Full Paper
versus 65.5:34.5, entries 6 and 1, respectively). This is dimin-
ished in iminium salts derived from catalyst 1, where the intra-
molecular interaction dominates. The addition of a second aryl
ring in the form of a (diphenyl)methyl unit proved ineffective
(entry 7, e.r. 60.5:39.5). Consistent with the N-methylpyrrole
study, electronic modulation of the aryl shielding group gave
significantly different catalysis outcomes. The trimethoxyphen-
yl derivative outperformed the first-generation catalyst
(entry 8), whilst the pentafluorophenyl analogue led to the
second selectivity reversal hit (e.r. 40:60, entry 9). Again, this
may be a consequence of the decreased tendency of the elec-
tron deficient aryl ring to participate in an intramolecular inter-
action, thus placing the benzylic protons above the catalyst
core. A study of the second-generation imidazolidinone scaf-
fold (entries 10–16) once more confirmed the aminal configu-
ration-dependence of selectivity in both transformations (e.r.
88:12 and 50:50, entries 10 and 11 respectively). This was a pre-
dominant factor in catalysts irrespective of the electron rich
nature of the shielding group (entries 12/13 and 14/15). Con-
sistent with the N-methylpyrrole results, the syn-diastereomers
furnished higher levels of enantioselectivity than the corre-
sponding anti systems. However, the pentafluorophenyl cata-
lyst delivered selectivities that approach those of the MacMil-
lan second-generation catalyst. Importantly, complete deletion
of the benzyl substituent was remarkably well tolerated (e.r.
23.5:76.5, entry 16).
of the working hypothesis, catalysts 21, 22 and 23 were con-
ceived (Table 2). It was envisaged that by progressively remov-
ing aromaticity (entry 1), and subsequently the steric footprint
of the shielding arm (entries 2 and 3), it would be possible to
enhance the tentative intermolecular cation–p interaction that
pre-organises the ensemble prior to CÀC bond formation.
To that end, imidazolidinones 21, 22 and 23 were prepared
from the constituent amino acids: gratifyingly the structures of
compounds 21 and 23 could be unequivocally established by
X-ray crystallography (Figure 7). The three catalysts were inde-
Figure 7. X-ray crystal structure analysis of catalysts 21 (HCl salt) and 23 (HCl
salt). Thermal ellipsoids shown at the 50% probability level.[28]
pendently exposed to trans-cinnamaldehyde and N-methylin-
dole at ambient temperature (Table 2). The analogous reactions
with N-methylpyrrole were performed in parallel as a control.
As expected, catalyst 21–23 proved to be perfectly competent
catalysts in the alkylation of N-methylpyrrole, albeit with
modest levels of enantiocontrol (up to e.r. 69:31).
Having identified catalysts 11 and 14 as lead structures in in-
verting the intrinsic sense of enantioinduction in organocata-
lytic Friedel–Crafts alkylation of N-methylindole (e.r. 36.5:63.5
and 40:60, respectively), a second iteration of molecular edit-
ing was performed (Table 2). Common to both structures is the
likely participation of the C2 C-H group (H-C-C-N+) in directing
the N-methylindole to the upper face of the p system either as
a consequence of homologation (11) or conformation (14). In
an attempt to augment the tentative aromatic interactions be-
tween the catalyst core and the substrate that forms the basis
However, switching to N-methylindole resulted in a general
inversion of the sense of enantiocontrol. This was most pro-
nounced with the l-valine derivative 22 for which an enantio-
meric ratio of 25:75 was obtained. Remarkably, this could be
enhanced to 14.5:85.5 at À558C.
The comparative analysis of N-methylpyrrole and N-methyl-
indole in Friedel–Crafts alkylations is consistent with the
notion that two distinct induction pathways are operational,
(Figure 2). This difference may be rationalised by invoking aro-
matic interactions between the substrate and the more steri-
cally congested face of the electrophile. Consequently, the C2
and C5 substituents of the imidazolidinone core pre-organise
the electron-rich N-methylindole prior to addition, thus form-
ing the basis of an induction model. Moreover, this would also
serve to increase the proximity of the reactants; a quintessen-
tial feature of enzyme catalysis.
Table 2. Application of catalysts 21, 22 and 23 in the enantioselective
Friedel–Crafts alkylation of N-methylpyrrole and N-methylindole.[a]
Catalyst
e.r.[b]
e.r.[b]
N-Me pyrrole 5
N-Me indole 4
1
69:31
39:61[c]
Compelling experimental evidence suggests that the area
above the catalyst core is key to understanding this selectivity
difference (Figure 2). However, the geometrical constraints of
this intermolecular interaction are not immediately obvious. In-
itially, it was assumed that a pincer-type model may be opera-
tional, such that several cation/CH–p interactions[9c,10a,21] would
operate synergistically to pre-organise the ensemble. However,
this would necessarily position the two p systems orthogonal
to each other, thus introduce orbital constraints which would
require a process of realignment prior to productive bond for-
mation. Alternatively, a “sticky surface” model can be envis-
aged in which multiple the CÀH bonds can interact with the
same face of the electron-rich heterocycle.
25:75[c]
2
3
63:37
14.5:85.5[d]
63.5:36.5
33:67[c]
[a] Full experimental details are provided in the Supporting Information.
[b] The product aldehydes were reduced in situ and the enantioselectivi-
ties were determined for the corresponding alcohols. [c] Selectivity rever-
sal observed. [d] Reaction performed at À558C.
Chem. Eur. J. 2015, 21, 10031 – 10038
10035
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim