Hz). While both LiTMP and LDA mediated dienolate
formation, the isolated yields of 8a were found to be higher
if anhydrous LiCl (6 equiv) was included in the reaction
mixture (entries 1/3, cf. entries 2/4). The role of the LiCl
may be to form a more reactive base by deaggregation of
the lithium amide,15 or it may be involved in solvation and
stabilization of the dianionic dienolates.16
products with benzylic, heterobenzylic, or allylic substituents
on the vinylglycine, no isomerization of the olefin into
conjugation with the adjacent π-system was observed (entries
1-4 and 8). Also noteworthy is the fact that the reaction
works for both N-Cbz and N-Boc carbamates (entries 1 and
2).
Substrate 6f returned a 6:1 mixture of the deconjugated
product 8f and starting material. This could be a consequence
of (a) incomplete deprotonation of 6f, (b) a mixture of R-
and γ-protonations, or (c) isomerization of the initially
formed kinetic product on workup or isolation. This was
probed by quenching the reaction with H2O and D2O. The
latter reaction gave 2-deuterated 8f exclusively, whereas the
former gave a ca. 6:1 mixture of 8f/6f. Further, on standing
(21 days), the crude NMR samples equilibrated to a ca. 2:3
mixture of 8f/6f. This equilibration could be accelerated by
addition of triethylamine (equilibration complete in ca. 2
days). Additionally, treatment of 6f with triethylamine also
resulted in the formation of a final 2:3 mixture of 8f/6f,
demonstrating that the process is truly at equilibrium. Taken
together, these results suggest that the initial proton quench-
ing occurs exclusively at the R-position, but that base-
mediated equilibration through enolization is facilitated by
the presence of the phenyl group as an electron-sink (by
comparison, no equilibration of 8a to 6a was seen on
treatment with triethylamine over 3 days). In the case of D2O
quenching, presumably a kinetic isotope effect slows the
equilibration sufficiently for the single regioisomer product
8f to be observed initially. A similar phenomenon was
observed in the deconjugation of 6h, where 49% of a 7:1
mixture of 8h and the isomerized dehydroamino acid methyl
2-(Cbz-amino)hexa-2,5-dienoate 9h was isolated, together
with 39% of 6h. The instability of the mixture of 8h/9h
prevented investigation of the origin of the isomers, but it is
possible that the terminal ethenyl substituent activates initially
formed 8h toward isomerization, analogously to the phenyl
substituent in 6f. Finally, attempted formation of protected
vinylglycine itself by deconjugation of 6i gave a 1:1 mixture
of 6i/8i. Quenching with D2Ã followed by aqueous workup
gave a 2:1 mixture of 2-deuterated 8i and nondeuterated 6i,
which isomerized to a 1:7 mixture on standing. Again, it
seems likely that a completely or largely regioselective
protonation at C2 is followed by isomerization.
Attempted reaction with lithium hexamethyldisilazide as
the base returned only starting material (entry 5), perhaps
reflecting the lower basicity of this reagent by comparison
with LDA and LiTMP.17 The crucial need to employ
dianionic species to mediate the deconjugation was illustrated
by the attempted deconjugation of the N-methyl derivative
of 6a, which returned only a 5% yield of the target
vinylglycine (entry 6). Mass spectral evidence for the
formation of Michael adducts of the lithium amide base to
this substrate was observed. Clearly, the initial deprotonation
of the secondary carbamate in the parent substrate 6a
generates a metalated enamine, which is resistant to Michael
addition and hence an efficient substrate for dienolate
formation. Finally, although good results were obtained in
the above series using 2.1 equiv of base, to ensure complete
deconjugation across a range of substrates, we found that
the use of 3 equiv of LiTMP was beneficial and allowed the
reactions to be scaled reliably (entry 7).
We next examined the scope and generality of the method
(Table 2) with respect to the dehydroamino acid substituents.
Table 2. Scope of the Synthesis of (E)-Vinylglycines 8
entry
product
R
yield (%)
1
2
3
4
5
6
7
8
9
10
8a
-CH2Ph
-CH2Ph
-CH2(p-MeOC6H4)
-CH2(5-Me-2-furyl)
-C8H19
-4-N-Boc-piperidinyl
-Ph
-CH2CHdCH2
-CHdCH2
82
70
75
81
99
71
98
90
49
e
8aaa
8b
8c
8d
8e
8f/6fb
8g
8h/9hc
8i/6id
As stated above, in all cases, a single (E)-olefin was
formed as the product. This can be rationalized by consider-
ing deprotonation of a lithium-coordinated intermediate
lithiated carbamate 10 through the transition state a, which
minimizes A1,3-strain by comparison with the alternative
transition state b, which would give rise to (Z)-vinylglycines
(Figure 3).
-H
a Substrate 6aa/product 8aa are N-Boc derivatives. b 6:1 mixture of 8f/
6f. c 7:1 mixture of 8h/9h plus 39% 6h isolated. d 1:1 mixture of 8i/6i by
1H NMR. e Isolated yield not recorded; reaction analyzed by crude 1H NMR.
Under the optimized conditions described (Table 1, entry
7), alkyl-substituted vinylglycines were isolated in excellent
yields as single (E)-stereoisomers (Table 2, entries 1-6 and
8). Particularly reassuring was the observation that, in
We also attempted to prepare trisubstituted vinylglycines
11 by deconjugation of 6j and 6k. Treatment under the
standard deconjugation conditions simply returned starting
material, and attempts to promote dienolate formation by
extended reaction times led to the formation of byproducts.
(15) De Pue, J. S.; Collum, D. B. J. Am. Chem. Soc. 1988, 110, 5518-
5524; ibid., 5524-5533.
(16) Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624-1654.
One of these (from 6k) was isolated and identified as 12,
apparently arising through a directed metalation of the
i
(17) pKa values in THF/HMDS ) 29.5; Pr2NH ) 35.7; TMP ) 37.3.
Fraser, R. R.; Mansour, T. S. J. Org. Chem. 1984, 49, 3442-3443.
Org. Lett., Vol. 7, No. 24, 2005
5435