Electrochemical, FT-IR and UV/VIS spectroscopic properties of the
caa
3
oxidase from
T. thermophilus
Petra Hellwig
1
, Tewfik Soulimane
2
* and Werner Ma¨ ntele
1
1
Institut fu
¨
r Biophysik der Johann-Wolfgang-Goethe-Universita
¨
t, Frankfurt/M., Germany;
2
Institut fu
¨
r Biochemie der Rheinisch-
Westfa
¨
lischen-Technischen Hochschule, Aachen, Germany
The caa
3
-oxidase from Thermus thermophilus has been
studied with a combined electrochemical, UV/VIS and
Fourier-transform infrared (FT-IR) spectroscopic
approach. In this oxidase the electron donor, cytochrome c,
is covalently bound to subunit II of the cytochrome c

and 0.378 V vs. SHE¢, respectively). The three phases of the
potential-dependent development of the difference signals
can be attributed to the cooperativity between the hemes a,
a
3
and the Cu
B
center, showing typical behavior for cyto-
chrome c oxidases. A stronger cooperativity of Cu
B
is dis-
cussed to reflect the modulation of the enzyme to the
different key residues involved in proton pumping. We thus
studied the FT-IR spectroscopic properties of this enzyme to
identify alternative protonatable sites. The vibrational
modes of a protonated aspartic or glutamic acid at
1714 cm
)1
concomitant with the reduced form of the protein
can be identified, a mode which is not present for other
cytochrome c oxidases. Furthermore modes at positions
characteristic for tyrosine vibrations have been identified.
Electrochemically induced FT-IR difference spectra after
inhibition of the sample with cyanide allows assigning the
formyl signals upon characteristic shifts of the m(C¼O)
modes, which reflect the high degree of similarity of heme a
3
to other typical heme copper oxidases. A comparison with
previously studied cytochrome c oxidases is presented and
on this basis the contributions of the reorganization of the

/e
–
across the membrane contribute to the proton
gradient needed to synthesize ATP.
Two pathways have been proposed to serve for consumed
and pumped protons on the basis of site-directed mutagen-
esis [1,2] and later using the crystal structures [3–5]. These
pathways are highly conserved among most studied cyto-
chrome oxidases [2,6]. However, cytochrome oxidases have
been reported that lack amino acids disputed to be essential
in proton translocation. In the case of caa
3
-oxidases from
T. thermophilus, for example, as well as from Rhodothermus
marinus, the amino acid Glu278 (numbering for Paracoccus
denitrificans), which is proposed to pass protons in the
D-pathway to the binuclear center, is missing, but proton-
pumping activity is observed [3,7–9]. A highly conserved
Tyr–Ser couple was suggested to replace Glu278 [9]. In the
current understanding, two pathways are necessary for the
catalytic activity, but different residues may be involved. In
an important step for the understanding of the essentials for
cytochrome c oxidase activity and coupled proton pump-
ing, the crystal structure of the aberrant ba
3
-oxidase from
T. thermophilus was determined [10] and alternative path-
ways discussed.
Correspondance to P. Hellwig, Institut fu
¨

oxidases, however, including a covalently bound type-c
heme [11]. This is currently only found in a few bacteria
[9,12]. Recent results showed that the enzyme is made of two
fusion proteins. The smaller protein consists of a typical
oxidase subunit II sequence, which provides the homonu-
clear Cu
A
binding site and is fused to a cytochrome c
domain [11,13]. The larger protein is a fusion product of
subunit I, that has the hemes a, a
3
and the Cu
B
sites, and
subunit III [8,12,13]. The heme c center in the caa
3
-oxidase is
proposed to serve as the first electron acceptor from a bc
1
complex [14]. We note, however, that no bc
1
complex has
yet been described for T. thermophilus. No activity was
detected for a reaction with soluble horse heart cyto-
chrome c, c
552
from T. thermophilus and yeast iso1 cyto-
chrome c, which serve as natural reductands for
cytochrome c oxidases, but a reduction can be noted for
nonphysiological reducing agents such as N,N,N¢,N¢-tetra-

For electrochemistry the sample was solubilized in
n-decyl-b-
D
-maltopyranoside, 100 m
M
phosphate buffer
(pH 7) containing 100 m
M
KCl and concentrated to
approximately 0.5 m
M
using Microcon ultrafiltration cells
(Millipore). Exchange of H
2
O against D
2
O was performed
by repeatedly concentrating the enzyme and rediluting it in a
D
2
O phosphate-buffer. H/D exchange was better than 80%
as judged from the shift of the amide-II mode at 1550 cm
)1
in the FT-IR absorbance spectra (data not shown). For
inhibition with cyanide, samples were diluted with 500 lL
of 100 m
M
phosphate buffer containing 20 m
M
KCN

)1
and 1000 cm
)1
.Asa
supporting electrolyte, 100 m
M
KCl was added. Approxi-
mately 5–6 lL of the protein solution were sufficient to fill
the spectroelectrochemical cell. Potentials quoted with the
data refer to the Ag/AgCl/3
M
KCl reference electrode,
adding + 208 mV for SHE¢ (pH 7) potentials. Midpoint
potentials are described for both electrode types.
Spectroscopy
FT-IR and UV/VIS difference spectra as a function of the
applied potential were obtained simultaneously from the
same sample with a setup combining an IR beam from
the interferometer (modified IFS 25, Bruker, Germany) for
the 4000–1000 cm
)1
range and a dispersive spectrometer
for the 400–900 nm range as reported previously [18]. First,
the protein was equilibrated with an initial potential at the
electrode, and single beam spectra in the VIS and IR range
were recorded. A potential step to the final potential was
then applied, and single beam spectra of this state were
again recorded after equilibration. Difference spectra as
presented here were then calculated from the two single
beam spectra with the initial single beam spectrum taken as

ted Nernst curve to the measured absorbance change at a
Ó FEBS 2002 Characterization of T. thermophilus caa
3
oxidase (Eur. J. Biochem. 269) 4831
single wavelength by an interactive fit. All parameters have
to be adjusted manually until the theoretical Nernst curve
and the measured data match well (fit by eye).
RESULTS AND DISCUSSION
UV/VIS difference spectra
Figure 1A shows the oxidized-minus-reduced UV/VIS
difference spectra of the caa
3
-oxidase from T. thermophilus
obtained for a potential step from )0.5 V to 0.5 V (vs. Ag/
AgCl/3
M
KCl). In the oxidized-minus-reduced spectra the
positive signals correlate with the oxidized and the negative
signals with the reduced form of the enzyme. For the
reduced form, the Soret band can be observed at 415 and
442 nm, and for the oxidized form at 403 and 422 nm. The
b–band can be seen at 517 nm and the a–band at 547 nm
and 603 nm.
The difference signals that can be observed between 400
and 700 nm include the contributions of the hemes c, a and
a
3
. The difference signals observed at 403, 415, 517 and
547 nm are characteristic for heme c. In electrochemically
induced difference spectra of horse heart cytochrome c the

.
UV/VIS redox titrations
In Fig. 2A the potential dependent development of the
a–band from heme c at 548 nm in an oxidative titration can
be seen (filled circles). The theoretical Nernst fit described
in Materials and methods yields a midpoint potential of
E
m
¼ )0.01 ± 0.01 V (vs. Ag/AgCl, and 0.218 V vs.
SHE¢) for heme c. This value was also reported by Yoshida
and Fee [16]. The midpoint potential of soluble horse heart
cytochrome c is 0.048 V (0.256 V vs. SHE¢, as obtained
with the same method as described here and in [18]); other
cytochrome c types show a close midpoint potential. The
midpoint potential of the cytochrome c in the cyto-
chrome c–cytochrome c oxidase complex is unknown and
may be the origin for this downshift. Alternatively, the
electron transfer directly from the bc
1
complex, as suggested
for a possible mechanism, could require a lower potential.
Figure 2B (filled circles) shows the potential-dependent
development of the difference signal at 443 nm from )0.4 V
to + 0.6 V. Three phases can be clearly discriminated. The
theoretical Nernst fit yields midpoint potentials
Em
1
¼ )0.075 ± 0.01 V, Em
2
¼ 0.04 ± 0.01 V and

3
.Onthisbasis,Cu
B
and
heme a
3
have been suggested to act as a two-electron
acceptor [12] in contrast to bovine heart oxidase, where
subsequent one-electron transfer is reported. The three
phasic curve, with a step of n ¼ 1 for each step as found
here, shows a significantly more comparable titrimetric
behavior in comparison to other typical oxidases, but in
contrast to the work by Yoshida and Fee [16]. The small
difference to other oxidases found here, reflected in the
stronger second step at 40 mV, may be attributed to the
Fig. 1. Oxidized-minus-reduced UV/VIS difference spectra of the caa
3
-
oxidase from T. thermophilus. Results obtained for a potential step
from )0.5 V to 0.5 V (vs. Ag/AgCl/3
M
KCl) in the absence (A) or the
presence (B, solid line) of cyanide, and for a potential step from )0.5 to
0.05 V in the presence of cyanide (B, dotted line).
4832 P. Hellwig et al.(Eur. J. Biochem. 269) Ó FEBS 2002
covalently attached heme c or to a generally changed
cooperativity of the other redox centers.
In order to discriminate the contributions of the cofac-
tors, inhibitors uncoupling or changing the cooperativity
can be used. Addition of cyanide strongly shifts the heme a

ates the a-band. Figure 2C shows a comparison of the
potential-dependent development of the modes at 599 nm
(triangles) and 442 nm (open circles) in the presence of
cyanide. As seen for the curve that represents the titration
curve at 602 nm (triangles) a smaller ratio is present for the
potential at )220 mV than for the titration curve measured
at 442 nm in the same conditions, supporting the assign-
ment to heme a
3
.
Heme c, however, shows a relatively small difference in
midpoint potential of 15 mV in the presence of cyanide
(Fig. 2A, empty circles) and thus does not indicate a
noteworthy cooperativity between heme a
3
and the heme c
centers. Heme c can be ruled out as origin for the distinct
second phase at 40 mV in the titration curve for the hemes a
and a
3
. It may be suggested that the different cooperativity,
as well as the lower heme a
3
potential, is necessary to
compensate the differences caused by the presence of
different key residues in the D-pathway, since the potentials
are assumed to be crucial for the coupling of electron and
proton transfer. Interestingly, for the caa
3
-oxidase from

absorbance units
at frequencies above 1750 cm
)1
, where no signals appear.
Only in regions of strong absorbance of the sample, such as
Fig. 2. Potential dependent development of the hemes in the caa
3
-oxid-
ase from T. thermophilus . Heme c was monitored at 548 nm in the
absence (filled circles) and presence (open circles) of cyanide (A) and a
midpoint potential of )0.01 V ± 0.01 V (vs. Ag/AgCl/3
M
KCl or
0.218 V vs. SHE¢) was obtained by a theoretical Nernst fit (solid line).
The hemes a and a
3
were monitored at 442 nm in the absence
(filled circles) and presence (open circles) of cyanide. Midpoint poten-
tials of Em
1
¼ )0.075V±0.01V, Em
2
¼ 0.04 V ± 0.01 V and
Em
3
¼ 0.17 V ± 0.02 V were determined (these values correspond to
0.133 V, 0.248 V and 0.378 V vs. SHE¢, respectively). After addition of
the inhibitor cyanide (open circles) a midpoint potential of
)0.22 V ± 0.01 V for heme a
3

3
. In addition to the signals of the hemes, the reorgan-
ization of the polypeptide backbone and amino acid side
chains occurring upon electron transfer of the five redox
active centers heme c, a, a
3
,Cu
A
and Cu
B
, and coupled
processes such as proton transfer can be expected to
manifest themselves in the spectra. In the following
paragraph the difference spectra will be described and
discussed. Tentative assignments will be presented on the
basis of the comparison to IR and Raman spectra of heme
model compounds, other oxidases, spectra of isolated
amino acids as model compounds and information on
contributions from the secondary structure from infrared
absorbance spectra and the deconvolution of the amide-I
region.
A particular problem of the assignment in the difference
spectra is the superposition of signals from different
constituents of the oxidase, which can lead to the possibility
of multi component bands and may present ambiguities in
the assignment. A spectral region particularly susceptible for
overlapping bands is the amide-I range. Although in this
range (approx. 1690–1610 cm
)1
) typical contributions from

and 1646 cm
)1
, and prominent negative differ-
ence modes are present at 1660 cm
)1
, 1626 cm
)1
and
1614 cm
)1
. After H/D exchange (Fig. 3B) the increase of
the signal at 1696 cm
)1
and 1626 cm
)1
can be observed. A
clear shift from 1634 cm
)1
to 1658 cm
)1
and to 1650 cm
)1
is visible. The modes involved in the signals at 1660 cm
)1
contribute in the range characteristic for the absorbance
from a–helical secondary structure elements. However,
absorbance changes induced by the reorganization of
a-helical secondary structure elements are expected to show
very small shifts after H/D exchange at most (2–10 cm
)1

spectrum of the caa
3
-oxidase from T. thermophilus to
amide-II modes, however, appears less probable since little
or no shift for H/D exchange is observed.
Assignment of heme vibrational modes
Formyl substituent. The C¼O bond of the formyl group at
the porphyrin ring of hemes a and a
3
can be expected to
contribute between 1680 cm
)1
and 1606 cm
)1
, depending
on hydrogen bonding with neighboring amino acids. The
formyl substituent of heme a is predicted to form a
Fig. 3. Oxidized-minus-reduced FT-IR difference spectra of the caa
3
-
oxidase from T. thermophilus. Results obtained for a potential step
from )0.5 V to 0.5 V (vs. Ag/AgCl/3
M
KCl) equilibrated in H
2
O(A)
and D
2
O buffer (B).
4834 P. Hellwig et al.(Eur. J. Biochem. 269) Ó FEBS 2002

electrochemically induced FT-IR difference spectra in
Fig. 3A corresponding bands can be seen at 1678 cm
)1
(oxidized form) and 1660 cm
)1
(reduced form). These
modes have been previously attributed to the formyl side
chain from cytochrome c oxidase from bovine heart [27–29]
reportedtobesensitivetoCN
–
binding in a characteristic
way [27]. In Fig. 4A the spectra in the presence of cyanide
clearly reflect a shift of the mode at 1678–1668 cm
)1
and of
the mode at 1660–1652 cm
)1
, supporting the assignment to
the heme a
3
formyl mode. In a direct comparison of these
vibrational modes to those observed for the cytochrome c
oxidase from P. denitrificans, an analog environment of the
protein site of the heme a
3
formyl group in the absence and
presence of cyanide can be concluded.
Porphyrin ring vibrations. Porphyrin ring vibrations of
the heme centers, for example the CaCm vibration (m
37

39
) will contribute to the electrochemically induced
FT-IR difference spectra. However, we refrain from dis-
cussing and assigning these modes on the basis of the data
presented here, in spite of the fact that bands in the
difference spectra are observed in the region where the
modes were attributed.
The vibrational modes of bound cyanide
Electrochemically induced FT-IR difference spectra of
cyanide bound to heme a
3
were characterized to specify
possible variations of the binuclear center in direct compari-
son to other oxidases, as for example the cytochrome c
oxidase from P. denitrificans. In the spectral range from
2200–2000 cm
)1
contributions from cyanide ligand bound
to heme a
3
can be expected. In the inset in Fig. 4 a strong
positive mode can be seen at 2148 cm
)1
and a negative
signal at 2040 cm
)1
for a potential step from )0.5 to 0.5 V
(unbroken line) and for )0.5 to 0.05 V (dotted line). A small
mode at 2092 cm
)1

completely developed. To allow the above-mentioned
bridged structure to be present, Cu
B
must be oxidized at
this potential in the presence of cyanide, since the contri-
bution of the unbridged Fea
3
3+
–CN
–
structure was repor-
ted to be observable at 2132 cm
)1
.
The position of the cynide vibrational modes are
essentially identical to the ones observed for bovine heart
oxidase [31] and from P. denitrificans (Hellwig et al.
unpublished results) reflecting a close environment and
ligand binding properties of the binuclear heme a
3
–Cu
B
center.
Identification of protonable sites
Aspartic and Glutamic acid side chains. The m(C¼O)
mode of protonated aspartic and glutamic side chains
absorb typically above 1710 cm
)1
, the exact absorption
depending on the hydrogen bonding. A negative mode is

upon H/D exchange. The signal
may originate from heme c reduction, but also reflect a
distinct protonable site in subunit I, involved in a different
proton pathway.
For the cytochrome c oxidase from P. denitrificans
difference modes at 1746 cm
)1
and 1734 cm
)1
have been
observed and attributed to Glu278 [19,34], and to the
equivalent residues in the cytochrome bo
3
quinol oxidase
from E. coli [35,36], a residue which lacks the caa
3
-oxidase
as mentioned above. Correspondingly no analogous con-
tribution can be seen here.
Tyrosines. Pereira et al. [9] recently suggested a Tyr–Ser
motif, conserved in several of the cytochrome c oxidases
which lack the above-mentioned Glu278 residue, to be
involved in proton pumping. For tyrosine side chains, the
m
19
(CC) ring mode for the protonated form of tyrosines is
proposed to absorb with an strong signal at 1518 cm
)1
and
for the deprotonated form at 1496–1486 cm

oxidase from T. thermophilus.
caa
3
FT-IR
caa
3
RR [17] Redox state Tentative assignments
Comparable modes for
aa
3
P. denitrificans [30]
1744 ox m(C¼O) Glu278 for P. denitrificans 1746
– red m(C¼O) Glu278 for P. denitrificans 1734
1714 – red m(C¼O) Asp/Glu –
1708 ox m(C¼O) Asp/Glu 1708
1694 ox
1692 red amide-I (b-sheet) 1694
1684 ox amide-I (b-sheet) 1688
1682 red amide-I (b-sheet, loops) 1684
1678 1674 ox m(C¼O) CHO heme a
3
m(C¼O) heme propionates
m(C¼O) Asn/Gln
m(CN
3
H
5
)
as
Arg

heme c
1602 1604 ox m
37
heme a
(m
8a
/
8b
(CC) Tyr-OH)
1608 1610 red m(C¼O) CHO heme a 1606
1597 ox m
37
heme c
1580 1585 ox m
37
heme a
3
1588
1562 1567/1558 ox m
38x
heme a/a
3
m(COO
–
)
as
heme propionate
m(COO
–
)

–
1245 – red m
7¢a
(CO) and d(COH) Tyr-OH
4836 P. Hellwig et al.(Eur. J. Biochem. 269) Ó FEBS 2002
during catalytic cycle [39]. The contributions of protonated
and ionized carboxylic groups of the heme propionates
for the cytochrome c oxidase from P. denitrificans were
assigned by specific
13
C-labelling of the carboxylic groups
of the four heme propionates and site-directed mutagenesis
in the vicinity of its site [40,41]. A signal at 1676 cm
)1
was attributed to contributions of protonated carboxylic
groups. Difference bands at 1570 cm
)1
and 1538 cm
)1
were
assigned to the m(COO
–
)
as
vibration and at 1380 cm
)1
to the
m(COO
–
)

Interestingly potential titrations of the enzyme show a
slightly different redox-dependent behavior. It may be
suggested that the stronger cooperativity displays the
modulation of the enzyme to the different residues involved.
This is in line with the observation reported previously for
the caa
3
-oxidase from R. marinus and ba
3
-oxidase from
T. thermophilus [9,26]. An influence of the attached heme c
center is less likely on the basis of titrations in the presence
of cyanide.
The electrochemically induced FT-IR difference spectra
also include the contributions of the heme centers c, a and
a
3
. Together with the spectra in the presence of cyanide and
in direct comparison to previous resonance Raman data it
can be concluded that the hemes a and a
3
have a similar
structural environment comparised with bovine heart and
P. denitrificans oxidases [17].
In summary, the caa
3
-cytochrome c oxidase shows the
characteristic complex redox behavior and shows several
structural properties of a typical cytochrome c oxidase.
The presence of the two proton pathways is discussed as

Structure at 2.7 A
˚
resolution of the Paracoccus denitrificans two-
subunit cytochrome c oxidase complexed with an antibody FV
fragment. Proc. Natl Acad. Sci. 94, 10547–10553.
6. Pereira, M.M., Santana, M. & Teixeira, M. (2001) A novel
scenario for the evolution of haem-copper oxygen reductases.
Biochim. Biophys. Acta. 1505, 185–208.
7. Yoshida, T. & Fee, J.A. (1984) Studies on cytochrome c oxidase
activity of the cytochrome c
1
aa
3
complex from Thermus thermo-
philus. J. Biol. Chem. 259, 1031–1036.
8.Mather,M.W.,Springer,P.,Hensel,S.,Buse,G.&Fee,J.A.
(1993) Cytochrome oxidase genes from Thermus thermophilus.
Nucleotide sequence of the fused gene and analysis of the deduced
primary structures for subunits I and III of cytochrome caa
3
.
J. Biol. Chem. 268, 5395–5408.
9. Pereira, M.M., Verkhovskaya, M.L., Teixeira, M. & Verkhovsky,
M.I. (2000) The caa
3
terminal oxidase of Rhodothermus marinus
lacking the key glutamate of the
D
-channel is a proton pump.
Biochemistry 39, 6336–6340.

15. Soulimane,T.,vonWalter,M.,Hof,P.,Than,M.E.,Huber,R.&
Buse, G. (1997) Cytochrome-c552 from Thermus thermophilus:a
functional and crystallographic investigation. Biochem. Biophys.
Res. Comm. 237, 572–576.
16. Yoshida, T. & Fee, J.A. (1985) Potentiometric study of cyto-
chrome c
1
aa
3
from Thermus thermophilus. J. Inorg. Biochim. 23,
279–288.
17. Gerscher, S., Hildebrandt, P., Soulimane, T. & Buse, G. (1998)
Resonance Raman spectroscopic study of the caa
3
oxidase from
Thermus thermophilus. Biospectroscopy 4, 365–377.
Ó FEBS 2002 Characterization of T. thermophilus caa
3
oxidase (Eur. J. Biochem. 269) 4837
18. Moss, D.A., Nabedryk, E., Breton, J. & Ma
¨
ntele, W. (1990)
Redox-linked conformational changes in proteins detected by a
combination of infrared spectroscopy and protein electro-
chemistry. Evaluation of the technique with cytochrome c. Eur. J.
Biochem. 187, 565–572.
19. Hellwig, P., Behr, J., Ostermeier, C., Richter, O.M., Pfitzner, U.,
Odenwald, A., Ludwig, B., Michel, H. & Ma
¨
ntele, W. (1998)

26. Hellwig, P., Soulimane, T., Buse, G. & Ma
¨
ntele, W. (1999) Elec-
trochemical, FTIR, and UV/VIS spectroscopic properties of the
ba
3
oxidase from Thermus thermophilus. Biochemistry 38, 9648–
9658.
27. Babcock, G.T. (1988) Raman scattering by cytochrome oxidase
and by heme a model compounds. In Biological Applications of
Resonance Raman Spectroscopy (T.Spiro,ed.),pp.294.Wileyand
Sons, New York.
28. Harmon, P.A., Hendler, R.W. & Levin, I.W. (1994) Resonance
Raman and optical spectroscopic monitoring of heme a redox
states in cytochrome c oxidase during potentiometric titrations.
Biochemistry 33, 699–707.
29. Ching, Y.C., Argade, P.V. & Rousseau, D.L. (1985) Resonance
raman spectra of CN-bound cytochrome oxidase: spectral isola-
tion of cytochromes a2+, a3(2+), and a3(2+)(CN-). Biochem-
istry. 27, 4938–4946.
30. Hellwig, P., Grzybek, S., Behr, J., Ludwig, B., Michel, H. &
Ma
¨
ntele, W. (1999) Electrochemical and ultraviolet/visible/infra-
red spectroscopic analysis of heme a and a
3
redox reactions in the
cytochrome c from Paracoccus denitrificans:separationofhemea
and a
3

36. Yamazaki, Y., Kandori, H. & Mogi, T. (1999) Effects of subunit I
mutations on redox-linked conformational changes of the
Escherichia coli bo-type ubiquinol oxidase revealed by Fourier-
transform infrared spectroscopy. J. Biochem. (Tokyo) 126,
194–199.
37. Venyaminov, S.Y. & Kalnin, N.N. (1990) Quantitative IR spec-
trophotometry of peptide compounds in water (H
2
O) solutions. I.
Spectral parameters of amino acid residue absorption bands.
Biopolymers 30, 1243–1257.
38. Hienerwadel, R., Boussac, A., Breton, J., Diner, B. & Berthomieu,
C. (1997) Fourier transform infrared difference spectroscopy of
photosystem II tyrosine D using site-directed mutagenesis and
specific isotope labeling. Biochemistry 36, 14712–14723.
39. Michel, H., Behr. J., Harrenga, A. & Kannt, A. (1998) cytochrome
c oxidase: structure and spectroscopy. Ann. Rev. Biophys. Biomol.
Struct. 27, 329–356.
40. Behr, J., Hellwig, P., Ma
¨
ntele, W. & Michel, H. (1998) Redox
dependent changes at the heme propionates in cytochrome c oxi-
dase from Paracoccus denitrificans: direct evidence from FTIR
difference spectroscopy in combination with heme propionate
13
C
labeling. Biochemistry 37, 7400–7406.
41. Behr, J., Michel, H., Ma
¨
ntele, W. & Hellwig, P. (2000) Functional