The synthesis of a corrole analogue of aquacobalamin (Vitamin B 12a) and its ligand substitution reactions

Article abstract of DOI:10.1021/ic5000793

The synthesis of a Co(III) corrole, [10-(2-[[4-(1H-imidazol-1-ylmethyl) benzoyl]amino]phenyl)-5,15-diphenylcorrolato]cobalt(III), DPTC-Co, bearing a tail motif terminating in an imidazole ligand that coordinates Co(III), is described. The corrole therefor

Full text of DOI:10.1021/ic5000793

Article
pubs.acs.org/IC
The Synthesis of a Corrole Analogue of Aquacobalamin
(Vitamin B12a) and Its Ligand Substitution Reactions
Caitlin F. Zipp,† Joseph P. Michael,† Manuel A. Fernandes,† Sadhna Mathura,† Christopher B. Perry,†
Isabelle Navizet,† Penny P. Govender,‡ and Helder M. Marques*,†
†Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, P.O. Wits, Johannesburg, 2050 South Africa
‡Department of Applied Chemistry, University of Johannesburg, P.O. Box 17011, Doornfontein, Johannesburg, 2028 South Africa
S
* Supporting Information
ABSTRACT: The synthesis of a Co(III) corrole, [10-(2-[[4-(1H-imidazol-
1-ylmethyl)benzoyl]amino]phenyl)-5,15-diphenylcorrolato]cobalt(III),
DPTC-Co, bearing a tail motif terminating in an imidazole ligand that
coordinates Co(III), is described. The corrole therefore places Co(III) in a
similar environment to that in aquacobalamin (vitamin B12a, H2OCbl+) but
with a different equatorial ligand. In coordinating solvents, DPTC-Co is a
mixture of five- and six-coordinate species, with a solvent molecule occupy-
ing the axial coordination site trans to the proximal imidazole ligand.
In an 80:20 MeOH/H2O solution, allowed to age for about 1 h, the pre-
dominant species is the six-coordinate aqua species [H2O−DPTC-Co]. It
is monomeric at least up to concentrations of 60 μM. The coordinated H2O has a pKa = 9.76(6). Under the same conditions
H2OCbl+ has a pKa = 7.40(2). Equilibrium constants for the substitution of coordinated H2O by exogenous ligands are reported
as log K values for neutral N-, P-, and S-donor ligands, and CN−, NO2 , N3 , SCN−, I−, and Cys in 80:20 MeOH/H2O solution
at low ionic strength. The log K values for [H2O−DPTC-Co] correlate reasonably well with those for H2OCbl+; therefore,
Co(III) displays a similar behavior toward these ligands irrespective of whether the equatorial ligand is a corrole or a corrin.
Pyridine is an exception; it is poorly coordinated by H2OCbl+ because of the sterically hindered coordination site of the corrin.
With few exceptions, [H2O−DPTC-Co] has a higher affinity for neutral ligands than H2OCbl+, but the converse is true for
anionic ligands. Density functional theory (DFT) models (BP86/TZVP) show that the Co−ligand bonds tend to be longer in
corrin than in corrole complexes, explaining the higher affinity of the latter for neutral ligands. It is argued that the residual charge
at the metal center (+2 in corrin, 0 in corrole) increases the affinity of H2OCbl+ for anionic ligands through an electrostatic
attraction. The topological properties of the electron density in the DFT-modeled compounds are used to explore the nature of
the bonding between the metal and the ligands.
−
−
thermodynamics and kinetics of their ligand substitution
reactions.10,16−19
1. INTRODUCTION
Co(III) is inert in many of its complexes. For example, NMR
−
When the ambidentate nucleophiles SeCN−, NO2 , and
methods show that the residence time of H2O in Co(III)aq is of
S2O32− replace H2O in aquacobalamin H2OCbl+ (see Figure 1),
they bind through the softer of the two donor atoms;11 SCN−
is N-bound in the solid state,11,20 but there is NMR evidence
that it exists as an approximately equal mixture of the S- and
N- bound species in solution.11 The preference for the softer
donor atom of the ligand suggests that Co(III) in the cobal-
amins is relatively soft. However, the kinetically favored product
in some cases is the harder of two donor atoms of an ambident
nucleophile: H2OCbl+ reacts initially through N of SCN− and
O of OCN− before equilibrating to a mixture of S- and
N-bound SCN− and N-bound OCN−.16
the order of 105 s,1 compared to ca. 10−8 s in Co(II)aq and 10−5
s
2
in Fe(III)aq.3 However, Co(III) is surprisingly labile in the cobalt−
corrins (derivatives of vitamin B12) in which the equatorial ligand
is a corrin (for example, references 4−8, see Figure 1). On the
basis of somewhat limited data, the approximate lability ratio of
the metal ion toward substitution in corrin, porphyrin, cobaloxime,
and tetraammine systems, where the equatorial ligand is a N4
donor system (eq 1, where the entering ligand X replaces L trans
to Y), is estimated to be 109:106:104:1.9,10
[Y − Co3+(N ) − L] + X → [Y − Co3+(N ) − X] + L
4
4
There is ample spectroscopic evidence that the positions of
the principal π → π* transitions of the corrin depend on the
donor power of the axial ligand.11,13,21,22 When the H atom at
the C10 position of the corrin (Z in Figure 1) is replaced by Cl
(1)
We have been interested for some time in exploring how the
equatorial ligand in the cobalt−corrins controls and modifies
the properties of Co(III). Toward this end we have reported
structural,9,11,12 spectroscopic,12,13 and density functional
theory (DFT) investigations14,15 into cobalt−corrins in which
the equatorial corrin ligand has been modified, as well as the
Received: January 13, 2014
Published: April 10, 2014
© 2014 American Chemical Society
4418
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
softer and more covalent character.12 The determination of
equilibrium constants (log K) for substitution of axially
coordinated H2O in the aquacyano derivatives of these
complexes, ACCbs and ACSYCbs, in which one of the cyanide
ligands is replaced by H2O by a number of anionic ligands,
showed not only that the softer metal in ACCbs has a pre-
ference for softer ligands but also that the harder metal in
ACSYCbs prefers the harder ligands, while the kinetics of the
substitution of H2O by CN− demonstrated that ACCbs is more
labile than ACSYCbs.18 Log K values for the reaction of these
compounds with neutral N-donor ligands showed that the
harder Co(III) in ACSYCbs prefers aliphatic amines with a
harder N donor, while the softer Co(III) in ACCbs prefers the
softer N donors of aromatic amines.19 We now take our
approach of modifying the electronic structure of the equatorial
ligand in the cobalt−corrins a step further by preparing a tailed
corrole, culminating in an imidazole ligand, which places
Co(III) in a similar coordination environment as in H2OCbl+
but with a corrole rather than a corrin as the equatorial ligand.
Corroles are aromatic tetrapyrrole macrocycles containing,
like corrins, a direct C1−C19 pyrrole−pyrrole bridge,21 re-
sulting in a macrocyclic cavity topologically similar to that of
corrins;22 however, with an aromatic 18 π-electron system, they
are more electron-rich.24 Corroles are stronger acids than
corrins because of the presence of three pyrrole-like and one
imine-like nitrogen donors; on coordination of metal ions,
corroles form trianions, whereas corrins form monoanions.
Corroles are therefore capable of binding metal ions in both
high and low oxidation states.24,25
There are a number of routes to the synthesis of corroles.
The independent announcement by two groups of a facile one-
pot synthesis26,27 in 1999 led to a significant increase in
the number of reports on corroles (Scheme 1, (i)). In this approach,
the first step is an acid-catalyzed electrophilic substitution reaction
between pyrrole and benzaldehyde derivatives, which produces
polymers of various lengths with alternating pyrrole and
aldehyde subunits. The second step involves use of an oxidant
such as 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) or
tetrachloro-1,4-benzoquinone (p-chloranil) in, for example, chloro-
form or dichloromethane, to effect an oxidative ring closure of the
polymers (pyrrole−pyrrole coupling) and subsequent aromatiza-
tion to yield a variety of macrocycles based on the length and
identity of the polymer chain produced in the first step.28 The
desired polymer for corrole synthesis is a bilane or tetrapyrrane
formed by the condensation of four pyrrole and three aldehyde
subunits.28 This approach yields A3 corroles (A is the substituent
at the 5, 10, and 15 positions) in low yields (5−15%).26,27,29,30
A novel method for the synthesis of mesosubstituted corroles
was introduced by Koszarna and Gryko.28 The reaction occurs
in water/methanol in the presence of 0.1 equiv of HCl and
exploits the difference in solubility of the starting materials and
the bilane product; whereas the starting aldehyde and pyrrole,
and the dipyrromethane intermediate, are soluble, the bilane
tends to precipitate, effectively halting the reaction at the this
stage. The isolated bilane is then oxidized to the corrole macro-
cycle. This procedure reduces the formation of porphyrins and
other products. The use of p-chloranil rather than DDQ tends
to increase corrole yields (typically between 13% and 56%).
A stepwise approach to a corrole usually begins with the syn-
thesis of a dipyrromethane28,31,32 from pyrrole and an aldehyde
in the presence of a catalytic amount of acid, often trifluoro-
acetic acid (TFA).33 The synthesis requires a large excess of
pyrrole (typically 40:1 pyrrole/aldehyde), short reaction times,
Figure 1. The standard view and numbering of the cobalamins.
The axial ligand L is on the upper or β face of the corrin. In vitamin
B12 itself (cyanocobalamin, CNCbl) Z = H, L = CN−. Aquacobalamin
(H2OCbl+) has Z = H, L = H2O. The trans ligand is the base 5,6-
dimethylbenzimidazole (dbzm). In cobyrinic acid, all side chain are
hydrolyzed to carboxylic acids; the cobesters are esters (usually methyl
esters) of cobyrinic acid.
in [L−(Co3+Cbl)], L = H2O, CN−, Me−, the C10−Cl bond
length depends on the polarizability of L.9 The replacement of
the C10 H in H2OCbl+ by NO deactivates the axial coordina-
tion site toward ligand substitution.17 Its replacement by Cl
perturbs both the thermodynamics and kinetics of the substitu-
tion of axial H2O.10
This accumulated evidence indicates that there is electronic
communication between the axial coordination site and the
equatorial ligand such that the structure of the latter can control
and modify the chemistry of the former, and that Co(III) in the
cobalamins is relatively soft. We have suggested9,10,18,23 that the
electron-rich equatorial corrin ligand transfers electron density
to Co(III), confers on it some labile Co(II) character, and that
this is the origin of the unusual kinetic lability of Co(III) in the
cobalt−corrins.
We recently reported on the characterization of a dicyano
stable yellow derivative of cobyrinic acid heptamethyl ester,
(5R,6R)-Coα,Coβ-dicyano-5,6-dihydro-5-hydroxy-heptamethylcob-
(III)yrinate-c,6-lactone (DCSYCbs), in which the C5 carbon is
oxidized such that the 13 atom, 14 π-e− delocalized system of
corrins is interrupted, giving a triazamethine system with four
conjugated double bonds between N22 and N24 and an
isolated double bond between N21 and C4 (see Figure 1 for
numbering).12 On the basis of 59Co NMR data and the
stretching frequency υCN of axially coordinated CN−, we
showed that Co(III) interacts less strongly with the equatorial
ligand than in dicyano cobyrinic acid heptamethyl ester
(DCCbs) itself. DFT calculations confirmed there is greater
overlap between the corrin and the metal orbitals in DCCbs
compared to DCSYCbs, which gives the metal in the former a
4419
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
a
Scheme 1. Three Synthetic Routes to Corroles
a
(i) Acid-catalyzed reaction between pyrrole and a benzaldehyde produces inter alia a bilane. Conducting the reaction in a water/methanol solvent
system causes precipitation of the bilane, halting the condensation reaction at this stage. (ii) There is greater control over the synthesis of the bilane
by proceeding through dipyrromethane intermediate (cat. TFA, 15 min, large excess of pyrrole). (iii) CH3OH/H2O 1:1, cat. HCl. (iv) p-Chloranil
(1 equiv) (v) EtMgBr (5 equiv), R′COCl (2.5 equiv). (vi) NaBH4 (50 equiv), THF/CH3OH 3:1 or NaBH4 (20 equiv), THF/CH3OH 10:1
(anhydrous). (vii) NH4Cl (2.5 equiv), BF3Et2O (0.2 equiv), CH3CN.
and low temperature (typically 15 min, room temperature) to
minimize the formation of oligomers or pyrrole polymerization.
Yields are usually better than 50%. Dipyrromethane units can
then be condensed with an aldehyde under acidic conditions
to form a bilane. Alternatively, the corrole macrocycle is con-
structed from the acylation and subsequent reduction of
dipyrromethane to form a dicarbinol, which is then coupled
with 2,2′-bipyrrole in a [2 + 2] condensation to yield a
corrole,32,34 with decreased chance of porphyrin formation.
Yields of corrole are typically up to 12%. There are several
routes to the synthesis of 2,2′-bipyrrole from pyrrole.35,36
Metalation of the corrole is usually effected by refluxing in an
appropriate solvent such as methanol with a metal acetate,
chloride, or carbonyl.37
There are relatively few studies on the ligand-substitution
reactions of corroles, and to our knowledge there is only one
example of a substitution reaction that has been studied in
aqueous solution, namely, the binding of human serum albumin
(HSA) to the gallium and manganese complexes of the water-
soluble 2,17-bis-sulfonato-5,10,15(trispentafluorophenyl)-
corrole.38 Both the free base and metalated corroles bind
HSA very strongly, such that the equilibrium constants for
these reactions could not be determined by ultraviolet−visible
(UV−vis) spectrophotometric titrations. A brief review of the
available equilibrium constants is given in Section S2.1 of the
Supporting Information.
variety of ligands and compare the results to those obtained
with H2OCbl+ under the same conditions to assess the effect of
the equatorial ligand on the coordination chemistry of Co(III).
2. RESULTS
2.1. Synthesis of a Tailed Corrole Mimic of Aqua-
cobalamin. We explored various routes to simple corroles39
and in our experience found Gryko’s method28 to be the most
generally successful. We then set out to design a corrole bearing
a “tail” substituent such that the terminus of the tail terminated
in an imidazole-type ligand for Co(III), so as to have similar
electronic properties to the 5,6-dimethylbenzimidazole
(dmbzm) proximal ligand of H2OCbl+. We employed
molecular mechanics methods (see Experimental Section,
Table S3.1 and Figure S2 of the Supporting Information) to
investigate the probable relative stability of a number of model
complexes. A variety of systems with the tail extending from a
C10 phenyl substituent and culminating in an aromatic N-
donor were evaluated (a few examples are given in Figure S3 of
the Supporting Information). Systems were evaluated based on
the value of the Zindo single-point energy of the molecular
mechanics energy-minimized structure, the tilt angle of the N-
donor base to the corrole plane, and the requirement that the
metalated corrole ring remain relatively planar, as found in the
crystal structures of a number of Co(III) corroles.
The system chosen consisted of an amide and a phenyl ring
(in a meta arrangement) between the meso phenyl and a
terminal imidazole (Figure 2). Molecular modeling showed this
system to have a relatively low energy with all structural metrics
within the expected values. There was little distortion of
the plane of the corrole macrocycle, and the cobalt-imidazole
bond was nearly perpendicular to the mean plane of the macro-
cycle. The tail was less flexible than in aliphatic-based systems
(Supporting Information, Figure S3) but less rigid than the
To realistically mimic the coordination environment of
Co(III) in H2OCbl+, the corrole model should (i) be soluble in
water or mixed solvents with a significant amount of water, (ii)
contain an aromatic N-donor as one of the axial ligands, and
(iii) have H2O as the other axial ligand. The design, synthesis,
characterization, and solution behavior of such a corrole is
described in this Paper. We also report equilibrium constants
(log K values) for the substitution of coordinated H2O by a
4420
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
presence of triethylamine in 57% yield after 2 h reaction;
diffraction-quality crystals of 3-(chloromethyl)-N-[2-(hydroxymethyl)-
phenyl]benzamide were obtained, and the molecular structure
was determined by X-ray diffraction (XRD) methods (Support-
ing Information, Section S2.4, CCDC deposition #980627). If
the reaction was allowed to proceed for more than 3 h, 2-[3-
(chloromethyl)benzamido]benzyl 3-(chloromethyl)benzoate was
produced instead. This too crystallized and was characterized by
XRD (Supporting Information, Section S2.4, CCDC deposition
#980628).
Imidazole was coupled to 3-(chloromethyl)-N-[2-
(hydroxymethyl)phenyl]benzamide in dimethylformamide
(DMF), based on Berto’s synthesis,44 to yield the alcohol N-
[2-(hydroxymethyl)phenyl]-3-(1H-imidazol-1-ylmethyl)-
benzamide. The use of ammonia solution rather than triethylamine
to basify the work-up solution produced the alcohol in good yield
and high purity. Removal of all traces of DMF was essential for the
oxidation of the alcohol to the aldehyde, N-(2-formylphenyl)-3-
(1H-imidazol-1-ylmethyl)benzamide, using manganese dioxide in
chloroform over a 20 h period. Both the alcohol and the
aldehyde were found to be crystalline and polymorphic solids;
we have reported their XRD structures.51
Figure 2. The basic scaffold of the chosen Co(III) corrole model for
H2OCbl+, [10-(2-[[4-(1H-imidazol-1-ylmethyl)benzoyl]amino]-
phenyl)-5,15-diphenylcorrolato]-cobalt(III), abbreviated as diphenyl-
tailed corrole Co(III), DPTC-Co. In an 80:20 MeOH/H2O solution,
the axial coordination site is occupied by a solvent molecule (see text),
and the compound is referred to as [H2O−DPTC-Co]. Future
elaboration of the two phenyl substituents with water-solubilizing
groups is envisaged.
anthracene systems. This suggested that the tail would be able
to rotate in solution and move into a position favorable for
coordination to cobalt. Moreover, there is literature precedence
for this substituent in both porphyrin40−45 (or bearing sub-
stituents on the imidazole moiety42,46−48) and corrole49 chemistry.
As the yields of corrole are usually quite low, our approach
entailed the preparation of intermediates bearing the elements
of the target molecule before cyclization to produce the final
corrole. Therefore, an aldehyde bearing the tail motif and a
dipyrromethane bearing a phenyl were prepared.
The initial idea was to prepare an imidazole acid chloride by
a route described by Collman50 (Supporting Information,
Scheme S1, Section S2.2); condensation with 2-aminobenzyl
alcohol and oxidation should have afforded the required tail-
bearing aldehyde. As described in the Supporting Information,
we were unable to prepare the acid chloride.
This aldehyde and 2,2-(phenylmethylene)bis(1H-pyrrole)
were used in Gryko’s methanol/water corrole synthesis28 to
produce the corrole 10-(2-[[4-(1H-imidazol-1-ylmethyl)-
benzoyl]amino]phenyl)-5,15-diphenylcorrole, referred to here
as diphenyl tailed corrole (DPTC), a very dark green/purple
solid that appears to be microcrystalline but did not yield
diffraction-quality crystals, despite many crystallization at-
tempts. The UV−vis spectrum of a 0.6 mM solution in hexane
shows a split Soret band at 414.5 nm (ε = 3.58 × 103 M−1
cm−1) and 435 nm and two bands in the visible region
(579, 610 nm). The molecular ion peak was observed at m/z =
726 ([M + H]+). The NH amide signal was observed at
1
9.04 ppm in the H NMR spectrum, and the amide carbonyl
signal was seen at 163.61 ppm in the 13C NMR spectrum. Signals
This difficulty was overcome by using 3-(chloromethyl)-
benzoyl chloride as starting material (Scheme 2). It was first
coupled to 2-aminobenzyl alcohol in dichloromethane in the
from the methylene group were present in both the 1H and 13
C
NMR spectra at 3.09 and 21.03 ppm, respectively. The typical
a
Scheme 2. Synthesis of the Tailed Aldehyde from 3-(Chloromethyl)benzoyl Chloride
a
(i) Dry CH2Cl2, triethylamine, 0°C, 10 min, then room temperature, 2 h; (ii) DMF, 2 h; (iii) MnO2, CHCl3, 20 h; (iv) as in (ii) but with reaction
time >3 h.
4421
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
corrole NMR signals of the β protons were present at 8.85,
8.83, 8.46, and 8.40 ppm.
In the calculated spectrum of [H2O−DCTC-Co] in which H2O
occupies the sixth axial coordination site, the two most intense
transitions in the Soret region are red-shifted and slightly fur-
ther apart (8 nm). The Q band transition is also red-shifted
(Supporting Information, Figure S6). The transitions undergo a
further shift to the red in [CN−DCTC-Co]−; the two principal
Soret transitions are now quite far apart (25 nm) and might
present in the experimental spectrum as a split Soret band.
These calculations afford a rationalization of the experimental
observations shown in Figure 3. The more intense, higher-
energy Soret band of DPTC-Co in DCM is consistent with a
five-coordinate species. In the presence of water, the Soret
broadens and shifts to the red, consistent with [H2O−DPTC-Co]
being the predominant species in solution. (In dried methanol,
there is a band at 386 nm and a band at 436 nm, with a
shoulder at 453 nm; see Figure S7 of the Supporting
Information. This suggests a mixture of five- and six-coordinate
species in solution.) In acetonitrile, the intensity at 390 nm and
the split Soret at 442 and 455 nm are also indicative of a
mixture of five- and six-coordinate species in solution. We shall
use the nomenclature DPTC-Co to refer to the cobalt corrin in
the solid state or in solvents where it exists either as a five-
coordinate species or as a mixture of five- and six-coordinate
species. In 80:20 MeOH/aqueous buffer solutions that have
been allowed to equilibrate (see above), the predominant
species is six-coordinate, and we shall refer to it as [H2O−
DPTC-Co].
Refluxing DPTC in methanol with excess cobaltous acetate
and sodium acetate for 2 h produced the Co(III) corrole, [10-
(2-[[4-(1H-imidazol-1-ylmethyl)benzoyl]amino]phenyl)-5,15-
diphenylcorrolato]cobalt(III), DPTC-Co (m/z = 782.4
([M + H]+)). Metalation resulted in a general upfield shift of
1
the H NMR signals as a result of the shielding effect of low
spin Co(III);39 the β protons shifted from 8.85, 8.83, 8.46, and
8.40 ppm in the metal-free corrole to 8.80, 8.40 (integrating for
four β protons), and 8.19 ppm in the metalated species.
2.2. The Solution Chemistry of DPTC-Co. 2.2.1. UV−Vis
Spectra. The UV−vis spectrum of DPTC-Co depends on the
nature of the solvent. Representative spectra in dichloro-
methane, in acetonitrile, and in aqueous methanol are shown in
Figure 3. The spectra in other solvents are given in Figure S4 of
the Supporting Information.
2.2.2. Behavior in Aqueous Methanol Solution. The UV−
vis spectrum of [H2O−DPTC-Co] in 80:20 MeOH/H2O
changes with pH, as shown in Figure 4. This is attributed to
Figure 3. The UV−vis spectra of ca. 50 μM DPTC-Co in
dichloromethane (red), in acetonitrile (green), and in neutral 80:20
methanol/H2O (blue) after equilibration.
The spectrum of DPTC-Co in noncoordinating solvents
such as dichloromethane, hexane, and ethyl acetate has a
prominent Soret band around 390 nm, with a shoulder around
425 nm, and a broad Q band around 555 nm. The intensity of
the Soret band is solvent-dependent. Coordinating solvents
including acetone, methanol, acetonitrile, and tetrahydrofuran
(THF) cause the Soret band to split and move to around
450 nm. In many cases (Figure 3) significant absorbance at
around 390 nm remains. The broad Q band around 550 nm
becomes a series of bands, with the most prominent being the
longest wavelength band between 600 and 650 nm (Figure 3
and Supporting Information, Figure S4). When the compound
was dissolved in 80:20 methanol/water mixtures, the intensity
of its Soret at 390 nm decreased and moved to 440 nm, with a
half-life of approximately 10 min at 25 °C.
To rationalize these observations, time-dependent (TD)
DFT calculations were carried out on a model of DPTC-Co
(Figure S5 of the Supporting Information). According to this
model, two higher-intensity transitions and several lower-
intensity transitions constitute the Soret region (Figure S6 of
the Supporting Information); there is one principal transition in
the visible region which makes up the Q band, and there are
multiple transitions in the UV region. (This is discussed more
fully in Section S2.5 of the Supporting Information). In the case
of five-coordinate DPTC-Co, the two higher-intensity tran-
sitions in the Soret region are close together (separated by
7 nm); these will clearly overlap in the experimental spectrum
and present as a fairly intense band in the near-UV region.
Figure 4. The dependence of the UV−vis spectrum of [H2O−DPTC-
Co] on pH in 80:20 MeOH/H2O; pH 6.9 (blue) and pH 10.5 (red).
ionization of coordinated H2O to form [OH−DPTC-Co]−.
The absorbance at the Soret maximum and the Q-band max-
imum at pH 9.5 were determined as a function of [H2O−
DPTC-Co] up to concentrations of 60 μM; Beer’s law is strictly
obeyed, which is a strong indication52−55 that the species is
monomeric in solution at least up until these concentrations.
The pKa for the ionization of coordinated H2O in [H2O−
DPTC-Co] was determined by titrating between pH 6.9 and
11.7 in an 80:20 MeOH/buffer solution, where the buffer
consisted of 1 mM each of 3-morpholinopropane-1-sulfonic
acid (MOPS), 2-amino-2-hydroxymethylpropane-1,3-diol
(TRIS), and 2-(cyclohexylamino)ethanesulfonic acid (CHES).
The ionic strength (μ) of the solution was very low, on average
0.8 mM across the pH range studied. The spectroscopic
4422
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
changes (Figure S8 of the Supporting Information) are
consistent with a single pKa = 9.76 0.06.
dependent on pH, but become pH-independent, 3.02(10),
once they are corrected for the pKa of [H2O−DPTC-Co].
If the log K values were too large for direct spectrophoto-
metric determination, they were determined in a competition
experiment. For example, log K for binding of trimethylphos-
phite (TMP) by [H2O−DPTC-Co] was determined by titrating
aliquots of this ligand into a solution of the DMAP complex of the
corrole. Log K(TMP) could then be determined from the pre-
viously measured value of log K(DMAP) and the observed log K
from the titration since log K(obs) = log K(DMAP) + log
K(TMP). The values of log K obtained are summarized in Table 3.
The pKa of H2O in H2OCbl+ has been reported by several
workers,5,6,9,56−62 with values varying between 7.561 and 8.1.6
This variability is probably because of differences in ionic
strength and, in some cases, the unfortunate use of chloride
salts to adjust ionic strength (chloride will coordinate Co(III)
in H2OCbl+, albeit weakly16,63). We determined the pKa of
H2OCbl+ using the same buffer system as for [H2O−DPTC-Co],
namely, μ = 0.8 mM in 80:20 MeOH/H2O, as well as in aqueous
solution. In aqueous and aqueous methanol solutions, we found
pKa = 7.462(7) and 7.40(2), respectively (Supporting Information,
Figure S9).
3. DISCUSSION
The pKa values of coordinated H2O in H2OCbl+ and DPTC-Co
in the same solvent and ionic strength (80:20 MeOH/H2O,
μ = 0.8 mM), namely, 7.462(7) and 9.76(6), respectively, are
indicative of the somewhat different nature of the Co−O bond
in these two compounds. We have previously shown that
replacing the C10 H with electron-donating Cl in H2O(10Cl-
Cbl)+ decreases the pKa of coordinated H2O from 8.09
(aq solution, μ = 0.5 M, NaNO3) to 7.65,9 while replacing it
with electron-withdrawing NO increases the pKa to 10.71 (aq
solution, μ = 2.2 M, NaClO4). We used semiempirical MO
calculations to rationalize this observation, showing that the
Co−O bond in H2O(10Cl-Cbl)+ and H2O(10NO-Cbl)+ is more
covalent and more ionic, respectively, than it is in H2OCbl+ itself.
A more ionic Co−O bond makes the lone pair on coordinate
OH− more readily available, resulting in its protonation at a
higher pH of the solution. We find a similar trend in our models
of H2OCbl+ and [H2O−DPTC-Co], on which we performed
DFT calculations. The value of the electron density at the bond
critical point ρb is a measure of the strength of the bond.73−77
The ratio |Vb|/Gb, where Vb and Gb are the potential and kinetic
energy densities at the bond critical point of a chemical bond, is
useful to characterize the nature of the bond:78 |Vb|/Gb < 1 is char-
acteristic of an ionic bond, and |Vb|/Gb > 2 is characteristic of a co-
valent bond; a bond of intermediate character has 1 < |Vb|/Gb < 2.
Moreover, for a metal−ligand interaction, ρ is usually close to zero
and the Laplacian at the bond critical point, ∇2ρ > 0. The Co−O
bond in H2OCbl+ is shorter (2.107 Å), stronger (ρb = 0.0551 au,
where 1 au of ρb = 6.7483 e Å−3, ∇2ρ = 0.2968 au, where 1 au of
∇2ρ = 24.099 e Å−5), and more covalent (|Vb|/Gb = 1.0873) than
that in [H2O−DPTC-Co] (2.126 Å; 0.0508 au; 0.3785 au; 1.0695).
Therefore, the more ionic Co−O bond in [H2O−DPTC-Co] is
associated with a higher pKa value for coordinated H2O.
The solvent can have a significant effect on the value of log K
(Table 1). The log K values for neutral ligands tend to be
higher in aqueous solution than in 80% MeOH, but the
converse is usually true for anionic ligands (Table 4). Others
have reported similar observations. Balt and co-workers found
that in acetonitrile−water mixtures the equilibrium constants
for the binding of thiocyanate to H2OCbl+ increased from log
K = 3.03 to 4.17 as the fraction of acetonitrile increased from 0
to 80%;79 Moreno-Esparza and co-workers reported that log K
for the binding of iodide increased from 1.41 to 3.95 as the
methanol fraction increased from 0 to 100%.72 The effect is
presumably a consequence of differential solvation of the
entering ligand. This should manifest itself as a significant
dependence of log K on the ΔS of the reaction and will require
a study of the effect of temperature and solvent composition on
log K; this is beyond the scope of the present work.
2.3. Equilibrium Constants. While many equilibrium
constants (log K) for substitution of coordinated H2O in
H2OCbl+ are available (for example, reference 64) these have
all been determined in aqueous solution. To determine whether
solution composition materially affected the results, we studied
the substitution of coordinated H2O by N-methylimidazole in
very low ionic strength solutions with between 0 and 80% (v/v)
MeOH (50 mM MOPS). The log K values, corrected for the
effect of pH (eq 5, vide infra), are summarized in Table 1.
Table 1. Effect of Solvent Composition on the log K Value
for Substitution of H2O in H2OCbl+ by N-Methylimidazole
(25 °C)
a
% MeOH (v/v)
polarity (ET(30))65,66
log K
μ (M)
0
20
40
60
80
63.1
61.0
59.3
57.3
56.3
4.32(4)
4.32(4)
4.21(5)
4.04(6)
3.99(5)
0.019
0.021
0.019
0.022
0.022
a
Corrected for pH (eq 5).
The composition of the solvent has a small but significant
effect on the values of log K. A similar effect, for the substitu-
tion of H2O by pyridine in aquacyanocobyrinic acid, has been
reported by Hamza.67 We therefore determined all log K values
for both [H2O−DPTC-Co] and H2OCbl+ under the same
conditions (25 °C, 80% MeOH, 20% CAPS, CHES, MOPS, or
MES buffer, v/v, μ ≈ 0.02 M).
Figure S10 of the Supporting Information shows, as an
example, the spectroscopic changes associated with the sub-
stitution of H2O in [H2O−DPTC-Co] by dimethylaminopyr-
idine (DMAP), and Supporting Information, Figure S11 shows
−
the spectroscopic changes where the incoming ligand is N3 .
To verify the validity of the correction for the pKa of [H2O−
DPTC-Co] (eq 5), we determined log K values for
−
coordination of NO2 at pH values above and below the pKa
value of [H2O−DPTC-Co], which is 9.76(6)). The results are
listed in Table 2. The observed log K values are clearly
Table 2. Effect of pH on the log K Value for Substitution of
−
H2O in [H2O−DPTC-Co] by NO2 (25 °C)
a
pH
log Kobs
log K
10.91
10.02
9.51
1.77(12)
2.60(7)
2.86(6)
2.97(17)
2.95(12)
3.05(7)
3.06(6)
3.01(17)
8.78
In Figure 5 the log K value for coordination of a ligand by
[H2O−DPTC-Co] is plotted against log K for its coordination
a
Corrected for the pKa of DPTC-Co (eq 5)
4423
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
Table 3. Stability Constants for the Coordination of Ligands by DPTC-Co and H2OCbl+ (80% MeOH, 25 °C, μ ≈ 0.02 M)
H2OCbl+
log K
[H2O−DPTC-Co]
a
a
b
ligand
pH
log K (aq sol)
Neutral Ligands
pH
log K
ligand stoichiometry
ethanolamine
pyridine
8.12
7.20
9.10
7.07
9.14
7.08
7.07
3.60(5)
0.48(4)
3.99(2)
3.99(5)
1.0(2)
9.50
9.16
9.51
9.46
9.52
9.47
9.44
9.50
5.15(2)
4.11(2)
5.32(2)
5.41(4)
1.09(4)
2.1(1)
1.19(2)
1.04(1)
1.10(1)
1.2310
4.4468
dimethylaminopyridine (DMAP)
N-methylimidazole
thiophene
thiourea
1.76(2)
4.77(7)
1.169
5.070
c
trimethylphosphite (TMP)
trimethylphosphite (TMP)
11.8(1)
10.4(1)
d
Anionic Ligands
cysteine
cyanide
6.01
7.09
7.07
7.12
7.11
7.12
6.99
6.5(1)
6.071
9.87
9.59
9.51
8.78
9.51
9.50
9.63
5.7(2)
1.07(4)
e
f
12.53(7)
>1260
13.1(7)
13.2(6)
3.0(1)
g
h
11.9(5)
nitrite
6.7(4)
5.3410
4.8510
3.0310
1.410
0.96(2)
0.98(2)
azide
6.35(5)
4.73(4)
3.96(2)
2.82(2)
1.31(5)
2.33(3)
thiocyanate
iodide
a
b
c
The pH at which log K was determined; corrected log K values (eq 5) are reported in this table. See Experimental Section. Competition
d
e
f
experiment with DMAP. Competition experiment with pyridine. Competition experiment with azide. Competition experiment with DMAP, in
cell. Competition experiment with nitrite. Competition experiment with DMAP, out of cell.
g
h
by H2OCbl+. The log K values are reasonably correlated (i.e.,
the general behavior of Co(III) toward the range of ligands
studied is similar whether the equatorial ligand is corrin or
corrole). Pyridine is an exception: it is poorly coordinated by
H2OCbl+, and log K is anomalously low (0.48(4)); however,
log K for coordination of DMAP (3.99(2)) is significantly
higher. In the absence of a crystal structure, we used DFT
calculations to rationalize this observation. As shown in Figure
S12 of the Supporting Information, the sentinel methyl groups
C54 and C46 and the methylenes C26 and C37 of the a and c
acetamide side chains on the sterically crowded β face confine
pyridines to the C5C15 vector, and there is close contact
between the C2 and C5 hydrogens on pyridine and C5 and
C15 of the corrin, respectively. This results in a relatively long
Co−Npy bond (2.045 Å). The presence of electron-donating
NMe2 in the para position decreases the bond length to Co(III)
(2.027 Å); the corrin ring is flattened, and the nonbonded
contacts become marginally longer. In DMP, therefore, the
electronic influence of the para substituent overrides the steric
constraints of the crowded β face of the corrin.
The ligands fall into two classes. [H2O−DPTC-Co] has a
higher affinity for neutral ligands than H2OCbl+, but the
converse is true for anionic ligands. CN− is an exception, but it
must be appreciated that the very large values of log K for the
coordination of TMP by [H2O−DPTC-Co] and for CN− by
both [H2O−DPTC-Co] and H2OCbl+ makes it difficult to
obtain accurate values (Table 4). [H2O−DPTC-Co] is neutral,
whereas H2OCbl+ has a +2 charge at the metal center (if it is
assumed that the charge on phosphate is too remote to affect
matters). This therefore suggests that an electrostatic
interaction between the metal center and an incoming anionic
ligand is a significant contributor to the affinity of H2OCbl+
compared to [H2O−DPTC-Co] for anionic ligands. This
observation is in agreement with our recent observations in a
comparison in aqueous solution of the coordination chemistry
of Co(III) in H2OCbl+ and in a Co(III) analogue in which
corrin was replaced by a porphyrin;23 we found that H2OCbl+,
with its +2 charge at the metal center, binds anionic ligands
with a greater affinity than the porphyrin analogue (the charge
at the metal center of a Co(III) porphyrin is +1). Conversely,
the Co(III) porphyrin tended to bind neutral ligands with a
higher affinity than the Co(III) corrin.
A quantum theory of atoms in molecules (QTAIM)
analysis80 of the topological properties of the electron density
at the bond critical points is a useful way of exploring the nature
of the bonding in transition-metal complexes.81−86 To explore
this in these Co(III) corrole and corrin complexes we chose
pyridine, imidazole, and TMP as representative neutral ligands;
CN−, NO2 , and CH3S− as representative anionic ligands; and
−
H2O to model the starting complexes of the ligand-substitution
reactions under consideration. The data are summarized in
Table 4.
As expected,73−77 there is a strong correlation between ρb
and the bond length between Co(III) and the axial ligands
(Supporting Information, Figure S13). In the aqua complex of
both corrole and corrin, the bond to axial H2O is long and
weak, which is clearly the reason why it is readily displaced by
other ligands. There is a normal trans influence in both
systems; that is, there is an inverse correlation between the
length of the Co−L bond length and the length of the Co−Nax
band to the proximal imidazole ligand (Supporting Informa-
tion, Figure S14), although the two ligands with S and P donors
do not follow this trend.
With the exception of H2O, the bond lengths between
Co(III) and L are longer in the corrin complexes than they are
in the corrole complexes (i.e., Δrcorrin−corrole > 0, and on average
ρL−Co (corrin) < ρL−Co(corrole)); however, the difference is greater
for the neutral ligands (Δrcorrin−corrole = 0.038(15) Å) than
it is for the anionic ligands (Δrcorrin−corrole = 0.013(7) Å), see
Table S2 of the Supporting Information. This is precisely what
we found in our recent comparison of the coordination
chemistry of Co(III) corrins and porphyrins23 and suggests that
the electrostatic attraction between an anionic ligand and the
residual +2 charge at the metal center of a corrin is one of the
reasons why the stability of the L−Co−Cbl+ when L is an anion
is increased relative to that of the corrole complex.
4424
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
Figure 5. Correlation of log K for coordination of neutral (blue
circles) and anionic (red squares) by H2OCbl+ and [H2O−DPTC-Co]
in 80:20 MeOH/H2O at 25 °C. The dotted line is the equivalue line.
Abbreviations: tu = thiourea; tp = thiophene; TMP = trimethylphos-
phite; DMAP = N,N-dimethylaminopyridine; NMeIm = N-methyl-
imidazole; Py = pyridine.
The data in Table 4 show that the metal−ligand bonds in
both the corrole and corrin systems are predominantly ionic
(ρ close to zero, ∇2ρ > 0, and |V|/G close to 1), although there
is increasing covalent character (assessed by |V|/G and by the
delocalization index DI, a measure of the average number of
electrons shared between Co and the donor atom of the ligand)
as the richness of electron density on the ligand increases
−
(H2O < imidazole < pyridine < NO2 < CN− < TMP <
CH3S−). CN− (corrole, corrin) and TMP (corrole) have very
high log K values for coordination to Co(III), and the bond
between their donor atom and the metal has considerable
covalent character. Yet cysteine (modeled by CH3S− in our
DFT calculations) also has a Co−S bond with very significant
covalent character but considerably smaller log K in both sys-
tems. Thus, while there may be some dependence of log K on
the metal−ligand bond, that dependence is weak (Figure S15 of
the Supporting Information, r2 = 0.32, p = 5.3%).
The ellipticity of a bond (ε = |λ1|/|λ2| − 1, where λ1 and λ2
are the eigenvalues of the Hessian matrix of the electron density
at the bond critical point (bcp) along mutually perpendicular
axes, which are themselves perpendicular to the bond path) is a
function of the ratio of the rate of electron density decrease in
the two directions perpendicular to the bond path at the bcp.80
If there is significant π bonding between two atoms then |λ1| >
|λ2| and ε will be significantly greater than 0. For example, in
our model of the imidazole complex of the Co(III) corrin, the
C5−C6 bond, part of the delocalized system of the corrins,
presents with λ1 = −0.6720 au and λ2 = −0.5299 au such that
ε = 0.2727. By contrast, for the C7−C8 single bond, λ1 =
−0.4339 au, λ2 = −0.4334 au, and ε = 0.0010. The values of
λ1, λ2, and λ3 for all Co−L bonds modeled in this work are
summarized in Table S3 of the Supporting Information.
When compared to the other Co−L bonds, we find that
−
the bond between NO2 and Co(III) in both the corrole and
the corrin system has significant ellipticity; this is indicative of
the participation of a frontier π* orbital in the bonding with the
4425
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
Low sample concentrations (1 mg mL−1) had to be used; at higher
concentrations, broad signals with no multiplet resolution were observed. It
is thought that this arose from aggregation of macrocycles in solution due to
significant π···π interaction. The available solid-state structures of corroles
related to DPTC show such interactions in the solid state.39
metal. In the case of CN−, while ε is small, this masks the
relatively large values of both λ1 and λ2, suggesting that both
frontier π* orbitals are involved in the bonding. We found
similar effects in the bonding of NO2− and CN− with Co(III) in
a Co(III) porphyrin.23
Determination of the Molar Absorptivity of DPTC-Co. The
absence of a crystal structure of DPTC-Co required an indirect
determination of the molar absorptivity ε, which was required for
determining equilibrium constants (see below, eq 4), since the solid
state could well include solvent molecules as found for example in all
crystal structures of the cobalt−corrins. A series of cobalt standards
was prepared using dicyanocobalamin, prepared by treating a solution
of aquacobalamin with excess cyanide and waiting until there was no
further change in the UV−vis spectrum. Using the well-established
value of ε367 = 3.04 × 104 M−1 cm−1,89 the concentration of the
solutions could be determined. The concentration of Co was
determined by atomic absorption spectroscopy on a PG-900 atomic
absorption spectrophotometer using a graphite furnace fitted with a
pyrolytically coated HGA-76 graphite tube and a Photron hollow-
cathode lamp single element cobalt lamp. The absorption line
characteristic of Co at 240.7 nm was integrated using the AAWin
software suite.90 Nonspecific absorption of radiation was corrected for
by subtracting the absorbance recorded from a deuterium lamp.91 We
obtained a linear response up to a concentration of 0.2 ppm of Co.
Samples of DPTC-Co in dichloromethane were prepared from a
sample of DPTC-Co dried on a high vacuum line, which was weighed
and diluted to give an approximate Co concentration of 0.1 ppm. The
relative standard deviation of the measurements of the samples
4. CONCLUSION
We report the synthesis and characterization of a Co(III)
corrole that places the metal ion in a similar coordination
sphere as in aquacoabalamin, H2OCbl+; Co(III) is coordinated
in the proximal coordination site by imidazole (cf. 5,6-
dimethylbenzimidazole in H2OCbl+) and has a weakly bound
H2O ligand in the sixth coordination site that is readily replaced
by exogenous ligands. Equilibrium constants (log K) values for
replacement of H2O by a variety of neutral and anionic ligands
were determined in aqueous methanol solutions under
conditions where the Co(III) corrole is monomeric. We find
that Co(III) behaves similarly toward ligands whether it is in a
corrin or a corrole. However, based on DFT modeling, bonds
between it and axial ligands are longer in a corrin than in a
corrole, explaining why in general log K values for coordination
of neutral ligands are larger in a Co(III) corrole than they are in
a Co(III) corrin. If L is an anion, then the difference in bond
lengths becomes significantly smaller, and we suggest that an
electrostatic interaction between the ligand and the residual +2
charge at the metal center in the corrin enhances the affinity of
Co(III) for the ligand such that log K values for coordination of
anions is larger in the corrin system than it is in the corrole
system. This work highlights how the nature of the equatorial
ligand in Co(III) chemistry can significantly modify the chem-
istry of the axial coordination site of these compounds and
provides an indirect answer to the otherwise surprising fact that
normally inert Co(III) is used as coenzyme in biology. The
influence this has on the kinetics of ligand substitution has also
been investigated and will be reported elsewhere.
and standards was <2%. We found that in dichloromethane ε389
=
5.14 × 104 M−1 cm−1 for DPTC-Co. Values of ε in other solvents (and
hence the concentration of DPTC-Co) could be determined by
comparing the spectra of aliquots of a stock solution of DPTC-Co in
dichloromethane and in the solvent of interest.
Equilibrium Constants. The equilibrium constants, K, for the
coordination of the ligands by [H2O−DPTC-Co] and aquacobalamin
(vitamin B12a, H2OCbl+), eq 1, where coordinated H2O (or possibly
MeOH) is displaced by the exogenous ligand L, Nax is the trans N-
donor ligand (imidazole in DPTC-Co and 5,6-dimethylbenzimidazole
in H2OCbl+), N4 represented the equatorial macrocycle corrole or
corrin, and the overall charge is omitted for convenience, were deter-
mined by addition of aliquots of a stock solution of the appropriate
ligand to a 20−50 μM solution of the Co(III) complex contained in a
1.00 cm path length cuvette housed in the thermostated cell block of a
UV−vis spectrophotometer. The solvent was 80:20 MeOH/H2O and
0.05 M in the appropriate buffer (CAPS, for pH > 10; CHES, for pH
8.5−10; MOPS for pH 7−8; MES for pH 6−7). Typically between 12
and 18 aliquot additions were made in each titration, and absorbance
readings were corrected for dilution. For cases where log K is relatively
small (log K < 4), the absorbance data at wavelength λ were fitted
using nonlinear least-squares methods to a simple binding isotherm
(eq 3) as objective function, with A0, A1, and K as the parameters to be
optimized. A0 and A1 are the absorbance values at λ corresponding to 0
and 100% complex formation, respectively.
5. EXPERIMENTAL SECTION
Experimental details of materials and methods used in the synthesis of
DPTC-Co are given in the Supporting Information (Section S2).
Deionized water, purified using a Direct Q UV 3 Millipore system and
further purified using a Millipore Milli-Q unit (>18 mΩ), was used
throughout this study. Hydroxocobalamin was from Roussel. pH
measurements were performed using a Metrohm High-precision
780 pH meter and a Metrohm Unitrode combination pH electrode.
Metrohm ready-to-use buffer solutions at pH 4, 7, and 9 were used to
calibrate the electrode. The pH readings in methanol/water mixtures
are uncorrected. UV−vis spectra were recorded using 1.00 cm path
length cells on either a Cary 1E UV−vis or a Cary 300 Bio UV−vis
Spectrophotometer, fitted with a Cary Dual Cell Peltier accessory from
Varian, set at 25.0 °C, or a Cary 3E UV−vis spectrophotometer, fitted
with a circulating water bath maintained at 25.0 0.1 °C. Spectra were
recorded with a spectral bandwidth of 2 nm, signal averaging time of
0.5 s, data interval of 1.00 nm, and scan rate of 120 nm min−1. Ligand-
binding isotherms were fitted with standard nonlinear least-squares
methods using SigmaPlot.87
Proton NMR data were recorded at 300.13 MHz, and carbon
NMR data were recorded at 75.47 MHz on a Bruker AVANCE
300 spectrometer. Spectra of corrole macrocycles were recorded on a
Bruker ULTRASHIELD 500 PLUS spectrophotometer where proton
NMR data were recorded at 500.13 MHz and carbon NMR data were
recorded at 125.77 MHz. All samples were dissolved in CDCl3.
Chemical shifts are reported on the δ scale (parts per million (ppm))
relative to the internal standard of 0.03% tetramethylsilane (1H) and
the central CDCl3 line at 77.00 ppm (13C). NMR spectroscopic data
were processed using the MestReNova software suite of programs.88
[Nax − Co(N ) − OH2] + L ⇌ [Nax − Co(N ) − L] + H2O
4
4
(2)
Aλ = (A0 + KA1[L])/(1 + K[L])
(3)
When log K is relatively large (>4), the implicit assumption in eq 3
that [L]free ≈ [L]total is not true because a significant fraction of the
added ligand will be complexed to the metal ion. In this case [L] in
eq 3 has to be replaced by an explicit expression for [L]free. It can be
shown92 that [L]free is given by eq 4, where only one root has physical
meaning. In eq 4, [M]total is the total metal ion concentration.
−a2
a22 − 4a1a3
2a1
[L]free
=
a1 = K; a2 = 1 + K[M]total − K[L]total ; a3 = −[L]total
(4)
4426
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
It is well-established in cobalt−corrin chemistry that hydroxide in
hydroxocobalamin will not be displaced by an exogenous ligand L.5,93,94
Moreover, it is usually the case that the protonated form of a ligand
(for example, imidazole) will not coordinate to a metal ion.
Conditional stability constants Kobs, determined at some pH value,
were therefore converted to pH-independent equilibrium constants K,
using eq 5, where pKaL refers to the acid dissociation constant of the
conjugate acid of ligand L and the pKaM value, to the acid dissociation
constant of coordinated H2O in DPTC-Co and H2OCbl+.
Co(III) corrole modeled by molecular mechanics methods as
potential targets (Figure S3); UV−vis spectra of DPTC-Co in a
variety of solvents (Figure S4); representation of the simplified
model of DPTC-Co used for DFT calculations (Figure S5);
CAM-B3LYP/TZVP TD-DFT spectra of five-coordinate
DPTC-Co, [H2O−DPTCCo], and [CN−DPTCCo]− (Figure S6);
UV−vis spectrum of DPTC-Co in dried methanol and in 80:20
MeOH/H2O (Figure S7); dependence of the absorbance at
621 nm of [H2O−DPTC-Co] on pH in 80:20 MeOH/H2O
(Figure S8); dependence of the absorbance at 350 nm of
H2OCbl+ on pH in aqueous solution and in 80:20 MeOH/H2O
(Figure S9); spectroscopic changes observed on titrating
[H2O−DPTC-Co] in 80% MeO/H2O with dimethyalminopyr-
K = Kobs(1 + 10pK −pH)(1 + 10pH−pK
)
aL
aM
(5)
The spectroscopic changes were recorded between 300 and 800 nm;
typically between 4 and 6 wavelengths where absorbance changes were
largest were fitted to eq 3 or eq 4, as appropriate. A weighted average
(with the weighting as the reciprocal of the relative percentage error of
the fit) of the results obtained at the monitoring wavelengths was
calculated.
The ligand stoichiometry associated with some titrations of
[H2O−DPTC-Co] was determined from the slope of a Hill plot, a
plot of log [L] against log ((A0 − A)/(A − A1)), where A is the
absorbance associated with each concentration of the incoming ligand
L; the other variables have been defined (vide supra).
Molecular Mechanics. The Generalized AMBER Force Field
(GAFF)95 as implemented in HyperChem 7,96 and expanded for the
cobalt−corrins,97 was used for exploring potential Co(III) corrole
models of H2OCbl+. The force field was augmented for modeling
corroles as follows. A search of the Cambridge Structural Database98
using ConQuest99 for Co(III) corroles yielded 10 structures, for which
eight new atom types were defined as illustrated in Figure S2 of the
Supporting Information. Initial values of both lengths, bond angles,
torsional angles, and the bond stretch and angle-bending force
constants involving the metal ion were arrived at by analogy with the
cobalt−corrins force-field values97 and informed by the crystallo-
graphic data. Parameters for the pyrrolic and meso carbon atom were
based on standard GAFF values for organic systems. These initial
parameters were then used to model Co(III) triphenylcorrole. All
bond lengths, angles, and torsions were measured and compared to the
crystallographic data. If the parameters produced structural metrics within
one standard deviation of the mean of the 10 crystallographic structures,
they were accepted. Otherwise they were adjusted on a trial-and-error
basis until this level of agreement was reached. The final parameters are
listed in Table S1 of the Supporting Information. Geometry optimization
−
idine (Figure S10) and with N3 (Figure S11); depiction of
DFT-optimized B12 complexes with pyridine and dimethylami-
nopyridine (Figure S12); correlation in DPTC-Co and Co(III)
corrin complexes between the axial bond length to the proximal
imidazole ligand and the trans ligand L, and the electron
density at the Co−imidazole and Co−L bond critical points
(Figure S13); dependence of the trans Co−Nax bond length to
imidazole in DFT models of [L−DPTC-Co] and)
[L−Co(corrin)−Im] on the Co−L bond length (Figure S14);
correlation between |V|/G at the bond critical point of the
Co−ligand bond in DFT models (BP86/TZVP) of Co(III) corrole
and Co(III) corrin complexes, and the experimentally determined
log K values for substitution of axial H2O (Figure S15); a brief
review of the ligand-substitution reactions of Co(III) and Mn(III)
corroles (Section S2.1); account of the attempted synthesis of an
aldehyde bearing the tail motif (Section S2.2); detailed description
of the synthesis of DPTC-Co (Section S2.3); crystallographic
details of molecular structures reported in this work (Section S2.4);
account of the TD-DFT calculations (Section S2.5); tables of new
AMBER parameters for modeling Co(III) corroles (Table S3.1);
comparison of axial bond length distances in DFT (BP86/TZVP)
models of [L−Co(N4)−Nax] complexes (Table S3.2); and
eigenvalues of the Hessian of the electron density at the bond
critical point of Co−L bonds, and bond ellipticity, in DFT Models
(BP86/TZVP) in corrole and corrin complexes (Table S3.3); 37
pages. This material is available free of charge via the Internet at
http://pubs.acs.org.
́
was carried out using the Polak−Ribiere conjugate gradient algorithm
with a convergence criterion of 0.05 kcal Å−1 mol−1. Single-point energy
calculations on the molecular-mechanics optimized geometry were carried
out using the semiempirical Zindo method.100
AUTHOR INFORMATION
Corresponding Author
*E-mail: Helder.Marques@wits.ac.za. (H.M.M.)
■
DFT and QTAIM Calculations. DFT calculations were performed
on a simplified model of DPTC-Co (Figure S5 of the Supporting
Information) in vacuo. The tail was truncated at the amide moiety, and
imidazole was used as the proximal axial ligand. Calculations were also
carried out on a simplified model of H2OCbl+ with all substituents on
the corrin replaced by H, and the dimethylbenzimidazole base replace
by imidazole. Calculations were carried out using Gaussian-09,101 the
BP86 functional,102,103 and the TZVP basis set.104,105 In addition, to
help speed up the calculations we made use of the density-fitting
approximation,106,107 and the density-fitting TZVP basis set as
well.108,109 TD-DFT single-point excitation energies for the first 30
excited states were calculated using the hybrid density functional
CAM-B3LYP110 with TZVP basis set. The electron density at the
Co−L bond critical points was analyzed using Bader’s QTAIM80 as
implemented in AIMALL;111 for this purpose, the wave function files
were generated using Gaussian-09 for input into AIMALL.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The financial assistance of the Department of Science and
Technology, the National Research Foundation, Pretoria,
through SARChI, the South African Research Chairs Initiative,
the University of the Witwatersrand, Johannesburg, and the
Andrew P. Mellon Foundation is gratefully acknowledged.
REFERENCES
■
(1) Rutenberg, A. C.; Taube, H. J. Chem. Phys. 1952, 20, 825−826.
(2) Swift, T. J.; Connick, R. E. J. Chem. Phys. 1962, 37, 307−320.
(3) Fiat, D.; Connick, R. E. J. Am. Chem. Soc. 1968, 90, 608−615.
(4) Poon, C. K. Coord. Chem. Rev. 1973, 10, 1−35.
(5) Reenstra, W. W.; Jencks, W. P. J. Am. Chem. Soc. 1979, 101,
5780−5791.
ASSOCIATED CONTENT
* Supporting Information
■
S
Dependence of equilibrium constants of coordination by Co(III)
corroles on the pKa of the entering ligand (Figure S1);
new atom types defined for molecular mechanical
calculations of cobalt−corroles (Figure S2); a selection of
(6) Marques, H. M.; Brown, K. L.; Jacobsen, D. W. J. Biol. Chem.
1988, 263, 12378−12383.
4427
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
(7) Stochel, G.; van Eldik, R.; Kunkely, H.; Vogler, A. Inorg. Chem.
1989, 28, 4314−4318.
(47) Didier, A.; Michaudet, L.; Ricard, D.; Baveux-Chambenoît, V.;
Richard, P.; Boitrel, B. Eur. J. Org. Chem. 2001, 1917−1926.
(48) Collman, J. P.; Sunderland, C. J.; Boulatov, R. Inorg. Chem.
2002, 41, 2282−2291.
(8) Hamza, M. S. A.; Zou, X.; Brown, K. L.; van Eldik, R. Inorg. Chem.
2001, 40, 5440−5447.
(9) Brown, K. L.; Cheng, S.; Zou, X.; Zubkowski, J. D.; Valente, E. J.;
Knapton, L.; Marques, H. M. Inorg. Chem. 1997, 36, 3666−3675.
(10) Knapton, L.; Marques, H. M. Dalton Trans. 2005, 889−895.
(11) Perry, C. B.; Fernandes, M. A.; Brown, K. L.; Zou, X.; Valente,
E. J.; Marques, H. M. Eur. J. Inorg. Chem. 2003, 2095−2107.
(12) Chemaly, S. M.; Brown, K. L.; Fernandes, M. A.; Munro, O. Q.;
Grimmer, C.; Marques, H. M. Inorg. Chem. 2011, 50, 8700−8718.
(13) Perry, C. B.; Marques, H. M. S. Afr. J. Chem. 2005, 58, 9−15.
(14) Navizet, I.; Perry, C. B.; Govender, P. P.; Marques, H. M. J. Phys.
Chem. B 2012, 116, 8836−8846.
(49) Collman, J. P.; Decrea
́
u, R. A. Org. Lett. 2005, 7, 975−978.
(50) Collman, J. P.; Broring, M.; Fu, L.; Rapta, M.; Schwenninger, R.
̈
J. Org. Chem. 1998, 63, 8084−8085.
(51) Zipp, C. F.; Dirr, H. W.; Fernandes, M. A.; Marques, H. M.;
Michael, J. P. Cryst. Growth Des. 2013, 13, 3463−3474.
(52) White, W. I. In The Porphyrins; Dolphin, D., Ed.; Academic
Press: New York, 1978; Vol. V.
(53) Pasternack, R. F.; Huber, P. R.; Boyd, P.; Engasser, G.;
Francesconi, L.; Gibbs, E.; Fasella, P.; Venturo, G. C.; de Hinds, L. C.
J. Am. Chem. Soc. 1972, 94, 4511−4517.
(15) Govender, P. P.; Navizet, I.; Perry, C. B.; Marques, H. M. Chem.
Phys. Lett. 2012, 550, 150−155.
(54) Aron, J.; Baldwin, D. A.; Marques, H. M.; Pratt, J. M.; Adams, P.
A. J. Inorg. Biochem. 1986, 27, 227−243.
(16) Marques, H. M.; Knapton, L. J. Chem. Soc., Dalton Trans. 1997,
3827−3833.
(55) Munro, O. Q.; Marques, H. M. Inorg. Chem. 1996, 35, 3752−
3767.
(17) Marques, H. M.; Knapton, L.; Zou, X.; Brown, K. L. J. Chem.
Soc., Dalton Trans. 2002, 3195−3200.
(56) Betterton, E. A. Ph.D. Thesis, University of the Witwatersrand:
Johannesburg, 1982.
(57) Hanania, G. I. H.; Irvine, D. H. J. Chem. Soc. 1964, 5694−5697.
(58) Randall, W. C.; Alberty, R. A. Biochemistry 1967, 6, 1520−1525.
(59) Rubinson, K. A.; Parekh, H. V.; Mark, H. B. Inorg. Chem. 1983,
22, 458−463.
(60) Hayward, G. C.; Hill, H. A. O.; Pratt, J. M.; Vanston, N. J.;
Williams, R. J. P. J. Chem. Soc. A 1965, 6485−6493.
(61) Smith, E. L.; Fantes, K. H.; Ball, S.; Waller, J. G.; Emery, W. B.;
Anslow, W. K.; Walker, A. D. Biochem. J. 1952, 52, 389−395.
(62) Fantes, K. H.; Page, J. E.; Parker, L. F. J.; Smith, E. L. Proc. R.
Soc. B 1950, 136, 592−609.
(63) Prinsloo, F. F.; Breet, E. L. J.; van Eldik, R. J. Chem. Soc., Dalton
Trans. 1995, 685−688.
(64) Pratt, J. M. The Inorganic Chemistry of Vitamin B12; Academic
Press: London, 1972.
(65) Reichardt, C.; Welton, T. Solvents and Solvent Effects in Organic
Chemistry, 4th ed.; Wiley Verlag Chemie: Weinheim, Germany, 2011.
(66) Smithrud, D. B.; Sanford, E. M.; Chao, I.; Ferguson, S. B.;
Carcanague, D. R.; Evanseck, J. D.; Houk, K. N.; Diederich, F. Pure
Appl. Chem. 1990, 62, 2227−2236.
(67) Hamza, M. S. A. J. Inorg. Biochem. 1998, 69, 269−274.
(68) Mathura, S. PhD Thesis, University of the Witwatersrand: 2013.
(69) Firth, R. A.; Hill, H. A. O.; Pratt, J. M.; Thorp, R. G.; Williams,
R. J. P. J. Chem. Soc. A 1969, 381−386.
(70) Chemaly, S. M. J. Inorg. Biochem. 1991, 44, 17−25.
(71) Hill, H. A. O.; Pratt, J. M.; Thorp, R. G.; Ward, B.; Williams, R.
J. P. Biochem. J. 1970, 120, 263−269.
(72) Moreno-Esparza, R.; Lopez, M.; Peannell, K. H. J. Chem. Soc.,
Dalton Trans. 1992, 1791−1795.
(18) Chemaly, S. M.; Florczak, M.; Dirr, H.; Marques, H. M. Inorg.
Chem. 2011, 50, 8719−8727.
(19) Chemaly, S. M.; Kendall, L.; Nowakowska, M.; Pon, D.; Perry,
C. B.; Marques, H. M. Inorg. Chem. 2013, 52, 1077−1083.
(20) Garau, G.; Geremia, S.; Marzilli, L. G.; Nardin, G.; Randaccio,
L.; Tauzher, G. Acta Crystallogr., Sect. B 2003, 51−59.
(21) Gross, Z. J. Biol. Inorg. Chem. 2001, 6, 733−738.
(22) Gryko, D. T. Eur. J. Org. Chem. 2002, 1735−1743.
(23) Mathura, S.; Sannasy, D.; de Sousa, A. S.; Perry, C. B.; Navizet,
I.; Marques, H. M. J. Inorg. Biochem. 2013, 123, 66−79.
(24) Aviv-Harel, I.; Gross, Z. Chem.Eur. J. 2009, 15, 8382−8394.
(25) Joseph, C. A.; Ford, P. C. J. Am. Chem. Soc. 2005, 127, 6737−
6743.
(26) Paolesse, R.; Jaquinod, L.; Nurco, D. J.; Mini, S.; Sagone, F.;
Boschi, T.; Smith, K. M. Chem. Commun. 1999, 1307−1308.
(27) Gross, Z.; Galili, N.; Saltsman, I. Angew. Chem., Int. Ed. 1999, 38,
1427−1429.
(28) Koszarna, B.; Gryko, D. T. J. Org. Chem. 2006, 71, 3707−3717.
(29) Paolesse, R.; Nardis, S.; Sagone, F.; Khoury, R. G. J. Org. Chem.
2001, 66, 550−556.
(30) Gross, Z.; Galili, N.; Simkhovich, L.; Saltsman, I.; Botoshansky,
M.; Blaser, D.; Boese, R.; Goldberg, I. Org. Lett. 1999, 1, 599−602.
(31) Gryko, D. T.; Jadach, K. J. Org. Chem. 2001, 66, 4267−4275.
(32) Roa, P. D.; Dhanalekshmi, S.; Little, B. J.; Lindsey, J. S. J. Org.
Chem. 2000, 65, 7323−7344.
(33) Lee, C.-H.; Lindsey, J. S. Tetrahedron 1994, 50, 11427−11440.
(34) Decrea
3327.
́
u, R. A.; Collman, J. P. Tetrahedron Lett. 2003, 44, 3323−
(35) Itahara, T. J. Chem. Soc., Chem. Commun. 1980, 49−50.
(36) Fu, L.; Gribble, G. W. Tetrahedron Lett. 2008, 49, 7352−7354.
(37) Erben, C.; Will, S.; Kadish, K. M. In The Porphyrin Handbook;
Kadish, K. M., Smith, K. M., Guilard, R., Eds.; Academic Press: New
York, 2000; Vol. 2, pp 233−300.
(73) Bader, R. F. W.; Matta, C. F.; Cortes
metallics 2004, 23, 6253−6263.
́ ́
-Guzman, F. Organo-
(74) Espinosa, E.; Souhassou, M.; Lachekar, H.; Lecomte, C. Acta
Crystallogr., Sect. B 1999, 55, 563−572.
(75) Howard, S. T.; Krygowski, T. M. Can. J. Chem. 1997, 75, 1174−
1181.
(38) Mahammed, A.; Gray, H. B.; Weaver, J. J.; Sorasaenee, K.;
Gross, Z. Bioconjugate Chem. 2004, 15, 738−746.
(76) Grabowski, S. J. J. Phys. Chem. A 2000, 105, 5551−5557.
(77) Vidal, I.; Melchor, S.; Alkorta, I.; Elguero, J.; Sundberg, M. R.;
Dobado, J. A. Organometallics 2006, 25, 5638−5647.
(78) Espinosa, E.; Alkorta, I.; Elguero, J.; Molins, E. J. Chem. Phys.
2002, 117, 5529−5542.
(79) Balt, S.; de Bolster, M. W. G.; van Herk, A. M. Inorg. Chim. Acta
1985, 107, 13−17.
(80) Bader, R. F. Atoms in Molecules: A Quantum Theory; Oxford
University Press: Oxford, U.K., 1990.
(39) Zipp, C. F.; Michael, J. P.; Fernandes, M. A.; Marques, H. M. S.
Afr. J. Chem. 2013, 66, 158−166.
(40) Young, R.; Chang, C. K. J. Am. Chem. Soc. 1985, 107, 898−909.
(41) Liu, J.-G.; Naruta, Y.; Tani, F. Angew. Chem., Int. Ed. 2005, 44,
1836−1840.
(42) Collman, J. P.; Decrea
3549.
́
u, R. A. J. Org. Chem. 2004, 69, 3546−
(43) Liu, J.-G.; Ohta, T.; Yamaguchi, S.; Ogura, T.; Sakamoto, S.;
Maeda, Y.; Naruta, Y. Angew. Chem., Int. Ed. 2009, 48, 9262−9267.
(44) Berto, T. C.; Praneeth, V. K. K.; Goodrich, L. E.; Lehnert, N. J.
Am. Chem. Soc. 2009, 131, 17116−17126.
(81) Varadwaj, P. R.; Cukrowski, I.; Marques, H. M. J. Phys. Chem. A
2008, 112, 10657−10666.
(82) Varadwaj, P. R.; Cukrowski, I.; Perry, C. B.; Marques, H. M. J.
Phys. Chem. A 2011, 115, 6629−6640.
(45) Karlin, K. D. Nature 2010, 463, 168−169.
(46) Decrea
J. Org. Chem. 2007, 72, 2794−2802.
́
u, R. A.; Collman, J. P.; Yang, Y.; Yan, Y.; Devaraj, N. K.
(83) Varadwaj, P. R.; Marques, H. M. Phys. Chem. Chem. Phys. 2010,
12, 2126−2138.
4428
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429

Inorganic Chemistry
Article
(84) Varadwaj, P. R.; Marques, H. M. Theor. Chem. Acc. 2010, 127,
711−725.
(85) Varadwaj, P. R.; Varadwaj, A.; Marques, H. M. J. Phys. Chem. A
2011, 115, 5592−5601.
(86) Varadwaj, P. R.; Varadwaj, A.; Peslherbe, G. H.; Marques, H. M.
J. Phys. Chem. A 2011, 115, 13180−13190.
(87) Systat Software, 11.0.0.75; SigmaPlot: San Jose, CA, 2008.
(88) MestReNova, v. 6.0.2−5475; Mestrelab Research : Santiago de
Compostela, Spain, 2009.
(89) Hill, J. A.; Pratt, J. M.; Williams, R. J. P. J. Chem. Soc. A 1964,
5149−5153.
(90) PG Instruments, Leicestershire, UK, 2012.
(91) Kirino, O.; Yamamoto, S.; Kato, T. Agric. Biol. Chem. 1980, 44,
2143−2147.
(92) Marques, H. M.; Munro, O. Q.; Crawcour, M. L. Inorg. Chim.
Acta 1992, 196, 221−229.
(93) Conn, J. B.; Wartman, T. G. Science 1952, 115, 72−73.
(94) Marques, H. M.; Brown, K. L.; Jacobsen, D. W. J. Biol. Chem.
1988, 263, 12378−12383.
(95) Wang, J.; Wolf, R.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J.
Comput. Chem. 2004, 25, 1157−1174.
(96) Hypercube, Inc.: 7.03, Gainesville, FL, 2002.
(97) Marques, H. M.; Ngoma, B.; Egan, T. J.; Brown, K. L. J. Mol.
Struct. 2001, 561, 71−91.
(98) Allen, F. H. Acta Crystallogr., Sect. B 2002, B58, 380−388.
(99) Bruno, I. J.; Cole, J. C.; Edgington, P. R.; Kessler, M.; Macrae, C.
F.; McCabe, P.; Pearson, J.; Taylor, R. Acta Crystallogr. 2002, B58,
389−397.
(100) Anderson, W. P.; Edwards, W. D.; Zerner, M. C. Inorg. Chem.
1986, 25, 2728−2732.
(101) Gaussian 09, Revision A.02; Gaussian, Inc: Wallingford, CT,
2009.
(102) Becke, A. D. Phys. Rev. A 1988, 38, 3098−3100.
(103) Perdew, J. P.; Wang, Y. Phys. Rev. B 1986, 33, 8822−8824.
(104) Schafer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97,
̈
2571−2577.
(105) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,
̈
5829−5835.
(106) Dunlap, B. I. J. Chem. Phys. 1983, 78, 3140−3142.
(107) Dunlap, B. I. J. Mol. Struct.: THEOCHEM 2000, 529, 37−40.
(108) Eichkorn, K.; Treutler, O.; Ohm, H.; Haser, M.; Ahlrichs, R.
̈
̈
Chem. Phys. Lett. 1995, 240, 283−290.
(109) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor.
Chem. Acc. 1997, 97, 119−124.
(110) Yanai, T.; Tew, D. P.; Handy, N. C. Chem. Phys. Lett. 2004,
393, 51−57.
(111) AIMAll (Version 11.12.19), http://aim.tkgristmill.com,
08.05.04, Overland Park, KS, 2011.
4429
dx.doi.org/10.1021/ic5000793 | Inorg. Chem. 2014, 53, 4418−4429