plant Carduus crispus (welted thistle) along with crispine B (4) and
three other bicyclic isoquinoline alkaloids (Fig. 3).5
Crispine A (3) has been shown to possess certain biological
activity, i.e. it shows cytotoxic activity against HeLa human cancer
cell lines,5 whilst compound 5 and its 5-phenyl derivatives exhibit
antidepressant-like activity.6 The deoxygenated compound 5,
which constitutes the parent framework of crispines A and B, is
a known degradation product of the alkaloid norsecurinine7 and
has been previously synthesized, including an elegant route to both
enantiomers from L-malic or L-tartaric acid.8 A number of
syntheses of racemic crispine A (3) have been reported, which
utilize different protocols for the construction of the pyrrolidine
ring.9–11 Recently, the first enantioselective syntheses of this
alkaloid were described, whereby an asymmetric transfer hydro-
genation was employed in the key step,12 or an asymmetric
allylboration of cyclic imines was utilised.13
Scheme 1 Synthesis of (¡)-crispine A (3) and the parent framework
(¡)-5. Reagents and conditions: For R = H: (i) c-butyrolactone, neat,
120 uC, then (ii) SOCl2, DCM, r.t. (87% over 2 steps); (iii) t-BuOK, EtOH,
78 uC (67%); (iv) P2O5, tetraline, 207 uC then NaBH4, AcOH, EtOH, 0 uC
(45% over 2 steps). For R = OMe: (i) c-butyrolactone, PhMe, 110 uC
(72%); (iii) and (iv) POCl3, PhMe, 110 uC then NaBH4, AcOH, EtOH, 0 uC
(67% over 4 steps).
Both crispine A (3) and the less functionalised analogue 5 were
found to map well onto the template shown in Fig. 2 and hence
were predicted to be suitable candidates for chemo-enzymatic
deracemisation by MAO-N. In the case of crispine A (3), the two
methoxy groups occupy the space depicted by one of the green
arrows and the black arrow and hence it was felt that this substrate
would be a good probe of the steric limits imposed on the substrate
by the available space at the enzyme active-site.
The requisite racemic substrates (¡)-3 and (¡)-5 were prepared
as shown in Scheme 1 based upon known procedures.14–17 Each
substrate was then tested in turn with the MAO-N-5 variant amine
oxidase using previously reported assay procedures.4 In agreement
with the prediction by the model, both crispine A (3) and the
deoxygenated analogue 5 proved to be good substrates for the
variant amine oxidase MAO-N-5. Initial deracemisation reactions
were carried out at 10 mM substrate concentration with 3–4
equivalents of ammonia–borane as the reducing agent and washed
whole cells (Escherichia coli) expressing the MAO-N-5 variant
amine oxidase. The progress of the reactions was monitored by
chiral HPLC as shown in Fig. 4. The deracemisation of (¡)-5 took
only 6 hours to reach completion (e.e. = 97%) whereas (¡)-
crispine A (3) required 40 hours to reach the same enantiomeric
purity (e.e. = 97%). The longer reaction time for deracemisation of
(¡)-3 is presumably a consequence of the increased steric demand
of the substrate and might be improved by further directed
evolution of MAO-N in order to select for variants with enhanced
activity towards (¡)-3.
Fig. 4 Deracemisation of racemic 5 to (R)-(+)-5 using the MAO-N-5
variant, as monitored by chiral HPLC.
selective for oxidation of the (S)-enantiomer of the substrate as
shown in Fig. 1.1–3 Similarly the specific optical rotation of 5
([a]D = +97.0) matches very well with the value previously reported
([a]D = +97.6).8
In summary we have developed a simple template-based
mnemonic for the enzyme monoamine oxidase (MAO-N) which
we hope will serve as a useful guide to identify racemic amine
substrates that can be subjected to chemo-enzymatic deracemisa-
tion. The template is simply based upon the steric demand of the
substrate and at this stage it is not designed to take into account
other properties of the substrate which could affect reactivity.
However, as demonstrated above with the example of crispine A
(3), this mnemonic can aid in the application of deracemisation as
a tool not only to prepare enantiomerically pure chiral amine
building blocks, but also biologically active target molecules in
their own right.
The specific optical rotation obtained for crispine A (3) ([a]D
=
+88.4) compares favourably with the value reported for the
natural material ([a]D = +91.0)12 assigning the product from the
deracemisation reaction as the (R)-enantiomer. This result is in
accordance with previous observations that MAO-N is highly
We are grateful to the Centre of Excellence for Biocatalysis,
Biotransformations and Biocatalytic Manufacture (CoEBio3) at
the University of Manchester for funding (AJE, TJS),
GlaxoSmithKline (PhD studentship award to RR) and the
Biotechnology and Biological Sciences Research Council
(BBSRC) for a postdoctoral fellowship to KRB.
Fig. 3 The pyrrolo[2,1-a]isoquinoline alkaloids (3 and 4) from C. crispus
and the parent framework (+)-5.
This journal is ß The Royal Society of Chemistry 2007
Chem. Commun., 2007, 3640–3642 | 3641