Enhanced Stability of the FeII/MnII State in a Synthetic Model of Heterobimetallic Cofactor Assembly

Article abstract of DOI:10.1021/acs.inorgchem.5b02322

Heterobimetallic Mn/Fe cofactors are found in the R2 subunit of class Ic ribonucleotide reductases (R2c) and R2-like ligand binding oxidases (R2lox). Selective cofactor assembly is due at least in part to the thermodynamics of M^II binding to the apoprotein. We report here equilibrium studies of Fe^II/Mn^II discrimination in the biomimetic model system H₅(F-HXTA) (5-fluoro-2-hydroxy-1,3-xylene-α,α′-diamine-N,N,N′,N′-tetraacetic acid). The homobimetallic F-HXTA complexes [Fe(H₂O)₆][1]₂·14H₂O and [Mn(H₂O)₆][2]₂·14H₂O (1 = [Fe^II₂(F-HXTA)(H₂O)₄]^-; 2 = [Mn^II₂(F-HXTA)(H₂O)₄]^-) were characterized by single crystal X-ray diffraction. NMR data show that 1 retains its structure in solution (2 is NMR silent). Metal exchange is facile, and the heterobimetallic complex [Fe^IIMn^II(F-HXTA)(H₂O)₄]^- (3) is formed from mixtures of 1 and 2. ¹⁹F NMR was used to quantify 1 and 3 in the presence of excess M^II(aq) at various metal ratios, and equilibrium constants for Fe^II/Mn^II discrimination were calculated from these data. Fe^II is preferred over Mn^II with K₁ = 182 ± 13 for complete replacement (2 21). This relatively modest preference is attributed to a hard-soft acid-base mismatch between the divalent cations and the polycarboxylate ligand. The stepwise constants for replacement are K₂ = 20.1 ± 1.3 (2 23) and K₃ = 9.1 ± 1.1 (3 21). K₂ > K₃ demonstrates enhanced stability of the heterobimetallic state beyond what is expected for simple Mn^II Fe^II replacement. The relevance to Fe^II/Mn^II discrimination in R2c and R2lox proteins is discussed.

Full text of DOI:10.1021/acs.inorgchem.5b02322

Article
pubs.acs.org/IC
Enhanced Stability of the FeII/MnII State in a Synthetic Model of
Heterobimetallic Cofactor Assembly
William D. Kerber,*,† Joshua T. Goheen,† Kaitlyn A. Perez,† and Maxime A. Siegler‡
†Department of Chemistry, Bucknell University, Lewisburg, Pennsylvania 17837, United States
‡Department of Chemistry, Johns Hopkins University, Baltimore, Maryland 21218, United States
S
* Supporting Information
ABSTRACT: Heterobimetallic Mn/Fe cofactors are found in the R2 subunit of
class Ic ribonucleotide reductases (R2c) and R2-like ligand binding oxidases
(R2lox). Selective cofactor assembly is due at least in part to the
thermodynamics of MII binding to the apoprotein. We report here equilibrium
studies of FeII/MnII discrimination in the biomimetic model system H5(F-
HXTA) (5-fluoro-2-hydroxy-1,3-xylene-α,α′-diamine-N,N,N′,N′-tetraacetic
acid). The homobimetallic F-HXTA complexes [Fe(H2O)6][1]2·14H2O and
[Mn(H2O)6][2]2·14H2O (1 = [FeII (F-HXTA)(H2O)4]−; 2 = [MnII (F-
2
2
HXTA)(H2O)4]−) were characterized by single crystal X-ray diffraction. NMR
data show that 1 retains its structure in solution (2 is NMR silent). Metal
exchange is facile, and the heterobimetallic complex [FeIIMnII(F-HXTA)(H2O)4]− (3) is formed from mixtures of 1 and 2. 19
F
NMR was used to quantify 1 and 3 in the presence of excess MII(aq) at various metal ratios, and equilibrium constants for FeII/
MnII discrimination were calculated from these data. FeII is preferred over MnII with K1 = 182 13 for complete replacement (2
⇌ 1). This relatively modest preference is attributed to a hard−soft acid−base mismatch between the divalent cations and the
polycarboxylate ligand. The stepwise constants for replacement are K2 = 20.1 1.3 (2 ⇌ 3) and K3 = 9.1 1.1 (3 ⇌ 1). K2 > K3
demonstrates enhanced stability of the heterobimetallic state beyond what is expected for simple MnII → FeII replacement. The
relevance to FeII/MnII discrimination in R2c and R2lox proteins is discussed.
originally identified in the pathogen Chlamydia trachomatis
based on the replacement of a tyrosine residue, normally
oxidized to a tyrosyl radical, with a redox inactive phenyl-
alanine.7 The Mn/Fe cofactor was eventually identified, and the
extra oxidizing equivalent from the missing organic radical was
found to be stored on Mn.8 A family of ligand-binding oxidases
homologous with R2c (R2lox) was subsequently discovered.9
R2lox has been structurally characterized in Mycobacterium
tuberculosis and Geobacillus kaustophilus and is thought to be
involved in fatty acid biosynthesis. Although the protein
structure is modified to create a larger substrate-binding
pocket, the heterobimetallic Mn/Fe cofactor in R2lox proteins
is conserved.
In vitro metalation of both Ct R2c and Gk R2lox has been
investigated, and the thermodynamics of metal binding play a
role in each case.9c,10 Both apoenzymes initially load divalent
metal ions with at least partial discrimination for the correct
{MnII(1)FeII(2)} cofactor (Figure 1). A cooperative binding
scheme has been proposed for Gk R2lox: site 1 is disordered in
the apoprotein and is not loaded until FeII preferentially binds
to site 2. In contrast, metal binding in Ct R2c does not appear
to be cooperative. Both Ct R2c sites are ordered in the
apoprotein and are sterically isolated by a phenylalanine
residue. In the presence of both metals Ct R2c selectively
INTRODUCTION
■
Protein metallocofactors play an important and diverse role in
nature and range in complexity from single metal ions to large
multimetallic clusters.1 Multinuclear metallocofactors are
intriguing because metal ions in close proximity may work
cooperatively together to accomplish otherwise difficult
chemical transformations.2 These cofactors may be assembled
from other proteins (metallochaperones)3 or from small-
molecule complexes found in labile cellular metal pools.4
Frequently the mechanism of metallocofactor assembly is
uncertain. Self-assembly of metallocofactors in vitro from
apoenzymes and solvated metal ions is often possible, although
care must be taken to avoid mismetalation.2c,5 It is of course
still possible for metallochaperones to be involved in vivo, but it
seems plausible in these cases that the thermodynamics of
metal−ligand interactions in the protein active site play an
important role in cofactor assembly.
One such example is class I ribonucleotide reductases
(RNRs), which catalyze the conversion of RNA to DNA.2a
The R2 subunit contains a dinuclear metallocofactor, bound by
carboxylate and histidine residues at the center of a four helix
bundle, which uses O2 to generate a radical that is transferred to
the active site of the R1 subunit. Class I RNRs are divided into
three subclasses based on the identity of the R2 subunit and its
bimetallic cofactor:5 class Ia utilizes iron, class Ib utilizes
manganese, and the more recently discovered class Ic contains a
heterobimetallic Fe/Mn cofactor.6 Class Ic RNR (R2c) was
Received: October 7, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Chart 1. Dinucleating Ligand Used to Model Metal Binding
to R2c/R2lox Proteins
19F NMR was developed to measure equilibrium constants for
FeII/MnII discrimination. Our model shows only a modest
preference for FeII, which means MnII can compete effectively
for the ligand binding sites at similar concentrations. We also
observed a preference for the heterobimetallic FeII/MnII
complex that is greater than predicted by simple Mn → Fe
replacement. Both of these observations support the proposed
mechanisms of cofactor assembly in R2c and R2lox proteins.
Cytoplasmic concentrations of FeII and MnII are similar, and
therefore weak discrimination is necessary to load both metals
simultaneously.4c An intrinsic preference for the heterobime-
tallic state is also beneficial because selective loading of FeII into
site 2 will then help direct MnII to site 1.
Figure 1. Schematic drawing of the reduced {MnII(1)FeII(2)} active sites
in C. trachomatis R2c protein and G. kaustophilus R2lox homologue 1
protein.
RESULTS
■
assembles {MnII(1)FeII(2)} while Gk R2lox loads a mixture of
{MnII(1)FeII(2)} and {FeII(1)FeII(2)}. The metal ions are substitu-
tionally labile in the reduced state and become fixed in place
after O2 activation.11 Kinetic selectivity plays a role in Gk R2lox
loading as faster oxidation of {MnII(1)FeII(2)} drives the system
toward the correct heterobimetallic cofactor.
The dinucleating ligand 5-fluoro-2-hydroxy-1,3-xylene-α,α′-
diamine-N,N,N′,N′-tetraacetic acid, H5(F-HXTA), was synthe-
sized by Mannich alkylation of 4-fluorophenol with iminodi-
acetic acid. Homobimetallic F-HXTA complexes were prepared
by addition of MCl2 (3 equiv; M = Fe, Mn) and NaOH (5
equiv) to H5(F-HXTA). The resultant colorless solutions were
slightly basic (pH 7−8) and air sensitive. Exposure of both FeII
and MnII F-HXTA complexes to O2 produced darkly colored,
uncharacterized products. In the case of iron, the UV−vis
spectrum of the oxidized product displayed a strong absorption
at 360 nm and a weak shoulder at 500 nm, which is analogous
to the spectrum reported for diferric complexes of Me-
HXTA.17 Under anaerobic conditions, the complex anion
Although a definitive explanation for MnII/FeII discrim-
ination in Ct R2c and Gk R2lox remains elusive, it is clear that
both proteins are tuned to bias metal binding equilibria toward
the correct {MnII(1)FeII(2)} cofactor. This is a particularly
impressive achievement given the similarity of the residues in
the binding sites and the properties of the metal ions.12 FeII and
MnII both tend to form high-spin octahedral complexes that are
substitutionally labile, but FeII complexes have greater ligand
field stabilization energy (LFSE) because of an additional
valence electron (d6 vs d5). FeII also has a slightly smaller ionic
radius. The net result is that FeII usually forms stronger bonds
than MnII, as observed in the classic Irving−Williams series.
Both metals are borderline-hard Lewis acids with a preference
for Lewis bases such as imidazole and pyridine. In mononuclear
complexes, the thermodynamic preference for FeII over MnII is
well established,13 and the same general trend is expected to be
true for dinuclear complexes, although fewer examples are
known.14 What is not clear is the stability of the FeII/MnII state
relative to the homobimetallic species. The few known
examples of synthetic FeII/MnII complexes have been prepared
by exploiting nonequilibrium conditions to avoid formation of
undesirable homobimetallic species.15 Given the importance of
metal exchange equilibria in R2c and R2lox proteins, we sought
to develop a synthetic model system from which FeII/MnII
binding could be quantified. The F-HXTA platform (Chart 1)
was chosen because, like the binding sites in R2c and R2lox, it
is dominated by hard carboxylate donors and because dinuclear
HXTA complexes with other MII cations are known.16 The
symmetry of F-HXTA makes the two binding sites identical,
which allows the heterobimetallic complex to be investigated
without the added complexity of coordination isomers. A
method for quantifying paramagnetic F-HXTA complexes by
[MII (F-HXTA)(H2O)4]− (1, M = Fe; 2, M = Mn) crystallized
2
with [M(H2O)6]+2 to yield colorless single crystals suitable for
X-ray structure determination. Displacement ellipsoid plots of 1
and 2 are shown in Figure 2. Crystallographic data are given in
Table 1 along with selected bond lengths and angles in Table 2.
Crystals of [Fe(H2O)6][1]2·14H2O and [Mn(H2O)6][2]2·
14H2O were isolated in an anaerobic chamber and dried over
P2O5 for 1 week. Elemental analyses (C,H,N) on replicate
samples were consistent with the loss of 22 water molecules in
both cases. The isolated products are therefore formulated as
[M(H2O)6][MII (F-HXTA)]2 (1′, M = Fe; 2′, M = Mn) with
2
the six remaining waters shown in the complex cation, although
other coordination isomers are of course possible. Dissolution
1
of 1′ in D2O gave a colorless solution, and the H NMR
spectrum shown in Figure 3a was recorded. Resonances are
broad and dispersed over 170 ppm because of the paramagnetic
FeII atoms.18 Seven peaks of equal area are consistent with the
C2 symmetry of 1 observed in the solid state. Paramagnetic line
broadening obscures geminal CH2 coupling and three-bond
H−F coupling. Longitudinal relaxation time constants (T1) are
given in Table 3, along with values calculated from the crystal
structure of 1.19 The 19F NMR spectrum of 1′ shows a single
broad resonance at −74.3 ppm (Figure S4 in the Supporting
Information). Identical 1H and 19F spectra were obtained when
1 was prepared in situ from a mixture of H5(F-HXTA), FeCl2,
B
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 2. Displacement ellipsoid plots of [MII (F-HXTA)(H2O)4]− anions (1, M = Fe; 2, M = Mn). F-HXTA hydrogen atoms are shown to aid
2
assignment of NMR spectra. The same atom-labeling scheme is used consistently for both complexes.
Table 1. Crystallographic Data for [Fe(H2O)6][1]2·14H2O
and [Mn(H2O)6][2]2·14H2O
Table 2. Selected Bond Lengths and Angles for [MII (F-
2
HXTA)(H2O)4]− (1, M = Fe; 2, M = Mn)
[Fe(H2O)6][1]2·
[Mn(H2O)6][2]2·
1
2
14H2O
14H2O
Bond Lengths (Å)
2.2016(19)
2.2113(19)
2.1025(16)
2.0943(16)
2.1039(17)
2.1127(16)
2.1599(16)
2.1260(16)
2.0880(17)
2.0742(17)
2.1861(17)
2.1829(17)
Bond Angles (deg)
129.58(7)
91.04(7)
Crystal Data
C32H84F2Fe5N4O46
1578.28
M1−N1
M2−N2
M1−O5
M2−O5
M1−O3
M2−O6
M1−O1
M2−O8
M1−O1W
M2−O4W
M1−O2W
M2−O3W
2.2783(17)
2.2935(17)
2.1415(14)
2.1332(14)
2.1535(14)
2.1500(15)
2.2414(14)
2.1940(14)
2.1345(16)
2.1252(16)
2.2083(16)
2.2111(15)
empirical formula
formula wt
cryst syst
space group
a [Å]
C32H84F2Mn5N4O46
1573.73
triclinic
triclinic
P1
̅
P1
̅
10.8238(3)
11.3967(3)
13.4092(4)
92.709(2)
113.420(3)
90.301(2)
1515.61(8)
1
10.8657(4)
11.4436(4)
13.5006(4)
92.535(3)
113.243(3)
90.364(3)
1540.43(10)
1
b [Å]
c [Å]
α (deg)
β (deg)
γ (deg)
V (Å3)
Z
T (K)
110(2)
110(2)
M1−O5−M2
N1−M1−O5
N2−M2−O5
N1−M1−O2W
N2−M2−O3W
O5−M1−O1W
O5−M2−O4W
O1W−M1−O2W
O3W−M2−O4W
O1−M1−O3
127.58(6)
88.59(6)
88.93(6)
90.97(6)
90.14(6)
88.34(6)
90.95(6)
91.84(6)
90.16(6)
148.97(5)
149.39(5)
cryst size (mm3)
0.26 × 0.20 × 0.16
1.290
0.16 × 0.11 × 0.09
1.116
μ (mm−1
)
91.30(7)
Data Collection
0.794, 0.860
27964
90.29(7)
Tmin, Tmax
0.883, 0.927
23903
7077
89.90(7)
measd reflns
unique reflns
obsd reflns [I > 2σ(I)]
Rint
87.33(7)
6967
89.37(7)
6627
6433
91.37(7)
0.034
0.031
89.68(7)
(sin θ/λ)max (Å−1
)
0.650
0.650
152.63(6)
153.09(6)
Refinement
0.031
O6−M2−O8
R[F2 > 2σ(F2)]
wR(F2)
0.030
0.072
0.069
broadening so extreme that 2 is effectively NMR silent.18
Solution magnetic susceptibility measurements were made on
isolated samples of 1′ and 2′ dissolved in D2O using the Evans
method.20 The effective magnetic moments (μeff) were found
to be 12.4μB for 1′ and 13.5μB for 2′. These measurements
agree with the theoretical values of μeff calculated for complexes
containing five noninteracting high-spin MII ions (12.05μB for
FeII and 13.23μB for MnII).21
A metal ion exchange experiment was performed by
dissolving approximately equimolar amounts of 1′ and 2′ in
D2O. The heterobimetallic complex [FeMn(F-HXTA)-
(D2O)4]− (3) was expected to form if metal exchange is facile.
The 1H NMR spectrum of this mixture is shown in Figure 3b; 1
is clearly still present, and a new set of resonances has emerged
that is consistent with 3. NMR can usually resolve MnII/FeII
S
1.05
1.11
params (restraints)
Δρmax, Δρmin (e Å−3
488 (42)
0.93, −0.62
488 (42)
0.77, −0.33
)
and NaOD. A titration of F-HXTA with FeCl2 monitored by
19F NMR is shown in Figure S5. At substoichiometric FeII
concentrations, a second resonance was observed at −84.7
ppm. This species, which decreased with increasing iron
concentration, is attributed to a mononuclear FeII(F-HXTA)
complex. When 2 equiv of FeII was added, only the dinuclear
1
complex 1 was observed. In contrast to 1′, the H spectrum of
2′ dissolved in D2O showed only a broad solvent resonance,
and the 19F spectrum was completely featureless. The slower
electronic relaxation rate of MnII compared with FeII causes line
C
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
1
Figure 3. H NMR spectra of isolated [M(H2O)6][M2(F-HXTA)]2 complexes dissolved in D2O. (a) [Fe2(F-HXTA)(D2O)4]− from 1′; (b) a
mixture of 1′ and 2′ gives [Fe2(F-HXTA)(D2O)4]− and [FeMn(F-HXTA)(D2O)4]− after metal ion exchange; [Mn2(F-HXTA)(D2O)4]− is present
but not observed. Presaturation was used to suppress residual H2O in both spectra. Inset: one of the aryl protons of 3 overlaps with 1.
1
K2=β /β2
2 + FeII(aq) Hooooo3ooooI 3 + MnII(aq)
Table 3. H-NMR Data for Isolated [Fe(H2O)6][Fe2(F-
(5)
HXTA)]2 (1′) Dissolved in D2O
K3=β1/β3
c
d
e
3 + FeII(aq) HooooooooI 1 + MnII(aq)
δ
ν1/2
T1, obsd
(ms)
T1, calcd
(ms)
a
b
(6)
protons
position
(ppm)
(Hz)
H5B, H12B
H2B, H15B
H5A, H12A
H3A, H13A
H3B, H13B
H7, H9
eq
eq
ax
eq
ax
165.9
88.0
78.9
58.8
25.4
17.6
−5.2
187
244
443
197
294
64
2.1
2.3
0.63
2.4
1.0
9.4
1.1
2.4
2.2
0.59
2.1
1.3
We have evaluated K1−3 in a series of metal competition
experiments by preparing equilibrium mixtures of F-HXTA
with excess FeCl2/MnCl2 at pH 7.29(3). The concentrations of
1 and 3 were measured directly by 19F NMR and the
concentrations of 2, FeII(aq), and MnII(aq) were calculated
from mass balance equations. Although K1−3 may be
determined at a single Fe/Mn ratio, we repeated the
experiment at multiple ratios to test our proposed speciation
model. The results were averaged to give K1 = 182 13, K2 =
20.1 1.3, and K3 = 9.1 1.1. Data are shown in Figure 4; the
H2A, H15A
ax
336
0.95
a
b
Atom labeling scheme from Figure 2. Axial-like or equatorial-like
c
d
disposition in the chelate ring. Line width at half height. Measured
by inversion−recovery. Calculated relative to T1(H7/9) using H/Fe
distances from the crystal structure of 1; see ref 19 and Supporting
e
Information for a discussion.
complexes because magnetic coupling to FeII increases the
electronic relaxation rate of MnII.15c,18 Complex 3 has C1
1
symmetry and therefore twice as many H resonances as 1. A
19F NMR spectrum of the 1′/2′ mixture was recorded, and a
new resonance attributed to 3 is seen at −65.5 ppm. (Figure S6
in the Supporting Information). 19F NMR was also used to
investigate the kinetics of metal ion exchange: The amount of 3
increased for about 2 h after mixing, and then the peak area
ratios remained constant. Identical equilibrium mixtures could
be prepared in situ from H5(F-HXTA), FeCl2, MnCl2, and
NaOD.
The speciation of F-HXTA in the presence of both FeII and
MnII is governed by the equilibria shown in eqs 1−3. Although
we have not measured the individual stability constants β1−3
,
□
□
○
Figure 4. Concentrations of 1 ( ), 2 ( ), and 3 ( ) as a function of
FeII/MnII ratio. [F-HXTA] = 9.03(5) mM; [MCl2]total = 40.16(9)
mM; pH = 7.29(3). Curves are calculated from K1−3 (this work) and
βMCl+ (ref 22).
they must be relatively large because there is no evidence for
free ligand or mononuclear MII(F-HXTA) complexes at
stoichiometric metal loadings. Equations 4−6 describe F-
HXTA speciation in the presence of excess MII, with K1−3
defining the relative stabilities of the three bimetallic complexes.
open symbols are concentrations measured by experiment, and
the curves are concentrations calculated from K1−3 and the
formation constants for [FeCl]+(aq) and [MnCl]+(aq).22 The
complete data set, including individual NMR spectra, is given in
the Supporting Information.
Conditions for metal competition experiments were chosen
such that all species were present above 1 mM. All samples
were equilibrated at 25 °C for 24 h prior to analysis, and pH
was maintained at 7.29(3) with buffered N-methylmorpholine
(NMM). NMM was chosen because its steric bulk inhibits
β1
[F‐HXTA]−5 + 2FeII(aq) ⇌ 1
(1)
β2
[F‐HXTA]−5 + 2MnII(aq) ⇌ 2
(2)
β3
[F‐HXTA]−5 + FeII(aq) + MnII(aq) ⇌ 3
(3)
K1=β1/β2
2 + 2FeII(aq) HooooooooI 1 + 2MnII(aq)
(4)
D
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
metal ion coordination, which was confirmed when identical
NMR spectra were obtained from unbuffered solutions
manually adjusted to pH 7.3 with NaOH. Identical spectra
were also obtained when FeCl2 was replaced with Fe-
(NH4)2(SO4)2. Because pH and MCl2 concentration were
held constant, all measurements were made at the same ionic
strength (I = 0.111 M) and activity coefficients were neglected.
hyperfine shift (the paramagnetic contribution to the chemical
shift) and nuclear relaxation time constants (T1 and T2) are
affected by dipolar contributions from unpaired metal electrons,
and by contact contributions from spin density delocalized onto
the resonanting nuclei. Dipolar coupling of protons to a
magnetically anisotropic metal like high-spin FeII is orientation
dependent, but the interaction is averaged by molecular
motion. The averaged dipolar shift is therefor isotropic and is
more properly called a pseudocontact shift. Because dipolar
coupling is distance dependent, hyperfine shifts, line widths,
and relaxation times are all a rough measure of proton−metal
distance. The aromatic protons in 1 (H7/9) are the furthest
from FeII and are easily assigned to the resonance at 17.6 ppm
by their narrower line width, longer T1, and smaller hyperfine
shift. The CH2 groups in 1 are found in chelate rings with
protons in equatorial-like and axial-like positions. The
equatorial protons are further from the metal and therefore
have narrower line widths and longer T1 times than the axial
protons.19 This is evidenced by the three sharp and three broad
CH2 resonances in Figure 3a and three long (≥2.1 ms) and
three short (≤1.1 ms) T1 times in Table 3. Proton−metal
distances from the crystal structure of 1 were also used to
calculate T1 times, and these values are in good agreement with
the experimentally determined time constants (see Supporting
Information for a discussion of this procedure). The CH2
hyperfine shifts are more difficult to interpret because of the
dependence of the contact contribution on the Fe−N−C−H
dihedral angle, but a given equatorial proton is generally shifted
further downfield than its axial partner.18 Taken together the
chemical shifts, line widths, and T1 times allow the tentative
assignments of CH2 protons shown in Table 3.
4-Fluorophenol (4-FP) was used as an internal standard for 19
F
NMR and its presence did not affect the peak area ratio of 1/3.
Although 4-FP did exhibit paramagnetic line broadening and
faster relaxation (T1 = 60−80 ms depending on Mn/Fe ratio),
its chemical shift was unaffected. These data indicate the 4-FP
does not form metal complexes to a significant degree at the
low concentrations employed in this work.
DISCUSSION
■
F-HXTA Complexes in the Solid State. Single crystals of
[Fe(H2O)6][1]2·14H2O and [Mn(H2O)6][2]2·14H2O suitable
for X-ray structure determination were isolated, and the two
complexes share nearly identical crystal lattices. The small
differences observed are attributable to slightly but statistically
longer metal−ligand bonds in 2 versus 1 (ΔM−L = 0.022−0.082
Å). The largest difference in bond lengths is observed with the
amine (ΔM1−N1 = 0.079 Å; ΔM2−N2 = 0.082 Å) and the
carboxylates lying in the plane of the phenolate (ΔM1−O1
=
0.082 Å; ΔM2−O8 = 0.068 Å). Complexes 1 and 2 have
approximate C2 symmetry, with the aromatic ring canted from
the M1−O5−M2 plane by 52.2° in 1 and 53.6° in 2. Each
metal atom is found in a distorted octahedral geometry. The
angles in those octahedra range from 76.5° to 108.1° for 1 and
from 74.6° to 105.4° for 2. The metal atoms are displaced from
mean square planes (defined by N1, O1W, O2W, and O5 and
N2, O3W, O4W, and O5) by ca. 0.05 and 0.03 Å in 1 and ca.
0.08 and 0.01 Å in 2. The carboxylate ligands are trans across
the square planes, with substantial deviations from linearity
observed in the O1−M1−O3 and O6−M2−O8 angles;
152.63(6)° and 153.09(6)° for 1 and 148.97(5)° and
149.39(5)° for 2.
Dinuclear FeII2 and MnII2 complexes in hard ligand fields are
generally high-spin and exhibit weak magnetic coupling.9c,23
Magnetic susceptibility measurements on solutions of 1′ and 2′
are consistent with this trend, and the complexes are described
as ensembles of noninteracting high-spin MII ions. The
formation of 3 from a mixture 1′ and 2′ demonstrates that
MII(aq) exchange with [M2(F-HXTA)(H2O)4]− is facile. The
19F NMR spectrum of the mixture shows a single resonance
each for 1 and 3 (Figure S6). The 1H spectrum is more difficult
to interpret because 3 has 14 unique F-HXTA protons (Figure
3b). The aryl protons in 3 are relatively sharp and easy to
identify, but unlike 1, separate resonances are expected for H7
and H9. One of these protons is clearly seen at 22.4 ppm, and
examination of the inset in Figure 3b shows that the other
overlaps with the H7/9 signal from 1. The 12 broad CH2
resonances are present but not readily assigned. Similar spectra
have been reported for MnII(μ-OAr)FeII clusters supported by
other dinucleating ligands,15 and the isoelectronic FeII/FeIII
complex [Fe2(Me-HXTA)(μ-OAc)2]−2.24
Drying the isolated crystals caused loss of 22 of 28 water
molecules in both cases. The formulation of these products as
[M(H2O)6][M2(F-HXTA)]2 (1′, M = Fe; 2′, M = Mn) is
convenient but probably not entirely accurate. Other
coordination isomers are possible, and it also seems likely in
the solid state that at least some of the sites vacated by water
are reoccupied by bridging interactions with ligands from
adjacent complex ions. Regardless, the observation that NMR
spectra of 1′ are identical to samples prepared in situ indicates
that hydration of the isolated material regenerates [FeII (F-
2
HXTA)(H2O)4]− in solution. This is presumably true of 2′ as
well, although the MnII complex is NMR silent.
FeII/MnII Exchange Equilibria. Paramagnetic 19F NMR has
increasingly been used to investigate both metalloproteins and
synthetic model systems.25 We employed 19F NMR to measure
equilibrium concentrations of paramagnetic F-HXTA com-
plexes because of the better resolution and fewer lines
1
F-HXTA Complexes in Solution. The H NMR spectrum
of isolated 1′ exhibits seven resonances of equal intensity,
consistent with a C2 symmetric F-HXTA ligand as observed in
the solid state. Although FeII is substitutionally labile, chelation
slows ligand exchange sufficiently that fluxional behavior on the
NMR time scale is not observed. Our data is consistent with the
structurally similar diiron(II) complex, [Fe2(BPMP)(μ-O2P-
(OPh)2)2]+ (BPMP = 2,6-bis[(bis(2-pyridylmethyI)amino)-
methyl]-4-methylphenol), which has been thoroughly charac-
1
compared with H spectra. Equilibrium constants for FeII/
MnII exchange show that FeII binds more strongly than MnII
(K1−3). Although this trend is typical, the magnitude of the
preference is relatively small. Complete metal ion replacement
for a bimetallic complex (eq 4) is equal to the sum of the
stepwise reactions (eqs 5 and 6) and K1 = K2K3. For
comparison to mononuclear complexes, we will define the
FeII/MnII discrimination constant (βFe/βMn) as √K1 = 13, or
19
1
terized by 1-D and 2-D H NMR. Coupling of electron and
nuclear magnetic moments disperses the proton resonances for
1 over 170 ppm and produces substantial line broadening by
providing an efficient mechanism for nuclear relaxation.18 The
E
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
the average equilibrium constant for a single MnII → FeII
substitution. EDTA (ethylenediamine tetraacetate) is structur-
ally similar to F-HXTA, and [MII(EDTA)]−2 also shows a
modest preference for FeII (βFe/βMn = 1.4).26 FeII affinity
increases dramatically with pyridine ligands, for example, βFe/
βMn = 1.5 × 104 for [MII(pic)3]− (pic = pyridine-2-
carboxylate).27 The trend continues for complexes with only
pyridyl donors: [MII(bpy)3]+2 (bpy = 2,2′-bipyridine) has a
FeII/MnII discrimination constant 10 orders of magnitude
greater than F-HXTA (βFe/βMn = 5.4 × 1011).28 Hard−soft
acid−base theory (HSAB) provides an explanation for these
data. Both FeII and MnII are moderately hard Lewis acids that
form strong bonds with similar Lewis bases, such as pyridine
and imidazole. Maximizing metal−ligand orbital overlap with a
good HSAB match increases βFe/βMn by magnifying the effect
of the differences between the two metals (LFSE, ionic radii, d6
vs d5 geometric preferences). Conversely, a HSAB mismatch
with very hard carboxylate ligands minimizes these differences
and produces a small FeII/MnII discrimination factor because
the slightly harder MnII ion competes more effectively with FeII
for available metal binding sites.
The relationship between K2 and K3 determines the relative
stability of the heterobimetallic complex 3. If the two metal
binding sites in F-HXTA were completely independent from
one another, each MnII → FeII substitution would be
isoenergetic and the stepwise replacement constants would be
equal (K3 = K2 = √K1). If this were true, the preference for FeII
could be explained by the same arguments used for
mononuclear complexes. Although these factors are certainly
still important, the observation that K2 > K3 (20.1 1.3 vs 9.1
1.1) indicates an additional favorable interaction in 3 specific
to the heterobimetallic state. Since the crystal structures of 1
and 2 are nearly identical, it is reasonable to assume that 3
would be similar as well, which means there is not a substantial
change in ligand geometry that can be used to explain the
observed stability enhancement. MII(μ-OR)MII clusters of FeII
and MnII ions generally experience weak antiferromagnetic
coupling for both homo- and heteronuclear combinations,9c,23
so differences in magnetic coupling do not provide an obvious
explanation for the stability of 3. Because K2 is only slightly
larger than K3, only a relatively small steric or electronic effect is
necessary to create the observed bias toward 3, but as discussed
below for Ct R2c and Gk R2lox, even a small effect may be
important for heterobimetallic cofactor assembly.
neglected. On the other hand, protonation of carboxylate
residues in 1−3 should be considered because this has been
observed in the solid state for other MII (HXTA) complexes.16a
2
Unbuffered solutions prepared from H5(F-HXTA), MCl2, and
stoichiometric NaOH (5 equiv) generally had a pH of 7−8
suggesting the metal complexes are very weakly basic.
Nevertheless, if the pKa of H[M(F-HXTA)(H2O)4] varies
slightly with metal identity, a small systematic error in the
calculated concentrations of 1−3 will be introduced.
The effect of chloride in the reaction medium is more
significant. MCl2 salts were utilized because they are available in
high purity for both metals and it was desirable to keep the
counterion identity constant. Chloride is a coordinating anion,
and the concentration of [M(H2O)5Cl]+ is not negligible, but
fortunately it does not vary much with the FeII/MnII ratio
because the formation constants for the two metals are similar
(log βFeCl = −0.16; log βMnCl = −0.61).22 When βFeCl and
+
+
+
+
βMnCl were included in the speciation model, the calculated
fraction of the total metal pool found as [M(H2O)5Cl]+ varied
from 2.9% to 1.1% as the mole fraction of MnII varied from 0 to
1. This change makes a small difference in the calculated
concentrations of 1−3 and is at least partially responsible for
the systematic error seen in Figure 4. Chloride may also
coordinate directly to F-HXTA complexes, but the importance
of species like [MII (F-HXTA)(H2O)3Cl]−2 cannot be
2
determined from our data. The magnitude of the effect is
probably smaller than for MII(aq) because addition of Cl− to
anionic [MII (F-HXTA)]− complexes is presumably less
2
favorable than addition to dicationic hexaaquo ions.
Relevance to Protein Cofactor Assembly. Comparison of
the active sites of Ct R2c and Gk R2lox with the crystal
structures of 1 and 2 shows that F-HXTA is a reasonable small-
molecule analog of the natural systems. The metal ions in 1 and
2 are found in distorted octahedrons with a single neutral
nitrogen donor and five anionic oxygen donors, creating a hard
ligand field that exhibits a relatively modest ability to
discriminate between FeII and MnII. This HSAB mismatch
also causes FeII and MnII F-HXTA complexes to oxidize very
rapidly in air. Ct R2c and Gk R2lox contain six-coordinate metal
ions as well, each with one neutral nitrogen donor (histidine)
and a collection of anionic oxygen donors (carboxylates and
possibly hydroxide in the case of Ct R2c). This abundance of
hard Lewis bases has interesting consequences for the assembly
of bimetallic cofactors in Ct R2c and Gk R2lox proteins. The
catalytically active forms of both enzymes contain high-valent
metals that are stabilized by a hard ligand field,11 but the metals
are loaded in the divalent state. This creates a HSAB mismatch
between the metal ions and the binding site during cofactor
assembly, which in the case of simple coordination compounds
reduces the FeII/MnII discrimination factor (βFe/βMn). Because
cytoplasmic concentrations of FeII and MnII are similar,4c
carboxylate-rich binding sites should enhance competitive
formation of {MnIIFeII} cofactors.
Speciation Model Selection. Equations 4−6 outline a
relatively simple speciation model that assumes complete
formation of anionic [MII (F-HXTA)]− complexes in the
2
presence of excess MII(aq). Each species is an aggregate of
multiple states (protonation, chloride complexation), and in
order to calculate K1−3 it must be assumed that the composition
of these states is invariant across all Fe/Mn ratios. Calculating
the expected concentrations of 1−3 from K1−3 (the curves in
Figure 4) is a way to test this assumption. A slight systematic
error between the measured and calculated concentrations is
apparent in Figure 4, and although the discrepancy is small
enough to not affect the conclusion that K2 > K3,29 further
discussion of the speciation model is warranted.
Equilibrium binding of MnII/FeII mixtures to apoenzymes
has been investigated for both Ct R2c and Gk R2lox. Griese et
al.9c reported crystal-soaking experiments of metal-free Gk
R2lox in excess MII at a 1:1 Fe/Mn ratio. Analysis of the metal
content by X-ray anomalous dispersion revealed that site 1
contained approximately equimolar amounts of each metal
while site 2 contained excess FeII in about a 4:1 ratio. In terms
of macroscopic selectivity, Gk R2lox preferentially binds FeII in
a 65:35 ratio from a 1:1 mixture of aqueous metal salts. This
level of MII discrimination is comparable to our model system:
All samples were buffered at pH 7.29, and protonation
equilibria were neglected. This is reasonable for [M(H2O)6]+2
+
because the vanishingly small hydrolysis constants (log βFeOH
=
−9.88; log βMnOH = −10.59)30 cause an insignificant change
when included in the speciation model. If the aquo ligands in
1−3 have comparable acidities, their hydrolysis is also safely
+
F
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Under similar conditions,31 F-HXTA preferentially binds FeII in
an 83:17 ratio. Equilibrium binding to Ct R2c has been
reported by Dassama et al.10 at stoichiometric MII loadings (2
equiv per polypeptide; 1:1 Fe/Mn ratio). Macroscopic
discrimination cannot be determined at these conditions (the
bound Fe/Mn ratio is necessarily 1:1 if all sites are occupied),
(K2 = 20.1 1.3 for 2 ⇌ 3; K3 = 9.1 1.1 for 3 ⇌ 1). The fact
that K2 is greater than K3 indicates that there is an additional
favorable interaction in 3 specific to the heterobimetallic state.
Although there are not any obvious steric or electronic
explanations for this effect in the F-HXTA model, if found to
be generally true for MnII/FeII complexes, this effect provides
another explanation for how proteins like R2c and R2lox can
selectively assemble heterobimetallic cofactors. We are
currently investigating the stability of FeII/MnII complexes in
other ligand fields, including a nonsymmetric F-HXTA
derivative that will mimic the high and low affinity sites in
R2c and R2lox proteins.
but a very strong microscopic preference for the {MnII(1)FeII
}
(2)
cofactor was observed. In fact, within the limits of detection by
X-ray anomalous dispersion, MnII was found exclusively in site
1 and FeII exclusively in site 2. Although stoichiometric MII also
maximizes the concentration of 3 in our model,32 the
impressive thermodynamic preference for the heterobimetallic
cofactor in Ct R2c is clearly greater than that in F-HXTA.
The explanation for selective formation of a {MnII(1)FeII
}
(2)
EXPERIMENTAL PROCEDURES
■
cofactor in R2c and R2lox proteins has been discussed
extensively. What is particularly intriguing is the mechanism
by which site 1 subverts the usual preference for FeII over MnII.
A comparison of binding sites in a series of RNR R2 class Ia, b,
and c proteins by Dassama et al.10 suggests that MnII selectivity
may result from the presence of exogenous water ligands that
allow a more flexible coordination geometry. Similar arguments
have been made for small-molecule coordination complexes
that prefer MnII to FeII.13 Chelating ligands that enforce a
subtle distortion or a reduction in coordination number may
alter metal binding preference.33 A rigid binding pocket can
select the slightly larger MnII cation in favor of the smaller FeII
cation.34 Although there are probably many contributing factors
to selective cofactor assembly in R2c and R2lox proteins, our
results suggest a new possibility should be investigated: Binding
of MnII in site 1 may be affected by the presence of FeII in site
2, not just through the rearrangement of residues in the
adjacent site9c but also through an intrinsic favorable
interaction between the metal ions in a {MnIIFeII} cluster.
General. Aqueous solutions were prepared from Type 1 ultrapure
water with a resistivity of at least 18 MΩ·cm−1. FeCl2·4H2O from
Strem was stored under N2(g), and anhydrous MnCl2 from Alfa Aesar
was stored in a desiccator. N-Methylmorpholine from Aldrich was
distilled from sodium metal under N2(g). 4-Fluorophenol from
Aldrich was sublimed under vacuum when used as an internal
standard. All other reagents were obtained from Aldrich or Acros and
used as received.
FeII and MnII complexes of F-HXTA are air sensitive and were
prepared and handled in a COY Laboratories anaerobic chamber
containing <1 ppm of O2. All solvents and liquid reagents were
freeze−pump−thaw degassed prior to use. NMR samples in D2O were
sealed in J-Young-style NMR tubes. NMR samples in mixed H2O/
D2O media were frozen in liquid nitrogen and flame-sealed under
vacuum in medium wall NMR tubes. Crystallization samples were
frozen in liquid nitrogen and flame-sealed under vacuum in glass
storage ampules. Caution! While no problems were encountered in this
work, flame-sealing vessels in liquid nitrogen may create an explosion
hazard. Always confirm that no gases have condensed in the vessel prior to
sealing and allow the samples to thaw completely behind a blast shield.
5-Fluoro-2-hydroxy-1,3-xylene-α,α′-diamine-N,N,N′,N′-tet-
raacetic acid, H5(F-HXTA). A solution of 0.616 g of NaOH (15.4
mmol) in 2.63 mL of H2O was cooled in an ice bath, and 1.051 g of
iminodiacetic acid (7.90 mmol) was added. After complete dissolution,
0.444 g of 4-fluorophenol (3.96 mmol) was added followed by 0.369 g
of paraformaldehyde (12.3 mmol). The cloudy yellow solution was
warmed to room temperature and then heated to 50 °C. After heating
for 48 h, the homogeneous orange solution was diluted with 23 mL of
methanol and cooled in an ice bath. The crude product was
precipitated with 0.320 mL of 12.1 M HCl (3.87 mmol). The solid
was isolated by filtration and dried under vacuum to yield 1.189 g (ca.
2.54 mmol) of Na3H2(F-HXTA). The crude product was dissolved in
12.7 mL of H2O and addition of 3.13 mL of 2.42 M HCl (7.57 mmol)
caused H5(F-HXTA) to crystallize slowly at room temperature. The
solid was isolated by filtration and dried under vacuum to yield 0.914 g
CONCLUSIONS
■
Equilibrium binding of FeII/MnII mixtures to the dinucleating
ligand F-HXTA was used as a chemical model of hetero-
bimetallic cofactor loading in R2c and R2lox proteins.
Homobimetallic species containing the complex anion
[MII (F-HXTA)(H2O)4]− (1, M = Fe; 2, M = Mn) were
2
characterized by single crystal X-ray diffraction. The two
structures are nearly identical except for small differences
attributed to slightly longer bonds for MnII. The complexes are
small-molecule analogs of R2c and R2lox proteins with each
metal in a distorted octahedral ligand field dominated by hard
carboxylate residues. NMR data shows that 1 retains its
structure in solution, and the same is presumed for 2 although
the MnII complex is NMR silent. The heterobimetallic variant 3
is formed from mixtures of 1 and 2, demonstrating that MII
exchange in F-HXTA complexes is facile. Metal competition
experiments were performed using 19F NMR to measure the
concentrations of 1 and 3 at various FeII/MnII ratios and
equilibrium constants for FeII/MnII exchange were determined.
Complete replacement of MnII with FeII is favorable (K1 = 182
13 for 2 ⇌ 1), but the magnitude of the preference is
relatively small. This is attributed to a HSAB mismatch between
the carboxylate ligands in F-HXTA and divalent transition
metal ions. This demonstrates how a binding site dominated by
carboxylate residues may be advantageous for heterobimetallic
proteins like R2c and R2lox by allowing the usually less
favorable {MnIIFeII} cofactor to compete with {FeIIFeII}. The
relative stability of the heterobimetallic complex 3 was assessed
by comparison of stepwise MnII → FeII equilibrium constants
1
(2.27 mmol, 57%) of pale yellow crystals. H NMR (DMSO, 400
3
MHz): δ 6.96 (d, 2H, JHF = 9.2 Hz), 3.83 (s, 4H), 3.43 (s, 8H). 19F
NMR (DMSO, 376 MHz): δ −128.5 (t, 3JFH = 9.2 Hz, 1JFC = 234 Hz).
13C{1H} NMR (DMSO, 100 MHz) δ 172.5, 155.0 (d, 1JCF = 233 Hz),
3
151.1, 125.0, 114.6 (d, JCF = 22 Hz), 54.0, 53.0. Anal. Calcd for
C16H19FN2O9: C, 47.77; H, 4.76; N, 6.96. Found: C, 47.48; H, 4.69;
N, 6.88.
[Fe(H2O)6][Fe2(F-HXTA)(H2O)4]2·14H2O. In an anaerobic cham-
ber, 67 mg of H5(F-HXTA) (0.17 mmol) was dissolved in 1.49 mL of
H2O with 0.670 mL of NaOH (0.995 M, 0.666 mmol), and to this
solution was added 0.960 mL of FeCl2 (0.5221 M, 0.501 mmol). The
solution was titrated to pH 7.13 with 0.975 mL of NaOH (0.166 M,
0.162 mmol). A fine blue precipitate formed immediately and was
removed with a 0.45 μm nylon syringe filter. The supernatant was
frozen in liquid nitrogen and sealed in a glass ampule. The product
precipitated after thawing, but heating to 50 °C caused it to redissolve.
The homogeneous solution was stored at 5 °C, and colorless crystals
suitable for X-ray structure determination grew in 1 week. The product
was isolated by filtration in an anaerobic chamber and dried over P2O5
G
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
1
for 1 week. H NMR (D2O, 400 MHz): δ 165.9 (br, 2H), 88.0 (br,
acquisition was mitigated by increasing the amplifier blanking delay
(rof2) to 40 ms.
2H), 78.9 (br, 2H), 58.8 (br, 2H), 25.4 (br, 2H), 17.6 (br, 2H), −5.2
(br, 2H). 19F NMR (D2O, 376 MHz): δ −74.3. Anal. Calcd for
C32H40F2Fe5N4O24 (loss of 22 H2O): C, 32.52; H, 3.41; N, 4.74.
Found: C, 32.28; H, 3.48; N, 4.60.
Procedure for Equilibrium Measurements. The wide sweep width
used in 19F experiments (33.8 kHz) made it impossible to produce a
uniform 90° pulse for all three resonances of interest (1, 3, and the
internal standard 4-FP). Two acquisitions were performed per sample,
varying the transmitter offset frequency to be equidistant from 4-FP
and either 1 or 3. A 90° pulse was calibrated separately for each
experiment. A total of 51 200 transients were collected with a total
recycle time of 0.48 s (0.08 s acquisition; 0.4 s delay), which was at
least 6× greater than the longest T1 (dependent on FeII/MnII ratio).
Probe temperature was maintained at 25.0 °C. The FID was zero-filled
to 32 000 points, and acoustic ringing was removed in the first four
points using backward-linear prediction. Twenty hertz of exponential
line broadening was applied, and resonances were integrated over 18
times the peak-width at half-height to cover 99% of the area of a
Lorentzian distribution.
[Mn(H2O)6][Mn2(F-HXTA)(H2O)4]2·14H2O. In an anaerobic cham-
ber, 97 mg of H5(F-HXTA) (0.24 mmol) was dissolved in 0.970 mL of
NaOH (0.995 M, 0.965 mmol), and to this solution was added 1.19
mL of MnCl2 (0.6086 M, 0.7248 mmol). The solution was titrated to
pH 8.42 with 0.780 mL of NaOH (0.291 M, 0.227 mmol). After 2 h, a
fine white precipitate had formed, which was removed with a 0.45 μm
nylon syringe filter. The supernatant was frozen in liquid nitrogen and
sealed in a glass ampule. Colorless crystals suitable for X-ray structure
determination grew in 1 week at 5 °C. The product was isolated by
filtration in an anaerobic chamber and dried over P2O5 for 1 week.
Anal. Calcd for C32H40F2Mn5N4O24 (loss of 22 H2O): C, 32.64; H,
3.42; N, 4.76. Found: C, 32.44; H, 3.46; N, 4.46.
Speciation Modeling. Concentration curves for 1−3 in Figure 4
were calculated in Hyss200937 at the following initial conditions: [FeII]
= 40.18−0 mM; [MnII] = 0−40.24 mM; [F-HXTA] = 9.03 mM; [Cl−]
= 80.4 mM. The model used for this calculation consisted of formation
constants (β) for 1, 2, 3, [FeCl]+(aq), and [MnCl]+(aq). β1−3 were
calculated from K1−3 by choosing an arbitrarily large value of β1 (1020)
such that essentially no F-HXTA was present and then calculating
Single Crystal X-ray Crystallography. All reflection intensities
were measured at 110(2) K using a SuperNova diffractometer
(equipped with Atlas detector) with Mo Kα radiation (λ = 0.71073
Å) under the program CrysAlisPro (version 1.171.36.32, Agilent
Technologies, 2013). The same program was used to refine the cell
dimensions and for data reduction. Structures were solved with the
program SHELXS-2013 (Sheldrick, 2008) and refined on F2 with
SHELXL-2013 (Sheldrick, 2008). Analytical numeric absorption
correction based on a multifaceted crystal model was applied using
CrysAlisPro. The temperature of the data collection was controlled
using the system Cryojet (manufactured by Oxford Instruments). The
H atoms were placed at calculated positions (unless otherwise
specified) using the instructions AFIX 23 or AFIX 43 with isotropic
displacement parameters having values 1.2Ueq of the attached C atoms.
The H atoms of the coordinated (OnW, n = 1−7) and lattice (OnW, n
= 8−14) water molecules were found from difference Fourier maps,
and their coordinates were refined freely. Both structures are ordered
with the [M(H2O)6]+2 cations found at sites of inversion symmetry,
thus only one-half of the cation is crystallographically independent.
The crystals of 1 and 2 were pseudomerohedrally twinned. Twinning
was checked with the TwinRotMat algorithm from Platon.35 The twin
relationship corresponds to a 2-fold axis along the a direction with the
+
+
relative values of β2 and β3 from eqs 4−6. βFeCl and βMnCl were taken
from the literature.22
Magnetic Susceptibility. Measurements were made at 25 °C
using the Evans method adapted to the geometry of superconducting
magnets.20 Accurately weighed samples of 1′ and 2′ were dissolved in
D2O in volumetric flasks to a target concentration of ca. 10 mg/mL.
The solutions were prepared to contain 1.0 mM tert-butanol and were
loaded into the outer chambers of coaxial NMR tubes. The inner
chamber contained a reference solution of 1.0 mM tert-butanol in
D2O. The molar susceptibilities of the dissolved complexes (χM) were
calculate from the equation
⎛
⎜
⎝
⎞
⎟
⎠
3Δν
χM = MW
χ +
0
solute
4πν c
(7)
o
matrix M: 1 0 0/0 1 0/1 0 1. The BASF scale factors refine to
0.1331(7) for 1 and 0.0416(5) for 2.
where MWsolute is the molecular weight of the metal complex, χ0 is the
mass susceptibility of the solvent D2O (−0.65 × 10−6 cm3/g), Δν is
the difference in hertz between the tert-butanol resonances in the inner
and outer chambers, νo is the operating frequency of the spectrometer
in hertz, and c is the concentration of the metal complex in grams per
milliliter. The paramagnetic contribution to the observed molar
susceptibility (χP) was calculated by subtracting the diamagnetic
contribution (χD) from χM. Pascal constants were used to estimate χD
for 1′ (−508 × 10−6 cm3/mol) and 2′ (−513 × 10−6 cm3/mol).38 The
paramagnetic susceptibilities were used to determine effective
magnetic moments (μeff) expressed in units of the Bohr magneton
(μB)
̅
̅
̅
Metal Exchange Experiments. Solutions were prepared by
diluting aliquots of aqueous stock solutions to the final target
concentrations. Calibrated automatic pipettes and volumetric flasks
were used throughout. Because H5(F-HXTA) is insoluble in water,
stock solutions were prepared as Na3H2(F-HXTA) by adding 3 equiv
of NaOH.
Representative Procedure. To a 5 mL volumetric flask containing
1.000 mL of D2O was added 1.494 mL of Na3H2(F-HXTA) (0.03055
M, 0.0456 mmol), 0.400 mL of FeCl2 (0.2010 M, 0.0804 mmol), 0.598
mL of MnCl2 (0.2012 M, 0.120 mmol), 0.120 mL of 4-fluorophenol
(0.2070 M, 0.0248 mmol), and 0.494 mL of N-methylmorpholine
(0.302 M, 0.149 mmol). The flask was diluted to the mark with H2O.
An aliquot of this solution was sealed in an NMR tube and equilibrated
at 25.0 °C for 24 h prior to analysis. The pH of the remaining solution
was measured after 24 h, and the raw meter reading was adjusted for
isotopic composition using the formula pHactual = pHmeter − n(0.073 ×
pHmeter − 0.42) where n is the deuterium mole fraction of the
sample.36
μ
= 2.84 χ T
P
(8)
eff
where T is the temperature in Kelvin. These values were compared
with the predicted magnetic moments for complexes containing n
noninteracting metal ions
i=1
μ2 = g2( S(S + 1))
∑
i
i
NMR Spectroscopy. Spectra were acquired on a Varian 400 MHz
DirectDrive spectrometer equipped with a 5 mm ASW PFG probe.
Data were processed in iNMR or MNova. Chemical shifts are reported
(9)
n
where g is the Lande factor (2.2 for high-spin FeII and 2.0 for high-spin
́
1
relative to residual solvent resonances for H and 13C and relative to
MnII) and Si is the total electron spin quantum number for each
individual metal ion.21
NaPF6 for 19F. T1 measurements were made with a calibrated 90°
pulse using a standard inversion−recovery pulse sequence. Several
changes to standard Varian pulse sequences were necessary for
paramagnetic samples: The automatic gradient shimming routine was
modified by increasing the number of transients acquired (nt) to 64,
decreasing the relaxation delay (d1) to 0.1 s, and decreasing the
incremental delay array (d3) to [0, 0.01] s. Acoustic ringing during 19F
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
H
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Branca, R. M. M.; Lehtio, J.; Graslund, A.; Lubitz, W.; Siegbahn, P. E.;
̈
̈
chem.5b02322. CIFs have been deposited in the Cambridge
Structural Database.
Hogbom, M. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 17189−17194.
̈
(10) Dassama, L. M. K.; Krebs, C.; Bollinger, J. M., Jr; Rosenzweig, A.
C.; Boal, A. K. Biochemistry 2013, 52, 6424−6436.
NMR spectra of H5(F-HXTA), 1′, a titration of F-HXTA
with FeCl2, and of a 1′/2′ mixture. Discussion of the
method used to calculate T1 time constants for 1, and
complete data set and results of F-HXTA metal
competition experiments (PDF)
Crystallographic information file (CIF) for 1 (CIF)
Crystallographic information file (CIF) for 2 (CIF)
(11) The oxidized states are believed to be {MnIV(1)FeIII(2)} for Ct
R2c and {MnIV(1)FeIV(2)} for Gk R2lox, see ref 6.
(12) Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry;
Wiley: New York, 1988; pp 697−724.
(13) A search of the IUPAC stability constant database yields 347
entries for ligands with known stability constants for both FeII and
MnII. In only 20 cases is the corresponding MnII complex more stable.
See http://www.iupac.org/index.php?id=410 (accessed September 24,
2015).
AUTHOR INFORMATION
Corresponding Author
*E-mail: will.kerber@bucknell.edu.
■
(14) Wang, J.; Martell, A. E.; Motikatis, R. J. Inorg. Chim. Acta 2001,
322, 47−55.
(15) (a) Tereniak, S. J.; Carlson, R. K.; Clouston, L. J.; Young, V. G.,
Jr.; Bill, E.; Maurice, R.; Chen, Y.-S.; Kim, H. J.; Gagliardi, L.; Lu, C. C.
Notes
The authors declare no competing financial interest.
J. Am. Chem. Soc. 2014, 136, 1842−1855. (b) Carboni, M.; Clem
́
ancey,
M.; Molton, F.; Pecaut, J.; Lebrun, C.; Dubois, L.; Blondin, G.; Latour,
́
J. M. Inorg. Chem. 2012, 51, 10447−10460. (c) Wang, Z.; Holman, T.
R.; Que, L., Jr. Magn. Reson. Chem. 1993, 31, S78−S84. (d) Holman,
T. R.; Wang, Z.; Hendrich, M. P.; Que, L., Jr. Inorg. Chem. 1995, 34,
134−139.
ACKNOWLEDGMENTS
■
We are grateful for financial support provided by Bucknell
University and the donors of the American Chemical Society
Petroleum Research Fund (Grant 50877-UNI3). W.D.K.
thanks Dr. David Rovnyak for helpful discussion of quantitative
paramagnetic NMR.
́
(16) (a) Laborda, S.; Clerac, R.; Anson, C. E.; Powell, A. K. Inorg.
Chem. 2004, 43, 5931−5943. (b) Meng, B.-H.; Gao, F.; Zhu, M.-L.
Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 2004, 60, m308−
m310. (c) Gao, F.; Meng, B.-H.; Wei, Y.-B. Acta Crystallogr., Sect. C:
Cryst. Struct. Commun. 2004, 60, m360−m362. (d) Ma, L.; Lu, L.; Zhu,
M.; Wang, Q.; Gao, F.; Yuan, C.; Wu, Y.; Xing, S.; Fu, X.; Mei, Y.;
Gao, X. J. Inorg. Biochem. 2011, 105, 1138−1147.
REFERENCES
■
(1) Crichton, R. R. Biological Inorganic Chemistry; Elsevier:
Amsterdam, 2012.
(17) Murch, B. P.; Bradley, F. C.; Boyle, P. D.; Papaefthymiou, V.;
Que, L., Jr. J. Am. Chem. Soc. 1987, 109, 7993−8003.
(18) (a) Bertini, I.; Luchinat, C. NMR of Paramagnetic Molecules in
Biological Systems; Benjamin/Cummings: Menlo Park CA, 1986.
(b) Bertini, I.; Turano, P.; Vila, A. J. Chem. Rev. 1993, 93, 2833−2932.
(19) Ming, L. J.; Jang, H. G.; Que, L., Jr. Inorg. Chem. 1992, 31, 359−
364.
(2) (a) Stubbe, J.; Nocera, D. G.; Yee, C. S.; Chang, M. C. Y. Chem.
Rev. 2003, 103, 2167−2201. (b) Sazinsky, M. H.; Lippard, S. J. Acc.
Chem. Res. 2006, 39, 558−566. (c) Schenk, G.; Mitic, N.; Gahan, L. R.;
́
Ollis, D. L.; McGeary, R. P.; Guddat, L. W. Acc. Chem. Res. 2012, 45,
1593−1603. (d) Peterson, R. L.; Kim, S.; Karlin, K. D. In
Comprehensive Inorganic Chemistry II; Reedijk, J. K. P., Ed.; Elsevier:
Oxford, 2013; Vol. 3, pp 149−177. (e) Grazina, R.; Pauleta, S. R.;
Moura, J. J. G.; Moura, I. In Comprehensive Inorganic Chemistry II;
Reedijk, J. K. P., Ed.; Elsevier: Oxford, 2013; Vol. 3, pp 103−148.
(3) (a) Rae, T. D.; Schmidt, P. J.; Pufahl, R. A.; Culotta, V. C.;
O’Halloran, T. V. Science 1999, 284, 805−808. (b) Rosenzweig, A. C.
Chem. Biol. 2002, 9, 673−677. (c) Philpott, C. C. J. Biol. Chem. 2012,
287, 13518−13523.
(20) (a) Evans, D. F. J. Chem. Soc. 1959, 2003−2005. (b) Sur, S. K. J.
Magn. Reson. 1989, 82, 169−173.
(21) Girerd, J.-J.; Journaux, Y. In Physical Methods in Bioinorganic
Chemistry: Spectroscopy and Magnetism; Que, L., Jr., Ed.; University
Science Books: Sausalito, CA, 2000; pp 321−374.
(22) (a) Heinrich, C. A.; Seward, T. M. Geochim. Cosmochim. Acta
1990, 54, 2207−2221. (b) Gammons, C. H.; Seward, T. M. Geochim.
Cosmochim. Acta 1996, 60, 4295−4311.
(4) (a) Kakhlon, O.; Cabantchik, Z. I. Free Radical Biol. Med. 2002,
33, 1037−1046. (b) Tottey, S.; Waldron, K. J.; Firbank, S. J.; Reale, B.;
Bessant, C.; Sato, K.; Cheek, T. R.; Gray, J.; Banfield, M. J.; Dennison,
C.; Robinson, N. J. Nature 2008, 455, 1138−1142. (c) Hider, R. C.;
Kong, X. L. BioMetals 2011, 24, 1179−1187.
(23) Que, L., Jr.; True, A. E. Prog. Inorg. Chem. 1990, 38, 97−200.
(24) Borovik, A. S.; Murch, B. P.; Que, L., Jr.; et al. J. Am. Chem. Soc.
1987, 109, 7190−7191.
́
(25) Belle, C.; Beguin, C.; Hamman, S.; Pierre, J.-L. Coord. Chem.
(5) (a) Cotruvo, J. A., Jr.; Stubbe, J. Annu. Rev. Biochem. 2011, 80,
733−767. (b) Stubbe, J.; Cotruvo, J. A., Jr. Curr. Opin. Chem. Biol.
2011, 15, 284−290. (c) Cotruvo, J. A., Jr.; Stubbe, J. Metallomics 2012,
4, 1020−1036.
Rev. 2009, 253, 963−976.
(26) Brunetti, A. P.; Nancollas, G. H.; Smith, P. N. J. Am. Chem. Soc.
1969, 91, 4680−4683.
(27) Anderegg, G. Helv. Chim. Acta 1960, 43, 414−424.
(28) Irving, H.; Mellor, D. H. J. Chem. Soc. 1962, 5222−5237.
(29) If K2 = K3, there would necessarily be a FeII/MnII ratio that
produces equimolar amounts of 1, 2, and 3. Inspection of Figure 4
shows that this is clearly not the case.
(6) (a) Griese, J. J.; Srinivas, V.; Hogbom, M. JBIC, J. Biol. Inorg.
̈
Chem. 2014, 19, 759−774. (b) Bollinger, J. M., Jr; Jiang, W.; Green, M.
T.; Krebs, C. Curr. Opin. Struct. Biol. 2008, 18, 650−657. (c) Carboni,
M. L.; Latour, J.-M. Coord. Chem. Rev. 2011, 255, 186−202.
(7) Hogbom, M.; Stenmark, P.; Voevodskaya, N.; McClarty, G.
̈
(30) (a) Millero, F. J.; Yao, W.; Aicher, J. Mar. Chem. 1995, 50, 21−
39. (b) Perrin, D. D. J. Chem. Soc. 1962, 2197−2200.
(31) Initial conditions: 9 mM F-HXTA; 18 mM FeCl2; 18 mM
MnCl2. Calculated speciation: 2.62 mM 3, 5.86 mM 1, and 0.525 mM
2.
Science 2004, 305, 245−248.
(8) (a) Jiang, W.; Yun, D.; Saleh, L.; Barr, E. W.; Xing, G.; Hoffart, L.
M.; Maslak, M. A.; Krebs, C.; Bollinger, J. M. Science 2007, 316, 1188−
1191. (b) Voevodskaya, N.; Lendzian, F.; Ehrenberg, A.; Graslund, A.
̈
̈
FEBS Lett. 2007, 581, 3351−3355. (c) Andersson, C. S.; Ohrstrom,
̈
(32) Initial conditions: 9 mM F-HXTA; 9 mM FeCl2; 9 mM MnCl2.
Calculated speciation: 3.85 mM 3, 2.58 mM 1, and 2.58 mM 2.
M.; Popovic-Bijelic, A.; Graslund, A.; Stenmark, P.; Hogbom, M. J. Am.
́
́
̈
̈
Chem. Soc. 2012, 134, 123−125. (d) Dassama, L. M. K.; Boal, A. K.;
Krebs, C.; Rosenzweig, A. C.; Bollinger, J. M., Jr. J. Am. Chem. Soc.
2012, 134, 2520−2523.
(33) (a) Delgado, R.; Frausto da Silva, J. J. R.; Vaz, M. C.; Paoletti,
́
P.; Micheloni, M. J. Chem. Soc., Dalton Trans. 1989, 133−137.
(b) Paoletti, P.; Ciampolini, M. Inorg. Chem. 1967, 6, 64−68.
(34) Miyake, H.; Kato, N.; Kojima, Y.; Sugihara, A. Inorg. Chim. Acta
1994, 223, 121−124.
(9) (a) Andersson, C. S.; Hogbom, M. Proc. Natl. Acad. Sci. U. S. A.
̈
2009, 106, 5633−5638. (b) Hogbom, M. JBIC, J. Biol. Inorg. Chem.
2010, 15, 339−349. (c) Griese, J. J.; Roos, K.; Cox, N.; Shafaat, H. S.;
̈
I
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(35) Spek, A. L. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2009, D65,
148−155.
(36) Kręzel, A.; Bal, W. J. Inorg. Biochem. 2004, 98, 161−166.
̇
(37) (a) Alderighi, L.; Gans, P.; Ienco, A.; Peters, D.; Sabatini, A.;
Vacca, A. Coord. Chem. Rev. 1999, 184, 311−318. (b) Hyperquad
Simulation and Speciation (HySS2009), Protonic Software, 2009;
http://www.hyperquad.co.uk/hyss.htm (accessed September 24,
2015).
(38) Bain, G. A.; Berry, J. F. J. Chem. Educ. 2008, 85, 532−536.
J
DOI: 10.1021/acs.inorgchem.5b02322
Inorg. Chem. XXXX, XXX, XXX−XXX