379

A R T I C L E S
O’Hagan et al.
Scheme 1
Scheme 2
which is fluorinated toward the end of the biosynthetic pathway.
Additionally, the observed retention of both deuteriums from
(2R)-[2H2]-glycerol 3b at the fluoromethyl groups of 1 and 2
indicates that there is no requirement for oxidation at that carbon
during metabolic conversion of the C-O to C-F bond during
fluorometabolite biosynthesis. Recently, the nature of the
fluorination enzyme in S. cattleya has been revealed13 and it
has been shown that inorganic fluoride reacts with S-adenosyl-
L-methionine (SAM) 6 to generate 5′-fluoro-5′-deoxyadenosine
(5′-FDA) 7 (Scheme 2).
SAM is of course derived from a reaction between ATP and
L-methionine mediated by methionine adenosyltransferase; thus,
glycerol becomes incorporated into the ribose ring of adenosine,
via glyceraldehyde-3-phosphate incorporation into sedoheptu-
lose-7-phosphate by transaldolase14 and then to ribose-5-
phosphate by the action of transketolase.15 Subsequent involve-
ment in ATP biosynthesis delivers the original pro-R hydroxy-
methyl group of glycerol to the triphosphate activated 5′-carbon
of ATP 5. Methionine adenosyltransferase mediates a nucleo-
philic attack of the sulfur of L-methionine to 5′-carbon center
of ATP to generate SAM 6 with inversion of configuration.16
The stereochemical course of the fluorinase reaction between
SAM and fluoride ion to generate 5′-FDA 7, remains to be
determined.
Fluoroacetaldehyde 8 has been identified12,17 as the last
common biosynthetic intermediate to both fluoroacetate 1 and
4-fluorothreonine 2. Resting cells of S. cattleya can efficiently
oxidize fluoroacetaldehyde 8 to fluoroacetate 1 and a feeding
experiment also in resting cells of S. cattleya with [1-2H1]-
fluoroacetaldehyde 8a resulted in a high incorporation (60%)
(15) Gunsalus, I. C.; Horecker, B. L.; Wood, W. A. Bacteriol. ReV. 1955, 19,
79-128.
(16) Parry, R. J.; Minta, M. J. Am. Chem. Soc. 1982, 104, 871-872.
(17) Moss, S. J.; Murphy, C. D.; Hamilton, J. T. G.; McRoberts, W. C.; O’Hagan,
D.; Schaffrath, C.; Harper, D. B. Chem. Commun. 2000, 2281-2282.
(14) Kamada, N.; Yasuhara, A.; Takano, Y.; Ikeda, M. Appl. Microbiol. Biotech.
2001, 56, 710-717.
9
380 J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003

Enantiomeric Analysis of [2H ]-Fluoroacetic Acid
A R T I C L E S
1
Scheme 3
of the deuterium atom into C-3 of 4-fluorothreonine 3, indicating
the direct incorporation of fluoroacetaldehyde 8 into C-3 and
C-4 of 4-fluorothreonine 2 as illustrated in Scheme 2. This
pivotal role for fluoroacetaldehyde 8 has been reinforced more
recently by the isolation of the two enzymes that convert
fluoroacetaldehyde 8 to fluoroacetate 118 and fluoroacetaldehyde
8 to 4-fluorothreonine 2.19 Also fluoroacetaldehyde has been
identified in vitro in incubations of 5′-FDA with a cell free
extract of S. cattleya.12 Thus the transformation of fluoro-
acetaldehyde 8 to fluoroacetate 1 and 4-fluorothreonine 2 is on
a secure biochemical basis. To explore the process of biological
fluorination in greater detail, it became pertinent to assess the
stereochemical course of the fluorination event. Clearly, an SN2
substitution reaction by fluoride ion on SAM 6 will generate
5′-FDA 7 with inversion of configuration to generate the
fluoromethyl group found in the fluorometabolites 1 and 2.
Alternatively, a sequence of events involving two SN2 substitu-
tion reactions perhaps involving a nucleophilic substituent on
the enzyme, would result in an overall retention of configuration.
Finally, there is the possibility of an elimination reaction
ocurring on SAM 6 to generate an enzyme bound substrate with
sp2 hybridization at C-5′. Fluoride addition could then occur to
one or other face of the double bond and this would result in
either an inversion or retention of configuration depending on
the stereochemical course of the reaction.
To trace the stereochemical fate the pro-R hydroxymethyl
carbon of glycerol during fluorometabolite biosynthesis, two
criteria had to be met. First appropriate deuterium labeled
precursors were required for feeding experiments to generate
samples of chiral [2H1]-fluoroacetate 1a in vivo. As (R)-[2H2]
glycerol 3b had already proven11 to be good substrates for
incorporation into the fluorometabolites 1 and 2, appropriately
labeled glycerols carrying single and stereochemically defined
deuterium atoms on the pro-R hydroxymethyl group, 3d and
3e, emerged as ideal candidate precursors for this stereochemical
investigation.
(1R, 2R)- and (1S, 2R)-[1-2H1]-glycerols 3d and 3e respectively,
for such feeding experiments were accessible from a previously
developed synthesis protocol.20 [2H4]-Succinate 4a also emerged
as an alternative biosynthetic precursor in which to explore the
stereochemical course of fluorination during fluorometabolite
biosynthesis. It has previously been shown7 that when [2H4]-
succinate 4a is administered to resting cells of S. cattleya the
resultant fluoroacetate 1 and 4-fluorothreonine 2 become labeled
with a single deuterium atom in each of the fluoromethyl groups.
This is rationalized by the processing of succinate 4 along
enzymes of the citric acid cycle via fumarate 10 and L-malate
11 to oxaloacetate 12. The enzyme phosphoenylpuruvate(PEP)
carboxykinase,21 converts oxaloacetate 12 to PEP 13 and then
enolase22 generates 2-phosphoglycerate 14, before further
processing to glyceraldehyde-3-phosphate 9 as shown in
Scheme 3.
The enzymatic transformations mediated by fumarase, PEP-
carboxykinase and enolase all impact on the absolute config-
uration at the ultimate fluoromethyl groups in 1 and 2. The
stereochemical course of all four of these enzymatic reactions21-23
is already established, and therefore, it can be deduced that the
(2R, 3S)-[3-2H1]-glyceraldehyde-3-phosphate 9a generated in
vivo from [2H4]-succinate 4a will have predominantly the (3R)
configuration at the isotopically labeled carbon as indicated in
Scheme 3. Accordingly, if the configuration of the resultant
fluoroacetate 1a can be determined then the stereochemical
relationship between (2R, 3R)-[3-2H1]-glyceraldehyde-3-phos-
phate 9a and the fluoromethyl carbon of [2H1]-fluoroacetate 1,
can be evaluated.
Second a method was required to assess the enantiomeric
enrichment of the resultant samples of chiral [2H1]-fluoroacetates
1a. Circular dichroism (CD) has been used to determine24 the
absolute configuration of [2H1]-fluoroacetate 1a. The method
is effective; however, such analysis requires a chemically pure
and highly isotopically enriched sample of either (R) or (S)-
[2H1]-fluoroacetate 1a. For this biosynthetic study, a method
of analysis was required that could reliably assay sub-milligram
quantities of biosynthetically derived fluoroacetate which
contained only a moderate level of isotopic enrichment (∼15%).
Chiral [2H1]-fluoroacetate 1a has been used to explore the
stereospecificity of the citrate synthase mediated condensation
of oxaloacetate and fluoroacetyl-CoA to generate 2-(2R, 3R)-
It was anticipated that the resultant fluoroacetates 1a would
retain a single, stereochemically defined deuterium atom after
feeding experiments with each of these substrates. The required
(20) Nieschalk, J.; O’Hagan, D. Tetrahedron Asymmetry 1997, 8, 2325-2330.
(21) Rose, I. A.; O’Connell, E. L.; Noce, P.; Utter, M. F.; Wood, H. G.; Willard,
J. M.; Cooper, T. G.; Benziman, M. J. Biol. Chem. 1969, 244, 6130-
6133.
(22) Cohn, M.; Pearson, J. E.; O’Connell, E. L.; Rose, I. A. J. Am. Chem. Soc.
1970, 92, 4095-4098.
(18) Murphy, C. D.; Moss, S. J.; O’Hagan, D. Appl. EnViron. Microbiol. 2001,
67, 4919-4921.
(19) Murphy, C. D.; O’Hagan, D.; Schaffrath, C. Angew Chem, Int. Ed. Engl.
2001, 40, 4479-4481.
(23) Gawron, O.; Glaid, A. J.; Fondy, T. P. J. Am. Chem. Soc. 1960, 83, 3634-
3640.
(24) Gartz, D.; Reed, J.; Re´tey, J. HelV. Chim. Acta. 1996, 79, 1021-1025.
9
J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003 381

A R T I C L E S
O’Hagan et al.
Scheme 4 a
a i. LiAl2H4, THF reflux, 6 h, 98%; ii. PCC, 3 Å mol sieves, CH2Cl2, 4 h, 25 °C, 81%; iii (-)Alpine Borane, THF reflux, 2 h, 66%; iv Et2NSF3, CH2Cl2,
-78 °C to r.t, 10 h, 50%; v. KMnO4/KIO4, H2O, 24 h, then H2SO4, lyophilize, dil. NaOH then freeze-dry, add unlabeled NaFAc; vi. SOCl2, C6H13OH,
0 °C to reflux, 2 h, 50%.
fluorocitrate.25-27 The enzyme abstracts the pro-S hydrogen of
fluoroacetyl-CoA with a high level of selectivity and delivers a
single stereoisomer of fluorocitrate. Citrate synthase offered a
potential enzymatic assay for this biosynthetic study; however,
in view of the small amounts of fluoroacetate that are generated
from the biosynthetic experiments and the requirement to
prepare samples of fluoroacetyl-CoA from that material, this
approach proved impractical and was unsuccessful in our hands.
3d and 3e and [2H4]-succinate 4a. The experiments gave rise
to samples of [2H1]-fluoroacetate 1a. Derivatization of 1a as
2
its hexyl ester 19a and then H{1H}-NMR analysis in a chiral
liquid crystalline medium (PBLG/chloroform) resulted in the
successful stereochemical analyses of the fluoromethyl group.
Results and Discussion
Enantiomeric Assay of [2H1]-Fluoroacetate for Biosyn-
thetic Experiments. To develop an assay for [2H1]-fluoroacetate
1a a suitable reference sample of known chirality was required.
The hexyl ester of fluoroacetate was prepared as shown in
Scheme 4.
A sample of [1-2H]-benzaldehyde 17a was generated by the
reduction of ethyl benzoate 15 to 16a with lithium aluminum-
[2H4]-hydride and then oxidation of 16a to 17a with pyridinium
chlorochromate. Asymmetric reduction of [1-2H]-benzaldehyde
17a using â-3-pinanyl-9-BBN ((S)-Alpine Borane)32 has previ-
ously been shown33 to give the (R)-[7-2H1]-benzyl alcohol 16b
as the predominant product. The enantiomeric excess of this
product was conveniently determined by chiral liquid crystal
2H{1H}-NMR28 (Figure 1a) and shown to be 88%ee.
In this paper, a novel method for the assay of [2H1]-
fluoroacetate 1a chirality is described using 2H NMR in a chiral
lyotropic liquid crystalline phase prepared using poly-γ-benzyl-
L-glutamate (PBLG).28-30 When oriented in such a mesophase
it has been shown that enantiotopic nuclei become non
equivalent as the order parameters along enantiotopic directions
2
are different. H{1H}-NMR emerges as a powerful analytical
tool for enantiomeric analysis in liquid crystalline medium.31
The sensitivity of the technique is maximal for deuterium labeled
compounds as the large quadrupolar interaction (150-200 kHz
2
for this nucleus) induces large quadrupolar splittings. The H-
{1H}-NMR spectrum of a racemic mixture of a monodeuterated
compound dissolved in PBLG/cosolvent exhibits four peaks
which are to be viewed as two sets of doublets. As the chemical
shift anisotropy of deuterium is small, the two doublets are
centered on the same frequency, the chemical shift of the
deuterium. The inner doublet is assigned to one enantiomer and
the outer one, to the other enantiomer. The quadrupolar splitting
is the difference (in Hz) between the two peaks of each doublet.
This splitting is extremely sensitive to temperature and to the
PBLG/solvent and the PBLG/solute ratios; however, although
the magnitude of the quadrupolar splittings can vary, there is
generally no peak inversion/crossover for the enantiomer
doublets. We have used this method to assay the chirality of
fluoroacetate from three biosynthetic feeding experiments with
S. cattleya cells, using (1R, 2R)- and (1S, 2R)-[1-2H1]-glycerols
Treatment of this alcohol with diethylaminosulfur trifluoride
(DAST)34 generated a sample of [7-2H1]-benzylfluoride 18a.
This compound, which now carries a chiral fluoromethyl group,
was assayed for its enantiomeric excess (38%ee) using chiral
liquid crystal 2H{1H}-NMR analysis. The enantiomers of
[7-2H1]-benzylfluoride 18a were readily resolved as shown in
Figure 1b. The spectrum contains an “extra” set of doublets
due to total D-F coupling (9.6 Hz). This established for the first
2
time that chiral liquid crystal H{1H}-NMR was sufficiently
discriminating to resolve a chiral fluoromethyl group. DAST
reactions on secondary alcohols are well-known to suffer from
a lack of stereochemical integrity,35 although the predominant
stereoisomer generally results from a configurational inversion.
In this case, stereochemical inversion will generate the (S)
isomer of 18a as the major product enantiomer. The loss in
stereochemical integrity can be attributed to a significant level
(25) Dummel, R. J.; Kun, E. J. Biol. Chem. 1969, 244, 2966-2969.
(26) Keck, R.; Haas, H.; Re´tey, J. FEBS Letts. 1980, 114, 287-290.
(27) Marletta, M. A.; Srere, P. A.; Walsh, C. Biochemistry 1981, 20, 3719-
3723.
(28) Meddour, A.; Canlet, C.; Blanco, L.; Courtieu, J. Angew. Chem., Int. Ed.
1999, 38, 2391-2393.
(32) Midland, M. M.; Tramontano, A.; Kazubski, A.; Graham, R. S.; Tsai, D.
J. S.; Cardin, D. Tetrahedron 1984, 40, 1371-1380.
(33) Midland, M. M.; Greer, S.; Tramontano, A.; Zedric, S. A. J. Am. Chem.
Soc. 1979, 101, 2352-2355.
(34) Middleton, W. J. J. Org. Chem. 1975, 40, 574-578.
(35) Madiot, V.; Lesot, P.; Gree, D.; Courtieu, J.; Gree, R. Chem. Comm. 2000,
169-170.
(29) Lesot, P.; Merlet, D.; Loewenstein, A.; Courtieu, J. Tetrahedron Asymmetry
1998, 9, 1871-1881.
(30) Canet, I.; Courtieu, J.; Loewenstein, A.; Meddour, A.; Pechine, J. M. J.
Am. Chem. Soc. 1995, 117, 6520-6526.
(31) Tavasli, M.; Courtieu, J.; Goss, R. J. M.; Meddour, A.; O’Hagan, D. Chem.
Commun. 2002, 844-845.
9
382 J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003

Enantiomeric Analysis of [2H ]-Fluoroacetic Acid
A R T I C L E S
1
Figure 2. 2H{1H}-NMR spectra of racemic hexyl [1-2H1]-fluoroacetate
19a in a chiral liquid crystalline medium of chloroform and PBLG showing
clear resolution of the enantiomers. The natural abundance signals from
chloroform are labeled with an asterisk.
including the solvent (natural abundance CDCl3 signals are
marked with an asterix). Each of the doublets associated with
the enantiomers is clearly resolved.
The corresponding 2H{1H}-NMR spectrum for the synthetic
sample prepared as Scheme 4 and enriched with the (S)- 19a is
shown in Figure 3a. Again the enantiomers of hexyl [2H1]-
fluoroacetate 19a are clearly resolved in the 2H{1H}-NMR
spectrum using this technique and indicate that the sample of
hexyl [2H1]-fluoroacetate has an enantiomeric excess of (S)-
38%ee. There was no further racemisation of the sample during
the permanganate/periodate oxidation and it is assumed that the
predominant enantiomer of 1a has the (S) configuration fol-
lowing through from (S)-[2H1]-benzylfluoride 18a (38%ee). The
ease of the 2H{1H}-NMR assay method and its ability to resolve
the enantiomers of hexyl [2-2H1]-fluoroacetate 19a in such a
straightforward manner now offered a practical method by which
to assay the enantiomeric purity of [2H1]-fluoroacetates 1a
derived from biosynthetic feeding experiments.
Figure 1. 2H{1H}-NMR spectra of (a); (R)-[1-2H1]-benzyl alcohol (88%ee)
16b and (b) (S)-[1-2H1]-benzylfluoride (38%ee) 18a, in a chiral liquid
crystalline medium of chloroform and poly-γ-benzyl-l-glutamate (PBLG).
In (b) the additional doublets arises from 19F-2H coupling (9.6 Hz).
of SN1 character during the DAST reaction, a feature that is
almost certainly accommodated by the benzylic nature of the
substrate. There was no improvement in the enantiomeric excess
when this reaction was carried out at -50 °C, although the
reaction was much slower. Fluorination using Ishikawa’s reagent
(N,N-diethyl-1,1,2,3,3,3-hexafluorylpropylamine)36 was also
explored, however the product [7-2H1]-benzylfluoride 18a,
which was subjected to the same stereochemical analysis, had
an identical enantiomeric enrichment of 38%ee also in favor of
(S) 18a. Thus, there was no advantage in using this alternative
fluorination reagent. Oxidation of the aromatic ring of [7-2H1]-
benzylfluoride 18a to generate a sample of [2H1]-fluoroacetate
1a, was accomplished using aqueous KMnO4/KIO4. The product
was secured after acidification and lyophilisation. The lyophi-
lisate was then neutralized with dil. NaOH and the neutral
solution freeze-dried to leave sodium [2H1]-fluoroacetate 1a as
a clean white amorphous residue. This material was diluted with
unlabeled sodium fluoroacetate 1 such that the 2H-isotope
enrichment was ∼1%. The increased mass provided by the
unlabeled “carrier” material allowed for easier chemical de-
rivatization. Dilution of the deuterium label in the sample was
not a significant limitation and demonstrates that this assay
method can be used very effectively for low levels of deuterium
enrichment because the enantiomeric assay is based on 2H{1H}-
NMR analysis of the isotopically labeled molecules only. Hexyl
[2-2H1]-fluoroacetate 19a was prepared by direct treatment of
sodium [2H1]-fluoroacetate 1a with thionyl chloride followed
by the addition of hexanol. The product 19a was purified by
chromatography prior to 2H{1H}-NMR analysis in a chloroform/
Synthesis and Stereochemical Analysis of (1R, 2R)- and
(1S, 2R)- [1-2H1]-Glycerols 3d and 3e. The synthesis of (1R,
2R)- and (1S, 2R)- [1-2H1]-glycerols 3d and 3e has previously
2
been described,20 however access to chiral liquid crystal H-
{1H}-NMR analysis of the product glycerols has provided a
more penetrating insight into the nature of the products formed
than that previously evaluated. The synthesis of the glycerols
relied on either the reduction of [3-2H1]-propargyl alcohol 20a
with LiAlH4 and then a H2O quench, or the reduction of
unlabeled propargyl alcohol 20 with LiAlH4 followed by a D2O
quench, as shown in Scheme 5.
The preparation of (1R, 2R)-[1-2H1]-glycerol 3d gives a
satisfactory product sample (>95%ee). In the resultant 2H{1H}-
NMR of glycerol 3d shown in Figure 4a, there is only one
observable deuterium labeled species, with none of the corre-
sponding (1S, 2R) diastereoisomer 3e apparent. This is consistent
with our earlier observation20 where reduction of [3-2H1]-
propargyl alcohol 20a with LiAlH4 generated a single product
21a, whereas reduction of propargyl alcohol followed by a D2O
quench (Scheme 5b) gave a more complex product mixture.
2
Accordingly in the case of (1S, 2R)-[1-2H1]-glycerol 3e, H-
{1H}-NMR analysis (Figure 4b) of the product mixture derived
from LiAlH4 reduction of propargyl alcohol 20 followed by a
D2O quench generated a predominant cis product 21b (78%)
and a minor trans product 21a (12%). 2H{1H}-NMR also
confirmed a significant level (10%) of allyl alcohol 21c with
isotope delivered to the C-2 position as shown in Scheme 5.
Compound 21c arises from hydride delivery to the terminal
acetylene carbon of propargyl alcohol 20, with deuterium
2
PBLG mesophase. Figure 2 shows the resultant H{1H}-NMR
spectrum of a separately prepared racemic sample to 19a. The
spectrum contains three sets of signals (doublets) which arise
from the quadrupolar splitting from three different species
(36) Takaoka, A.; Iwakiri, H.; Ishikawa, N. Bull. Chem. Soc., Jpn. 1979, 52,
3377-3380.
9
J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003 383

A R T I C L E S
O’Hagan et al.
Figure 3. 2H{1H}-NMR spectra of hexyl [1-2H1]-fluoroacetates 19a in a chiral liquid crystalline medium of chloroform and PBLG. (a) Synthetic (S)-19a
(38%ee) prepared as Scheme 4; (b) (R)-19a (75%ee) after a feeding experiment to S. cattleya cells with (1R, 2R)-glycerol 3d; (c) (S)-19a (30-40%ee) after
a feeding experiment to S. cattleya cells with (1S, 2R)-glycerol 3e; (d) (R)-19a (25%ee) after a feeding experiment to S. cattleya cells with [2H4]-succinate
4a: The natural abundance signals from chloroform are labeled with an asterisk.
delivered to the central carbon after the D2O quench. The
resultant [2H1]-allyl alcohol mixture 21a-c was then converted
to the corresponding benzoyl esters 22a-c followed by AD-
mix-â dihydroxylation.37 This process afforded the benzoate
diols 23a-c as a solid material. As previously described,
recrystallization of this benzoate product significantly improved
the stereoisomeric excess of the major (2R) alcohol over the
(2S) products. The labeled glycerols were then generated by
hydrolysis of the benzoate diols 23a-c, and were purified by
although undesirable in general terms, was judged not to present
a significant complication to our interpretation of the biosyn-
thetic experiment because the C-2 carbon atom of glycerol
becomes oxidized during the biosynthesis of fluoroacetate.
2
Accordingly, H-isotope at the C-2 position of glycerol 3f is
liberated in a biosynthetic experiment. However, the presence
of the two diastereoisomers 3e and 3d in a ratio of 93.5:6.5
indicated a stereochemical excess (de) of 87% at the pro-R
hydroxymethyl group of glycerol. It is not clear why the
stereochemical integrity of the hydroxymethyl group is lower
in this case, but the conclusion is consistent with the experi-
mental observation after several preparations. With the samples
of (1R, 2R)-[1-2H1]-glycerol (>95% de) 3d and (1S, 2R)-[1-
2H1]-glycerol of 3e (87% de) in hand, biosynthetic feeding
experiments to resting cells of S. cattleya were carried out and
the results are described below.
Feeding Experiments of (1R, 2R)- and (1S, 2R)-[1-2H1]-
glycerols 3d and 3e into Fluoroacetate 1 in Resting Cells of
S. cattleya. The isotopically labeled (1R, 2R)- and (1S, 2R)-[1-
2H1]-glycerols 3d and 3e were separately incubated with washed
resting cells of S. cattleya prepared from a batch culture which
had grown for 8 d.9 The washed cells were then incubated with
the glycerols 3d and 3e to a final concentration of 9mM. After
48 h, each experiment was worked up by freeze-drying the cell
supernatant, followed by acidification and lyophilization. The
2
silica gel chromatography. The chiral liquid crystal H{1H}-
NMR spectrum for the resultant (1S, 2R)-[1-2H1]-glycerols 3e
(3d and 3f) is shown in Figure 4b.
In Figure 4b, there are three clear doublets arising from three
deuterium labeled glycerols in the product mixture. The intensi-
ties of these signals correlate closely with those found in the
2H{1H}-NMR of the [2H1]-allyl alcohol intermediates, prepared
upstream of the glycerol products. Accordingly the signals are
assigned to (1S, 2R)-[1-2H1]-glycerol 3e (85.7%, 97 Hz), the
corresponding diastereoisomer (1R, 2R)-[1-2H1]-glycerol 3d
(6%, 133.5 Hz) and then the product of C-2 labeling, [2-2H1]-
glycerol 3f (8.3%) which has the largest quadrupolar splitting
(262.6 Hz). The presence of this latter glycerol in the sample,
(37) Sharpless, K. B.; Amberg, W.; Bennani, Y. L.; Crispino, G. A.; Hartung,
J.; Jeing, K.-S.; Kwong, H.-L.; Morikawa, K.; Wang, Z.-M.; Xu, D.;
Zhang, X.-L. J. Org. Chem. 1992, 57, 2768-2771.
9
384 J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003

Enantiomeric Analysis of [2H ]-Fluoroacetic Acid
A R T I C L E S
1
Scheme 5 a
2
a i. BuLi, THF, -78 °C; H2O quench, r.t. ii for (a). LiAlH4, THF, 0 °C; stir, r.t., 3 h; H2O quench, trans 100%. Ii for (b). LiAlH4, THF, 0 °C; stir, r.t.
2
3 h; H2O quench, cis 78%, trans 12%, CH2CH2HCH2OH 10%. iii. PhCOCl, 30%NaOH, Et3BnN+Cl-, DCM; for (a), steps ii-iii 32%; for (b), steps i-iii
38%. iv. AD-mix-â, t-BuOH/H2O 1:1, 0 °C, 4 h; crystallization Et2O, petroleum ether (40-60), for (a), 65%; for (b), 64%. v. NaOH (1 equiv), H2O, acetone,
for (a), 81%, 69%de; for (b), 78%, >95%de.
lustrated in Figure 3b. The quadrupolar splitting in Figure 3b
is larger than that observed in Figure 3a, 3c, and 3d as the PBLG
wt % in CHCl3 was higher (14.3%) in the sample recorded in
Figure 3b than those samples (11.3%) recorded in Figure 3a,
3c, and 3d. By comparison with the synthetic reference sample
(Figure 3a) it is concluded that this sample of 19 had a
predominant (R) configuration. The complementary feeding
experiment using (1S, 2R)-[1-2H1]-glycerol 3e gave rise to a
sample of hexyl [2-2H1]-fluoroacetate 19a with a lower enan-
tiomeric excess of 30-40%ee in favor of the (S) configuration
as illustrated in Figure 3c. Despite a difference in the calculated
%ee of the resultant hexyl [2H1]-fluoroacetates 19a and 19b
recovered from each experiment, the predominant absolute
stereochemistry was opposite in each case and thus the experi-
ments are mutually reinforcing. It can be concluded that the
replacement of the original C-O bond of the pro-R hydroxy-
methyl group of glycerol by the C-F bond in fluoroacetate 1,
has occurred with a predominant retention of configuration in
each case.
Enantiomeric Assay of [2H1]-Fluoroacetate from a Potas-
sium [2H4]-Succinate Feeding Experiment to S. cattleya.
Potassium [2H4]-succinate 4a was administered at a final
concentration of 9 mM to resting cells prepared from an 8 day
old culture of S. cattleya.9 The cells were incubated for 48 h
after which time the cells were worked up as before. Lyophi-
lization after acidification and then neutralization (dil NaOH)
followed by freeze-drying generated a sample of sodium [2-2H1]-
fluoroacetate 1a. 19F NMR analysis indicated a deuterium
enrichment of 15% in the resultant sodium [2H1]-fluoroacetate
as evidenced by a deuterium induced shift (δ ≈ 0.5 ppm) to
lower frequency of the natural abundance 19F NMR signal
(-218 ppm) corresponding to fluoroacetate (and also to that
corresponding to 4-fluorothreonine at -231 ppm). Cold “carrier”
sodium fluoroacetate (50 mg) was added to this material to
Figure 4. 2H{1H}-NMR spectra, in a chiral liquid crystalline medium of
DMF and PCBLL, of the synthetic samples of [1-2H1]-glycerols 3 prepared
as Scheme 5. (a) The 2H{1H}-NMR spectrum of (1R, 2R)-[1-2H1]-glycerol
3d, de >95%. (b) The 2H{1H}-NMR spectrum has signals for (1S, 2R)-
[1-2H1]-glycerol 3e (de 87%) as well as 3d and 3f.
lyophilizate was then neutralized (dil NaOH) and freeze-dried
to secure the resultant fluoroacetates as their sodium salts. After
the addition of cold carrier, the fluoroacetates were converted
to their corresponding hexyl [2-2H1]-fluoroacetates 19a, and each
sample was analyzed by chiral liquid crystal 2H{1H}-NMR. The
resultant spectra are shown in Figure 3b and 3c. The feeding
experiment with (1R, 2R)-[1-2H1]-glycerol 3d gave rise to a
sample of hexyl [2-2H1]-fluoroacetate 19a of 75%ee as il-
9
J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003 385

A R T I C L E S
O’Hagan et al.
adjust the level of isotope incorporation to ∼2% enrichment.
The material was then converted to hexyl [2-2H1]-fluoroacetate
19a as previously described and was dissolved in a chloroform
solution of PBLG for 2H{1H}-NMR analysis. The resultant 2H-
{1H}-NMR spectrum is shown in Figure 3d. The enantiomeric
purity of the [2H1]-fluoromethyl group of fluoroacetate 1 isolated
from the [2H4]-succinate feeding experiment is low, however it
can be estimated that there is a bias in favor of the (R)
enantiomer (25%ee) of [2-2H1]-fluoroacetate by comparison with
the spectrum of the synthetic (S)-[2H1]-fluoroacetate (38%ee)
(Figure 3a). In view of the known stereochemical course of the
various enzymes that process succinate (Scheme 3) such an
outcome indicates again a predominant retention of configura-
tion in going from a (R)-[1-2H1]-glycolytic phosphate intermedi-
ate such as 14 or 9 to (R)-[2H1]-fluoroacetate. The low
enantiomeric excess most probably arises from the extent of
biochemical processing that succinate has undergone as it is
directed toward fluorometabolite biosynthesis (Scheme 3). For
example, [3-2H1]-oxaloacetate 12a shown in Scheme 3 is an
intermediate between [2H4]-succinate 4a and the glycolytic
phosphates 14 and 9a. It is anticipated that the stereogenic center
at C-3 of [3-2H1]-oxaloacetate 12a will be susceptible to
significant in vivo racemization and could account for the low
%ee in this experiment.
type reaction. These results do not exclude the intermediacy of
a double bond in an elimination process, followed by fluoride
ion addition; however, this appears to be a less likely scenario.
This stereochemical study sheds some light on the mechanism
of biological fluorination but a more detailed analysis of the
reaction mechanism must await studies on the purified fluorina-
tion enzyme.
Experimental Section
General Procedures. All reagents were obtained from Aldrich
Chem. Co. Ltd. unless otherwise stated, and were used without further
purification. Solvents were dried and distilled prior to use and all
1
reactions were carried out under an atmosphere of N2. H- and 13C-
NMR spectra were recorded on a Varian Gemini 300 MHz (1H at 299.98
2
MHz, 13C at 75.431 MHz) spectrometer. H{1H}-NMR spectra were
recorded on Bruker DRX-400 NMR and Unity Inova 500 MHz
spectrometers.
Synthesis of a Reference Sample of Hexyl (S)-[2-2H1]-Fluoro-
acetate 19a. (Caution. All derivatives of fluoroacetate are extremely
toxic!) (R)-[7-2H1]-Benzyl alcohol 16b was prepared after treatment
of [7-2H]-benzaldehyde 17a with (S)-Alpine Borane following the
method of M. M. Midland et al.32 This generated a sample of (R)-[7-
2H1]-benzyl alcohol 16b with a 88%ee as judged by chiral liquid crystal
2H{1H}-NMR analysis (Figure 1a).31 The resultant (R)-[7-2H1]-benzyl
alcohol (0.49 g, 4.5 mmol) was added to a stirred and cooled (-78
°C) solution of diethylaminosulfur trifluoride (DAST) (0.7 mL, 5.8
mmol) in dichloromethane (40 mL). The reaction was allowed to stir
and come to ambient temperature over 10 h. Water (40 mL) was added
and the organic phase was washed with water (3 × 30 mL), dried over
MgSO4 and the solvent removed under reduced pressure on a cool rotary
evaporator. Care was taken not to exceed a water bath temperature of
20 °C to minimize evaporation of the benzyl fluoride product. (δF
(CDCl3): -207.20 (1F, d.t, JHF 47.6, JDF 7.3, CDHF); δH(CDCl3): 5.4-
(1H, d, JHF 47.8, CDHF) 7.41 (5H, s, Ar-H). The resultant (S)-[7-
2H]-benzylfluoride 18a had a 38%ee as judged by chiral liquid crystal
2H{1H}-NMR analysis (see Figure lb). This material was not purified
further and was taken directly through the oxidation reaction. Accord-
ingly, (S)-[7-2H]-benzylfluoride (398 mg, 2.2 mmol) was added to a
stirred solution of potassium permanganate (105 mg, 0.67 mmol) and
potassium periodate (39.2 g, 170 mmol) in water (200 mL). The mixture
was stirred at ambient temperature for 48 h and then most of the water
was removed under reduced pressure. The remaining solution was
acidified with sulfuric acid (2 M), and then the [2H1]-fluoroacetic acid
product was lyophilized. The lyophilizate was neutralized using sodium
hydroxide solution (0.1 M) and the water removed by freeze-drying to
furnish sodium (S)-[2-2H1]-fluoroacetate 1a as a white amorphous
residue. δF(CDCl3): -217.4 (1F, d.t, JHF 47.8, JDF 7.15, CDHF). This
material was diluted with unlabeled sodium fluoroacetate (400 mg, 4
mmol), thionyl chloride (0.5 mL, 5 mmol) was added and the suspension
heated under reflux for 2 h. Hexanol (2 mL, 16 mmol) was added and
the reaction stirred for a further 20 min. The reaction was quenched
by the addition of water and the product extracted into dichloromethane.
The solvent was dried (MgSO4) and hexyl fluoroacetate 19 was purified
over silica gel (100%, dichloromethane) to afford hexyl fluoroacetate
containing ∼1% hexyl (S)-[2-2H1]-fluoroacetate 19a as a colorless oil
(405 mg, 50%). δF (CDCl3): -230.4 (1F, d.t, JHF 46.5, JDF 7.34, CDHF,
1%), -229.9 (1F, t, JHF 47.6, CH2F, 99%), δH (CDCl3): 0.79 (3H, t,
J 6.0, CH3), 1.20 (6H, m), 1.56 (2H, m), 4.09 (2H, t, J 6.8, CH2O),
4.72 (2H, d, JHF 47.0, CH2F). δC (CDCl3): 13.8, 22.4, 25.3, 28.4, 31.3,
Conclusion
An assay to determine the configuration of (R) and (S)
[2-2H1]-fluoroacetates 1a has been developed based on 2H{1H}-
NMR in a lyotropic liquid crystalline phase. The assay has been
used to determine the enantiomeric excess of samples of [2-2H1]-
fluoroacetate of low isotope incorporation (∼1%), as the method
analyses only those molecules carrying a deuterium atom. The
method has been used to explore the stereochemical course of
fluorination during the biosynthesis of fluoroacetate 1 from
glycerol and succinate in S. cattleya. Three isotopically labeled
precursors, (1R, 2R)-[1-2H1]-glycerol 3d, (1S, 2R)-[1-2H1]-
glycerol 3e and [2H4]-succinate 4a, all gave rise to samples of
[2-2H1]-fluoroacetate 1a, generated in vivo during incubations
with washed resting cells of S. cattleya. In all cases, there was
some loss of stereochemical integrity in that the resultant
[2-2H1]-fluoroacetates were not enantiomerically pure. However,
the configurational data obtained after the feeding experiments
with (1R, 2R) and (1S, 2R)-[1-2H1]-glycerols 3d and 3e were
complementary and support a predominant overall retention of
configuration as the pro-R hydroxymethyl group of glycerol is
biotransformed to the fluoromethyl group of fluoroacetate 1.
In the case of 3e, the lower calculated %ee relative to 3d (30%ee
versus 75%ee) may be due to the low signal-to-noise ratio of
this sample. In light of the recent identification of the fluori-
nase13 from S. cattleya which mediates a reaction between
fluoride ion and SAM (Scheme 2), it follows that the admin-
istered glycerols became incorporated into the ribose ring of
ATP. SAM was then generated from ATP by combination with
L-methionine with concomitant stereochemical inversion at the
original pro-R hydroxymethyl carbon of glycerol. The overall
retention of configuration observed in these whole cell experi-
ments from glycerol or by labeling glycolytic intermediates from
succinate forces the conclusion that enzymatic C-F bond
formation also occurs with an inVersion of configuration
between SAM and fluoroacetate, perhaps as a result of an SN2
1
2
65.3, 77.4 (d, JCF 181.6, CH2F) 167.8 (d, JCF 22.3, CdO).
Glycerols 3d and 3e were Prepared as Previously Described.20
Preparation of Washed Resting Cells of S. cattleya. A strain of S.
cattleya NRRL 8057 was obtained from The Queen’s University of
Belfast, Microbial Biochemistry Section, Food Science Department,
Belfast (originally from United States Department of Agriculture,
9
386 J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003

Enantiomeric Analysis of [2H ]-Fluoroacetic Acid
A R T I C L E S
1
Agricultural Research Service, Midwest Area Northern Regional
Research Laboratories, Peoria, Illinois, USA). Cultures were maintained
on agar slants containing soybean flour (2% w/v), mannitol (2% w/v),
agar (1.5% w/v) and tap water, grown at 28 °C until sporulation could
be detected as previously described. The resultant static cultures were
stored at 4 °C. Seed cultures were prepared by transfer of spores and
aerial mycelia from a static culture, into a conical flask (500 mL)
plugged with cotton wool and containing chemically defined medium
(90 mL). After being shaken for 4 d at 28 °C and 180 rpm, an aliquot
(0.3 mL) of the vegetative mycelium was used to inoculate the batch
cultures. The cultures were then incubated at 28 °C and 180 rpm for
6-8 days. After growth for 6-8 d, cells were harvested by centrifuga-
tion, and the pellet was washed three times with 50 mM sterile MES
buffer pH 6.5. After the final wash the bacterial pellet was resuspended
in 50mM MES buffer (45 mL) pH 6.5 at a concentration of 0.18 wet
wt mL-1. This cell suspension was used for the subsequent isotopically
labeled precursor feeding experiments.
Feeding Experiments with (1R, 2R)-[1-2H1]-Glycerol 3d, (1S, 2R)-
[1-2H1]-Glycerol 3e and [2H4]-Succinate 4a. A solution of NaF (3
mL, 40 mM) was added to the washed cells (45 mL) prepared as
described above and then the suspension was diluted with MES buffer
(42 mL). The labeled precursors (1R, 2R)-[1-2H1]- 3d and (1S, 2R)-
[1-2H1]- glycerol 3e dissolved in 50 mM MES buffer and the pH
adjusted to 6.5 with 2M KOH solution. This solution was then added
to the cell suspension of S. cattleya such that the final glycerol
concentration was 9 mM. Similarly a solution of potassium [2H4]-
succinate 4a (3 mL, 40 mM) prepared from [2H4]-succinic acid dissolved
in 50 mM MES buffer and the pH adjusted to 6.5 with 2 M KOH
solution, was added to the S. cattleya cell supension by filter sterilisation
such that the final concentration of labeled succinate was 9 mM. In
each case incubations were carried out at 28 °C for 48 h on an orbital
shaker (175 rpm). For resting cell experiments, an aliquot (5 mL) of
the cell suspension was used and supplemented with a putative precursor
by filter sterilisation. The total volume was adjusted to 23 mL with 50
mM MES buffer pH 6.5 and incubated in 75 mL conical flasks plugged
with cotton wool at 28 °C on an orbital shaker at 200 rpm for 48 h.
Chiral Liquid Crystal 2H{1H}-NMR Analyses of Hexyl [2-2H1]-
Fluoroacetates 19. Poly-γ-benzylglutamate (120 mg, MWt 150 000-
350 000, Sigma Chem. Co. Ltd.,) and a sample (∼50-60 mg) of hexyl
[2-2H1]-fluoroacetate 19a were placed into a 5 mm NMR tube. Dry
chloroform (0.8-0.54 mL) was added until a 10-15wt % of PBLG
was achieved. To aid complete dissolution of the polymer and dispersion
of the solute through the gel the sample was repeatedly centrifuged
(20 times) on a low speed (600 rpm, ∼40 relative centrifugal force)
benchtop centrifuge, turning the tube upside-down between each
centrifugation. 2H{1H}-NMR spectra were recorded at 61.4 Hz on
Bruker DRX-400 spectrometer with a 2H-probe. The temperature was
held at 303 K with proton broad band decoupling and samples were
spun at 14 Hz. In a typical experiment eg (R)-19a, Figure 3b, the sample
contained 11 mgs of (R)-19a and had a 14.3wt % fraction of PBLG in
chloroform. The sample was run for 6 h involving 10 000 transients.
The quadrupolar splitting varies with the PBLG/CHCl3 wt fraction,
temperature and concentration of the solute, thus these are not constant
for every experiment. The experiments shown in Figure 3a, 3c and 3d
had PBLG/CHCl3 wt fractions of 11-12% and thus the quadrupolar
splittings are smaller (i.e., ∼230 Hz versus ∼440 Hz for the (R)-
enantiomer) in these three spectra compared with that observed (R)-
19a in Figure 3b.
Acknowledgment. We thank the Isle of Man Education
Authority for a studentship (R.J.M.G.) and the EPSRC/CNRS
CERC-3 program for an Anglo-French research grant.
JA026654K
9
J. AM. CHEM. SOC. VOL. 125, NO. 2, 2003 387