RESEARCH
| REPORT
oxidation pathway (Fig. 1B). This could be achieved
by stabilizing the reactive intermediates suffi-
ciently to promote a 1,2-hydride migration (step v
in Fig. 1B) and selectively generate the anti-
Markovnikov oxidation product (fig. S2). Here
we present the directed evolution of an iron-
heme–dependent aMOx that oxidizes a range
of styrenes to their anti-Markovnikov carbonyl
compounds, using O2 as the terminal oxidant.
To identify a suitable starting point for evolu-
tion of an aMOx, we examined a P450 from the
rhodobacterium Labrenzia aggregata (P450LA1).
P450LA1 has been described as a promiscuous
sulfoxidation catalyst that also generates the
anti-Markovnikov carbonyl compound in the ep-
oxidation of styrene (18). In contrast to this
report, which proposed that an epoxidation-
isomerization sequence led to side-product for-
mation, we discovered that P450LA1 catalyzes
direct anti-Markovnikov oxidation without an
epoxide intermediate (fig. S3). This direct anti-
Markovnikov oxidation of styrene proceeds with
a total turnover number (TTN) of 100 for alde-
hyde product formation, which exceeds the pro-
ductivity of most synthetic catalysts reported
(table S1). However, performance is limited
by poor selectivity, because in addition to anti-
Markovnikov oxidation, the epoxide makes up
more than half of the product (55%).
tables S2 and S3). This aMOx performs direct
anti-Markovnikov oxidation of styrene with a
TTN of 3800 and 81% selectivity (Fig. 2A). The
enzyme’s activity is improved 38-fold over the
wild-type protein and is two orders of magnitude
more efficient than previously reported catalysts
(table S1). In contrast to small-molecule catalysts,
aMOx can catalyze this aerobic oxidation using
earth-abundant iron in its native heme cofactor.
In agreement with the mechanism proposed
by Groves and Myers (7), we suggest that a 1,2-
hydride migration is the central element of the
aMOx catalytic cycle. We obtained support for
the overall mechanism by determining that the
corresponding epoxides (R)-3 and (S)-3 are not
converted by aMOx and are therefore not inter-
mediates in the cycle (Fig. 2B and fig. S6). The time
course of the reaction also shows constant product
ratios over time (fig. S7). Thus, we are confident
that aMOx catalyzes a direct anti-Markovnikov
oxo transfer and that directed evolution did
not optimize the catalyst for an epoxidation-
isomerization sequence. We also confirmed the
1,2-hydride migration by converting the isotopi-
cally labeled substrate 4 and analyzing the prod-
lective anti-Markovnikov oxidation of prochiral
1,1-disubstituted alkenes, such as a-methylstyrene
(6), which are particularly challenging starting
materials in asymmetric synthesis (22). Even
though 6 was not used for screening, and the
enzyme therefore did not undergo selection for
enantioselectivity during directed evolution, aMOx
performed the reaction with good selectivity to
the corresponding (S)-enantiomer (enantiomeric
ratio, 91:9; Fig. 2D) (23, 24). We propose that
the asymmetric induction occurs during the 1,2-
hydride migration. Catalytic enantioselective 1,2-
migrations of prochiral carbocations have been
demonstrated only for the migration of alkyl
groups, and often in ring-strain releasing pro-
cesses (25, 26). Such asymmetric 1,2-migrations
are difficult because the catalyst must differenti-
ate the prochiral faces of the highly reactive,
planar carbocation intermediates. The observed
enantioselectivity suggests that aMOx locks the
substrate in a specific conformation that aligns
one of the C–H bonds coplanar to the empty p
orbital of the carbocation intermediate (Fig. 2E).
These results further suggest that the mutations
identified during evolution enhanced the aMOx
cycle by fine-tuning the orbital alignment for the
1,2-hydride migration. A role for precise optimi-
zation is supported by the subtle steric changes
introduced by many of the active-site amino acid
1
uct by H and 13C nuclear magnetic resonance
(Fig. 2C and fig. S8).
Because enzyme active sites are chiral, we
reasoned that aMOx should catalyze enantiose-
We sought to improve the P450LA1 enzyme’s
activity and selectivity for anti-Markovnikov oxi-
dation of styrene by directed evolution (19). An
aldehyde-specific colorimetric reagent (fig. S4)
was used to screen libraries in which the enzyme’s
heme domain was randomly mutated using error-
prone polymerase chain reaction. Four rounds of
mutagenesis and screening yielded the quintuple
mutant P7 (P450LA1 T121A-N201K-N209S-Y385H-
E418G), and TTN for anti-Markovnikov oxidation
increased from 100 to 1200 (Fig. 2A and tables S2
and S3). All five mutations of P7 contribute to the
improved activity, but only the T121A mutation,
believed to be in the active site based on a homol-
ogy model of the heme domain structure (fig. S5),
enhanced selectivity for the anti-Markovnikov
product (from 45 to 55%). Apparently, screening
libraries solely based on desired aldehyde forma-
tion increased overall enzyme activity but pro-
vided only limited control over the competing
epoxidation pathway. (Single-letter abbreviations
for the amino acid residues are as follows: A, Ala;
E, Glu; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M,
Met; N, Asn; R, Arg; S, Ser; T, Thr; V, Val; Y, Tyr.)
To exert greater pressure against epoxidation,
we used high-performance liquid chromatogra-
phy (HPLC) to screen for increases in the ratio of
anti-Markovnikov oxidation to epoxidation prod-
ucts. We also switched from random mutagene-
sis of the whole heme domain to site-saturation
mutagenesis of amino acids predicted to be in
the active site and heme-binding pocket (20, 21).
Six additional rounds of site-saturation muta-
genesis and screening generated the variant that
we call aMOx. It has eight additional amino acid
mutations (A103L, M118L, R120H, V123I, I326V,
V327M, H385V, and M391L), overall substituting
3% of the heme domain amino acids (fig. S5 and
O
O
aMOx, O
2
aMOx
No conversion
NAD(P)H
O , NADH
2
3
1
2
(R) or (S)
4500
100
90
80
70
60
50
40
30
20
10
D
O
D
aMOx
4000
3500
3000
2500
2000
1500
1000
500
D
D
O , NADH
2
4
5
Me
Me
O
aMOx
O , NADH
2
6
7
91:9 (S):(R)
Fe
R1
R2
H
H
O
H
Fe
O
H
R1
R2
0
•
•
locked substrate conformation
enantioselective 1,2-hydride migration
Directed evolution
P450LA1
aMOx
Fig. 2. Directed evolution of aMOx and mechanistic insights. (A) aMOx was engineered in the
laboratory in 10 rounds of directed evolution using styrene 1 as a substrate (tables S2 and S3).
The error bars represent the standard deviation of the total turnover number (TTN) for anti-
Markovnikov oxidation in reactions run in at least two independent triplicates (minimum of
six reactions). (B) The aMOx-catalyzed anti-Markovnikov oxidation is a direct oxidation without
an epoxide intermediate (fig. S6). (C) Isotopic labeling experiments support a 1,2-hydride
migration in the catalytic cycle (fig. S8). (D) aMOx-catalyzed enantioselective anti-Markovnikov
oxidation (fig. S9). (E) Enantiocontrol of aMOx derives from the enzyme’s capacity to lock the
substrate in a specific conformation. Carbocation 1,2-rearrangements are best described as
concerted reactions via a suprafacial 1,2-shift (25, 44). NADH, reduced form of NAD+ (nicotinamide
adenine dinucleotide, oxidized form); NADPH, reduced form of NADP+ (NAD+ phosphate); D,
deuterium; Me, methyl.
Hammer et al., Science 358, 215–218 (2017)
13 October 2017
2 of 4