Catalytic iron-carbene intermediate revealed in a
cytochrome
c
carbene transferase
Russell D. Lewis
a,1
, Marc Garcia-Borràs
b,1
, Matthew J. Chalkley
c
, Andrew R. Buller
c,2
, K. N. Houk
b,3
, S. B. Jennifer Kan
c,3
,
and Frances H. Arnold
a,c,3
a
Division of Biology and Bioengineering, California Institute of Technology, Pasadena, CA 91125;
b
Department of Chemistry and Biochemistry, University of
California, Los Angeles, CA 90095; and
c
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125
Contributed by Frances H. Arnold, May 31, 2018 (sent for review April 26, 2018; reviewed by Kara L. Bren and Ryan G. Hadt)
Recently, heme proteins have been discovered and engineered by
directed evolution to catalyze chemical transformations that are
biochemically unprecedented. Many of these nonnatural enzyme-
catalyzed reactions are assumed to proceed through a catalytic
iron porphyrin carbene (IPC) intermediate, although this interme-
diate has never been observed in a protein. Using crystallographic,
spectroscopic, and computational methods, we have captured and
studied a catalytic IPC intermediate in the active site of an enzyme
derived from thermostable
Rhodothermus marinus
(
Rma
) cyto-
chrome
c
. High-resolution crystal structures and computational
methods reveal how directed evolution created an active site for
carbene transfer in an electron transfer protein and how the
laboratory-evolved enzyme achieves perfect carbene transfer ster-
eoselectivity by holding the catalytic IPC in a single orientation.
We also discovered that the IPC in
Rma
cytochrome
c
has a singlet
ground electronic state and that the protein environment uses
geometrical constraints and noncovalent interactions to influence
different IPC electronic states. This information helps us to under-
stand the impressive reactivity and selectivity of carbene transfer
enzymes and offers insights that will guide and inspire future
engineering efforts.
carbene
|
reactive intermediate
|
heme
|
metalloenzyme
|
carbene transferase
C
arbenes are formally neutral, divalent carbon species with
two nonbonded electrons. They are typically reactive due to
their incomplete octet electron configuration (1). Highly stabi-
lized (
“
persistent
”
) carbenes exist in biological systems, most
notably as the deprotonated thiazolium group of thiamine in
pyruvate oxidases (2). In contrast, reactive metal-carbenes have
not been found in nature, despite their versatility and utility in
chemical synthesis (3
–
5). Recent protein engineering efforts
have revealed that heme proteins can catalyze a myriad of formal
carbene transfer reactions, including alkene cyclopropanation
(6
–
9), alkyne cyclopropenation and bicyclobutanation (10), and
carbonyl olefination (11) as well as carbon
−
boron (12),
–
nitrogen
(13),
–
silicon (14), and
–
sulfur (15) bond formation (Fig. 1). The
proteins enable the heme center to make products that iron
porphyrin alone does not (10, 12), and their genetically encoded
structures direct selectivity that can be switched by evolution (6, 8
–
10, 12, 14), making it possible to create whole new biocatalytic
pathways to important molecules.
By analogy to known transition metal-catalyzed reactions that
use diazo reagents as carbene precursors (3
–
5), the emerging
enzymatic carbene transfer activities of heme proteins are all
thought to involve a reactive iron porphyrin carbene (IPC) in-
termediate; transfer of the carbene to a second substrate and
product release regenerate the protein catalyst (6
–
15). Our
current understanding of enzyme-catalyzed carbene transfer re-
actions comes primarily from experimental characterizations of
model IPC compounds (16
–
21) and their computational studies
(22
–
27). Determining the structures and properties of reactive
intermediates is central to deciphering how enzymes enhance
reaction rates and dictate selectivity (28), and recent mechanistic
studies of cyclopropanation enzymes (24, 26, 27) have also
underscored the importance of the IPC intermediate and the
need to characterize it within a protein scaffold.
Results and Discussion
Structural Characterization of En
gineered Carbene Transferase,
Rhodothermus marinus
cytochrome
c
V75T M100D M103E.
Rhodothermus
marinus
(
Rma
) cytochrome c (cyt
c
)isanelectrontransferprotein
[Protein Data Bank (PDB) ID code 3CP5] (29) previously engineered
to catalyze the formation of carbon
−
silicon (14) and carbon
−
boron
(12) bonds by inserting a
reactive iron-carbene
into the corresponding
Si
–
HandB
–
H bonds of a silane or borane substrate. We set out to
capture and characterize the IPC intermediate in
Rma
cyt
c
by turning
to
Rma
cyt
c
V75T M100D M103E (TDE), a variant previously
engineered to catalyze silylation re
actions with high efficiency (14).
Rma
TDE was discovered after three rounds of site saturation mu-
tagenesis and screening for enhanced
activity, which resulted in amino
acid substitutions V75T, M100D, a
nd M103E. This protein is capable
of catalyzing thousands of rounds
of carbene transfer using ethyl
2-diazopropanoate (Me-EDA) (Fig. 2
A
) as a carbene precursor,
Significance
Here, we capture and study a reactive iron porphyrin carbene
(IPC) intermediate in the heme binding pocket of an engi-
neered cytochrome
c
protein. IPCs have never before been di-
rectly characterized in a protein, although they are thought to
be the key catalytic intermediate common to an array of abi-
ological but synthetically useful carbene transfer reactions
catalyzed by wild-type and engineered heme proteins. Our
work provides insight into how a
“
carbene transferase
”
ac-
quired its new-to-nature function as well as how it facilitates
efficient and selective transfer of the carbene to a second
substrate. Knowledge gained by studying this versatile in-
termediate provides a foundation for studying the mechanisms
of carbene transfer reactions and will facilitate the engineering
of carbene transfer enzymes.
Author contributions: R.D.L., M.G.-B., K.N.H., S.B.J.K., and F.H.A. designed research; R.D.L.,
M.G.-B., M.J.C., and S.B.J.K. performed research; R.D.L., M.G.-B., M.J.C., A.R.B., K.N.H.,
S.B.J.K., and F.H.A. analyzed data; and R.D.L., M.G.-B., K.N.H., S.B.J.K., and F.H.A. wrote
the paper.
Reviewers: K.L.B., University of Rochester; and R.G.H., Argonne National Laboratory.
The authors declare no conflict of interest.
Published under the
PNAS license
.
Data deposition: The atomic coordinates and structure factors have been deposited in the
Protein Data Bank,
www.wwpdb.org
(PDB ID codes
6CUK
and
6CUN
).
1
R.D.L. and M.G.-B. contributed equally to this work.
2
Present address: Department of Chemistry, University of Wisconsin
–
Madison, Madison,
WI 53706.
3
To whom correspondence may be addressed.
Email: houk@chem.ucla.edu, sbjkan@
caltech.edu, or frances@cheme.caltech.edu.
This article contains supporting information online at
www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1807027115/-/DCSupplemental
.
Published online June 26, 2018.
7308
–
7313
|
PNAS
|
July 10, 2018
|
vol. 115
|
no. 28
www.pnas.org/cgi/doi/10.1073/pnas.1807027115
presumably through the intermediacy of IPC
1
(Fig. 2
A
). To visualize
the active site for carbene transfer, we obtained an X-ray crystal
structure of
Rma
TDE at 1.47 Å (PDB ID code 6CUK) (
SI Appendix
,
Table S1
A
). This structure reveals a pocket on the distal face of the
heme created by the three mutations (Fig. 2
B
). In
Rma
TDE, the
native distal methionine has bee
n mutated to an aspartate (M100D),
the
α
-carbonofwhichisdisplacedby4.8Årelativetoitspositionin
the wild-type structure (
SI Appendix
,Fig.S1
). This displacement
produces a cavity in the space previously occupied by the M100
sidechain and allows a water molecule (or hydroxide ion) to
coordinate to the heme iron in the
enzyme resting state. Overall,
directed evolution created an entirely new active site in
Rma
TDE,
where substrates bind, the carbene forms, and catalysis takes place.
Crystallographic Observation of an IPC in
Rma
TDE.
To observe an
IPC in the protein scaffold, crystals of
Rma
TDE were soaked
with Me-EDA at room temperature for 30 min under air, reduced
with sodium dithionite, and flash frozen in liquid nitrogen before
diffraction. Under these conditions, we determined an X-ray crystal
structure at 1.29 Å that shows a species bound to the iron porphyrin
(PDB ID code 6CUN) that was not p
resent in our previous struc-
ture of
Rma
TDE (PDB ID code 6CUK) (
SI Appendix
,TableS1
A
).
We observed contiguous electron density from the iron porphyrin to
an active site ligand, consistent with an enzymatic IPC intermediate,
rather than iron-bound Me-EDA (Fig. 2
C
). The observed electron
density is in agreement with a single carbene species in a single
conformation (
SI Appendix
, section IIH
), suggesting that this
structure represents the dominant form of the IPC intermediate in
solution during catalysis. The high occupancy of the carbene indi-
cates that the rate of carbene formation greatly exceeds the rate of
carbene decay in the absence of a second substrate, making the
direct characterization of an enzymatic IPC intermediate possible.
The crystal structure of carbene-bound
Rma
TDE displays
some similarities to previously studied small molecule IPC
complexes. The Fe
–
C bond length of 1.9 Å is consistent with the
previously determined structure of the IPC 1-methylimidazole-
ligated mesotetrakis(pentafluo
rophenyl)porphyrin (TPFPP)
diphenylcarbene, where the reported Fe
–
C bond length is 1.83 Å
(20). The Fe
–
N (imidazole) bond length of 2.1 Å in the protein is
also consistent with that in the small molecule structure (2.17 Å)
(20). Out-of-plane distortions, particularly ruffling distortions,
are observed in both the imidazole-ligated TPFPP diphenylcarbene
structure and the carbene-bound
Rma
TDE structure. Whereas
the mean out-of-plane deviation for TPFPP diphenylcarbene was
1.03 Å (
SI Appendix
,TableS2
), it is smaller for carbene-bound
Rma
TDE (0.6 Å; the deviation for
Rma
TDE alone is 0.7 Å). The
difference in distortion between both
Rma
TDE structures is small,
and both values are within the range reported for c-type cyto-
chrome proteins (0.3
–
1.2 Å) (30). Ruffling is known to be the main
distortion in these proteins and is thought to be induced by co-
valent attachment of the heme to the protein (31).
In contrast to what we observed in
Rma
TDE in the absence of
Me-EDA, amino acid residues D100, T101, and D102 are un-
resolved in the carbene-bound
Rma
TDE structure, indicating
that this loop region is flexible and the carbene is solvent ac-
cessible (Fig. 2
C
). The crystal structure also reveals close, non-
polar contacts between the iron-carbene and five amino acid
sidechains (T75, M76, P79, I83, and M89 are 3.5
–
4.3 Å from the
carbene) (Fig. 2
C
), suggesting that the active site is strongly
hydrophobic, despite its solvent accessibility.
IPC Electronic Structure and Spectroscopy.
When first proposed as a
reactive enzyme intermediate, the IPC in cytochrome P450 was
presented as analogous to oxoferryl intermediates found in the
natural P450 catalytic cycle (6). This assignment was supported
by the Fe(IV)-like Mössbauer parameters obtained for small
molecule IPC complexes (20, 21, 32). However, Zhang and co-
workers (24, 26, 27) have argued that IPCs are better described
as Fe(II) closed shell singlets, lower in energy than the triplet
states. Most recently, Shaik and coworkers (23) studied a model
IPC complex and concluded that the lowest energy state for their
IPC was an open shell singlet, best described as an
S
=
1/2 Fe(III)
that is antiferromagnetically coupled to a carbene radical.
Although the IPC open shell singlet is similar to the Fe(III)-
superoxide electronic structure in heme oxygen binding (33),
no experimental evidence has been published to support the
relevance of an open shell singlet configuration in IPCs. Com-
putational studies from both the Zhang group and the Shaik
group suggest that the energy differences between electronic
states are small (22
–
24, 27), and thus, the open shell singlet,
closed shell singlet, and triplet states should all be considered
plausible ground states for the IPC intermediate in the envi-
ronment of an enzyme active site.
Encouraged by our success in capturing the IPC in the enzyme
crystals, we sought to use spectroscopy to characterize the elec-
tronic structure of the IPC in
Rma
TDE. Under conditions
similar to those where
Rma
TDE is catalytically active, we pre-
pared solution samples of both
Rma
TDE and Me-EDA
–
treated
Rma
TDE and analyzed them with ultraviolet-visible (
SI Appendix
,
Fig. S2
), electron paramagnetic resonance (EPR), and Mössbauer
spectroscopy. Neither sample gave EPR signals in either parallel
(10 mW) or perpendicular (2 mW) mode at 4 K (
SI Appendix
, Fig.
S3
). While the absence of EPR signals under these conditions
does not prove that the samples are diamagnetic, these results are
consistent with our assignment of the resting enzyme and carbene-
bound enzyme as low-spin species (vide infra). Mössbauer spectra
of
Rma
TDE at 80 K were best fit as a single species with an
isomer shift (
∂
Fe
) of 0.45 mm
·
s
−
1
and a quadrupole splitting
(
j
Δ
E
Q
j
) of 1.18 mm
·
s
−
1
(
SI Appendix
,Fig.S4
), consistent with an
S
=
0 low-spin Fe(II) resting-state species. This observation is in
contrast to other Fe(II) hemes in proteins lacking an endogenous
distal ligand to iron, which have been shown to favor high-spin
(
S
=
2) states (34, 35). Mössbauer spectra of Me-EDA
–
treated
Rma
TDE at 80 K indicated the presence of a major species (88%
abundance) and a minor species (12
% abundance), the latter of which
wasassignedasferrous
Rma
TDE. The major species was assigned as
carbene-bound
Rma
TDE based on the Mössbauer parameters
Fig. 1.
The scope of laboratory-evolved biocatalytic carbene transfer reac-
tions. Engineered heme proteins catalyze a broad range of abiological re-
actions with diazo reagents. These reactions are assumed to proceed
through a catalytic IPC intermediate. X indicates porphyrin proximal ligand,
usually serine or histidine. All published examples of heme-dependent car-
bene transfer enzymes accept diazo reagents having one or more electron-
withdrawing groups (e.g., COOR, CF
3
) as carbene precursors. The active
resting state of the catalyst is Fe(II)-porphyrin.
Lewis et al.
PNAS
|
July 10, 2018
|
vol. 115
|
no. 28
|
7309
BIOCHEMISTRY
obtained (
∂
Fe
=
0.18 mm
·
s
−
1
,
j
Δ
E
Q
j
=
1.83 mm
·
s
−
1
)andthelackof
observed EPR signal (
SI Appendix
,Figs.S3andS4
). Thus, this re-
active intermediate can be made in high yields and studied in both
freeze-quenched solutions and crystals.
Computational Models of the IPC.
The symmetric Mössbauer line
shape of Me-EDA
–
treated
Rma
TDE and its lack of EPR signal
suggested that this is an
S
=
0 species or an
S
=
1 species with a
large zero-field splitting. To further differentiate these electronic
states, we used density functional theory (DFT) to characterize
model structures of IPC
1
in three possible electronic states. Our
calculations showed that the closed shell singlet (
S
=
0) state was
8.8 kcal
·
mol
−
1
more stable than the triplet state (
S
=
1) and
1.0 kcal
·
mol
−
1
less stable than the open shell singlet (Fig. 3
A
and
SI Appendix
, Fig. S5
). The geometrical features of the DFT-
optimized structures also suggested that the IPC in
Rma
TDE
exists as a singlet: the orientation of the ester group in the
carbene-bound
Rma
TDE crystal structure is similar to that
observed in the optimized closed shell and open shell singlet
structures but distinctly different from the triplet state (Fig. 3
A
).
The electronic structure of cytochrome proteins is known to cor-
relate with heme ruffling (31, 36), and the singlet ground state is
further supported by heme ruffling analysis (
SI Appendix
,Table
S2
): the ruffling out-of-plane distortion observed in the crystal
structure of carbene-bound
Rma
TDE (0.5 Å) is more similar to
that observed in the DFT-optimized closed shell singlet (0.401 Å)
and open shell singlet (0.421 Å) states than in the triplet state
(0.784 Å), where the heme is more distorted. While our DFT
calculations suggested that the triplet state was energetically and
geometrically unfavorable, the open shell and closed shell singlet
states are very close in energy and cannot be unambiguously
assigned to IPC
1
(
SI Appendix
,Fig.S5
). We also simulated the
Mössbauer parameters of the open shell and closed shell singlet
states but found that their predicted Mössbauer parameters were
within experimental error of the observed values (
SI Appendix
,
Fig. S4
E
). Spectroscopy and theory indicate that carbene-bound
Rma
TDE is a ground-state singlet.
The DFT models suggest that the observed end-on carbene is
thermodynamically more stable than the alternative (Fe, N)-
bridged carbene, which also requires a large distortion of the
porphyrin that is difficult to attain for the covalently ligated
heme in cyt
c
proteins (
SI Appendix
, Fig. S6
). This is in contrast
to the thiolate-ligated IPC computationally studied by Shaik and
coworkers (23), where the (Fe, N)-bridged mode was found to be
thermodynamically favored. Small molecule (Fe, N)-bridged
IPCs have been synthesized and crystallized (17
–
19, 25); their
experimental characterizations are distinct from that of end-on
IPCs. As far as we know, (Fe, N)-bridged carbenes are not active
intermediates in carbene transfer reactions (37).
Computational Modeling of the IPC Enzymatic Intermediate.
We next
used molecular dynamics (MD) to understand how the protein
scaffold of
Rma
TDE affects the conformation and properties of
the IPC intermediate. MD simulations were performed on sin-
glet IPC
1
docked into
Rma
TDE (PDB ID code 6CUK) using
various IPC
1
conformations as starting points and explicit water
as solvent. MD trajectories revealed that, regardless of the initial
carbene orientation, the IPC converges to a single well-defined
orientation (Fig. 3
B
and
SI Appendix
, Fig. S7
). This conforma-
tion is remarkably similar to that observed in the carbene-bound
AB
C
Fig. 2.
Carbene transfer silylation reaction and structures of resting and carbene-bound
Rma
TDE. (
A
)TheC
–
Si bond-forming reaction catalyzed by
Rma
cytochrome
c
proceeds via intermediate IPC
1
.(
B
) Mutations V75T, M100D, and M103E (
Rma
TDE) in
Rma
cytochrome
c
increased its carbene transfer activity
(14). The crystal structure of
Rma
TDE with no Me-EDA bound (1.47 Å; PDB ID code 6CUK) reveals a cavity distal to the heme (orange surfaces) that is not
present in wild-type
Rma
cyt
c
.(
C
) The crystal structure of carbene-bound
Rma
TDE (1.29 Å; PDB ID code 6CUN), with the carbene species in cyan. Residues
D100, T101, and D102 are unresolved. (
Inset
) Omit map in green (F
O
–
F
C
) of the carbene electron density contoured at 3
σ
. Interactions between the carbene
and amino acid residues are shown with yellow lines. (
i
) T75, 3.7 Å. (
ii
) M76, 4.3 Å. (
iii
) P79, 3.5 Å. (
iv
) I83, 3.6 Å. (
v
) M89, 3.5 Å.
7310
|
www.pnas.org/cgi/doi/10.1073/pnas.1807027115
Lewis et al.
Rma
cyt
c
crystal structure, indicating the catalytic relevance of
the latter in solution (Fig. 3
B
). Furthermore, MD simulations
highlighted the flexibility of the active site front loop (T98-E103)
(Fig. 3
C
), favoring an
“
open
”
conformation where the IPC is
solvent exposed and accessible to substrates with pro-
R
facial
selectivity, while keeping the pro-
S
face inaccessible (Fig. 3
D
).
This high degree of loop flexibility is consistent with the un-
resolved residues (D100
–
D102) observed in the carbene-bound
Rma
TDE crystal structure (Fig. 2
C
). To further investigate the
protein
’
s enantiopreference, we performed DFT calculations to
characterize the intrinsic geometric features of the pro-
R
transition
state for carbene Si
−
H insertion using a truncated computational
model and dimethylphenylsilane as substrate (Fig. 3
E
and
SI Ap-
pendix
,Fig.S8
). When comparing the lowest energy-optimized
pro-
R
transition state with the most populated structure of
carbene-bound
Rma
TDE, the space made available when
Rma
TDE is in its open conformation provides a perfect fit for the
silane substrate (Fig. 3
D
and
E
and
SI Appendix
,Fig.S8
). How-
ever, the pro-
S
transition state could only take place when the IPC
is rotated away from its preferred conformation, which is disfavored
by the active site of
Rma
TDE. Thus, in
Rma
TDE-catalyzed
carbon
−
silicon bond-forming reactions,
the protein dictates substrate
A
BC
D
EF
Fig. 3.
Computational models of carbene-bound
Rma
TDE. (
A
) DFT optimized structures of model IPC
1
in the closed shell singlet, open shell singlet, and
triplet electronic states. DFT calculations predict that the singlet state is 8.8 kcal
·
mol
−
1
more stable than the triplet state, which is consistent with previously
characterized small molecule IPCs. The calculated Fe
–
C bond lengths are 1.79, 1.84, and 1.97 Å for the closed shell singlet, open shell singlet, and triplet,
respectively. The Fe
–
N (imidazole) bond lengths are 2.21, 2.21, and 2.19 Å, respectively. These numbers are within experimental error of the lengths observed
in the carbene-bound
Rma
TDE crystal structure, where the Fe
–
C bond length is 1.9 Å and the Fe
–
N bond length is 2.1 Å. (
B
) Overlay of the carbene-bound
Rma
TDE crystal structure with a QM/MM-optimized structure of carbene-bound
Rma
TDE. QM/MM structure: protein scaffold is gray, and carbene is green.
Crystal structure (PDB ID code 6CUN): protein scaffold is light pink, and carbene is cyan. The preferred carbene conformation determined by MD simula
tions
and QM/MM calculations is almost identical to that determined by X-ray crystallography. (
C
) Overlay of 10 representative snapshots obtained from 1,000 ns of
MD simulation on the carbene-bound
Rma
TDE. Simulations show that the active site front loop (residues 98
–
103; highlighted in red) is highly flexible,
switching between
“
open
”
and
“
closed
”
conformations. Despite the front loop
’
s high flexibility, the carbene is stabilized in a single major conformation (
SI
Appendix
, Fig. S7
). (
D
) Snapshot obtained from an MD trajectory on carbene-bound
Rma
TDE with the front loop in the open conformation, showing that the
carbene pro-
R
face is solvent exposed. Solvent-accessible area is showed in orange. (
E
) DFT-optimized structure of the lowest-energy pro-
R
transition state
(closed shell singlet state) for Si
–
H insertion with IPC
1
(
SI Appendix
, Fig. S8
). (
F
) A representative view of an MD snapshot on carbene-bound
Rma
TDE
showing interactions between the bridging water with the heme carboxylate, the carbene carbonyl, and the amide bond of residue D100. Hydrogen-bondi
ng
interactions with the carbene have been shown to affect the stability of IPCs (25, 27, 37).
Lewis et al.
PNAS
|
July 10, 2018
|
vol. 115
|
no. 28
|
7311
BIOCHEMISTRY
approach to the IPC intermediate from the pro-
R
face, which on
carbene transfer to the Si
−
H bond of the silane substrate, gives
R
-configured organosilicon products exclusively as observed ex-
perimentally (
>
99% enantiomeric excess) (14).
MD simulations also revealed the presence of a water mole-
cule that forms persistent hydrogen bonds bridging one of the
heme carboxylate groups and the carbonyl group of the carbene
(Fig. 3
E
). This water molecule is highly conserved, with its po-
sition further stabilized by polar contacts with the front loop
residues. In the carbene-bound
Rma
TDE crystal structure, this
bridging water was not observed but was likely obscured due to
its proximity to the unresolved residues D100
–
D102. Hydrogen
bonding is known to affect the electronic structure and stability
of IPC intermediates (25, 27, 37). In view of this, we designed a
truncated computational model of an IPC
–
water complex, and
through DFT calculations, we found that the H-bonding water
molecule stabilizes the carbene complex by about 7 kcal
·
mol
−
1
in
its closed shell singlet state. This stabilizing effect is more sub-
stantial on the closed shell singlet than on the open shell singlet
or triplet states, contributing to determining the preferred
conformation adopted by the IPC in the protein (
SI Appendix
,
Fig. S9
). Hybrid quantum mechanics/molecular mechanics
(QM/MM) calculations showed that, within
Rma
TDE, the sin-
glet
–
triplet energy gap of the hydrogen-bonded IPC increases
by 2.9 kcal
·
mol
−
1
(8.8 kcal
·
mol
−
1
in the truncated model,
11.7 kcal
·
mol
−
1
when considering the full system) (
SI Appendix
,
Fig. S10
). This significant destabilization of the radical triplet
state is largely due to conformational restraints imposed by the
protein active site and the bridging water: to minimize steric
clash with the P79 and I83 sidechains, the carbene ester is forced
to adopt a conformation parallel to the heme, which is in-
trinsically preferred by the closed shell and open shell singlet
states but disfavors the radical triplet state, as this conformation
cannot stabilize the unpaired electron in the carbene
p
orbital
by the adjacent ester. Thus, mediated by amino acid residues
and an ordered water, the active site of
Rma
TDE imposes
multiple conformational restraints on the IPC intermediate to
stabilize the closed shell singlet state and disfavor the presence
of carbene radicals.
Summary and Conclusions
In conclusion, characterization of a catalytically active IPC in-
termediate inside the active site of an enzyme has provided a
foundation from which to study enzyme-catalyzed carbene trans-
fer reactions and explain the reactivity and selectivity of
Rma
TDE-catalyzed carbene transfer. Structural characterization of the
Rma
TDE enzyme showed that the three mutations introduced
during directed evolution led to dramatic changes in a loop distal
to the heme and generated a protein structure preorganized for
substrate binding and IPC formation. The experimental IPC
structure and computational modeling showed that
Rma
TDE
stabilizes a single conformation of the IPC that gives the enzyme
complete stereoselectivity control during carbene transfer. The
conformational restraints imposed by the enzyme also favor the
formation of a singlet iron-carbene. This convergent approach to
characterizing the emerging family of carbene transferases enables
informed engineering that will enhance the generality and utility
of these powerful catalysts.
Materials and Methods
Detailed experimental methods are presented in
SI Appendix
.
Cloning, Expression, and Purification of
Rma
TDE.
The gene encoding
Rma
TDE was obtained as a gBlock and cloned into pET22b(
+
). Expression was
performed in HyperBroth (AthenaES) using
Escherichia coli
BL21
E. cloni
EXPRESS (Lucigen).
57
Fe-labeled protein was produced using defined media
supplemented with
57
Fe.
Rma
TDE was purified via a HisTrap HP column
(GE Healthcare).
Crystallography.
Crystals of
Rma
TDE were grown using the sitting drop
vapor diffusion method and were cryoprotected before diffraction at the
Stanford Synchrotron Radiation Lightsource on beamline 12
–
2. Carbene-
bound crystals of
Rma
TDE were prepared by soaking preformed crystals
with concentrated solutions of Me-EDA and sodium dithionite. Structures
were determined by molecular replacement, and models were built using
standard procedures.
EPR Spectroscopy.
EPR spectra were recorded on a Bruker EMX X-band
CW-EPR spectrometer using Bruker Win-EPR software (version 3.0). Data
were collected in parallel mode with a power of 10 mW and in perpendicular
mode with a power of 2 mW, both at 4 K. Samples of
Rma
TDE were prepared
under anaerobic conditions in the presence of sodium dithionite before flash
freezing with liquid nitrogen.
Mössbauer Spectroscopy.
Mössbauer spectra were recorded using a spec-
trometer from SEE Co., and data analysis was performed using the WMOSS
program. Solution samples of
57
Fe-labeled
Rma
TDE were prepared under
anaerobic conditions in the presence of sodium dithionite before flash
freezing with liquid nitrogen. Samples were stored under liquid nitrogen
until being mounted on the cryostat, and data were recorded at 77 K.
DFT Calculations.
DFT calculations were carried out using Gaussian09 (38) and
Gaussian16 (39) using a truncated model. Geometry optimizations and fre-
quency calculations were performed using (U)B3LYP (40
–
42) functional with
the SDD basis set for iron and 6
–
31G(d) on all other atoms and within the
conductor-like polarizable continuum model (CPCM) (diethyl ether,
e
=
4)
(43, 44) to have an estimation of the dielectric permittivity in the enzyme
active site. Enthalpies and entropies were calculated for 1 atm and 298.15 K.
A correction to the harmonic oscillator approximation was also applied to
the entropy calculations by raising all frequencies below 100 cm
–
1
to 100 cm
–
1
(45, 46). Single-point energy calculations were performed using the dispersion-
corrected functional (U)B3LYP-D3(BJ) (47, 48) with the Def2TZVP basis set on
all atoms and with CPCM (diethyl ether,
e
=
4). All energy values discussed in
the manuscript correspond to the quasiharmonic corrected Gibbs energies
(
Δ
G-qh) at (U)B3LYP-D3(BJ)/Def2TZVP/PCM(diethyl ether)//(U)B3LYP/6
–
31G(d)
+
SDD(Fe)/PCM(diethyl ether) level if not otherwise noted. Further descriptions
and details of the methodology are provided in
SI Appendix
.
Hybrid QM/MM Calculations.
QM/MM calculations within the ONIOM ap-
proach (49, 50) were carried out using Gaussian09 (38). Geometry optimi-
zations were performed using (U)B3LYP (40
–
42) functional in combination
with the SDD basis set for iron and 6
–
31G(d) on all other atoms and
AmberFF14Sb force field (51) using a mechanical embedding scheme. Sta-
tionary points were verified as minima by a vibrational frequency analysis,
thermal corrections were calculated for 1 atm and 298.15 K, and a correction
to the harmonic oscillator approximation was included as described above.
Single-point energy calculations were performed at the (U)B3LYP/Def2TZVP:
AmberFF14Sb level and using an electrostatic embedding scheme. Snapshots
for QM/MM calculations were obtaine
d from classical MD trajectories.
Further descriptions and details of the methodology are provided in
SI Appendix
.
MD Simulations.
MD simulations were performed using the GPU code
(pmemd) (52) of the AMBER 16 package (53). Parameters for the carbene-
bound IPC were generated within the antechamber and MCPB.py (54)
modules in AMBER16 package using the general AMBER force field (gaff)
(55), with partial charges set to fit the electrostatic potential generated at
the B3LYP/6
–
31G(d) level by the RESP model (56). Protonation states of
protein residues were predicted using H
++
server. Each protein was im-
mersed in a preequilibrated truncated cuboid box with a 10-Å buffer of
TIP3P (57) water molecules using the leap module, and the systems were
neutralized by addition of explicit counterions (Na
+
and Cl
−
). All subsequent
calculations were done using the widely tested Stony Brook modification of
the Amber14 force field (ff14sb) (58). After minimization, heating, and
equilibration of the system (
SI Appendix
has a full description of the pro-
tocol used), production trajectories were then run for 1,000 ns (1
μ
s). The
obtained trajectories were processed and analyzed using the cpptraj (58)
module from Ambertools utilities.
ACKNOWLEDGMENTS.
We thank J. M. Bollinger Jr., K. Chen, X. Huang,
C. Krebs, C. J. Pollock, and R. K. Zhang for helpful discussions and A. Tang for
experimental assistance. This work was supported by National Science
Foundation Division of Chemistry Grant CHE-1361104 (to K.N.H.); the
Rothenberg Innovation Initiative (RI2) Program (S.B.J.K. and F.H.A.); the
7312
|
www.pnas.org/cgi/doi/10.1073/pnas.1807027115
Lewis et al.
Jacobs Institute for Molecular Engineering for Medicine at Caltech (S.B.J.K.
and F.H.A.); National Science Foundation Division of Molecular and Cellular
Biosciences Grant MCB-1513007 (to F.H.A.); and Office of Chemical, Bio-
engineering, Environmental and Transport Systems SusChEM Initiative Grant
CBET-1403077 (to F.H.A.). R.D.L. is supported by NIH National Research
Service Award Training Grant 5 T32 GM07616. M.G.-B. thanks the Ramón
Areces Foundation for a postdoctoral fellowship. M.J.C. thanks the Center
for Environmental Microbial Intera
ctions at Caltech for a fellowship.
Crystallography experiments were supported by J. Kaiser and the Caltech
Molecular Observatory. EPR experi
ments were performed with the assis-
tance of P. Oyala and supported by National Science Foundation Grant
NSF-1531940 (to the Caltech EPR Facility). Computational resources were
provided by the University of California, Los Angeles Institute for Digital
Research and Education and the Extreme Science and Engineering Dis-
covery Environment, which is suppo
rted by National Science Foundation
Grant OCI-1053575.
1. Bourissou D, Guerret O, Gabbaï FP, Bertrand G (2000) Stable carbenes.
Chem Rev
100:
39
–
92.
2. Meyer D, Neumann P, Ficner R, Tittmann K (2013) Observation of a stable carbene at
the active site of a thiamin enzyme.
Nat Chem Biol
9:488
–
490.
3. Kornecki KP, et al. (2013) Direct spectroscopic characterization of a transitory dirho-
dium donor-acceptor carbene complex.
Science
342:351
–
354.
4. Doyle MP (1986) Catalytic methods for metal carbene transformations.
Chem Rev
86:
919
–
939.
5. Ford A, et al. (2015) Modern organic synthesis with
α
-diazocarbonyl compounds.
Chem Rev
115:9981
–
10080.
6. Coelho PS, Brustad EM, Kannan A, Arnold FH (2013) Olefin cyclopropanation via
carbene transfer catalyzed by engineered cytochrome P450 enzymes.
Science
339:
307
–
310.
7. Bordeaux M, Tyagi V, Fasan R (2015) Highly diastereoselective and enantioselective
olefin cyclopropanation using engineered myoglobin-based catalysts.
Angew Chem
Int Ed Engl
54:1744
–
1748.
8. Gober JG, et al. (2016) Mutating a highly conserved residue in diverse cytochrome
P450s facilitates diastereoselective olefin cyclopropanation.
ChemBioChem
17:
394
–
397.
9. Knight AM, et al. (2018) Diverse engineered heme proteins enable stereodivergent
cyclopropanation of unactivated alkenes.
ACS Cent Sci
4:372
–
377.
10. Chen K, Huang X, Kan SBJ, Zhang RK, Arnold FH (2018) Enzymatic construction of
highly strained carbocycles.
Science
360:71
–
75.
11. Weissenborn MJ, et al. (2016) Enzyme-catalyzed carbonyl olefination by the
E. coli
protein YfeX in the absence of phosphines.
ChemCatChem
8:1636
–
1640.
12. Kan SBJ, Huang X, Gumulya Y, Chen K, Arnold FH (2017) Genetically programmed
chiral organoborane synthesis.
Nature
552:132
–
136.
13. Wang ZJ, Peck NE, Renata H, Arnold FH (2014) Cytochrome P450-catalyzed insertion
of carbenoids into N-H bonds.
Chem Sci (Camb)
5:598
–
601.
14. Kan SBJ, Lewis RD, Chen K, Arnold FH (2016) Directed evolution of cytochrome
c
for
carbon-silicon bond formation: Bringing silicon to life.
Science
354:1048
–
1051.
15. Tyagi V, Bonn RB, Fasan R (2015) Intermolecular carbene S-H insertion catalysed by
engineered myoglobin-based catalysts.
Chem Sci (Camb)
6:2488
–
2494.
16. Mansuy D, et al. (1978) Dichlorocarbene complexes of iron(II)-porphyrins-crystal and
molecular structure of Fe(TPP)(CCl2)(H2O).
Angew Chem Int Ed Engl
17:781
–
782.
17. Chevrier B, Weiss R, Lange M, Chottard JC, Mansuy D (1981) An iron(III)-porphyrin
complex with a vinylidene group inserted into an iron-nitrogen bond: Relevance to
the structure of the active oxygen complex of catalase.
J Am Chem Soc
103:
2899
–
2901.
18. Olmstead MM, Cheng RJ, Balch AL (1982) X-ray crystallographic characterization of an
iron porphyrin with a vinylidene carbene inserted into an iron-nitrogen bond.
Inorg
Chem
21:4143
–
4148.
19. Artaud I, Gregoire P, Leduc P, Mansuy D (1990) Formation and fate of iron-carbene
complexes in reactions between a diazoalkane and iron-porphyrins: Relevance to the
mechanism of formation of N-substituted hemes in cytochrome P-450 dependent
oxidation of sydnones.
J Am Chem Soc
112:6899
–
6905.
20. Li Y, Huang JS, Zhou ZY, Che CM, You XZ (2002) Remarkably stable iron porphyrins
bearing nonheteroatom-stabilized carbene or (alkoxycarbonyl)carbenes: Isolation,
X-ray crystal structures, and carbon atom transfer reactions with hydrocarbons.
JAm
Chem Soc
124:13185
–
13193.
21. Liu Y, et al. (2017) Electronic configuration and ligand nature of five-coordinate iron
porphyrin carbene complexes: An experimental study.
J Am Chem Soc
139:5023
–
5026.
22. Khade RL, et al. (2014) Iron porphyrin carbenes as catalytic intermediates: Structures,
Mössbauer and NMR spectroscopic properties, and bonding.
Angew Chem Int Ed Engl
53:7574
–
7578.
23. Sharon DA, Mallick D, Wang B, Shaik S (2016) Computation sheds insight into iron
porphyrin carbenes
’
electronic structure, formation, and N-H insertion reactivity.
JAm
Chem Soc
138:9597
–
9610.
24. Wei Y, Tinoco A, Steck V, Fasan R, Zhang Y (2018) Cyclopropanations via heme car-
benes: Basic mechanism and effects of carbene substituent, protein axial ligand, and
porphyrin substitution.
J Am Chem Soc
140:1649
–
1662.
25. Tatsumi K, Hoffmann R (1981) Metalloporphyrins with unusual geometries. 2. Slipped
and skewed bimetallic structures, carbene and oxo complexes, and insertions into
metal-porphyrin bonds.
Inorg Chem
20:3771
–
3784.
26. Khade RL, Zhang Y (2015) Catalytic and biocatalytic iron porphyrin carbene forma-
tion: Effects of binding mode, carbene substituent, porphyrin substituent, and pro-
tein axial ligand.
J Am Chem Soc
137:7560
–
7563.
27. Khade RL, Zhang Y (2017) C-H insertions by iron porphyrin carbene: Basic mechanism
and origin of substrate selectivity.
Chemistry
23:17654
–
17658.
28. Rittle J, Green MT (2010) Cytochrome P450 compound I: Capture, characterization,
and C-H bond activation kinetics.
Science
330:933
–
937.
29. Stelter M, et al. (2008) A novel type of monoheme cytochrome
c
: Biochemical and
structural characterization at 1.23 A resolution of
rhodothermus marinus
cytochrome
c
.
Biochemistry
47:11953
–
11963.
30. Graves AB, Graves MT, Liptak MD (2016) Measurement of heme ruffling changes in
MhuD using UV-vis spectroscopy.
J Phys Chem B
120:3844
–
3853.
31. Kleingardner JG, Bren KL (2015) Biological significance and applications of heme
c
proteins and peptides.
Acc Chem Res
48:1845
–
1852.
32. English DR, Hendrickson DN, Suslick KS (1983) Mössbauer spectra of oxidized iron
porphyrins.
Inorg Chem
22:367
–
368.
33. Wilson SA, et al. (2013) X-ray absorption spectroscopic investigation of the electronic
structure differences in solution and crystalline oxyhemoglobin.
Proc Natl Acad Sci
USA
110:16333
–
16338.
34. Ran Y, et al. (2010) Spectroscopic identification of heme axial ligands in HtsA that are
involved in heme acquisition by Streptococcus pyogenes.
Biochemistry
49:2834
–
2842.
35. Engler N, Prusakov V, Ostermann A, Parak FG (2003) A water network within a pro-
tein: Temperature-dependent water ligation in H64V-metmyoglobin and relaxation
to deoxymyoglobin.
Eur Biophys J
31:595
–
607.
36. Liptak MD, Wen X, Bren KL (2010) NMR and DFT investigation of heme ruffling:
Functional implications for cytochrome
c
.
J Am Chem Soc
132:9753
–
9763.
37. Dzik WI, Xu X, Zhang XP, Reek JNH, de Bruin B (2010)
‘
Carbene radicals
’
in Co(II)(por)-
catalyzed olefin cyclopropanation.
J Am Chem Soc
132:10891
–
10902.
38. Frisch MJ, et al. (2009) Gaussian 09, Revision A.02 (Gaussian Inc., Wallingford, CT).
39. Frisch MJ, et al. (2016) Gaussian 16, Revision B.01 (Gaussian, Inc., Wallingford, CT).
40. Becke AD (1988) Density-functional exchange-energy approximation with correct
asymptotic behavior.
Phys Rev A Gen Phys
38:3098
–
3100.
41. Becke AD (1993) Density
‐
functional thermochemistry. III. The role of exact exchange.
J Chem Phys
98:5648
–
5652.
42. Lee C, Yang W, Parr RG (1988) Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density.
Phys Rev B Condens Matter
37:
785
–
789.
43. Barone V, Cossi M (1998) Quantum calculation of molecular energies and energy
gradients in solution by a conductor solvent model.
J Phys Chem A
102:1995
–
2001.
44. Cossi M, Rega N, Scalmani G, Barone V (2003) Energies, structures, and electronic
properties of molecules in solution with the C-PCM solvation model.
J Comput Chem
24:669
–
681.
45. Ribeiro RF, Marenich AV, Cramer CJ, Truhlar DG (2011) Use of solution-phase vibra-
tional frequencies in continuum models for the free energy of solvation.
J Phys Chem
B
115:14556
–
14562.
46. Zhao Y, Truhlar DG (2008) Computational characterization and modeling of buckyball
tweezers: Density functional study of concave-convex pi...pi interactions.
Phys Chem
Chem Phys
10:2813
–
2818.
47. Grimme S, Ehrlich S, Goerigk L (2011) Effect of the damping function in dispersion
corrected density functional theory.
J Comput Chem
32:1456
–
1465.
48. Grimme S, Antony J, Ehrlich S, Krieg H (2010) A consistent and accurate
ab initio
parametrization of density functional dispersion correction (DFT-D) for the 94
elements H-Pu.
J Chem Phys
132:154104.
49. Dapprich S, Komáromi I, Byun KS, Morokuma K, Frisch MJ (1999) A new ONIOM
implementation in Gaussian98. Part I. The calculation of energies, gradients, vibrational
frequencies and electric field derivatives.
JMolStructTHEOCHEM
461
–
462:1
–
21.
50. Chung LW, et al. (2015) The ONIOM method and its applications.
Chem Rev
115:
5678
–
5796.
51. Maier JA, et al. (2015) ff14SB: Improving the accuracy of protein side chain and
backbone parameters from ff99SB.
J Chem Theory Comput
11:3696
–
3713.
52. Salomon-Ferrer R, Götz AW, Poole D, Le Grand S, Walker RC (2013) Routine micro-
second molecular dynamics simulations with AMBER on GPUs. 2. Explicit solvent
particle mesh Ewald.
J Chem Theory Comput
9:3878
–
3888.
53. Case DA, et al. (2016)
Amber 16
(University of California, San Francisco).
54. Li P, Merz KM, Jr (2016) MCPB.py: A python based metal center parameter builder.
J Chem Inf Model
56:599
–
604.
55. Wang J, Wolf RM, Caldwell JW, Kollman PA, Case DA (2004) Development and testing
of a general amber force field.
J Comput Chem
25:1157
–
1174.
56. Bayly CI, Cieplak P, Cornell W, Kollman PA (1993) A well-behaved electrostatic
potential based method using charge restraints for deriving atomic charges: The RESP
model.
J Phys Chem
97:10269
–
10280.
57. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML (1983) Comparison
of simple potential functions for simulating liquid water.
J Chem Phys
79:926
–
935.
58. Roe DR, Cheatham TE, 3rd (2013) PTRAJ and CPPTRAJ: Software for processing and
analysis of molecular dynamics trajectory data.
J Chem Theory Comput
9:3084
–
3095.
Lewis et al.
PNAS
|
July 10, 2018
|
vol. 115
|
no. 28
|
7313
BIOCHEMISTRY