ACS Catalysis
Research Article
side groups were not tolerated, most likely because of steric
factors, and so this was explored in more detail with α-
substituted cyclohexanones.
observed with aldehydes and NCS. Interestingly, in a previous
work, (3R)-methylcyclohexanone was accepted by A79F-
Tf NCS to give the (1R,3’R)-product, but with the methyl
group at the 3-position, different steric restraints will be acting
In a previous work, it was reported that α-substituted cyclic
ketones yielded no products most likely because of steric
34
within the active site.
3
4
reasons. However, as α-methyl-substituted aldehydes can be
A computational docking study also suggested that the imine
intermediate formed between dopamine and (2R)-5l was
preferred in the active site, compared to (2S)-5l, with the
ing Information Figure S1.6). Because A79 is located at the
entrance of the active site, the improved reactivity of A79F
toward 5l could be explained by the bulkier amino acid residue
helping to orientate the imine intermediate into a reactive
conformation, promoting the cyclization to give 6l. The
acceptance of 5l by NCS highlights its ability to achieve both
enantioselective and stereoselective PSRs, which is chemically
challenging to achieve by other methods. In addition to 5l, 2-
ethylcyclohexanone was also used and a small amount of the
corresponding THIA was detected by LC−MS but could not
be isolated.
3
2
accepted by NCS, and the acceptance of α-substituted
cyclohexanones by NCS could be useful in developing routes
to the erythrina alkaloids, this was explored further. Initial
attempts using 2-chlorocyclohexanone led to substrate
degradation, and no THIA product was generated. However,
with 2-methylcyclohexanone 5l and WT-NCS, THIA for-
mation was noted by LC−mass spectrometry (MS). Reactions
were optimized, particularly using higher equivalents of ketone,
higher enzyme concentrations, longer reaction times, different
cosolvents (DMSO gave the highest yields), and enzyme
variants. Variants A79I, A79F, Y108F, and WT that gave rise to
some of the higher conversions after 24 h were used in
reactions for longer periods of time, leading to improved
conversions: a preparative scale reaction, for example, with rac-
5
l (A79F, 7 d) gave 6l in 15% isolated yield; longer reaction
Next, more challenging bicyclic ketone substrates were
10
times reported higher conversions of >70%, but they were
isomers could be generated, but chiral HPLC analysis
suggested that there was predominantly one diastereoisomer
present. To confirm whether one isomer of 5l was accepted, a
two-step one-pot reaction cascade was developed. Compound
explored affording a family of otchosentine 1 natural product
analogues. Again, the use of a range of NCS variants was
Remarkably, when using β-tetralone 5m, 6m was formed in
79% conversion yield and 86% ee using A79I. Using other β-
tetralones 5n-5p, good conversions and isolated yields were
achieved, together with similar ees. Indeed, 6-substituted-β-
tetralones with electron-withdrawing chloro- or bromo-groups
gave 6n with WT-NCS and 6o with A79I in 50 and 54%
isolated yields, respectively. The use of 5p with an electron-
donating group at the 7-methoxy-position was also readily
accepted, giving 6p in 55% isolated yield with M97F. The
purification of THIAs 6m-6p was readily achieved by acid−
base extraction procedures without the requirement for
chromatographic purifications, enabling the facile scalability
of such approaches. All NCS variants used, however, failed to
(
2R)-5l was prepared from 2-methyl-2-cyclohexen-1-one 7
using the ene-reductase NCR from Zymomonas mobilis
41
expressed in E. coli (and NADPH and G6PDH cofactor
recycling system), and then A79F-TfNCS was added to give 6l
(Scheme 4a). Both products formed using (rac)-5l and (2R)-5l
Scheme 4. Use of 5l and Key NMR Analyses to Establish the
Stereochemistry of 6l
(
Supporting Information Figure S1.2), presumably because
steric reasons when entering the active site. Assignment of the
major enantiomer shown (Scheme 5) was tentatively made
based upon preference for formation of the (S)-isomer with
25,26
aldehydes.
2-Indanone 5q was readily accepted to give 6q
in 76% conversion yield and 59% isolated yield.
To understand better the results with these challenging
bicyclic ketones, some docking experiments were performed.
Here, both the affinity energies and key distances of the five
with imine intermediates toward 6m and 6o were determined.
Previous works have described the ‘dopamine-first mechanism’
where the Pictet−Spengler cyclization is triggered by the
deprotonation of the (dopamine-derived) meta-OH by Lys122,
(
a) A79F-TfNCS-catalyzed PSR between 4 and (rac)- 5l or (2R)-5l
generated from 2-methyl-2-cyclohexen-1-one and the ene-reductase
NCR; (b) C-2 Me signals in the H NMR spectra of 6l starting from
(
1
1
1
rac)-5l (green) and (2R)-5l (red); and (c) Key H- H NOESY
42
correlations in 6l produced from (rac)-5l.
and the crystal structure containing a mimic highlighted this
1
35,36
gave the same single product by H NMR spectroscopy (see
key interaction with bond distances of ∼3 Å.
Notably, with
1
the methyl signal region in Scheme 4b). H NMR
spectroscopic analysis revealed key Nuclear Overhauser effects
NOEs) between 2-H and 8’-H and 3’-H and 2-CH , and
because (2R)-5l was readily accepted by A79F-TfNCS and the
equatorial C-2-methyl substituent is more likely, the stereo-
chemistry of 6l was assigned as (1’S,2R), consistent with the
NOE data (Scheme 4c) and the S-selectivity normally
both intermediates modeled (Figure 2), they fitted into the
active site of all the structures, but A79I gave either the
shortest distance (∼3 Å) from meta-OH to the Lys122 residue
combined with a folded conformation (6o) or one of the best
binding affinities and a folded conformation (6m) with meta-
OH to Lys122 approaching ∼4.0 Å when a weak interaction
occurs. Indeed, the modeling with L76V to give 6m gave a less
(
3
1
34
ACS Catal. 2021, 11, 131−138