Journal of the American Chemical Society
Article
copper source instead of CuBr·SMe2 (Table 1, entry 8).
Adopting ligand L1 in combination with CuTc as the optimal
catalyst, we continued with screening of LAs. While Me3SiOTf
activates quinoline toward both the C-2 and C-4 attacks, we
hypothesized that increasing the steric bulk of the LA might
reduce the background C-2 attack and direct EtMgBr toward
C-4 addition. Testing several silyl-based LAs, all with TfO− as
the counterion but with different degrees of steric hindrance
(Table 1, entries 9−12), we observed that the catalytic system
is very sensitive to this feature of the LA. When bulky LAs are
used, the C-4 attack is the main reaction pathway, albeit at the
cost of dramatically decreased reactivity and enantioselectivity.
At this point we decided to switch from silicon-based LAs to
the slightly less reactive boron-based LA BF3·Et2O, hoping that
the diminished activation of the quinoline ring toward
nucleophilic attack would allow the Cu(I) catalyst to
outcompete the noncatalytic C-2 attack. Using BF3·Et2O as
the LA in the reaction without a copper catalyst (Table 1, entry
13), only 20% of the quinoline was converted to the mixture of
C-2- (major product) and C-4-addition products, with 80% of
the substrate remaining. Much to our delight, when combining
BF3·Et2O with the L1/Cu(I) complex, the reaction proceeded
with absolute C-4 regioselectivity, providing the corresponding
product 3a with 58% isolated yield and more than 99% ee
(Table 1, entry 14).
When BF3·Et2O was used together with L4/Cu(I), excellent
regioselectivity was obtained as well, with 88% ee, but much
lower conversion (Table 1, entry 15). Settling on BF3·Et2O
and L1/Cu(I) as the best LA and catalyst system we turned to
the screening of various solvents (Table 1, entries 16−20).
Remarkably, except for THF, all the solvents tested, including
2-methyltetrahydrofuran (2-Me-THF), t-BuOMe, toluene, and
Et2O, afforded absolute regioselectivity toward the desired
addition product 3a with almost full conversion and over 99%
ee. However, the NMR yields are lower than with CH2Cl2 (see
choice.
Having established the optimal reaction conditions for this
dearomatization protocol, we addressed the problems
associated with the isolated yields of the addition product
3a. Because of the instability of the addition product, only 58%
isolated yield of 3a with a complex mixture of byproducts was
obtained, despite achieving nearly full conversion of the
substrate (Table 1, entry 14). Continued attempts to optimize
the reaction to eliminate the byproducts did not lead to an
improved isolated yield of 3a. Thus, we decided to change the
isolation strategy and instead of trapping the addition product
with acetyl chloride we attempted to reduce the addition
intermediate with BH3·THF to form the tetrahydroquinoline
4a (Scheme 2).
This strategy worked well at −78 °C, avoiding the formation
of byproducts and providing an 83% isolated yield, thus
establishing this final optimized strategy: BF3·Et2O is used as
LA in the presence of 6 mol % of chiral ligand L1 and 5 mol %
of CuTc, with CH2Cl2 as solvent at −78 °C, and after addition
of Grignard reagents, BH3·THF is used to reduce the
dearomatized intermediate.
(4a−4e). Grignard reagents containing a terminal phenyl, a
terminal double bond, or a silyl functional group are also well-
tolerated, providing the corresponding products 4f, 4g, and 4h
with high yields and over 99% ee. It is remarkable that, except
for the C-4 product obtained with i-PrMgBr (80% enantio-
meric excess), the addition of all the tested Grignard reagents
to 1a led exclusively to the C-4-addition products with over
99% ee, proving that our catalytic system has excellent control
over the reactivity as well as the regio- and enantioselectivity.
Subsequently, we explored the scope of the substrates, focusing
on investigating the effect of different substitutions on the
quinoline upon addition with EtMgBr.
Evaluating the effect of substituents at the C-6 position, we
found that addition to substrates bearing either an electron-
withdrawing group (fluoro and bromo) or an electron-
donating group (methyl and methoxy) leads to the
corresponding C-4-addition products 5a−5d with excellent
results. The ester and cyano functional groups at the C-6
position are also tolerated, generating products 5e and 5f with
excellent yields and ee, indicating that this catalytic system can
efficiently control the regioselectivity to avoid C-2 addition as
well as chemoselectivity to avoid addition to the functional
groups. C-4-addition products 5g−5j, derived from the
addition of EtMgBr to substrates bearing bromo-, methyl-,
methoxy-, and vinyl-substituents at the C-7 position,
respectively, were obtained with high yields and enantiopurities
over 99%.
Our catalytic system can even tolerate the presence of
substituents at the sterically demanding 5- and 8-positions,
providing the corresponding addition products 5k−5n with
high yields and excellent ee. Also, an example of a 6,7-
disubstituted quinoline, bearing a conjugated phenyl ring, was
successfully subjected to our catalytic system, leading to the
formation of 1,4-addition product 5o with excellent regio- and
enantioselectivity. Significantly, quinolines bearing substituents
at the C-2 position are also excellent substrates for this
reaction, allowing access to tetrahydroquinolines 5p and 5q
with multiple stereocenters with 6:1 diastereoselectivity and
high enantiomeric excess. Contrary to these findings, no
conversion was observed with C-3- and C-4-substituted
quinolines. This kind of sensitivity to the substituents aligns
with our observations in copper-catalyzed additions to α,β-
unsaturated amides and carboxylic acids, where it was shown
that additional substitutions in the α- or β-positions with
respect to the electron-withdrawing group diminishes the
reactivity of the substrates drastically and prevents copper
catalysis from operating.11
It is also important to note that the reaction rate decreases
with increased steric hindrance of either the substrate or the
Grignard reagent, as well as with enhanced electron density of
the substrate. For these reasons, increased amounts of
Grignard reagent and BF3·Et2O (3 equiv each) are required
to reach full conversion toward products 4b, 4d−4h, 5d, 5i, 5l,
5n, 5o, and 5q. When using an aryl Grignard reagent, namely,
PhMgBr, no conversion to the addition product was observed,
even in the presence of a large excess of the corresponding
Grignard reagent and BF3·Et2O. Conversely, for substrate 5e,
the amount of LA had to be reduced from 2 to 1.2 equiv to
avoid possible binding to the ester group and thus to ensure
that the addition occurs selectively at the quinoline ring. For
substrate 5j, the amount of LA was also reduced to 1.2 equiv in
order to reduce the formation of byproducts.
With the optimized conditions in hand, we started to explore
the scope of this transformation, first concentrating on the
addition of various types of Grignard reagents (Scheme 2). We
were delighted to find that, when quinoline 1a is used as the
substrate, excellent results are obtained with both linear and
sterically demanding α-, β-, and γ-branched Grignard reagents
E
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX