RESEARCH
an ester group) might hinder access of the cyclo-
propene to the iron-carbenoid; and (iv) the protein
would also need to exert precise stereocontrol
over the second carbene transfer step, regard-
less of structural differences between the ini-
tial alkyne and the cyclopropene intermediate.
Despite these challenges, we decided to inves-
tigate whether a starting enzyme with this unusual
and non-natural activity could be identified, and
whether its active site could be engineered to
create a suitable environment for substrate bind-
ing, intermediate stabilization, and selective pro-
duct formation.
BIOCATALYSIS
Enzymatic construction of highly
strained carbocycles
Kai Chen, Xiongyi Huang, S. B. Jennifer Kan, Ruijie K. Zhang, Frances H. Arnold*
Small carbocycles are structurally rigid and possess high intrinsic energy due to their ring
strain. These features lead to broad applications but also create challenges for their
construction. We report the engineering of hemeproteins that catalyze the formation of
chiral bicyclobutanes, one of the most strained four-membered systems, via successive
carbene addition to unsaturated carbon-carbon bonds. Enzymes that produce
cyclopropenes, putative intermediates to the bicyclobutanes, were also identified. These
genetically encoded proteins are readily optimized by directed evolution, function
in Escherichia coli, and act on structurally diverse substrates with high efficiency and
selectivity, providing an effective route to many chiral strained structures. This
biotransformation is easily performed at preparative scale, and the resulting strained
carbocycles can be derivatized, opening myriad potential applications.
We first tested whether free heme [with or
without bovine serum albumin (BSA)], which
is known to catalyze styrene cyclopropanation
(27), could transfer carbene to an alkyne. Reac-
tions using ethyl diazoacetate (EDA) and phenyl-
acetylene (1a) as substrates in neutral buffer (M9-N
minimal medium, pH 7.4) at room temperature,
however, gave no cyclopropene or bicyclobutane
product. Next, a panel of hemeproteins—including
variants of cytochrome P450, cytochrome P411
(P450 with the axial cysteine ligand replaced by
serine), cytochrome c, and globins in the form
of E. coli whole-cell catalysts—were tested for the
desired transformation under anaerobic conditions
(32), but none were fruitful (Fig. 1C and table S1).
Interestingly, a P411 variant obtained in a previous
cyclopropanation study, P411-S1 I263W (see sup-
plementary materials for sources, sequences, and
mutations), afforded a furan product (3b) with
a total turnover number (TTN) of 210. Because
other furan analogs have been identified as ad-
ducts of carbenes and alkynes (33), we were
curious as to how furan 3b was generated. Pre-
liminary kinetic study of the enzymatic reaction
suggested that the enzyme first synthesized an
unstable cyclopropene (3a), which subsequently
rearranged to the furan either spontaneously or
with assistance from the enzyme (Fig. 1B and fig.
S5). This result provided strong evidence that the
P411 hemeprotein is capable of transferring a car-
bene to an alkyne, which is, to our knowledge, an
activity not previously reported for any protein
or even any iron complex.
To divert the enzymatic reaction to bicyclobu-
tane formation, the enzyme would have to trans-
fer a second carbene to cyclopropene intermediate
3a before the cyclopropene rearranges to the
undesired furan product (Fig. 1B). We thus tested
P411 variants closely related to P411-S1 I263W.
We reasoned that amino acid residue 263, which
resides in the distal pocket above the heme co-
factor, might modulate the rate of this step, and
that the bulky tryptophan (Trp) side chain at this
site may be blocking the second carbene transfer.
A P411-S1 variant with phenylalanine (Phe) in-
stead of Trp at this position (263F) in fact cata-
lyzed bicyclobutane formation at a very low level
(<5 TTN) (table S1). Variant P4 with three ad-
ditional mutations relative to P411-S1 I263F (V87A,
A268G, and A328V) (28) synthesized the desired
bicyclobutane 2a with 80 TTN and with the for-
mation of furan adduct substantially suppressed
(2a:3b > 50:1; Fig. 1C). Another related P411 var-
iant, E10 (= P4 A78V A82L F263L), which was
n cyclic organic molecules, ring strain arises
from distortions of bond angle and bond length,
steric clashes of nonbonded substituents, and
other effects (1). The simplest carbocycles, cy-
clopropanes and cyclobutanes, possess ring
ing, with multiple chiral centers generated at the
same time (14, 15) (fig. S2). Cyclopropene syn-
thesis through enantioselective single-carbene
addition to alkynes also requires chiral transition
metal catalysts based on rhodium (16, 17), irid-
ium (18), and cobalt (19). Development of a sus-
tainable catalytic system that performs with high
efficiency and selectivity under ambient condi-
tions would be a major advance for construction
of these useful, highly strained carbocycles.
I
strains of 26 to 28 kcal/mol (2). Introducing carbon-
carbon multiple bonds or bridges to these small
ring systems induces additional strain as well as
structural rigidity. For example, cyclopropenes
with an endo-cyclic double bond bear a strain
of 54 kcal/mol, whereas bicyclo[1.1.0]butanes,
folded into puckered structures, distinguish them-
selves as one of the most strained four-membered
systems, with strain of ~66 kcal/mol (fig. S1) (2).
These carbocycles are particularly attractive inter-
mediates in chemical and materials synthesis,
because they can undergo strain-release trans-
formations to furnish a myriad of useful scaffolds
Enzymes, the catalytic workhorses of biology,
are capable of accelerating chemical transforma-
tions by orders of magnitude while exhibiting
exquisite control over selectivity (20). Although na-
ture synthesizes various cyclopropane-containing
products (21), cyclopropene or bicyclobutane frag-
ments are extremely rare (fig. S3) (22, 23). This
may be attributed to the lack of biological machin-
ery for synthesizing these motifs and/or the in-
stability of these structures under biological or
natural product isolation/purification conditions.
Nonetheless, we envisioned that existing enzymes
could be repurposed to forge strained carbocycles
by taking advantage of their catalytic promiscuity
(24, 25) in the presence of non-natural substrates
and by using directed evolution to optimize the ac-
tivity and selectivity of these starting enzymes (26).
In the past several years, we and others have
engineered natural hemeproteins to catalyze re-
actions not known in nature (27–32). We hypoth-
esized that carbene transfer to triple bonds with
a heme-dependent enzyme might afford highly
strained cyclopropene and bicyclobutane struc-
tures and might do so enantioselectively. We
anticipated several challenges at the outset, es-
pecially in chiral bicyclobutane formation, as it
involves two sequential carbene additions to
the alkyne substrate: (i) The enzyme would need
to bind the alkyne in a specific conformation in
order to transfer the carbene enantioselective-
ly; (ii) the high-energy cyclopropene interme-
diate generated by the first carbene addition
would need to be accepted and stabilized by
the protein; (iii) relative to methylene carbene
used previously, a substituted carbene (e.g., with
(3–6). The structural rigidity imparted by strained
rings in supramolecular materials can lead to in-
teresting physical properties, such as mechanical
stability (7) and high glass transition temperature
(8). The intrinsic energy of these strained struc-
tures can also be relieved in response to exog-
enous force, which leads to radical changes in
physical properties (e.g., conductivity), a feature
highly desirable for stimulus-responsive materials
(9, 10).
High ring strain, however, greatly increases the
difficulty of synthesis. A commonly used method
for preparing bicyclobutanes starts from dibromo-
2-(bromomethyl)cyclopropane substructures and
uses organolithium reagents for lithium-halogen
exchange, followed by nucleophilic substitution
under rigorously anhydrous and cryogenic con-
ditions (3). An alternative route relies on the
double transfer of a carbene to alkynes, but the
few examples in the literature are mostly limited
to methylene carbene (11–13). Asymmetric bicy-
clobutane construction is particularly challeng-
Division of Chemistry and Chemical Engineering 210-41,
California Institute of Technology, Pasadena, CA 91125, USA.
*Corresponding author. Email: frances@cheme.caltech.edu
Chen et al., Science 360, 71–75 (2018)
6 April 2018
1 of 5