Angewandte
Chemie
acidic amide and a second equivalent can deprotonate either
the a- or the a’-position. Addition of a chelating metal salt
(MX2) should generate either an enolate B with an endocyclic
double bond or the enolates C and D with exocyclic double
bonds. Enolate D is expected to be favored since it has
significantly lower 1,3-allyl strain.[20] While allylation of
enolate B would generate a quaternary stereogenic center,
allylation of C and/or D should result in the diastereoselective
generation of a second tertiary stereogenic center. Amino
ketones of type A are known to be rather sensitive
compounds that readily epimerize.[19,21] Therefore, the a-
deprotonation must be suppressed completely if the a’-
allylation should be (highly) stereoselective.
On the other hand, from a synthetic point of view, this group is
not very popular, since it is rather difficult to remove.[22]
Therefore, we tested other, more easily removable protecting
groups. In amino acid and peptide allylations the best results
were obtained with the trifluoroacetyl (TFA) protecting
group, whose electron-withdrawing effect is comparable to
that of the tosyl group. Indeed, the reaction of the N-
trifluoroacetylated ketone 1d proceeded nicely in good yield
but surprisingly without significant diastereoselectivity
(entry 4). Also here only the required monoallylation product
was obtained, but the stereogenic center of the amino ketone
epimerized almost completely under the reaction conditions
used, explaining also the low diastereoselectivities observed.
Apparently, in this case the well-known lability of amino
ketones could not be suppressed. Therefore, we switched to
the N-Boc protecting group as a representative of the easily
removable carbamate protecting groups. With this protecting
group, by far the best yields could be obtained, combined with
excellent diastereoselectivities, especially with the sterically
demanding side chains (entries 6 and 7). Almost the same
selectivities were obtained with methallyl carbonate as the
allylic substrate (entries 8–10). With this allylcarbonate we
also investigated the influence of the substituent on the
enolate.
In our first experiment we investigated the allylic
alkylation of N-tosylated phenylalanine-derived benzyl
ketone 1a (Table 1). The N-tosyl protecting group was
Table 1: Allylic alkylations of various different protected a-amino ketones
1.
Up to this point, all reactions had been carried out with
the phenyl-substituted enolate to favor a’-deprotonation and
chelate complex D. Now we replaced the phenyl ring by
a small methyl group to test its influence on the regio- and
stereoselectivity of the allylation (entries 11 and 12). To our
surprise, we observed no significant difference with regards to
yield and selectivitiy.
Entry Ketone PG
R1
R2
R3
Prod. Yield [%][a] d.r.[b]
1
2
3
4
5
6
7
8
1a
1b
1c
1d
1e
1 f
1g
1e
1 f
1g
1h
1i
Tos
Tos
Tos
TFA Bn
Boc Bn
Boc iPr
Boc sBu Ph
Boc Bn
Boc iPr
Bn
Me Ph
sBu Ph
Ph
H
H
H
H
H
H
H
2a
2b
2c
2d
2e
2 f
2g
91
72
87
68
93
95
96
85
73
72
78
80
95:5
94:6
97:3
58:42
88:12
96:4
97:3
94:6
94:6
96:4
94:6
93:7
Ph
Ph
Ph
So far, only selectivity issues in the amino ketone frag-
ment had been addressed, since only symmetric p-allyl–Pd
complexes were involved. With more complex and unsym-
metrically substituted allyl complexes the situation gets more
complicated since also regioisomeric products in the “allyl
fragment” can be formed. To address this issue we inves-
tigated also reactions of an alkyl- and an aryl-substituted p-
allyl complex. For its generation, both the corresponding
branched (b) and linear (l) allyl carbonates 4 and 5 were used
(Table 2). Both allylic substrates gave the linear allylation
product with high preference. With the phenyl-substituted
derivatives the linear product was obtained almost exclu-
sively, independent of the ketone used. Major differences
were observed in the yields and diastereoselectivities. Here,
the aryl-substituted allylic systems 5 were definitely superior
in both aspects. High yields and excellent selectivities were
obtained, even with the ethyl ketones (1h and 1i) (entries 9–
11).
Finally, we also investigated allylations using chiral 1,3-
disubstituted allylic substrates to access b-branched substitu-
tion products (Scheme 4). In this case a double stereodif-
ferentiation (induced and simple diastereoselectivity) can be
expected. We chose 2-(4-phenyl-3-butenyl)carbonates 8 as
allylic substrates, since these substrates react with a high
degree of regioretention. With chelated glycine ester enolates
the reaction proceeds in a highly diastereoselective fashion
with perfect retention of configuration in the allyl fragment.
While the reaction with (R)-8 (96% ee) resulted in the
Ph
Ph
Me 3e
Me 3 f
Me 3g
9
10
11
12
Boc sBu Ph
Boc iPr
Boc sBu Me Me 3i
Me Me 3h
[a] Yield of isolated product. [b] Determined by HPLC and/or NMR
analysis of the crude products.
selected to guarantee deprotonation of the rather acidic
tosyl amide, and the benzyl ketone was chosen to facilitate the
a’-deprotonation, generating a conjugated enolate. In addi-
tion, owiong to the sterically demanding phenyl group, Z
enolate D should be significantly favored over the E enolate
C. During the optimization of the reaction we varied a range
of reaction parameters and the by far best results were
obtained when LHMDS was used as a base and ZnCl2 as the
chelating metal salt. Under these conditions the desired a’-
allylation product was formed exclusively in high yield and
diastereoselective fashion (entry 1).
With these optimized conditions in hand, we next inves-
tigated the influence of the side chain, of the N-protecting
group, and of the allylic substrate used. Even with the smallest
side chain (R = Me) excellent diastereoselectivity was
obtained (entry 2), which increased with the steric demand
of the side chain (entry 3). Apparently the N-tosyl group is an
excellent protecting group for this type of ketone allylations.
Angew. Chem. Int. Ed. 2015, 54, 9120 –9123
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9121