Experimental study of hydrogen bonding potentially stabilizing the 5'-deoxyadenosyl radical from coenzyme B12

Article abstract of DOI:10.1002/chem.201201840

Coenzyme B₁₂ can assist radical enzymes that accomplish the vicinal interchange of a hydrogen atom with a functional group. It has been proposed that the Co-C bond homolysis of coenzyme B₁₂ to cob(II)alamin and the 5'-deoxyadenosyl radical is aided by hydrogen bonding of the corrin C19-H to the 3'-O of the ribose moiety of the incipient 5'-deoxyadenosyl radical, which is stabilized by 30 kJ mol^-1 (B. Durbeej et al., Chem. Eur. J. 2009, 15, 8578-8585). The diastereoisomers (R)- and (S)-2,3-dihydroxypropylcobalamin were used as models for coenzyme B ₁₂. A downfield shift of the NMR signal for the C19-H proton was observed for the (R)-isomer (Δ=4.45 versus 4.01 ppm for the (S)-isomer) and can be ascribed to an intramolecular hydrogen bond between the C19-H and the oxygen of CHOH. Crystal structures of (R)- and (S)-2,3-dihydroxypropylcobalamin showed C19-H···O distances of 3.214(7) A? (R-isomer) and 3.281(11) A? (S-isomer), which suggest weak hydrogen-bond interactions (-ΔG<6 kJ mol^-1) between the CHOH of the dihydroxypropyl ligand and the C19-H. Exchange of the C19-H, which is dependent on the cobalt redox state, was investigated with cob(I)alamin, cob(II)alamin, and cob(III)alamin by using NMR spectroscopy to monitor the uptake of deuterium from deuterated water in the pH range 3-11. No exchange was found for any of the cobalt oxidation states. 3',5'-Dideoxyadenosylcobalamin, but not the 2',5'-isomer, was found to act as a coenzyme for glutamate mutase, with a 15-fold lower k_cat/K_M than 5'-deoxyadenosylcobalamin. This indicates that stabilization of the 5'-deoxyadenosyl radical by a hydrogen bond that involves the C19-H and the 3'-OH group of the cofactor is, at most, 7 kJ mol^-1 (-ΔG). Examination of the crystal structure of glutamate mutase revealed additional stabilizing factors: hydrogen bonds between both the 2'-OH and 3'-OH groups and glutamate 330. The actual strength of a hydrogen bond between the C19-H and the 3'-O of the ribose moiety of the 5'-deoxyadenosyl group is concluded not to exceed 6 kJ mol^-1 (-ΔG).

Full text of DOI:10.1002/chem.201201840

DOI: 10.1002/chem.201201840
Experimental Study of Hydrogen Bonding Potentially Stabilizing the
5’-Deoxyadenosyl Radical from Coenzyme B12
Peter Friedrich,[a, b] Ulrich Baisch,[a] Ross W. Harrington,[a] Fredrick Lyatuu,[b]
Kai Zhou,[c] Felix Zelder,[c] William McFarlane,[a] Wolfgang Buckel,[b, d] and
Bernard T. Golding*[a]
Dedicated to Professors Duilio Arigoni and Jꢀnos Rꢁtey
Abstract: Coenzyme B12 can assist radi-
cal enzymes that accomplish the vicinal
interchange of a hydrogen atom with a
functional group. It has been proposed
oxygen of CHOH. Crystal structures of
(R)- and (S)-2,3-dihydroxypropylcob-
oxidation states. 3’,5’-Dideoxyadenosyl-
cobalamin, but not the 2’,5’-isomer, was
found to act as a coenzyme for gluta-
À
G
3.214(7) ꢀ (R-isomer) and 3.281(11) ꢀ
(S-isomer), which suggest weak hydro-
mate mutase, with a 15-fold lower kcat/
À
that the Co C bond homolysis of coen-
KM than 5’-deoxyadenosylcobalamin.
This indicates that stabilization of the
5’-deoxyadenosyl radical by a hydrogen
zyme B12 to cob(II)alamin and the 5’-
gen-bond
interactions
(ÀDG<
deoxyadenosyl radical is aided by hy-
6 kJmolÀ1) between the CHOH of the
À
À
À
drogen bonding of the corrin C19 H to
dihydroxypropyl ligand and the C19
bond that involves the C19 H and the
À
the 3’-O of the ribose moiety of the in-
cipient 5’-deoxyadenosyl radical, which
is stabilized by 30 kJmolÀ1 (B. Durbeej
et al., Chem. Eur. J. 2009, 15, 8578–
8585). The diastereoisomers (R)- and
(S)-2,3-dihydroxypropylcobalamin
H. Exchange of the C19 H, which is
3’-OH group of the cofactor is, at most,
7 kJmolÀ1 (ÀDG). Examination of the
crystal structure of glutamate mutase
revealed additional stabilizing factors:
hydrogen bonds between both the 2’-
OH and 3’-OH groups and glutamate
330. The actual strength of a hydrogen
dependent on the cobalt redox state,
was investigated with cob(I)alamin, co-
b(II)alamin, and cobACTHNUGTRNEUNG(III)alamin by
using NMR spectroscopy to monitor
the uptake of deuterium from deuterat-
ed water in the pH range 3–11. No ex-
change was found for any of the cobalt
were used as models for coenzyme B12.
A downfield shift of the NMR signal
for the C19 H proton was observed
for the (R)-isomer (d=4.45 versus
4.01 ppm for the (S)-isomer) and can
be ascribed to an intramolecular hydro-
À
bond between the C19 H and the 3’-O
of the ribose moiety of the 5’-deoxy-
ACHTUNGTRNEaNUNG denosyl group is concluded not to
À
Keywords: coenzyme B12
·
5’-de-
exceed 6 kJmolÀ1 (ÀDG).
oxyadenosyl radical · enzymes · hy-
drogen bonds · mutase · radical ions
À
gen bond between the C19 H and the
Introduction
Coenzyme B12-dependent enzymes comprise of mutases,
which catalyze either carbon-skeleton rearrangements or
1,2-shifts of amino groups, and eliminases, which bring
about the loss of ammonia or water from a substrate mole-
cule[1] and include class II ribonucleotide reductase.[1e,2] The
catalytic activity of coenzyme B12-dependent enzymes de-
pends on homolytic cleavage of the relatively weak cobalt-
[a] Dr. P. Friedrich, Dr. U. Baisch, Dr. R. W. Harrington,
Prof. Dr. W. McFarlane, Prof. Dr. B. T. Golding
School of Chemistry, Newcastle University
Newcastle upon Tyne, NE1 7RU (UK)
(III)–carbon
s-bond
(bond
dissociation
energy
E-mail: bernard.golding@newcastle.ac.uk
ꢀ130 kJmolÀ1)[3] to provide cob(II)alamin and the 5’-deoxy-
adenosyl (Ado) radical. The primary organic Ado radical
abstracts a hydrogen atom from the substrate to form 5’-de-
oxyadenosine and a substrate-derived radical, which rear-
ranges to a form a product-related radical. Abstraction of a
hydrogen atom from the methyl group of 5’-deoxyadenosine
by the product-related radical then gives the product with
regeneration of the Ado radical.[1d,h,4] Among the important
[b] Dr. P. Friedrich, F. Lyatuu, Prof. Dr. W. Buckel
Laboratorium fꢁr Mikrobiologie, Fachbereich Biologie
Philipps-Universitꢂt, 35032 Marburg (Germany)
[c] K. Zhou, Dr. F. Zelder
Institute of Inorganic Chemistry, University of Zꢁrich
Winterthurerstrasse 190, 8057, Zꢁrich (Switzerland)
[d] Prof. Dr. W. Buckel
Max-Planck-Institut fꢁr terrestrische Mikrobiologie
Karl-von-Frisch-Str. 10, 35043 Marburg (Germany)
À
unresolved issues with this pathway are how the Co C bond
homolysis is achieved at a rate appropriate for the enzymat-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201201840.
16114
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16114 – 16122

FULL PAPER
ic processes and whether the intermediate radicals are stabi-
lized by cob(II)alamin.[1e]
For mutases (e.g., glutamate mutase), it has been pro-
posed that cob(II)alamin stabilizes the intermediate methyl-
ene radicals, thus acting as a so-called conductor and lower-
ing the transition-state energy barrier for radical formation
and rearrangement.[1e] The energy difference between a con-
certed mechanism for the formation of a substrate-derived
radical that is stabilized by cob(II)alamin (i.e., conductor)
and a stepwise reaction sequence with cob(II)alamin as a
spectator was calculated to favor the former by about
30 kJmolÀ1.[5] For the eliminases, cob(II)alamin is a specta-
tor to the rearrangement step.[6] There are also some elimi-
nases in which coenzyme B12 is replaced by S-adenosylme-
thionine (SAM), a source of the 5’-deoxyadenosyl radical
but not of cob(II)alamin, for example, class III ribonucleo-
tide reductase from Escherichia coli or bacteriophage T4
and glycerol dehydratase from Clostridium butyricum.[2,7]
The conductor–spectator subdivision is in accord with EPR
studies and protein crystallography. EPR spectroscopy
showed that the distance between cob(II)alamin and an in-
termediate radical is about 10 ꢀ in the eliminases, for exam-
ple, ethanolamine ammonia lyase (EAL).[8] The crystal
structures of EAL[9] and diol dehydratase[10] showed a simi-
lar distance from the cobalamin to the bound substrate.
However, according to EPR data[11] and protein crystallogra-
phy[12] for the mutases, the distance between cob(II)alamin
and a bound substrate-derived radical is about 6 ꢀ. The
EPR spectra demonstrated an interaction between the two
radicals, albeit a very weak one.[11b]
Scheme 1. Alkylcobalamins (R2, S2, 3–5) obtained from cyanocobalamin
(CNCbl, R=CN, 6) by using chlorinated adenine precursors; hydroxo-
AHCTUNGTREGcNNUN obalamin (HOCbl, R=OH, 7). Numbering system for cobalamin pro-
tons is adapted from that of Calafat and Marzilli.[13]
The application of density functional theory by using ribo-
sylcobalamin as a model for coenzyme B12 led to the propos-
al of a hydrogen bond between the 3’-oxygen atom of the
are generally weaker (DGꢀ2–5 kJmolÀ1)[18] than those in
which an OH or NH group acts as a donor to the C=O
oxygen (DGꢀ15 kJmolÀ1), although the DG values for hy-
drogen bond formation depend strongly on the environment
(gas, liquid, or solid phase).[15]
A study of the diastereoisomers (R)- and (S)-2,3-dihy-
droxypropylcobalamin (R2 and S2, respectively) by NMR
spectroscopy revealed a surprising downfield shift of the
À
ribose moiety and the cobalamin C19 H (protons are desig-
nated according to the system of Marzilli et al.;[13]
Scheme 1). This interaction was supposed to stabilize the ri-
bosyl methylene radical/cob(II)alamin pair.[14] After fission
À
of the Co C bond, pseudorotation of the ribose from the
C2’-endo to C3’-endo conformation[12b] was required to bring
the 3’-OH into a suitable position (3.5 ꢀ distance; 5.3 ꢀ
without the conformational change) to interact with the
C19 H proton for the (R)-isomer (4.60 ppm versus
À
4.17 ppm for the (S)-isomer).[19] These are model compounds
for coenzyme B12, with the (S)-isomer exactly overlaying the
structure of the coenzyme. One aspect of the present study
was to investigate the possibility that the difference in chem-
À
C19 H. The degree of stabilization was calculated to be
about 30 kJmolÀ1 and was proposed to contribute to the
lowering of the transition state energy for hydrogen atom
abstraction from a substrate molecule (e.g., glutamate).[14]
Furthermore, the Co atom of cob(II)alamin was deemed im-
À
ical shift of the signal for the C19 H in R2 compared to S2
À
is due to a specific hydrogen bond between the C19 H and
the 2’- or 3’-O of R2. This question has been addressed by
À
À
portant for enhancing the polarization of the C19 H bond
through the intervening nitrogen atom, thus facilitating hy-
searching for coupling between the C19 H and protons in
the dihydroxypropyl group. The detection of hydrogen
À
drogen bonding with 3’-OH.
The putative interaction of the C19 H with the 3’-oxygen
bonding between an oxygen of R2 and/or S2 and the C19 H
would indicate that, in principle, this type of interaction is
À
À
of Ado is an example of a C H donor to an O acceptor hy-
possible for the 3’-O of the Ado radical that is formed by
homolysis of the Co C bond of coenzyme B12. During an at-
drogen bond,[15] classically defined by the chloroform
donor–acetone acceptor system.[16] Recently studied exam-
ples include the interaction of a hydroxyl group with a
À
tempt to redetermine the crystal structures of R2 and S2 at
a higher resolution than previously reported[20] to enable the
detection of hydrogen bonds, new crystal forms of R2 and
S2 were discovered and a better insight into the geometry
and hydrogen-bond interactions of these compounds was ob-
neighboring methyl group[17] and a glutamine (amide CO)
[18]
À
residue that hydrogen bonds (2.0 ꢀ) to a histidine C H.
Hydrogen bonds between a C H donor and an O acceptor
À
Chem. Eur. J. 2012, 18, 16114 – 16122
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16115

B. T. Golding et al.
tained. In another aspect of this study, exchange of the
4.17 ppm, respectively. This observation could be interpreted
À
À
C19 H with solvent protons, which is dependent on the
cobalt redox state, was investigated. Finally, the cofactor
as evidence for a hydrogen bond between the C19 H and
one of the propyl hydroxyl groups of R2. To obtain further
evidence, additional NMR experiments on R2 and S2 were
undertaken.
properties of 2’,5’-dideoxyadenosyl- (3) and 3’,5’-dideoxy-
ACHTUNGTRENNUNGadenosyl-cobalamin (4), in which the 2’- and 3’-OH groups
have been deleted in turn, and of peptidoadenylcobalamin
(5) have been determined. Thus, four complementary exper-
imental techniques have been used to probe the reality or
Neither two-dimensional (2D) nuclear Overhauser effect
spectroscopy (NOESY) nor 2D rotating-frame Overhauser
effect spectroscopy (ROESY) showed any NOE between
À
À
otherwise of hydrogen bonding between C19 H and a
neighboring hydroxyl group.
the C19 H and protons of the 2,3-dihydroxypropyl group.
Double-pulsed field gradient spin-echo 1D NOE and ROE
The complementary studies described in this paper show
experiments[21] also failed to detect NOE interactions involv-
À
that although there is experimental support for weak hydro-
ing the C19 H and the 2,3-dihydroxypropyl ligand. Field
gradient HMBC experiments showed no scalar coupling be-
À1
À
gen bonding (DG<6 kJmol ) between the C19 H of a cob-
tween the C19 H and 13C nuclei in the 2,3-dihydroxypropyl
À
alamin and a hydroxyl substituent on a cobalt–alkyl group,
À
the polarization of the C19 H is insufficient to enable its ex-
change with solvent protons in any cobalamin oxidation
state. The 15-fold reduced ability of 3’,5’-dideoxyadenosyl-
group, or between 13C19 and protons of the 2,3-dihydroxy-
propyl group.
R2 and S2 crystallized in different forms (designated R2b
and S2b) compared to the crystal structures reported previ-
ously (R2a and S2a).[20] Different unit-cell dimensions and
numbers of disordered solvent molecules were observed for
the new structures. Because of the high positional disorder
of water and acetone in the cavities between cobalamin mol-
ecules, the resolution of these structures was not as high as
expected and the statistics are only slightly better than those
reported previously.[20] The major difference between S2a
and S2b lies in the orientation of the dihydroxypropyl
ligand. In both structures, water molecules surround this
fragment and engage in hydrogen bonding. In S2a, no hy-
drogen bond is present between the C19 H and the CHOH
group of the dihydroxypropyl ligand (distance>4 ꢀ); how-
ever, S2b shows a weak hydrogen bond of 3.281(11) ꢀ be-
tween the C19 H and the CHOH, which is facilitated by ro-
tation of the ligand around its axis (Figure 2).
Surprisingly, R2a and R2b
ACHTUNGTRENNUNGcobalamin (4) to act as a coenzyme with glutamate mutase
compared with coenzyme B12 corresponds to a change in
transition state energy of 7 kJmolÀ1, which may be due, in
À
part, to the loss of the C19 H/Ado interaction.
Results and Discussion
(R)- and (S)-2,3-Dihydroxypropylcobalamin: These com-
pounds (designated as R2 and S2) are models for coenzy-
me B12 (1; note that the configuration at C4’ in coenzy-
me B12 corresponds to that of the dihydroxypropyl group of
À
S2) and exhibit a surprising difference in the chemical shift
1
À
of the C19 H signal in their H NMR spectra. Experimental
À
values for R2 and S2 are 4.45 ppm and 4.01 ppm, respective-
ly (Figure 1), and literature values are[19] 4.60 ppm and
exhibited different intermolecu-
lar interactions, in particular
between the dihydroxypropyl
and acceptor groups of the ad-
jacent molecule. In R2b, the
OH groups of the dihydroxy-
propyl ligand show hydrogen
bonding to one phosphate
oxygen
(O4a(H)···O2P=
2.738(6) ꢀ) as well as further
interactions to the N(59)H
group attached to carbon PR1
and C57, and to one acetamide
carbonyl (C61=O62). In R2a,
no hydrogen bonding is ob-
served, with only solvent water
molecules acting as hydrogen-
bond acceptors. The C19···O
distances are 3.29(3) ꢀ for R2a
and 3.214(7) ꢀ for R2b, which
À
imply that a weak C19 H···O
interaction with one of the
propyl hydroxyl groups is pres-
Figure 1. 1H NMR (400 MHz, in D2O) spectra of a) (R)- and b) (S)-2,3-dihydroxypropylcobalamin (R2, S2).
À
À
Chemical shifts of the C19 H and C8 H are indicated.
16116
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16114 – 16122

Potential Hydrogen-Bonding Stabilization of a Radical from Coenzyme B12
FULL PAPER
removed by freeze-drying after equilibration in H2O.
MALDI-TOF-MS and 1D- and 2D-NMR spectroscopy were
used to analyse the samples for exchanged hydrogen atoms
(see the Supporting Information for full details). Within ex-
perimental error, the NMR integrals matched the MALDI-
TOF-MS analysis of the samples.
1
The H NMR chemical shifts assigned for previous B12 de-
rivatives[1b,13,19,23] assisted the data analysis given here. As a
control, 7 was incubated in D2O for up to one week at 208C
with pH values of 3, 6, 7, 8, and 10.8 without significant ex-
À
À
À
change of the C19 H, C10 H, B2 H, or any other CH
proton being observed (see the Supporting Information, Fig-
À
ures S2 and S3). C10 H was reported to exchange at a
Figure 2. Fragments of the crystal structures of a) R2b and b) S2b. Ther-
mal ellipsoids are depicted at a 50% probability level.
strongly acidic pH,[24] but at pH 3, no significant exchange
was visible (Figure 3 (bottom) and Figure S2 in the Support-
ing Information).
ent in each structure. For lists of the hydrogen-bond interac-
tions, bond distances, and
angles, see the Supporting In-
formation.
Exchange of cobalamin protons
dependent on cobalt oxidation
state: CobACHTUNGTRENNUNG(III)alamin, cob(II)-
ACHTUNGTRENNUNGalamin, and cob(I)alamin were
examined for the exchange of
À
C19 H and other protons in
the pH range 3.0–10.8 (see the
Supporting Information, Fig-
ure S3). Note that hydroxo-
A
ACHTUNGTRENUN(NG III)alamin
A
258C) at pH<7, and many of
the resonances that arise from
À
À
C H bonds, for example, B2 H
À
and C10 H, shift in response to
the pH of the sample (see the
Supporting Information, Fig-
ure S2, left arrows). It is as-
sumed that the NMR spectra
Figure 3. The exchange of cob(II)alamin (from the reduction of 7 with with PtO2/H2) in D2O at pH 3, 7, and
1
10.8 as monitored by 600 MHz H NMR. The water signal is omitted.
show aquocob
G
G
Similar experiments with cob(II)alamin showed exchange
À
pH 9. Compound 7 was reduced to cob(II)alamin by dithio-
threithol (DTT) or PtO2/H2. Light-driven homolytic cleav-
age of 5’-deoxyadenosylcobalamin (1) was also employed to
generate cob(II)alamin. Cob(I)alamin was obtained by the
reduction of 7 with sodium dithionite.[22] The typical UV/Vis
absorption maxima at around lmax =550 nm for cob-
for C10 H to occur primarily at pH 3 (Figure 3), which was
limited to the incorporation of a single deuterium atom at
values up to pH 6 (see the Supporting Information, Fig-
AHCTUNGTREGuNNUN res S1, S4, S6, and S8). Exchange at pH 10.8 (see the Sup-
porting Information, Figures S1 and S5) could not be attrib-
uted to a single position and is due to side reactions at ele-
À
G
vated pH (see the Supporting Information). The C19 H
doublet was clearly visible in one-dimensional NMR and
R
signal was obtained only after reduction by PtO2/H2 or DTT.
The control sample of untreated 7 and the dithionite-re-
duced sample displayed no EPR signal. After conducting an
exchange experiment, the samples were reoxidized by air to
7 prior to analysis, except for the DTT reductions and kinet-
ic studies with short incubation times, which required oxida-
1H-1H-DFQ-COSY spectra showed the known correlation
network to C18 H and C60 H, regardless of pH, without
any significant decrease in the C19 H peak intensity com-
À
À
À
pared to the cob
(III)alamin control (see the Supporting In-
formation, Figure S3).
Cob(I)alamin showed exchange at every pH level tested
(see the Supporting Information, Figures S1 and S9) and
tion by potassium hexacyanoferrate
ACHTUNGERTN(NUNG III). Deuterium that re-
À
mained in readily exchangeable positions (OD and ND) was
could be attributed to two positions: C10 H (predominant
Chem. Eur. J. 2012, 18, 16114 – 16122
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16117

B. T. Golding et al.
AHCTUNGTRENNUNG
values:
1
0.52Æ0.06 mm;
4
4
0.56Æ0.02 mm). However,
showed a 15-fold lower catalytic
efficiency (kcat =0.089Æ0.01 sÀ1
and kcatKMÀ1 =1.59ꢄ105 sÀ1 mÀ1)
than 1 (kcat =1.24Æ0.36 sÀ1 and
kcat
KMÀ1 =2.38ꢄ106 sÀ1 mÀ1)
when a 14-fold excess of com-
ponent S over E was used. No
catalytic activity was detected
when the hologlutamate mutase
was assembled by using 3 or 5.
The subsequent addition of 1 to
the holoenzyme that was as-
sembled with 3 or 5 did not re-
store the catalytic activity, indi-
cating relatively strong binding
of these artificial cofactors to
glutamate mutase. HOCbl (7)
Figure 4. The exchange of cob(I)alamin (from the reduction of 7 with dithionite) in D2O at pH 3, 7, and 10.8
1
À
À
as monitored by 600 MHz H NMR at pH 9. Highlighted resonances: C19 H (broad doublet); C10 H (s; d=
À
6.00 ppm); B2 H (s, d=6.63 ppm). The water signal is omitted.
À
at pH 3) and B2 H (complete at pH 7 and 10.8, Figure 4).
also assembled the holoenzyme and thus inhibited enzyme
activity in a similar fashion.
À
The C19 H doublet (d=4.2 ppm) was clearly visible in one
1
dimensional H NMR (Figure 4).
2,3-DihydroxypropylcobACTHNUGRTENUNGalamins: (R)- and (S)-2,3-dihydroxy-
Kinetic and binding studies of glutamate mutase with cob-
ACHTUNGTRENNUNGalamin derivatives: To investigate the role of the 3’-hydroxyl
propylcobalamins (R2, S2) were used as models for 5’-deox-
yadenosylcobalamin (1) in experiments to probe for spatial
À
group of the 5’-deoxyadenosyl moiety of coenzyme B12 (1)
during the reaction of glutamate mutase, 2’,5’-dideoxyadeno-
syl- (3), 3’,5’-dideoxyadenosyl- (4), and peptidoadenyl-cob-
interactions of the C19 H. Neither 1D- and 2D-NMR ex-
periments based on the NOE, nor HMBC experiments,
À
showed spatial interactions of the C19 H with the upper
ligand of R2 or S2. The failure
of these measurements to pro-
vide evidence for hydrogen
À
bonding between the C19 H
and a hydroxyl group may be a
consequence of the weakness of
the interaction that is indicated
by the other approaches descri-
bed in this paper.
Crystal structures of R2 and
S2 (except the reported form of
S2a)[20] show C19 H···O2 dis-
À
tances that suggest weak hydro-
gen bond interactions between
the dihydroxypropyl ligand and
À
the C19 H. Bond-dissociation
energies for these hydrogen
bonds are estimated as less
than 6 kJmolÀ1,[15,25] with the
distances lying in the predicted
range of 3.3 to 3.5 ꢀ. In the
crystal form of S2a, hydrogen
bonding of the dihydroxypropyl
Scheme 2. Synthesis of adenine derivatives. Reagents and conditions: i) Cs2CO3, K2CO3, methyl bromoacetate,
DMF, 2.5 h at RT; ii) aqueous KOH, MeOH, H2O, 45 min at RT; iii) 2-chloroethanamine hydrochloride, DMF,
4-dimethylaminopyridine (DMAP), 5 min at 08C followed by 1-ethyl-3-(3-dimethylaminopropyl) carbodi-
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
16118
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16114 – 16122

Potential Hydrogen-Bonding Stabilization of a Radical from Coenzyme B12
FULL PAPER
ligand with the acetamide group of ring B of the next coba-
zimidazole from cobalt, followed by a base-mediated rever-
sible deprotonation at B2. The dissociation is facilitated be-
cause in cob(I)alamin the cobalt is predominantly in the
four-coordinated base-off state.[26] Despite an extensive
lamin molecule is pronounced, whereas the CHOH groups
[20]
À
in R2a, R2b, and S2b are positioned towards the C19 H.
Both R2b and S2b show very similar geometrical parame-
ters around the metal atom, and no significant difference in
intramolecular interactions were observed between the two
diastereoisomers. However, there are significant differences
in intermolecular interactions between neighboring cobala-
min molecules.
À
search, there was no detectable exchange at the C19 H.
Durbeej et al.[14] suggested that a difference in the electron
density of the corrin ring might be able to influence the po-
[5]
À
larization of the C19 H. The carbanion that is derived
À
from the deprotonation of the C19 H is expected to be sta-
bilized by delocalization into the corrin p-system. Intuitively,
The energy calculations reported by Durbeej et al. as-
À
sumed a Co C distance between 3.200 and 4.492 ꢀ to simu-
it seems unlikely that an increase in electron density by the
À
late an intermediate state shortly before complete Co C
reduction of cobCAHTUNGTREN(NUNG III)alamin to cob(II)alamin or cob-
bond rupture.[14] This is in sharp contrast to the much short-
er bond and therefore more rigid situation that exists in the
solid state for R2 and S2. However, it has been shown that
weak CH···O interactions do exist for both R2 and S2, but
1) they are subject to different packing variations in the
solid, which are primarily directed by intermolecular inter-
actions, and 2) there is only a relatively small gain in the
energy associated with the interaction. It is interesting that
in all four crystal structures (R2a, R2b, S2a, S2b) the dihy-
droxypropyl ligand has a different orientation. There is
enough space for a certain degree of freedom as to how this
ligand and its OH groups can be oriented when the com-
pound crystallizes. This can be seen particularly for S2b, in
which the ligand is rotated about 1808 compared to S2a. As
a consequence, intramolecular hydrogen bonding can now
take place. The freedom of orientation of the ligand when it
is fully coordinated to the Co centre (1.95–2.07 ꢀ) suggests
AHCTUNGTREGNUN(N I)alamin could significantly lower the pKa of the C19 H.
À
À
Based on our observations, the pKa of the C19 H is likely
to be higher than that of carbon acids such as chloroform
(pKa 21) or even fluoroform (pKa 27), both of which under-
go exchange in alkaline deuterium oxide.[27]
Biochemical experiments: The strength of the binding of co-
factor analogues 2’,5’-dideoxyadenosyl- (3) and 3’,5’-di-
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
binding. The difference in catalytic efficiency between 4 and
1 converts into a maximum stabilization of the Ado radical
by the ribose 3’-OH (7 kJmolÀ1) during substrate activa-
tion.[14,28]
An additional, hitherto unrecognized, factor in the stabili-
zation of the Ado radical could be due to an interaction of
the 2’-OH of ribose, because compound 3 and peptidoade-
nylcobalamin (5) failed to produce the active holoenzyme.
The crystal structure of glutamate mutase[12b] (Figure 5) re-
veals the possibility of a hydrogen bond between the 2’-OH
and 3’-OH groups with glutamate 330, which could occur
À
that when the Co C distance increases, as in the catalytic
cycle of a mutase, then an OH group could adopt a better
geometrical arrangement and hence form a stronger hydro-
À
gen bond to the C19 H.
Exchange phenomena for cobalamin oxidation states: The
NMR (Figures 3 and 4) and mass spectrometric data (see
the Supplementary Information) presented in this paper
show a sharp distinction in exchange behavior between the
three cobalamin oxidation states, with a pH influence pres-
ent in all cases. For cob-
À
before and after homolysis of the Co C bond. These inter-
actions could act as stabilizing elements for pseudorotation,
lowering the energy barrier for radical generation upon sub-
strate addition. Similar conclusions regarding the joint roles
ACHTUNGTRENNUNG(III)alamin, no exchange was
observed in the pH range of 3–
10.8. For cob(II)alamin, we ob-
À
served C10 H exchange even
in the pH range 3–6, with an in-
crease in rate at lower pH. This
behavior can be ascribed to the
higher electron density in the
corrin ring, which enhances the
propensity for protonation at
C10. This phenomenon was also
observed for cob(I)alamin. In
addition, exchange at B2 occurs
with a faster rate at higher pH.
The mechanism of this ex-
change probably requires disso-
À
Figure 5. Substrate activation at glutamate mutase active site: a) Co C bond intact; b) Ado radical is formed
À
and abstracts a substrate hydrogen. Note that pseudorotation of the ribose after Co C bond cleavage positions
À
ribose 3’-OH within a hydrogen-bonding distance of the C19 H and E330, which can engage in an additional
ciation of the 5,6-dimethylben- hydrogen bond with the ribose 2’-OH before and after 5’-deoxyadenosylcobalamin homolysis.
Chem. Eur. J. 2012, 18, 16114 – 16122
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16119

B. T. Golding et al.
of ribose 2’-OH and 3’-OH have been reached based on
computational studies with methylmalonyl–CoA mutase[28]
and glutamate mutase (interactions with ribose were tested
for K326, E330, and an amide side chain of the corrin
ring).[29] Previous experiments with methylmalonyl–CoA
mutase in the presence of 2’,5’-dideoxy-adenosylcobalamin
(3) showed only 1–2% residual activity.[30]
Experimental Section
General: Chemicals and deuterated solvents were obtained from reputa-
ble suppliers. Cyanocobalamin (CNCbl, vitamin B12, 6) was a generous
gift from DSM Nutritional Products (Basel/Switzerland) and Professor Y.
Murakami (Fukuoka, Japan). The chemicals were of reagent grade and
used without further purification. Solutions used for reductions were
sodium acetate (100 mm, pH 3), potassium phosphate (100 mm, pH 7.0),
or potassium carbonate (pH 10.8). The pH of the samples was adjusted
to other required values by using pH paper or a pH microelectrode
before reduction. Because it was found that pH electrodes react with hy-
drogen and deuterium in a similar manner, the shift between pD and pH
values was compensated for by adjusting the pH of the samples after dis-
solving the appropriate amount of buffer salts in the respective solvent,
according to the standards given in the literature.[31] Note that for reasons
of simplicity, pH is used to describe the samples that were prepared in
D2O. All incubations under reducing conditions were done in an anaero-
bic chamber (5% H2 in N2). All reductions and reoxidations were moni-
tored for completion by using UV/Vis spectroscopy. The samples were
tempered at 208C in air; for incubations at 378C, a heating block was
used.
Conclusion
The experiments presented in this paper detected a weak in-
À
teraction between the C19 H and a hydroxyl group of the
cobalt-bound 2,3-dihydroxypropyl group in R2 or S2, but
À
failed to find exchange of the C19 H in the pH range 3–
10.8 for cobalamin in any oxidation state. However, the re-
sults demonstrate the influence of the redox state of the
cobalt atom in cobalamins on the lower ligand (base-mediat-
Organic synthesis: Three adenine derivatives were prepared as chlorinat-
ed alkylating agents (Scheme 2). N6-Carbobenzoxyadenine (8) was syn-
thesized as described.[32] Methylbromoacetate was use to convert 8 to
methyl N6-carbobenzoxy-2-(6-amino-9H-purin-9-yl)acetate (i; 9),[32]
which was subsequently deprotected to form N6-carbobenzoxy-2-(6-
amino-9H-purin-9-yl)acetic acid (ii; 10).[33] Introduction of a chloroethyl
group gave N6-carbobenzoxy-2-(6-amino-9H-purin-9-yl)-N-(2-chloroeth-
yl) acetamide (iii; 11),[34] and the subsequent Pd-catalyzed deprotection
of the amine gave the first alkylating agent 2-(6-amino-9H-purin-9-yl)-N-
(2-chloroethyl)acetamide (iv; 12).[35] The second chlorinated adenine, 5-
chloro-2’,5’-dideoxyadenosine (v; 14), was prepared from 2’-deoxyadeno-
sine (13) according to published procedures with slight modifications.[36]
The third chlorinated adenine, 5-chloro-3’,5’-dideoxyadenosine (vi; 16)
was prepared similarly from 3’-deoxyadenosine (15). These three alkylat-
ing agents were used to generate alkylcobalamins from CNCbl (6)
(Scheme 1) by using modified, published procedures.[37] For the synthesis
of (R)- and (S)-2,3-dihydroxypropylcobalamin (R2, S2), a previously pub-
lished method,[23a] in which CNCbl (6) was reacted with (R)- or (S)-1-
chloropropane-2,3-diol (Scheme 1), was modified. For details of the or-
ganic synthesis and NMR spectra, see the Supporting Information (NMR
data are found in Table S2). For UV/Vis characterisation of 1, 3, 4, and 5,
see the Supporting Information (Figure S11; in addition, see ref. [38] for
3 and 4).
À
ed exchange of the B2 H in cob(I)alamin) and the electron
density in the corrin ring (exchange by reversible protona-
À
tion at C10 H for cob(II)alamin and cob(I)alamin).
The experiments with 2’,5’-dideoxyadenosylcobalamin (3),
3’,5’-dideoxyadenosylcobalamin (4), and peptidocobalamin
(5) indicate how the 2’-OH and 3’-OH of the ribose moiety
may assist in the generation of the Ado radical and hydro-
gen abstraction from the substrate. Experimental evidence
has been obtained that shows that the 3’-OH group of the
ribose could stabilize the Ado radical by a hydrogen bond
À
with the C19 H of cob(II)alamin. However, the experimen-
tal estimate for the strength of this hydrogen bond (DG
ꢀ6 kJmolÀ1) is significantly less than that calculated previ-
ously.[14,28] Hence, the proposal by Durbeej et al.[14] that cob-
À
ACHTUNGTRENNUNG(II)alamin acts as a conductor through a C19 H hydrogen
bond to the 3’-OH of Ado is barely supported. Based on
measurements of the activities of 3 and 4 relative to 5’-
deoxyACHTUNGTRENNUNGadenosylcobalamin, it is suggested that the Ado radi-
cal also benefits from an interaction between the ribose 2’-
OH and 3’-OH and the glutamate 330 of glutamate mutase.
Our conclusions regarding the symbiotic role of the 2’-OH
and 3’-OH are consistent with new computational stud-
ies.[28,29]
The question of whether there is any stabilizing orbital in-
teraction between cob(II)alamin and intermediate radicals
in mutase reactions remains unresolved.[1e,5] Durbeej et al.
did not find evidence for an orbital interaction between cob-
Analytical methods: NMR spectra for HOCbl (7) and 5’-deoxyadenosyl-
cobalamin (1) were recorded on Bruker AVANCE 300 B and AVANCE
600 spectrometers with 10 mgmLÀ1 samples (standard 1D-1H NMR and
1H-1H-DQF-COSY spectroscopy). NMR spectra for (R)- and (S)-2,3-di-
hydroxypropylcobalamin (R2, S2) (30 mgmLÀ1 at 458C to increase solu-
bility) were recorded on Jeol ECS-400 MHz (only for standard 1H NMR
spectroscopy) and Jeol Lambda 500 MHz machines (advanced measure-
ments), including 2D NOESY (mixing times of 400 and 600 ms), 2D
ROESY (spin-locking (mixing) times of 150 and 250 ms), double pulsed
field gradient spin-echo 1D NOE and ROE experiments (mixing times of
500 and 750 ms)[21] performed with selective excitation of the C19 H or
ACHTUNGTRENNUNG(II)alamin and intermediate radicals as proposed by Buckel
À
et al.,[1e] and there is no experimental evidence for this inter-
action. In a further computational study focusing on methyl-
malonyl–CoA mutase, Bucher et al.[28] used Car–Parrinello
molecular dynamics to explore the role of both 2’-OH and
3’-OH in stabilizing the Ado radical. They stated that the
of target protons of the 2,3-hydroxypropyl group, and field gradient
HMBC experiments with timings optimized for long range couplings be-
tween 1H and 13C of 2, 4, 6, or 8 Hz. NMR spectra of the samples (com-
pounds 3–5 and 8–16) for biochemical testing with glutamate mutase
were recorded on a Bruker AV-500 spectrometer (Karlsruhe, Germany)
and their data processing was carried out with ACD/SpecManager. The
chemical shifts are given in ppm relative to the signal from the deuterat-
ed solvent. EPR measurements were recorded on a Bruker X-band EPR
spectrometer EMX-6/1 with a rectangular standard cavity (77 K). The
sample concentration was 10 mgmLÀ1. After reduction, the samples were
incubated for 3 h before being stored in liquid nitrogen.
À
magnitude of the interaction of the 3’-OH with the C19 H
was “somewhat less in the enzymatic environment than in
the gas phase”. The experimental results given in this paper
show that this interaction must be significantly less than that
previously calculated.
The cobalamins (3, 4, and 5) that were intended for enzymatic assays
with glutamate mutase were purified and analysed by HPLC and HPLC-
16120
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16114 – 16122

Potential Hydrogen-Bonding Stabilization of a Radical from Coenzyme B12
FULL PAPER
ESI-MS. HPLC solvents were 0.1% aqueous trifluoroacetic acid (A) and
methanol (B). HPLC analyses were performed on a Merck-Hitachi L-
7000 system that is equipped with a diode array UV/Vis spectrometer
and Macherey–Nagel Nucleosil C-18ec RP columns (particle size: 5 mm;
pore size: 100 ꢀ; 250ꢄ3 mm; flow rate: 0.5 mLminÀ1). Preparative
HPLC separations were performed on a Varian Prostar system equipped
with two Prostar 215 pumps, a Prostar 320 UV/Vis detector and Macher-
ey–Nagel Nucleosil C-18ec RP columns (particle size: 7 mm; pore size:
100 ꢀ; 250ꢄ40 mm; flow rate: 40 mLminÀ1). HPLC-ESI-MS spectra
were measured on a Bruker HCT spectrometer equipped with an Aquini-
ty UPLC (Waters) by using Nucleosil C-18ec RP columns (particle size:
5 mm; pore size: 100 ꢀ; 250ꢄ3 mm; flow rate: 0.3 mLminÀ1). HPLC-ESI-
MS solvents were 0.1% formic acid (A) and methanol (B). The following
gradient (A) was used for all HPLC, HPLC-ESI-MS, and preparative
HPLC measurements: 25% B for 5 min, then to 100% B over 25 min,
then 100% B for 10 min. The following gradient (B) was used only for
compound 4: 10% B for 2 min, then to 100% B over 10 min, then 100%
B for 13 min.
Cob(I)alamin: Compound 7 (10 mgmLÀ1) was reduced by using sodium
dithionite (1m) in D2O.[22] The samples were reoxidized in air (10 h) and
purified on a C18 column (see details above). For kinetic studies moni-
tored by MALDI-TOF-MS, the samples were first reoxidized with a
slight excess of potassium hexacyanoferrateACTHNUTRGNEUGN(III). The samples were
freeze-dried and redissolved in H2O for MALDI-TOF-MS. For NMR
spectroscopy, the samples were recrystallized from water/acetone (1:10).
The NMR samples in D2O were adjusted to pH 9 with a pH microelec-
trode.
X-ray crystal-structure determination: X-ray crystallographic data were
collected on an Oxford Diffraction Gemini A Ultra diffractometer at
150 K by using CuKa radiation (l=1.54184 ꢀ). Analytical numeric ab-
sorption corrections that use a multifaceted crystal model which is based
on expressions derived by R. C. Clark and J. S. Reid[39] were applied,
based on symmetry-equivalent and repeated reflections. Structures were
solved by direct methods and refined on all unique F2 values, with aniso-
tropic non-H atoms and constrained riding isotropic H atoms. High ani-
sotropy of some atoms in solvent acetone molecules, which are present in
the crystal structure of R2b, indicate disorder, but this could not be re-
solved satisfactorily by split positions and has been ignored. Disordered
water molecules that could not be modeled as discrete atoms in R2b and
S2b were treated by the SQUEEZE procedure of PLATON.[40] Programs
used were CrysAlisPro for data collection, integration, and absorption
corrections,[41] and OLEX2[42] or SHELXTL[43] for structure solution, re-
finement, and graphics. Full details about crystallographic experimental
information is provided in the Supporting Information, together with a
list of bond lengths and angles. CCDC-864775 (R2b) and 864776 (S2b)
contain the supplementary crystallographic data for this paper. These
data can be obtained free of charge from The Cambridge Crystallograph-
ic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
All MALDI-TOF-MS measurements of the deuterium exchange experi-
ments were done on an Applied Biosystems 4800Plus MALDI-TOF/TOF
mass spectrometer. The samples were measured directly upon being dis-
solved in water and after 6 h to check for a difference in exchange. a-
Cyano-4-hydroxycinnamic acid in 50% acetonitrile (0.1% trifluoroacetic
acid in H2O) as a matrix was mixed in a dilution series (2-, 4-, 8-, 16-, 32-
fold) for sample preparation. Sheffield ChemPuter (webpage retrieved
on
1 February 2012: http://winter.group.shef.ac.uk/chemputer/isotopes.
html) was used to generate theoretical values of natural-abundance iso-
topic distribution patterns for MALDI-TOF-MS. The Data Explorer
from Applied Biosystems was used to process MALDI-TOF-MS data
and extract integrals of individual signals. The monoisotopic signal of un-
Enzymatic purification and assays for glutamate mutase kinetic studies:
Glutamate mutase (components S and E) from Clostridium cochlearium
was recombinantly produced and purified by following published proce-
dures.[11a] Methylaspartase was purified from cell-free extracts of Clostri-
dium tetanomorphum.[44] The components S and E were assembled into
the holoenzyme in the presence of 1 or 3, 4, and 5. A standard coupled
enzyme assay with methylaspartase was used for the measurement of glu-
tamate mutase kintetic parameters (378C, 50 mm Tris, pH 8.3, 0.05 mm
mercaptoethanol). The assays for enzymatic activity were performed by
using component E (GlmE) and a 14-fold excess of component S (GlmS)
together with the cofactor analogue at varying concentrations. All alkyl-
cobalamins (1, 3, 4, and 5) were shielded from light and measurements
were performed under a protective red light. The initial specific activities
of the partially purified enzymes were: methylaspartase (44 UmgÀ1),
GlmS (66 UmgÀ1), and GlmE (18 UmgÀ1). The specific activity of GlmS
was determined by using an excess of GlmE, and vice versa. When meas-
uring KM and Vmax for 5’-deoxyadenosylcobalamin (1), GlmS (5 mg),
GlmE (2.6 mg), methylaspartase (36 mg), and glutamic acid (20 mm) were
used, and the cofactor concentration was varied (0.32 mm–25 mm). In the
case of cobalamin derivatives 3, 4, and 5, ten times the amount of apo-
glutamate mutase was used. 3’,5’-Dideoxyadenosylcobalamin (4) concen-
trations were varied between 0.35 mm and 70 mm. The concentration of
component E was used to calculate kcat for compounds 1 and 4. After
mixing all ingredients together, the cuvette was incubated for 5 min at
378C to ensure the formation of the holoenzyme composed of E, S, and
either 1 or one of the analogues 3, 4, or 5. Then the reaction was started
with (S)-glutamate, and the formation of (2S,3S)-methylaspartate was
measured continuously at 240 nm by conversion to mesaconate by using
mesaconase as an auxiliary enzyme. In this assay, the inclusion of 1 or
the analogue had two different functions: the assembly of the holoen-
zyme, and the action as a cofactor for catalysis.
labeled cobACHTUNGTRENNUNG(III)alamin without an upper ligand was used to calculate its
whole isotopic distribution pattern (based on natural abundance and a
reference measurement of unlabeled hydroxocobalamin; see the Support-
ing Information, Table S1). This was then subtracted from the signals
with higher mass to reveal the pattern caused by +1 Da labeled cob-
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
ACHTUNGTRENNUNG
the Supporting Information). All other mass spectra were recorded
either in the positive or negative mode on an Esquire HCT from Bruker
(Bremen, Germany).
Deuterium exchange of cobalamins under reducing conditions:
Cob(II)alamin reduction with dithiothreitol (DTT): Hydroxocobalamin
(HOCbl, 7, 10 mgmLÀ1) was reduced by using DTT (0.1m) in D2O. The
reoxidation was done by using potassium hexacyanoferrateACTHNUTRGENUGN(III) (0.2m) in
D2O. For analysis, the samples were desalted on C18 columns (Sep-Pak
Vac from Waters). The columns were conditioned with methanol, equili-
brated with D2O, loaded with solutions of the sample in D2O, and then
eluted with acetonitrile in D2O (50%). The samples were freeze-dried
and suspended in H2O prior to MALDI-TOF-MS measurements. Initial-
ly, the samples were dissolved in D2O and used for NMR spectroscopy,
but no clear spectra could be obtained. Additional purification by using
HPLC (RP18 reverse-phase semipreparative column, solution gradient of
0 to 80% acetonitrile in H2O) or recrystallization from water/acetone
(1:10) did not lead to sufficiently clear NMR spectra.
Cob(II)alamin reduction with PtO2/H2: Compound 7 (10 mgmLÀ1) was
stirred with PtO2 (0.5 mgmLÀ1) under a H2/N2 atmosphere (5%). For re-
oxidation, the catalyst was filtered off and the sample was incubated in
air (10 h). No further purification for NMR analysis was needed. For
MALDI-TOF-MS, the samples were freeze-dried and dissolved in H2O.
Cob(II)alamin generation by photolysis: 5’-Deoxyadenosylcobalamin (1,
10 mgmLÀ1) in D2O was irradiated for 6 h (120 W incandescent light
bulb) for three successive days in an NMR tube in an anaerobic chamber.
The control sample was kept in the dark under air. After reoxidation in
air (10 h), no further purification was done for NMR analysis and
MALDI-TOF-MS.
Acknowledgements
We are indebted to Professor Christoph Kratky, Graz, Austria for provid-
ing the crystal structure presented in Figure 5. We thank the German Re-
Chem. Eur. J. 2012, 18, 16114 – 16122
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
16121

B. T. Golding et al.
search Foundation (W.B. and B.T.G.), the EPSRC (B.T.G. and W.M.) and
the Fonds der chemischen Industrie (W.B.) for generous financial sup-
port. We thank Dr. Antonio Pierik for EPR spectra, Jçrg Kahnt for
MALDI-TOF-MS, Dr. Ronald Wagner and Dr. Xiulan Xie for NMR (all
from Marburg, Germany), and Dr. Karel Zelenka (Zꢁrich) for the syn-
thesis of compounds 3 and 4.
[19] R. J. Anderson, R. M. Dixon, B. T. Golding, J. Organomet. Chem.
1992, 437, 227–237.
[20] N. W. Alcock, R. M. Dixon, B. T. Golding, J. Chem. Soc. Chem.
Commun. 1985, 603–605.
[21] a) K. Stott, J. Keeler, Q. N. Van, A. J. Shaka, J. Magn. Reson. 1997,
125, 302–324; b) W. Bauer, A. Soi, A. Hirsch, Magn. Reson. Chem.
2000, 38, 500–503.
[22] J. M. Pratt, in Inorganic Chemistry of Vitamin B12, Academic Press,
New York, 1972, p. 201.
[23] a) R. M. Dixon, B. T. Golding, O. W. Howarth, J. L. Murphy, J.
Chem. Soc. Chem. Commun. 1983, 243–245; b) K. Kurumaya, M.
Kajiwara, Chem. Pharm. Bull. 1989, 37, 9–12.
[24] a) R. Bonnett, D. G. Redman, Proc. R. Soc. Lond. A 1965, 288,
342–343; b) H. A. O. Hill, B. E. Mann, J. M. Pratt, R. J. P. Williams,
J. Chem. Soc. A 1968, 564–567; c) S. A. Cockle, H. A. O. Hill,
R. J. P. Williams, B. E. Mann, J. M. Pratt, Biochim. Biophys. Acta
Gen. Subj. 1970, 215, 415–418.
[25] S. Scheiner, Phys. Chem. Chem. Phys. 2011, 13, 13860–13872.
[26] D. Lexa, J. M. Saveant, Acc. Chem. Res. 1983, 16, 235–243.
[27] a) E. A. Symons, M. J. Clermont, J. Am. Chem. Soc. 1981, 103,
3127–3130; b) I. F. Tupitsyn, A. S. Popov, Russ. J. Gen. Chem. 2001,
71, 89–101.
[28] D. Bucher, G. M. Sandala, B. Durbeej, L. Radom, D. M. Smith, J.
Am. Chem. Soc. 2012, 134, 1591–1599.
[29] a) K. P. Jensen, U. Ryde, J. Am. Chem. Soc. 2005, 127, 9117–9128;
b) J. B. Rommel, J. Kꢂstner, J. Am. Chem. Soc. 2011, 133, 10195–
10203.
[30] A. M. Calafat, S. Taoka, J. M. Puckett, C. Semerad, H. Yan, L. Luo,
H. Chen, R. Banerjee, L. G. Marzilli, Biochemistry 1995, 34, 14125–
14130.
[1] a) J. Halpern, Science 1985, 227, 869–875; b) R. Banerjee, Chemistry
and Biochemistry of B12, Wiley, New York, 1999; c) E. N. G. Marsh,
C. L. Drennan, Curr. Opin. Chem. Biol. 2001, 5, 499–505; d) K. L.
Brown, Chem. Rev. 2005, 105, 2075–2150; e) W. Buckel, C. Kratky,
B. T. Golding, Chem. Eur. J. 2006, 12, 352–362; f) E. N. G. Marsh,
D. P. Patterson, L. Li, ChemBioChem 2010, 11, 604–621; g) K.
Gruber, B. Puffer, B. Krꢂutler, Chem. Soc. Rev. 2011, 40, 4346–
4363; h) W. Buckel, B. T. Golding, Radical Enzymes. Encyclopedia
of Radicals in Chemistry, Biology and Materials, Wiley, New York,
2012.
[2] P. Nordlund, P. Reichard, Annu. Rev. Biochem. 2006, 75, 681–706.
[3] a) B. P. Hay, R. G. Finke, J. Am. Chem. Soc. 1986, 108, 4820–4829;
b) B. P. Hay, R. G. Finke, J. Am. Chem. Soc. 1987, 109, 8012–8018.
[4] a) R. Banerjee, Chem. Rev. 2003, 103, 2083–2094; b) T. Toraya,
Chem. Rev. 2003, 103, 2095–2127.
[5] P. Kozlowski, T. Kamachi, T. Toraya, K. Yoshizawa, Angew. Chem.
2007, 119, 998–1001; Angew. Chem. Int. Ed. 2007, 46, 980–983.
[6] B. T. Golding, L. Radom, J. Am. Chem. Soc. 1976, 98, 6331–6338.
[7] a) C. Raynaud, P. SarÅabal, I. Meynial-Salles, C. Croux, P. Soucaille,
Proc. Natl. Acad. Sci. USA 2003, 100, 5010–5015; b) J. R. OꢅBrien,
C. Raynaud, C. Croux, L. Girbal, P. Soucaille, W. N. Lanzilotta, Bio-
chemistry 2004, 43, 4635–4645.
[31] a) R. Gary, R. G. Bates, R. A. Robinson, J. Phys. Chem. 1964, 68,
3806–3809; b) M. Paabo, R. G. Bates, Anal. Chem. 1969, 41, 283–
285.
[32] S. A. Thomson, J. A. Josey, R. Cadilla, M. D. Gaul, C. F. Hassman,
M. J. Luzzio, A. J. Pipe, K. L. Reed, D. J. Ricca, R. W. Wiethe, S. A.
Noble, Tetrahedron 1995, 51, 6179–6194.
[33] K. L. Dueholm, M. Egholm, C. Behrens, L. Christensen, H. F.
Hansen, T. Vulpius, K. H. Petersen, R. H. Berg, P. E. Nielsen, O. Bu-
chardt, J. Org. Chem. 1994, 59, 5767–5773.
[34] K. Zhou, F. Zelder, Angew. Chem. 2010, 122, 5305–5307; Angew.
Chem. Int. Ed. 2010, 49, 5178–5180.
[35] G. M. Anantharamaiah, K. M. Sivanandaiah, J. Chem. Soc. Perkin
Trans. 1 1977, 490–491.
[36] a) D. W. Jacobsen, R. J. Holland, Y. Montejano, F. M. Huennekens,
J. Inorg. Biochem. 1979, 10, 53–65; b) K. L. Brown, S. Cheng, X.
Zou, J. Li, G. Chen, E. J. Valente, J. D. Zubkowski, H. M. Marques,
Biochemistry 1998, 37, 9704–9715.
[8] a) V. Bandarian, G. H. Reed, Biochemistry 2002, 41, 8580–8588;
b) S. C. Ke, Biochim. Biophys. Acta Gen. Subj. 2003, 1620, 267–272;
c) J. M. Canfield, K. Warncke, J. Phys. Chem. B 2005, 109, 3053–
3064; d) L. Sun, O. A. Groover, J. M. Canfield, K. Warncke, Bio-
chemistry 2008, 47, 5523–5535.
[9] N. Shibata, H. Tamagaki, N. Hieda, K. Akita, H. Komori, Y. Sho-
mura, S. Terawaki, K. Mori, N. Yasuoka, Y. Higuchi, T. Toraya, J.
Biol. Chem. 2010, 285, 26484–26493.
[10] N. Shibata, Y. Nakanishi, M. Fukuoka, M. Yamanishi, N. Yasuoka,
T. Toraya, J. Biol. Chem. 2003, 278, 22717–22725.
[11] a) O. Zelder, B. Beatrix, U. Leutbecher, W. Buckel, Eur. J. Biochem.
1994, 226, 577–585; b) H. Bothe, D. J. Darley, S. P. J. Albracht, G. J.
Gerfen, B. T. Golding, W. Buckel, Biochemistry 1998, 37, 4105–
4113; c) G. H. Reed, S. O. Mansoorabadi, Curr. Opin. Struct. Biol.
2003, 13, 716–721; d) S. O. Mansoorabadi, R. Padmakumar, N. Fa-
zliddinova, M. Vlasie, R. Banerjee, G. H. Reed, Biochemistry 2005,
44, 3153–3158.
[37] a) B. Zagalak, J. Pawelkiewicz, Acta Biochim. Pol. 1965, 12, 219–
228; b) S. Gschçsser, R. B. Hannak, R. Konrat, K. Gruber, C. Mikl,
C. Kratky, B. Krꢂutler, Chem. Eur. J. 2005, 11, 81–93.
[38] a) M. Ichikawa, T. Toraya, Biochim. Biophys. Acta Protein Struct.
Mol. Enzymol. 1988, 952, 191–200; b) M. P. Jensen, J. Halpern, J.
Am. Chem. Soc. 1999, 121, 2181–2192.
[12] a) F. Mancia, N. H. Keep, A. Nakagawa, P. F. Leadlay, S. McSwee-
ney, B. Rasmussen, P. Bçsecke, O. Diat, P. R. Evans, Structure 1996,
4, 339–350; b) K. Gruber, R. Reitzer, C. Kratky, Angew. Chem.
2001, 113, 3481–3484; Angew. Chem. Int. Ed. 2001, 40, 3377–3380.
[13] A. M. Calafat, L. G. Marzilli, J. Am. Chem. Soc. 1993, 115, 9182–
9190.
[39] R. C. Clark, Acta Crystallogr. Sect. A 1995, 51, 887–897.
[40] a) A. L. Spek, J. Appl. Crystallogr. 2003, 36, 7–13; b) A. L. Spek,
Acta Crystallogr., Sect. D: Biol. Crystallogr. 2009, 65, 148–155.
[41] CrysAlisPro, Version 1.171.35 ed., Agilent Technologies, 2010, Xcali-
bur CCD system.
[14] B. Durbeej, G. M. Sandala, D. Bucher, D. M. Smith, L. Radom,
Chem. Eur. J. 2009, 15, 8578–8585.
[15] C. A. Hunter, Angew. Chem. 2004, 116, 5424–5439; Angew. Chem.
Int. Ed. 2004, 43, 5310–5324.
[16] a) E. A. Guggenheim, Mixtures: The Theory of the Equilibrium
Properties of Some Simple Classes of Mixtures, Solutions and Alloys,
Clarendon Press, Oxford, 1952; b) R. D. Green, Hydrogen Bonding
[42] O. V. Dolomanov, L. J. Bourhis, R. J. Gildea, J. A. K. Howard, H.
Puschmann, J. Appl. Crystallogr. 2009, 42, 339.
[43] G. M. Sheldrick, Acta Crystallogr. Sect. A 2008, 64, 112.
[44] H. A. Barker, R. D. Smyth, H. J. Bright, L. L. Ingraham, Methods
Enzymol. 1962, 5, 827–832.
À
by C H Groups, Macmillan, London, 1974.
[17] a) A. Kolocouris, J. Org. Chem. 2009, 74, 1842–1849; b) S. Horowitz,
J. D. Yesselman, H. M. Al-Hashimi, R. C. Trievel, J. Biol. Chem.
2011, 286, 18658–18663.
Received: May 25, 2012
Revised: August 21, 2012
Published online: October 18, 2012
[18] D. A. Boudreaux, J. Chaney, T. K. Maiti, C. Das, FEBS J. 2012, 279,
1106–1118.
16122
www.chemeurj.org
ꢃ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 16114 – 16122