Crystallization-induced asymmetric transformations and self-regeneration of stereocenters (SROSC): Enantiospecific synthesis of α-benzylalanine and hydantoin BIRT-377

Article abstract of DOI:10.1016/S0040-4039(01)00403-8

N-Isobutoxycarbonyl protected L-alanine was condensed with 4-phenylbenzaldehyde in a crystallization-controlled process to give the corresponding cis-oxazolidinone derivative as the sole product in high yield; this underwent enolization and benzylation wi

Full text of DOI:10.1016/S0040-4039(01)00403-8

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 42 (2001) 3231–3234
Crystallization-induced asymmetric transformations and
self-regeneration of stereocenters (SROSC): enantiospecific
synthesis of a-benzylalanine and hydantoin BIRT-377
Elio Napolitano* and Vittorio Farina
Department of Chemical Development, Boehringer Ingelheim Pharmaceuticals Inc., 900 Ridgebury Road, Ridgefield,
CT 06877, USA
Received 25 January 2001; revised 2 March 2001; accepted 8 March 2001
Abstract—N-Isobutoxycarbonyl protected
L-alanine was condensed with 4-phenylbenzaldehyde in a crystallization-controlled
process to give the corresponding cis-oxazolidinone derivative as the sole product in high yield; this underwent enolization and
benzylation with the typical high degree of stereoselectivity observed in the SROSC involving this class of compounds. © 2001
Elsevier Science Ltd. All rights reserved.
Because of their important applications in medicinal
chemistry, quaternary a-amino acids (and particularly
methylated ones) have received much attention from
the synthetic standpoint.1 Among the many approaches
to enantiomerically pure quaternary a-amino acids, the
protocol called Self-Regeneration of Stereocenters
(SROSC), devised and extensively studied by Seebach
and co-workers, is especially attractive because it starts
with natural amino acids (a class of compounds readily
available in enantiomerically pure form), it does not
employ any other chiral reagents or catalysts, and it
enjoys a very broad generality.2
BIRT-377 (1, a hydantoin endowed with promising
anti-inflammatory properties),3 the embodiment of this
principle would involve the benzylation of either the cis
oxazolidinone 3 or its trans isomer 4, resulting respec-
tively from condensation of -alanine (5) or its enan-
L
tiomer (6) with an aldehyde and an acylating agent.
Upon careful examination of the literature and some
preliminary experiments, however, it becomes readily
apparent that what may be hindering the general appli-
cability of SROSC to large scale preparations of qua-
ternary a-methylated amino acids is the rather
awkward experimental protocol,4 the risk of racemiza-
tion upon condensation of the amino acid salt with an
aldehyde, and the difficulty in producing either a cis or
trans oxazolidinone (3 or 4) selectively. Apparently, no
In the specific case of a-benzylalanines, such as the one
(2, X=Br) we required for a large scale preparation of
Keywords: dynamic resolution; alkylation; quaternary amino acids; oxazolidinones; hydantoins.
* Corresponding author. On leave from the Scuola Normale Superiore di Pisa. Present address: Dipartimento di Chimica Bioorganica e
Biofarmacia, Via Bonanno 33, 56100 Pisa, Italy. Fax: +39-050-43321; e-mail: eliona@farm.unipi.it
0040-4039/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(01)00403-8

3232
E. Napolitano, V. Farina / Tetrahedron Letters 42 (2001) 3231–3234
attempt has ever been made at understanding the mech-
anistic basis of the stereochemical outcome of C(2) in
the step in which the oxazolidinone ring is formed,
which is where the relay of chirality begins, and a
recent review simply concludes that it is highly sub-
stituent-dependent.1 As a matter of fact, depending on
the nature of the aldehyde and the acylating agent, the
Seebach protocol is reported to yield predominantly the
cis isomer (4:1, R=tBu, Z=Ph),5 mostly trans (7.5:1,
R=Z=Ph),6 (2.5:1, R=Ph, Z=BnO),7 exclusively cis
(R=ferrocenyl, Z=tBu),8 and exclusively trans (R=
Z=Ph).9 We wish to report our approach to a rational
and efficient solution of problems related to alkylation
of alanine by means of the SROSC, which will hope-
fully make this method more suitable for large scale
preparations of enantiomerically pure methylated qua-
ternary a-amino acids.
A, isobutoxycarbonyl alanine (8) was condensed with
4-phenylbenzaldehyde (9) under the typical conditions
of acetal formation (reflux in a hydrocarbon solvent in
the presence of p-toluenesulfonic acid under removal of
water by azeotropic distillation);13 an equilibrium mix-
ture of 10 and its C(2) epimer resulted when a homoge-
neous solution was maintained throughout the reaction;
but we were pleased to see that, when the condensation
between 8 and 9 was carried out in the appropriate
volume of cyclohexane or iso-octane, 10 precipitated as
a nicely crystalline solid from the reaction mixture and
could be isolated in 91% yield (with a negligible amount
of its C(2) epimer in the mother liquors), thus demon-
strating that oxazolidinone formation and its dynamic
resolution could be very conveniently accomplished in
the same step (Scheme 1). Alternatively, 8 was con-
verted to the acid chloride (oxalyl chloride in methylene
chloride) and this intermediate added to 9 in the pres-
ence of a catalytic amount of ZnCl2;14 an equilibrium
mixture of 10 and its C(2) epimer resulted from the
condensation of 8 and 9 in homogeneous solution:
however, when most of the solvent of the reaction was
removed and the product allowed to precipitate by
addition of methyl t-butyl ether, pure 10 (88% yield)
was isolated without precipitation of the epimer, which
was present in the mother liquors in the thermodynamic
proportions, thus showing that a second-order asym-
metric transformation was operating under these condi-
We started by analyzing the thermodynamic profile of
oxazolidinone formation, which was an easy task after
finding that N-acylated-2-arylsubstituted derivatives
undergo epimerization at the acetal position C(2) by
exposure to a variety of acidic reagents (such as ZnCl2
in methylene chloride, or trifluoroacetic acid, p-toluene-
sulfonic acid in refluxing benzene) to yield an equi-
librium mixture of cis and trans isomers (3 and 7).
Some equilibrium data illustrating the effect of the
substituents are listed in Table 1. The cis isomer is
generally favored; the cis isomer is more favored when
the N-acyl substituent is an alkoxycarbonyl group
rather than a benzoyl group (entries 1 and 2); the
nature of the alkoxy group in the alkoxycarbonyl
derivatives has little influence; a large aryl substituent
at position 2 destabilizes the cis isomer.10
tions, as well. Method
A was not suitable for
condensation of 8 with 9-anthraldehyde. Both method
A and B were not suitable for use with benzyloxycar-
bonyl-protected alanine.
With a very convenient approach to 10 in hand, we
turned to the alkylation reaction. We found that 10
could be deprotonated with lithium hexamethyldisi-
lazide and alkylated at a temperature of −27°C (much
more convenient from an industrial standpoint than
−78°C, as generally adopted in the alkylation step of
SROSC) provided the base was added slowly to a
mixture of nucleophile and electrophile. This gave 11 in
excellent yield and high (>98%) diastereomeric purity.
Much lower yields were obtained if the enolate was
pre-formed at −27°C in the absence of the electrophile,
suggesting that the lithium enolate of 10 is stable at
−78°C, but not at −27°C. Treatment of 11 with lithium
methoxide, followed by aqueous bisulfite work-up,15
gave 12, whereas 4-phenylbenzaldehyde precipitated as
its bisulfite adduct 13 and could be filtered off. Crude
12a was hydrolyzed to the free benzylated amino acid
On the basis of the above observations, we decided that
a suitable strategy to a single oxazolidinone could be
the following: (a) improve the chemical synthesis of the
oxazolidinone by maximizing the chemical yield and
minimizing the risk of racemization, without concern
for the stereochemical outcome and the ratio of iso-
mers; (b) obtain one pure stereoisomeric oxazolidinone
as a solid by devising a first or a second order asymmet-
ric transformation.11 We focused on N-alkoxycarbonyl-
2-aryloxazolidinones, rather than N-acyl analogues
because of their higher preference for the cis isomer and
a possibly easier conversion to hydantoins. A derivative
of suitable crystalline properties turned out to be 10,
for which two distinct highly efficient methods of syn-
thesis (A and B) were devised.12 According to method
Table 1. Equilibration of cis and trans oxazolidinones by C(2) epimerization (equilibration carried out with ZnCl2 in CH2Cl2
1
at rt overnight; ratios by H NMR)

E. Napolitano, V. Farina / Tetrahedron Letters 42 (2001) 3231–3234
3233
Scheme 1. (a) (COCl)2 (cat. DMF) CH2Cl2, then 4-phenylbenzaldehyde (9) and ZnCl2, 82%; (b) p-TsOH, cyclohexane reflux
(−H2O), 91%; (c) 4-bromobenzyl bromide or benzyl bromide, LiN(SiMe3)2, THF/hexane, −27°C, 100%; (d) MeOLi, MeOH, then
saturated aqueous NaHSO3, 97%; (e) 3,5-dichloroaniline, NaOMe, refluxing toluene, 64%; (f) MeI, NaOH, TBA–HSO4, 95%; (g)
HCl H2O/AcOH, 100°C, 100%.
14 (as the hydrochloride) identical with an authentic
sample. Also, the protected aminoacid esters 12 turned
out to be excellent precursors to hydantoins; 12b was in
fact converted to 1 by treatment with the sodium
derivative of 3,5-dichloroaniline (obtained in situ from
the reaction of the aniline and sodium methoxide) in
refluxing toluene,16 followed by N-methylation; this
approach to 1 (which includes only 5 steps from readily
available 8, has a 40% overall yield and >99.9%
stereoselectivity) competes favorably with the previ-
ously reported ones.3b
Chem. 1994, 59, 7671; (b) Karady, S.; Amato, J. S.;
Weinstock, L. M. Tetrahedron Lett. 1984, 25, 4337; (c)
Zydowski, T. M.; de Lara, E.; Spanton, S. G. J. Org.
Chem. 1990, 55, 5437. These approaches gave low yields
when applied to our system. See also Ref. 1.
5. Seebach, D.; Fadel, A. Helv. Chim. Acta 1985, 68, 1243.
6. Fadel, A.; Salau¨n, J. Tetrahedron Lett. 1987, 28, 2243.
7. Altmann, E.; Nebel, K.; Mutter, M. Helv. Chim. Acta
1991, 74, 800.
8. Alonso, F.; Davies, S. G.; Elend, A. S.; Haggitt, J. L. J.
Chem. Soc., Perkin Trans. 1 1998, 257.
9. Nebel, K.; Mutter, M. Tetrahedron 1998, 44, 4793.
10. These data can be rationalized if one considers, as can be
seen by simple MM2 calculation, that the relative stabil-
ity of compounds 3 and 7 is not primarily controlled by
the direct interaction of substituents at C(2) and C(4) but
by their interaction with the N-acyl substituent which
tends to be coplanar with all the carbon atoms bound to
the nitrogen. When the two substituents are cis to each
other, their steric repulsion with the N-acyl substituent is
readily minimized by a slight bending of the cycle; when
the two substituents are trans, such a possibility for
minimizing the steric repulsion is not effective. Only when
the aryl substituent at C(2) is especially large (Table 1,
entry 6), does the direct interaction with the substituent
at C(4) become significant and the cis isomer is somewhat
destabilized.
11. (a) Jacques, J.; Colbert, A. Enantiomers, Racemates and
Resolutions; Wiley-Interscience: New York, 1981; (b) See
also: Eliel, E. Stereochemistry of Carbon Compounds;
MacGraw Hill: New York, 1994; (c) For a recent review
of dynamic resolutions, see: Caddick, S.; Jenkins, K.
Chem. Soc. Rev. 1996, 447.
12. Oxazolidinone 10: Method A. The protected aminoacid 3
(10 g, 53 mmol), 4-phenylbenzaldehyde (10 g, 55 mmol)
and p-toluenesulfonic acid (0.6 g) were heated to a rapid
reflux in cyclohexane (150 mL) with continuous removal
of the water formed by means of a Dean–Stark trap.
After 18 h, a tan crystalline precipitate was formed; if
not, an authentic sample of 10 was added to induce
crystallization and the reflux continued for 8 h to ensure
equilibration. Heating was discontinued, the mixture was
allowed to equilibrate slowly at room temperature and
then cooled in an ice bath; the solid was collected by
suction; trituration with methyl tert-butyl ether gave pure
In conclusion, a crystallization-induced process was
devised by which
L-alanine was converted to the cis
oxazolidinone 10 (virtually free from its trans isomer)
suitable as an intermediate for benzylation of alanine
via the SROSC approach; the possibility of conducting
the alkylation of 10 at −27°C rather than at −78°C was
demonstrated; these findings should allow the applica-
tion of SROSC to large scale preparation of alkylated
alanines. Also, a new convenient entry to biologically
important hydantoins is disclosed.
References
1. Cativiela, C.; D´ıaz-de-Villegas, D. M. Tetrahedron:
Asymmetry 1998, 9, 3517.
2. Seebach, D.; Sting, A. R.; Hoffmann, M. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 2708.
3. (a) Kelly, T.; Jeanfavre, D. D.; McNeil, D. W.; Woska, J.
R.; Reilly, P. L.; Mainolfi, E. A.; Kishimoto, K. M.;
Nabozny, G. H.; Zinter, R.; Bormann, B. J.; Rothlein, R.
J. Immunol. 1999, 163, 5173; (b) For previous approaches
from these laboratories, see: Spero, D. M.; Kapadia, S. J.
Org. Chem. 1996, 61, 7398; Yee, N. K. Org. Lett. 2000, 2,
2781.
4. We found the Na salt of alanine to be very hygroscopic
and prone to forming a thick oil. Its conversion to the
Schiff base under heterogeneous conditions in pentane, or
higher hydrocarbons, was plagued by very long reaction
times and erratic results. Acylation gave products with
highly variable cis:trans ratios. Alternative condensation
protocols: (a) Cheng, H.; Keitz, P.; Jones, J. B. J. Org.

3234
E. Napolitano, V. Farina / Tetrahedron Letters 42 (2001) 3231–3234
10 (17 g, 91%) as off-white prisms. A sample was recrys-
then methyl t-butylether (100 mL) was added and the
slurry stirred at 4°C for 18 h. Filtration gave 10 as an
off-white solid (16.5 g, 88%).
tallized from methanol: mp 154–155°C; 1H NMR
(CDCl3) l 7.65–7.35 (m, 9H), 6.70 (s, 1H), 4.50 (q, J=7
Hz, 1H), 3.93 (d, J=6.5 Hz, 2H), 1.91 (m, 1H), 1.63 (d,
J=7 Hz, 3H), 0.90 (m, 6H); 13C NMR (CDCl3) l 173.23,
154.36, 140.82, 136.49, 129.45, 128.33, 128.06, 127.74,
127.31, 89.59, 72.99, 52.75, 28.41, 19.53, 18.89. Anal.
calcd for C21H23NO4: C, 71.36; H, 6.56; N, 3.96. Found:
C, 71.30; H, 6.50; N, 3.82.
13. This quite obvious approach is not unprecedented (see
Ref. 4b) but it has not found extensive application prob-
ably because of the poor results obtained with benzyloxy-
carbonyl protected alanine.
14. Although chlorides from N-alkoxycarbonyl amino acids
are known to be configurationally stable in the presence
of Lewis acids (Cupps, T. L.; Boutin, R. H.; Rapoport,
H. J. Org. Chem. 1985, 50, 3972) and acid chlorides are
known to give addition compounds with aldehydes (Neu-
enschwander, M.; Bigler, P.; Christen, K.; Iseli, R.;
Kyburz, R.; Muehle, H. Helv. Chim. Acta 1978, 61, 2047)
the combination of these reactions has never been used
before in the preparation of enantiopure oxazolidinones.
15. For the regeneration of aldehydes from their bisulfite
adducts, see: Kjell, D. P.; Slattery, B. J.; Semo, M. J. J.
Org. Chem. 1999, 64, 5722.
Method B. A solution of 8 (10 g, 52.8 mmol) and
dimethyl formamide (0.5 mL) in dichloromethane (50
mL) was stirred at 10°C and oxalyl chloride (7 g, 4.8 mL,
55.2 mmol) was added over 15 min. The mixture was
allowed to reach room temperature. After 1 h, 4-phenyl-
benzaldehyde (10 g, 54.9 mmol) was added in one por-
tion, followed by anhydrous ZnCl2 (0.50 g, 3.67 mmol).
After a brief induction period, a mild exothermic reaction
ensued and the solution turned from yellow to red. The
solution was stirred for 4 h at rt while purging with a
slow stream of nitrogen to remove the HCl formed. The
solution was concentrated to about half of its volume,
16. De Feoand, R. J.; Strickler, P. D. J. Org. Chem. 1963, 28,
2915.
.
.