The aza-[2,3]-Wittig sigmatropic rearrangement of acyclic amines: Scope and limitations of silicon assistance

Article abstract of DOI:10.1021/jo0056343

The inclusion of a C-2 trialkylsilyl substituent into allylic amine precursors allows the base-induced aza-[2,3]-Wittig sigmatropic rearrangement to proceed in excellent yield and diastereoselectivity. The rearrangement precursors require a carbonyl-based nitrogen protecting group that must be stable to excess of strong base required for the reaction. The N-Boc and N-benzoyl group are very good at stabilizing the product anion and initiating deprotonation. The migrating groups (G) need to stabilize the intial anion by resonance and require G-CH₃ pK_a > 22 in order for the initial anion to be reactive enough for rearrangement. Products 7, 29b-d,f,g, and 23 are formed with high (10-20:1) anti diastereoselectivity. Product 23 containing the morpholine amide group is useful for preparing other carbonyl derivatives.

Full text of DOI:10.1021/jo0056343

9152
J . Org. Chem. 2000, 65, 9152-9156
Th e Aza -[2,3]-Wittig Sigm a tr op ic Rea r r a n gem en t of Acyclic
Am in es: Scop e a n d Lim ita tion s of Silicon Assista n ce
J ames C. Anderson,* Alice Flaherty, and Martin E. Swarbrick
School of Chemistry, University of Nottingham, Nottingham NG7 2RD, U.K.
j.anderson@nottingham.ac.uk
Received September 4, 2000
The inclusion of a C-2 trialkylsilyl substituent into allylic amine precursors allows the base-induced
aza-[2,3]-Wittig sigmatropic rearrangement to proceed in excellent yield and diastereoselectivity.
The rearrangement precursors require a carbonyl-based nitrogen protecting group that must be
stable to the excess of strong base required for the reaction. The N-Boc and N-benzoyl group are
very good at stabilizing the product anion and initiating deprotonation. The migrating groups (G)
need to stabilize the intial anion by resonance and require G-CH3 pKa > 22 in order for the initial
anion to be reactive enough for rearrangement. Products 7, 20b-d ,f,g, and 23 are formed with
high (10-20:1) anti diastereoselectivity. Product 23 containing the morpholine amide group is useful
for preparing other carbonyl derivatives.
In tr od u ction
rearrangement proceeds with inversion of configuration
at the lithium-bearing carbon,7 in accord with precedent
in the oxygen series.8 We characterized the first acyclic
example of the aza-[2,3]-Wittig sigmatropic rearrange-
ment, and from the outset argued that the resulting aza
anion from the rearrangement needed to be stabilized
in some way to provide some thermodynamic driving
force for the reaction (eq 2, X ) H).9 The Boc protecting
group seemed to achieve the best results, although
Manabe also showed the efficacy of the tetramethylphos-
phordiamide group, with similar substrates, in terms of
yield.10 Our first example proved to be very limited, and
it was only after incorporating an anion stabilizing group
at the C-2 position of our substrate, to try and stabilize
the transition state and dictate diastereoselection, did
the reaction become more versatile, high yielding and
diastereoselective (eq 2, X ) SiMe3).11 Among the anion
stabilizing groups surveyed,11b the silyl group provided
the best rate acceleration and diastereocontrol in this
rearrangement.
Since the first observation of the [2,3]-Wittig sigmat-
ropic rearrangement (eq 1, X ) O) in only 1960,1 this
rearrangement has developed into a synthetically power-
ful and highly stereoselective carbanion rearrangement.2
Despite the seemingly trivial replacement of oxygen for
nitrogen, the aza analogue (eq 1, X ) N) remained under
developed until the beginning of the 1990s.
When trying to prepare the 3-lithio derivative of
1-benzyl-4-vinyl-2-azetidinone, Durst et al. characterized
the first example of an aza-[2,3]-Wittig rearrangement
and verified that the driving force for the rearrangement
was the relief of ring strain in going from a four- to a
seven-membered ring.3 Over 20 years later, this seminal
observation was capitalized upon first by Somfai4 and
then Coldham5 by the use of N-substituted vinylaziri-
dines as rearrangement precursors. These facile rear-
rangements were again driven by the relief of ring strain,
this time in going from an aziridine to a tetrahydropy-
ridine. The anionic aza-[2,3]-Wittig rearrangement of
acyclic precursors, which do not have a latent thermo-
dynamic driving force, have proven to be considerably
more difficult. Initial attempts were thwarted by the
competing [1,2] rearrangement pathway,6 although the
most successful attempt verified that the aza-[2,3]-Wittig
(6) (a) Reetz, M. A.; Schinzer, D. Tetrahedron Lett. 1975, 3485. (b)
Broka, C. A.; Shen, T. J . Am. Chem. Soc. 1989, 111, 2981. (c) Murata,
Y.; Nakai, T.Chem. Lett. 1990, 2069. (d) Coldham, I. J . Chem. Soc.,
Perkin Trans. 1 1993, 1275. (e) Gulea-Purcarescu, M.; About-J audet,
E.; Collignon, N.; Saquet, M.; Masson, S. Tetrahedron 1996, 52, 2075.
(7) Gawley, R. E.; Zhang, Q.; Campagna, S. J . Am. Chem. Soc. 1995,
117, 11817.
(8) (a) Verner, E. J .; Cohen, T. J . Am. Chem. Soc. 1992, 114, 375.
(b) Hoffmann, R.; Bru¨ckner, R. Angew. Chem., Int. Ed. Engl. 1992,
31, 647. (c) Tomooka, K.; Igarashi, T.; Watanabe, M.; Nakai, T.
Tetrahedron Lett. 1992, 33, 5795.
(9) Anderson, J . C.; Siddons, D. C.; Smith, S. C.; Swarbrick, M. E.
J . Chem. Soc., Chem. Commun. 1995, 1835.
(10) Manabe, S. Tetrahedron Lett. 1997, 38, 2491.
(11) (a) Anderson, J . C.; Siddons, D. C.; Smith, S. C.; Swarbrick, M.
E. J . Org. Chem. 1996, 61, 4820. (b) Anderson, J . C.; Smith, S. C.;
Swarbrick, M. E. J . Chem. Soc., Perkin Trans. 1 1997, 1517. (c)
Anderson, J . C.; Roberts, C. A. Tetrahedron Lett. 1998, 39, 159. (d)
Anderson, J . C.; Dupau, P.; Siddons, D. C.; Smith, S. C.; Swarbrick,
M. E. Tetrahedron lett. 1998, 39, 2649.
(1) Cast, J .; Stevens, T. S.; Holmes, J . J . Chem. Soc. 1960, 3521.
(2) For the most recent review, see: Nakai, T.; Mikami, K. Org.
React. 1994, 46, 105.
(3) Durst, T.; Elzen, R. V. D.; LeBelle, M. J . J . Am. Chem. Soc. 1972,
94, 9261.
(4) (a) Åman, J .; Somfai, P. J . Am. Chem. Soc. 1994, 116, 9781. (b)
Åhman, J .; J arevång, Somfai, P. J . Org. Chem. 1996, 61, 8148 and
references therein.
(5) (a) Coldham, I.; Collis, A. J .; Mould, R. J .; Rathmell, R. E.
Tetrahedron Lett. 1995, 36, 3557. (b) Coldham, I.; Collis, A. J .; Mould,
R. J .; Rathmell, R. E. J . Chem. Soc., Perkin Trans. 1 1995, 2739.
(12) (a) Wu, Y.-D.; Houk, K. N.; Marshall, J . A. J . Org. Chem. 1990,
55, 1421 and references there in. See also: (b) Mikami, K.; Uchida,
T.; Hirano, T.; Wu, Y.-D.; Houk, K. N. Tetrahedron 1994, 50, 5917.
(13) Ishikawa, A.; Uchiyama, H.; Katsuki, T.; Yamaguchi, M.
Tetrahedron Lett. 1990, 31, 2415-8.
(14) Anderson, J . C.; Flaherty, A. J . Chem. Soc., Perkin Trans. 1
2000, 3025-7.
10.1021/jo0056343 CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/18/2000

Sigmatropic Rearrangement of Acyclic Amines
J . Org. Chem., Vol. 65, No. 26, 2000 9153
N-benzoyl analogue 1a reacted cleanly with LDA (eq 3)
to give rearranged material in 84% yield with a >20:1
1
diastereomeric ratio as judged by H NMR. This result
is very similar to the rearrangement of the corresponding
N-Boc material. The N-diphenylphosphinyl 1b and N-
tosyl 1c analogues both decomposed under either of the
basic reaction conditions. Attempted rearrangement of
a precursor possessing a methyl group instead of a
protecting group led to complete recovery of starting
material.20
We believe the silyl substituent stabilizes R-negative
charge, which develops at the central, C-2 vinyl carbon
atom, during the rearrangement. Theoretical calculations
of the oxy-[2,3]-Wittig process support this,12 although
other stereoelectronic arguments could apply.13 We have
also found that the phenyldimethylsilyl substituent al-
lows facile removal of the silicon from the product
terminal alkene with a combination of t-BuOK-18-
crown-6-TBAF.14 In this paper, we wish to report the
scope and limitations of the silicon-assisted aza-[2,3]-
Wittig sigmatropic rearrangement of acyclic allylic amines.
Resu lts a n d Discu ssion
Th e Na tu r e of th e Nitr ogen P r otectin g Gr ou p . A
series of aza-[2,3]-Wittig precursors 1a -c, based upon
our initial precursor,9 but differing only by the choice of
nitrogen protecting group, were prepared according to
Scheme 1. We now use a convergent synthesis for our
precursors instead of the linear route reported earlier.11a,b
Diastereomerically pure allylic alcohol 2 was prepared
by hydroalumination15 of phenyldimethylsilylpropyne
(3),16 formation of the ate complex with MeLi, and
addition of paraformaldehyde. Treatment of 2 with a
preformed suspension of triphenylphosphine dibromide
and triethylamine in CH2Cl2 gave the corresponding
allylic bromide 4, which upon addition of the potassium
anion of the migrating group 5a -c17 furnished the
rearrangement precursors with protecting groups N-
benzoyl 1a ,18 N-diphenylphosphinyl 1b,19 and N-tosyl 1c.
From our work it would appear that a carbonyl-based
protecting group on nitrogen, inert to the basic reaction
conditions and which can provide some degree of stabi-
lization of the product anion, is essential for a successful
rearrangement.
As an extension to this methodology we speculated
whether we could rearrange cyclic substrates where the
protecting group and migrating group were tethered, as
in compounds such as 8 (Scheme 2). This could lead to
the functionalization of certain nitrogen heterocycles. The
synthesis of N-benzylisoindolinone (8, R ) Bn) is known
and proceeds via the carbonylative cyclization of N-ben-
zyl-2-bromobenzylamine using palladium(II) acetate ca-
talysis.21 Cyclization of the corresponding primary amine
is low yielding, so we hoped that the but-2(E)-enyl group
could be introduced in place of the benzyl group in the
documented synthesis of N-benzyl-2-bromobenzylamine.
It was necessary to prepare the required secondary amine
using a protection-alkylation-deprotection route. Protec-
tion of 2-bromobenzylamine as its trifluoroacetyl deriva-
tive, alkylation with crotyl bromide, and deprotection22
gave N-but-2(E)-enyl-2-bromobenzylamine (9) in 58%
overall yield (Scheme 2). An unoptimized carbonylative
cyclization furnished 8, which when treated with t-BuLi,
in an attempt to initiate an aza-[2,3]-Wittig rearrange-
ment, gave only the enamine 10 as a mixture of isomers
and recovered starting material (55%). This suggests that
although the benzylic position is more acidic, a complex-
induced proximity effect23 involving the carbonyl group
has overidden the natural thermodynamic bias of the
precursor to give kinetic deprotonation. This nonproduc-
tive deprotonation upon quenching leads to double bond
Sch em e 1a
a
Reagents: (i) DIBAL, MeLi, (HCHO)n; (ii) Br2, Ph3P, Et3N.
The aza-[2,3]-Wittig rearrangement precursors 1a -c
were subjected to our standard conditions for rearrange-
ment. Bases n-BuLi and LDA were both explored with
addition at -78 °C in the solvent mixture THF/HMPA
(4:1) and then stirring at -40 °C overnight. Only the
Sch em e 2a
(15) Metz, P.; Linz, C. Tetrahedron Lett. 1994, 50, 3951.
(16) Mao, S. S. H.; Liu, F.-Q.; Tilley, T. D. J . Am. Chem. Soc. 1998,
120, 1193.
(17) For preparation of 5a , see: Yokomatsu, T.; Arakawa, A.;
Shibuya, S. J . Org. Chem. 1994, 59, 3506. For preparation of 5b, see:
Coulton, S.; Moore, G. A.; Ramage, R. Tetrahedron Lett. 1976, 4005.
For preparation of 5c, see: Lerch, U.; Moffatt, J . G. J . Org. Chem.
1971, 36, 3686.
(18) White, E. Organic Syntheses; Wiley: New York, 1978; Collect.
Vol. VI, p 336.
a
Reagents: (i) aq NaOH; (CF3CO)2O, Py; (ii) KH, (E)-
CH3CHdCHCH2Br; (iii) LiOH‚H2O; (iv) CO, Pd(OAc)2, n-Bu3N;
(v) t-BuLi.
(19) Osborn, H. M. I.; Sweeney, J . B. Synlett 1994, 145.

9154 J . Org. Chem., Vol. 65, No. 26, 2000
Anderson et al.
Sch em e 3a
migration and starting material. This result suggests
that a N-carbonyl protecting group, in addition to provid-
ing stabilization of the product aza anion, also plays an
important part in the deprotonation step in aza-[2,3]-
Wittig sigmatropic rearrangements.
Th e Con figu r a t ion a n d Su b st it u t ion of t h e Al-
k en e. Our standard substrate during the development
of this methodology was normally an E(C)-alkene, and
we have shown that primary or secondary alkyl substit-
uents give good rearrangements in terms of yield and
diastereoselection.11a The Z(C)-alkenes are much more
difficult to synthesize, but we have shown that rear-
rangement of N-Boc-N-(2-trimethylsilylpent-2(E)-enyl)-
benzylamine gave rearranged product in 62% yield with
a diminished diastereoselection of 3:1 in favor of the syn
diastereoisomer.11d According to our transition-state model,
efficient access to the alternate syn diastereoisomer, from
the Z(C)-alkene, is complicated by conflicting steric
effects.11d However, similar types of compounds possess-
ing the relative stereochemistry this methodology finds
difficult to deliver can be prepared using Kazmaier’s [3,3]-
sigmatropic rearrangement of glycine ester enolates.24 To
complete this study, we have investigated the [2,3]-Wittig
rearrangement of terminal alkene (11) rearrangement
precursor. The synthesis of 11 followed a route analogous
to that described for 1, but started from phenyldimeth-
ylsilylethyne, in 52% overall yield. Rearrangement of the
terminal alkene precursor 11 proceeded smoothly in 78%
yield (eq 4). These results show that the rearrangement
works well for terminal alkenes and E(C)-alkenes to give
anti diastereomers such as 7.
a
Reagents: (i) G ) CN, KH; G ) CtC-CH3 or CtC-TMS, from
corresponding OH compounds with Br2, PPh3, DMF then K2CO3
and Boc2NH; (ii) TFA; (iii) KH; (iv) Me(MeO)NH‚HCl, DIEA,
DMAP, DCC; (v) PhMgCl; (vi) MeMgBr; (vii) LiAlH4.
In the first cases we looked at simple carboxyl deriva-
tives 13a -d . Precursors 13a -c were all derived from the
alkylation of a glycine derivative 14 with 4 (Scheme 3).
The oxazoline derivative 13d was synthesized from the
alkylation of N-Boc-2-(trimethylsilyl)but-2(Z)-enylamine
(15)11a with 2-chloromethyl-4,4-dimethyl-1,3-oxazoline
(16, eq 5),26 although we have prepared similar com-
pounds from the alkylation of 17, itself derived from the
condensation of N-Boc-glycine and 2-amino-2-methylpro-
pan-1-ol (eq 6). Alternative anion stabilizing groups
13e-g were synthesized in a manner similar to that for
13a -c. The amido nucleophiles were derived from mon-
odeprotection of the product from alkylation of di-tert-
butylimidodicarbonate27 with suitably activated methyl-
ene derivatives 18e-g (Scheme 3).28 It was not possible
to prepare the corresponding ketone/aldehyde derivatives
in an analogous fashion, so 13i-k were prepared from
the Weinreb amide 13h , itself derived from 13a . To
generate a totally resonance unstabilized anion we
required the tri-n-butyltin derivative 13l, which was
prepared via the alkylation of 1911b with tributylstan-
nylmethyl methanesulfonate (eq 7).29
Th e Migr a tin g Gr ou p . In our preliminary studies,9
a phenyl substituent on the migrating group was optimal.
Additional substituents on the phenyl group led to
decreased yields with no improvement in diastereoselec-
tivity.25 Other more common activating substituents,
upon deprotonation, were dormant toward rearrange-
ment, which verified that the anion at the migrating
carbon had to be as reactive as possible in these simple
systems.6c,9 Our main goal in this project was to be able
to synthesize unnatural amino acids, which meant that
the migrating groups’ substituent had to be convertible
to a carboxyl group, which we have demonstrated for the
phenyl substituent.9 The inclusion of a silyl group at C-2
greatly accelerated the rate of the reaction,11 and we have
subsequently explored the range of activating substitu-
ents on the migrating group which allow the aza-[2,3]-
Wittig rearrangement to proceed.
Rearrangement of precursors 13a -l was investigated
by treatment with strong bases n-BuLi or LDA in THF
with HMPA cosolvent at -78 °C and warming to -40 °C
or room temperature for 14 h (eq 8). The amide base was
(20) Anderson, J . C.; Padox, G. Unpublished results.
(21) Mori, M.; Chiba, K.; Ban, Y. J . Org. Chem. 1978, 43, 1684.
(22) Baker, R.; Castro, J . L. J . Chem. Soc., Perkin Trans. 1 1990,
47.
(23) Beak, P.; Meyers, A. I. Acc. Chem. Res. 1986, 19, 356.
(24) (a) Kazmaier, U. Angew. Chem., Int. Ed. Engl. 1994, 33, 998.
(b) Kazmaier, U.; Krebs, A. Angew. Chem., Int. Ed. Engl. 1995, 34,
2012. (c) Kazmaier, U.; Maier, S. J . Chem. Soc., Chem. Commun. 1995,
1991. (d) Kazmaier, U.; Schneider, C. Tetrahedron Lett. 1998, 39, 817.
(e) Kazmaier, U.; Krebs, A. Tetrahedron Lett. 1999, 40, 479.
(25) Siddons, D. C. PhD thesis, University of Sheffield, 1998.
(26) Falorni, M.; Lardicci, L. J . Org. Chem. 1986, 51, 5291.
(27) Grehn, L.; Ragnarsson, U. Synthesis 1987, 275.
(28) Walters, M. A.; Hoem, A. B. J . Org. Chem. 1994, 59, 2645.
(29) Seitz, D. E.; Carroll, J . J .; Lee, S. H.; Synth. Commun. 1983,
13, 129.

Sigmatropic Rearrangement of Acyclic Amines
J . Org. Chem., Vol. 65, No. 26, 2000 9155
Ta ble 1. Scop e of Su bstitu en ts G on Aza -[2,3]-Wittig
Rea r r a n gem en t
material for attempted rearrangements of 13i and 13j
suggested that the anions generated in these particular
substrates were too stable for rearrangement. Aldehyde
13k was too sensitive and was destroyed under the basic
rearrangement conditions. Transmetalation of 13l with
n-BuLi gave protodestannylated material at -78 °C.
Upon warming to -40 °C for 3 h only [1,2] migration of
the tert-butoxycarbonyl group could be detected (15%)
along with a trace of protodestannylated material (<5%)
and degradation.
equiv of
basea
% yieldc
(dr)d
precursor
G
conditionsb
In this survey, HMPA was essential for the highest
yield of rearrangement. We also found that the optimal
amount of base varied with substrate, but in general
excess base did not appear to be detrimental to the
reaction yield or diastereoselection. Presumably the aza
anion in the product protects the R-center from depro-
tonation/epimerisation.31
13a
13b
13c
13d
13e
13f
13g
13h
13i
13j
13k
13l
3.5
1.9
1.9
1.2
1.5
1.6
1.5
1.3
2.0
1.3
1.2
1.5
CO2H
B
B
C
15 (>20:1)
70 (>20:1)
71 (>20:1)
70e (>20:1)
67 (3:1)
43 (10:1)
77 (10:1)
20;h 22i
100i
CO2Me
CONMe2
oxazoline
CN
CtC-Me
CtC-TMS
CON(OMe)Me
COPh
A
Bf
Af,g
Bf
B/C
B/C
B/C
B/C
Aj
The results from the rearrangements with different
activating groups G define a level of stabilization for the
carbanion above which there is no rearrangement. From
comparing the pKa data of G-CH3 compounds and as-
suming that the effect of the NBoc(R) substituent for all
of the precursors 13 is identical, anion stabilizing groups
G that exert a pKa of less than about 22 on an adjacent
methyl group do not facilitate the rearrangement.32 This
can be explained by the carbanion being too stable in
these cases. However there remains an anomally that the
unstabilized anion (13l, G ) SnBu3) did not rearrange.
The failure of the carbonyl precursors 13i-k to rear-
range was felt to be a limitation to the potential synthetic
usefulness of these products. If the Weinreb amide
derivative 13h had rearranged then standard addition
of an alkyl group or hydride would have furnished the
required carbonyl derivative of the product. However,
methyl ester 20b can be reduced with DIBAL to give the
corresponding aldehyde 20k (eq 9) in 59% yield with a
small erosion in diastereomeric ratio (>20:1 to 10:1)
COMe
CHO
SnBu3
70i
0
92k
a
b
Optimal equivalents. General procedures: (A) n-BuLi, THF/
HMPA (4:1), -78 to -40 °C, 14 h; (B) as A, but with LDA; (C) as
B, but warming to room temperature, 3 h. c Isolated yield. Ratios
d
of unpurified products determined by 1H NMR. e In addition
recovered 13d , 18%. f Reaction quenched with pH 7 buffer instead
of standard MeOH. tBuLi. De-N-methoxylated precursor 13h .
g
h
i
j
k
Starting material. No HMPA, 2 h. Protodestannylated 13l.
necessary for activating groups G, which were inherently
reactive toward n-BuLi. Table 1 summarizes the results
and gives the optimum reaction conditions for the most
successful rearrangements.
Rearrangement of the dianion of 13a was not particu-
larly successful under any of the reaction conditions,
although the small amount of product formed under
general procedure A was essentially one diastereoisomer.
The simple ester and amide function (13b and 13c,
respectively) rearranged in high yield and excellent
diastereoselectivity. The major diastereoisomer was as-
signed the structure 20 based upon our transition-state
model and by analogy to our previous rearrangements.11
Additional proof was provided by a single-crystal X-ray
structure determination of the major diastereoisomer of
20c.30 Rearrangement of the oxazoline substituent 13d
proceeded in good yield and excellent diastereoselectivity,
but this product is extremely sensitive to epimerization
by acid or base and could not be purified by silica gel
chromatography. Cyano and acetylene functions all rear-
ranged in moderate to good yields, but with varying levels
of diastereoselection. The small linear cyano group of 13e
gave a diastereomeric ratio of 3:1, compared to ∼10:1 for
the alkynyl substrates 13f and 13g, which are of com-
parable size. It has been documented in related oxy-[2,3]-
Wittig rearrangement chemistry that the cyano group
can interact with the developing negative charge at C-2
in the pericyclic transition state.12 This type of interaction
would electronically favor the alternate diastereomer to
that drawn for 20. The moderate yield for 13f can be
attributed to the basic reaction conditions not being
wholly compatible with propargylic protons. Surprisingly,
the Weinreb amide analogue 13h would not rearrange
under either set of reaction conditions. Aside from
recovered starting material, only de-N-methoxylated
precursor was isolated in 20% yield. Recovered starting
To access the corresponding keto products, we sought
a more robust, but similar, activating substituent to the
Weinreb amide. Morpholine amides have been introduced
as a cheaper alternative to Weinreb amides for large-
scale preparations of ketones from acyl derivatives.33 The
morpholine amide rearrangement precursor 21 was
prepared by the standard method from the ester 14b.
Heating 14b in anhydrous morpholine for 3 days at 80
°C gave 22, which could be alkylated with 4 in 59%
overall yield (Scheme 4). Treatment of this substrate
under our normal rearrangement conditions using n-
BuLi or LDA resulted in low yields of product. We found
that moderately yielding reactions could sometimes give
(31) Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624, (see
especially pp 1640-6).
(32) We know G ) ester rearranges and that simple esters have
pKa(H2O) ∼24. G ) carbonyl does not rearrange, and simple ketones/
aldehydes have pKa(H2O) ∼20 (for pKa values, see: Bordwell, F. G.
Acc. Chem. Res. 1988, 21, 456 and references therein). So we make
the assumption that the line of reactivity vs stability is somewhere
between these two values.
(33) Martin, R.; Romea, P.; Tey, C.; Urp´ı, F.; Vilarrasa, J . Synlett
1997, 1414.
(30) Anderson, J . C.; Flaherty, A. 2000, submitted.

9156 J . Org. Chem., Vol. 65, No. 26, 2000
Anderson et al.
Sch em e 4a
migrating group (G) is also important. A migrating group
(G) which stabilizes the initial carbanion by resonance
is favorable, as G ) carboxylate anion (13a ) and unsta-
bilized carbanions derived from transmetalation of tri-
n-butylstannane 13l were ineffective in the rearrange-
ment. We also propose that the initial carbanion should
not be too stabilized. Migrating groups with G-CH3 pKa
< 22 do not rearrange. Excess base (1.2-2.0 equiv) is
optimal for high yields and a selection from n-BuLi,
t-BuLi, LDA, or KH should be made according to the
substrate and optimization. The lithium bases require
HMPA as a cosolvent (20%) for good yields, but use of
KH requires only 18-crown-6 as an additive. We believe
these additives make the initial anions more reactive by
sequestering the metal counterions. In the case of E(C)
alkenes, anti diastereoisomers (7, 20, and 23) are formed
with good stereocontrol (10-20:1), except G ) CN (13e),
which gives a diastereomeric ratio of 3:1, probably due
to deleterious stereoelectronic interactions.12 The prod-
ucts should be useful as building blocks in the synthesis
of non proteinogenic amico acids and other nitrogen
containg target molecules. Products containing the mor-
pholine amide group (23) could be useful for synthesising
other carbonyl derivatives not accessible from the rear-
rangement of their respective precursors. Enantioselec-
tive studies and applications to target synthesis are in
progress and will be reported in due course.
a
Reagents: (i) morpholine 80 °C, 3 d; (ii) KH, 4; (iii) KH, 18-
C-6; (iv) MeLi.
higher yields if a KH/18-crown-6 combination in THF was
used. For example, treatment of 13c with KH (2.4 equiv),
18-C-6 (0.5 equiv) in THF at 0 °C, with warming to room
temperature for 2 h, gave a 94% (dr >20:1) of rearranged
product 20c (cf. 13c plus LDA, 71% (dr > 20:1), Table
1). Rearrangement of 21 using KH/18-C-6 furnished the
rearranged product 23 in 72% yield with essentially
1
complete diastereoselection (>20:1 by H NMR). Treat-
ment with a representative alkyllithium, methyllithium,
gave the methyl ketone 20j in 83% yield with no erosion
in diastereoselection.
Su m m a r y. Building upon our findings that a C-2
trialkylsilyl substituent assists the aza-[2,3]-Wittig sig-
matropic rearrangement of certain acyclic allylic amines,11
we have now determined that the nitrogen protecting
group in these systems should be carbonyl based and
resistant to the basic conditions employed for this rear-
rangement. The protecting group is implicated in the
initial deprotonation of the rearrangement precursor and
stabilization of the product aza anion. The rearrangement
works well for terminal alkenes and primary and second-
ary alkyl substituted E(C) alkenes. The nature of the
Ack n ow led gm en t. We thank the EPSRC, Zeneca
Agrochemicals, and Merck Sharp and Dohme for finan-
cial support.
Su p p or tin g In for m a tion Ava ila ble: Syntheses and spec-
troscopic data of compounds 1, 2-4, 7-13, 14e-g, 17, 18,
1
20a -g,j,k , 23, including H NMR spectra of compounds 1a -
c, 4, 7, 8, 10-12, 13f,g,i-l, 20a ,b,e-g,j,k , and 23. This
material is available free of charge via the Internet at
http://pubs.acs.org.
J O0056343