Efficient synthesis of (2R,3S)-2-amino-3-(benzyloxy)-4,4,4- trifluorobutanoic acid (4,4,4-trifluoro-OBn-d-allothreonine)

Article abstract of DOI:10.1016/j.tetlet.2010.07.147

An efficient synthesis of the non-proteinogenic amino acid (2R,3S)-4,4,4-trifluoro(OBn)-threonine is described. Starting with commercially available (S)-Garner's aldehyde, the desired amino acid was prepared as its hydrochloride salt in five steps and an overall yield of 33% (59% based on recovered starting material). The utility of this unusual amino acid was demonstrated by its elaboration into a potent and selective androgen.

Full text of DOI:10.1016/j.tetlet.2010.07.147

Tetrahedron Letters 51 (2010) 5361–5363
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Efficient synthesis of (2R,3S)-2-amino-3-(benzyloxy)-4,4,4-trifluorobutanoic
acid (4,4,4-trifluoro-OBn-D-allothreonine)
Chun-min Zeng a, Sean A. Kerrigan a, John A. Katzenellenbogen b, Connie Slocum c, Kyla Gallacher c,
Maysoun Shomali c, C. Richard Lyttle c, Gary Hattersley c, Chris P. Miller c,
*
a Obiter Research, 2809 Gemini Court, Champaign, IL 61822-9647, United States
b University of Illinois, Department of Chemistry at Urbana-Champaign, 600 South Mathews Avenue, Urbana, IL 61801, United States
c RADIUS Health, 300 Technology Square, Cambridge, MA 02139, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
An efficient synthesis of the non-proteinogenic amino acid (2R,3S)-4,4,4-trifluoro(OBn)-threonine is
described. Starting with commercially available (S)-Garner’s aldehyde, the desired amino acid was pre-
pared as its hydrochloride salt in five steps and an overall yield of 33% (59% based on recovered starting
material). The utility of this unusual amino acid was demonstrated by its elaboration into a potent and
selective androgen.
Received 30 March 2010
Revised 22 July 2010
Accepted 26 July 2010
Available online 12 August 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Incorporating fluorine into molecules of pharmacological inter-
est has played a remarkably important role in drug design and dis-
covery.1 Despite its size similarity to hydrogen, fluorine varies
significantly from hydrogen with regard to its electronegativity,
hydrophobicity, hydrogen-bond acceptor capability (when bonded
to carbon), and resistance to metabolism. Not surprisingly, incor-
porating fluorine into otherwise natural biomolecules can be par-
ticularly interesting. Beyond the utility of such compounds per
se, fluorinated biomolecules can also be powerful extensions to
the extant chiral pool. Many fluorinated analogs of otherwise pro-
teinogenic amino acids have been prepared and studied in a variety
of contexts. In the course of our own research, we needed to access
useful quantities of the O-protected, non-proteinogenic amino acid
(2R,3S)-2-amino-3-benzyloxy-4,4,4-trifluorobutanoic acid hydro-
chloride (Fig. 1).
comprising a Sharpless asymmetric dihydroxylation to an olefin
and significant further elaboration to yield the desired amino acid
has been described.5
We either perceived or directly experienced challenges in pro-
ceeding with the literature routes, especially in view of our desire
to obtain the O-protected amino acid in gram quantities. Our alter-
nate route is shown below (Scheme 1). We started with the com-
mercially available (S)-Garner’s aldehyde 1 and thus were able to
efficiently leverage the extant chiral pool. We added the trifluoro-
methyl anion equivalent generated by TMSCF3 and TBAF to give 2
as a mixture of two diastereomers in an approximate ratio of 9:1.6
After benzylation and column chromatography, the major and de-
sired stereoisomer 3 was obtained.7 Treatment of 3 with p-TsOH
gave the alcohol 4 in 90% yield. Jones oxidation of 4, followed by
removal of the Boc group using saturated AcOEt with hydrogen
chloride provided 5 as a white solid in 84% yield over two steps.
Having developed a scheme that could produce this unusual
amino acid in synthetically useful quantities, we proceeded to
incorporate the molecule into our target structure 7 (Fig. 2). We
were curious to see whether the alteration in hydrophobicity and
hydrogen acidity compared to the non-fluorinated structure 8
could affect its activity at our biological target, the androgen
receptor.7a
There have been a few syntheses of (2R,3S)-2-amino-3-hydro-
xy-4,4,4-trifluorobutanoic acid or its enantiomer (2S,3R)-2
-amino-3-hydroxy-4,4,4-trifluorobutanoic acid reported in the lit-
erature. For example, the racemic and non-diastereoselective
aldol-condensation between
a glycine–Schiff-base condensate
and trifluoroacetaldehyde has been reported. The diastereomers
were then separated by column chromatography, the products
acetylated, and the enantiomers resolved by a lipase-mediated
saponification.2 Seebach has reported that aldol reaction between
a chiral imidazolidinone and trifluoroacetaldehyde proceeds with
good enantioselectivity and moderate diastereoselectivity.3 It has
also been shown that the desired amino acid can be prepared by
aldol reaction of a glycine, chiral Schiff-base condensate, and tri-
fluoroacetaldehyde.4 Additionally, a ‘from the ground-up’ approach
The synthesis of 7 is shown in Scheme 2. The key step in the
scheme is an ipso-fluoro substitution by the amino acid 6, afford-
ing the adduct 10 in modest yield. Intermediate 10 was taken for-
ward by sequentially coupling with the acyl hydrazide 11,
8
dehydration with TPP/I2, and deprotection with BBr3 to yield the
desired product 7. The binding affinity of compound 7 for the
androgen receptor was assayed and determined to have an IC50
of 290 nM, indicating a significantly lower affinity than that previ-
* Corresponding author. Tel.: +1 617 551 4702; fax: +1 617 551 4701.
E-mail address: cmiller@radiuspharm.com (C.P. Miller).
ously obtained for the non-fluorinated analog compound
8
0040-4039/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2010.07.147

5362
C. Zeng et al. / Tetrahedron Letters 51 (2010) 5361–5363
NH2.HCl
HO
CF3
O
OBn
Figure 1. (2R,3S)-2-Amino-3-benzyloxy-4,4,4-trifluorobutanoic acid hydrochloride.
(0.7 nM).9 Surprisingly, despite its relatively low binding affinity,
compound 7 demonstrated significant in vivo activity in the Rat
Herschberger assay.10 At a dose of just 3 mg/kg (orally), compound
7 increased levator ani muscle (‘LABC’) in a castrated rat to a level
significantly above the non-castrated control (‘Sham’) and testos-
terone propionate-treated (‘TP’) controls (Fig. 3). In contrast, com-
pound 7 increased the weight of the prostate to a significantly
lesser degree than that observed for the testosterone-treated or
sham control (Fig. 3). Selectivity of anabolic action on muscle over
androgenic action on prostate is considered a hallmark of tissue-
selective androgens, often referred to as selective androgen recep-
tor modulators (SARMs).
first requirement, the ultimate biological activity of the ligand–
receptor complex is multi-factorial due to the interaction of the li-
gand–complex with various co-regulatory proteins that can modu-
late the transactivation efficacy of the ligand. Further, in vivo
results are also dependent on numerous additional variables such
as solubility, absorption, and metabolism which might decrease
or increase a given molecule’s final efficacy and potency. As men-
tioned previously, because fluorine can affect a number of physical
parameters, the ultimate biological results from fluorine substitu-
tion are often not predictable a priori. In the instant case, the incor-
poration of fluorine atoms into a compound with high androgen
receptor binding affinity resulted in a compound with low receptor
affinity but potent in vivo activity.
In conclusion, we found this chemistry to be a powerful tool for
our purposes and thought it worth sharing. The other enantiomer
of the described compound could be prepared by starting with
(R)-Garner’s aldehyde. We think it would be interesting to see
whether our described method could be used to readily obtain
the other diastereomer as well. While a small amount could be
available as the minor diastereomer from the extant scheme, we
believe that a more productive approach might include a direct
inversion at the hydroxyl center, perhaps via a Mitsunobu-type hy-
droxyl inversion. In this way, the presently described chemistry
could be readily expanded to provide all four possible
stereiosomers.
The androgen receptor is a nuclear hormone receptor. While
affinity of a ligand for its cognate nuclear receptor is an important
Boc
N
Boc
N
O
H
1) TMSCF3/TBAF
92%
OH
NaH/DMF
O
O
BnBr
85%
CF3
2
1
NHBoc
CF3
Boc
N
CrO3
p-TsOH
OBn
H
HO
CF3
MeOH
50%
O
acetone
OBn
3
4
(90% based on st. mat.)
NH2.HCl
CF3
NHBoc
CF3
HCl
HO
HO
AcOEt
O
OBn
O
OBn
84% for two steps
5
6
LABC
*^
140
Scheme 1. Stereoselective synthesis of (2R,3S)-4,4,4-trifluoro(OBn)-threonine from
(S)-Garner’s aldehyde in five steps.
120
*
*
100
80
60
40
20
0
CF3
H
N
H
N
Cl
Cl
OH
OH
O
O
NC
NC
N
N
N
N
CN
CN
Veh
TP (1 mg/kg) cpd 7 (3
mg/kg)
Sham
*
7
8
Figure 2. Structures of target compound 7 and non-fluorinated analog 8.
Prostate
70
60
50
40
30
20
10
0
CF3
NH2.HCl
CF3
H
Cl
F
Cl
N
OBn
CO2H
Cs2CO3, Heat
DMSO
O
H
+
*
O
OBn
NC
NC
6
10
9
*^
22%
CF3
OH
H
N
1- EDCI/HOBt
Cl
NC
Et3N
O
O
N
NH2
N
N
H
7
Veh
TP (1 mg/kg) cpd 7 (3
mg/kg)
Sham
NC
CN
25% from 10
11
2 - TPP/I2, Et3N
3 - BBr3
Figure 3. Effect of compd 7 on levator ani muscle (LABC) and prostate weight in
treated, castrated rats. Organ and muscle weight is on Y-axis and the numbers refer
*
q <0.05 compared to vehicle. ^q <0.05 compared to TP and
Scheme 2. Synthesis of non-steroidal androgen 7 from non-proteinogenic amino
acid 6.
to weight in milligrams.
sham.

C. Zeng et al. / Tetrahedron Letters 51 (2010) 5361–5363
5363
(400 MHz, CDCl3) 5.21 (d, J = 7.6 Hz, 1H), 4.87 (d, J = 11.1 Hz, 1H), 4.58 (d,
J = 11.1 Hz, 1H), 4.13 (m, 1H), 4.00 (d, J = 11.7 Hz, 1H), 3.83 (m, 1H), 3.65 (d,
J = 11.7 Hz, 1H), 1.40 (s, 9H); 13C NMR (100 MHz, CDCl3) 155.7, 136.2, 129.0,
128.9, 128.7, 125.2 (q), 80.4, 78.1 (q), 76.0, 62.1, 50.7, 28.5; HRMS (ES) calcd for
Acknowledgment
We wish to acknowledge Dr. Bill Boulanger, the owner of Obiter
Research, for his kind support of this work.
C
16H22F3NO4Na [M+Na]+ 372.1399, found 372.1403.
Compound 5: To a solution of tert-butyl (2S,3S)-3-(benzyloxy)-4,4,4-trifluoro-1-
hydroxybutan-2-ylcarbamate (1.47 g, 4.2 mmol) in acetone (80 mL) was added
Jone’s reagent (8.2 mL) at 0 °C. The mixture was stirred at the same
temperature for 3 h until the starting material was completely consumed,
then the reaction was quenched by adding isopropanol (5 mL). The reaction
mixture was extracted with AcOEt. The AcOEt extracts were washed with
water, brine and dried over Na2SO4. Removal of the solvent gave the crude
(2R,3S)-3-(benzyloxy)-2-(tert-butoxycarbonylamino)-4,4,4-trifluorobutanoic
acid (1.56 g), which was used directly for the following reaction. Mp 118–
References and notes
1. (a) Hagmann, W. K. J. Med. Chem. 2008, 51, 4359; (b) Muller, K.; Faeh, C.;
Diederich, F. Science 2007, 317, 1881.
2. Kitazume, T.; Lin, J. T.; Yamazaki, T. Tetrahedron: Asymmetry 1991, 2, 235.
3. Seebach, D.; Juaristi, E.; Miller, D. D.; Schickli, C.; Weber, T. Helv. Chem. Acta
1987, 70, 237.
4. Soloshonok, V. A.; Kukhar’, V. P.; Galushko, S. V.; Svistunova, N. Y.; Avilov, D. V.;
Kuz’mina, N. A.; Raevski, N. I.; Struchkov, Y. T.; Pysarevsky, A. P.; Belokon’, Y. N.
J. Chem. Soc., Perkin Trans. 1 1993, 3143.
120 °C; ½a 2D5
ꢀ
= +13.1 (c 0.79, MeOH); 1H NMR (400 MHz, CD3OD) 7.25–7.34 (m,
5H), 4.76 (d, J = 10.9 Hz, 1H), 4.70 (d, J = 10.9 Hz, 1H), 4.53 (d, J = 6.4 Hz, 1H),
4.32 (m, 1H), 1.41 (s, 9H); 13C NMR (100 MHz, CD3OD) 170.7, 156.1, 136.9,
128.2, 128.1, 128.0, 125.0 (q), 79.8, 76.5 (q), 74.8, 53.1, 27.4; HRMS (ES) calcd
for C16H21F3NO5 [M+H]+ 364.1372, found 364.1376.
5. Jiang, Z.-X.; Qin, Y.-Y.; Qing, F.-L. J. Org. Chem. 2003, 68, 7544.
6. The addition of
a trifluoromethyl anion to Garner’s aldehyde has been
Compound
6:
To
a
solution
of
(2R,3S)-3-(benzyloxy)-2-(tert-
described but the diastereomer ratio was not reported. See Qing, F.-L.; Peng,
S.; Hu, C.-M. J. Fluor. Chem. 1998, 88, 79–81.
7. Experimental procedures and selected analytical data for the preparation of
(2R,3S)-4,4,4-trifluoro(OBn)-threonine.
butoxycarbonylamino)-4,4,4-trifluorobutanoic acid (1.56 g, 4.2 mmol) in
AcOEt (30 mL) was added AcOEt saturated with hydrogen chloride (30 mL) at
0 °C. Then, the mixture was stirred at room temperature for 1.5 h. After
concentration of the reaction mixture to about 15 mL, the white solid was
collected by filtration and washed with AcOEt, dried in vacuum to give the title
Compound 2: To
a mixture of (S)-4-formyl-2,2-dimethyl-oxazolidine-3-
carboxylic acid tert-butyl ester (S-Garner’s aldehyde) (1.0 g, 4.36 mmol) and
(trifluoromethyl)trimethylsilane (2.0 M solution in THF, 2.6 mL, 5.20 mmol) in
THF (10 mL) was added tetrabutylammonium fluoride (1.0 M solution in THF,
0.1 mL, 0.10 mmol) at 0 °C. After addition, the mixture was stirred at room
temperature for a weekend (2.5 days). Then, tetrabutylammonium fluoride
(1.0 M solution in THF, 9 mL, 9.0 mmol) was added, the mixture was stirred for
another 6 h, then quenched by adding saturated aq NaHCO3 solution and
extracted with AcOEt. The AcOEt extracts were washed with saturated aq
NaHCO3 solution, water, brine and dried over Na2SO4. Concentration and
compound
(2R,3S)-2-amino-3-(benzyloxy)-4,4,4-trifluorobutanoic
acid
= ꢁ13.8
hydrochloride salt (1.06 g, 84% for two steps). Mp 194–196 °C; ½a D25
ꢀ
(c 1.25, MeOH); 1H NMR (400 MHz, CD3OD) 7.32–7.44 (m, 5H), 4.86 (d,
J = 2.0 Hz, 2H), 4.70 (dq, J = 3.6, 6.6 Hz, 1H), 4.50 (d, J = 3.6 Hz, 1H); 13C NMR
(100 MHz, CD3OD) 166.0, 136.0, 128.6, 128.5, 128.4, 123.9 (q), 74.5, 74.2 (q),
52.4; HRMS (ES) calcd for C11H13F3NO3 [MꢁCl]+ 264.0848, found 264.0843.
(a) US 2010/00417121.
8. Wipf, P.; Miller, C. P. J. Org. Chem. 1993, 58, 3604–3606.
9. The AR-binding assay was performed as specified by the manufacturer
purification
gave
(S)-tert-butyl
2,2-dimethyl-4-(2,2,2-trifluoro-1-
(Invitrogen, Madison, WI). Briefly, 1 ll of 10 mM compound was added to
hydroxyethyl)oxazolidine-3-carboxylate (1.21 g, 92%), which showed
complicated NMR spectrum because of the different conformers.
a
500
l
l of AR screening buffer in a 1.5 ml eppendorf tube to make a 2 ꢂ 10ꢁ5
M
stock. 10-fold serial dilutions of the test compounds were prepared ranging in
concentration from 10ꢁ5 M to 10ꢁ12 M. Each dilution was added in triplicate to
a black 384-microtiter plate. The test compounds are diluted twofold in the
final reaction. 2ꢂ AR-Fluormone™ complex was prepared with 2 nM
Compound 3: To a suspension of NaH (60% in mineral oil, 300.0 mg, 7.50 mmol)
in DMF (20 mL) was added (S)-tert-butyl 2,2-dimethyl-4-(2,2,2-trifluoro-1-
hydroxyethyl)oxazolidine-3-carboxylate (1.09 g, 3.64 mmol) in DMF (20 mL),
the mixture was stirred at room temperature for 1 h. Then, benzyl bromide
(0.86 mL, 7.23 mmol) was added and the mixture was stirred at room
temperature overnight, then quenched with ice-water and extracted with
AcOEt. The AcOEt extracts were washed with water, brine and dried over
Na2SO4. Removal of the solvent gave a residue, which was purified by silica gel
Flourmone AL Green™ and 30 nM AR. 25
l
l of 2ꢂ complex was aliquoted to
each reaction well, such that the final reaction volume was 50
l
l per well. Plate
was sealed with a foil cover and incubated in the dark at room temperature for
4 h. Polarization values for each well were measured. The polarization values
were plotted against the concentration of the test compound. The
concentration of the test compound that results in half-maximum shift
equals the IC50 of the test compound. As a control, a competition curve for
R1881(methyltrienolone) was performed for each assay. Curve Fitting was
performed using GraphPad PrismÒ software from GraphPad™ Software Inc.
10. The Herschberger assay looks primarily at the ability of androgens to increase
muscle size in an immature, castrated rat. In addition, androgenic effects are
looked at primarily by weighing the prostate. Selective compounds will show a
greater increase in the levator ani relative to the prostate when compared to
testosterone treated, castrated animals or to intact animals that have not been
treated. Immature Sprague Dawley male rats were obtained Charles River
Laboratories (Stoneridge, NY). All animals were maintained in a temperature
and humidity controlled room with a 12 h light: 12 h dark cycle, with ad lib
access to food (TD 291615, Teklad, Madison, WI) and water. Rats were
anesthetized and orchidectomized (GDX) or sham surgery (SHAM) was
chromatography
to
give
(S)-tert-butyl
4-((R)-1-(benzyloxy)-2,2,2-
trifluoroethyl)-2,2-dimethyloxazolidine-3-carboxylate (1.21 g, 85%), which
showed a complicated NMR spectrum because of the different rotamers. 1H
NMR (400 MHz, CDCl3) 7.28–7.33 (m, 5H), 4.60–4.85 (m, 3H), 3.93–4.23 (m,
3H), 1.42–1.65 (m, 15H).
The assignment of the stereochemistry of the major isomer 3 was based on (1)
the preferential formation of the anti-adduct from a nonchelation controlled
addition of an
addition of trichloromethyl anion to
aldehyde gave the anti-adduct as the only diastereomer;12 and (3)
a
-substituted aldehyde and a carbanion;11 (2) the reported
a
very similar aldehyde as Garner’s
a
comparison of 1H NMR spectrum of the amino acid derived from our
debenzylated product (the free amino acid) to the data for the same
diastereomer as reported in the literature. In particular, the chemical shift of
the b-hydrogen (CHCF3) of the anti-isomer as obtained by us at d 4.24 (DMSO-
d6) closely matches the reported value of d 4.20 (DMSO-d6) whereas the
chemical shift of the b-hydrogen (CHCF3) of the syn-isomer was reported at d
4.70 (DMSO-d6).13
performed. After
a 7-day recovery period, the animals were randomized
according to weight and assigned to treatment groups (n = 5), SHAM,
OVX + vehicle, OVX + compd treated. Testosterone propionate (TP 1 mg/kg in
5% DMSO/95% corn oil) was administered by once daily subcutaneous
injections, while the compound was dosed in vehicle (0.5%
carboxymethylcellulose) and administered by once daily oral gavage. The
rats were then dosed once daily for 4 days. All animals were euthanized via
carbon dioxide inhalation 24 h after the last dose. The prostate and levator ani
and bulbocavernosus (LABC) tissues were removed, weighed and recorded.
11. Cee, V. J.; Cramer, C. J.; Evans, D. A. J. Am. Chem. Soc. 2006, 128, 2920.
12. Beaulieu, P. L. Tetrahedron Lett. 1991, 32, 1031.
Compound 4: To
a solution of (S)-tert-butyl 4-((R)-1-(benzyloxy)-2,2,2-
trifluoroethyl)-2,2-dimethyloxazolidine-3-carboxylate (1.21 g, 3.1 mmol) in
methanol (100 mL) was added p-toluenesulfonic acid monohydrate (90 mg,
0.47 mmol). The mixture was stirred at room temperature for 5 days, then the
methanol was removed and the residue was dissolved with AcOEt. The AcOEt
solution was washed with saturated aq. Na2CO3 solution, water, brine and
dried over Na2SO4. Removal of the solvent gave a residue, which was purified
by flash chromatography to afford the title compound tert-butyl (2S,3S)-3-
(benzyloxy)-4,4,4-trifluoro-1-hydroxybutan-2-ylcarbamate (0.54 g, 50%, 90%
based on recovered starting material) and recovered starting material (0.54 g),
13. Scolastico, C.; Conca, E.; Prati, L.; Guanti, G.; Banfi, L.; Berti, A.; Farina, P.;
Valcavi, U. Synthesis 1985, 850.
which can be recycled. Mp 90–92 °C; ½a D25
ꢀ
= ꢁ66.9 (c 0.71, CHCl3); 1H NMR