A practical synthesis of enantiopure 4,4,4-trifluoro-allo-threonine from an easily available fluorinated building block

Article abstract of DOI:10.1016/j.tetlet.2014.11.041

A practical method to prepare enantiopure 4,4,4-trifluoro-allo-threonine was developed by using an easily available fluorinated building block and a chiral auxiliary as starting materials. Trifluoroacetic anhydride was annulated with a ketene, derived fro

Full text of DOI:10.1016/j.tetlet.2014.11.041

Accepted Manuscript
A practical synthesis of enantiopure 4,4,4-trifluoro-allo-threonine from an easi-
ly available fluorinated building block
Joonil Cho, Siho Irie, Nobutaka Iwahashi, Yoshimitsu Itoh, Kazuhiko Saigo,
Yasuhiro Ishida
PII:
S0040-4039(14)01934-0
DOI:
http://dx.doi.org/10.1016/j.tetlet.2014.11.041
TETL 45425
Reference:
Tetrahedron Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
22 August 2014
4 November 2014
10 November 2014
Please cite this article as: Cho, J., Irie, S., Iwahashi, N., Itoh, Y., Saigo, K., Ishida, Y., A practical synthesis of
enantiopure 4,4,4-trifluoro-allo-threonine from an easily available fluorinated building block, Tetrahedron
Letters (2014), doi: http://dx.doi.org/10.1016/j.tetlet.2014.11.041
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
To create your abstract, type over the instructions in the template box below.
Fonts or abstract dimensions should not be changed or altered.
Leave this area blank for abstract info.
A practical synthesis of enantiopure 4,4,4-
trifluoro-allo-threonine from an easily
available fluorinated building block
Joonil Cho, Siho Irie, Nobutaka Iwahashi, Yoshimitsu Itoh, Kazuhiko Saigo, Yasuhiro Ishida
RIKEN Center for Emergent Matter Science, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
O
O
F3C
F3C
O
O
O
O
O
F3C
OH
OH
F3C
Cl
O
CF3
NH
Ph
O
N
O
N
O
O
O
H2N
OEt
O
O
Ph
O
Ph
Ph
Ph
Ph
O
O
Easily available materials
High stereoselectivity (99% d.r.)
7steps, 51% overall yield

1
Tetrahedron Letters
journal homepage: www.els evier.com
A practical synthesis of enantiopure 4,4,4-trifluoro-allo-threonine from an easily
available fluorinated building block
Joonil Cho a , Siho Irie b , Nobutaka Iwahashi b , Yoshimitsu Itoh b , Kazuhiko Saigo c , Yasuhiro Ishida a,∗
a RIKEN Center for Emergent Matter Science, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
b Department of Chemistry and Biotechnology, Graduate School of Engineering, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
c Equipment Support Planning Office, Kochi University, Kohasu, Oko-cho, Nankoku, Kochi 783-8505, Japan
ARTICLE INFO
ABSTRACT
Article history:
Received
A practical method to prepare enantiopure 4,4,4-trifluoro-allo-threonine was developed by using
an easily available fluorinated building block and a chiral auxiliary as starting materials.
Trifluoroacetic anhydride was annulated with a ketene, derived from a glycine equivalent
bearing a chiral oxazolidinone [(4S,5R)-4,5-diphenyloxazolidin-2-one], and acetone to afford a
trifluoromethylated α,β-unsaturated lactone. The asymmetric hydride reduction of the α,β-
unsaturated lactone proceeded with excellent stereoselectivity to give the corresponding (2R,3S)-
4-oxapentan-5-olide derivative exclusively (diastereopurity, 99%). From the reduced product
thus obtained, protecting groups were readily removed by acid treatment and subsequent
catalytic hydrogenolysis to afford enantiopure (2R,3S)-4,4,4-trifluoro-allo-threonine in an
excellent yield (7 steps, 51% overall yield).
Received in revised form
Accepted
Available online
Keywords:
Amino acids
Asymmetric synthesis
Cycloaddition
Fluorinated amino acids
Hydride reduction
2009 Elsevier Ltd. All rights reserved.
Fluorinated analogues of proteinogenic α-amino acids have
attracted considerable attention from various fields, such as
pharmacology, agrochemistry, and peptide engineering.1 Among
fluorinated α-amino acids reported hitherto, most extensively
studied are δ- and γ-fluorinated ones, owing to their chemical
stability, tolerance toward racemization, and properly reactive
amino group.2 In particular, fluorinated analogues of threonine
[(2S,3S)- and (2R,3R)-4,4,4-trifluorothreonines; L- and D-
isomers] and its allo-forms [(2S,3R)- and (2R,3S)-4,4,4-trifluoro-
allo-threonines; L- and D-isomers] are of interest. They possess a
modifiable hydroxyl group in their side chain and are useful as
precursors of various fluorinated chiral compounds, including
other fluorinated α-amino acids. In addition, because of the
electron-withdrawing nature of the trifluoromethyl group,3 the
hydrogen-bond donating ability of the hydroxyl group is highly
enhanced, thereby bringing significant effects on the
conformation of oligo-peptides when these fluorinated amino
acids are incorporated in peptide sequences.4 Also as a synthetic
challenge, the stereocontrolled preparation of these amino acids
is worthwhile, because the introduction of multiple
functionalities (amino, carboxyl, hydroxyl, and trifluoromethyl
groups) is required in a stereocontrolled manner at two
stereogenic centers.
So far, several groups have developed methods to prepare
(i) Annulation
(ii) Reduction
(iii) Deprotection
R3
O
O
R1*
F3C
O
F3C
H2N
OH
OH
N
R2*
R4
O
F3C
O
CF3
R1*
R3
O
Trifluoroacetic anhydride
Fluorinated building block
N
R2*
R3 R4
F3C
O
O
F3C
O
C
O
O
O
4,4,4-Trifluoro-allo-Thr
R4
O
R1*
R1*
O
O
H
N
R3
O
N
O
Ph
N
R1*
R3 R4
R2*
F3C
F3C
H2N
OH
OH
R2*
C
N
R4
α,β-Unsaturated lactone
Key intermediate
O
R1*
Ph
R2*
O
N
R2*
Oxazolidinone
Chiral N-source
Ketene
Gly equivalent
Ketone
Protective group
O
O
4,4,4-Trifluoro-Thr
Fluorinated α-amino aci
d
Scheme 1. Strategy for the asymmetric synthesis of 4,4,4-trifluoro-allo-threonine and 4,4,4-trifluorothreonine using trifluoroacetic anhydride
as a fluorinated building block.
———
∗ Corresponding author. Tel.: +81-48-462-1111 (ext. 6351); fax: +81-48-467-8214; e-mail: y-ishida@riken.jp (Y. Ishida)

2
Tetrahedron
O
O
Step-I
Step-II
H
NaOH
(excess)
O
Rf
O
NH
Ph
O
N
OEt
trans-addition
H
O
THF/DMF
rt
94%
water/THF
rt
93%
O
H
(2R,3S)-6a–c
(2S,3S)-6a–c
O
O
N
Ph
Ph
Ph
Rf
1
2
O
O
H
re-attack
O
H
N
Rf
O
O
(COCl)2
(2.8 eq)
N
O
O
O
(3S)-enolate
cis-addition
4
O
N
OH
H
O
N
Cl
Rf
O
3
5
O
O
H
O
toluene
rt
2
O
Ph
Ph
O
Rf
H
Ph
Ph
3
N
6
1
4
trans-addition
O
N
Rf
O
i) (RfCO)2O (3.0 eq)
pyridine (3.9 eq)
ii) (CH3)2CO (excess)
(2S,3R)-6a–c
(2R,3R)-6a–c
5a–c
H
Rf
H
O
O
O
O
H
a: Rf = CF3
b: Rf = CF2H
c: Rf = C2F5
O
Rf
H
N
O
N
si-attack
O
diethyl ether
i) rt, ii) 0 °C
a: 80%; b: 88%; c: 97%
(2 steps from 3)
O
(3R)-enolate
O
H
Ph
Ph
cis-addition
N
5a–c
O
Scheme 2. Synthesis of the key intermediates 5a–c.
Scheme 3. Hydride addition to the key intermediates 5a–c (Step-I)
and protonation of the resultant enolates (Step-II).
nonracemic 4,4,4-trifluoro-allo-threonine, which can be
classified into the following three categories: (i) The
enantioseparation of 4,4,4-trifluoro-3-hydroxybutanoic acid, the
introduction of an amine-equivalent functional group, and the
transformation of the functional group.5 (ii) The cross-coupling
of a bromoalkene and trifluoromethyl iodide, the asymmetric
dihydroxylation of the trifluoromethylated alkene, and the
transformation of the functional group.6 (iii) The asymmetric
aldol addition of a glycine equivalent with trifluoroacetaldehyde
and subsequent deprotection.7 However, these methods often
have problems in accessibility and handling of fluorinated
building blocks, stereocontrol, and lengthy synthetic steps.
Therefore, it remains a challenge to develop more practical
methods for the preparation of enantiopure 4,4,4-trifluoro-allo-
threonine.
For this aim, we focused on a three-component annulation
involving trifluoroacetic anhydride to form a trifluoromethylated
cyclic ester, reported by Zard et al.8,9 Trifluoroacetic anhydride
(boiling point of 40 °C, ~US$300/500 g, half lethal dose of 100
mg kg–1) is one of the most ideal fluorinated building blocks in
terms of cost, safety, and handling. In this reaction, a ketene and
trifluoroacetic anhydride generate a reactive trifluoroacetylketene
(and/or its synthetic alternatives), which readily undergoes the [4
+ 2] cycloaddition with a ketone to afford an α,β-unsaturated
lactone substituted with a trifluoromethyl group (Scheme 1, i).
By using this powerful one-pot reaction, a six-membered ring
bearing multiple functionalities, including trifluoromethyl, ester,
and enol ether units, can be constructed from easily available
materials.8,9
In the present study, we envisioned that, if a glycine
equivalent having a chiral N-substituent is used as the source of a
ketene (Scheme 1, left), the resultant α,β-unsaturated lactone
might serve as a key intermediate for the stereoselective
synthesis of 4,4,4-trifluoro-threonine and/or 4,4,4-trifluoro-allo-
threonine, through the asymmetric reduction of its α,β-
unsaturated carbonyl unit (Scheme 1, ii) and subsequent
deprotection (Scheme 1, iii). As a nitrogen source having a
chiral substituent, we chose an enantiopure oxazolidinone
[(4S,5R)-4,5-diphenyloxazolidin-2-one (1); Scheme 1, left].10
Owing to its rigid five-membered-ring structure bearing sterically
demanding two phenyl groups, this oxazolidinone serves as a
promising chiral auxiliary in asymmetric synthesis.11 In addition,
the proper location of benzylic nitrogen and oxygen in this chiral
auxiliary enables the efficient cleavage of the chiral N-protecting
units under mild conditions. Based on this synthetic strategy,
here we report a highly practical synthesis of enantiopure 4,4,4-
trifluoro-allo-threonine from an easily available fluorinated
building block.
The key intermediate 5a, the trifluoromethylated α,β-
unsaturated lactone bearing the oxazolidinone chiral auxiliary,
was prepared as shown in Scheme 2. Firstly, the chiral
oxazolidinone 1 was converted into the ester 2 having a glycine
skeleton by treating with sodium hydride and ethyl bromoacetate.
Through alkali-mediated hydrolysis and subsequent chlorination
using oxalyl chloride, the ester 2 was converted into the
corresponding acyl chloride 4. In a manner similar to the
procedure reported by Zard et al.,9b the annulation of
trifluoroacetic anhydride, acetone, and the ketene in situ
generated from 4 proceeded smoothly to afford the key
intermediate 5a. Interestingly, 5a exhibited notable signal
1
broadening in H and 19F NMR measurements (in CDCl3 at 25
°C; Supplementary Figure S1). As discussed later, this signal
broadening is most likely due to the restricted rotation around the
C–N bond that connects sterically demanding five-membered
(oxazolidinone) and six-membered (α,β-unsaturated lactone)
rings in 5a. In order to confirm the generality of this reaction,
difluoroacetic anhydride and pentafluoropropionic anhydride
were used in place of trifluoroacetic anhydride, and the
corresponding fluorinated α,β-unsaturated lactones 5b and 5c,
bearing
difluoromethyl
and
pentafluoroethyl
groups,
respectively, were obtained in excellent yields (88% for 5b and
97% for 5c). Worth noting is that the key intermediates 5a–c,
characterized with multiple functionalities and a well-defined
chiral structure, could be efficiently prepared from commercially
available materials (4 steps, >70% overall yield).
With the key intermediates 5a–c in hand, we carried out the
asymmetric hydride reduction of their α,β-unsaturated carbonyl
unit by using NaBH4 as a hydride source. The reduction involves
the following two consecutive asymmetric transformations
(Scheme 3). Step-I: the addition of a hydride to the α,β-
unsaturated carbonyl unit. Step-II: the protonation of the
resulting enolate. Because two stereogenic centers are generated
through these steps, four stereoisomers are possible [(2R,3S),
(2S,3R), (2R,3R), and (2S,3S)] for the resultant saturated lactone
6a–c. Quite interestingly, when the hydride reduction of 5a was
performed in methanol, followed by the treatment with aqueous
ammonium chloride, the reaction smoothly proceeded with
excellent diastereo- and enantioselectivities to exclusively yield
one of the trans isomers [(2R,3S)-6a: Table 1, entry 1; 99%
ratio]. Although a trace amount of the other trans isomer was

3
Table 1. Asymmetric hydride reduction of 5a–c.
Rf
N
Rf
N
i) NaBH4 (0.8 eq)
medium
–78 °C, 6 h
Rf
N
O
O
O
O
O
O
O
O
O
O
O
O
ii) NH4Claq
O
O
O
Ph
Ph
Ph
Ph
Ph
Ph
5a–c
–78 °C r.t.
(2R,3S)-6a–c
(2S,3S)-6a–c
a: Rf = CF3
b: Rf = CF2H
c: Rf = C2F5
Rf
Rf
O
O
O
O
O
N
N
O
O
O
O
O
Ph
Ph
Ph
Ph
(2S,3R)-6a–c
(2R,3R)-6a–c
entry substrate (Rf) medium
diastereomer ratioa)
(2R,3S) (2S,3R) (2S,3S) (2R,3R)
yieldb)
(2R,3S)
1
2
3
4
5
5a (CF3)
5a (CF3)
5a (CF3)
MeOH
EtOH
iPrOH
99
94
71
1
6
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
82%
77%
55%
92%
84%
29
1
5b (CF2H) MeOH/THFc) 99
5c (C2F5) MeOH 97
3
a) Determined by 19F NMR. b) Isolated yield. c) MeOH/THF = 1:4, v/v.
Absolute MeOH could not dissolve 5b sufficiently at –78 °C.
formed [(2S,3R)-6a; 1% ratio], these stereoisomers could be
easily separated by silica gel column chromatography. The
stereochemistry of these isomers was unambiguously determined
by X-ray crystallography of the isomer (2S,3R)-6a itself or a
derivative of (2R,3S)-6a (Supplementary Figures S7 and S8).12,13
In the present hydride reduction, the choice of reaction medium
brought a crucial effect on the ratio of the trans isomers [(2R,3S)-
6a:(2S,3R)-6a = 71:29–99:1, entries 1–3]. Irrespective of the
reaction medium used, the reaction mixtures always exhibited, in
19F NMR measurement, only two signals for the reduced species
that were attributable to the trans isomers [(2R,3S)- and (2S,3R)-
6a] (Supplementary Figure S2). Therefore, we concluded that
the cis isomers [(2R,3R)- and (2S,3S)-6a] were not generated at a
detectable level in the present reaction. Under the conditions
optimized for 5a, the hydride reduction of the other fluorinated
α,β-unsaturated lactones (5b and 5c) proceeded in good yield
with excellent stereoselectivity (entries 4 and 5),14 suggesting the
general utility of the present stereocontrolled reaction.
Figure 1. Optimized structures (B3LYP/6-31G*) of (a) the syn- and
anti-conformers of 5a (see Supplementary Figure S3 for details) and
(b) (2R,3S)- and (2S,3S)-6a viewed from top (left) and side (right) of
the six-membered ring.
(3S)-enolate (Scheme 3), which is in good agreement with the
stereochemistry at the C(3) position of the major product
[(2R,3S)-6a]. This mechanism can also explain the marked
solvent effect on the stereoselectivity (Table 1), taking into
account of the bulkiness of the hydride source. When the
reduction was performed in reactive alcohols such as methanol
(entry 1), the original hydride source NaBH4 was likely to be
converted into bulkier NaB(OR)nH4-n (n = 1–3), leading to higher
stereoselectivity. Contrary to this, when a less reactive alcohol
such as 2-propanol was used as a reaction medium (entry 3), the
conversion of the hydride source would be rather slow, so that
less-bulky NaBH4 served as the main hydride source to result in
lower stereoselectivity.15
Supposing that Step-I carried out in methanol exclusively
afford the (3S)-enolate from 5a, possible products, obtained by
the following protonation (Step-II), are (2R,3S)- and (2S,3S)-6a
(Scheme 3). Concerning its stereocontrol mechanism, we found
that the elimination and addition of the proton at the C(2)
position are reversible, as proved by the quick H/D exchange of
this proton in deuterated methanol at 25 °C (Supplementary
Figure S4). This is most likely because the acidity of the proton
at the C(2) position was highly enhanced by the electron-
withdrawing nature of the carbonyl and trifluoromethyl groups,
together with the effect of the constrained cyclic structure of the
lactone unit. Therefore, the stereochemistry of Step-II was
To elucidate this excellent stereoselectivity, reaction courses
of the elementary steps (Scheme 3, Step-I and Step-II) were
considered by using 5a as the representative. Step-I is an
irreversible and kinetically controlled reaction. As suggested by
1
the H and 19F NMR signal broadening (Supplementary Figure
S1), the molecular motion of 5a is highly restricted, most likely
in terms of the rotation around the C–N bond, due to the steric
congestion of the directly connected five-membered
(oxazolidinone) and six-membered (α,β-unsaturated lactone)
rings. Therefore, two kinds of conformers (syn and anti) are
supposed for 5a, where syn and anti represent the relative
orientation between the phenyl groups in the oxazolidinone unit
and the trifluoromethyl group in the α,β-unsaturated lactones
unit. Based on the molecular modeling study (MM2 and
B3LYP/6-31G*), the syn conformer is more stable by 1.6 kcal
mol–1 than the anti conformer (Figure 1a and Supplementary
Figure S3). In the syn-conformer, the si-face of the α,β-
unsaturated lactone is shielded by one of the phenyl groups of the
oxazolidinone unit, while the re-face is exposed for an
nucleophilic attack (Figure 1a). Therefore, the hydride addition
is considered to proceed preferentially at the re-face to afford the
controlled in a thermodynamic manner.
According to a
molecular modeling study (MM2 and B3LYP/6-31G*), (2R,3S)-
6a is much more stable than (2S,3S)-6a (∆G = 4.5 kcal mol–1). In

4
Tetrahedron
Acknowledgments
F3C
N
F3C
N
F3C
HCl
(3 M)
H2 (1 atm)
Pd/C
O
OH
OH
OH
OH
O
O
A part of this research is financially supported by the Grant-
in-Aid for Young Scientists (B) 21750137 (MEXT, Japan). The
generous allotment of computational time from the Institute for
Molecular Science (Okazaki, Japan) is gratefully acknowledged.
O
O
H2N
O
water/iPrOH
MeOH
rt
97 %
O
O
O
rt
Ph
Ph
(2R,3S)-6a
Ph
Ph
(2R,3S)-7a
91%
(2R,3S)-
4,4,4-Trifluoro-
allo-Thr
Supplementary Data
Scheme 4. Deprotection of (2R,3S)-6a to form (2R,3S)-4,4,4-
trifluoro-allo-threonine.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/–designated by journal--
.
the six-membered ring of the (2S,3S)-isomer, the oxazolidinone
and trifluoromethyl substituents are arranged near to each other
(Figure 1b, lower) to cause a sterically congested and unstable
structure, compared with the (2R,3S)-isomer. This stability order
is in good agreement with that of the trans and cis isomers of
general cyclic compounds. The calculated ∆G value of 4.5 kcal
mol–1 corresponds to the isomer ratio of 99.95:0.05 and can
elucidate the observed high stereoselectivity in Step-II.
References and notes
1. (a) Yoder, N. C.; Kumar, K. Chem. Soc. Rev. 2002, 31, 335–341;
(b) Bégué, J. -P.; Bonnet-Delpon, D. J. Fluorine Chem. 2006, 127,
992–1012; (c) Müller, K.; Faeh, C.; Diederich, F. Science 2007,
317, 1881–1886; (d) Filler, R.; Saha, R. Future Med. Chem. 2009,
1, 777–791; (e) Fluorine in Medicinal Chemistry and Chemical
Biology; Ojima, I., Ed.; Wiley-Blackwell: Chichester, 2009; (f)
Salwiczek, M.; Nyakatura, E. K.; Gerling, U. I. M.; Ye, S.;
Koksch, B. Chem. Soc. Rev. 2012, 41, 2135–2171.
As we envisioned, two-stage deprotection of (2R,3S)-6a
proceeded smoothly without losing diastereopurity to afford
(2R,3S)-4,4,4-trifluoro-allo-threonine (Scheme 4). In the first
stage, the acetal moiety of (2R,3S)-6a was removed by acid
2. (a) Sutherland, A.; Willis, C. L. Nat. Prod. Rep. 2000, 17, 621–
631; (b) Qiu, X. L.; Qing, F. L. Eur. J. Org. Chem. 2011, 3261−
3278; (c) Aceña, J. L.; Sorochinsky, A. E.; Moriwaki, H.; Sato, T.;
Soloshonok, V. A. J. Fluorine Chem. 2013, 155, 21–38.
3. (a) Schlosser, M. Angew. Chem. Int. Ed. 1998, 37, 1496–1513; (b)
O’Hagan, D. Chem. Soc. Rev. 2008, 37, 308–319.
4. (a) Kitamoto, T.; Marubayashi, S.; Yamazaki, T. Chem. Lett.
2006, 35, 1264–1265; (b) Kitamoto, T.; Marubayashi, S.;
Yamazaki, T. Tetrahedron 2008, 64, 1888–1894.
5. (a) Guanti, G.; Banfi, L.; Narisano, E. Tetrahedron 1988, 44,
5553–5562; (b) Von dem Bussche-Hunnefeld, C.; Seebach, D.
Chem. Ber. 1992, 125, 1273–1281; (c) Sting, A. R.; Seebach, D.
Tetrahedron 1996, 52, 279–290.
6. Jiang, Z. X.; Qin, Y. Y.; Qing, F. L. J. Org. Chem. 2003, 68,
7544–7547.
7. (a) Seebach, D.; Juaristi, E.; Miller, D. D.; Schickli, C.; Weber, T.
Helv. Chim. Acta 1987, 70, 237–261; (b) Soloshonok, V. A.;
Kukhar, V. P.; Galushko, S. V.; Svistunova, N. Y.; Avilov, D. V.;
Kuzmina, N. A.; Raevski, N. I.; Struchkov, Y. T.; Pysarevsky, A.
P.; Belokon, Y. N. J. Chem. Soc. Perkin Trans. 1 1993, 3143–
3155.
8. Zard, S. Z. Org. Biomol. Chem. 2007, 5, 205–213.
9. (a) Boivin, J.; El Kaim, L.; Zard, S. Z. Tetrahedron Lett. 1992, 33,
1285–1288; (b) Boivin, J.; El Kaim, L.; Zard, S. Z. Tetrahedron
1995, 51, 2573–2584; (c) Boivin, J.; El Kaim, L.; Zard, S. Z.
Tetrahedron 1995, 51, 2585–2592.
1
treatment to give (2R,3S)-7a in 91% yield. Because the H and
19F NMR spectra of the product showed one set of signals, we
concluded that the stereochemistry at the C(2) and C(3) positions
of the product was preserved through the acid treatment
(Supplementary Figure S5). Otherwise, the product must be a
mixture of multiple diastereoisomers, owing to the presence of
chiral oxazolidinone unit, and would give multiple sets of NMR
signals. Although the stereochemical stability at the C(2)
position of (2R,3S)-7a was quite unlike to the above-described
nature of its precursor (2R,3S)-6a, of which the C(2) proton can
exchange with solvent protons at ambient conditions
(Supplementary Figure S4), this is most likely because the
scission of the cyclic structure in (2R,3S)-6a released the
structural constraint and reduced the acidity of the C(2) proton.
In the next stage, the oxazolidinone unit of (2R,3S)-7a was
cleaved by catalytic hydrogenolysis to give the target (2R,3S)-
1
4,4,4-trifluoro-allo-threonine in 97% yield (Scheme 4). The H
and 19F NMR spectra of the product again showed one set of
signals, thereby excluding the formation of the diastereoisomers
(Supplementary Figure S6). Furthermore, the specific rotation of
26
the product ([α]D = 11.6 deg cm2 g–1, (c 0.4, H2O)) is in good
10. Ager, D. J.; Prakash, I.; Schaad, D. R. Chem. Rev. 1996, 96, 835–
875.
agreement with those reported,6 indicating that the
stereochemistry at the C(2) and C(3) positions of the product was
preserved through the hydrogenolysis.
11. (a) Saigo, K.; Kasahara, A.; Ogawa, S.; Nohira, H. Tetrahedron
Lett. 1983, 24, 511–512; (b) Hashimoto, Y.; Takaoki, K.; Sudo,
A.; Ogasawara, T.; Saigo, K. Chem. Lett. 1995, 235–236; (c)
Kagoshima, H.; Hashimoto, Y.; Oguro, D.; Saigo, K. J. Org.
Chem. 1998, 63, 691–697; (d) Matsui, S.; Hashimoto, Y.; Saigo,
K. Synthesis 1998, 1161–1166.
12. The minor isomer [(2S,3R)-6a] afforded large single crystals,
which allowed for the determination of its stereochemistry by X-
ray crystallography. On the other hand, the major isomer
[(2R,3S)-6a] was of low crystallinity. Therefore, (2R,3S)-6a was
In conclusion, we developed a highly practical and efficient
method to prepare enantiopure (2R,3S)-4,4,4-trifluoro-allo-
threonine (7 steps, 51% overall yield), using an easily available
fluorinated building block and a chiral auxiliary. Trifluoroacetic
anhydride is one of the most ideal fluorinated building blocks, in
terms of cost, safety, and handling. Because all reactions
proceeded in high yields without using expensive/toxic materials
and special facilities/techniques, the present method is
advantageous for large-scale synthesis. Moreover, this method is
certainly applicable for the preparation of (2S,3R)-4,4,4-trifluoro-
allo-threonine, because both enantiomers of the starting material,
erythro-2-amino-1,2-diphenylethanol, for the chiral auxiliary are
commercially available. Considering the possibility of further
asymmetric transformation of the key intermediates 5a–c, such as
the 1,4-addition to the α,β-unsaturated carbonyl unit using
various nucleophiles, the present method would lead to the large-
scale stereocontrolled preparation of other fluorinated α-amino
acids.
converted into a more crystalline derivative [(2R,3S)-8a; see
Supplementary Data], and its stereochemistry was determined by
X-ray crystallography.
13. CCDC-1012850 [(2R,3S)-8a] and CCDC-1012851 [(2S,3R)-6a]
contain the supplementary crystallographic data for this paper.
These data can be obtained free of charge from the Cambridge
Crystallographic
www.ccdc.cam.ac.uk/data_request/cif.
Data
Centre
via
14. Through the hydride reduction of 5b and 5c, two kinds of products
were exclusively obtained at ratios of 99:1 and 97:3. Judging
from the 1H NMR coupling constants between C(2)H and C(3)H
in their lactone unit (~10 Hz), these products are deduced to be the
trans isomers. Considering the analogy to 5a, the major and

5
minor products were tentatively assigned to be the (2R,3S)- and
(2S,3R)-isomers, respectively.
15. (a) Haubenstock, H.; Eliel, E. L. J. Am. Chem. Soc. 1962, 84,
2368–2371; (b) Tsuda, Y.; Sakai, K.; Akiyama, K.; Isobe, K.
Chem. Pharm. Bull. 1991, 39, 2120–2123.

Table 1. Asymmetric hydride reduction of 5a–c.
Rf
N
Rf
N
i) NaBH4 (0.8 eq)
medium
–78 °C, 6 h
Rf
N
O
O
O
O
O
O
O
O
O
O
O
O
ii) NH4Claq
O
O
O
Ph
Ph
Ph
Ph
Ph
Ph
5a–c
–78 °C r.t.
(2R,3S)-6a–c
(2S,3S)-6a–c
a: Rf = CF3
b: Rf = CF2H
c: Rf = C2F5
Rf
Rf
O
O
O
O
N
O
O
N
O
O
O
O
Ph
Ph
(2R,3R)-6a–c
Diastereomer ratio (%)a
Ph
Ph
(2S,3R)-6a–c
Entry
Substrates (Rf)
Medium
Yield (%)b
(2R,3S)
99
(2S,3R)
(2S,3S)
n.d.
(2R,3R)
n.d.
(2R,3S)
82
1
2
3
4
5
5a (CF3)
5a (CF3)
5a (CF3)
5b (CF2H)
5c (C2F5)
MeOH
EtOH
1
6
94
n.d.
n.d.
77
iPrOH
71
29
1
n.d.
n.d.
55
MeOH/THFc
MeOH
99
n.d.
n.d.
92
97
3
n.d.
n.d.
84
aDetermined by 19F NMR. bIsolated yield. cMeOH/THF = 1:4, v/v. Absolute
MeOH could not dissolve 5b sufficiently at –78 ºC.