Engineering of RuMb: Toward a Green Catalyst for Carbene Insertion Reactions

Article abstract of DOI:10.1021/acs.inorgchem.6b03148

The small, stable heme protein myoglobin (Mb) was modified through cofactor substitution and mutagenesis to develop a new catalyst for carbene transfer reactions. The native heme was removed from wild-type Mb and several Mb His64 mutants (H64D, H64A, H64V), and the resulting apoproteins were reconstituted with ruthenium mesoporphyrin IX (RuMpIX). The reconstituted proteins (RuMb) were characterized by UV-vis and circular dichroism spectroscopy and were used as catalysts for the N-H insertion of aniline derivatives and the cyclopropanation of styrene derivatives. The best catalysts for each reaction were able to achieve turnover numbers (TON) up to 520 for the N-H insertion of aniline, and 350 TON for the cyclopropanation of vinyl anisole. Our results show that RuMb is an effective catalyst for N-H insertion, with the potential to further increase the activity and stereoselectivity of the catalyst in future studies. Compared to native Mb ("FeMb"), RuMb is a more active catalyst for carbene transfer reactions, which leads to both heme and protein modification and degradation and, hence, to an overall much-reduced lifetime of the catalyst. This leads to lower TONs for RuMb compared to the iron-containing analogues. Strategies to overcome this limitation are discussed. Finally, comparison is also made to FeH64DMb and FeH64AMb, which have not been previously investigated for carbene transfer reactions.

Full text of DOI:10.1021/acs.inorgchem.6b03148

Article
pubs.acs.org/IC
Engineering of RuMb: Toward a Green Catalyst for Carbene Insertion
Reactions
Matthew W. Wolf, David A. Vargas, and Nicolai Lehnert*
Department of Chemistry and Department of Biophysics, University of Michigan, Ann Arbor, Michigan 48109, United States
*
S Supporting Information
ABSTRACT: The small, stable heme protein myoglobin (Mb)
was modified through cofactor substitution and mutagenesis to
develop a new catalyst for carbene transfer reactions. The native
heme was removed from wild-type Mb and several Mb His64
mutants (H64D, H64A, H64V), and the resulting apoproteins were
reconstituted with ruthenium mesoporphyrin IX (RuMpIX). The
reconstituted proteins (RuMb) were characterized by UV−vis and
circular dichroism spectroscopy and were used as catalysts for the
N−H insertion of aniline derivatives and the cyclopropanation of
styrene derivatives. The best catalysts for each reaction were able to
achieve turnover numbers (TON) up to 520 for the N−H
insertion of aniline, and 350 TON for the cyclopropanation of vinyl
anisole. Our results show that RuMb is an effective catalyst for N−
H insertion, with the potential to further increase the activity and
stereoselectivity of the catalyst in future studies. Compared to native Mb (“FeMb”), RuMb is a more active catalyst for carbene
transfer reactions, which leads to both heme and protein modification and degradation and, hence, to an overall much-reduced
lifetime of the catalyst. This leads to lower TONs for RuMb compared to the iron-containing analogues. Strategies to overcome
this limitation are discussed. Finally, comparison is also made to FeH64DMb and FeH64AMb, which have not been previously
investigated for carbene transfer reactions.
I. INTRODUCTION
use of biotin-modified catalysts in avidin or streptavidin) and5
covalent attachments to residues located inside the protein.
Protein structure can also be modified through mutagenesis
either by directed mutagenesis of a few key residues near the
active site of the protein (rational design) or through directed
evolution, an iterative approach with many rounds of sequential
Small-molecule transition-metal complexes have proven to be
effective catalysts in industrial organic syntheses, often
catalyzing thousands of turnovers with high turnover
frequencies (TOF) and long catalyst lifetimes. However,
many of these catalysts require environmentally detrimental
solvents to carry out these reactions and bulky ligands to induce
product stereoselectivity. By contrast, metalloenzymes catalyze
a variety of reactions in biology with high turnover numbers
1
−4,6
mutagenesis, and catalyst screening.
A directed evolution
approach often initially targets key active-site residues, but this
approach still trades a greater understanding of the changes to
the catalytic system with each modification for a more
expedient route to higher yields and stereoselectivities. There
are multiple examples of one or the other approach being used
for abiological catalysis, and a few studies have also used the
(TON) and stereo- and enantiospecificity, conferred by the
chiral active site around the metal center. Modern advances in
protein modification via facile mutagenesis and the introduction
of new metal cofactors have allowed for the preparation of new
artificial enzymes to catalyze different organic transforma-
7
two approaches in tandem.
1
,2
Carbene transfer reactions are of particular interest in the
quickly developing field of organometallic biocatalysis. These
reactions can produce new C−C, C−N, and C−S bonds and
9,10
tions. By mutating key active-site residues around the metal
center, product stereoselectivity can be altered. Starting with
the same protein scaffold, sequential rounds of mutagenesis
allow for multiple unique enzymes that catalyze the same
3
,4,6−8
useful organic products such as cyclopropanes,
amines,
11
12
3
,4
and thioethers, among others. A number of transition-metal
complexes catalyze the decomposition of diazo reagents to
form an electrophilic metal−carbene unit that is then attacked
by a nucleophilic reagent such as an olefin (for cyclo-
organic transformation but favor opposite stereoisomers. In
this way, a variety of organometallic catalysts can be produced
for different organic transformations and for specific stereo-
isomers of a desired product. In general, two approaches are
used to accomplish this: (a) the anchoring of catalysts in a
13
propanation) or an amine (for N−H insertion). In particular,
5
protein matrix and (b) modifying existing protein active
1
−4,6
sites.
Transition-metal catalysts can be anchored inside
Received: December 29, 2016
protein scaffolds through supramolecular contacts (such as the
©
XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
a variety of metalloporphyrin complexes including metals such
reactions. We also studied the pH dependence of these
reactions and compared the results to [RuMpIX] outside of the
protein scaffold. Our results show that the RuMb catalysts
demonstrate moderate to high yields for the N−H insertion
reactions and low to moderate yields for cyclopropanation
reactions, but yield is significantly limited by catalyst stability.
Enhanced diastereoselectivity of the cyclopropane products can
be induced by RuH64DMb and some of the other variants, but
this selectivity is lower compared to Mb catalysts with multiple
point mutations and compared to corresponding FeMb
variants. Future studies are aimed at optimizing our RuMb
catalysts via additional mutagenesis experiments to further
decrease the polarity of the Mb active site and, in addition, via
porphyrin modifications to increase the stability of our RuMb
catalysts.
as Fe, Co, Ru, Rh, and Ir have been effective toward catalyzing
14−26
carbene transfer reactions.
These complexes catalyze
thousands of turnovers, and the second- and third-row
transition-metal complexes typically have long catalyst lifetimes,
but also exhibit the aforementioned drawbacks of small-
molecule catalysts, where stereoselectivity is difficult to control
and reactions are often performed in environmentally
detrimental organic solvents. Chiral porphyrins can induce
stereoselectivity in the catalytic products, but these porphyrins
usually require more arduous syntheses and still exhibit lower
27−29
enantioselectivity compared to enzymes.
multiple groups have sought to modify heme proteins
containing Fe porphyrin cofactors) to produce new types of
stereoselective and water-soluble catalysts for these trans-
In recent years,
(
3
,4,6−12
formations.
The Arnold and Fasan groups have demonstrated that heme
proteins such as cytochrome (Cyt) P450BM3 and myoglobin
II. EXPERIMENTAL PROCEDURES
General Procedures. All chemicals and reagents were purchased
from commercial suppliers (Sigma-Aldrich, Fisher Scientific, Acros,
Frontier Scientific) and used without further purification unless
(
Mb) can be modified to produce very active and stereo-
3
,4,6,8−12
selective carbene transfer catalysts.
Multiple rounds of
1
active-site mutagenesis and reactivity screening through
directed evolution can lead to improved product yield and
stereoselectivity, though at the expense of gaining a more
fundamental understanding of how these modifications
otherwise noted. H NMR spectra were measured on either a Varian
MR400 (operating at 400 MHz), a Varian Inova 500 (operating at 500
MHz), or a Varian VNMRS 500 (operating at 500 MHz).
Tetramethylsilane (TMS) served as the internal standard (0 ppm)
1
3
,4
for H NMR measurements. Alumina gel chromatography purifica-
improve catalysis. Rational design of a catalyst through
active-site mutations and cofactor modification is a more time-
intensive approach, but it can also lead to increased activity and
selectivity of the system, and a better understanding of how the
tions were performed using neutral aluminum oxide 90. Gas
chromatography/mass spectrometry (GC/MS) analyses were per-
formed using a Shimadzu QP-2010 GC/MS equipped with a 30 m
long DB-5 column with 0.25 mm I.D. Separation method: 1 μL
injection, injector temperature: 200 °C, detector temperature: 250 °C.
Gradient: column temperature set to 60 °C for 3 min, then to 250 °C
at 20 °C/min, and held at 250 °C for 2.5 min (7.5 min for derivatives).
Total run time was 15.00 min for the aniline and styrene reactions, and
it was extended to 20.00 min for the aniline and styrene derivatives.
Protein mass spectrometry analyses were performed using an Agilent
Q-TOF HPLC-MS equipped with a Poroshell 300SB-C8 column.
Separation method: 20 μL injection, flow rate: 0.5 mL/min, gradient:
95:5 water/acetonitrile for 3 min, followed by an increase over 10 min
to 100% acetonitrile to elute the protein. Total run time was 13 min.
Porphyrin mass spectrometry analyses were performed using an
Agilent 6230 TOF HPLC-MS.
Protein Expression and Purification. Wild-type sperm whale
myoglobin (Mb), and H64A and H64D Mb variants were generously
expressed and purified by Dr. T. Matsuo’s group at the Nara Institute
of Science and Technology (NAIST) in Ikoma, Japan. The H64V
mutant was also expressed in the laboratory of Dr. Matsuo in E. coli
BL21(DE3) cells, modified from standard procedures. Briefly, cells
were grow in lysogeny broth (LB) media, (ampicillin, 50 μg/L) at 37
°C and 160 rpm until OD600 reached ∼0.8. The cell-containing media
were placed on ice immediately after expression, and cells were
pelleted by centrifugation at 8800 rpm and 4 °C using a Beckman
Coulter JA-10 fixed-angle rotor. Cell pellets were suspended in ∼100
mL of 50 mM Tris buffer, pH = 8, and 10 mL of lysis buffer were
added prior to sonication. Cells were sonicated using a Fisher
Scientific model 550 Dismembrator. The cells were pulsed six times
for 15 s each time with 1 min intervals. After a stock solution of DNase
6
,10−12
second coordination sphere influences reactivity.
Arnold
and Fasan’s catalysts are effective at catalyzing cyclopropana-
tion, N−H insertion, and other carbene transfer reactions. Cyt
P450
mutants can catalyze carbene transfer reactions in
BM3
vitro and in vivo and can achieve tens of thousands of
3
,4
turnovers. In addition, the Arnold group has developed Cyt
P450 mutants for selectively producing trans or cis cyclo-
3
,4
propanes. The Fasan group has shown that a double mutant
of Mb (H64V, V68A) can achieve over 45 000 total turnovers
6
in the cyclopropanation reaction. However, the use of the
native Fe cofactor in these cases means that many of these
catalysts are highly dioxygen sensitive, as ferrous hemes bind O2
tightly in Mb and native Cyt P450s. Compared to Fe, second-
and third-row transition-metal porphyrins such as Ru, Rh, and
Ir are less dioxygen sensitive and are highly active toward
20,23−25
3
4
carbene transfer reactions as small molecule catalysts.
Through reconstitution of the apoprotein with a new
metalloporphyrin, new catalysts for organometallic reactions
can be prepared. Recently, this approach has been successfully
7
,30
used to generate CoMb, MnMb, and IrMb catalysts.
In
parallel investigations, our work has focused on Ru catalysts for
carbene insertion reactions. We speculated that modifying
myoglobin through the insertion of a Ru porphyrin into the
active site would lead to a more active and less dioxygen-
sensitive catalyst for carbene transfer reactions.
(1 μg/mL) was added and the solution was stirred for 1 h, the
In this study, we follow the rational design approach, and we
report the design of Ru-modified Mb catalysts and their
catalytic activity toward cyclopropanation and N−H insertion.
For this purpose, we incorporated Ru mesoporphyrin IX
suspended cells were subjected to the sonication procedure again. The
lysed cells and protein mixture were centrifuged at 8800 rpm and 4 °C
for 80 min to extract the protein. The protein was initially purified by
adding ammonium sulfate up to 50% saturation, centrifuging, and then
adding more ammonium sulfate up to 90% saturation to precipitate
myoglobin. After another round of centrifugation, the protein pellet
was collected, suspended in 10 mM potassium phosphate (KPi) buffer,
pH = 6, and then dialyzed against the same buffer. The protein was
loaded onto a CM Sepharose column with 10 mM KPi buffer at pH 6,
and the collected protein fractions were then concentrated and loaded
onto a Sephadex G-50 column with 100 mM KPi buffer, pH = 7, and
(
RuMpIX) into wild-type (wt) Mb as has been previously
3
1−33
reported
and also into Mb mutants and characterized
these artificial enzymes using different spectroscopic methods.
For reactivity studies, we used ethyl diazoacetate (EDA) as a
carbene source, styrene derivatives for the cyclopropanation
reactions, and aniline derivatives for the N−H insertion
B
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
eluted with the same buffer. The proteins were characterized by UV−
vis and circular dichroism spectroscopy.
dithionite (100 mM stock solution) in potassium phosphate buffer
(100 mM, pH 7.0) was purged by bubbling argon through the solution
for 10 min in a sealed vial. All reagents and proteins were brought into
a Coy vinyl anaerobic chamber (30−50 ppm of O , 1.5−3.0% H )
Synthesis of Ruthenium Mesoporphyrin IX Dimethyl Ester
Carbonyl [RuMpIXDME(CO)]. [RuMpIXDME(CO)] was synthe-
2
2
35
III
sized as previously reported. Under an Ar atmosphere, a solution of
mesoporphyrin IX dimethyl ester (218 mg, 0.366 mmol) and
Ru (CO) (400 mg, 0.626 mmol) in dimethylformamide (DMF, 60
before running the reactions. The Ru protein was first reduced to
II
Ru through addition of 40 μL of Na S O solution, followed by the
2
2
4
3
12
addition of 4 μL of aniline or aniline derivatives from a 2 M stock
solution in methanol (with the exceptions of 4-nitroaniline and 4-
vinylaniline, which were less soluble in methanol and were therefore
dissolved in dimethyl sulfoxide), and 8 μL of EDA (2 M stock solution
in methanol), which translates to a total MeOH content of 2.65%. The
reactions were left under magnetic agitation for 18 h at room
mL) was refluxed at 160 °C. The conversion from free base porphyrin
to metalated complex was followed by thin-layer chromatography
(
TLC) and UV−vis spectroscopy. After ∼21 h, the heat was turned
off, and the solvent was removed in vacuo. The resulting solid was
dissolved in 50:50 CH Cl /ethanol (60 mL) and refluxed for another
2
2
II
3
0 min at 70 °C. The solvent was then removed by rotary evaporation.
temperature. Reactions with free [Ru MpIX] were performed in an
III
Unmetalated porphyrin was removed from the product on an alumina
column in CH Cl . The product was eluted as a bright red band after
identical procedure with the exception that the Ru Mb was replaced
III
by 400 μL of [Ru MpIX(X)] (20 μM) in potassium phosphate buffer
2
2
adding 2% methanol to the column. The product was isolated (181
(100 mM, pH 7.0). Reactions at pH 5 and 3 were also performed in
mg; 68% yield). The product was characterized by UV−vis
the same manner. The pH of the protein, buffer, and Na S O
2
2
4
−1
(
nanometers) in CH Cl (394, 517, 548), IR (cm ) in a KBr disk
solutions were all adjusted to pH 5 or 3 by addition of 3 M HCl
before combining the protein, reductant, and reagents in the anaerobic
chamber.
2
2
1
(
1924 (CO), 1718, 1634) and H NMR spectroscopy (parts per
million) in CDCl : 1.88 (t, ethyl-CH ), 3.12 (s), 3.29 (t, propionate-
3
3
CH −), 3.54, 3.57, 3.59 (s, methyl-CH ), 3.64, 3.67 (s, methoxy-CH ),
Cyclopropanation Reactions. Reactions were performed in the
same manner as the N−H insertion reactions, using 20 mM styrene
(or styrene derivatives) instead of aniline. All 2 M styrene solutions
were prepared in methanol apart from 4-nitrostyrene, which was more
soluble in dimethyl sulfoxide.
2
3
3
4
9
5
.00, (q, ethyl-CH −), 4.29 (t, propionate-CH −), 9.75, 9.80, 9.93,
2
2
.94 (s, 1 each, meso-H). Anal. Calcd (C H N O Ru): C 61.57, H
37
40
4
5
.59, N 7.76. Found: C 60.48, H 5.63, N 7.87%.
Photolysis of [RuMpIXDME(CO)]. CO was labilized from the Ru
porphyrin by measuring out ∼1 mg of [RuMpIXDME(CO)] and
dissolving the solid in ∼3 mL of dimethylformamide (DMF) in the
glovebox. The solution was irradiated with a 100 W mercury lamp
equipped with a UV filter to irradiate the sample with 351 nm light.
The solution was irradiated for ∼24 h, before removing the solvent in
vacuo. The product [RuMpIXDME] was characterized by UV−vis
Product Analysis. The reactions were analyzed by addition of 20
μL of internal standard (2-phenyl-ethanol, 1 M in methanol) to the
reaction mixture, followed by extraction with 1.5 or 3 mL of ethyl
acetate, depending on the reaction. The organic layers were dried with
magnesium sulfate overnight and later filtered and analyzed by GC/
MS (see General Procedures section for details on GC/MS analyses).
Calibration curves for quantification of the N−H insertion products of
aniline and aniline derivatives and the cyclopropanation products of
styrene and styrene derivatives were constructed using authentic
−1
(
nanometers) in DMF (392, 494, 520), and IR spectroscopy (cm ) in
a KBr disk by adding drops of the [RuMpIXDME] solution in DMF to
the KBr (1735, 1635).
Hydrolysis of [RuMpIXDME]. [RuMpIXDME] (∼1 mg) was
dissolved in 10 mL of 9:1 v/v methanol/water. NaOH (1% by
volume) was added to the solution, which was left refluxing for 1 h in
air. The reaction mixture was concentrated to ∼1 mL under vacuum,
and the pH was adjusted with HCl (3 M) to a pH of ∼7.0, which leads
standards produced synthetically (using 1−2 mol % Rh (OAc) as the
2 4
catalyst). The exception to this is the quantification of the products of
4-vinylaniline, which can undergo both N−H insertion and cyclo-
propanation. These products were quantified using the calibration
curves for the standard aniline and styrene products in the reaction
with EDA. The reactions to produce authentic standards were run
−
to the formation of a Ru(III) mesoporphyrin complex with OH or
6
,10
−
III
−
−
Cl bound. The [Ru MpIX(X)] (X = OH or Cl ) solution was
stored at −20 °C prior to reconstitution or reactivity studies. The
product was characterized by UV−vis spectroscopy (nm) in aqueous
according to the same methods previously reported.
All measure-
ments in the Results tables were performed in triplicate. Negative
control experiments were performed in the absence of catalyst and the
absence of reductant outside of the Coy glovebox.
1
00 mM pH 7 potassium phosphate (KPi) buffer (389, 517).
Preparation of Apoprotein. The heme cofactor was removed
Reactions with Ethyl Diazoacetate in a Sealed Cuvette.
Native sperm whale Mb, RuMb, and [RuMpIX(X)] (5 μM, pH 7)
were prepared in the Coy glovebox in a sealed, screw-top cuvette. In
the glovebox, 5000 equiv of Na S O in pH 7 KPi buffer and 2000
36
from Mb and the Mb mutants via Teale’s method. Briefly, the
protein (in 100 mM potassium phosphate buffer) was unfolded by
lowering the pH of the solution to ∼2.3. The heme was extracted with
2
2
4
equiv of EDA in a 0.5 M stock solution in methanol were added. For
each reaction, the cuvette was sealed and removed from the wetbox
before monitoring the kinetics by UV−vis spectroscopy for 1 h.
Protein Mass Spectrometry after Cyclopropanation. Native
horse heart Mb and RuMb (20 μM, pH 7) were prepared, and
reactions with EDA and styrene were performed in the same manner
as in the N−H insertion reactions, with the exception of the buffer
concentration being 0.2 mM instead of 100 mM to decrease the
amount of the salt added to the column during analysis. After 3 h, the
reactions were stopped, and the protein masses were analyzed by
HPLC-MS (see General Procedures). In a similar way, [RuMpIX(X)]
2
-butanone, and after heme removal the protein was dialyzed against
three buffers, each for 2 h: 10 mM potassium phosphate (KPi) buffer
at pH 6.0, 100 mM KPi buffer at pH 6.5, and 100 mM KPi buffer at
pH 7.0. The apoproteins were stored at 4 °C prior to reconstitution.
Yields typically range from 65% to 75%, calculated from the starting
concentration of the heme-bound protein.
III
Preparation of RuMb and RuMb Mutants. [Ru MpIX(X)]
(
7
1.1 equiv) was added to ApoMb (1 equiv) in KPi (100 mM, pH =
.0), and the reconstituted protein was incubated at 0 °C for ∼30−60
III
min before purification. The new Ru Mb solution was purified by
size-exclusion chromatography (Sephadex G-25) with 100 mM KPi,
pH = 7.0 as the mobile phase. The protein was concentrated, flash
frozen, and stored at −20 °C. The same procedure was applied for the
(
20 μM) was prepared and analyzed by LC-MS before and after a 15
min reaction with EDA and styrene.
III
preparation of the Ru Mb mutants (H64D, A, and V). Typical ratios
between the absorbance of the porphyrin Soret band (∼389−394 nm)
III. RESULTS AND ANALYSIS
and the absorbance at ∼280 nm of the protein were from 2.5:1 to
For the mutagenesis experiments performed here, we chose
His64 as the key target site due to the proximity of His64 to the
heme (Figure 1) and due to the enhancement in carbene
transfer reactivity detailed in previous reports upon modifica-
3
.0:1. Protein concentration was determined from the Soret band
absorbance in the UV−vis spectra, using an extinction coefficient of
−1
−1
31
9
5.7 mM cm that was previously reported.
N−H Insertion Reactions. Reactions were performed at a ∼450
μL scale using 20 μM RuMb, 20 mM aniline (or aniline derivative), 40
6
,10−12
tion of this amino acid.
The H64A and H64V mutants
mM ethyl diazoacetate (EDA), and 10 mM Na S O . The sodium
were chosen to increase the size of the protein active site and,
2
2
4
C
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
with Na S O , the Soret band sharpens and Q bands appear at
2
2
4
4
5
95 and 520 nm (Figure 2). In particular, the sharp Q-band at
20 nm is a hallmark for the Ru(II) state of RuMb. The CD
spectra show that various RuMb variants retain similar α-helical
structures as wild-type Mb after reconstitution (Figure S5).
II
3
.2. Reactivity of Free [Ru MpIX]. Initial work focused on
the catalysis of two different carbene transfer reactions by
II
[
Ru MpIX]: N−H insertion and cyclopropanation. Reactions
were run under microanaerobic conditions of 30−50 ppm of
O and 1.5−3.0% H . Control reactions without catalyst were
run for each substrate, and the results are listed in Table S1.
During the course of these studies, we realized that the Ru
was being reduced to Ru by the H inside the glovebox; thus,
2
2
III
II
2
the reactions could be catalyzed without external reductant
Figure 1. Enlarged view of the heme cofactor environment in wild-
type sperm whale Mb (image generated using PDB file 1MBO). The
His64 residue is in close proximity to the heme, and mutating this
residue can greatly impact the function and catalytic properties of the
protein. Several other nearby residues are highlighted in yellow.
added. Despite this finding, excess reductant (Na S O ) was
2
2
4
included in each of the subsequent reactions to keep the
II
reaction conditions consistent. At pH 7, the free [Ru MpIX]
complex converts aniline to 2-(phenylamino)acetate (3a) in
4
8% yield (Table 1). The yield increases to 49%, when running
II
the reaction at pH 5, and to 67% at pH 3. [Ru MpIX] further
catalyzes the cyclopropanation of styrene to ethyl 2-phenyl-
cyclopropanecarboxylate (5a) in 26% yield at pH 7 (Table 2).
The free porphyrin in buffer does not induce substantial
diastereoselectivity, with only 62% trans-cyclopropane product
observed. When the reaction is performed at pH 5, the yield
decreases to 16% and then further decreases to 11% at pH 3.
The stereoselectivity of the products increases slightly at lower
pH to 65% trans-cyclopropane product. In comparison with
[RuMpIX(X)], [FePpIX(Cl)] (hemin) exhibits lower yields of
27% and 16% for the N−H insertion and cyclopropanation
reactions, respectively. However, [FePpIX(Cl)] does demon-
strate significantly higher diastereoselectivity in the cyclo-
propanation reaction87% trans product compared to 62%
in particular, to provide a more hydrophobic environment for
the reactions investigated here. Since our substrates are rather
hydrophobic (styrene and aniline derivatives), a more hydro-
phobic active site pocket should allow for better substrate
binding and, hence, increased TONs. In contrast, the H64D
mutant places the more polar Asp residue in the active site,
making the active site more hydrophilic and potentially
allowing for stronger H-bonding interactions in the active site
compared to His.
3
.1. Protein Preparation and Analysis. The various
reconstituted proteins were characterized by UV−vis and CD
spectroscopy. The porphyrin Soret band for free [RuMpIX(X)]
at 389 nm shifts 1−5 nm to lower energy after incorporation
III
into the protein, and upon reduction of the resulting Ru Mb
Figure 2. Electronic absorption spectra comparing free [RuMpIX(X)] to the reconstituted Mb variants (a−d). UV−vis spectrum of Ru(II)Mb after
II
reduction with excess Na S O (e). Spectral comparison of the CO-bound complex [Ru(II)MpIX(CO)] and Ru Mb with CO bound (f).
2
2
4
D
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 1. Catalytic Activity of [RuMpIX], Wild-Type Sperm
Whale Ruthenium Myoglobin (RuMb), and the Mb Mutants
in the N−H Insertion Reaction with Aniline and EDA
Table 2. Catalytic Activity of [RuMpIX], Wild-Type Sperm
Whale Ruthenium Myoglobin (RuMb), and the Mb Mutants
in the Cyclopropanation Reaction with Styrene and EDA
a
b
catalyst
RuMpIX]
pH
yield (%)
TON
b
yield
(%)
% trans
diastereomer
a
[
7
5
3
7
5
3
7
5
3
7
5
3
7
5
3
7
8
7
7
48
49
67
36
17
27
32
33
30
39
19
17
52
34
35
27
21
46
55
482 ± 11
487 ± 58
671 ± 56
360 ± 37
172 ± 30
273 ± 39
320 ± 39
325 ± 58
296 ± 40
392 ± 37
191 ± 23
173 ± 16
518 ± 78
342 ± 22
346 ± 28
271 ± 38
210
catalyst
[RuMpIX]
pH
TON
7
5
3
7
5
3
7
5
3
7
5
3
7
5
3
7
7
7
7
8
7
26
16
263 ± 121
162 ± 16
114 ± 48
32 ± 12
12 ± 4.9
33 ± 5.7
5.8 ± 0.3
2.9 ± 1.1
7.4 ± 3.7
39 ± 7.0
68 ± 14
84 ± 8.0
17 ± 4.6
29 ± 9.0
22 ± 3.0
158 ± 64
248 ± 28
360 ± 22
305 ± 25
365
62
65
65
80
68
65
87
79
78
72
70
74
69
59
68
87
91
93
96
92
94
RuMb
11
RuMb
3.2
1.2
3.3
0.6
0.3
0.7
3.9
6.8
8.4
1.7
2.9
2.2
16
RuH64DMb
RuH64VMb
RuH64AMb
RuH64DMb
RuH64VMb
RuH64AMb
[FePpIX]
[
FePpIX]
c
FeMb
FeH64DMb
FeH64AMb
461 ± 35
553 ± 35
FeMb (sperm whale)
FeMb (horse heart)
FeH64DMb
25
36
a
4
50 μL scale reactions using 20 μM catalyst (0.1 mL %), 10 mM
Na S O , 20 mM 1a and 40 mM 2 conducted for 18 h at room
temperature under microaerobic conditions. Determined based on
31
c
FeH64VMb
73
2
2
4
b
FeH64AMb
29
287 ± 15
GC-MS conversion, using a calibration curve derived from isolated 3a.
Data from ref 10; reaction conditions are slightly different in this case
compared to this work.
a
4
50 μL scale reactions using 20 μM catalyst (0.1 mL %), 10 mM
c
Na S O , 20 mM 4a, and 40 mM 2 conducted for 18 h at room
temperature under microaerobic conditions. Determined based on
2
2
4
b
GC-MS conversion, using a calibration curve obtained with isolated
c
5
a. Data from ref 6; reaction conditions were slightly different in this
case compared to this work.
trans product produced by [RuMpIX(X)] at pH 7. Carbene
transfer reactions were also run with [Ru MpIX(CO)] in pH 7
II
buffer to compare to the CO-free porphyrin complex (data not
Scheme 1. Putative Mechanism for the N−H Insertion of
Aniline and the Cyclopropanation of Styrene with Ethyl
Diazoacetate Catalyzed by a Ru(II) Porphyrin with CO
Bound
II
shown). [Ru MpIX(CO)] catalyzes the N−H insertion
reaction, in up to 84% yield, and the cyclopropanation of
II
styrene in 27% yield. Using [Ru MpIX(CO)] as the catalyst
results in slightly higher diastereoselectivity relative to
II
[
Ru MpIX] with 73% versus 62% trans-cyclopropane product.
This result implies that the CO group is not replaced when the
RuCR carbene intermediate is formed in the first step of the
2
reaction, but instead, that the trans CO ligand influences the
reactivity of the bound carbene with substrate in the second
step of the reaction (Scheme 1). Hence, using porphyrin
derivatives, a further fine-tuning of the reactivity of the system
should be possible.
II
3
.3. N−H Insertion with RuMb and Mutants. Ru Mb
and the other RuMb variants catalyze the N−H insertion of
aniline in various yields (Table 1; all isolated yields). The N−H
insertion with aniline is catalyzed by wild-type Ru Mb in 36%
II
II
yield at pH 7, and in contrast to free [Ru MpIX], the yield
decreases in the lower pH reactions to 17% and 27% for pH 5
II
and 3, respectively. The Ru H64DMb catalyst exhibits similar
II
activity to Ru Mb, converting aniline to the desired product in
3
2% yield at pH 7 and 5, and in 30% yield at pH 3. Hence,
lowering the pH has no real effect on the N−H insertion
reaction catalyzed by Ru H64DMb. Ru H64VMb, which has a
larger and more hydrophobic pocket, exhibits a slightly higher
When the reactions were run at lower pH, the yield decreased
dramatically to 19% and 17% for pH 5 and 3, respectively. The
II
II
II
Ru H64AMb catalyst exhibits the highest yield of any of the
II
II
II
II
yield of 39% compared to Ru Mb and Ru H64DMb at pH 7.
catalysts at pH 7−52%. Like Ru H64VMb and Ru Mb, it also
E
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
catalyzes fewer turnovers at pH 5 and 3 (34% and 35% yield,
respectively).
producing 87% trans isomer at pH 7, significantly higher than
II
free [Ru MpIX]. The diastereoselectivity of the cyclopropane
II
Native myoglobin (termed FeMb for clarity) variants catalyze
the N−H insertion of aniline in relatively similar yields
product induced by Ru H64VMb is relatively small, ranging
II
between 70 and 74% at different pH values. Ru H64AMb
II
II
compared to Ru Mb. At pH 8, sperm whale Fe Mb catalyzes
exhibits the lowest diastereoselectivity of the different protein
catalysts, with only 69% trans isomer at pH 7, 59% at pH 5, and
II
the reaction in 21% yield, lower than the 36% yield that Ru Mb
1
0
II
II
II
achieves. Fe H64DMb and Fe H64AMb exhibit slightly
higher yields compared to their Ru-reconstituted counterparts,
catalyzing the reaction with 46% and 55% yields, respectively.
Wild-type horse heart apo-Mb was also reconstituted with
68% at pH 3 produced. In conclusion, Ru H64VMb and
II
Ru H64AMb, the two more hydrophobic proteins with the
least sterically congested active sites, demonstrate similar
II
stereoselectivity to the free [Ru MpIX] complex. These results
[
RuMpIX(CO)] to compare the N−H insertion activity to the
indicate that increased steric bulk in close proximity of the
heme, as in the H64D mutant, may increase stereoselectivity,
but at the same time, product yield is compromised in the
H64D mutant of our Ru-reconstituted catalyst. It remains to be
seen whether a bulky, but nonpolar amino acid at the H64
position could lead to increased product yield.
II
II
CO-free Ru Mb. Ru Mb(CO) achieved up to ∼10% yield,
much lower than Ru Mb. We speculate that this residual
activity results from CO loss from a small fraction of
Ru Mb(CO) during turnover. Therefore, CO inhibits activity,
likely due to the fact that in order to form the carbene
intermediate, the CO ligand would have to be replaced.
Removal of CO is inefficient, and this differs from the free
II
II
In comparison to RuMb and the RuMb variants, FeMb
variants demonstrate increased TON and diastereoselectivity
32
II
[
RuMpIX(CO)] complex, where the metal−carbene bond can
for the cyclopropanation of styrene. Fe Mb (sperm whale)
be formed in the presence of the CO ligand (in trans position
to CO).
exhibits 25% conversion to product at pH 7 and produces 91%
trans product. Horse-heart myoglobin actually achieves a higher
product yield (36%) compared to sperm-whale myoglobin,
though it exhibits the same diastereoselectivity (93% trans
3
.4. Cyclopropanation with RuMb and Mutants. The
various Ru catalysts also catalyze the cyclopropanation of
styrene with ethyl diazoacetate (EDA; see Table 2), though
mostly in relatively low yields. Ru Mb is a poor catalyst, only
catalyzing the reaction in 3.2% yield at pH 7. Yield decreases at
pH 5 to 1.2% and rebounds at pH 3 to 3.3%, but the Ru Mb
never exhibits high activity for this reaction. This is very
surprising, considering the relatively higher activity observed for
free [RuMpIX(X)] as described above. This is also different
from the N−H insertion reaction, where Ru Mb shows similar
activity compared to free [RuMpIX(X)]. This point is further
addressed in the Discussion section and originates from an
unexpected lack of stability of the catalyst. Ru H64DMb
demonstrates lower activity but higher stereoselectivity (87%
trans isomer) compared to Ru Mb. Unfortunately,
Ru H64DMb only achieves 0.6% yield at pH 7 and does not
improve at lower pH values (0.3% at pH 5 and 0.7% at pH 3).
As is the case with Ru Mb, the stereoselectivity decreases at
lower pH, but note that the stereoselectivity is difficult to
measure accurately from the GC chromatograms with such low
TON. The two variants with more hydrophobic active sites,
II
II
product) as expected. Fe H64DMb and Fe H64AMb achieve
II
II
31% yield and 29% yield, respectively. Similar to Ru H64DMb,
Fe H64DMb induces the highest diastereoselectivity of the
II
II
FeMb variants, and its selectivity is even more impressive
II
compared to Ru H64DMb, as it achieves up to 96% trans
II
product. The Fasan group previously investigated Fe H64VMb
and reported 73% conversion to product and 92% trans
II
selectivity at pH 8, although these reactions were run at higher
6
catalyst loading. Nevertheless, it is clear that FeMb variants are
much more effective at catalyzing the cyclopropanation of
styrene and EDA compared to their RuMb counterparts.
3.5. Substrate Scopes. To conduct a substrate scope
determination for the N−H insertion and cyclopropanation
reactions, several para-substituted aniline and styrene deriva-
II
II
II
II
tives were used. Ru H64AMb achieves the highest TON for the
II
N−H insertion reaction and provides a larger active-site pocket
for bulkier aniline and styrene derivatives, and thus
II
Ru H64AMb was chosen as the catalyst for investigating
II
substrate scopes. Ru H64AMb catalyzes the N−H insertion of
II
II
namely, Ru H64VMb and Ru H64AMb, exhibit higher TON
compared to Ru H64DMb. Ru H64VMb exhibits 3.9%
conversion at pH 7, which increases to 6.8% and 8.4%
4-methylaniline, 4-chloroaniline, 4-trifluoromethylaniline, and
4-vinylaniline in moderate yield (36%, 33%, 45%, and 15%,
respectively), whereas the substrates 4-nitroaniline and 4-
methoxyaniline exhibit very low yields (2.9% and 5.7%) for
their respective amine products (Table 3). The high yield for
the reaction with 4-trifluoromethylaniline is particularly
surprising, as it was expected that aniline derivatives with
II
II
II
conversion at pH 5 and 3, respectively. Ru H64AMb is a worse
II
II
catalyst compared to Ru H64VMb and Ru Mb, exhibiting a
yield of 1.7% at pH 7. Yields increase slightly to 3.0% and 2.3%
at pH 5 and 3, respectively, but still do not match the activity of
Ru H64VMb. In a separate experiment, EDA was added slowly
to Ru H64AMb at pH 7 over the course of 35 min, but this
II
more electron-withdrawing groups such as −CF would be
3
II
worse nucleophiles and thus worse substrates for interacting
causes the yield to drop even loweronly 0.8% product
with the electrophilic metal−carbene bond. Correspondingly,
conversion was observed.
4-nitroaniline with the electron-withdrawing −NO group
2
None of our first-generation artificial enzymes demonstrate
high product diastereoselectivity, but there are some notable
exhibits a very low yield, in agreement with these expectations.
In contrast, the 4-methoxyaniline with an electron-donating
II
differences between the proteins. Ru Mb exhibits good
−OCH group also exhibits a very low yield for the N−H
3
stereoselectivity80% trans isomer at pH 7, and the
stereoselectivity decreases at lower pH, down to 65% trans
isomer at pH 3. The stereoselectivity of the wild-type variant at
neutral pH is therefore higher compared to the intrinsic
insertion reaction, again negating the simple explanation that
product yield is solely correlated with the electronic nature of
the aniline. Clearly, other factors must also be at work (see
Discussion). As a substrate, 4-vinylaniline can potentially
undergo both N−H insertion and cyclopropanation. The N−
H insertion product is the only product observed in the GC/
II
stereoselectivity of free [Ru MpIX], which produces the trans
II
diastereomer at 62−65% yield. Ru H64DMb demonstrates the
II
best product stereoselectivity of all catalysts tested here,
MS chromatogram, which concurs with Ru H64AMb exhibit-
F
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
a
Table 3. Substrate Scope for the N−H Insertion Reaction
Catalyzed by RuH64AMb
Table 4. Substrate Scope for the Cyclopropanation
Reaction Catalyzed by RuH64AMb
a
a
Reaction conditions: 20 μM RuH64AMb (0.1 mol %), 20 mM 2b−f,
4
0 mM EDA, and 10 mM sodium dithionite, conducted for 18 h at
b
room temperature in microanaerobic conditions. Determined based
on GC-MS conversion, using calibration curves derived from isolated
4
b−4f.
exhibits only 52% generation of the trans cyclopropane product,
yielding nearly a 1:1 mixture of both diastereomers.
a
Reaction conditions: 20 μM RuH64AMb (0.1 mol %), 20 mM 1b−f,
4
0 mM EDA, and 10 mM sodium dithionite, conducted for 18 h at
3.6. Stability of RuMb. To better understand the low
b
room temperature in microaerobic conditions. Determined based on
GC-MS conversion, using calibration curves derived from isolated 3b−
3
II
catalytic activity of Ru Mb toward cyclopropanation, especially
in comparison to native FeMb as reported by the Fasan group
f.
(
see Introduction), the stability of the modified protein after
addition of EDA was investigated by UV−vis spectroscopy and
mass spectrometry. After the addition of reductant and EDA to
RuMb in a sealed cuvette, a rapid decrease in the Soret band
absorbance at 396 nm occurs (Figure 3b). The Soret band
absorbance decreases from 0.47 to 0.26 over the course of 20
min, and it decreases further to 0.22 after 60 min. In addition,
the Q-band absorbance at 522 nm also decreases over the
course of 60 min. In contrast, the Soret band of native sperm-
whale Mb decreases only slightly over 60 min from 0.56 to 0.45,
and minimal spectral changes occur for the Q-band at 556 nm
ing much higher product conversion when using aniline as the
substrate compared to styrene.
The cyclopropanation substrate scope was also investigated
II
using Ru H64AMb. Several styrene derivatives were examined
for cyclopropanation activity, with the functional groups
analogous to the aniline derivatives discussed earlier (Table
4
). All of the derivatives show higher activity compared to
styrene. 4-Methoxystyrene demonstrates by far the highest yield
36%) compared to the other substrates. 4-Chlorostyrene
(
II
shows a moderate conversion to the product (15%), and all
(Figure 3a). These results indicate that the Fe Mb catalyst is
other substrates exhibit lower yields7.3%, 7.1%, and 2.7% for
still active for over an hour after the addition of EDA, whereas
II
4
-methylstyrene, 4-nitrostyrene, and 4-trifluoromethylstyrene,
Ru Mb appears to be quickly degrading under the same
II
respectively. The lower yield is expected for 4-nitrostyrene and
conditions. This indicated to us that the Ru CR species that
2
4
-trifluoromethylstyrene with the electron-withdrawing −NO2
forms in the first step of catalysis may be attacking the
porphyrin itself or amino acids in the active site of the protein.
To investigate this further, mass spectra of FeMb and RuMb
were taken before and after the cyclopropanation reactions to
determine whether the metal−carbene unit could attack and
and −CF groups, but not for 4-methylstyrene with the slightly
3
electron-donating −CH group. Almost all the substrates
3
exhibit similar diastereoselectivity, yielding 67−72% of the
trans isomer. The only exception to this is 4-nitrostyrene, which
G
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 3. Electronic absorption spectra of (a) 5 μM native sperm-whale myoglobin after adding 5000 equiv of Na S O and 2000 equiv of ethyl
2
2
4
diazoacetate and (b) 5 μM RuMb under the same conditions.
Figure 4. Mass spectra of (a) horse-heart FeMb, (b) horse-heart FeMb after adding 500 equiv of reductant, 1000 equiv of styrene, and 2000 equiv of
EDA, (c) RuMb, (d) RuMb after adding 500 equiv of reductant, 1000 equiv of styrene, and 2000 equiv of EDA.
modify the protein backbone. The mass spectrum of horse-
heart myoglobin shows a deconvoluted peak at 16951.7 amu,
the mass of the protein scaffold without the native cofactor
into the reaction, which shows that the catalyst is still active
after this time period. The mass of RuMb (sperm whale)
appears at 17995.4 amu (Figure 4c). Interestingly, RuMpIX
appears to still be bound to the protein in this case. After
reaction with styrene and EDA, a larger mass peak appears at
18081.4 amu, followed by a forest of smaller peaks that appear
with higher masses, indicative of heavy protein modification
and degradation. The main signal at 18081.4 amu shows an
increased mass of 86.0 amu compared to RuMb, once again
(which is likely lost from the protein during ionization; Figure
4a). After a 3 h reaction with styrene and EDA, two larger mass
peaks are visibleat 16951.7 and 17037.9 amu (Figure 4b).
The mass of the native protein increased by 86.2 amu, which is
equivalent to the mass of EDA minus dinitrogen. Both the
unmodified and carbene-modified protein exist in solution 3 h
H
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
II
III
Figure 5. (a) Electronic absorption spectra of Ru Mb in 100 mM KPi buffer at different pH values. (b) Circular dichroism spectra of Ru Mb at
different pH values.
corresponding to the addition of one carbene unit to the
protein (or potentially the porphyrin). Some of the other
smaller peaks likely result from multiple additions of EDA to
the protein. It is unclear at this point which protein residue or
residues get modified by the metal−carbene species, but clearly
RuMb is heavily modified in the presence of EDA, with the
metal−carbene likely inserting into both the protein scaffold
and even the porphyrin, indicated by the fast bleaching of the
Soret band shown in Figure 3b. This leads to a quick loss of
catalytic competency of RuMb.
IV. DISCUSSION
II
In this paper, the reactivity of Ru Mb and corresponding distal
His (H64) mutants to mediate carbene transfer reactions in
water was investigated. Activity toward the N−H insertion of
II
II
aniline is comparable between free [Ru MpIX], Ru Mb
II
II
variants, and Fe Mb variants with Ru H64AMb and
II
Fe H64AMb achieving the highest TON (520 and 550 at pH
7
, respectively) for this reaction among our selected catalysts.
II
In contrast, TONs decrease among all of the Ru Mb variants
compared to free [Ru MpIX] and Fe Mb variants for the
cyclopropanation of styrene. The best Ru protein catalyst for
II
II
The degradation of free [RuMpIX(X)] in the presence of
EDA was also investigated by UV−vis spectroscopy and mass
spectrometry. The UV−vis spectrum demonstrates a rapid
decrease in the Soret band intensity over the course of 1 h after
the addition of 2000 equiv of EDA (Figure S9). A complete
disappearance of the free [RuMpIX(X)] mass at 666.18 m/z is
observed 15 min into the catalytic reaction between EDA and
styrene (Figure S10). From these results, clearly the Ru
porphyrin both inside and outside the protein scaffold quickly
degrades upon the addition of EDA.
II
the cyclopropanation of styrene is Ru H64VMb, achieving up
to ∼40 TON at pH 7 and ∼85 TON at pH 3, but the free
II
[Ru MpIX] demonstrates higher activity, catalyzing the
II
reaction in up to ∼260 TON at pH 7. Ru Mb and
II
Ru H64DMb induce increased stereoselectivity during the
cyclopropanation reaction, producing 80% and 87% trans
isomer, respectively. Neither of the more hydrophobic variants
II
II
Ru H64VMb and Ru H64AMb induce significant stereo-
selectivity compared to the intrinsic porphyrin stereoselectivity
of 62−65% trans isomer. Despite the low activity observed for
the cyclopropanation reaction, the Mb variants with larger and
II
The stability of Ru Mb was also investigated as a function of
pH to potentially gain insight into reactivity differences at pH 7,
II
more hydrophobic active sites (Ru H64AMb and
5
, and 3. Between pH 5.4 and 3.9, the electronic absorption
spectrum shows a significant change in the Soret band region
Figure 5a). The initial Soret band at 396 nm decreases, and a
II
Ru H64VMb) achieve higher TON for the N−H insertion
II
reaction, and Ru H64VMb also exhibits higher TON for the
(
II
cyclopropanation reaction compared to Ru Mb and
new band appears at 407 nm. This directly indicates
protonation of a basic residue in the active site of Mb, close
to the Ru-porphyrin unit. The two resulting species coexist in
the spectra at low pH. Importantly, neither one of the two
species in solution shows a UV−vis spectrum that matches with
II
Ru H64DMb (Tables 1 and 2). This is likely due to the
hydrophobic nature of the substrates aniline and styrene,
allowing for more facile entrance into the hydrophobic active
II
site of Ru H64VMb. This indicates that one way to prepare
more active catalysts (for the substrates considered here) would
be to further modify the active site to increase the
hydrophobicity. Mutagenesis of other active-site residues to
more hydrophobic amino acids will likely aid in this endeavor.
Active-site mutations around the porphyrin would also aim to
II
that of free Ru MpIX at low pH (see Figure S7), indicating that
the Ru-porphyrin is not released from the protein. The circular
III
dichroism spectra of Ru Mb show a significant decrease in the
well at 222 nm at pH 3.0 (Figure 5b). At pH 3, sperm-whale
myoglobin therefore exists in a partially unfolded intermediate
II
increase product stereoselectivity, as the Ru H64AMb and
3
7
state, and we observe a corresponding decrease in ellipticity.
II
Ru H64VMb catalysts only achieve 59−69% and 70−74% trans
II
However, it appears that the Ru MpIX cofactor is still bound in
the protein active site. Besides the UV−vis data, further
evidence that Ru-porphyrin is not released from Mb at pH 3.0
is provided by the fact that RuMpIX would catalyze the
cyclopropanation reaction in much higher yield compared to
RuMb. At pH 5.0, RuMb is still properly folded as expected.
isomer, respectively, which is comparable to the selectivity of
II
free [Ru MpIX] (62−65% trans isomer). The low stereo-
selectivity of the H64A and V mutants is likely caused by their
very open active sites with little steric strain. Addition of more
bulky, hydrophobic amino acids would be a straightforward way
to improve on this in the future.
I
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
It is also evident that the pH of the solution plays a role in
determining the TON for these carbene transfer reactions, in
particular, in the case of the N−H insertion reaction. For most
of the cyclopropanation reactions, it is difficult to discern a
trend in reactivity as a function of pH due to low catalytic
This does NOT, however, explain the discrepancy in TON and
stereoselectivity between RuMb and native sperm-whale Mb
and its variants that use iron in the active site (referred to as
FeMb here for clarity): as first reported by the Fasan group (see
Introduction), FeMb is a good catalyst for the cyclopropanation
of styrenes with EDA. For direct comparison, we used sperm-
II
activity. In the case of Ru H64VMb, however, there is a clear
increase in observed reactivity as pH decreases (4% yield at pH
II
whale Fe Mb and found that it catalyzes the cyclopropanation
7
to 8% yield at pH 3). Our results point toward two possible
of styrene in 25% yield with 91% trans product formed, which is
in stark contrast to the 3% yield and 80% trans product
produced by Ru Mb under equivalent conditions. We had
explanations for this observation. First, the acid titration study
of Ru Mb shows the appearance of a new Soret band at 407
II
II
nm at low pH, which indicates the formation of a protonated
form of RuMb below pH 5.4 (Figure 5), where the vicinity of
the Ru-porphyrin has been modified. We do not expect that
anticipated that FeMb would not catalyze carbene transfer
reactions under a 30−50 ppm O atmosphere due to the high
2
II
affinity of Fe Mb toward binding O , but the excess of
2
II
II
−
this originates from the simple Ru −OH /Ru −OH equili-
reductant allows for consumption of any dissolved O and for
2
2
brium, since the pK of water bound to divalent transition-
catalysis to proceed. The discrepancy in reactivity between
FeMb and RuMb was initially attributed to an intrinsically
lower carbene transfer activity of RuMpIX, but our data on the
reactivity of free RuMpIX versus FePpIX actually show that the
opposite is true and that the Ru-porphyrin is the more active
catalyst. We then turned to UV−vis and mass spectrometry to
determine whether this difference between RuMb and FeMb
could actually originate from differences in catalyst stability,
which is an often overlooked aspect in the field of biocatalysis
(but there is a notable exception from this: in a paper from the
Arnold group that just appeared in the literature, Cyt P450
A
metal porphyrin complexes usually lies in the 9−11 range. This
distal His may be protonated at this pH, but it is unclear if His
protonation would result in such a dramatic change in the
absorption spectrum. At pH = 3, unfolding of the protein is
further observed. Despite sperm-whale myoglobin existing in a
partially unfolded intermediate state below pH 5, our data show
that the RuMpIX cofactor is not released at pH = 3, since free
RuMpIX has characteristic UV−vis features, though it is not
entirely clear if the RuMpIX cofactor is binding at other sites of
the slightly unfolded protein instead of just at His93 at lower
pH. In any case, the protonation of the protein in the vicinity of
the Ru-porphyrin unit and/or the more dynamic structure of
the protein around pH = 3 are responsible for the observed
changes in catalytic activity at low pH, leading to an increase in
cyclopropanation activity. Second, for the N−H insertions at
lower pH, the amine group of the styrene aniline substrate will
become protonated (and this is also a function of the exact
nature of the aniline derivative), forming an anilinium cation.
4
0
degradation under carbene transfer conditions was reported).
II
II
After adding EDA to both Fe Mb and Ru Mb under identical
conditions, we observed a rapid decrease in the Soret band of
II
Ru Mb via UV−vis measurements, which we propose is due to
the metal−carbene moiety modifying the active site of RuMb
and the RuMpIX catalyst itself; the latter being indicated by the
rapid bleaching of the Soret band (Figure 3). In stark contrast,
II
Fe Mb appears relatively stable over the course of at least 1 h
Aniline has a pK of 4.58, and therefore, at pH < 5 a significant
by UV−vis spectroscopy. Further insight into this issue is
provided by mass spectrometry: after the reaction with styrene
and EDA, RuMb shows that the mass of EDA (minus
dinitrogen) has been added to the protein. The corresponding
signal in the mass spectrum is accompanied by a forest of peaks
at higher and lower masses that indicate the presence of species
where multiple EDA units have been added to the protein, plus
heavy degradation of the protein itself (Figure 4). In contrast,
a
percentage of the substrate is in the protonated anilinium form.
Since the Mb active site is hydrophobic (especially in the H64A
and H64V mutants), we can expect that the protonated
anilinium ion cannot readily enter the Mb active site, and
hence, the second step of the reaction now becomes
unfavorable at lower pH. This is exactly what we observe for
the N−H insertion reaction. In further agreement with this
II
II
idea, this trend is reversed for free [Ru MpIX], because active
the spectrum of Fe Mb under the same conditions shows only
site access is not a factor for the carbene insertion reactions of
the free Ru-porphyrin complex. In addition, in the H64D
mutant, an anionic carboxylate group is present in the active
site. This might explain why H64D does not show any
dependence of the TON on pH: in this case, the anionic
carboxylate group might allow for a facile entrance of the
anilinium cation into the active site, forming an ion pair, and
the carboxylate side chain could assist in deprotonating the
anilinium group during the N−H insertion step.
two species: the unmodified (original) protein, plus a species
where the mass of one EDA (minus dinitrogen) has been added
to the protein matrix. These results demonstrate that opposite
II
to our intuition, the Ru CR unit is actually the more
2
reactive carbene transfer catalyst compared to ironleading to
carbene transfer to the Ru porphyrin and the protein matrix
and, in this way, causing relatively fast catalyst degradation.
Therefore, the low TON observed for cyclopropanation with
RuMb and its variants is caused by the low concentration of
styrene (due to solubility) under the reactions conditions,
making this reaction slow and unable to compete with catalyst
degradation. In contrast, the more stable FeMb is able to
II
Clearly, the Ru Mb catalysts are much better at catalyzing
the N−H insertion of aniline compared to the cyclo-
propanation of styrene, despite both reactions using EDA as
the carbene source and the fact that both reactions are catalyzed
sustain catalysis much longer, resulting in higher TON after the
II
II
by free [Ru MpIX] in good yields. However, aniline is 120
typical 18 h of reaction time. It is likely that the Ru CR
2
times more soluble than styrene in water, and thus aniline has
easier access to the protein active site in comparison to
moiety also quickly modifies residues in the active site of the
protein, changing the geometry of the active site and, in this
way, contributing to the observed difference in product
stereoselectivity. On the basis of our results, all RuMb catalyst
is in fact degraded in under 1 h. On the other hand, the Ru Mb
catalysts exhibit similar reactivity to the free [Ru MpIX] for
3
8,39
styrene.
Addition of styrene to the buffer and protein
solution causes the solution to become slightly cloudy,
indicating the poor solubility of styrenes in the aqueous
phase. This would explain why N−H insertion reactions with
RuMb give much higher TON compared to cyclopropanations.
II
II
N−H insertion, since in this case the more soluble and more
J
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
II
II
reactive aniline substrate is able to intercept the Ru CR unit
Fasan and co-workers, as well as to our Fe H64AMb and
2
II
before carbene transfer to the protein matrix or the porphyrin
can take place. We therefore conclude that the low solubility
and reactivity of styrene and some of the styrene derivatives
under our reaction conditions is a major factor in the low
activity for the cyclopropanation reaction of RuMb. RuMb
therefore has great potential to be developed into a highly
reactive carbene transfer catalyst, if these stability issues can be
overcome. This is a main focus of our ongoing work. The
recent paper by Arnold and co-workers on Cyt P450s is
certainly encouraging in this regard, showing that catalyst
stability may be able to be improved.
The substrate scope studies yielded a surprising result for the
N−H insertion reaction: 4-trifluoromethylaniline exhibits a
higher yield (45%) compared to the other substrates, with the
Fe H64DMb variants. This is largely due to the stability issues
6
II
discussed above. Ru H64DMb achieves the highest diaster-
eoselectivity (87% trans product) but exhibits low activity
regardless of pH and lower diastereoselectivity compared to
every one of the FeMb variants (all >90% trans product).
Ru H64VMb exhibits higher cyclopropanation activity relative
to the other Ru Mb catalysts but still relatively low activity
overall (only 4−8% yield depending on catalyst and pH) and
diastereoselectivity (70−74%). The only substrate that is able
to be efficiently converted to the cyclopropane product by
Ru H64AMb is 4-methoxystyrene, the most electron-rich and
most soluble styrene derivative studied in this work.
II
II
40
II
In summary, wild-type Mb and three His64 mutants of Mb
were reconstituted with [RuMpIX(X)]. These catalysts exhibit
moderate activity in the N−H insertion reaction of aniline
derivatives and low activity in the cyclopropanation of most of
the styrene derivatives studied. RuH64DMb is able to achieve
the best diastereoselectivity, but it exhibits low activity. Our
results show that one must be cautious when analyzing the
outcomes of organometallic reactions in water: besides the
reactivity differences that result from electronic effects,
solubility and acid/base equilibria can influence the outcome
of reactions dramatically. We also observed an interesting
difference between our RuMb catalysts and analogous iron
systems studied in the literature, where application of Ru leads
to significantly suppressed cyclopropanation reactivity. Our
studies show that this is not due to a lack of carbene transfer
reactivity of RuMb, but to the contrary, that this is caused by
the catalyst being too reactive, leading to degradation of the
protein and the porphyrin and, hence, fast catalyst deactivation.
This is especially a factor when styrene substrates are used (for
cyclopropanation) that have a low solubility in water. This
result is encouraging, as it shows that RuMb has a great
potential as a highly active carbene transfer catalyst, if the
stability issues can be overcome. This work is in progress. Our
results also show that Mb mutants with hydrophobic active-site
pockets show moderate increases in activity, at least in
conjunction with the nonpolar styrene and aniline derivatives
studied here. Hence, even more active catalysts should result
from double or triple Mb mutants that further increase the
hydrophobicity of the active site. To induce increased
stereoselectivity in the cyclopropane products, bulky hydro-
phobic groups should be used in these mutants. This is the next
logical step in rational catalyst design that should lead to
improved carbene transfer catalysts, and corresponding
investigations are in progress.
exception of aniline. The electron-withdrawing −CF group
3
was expected to decrease the nucleophilicity of the aniline and
thus exhibit less reactivity with the electrophilic Ru-carbene
unit. On the other hand, the 4-methoxyaniline substrate
exhibits a yield of only ∼6% for the reaction, which is much
lower than expected for a substrate with the more electron-
donating −OCH group. For the styrene derivatives with the
3
same para substitutions, the relative reactivity is inverted: 4-
methoxystyrene is the most reactive substrate (35% yield),
whereas 4-trifluoromethylstyrene is the least reactive (∼3%
yield). Clearly, besides electronic effects of the substrate, there
are additional effects at work that influence the outcome of the
reaction as discussed above. This includes, in particular,
differences in pK (anilines) and water solubility. In the case
of the −OCH group, 4-methoxystyrene may be more water-
soluble compared to styrene and some of the other substrates,
which would make it more available (higher effective
concentration) at the very beginning of the reaction, increasing
the yield before catalyst degradation occurs. On the other hand,
A
3
4
-methoxyaniline acts as a poorer substrate in the N−H
insertion reaction, because the methoxy group increases
basicity, which leads to a higher pKA (pKA = 5.29) and
increased formation of the anilinium cation, hindering access of
the substrate to the active site. It seems therefore evident that
substrate solubility under catalytic conditions may play a
significant role in reactivity. Despite 4-methylstyrene possessing
a slightly electron-donating −CH group, it is expected to have
lower solubility compared to some of the other substrates, and
the yield with this substrate is comparable to 4-nitrostyrene
3
(
both ∼7%), much lower than for 4-chlorostyrene (15%). On
the other hand, the size of the substrate does not seem to be an
issue for reactivity, as the larger −CF3 group in 4-
trifluoromethylaniline does not lead to decreased reactivity
for this substrate.
ASSOCIATED CONTENT
■
In comparison to the work of the Fasan group, which
investigated carbene transfer reactions with iron-containing Mb
and several Mb mutants, some significant differences between
*
S
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.6b03148.
II
II
the two systems are evident. [Ru MpIX] and Ru Mb achieve
similar TON in the N−H insertion of aniline compared to free
hemin or wt Mb. However, the catalyst that the Fasan group
III
Additional spectroscopic characterization of [Ru MpIX-
(X)], [Ru MpIX], and RuMb and its variants studied
here, GC-MS traces for cyclopropanation and N−H
insertion reactions, and calibration curves (PDF)
II
II
tested for the N−H insertion of aniline was the Fe H64V/
V68A Mb double mutant, which is able to achieve up to 6150
TON. Future studies on a corresponding double mutant of
RuMb will show if the Ru system can match or surpass this
result for Fe. [Ru MpIX] exhibits slightly higher cyclo-
1
0
II
AUTHOR INFORMATION
■
propanation activity to free hemin (∼25% yield vs ∼15%
yield, respectively), but Ru Mb and our His64 mutants all fare
II
Corresponding Author
worse compared to wild-type Mb and the Mb mutants used by
*E-mail: lehnertn@umich.edu.
K
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
ORCID
(17) Doyle, M. P. Exceptional Selectivity in Cyclopropanation
Reactions Catalyzed by Chiral Cobalt(II)-Porphyrin Catalysts. Angew.
Chem., Int. Ed. 2009, 48, 850−852.
Nicolai Lehnert: 0000-0002-5221-5498
Notes
The authors declare no competing financial interest.
(18) O’Malley, S.; Kodadek, T. Asymmetric Cyclopropanation of
Alkenes Catalyzed by a Chiral Wall Porphyrin. Tetrahedron Lett. 1991,
2, 2445−2448.
19) Collman, J. P.; Rose, E.; Venburg, G. D. Reactivity of
3
(
ACKNOWLEDGMENTS
Financial support for this work from the Univ. of Michigan,
Associate Professor Support Fund, is gratefully acknowledged
■
Ruthenium 5,10,15,20-Tetramesitylporphyrin Towards Diazoesters -
Formation of Olefins. J. Chem. Soc., Chem. Commun. 1993, 934−935.
(20) Galardon, E.; Le Maux, P.; Simonneaux, G. Cyclopropanation of
(
to N.L.). M.W.W. thanks the Rackham Graduate School and
alkenes with ethyl diazoacetate catalysed by ruthenium porphyrin
the Department of Chemistry (both Univ. of Michigan) for
financial support in the form of Graduate Student Fellowships.
Special thanks is given to Dr. T. Matsuo and Mr. M. Ishida of
the Nara Institute for Science and Technology for providing the
various Mb mutants and their corresponding plasmids. We also
thank J. Kumutima (Lehnert laboratory) for expressing some of
the H64V myoglobin mutant that was used in this study.
complexes. Chem. Commun. 1997, 927−928.
(21) Lo, W. C.; Che, C. M.; Cheng, K. F.; Mak, T. C. W. Catalytic
and asymmetric cyclopropanation of styrenes catalysed by ruthenium
porphyrin and porphycene complexes. Chem. Commun. 1997, 1205−
1
206.
(22) Galardon, E.; LeMaux, P.; Simonneaux, G. Insertion of ethyl
diazoacetate into N-H and S-H bonds catalyzed by ruthenium
porphyrin complexes. J. Chem. Soc., Perkin Trans. 1 1997, 2455−2456.
(23) Che, C. M.; Huang, J. S. Ruthenium and osmium porphyrin
carbene complexes: synthesis, structure, and connection to the metal-
mediated cyclopropanation of alkenes. Coord. Chem. Rev. 2002, 231,
REFERENCES
■
(
1) Prier, C. K.; Arnold, F. H. Chemomimetic Biocatalysis: Exploiting
the Synthetic Potential of Cofactor-Dependent Enzymes To Create
1
(
51−164.
24) Anding, B. J.; Ellern, A.; Woo, L. K. Olefin Cyclopropanation
Catalyzed by Iridium(III) Porphyrin Complexes. Organometallics
012, 31, 3628−3635.
25) Intrieri, D.; Caselli, A.; Gallo, E. Cyclopropanation Reactions
Mediated by Group 9 Metal Porphyrin Complexes. Eur. J. Inorg. Chem.
011, 2011, 5071−5081.
26) Nicolas, I.; Le Maux, P.; Simonneaux, G. Asymmetric catalytic
cyclopropanation reactions in water. Coord. Chem. Rev. 2008, 252,
27−735.
27) Che, C. M.; Huang, J. S.; Lee, F. W.; Li, Y.; Lai, T. S.; Kwong, H.
New Catalysts. J. Am. Chem. Soc. 2015, 137, 13992−14006.
(
2) Ward, T. R. Artificial enzymes made to order: Combination of
computational design and directed evolution. Angew. Chem., Int. Ed.
008, 47, 7802−7803.
3) Coelho, P. S.; Brustad, E. M.; Kannan, A.; Arnold, F. H. Olefin
2
(
2
(
Cyclopropanation via Carbene Transfer Catalyzed by Engineered
2
(
Cytchrome P450 Enzymes. Science 2013, 339, 307−310.
(
4) Coelho, P. S.; Wang, Z. J.; Ener, M. E.; Baril, S. A.; Kannan, A.;
Arnold, F. H.; Brustad, E. M. A serine-substituted P450 catalyzes
highly efficient carbene transfer to olefins in vivo. Nat. Chem. Biol.
7
(
2
013, 9, 485−487.
5) Steinreiber, J.; Ward, T. R. Artificial metalloenzymes as selective
catalysts in aqueous media. Coord. Chem. Rev. 2008, 252, 751−766.
6) Bordeaux, M.; Tyagi, V.; Fasan, R. Highly Diastereoselective and
L.; Teng, P. F.; Lee, W. S.; Lo, W. C.; Peng, S. M.; Zhou, Z. Y.
Asymmetric inter- and intramolecular cyclopropanation of alkenes
catalyzed by chiral ruthenium porphyrins. Synthesis and crystal
structure of a chiral metalloporphyrin carbene complex. J. Am. Chem.
Soc. 2001, 123, 4119−4129.
(
(
Enantioselective Olefin Cyclopropanation Using Engineered Myoglo-
bin-Based Catalysts. Angew. Chem., Int. Ed. 2015, 54, 1744−1748.
(28) Cui, X.; Xu, X.; Jin, L. M.; Wojtas, L.; Zhang, X. P.
(
7) Key, H. M.; Dydio, P.; Clark, D. S.; Hartwig, J. F. Abiological
Stereoselective radical C-H alkylation with acceptor/acceptor-sub-
stituted diazo reagents via Co(II)-based metalloradical catalysis. Chem.
Sci. 2015, 6, 1219−1224.
catalysis by artificial haem proteins containing noble metals in place of
iron. Nature 2016, 534, 534−537.
(
8) Renata, H.; Wang, Z. J.; Kitto, R. Z.; Arnold, F. H. P450-catalyzed
(29) Xu, X.; Zhu, S. F.; Cui, X.; Wojtas, L.; Zhang, X. P. Cobalt(II)-
asymmetric cyclopropanation of electron-deficient olefins under
Catalyzed Asymmetric Olefin Cyclopropanation with alpha-Ketodia-
aerobic conditions. Catal. Sci. Technol. 2014, 4, 3640−3643.
zoacetates. Angew. Chem., Int. Ed. 2013, 52, 11857−11861.
(
9) Wang, Z. J.; Peck, N. E.; Renata, H.; Arnold, F. H. Cytchrome
(30) Bordeaux, M.; Singh, R.; Fasan, R. Intramolecular C(sp3)-H
P450-catalyzed insertion of carbenoids into N-H bonds. Chem. Sci.
014, 5, 598−601.
10) Sreenilayam, G.; Fasan, R. Myoglobin-catalyzed intermolecular
carbene N-H insertion with arylamine substrates. Chem. Commun.
015, 51, 1532−1534.
11) Tyagi, V.; Bonn, R. B.; Fasan, R. Intermolecular carbene S-H
insertion catalysed by engineered myoglobin-based catalysts. Chem. Sci.
015, 6, 2488−2494.
12) Tyagi, V.; Fasan, R. Myoglobin-Catalyzed Olefination of
Aldehydes. Angew. Chem., Int. Ed. 2016, 55, 2512−2516.
13) Doyle, M. P. Catalytic Methods for Metal Carbene Trans-
formations. Chem. Rev. 1986, 86, 919−939.
14) Li, Y.; Huang, J. S.; Zhou, Z. Y.; Che, C. M.; You, X. Z.
amination of arylsulfonyl azides with engineered and artificial
2
(
myoglobin-based catalysts. Bioorg. Med. Chem. 2014, 22, 5697−5704.
(31) Paulson, D. R.; Addison, A. W.; Dolphin, D.; James, B. R.
Preparation of Ruthenium(II) and Ruthenium(III) Myoglobin and the
Reaction of Dioxygen, Carbon Monoxide, with Ruthenium(II)
Myoglobin. J. Biol. Chem. 1979, 254, 7002−7006.
2
(
2
(
(32) Srivastava, T. S. A carbon monoxide derivative of ruthenium (II)
myoglobin probe of heme protein conformation. Biochim. Biophys.
Acta, Protein Struct. 1977, 491, 599−604.
(
(33) Winter, M. B.; McLaurin, E. J.; Reece, S. Y.; Olea, C.; Nocera, D.
G.; Marletta, M. A. Ru-porphyrin protein scaffolds for sensing O . J.
2
(
Am. Chem. Soc. 2010, 132, 5582−5583.
Remarkably stable iron porphyrins bearing nonheteroatom-stabilized
carbene or (Alkoxycarbonyl) carbenes: Isolation, X-ray crystal
structures, and carbon atom transfer reactions with hydrocarbons. J.
Am. Chem. Soc. 2002, 124, 13185−13193.
(34) Springer, B. A.; Sligar, S. G. High-Level Expression of Sperm
Whale Myoglobin in Escherichia-Coli. Proc. Natl. Acad. Sci. U. S. A.
1987, 84, 8961−8965.
(35) Galardon, E.; Le Maux, P.; Simonneaux, G. Cyclopropanation of
Alkenes, N−H and S−H Insertion of Ethyl Diazoacetate Catalysed by
Ruthenium Porphyrin Complexes. Tetrahedron 2000, 56, 615−621.
(36) Teale, F. W. J. Cleavage of the Haem-Protein Link by Acid
Methylethylketone. Biochim. Biophys. Acta 1959, 35, 543.
(37) Yang, A.-S.; Honig, B. Structural Origins of of pH and Ionic
Strength Effects on Protein Stability, Acid Denaturation of Sperm
Whale Apomyoglobin. J. Mol. Biol. 1994, 237, 602−614.
(
15) Intrieri, D.; Le Gac, S.; Caselli, A.; Rose, E.; Boitrel, B.; Gallo, E.
Highly diastereoselective cyclopropanation of alpha-methylstyrene
catalysed by a C-2-symmetrical chiral iron porphyrin complex. Chem.
Commun. 2014, 50, 1811−1813.
(
16) Huang, L. Y.; Chen, Y.; Gao, G. Y.; Zhang, X. P.
Diastereoselective and enantioselective cyclopropanation of alkenes
catalyzed by cobalt porphyrins. J. Org. Chem. 2003, 68, 8179−8184.
L
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(
38) Yalkowsky, S. H.; He, Y. Handbook of Aqueous Solubility Data:
An Extensive Compilation of Aqueous Solubility Data for Organic
Compounds Extracted from the AQUASOL DATABASE; CRC Press
LLC: Boca Raton, FL, 2003; p 262.
(
39) Yalkowsky, S. H.; He, Y.; Jain, P. Handbook of Aqueous Solubility
Data, 2nd ed.; CRC Press: Boca Raton, FL, 2010; p 468.
40) Renata, H.; Lewis, R. D.; Sweredoski, M. J.; Moradian, A.; Hess,
(
S.; Wang, Z. J.; Arnold, F. H. Identification of Mechanism-Based
Inactivation in P450-Catalyzed Cyclopropanation Facilitates Engineer-
ing of Improved Enzymes. J. Am. Chem. Soc. 2016, 138, 12527.
M
DOI: 10.1021/acs.inorgchem.6b03148
Inorg. Chem. XXXX, XXX, XXX−XXX