The role of a dipeptide outer-coordination sphere on H2- production catalysts: Influence on catalytic rates and electron transfer

Article abstract of DOI:10.1002/chem.201202849

The outer-coordination sphere of enzymes acts to fine-tune the active site reactivity and control catalytic rates, suggesting that incorporation of analogous structural elements into molecular catalysts may be necessary to achieve rates comparable to thos

Full text of DOI:10.1002/chem.201202849

DOI: 10.1002/chem.201202849
The Role of a Dipeptide Outer-Coordination Sphere on H2-Production
Catalysts: Influence on Catalytic Rates and Electron Transfer
Matthew L. Reback,[a] Bojana Ginovska-Pangovska,[a] Ming-Hsun Ho,[a] Avijita Jain,[b]
Thomas C. Squier,[a] Simone Raugei,*[a] John A. S. Roberts,*[a] and Wendy J. Shaw*[a]
Abstract:
The
outer-coordination
was either 3-(meta- or para-aminophe-
nyl)propionic acid terminated as an
acid, an ester or an amide. Dipeptides
consisted of one of the non-natural
amino acids coupled to one of four
amino acid esters: alanine, serine, phe-
nylalanine or tyrosine. All of the cata-
lysts are active for hydrogen produc-
tion, with rates averaging ~1000 sꢀ1,
40% faster than the unmodified cata-
lyst. Structure and polarity of the ali-
phatic or aromatic side chains of the C-
terminal peptide do not strongly influ-
ence rates. However, the presence of
an amide bond increases rates, suggest-
ing a role for the amide in assisting cat-
alysis. Overpotentials were lower with
substituents at the N-phenyl meta posi-
tion. This is consistent with slower elec-
tron transfer in the less compact, para-
substituted complexes, as shown in dig-
ital simulations of catalyst cyclic vol-
tammograms and computational mod-
eling of the complexes. Combining the
current results with insights from previ-
ous results, we propose a mechanism
for the role of the amino acid and di-
sphere of enzymes acts to fine-tune the
active site reactivity and control cata-
lytic rates, suggesting that incorpora-
tion of analogous structural elements
into molecular catalysts may be neces-
sary to achieve rates comparable to
those observed in enzyme systems at
low overpotentials. In this work, we
evaluate the effect of an amino acid
and
dipeptide
outer-coordination
sphere on [Ni(PPh2NPh-R2)2]2+ hydrogen
production catalysts. A series of 12
new complexes containing non-natural
amino acids or dipeptides was prepared
to test the effects of positioning, size,
polarity and aromaticity on catalytic
activity. The non-natural amino acid
Keywords: electrochemistry
mogeneous catalysis hydrogen
production outer coordination
sphere · peptide catalysts
· ho-
·
peptide
based
outer-coordination
·
sphere in molecular hydrogen produc-
tion catalysts.
Introduction
matched with synthetic analogs. Acidic, basic, polar and aro-
matic residues have all been implicated in modulating or
tuning enzyme active sites.[1–2]
The protein scaffold or outer-coordination sphere of metal-
loenzymes plays a critical role in their resulting activity.
Channels control the timed transport of electrons, substrates
and products to and from the buried active site, and an envi-
ronment is created around the active site that stabilizes li-
gands, influences redox potentials and controls the electric
polarizability.[1] This environment can be achieved with ster-
ics, hydrogen bonding, and/or by global effects such as the
overall dielectric, the result of which creates a fine-tuned en-
vironment for achieving rates and efficiencies that are un-
The importance of the environment created by the outer-
coordination sphere has been demonstrated by DeGradoꢀs
group, who used artificial enzymes to show the stabilizing
influence of outer-coordination sphere hydrogen bonds,[3]
and in work by Luꢀs group, who have demonstrated the abil-
ity to tune redox potentials over a range of 700 mV with
modest changes in the environment near the active site of a
copper azurin protein.[4] The importance of the active site
environment in enzyme catalysts suggests that controlling
the environment around molecular catalysts may also pro-
vide significant benefit. This approach also has the potential
to provide insight into the role of the enzymatic scaffold be-
cause molecular catalysts are likely to have fewer complica-
tions due to their comparative simplicity.
[a] Dr. M. L. Reback, M. Sc. B. Ginovska-Pangovska, Dr. M.-H. Ho,
Dr. T. C. Squier, Dr. S. Raugei, Dr. J. A. S. Roberts, Dr. W. J. Shaw
Pacific Northwest National Labs
Hydrogenases are a class of enzymes of interest for their
ability to efficiently and reversibly reduce H+ to H2.[5]
A
Richland, WA 99354 (USA)
E-mail: wendy.shaw@pnnl.gov
second-coordination sphere pendant amine has been identi-
fied from the crystal structure of the active site of [FeFe]-hy-
drogenase (Ia).[6] Studies of functional mimics have shown
this to be essential for catalysis, and have led to catalysts
that can meet (Ib) or exceed (Ic) the rates of the enzyme,
but at higher overpotentials (300–600 mV).[7] These results
suggest a pivotal role of the hydrogenase outer-coordination
[b] Dr. A. Jain
Current address:
Department of Chemistry
Indiana University of Pennsylvania
Indiana, PA 15705 (USA)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201202849.
1928
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

FULL PAPER
sphere in achieving energy efficiency in producing hydrogen,
but the influence of the outer-coordination sphere on these
mimics has been largely unexplored.
acid, an ethyl ester or an amide. The dipeptide systems have
an additional esterified amino acid residue coupled to the
non-natural amino acid, including alanine ethyl ester
(AlaOEt, with a methyl side chain at the central carbon),
serine methyl ester (SerOMe, with a methoxy group at the
central carbon), phenylalanine methyl ester (PheOMe, with
a benzyl at the central carbon), or tyrosine methyl ester
(TyrOMe, with a para-hydroxybenzyl at the central carbon).
This set of complexes presents a matrix of comparisons: of
the role of terminal functionality with the monopeptide
complexes; of size effects with the amino acid versus dipep-
tide complexes; of the effects of aromaticity and polarity
with the different esterified dipeptides; and of functional
group positioning by comparison of para- and meta-substi-
tuted complexes. The comparisons are based on measure-
ments of turnover frequency (TOF) at different potentials,
voltammetric studies of the catalysts themselves, and com-
plementary molecular dynamics (MD) studies. This paper
addresses each of these comparisons to provide a mechanis-
tic understanding of the role of the environment created by
the outer-coordination sphere on molecular catalysts.
The environment around the
active site of hydrogenase,
along with many metalloen-
zymes, is highly polar and
charged.[1,8] Within 9 ꢂ of the
bimetallic cluster in the active
site of [NiFe]-hydrogenase,
there are 60% acidic, basic or
polar residues, compared to
typical proteins which have a
core consisting of 60% hydro-
phobic residues.[9] Although sys-
tematic mutations to test the
role of the environment around
the active site have not been
Figure 1. Amino acid and dipeptide complexes investigated in this study.
explored, we have been study-
ing the effect of outer-coordina-
tion sphere amino acids in mo-
lecular catalysts to determine if they can enhance catalysis,
as well as provide insight into the enzyme. Initial investiga-
tions on the role of acidic and basic residues around func-
tional mimics of hydrogenase (Ib), attached through a non-
natural amino acid linker to generate dipeptide containing
catalysts, [Ni(PPh2NPh-dipeptide2)2]2+, showed an order of magni-
tude variation in rates, as much as five times faster than the
unmodified catalyst (Ib).[10] The enhancement in rates was
thought to be due to an ability to concentrate protons
around the active site, although a clear mechanism was not
easily established.
This paper presents a series of functional hydrogenase
mimics having the formula [Ni(PPh2NPh-R2)2]2+ (IIa and IIb,
Figure 1), featuring a series of amino acid or dipeptide sub-
stituents at the N-phenyl rings attached at either the meta-
or para- positions. The non-natural amino acid systems (3-
(3-or 4-aminophenyl)propionic acid) are terminated as an
Results
Synthesis and characterization: Cyclic diphosphines contain-
ing the non-natural amino acid residues 3-(meta- or para-
aminophenyl)propionic acid were prepared with an amide
(PPh2Nm/pNN-NH 2), ethyl ester (PPh2Nm/pNN-OEt2), or carboxylic
2
acid termination (PPh2Nm/pNNA2), following previously descri-
bed protocols.[10a] The dipeptide-functionalized PPh2NNNA–amino
ligands were prepared from PPh2Nm/pNNA by coupling
acid ester
2
2
the amino acid esters AlaOEt, SerOMe, PheOMe, or
TyrOMe using standard TBTU/HOBT/DIPEA peptide cou-
pling protocols, as shown in Scheme 1, reaction 1.[11] The li-
gands were isolated as off-white solids in low to moderate
yields (20–60% for the dipeptides) and high purity, and
were characterized using elemental analysis, 31P{1H}, H and
1
1H-TOCSY NMR spectroscopy, affording data consistent
with the proposed structures. Nickel complexes were pre-
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1929

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
Electrochemical studies: Each
of the dipeptide ester, amino
acid ester and amino amide
complexes reported in this
study showed two distinct, re-
versible reduction waves as-
signed to the NiII/I and the NiI/0
couples (Figure 3). Diffusion
control was demonstrated by a
linear increase in the non-cata-
lytic current (ip) with the square
1
=
2
root of the scan rate (u ) for
the NiII/I couple (Figure S1).
The meta- and para-substituted
non-natural amino acid com-
plexes, m/p-NNA, show quasi-
reversible NiII/I and irreversible
NiI/0 couples. Redox potentials
acid ester
acid
Scheme 1. Synthesis of the PPh2NNNA–amino
ligands (top). Synthesis of the [Ni(PPh2NNNA–amino
ester)2]2+
2
(E1/2), reported vs. the Cp2Fe+
complexes (bottom). R represents the amino acid side chain.
/Cp2Fe couple at 0.0 V, and
peak-to-peak separations (DEp)
are presented in Table S2.
pared by reacting two equivalents of a PPh2NNNA-aminoacidester
The modifications to the outer-coordination sphere were
specifically selected to minimize the impact on the electron-
ics of the metal. Differences in redox potentials are small, as
expected given the distance and number of bonds separating
the structural modifications from the metal center. E1/2
values are within 20 mV for both the NiII/I potentials (ꢀ0.83
2
ligand with one equivalent of [NiACHTUNGTRNEUNG
(MeCN)6]2+ in acetonitrile,
resulting in a red solution from which a moderately air sen-
sitive metal complex was obtained in high yield (70–90%)
and high purity, Scheme 1, reaction 2. The [Ni(PPh2Nm/pNNA-
aminoacidester
1
2)2)2+ complexes were characterized by 31P{1H}, H
1
and H-TOCSY NMR spectroscopy, mass spectrometry and
electrochemistry, all of which are consistent with the pro-
posed structures and similar to other reported complexes of
this type.[5c,7b,c,10] Hereafter the nickel complexes are refer-
red to by the position and identity of the amino acid or di-
peptide substituent: for example m-NNA-SerOMe refers to
the nickel complexes that are substituted off the phenyl ring
in the meta-position and contain the non-natural amino acid
coupled to serine methyl ester.
The 31P{1H} NMR spectra of the metal complexes all
show a single phosphorous resonance between 4 and 6 ppm
(Figure 2A), as previously reported.[10a] The 31P resonances
for the meta-substituted complexes (~4.6 ppm) are consis-
tently shifted upfield as compared to the para-substituted
complexes (~5.2 ppm), indicating increased electron density
at the phosphorous atom, as expected (Table S1). The re-
stricted exchange process observed for other dipeptide com-
plexes of this type was not observed for the ester and amide
complexes reported here.[10]
1
Two-dimensional H-TOCSY spectra were used to assign
the 1D-1H NMR spectra (Figure 2B). In particular, the pres-
ence of the amide, identified by its coupling with the Ca
proton of the amino acid ester, confirms that the added
amino acid ester was coupled to the non-natural amino acid
Figure 2. A) 31P{1H} spectrum of m-NN-OEt shown is representative of
all of the nickel complexes reported in this paper. B) 1H TOCSY was
used to assign the 1H NMR spectra (m-NNA-AlaOEt is shown). The
presence of coupling between the amide resonance (dark asterisk) and
the Ca resonance (light asterisk), shown by the solid lines, provided
direct confirmation that the amino acid was coupled.
of the PPh2NNNA ligand (Scheme 1). Comparing the integra-
2
tion of the amide protons to the Ca protons, as well as
other protons in the ligand, confirms complete coupling.
The amide-NH and Ca proton chemical shifts are reported
in Table S1.
1930
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
Figure 3. Cyclic voltammogram of p-NNA-AlaOEt (0.7 mm, scan rate u=
1 Vsꢀ1) in acetonitrile (0.2m Et4N+BF4ꢀ) showing reversible NiII/I and
NiI/0 redox couples typical of all the complexes reported with the excep-
tion of the meta- and para-non-natural acid complexes showing quasire-
versible NiII/I couples.
to ꢀ0.85 V) and the NiI/0 potentials (ꢀ1.03 to ꢀ1.05 V). Po-
tentials are uniformly more positive for the meta- versus
para-substituted complexes, by 10 mV for the NiII/I couples
and by 20 mV for the NiI/0. Values of DEp for the NiI/0 redox
couples are between 60 and 66 mV, however, the DEp values
for the NiII/I couples range from 63 to 120 mV (scan rate u=
1.0 Vsꢀ1), and are generally larger for the para-substituted
complexes. This reflects differences in electron transfer ki-
netics, as discussed in detail below.
Figure 4. Addition of acid followed by the addition of water into a solu-
tion of A) p-NNA-SerOMe and B) m-NNA-SerOMe, results in an in-
crease in current, indicative of catalytic activity. These results are typical
for those observed for each of the catalysts.
each non-natural amino acid or dipeptide substituted nickel
catalyst until the ratio of icat/ip stopped increasing (the acid
independent region,[5c, 7b, c,10] requiring ~0.08 to 0.11m acid
for the catalyst concentrations employed, Figure S2). Fig-
ure 4A and B show voltammograms recorded at acid inde-
pendence for the para- and meta-substituted serine com-
plexes, p-NNA-SerOMe and m-NNA-SerOMe,
Catalytic hydrogen production: All of the reported nickel
complexes are active electrocatalysts for the production of
H2 (Table 1). Protonated dimethylformamide (DMFH+, pKa
6.1 in acetonitrile)[12] was added to acetonitrile solutions of
which are representative of the electrochemistry for
Table 1. Turnover frequencies (TOF, sꢀ1) for hydrogen production by [Ni(PPh2Nm/pNNA–
each of the complexes. As has been reported pre-
viously,[7b, c, 10a] adding water in limited quantities in-
creases the catalytic current until a new maximum
is attained at water concentrations between 0.5 and
1.5m (Figure 4 and Table 1). Catalytic cyclic voltam-
mograms were collected at 1 Vsꢀ1, in the scan rate
independent region (Figure S3).
2)2]2+ at overpotentials of 300 and 400 mV, and the percent increase in rate
amino acid ester
between the overpotentials.
icat-A (ꢀ0.9 V)
icat-B (ꢀ1.1 V)
OP (~300 mV)
OP (~400 mV)
Metal complex
TOF
TOF with
H2O
TOF
TOF with
H2O
[%] increase
in TOF vs
OP
[sꢀ1
]
[sꢀ1
]
(ꢁ10%) (ꢁ10%)
(ꢁ10%) (ꢁ10%)
w/ H2O[a]
Catalytic rates were determined at two potentials,
icat-A and icat-B, as indicated with the arrows in
Figure 4, due to the distinctly different shape of the
catalytic wave for the para-substituted complexes.
Catalytic currents either increased gradually (para-
substituted complexes, Figure 4A), or rapidly, lead-
ing to a plateau (meta-substituted complexes, Fig-
ure 4B). Turnover frequencies were determined
using equation 1 (Methods Section) and are report-
ed in Table 1. At their fastest, all of these catalysts
have similar TOFꢀs, with an average of ~1000 sꢀ1 at
the larger applied potentials in the presence of both
acid and water (Table 1, icat-B), ~40% faster than
the fastest rate for the unmodified parent complex
(having phenyl substituents at the P and N atoms).
para
p-NNA-TyrOMe
p-NNA-AlaOEt
p-NNA-SerOMe
p-NN-NH2
110
120
110
180
177
640
670
600
640
700
440
650
290
230
210
290
250
200
610
1200
1100
1100
1030
950
87
64
83
60
35
107
29
p-NN-OEt[b]
p-NNA-PheOMe 90
910
840
p-NNA[b,c]
280
meta
m-NNA-PheOMe 160
m-NNA-SerOMe 160
m-NN-NH2
m-NNA-AlaOEt 120
m-NNA-TyrOMe 130
930
880
918
920
770
640
560
720
220
190
230
180
220
220
670
NA
1180
1080
1100
1040
1040
780
27
23
20
13
35
21
41
NA
182
m-NN-OEt
160
500
m-NNA[c]
790
NA
[Ni
N
[a] ((icat-Bꢀicat-A)/icat-A ꢃ100). [b] Data for these complexes were originally reported by
Jain et al.[10a] The values reported in this paper were re-run under identical catalytic
conditions used for the amino acid ester complexes for consistency. [c] These com-
plexes showed quasi reversible II/I and I/0 couples and were run under slightly differ-
ent conditions (see Experimental Section). [d] Data is taken from Kilgore et al.[7b]
The role of the amide functional group: While dif-
ferences in catalytic rates are not large, an increase
in rate for any complex with an amide bond is ob-
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1931

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
Figure 5. TOF values of the meta-substituted complexes are displayed as
a function of amino acid or dipeptide substituent, showing a moderate
but clear enhancement of the rate when an amide bond is present.
served for the meta-substituted complexes (Figure 5); meas-
urably slower rates are observed for the amino acid and
amino acid ester substituted complexes. A similar trend is
found for the para-substituted complexes, where the amino
acid and amino acid ester complexes have slower rates than
the other para-substituted catalysts, but the step function is
not observed, possibly due to other effects such as slower
electron transfer, described in more detail below. This obser-
vation suggests a clear role for the amide functionality in hy-
drogen production. While the amide bond appears to have
an influence on rates, the ligand size, polarity and aromatici-
ty do not contribute significantly to the rates of hydrogen
production, and suggest that design principles that attempt
to modify the dielectric constant near the metal will not
result in substantial rate enhancements due to this effect
alone.
Figure 6. Concentration of water as a function of the distance from the
solvent-accessible surface of the catalyst for a 1.5m water solution in ace-
tonitrile (top), and snapshots from the MD simulations showing the dis-
tribution of water within 3 ꢂ of the complex for p-NNA-Lys (bottom
left) and p-NNA-AlaOEt (bottom right). The m/p-NNA-Lys concentrates
significantly more water near the surface of the molecule than the amide
containing p-NNA-AlaOEt or the ester containing p-NN-OEt complexes.
Similar trends were observed for other concentrations of water (Fig-
ure S4).
complex surface (blue and red lines in Figure 6), with any
differences being within our statistical uncertainty. Movies
from the MD simulations showing the distribution of water
around p-NNA-Lys and p-NNA-AlaOEt are provided in the
Supporting Information.
A possible role of the amide functional group could be to
concentrate water and/or protons around the catalyst. This
could also be the role of dipeptide complexes containing
acidic and basic functional groups reported previously.[10b]
To further investigate this possibility, MD simulations were
performed on three representative complexes in acetonitrile
at various water concentrations. Complexes with no amide,
m/p-NN-OEt, and with an amide, m/p-NNA-AlaOEt, were
studied. Additionally, a dipeptide complex containing the
basic amino acid lysine, m/p-NNA-Lys, the fastest dipeptide
catalyst reported to date,[10b] was also studied, in its proto-
nated form.
Water was observed to interact extensively with the proto-
nated amino group of lysine, and, consequently, concentrate
more strongly near the surface of the m/p-NNA-Lys catalyst
(green lines in Figure 6). This finding provides evidence
that, in acetonitrile solutions, charged amino groups are able
to concentrate water, consistent with our hypothesis. In con-
trast, the water distribution around the m/p-NN-OEt and m/
p-NNA-AlaOEt complexes is not significantly different
from the bulk, aside from small fluctuations close to the
The effect of water on catalytic rates: The addition of water
resulted in more significant rate enhancements for the func-
tionalized amino acid and dipeptide substituted complexes
reported here (as much as 7.5 times) than observed for the
parent complex (1.3 times), and are reminiscent of the pre-
viously reported complex having a -CH2P(O)ACTHNUTRGNE(UNG OEt)2 sub-
stituent at the N-phenyl para-position (rates were reported
as 1850 vs 500 sꢀ1 under dry conditions).[7b] The rates for all
of these complexes in the absence of water are slower than
for the parent catalyst, suggesting a limitation in these cata-
lysts, possibly steric, that is not present in the parent catalyst
and that is overcome by the presence of water.
The role of substituent size and positioning: Electrochemis-
try studies: As shown in Figure 4, catalytic currents, and
thus TOFꢀs, increase more rapidly with increasingly negative
applied potential for the meta-substituted (Figure 4B) versus
the para-substituted (Figure 4A) complexes. This correla-
tion, observed for all of the systems examined under both
wet and dry conditions, is illustrated in Figure 7 and Table 1
1932
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
Figure 7. Comparing the % increase in TOF as a function of overpoten-
tial (Table
1 and equation therein), the para-substituted complexes
(black) show that larger complexes are more inhibited at lower overpo-
tentials, indicating that more energy is required to turn larger catalysts
over. The same trend is not observed for the meta-substituted complexes
(blue).
by a comparison of TOF values at ꢀ0.9 and ꢀ1.1 V versus
Cp2Fe+/0. This is not due to resistance effects since the maxi-
mum catalytic currents for the meta- and para-complexes
are the same. The meta-substituted complexes show little in-
hibition at moderate potentials, and the degree of inhibition
shows essentially no substituent size dependence. In con-
trast, the para-substituted complexes show a marked inhibi-
tion that is more severe with larger substituents, suggesting
that the effect is unique to the para-substituted systems and
is steric in origin.
The observation of plateau currents in voltammograms of
electrocatalytic systems, such as that seen in Figure 4B, is
consistent with steady-state reaction kinetics with rapid elec-
tron transfer. Electron transfer rates generally increase ex-
ponentially with increasing applied potential.[13] At suffi-
ciently high potentials, electron transfer is rapid compared
to turnover and the current becomes a function of the rate
of chemical steps in the catalytic cycle.[14] For electron trans-
fers with larger barriers, a larger overpotential is required;
for an electrocatalytic system, this will register as a more
gradual increase in current with applied potential, such as
that seen in Figure 4A, consistent with the observations re-
ported here.
To test the hypothesis that slow electron transfer is re-
sponsible for the observed substituent positioning effect on
the increase of current with applied potential (Figure 4,
Table 1), we examined the voltammetry of m/p-NNA-
TyrOMe and m/p-NNA-SerOMe catalysts without added
substrate as a function of the scan rate (scan rate u=1–
40 Vsꢀ1). Figure 8 shows voltammograms obtained at
10 Vsꢀ1 along with digital simulations for m/p-NNA-
TyrOMe. Experimental and simulated voltammograms of
the m/p-NNA-TyrOMe complexes at the lowest and highest
scan rates are shown in Figure S5.
Figure 8. Cyclic voltammograms of A) meta- and B) para-NNA-TyrOMe
complexes (0.7 mm) in acetonitrile (0.2m nBu4N+BF4ꢀ) without added
acid. Experimental data (background-subtracted, black lines) was collect-
ed at a 1 mm dia. glassy carbon electrode; simulated voltammograms
(red circles) were obtained by least-squares refinement of electron trans-
fer kinetic parameters.
tron transfer rates, as borne out in digital simulations of
these voltammograms. The estimated values of ks (the elec-
tron transfer rate constant at E=E8 for the NiII/I couple)
were 0.036(4) and 0.0132(2) cmsꢀ1, respectively, for the
meta- and para-substituted complexes (confidence intervals
at the 2 ꢃ s level), with Butler–Volmer transfer coefficients
a of 0.214(9) and 0.270(5) (Table S3). Electron transfer was
found to be considerably faster for the NiI/0 redox couples of
both the meta- and para-substituted complexes; kinetic pa-
rameters were statistically indistinguishable, with ks =
0.126(9) and 0.136(9) cmsꢀ1 and a=0.34(3) and 0.30(3), re-
spectively. The digital simulation and least-squares refine-
ment procedures are described in the Experimental Section.
To examine the influence of electron transfer kinetics on
the catalytic response, we used the above estimates of ks
and a to model a prototypical EC’ catalytic reaction. The
model consisted of a reduction of the catalyst by one elec-
tron at the electrode and a chemical re-oxidation of the cat-
alyst by a substrate molecule occurring at a rate of 250 sꢀ1,
approximately as observed for these catalysts without added
water. Complete simulation parameters are given in Table
S5. The resulting simulated catalytic waves, along with the
response expected for fast electron transfer, are shown in
Figure 9A. The experimentally observed responses (catho-
dic sweep only) are shown in Figure 9B. The qualitative sim-
ilarity between the simulated and experimental catalytic
waves shown in Figure 9 indicates that the difference in
electron transfer kinetics determined for these catalysts will
result in differences in the catalytic responses that are simi-
lar in kind and in magnitude to those observed experimen-
tally.
These voltammograms show a substantial difference in
the shape of the NiII ! NiI reduction wave, with the peak of
this wave merging into the NiI ! Ni0 wave for the para-sub-
stituted complex. This is consistent with a difference in elec-
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1933

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
Figure 10. Probability distribution of the radius of gyration for the m/p-
NNA-AlaOEt complexes shows that the meta-substituted complex is
more compact than the para-substituted complex. Representative struc-
tures from the molecular dynamics trajectories are shown below (the m-
NNA-AlaOEt structure is gray and transparent).
Figure 9. A) Simulated cyclic voltammograms for a prototypical one-elec-
tron electrocatalytic process (TOF=250 sꢀ1) with electron transfer kinet-
ic parameters ks and a as estimated by digital simulation for the NiII/I
redox couple of meta-NNA-TyrOMe (dashed trace), para-NNA-TyrOMe
(solid trace), and for a fast electron transfer (ks =1ꢃ104 cmsꢀ1, a=0.5;
dotted trace; see Table S5 for details). B) Corresponding experimental
cyclic voltammograms (cathodic sweep only) for meta- (dashed trace)
and para-NNA-TyrOMe (solid trace) in MeCN (0.1m nBu4N+PF6ꢀ) with
sufficient acid to afford pseudo-first order kinetics, and with no added
water. The simulated voltammograms shown in panel A were not ob-
tained by refinement against the data in panel B. These data provide
strong evidence that electron transfer kinetics explain the observed dif-
ferences in the catalytic waves for meta- and para-substituted complexes.
core. However, the radius of gyration,[15] which provides a
measure of the compactness of the molecules, was found to
be different for the two complexes, where the meta-substi-
tuted complex was observed to be more compact
(Figure 10).
The compactness of the meta-substituted complexes is due
to the substituents folding around the outside of the catalyst
instead of pointing outward from the catalyst as observed in
the para-substituted complexes. In addition, the meta-substi-
tuted complex displays a larger flexibility of the dipeptide
chains as indicated by the root mean square displacement
from the average structure of 3.47 ꢂ when compared to the
corresponding quantity for the para-substituted complex of
2.50 ꢂ. The different flexibility of the two complexes is
shown with an overlay of the MD trajectories in Figure 11.
NiI complexes show similar properties (data not shown).
A similar analysis was performed with the m/p-SerOMe
complexes. A smaller but statistically significant difference
in the NiII/I electron transfer kinetic parameters for the meta
and para-substituted complexes was observed for these com-
plexes as well. Estimates of ks were 0.085(4) and 0.043(3)
cmsꢀ1 for the meta- and para-substituted complexes, respec-
tively, with a values of 0.18(2) and 0.24(1) (Table S4). For
the NiI/0 couples, initial refinements afforded ks > 0.2 cmsꢀ1,
and large confidence intervals for both ks and a. This sug-
gested that electron transfer for the NiI/0 was under equili-
brium control for both complexes at all scan rates. Experi-
mental and simulated voltammograms of the m/p-NNA-
SerOMe complexes at the lowest and highest scan rates are
shown in Figure S6.
Amino acids as proton relays: To explore whether the di-
peptide substituents can act as proton relays, we carried out
umbrella sampling free energy simulations for the m/p-
NNA-Lys and m/p-NNA-AlaOEt (Figure 12). In the para-
substituted complexes, the energetic cost to bring the car-
boxylic acid functionality of p-NNA-Lys or the carbonyl
oxygen of the ester group of p-NNA-AlaOEt (N-to-O con-
tact) within hydrogen bonding distance (~3.5 ꢂ) of the
pendant amine is 7.5 and 5.6 kcalmolꢀ1, respectively. Both
of the meta-substituted complexes, m-NNA-Lys and m-
NNA-AlaOEt complexes show even smaller energy costs
The role of substituent positioning: Computational model-
ing: To provide insight into the role of substituent position-
ing (meta vs. para) on the rate of electron transfer, geomet-
rical parameters from MD simulations on the m/p-NNA-
AlaOEt complexes were analyzed. Our simulations showed
minimal differences in the geometry of the [NiACTHNUTRGNEUNG
(PPh2NPh2)2]2+
1934
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
Figure 12. Relative free energy cost to bring the functional groups of the
peptide substituents within hydrogen bonding distance (3.5 ꢂ) of the
pendant amine for (top) para-substituted complexes and (bottom) meta-
substituted complexes. The energy of the complex as a function of the N-
to-O distance, where N is the pendant amine and O is the carbonyl
oxygen on the alanine ester or on the carbonyl oxygen of the acid on
lysine, are shown in red and blue, respectively. The green line shows the
free energy as a function of the N-to-N distance for the m/p-NNA-Lys
complexes (between the pendant amine and the protonated amino group,
shown in the insert for p-NNA-Lys). These data show that the outer-coor-
dination sphere amino acids can act as proton relays to deliver protons to
the pendant amine.
Figure 11. Overlay of the structures from MD trajectories of the meta-
(top) and para- (bottom) substituted NNA-AlaOEt catalysts. The uncon-
strained nature of the dipeptide ligands is evident based on the large
phase space covered. The meta-substituted complex is more flexible then
the para-substituted complex, evidenced by the meta-substituted ligands
2+
2
accessing the full space around the Ni
(PPh2NPh
)
core.
2
for the N-to-O contact than the corresponding para-substi-
tuted counterparts, with values of 2.8 and 3.6 kcalmolꢀ1, re-
spectively. The reduced energy cost might be a consequence
of the more compact nature of the meta-substituted ligands.
While the barriers for both meta- and para-substituted com-
plexes are low enough to be accessed by structural fluctua-
tions of the molecule, in no case were stable interactions
(minima on the free energy surface) observed between the
functional groups of the substituents and the pendant amine.
The energy cost to bring the protonated amino group of
the lysine for the m/p-NNA-Lys complex within ~3.5 ꢂ of
the pendant amine was also investigated. This was found to
be 8.5 kcalmolꢀ1 for both the meta- and para-substituted
complex (Figure 12, N-to-N contact), also reasonably acces-
sible via thermal fluctuations of the molecule in solution.
The similarity in this case between the meta- and para-sub-
stituted complexes is due to the protonated amino group of
the lysine in both complexes preferring to interact with the
surrounding medium (acetonitrile and water). The larger en-
ergies observed for the N-to-N contact as compared to the
N-to-O contacts are most likely due to the insufficient
energy gain from forming a hydrogen bond with the pendant
amine, that does not compensate for the loss of the confor-
mational entropy and the reduced interaction with the envi-
ronment.
The significant difference in the energy cost between the
meta- and para-substituted complexes to achieve an N-to-O
contact indicates that the meta-substituted complexes have
easier access to the pendant amines. If protonation of the
pendant amines involves a proton transfer from protonated
carboxyl or carbonyl groups, a faster catalytic rate would be
expected for the meta-substituted complexes. Since this is
not observed experimentally (Table 1), we propose that the
protons are not likely delivered to the pendant amine in this
manner. Proton transfer from the protonated amino group
of m/p-NNA-Lys complexes is accessible since the free
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1935

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
energy cost of 8.5 kcalmolꢀ1 to achieve the N-to-N contact
can be thermally overcome, and is a possible proton transfer
pathway.
sphere p-interactions from phenylalanine and tyrosine resi-
dues,[2e] and polar groups are shown to be important in stabi-
lizing active sites.[2f] The dielectric effect of the outer-coordi-
nation sphere has also been proposed to be important in en-
zymes.[1,2d] The results obtained here suggest that the global
dielectric around the active site is not important for influ-
encing catalysis in the present systems.
Discussion
Metalloenzymes are highly efficient catalysts, performing a
wide variety of chemically challenging reactions with an
overall efficiency typically exceeding that of their molecular
catalyst counterparts. The active site of the enzyme (the pri-
mary-coordination sphere) unquestionably plays an essential
role in catalysis. However, the secondary-coordination
sphere, including atoms close to but not directly interacting
with the metal, and the outer-coordination sphere which in-
cludes the rest of the protein scaffold, also play essential
roles in the resulting catalytic process. In enzymes, the
outer-coordination sphere participates in catalysis by con-
trolling selective substrate transport and electron transfer as
well as by controlling reactivity with subtle influences such
as hydrogen bonding stabilizing unusual ligand structures,
controlling the redox potential of the metal or the pKa of a
given residue.[1,16] NMR and X-ray crystallography have elu-
cidated many of the complex features of the protein scaffold
in enzymes, but there is still much to be learned about how
enzymes function with such apparent simplicity, with poten-
tial applications to improving molecular catalysts. Because
they contain fewer functionalities, molecular catalysts pro-
vide a reasonable scaffold with which to test some of these
features.
Understanding the role that chemically attached function-
al groups in the outer-coordination sphere can have on con-
trolling the environment around the active site of molecular
catalysts and developing mechanistic insight into the ob-
served phenomenon are important goals. The studies report-
ed herein were undertaken to evaluate the role of individual
amino acid functional groups, to increase the diversity of
amino acid functional groups studied so further trends could
be established, and to assess whether re-positioning the di-
peptide would influence catalysis. Combining the studies
here with the observations from previous work allows us to
develop a mechanism for the contribution of the uncon-
strained outer-coordination sphere introduced into these
molecular catalysts.
Non-natural amino acid esters in either the meta- or para-
position resulted in rates only slightly faster than the parent
catalyst. The presence of an amide bond results in a nearly
uniform increase in rate of 40% (Figure 5 and Table 1). Be-
cause we investigated the C-terminal esterified version of
the dipeptide, eliminating an influence from a free acid
group, and because we see no influence on catalysis with the
addition of polar or aromatic groups, the increase in rate
can be attributed to the presence of the amide functional
group.
Consideration of the present data alongside prior results
reveals interesting trends. The order of rate enhancement
versus functional group is summarized as follows: bases >
acids > amides > esters. Amide functional groups enhance
catalysis while ester groups appear to have a negligible
effect. For instance, previously reported data (Table S6)[10b]
show that the protected aspartic acid and glutamic acid sub-
stituted complexes have similar rates to the protected ala-
nine and phenylalanine complexes reported in this paper.
They both have two esters instead of one, neither of which
shows an enhancement in rate. Multiple amides, found in
the lysineACTHNUGRTNEUNG(Boc)OMe substituted complex (Table S6), en-
hance rates even more than a single amide. In contrast to
esters, free acids do enhance catalysis, increasing the ob-
served rates by 1.5 to 3 times. Free amines increase the rates
even more than free acids, as the lysine substituted complex
demonstrates.
Our mechanistic interpretation is that the amino acid/
ester complexes help to concentrate protons and water, and
in some cases enhance proton delivery. This is demonstrated
with our computational models, which show that the lysine
substituted complex, the fastest amino acid substituted cata-
lyst to date, shows a significant increase in water concentra-
tion compared to the p-NN-OEt or the p-NNA-AlaOEt.
The increased local concentration of protons and water
could enhance catalysis alone, since both water and protons
are shown to increase catalytic rates.[7b,10a] Additionally, MD
studies of the p-NNA-Lys complex suggest that thermal fluc-
tuations in the dipeptide substituents may bring them close
enough to deliver a proton directly to the pendant amine.
The unconstrained nature of the ligands necessitates that
this would be a transient conformation, however, the low
energy cost to achieve this structural rearrangement suggests
that proton delivery may be occurring through this pathway
some of the time, at least for the basic and possibly acidic
complexes, if not for the ester terminated complexes. The
faster rates observed for complexes with basic amino acids
compared to those with acidic amino acids may be partially
due to the lower pKa of amines in acetonitrile (~17–20 vs
~20–25)[17] lowering the barrier for proton delivery.
The role of functional groups on hydrogen production: The
presence of polar or aromatic side chains has no measurable
impact on rate when compared to their non-polar or non-ar-
omatic counterparts (serine and alanine, or tyrosine and
phenylalanine, for example; Table 1). This is in contrast to
previous findings that show the importance of similarly posi-
tioned acidic or basic groups in enhancing catalytic rates
(Table S6, taken from Jain et al.).[10b] This suggests that the
global dielectric or p-interactions created by the nearby, but
unconstrained polar or aromatic groups, do not have a sig-
nificant influence on the catalytic activity. Metalloenzymes
have been shown to be influenced by outer-coordination
1936
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
A similar ability to concentrate water was not computa-
tionally predicted for the p-NNA-AlaOEt, a complex con-
taining an amide bond that is shown to enhance catalysis.
We speculate that the amide bond is enhancing catalysis by
a similar mechanism, but that the computational studies are
insensitive to the smaller changes in rate for complexes con-
taining an amide (~40%) compared to the lysine substituted
complex (350%), due to the statistical error inherent to
those studies. It is also possible that the neutral amide is
more able to concentrate protons than water, an investiga-
tion outside the scope of this study.
comparing the larger tyrosine complexes with the smaller
serine complexes. Computational modeling suggests that
there is a comparative compactness of the meta-substituted
complexes, however, this structural difference could also in-
fluence both rearrangement and proximity. Therefore, the
present results indicate that ligand size effects on proximity
of the Ni center to the electrode and intramolecular interac-
tions influencing barriers to rearrangement cannot be distin-
guished and may both contribute to the observed electron
transfer rate. Nonetheless, the electrochemistry experiments
provide compelling evidence that the higher overpotential
needed for the para-substituted complexes is due to slower
electron transfer than for the meta-substituted complexes.
Of significance is that the overpotential can be recovered by
modifying the positioning; this is a significant advancement
in our understanding of the outer-coordination sphere on
molecular catalysts.
Effect of substituent positioning on H2 production: Dipep-
tide substituents were attached in both meta- and para-posi-
tions of the catalyst N-phenyl groups to examine the role of
the outer-coordination sphere structure and connectivity on
catalytic rates and overpotentials. As seen in Table 1, sub-
stituent positioning did not influence the rates of hydrogen
production at large applied potentials but showed a substan-
tial effect on the potentials needed to achieve these rates.
All of the para-substituted complexes required larger ap-
plied potentials to attain maximal turnover frequencies, as
noted previously.[10b]
This observation is hypothesized to reflect a difference in
electron transfer kinetics. If electron transfer is fast com-
pared to turnover, the current will rise rapidly over a
narrow range of potentials, as illustrated in the simulated
catalytic wave shown in Figure 9A, dotted trace. Slow elec-
tron transfer can be overcome by the application of larger
potentials, so the rate of increase in current with applied po-
tential reports on the rate of electron transfer compared to
turnover. Digital simulation of CVs collected with m/p-
NNA-SerOMe and m/p-NNA-TyrOMe complexes with scan
rates from 1 to 40 Vsꢀ1 in the absence of acid revealed
faster electron transfer rates for the meta-substituted serine
and tyrosine complexes than the para-substituted complexes
(Figure 8, S5 and S6). Simulated cyclic voltammograms of a
prototypical one electron catalytic process using these elec-
tron transfer rate estimates showed that overpotentials at
Ep/2 increase with slower electron transfer, in agreement
with the experimental observations (Figure 9).
Redox kinetics could be influenced either by ligand size,
with larger ligands limiting the proximity of the metal
center to the electrode, or through steric or hydrogen bond-
ing interactions between ligands that could influence rear-
rangement barriers. The reduction of NiII to NiI in particular
forces a change in the phosphine binding geometry from
roughly square planar to tetrahedral,[18] so electron transfer
kinetics would be more evident for the NiII/I couple than for
the NiI/0 couple. The NiI/0 couples for the serine substituted
complexes were too fast to measure with the present data,
however with the tyrosine-substituted systems, the NiII/I
couple shows a positioning effect whereas the NiI/0 couple
does not, supporting the rearrangement hypothesis. If prox-
imity to the electrode were the dominant effect, the larger
complexes would be expected to have slower electron trans-
fer for both the NiII/I and the NiI/0, as is also observed in
Providing insight into the enzyme: The lack of dependence
of the rates on the polarity of the dipeptide is somewhat un-
expected given the ability of these groups to form hydrogen
bonds which would allow them to concentrate water or pro-
tons. It is also unexpected in light of the highly hydrophobic
nature of the [NiFe]-hydrogenase active site, 30% of which
is due to polar side chains such as serine and tyrosine. It
does suggest that a global dielectric effect is not important
to catalysis in the molecular catalysts mimic, and may also
not play an important role in the enzyme. However, the
local dielectric created by the precise positioning afforded
by the large protein scaffold in hydrogenase is likely an es-
sential aspect of the outer-coordination sphere that is not
achieved in the molecular catalysts reported here, demon-
strated by the conformational freedom of the dipeptides
shown in Figure 11. The impact of positioning in the meta-
versus para- orientation highlights the influence of position-
ing within the chemical bonding framework. Positioning
within the 3-dimensional framework is expected to be equal-
ly important, as demonstrated by the transient proton deliv-
ery from the flexible dipeptide to the pendant amine. Future
efforts to constrain the position of specific side chains will
be important in evaluating this effect in hydrogenase en-
zymes, as well as in molecular catalyst mimics. Overall, the
insights gained here suggest that the structured protein scaf-
fold provides advantages in both atomic control of the envi-
ronment around the active site as well as maintaining facile
electron transfer.
Conclusion
Twelve new catalysts for hydrogen production containing a
peptide based outer-coordination sphere have been synthe-
sized and characterized electrochemically, structurally and
computationally. The outer-coordination sphere has been
shown to modulate both rates and overpotentials of hydro-
gen production. The bulky dipeptide ligands increased barri-
ers to electron transfer and thus required higher overpoten-
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1937

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
under vacuum to yield a dark yellow oil (216 mg, 1.19 mmol, 36%).
1H NMR (CD2Cl2): d = 7.11, 6.61, 6.54 (m, 4H, N-Ph), 4.14 (q, 2H,
COOCH2), 3.73 (b, 2H, NH2), 2.87 (dd, 2H, N-PhCH2), 2.61 (dd, 2H,
CH2COO), 1.27 ppm (t, 3H, CH2CH3).
tials for catalysis compared to the unmodified complex, with
the largest effect seen for the para-substituted complexes.
Re-positioning the dipeptide resulted in recovering the
lower operating potential, due to the more compact struc-
ture of the meta-substituted species. Polar or aromatic side
chains, in the absence of acidic backbone groups, did not
appear to enhance catalysis beyond the addition of the
amide group, implying that a global dielectric does not en-
hance catalysis. However, amide, acidic or basic groups do
enhance catalysis up to five-fold, which we attribute to the
ability of these functional groups to concentrate water and
protons near the active site and to facilitate proton transfer.
While the global effects observed here are important, the re-
sults suggest that the precise positioning afforded by the
protein scaffold is important in affording fast rates at low
overpotentials.
PPh2NmNNA2: To ethanol (100 mL) was added bismethylhydroxy phenyl-
phosphine (0.515 g, 3.03 mmol) and 3-(3-aminophenyl)propionic acid
(0.500 g, 3.03 mmol). The solution was slowly heated to 768C and allowed
to stir overnight. The resulting white precipitate was collected and
washed with ethanol and diethyl ether and dried in vacuo (0.81 g,
1.35 mmol, 89%). The compound was not soluble in common organic sol-
vents but was soluble in basic chloroform (pH>10). 31P{1H} NMR (basic
CDCl3): d = ꢀ54.66 ppm (s); 1H NMR (basic CDCl3): d = 7.66–6.50 (m,
18H, P-Ph, N-Ph), 4.45 (dd, 4H, PCH2N), 3.95 (dd, 4H, PCH2N), 2.86
(m, 4H, CH2C(O)N), 2.62 ppm (m, 4H, N-PhCH2); PPh2NmNN-OEt was
2
prepared in an analogous manner, see Supporting Information.
PPh2NpNN-NH22: 3 equiv TBTU (241 mg, 0.752 mmol) and 3 equiv HOBT
(102 mg, 0.752 mmol) were added to a dichloromethane solution contain-
ing one equivalent of PPh2NpNNA (150 mg, 0.251 mmol) and 8 equiv
2
DIPEA (258 mg, 2.0 mmol). The mixture was stirred for 20 min and then
4 equiv ammonium chloride (54.0 mg, 1.0 mmol) were added and stirred
overnight. The resulting white solid was filtered off and washed with co-
pious amounts of water to remove any residual ammonium chloride, fol-
lowed by washing with acetonitrile and diethyl ether. The resulting white
solid was collected and dried in vacuo (121.9 mg, 0.205 mmol, 82%).
31P{1H} NMR ([D6]DMSO): d = ꢀ51.10 ppm (s); 1H NMR ([D6]DMSO):
d = 7.68–7.5 (m, 10H, P-Ph), 7.22 (s, 2H, C(O)NH2), 7.00 (dd 4H, N-
Ph(m)), 6.69 (s, 2H, C(O)NH2), 6.59 (d, 4H, N-Ph(o)), 4.53 (m, 4H,
PCH2N), 4.12 (m, 4H, PCH2N), 2.64 (m, 4H, CH2C(O)N), 2.24 (m, 4H,
N-PhCH2); elemental analysis (%) calcd for: C 68.44, H 6.42, N 9.39;
Experimental Methods
1
Instrumentation and methods: H and 31P{1H} NMR spectra were record-
ed on Varian VNMRS or Inova spectrometers operating at 300 MHz or
1
1
500 MHz H frequency at 238C. All H signals were internally referenced
to the residual solvent protons and 1H TOCSY were used to assign the
1H spectra. The 31P{1H} NMR spectra were externally referenced to phos-
phoric acid (0 ppm). Electrospray ionization (ESI) and chemical ioniza-
tion (CI) mass spectra were collected at the Indiana University Mass
Spectrometry Facility on a Waters/Micromass LCT Classic using anhy-
drous solvents and inert atmosphere techniques. MALDI-MS were col-
lected on a Waters LR used in reflectron mode, with a-cyano-4-hydroxy-
cinnamic acid as the matrix, at the Protein Chemistry Technology Center,
University of Texas, Dallas, TX. Elemental analyses were carried out by
Atlantic Microlab, Norcross, Ga.
found: C 68.14, H 6.23, N 9.48; PPh2NmNN-NH2 was prepared in an analo-
2
gous manner; see Supporting Information.
PPh2NpNNA-AlaOEt (AlaOEt=dl-alanine ethyl ester): 2 equiv TBTU (161 mg,
2
0.502 mmol) and 2 equiv HOBT (60 mg, 0.502 mmol) were added to a di-
chloromethane solution containing one equivalent of PPh2NpNNA (150 mg,
2
0.251 mmol) and 2.2 equiv DIPEA (71.1 mg, 0.552 mmol). The mixture
was stirred for 20 min and then 2 equiv DL-alanine ethyl ester hydro-
chloride (77.0 mg, 0.502 mmol) were added and stirred overnight. The
solution was extracted with water to remove residual chloride and dried
over anhydrous magnesium sulfate. The solution was passed through
celite and the solvent was removed under vacuum. The resulting yellow/
white solid was washed with acetonitrile and recrystallized from diethyl
ether/dichloromethane (4:1). The resulting white solid was collected and
dried in vacuo (80.3 mg, 0.10 mmol, 40%). 31P{1H} NMR (CDCl3): d =
Materials: All reactions were performed under an inert atmosphere of ni-
trogen using standard Schlenk techniques or in a glove box. Solvents
were de-oxygenated and purified using an Innovative Technology, Inc.
PureSolv solvent purification system. [D3]-Acetonitrile (Cambridge Iso-
tope Laboratories, 99.5% D) was vacuum-distilled from P2O5. CDCl3
(Cambridge Isotope Laboratories, 99.5% D) was degassed and stored
over molecular sieves. Water was dispensed from a Millipore MilliQ puri-
fier at 18 MW and sparged with nitrogen. 2-(1H-Benzotriazole-1-yl)-
1,1,3,3-tetramethyluronium tetrafluoroborate (TBTU), 3-(4-aminophe-
nyl)propionic acid, 3-(3-aminophenyl)propionic acid, ammonium chloride
and trimethylsilyl chloride (Aldrich) were used as received. Diisopropy-
lethyl amine (DIPEA) (Aldrich) was degassed prior to use by the freeze-
pump-thaw method. The amino acid esters were purchased from Nova
Biochem or Sigma Aldrich and used as received. Tetraethylammonium
tetrafluoroborate (Alfa-Aesar) was recrystallized twice from hot ethanol;
the crystals obtained were dried under vacuum. Ferrocene (Aldrich) was
sublimed under vacuum before use. Dimethylformamide-trifluorometha-
nesulfonic acid, DMFH+, was prepared by the method of Favier and
1
ꢀ50.05 ppm (s); H NMR (CDCl3): d = 7.60 (m, 2H, P-Ph(p)), 7.50, 7.48
(m, 8H, P-PhACTHNUTRGENUG(N o,m)), 7.03 (dd 4H, N-Ph(m)), 6.64 (d, 4H, N-Ph(o)), 5.99
(d, 2H, C(O)NH), 4.53 (m, 2H, aCH), 4.44 (m, 4H, PCH2N), 4.17 (q, 4,
COOCH2CH3), 4.00 (dd, 4H, PCH2N), 2.83 (m, 4H, CH2C(O)N), 2.43
(m, 4H, N-PhCH2), 1.31 (d, 6H, bCH3), 1.23 ppm (t, 6H, CH2CH3); ele-
mental analysis (%) calcd for: C 6.32, H 6.83, N, 7.03; found: C 66.26, H
6.95, N 7.07.
All of the following dipeptide ligands were prepared in an analogous
manner to the PPh2NpNNA-AlaOEt preparation described above, with com-
2
plete details in the Supporting Information: PPh2Nm/pNNA-SerOMe
2
(SerOMe=l-serine methyl ester), PPh2Nm/pNNA-PheOMe (PheOMe=l-phe-
2
nylalanine methyl ester), PPh2Nm/pNNA-TyrOMe (TyrOMe=l-tyrosine methyl
2
DuÇach.[19] [Ni (BF4)2 and PPh2Nm/p--NNA (where p-NNA=3-(4-
ACHTUNGTRENNUNG(MeCN)6]ACHTUNGTRENNUNG
2
ester), PPh2NmNNA-AlaOEt2 (AlaOEt=dl-alanine ethyl ester).
aminophenyl)propionic acid and m-NNA=3-(3-aminophenyl)propionic
Metal complexes:[Ni(PPh2NpNN-NH22)2]
ligand (68.4 mg, 0.115 mmol) was added to an acetonitrile solution con-
taining 0.5 equiv [Ni(MeCN)6](BF4)2 (27 mg, 0.057 mmol) and stirred for
(BF4)2: The purified PPh2NpNN-NH2
2
acid) were prepared following literature methods.[20]
Syntheses
N
ACHTUNGTRENNUNG
24 h. The resulting red solution was filtered through celite and the sol-
vent was removed under vacuum. The residual red oil was then re-dis-
solved in a minimal amount of acetonitrile (~2 mL) and added dropwise
to 08C diethyl ether (~30 mL) and stirred in an ice bath until all of red
solid had precipitated out of the acetonitrile/diethyl ether solution. The
resulting red solid was filtered, washed with diethyl ether and dried in
vacuo (78 mg, 0.055 mmol, 95%). 31P{1H} NMR (CD3CN): d = 5.32 ppm
Ligands: 3-(3-Aminophenyl)propionic acid ethyl ester: This compound
was synthesized following a previously published procedure. Excess tri-
methylsilyl chloride (15.6 mmol) was added dropwise to a stirred solution
of 3-(3-aminophenyl)propionic acid (0.50 g, 3.27 mmol) in ethanol
(10 mL) at 08C. This mixture was then allowed to warm to room temper-
ature and let stir for an additional three hours. Then diethyl ether
(10 mL) was added and washed with a saturated sodium bicarbonate sol-
ution. The organic layer was collected and dried over anhydrous magne-
sium sulfate. The solution was filtered and the solvent was removed
1
(s); H NMR (CD3CN): d = 7.44–7.19 (m, 36H, P-Ph, N-Ph), 6.18 (s, 4H,
C(O)NH2), 5.67 (s, 4H, C(O)NH2), 4.24 (m, 8H, PCH2N), 3.93 (m, 8H,
1938
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
PCH2N), 2.91 (dd, 8H, CH2C(O)N), 2.50 ppm (dd, 8H, N-PhCH2);
MALDI MS: m/z: calcd for {[Ni(PPh2NpNN-NH22)2]}2+: 1251.58; found:
1250.41 [M2+ꢀH+].
where R is the gas constant, T is the temperature in Kelvin, F is the Fara-
day constant, and u is the scan rate (Vsꢀ1).[21] To determine icat/ip, the
cathodic current (ip) was corrected for dilution. The value of icat was
taken where the catalytic current plateaus, where such a plateau was ob-
served, however, all of the para-substituted complexes reported here
show a nearly constant increase in current with no clear plateau (Fig-
ure S5), so icat was measured at both ꢀ0.9 V (icat-A) and ꢀ1.1 V (icat-B) for
all complexes to allow for direct comparison. Each measurement was re-
peated at least three times for statistical purposes. Overpotentials were
All of the metal complexes were prepared in an analogous manner to the
[Ni(PPh2NpNN-NH22)2]
ACHTUNGTRENNUNG
in
the
ACHTUNGTRENNUNG
[Ni(PPh2Nm/pNNA-SerOMe2)2]
U
ACHTUNGTRENNUNG
[Ni(PPh2Nm/pNNA-TyrOMe2)2]
G
ACHTUNGTRENNUNG
2)2]ACHTUNGTRENNUNG ACHTUGNTRENNNUG
(BF4)2; [Ni(PPh2Nm/pNN-NH22)2]
OEt
calculated using the method of Evans et al. [Eq. (2)],[22] with the EHꢃ þ
ꢀ0.14 V and the pKa 6.1.
=
Electrochemistry: All electrochemical studies were done in a N2 glove
ꢀ
box in 0.2m Et4N+BF4 acetonitrile solution at the ambient temperature
of the glovebox, 24–278C, using ferrocene as an internal standard unless
otherwise stated. Cyclic voltammetry experiments were performed on a
CH Instruments 1100A or 600D electrochemical analyzer using a stand-
ard three-electrode configuration. The working electrode was a 1 mm di-
ameter glassy carbon disk coated in PEEK (Cypress Systems EE040),
polished between scans using 0.25 mm MetaDi II diamond polishing paste
(Buehler) on a Buehler MicroCloth lubricated with 18 MW water, and
the counter electrode was a 3 mm glassy carbon rod (Alfa-Aesar). The
reference electrode was a platinum wire (Alfa-Aesar). All couples were
referenced to the ferrocenium/ferrocene couple at 0.0 V. Typical experi-
ments are described.
2:303RT
Eꢃ1=2 ¼ EꢃHþ ꢀ ð
ÞpKa
ð2Þ
F
Digital simulation of cyclic voltammograms: Kinetic parameters for elec-
tron transfer (ks, cmsꢀ1; a) were determined by simulating cyclic voltam-
mograms (CVs) of m-NNA-TyrOMe, p-NNA-TyrOMe, m-NNA-SerOMe,
and p-NNA-SerOMe, in acetonitrile (0.2m Et4N+BF4ꢀ). For each com-
plex, solutions were prepared with [Ni]=0.7 mm and [ferrocene]=
0.3 mm. CVs were collected with scan rates of u=1, 2, 4, 10, 20, and
40 Vsꢀ1, beginning at a potential positive of the NiII/I couple, traversing
the NiII/I and NiI/0 couples, and then traversing the Cp2Fe+/0 couple. CVs
were referenced to this couple, which was also used to estimate the solu-
tion resistance (500 W). CVs were truncated after referencing to show
only the NiII/I and NiI/0 couples.
Scan rate dependence during catalysis: Catalytic waves were obtained for
[Ni(PPh2NpNNA-SerOMe2)2](BF4)2 and [Ni(PPh2NpNNA-PheOMe2)2]
ACHTUNGTRENNUNG ACHUTGNTREN(NUGN BF4)2 at several
acid concentrations at scan rates ranging from 0.1 to 15 Vsꢀ1 to deter-
mine the optimal scan rate at which to collect data. The catalytic activity
was independent of scan rate between 1 to 8 Vsꢀ1, therefore all studies
CVs of the electrolyte solution collected using the same parameters as
for the complexes were used for background subtraction. The back-
ground-subtracted CVs showed unexpected nonzero anodic currents in
the initial portion of the cathodic scans, along with a crossing of the
cathodic and anodic traces near the turnaround voltage; this was attribut-
ed to excess capacitance currents in the background scans. A multiplier
m, where icorrected = iobserved ꢀ m ꢃ ibackground was refined to minimize the
mean squared current (icorrected)2 over the first 50 mV of data for each vol-
were done using a scan rate of 1 Vsꢀ1
.
Catalysis: In a 4 mL glass vial, 2 mL of a 0.70 mm catalyst solution was
ꢀ
made in 0.2m Et4N+BF4 acetonitrile solution. A small amount of ferro-
cene was added as an internal standard and the solution was stirred for
5 min. A voltammogram was measured for the analyte solution at 1 Vsꢀ1
.
The analyte solution was then titrated with 20 mL aliquots of 1.0m
DMFH+ in acetonitrile using a 50 mL syringe. The working electrode was
polished before each addition of acid and a cyclic voltammogram was
measured after each addition of acid. Acid was added to the analyte solu-
tion until no further increase in catalytic current was observed (0.08 to
0.11m DMFH+). Water was then added in 5 mL aliquots using a 10 mL sy-
ringe until the catalytic current reached its maximum (~0.70–0.85m
water).
tammogram. Values determined for
m were 0.702–0.835 (m-NNA-
TyrOMe), 0.670–0.814 (p-NNA-TyrOMe), 0.795–0.946 (m-NNA-
SerOMe), and 0.739–0.882 (p-NNA-SerOMe). The background-subtract-
ed data generally showed some residual capacitance; a value of Cdl =2ꢃ
10ꢀ8 F was determined by manual adjustment.
The background-subtracted, referenced voltammograms were simulated
using DigiElch Version 7.F. Refinement parameters for the NiII/I and NiI/0
redox couples and for the diffusion coefficients of the Ni species (as-
sumed to be equal for all oxidation states of a given complex) were deter-
mined by successive least-squares refinement of calculated versus experi-
mental voltammograms. Initial estimates for the diffusion coefficient D,
The meta- and para-substituted non-natural amino acid nickel complexes,
[Ni(PPh2Nm/pNNA2)2]
ACHTUNGTRENNUNG(BF4)2, had different solubility properties in acetoni-
trile compared to the ester complexes. A typical experiment for catalysis
was slightly different for this compound. In a 2 mL glass vial, the catalyst
(0.0146 g, 0.01 mmol) was dissolved in acetonitrile (1 mL) and syringe fil-
ꢀ
tered into a clean 2 mL glass vial. To this solution was added Et4N+BF4
and for E8
A
ACHTUNGTRENNUNG
(0.15 g, 0.7 mmol), stirred for 5 min and then syringe filtered into a clean
2 mL glass vial. Then 0.5 mL of this solution was diluted to 1 mL and sy-
ringe filtered, this was repeated one more time. The resulting catalyst
concentration was approximately 0.5 to 0.7 mm. A small amount of ferro-
cene was added as an internal standard and the solution was stirred for
4 Vsꢀ1, assuming k
C
CHTUNGTRENNUNG
then afforded estimates of E8, a, and ks for the NiII/I and NiI/0 couples,
and D. For m/p-NNA-TyrOMe, this refinement converged, with s=0.027
and 0.025 for the meta- and para-substituted complexes. Parameter esti-
mates are provided in Table S3.
5 min. A voltammogram was measured for the analyte solution at 1 Vsꢀ1
.
Refinement using all data with E8, a, and ks for the NiII/I and NiI/0 couples
and D allowed to vary freely did not converge for m/p-NNA-SerOMe.
Confidence intervals for a and ks for the NiI/0 couple were much larger
than for the NiII/I couple, and these parameters were taken to be ill-deter-
mined. On the assumption that NiI/0 is Nernstian for these complexes, a-
The analyte solution was then titrated with 10 mL aliquots of 1.0m
DMFH+ in acetonitrile using a 50 mL syringe. The working electrode was
polished before each addition of acid and a cyclic voltammogram was
measured after each addition of acid. Acid was added to the analyte solu-
tion until no further increase in catalytic current was observed (0.15m
DMFH+). Water was then added in 2 mL aliquots using a 10 mL syringe
until the catalytic current reached its maximum (1.3m water).
ACHUTNGRENNUG sCAHTUNGTRENNUGN
(NiI/0) was set to 0.5 and k (NiI/0) was set to 1 cmsꢀ1. Refinement on the
remaining parameters converged, with s=0.027 and 0.031 for the meta-
and para-substituted complexes. Parameter estimates are provided in
Table S4.
Calculating the rate of H2 production: The turnover frequency or TOF
(kobs, sꢀ1), was calculated from the ratio of the catalytic current (icat
)
Computational methods: Classical molecular dynamics (MD) simulations
were carried out using replica-exchange methodology[23] (REMD) to ex-
plore the differences in conformational space of the meta- and para-sub-
stituted alanine containing complexes, m/p-NNA-AlaOEt. The
Amber10[24] MD simulation code was used. The force field parameters
measured in the acid independent region (Figure S1) to the current ob-
tained for a diffusive one-electron reduction with the catalyst in the ab-
sence of acid (ip), using Equtaion (1):
for the core metal complex, [Ni
FF99SB[25] force field was used for the ligands. In the [Ni
moiety, the Ni and P atoms were assigned Blçchl charges,[26] and all other
ACHTUNGTRENNUNG
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
icat
iP
2
RT ꢂ kobs
Fu
AHCTUNGTRENNUNG
ð1Þ
¼
0:4463
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1939

S. Raugei, J. A. S. Roberts, W. J. Shaw et al.
atoms were given RESP charges.[27] Details about the development of the
force field are available in the Supporting Information. The solvent (ace-
tonitrile) was treated implicitly using a modified Generalized Born (GB)
model.[28] The REMD simulation utilized 20 replicas and a temperature
range of 280.0–930.0 K. The simulations ran for a total of 95 ns, generat-
ing ~13000 structures for conformational analysis of the 300 K trajectory.
M. Dudev, C. Lim, J. Am. Chem. Soc. 2003, 125, 3168–3180; d) S.
Karlin, Z. Y. Zhu, K. D. Karlin, Biochemistry 1998, 37, 17726–
17734; e) C. Yang, X. Li, W. Wei, Y. Li, Z. Duan, J. Zheng, T.
Huang, Chem. Eur. J. 2007, 13, 3120–3130; f) R. Wolfenden, M. J.
Snider, Acc. Chem. Res. 2001, 34, 938–945.
[3] I. V. Korendovych, S. J. Shandler, G. L. Montalvo, W. F. DeGrado,
Org Lett. 2011, 13, 3474–3477.
The interaction of water was also studied by modeling the solvent explic-
itly in three different systems: m/p-NN-OEt, m/p-NNA-AlaOEt, and m/
p-NNA-Lys. Simulations included explicit acetonitrile or acetonitrile/
water mixtures representing water concentrations between 0.5 and 3m.
TIP3P water parameters[29] and acetonitrile parameters published by Ni-
kitin and Lyubartsev[30] were used. The modeled systems contained ~900
solvent molecules, with a periodic box of 45 ꢂ ꢃ 45 ꢂ ꢃ 45 ꢂ for the m/
p-NN-OEt and m/p-NNA-AlaOEt solutes, and ~3600 solvent molecules
in a periodic box of 70 ꢂ ꢃ 70 ꢂ ꢃ 70 ꢂ for m/p-NNA-Lys. Statistics
were accumulated from ~30 ns trajectories for the m/p-NN-OEt and m/p-
NNA-AlaOEt, and ~20 ns for m/p-NNA-Lys, obtained at ambient condi-
tions (pressure of 1 atm and temperature of 300 K). To evaluate the con-
centration of water molecules near the catalysts, the number of water
molecules (N) was counted around the solvent accessible surface as a
function of the distance from the surface. The volume (V) around the cat-
alyst was calculated for concentric shells with 1 ꢂ thickness, and the con-
centration of water was calculated for each shell (N/VꢃNA).
[4] a) Y. Lu, N. Yeung, N. Sieracki, N. M. Marshall, Nature 2009, 460,
855–862; b) N. M. Marshall, D. K. Garner, T. D. Wilson, Y.-G. Gao,
H. Robinson, M. J. Nilges, Y. Lu, Nature 2009, 462, 113–116.
[5] a) M. Frey, ChemBioChem 2002, 3, 153–160; b) C. Lꢄger, A. K.
Jones, W. Roseboom, S. P. J. Albracht, F. A. Armstrong, Biochemis-
try 2002, 41, 15736–15746; c) A. D. Wilson, R. H. Newell, M. J.
McNevin, J. T. Muckerman, M. Rakowski Dubois, D. L. Dubois, J.
Am. Chem. Soc. 2006, 128, 358–366.
[6] a) Y. Nicolet, B. J. Lemon, J. C. Fontecilla-Camps, J. W. Peters,
Trends Biochem. Sci. 2000, 25, 138–143; b) Y. Nicolet, C. Piras, P.
Legrand, C. E. Hatchikian, J. C. Fontecilla-Camps, Structure 1999, 7,
13–23; c) J. W. Peters, W. N. Lanzilotta, B. J. Lemon, L. C. Seefeldt,
Science 1998, 282, 1853–1858.
[7] a) M. L. Helm, M. P. Stewart, R. M. Bullock, M. R. Dubois, D. L.
Dubois, Science 2011, 333, 863–866; b) U. J. Kilgore, J. A. S. Rob-
erts, D. H. Pool, A. M. Appel, M. P. Stewart, M. R. DuBois, W. G.
Dougherty, W. S. Kassel, R. M. Bullock, D. L. Dubois, J. Am. Chem.
Soc. 2011, 133, 5861–5872; c) U. J. Kilgore, M. P. Stewart, M. L.
Helm, W. G. Dougherty, W. S. Kassel, M. R. DuBois, D. L. Dubois,
R. M. Bullock, Inorg. Chem. 2011, 50, 10908–10918.
[8] A. Volbeda, L. Martin, C. Cavazza, M. Matho, B. W. Faber, W.
Roseboom, S. P. Albracht, E. Garcin, M. Rousset, J. C. Fontecilla-
Camps, J. Biol. Inorg. Chem. 2005, 10, 239–249.
[9] T. E. Creighton, Proteins: Structures and Molecular Properties, 2nd
ed., W. H. Freeman and Company, New York, 1993.
[10] a) A. Jain, S. Lense, J. C. Linehan, S. Raugei, H. Cho, D. L. Dubois,
W. J. Shaw, Inorg. Chem. 2011, 50, 4073–4085; b) A. Jain, M. L.
Reback, M. L. Lindstrom, C. E. Thogerson, M. L. Helm, A. M.
Appel, W. J. Shaw, Inorg. Chem. 2012, 51, 6592–6602.
[11] a) A. El-Faham, F. Albericio, Chem. Rev. 2011, 111, 6557–6602;
b) V. Dourtoglou, J.-C. Ziegler, B. Gross, Tetrahedron Lett. 1978, 19,
1269–1272; c) R. Knorr, A. Trzeciak, W. Bannwarth, D. Gillessen,
Tetrahedron Lett. 1989, 30, 1927–1930.
The free energy cost was calculated for the structural reorganization of
the complexes to allow for direct delivery of protons from the dipeptide
to the pendant amine in the catalysts. The simulations used the implicit
solvation model and umbrella sampling MD methodology, and the free
energy profiles were reconstructed using the WHAM (Weighted Histo-
gram Analysis Method) approach.[31] The umbrella sampling was done in
36 windows, for the m/p-NNA-Lys system, and 26 windows for m/p-
NNA-AlaOEt, varying the reaction coordinate by 0.5 ꢂ, from 3.0 to
15.5 ꢂ for m/p-NNA-AlaOEt, and from 3 to 20.5 ꢂ for NNA-Lys. Each
trajectory was equilibrated for 500 ps and statistics were collected from
1.5 ns long trajectories. A constraining harmonic potential with a force
constant of 20 kcalmolꢀ1 was used. Prior to the umbrella sampling, the
peptides involved in the dynamics were relaxed using simulated anneal-
ing, by gradually heating the system to 800 K then cooling it down and
equilibrating at 300 K.
[12] I. M. Kolthoff, M. K. Chantooni, S. Bhowmik, Anal. Chem. (Wash.)
1967, 39, 1627.
[13] A. J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals
and Applications, 2nd ed., Wiley, New York, 2001.
Acknowledgements
[14] R. Nicholoson, I. Shain, Anal. Chem. 1964, 36, 706.
The authors would like to thank Dr. Daniel L. Dubois for helpful discus-
sions. This work was funded by the US DOE Basic Energy Sciences,
Physical Bioscience program (M.L.R., T.C.S., W.J.S.), the US DOE Basic
Energy Sciences, Chemical Sciences, Geoscience and Biosciences Divi-
sion (A.J., W.J.S.), the Office of Science Early Career Research Program
through the Office of Basic Energy Sciences (W.J.S., B.G.P., M.H., S.R.),
and the Center for Molecular Electrocatalysis, an Energy Frontier Re-
search Center funded by the US Department of Energy, Office of Sci-
ence, Office of Basic Energy Sciences (J.A.S.R., S.R.). Part of the re-
search was conducted at the W. R. Wiley Environmental Molecular Sci-
ences Laboratory, a national scientific user facility sponsored by US De-
partment of Energyꢀs Office of Biological and Environmental Research
(BER) program located at Pacific Northwest National Laboratory
(PNNL). Pacific Northwest National Laboratory is operated by Battelle
for the US Department of Energy.
[15] a) The radius of gyration is a measure of the compactness of a mole-
n
P
2
1
n
cule, calculated as: R2g ¼
ðri ꢀ RcÞ , where Rc is the center of ge-
~
~
~
i¼1
~
ometry of the molecule, ri is the position of atom i, and n is the total
number of atoms in the molecule; b) A. Ahmed, F. Rippmann, G.
Barnickel, H. A. Gohlke, J. Chem. Inf. Model. 2011, 51, 1604–1622.
[16] a) J. Lee, N. M. Goodey, Chem. Rev. 2011, 111, 7595–7624; b) S. W.
Ragsdale, Chem. Rev. 2006, 106, 3317–3337.
[17] K. Izutsu, Acid–Base Dissociation Constants in Dipolar Aprotic Sol-
vents, Blackwell Scientific Publications, Oxford, 1990.
[18] E. S. Wiedner, J. Y. Yang, S. Chen, S. Raugei, W. G. Dougherty,
W. S. Kassel, M. L. Helm, R. M. Bullock, M. Rakowski DuBois,
D. L. DuBois, Organometallics 2012, 31, 144–156.
[19] I. Favier, E. Dunach, Tetrahedron Lett. 2004, 45, 3393–3395.
[20] A. Le Goff, V. Artero, B. Jousselme, P. D. Tran, N. Guillet, R.
Metaye, A. Fihri, S. Palacin, M. Fontecave, Science 2009, 326, 1384–
1387.
[21] a) J. M. Savꢄant, E. Vianello, Electrochim. Acta 1965, 10, 905–920;
b) J. M. Savꢄant, E. Vianello, Electrochim. Acta 1967, 12, 629–646;
c) D. H. Pool, D. L. DuBois, J. Organomet. Chem. 2009, 694, 2858–
2865.
[22] G. A. N. Felton, R. S. Glass, D. L. Lichtenberger, D. H. Evans, Inorg.
Chem. 2006, 45, 9181–9184.
[1] S. Karlin, Z. Zhu, K. Karlin, Proc Natl. Acad. Sci. USA 1997, 94,
14225–14230.
[2] a) L. Bobadilla, F. Nino, E. Cepeda, M. A. Patarroyo, Characterizing
and predicting catalytic residues in enzyme active sites based on local
properties: A machine learning approach 2007; b) H. R. Brodkin,
W. R. Novak, A. C. Milne, J. A. D’Aquino, N. M. Karabacak, I. G.
Goldberg, J. N. Agar, M. S. Payne, G. A. Petsko, M. J. Ondrechen,
D. Ringe, Biochemistry 2011, 50, 4923–4935; c) T. Dudev, Y. L. Lin,
[23] a) A. Mitsutake, Y. Sugita, Y. Okamoto, Biopolymers 2001, 60, 96–
123; b) H. Nymeyer, S. Gnanakaran, A. E. Garcꢅa, Meth. Enzymol.
1940
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 1928 – 1941

H2 Production
FULL PAPER
2004, 383, 119–149; c) X. Cheng, G. Cui, V. Hornak, C. Simmerling,
J. Phys. Chem. B 2005, 109, 8220–8230.
[28] A. Onufriev, D. Bashford, D. A. Case, Proteins Struct. Funct. Bioinf.
2004, 55, 383–394.
[24] Amber10, T. A. D. D. A. Case, T. E. Cheatham, III, C. L. Simmerl-
ing, J. Wang, R. E. Duke, R. Luo, M. Crowley, R. C. Walker, W.
Zhang, K. M. Merz, B. Wang, S. Hayik, A. Roitberg, G. Seabra, I.
Kolossvꢆry, K. F. Wong, F. Paesani, J. Vanicek, X. Wu, S. R. Brozell,
T. Steinbrecher, H. Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak,
G. Cui, D. H. Mathews, M. G. Seetin, C. Sagui, V. Babin, P. A. Koll-
man, University of California, San Francisco, 2008.
[29] W. L. Jorgensen, J. Chandrasekhar, J. Madura, M. L. Klein, J. Chem.
Phys. 1983, 79, 926–935.
[30] A. M. Nikitin, A. P. Lyubartsev, J. Comput. Chem. 2007, 28, 2020–
2026.
[31] S. Kumar, D. Bouzida, R. H. Swendsen, P. A. Kollman, J. M. Rosen-
berg, J. Comput. Chem. 1992, 13, 1011–1021.
[25] P. C. W. D. Cornell, C. I. Bayly, I. R. Gould, K. M. Merz, Jr., D. M.
Ferguson, D. C. Spellmeyer, T. Fox, J. W. Caldwell, P. A. Kollman. ,
J. Am. Chem. Soc. 1995, 117, 5179–5197.
Received: August 8, 2012
Published online: December 11, 2012
[26] P. E. Blçchl, J. Chem. Phys. 1995, 103, 7422–7428.
[27] C. I. Bayly, P. Cieplak, W. Cornell, P. A. Kollman, J. Phys. Chem. J.
Phys. Chem. B 1993, 97, 10269–10280.
Please note: Minor changes have been made to this manuscript since its
publication in Chemistry—A European Journal Early View. The Editor.
Chem. Eur. J. 2013, 19, 1928 – 1941
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1941