3124
A. J. Pearson, X. Wang / Tetrahedron Letters 46 (2005) 3123–3126
Fe(CO)3
Fe(CO)3
(CO)3Fe
(CO)3Fe
R
R
Me
R
Me
*
*
*
NP
NP
NP
R
R
Fe(CO)
N
N
3 O
O
O
O
O
P
6
P
5
6
8
7
Scheme 2.
with their rate of interconversion (which occurs by iron-
mediated hydride transfer), then we should observe pre-
dominant formation of 7 over 8. Indeed, if the equilibra-
tion between 5 and 6 is sufficiently facile, and the
cyclization rates are greatly different, a mixture of these
two complexes might afford a single product. Note that
the diene–Fe(CO)3 unit in 6 is enantiomeric with that in
5 (see the alternate drawing of 6, which is derived by
rotation about the diene–carbonyl C–C bond). Of
course, 7 will undergo the same rearrangement as noted
for 2, but this is also a controllable event.5 Should this
process occur as proposed, then the newly generated ste-
reocentres are controlled by the stereochemistry at C*,
and the requisite amide can be prepared from a racemic
cyclohexadienoic acid complex.
duce a pair of diastereomeric amides, 14 and 15, which
were separated (Scheme 4). Complex 14 was also pre-
pared from the amine and enantiomerically pure acid
of known stereochemistry,8 which allowed unambiguous
structural assignment of 14 and 15. Cyclization of 14
gave a 1:1 mixture of epimeric spirolactams 16 in 75%
isolated yield (4 h, 100% conversion; minor by-products
appear to be from demetallation of 16). It should be
noted that, according to our earlier observations on this
reaction, 16a is the initial product and this undergoes
thermal rearrangement to give 16b. Direct cyclization
of 15 did not afford any of the products 17. Instead,
upon heating 15 rearranged to give its diastereomer
14, which then cyclized to give products 16 in 65% yield
(8 h, 60% conversion). Cyclization of a 1:1 mixture of 14
and 15 gave the same pair of epimers 16 in 63% isolated
yield. The much slower cyclization of 15 compared with
14 is consistent with the proposition that complex 15
must undergo rearrangement of the diene, to form 14
first. The mixture of epimeric lactams 16 was demetal-
lated (Me3NO), followed by hydrogenation to give a
single product 18, thus confirming that efficient dynamic
kinetic resolution indeed occurs.
The amide derivatives (X = NR) are chosen for illustr-
ation here because the requisite amines can be prepared
from the corresponding readily available amino acids.
We also chose to use a trisubstituted pendant alkene
unit, for reasons that will become apparent later in the
discussion. For this purpose, we required amine deriva-
tive 11, which was prepared from phenylalanine as out-
lined in Scheme 3. Thus, the protected amino aldehyde 9
was prepared using a literature procedure,6 and sub-
jected to Horner–Wadsworth–Emmons reaction under
two sets of reaction conditions. The phosphonyl enolate
derived by treatment of the precursor phosphonoester
with n-BuLi afforded a mixture of E and Z alkenoic
esters 10 favouring the E isomer, but in only a 4:1 ratio,
though these can be separated chromatographically.
Use of DBU/LiCl7 produced essentially single E isomer
in good yield. N-Deprotection of (E)-10 afforded 11. It
should be noted that the minor (Z) isomer from the
BuLi-promoted reaction underwent spontaneous form-
ation of lactam 12 on deprotection, as expected, so
our study of the [6+2] ene cyclization reaction was con-
fined to the E isomer at this time.
The difference in reactivity between 14 and 15 can be
rationalized by considering the strain energies of puta-
tive intermediates, which can be calculated by molecular
mechanics using PC Spartan. According to the mecha-
nism, 14 and 15 have to be converted to the intermedi-
ates 18 and 19, respectively. Intermediate 18 is lower
in energy than 19 by 1.8 kcal/mol. Formation of 19 from
15 is therefore likely to be slow compared with isomeri-
zation of 15 to 14, which prevents formation of 17a. At
the same time, conversion of 14 to 18 occurs more easily
and 18 then cyclizes to form 16a.
As mentioned earlier, we considered that a trisubstituted
alkene residue would be needed for efficient dynamic
kinetic resolution. This supposition was based on the
fact that intermediate 19 has a destabilizing non-bonded
interaction between the methyl and benzyl groups, as
The chiral trisubstituted amino ester 11 was coupled
with the racemic acid 13, via its acyl mesylate, to pro-
1) 4-MeOC6H4COCl
Bn
Bn
Bn
CO2Et
Bn
(>95%)
EtO2C(Me)CHPO(OEt)2
+
O
BocN
PMB
BocN
PMB
BocN
-Cl +H3N
CO2Et
10 (E)
CO2Me
2) LiAlH4 (98%)
Conditions:
10 (Z)
3) Boc2O, dioxane (98%)
4) Swern (70%)
(a) n-BuLi, THF, -78 ºC, 3h (74%)
(b) DBU, CH3CN, 0 ºC, 2h (76%)
Conditions: (a) 4:1 E/Z; (b) >100:1 E/Z
OMe
9
Bn
TFA, CH2Cl2
TFA, CH2Cl2
10 (E)
10 (Z)
HN
PMB
O
0 ºC, 10 min.
>95%
CO2Et
0 ºC, 10 min.
>95%
Bn
N
PMB
12
11
Scheme 3.