Direct evidence for a hydrogen bond to bound dioxygen in a myoglobin/hemoglobin model system and in cobalt myoglobin by pulse-EPR spectroscopy

Article abstract of DOI:10.1002/anie.200705180

(Figure Presented) Hydrogen bond revealed: In a cobalt(II) porphyrin complex which serves as a model for the dioxygen binding site of myoglobin (Mb) and hemoglobin, a distal hydrogen bond to the bound O₂ was identified and characterized by pulse ENDOR spectroscopy. A similar but stronger hydrogen bond was revealed with the same methods in natural Co-Mb-O₂.

Full text of DOI:10.1002/anie.200705180

Communications
DOI: 10.1002/anie.200705180
Hydrogen Bonds in Proteins
Direct Evidence for a Hydrogen Bond to Bound Dioxygen in a
Myoglobin/Hemoglobin Model System and in Cobalt Myoglobin by
Pulse-EPR Spectroscopy**
Henry Dube, Besnik Kasumaj, Carlos Calle, Makoto Saito, Gunnar Jeschke, and
François Diederich*
In memory of Arthur Schweiger
Discrimination between dioxygen and carbon monoxide
binding to the respiratory proteins myoglobin (Mb) and
hemoglobin (Hb) is vital for aerobic life. There is still an
ongoing debate about the nature and molecular mechanism of
this discrimination.[1] It is widely believed that the distal
histidine stabilizes bound dioxygen by a hydrogen-bond
interaction, although a direct observation and characteriza-
tion of this proposed hydrogen bond has been difficult and
ambiguous.[1,2b] Functional CoII analogues of the natural, FeII-
containing Mb and Hb can be used to study the interactions of
bound dioxygen with its surroundings by EPR methods.[2] It
has been shown that the dioxygen adducts of Co-Mb and
natural Mb adopt very similar geometries.[3] Herein, we
present a new CoII-containing model complex 1-Co for the
dioxygen-binding site of Mb and Hb, together with a direct
pulse-EPR evidence for a distal hydrogen bond in the
corresponding dioxygen adduct 1-Co-O2. Furthermore, we
extended our study to Co-Mb and provide a comprehensive
pulse-EPR study of the distal hydrogen bonding in Co-Mb-O2
as well.
Model compound 1-Co was synthesized via the corre-
sponding ZnII derivative 1-Zn, using a classical porphyrin
condensation as the key step (Scheme 1).[4] Dipyrromethanes
2
[5] and 3[6] and aldehyde 4[7] were subjected to the condensa-
tion reaction to yield free-base porphyrin 5-2H. Subsequent
ZnII insertion gave 5-Zn, to which the axial base was attached
using 1-(6-bromohexyl)imidazole,[8] leading to formation of
6-Zn. Removal of the SiiPr3 protecting group furnished 7-Zn,
and Sonogashira cross-coupling with 5(6)-iodobenzimida-
zole[9] provided 1-Zn with the fully functionalized porphyrin
core. After the ZnII ion was removed with TFA, CoII was
inserted into the intermediate free-base porphyrin 1-2H using
CoCl2.
X-band (ca. 9.6 GHz) continuous wave (CW) EPR
spectra of 1-Co-O2 in frozen solution (120 K) were recorded
to demonstrate the spectroscopic purity of the obtained
samples. The spectra exhibit the common features of a low-
spin CoII species with dominating dz2 character in a porphyrin
environment (see Supporting Information). Dioxygen
adducts were obtained within seconds by exposing samples
of 1-Co to dioxygen at 208C. By forming the dioxygen
adducts, the spin population is transferred from the central
CoII ion to the attached dioxygen nuclei. This transfer results
in a different X-band CW EPR spectrum of the frozen
solution (120 K), exhibiting narrower features (see
Supporting Information).
Compound 1-Co consists of a CoII porphyrin core with an
alkyl-tethered imidazole base to mimic the proximal histidine
in Mb and Hb, and an alkyne-appended benzimidazole
residue mimicking the distal histidine (Scheme 1). Only the
distal hydrogen-bond-donating proton can be exchanged by
D2O in complex 1-Co.
Proton hyperfine splittings can be resolved by pulse-EPR
methods.[10] However, a variety of ESEEM (electron spin
echo envelope modulation) and ENDOR (electron nuclear
double resonance) methods failed to measure the complete
proton hyperfine splittings in 1-Co-O2 (see Supporting
Information) because of a severe cross suppression effect.[11]
The Q-band (ca. 35 GHz) Davies-ENDOR experiment[10d,e]
combined with short preparation pulses (which enhance the
relative intensity of larger hyperfine splittings[10d]) allowed the
complete determination of the hyperfine interactions caused
by protons close to bound dioxygen. These experiments were
carried out at low temperature (10 K) on a frozen solution of
1-Co-O2. Field-swept FID-integral-detected EPR spectra
were recorded to determine the magnetic field (observer)
positions for the pulse-EPR experiments. Subsequently,
proton ENDOR spectra were recorded at twelve different
observer positions for the complete detection of the proton
splittings. The largest proton hyperfine coupling for complex
1-Co-O2—ranging from 6.0 MHz at the single crystal posi-
[*] H. Dube,[+] Dr. M. Saito, Prof. Dr. F. Diederich
Laboratorium für Organische Chemie
ETH Zürich
Hönggerberg, HCI, 8093 Zürich (Switzerland)
Fax: (+41)44-632-1109
E-mail: diederich@org.chem.ethz.ch
Homepage: http://www.diederich.chem.ethz.ch
B. Kasumaj,[+] Dr. C. Calle
Laboratorium für Physikalische Chemie, ETH Zürich
Hönggerberg, HCI, 8093 Zürich (Switzerland)
Prof. Dr. G. Jeschke
Fachbereich Chemie, Universität Konstanz
Universitätsstrasse 10, 78457 Konstanz (Germany)
[+] These authors contributed equally to this work.
[**] This work was suppported by the Swiss National Foundation, the
NCCR “Nanoscale Science” Basel, and ETH Zurich. We are grateful
to Joerg Forrer for technical support.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
2600
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 2600 –2603

Angewandte
Chemie
associated with the hydrogen bond is dominated
by a through-space contribution with only a small
isotropic contribution. The hydrogen bond is thus
mainly dipolar in character.[10c]
Simulations of CW EPR and ENDOR spec-
tra of 1-Co-O2 were carried out with the main
emphasis on deriving the geometry parameters
and bonding properties of the exchangeable
proton. CW EPR spectra simulations of 1-Co-
O2 were performed to obtain the Co–O2 geom-
etry. The g-values were g = [2.0027 Æ 0.0005,
1.989 Æ 0.001, 2.0723 Æ 0.0005], the principal
values of the metal hyperfine tensor ACo
=
[À54, À28, À28]MHz (each Æ 2 MHz) and the
Euler angles (a, b, g) = (08 Æ 108, 708 Æ 108, 08 Æ
208) with respect to the g-frame, that is, the angle
between the O–O axis and the heme normal is
708 in complex 1-Co-O2 (Figure 2).
Subsequently, Davies-ENDOR simulations
were carried out at each observer field position
to fit the experimental difference spectra of the
exchangeable proton (Figure 1b) and the solvent
protons (see the Supporting Information). We
obtained a hyperfine tensor for the exchangeable
proton of AH = [À6.8, À6.0, 13.5]MHz (each
Æ 0.5 MHz), with Euler angles (a, b, g) = (08,
1058, 08) Æ 108 (Figure 2). This result indicates a
Scheme 1. Synthesis of 1-Co. a) TFA, CH2Cl2, 208C, 16 h, then chloranil, 408C, 2 h,
15%; b) Zn(OAc)2·2H2O, MeOH/CHCl3, 658C, 1 h, 96%; c) 1-(6-bromohexyl)imida-
zole, Cs2CO3, DMF, 208C, 14 h, 79%; d) nBu4NF, THF, 208C, 40 min, 93%; e) 5(6)-
iodobenzimidazole, [Pd(PPh3)4], CuI, NEt3, DMF, 1008C, 4 h, 51%; f) TFA, CHCl3,
208C, 12 min, quant.; g) CoCl2, 2,6-lutidine, THF, 208C, 12 h. TFA=trifluoroacetic
acid.
mainly dipolar character of the hydrogen bond,
H
with the strongest proton interaction (Az
)
directed towards dioxygen. The exchangeable
proton is positioned directly above the dioxygen,
with an angle between the distal proton and the
O–O axis of Æ 1058. Although the hyperfine
tions gz and gy (low and high field, respectively) to 14.0 MHz
along gx—disappeared upon D2O exchange (Figure 1a). This
hyperfine splitting for an exchangeable proton is exception-
ally large when compared to all the splittings for related CoII
dioxygen adducts reported to date.[2] It clearly demonstrates a
hydrogen-bonding interaction in complex 1-Co-O2 between
the bound dioxygen and a defined and exchangeable nearby
proton. Additionally, Davies-ENDOR spectra of 1-Co-O2
were recorded in CD2Cl2 as solvent, to probe possible solvent
interactions (see Supporting Information). The only signifi-
cant differences to the spectra recorded in CH2Cl2 appear at
smaller values for the proton hyperfine splittings (up to
6.0 MHz; see Supporting Information). Samples of 1-Co-O2
prepared in wet and dry solvent showed the same large
hyperfine splittings. Thus, a dominating interaction of solvent
molecules or residual solvent water with bound dioxygen in 1-
Co-O2 can be precluded. To elucidate whether the hyperfine
matrix of the exchangeable proton has both positive and
negative principal values, the X-band 6-pulse HYSCORE[12]
spectrum of 1-Co-O2 was measured at the field position
corresponding to the largest proton interaction. Although the
6-pulse HYSCORE spectrum did not allow the full exchange-
able splitting to be measured, its off anti-diagonal ridge was
clearly visible (see Supporting Information). This spectrum
demonstrates that both positive and negative principal values
exist, which in turn shows that the hyperfine coupling
tensor is not entirely axial, and the bulk electron spin
population is not localized at one atom but distributed
between the two oxygen nuclei, we assumed the spin density
to be centered at one point in space.[13] A distance of (2.3 Æ
0.2) Š between this spin-density center and the NH proton of
the distal benzimidazole was obtained by using the point-
dipole approximation with
a dipolar contribution T=
6.5 MHz.[10c] A preliminary modeling at the semiempirical
PM5 level of theory of the FeII substituted complex 1-Fe-O2
(for easier modeling) showed that the distal benzimidazole
NH proton is located close to dioxygen at a very similar
distance of approximately 2.40 Š (see the Supporting
Information). Taken together with the fact that the NH
proton of the distal benzimidazole is the only proton that can
be exchanged by D2O, we conclude that the hydrogen bond
observed indeed arises from interactions between this distal
NH proton and bound dioxygen.
Having found the suitable pulse-EPR method to measure
the complete proton hyperfine splittings of porphyrin CoII
dioxygen adducts, we completed our study with the examina-
tion of the hyperfine splittings in Co-Mb-O2. Co-Mb-O2 was
prepared in aqueous buffer solution and in deuterated buffer
solution using apo-Mb from equine skeletal muscle according
to standard methods.[14] An even larger hyperfine splitting,
ranging from 10.0 MHz (gz and gy at low and high field,
respectively) to 19.0 MHz (near gx), was found in the
Angew. Chem. Int. Ed. 2008, 47, 2600 –2603
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 2601

Communications
Figure 2. A distal hydrogen bond to bound dioxygen is formed in
complex 1-Co-O2. The angles and distance shown were obtained from
the frozen-solution EPR measurements of 1-Co-O2; b=1058.
Figure 1. a) Frozen-solution (CH2Cl2) Davies-ENDOR spectra recorded
at Q-band frequency at 10 K of 1-Co-O2 (black), and after D2O
exchange (red). Arrows indicate the largest and exchangeable proton
splitting. Observer positions 1–3 are indicated in the inset showing the
FID-integral-detected EPR spectra. The preparation pulse length at
observer position 1 (close to gz) and 3 (gy) is 160 ns, and at observer
position 2 (gx) 80 ns. b) Corresponding difference spectra of 1-Co-O2
before and after D2O exchange at the three different observer
positions 1–3 (black) and simulated spectra of the exchangeable
proton (blue). The orientations contributing to the experimental spec-
tra are projected on the unit spheres.
Figure 3. Frozen-solution Davies-ENDOR spectra recorded at Q-band
frequency of Co-Mb-O2 in H2O buffer (black), and in D2O buffer (blue)
using a 80 ns preparation pulse. Arrows indicate the largest and
exchangeable proton hyperfine splitting. Observer positions 1–3 are
indicated in the inset showing the FID-integral-detected EPR spectra.
complex 1-Co-O2 is the character of the hydrogen bond in Co-
Mb-O2, which is mainly dipolar in nature. This was again
demonstrated by the X-band 6-pulse HYSCORE spectrum
measured at the position of the largest extension of the
exchangeable proton hyperfine splitting (see Supporting
Information).
Although the proton hyperfine splittings of Co-Mb-O2
have been studied already, only one field position (corre-
sponding to gy) was measured and the observed exchangeable
proton splitting was considerably smaller (9.0 MHz).[2b] In the
light of our results, we suppose that this 9.0 MHz proton
hyperfine splitting is actually only the inner part of the
complete proton hyperfine splitting at this field position.
corresponding frozen-solution Q-band Davies-ENDOR spec-
tra of Co-Mb-O2 (13 K). Upon D2O exchange of the buffer
solution, this large hyperfine splitting disappeared (Figure 3).
In Co-Mb-O2, as in our model compound, a defined distal
hydrogen bond was detected. The evolution of the largest and
exchangeable proton hyperfine splitting over the different
field positions in Co-Mb-O2 is the same as in the model
complex 1-Co-O2, only the magnitude is greater in the
protein, which suggests a similar orientation of the two
hydrogen bonds but a smaller distance (2.0 Æ 0.2 Š) to the
interacting proton in Co-Mb-O2. Also similar to our model
2602 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 2600 –2603

Angewandte
Chemie
As the protein is studied in H2O as solvent, exchangeable
proton splittings are hard to assign to particular protein
residues or surrounding H2O molecules. However, our model-
complex study revealed the hydrogen-bonded benzimidazole
NH proton to be positioned above dioxygen. As the geometry
of the distal hydrogen bond is similar in our model and in Co-
Mb-O2, the interacting proton in the protein must also be
located above dioxygen. Only the distal histidine NH proton
is able to reach the bound dioxygen in such a position and at a
distance of 2.00 Š. H2O in the distal pocket has a completely
different position[15] and thus, it is very unlikely that H2O is
the cause of the largest and exchangeable proton hyperfine
splitting in our Davies-ENDOR spectra.
[1]a) J. P. Collman, R. Boulatov, C. J. Sunderland, L. Fu, Chem.
Rev. 2004, 104, 561 – 588; b) J. A. Lukin, V. Simplaceanu, M.
Zou, N. T. Ho, C. Ho, Proc. Natl. Acad. Sci. USA 2000, 97,
10354 – 10358; c) S. Borman, Chem. Eng. News 1999, 77(49), 31 –
36; d) M. Momenteau, C. A. Reed, Chem. Rev. 1994, 94, 659 –
698; e) B. A. Springer, S. G. Sligar, J. S. Olson, G. N. Phillips, Jr.,
Chem. Rev. 1994, 94, 699 – 714; f) L. Pauling, Nature 1964, 203,
182 – 183.
[2]a) H. C. Lee, E. Scheuring, J. Peisach, M. R. Chance, J. Am.
Chem. Soc. 1997, 119, 12201 – 12209; b) J. Huettermann, R.
Stabler in Electron Magnetic Resonance of Disordered Systems
(Ed.: N. D. Yordanov), World Scientific, Singapore, 1989,
pp. 127 – 148; c) F. A. Walker, J. Bowen, J. Am. Chem. Soc.
1985, 107, 7632 – 7635; d) B. M. Hoffman, D. H. Petering, Proc.
Natl. Acad. Sci. USA 1970, 67, 637 – 643.
As has been observed earlier in EPR crystal studies,[13b,16]
at low temperature (T< À508C) there are two different
dioxygen species of Co-Mb-O2 (species 1, species 2), differing
by approximately 908 in the orientation of bound dioxygen
with respect to the cobalt heme group. We can confirm these
findings, and were able to separate the FID-integral-detected
EPR spectra of both species using a longitudinal relaxation
time filter at Q-band frequency (see Supporting Information).
Additionally, by using the same filter in combination with the
Davies-ENDOR sequence, we were able to attribute the
largest and exchangeable proton hyperfine splitting exclu-
sively to the species 2 (longitudinal relaxation time of ca.
450 ms). Apparently, the dioxygen of species 1 (longitudinal
relaxation time of ca. 80 ms) does not strongly interact with an
exchangeable proton within 2.70 Š, such an interaction would
have been observable in our spectra. As the absolute geo-
metries of dioxygen in the two species have not been assigned
as of yet, we can only speculate about the reasons of this
finding. Further studies are being pursued to answer the open
questions.
In summary, we directly detected distal hydrogen bonding
to bound dioxygen in model complex 1-Co-O2 and in Co-Mb-
O2 by Q-band Davies-ENDOR and present the complete
EPR parameters for this interaction in 1-Co-O2 obtained by
simulations. The dipolar character of the hydrogen bond as
well as its orientation was found to be similar in the model
complex 1-Co-O2 and in Co-Mb-O2. Thus, we conclude that
complex 1-Co-O2 is an excellent functional model for the
dioxygen-binding site of Mb and Hb, and reproduces well
crucial properties of the natural proteins.
[3]E. A. Brucker, J. S. Olson, G. N. Phillips, Jr., Y. Dou, M. Ikeda-
Saito, J. Biol. Chem. 1996, 271, 25419 – 25422.
[4]All compounds were fully characterized by melting points, 1H
and 13C NMR spectroscopy (except for 1-Co), IR, UV/Vis, mass
spectrometry (MS), and high-resolution MS. The preparation of
1-Co is described in full detail in the Supporting Information.
[5]Dipyrromethane 2 was prepared as described for the related
compound lacking the OH group, see: B. Felber, F. Diederich,
Helv. Chim. Acta 2005, 88, 120 – 153.
[6]B. J. Littler, M. A. Miller, C.-H. Hung, R. W. Wagner, D. F.
OꢀShea, P. D. Boyle, J. S. Lindsey, J. Org. Chem. 1999, 64, 1391 –
1396.
[7]P. J. Dandliker, F. Diederich, A. Zingg, J.-P. Gisselbrecht, M.
Gross, A. Louati, E. Sanford, Helv. Chim. Acta 1997, 80, 1773 –
1801.
[8]P. Weyermann, F. Diederich, Helv. Chim. Acta 2002, 85, 599 –
617.
[9]Z. Kazimierczuk, L. Dudycz, R. Stolarski, D. Shugar, J.
Carbohydr. Nucleosides Nucleotides 1981, 8, 101 – 117.
[10]a) D. Goldfarb, D. Arieli, Annu. Rev. Biophys. Biomol. Struct.
2004, 33, 441 – 468; b) T. Prisner, M. Rohrer, F. MacMillan,
Annu. Rev. Phys. Chem. 2001, 52, 279 – 313; c) A. Schweiger, G.
Jeschke, Principles of Pulse Electron Paramagnetic Resonance,
Oxford University Press, Oxford, 2001; d) Pulsed Davies-
ENDOR is described in: C. Fan, P. E. Doan, C. E. Davoust,
B. M. Hoffman, J. Magn. Reson. 1992, 98, 62 – 72; e) E. R.
Davies, Phys. Lett. A 1974, 47, 1 – 2.
[11]S. Stoll, C. Calle, G. Mitrikas, A. Schweiger, J. Magn. Reson.
2005, 177, 93 – 101.
[12]B. Kasumaj, S. Stoll, J. Magn. Reson. 2008, DOI: 10.1016/
j.jmr.2007.11.001.
[13]a) See: S. Van Doorslaer, A. Schweiger, B. Kraeutler, J. Phys.
Chem. B 2001, 105, 7554 – 7563, and references therein; b) For
the stereoelectronic structure of Co-Mb-O2 see: L. C. Dickinson,
J. C. W. Chien, Proc. Natl. Acad. Sci. USA 1980, 77, 1235 – 1239.
[14]F. Ascoli, M. Rosaria, R. Fanelli, E. Antonini, Methods
Enzymol. 1981, 76, 72 – 87.
Received: November 9, 2007
Published online: January 25, 2008
[15]M. L. Quillin, R. M. Arduini, J. S. Olson, G. N. Phillips, Jr., J.
Mol. Biol. 1993, 234, 140 – 155.
[16]H. Hori, M. Ikeda-Saito, T. Yonetani, Nature 1980, 288, 501 –
502.
Keywords: ENDOR spectroscopy · hemoglobin ·
hydrogen bonding · myoglobin · porphyrinoids
.
Angew. Chem. Int. Ed. 2008, 47, 2600 –2603
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 2603