Influence of hydroxylamine conformation on stereocontrol in Pd-catalyzed isoxazolidine-forming reactions

Article abstract of DOI:10.1021/jo8027399

Palladium-catalyzed carboamination reactions between N-Boc-O-(but-3-enyl) hydroxylamine derivatives and aryl or alkenyl bromides afford cis-3,5- and trans-4,5-disubstituted isoxazolidines in good yield with up to >20:1 dr. The diastereoselectivity observe

Full text of DOI:10.1021/jo8027399

Influence of Hydroxylamine Conformation on Stereocontrol in
Pd-Catalyzed Isoxazolidine-Forming Reactions
Georgia S. Lemen, Natalie C. Giampietro, Michael B. Hay, and John P. Wolfe*
Department of Chemistry, UniVersity of Michigan, 930 North UniVersity AVenue, Ann Arbor, Michigan,
48109-1055
jpwolfe@umich.edu
ReceiVed December 15, 2008
Palladium-catalyzed carboamination reactions between N-Boc-O-(but-3-enyl)hydroxylamine derivatives
and aryl or alkenyl bromides afford cis-3,5- and trans-4,5-disubstituted isoxazolidines in good yield with
up to >20:1 dr. The diastereoselectivity observed in the formation of cis-3,5-disubstituted isoxazolidines
is superior to selectivities typically obtained in other transformations, such as 1,3-dipolar cycloaddition
reactions, that provide these products. In addition, the stereocontrol in the C-N bond-forming Pd-catalyzed
carboamination reactions of N-Boc-O-(but-3-enyl)hydroxylamines is significantly higher than that of related
C-O bond-forming carboetherification reactions of N-benzyl-N-(but-3-enyl)hydroxylamine derivatives.
This is likely due to a stereoelectronic preference for cyclization via transition states in which the Boc
group is placed in a perpendicular orientation relative to the plane of the developing ring, which derives
from the conformational equilibria of substituted hydroxylamines.
Introduction
pyrrolidines (3b, eq 2). Both of these reactions are believed to
occur via intramolecular syn-oxypalladation or syn-aminopal-
ladation of the alkene through organized cyclic transition states
in which substituents are positioned to minimize nonbonding
interactions.6 As shown below, pseudoequatorial orientation of
the R group in transition state 2a leads to the formation of 3a.1,2a
The formation of cis-disubstituted product 3b likely occurs via
transition state 2b, in which the Boc group is positioned in the
same plane as the developing ring and the R group is placed in
a pseudoaxial position to minimize A(1,3)-strain.1,3,7
Over the past several years, our group has been involved in
the development of Pd-catalyzed alkene carboetherification and
carboamination reactions for the stereoselective synthesis of
5-membered oxygen and nitrogen heterocycles,1 including
tetrahydrofurans,2 pyrrolidines,3 imidazolidin-2-ones,4 and pyra-
zolidines.5 For example, we have transformed substrates of
general structure 1a into trans-2,5-disubstituted tetrahydrofurans
(3a) with >20:1 dr (eq 1). Similarly high diastereoselectivities
are obtained in reactions of 1b that generate cis-2,5-disubstituted
(1) For reviews, see: (a) Wolfe, J. P. Eur. J. Org. Chem. 2007, 571–582. (b)
Wolfe, J. P. Synlett 2008, 2913-2937.
(2) (a) Hay, M. B.; Hardin, A. R.; Wolfe, J. P. J. Org. Chem. 2005, 70,
3099–3107. (b) Hay, M. B.; Wolfe, J. P. J. Am. Chem. Soc. 2005, 127, 16468–
16476, and references cited therein.
(3) (a) Bertrand, M. B.; Wolfe, J. P. Tetrahedron 2005, 61, 6447–6459. (b)
Bertrand, M. B.; Leathen, M. L.; Wolfe, J. P. Org. Lett. 2007, 9, 457–460, and
references cited therein. (c) Bertrand, M. B.; Neukom, J. D.; Wolfe, J. P. J.
Org. Chem. 2008, 73, 8851–8860.
(4) Fritz, J. A.; Wolfe, J. P. Tetrahedron 2008, 64, 6838–6852, and references
cited therein.
(5) Giampietro, N. C.; Wolfe, J. P. J. Am. Chem. Soc. 2008, 130, 12907–
12911.
10.1021/jo8027399 CCC: $40.75
Published on Web 02/24/2009
2009 American Chemical Society
J. Org. Chem. 2009, 74, 2533–2540 2533

Lemen et al.
TABLE 1. Limitations of Stereocontrol2a,3a,8
and instead afforded 8 (19% yield) as the major product, along
with numerous side products.
In this paper we describe the development of conditions to
effect the conversion of N-Boc-O-(but-3-enyl)hydroxylamines
to isoxazolidines via Pd-catalyzed carboamination reactions.14,15
The reactions proceed in good yield and provide cis-3,5- or
trans-4,5-disubstituted isoxazolidines with excellent diastereo-
selectivity. Importantly, the selectivities obtained in the forma-
tion of cis-3,5-disubstituted products exceed those typically
observed in other transformations, such as dipolar cycloaddition
reactions, that generate these products.16 We also describe
experiments that suggest the high diastereoselectivities in these
transformations result from a stereoelectronic preference for
cyclization via transition states in which the Boc group is placed
in a perpendicular orientation relative to the plane of the
developing ring. This preference is in sharp contrast to other
carboamination and carboetherification processes, which appear
to proceed Via transition states similar to 2b wherein carbamate
groups are positioned in the same plane as the ring being
formed. In addition, this stereoelectronic effect, which likely
derives from the relative energies of substituted hydroxylamine
conformers, has not previously been noted in alkene difunc-
tionalization processes.
We recently reported a variant of this chemistry in which
N-benzyl-N-(but-3-enyl)hydroxylamine derivatives were con-
verted to isoxazolidine products via Pd-catalyzed alkene
carboetherification.8,9 The isoxazolidine products are of signifi-
cance due to the fact that substituted isoxazolidines are displayed
in biologically active compounds,10 and also serve as precursors
to 1,3-amino alcohols.11 However, the synthetic utility of our
carboetherification reactions that generate cis-3,5-disubstituted
products is somewhat limited due to the modest diastereose-
lectivities that are obtained. For example, the coupling of 4c
with 3-bromopyridine generated 5c in 77% yield, but with only
3:1 dr (Table 1, entry 3). A similar limitation is observed in
related tetrahydrofuran- and pyrrolidine-forming reactions of 4a
and 4b, which generated 5a and 5b with 2-3:1 dr (entries 1
and 2).
Results
In light of our prior results, we were intrigued by a report
from Dongol on the synthesis of isoxazolidines via Pd-catalyzed
carboamination reactions of N-Boc-O-(but-3-enyl)hydroxy-
lamine derivatives (e.g., 6) with aryl iodides, which generated
cis-3,5-disubstituted isoxazolidines (e.g., 7) with up to 10:1
diastereoselectivity (eq 3).12,13 Unfortunately, in our hands the
conditions described by Dongol failed to provide more than trace
Optimization and Scope. In initial experiments we examined
coupling reactions between 9 and 2-bromonaphthalene using
catalysts generated in situ from mixtures of Pd2(dba)3 and
phosphine ligands. We elected to employ NaOtBu as the base
for these studies, as NaOtBu has provided good results in related
transformations that generate other heterocycles.1,17 As shown
in Table 2, the nature of the phosphine ligand had a significant
effect on both the chemical yields and the diastereoselectivities
of these reactions. Use of the chelating phosphines dppe or (()-
BINAP, which have relatively small bite angles (e93°),18 failed
to provide significant amounts of the desired product 10 at 65
°C and gave poor yields and diastereoselectivities at 110 °C
(entries 1 and 2). Use of chelating ligands with wider bite angles
led to better results, and good stereocontrol was obtained with
Dpe-phos and Xantphos (entries 5 and 6). Both dppf and dppf-
1
amounts (ca. 4% by H NMR) of the isoxazolidine product 7,
(6) Alternative mechanisms involving alkene carbopalladation have been ruled
out on the basis of deuterium labeling experiments and observed side products. For
further discussion see: (a) Reference 2b. (b) Reference 3c. (c) Ney, J. E.; Wolfe,
J. P. J. Am. Chem. Soc. 2005, 127, 8644-8651.
(7) Bertrand, M. B.; Wolfe, J. P. Org. Lett. 2006, 8, 2353–2356.
(8) Hay, M. B.; Wolfe, J. P. Angew. Chem., Int. Ed. 2007, 46, 6492–6494.
(9) For related studies, see: Jiang, D.; Peng, J.; Chen, Y. Tetrahedron 2008,
64, 1641–1647.
(10) (a) Minter, A. R.; Brennan, B. B.; Mapp, A. K. J. Am. Chem. Soc. 2004,
126, 10504–10505. (b) Ishiyama, H.; Tsuda, M.; Endo, T.; Kobabyashi, J.
Molecules 2005, 10, 312–316.
(11) (a) Iida, H.; Kasahara, K.; Kibayashi, C. J. Am. Chem. Soc. 1986, 108,
4647–4648. (b) Fredrickson, M. Tetrahedron 1997, 53, 403–425.
(12) Dongol, K. G.; Tay, B. Y. Tetrahedron Lett. 2006, 47, 927–930.
(13) Six examples were described in ref 12 with GC yields ranging from
8% to 74% and dr from 2 to 10:1. An isolated yield was reported for only one
example.
(14) For related Pd-catalyzed carboamination reactions of N-aryl-O-(but-3-
enyl)hydroxylamines with aryl bromides that afford N-arylisoxazolidines, see:
(a) Peng, J.; Jiang, D.; Lin, W.; Chen, Y. Org. Biomol. Chem. 2007, 5, 1391–
1396. (b) Peng, J.; Lin, W.; Yuan, S.; Chen, Y. J. Org. Chem. 2007, 72, 3145–
3148.
(15) For Pd-catalyzed Wacker-type carbonylative cyclofunctionalization
reactions of N-Boc-O-(but-3-enyl)hydroxylamines, see: Bates, R. W.; Sa-Ei, K.
Org. Lett. 2002, 4, 4225–4227.
(16) Intermolecular 1,3-dipolar cycloaddition reactions between nitrones and
terminal alkenes that afford 3,5-disubstituted isoxazolidine products typically
proceed with e3:1 diastereoselectivity, and formation of mixtures of regioisomers
can also be problematic. For reviews, see: (a) Confalone, P. N.; Huie, E. M.
Org. React. 1988, 36, 1–173. (b) Gothelf, K. V.; Jorgensen, K. A. Chem. ReV.
1998, 98, 863–909. (c) Kanemasa, S. Synlett 2002, 1371-1387. For representative
examples, see: (d) DeShong, P.; Leginus, J. M.; Lander, S. W. J. Org. Chem.
1986, 51, 574–576. (e) Ali, S. A.; Senaratne, P. A.; Illig, C. R.; Meckler, H.;
Tufariello, J. J. Tetrahedron Lett. 1979, 20, 4167–4170.
(17) In ref 12 Dongol reported that use of this base led to N-arylation of the
substrate. However, during the course of our studies we have not observed
significant amounts of side products resulting from competing N-arylation.
(18) For a comprehensive list of bite angles for chelating bis-phosphine
ligands, see: Fey, N.; Harvey, J. N.; Lloyd-Jones, G. C.; Murray, P.; Orpen,
A. G.; Osborne, R.; Purdie, M. Organometallics 2008, 27, 1372–1383.
2534 J. Org. Chem. Vol. 74, No. 6, 2009

Pd-Catalyzed Isoxazolidine-Forming Reactions
TABLE 2. Ligand Effectsa
13-15). Diastereomeric ratios did not change during the course
of reaction, and the ratio of ligand to metal did not have any
effect on diastereoselectivity for cases examined (PCy3, P(tBu)3,
and P(o-tol)3).
Interestingly, the influence of ligand structure on diastereo-
selectivity in the isoxazolidine-forming transformations contrasts
with tetrahydrofuran- and pyrrolidine-forming reactions, wherein
the ligand has not been observed to affect dr.2,3 In addition, the
effect of ligand structure on the diastereoselectivity of reactions
involving 9 is much larger than that in related C-O bond-
forming carboetherification reactions of 4c that generate isox-
azolidines (e.g., Table 1, entry 3). In the latter transformations,
diastereoselectivities observed with different phosphines are
typically in the range of 1 to 3:1.8,9,19
Following our optimization studies, we explored the scope
of the isoxazolidine-forming reactions using several different
combinations of 1-substituted N-Boc-O-(but-3-enyl)hydroxy-
lamines20 and electrophilic coupling partners. As shown in Table
3, a number of different cis-3,5-disubstituted isoxazolidines were
prepared in good yields with diastereoselectivities ranging from
6:1 to >20:1 (entries 2-11). Significantly, the diastereoselec-
tivities obtained in these reactions are superior to those typically
observed when 3,5-disubstituted isoxazolidines are prepared via
nitrone/alkene dipolar cycloaddition reactions.16 A broad range
of electron-rich, electron-poor, electron-neutral, sterically hin-
dered, and heterocyclic aryl bromides proved to be suitable
coupling partners, and reactions of alkenyl halides also provided
good results. However, the reaction of 9 with tert-butyl
2-bromobenzoate proceeded in low yield (44%) due to compet-
ing Heck arylation of the substrate, and relatively low diaste-
reoselectivity (6:1) was also observed in this reaction (entry
5). Use of iodobenzene in the conversion of 9 to 19 provided
similar results as obtained with bromobenzene (entry 3), and a
substrate protected as a methyl carbamate (12) was also
efficiently transformed with high stereoselectivity (entry 6). In
most cases efforts to employ substrates bearing disubstituted
alkenes were not successful. However, the coupling of 4-bro-
mobiphenyl with 16, which contains both a disubstituted alkene
and a substituent adjacent to the oxygen atom, proceeded with
4:1 diastereoselectivity. Upon isolation, 30 was obtained in 58%
yield as a single stereoisomer (entry 15). The hydroxylamine
carboamination reactions were also effective for the stereose-
lective preparation of trans-4,5-disubstituted products (entries
12-14) from a substrate substituted at C2.
a Conditions: 1.0 equiv of 9, 1.2 equiv of 2-bromonaphthalene, 1.2
equiv of NaOtBu, 1 mol % of Pd2(dba)3, 2 mol % of ligand (chelating
bis-phosphines) or 4 mol % of ligand (monodentate phosphines), toluene
To examine the feasibility of generating enantiopure disub-
stituted isoxazolidines via Pd-catalyzed carboamination, enan-
tiomerically enriched substrate (+)-9 was prepared from (S)-
1-phenylbut-3-en-2-ol (96% ee).21 As shown below (eq 5), the
coupling of (+)-9 with bromobenzene proceeded in 90% yield
with no erosion of enantiomeric purity.22
b
(0.25 M), 65 °C, 2 h. 1H NMR yield against an internal standard
(average of two experiments). Incomplete mass balance is attributed to
small errors in NMR integration unless otherwise noted. c The reaction
was conducted at 110 °C. d A product resulting from Heck arylation of
the substrate was also formed. e The reaction was conducted for ca.
45 h, as very low conversion (<10%) was observed after 2 h. f Isolated
yield (average of two experiments). g An unidentified side product was
also formed in ca. 13% yield (NMR).
iPr, which have identical backbones but different substituents
on the phosphorus atoms, provided similar diastereoselectivities
(entries 3 and 4).
The best yield and diastereoselectivity (87%, 28:1 dr) was
obtained with the bulky, monondentate ligand P(tBu)3 (entry
12). Use of smaller trialkyl phosphines led to lower diastereo-
selectivity (entries 9-11), and monodentate triaryl phosphines
also provided unsatisfactory results (entries 7 and 8). Biphenyl-
derived phosphines did not perform as well as P(tBu)3, and
provided diastereoselectivities ranging from 7 to 13:1 (entries
N-Substituent Effects. The diastereoselectivities observed
in Pd-catalyzed carboamination reactions of 1-substituted N-Boc-
O-(but-3-enyl)hydroxylamines 6, 9, 13, and 14 (Table 3, up to
>20:1 dr) are significantly higher than those obtained in our
previously reported carboetherification reactions of 1-substituted
N-benzyl-N-(but-3-enyl)hydroxylamines (e.g., Table 1, 4c, 3:1
J. Org. Chem. Vol. 74, No. 6, 2009 2535

Lemen et al.
TABLE 3. Synthesis of Isoxazolidinesa
TABLE 4. Diastereoselectivity in Carboetherification vs.
Carboaminationa
a Conditions: 1.0 equiv of hydroxylamine, 1.2-1.4 equiv of ArBr,
1.2-1.4 equiv of NaOtBu, toluene (0.25 M), 65 or 110 °C.
b Diastereomeric ratio observed in the crude reaction mixture. c Isolated
yield of the major diastereomer (average of two experiments). d Ar )
Ph. e Ar ) 2-naphthyl. f From ref 9. We have obtained nearly identical
results. g A mixture of products resulting from either Heck arylation or
isomerization of the starting alkene was obtained. h Significant amounts
of products derived from substrate N-O bond cleavage were also
formed. i The major product was benzaldehyde O-1-phenylbut-3-enyl
oxime. j Isolated as a 12:1 mixture of diastereomers. k Isolated as a 12:1
mixture of diastereomers.
dr). To further probe the factors that influence diastereoselec-
tivity in isoxazolidine-forming reactions that generate C-N vs.
C-O bonds, we sought to determine whether the nature of the
substrate N-substituent influenced stereocontrol. To this end,
we examined the carboamination and carboetherification of 4c,
9, and 31-34 with simple aryl bromides (Table 4). We quickly
discovered that P(tBu)3, which provided optimal results in
reactions of 9, was not effective for carboetherification reactions
of 4c. As such, we elected to employ Dpe-phos and Xantphos
as ligands for the experiments shown in Table 4, as these ligands
provided the desired products in most transformations of interest,
a Conditions: 1.0 equiv of substrate, 1.2 equiv of R-Br, 1.2 equiv of
NaOtBu, 1 mol % of Pd2(dba)3, 4 mol % of P(tBu)3 ·HBF4, toluene
(0.25 M), 65 °C. b Average isolated yields obtained from two or more
experiments. c Diastereomeric ratio of the isolated material. Numbers in
parentheses represent diastereomeric ratios observed in crude reaction
mixtures. d The reaction was conducted with 1.5 mol % of Pd2(dba)3 and
9 mol % of P(tBu)3 ·HBF4 at 90 °C. e The reaction was conducted with
2.0 equiv of ꢀ-bromostyrene and NaOtBu.
2536 J. Org. Chem. Vol. 74, No. 6, 2009

Pd-Catalyzed Isoxazolidine-Forming Reactions
SCHEME 1. Mechanism
and had previously been used in the reactions of 4a-c described
in Table 1. Use of Dpe-phos as ligand for the reactions of
substrates of 4c, 9, and 31-34 provided slightly different results
than were obtained with Xantphos. However, in most cases
similar diastereoselectivities were observed with both catalyst
systems.
As shown in Table 4, the selectivities and yields of these
transformations were significantly influenced by the type of
substituent on nitrogen. For example, Pd2(dba)3/Xantphos-
catalyzed carboetherification reactions of N-benzyl- or N-phenyl
substituted N-(but-3-enyl)hydroxylamines 4c and 31 proceeded
with low diastereoselectivity (entries 2 and 4, 1-2:1 dr). The
analogous carboamination reaction of N-phenyl-O-(but-3-enyl-
)hydroxylamine 33 also proceeded with low selectivity (entry
8, 2:1 dr)23 whereas the reaction of the N-benzyl-protected
derivative 34 failed to generate an isoxazolidine product. In
contrast, the Pd-catalyzed carboetherification of N-Boc-N-(but-
3-enyl)hydroxylamine derivative 32 with 2-bromonaphthalene
afforded trans-disubstituted product 38 with relatively high
selectivity (entry 6, 7:1 dr),24,25 and the Pd2(dba)3/Xantphos-
catalyzed carboamination of N-Boc-O-(but-3-enyl)hydroxy-
lamine 9 proceeded with good selectivity for formation of cis-
disubstituted product 10 (entry 12, 12:1 dr).26 Thus, the presence
of the N-Boc protecting group appears to have a large influence
on stereoselectivity in both the carboetherification and the
carboamination reactions.
Mechanism of Heterocycle Formation and Origin of
Stereocontrol. Palladium-catalyzed carboetherification and car-
boamination reactions of γ-hydroxyalkenes (e.g., 4a) and
γ-aminoalkenes (e.g., 4b) that provide tetrahydrofuran or
pyrrolidine products are believed to proceed through the
mechanism outlined in Scheme 1.1-3,6 In these transformations,
formation of the carbon-heteroatom bond occurs via intramo-
lecular alkene syn-heteropalladation from intermediate 40 (A
) CH2, B ) O or NR2) to generate 41. This complex can then
undergo C-C bond-forming reductive elimination to afford the
observed products. We have previously suggested that the
conversion of N-benzyl-N-(but-3-enyl)hydroxylamine substrates
(e.g., 4c) to isoxazolidine products (e.g., 5c) occurs via a similar
mechanism.8 This hypothesis is supported by the observation
that transformations of a substrate bearing a cyclic internal
alkene proceed with net syn-addition of the oxygen atom and
the aryl group to the alkene.8 Moreover, the fact that reactions
involving 4a, 4b, or 4c all provided cis-disubstituted products
with similar diastereoselectivity (Table 1, ca. 2-3:1) is also
suggestive of mechanistic similarities between the three processes.
As described above, Pd-catalyzed carboamination reactions
of 9 also provide cis-disubstituted products. However, the
diastereoselectivity of these transformations is quite high (up
to >20:1) relative to the reactions of 4a-c shown in Table 1.
To rule out the possibility that the differences in diastereose-
lectivity observed in reactions of 9 were due to a change in
mechanism, we conducted deuterium-labeling experiments to
determine the stereochemistry of alkene addition. As shown in
eq 6, treatment of 42 with tert-butyl 2-bromobenzoate under
our optimized conditions afforded 43 in 39% yield as a 7:1
mixture of 3,5-cis:trans diastereomers (eq 6).27,28 Similarly, the
coupling of 44 with tert-butyl 2-bromobenzoate (eq 7) provided
45 in a modest 31% yield with 3:1 dr (3,5-cis:trans).27
Importantly, syn-addition of the aryl group and the heteroatom
across the alkene was observed in both the carboamination of
42 and the carboetherification of 44, which suggests that both
transformations proceed via the mechanism depicted in Scheme
1.
(19) Chen has reported that reactions of 4c with bromobenzene in the presence
of various ligands proceed with the following diastereoselectivities (trans:cis):
(a) Dpe-phos, 2:1; (b) Xantphos, 1:2; (c) Ph3P, 1:1; (d) dppf, 1:1; (e) BINAP,
1:1. See ref 9. We have observed that three reactions of 4c with different aryl
bromides all proceed with 3 to 4:1 dr when Dpe-phos is used as ligand. See ref
8 and Table 4, entry 1.
(20) Substrates were prepared in 45-85% overall yields from the corre-
sponding alcohols via Mitsunobu reaction with N-hydroxyphthalimide followed
by sequential treatment with hydrazine and (Boc)2O.
(21) The enantiomerically enriched alcohol was prepared via lipase-catalyzed
kinetic resolution of racemic 1-phenylbut-3-en-2-ol. See: Chalecki, Z.; Guibe´-
Jampel, E.; Plenkiewicz, J. Synth. Commun. 1997, 27, 1217–1222.
(22) The small difference between enantiomeric purity of the product (97%
ee) and the substrate (96% ee) is likely due to small errors in the integration of
HPLC traces.
(23) The Pd-catalyzed carboamination reactions of 33 also generated
significant amounts of 1-phenylbut-3-en-1-ol and other side products resulting
from substrate N-O bond cleavage. Control experiments indicated that the N-
O bond cleavage occurred upon heating the substrate in the presence of NaOtBu
with no palladium catalyst.
Discussion
Stereoselectivity Differences in Isoxazolidine-Forming
Carboetherification vs. Carboamination Reactions. As noted
above in Table 4, for a series of transformations involving
identical catalysts (e.g., Pd/Dpe-phos), the nature of the N-
substituent has a large effect on diastereoselectivity in Pd-
catalyzed isoxazolidine-forming reactions of N- and O-butenyl
(24) For related results in the synthesis of pyrazolidines via Pd-catalyzed
carboamination reactions of N-(but-3-enyl)hydrazine derivatives see ref 5.
(25) The conditions originally developed for Pd-catalyzed carboetherification
of N-benzyl-N-(but-3-enyl)hydroxylamines were not effective with N-Boc-N-
(but-3-enyl)hydroxylamine substrates (ref 8). However, use of Xantphos as ligand
with a reaction temperature of 110 °C led to the successful conversion of 32 to
38.
(26) Similar results were obtained with methyl carbamate substrate 12. The
reaction of 12 with 2-bromonaphthalene provided 98% NMR yield (with
phenanthrene as internal standard) of 22 with 12:1 dr with Xantphos as ligand,
and 96% NMR yield and 13:1 dr with Dpe-Phos as ligand.
(27) tert-Butyl 2-bromobenzoate was employed as the aryl halide coupling
partner to facilitate assignment of stereochemistry through subsequent deriva-
tization. However, use of this aryl bromide led to the formation of relatively
large amounts of side products resulting from competing Heck-arylation of the
substrates.
(28) The diastereoselectivities observed in reactions of 9 (Table 3, entry 5)
or 42 (eq 6) with tert-butyl 2-bromobenzoate were significantly lower than
diastereoselectivities obtained in the reaction of 14 with 2-bromomesitylene
(Table 3, entry 10). The origin of this difference is not completely clear, but
may be due to coordination of the ester carbonyl group to the metal during key
steps in the catalytic cycle.
J. Org. Chem. Vol. 74, No. 6, 2009 2537

Lemen et al.
SCHEME 3. Transition States for Aminopalladation of 9
SCHEME 2. Transition States for Oxypalladation of 4c
hydroxylamine derivatives. To account for these results, we
suggest that the differences observed in diastereoselectivity of
isoxazolidine-forming carboetherification vs. carboamination
reactions may derive from conformational differences in the
transition states for the heteropalladation step. In carboetheri-
fication reactions of 4c, the most favorable transition state 47
(Scheme 2) would contain pseudoequatorial C1-phenyl and
N-benzyl groups, and the nonbonded electrons on each hetero-
atom would be eclipsed with the substituents on the adjacent
heteroatom.29 An alternative transition state (46), which leads
to the trans-disubstituted product, should be higher in energy
due to axial orientation of the phenyl group. However, the
difference in energy between transition states 46 and 47 should
be relatively small, as 46 is destabilized by only a single 1,3-
diaxial interaction (between the C1 phenyl group and the alkene
C3 hydrogen atom).30 This small energy difference is consistent
with the modest diastereoselectivity (ca. 2-4:1 dr favoring cis-
disubstitution) that is observed in Pd/Dpe-phos-catalyzed reac-
tions of these substrates.8,19
SCHEME 4. Transition States for Oxypalladation of 32
The relatively high diastereoselectivity obtained in Pd/Dpe-
phos-catalyzed carboamination reactions of 9 and related
N-Boc protected substrates may be due to the fact that O-alkyl
hydroxylamines bearing N-carbonyl groups prefer to assume
conformations in which the O-alkyl substituent is oriented
orthogonally to the Boc group.31 Thus, cyclization of 9 via
transition state 50 (Scheme 3) would minimize unfavorable
eclipsing interactions between the nonbonded electrons on
the oxygen and nitrogen atoms as well as 1,3-diaxial
interactions, and would generate the observed major (cis-
disubstituted) stereoisomer. Two transition states that would
lead to the formation of the trans-disubstituted minor product
are disfavored due to a combination of steric and stereoelec-
tronic effects. Transition state 48 appears to be particularly
unfavorable as it contains a 1,3-diaxial interaction between
the phenyl group and the alkene C3 H-atom, and also suffers
from electron-repulsion between the eclipsed nonbonding
electrons on the adjacent N and O atoms. An alternative
transition state 49, which leads to the minor (trans) stereo-
isomer, minimizes this electron repulsion. However, 49 is
destabilized by two 1,3-diaxial interactions, Ph-H and Ph-
Boc, that are not present in the more favorable transition
state 50.52 Thus, the difference in energy between transition
states 48-49 and 50 is expected to be greater than that
between 46 and 47, which leads to the observed higher
diastereoselectivity in the reactions of 9 relative to 4c.33,34
The stereochemical outcome of the Pd/Xantphos-catalyzed
carboetherification of N-Boc-N-(but-3-enyl)hydroxylamine 32,25
which provides trans-disubstituted isoxazolidine 38, may also
be due to the preferred conformation of the N-Boc-hydroxy-
lamine moiety. As shown in Scheme 4, cyclization through
transition state 51 in which the C1-phenyl group is oriented in
a pseudoaxial position would minimize allylic strain interactions
with the N-Boc group, which would be oriented parallel to the
forming ring.24 In contrast, transition state 52 would suffer from
significant A(1,3)-strain.
Effect of Phosphine Ligand Structure on Diastereo-
selectivity in the Pd-Catalyzed Carboamination Reactions
of 9. As illustrated above in Table 2, the structure of the
(32) Examination of molecular models suggests there is no significant 1,3-
diaxial steric interaction between the Boc group and the alkene C3 H-atom. Both
the alkene and the N-atom are sp2-hybridized, and the π-system of the Boc group
is oriented parallel to the alkene π-system. As such, the steric bulk of the Boc
group (the carbonyl oxygen atom and the OtBu substituent) is not in close
proximity to the C3 H-atom.
(33) The low diastereoselectivity observed in carboamination reactions of
N-phenyl-substituted substrate 31 may result from twisting of the N-aryl group
out of conjugation with the nonbonding electrons on nitrogen, which leads to
pyramidalization of the nitrogen atom and cyclization via a transition state similar
to 47.
(29) Hydroxylamines prefer to adopt a conformation in which the substituents
on the nitrogen atom are eclipsed with the nonbonding electrons on the oxygen
group in order to minimize repulsion between nonbonding electron pairs on the
adjacent heteroatoms. See: Riddell, F. G Tetrahedron 1981, 37, 849–858.
(30) The observation that most ligands examined provide ca. 1-2:1 dr in
the reaction of 4c with bromobenzene is also consistent with a very small
difference in energy between these two transition states. See ref 9 and footnote
19.
(34) The similar diastereoselectivities obtained with N-Boc and N-CO2Me
protected starting materials (Table 3, entry 6 and footnote 26) presumably result
from orientation of the O-R group away from the center of reactivity. This would
minimize the effect of the R group size on reactivity and selectivity.
(31) Hartung, J.; Svoboda, I.; Fuess, H.; Duarte, M. T. Acta Cryst. Sect. C
1997, 53, 1629–1631.
2538 J. Org. Chem. Vol. 74, No. 6, 2009

Pd-Catalyzed Isoxazolidine-Forming Reactions
SCHEME 5
relatively high diastereoselectivities obtained with biaryl phos-
phines are also consistent with this model due to the hemilabile
chelating nature of these ligands.35
The effect of monodentate triaryl and trialkyl phosphine
ligand structure on diastereoselectivity in reactions of 9 appears
to result from a combination of electronic and steric properties
of the ligand. For example, both P(tBu)3 and P(o-tol)3 are large
ligands, but use of the electron-rich trialkylphosphine leads to
excellent stereocontrol (28:1 dr), whereas use of the less
electron-donating triarylphosphine provides low diastereoselec-
tivity (2:1 dr). Smaller trialkyl phosphine ligands (e.g., PCy3)
also lead to poor diastereoselectivity, but the effect of ligand
size on selectivity is subtle, as the difference in phosphine cone
angle between PCy3 (174°, 2:1 dr) and P(tBu)3 (182°, 28:1 dr)
is relatively small. At the present time, the precise origin of
these effects is unclear.
phosphine ligand has a large influence on the diastereoselectivity
of Pd-catalyzed carboamination reactions of N-Boc-O-(1-
phenylbut-3-enyl)hydroxylamine 9. This effect has not been
previously observed in related Pd-catalyzed carboamination or
carboetherification reactions that afford pyrrolidine, pyrazolidine,
imidazolidin-2-one, or tetrahydrofuran products.1-7 In addition,
although ligands do affect diastereoselectivities in isoxazolidine-
forming carboetherification reactions of N-benzyl-N-(but-3-
enyl)hydroxylamine derivatives, the observed changes are
modest.19
To summarize the ligand effects shown in Table 2, we have
grouped the ligands examined into three categories: chelating
ligands, monodentate ligands, and biaryl phosphine derivatives.35
We have also defined the observed diastereoselectivities as either
“relatively poor (2-3:1)” or “relatively good (g7:1)”.36 Using
these categories, the trends can be summarized as follows:
(a) Chelating ligands with small bite angles (e93°) give
relatively poor dr. Chelating ligands with large bite angles
(g97°) give relatively good dr.
(b) Monodentate triaryl phosphines give relatively poor dr
regardless of size. However, with monodentate trialkyl phos-
phines dr increases as ligand size increases, and large ligands
give relatively good dr.
(c) Biarylphosphine ligands give relatively good dr regardless
of size or electronics.
The effect of bite angle on diastereoselectivity is consistent
with two pathways for alkene insertion (Scheme 5): via a five-
coordinate intermediate (54) or a four-coordinate intermediate
(55), with higher selectivity occurring through the four-
coordinate species.37 Examination of molecular models suggests
that the five-coordinate species 54 suffers from an unfavorable
steric interaction between the Boc group and either the Pd-Ar
moiety (as depicted) or the ligand, which could be expected to
raise the energy of the most favorable transition state 50
(Scheme 3) and lead to lower selectivity.38 When ligands with
large bite angles are employed, the four-coordinate intermediate,
which is generated via associative ligand substitution of alkene
for one arm of the bis-phosphine ligand (54 to 55),39 is more
accessible than when tightly chelating ligands are used. The
Summary and Conclusion
In conclusion, we have developed a highly stereoselective
synthesis of disubstituted isoxazolidines via Pd-catalyzed car-
boamination reactions of N-Boc-O-(but-3-enyl)hydroxylamine
derivatives. The majority of these products are formed with
diastereoselectivities that are superior to those typically observed
in other common methods for isoxazolidine synthesis, and a
number of derivatives are easily accessible. In addition, a
combination of deuterium labeling studies coupled with sys-
tematic variation of N-substituents has revealed two significant
mechanistic details: (a) despite differences in stereoselectivity,
carboamination and carboetherification reactions that generate
tetrahydrofurans, pyrrolidines, and isoxazolidines appear to
proceed via very similar mechanisms and (b) the effect of
hydroxylamine N-substituent on conformation has a large effect
on stereocontrol. This latter observation may be relevant to
previously reported cyclization reactions of similar substrates,15
and will likely be of use in the future design and planning of
other stereocontrolled transformations.
Experimental Section
Representative Procedure for the Synthesis of Isoxazolidines.
(()-(3S*,5S*)-3-Naphthalen-2-ylmethyl-5-phenylisoxazolidine-
2-carboxylic Acid tert-Butyl Ester (10). A flame- or oven-dried
Schlenk tube equipped with a magnetic stirbar was cooled under a
stream of nitrogen and charged with Pd2(dba)3 (2.3 mg, 0.0025
mmol), P(tBu)3 ·HBF4 (2.9 mg, 0.01 mmol), sodium tert-butoxide
(29 mg, 0.3 mmol), and 2-bromonaphthalene (62 mg, 0.3 mmol).
The Schlenk tube was evacuated and refilled with nitrogen (3×).
A solution of 9 (66 mg, 0.25 mmol) in toluene (4 mL) was added
and the resulting mixture was heated to 65 °C until the starting
material was consumed as judged by GC analysis (1.25 h). The
reaction mixture was cooled to rt and treated with saturated aqueous
ammonium chloride (2 mL) and ethyl acetate (5 mL). The layers
were separated, the aqueous layer was extracted with ethyl acetate
(2 × 5 mL), and the combined organic layers were dried over
anhydrous sodium sulfate, filtered, and concentrated in vacuo. The
crude product was purified by flash chromatography on silica gel
(35) The biaryl phosphine derivatives differ from the other two classes of
phosphine ligands examined, as the nonphosphorylated aromatic ring can act as
a hemilabile coordinating group with metal-ligand bonding that differs from
that of a chelating (bis)phosphine. For further discussion, see: (a) Surry, D. S.;
Buchwald, S. L. Angew. Chem., Int. Ed. 2008, 47, 6338-6361 and reference
cited therein. (b) Barder, T. E.; Buchwald, S. L. J. Am. Chem. Soc. 2007, 129,
12003–12010. (c) Barder, T. E.; Biscoe, M. R.; Buchwald, S. L. Organometallics
2007, 26, 2183–2192.
1
to afford 90 mg (92%) of the title compound as a clear oil. H
NMR analysis of the crude reaction mixture indicated the product
(36) In most instances ligands that gave poor dr also gave relatively low
chemical yield.
(37) Analogous alkene insertions into Pd-C bonds are believed to occur
from four-coordinate 16-electron species. Direct insertion from five-coordinate
intermediates is thought to be significantly higher in energy. See: (a) Ashimori,
A.; Bachand, B.; Calter, M. A.; Govek, S. P.; Overman, L. E.; Poon, D. J. J. Am.
Chem. Soc. 1998, 120, 6488–6499. (b) Samsel, E. G.; Norton, J. R. J. Am. Chem.
Soc. 1984, 106, 5505–5512. (c) Thorn, D. L.; Hoffmann, R. J. Am. Chem. Soc.
1978, 100, 2079–2090.
(38) This effect would also be expected to raise the energy of transition state
49 leading to the minor diastereomer. However, it does not appear that the energy
of 48, which also leads to the minor diastereomer, would be significantly
perturbed.
(39) (a) Tober, M. L. In ComprehensiVe Coordination Chemistry; Wilkinson,
G., Gilard, R. D., McCleverty, J. A., Eds.; Pergamon: Oxford, UK, 1987; Vol.
1, pp 281-329. (b) Cross, R. J. AdV. Inorg. Chem. 1989, 34, 219–292.
J. Org. Chem. Vol. 74, No. 6, 2009 2539

Lemen et al.
was formed as a 28:1 mixture of diastereomers; the isolated product
was obtained with >20:1 dr following purification. Data are for
the major diastereomer. 1H NMR (400 MHz, CDCl3) δ 7.83-7.75
(m, 3 H), 7.68 (s, 1 H), 7.48-7.38 (m, 3 H), 7.35-7.30 (m, 5 H),
4.89 (dd, J ) 6.4, 10.0 Hz, 1 H), 4.67-4.58 (m, 1 H), 3.37 (dd, J
) 6.4, 13.2 Hz, 1 H), 2.99 (dd, J ) 8.0, 13.2 Hz, 1 H), 2.73-2.64
(m, 1 H), 2.12 (m, 1 H), 1.42 (s, 9 H); 13C NMR (100 MHz, CDCl3)
δ 157.5, 137.4, 135.5, 133.4, 132.2, 128.4, 128.3, 128.0, 127.8,
127.7, 127.52, 127.46, 126.5, 125.9, 125.4, 82.9, 81.8, 61.9, 42.8,
42.4, 28.1; IR (film) 2978, 1729 cm-1; MS (ESI) 412.1880
(412.1889 calcd for C25H27NO3, M + Na+).
Additional funding for these studies was provided by the Camille
and Henry Dreyfus Foundation (New Faculty Award, Camille
Dreyfus Teacher Scholar Award), Research Corporation (In-
novation Award), Eli Lilly, Amgen, GlaxoSmithKline, and 3M.
The authors also thank Wei Li for conducting preliminary
studies in this area.
Supporting Information Available: Experimental proce-
dures, spectroscopic data, descriptions of stereochemical as-
1
signments, and copies of H and 13C NMR spectra for all new
compounds reported in the text. This material is available free
of charge via the Internet at http://pubs.acs.org.
Acknowledgment. The authors acknowledge the NIH-
NIGMS (GM 071650) for financial support of this work.
JO8027399
2540 J. Org. Chem. Vol. 74, No. 6, 2009