THE JOURNAL OF BHXLXICAL

CHEMISTRY
Vol. 254, No. 22, Issue of November 25, pp. 11288-11293,
Printed m USA.

Potentiometric
Absorbance

1379

Analysis
of Escherichia
coli Cytochromes
Range of 500 nm to 700 nm*

in the Optical

(Received for publication, April 2, 1979, and in revised

Richard

May

29, 1979)

W. Hendler

From the Laboratory

Maryland
20014

Richard

form,

of Cell Biology,

National

Heart,

Lung,

and Blood

Institute,

National

Institutes

of Health,

Bethesda,

I. Shrager

From the Laboratory

Maryland
20014

of Statistical

and Mathematical

Methodology,

The oxidation-reduction
potentials
of respiratory
components can reveal the magnitude of energy liberation
accompanying electron transfer, the most likely sequential reaction
order of the components, the existence of multiple components
having indistinguishable
optical properties,
the number of
electrons transferred, and the possible existence of energized
members arising during the electron transfer process. Because
of the importance
of obtaining accurate oxidation-reduction
potential data for respiratory components we have devoted a
considerable effort to improving the reliability of the collection
of such information.
We have found that many of the assumptions used in the past to obtain oxidation-reduction
potentials
are unjustified
and that some of the techniques are prone to
yielding unreliable
information.
This paper presents a new
approach to the electrochemistry
and data analysis for oxidation-reduction
potentials and presents results for the electron transport chain of Escherichia
cd.
EXPERIMENTAL

PROCEDURES

Materials-The
mediators
used in this work were: potassium
ferricyanide,
Merck and Co., Rahway, N. J. (I?,,, = 435 mV); quinhydrone, Fisher Scientific Co., Fair Lawn, N. J. (E,,, = 280 mV); 1,2naphthoquinone
(B,,, = 143 mV) and pyocyanine perchlorate (IX,,,
= -34 mV), K and K Laborabories,
Plainview,
N. J.; phenazine
methosulfate,
Calbiochem,
LaJolla,
Calif. (E,
= 80 mV); and 2hydroxy-1,4-naphthoquinone,
Eastman
Organic
Chemicals,
Rochester, N. Y. (E,n = -145 mV). The mediator
solutions
were freshly
prepared
for each experiment
in stock solutions
of 12 mM for potas* The costs of publication
of this article were defrayed
in part by
the payment
of page charges. This article must therefore
be hereby
marked
aduertisement
in accordance
with 18 U.S.C. Section 1734
solely to indicate
this fact.

National

Institutes

of Health,

Bethesda,

sium ferricyanide
and 6 !nM for each of the other five. E. coli
membranes
(T-fraction)
were prepared
as previously
described
(1).
Protein
concentration
was determined
by the Lowry
procedure
with
bovine serum albumin as standard (2). Mediators f E. coli cell
membranes
were placed in 125 mM KCl, 62.5 mM potassium
phosphate
at pH 7.0 for analysis.
Equipment
and Procedures-The
equipment
and procedures
have
been previously
described
(3-5). Analog signals for electrode
voltages
and optical density were passed through
20.Hz low pass active Butterworth
filters
(Frequency
Devices,
Inc., Haverhill,
Mass.) before
amplification
to remove unwanted
components,
particularly
at 60 Hz.
Additional
details relevant
to the current
studies are as follows. Prior
to the experimental
titration,
the system was cycled through
a preliminary phase of air oxidation
followed
by endogenous
substrate
reduction. Although
equilibration
can be obtained from either direction
(4),
the system returns to equilibrium
much more rapidly after a reductive
pulse than an oxidative
pulse. Therefore,
all titrations
were performed
by fist raising the voltage of the suspension
to the highest value and
then proceeding
to a stepwise reduction.
Optical transmittances
across
a spectrum
of wavelengths
were measured
and conversion
to absorbantes was performed
outside of the spectrophotometer.
In this way,
a series of spectra as a function
of voltage was obtained.
The average
titration
rate was -30 mV/h
and the average
evaporation
rate was
0.03 ml (1% of total volume)/h.
The time to titrate
90% of a single
component
was 4 h divided by the number
of electrons
transferred.
The pH of the titration
mixture
was 7.0 at the start and finish of all
titrations
where the highest voltage obtained
did not exceed 300 mV
(i.e. for cytochrome
bl). When the voltage
was raised through
the
range of cytochrome
d, however,
an acidification
of the medium
occurred.
The pH of the medium
did not change further
through
the
reductive
titration.
The average pH during the titration
of cytochrome
d was about 6.3. Current
studies to be reported
separately
indicate
that the process of acidification
of the external
medium
is related to
the oxidative
titration
of cytochrome
d.
Resolution
of Spectra versus Voltage Data into Individual
Oxidation-Reduction
Components-In
a previous
advancement
in the
treatment
of spectral data, Butler
and Hopkins
(6, 7), and Shipp (8)
used fourth
differences
to sharpen
and separate
clusters
of peaks.
Shipp was thus able to observe several individual
peaks in clusters
that had looked like one peak. Our aim, in contrast,
was the detection
of transitions
with respect to voltage, which was done more simply by
following
the height of the composite
peaks. Since each experiment
involves
many complete
spectra,
any simplification
is desirable.
It
should be noted that a transition
may be hidden by simultaneous
growth
and shrinkage
of two peaks in a cluster. Still, no method
is
without
some shortcoming,
as emphasized
by Butler and Hopkins
in
their discussion
of the fourth difference
method.
In order to isolate spectral changes due to the change of oxidationreduction
state of an oxidation-reduction
component
from those due
to changes of background
shape in our work, either of two analytic
techniques
were used. When a peak was isolated in the sense that the
second derivative
of the peak was of much greater
magnitude
than
those in the nearby
background,
the second derivative
at the wavelength of maximum
absorbance
was determined
by a least squares
procedure.
A parabola
was fit through
seven points composed
of the
peak wavelength
and three evenly spaced points on each side close to

11288

Downloaded from www.jbc.org by guest, on March 19, 2011

The oxidation-reduction
potentials
of Escherichia
coli
cytochromes
have been studied by a recently
described
technique
for automated
electrodic
potentiometry
(Hendler,
R. W., Songco, D., and Clem, T. R. (1977) Anal.
Chem 49,1908-1913;
Hendler,
R. W. (1977) Anal. Chem.
49, 1914-1918),
where entire spectra are recorded
at a
series of solution
potentials.
New techniques
for resolution
of the spectra
uersus voltage
data have been
applied. The results indicate
that a l-electron
transport
chain conducts
electrons
from substrate
to cytochrome
d, which is the cytochrome
oxidase. Cytochrome
d contains several
components
which
appear
to increase
electron
transfer
first to a a-electron
stage and then to
a 4-electron
stage for the final reduction
of a molecule
of oxygen to 2 molecules
of water.

DCRT,

Potentiometric

Analysis

of E. coli Cytochromes
2

the midpoint.
From principles
described
in Chapter
Six of Ralston
(9), one can derive the formula
for this approach
using a central point
and n equally spaced points on each side.

11289
1
PYOCYANINE

METHOSULFATE

Let
xZ = central

wavelength

A = x:, - x,
and
N=2n+Sums not involving

y:

FERRICYANIDE

QUINHYDRONE

S2 = 2A2 i i2
,=I

S, = 2A4 C i4
r=l
and
u = (NS,
Sums

involving

- &)I2
A

y:

Then:
d2y/clx2

= (NS., - S&)/u.

In our three-point
method,
n = 1 was used.
When a peak was not isolated
but was overlapped
by peaks of
similar widths,
the spectrum
was resolved
into individual
Gaussian
components
and a flat base level. It is assumed that peaks may grow
or shrink but may not change midpoint
or half-width.
A good choice
of peaks is one with the fewest number
of Gaussians
and in which the
synthesized
curve fits the experimental
curve throughout
the whole
voltage titration
range.
Once the data have been resolved
into values for the second
derivative
or height of a Gaussian component
as a function
of voltage,
a second procedure
of numeric
analysis involving
computer
fitting, is
employed.
In this stage, the number
of components
obeying
the
Nernst law is determined
as well as their relative
amounts
and the
number
of electrons
transferred
by each (3-5).
RESULTS

Fig. 1 shows absorbance

spectra
for the components
present
in the potentiometric
titration
mixtures
studied
in this work.
The identification
of cytochrome
components
in the E. coli
respiratory
chain
has been traditionally
based
on reduced
versus oxidized
difference
spectra
as shown
in Fig. 2. The
points
of intersection
of the reduced
and oxidized
spectra
are
called isosbestic
points and are believed
to represent
reference
wavelengths
where
the absorbance
is constant
for reduced
and oxidized
conditions.
Quantitative
analysis
for individual
cytochromes
is based on selecting
a peak and a nearby
reference wavelength,
preferably
an isosbestic
point.
For cytochrome
bl(cu), the wavelength
pair 560 nm and 550 nm are
used. For cytochrome
bl (p), 530 nm and 540 nm could be used
and for cytochrome
a,, 596 nm and 585 nm might
be taken.
The difference
spectrum
for cytochrome
d presents
a different
shape than
a true Gaussian
but the peak and trough
wavelengths
at 630 and 650 nm present
the most logical
choice
for
a two-point
analysis.
In the current
approach,
we have not
been limited
to only two wavelengths
for each oxidizing
potential
but instead
have collected
entire
spectra.
Fig. 3 shows
surfaces
generated
by plotting
spectra
as a function
of voltage
for a solution
containing
six mediators
alone
and for one
containing
the mediators
plus a suspension
of E. coli cell
membranes.

OYl.J
1
320

510

t..

700 320
WAVELENGTH

510

700

(nm)

FIG. 1. Optical
components
present
in titration
mixtures.
The top six panels show spectra
for the mediators
used in these
studies.
The symbol
0 denotes
the oxidized
spectrum
and R the
reduced.
Oxidation
and reduction
were accomplished
electrically
in
the case of mediators.
For the E. coli membranes,
air was the oxidant
and sodium dithionite,
the reductant.
Dithionite
contributes
to the
absorbance
at 355 nm. The bottom
line in the top six panels shows
the absorbance
for the assembled cuvette containing
clear buffer prior
to the addition
of mediator.
In the panel showing E. coli membranes,
the bottom line shows the spectrum
for the polypropylene
used as an
optical reference.
Phenazine
methosulfate
was used at two concentrations, 0.1 miw (0, and R,) and 0.04 mrvr (02 and RJ. Potassium
ferricyanide
was present at 0.2 mrvr, quinhydrone
at 0.6 mrvr, and the
other three mediators
at 0.1 mrvr each. The E. coli membranes
were
present at a concentration
of 6.2 mg of protein/ml.
The buffer was
0.125 M KC1 and 0.063 M potassium
phosphate
at pH 7.0. The optical
reference
was water for mediators
alone and was polypropylene
and
frosted glass for light-scattering
suspensions
analyzed
for respiratory
components.
To convert
spectra
referenced
to plastic and glass to
absolute
spectra,
the absorbances
shown in the lower right-hand
panel should be added to the appropriate
relative
spectra. Thus, the
solid line spectrum
was the reference
for the surface shown in Fig. 5,
the short dashed line spectrum
for the surface in Fig. 6, and the long
dashed line for the surface in Fig. 7.

As expected
from
the spectra
shown
in Fig. 1, most of the
features
in the surface
obtained
with
mediators
alone
are
found
in the lower wavelength
regions.
Pyocyanine,
however,
does create
a hump
in the higher
wavelength,
higher
voltage

Downloaded from www.jbc.org by guest, on March 19, 2011

S.I= A21 j'(y, , + y,+,)

,=I

11290

Potentiometric
REDUCED

ABSOLUTE

MINUS

Analysis
OXIDIZED

[-

FIG. 2. Spectra
for E. coli membranes.
The a and /3 absorbances
for the recognized
E. coli cytochromes
are most commonly
seen in
difference
spectra from 500 to 700 nm. This figure shows a typical
difference
spectrum
(rightpanel)
and the original spectra from which
the difference
spectrum
was derived (leftpanel).
Isosbestic
points are
the cross-over
points for the oxidized
(0) and reduced (R) spectra
which show up as 0 AA in the difference
spectrum. The oxidized and

reduced spectra shown here are enlargements of a portion of the

spectra shown in Fig. 1, lower left panel. The divisions shown on the
0 AA line in the right panel are placed at 5-nm intervals.
is most evident in Fig. 3, Panels 6, c,
E. coli membranes markedly alters
voltage region with a massive uniat 370 nm and the Soret absorption
apparent in the surface from E. coli

membranes
is the relative
smallness
(Y and /3 bands
of the cytochromes

of the absorbances
and the fact that

they

of the
are

situated as small features on an extensively slanting and

changing background. These facts are lost in difference spectra.

Another

feature

which

will

become

much

more

obvious

in

subsequent figures is the curvature or hump in the surface

seen from the high wavelength
side especially in Panels 6
and c. The analysis of the spectral features in the part of the
surface containing
the characteristic
absorbances of the cytochromes was complicated
by the presence of pyocyanine,
which is an effective mediator only at relatively high concentrations. Although
pyocyanine
is commonly used, we found
that its replacement by succinate resulted in better mediation
and the total elimination
of the interfering optical properties.
This can be seen in the surfaces of Fig. 4, Panels a to d,
compared to the corresponding
panels in Fig. 3. Although the
hump in the surfaces containing E. coli membranes is also
a to d in the two figures), it is still
reduced (compare Panels
an important
feature as will be seen subsequently.
For the analysis of the bl cytochromes, the spectrum was
started at 490 nm where, because of the lower levels of
FIG. 3. Spectra-voltage
surfaces
for six mediators
f E. coli
membranes.
Each surface is presented
in four aspects. The a panels
show a full face view from a low voltage position.
The axis running
perpendicularly
into the depth of the picture
represents
increasing
voltage and is labeled E. The horizontal
axis seen in full view represents wavelength
and the vertical
axis, absorbance
(A). The a panels
give scaled information
for the surface. For example,
Panel a shows
that the wavelength
range extends from 320 to 695 nm. The divisions
are in steps of 5 nm each. The E-axis extends from -203 mV to 349
mV. The voltage
steps are uneven
and reflect oxidation-reduction
buffering
activity.
Mediators
or oxidation-reduction
buffers are chosen to give closer steps in areas of maximum
interest.
The optical
reference
for mediators
alone (Panels
a to d) was water. The reference for the suspension
of membranes
(Panels
a to d) was polypropylene and frosted
glass. Panel a shows the absorbance
values for
several of the maxima
in the surface to indicate
the range of absorbantes in the figure. Panel a shows that the voltage range covered for
this surface extends from -214 mV to 377 mV. The absorbance
scale
shows AA relative
to the light-scattering
reference
which had an

absorbance and light scattering, the concentration

of E. coli
membranes could be increased from 2 to 3.4 mg/ml and the
level of phenazine methosulfate
increased from 0.04 to 0.1 InM.
Furthermore,
the wavelength resolution was improved from
5 nm to 2 nm and much closer voltage steps were taken. By
these procedures, the cytochrome
bl(a) and bl(/3) features
stand out clgarly from the surface in all four views (Fig. 5). In
spite of the absence of pyocyanine,
pronounced
localized
curvature in the background
surface is seen as indicated by
the asterisks in Fig. 5, Panels a, b, and c. This feature which
is ever present, appears as perhaps a side of an extremely
broad Gaussian component
centered in the 700 to 800 nm
region. It decreases in absorbance with decreasing voltage
with an apparent E, of about 30 mV and an n value of 2
electrons. It was not characterized
further. A bracket is shown
in Panel a to show the peak and isosbestic point traditionally
used for the quantification
of reduced cytochrome
b,(a). Although the completely
oxidized and reduced spectra have
equal absorbances at the isosbestic point of 550 nm (see Fig.
2), the belief that such a point represents a useful reference
wavelength for the assay of reduced cytochrome
bl(a) is seen
in the figure to be unwarranted.
The kinds of surfaces used for analysis of cytochromes
d
and a, are shown in Figs. 6 and 7. Because these spectra were
started at still higher wavelengths than those in Fig. 5, with
consequently
less light scattering, the level of E. coli membranes could be increased to 6.9 mg of protein/ml.
At the
highest voltage, 394 mV, an absorption
peak at 650 nm is
present (Fig. 6, Panel d). With decreasing voltage, this peak
appears to shift its location to lower wavelengths centered at
about 635 nm (Fig. 6c, at the arrow). With still lower voltages
the peak gradually shifts back to a higher wavelength.
An
alternative view, which our analysis favors, is that changes in
individual
peaks centered at 650 and 634 nm account for an
apparent shift in location of a single peak. It can also be
noticed that in the voltage range covered, no absorbance for
cytochrome a, is seen in the wavelength region near 600 nm.
The surface in Fig. 7 overlaps the region shown in Fig. 6 and
extends into a lower voltage region. A cytochrome al absorbance can be discerned in this surface at the lower voltage end,
particularly
in Panels a and b. The broad composite cytochrome d feature extends throughout
the whole voltage range.
Using
titrations

the

same

mediators

as previously

used

in chemical

of cytochrome
b,(a) (4, 5), electrodic titrations revealed the same components, although the relative percent of
each was different in the two preparations
(Table I). Because
of the changing nature of the general absorbance-light-scattering background as voltage is changed (shown in Fig. 5), the
apparent
absorbance
of 1.76 at 320 nm and 0.67 at 695 nm. The b
panels show the surface viewed from the high-wavelength,
low-voltage
corner of the surface. The c panels view the surface from a position
of high wavelength
and high voltage.
The d panels present
the
opposite
face from that in the a panels, viewed from a position
of
relatively
high voltage.
The irregular
or jagged nature
of a surface
seen at high absorbance
values reflects the limitation
of the 12-bit A/
D converter
which admits
4096 possible
transmittance
values.
A
variation
in signal of 0.1% or 4 units represents
an uncertainty
of less
than 0.004 A in the absorbance
range of 0 to 1.0, of 0.043 A at 2.0 A
and greater than 0.3 A at 3.0 A. For this reason, quantitative
analysis
of spectral surfaces is done with data collected in the lower absorbance
ranges. Concentrations
for the mediators
were those shown in Fig. 1
with phenazine
methosulfate
at the lower value. E. coli membranes
were present at 2 mg of protein/ml.
The symbols, ?, S, b,(p), bl(a), aI,
and d refer, respectively,
to components
contributed
by the E. coli
membranes,
namely unknown,
Soret, cytochromes
bl(/3), bl(cu), al,
and d.

Downloaded from www.jbc.org by guest, on March 19, 2011

part of the surface. This

and d. The presence of
the low wavelength,
low
dentified feature centered
at 430 nm. What is most

of E. coli Cytochromes

Potentiometric

-Ecoli

Analysis

MEMBRANES

of E.

coli Cytochromes

11291

+ Ecoli MEMBRANES

-214
WAVELENGTH

hn)

695

Downloaded from www.jbc.org by guest, on March 19, 2011

11292

Potentiometric

of E. coli Cytochromes

Analysis

--E. co/i MEMBRANES

i-ho/i

MEMBRANES

a
1.18
.77

0, '
ii0

WAVELENGTH

(nm)

695
b
Downloaded from www.jbc.org by guest, on March 19, 2011

Fro.
general

4. Spectra-voltage
surfaces
information
and concentrations.

for

five

mediators

(i.e. minus

two-point
method
is not reliable.
A three-point
method
leads
to a dramatic
change
in the resolution
of the E, values
of the
components.
The two higher
E,, values are markedly
reduced
and the relative
amounts
of the highest
and lower
E,,, com-

pyocyanine)

f E. coli

membranes.

Refer

to the legend

of Fig. 3 for

ponents
are greatly
altered.
The second
derivative
method
yields
results
in close agreement
to those of the three-point
method.
When
pyocyanine
is replaced
by succinate
as mediator, the electrodic
titration,
even by the two-point
methods,

Potentiometric

01
490

WAVELENGTH

Analysis

11293

of E. coli Cytochromes

630

Cnm)

and /3).
at the
was at
E. coli
0.1 mM.

Wavelength
steps were at 2 nm each. The optical
reference
was
polypropylene
and frosted glass as shown in Fig. 1. Optical features
for cytochromes
b,(a) and b,(P) are indicated
by arrows. A prominent
curvature
of the surface is indicated
by *. Further
details are presented in Fig. 3.

I
680

WAVELENGTH

(nm)

694

FIG. 6. Spectra-voltage
surface
for cytochromes
d. Four of the six mediators
(-pyocyanine
and 2-hydroxy-1,4-naphthoquinonej
-were
present.
Phenazine
methosulfate
and 1,2-naphthoquinone
were at 0.1 mM, ferricyanide
was at 0.2 mM, quinhydrone
at 0.6 mM, and E. coli
membranes
at 6.9 mg of protein/ml.
Wavelength
steps were at 2 nm. The optical reference
was polypropylene
and frosted glass as shown in
Fig. 1. The arrows indicate
features
discussed in the text. Further
information
is provided
in Fig. 3.

Downloaded from www.jbc.org by guest, on March 19, 2011

FIG. 5. Spectra-voltage
surface
for cytochromes
bl (a
Five of the six mediators
(--pyocyanine)
were
present
concentrations
shown in Fig. 1 except for quinhydrone
which
0.4 mM. Phenazine
methosulfate
was present at 0.1 mM, and
membranes
at 3.4 mg of protein/ml.
Succinate
was present at

Potentiometric Analysis of E. coli Cytochromes

11294

WAVELENGTH

576

694

hm)

steps were 2 nm. The optical reference

was polypropylene
and frosted
glass as shown in Fig. 1. The symbols a, and d are used to indicate
the location
of absorbances
for cytochromes
aI and d. Further
information is provided
in Fig. 3.

TABLE
Cytochrome
2-point
Err,

Chemical
titration
(+ pyocyanine)
Refs. 4, 5
No. of experiments
Electrodic
titration
(+ pyocyanine)

bi (4

fit
%

3.point
E,

fit

2nd

222 + 5.3
107 + 3.7
-47 t 5.1

39 f 1.2
33 + 1.9
28 + 1.1

214 f 17.9
118 -c 1.5
-39 -e 15.6

23 + 5.8
37 -L 6.2
40 f 1.7

161 + 15.6
66 + 9.9
-37 -e 7.5

40 +- 7.7
38 -c 0.9
21 f 5.7

180 + 3.3
49 -t 4.0
-88 + 5.7

30 + 1.2
35 + 5.3
35 f 4.3

182 -I 3.7
54 + 2.8
-108 + 14.4

25 k 2.0
55 k 3.7
20 + 1.7

derivative

E,

fit
%

(14)

No. of experiments
Electrodic
titration
(no pyocyanine)

(3)

No. of experiments

(3)

Electrodic
titration
(no pyocyanine)
No. of experiments

bl components

Not done
(3)

n E, values are in millivolts

fit, 570 nm was also used. The
b The I? shown above and
whether
the higher and lower

157 + 15.3
63 f 9.7
-35 -e 7.3
186 + 3.0
57 c 4.7
-105 f 16

?h
60 +- 6.7
?h

41 + 7.8
38 + 7.5
22 + 1.3
24 +- 1.2
60 f 5.7
16 + 4.5

(70-100)

f SE. For the 2-point fit, the peak at 560 nm and the isosbestic
point
at 550 nm were used. For the 3-point
2nd derivative
method was based on seven points as described
under Experimental
Procedures.
below the E, of 60 mV for cytochrome
b,(P)
indicates
that there was insufficient
absorbance
to determine
E, components
seen for cytochrome
bl(a)
were present.

yields results more like those of the three-point

and second
derivative fits obtained in the presence of pyocyanine. The
lowest E, component,
however, is very difficult to mediate
fully. In a previous publication
using electrodic titrations
(4),
the lowest E, component found by the two-point
method in
the presence of pyocyanine was -72 mV and in the absence of
pyocyanine, it was -34 mV. The two higher E, components
were not significantly
affected by omitting pyocyanine. In the
current data, two of the three titrations in the presence of
pyocyanine
were at a concentration
ratio of 8.7 nmol of

pyocyanine/mg
of protein and yielded E, values of -17 mV
and -30 mV. The third had 29 nmol of pyocyanine/mg
of
protein and yielded an E, value of -69 mV. Therefore, the
apparent difference for the lowest E, component seen in the
titrations where pyocyanine was replaced by succinate is due
to the superiority of succinate as a mediator for this component. Because of this and because pyocyanine
contributes
appreciable
absorbance in the regions of interest, the data
obtained in the absence of pyocyanine
are deemed more
reliable. Fig. 8a shows b,(a) species for a succinate-mediated

Downloaded from www.jbc.org by guest, on March 19, 2011

FIG. 7. Spectra-voltage
surface
for cytochromes
d and aI.
Five of the six mediators
(-pyocyanine)
were present at the concentrations shown in Fig. 1 except for quinhydrone
which was at 0.5 mM.
Phenazine
methosulfate
was present at 0.1 mrvr and E. coli membranes
at 6.9 mg of protein/ml.
Succinate
was present at 0.1 mM. Wavelength

Potentiometric
CYTOCHROME

Analysis

of E. coli Cytochromes

11295

CYTOCHROME

bl ALPHA

bl BETA

.001232

.001232

- .0002549.

# - .000254-9,a

-190

55

300

55

300

MILLIVOLTS

FIG. 8. Cytochromes
ysis. For each spectrum

bl titrations

by second derivative

anal-

at a different
voltage
such as shown in Fig.
5, the second derivative
at 560 nm (cytochrome
b,(a)) and at 530 nm
(cytochrome
b,(p))
were determined
by the least squares procedure
using seven points for each determination.
The negative
of the second

Absorbance
cytochrome

TABLE
values
Protein

Mediators

Total A

ma/d
3.4
3.4
6.8
6.8
6.8

;ii

bl(/3)

0.294

1.16

560
600
634
650

h(a)
a:
da4

0.240
0.200
0.184
0.178

1.01
1.19

d,so

II
of components

1.13
1.10

Due to cytochrome
AA

1.45
1.25
1.39
1.31
1.28

0.009
0.046
0.006
0.017
0.015

% total
0.62
3.7
0.43
1.3
1.2

-l
Q
.0X62

.01432

by
97
MILLIVOLTS

FIG. 9. Cytochrome d titration by two-point method. Spectra

such as those comprising
the surfaces shown in Figs. 6 and 7 were
used to determine
the AA for 630 nm minus 650 nm and these AA
plotted
as a function
of voltage.
The open symbols were obtained
from experiments
in the absence of pyocyanine.
The + symbols are
from an experiment
in the presence of pyocyanine.

A full resolution of the spectra versus voltage surface into

background
and Gaussian components, leads to a more complete understanding
of the changes in absorbance as influenced by changes in the oxidizing potential of the medium.
The overall background is fit by two broad Gaussians centered
at 500 nm and 700 nm and a flat base level which rises and
falls. In addition, two major Gaussian components centered at
634 nm and 650 nm and up to three minor Gaussian components at 590,660, and 680 nm account for the spectra observed
from 570 nm to 695 nm. The absorbance at 590 nm accounts
for the aI component observed in difference spectra of the E.
coli respiratory
chain. The ability of these resolved components to account for the spectral features observed in the
experimental
system is shown in Fig. 10 for several different
voltages. The component centered at 634 nm in the absorption
spectrum was fit by a 4-electron transfer (Fig. 11). The fits
based on l- or 2-electron transfers were quite poor. When the

Downloaded from www.jbc.org by guest, on March 19, 2011

experiment as extracted from spectra by the second derivative

approach and fit by one, two, and three components.
In
consideration
of all of the data obtained for cytochromes bl(o),
we believe that three components are present and that within
experimental
error of about -+lO mV, their E, values are 186,
57, and -105 mV.
The total amount of change of absorbance due to titration
of the component
known as cytochrome
I%(/?) is only 0.009
unit on a starting background absorbance of about 1.45 (Table
II). This is in contrast to about 5 times as much absorbance
due to cytochromes
&(a). Therefore
the analysis of cytochrome bl(P) is subject to relatively much more noise. Fig. 8b
shows data for cytochrome
&(/I) fit to one, two, and three
components. We are most certain that the middle E, component seen for cytochrome bl(a) is present and accounts for
at least two-thirds
of the total signal. Because of the relative
smallness in the amounts of signal expected from the highest
and lowest E, components
seen with cytochrome
&(a), the
amount of signal expected from these two components in bl(p)
is too small to be resolved. However, because of the correspondence of the middle E, component in the (Yand /I species
and because both the cytochrome bl(a) and cytochrome bl(/?)
absorbances were present in isolated, purified cytochrome
61
(lo), we feel that our data further establish that both absorbantes do originate
from cytochrome
bl. Another
point of
interest shown in the data of Table I is the fact that the
relative amount of the lowest E, species of cytochrome bl(oc)
markedly decreases as the number of points in the analysis
increase. This illustrates
the contribution
of background
changes in the two-point
analysis.
The major two wavelengths which appear to be associated
with cytochrome
d in the usual difference spectrum of the
respiratory
components
are 630 nm (peak) and 650 nm
(trough) (Fig. 2). When the difference in absorbance for these
two wavelengths is plotted against the voltage of the medium,
a rapid rise is seen with decreasing voltage from voltages
above 390 mV down to about 312 mV (Fig. 9a). The difference
in absorbance then drops to a low at about 50 mV and then
starts to rise again with further reduction of voltages (Fig. 96).
According to traditional
considerations,
this behavior could
be taken to indicate that the midpoint potential of cytochrome
d is under the influence
of the oxidation-reduction
state of
other components of the respiratory
chain so that upon reduction, the midpoint
potential of cytochrome
d is lowered
sufficiently to cause its subsequent reoxidation
and corresponding lessening of the 630 nm minus 650 nm AA.

derivative
is plotted as a function
of voltage with the points showing
the experimentally
determined
values. The computer
best fit is shown
for a single component
by a long dashed line, for two components
by
a short dashed line, and for three components
by a solid line. The
results of such analysis are shown in Tables I and III.

Potentiometric Analysis of E. coli Cytochromes

11296

311 mV
transition

Hf.
.02728

01301 1
100

139 mV

250

400

-150

75

MILLIVOLTS

WAVELENGTH

(nml

.03214

7-p

.45,

15

P
CI ID

Ht.

.02081

WI
WAVELENGTH

~ i
2%

325

400
MILLIVOLTS

4cxl

FIG. 11. Cytochrome

d (634) titration
by Gaussian
analysis.
Spectra such as those comprising
the surface in Fig. 6 were resolved
into Gaussian
components
plus a base level to reconstitute
experimentally
derived
curves (see Fig. 10). The heights of the component
centered
at 634 nm, in absorbance
units, plotted
as a function
of
voltage, are shown as open symbols. The computer
best fit for a single
component
transferring
1 electron
is shown in Panel a by a long
dashed line, a 2-electron
transfer
is represented
by a short dashed
line, and a 4-electron
transfer
by a solid line. The fit shown in Panel
b uses a minor 2-electron
component
at E, = 282 mV and a major 4.
electron
component
at E, = 310 mV.

computer was asked to fit best the number of electrons transn = 4.3. The
ferred with the initialization
at n = 2, it selected
root mean square error was essentially equal for a fit with n
equal to either 4.0 or 4.3. In most cases (i.e. 4 out of 5) the
fitting of the component at 634 nm was slightly improved by
including a minor 2-electron transfer component with an E,
value about 20 mV lower than the main 4-electron transfer
component (Fig. llb). The component absorbing at 650 nm
responded in a very complicated
way to changes in voltage
(Fig. 12). It was extremely difficult to obtain a convergence
for a fit that would accommodate
the dip occurring at about
326 mV. The best fit in all cases for the voltage region of about
150 to 400 mV was obtained with four components transferring
1 electron, 2 electrons, 4 electrons, and 4 electrons. The voltage

6i20

.05

(nm)

FIG. 13. Enhancement

of apparent
cytochrome
al absorbance by difference
spectroscopy.
The curves labeled 0 and R are
absorption
spectra for oxidized
and reduced
samples of a suspension
of E. coli membranes
at 6.8 mg of protein/ml.
The large dots
represent
the derived difference
spectrum.
The scale for AA is twice
that of A.

region from 150 mV down to about -100 mV required an

additional
l-electron
component.
These components
along
with the resolved cytochrome
b, components
are listed in
Table III.
The component
known as cytochrome
ai from reduced
versus
oxidized difference spectra contributes only about 0.006
A to a background
of about 1.4 units (Table II) and is therefore
very difficult to characterize
accurately. Resolution
of this
component
by both Gaussian and second derivative
techniques was carried out. Greater confidence is attached to the
second derivative fits even though the scatter is higher. In two
experiments, the Gaussian technique gave an E, of -9 f 20
mV; the second derivative technique gave an E, of 25 + 5
mV. The component
called cytochrome
aI is artifactually
accentuated in a difference spectrum as shown in Fig. 13. The
oxidized background
spectrum has a concavity at about 588
nm to 602 nm. This augments the initial rise in the cytochrome
a, component seen in the difference spectrum. It also adds
absorbance to that of the real but smaller component
that
arises upon reduction. The right side of the difference spec-

Downloaded from www.jbc.org by guest, on March 19, 2011

FIG. 10. Experimental

and reconstituted
spectra
for cytochrome
d analysis.
Spectra such as those comprising
the surface in
Fig. 6 were resolved
by Gaussian
analysis into a flat base-line
level,
two broad Gaussian
components
centered
at 500 and 700 nm, one
minor component
at 680 nm, and the cytochrome
d components
at
634 and 650 nm. Reconstituted
spectra were constructed
by recombining the absorbances
of the resolved components.
The figure shows
the superposition
of experimental
and reconstituted
curves at four
critical voltages.
At 383 mV, a component
at 650 nm is present but
the major increase in absorbance
of the component
at 634 nm has not
occurred
(refer to Figs. 11 and 12). At 311 mV the transition
from a
major absorbance
at 650 nm to one at 634 nm is occurring.
At 278
mV, there has been a recovery
of the absorbance
at 650 nm. At 139
mV, still more of the 650 nm component is present.

FIG. 12. Cytochromes

d (650) titration
by Gaussian
analysis.
By the same procedure
described
in Fig. 11, the heights
of the
Gaussian
component
centered
at 650 nm are plotted
as a function
of
voltage. The points in Panel a were obtained
from a surface like the
one shown in Fig. 6 and those in Panel b from a surface like the one
in Fig. 7. The solid line on Panel a was the computer
best fit for four
components
as follows: n = 1, E, = 187; n = 2, E, = 268; n = 4, E,,
= 309; n = 4, E, = 311. The dashed line includes four components
as
follows: n = 1, E, = 189; n = 2, Em = 265; n = 4, E, = 309; n = 8, E,
= 317. The b panel extends the analysis to lower voltages where an
additional
component
with n = 1 and E, = 28 is indicated.

Potentiometric

Analysis

of E. coli Cytochromes

11297

TABLE
III
Resolved
cytochromes
For the b, cytochromes,
the traditional
designations
of a and p
chrome d components
as discussed more fully in the text. Columns
3
have been retained.
For the d cytochromes,
two classes have been
and 4 show the number
of electrons
transferred
by the component
identified,
one absorbing
at 634 nm and the other at 650 nm. To
and whether
the individual
absorbance
is associated
with the reduced
or oxidized
state. Amounts
are in units which reflect the resolution
distinguish
multiple
components
among the individual
cytochrome
classes, superscripts
are used with the numbering
proceeding
from
technique
used. For the second derivative
technique,
this is absorbthe highest E,, species to the lowest. The cytochromes
listed in Group
ante units per nm of wavelength
squared at the peak wavelength.
For
the Gaussian technique
it is absorbance
units at the peak wavelength.
1 have been resolved
from the cytochromes
providing
the major
optical absorbance
and represent
solutions
to which we attach greater
The values following
-c are standard
errors of the mean which were
confidence.
Group 2 contains
components
providing
a lesser amount
obtained
for three experiments
in the case of cytochrome
bl and five
of absorbance
and which therefore
are resolved
with somewhat
less
experiments
in the case of d.
confidence.
Group 3 contains
an alternative
resolution
for the cyto-

Cytochrome

FOrin
R = reduced
0 = oxidized

No. of electrons

lLrz

Group
+
k
+
f
+
f
-c
&

3
5
16
5
5
5
5
7

1
1
1
4
4
4
2
1

$634)
d(650)

60
25
302
24

rf:
f
+
f

6.7
5
10
4

1
1
2
1

d(634)
d(634)
d(650)
d(650)
d(650)3
dC65014

329
313
330
320
278
195

+ 5
I!z 5
+ 5
+ 5
+ 5
-I 7

Group
b,(P)

Group
8
4
8
4
2
1

of the

R
R
R
R
0
R
R
R

0.316 x lo-
0.769 x lo-
0.204 x lo-
0.017
0.132
0.136
0.0054
0.0044

rt
rt
+
f
+
-t
*
f

0.031 x lo-
0.086 x lo-
0.060 x lo-
0.0014
0.009
0.009
0.0012
0.0014

R
R
R
0

0.23 X lo- C 0.025 x lo-

0.095 x lo- + 0.001 x 1o-3
0.0024 f 0.0010
0.0084 + 0.0004

2D
2D
2D
G
G
G
G
G

2
2D
2D
G
G

CYTOCHROME

0.0124
0.0061
0.0183
0.0227
0.0057
0.0044

OXIDASE

+
f
f
+
+
f

0.0011
0.0015
0.0008
0.0018
0.0012
0.0014

G
G
G
G
G
G

SCHEME

trum for cytochrome a1 is accentuated both by the concavity

of the oxidized spectrum and by the appearance of the reduced
absorbing component
at 634 nm which emphasizes the minimum to the right of the cytochrome ai peak in the difference
because

R
R
0
R
R
R

FIG. 14. Possible

sequence
for E.
coli respiratory
chain of cytochrome
components.
The
scheme
is constructed
from the components
deduced
in the current
study assuming
a linear
order arranged
in increasing
values of
the midpoint
potentials.
Cytochrome
designations
are by lower
case letters
with identifying
subscripts,
superscripts,
and wavelengths
where
needed.
RED
and OX designate
whether
the major
absorbance
is associated
with the reduced or oxidized
forms
of the cytochrome.
Evidence
for the branching
is
based on the fitted n values for number
of electrons
transferred.

spectrum. Furthermore,

Amount

small

magnitude

of

Downloaded from www.jbc.org by guest, on March 19, 2011

186
57
-105
323
324
322
276
196

bl(d
bl(d
bl(d
d(634)
d(650)
d(650)
d(650)3
d(650)4

Resolution tech- nique

2D = second derivative
G = Gaussian

0,
+
4e
+
4H

*ti,o
I

AA compared to the individual

absorbances,
an expanded
scale is used, which further magnifies the peak.
If the components of Groups 1 of Table III plus the traditionally recognized cytochrome
al are arranged in a linear
sequence of increasing E, values, a possible electron transport

Potentiometric

11298

of E. coli Cytochromes

Analysis
OX

d16501

4e
P
RED

dt6501

le

*I

REDd,66,,,

( flEDd(6501

20,

8e

FIG.
oxidase

15. Alternative
scheme.
This

cytochromes
scheme for cyconstructed in acthat an n value of

tochrome oxidase is
cord with the finding
8 also fits the data for cytochrome d
components centered at 634 and 650 nm.

RED
mv

+ 330

chain for E. coli is constructed

as shown in Fig. 14. The
components with E,, values below 200 mV form a l-electron
transport chain. The components of cytochrome d (i.e. cytochrome oxidase), however, multiplex
in two stages up to 4electron transfers, which is the number required to reduce a
molecule of oxygen. Whether all of the recognized components
are acting in a single electron transport chain and whether
they are arranged strictly in accord with their E, values has
not been established in the current work. Group 3 in the table
has been arrived at by setting the parameter for number of
electrons transferred
in the steepest slope regions to eight.
The closeness of fits as indicated by root mean square deviations was noticeably better in these cases than when n was set
to 4. An alternative
cytochrome
oxidase sequence based on
these resolutions is shown in Fig. 15.
DISCUSSION

We have found in these studies that many of the techniques

and basic assumptions commonly used to study quantitative
changes in cytochromes in membrane suspensions can lead to
serious errors. The quantification
for individual
cytochromes
is based on split beam spectrophotometry
and dual wavelength spectrophotometry
of light-scattering
suspensions, using a peak and reference wavelength.
We have defined and
discussed in detail four situations under normal usage of these
techniques which lead to nonlinearity
of response (11).
In using the AA between two characteristic
wavelengths to
quantify a given cytochrome it is assumed that: 1) The background changes between these two wavelengths are negligible
compared to the changes for the cytochrome;
2) reference
points defined by those wavelengths where a totally reduced
spectrum
crosses a totally
oxidized
spectrum
represent
isosbestic points where the absorbance is constant regardless
of the oxidation-reduction
state of the system; 3) characteristic
optical absorbance peaks for a particular
cytochrome always
increase on reduction so that a reduced minus oxidized difference spectrum shows the reduced spectrum for all of the
cytochromes
with the light-scattering
background
canceled
out.
We have found that these assumptions do not necessarily
hold. The absorbance background
is not relatively flat and

d163-41
mv

does not remain at a constant height nor retain the same

shape throughout
a range of different oxidation-reduction
potentials. Therefore, following absorbance at a peak wavelength or AA between a peak and reference wavelength does
not provide reliable data on changes of oxidation-reduction
state of a component of interest. We have found further that
the oxidized form of a cytochrome
can display prominent
absorbance features which occur at different wavelengths and
have different widths than the reduced absorbance
peaks.
Therefore,
a reduced minus oxidized difference spectrum
shows a composite of new features arising on reduction augmented by changes due to the subtraction
of peaks and
troughs present in the oxidized spectrum.
In our work, the two-point analysis of cytochrome bl led to
serious errors in the determination
of E, values for the species
present. We feel that the current values based on determining
the second derivative at the peak, from a seven-point analysis,
provide much more dependable
data. In the case of cytochrome al, the subtraction
of an oxidized spectrum and the
magnification
of the difference spectrum augment the apparent size of this feature, which upon analysis by either the
Gaussian or second derivative
approach represents an extremely small absorbance peak. The most interesting findings
in the current study apply to the cytochrome
oxidase (i.e.
cytochrome d) region of the spectrum. The two-point analysis
leads to a totally incomprehensible
response of the cytochrome to changes in oxidation-reduction
potential
of the
medium.
The cytochrome d absorbance feature in the three-dimensional surface contour displayed most unconventional
behavior. It seemed to shift its location from 650 nm to about 635
nm and then back again to about 645 nm. In fitting the data
to behavior of oxidation-reduction
entities we had to consider
the possibility of whether a single chromophore
could display
such erratic behavior as a function of its oxidation-reduction
state. We rejected this hypothesis on two counts. Firstly, there
should be two separate absorbance peaks for oxidized and
reduced heme centers and secondly, when computer fits were
obtained based on a one-chromophore
model, they were decidedly inferior to the fits based on the existence of two fixed
chromophore
centers at 634 nm and 650 nm. Further detailed

Downloaded from www.jbc.org by guest, on March 19, 2011

+ 320

Potentiometric

Analysis

11299

a coordination
of several molecules of cytochrome oxidase can
be accomplished,
the possibility that 2 molecules of oxygen
can be reduced at one time cannot be completely eliminated.
Whether all of the major components revealed in this work
participate
in an electron transport chain passing electrons
from a substrate to oxygen has not been established nor has
the actual biological
sequence of their operation.
In subsequent work using a rapid scanning spectrometer
plus natural
biological electron donor substrates these questions will be
studied in mitochondrial
systems as well as in E. coli.
Acknowledgments-We
appreciate and acknowledge
the continued
support of David Songco of the Computer Systems Laboratory and
Thomas
Branch.

R. Clem of the Biomedical

Engineering

and Instrumentation

REFERENCES
1. Hendler,

R. W., and Burgess,

A. H. (1972)

J. Cell Biol.

55, 266-

281
2. Lowry,

0. H., Rosebrough,
N. J., Farr, A. L., and Randall,
R. J.
(1951) J. Biol. Chem. 193,265-275
3. Hendler,
R. W., Songco, D., and Clem, T. R. (1977) Anal. Chem.
49, 1908-1913
4. Hendler, R. W. (1977) Anal. Chem. 49, 1914-1918
5. Hendler, R. W., Towne, D. W., and Shrager, R. I. (1975) Biochim.
Biophys.
Acta 376, 42-62
6. Butler, W. L., and Hopkins,
D. W. (1970) Pkotochem.
Photobiol.
12,451-456
7. Butler, W. L., and Hopkins, D. W. (1970) Photochem.
Photobiol.

12,439-450
8. Shipp, W. S. (1972) Arch. Biochem.
Biophys.
150,459-472
9. Ralston,
A. (1965) A First Course in Numerical
Analysis,
pp.
228-270, McGraw-Hill,
N. Y.
10. Deeb, S. S., and Hager, L. P. (1964) J. Biol. Chem. 239, 10241031
11. Hendler,
R. W. (1979) Anal. Biochem.
94,450-464
12. Caughey,
W. S., Wallace,
W. J., Volpe, J. A., and Yoshikawa,
S.
(1976) in The Enzymes
(Boyer,
P. D., ed), pp. 299-344,
Academic Press, N. Y.
13. George, P. (1965) in Oxidases
and Related Redox Systems (King,
T. E., Mason,
H. S., and Morrison,
M., eds) pp. 3-33, Wiley,
N. Y.
14. van Gelder, B. F., and Beinert,
H. (1969) Biochim.
Biophys.
Acta

189, l-24
15. Babcock,
G. T., Vickery,
L. E., and Palmer,
G. (1978) J. Biol.
Chem. 251,7907-7919
16. Babcock,
G. T., Vickery,
L. E., and Palmer,
G. (1978) J. Biol.
Chem. 253, 2400-2411
17. Tweedle,
M. F., Wilson, L. J., Garcia-Ifiiquez,
L., Babcock,
G. T.,
and Palmer,
G. (1978) J. Biol. Chem. 253,8065-8071
18. Moss, T. H., Shapiro,
E., King, T. E., Beinert,
H., and Hartzell,
C.
(1978) J. Biol. Chem. 253,8072-8073
19. Lindsay,
J. G. (1974) Arch. Biochem.
Biophys.
163, 705-715
20. Lindsav.
J. G.. and Wilson. D. F. (1974) FEBS Lett. 48.45-49
21. Wilson; D. F.,and Dutton;
P. L. (197Oj Arch. Biochem:
Biophys.

136,583-584
22. Anderson,
J. L., Kuwana,
T., and HartzelI,
C. R. (1976) Biochemistry 15,3947-3955
23. Heineman,
W. R., Kuwana,
T., and Hartzell,
C. R. (1972) Biothem. Biophys.
Res. Commun.
49, l-8
24. Chance, B., Saronio, C., and Leigh, J. S., Jr. (1975) J. Biol. Chem.
250,9226-9237
25. Palmer,
G., Babcock,
G. T., and Vickery,
L. P. (1976) Proc. Nutl.
Acad. Sci. U. S. A. 73,2206-2210
26. Lanne, B., and Vanngkd, T. (1978) Biochim.
Biophys.
Acta 501,
449-452

Downloaded from www.jbc.org by guest, on March 19, 2011

analyses of the cytochrome

oxidase components
were then
based on the two-chromophore
model. An unexpected finding
was that the absorbance of the component centered at 650 nm
started high in the oxidized state, then sharply dropped, then
rose sharply at first and then more gradually with decreasing
voltage to about 150 mV, then dropped again at lower voltages.
Another unexpected
finding was that the steepness of the
absorbance versus voltage plots for both the feature absorbing
at 634 nm and that at 650 nm could be fit only by Nernstian
functions involving more than single electron transfers. In
fact, the resolution
of components
absorbing
in the cytochrome oxidase region leads to a representation
of a possible
branched system which collects electrons from four single
electron transfer chains and multiplexes them for the eventual
reduction of a molecule of oxygen.
Cytochrome
oxidase has not been purified from E. coli. If
it is similar, however, to that of mitochondria
which have
been rigorously studied (12), it will contain two iron and two
copper centers. It has long been appreciated that a sequential
single electron transfer to oxygen is thermodynamically
unlikely because of the instability
of 02- (13). Therefore, either
a 2-electron reduction of % O2 or a 4-electron reduction of O2
is anticipated. Early electron paramagnetic
resonance studies
by van Gelder and Beinert (14) suggested an antiferromagnetic coupling between iron in cytochrome
a3 and another
oxidation-reduction
component
in the cytochrome
oxidase.
Many recent studies employing magnetic susceptibility
measurements and magnetic circular dichroism
in addition
to
electron paramagnetic
resonance support the idea of a magnetic coupling of the iron in cytochrome
a3 to a copper,
resulting in a functional 2-electron transfer component
(1518). Lindsay and Wilson have reported that chemical potentiometric titration of cytochrome oxidase liganded to carbon
monoxide results in a Nernstian
value of n. = 2 electrons
transferred (19, 20). The native enzyme was reported earlier
to titrate with an n = 1 for each iron center (21). Anderson et
al. (22) using a coulometric
titration technique disagreed with
the results of Lindsay and Wilson indicating the involvement
of carbon monoxide with 2-electron acceptors and interpreted
their findings in terms of a l-electron
process. Heineman
et
al. have shown that natural cytochrome
oxidase requires 4
electrons for complete reduction
(23) although their coulometric technique does not tell whether the transfers are for
single electrons or multiples. Arguments for a concerted multielectron transfer from the fully reduced enzyme to oxygen
have been presented (24-26). The absorbances we have studied at 634 nm and 650 nm most likely represent iron heme
groups which individually
can transfer only single electrons.
Although we are not proposing a particular mechanism at this
point, there would have to be a way in which more than a
single cytochrome
oxidase molecule could interact in the
membrane so that reduction of the multimer could occur in a
single concerted step when the required number of electrons
was amassed. The model with a final 4-electron transfer was
presented over that with an &electron
transfer although the
data more closely fit the 8-electron case, because it makes
more sense in terms of the 4 electrons required to reduce a
molecule of oxygen. However, if our analysis is correct and if

of E. coli Cytochromes