to 28 °C over 15 min, then GST fusion protein expression was
induced by adding isopropylthio-â-D-galactoside to a final con-
centration of 0.1 mM. After being stirred for an additional 6 h
under these conditions, the cells were collected by centrifugation
(6000g for 10 min at 4 °C). These could be stored at 4 °C for
later use, if desired. Approximately 25 g (wet weight) of cells
was resuspended in 1 L of 10 mM KPi (pH ) 5.6) containing 4
g/L glucose in a Braun Biostat B fermenter vessel (2 L total size)
containing 0.5 g of XAD-4 resin. The temperature, pH, and
dissolved oxygen level were maintained at 30 °C, 5.6 (3 M NaOH
titrant), and 75% saturation (fixed air flow of 0.25 vvm and
variable stirring rate), respectively. Aliquots of neat 5 (0.2 mL)
were added approximately every hour over a total of 12 h to a
final concentration of 6 mM. Glucose was added portionwise after
3 and 6 h to maintain a glucose concentration of approximately
4 g/L. Consumption of both 5 and glucose slowed significantly
after 8 h. After 24 h, the entire reaction mixture was gently
extracted with CH2Cl2 (2 × 300 mL) to avoid an emulsion, using
a procedure described in detail elsewhere.22 The organic layer
was dried with MgSO4, concentrated in vacuo, and purified by
flash chromatography (cyclohexane/Et2O 85:15) to afford 1.3 g
of 6 as a colorless oil (91% yield). [R]D -3.0°, c 0.68, CHCl3; lit.33
[R]D -3.0°, c 1.7, CHCl3. Anal. Calcd for C11H13O3Cl: C, 57.78;
H, 5.73. Found: C, 57.92; H, 5.86. Spectral data matched those
reported in the literature.35
Epoxides 7 and ent-7. To a solution of chlorohydrin 6 (1.3
g, 5.7 mmol) in DMF (28 mL) was added K2CO3 (2.2 g, 17 mmol,
3 equiv) and water (525 µL). After being stirred for 5 h at room
temperature, the mixture was diluted with water (75 mL) and
extracted with ether (3 × 75 mL). The combined organics were
washed with water to remove residual DMF (6 × 5 mL), dried
with MgSO4, and concentrated in vacuo to yield 1.1 g of 7 as a
colorless oil (99% yield). No further purification was necessary.
[R]D +24°, c 1.5, CHCl3; lit.26 [R]D +25°, c 1.1, CHCl3. Anal. Calcd
for C11H12O3: C, 68.74; H, 6.29. Found: C, 68.34; H, 6.57.
Spectral data matched those reported in the literature.26 Crude
ent-7 was prepared in the same way and subjected to silica gel
chromatography to remove a small amount of trans-epoxide,
affording 0.86 g of ent-7 as a colorless oil (66% overall yield from
5). [R]D -29°, c 2.9, CHCl3.
Oxazolines 2 and ent-2. A solution of glycidic ester 7 (0.50
g, 2.6 mmol) in benzonitrile (4 mL) was cooled to 0 °C under
argon atmosphere, then BF3‚OEt2 (330 µL, 2.6 mmol, 1.0 equiv)
was added over 10 min. The ice bath was removed, and the
reaction was stirred for 3 h before aqueous saturated NaHCO3
(4 mL) was added. After being stirred for an additional 2 h, the
reaction mixture was diluted with water (20 mL) and extracted
with CH2Cl2 (3 × 20 mL). The combined organics were dried
with MgSO4, concentrated in vacuo, and purified by silica gel
chromatography (cyclohexane/ether 85:15) to afford 0.42 g of 2
as a colorless oil (55% yield). 2: [R]D +11°, c 1.1, CHCl3. ent-2:
[R]D -12°, c 1.7, CHCl3. Spectral data for both 2 and ent-2
matched those reported in the literature for the racemic com-
pound.36,37
Taxol Side Chain Ethyl Esters 8 and ent-8. A solution of
oxazoline 2 (0.32 g, 1.1 mmol) in 0.5 M HCl (2.5 mL) and EtOH
(7 mL) was heated at reflux for 6 h. The solvent was removed
under reduced pressure, and then the residue was dissolved in
CH2Cl2 (10 mL), washed with water (2 × 7.5 mL), dried with
MgSO4, and concentrated under reduced pressure. The residue
was purified by silica gel chromatography (hexanes/Et2O 85:15)
to afford 0.30 g of 8 (85% yield) as a white solid, mp 162-163
°C, lit.38 mp 164-165 °C. [R]D -12 °, c 2.0, CHCl3; 1H NMR
(CDCl3) δ 1.30 (t, 3H, J ) 7.1), 3.30 (d, 1H, J ) 3.9), 4.30 (m,
Lewis acid) failed to improve this ratio. Fortunately, pure
trans-oxazoline 2 could be isolated from the crude mix-
ture by column chromatography in 55% yield. This was
hydrolyzed under mildly acidic conditions to yield the
natural enantiomer of the Taxol side chain ethyl ester
1
8. Its syn-configuration was evident from the H NMR
spectrum, particularly by the chemical shift (5.76 ppm)
and J2,3 value (1.8 Hz) of the C3 proton (in the case of
the trans-analogue, this signal lies 0.16 ppm upfield and
the corresponding J2,3 value is 3.5 Hz).34 The optical
purity of 8 was assessed by 1H NMR after derivatization
with (S)-R-methoxy-R-phenylacetic acid.
A complementary route was used to prepare the
antipode of the Taxol side chain. In this case, an E. coli
strain overexpressing the yeast YGL039w short chain
dehydrogenase was used for the bioconversion. Because
the C2-epimeric chlorohydrins resulting from this reaction
were inseparable by column chromatography, the mix-
ture was subjected to mild base-mediated ring closure.
The desired cis-(2S,3S)-epoxide ent-7 was isolated from
the crude product mixture in 66% overall yield from 5.
Its stereochemistry and absolute configuration were
confirmed by proton coupling and optical rotation values.
This material was carried on to ent-2 and ent-8 in the
same manner described above.
The ability to produce directly either enantiomer of the
chlorohydrin cis-epoxide precursor was a direct conse-
quence of the highly syn-selective biocatalysts used to
reduce 5 and our success in identifying individual yeast
reductases with the required stereoselectivities. These
results further underscore the power of genome-wide
expression libraries for solving synthetic problems. In
addition, the observation that two enzymes within the
same organism possessed complementary enantioselec-
tivities re-emphasizes the importance of screening indi-
vidual enzymes, rather than whole cells where such
enzymes would compete with one another.21
Using the Ritter reaction to install the nitrogen has
significant safety advantages over earlier methods that
require azide opening of epoxide 7. Our approach also
eliminated the need for a separate conversion to the
oxazoline, thereby enhancing the brevity and atom
economy of our synthesis. In fact, only the chlorine atom
was lost in the conversion of 5 to the target compounds.
It should be noted that the Taxol side chain is only
one member of a large family of R-hydroxy-â-amino acids
that are key substructures in pharmaceutically active
compounds, and our chemoenzymatic strategy may also
be useful for other members of this family. Our success
in the present case bodes well for such future applica-
tions.
Experimental Section
(2S,3R)-Ethyl 2-Chloro-3-hydroxy-3-phenylpropionate 6.
A 45-mL portion of LB medium supplemented with 30 µg/mL
kanamycin was inoculated with a single colony of E. coli BL21-
(DE3)(pIK8). After being shaken overnight at 37 °C, 40 mL of
this preculture was added to 4 L of the same medium containing
4 g/L glucose in a New Brunswick M19 fermenter. The culture
was grown for 2 h at 37 °C with a stir rate of 800 rpm and an
air flow of 0.5 vessel volumes per min (vvm) until it reached an
optical density at 600 nm of 0.6. The cell suspension was cooled
(35) Cabon, O.; Buisson, D.; Larcheveˆque, M.; Azerad, R. Tetra-
hedron: Asymmetry 1995, 6, 2199-2210.
(36) Yokoyama, M.; Menjo, Y.; Watanabe, M.; Togo, H. Synthesis
1994, 1467-1470.
(37) Repeated attempts to obtain consistent elemental analysis data
for 2 and ent-2 were unsuccessful, probably because of spontaneous
hydrolysis.
(38) Patel, R. N.; Banerjee, A.; Howell, J. M.; McNamee, C. B.;
Brozozowski, D.; Mirfakhrae, D.; Nanduri, V.; Thottathil, J. K.; Szarka,
L. J. Tetrahedron: Asymmetry 1993, 4, 2069-2084.
(34) Mader, M. M.; Edel, J. C. J. Org. Chem. 1998, 63, 2761-2764.
9656 J. Org. Chem., Vol. 70, No. 23, 2005