Journal of the American Chemical Society
Communication
ligands; however, no significant enantioselectivity was observed
(Table 1, entry 1; also see Table S1 in the Supporting
L4 were ineffective for this reaction, the other Box derivatives
K-L3, K-L5-7 worked well, and good to high levels of
enantioselectivities were obtained. Among them, the best K-
Box was K-L1, which was selected for further investigations
S5 in SI). Catalyst loading and the amount of 2a were
optimized, and the use of 1.1 equiv of 2a with 5 mol %
KHMDS and 5 mol % K-L1 system was found to be the best
conditions. The use of excess KHMDS to K-L1 still gave good
enantioselectivity. KHMDS itself also promoted the reaction,
but the yield was very low. The effect of the N-aryl group was
also investigated (Table S6 in SI). It was found that the 2-
methoxy group on the N-aryl group was important to achieve
high enantioselectivity in this reaction.
Table 1. Initial Investigation of Asymmetric Mannich
Reaction
a
b
entry
ligand
yield (%)
anti/syn
ee (%, anti)
c
1
L06
L07
L1
70
67:33
−
−
18
−
−
2
3
NR
NR
98
The substrate scope of the reaction was then examined
(Table 2). Tolyl imines were used in the reaction, and high
yields and high diastereo- and enantioselectivities were
obtained (entries 2−4). Ethyl and phenyl substituents on the
phenyl group were also effective (entries 5 and 6). In reactions
using imines bearing a methoxy group, high selectivities were
observed, but the reactivity of the p-methoxyphenyl imine was
lower (entries 7−9). Halogen-substituted imines were also
successfully employed, and the desired products were obtained
in good yields (entries 10−14). An imine bearing a CF3 group
and a 2-naphthyl imine gave the products in high yields with
high selectivities (entries 15 and 16). The pyridyl imines were
also effective; however, the 2-pyridyl imine showed lower
enantioselectivity (entries 17−19). Alkyl imines were also
available; cyclopropyl, tert-butyl, and 2-phenyl-1,1-dimethy-
lethyl imines reacted with 2a, and high enantioselectivities
were obtained (entries 20−22).69 The scope of the amide
structure was also investigated. It was found that less hindered
Box salt K-L5 was effective when longer alkylamides (2b−e)
were used, and high diastereo- and enantioselectivities were
obtained (entries 23−26). N,N-Dimethylacetamide (2f) also
showed good enantioselectivity using K-L1 as the ligand (entry
27). Other propionamides 2h−2j were further tested, and
good to high enantioselectivities were obtained (entries 28−
30). It was also found that tert-butyl propionate (2k) gave the
desired product in high yield with high enantioselectivity;
however, the diastereoselectivity was moderate (entry 31).
The DMP group on the nitrogen atom of the product was
successfully removed using cerium ammonium nitrate (CAN),
and 5aa was obtained in high yield after benzoylation without
any loss of enantioselectivity (Scheme 1, eq 1).70,71 On the
other hand, the amide part was converted into the ester under
acidic conditions to obtain β-amino ester 6aa in high yield (eq
2). In addition, the asymmetric Mannich reaction in a gram-
scale also proceeded smoothly without a decrease in reactivity
or selectivity (eq 3).
d
4
K-L1
99:1
91
a
b
Isolated yield. Determined by 1H NMR analysis of the crude
c
mixture. KHMDS (15 mol %) and L06 (16.5 mol %) in toluene.
d
The catalyst prepared from KHMDS (20 mol %) and L1 (10 mol
%).
Information (SI)). We then examined the use of typical chiral
bis(oxazoline) (Box) ligands L07, with a disubstituted
methylene tether, and L1, with an unsubstituted methylene
tether, and it was found that the reactions did not proceed at
all in either case (entries 2 and 3; see also Scheme S1 in SI).
Indeed, KHMDS reacted with L1 to form a potassium salt K-
L1 (See Chart S1 in SI). Unexpectedly, it was found that K-L1
itself was effective for chiral modification of KHMDS, and the
desired reaction proceeded smoothly in THF at −78 °C to
obtain the product 3aa in high yield with high diastereo- and
enantioselectivity (entry 4). It seemed that the chiral
potassium salt K-L1 worked as a chiral ligand; however, such
a chiral metal salt ligand was unprecedented, and it is generally
thought to be difficult to create a strict asymmetric
environment around a potassium enolate without a significant
Lewis basic coordination site.
Using an interesting catalyst system, optimization of the
reaction conditions was conducted (Table S3 in SI). The effect
of solvents on the reaction was first examined. While
cyclopentyl methyl ether (CPME) and tert-butyl methyl
ether (TBME) gave good enantioselectivities among the
ether solvents, THF was found to be the best solvent. We
then investigated potassium-Box salts K-L. While K-L2 and K-
Synthesis of SCH-48462, a cholesterol absorption inhibitor
possessing a β-lactam core, was then performed (Scheme
2).72−75 The reaction of imine 1g with amide 2g was
conducted using KHMDS and K-L5. The desired adduct
3gg was obtained in high yield with high diastereo- and
enantioselectivities. The adduct 3gg was treated with Tf2O
followed by NaOH to obtain β-lactam 7gg in good yield.76,77
After the DMP group was removed, NH-free β-lactam 8gg was
obtained, the optical purity of which was enhanced by
recrystallization. After the introduction of a p-methoxyphenyl
group, SCH-48462 was obtained in optically pure form.75
The structure of the K enolate−K-L1 complex was
investigated by DFT calculations (Figure 2).78,79 The results
5599
J. Am. Chem. Soc. 2021, 143, 5598−5604