N epsilon,N epsilon-dimethyl-lysine cytochrome c as an NMR probe for lysine involvement in protein-protein complex formation.

Abstract

The reductively dimethylated derivatives of horse and yeast iso-1-ferricytochromes c have been prepared and characterized for use as NMR probes of the complexes formed by cytochrome c with bovine liver cytochrome b5 and yeast cytochrome c peroxidase. The electrostatic properties and structures of the derivatized cytochromes are not significantly perturbed by the modifications; neither are the electrostatics of protein-protein complex formation or rates of interprotein electron transfer. Two-dimensional 1H-13C NMR spectroscopy of the complexes formed by the derivatized cytochromes with cytochrome b5 and cytochrome c peroxidase has been used to investigate the number and identity of lysine residues of cytochrome c that are involved in interprotein interactions of cytochrome c. The NMR data are incompatible with simple static models proposed previously for the complexes formed by these proteins, but are consistent with the presence of multiple, interconverting complexes of comparable stability, consistent with studies employing Brownian dynamics to model the complexes. The NMR characteristics of the Nepsilon,Nepsilon-dimethyl-lysine groups, their chemical shift dispersion, oxidation state and temperature dependences and the possibility of chemical exchange phenomena are discussed with relevance to the utility of Nepsilon, Nepsilon-dimethyl-lysine's being a generally useful derivative for characterizing protein-protein complexes.

Full text

Page 1

Biochem. J. (1998) 332, 439–449 (Printed in Great Britain)
439
Nε,Nε-Dimethyl-lysine cytochrome c as an NMR probe for lysine
involvement in protein–protein complex formation
Geoffrey R. MOORE*1Mark C. COX*, David CROWE*, Michael J. OSBORNE*, Federico I. ROSELL†, Jordi BUJONS†,
Paul D. BARKER†, Marcia R. MAUK† and A. Grant MAUK†
*School of Chemical Sciences, University of East Anglia, Norwich NR4 7TJ, U.K., and †Department of Biochemistry and Molecular Biology, University of British
Columbia, Vancouver, British Columbia V6T 1Z3, Canada
The reductively dimethylated derivatives of horse and yeast iso-
1-ferricytochromes c have been prepared and characterized for
use as NMR probes of the complexes formed by cytochrome c
with bovine liver cytochrome b?and yeast cytochrome c per-
oxidase. The electrostatic properties and structures of the
derivatized cytochromes are not significantly perturbed by the
modifications; neither are the electrostatics of protein–protein
complex formation or rates of interprotein electron transfer.
Two-dimensional ?H–??C NMR spectroscopy of the complexes
formed by the derivatized cytochromes with cytochrome b?and
cytochrome c peroxidase has been used to investigate the
number and identity of lysine residues of cytochrome c that are
INTRODUCTION
The characterization of protein–protein complexes formed be-
tween cytochrome c and its redox partners is central to a full
description of the intermolecular electron-transfer processes in
which they participate. Since the early modelling studies of
Salemme [1] considerable progress towards this goal has been
achieved with the cytochrome c–cytochrome b?
Salemme’s model predicted the structure of the cytochrome
c–cytochrome b?complex from the three-dimensional structures
of the individual proteins and implicated four lysine residues on
the surface of cytochrome c in the formation of this complex.
Such predictive methods subsequently evolved to incorporate
both complex electrostatic considerations and molecular dyn-
amics calculations that, together with the results of kinetic and
thermodynamic investigations of chemically modified derivatives
of cytochrome c, and variants produced by site-directed muta-
genesis, have led to considerable progress in the characterization
ofthecytochromec–cytochromeb?complexinsolution(reviewed
in [2]). Similar modelling [3–7], chemical modification [8,9] and
mutagenesis [10–12] studies of cytochrome c interacting with
cytochrome c peroxidase (CcP), together with X-ray crystallo-
graphic studies of several cytochrome c–CcP complexes [13],
have produced high-resolution structural models for this system.
However, as for the cytochrome c–cytochrome b?complex [2],
controversy persists concerning the structure of the cytochrome
c–CcP complex in solution [14], particularly in connection with
the identity of the key lysine residues of cytochrome c responsible
for recognition and binding of its reaction partner.
One approach to investigating interprotein complexes of
cytochrome c has been to use NMR spectroscopy [15–30].
complex.
Abbreviations used: 1D, one-dimensional; 2D, two-dimensional; a.m.u., atomic mass units; CcP, cytochrome c peroxidase; EXSY, chemical exchange
correlated spectroscopy; HSQC, heteronuclear single quantum coherence.
1To whom correspondence should be addressed (e-mail g.moore?uea.ac.uk).
involved in interprotein interactions of cytochrome c. The NMR
data are incompatible with simple static models proposed pre-
viously for the complexes formed by these proteins, but are
consistent with the presence of multiple, interconverting com-
plexes of comparable stability, consistent with studies employing
Brownian dynamics to model the complexes. The NMR charac-
teristics of the Nε,Nε-dimethyl-lysine groups, their chemical shift
dispersion, oxidation state and temperature dependences and the
possibility of chemical exchange phenomena are discussed with
relevancetotheutilityofNε,Nε-dimethyl-lysine’sbeingagenerally
useful derivative for characterizing protein–protein complexes.
However, neither one-dimensional (1D) ?H nor two-dimensional
(2D) ?H–?H NMR spectra provide sufficient resolution to allow
small perturbations of the interprotein contact residues, par-
ticularly the lysine residues, to be monitored in such complexes.
The large number of lysine residues (19 out of 104 for horse
cytochrome c) further complicates matters. To help overcome
this resolution problem, we have modified the lysine residues of
horse heart cytochrome c to produce the fully Nε-acetimidylated
derivative [21] and the fully Nε,Nε-dimethylated derivative
reported here. These modifications maintain the positive charge
on the derivatized lysine residues and provide each derivatized
residue with one (the Nε-acetimidylated protein) or two (the
Nε,Nε-dimethylated protein) methyl groups, whose resonances
can be observed in 1D ?H NMR spectra. This approach can be
enhanced further through the use of ??C-enriched modifying
reagents that label the derivatives with ??C and allow the
acquisitionof2D?H–??CCOSYorheteronuclearsinglequantum
coherence (HSQC) spectra. Perturbations of the derivatized
lysine methyl resonances resulting from complex formation
between the modified cytochrome c and its reaction partners can
then be used to identify the interprotein binding surface on
cytochrome c in a manner similar to that of Dick et al. [31], who
used reductive methylation of the lysine residues of the fd gene
5 DNA-binding protein to provide NMR probes for identifying
its DNA-binding surface. In the present study we have focused
on the Nε,Nε-dimethylated cytochrome c because this modi-
fication provides better NMR sensitivity, greater ease of ??C
labelling and a more extensive literature on the characterization
of proteins modified in this manner than does modification to the
Nε-acetimidylated cytochrome. Furthermore NMR studies of the
??C-enriched Nε,Nε-dimethylated derivatives of ribonuclease [32],

Page 2

440
G. R. Moore and others
lysozyme [33], fd gene 5 protein [31], ferredoxin [34], calmodulin
[35] and calbindin [36] have been reported, and the consensus
from these studies is that dimethylation is a relatively benign
modification in terms of altering protein structures.
In the present paper we describe the preparation of the Nε,Nε-
dimethylated derivative of ferricytochrome c and have studied
the effects of this modification on a range of chemical properties
of cytochrome c, including: its proton titration, the pH de-
pendence of the stability of the cytochrome c–cytochrome b?
complex, and the kinetics by which ferricytochrome c is reduced
by ferrocytochrome b?. Use of the fully dimethylated cytochrome
as a spectroscopic probe for identification of surface lysine
residues involved in the interaction of the cytochrome with
cytochrome b?and cytochrome c peroxidase has been initiated
through ?H and ??C NMR studies of the complexes formed by
the native and modified cytochrome.
EXPERIMENTAL
Protein preparation, purification and modification
Recombinant yeast cytochrome c peroxidase was prepared as
described previously [37,38] and concentrations determined on
the basis of ε????97000 M−??cm−? [39]. The yeast iso-1-cyto-
chromes c were prepared as described previously [40]. All yeast
cytochrome c samples carried a Cys???Thr substitution, which
prevents intermolecular disulphide bond formation and greatly
decreases the rate of autoreduction without affecting the struc-
tural or functional properties of the protein [41]. The Lys??Ala,
Lys??Ala, Lys??Ala and Lys??Ala variants of yeast iso-1-cyto-
chromescusedinthepresentworkhavebeendescribedelsewhere
([42], and F. I. Rosell, J. C. Ferrer and A. G. Mauk, unpublished
work). Recombinant trypsin-solubilized [44,45] and recombinant
lipase-solubilized [44] bovine liver microsomal cytochromes b?
expressed in Escherichia coli were prepared as described pre-
viously. Concentrations of ferricytochrome b?solutions were
based on ε???.??117000 M−??cm−? [46]. Horse heart cytochrome
c (Type VI; Sigma) was purified by cation-exchange chromato-
graphy to remove deamidinated protein by the procedure of
Brautigan et al. [47]. Concentrations of native and derivatized
ferricytochrome c solutions were determined on the basis of
ε???.??106000 M−??cm−? [48] and the assumption that the
spectrum of the modified cytochrome is the same as that of
the native protein.
Dimethylation of the lysine Nε-amino groups of cytochrome c
was performed by the method of Jentoft and Dearborn [49] as
modified by Wallace and Corte?sy [50]. Typically, to a solution of
50 mg of cytochrome c in 10 ml of 0.1 M sodium tetraborate
buffer at pH 9 were added five 20 µl volumes of 37% aqueous
formaldehyde at 2 min intervals and two 5 mg quantities of solid
NaBH?at 4 min intervals after alternate additions of form-
aldehyde. After a total of 10 min, 10 ml of 0.1 M HCl was added
to stop the reaction and the mixture was passed through a
Sephadex G-25 column (45 cm?1.5 cm) equilibrated with
0.05 M ammonium bicarbonate. The modified protein was then
oxidized with K?[Fe(CN)?] and chromatographed on a CM-
cellulose ion-exchange column (Whatman CM-52) equilibrated
with 0.005 M sodium phosphate buffer at pH 7. The modified
protein was eluted as a single sharp band with 0.5 M
NaCl?0.005 M sodium phosphate (pH 7.5); the NaCl was
removed by passage through a Sephadex G-25 column
equilibrated with 0.05 M sodium phosphate, pH 7.5. ??C-
labelled dimethylcytochrome c was obtained by using 21.3%
(v?v) [??C]formaldehyde (MSD Isotopes). Final purification of
methylated ferricytochrome c for potentiometric experiments
was achieved by FPLC (Pharmacia Mono-S HR 10?10 cation
exchange columns in 20 mM sodium phosphate buffer, pH 7.25,
witha 70–150 mM NaCl gradient).
Electrospray mass spectrometry
Mass spectra were recorded with a VG BIO Q quadrupole mass
spectrometer or with a triple quadrupole mass spectrometer
constructed in-house [51] that was fitted with a pneumatically
assisted electrospray interface. The sprayer and ion-sampling
orifice voltages were maintained at 5.5 and 0.2 kV. Samples were
introduced at 1 µl?min at ambient temperature; mass spectra
were obtained at a dwell time of 10 s per scan with 10 scans
summed per spectrum and with a step size of 0.2 atomic mass
units(a.m.u.).Proteinsamples(2–3 µM)werepreparedin10 mM
ammonium acetate buffer, pH 7.0, by dilution directly into this
buffer or by buffer exchange by centrifugal ultrafiltration
(Centricon-10; Amicon).
Potentiometric titrations
Potentiometric titrations were performed with a Radiometer
ABU93 Triburette operated under computer control as described
previously [45,52]. Protein samples used in these measurements
were deionized shortly before use by exhaustive exchange (at
least 10?-fold) into 10 mM NaCl by centrifugal ultrafiltration
(Centriprep-10; Amicon). For the protein–protein titrations
[45,53] the titrand protein concentration was 45–55 µM and the
titrant protein was 0.7–1.0 µM. For the continuous proton
titrations [45], initial methylated ferricytochrome c concen-
trations were 40–50 µM and the initial ferricytochrome b?con-
centrationwas0.7 µM.Thisratioofcytochromeb?tocytochrome
c was required to ensure that the proportion of cytochrome c
bound to cytochrome b?varied minimally over the range of pH
studied.ThecurvesofqagainstpHobtainedfromthecontinuous
proton titrations were integrated with the program Spectracalc
(Galactic Industries, Northampton, MA, U.S.A.) to generate
curves of log Kaagainst pH. The integration constants used in
this process were evaluated from the log Kavalues obtained from
protein–protein titrations. -Lysine dihydrochloride (L5751;
Sigma), Nε-methyl--lysine hydrochloride (M6004; Sigma),
Nε,Nε-dimethyl--lysine hydrochloride (E1810; Bachem) and
Nε,Nε,Nε-trimethyl--lysine hydrochloride (T1660; Sigma) were
purchased from the vendors indicated. Potentiometric titration
curves obtained for the lysine derivatives were analysed in terms
of two or three titratable groups as appropriate by a least-
squares process with the program SCIENTIST (MicroMath,
Salt Lake City, UT, U.S.A.). The pH titration curves of each
protein were reproducible to ?0.1 proton between pH 3 and 11.
Difference titration curves derived by subtracting one such
titration curve from another have an uncertainty of ?0.14
proton over this pH range.
Anaerobic stopped-flow kinetics
Second-order rate constants for the bimolecular reduction of
ferricytochrome c derivatives by ferrocytochrome b?were de-
termined under anaerobic conditions with a thermostatically
controlled Durrum stopped-flow spectrophotometer housed in a
glove box and interfaced to a microcomputer (Olis, Bogart, GA,
U.S.A.) as described previously [54,55]. Homogeneous second-
order reaction conditions (equal concentrations of reactants)
were required because the reduction of ferricytochrome c by
ferrocytochrome b?is too rapid to be studied under pseudo-first-
order conditions [54]. Rate data were analysed by least-squares
fitting to a homogeneous second-order rate expression with the
program MINSQ (MicroMath).

Page 3

441
Cytochrome c complexes with cytochrome b5and cytochrome c peroxidase
NMR spectroscopy
Samples were prepared for NMR spectroscopy by centrifugal
ultrafiltration (Centricon-10 devices; Amicon). Four cycles of
concentration and dilution with appropriate buffer in ?H?O were
required to achieve an adequate exchange of H?O for ?H?O. The
final protein solutions were diluted with buffer in ?H?O to give
the desired protein concentrations, and the pH was adjusted with
NaO?H or ?HCl. All reported pH values for ?H?O solutions,
indicated by p?H, are the direct meter readings uncorrected for
any isotope effect. The electrolyte solution in the pH electrodes
used for measurements with ?H?O solutions was not exchanged
with the corresponding ?H?O solution.
Mixtures of oxidized and reduced cytochromes were prepared
for exchange spectroscopy and saturation transfer experiments
by reducing the protein with an approx. 20% excess of sodium
dithionite followed by removal of the excess reducing agent by
centrifugal ultrafiltration. The reduced cytochrome was then
combined with a previously prepared oxidized sample. Ratios of
oxidized to reduced cytochrome were calculated from integrated
intensities of the ?H NMR signals of the Met?? methyl group of
the reduced cytochrome and the Pro?? δ-methylene proton of the
oxidized cytochrome. These groups have chemical shifts of ?3.3
and ?6.3 p.p.m. respectively [56,57].
NMR spectra were recorded with a Jeol GX-400 spectrometer.
The intensity of the residual HO?H peak was decreased by gated
irradiation for 0.5 s immediately before application of the 90?
pulse.Chemicalshifts arequotedrelativetothemethyl resonance
of 2,2-dimethylsilapentane-5-sulphonic acid at 0.000 p.p.m.
?H–??C COSY was performed with the procedure described by
Maudsley et al. [58]. Spectra were collected over a frequency
range of 1000 Hz in t?with 256 transients in t?or 4000 Hz in t?
with 512 transients in t?. The Jeol spectrometer used for this
work was not configured for HSQC or heteronuclear multiple
quantum coherence experiments of the type that has been found
to be useful for similar experiments on calmodulins [36], but the
COSY approach adopted, although requiring longer data acqui-
sition times, provides equivalent information.
The bimolecular electron self-exchange rate
dimethylcytochrome c was measured with the NMR saturation
transfer technique [59–61]. Ionic strength was varied by the
addition of NaCl. T?values were obtained for the Met?? methyl
resonance of native and modified horse heart cytochromes c by
using an inversion recovery sequence. ??C–??C chemical exchange
correlatedspectroscopy(EXSY)wasperformedwiththeNOESY
pulse sequence [62,63]. Spectra were collected over a range of
1000 Hz in each dimension with 256 transients in t?. 2D data
were processed with FELIX 2.30 (Biosym, Cambridge, U.K.).
To compare linewidths in 2D spectra, data were processed with
a skewed sinebell window function weighted to minimize dis-
tortions. Widths were estimated at half height with a 1D slice
taken through the centre of the peak in each dimension. This
method is adequate for the comparison of relative linewidths,
although not for the measurement of absolute values. Titrations
of cytochromes c with cytochrome c peroxidase were performed
by adding small volumes of an approx. 3 mM cytochrome c
peroxidase solution to approx. 1.4 mM cytochrome c solutions.
of
Nε,Nε-
RESULTS
Chemical characterization of methylated horse and yeast
cytochromes c
The observed masses of horse heart cytochrome c (12359
?3 a.m.u.) and the methylated derivative (12894?3 a.m.u.)
Table 1
ferricytochrome c derivatives
Parameters obtained from potentiometric titrations of horse heart
ProteinpH rangeNumber of titratable groupspH of zero net charge
Native
Native
Methylated
Methylated
3 to 11
11 to 3
3 to 11
11 to 3
28.6
30.5
25.6
24.8
10
9.6
Figure 1
horse heart ferricytochrome c [25 ?C, µ?0.1 M (KCl)]
Chemical modification-induced difference titration curves of
(A) The titration curve (from pH 11 to pH 3) of acetimidylated ferricytochrome c minus the
titration curve (from pH 11 to pH 3) of native ferricytochrome c. (B) The titration curve of
methylated ferricytochrome c (from pH 11 to pH 3) minus the titration curve (from pH 11 to
pH 3) of native ferricytochrome c.
were in good agreement with the values anticipated for the
unmodified(12360 a.m.u.)and
(12893 a.m.u.) proteins. The observed mass of the native yeast
cytochrome (12708?2 a.m.u.) was also in agreement with the
calculated value (12709 a.m.u.), and analysis of the modified
derivative provided a mass of 13130?1 a.m.u. as expected for
dimethylation of all 15 lysine residues (calculated mass
13130 a.m.u.).
exhaustively dimethylated
Potentiometric titrations of dimethylated horse cytochrome c
Potentiometric titrations of native and dimethylated horse cyto-
chromes c were performed to assess whether the modification
had substantially altered the electrostatic properties of the
cytochrome and to provide a control for the potentiometric
studies of dimethylated cytochrome c interacting with cyto-
chrome b?. The parameters extracted from the titration curves
obtained for native cytochrome c and its methylated derivative
are summarized in Table 1. As seen from these results, native
horse heart cytochrome c exhibited a small increase in the
number of groups that titrated between pH 3 and 11 after the

Page 4

442
G. R. Moore and others
Figure 2
association of native and dimethylated horse heart ferricytochrome c with
trypsin- and lipase-solubilized cytochrome b5
Dependence of q and log Kaon pH (µ?0.01 M, 25 ?C) for the
The uncertainties in q and pKaare smaller than the size of the data points. (A) Continuous proton
titrations of native ferricytochrome c (0.115 mmol) with tryptic ferricytochrome b5(1.95 mmol)
in 4.178 ml (solid line) and with methylated ferricytochrome c (0.128 mmol) with tryptic
ferricytochrome b5(1.81 mmol) in 3.969 ml (broken line). Data are also shown from
protein–protein titrations with native ferricytochrome c and lipase-solubilized ferricytochrome b5
(?) at pH 6.72 and 8.47; and with methylated ferricytochrome c and lipase-solubilized
cytochrome b5(?) at pH 6.75 and 8.25. (B) The curves in (A) were integrated to yield the
curves of log Kaagainst pH for complex formation between native ferricytochrome c and
tryptic ferricytochrome b5(solid line) and between methylated ferricytochrome c and tryptic
ferricytochrome b5(broken line). Representative log Kavalues from protein–protein titrations are
shown: native ferricytochrome c with tryptic cytochrome b5(?) or lipase-solubilized
cytochrome b5(?) and methylated cytochrome c with tryptic cytochrome b5(?) and lipase-
solubilized cytochrome b5(?).
protein was exposed to basic pH. Over the same pH range, three
fewer groups titrated in dimethylated ferricytochrome c than
observed for the native protein, although the pH of zero net
charge was virtually unchanged. As observed for the native
protein [52], the titration curve of the dimethylated cytochrome
exhibited hysteresis, but in this case exposure to alkaline pH
resulted in a slight decrease in the number of groups that titrated
rather than an increase.
The modification-induced difference titration curve obtained
for the fully methylated ferricytochrome is shown in Figure 1.
The titration curves of native and methylated ferricytochrome c
were identical between pH 4 and 8.5. Beyond this range, the
electrostatic properties of the modified protein began to deviate
significantly from those of the native protein. Importantly, the
shape of the curve between pH 9 and pH 11 indicates that the
modification had resulted in a decrease in the pKavalues of many
of the groups ionizing at alkaline pH. This finding is consistent
with Nε,Nε-dimethylation of the lysine residues and is incon-
sistent with their monomethylation or trimethylation, as shown
by the side chain pKavalues determined from titration curves
[25 ?C, 0.1 M KCl (results not shown)] of lysine (11.01?0.02),
Table 2
interaction of various forms of ferricytochrome c with ferricytochrome b5
The uncertainty in q values is approx. ?0.01, and the uncertainty in pKavalues is approx.
?0.05.
pH dependence of proton release and complex stability for
Proteins pH
q
pKa
Native cytochrome c–tryptic cytochrome b5
6.75
8.45
6.73
8.25
6.72
8.47
6.75
8.25
0.395.94
5.34
4.93
5.29
5.75
4.90
4.89
4.95
?0.84
0.59
?0.76
0.38
?1.15
0.45
?0.89
Methylated cytochrome c–tryptic cytochrome b5
Native cytochrome c–lipase cytochrome b5
Methylated cytochrome c–lipase cytochrome b5
Nε-monomethyl-lysine
(10.70?0.02);thequaternaryaminegroupofNε,Nε,Nε-trimethyl-
lysine did not ionize over the pH range 3–12. These pKavalues
are in good agreement with those reported by Jentoft and
Dearborn [49].
(11.34?0.02),
Nε,Nε-dimethyl-lysine
NMR characterization of the conformations of dimethyl-lysine
horse and yeast cytochromes c
Comparison of the modified and unmodified ferrocytochromes c
and ferricytochromes c by 1D and 2D NMR methods showed
that resonances of amino acid residues not directly attached to
the haem were either unaffected by the modification or were
shifted by no more than 0.05 p.p.m. ([21,64], and M. C. Cox, D.
Crowe and G. R. Moore, unpublished work). Some of the haem
resonances were shifted by substantially more than this, most
notably haem resonances of the ferricytochromes, but this effect,
which has been observed with other modifications of cytochrome
c [17,65,66], reflects small changes in the distribution of unpaired
electrons within the haem and not a major structural change.
Thus we conclude that the conformation of cytochrome c is not
altered significantly by the modification. The dynamic properties
of the structure, as manifested by the mobilities of aromatic
amino acids [67,68] and a haem-linked chemical exchange [69],
also seem not to have been perturbed [21,23,64].
Potentiometric characterization of the interaction of cytochrome
b5with horse Nε,Nε-dimethylcytochrome c
Continuous potentiometric titration of the complexes formed by
ferricytochrome b?with native and dimethylferricytochrome c
are shown in Figure 2(A). In addition, q values obtained from the
titration of native and dimethylated cytochrome c with lipase-
and trypsin-solubilized cytochrome b?at two values of pH are
shown. The parameters derived from titrations performed at
acidic and basic pH (Table 2) permit a quantitative comparison
of the electrostatic character of the complexes formed by lipase-
and trypsin-solubilized cytochrome b?with the native and
modified cytochrome c. From the results summarized in Figure
2 and Table 2, the electrostatic properties of all four protein–
protein complexes were virtually identical. Integration of the
continuous titration curves of the complexes (Figure 2A) to
determine the dependence of binding affinity (log pKa) on pH [45]
reveals that the shapes of the pH-dependence curves obtained for

Page 5

443
Cytochrome c complexes with cytochrome b5and cytochrome c peroxidase
Figure 3
native (?) and Nε,Nε-dimethyl-lysine ferricytochrome c (?) by tryptic
ferrocytochrome b5on ionic strength (25 ?C, sodium phosphate buffer)
Dependence of the second-order rate constant for reduction of
The uncertainties in the ln k12values are smaller than the size of the data points.
the native and modified cytochrome c were virtually indis-
tinguishable. Nevertheless, the modified cytochrome c exhibited
a significantly lower affinity for cytochrome b?between pH 5.5
and 9. We conclude from these results that the lipase- and
trypsin-solubilized cytochromes, which differ by the presence of
atworesidueextensionattheN-terminusofthelipase-solubilized
cytochrome, are, to a first approximation, sufficiently similar in
their interaction with the native and modified cytochrome c that
they could be used interchangeably in these studies.
Figure 4
ferricytochrome b5(5 mM phosphate at p2H 7, 25 ?C)
1H–13C COSY spectra of 3 mM13C-labelled Nε,Nε-dimethyl-lysine ferrricytochrome c in the absence (left) and the presence (right) of 3 mM
The acquisition parameters for the COSY spectrum were: spectral widths of 4000 and 1000 Hz respectively for the13C and1H dimensions; 512 t1increments of 2048 data points each; and 16
scans per increment.
Bimolecular electron transfer kinetics of horse Nε,Nε-
dimethylcytochrome c
The electron self-exchange rate of Nε,Nε-dimethyl-lysine cyto-
chrome c was 240 M−??s−? at zero added ionic strength, a value
only marginally lower than the 260 M−??s−? obtained for the
native cytochrome (results not shown). Both rate constants were
markedly,andsimilarly,dependentonionicstrength,asobserved
previously for cytochrome c [60,61].
The ionic strength dependence of the second-order rate
constants for the reduction of native and dimethylated ferri-
cytochrome c by ferrocytochrome b?is shown in Figure 3. Under
all conditions studied, the native ferricytochrome c was reduced
more rapidly than the modified cytochrome, consistent with their
respective self-exchange rates. The curves fitted to these results in
Figure 3 were based on the analytical procedure of van Leeuwen
[70] and employed estimates for the radii and dipole moments of
horse heart cytochrome c [16 A ? , 284 D (0.947?10−?? C?m)] and
cytochromeb?[17 A ? , 640 D(2.14?10−??C?m)]. Theresulting fits
indicatedinthefigure areconsistentwith valuesfor thelogarithm
of the rate constant at infinite ionic strength (ln kinf) and the
apparent charge products of 12.9?0.2 M−??s−?, ?240?90 for
the reaction of the two native proteins and 12.4?0.2 M−??s−?,
?70?50 for the reduction of modified cytochrome c by ferro-
cytochrome b?. Although the mechanistic significance of these
values is debatable, we note that the rate constants for the two
reactions are more similar at low ionic strength, where elec-
trostatic factors dominate, and diverge at higher ionic strength,
whereshieldingeffectsdiminish thecontributionofelectrostatics.
The inequivalent curvatures of the two fits (Figure 3) might
indicate that the nature of the rate-determining step changes with
ionic strength in a manner that is inequivalent for the reactions
of the native and modified cytochromes. This is consistent with
the ionic strength dependence of the rate constant for intra-
molecular electron transfer in the cytochrome c–cytochrome b?
and cytochrome c–CcP complexes noted previously [71,72].

Page 6

444
G. R. Moore and others
Figure 5
iso-1-ferricytochrome c variants in 50 mM sodium phosphate buffer, p2H 7.0
1H–13C COSY spectra of13C-labelled Nε,Nε-dimethyl-lysine yeast
(A) Wild type (Cys102Thr) at 25 ?C, (B) Lys73Ala at 25 ?C, (C) Lys79Ala at 25 ?C, (D) wild type
(Cys102Thr) at 35 ?C, (E) Lys86Ala at 35 ?C, (F) Lys87Ala at 35 ?C. The arrows in (B), (C) and
(E) indicate the locations of the peaks of Lys73, Lys79and Lys86respectively.
The kinetic results, in combination with results of the
potentiometric titrations described above, lead us to conclude
that the electrostatic properties of the cytochrome c–cytochrome
b?complex are unaffected by the dimethylation of surface lysine
residues of cytochrome c and that differences in complex stability
and reactivity involve primarily non-electrostatic factors. As a
result, dimethylated cytochrome c is a good NMR probe for
characterizing complexes formed by cytochrome c.
Characterization of the NMR resonances of dimethyl-lysine horse
and yeast cytochromes c
The ??C NMR signals of the lysine ε-methyl groups appear in the
chemical shift range expected for dimethylated species [73], with
no signals in the 30–35 or 55–60 p.p.m. ranges expected for
mono- or trimethylated lysine residues respectively (results not
shown), consistent with the mass spectrometric and potentio-
metric determination of the reaction product as dimethylated.
The?H–??CCOSYspectrumofdimethylhorseferricytochrome
c (Figure 4) contains cross-peaks from 15 of the 19 dimethyl-
lysine groups. Two further resonances are present in the 1D ?H
spectrum, at chemical shifts of 2.67 and 2.72 p.p.m. respectively
[23], but these do not appear in the 2D spectrum because they
have relatively short relaxation times, perhaps associated with
chemical exchange phenomena. Such effects seem to have a
significantroleintheappearanceofthe??Cspectrum.Resonances
1, 2 and 15 (Figure 4) exhibit a significant decrease in their
linewidths on increasing the temperature to 45 ?C, consistent
with a decrease in the contribution of a chemical exchange
process to the linewidth as the rate of the process increased
(results not shown). Two kinds of exchange process could be
responsible for this behaviour. If rapid rotation about the εC–N
bond of the dimethyl-lysine residue were prevented, the two
methyl groups could become inequivalent and have different
chemical shifts in either or both the ?H and ??C dimensions, as
has been observed with calmodulin [74] and the fd gene 5 protein
[31]. Alternatively, because lysine side chains are relatively
mobile, they could move between two or more positions in each
of which salt-bridging or hydrogen-bonding interactions with
carboxylate groups could occur. Such behaviour has previously
been suggested for Lys?? and Lys?? of horse cytochrome c [75].
Unambiguous assignment of the dimethyl-lysine resonances in
Figure 4 has not been possible. ?H–?H TOCSY and NOESY
spectra have not provided connectivities [64], and mutants of
horse cytochrome c in which lysine residues have been selectively
replaced are not available. Tentative assignments for some of the
resonances have been obtained from the effect of [Cr(CN)?]?−on
resonance linewidths [23,64] and the location of [Cr(CN)?]?−-
binding sites on the surface of the cytochrome [76] (Table 3).
Thechemical shiftdispersion ofthedimethyl-lysine resonances
(Figures 4 and 5) is partly a result of the presence of a
paramagnetic haem causing pseudocontact shifts; the corre-
sponding resonances of the diamagnetic reduced cytochromes
overlap to a greater extent [23,64]. 2D ??C–??C EXSY spectra of
mixtures of dimethylated oxidized and reduced cytochrome c
have allowed the ferrocytochrome c counterparts to some of the
resolved ferricytochrome c resonances to be identified, and all
but one are in a heavily overlapped region with ??C chemical
shifts between 44.9 and 45.4 p.p.m. (results not shown). Con-
sistent with this, many of the resonances of the modified
ferricytochromehavetemperature-dependentchemicalshiftsthat
move towards the ferrocytochrome c chemical shift as tem-
perature increases [64] (results not shown). Peak 15, with a ??C
chemical shift of 44.7 p.p.m. (Figure 4), is the only well-resolved
resonance that has a negligible temperature dependence, and
because there is a peak at the same position in the spectrum of
the ferrocytochrome c this resonance must arise from a group far
from the haem, consistent with the tentative assignment obtained
from the effect of [Cr(CN)?]?−binding (Table 3) [23].
As with horse cytochrome c, the ?H–??C COSY spectrum of
dimethyl yeast cytochrome c (Figure 5) improves on the
significant degree of overlap in the 1D spectra as well as showing
the chemical shift correlations. Eleven of the expected 15
dimethyl-lysine signals are resolved at 35 ?C, with resonances
beingsignificantlysharpenedbytheslightlyelevatedtemperature.
Chemical exchange leading to rapid relaxation in either or both
of the ?H and ??C dimensions might cause the other resonance to
be missing.
The ?H–??C COSY spectra of dimethylated site-directed
variants, Lys??Ala, Lys??Ala, Lys??Ala and Lys??Ala, of the yeast
cytochrome c are shown in Figure 5. Although the assignments

Page 7

445
Cytochrome c complexes with cytochrome b5and cytochrome c peroxidase
Table 3Effect of cytochrome b5and cytochrome c peroxidase on the dimethyl-lysine resonances of horse heart Nε,Nε-dimethyl-lysine ferricytochrome c
The tentative lysine assignments were obtained from the effect of [Cr(CN)6]3−binding on NMR spectra [23]. The digital resolution for peak width measurements is 1 Hz per point in the13C dimension
and 1 Hz per point in the1H dimension. Abbreviations: u, unaffected by cytochrome b5or by CcP, ?, affected by cytochrome b5but resolution insufficient to quantify the effect, or linewidth could
not be estimated owing to overlap.
Maximum ∆(δ) on binding cytochrome b5(p.p.m.)
∆(δ) on binding CcP (p.p.m.) Increase in peak width on binding CcP (Hz)
Lysine resonanceTentative lysine assignment
13C
1H
13C
1H
13C
1H
1
2
3
4
5
6
7
8
9
7/25/27
13/72/86
0.03
u
0.02
u
0.03
u
u
?
0.05
?
0.02
u
0.09
u
u
0.02
u
0.01
u
0.02
u
u
?
0.03
?
u
u
0.03
u
u
0.03
0.13
0.02
0.06
2
8
?
u
u
u
u
?
u
?
u
u
3
u
7
3
u
u
2
5
u
?
?
3
2
u
u
6
u
3
u
u
u
u
u
u
u
u
u
u
u
7/25/27
0.01
0.04 13/72/860.03
10
11
12
13
14
15
?0.02
u
u
u
u
?0.04
u
u
7/25/27
13/72/86
39/53/60
39/53/60
0.02
0.03
u
?0.02
of the signals for dimethyl-lysine residues 73 and 79 are im-
mediately apparent at 25 ?C, those for lysine residues 86 and 87
are not. Increasing the temperature to 35 ?C caused a narrowing
of the signals and allowed the identification of the dimethyl-
lysine 86 signal in the crowded region in the centre of the
spectrum. The spectrum of the Lys??Ala mutant contained all
the signals resolved in the wild-type derivative, suggesting
that the dimethyl-lysine 87 peak is one of those not appearing
in the spectrum. The assignments for lysine residues 73 and 86
are not apparent from 1D spectra owing to spectral overlap.
NMR characterization of the interaction of cytochrome b5with
Nε,Nε-dimethyl-lysine horse cytochrome c
The addition of ferricytochrome b?to dimethylferricytochrome c
causes ?H NMR peaks of the cytochrome c haem to shift (results
not shown). The most substantially affected were the haem
methyl-3 and haem methyl-8 resonances (Figure 6). Similar
perturbations have been observed for the interaction of ferri-
cytochrome b?
with both native cytochrome c and Nε-
acetimidylated ferricytochrome c [16,18,21,22]; and they are in-
dicative of the formation of a binary cytochrome c–cytochrome
b?complex. Whitford et al. [22] have proposed that the biphasic
plotsresultingfromsimilarexperimentsinwhichferricytochrome
b?is held constant and ferricytochrome c varied indicate the
formation of ternary complexes. However, in the absence of
supporting data we believe that the case for ternary complex
formation is unproved and suggest that the observations of
Whitford et al. are associated with a change in the binary
cytochrome c–cytochrome b?complex that perturbs the cyto-
chrome b?NMR spectrum but does not significantly affect the
cytochrome c NMR spectrum [2].
Theadditionofferricytochromeb?todimethylferricytochrome
c also causes small changes in the chemical shift of some of the
N-CH?resonances. However, the 1D ?H NMR spectrum is not
sufficiently well resolved for these perturbations to be monitored
easily [23]. To achieve adequate resolution we used ?H–??C
COSY spectra of the ??C-labelled dimethylferricytochrome c
Figure 6
ferricytochrome c N-CH3peaks (?, peak 9; ?, peak 13) and the
comparable1H chemical shift of its haem methyl-8 resonance (?) on the
addition of lipase-solubilized cytochrome b5to Nε,Nε-dimethyl-lysine ferri-
cytochrome c
Change in
13C chemical shifts of two Nε,Nε-dimethyl-lysine
(Figure 5). The incremental addition of small amounts of lipase-
solubilized bovine microsomal cytochrome b?to a sample of ??C-
labelled dimethylcytochrome c at low ionic strength (5 mM

Page 8

446
G. R. Moore and others
Figure 7
ferricytochrome c (A) and Nε,Nε-dimethyl-lysine horse heart ferricytochrome c (B)
1H–13C COSY spectra (50 mM sodium phosphate buffer, p2H 7.0, 25 ?C) of a 1:1 mixture of CcP and Nε,Nε-dimethyl-lysine horse heart
sodium phosphate buffer) resulted in a gradual increase in the
cytochrome b?to cytochrome c ratio that was accompanied by a
corresponding shift in the positions of the haem methyl
resonances of cytochrome c (Figure 6). Some of the dimethyl-
lysine ?H–??C COSY peaks also shifted throughout the titration.
The ?H–??C COSY spectrum of a 1:1 mixture of these two
proteins obtained at low ionic strength is shown in Figure 4, and
the change in ??C chemical shifts of peaks 9 and 13 as a function
of the ratio of cytochrome b?to cytochrome c is shown in Figure
6. In all, at least six of the ?H–??C methyl resonances were shifted
by the cytochrome b?. Increasing the buffer concentration to
50 mM signficantly attenuated the ability to detect complex
formation by NMR spectroscopy, with only peaks 9 and 13
shifting slightly, consistent with the acute inverse dependence of
the stability of the complex on ionic strength [77].
The interaction between horse Nε,Nε-dimethylated cytochrome c
and cytochrome c peroxidase
In similar experiments to those described above, the effect of CcP
complexation on Nε,Nε-dimethyl-lysine horse heart cytochrome c
was monitored. The addition of CcP caused a shift for haem
methyl resonances and for some of the lysine N-CH?signals. The
?H–??C COSY spectrum of a 1:1 mixture of Nε,Nε-dimethylated
horse heart cytochrome c and CcP is given in Figure 7. Seven
dimethyl-lysine N-CH?resonances were shifted by at least
0.02 p.p.m. in at least one dimension by interaction with CcP
and, as with the yeast cytochrome c, some of the resonances were
also broadened (Table 3). Linewidths for the unbound cyto-
chrome c were 12–13 Hz in the proton dimension and 3–9 Hz in
the carbon dimension, with five of the seven shifted resonances
being broadened by at least 3 Hz on complexation. Only one
resonance, peak 5, was broadened without a simultaneous
chemical shift change.
Interaction between yeast Nε,Nε-dimethylated iso-1-cytochrome c
and cytochrome c peroxidase
A titration was performed in which small volumes of a CcP
solution were added to a sample of ??C-labelled dimethyl yeast
cytochrome c, thus gradually increasing the ratio of CcP
to cytochrome c during the experiment. The addition of CcP to
dimethyl iso-1-ferricytochrome c caused ?H NMR peaks of the
Figure 8
yeast iso-1-ferricytochrome c in the presence and the absence of CcP,
showing the N-CH3resonance of trimethyl-lysine 72 (indicated by T) (50 mM
phosphate buffer, p2H 7.0)
Aliphatic region of the1H NMR spectrum of Nε,Nε-dimethyl-lysine
Bottom spectrum, cytochrome c in the absence of CcP; middle spectrum, a 0.5:1.0 mixture
of CcP and modified cytochrome; top spectrum, a 1.4:1.0 mixture of CcP and modified
cytochrome c.
ferricytochrometoshift(resultsnotshown), ashasbeenobserved
previously for cytochrome c–CcP complex formation [4,20,27].
Satterlee et al. [20] observed a change in chemical shift of
?0.7 p.p.m.inthehaemmethyl-3resonanceofhorsecytochrome
c in a solution containing equimolar concentrations of cyto-
chrome c and CcP in 0.01 M KNO?, p?H 6.7, in comparison with
?0.8 p.p.m. in the present work. These authors also observed
[20] a ?0.5 p.p.m. shift for the haem methyl-8 resonance, which
moved less than 0.05 p.p.m. in our experiments. This difference

Page 9

447
Cytochrome c complexes with cytochrome b5and cytochrome c peroxidase
Table 4 Chemical shift changes in the1H–13C COSY spectrum of13C-labelled Nε,Nε-dimethyl-lysine yeast iso-1-ferricytochrome c on binding CcP
The digital resolution for line width measurements is 0.6 Hz per point in the13C dimension and 1 Hz per point in the1H dimension. Abbreviation: ?, the signal was affected, but broadening or
shifting has obscured the signal in the bound protein, or a line width could not be estimated owing to overlap.
Dimethyl-lysine chemical shift in Nε,Nε-dimethyl-lysine
yeast iso-1-ferricytochrome c (p.p.m.)
∆δ with CcP (p.p.m.)Increase in line width on binding CcP (Hz)
13C
1HDimethyl-lysine assignment
13C
1H
13C
1H
45.46
45.33
45.27
45.23
45.19
45.19
44.88
43.84
43.59
3.00
2.92
2.85
2.81
2.82
2.75
2.79
2.18
2.79
73 0.01
0.04
u
?0.01
?0.02
?0.01
?
u
?0.01
?0.04
0.02
2
u
u
?
?
u
7
7
2
2
4
u
u
?
u
u
2
u
?0.02
?0.01
?
u
?0.03
u
0.02
86
79
between results might be due to the different buffer conditions in
the two studies. In the titration performed by Moench et al. [27],
in which CcP concentration was gradually increased in a solution
of yeast iso-1-cytochrome c, the haem methyl-3 resonance was
observed to split into bound and unbound signals at low protein
concentration in 10 mM KNO?, p?H 6.4, demonstrating that the
equilibrium was slow on the NMR timescale. In our study,
lowering the temperature of the cytochrome c?CcP mixture to
4 ?C did not split the resonance, indicating that under our
conditionsofionicstrengththecomplexremainedinfastchemical
exchange.
The addition of CcP to yeast iso-1-ferricytochrome c also
caused small changes in the chemical shifts of some of the lysine
N-CH?resonances. The methyl resonance of trimethyl-lysine 72,
which was present in the unmodified protein and was well
resolved at 3.4 p.p.m. in the ?H spectrum, was shifted and
broadened by the addition of CcP (Figure 8). A similar effect is
seen for the addition of ferricytochrome b?to Candida krusei
ferricytochrome c [18]. The ??C-labelled dimethyl-lysine peaks
were not sufficiently well resolved in 1D spectra to monitor
complexation-induced perturbations, so ?H–??C COSY spectra
were used (results not shown). In all, seven of the nine resolved
peaks in the spectrum were shifted by the addition of CcP by
0.01 p.p.m. or more in at least one dimension (Table 4). These
perturbations in chemical shift are significant in comparison with
the digital resolution (0.006 p.p.m. per point and 0.0025 p.p.m.
per point in the ??C and ?H dimensions respectively) but not all
are significant in comparison with the effect of cytochrome b?’s
binding to Nε,Nε-dimethylated horse heart cytochrome c. In this
case 6 out of 15 resonances shifted in one dimension by at least
0.02 p.p.m. The resonance of Nε,Nε-dimethylated yeast iso-1-
cytochrome c corresponding to dimethyl-lysine 79 shifted by
0.04 p.p.m. in the proton dimension, whereas those correspond-
ing to lysine residues 73 and 86 did not shift significantly on the
addition of CcP. In addition to shifting, some of the N-CH?
resonances of the yeast cytochrome c broadened on the addition
of CcP. Linewidths in the magnitude mode ?H–??C COSY
spectrum of the unbound yeast cytochrome c were in the range
9–14 Hz in the proton dimension and 4–11 Hz in the carbon
dimension and, as indicated in Table 4, the three peaks under-
goingthelargestchemicalshiftchanges,inatleastonedimension,
on binding CcP also broadened by 4 Hz or more in one or other
dimension. This result provides compelling evidence that these
lysine residues are directly involved in the complex with their
conformational flexibility, which causes them to have a relatively
sharplineintheunboundcytochrome,beingrestrainedsomewhat
by the complex formation. Taken together, the chemical shift
and linewidth changes (Table 4) indicate that at least six lysine
residues are involved in intermolecular contacts with CcP.
DISCUSSION
NMR is most useful for investigating protein structures when
resonances have been unambiguously assigned, but even though
some of the cytochrome c lysine residues cannot be monitored by
the approach adopted in this paper (4 of the 19 lysine residues of
horse cytochrome c and 4 of the 11 lysine residues of yeast
cytochrome c), and unambiguous assignments are available for
only three lysine residues of the yeast protein, important in-
formation about interprotein complexes can be obtained. In the
most extreme case with the unobserved lysine residues involved
in interprotein contacts, the argument made below for the
existence of dynamic complexes in which the proteins adopt a
multitude of different orientations is considerably strengthened.
The ?H–??C COSY spectra of ferricytochrome b?binding to
Nε,Nε-dimethyl-lysine ferricytochrome c (Figure 4) reveal that of
the 15 dimethyl-lysine residues observed, six are perturbed by
complex formation (Table 3). The most affected are the peak 9
and 13 lysine residues, which titrate with increasing ferricyto-
chrome b?in a similar fashion to the haem methyl resonances
(Figure 6). These lysine residues are associated with a
[Cr(CN)?]?−-binding site close to lysine residues 13, 72 or 86 [23],
although not all the lysine residues associated with this site are
perturbed by ferricytochrome b?as peak 2 is affected by
[Cr(CN)?]?−but not by the cytochrome (Table 3). Also, one of
the lysine residues associated with a second [Cr(CN)?]?−-binding
site, lysine 7, 25 or 27, is affected by ferricytochrome b?but the
other two lysine residues are not. Both of these [Cr(CN)?]?−-
binding sites are close to the exposed haem edge and thus it seems
that ferricytochrome b?binds to the region of cytochrome c that
includes its exposed haem edge, interacting with a minimum of
six lysine residues as it does so. The variable magnitude of the
chemical shift perturbations indicates that the six lysine residues
are not affected equally in NMR terms.
Chemical shifts are highly sensitive to the local magnetic
environments of nuclei and these might be altered when two
proteins form a complex because the nuclei being detected are
located at the interprotein binding region on the surface of one

Page 10

448
G. R. Moore and others
of the partners, or because the environments of nuclei not
located at the interprotein interface region are altered by a
conformational change induced by complex formation. All
available evidence points to the formation of a complex by
cytochromes c and b?without conformational changes [2], so the
chemical shift perturbations (Table 3) are because the N-CH?
groups are located in the interface region. There are no reported
studies directly related to ours, but in comparison with
complexation-induced shifts for peptide ?H–??N resonances used
to map intermolecular interaction regions for a number of
protein–protein complexes in which major conformational
changes did not accompany complexation [78–82], the chemical
shift perturbations we observed are small. This is because we
were looking at relatively weakly associated proteins for which
the perturbations were a weighted average of their values in the
complexed and uncomplexed states. It also seems likely that the
expected perturbations in chemical shift are spread over more
lysine residues than are involved in intermolecular interactions at
any moment as a consequence of the dynamic nature of the
complex. This is consistent with the observation that more lysine
residues are seen to be perturbed than expected from molecular
modelling studies of static complexes [1,83,84], and is consistent
with Brownian dynamics studies [55] and other NMR investi-
gations with different approaches [18,24]. A similar observation
of smaller perturbations in chemical shift than expected for the
complex of cytochrome c with plastocyanin has also been
accounted for by the assumption that the proteins adopt a
multitude of different orientations within the complex [30].
The crystal structures of CcP complexed with yeast iso-1-
cytochrome c and horse cytochrome c [13] have provided a focus
for discussions about these complexes in solution. Whereas the
crystallographic data provide static structures for the complexes,
the existence of alternative, more dynamic, structures has been a
topic of continuing debate [14], with much evidence pointing to
the formation of 2:1 cytochrome c–CcP complexes under a wide
range of solution conditions [85–88]. A further indication of the
complexity of the solution state interaction comes from energy
transfer measurements [89] and ?H NMR studies of the shielding
of amide ?H–?H exchange of cytochrome c by complex formation
with CcP [28]. These suggest there are several closely related
dynamically interconverting binding conformers, consistent with
Brownian dynamics simulations [90].
In this context of multiple interconverting binding conformers
our NMR findings are revealing in that they support a model in
which there is more than one binding orientation, with some
interfacial movement of the two proteins within the complex. A
significant observation from our work (Tables 3 and 4) is the
number of cytochrome c lysine methyl resonances shifting on
complexation with CcP: seven for of the horse heart cytochrome
and six for the yeast cytochrome. The crystal structures of the
cytochrome c–CcP complexes [13] show only three hydrogen
bonds with lysine residues (8, 72 and 87) for the horse protein
and two potential hydrogen bonds with lysine residues (73 and
87) of the yeast protein, with lysine 86 involved in an additional
intermolecular van der Waals interaction. Neither interaction
scheme is consistent with the NMR data. As with the cytochrome
c–cytochrome b?complex, the small magnitude of the pertur-
bations in chemical shift (Tables 3 and 4) is consistent with the
existence of several different complexes, each of which involves
only a few of the lysine residues. This observation is fully
consistent with the results of Brownian dynamics simulations of
Northrup et al. [5], who concluded that several sites might be
visited at each encounter between the proteins. During the
relativelylong-livedencounter,severalplausibleelectron-transfer
orientations were observed rather than a single dominant com-
plex. Northrup et al. [5] listed six lysine residues on horse
cytochrome c as being most frequently involved in ionic contacts
with CcP: residues 13, 25, 27, 72, 79 and 86. Interestingly, all
three of the NMR signals tentatively assigned to lysine residues
13,72and86arebroadenedandshiftedbythebindinginteraction
(Table 3). The perturbation to the NMR signals of trimethyl-
lysine 72 and dimethyl-lysine 79 of the yeast cytochrome c
demonstrates the involvement of these two groups in inte-
rmolecular contacts, but the signal of dimethyl-lysine 73 is only
minimally affected by complex formation (Table 4). Overall, the
yeast cytochrome c NMR data are more in keeping with the
theoretical predictions of Northrup et al. [5] for the horse
cytochrome c–CcP complex than with the X-ray structure of the
yeast cytochrome c–CcP complex.
Conclusions
Dimethylation of lysine residues provides an NMR spectroscopy
probe of intermolecular interactions. For cytochrome c inter-
acting with CcP and with cytochrome b?, NMR studies are
consistent with a model of the interprotein interactions that is
dynamic, with several sites on cytochrome c being visited at each
encounter between the proteins. 2D ?H–??C NMR has been
shown to be useful for monitoring the dimethyl-lysine groups but
relatively short relaxation times prevented all the resonances
from being observed. It is only the paramagnetism of the
ferrihaem of cytochrome c that permits the chemical shift
dispersion observed in this work; for diamagnetic proteins
containing a large number of lysine residues, resolution might be
a limiting factor.
We thank the BBSRC and EPSRC for support given to the UEA Centre for
Metalloprotein Spectroscopy and Biology by their Biomolecular Sciences Panel, the
NIH and MRC for support of this work by NIH Grant GM33804 and MRC Grant TM-
14021 (to A.G.M.), NATO for Travel Grant 870145 (to A.G.M. and G.R.M.), and the
Wellcome Trust for their award of a Research Leave Fellowship to G.R.M.
REFERENCES
1
2
Salemme, F. R. (1976) J. Mol. Biol. 102, 563–568
Mauk, A. G., Mauk, M. R., Moore, G. R. and Northrup, S. H. (1995) J. Bioenerg.
Biomemb. 27, 311–330
Poulos, T. L. and Kraut, J. (1980) J. Biol. Chem. 255, 10322–10330
Lum, V. R., Brayer, G. D., Louie, G. V., Smith, M. and Mauk, A. G. (1988) in Protein
Structure, Folding, and Design (Oxender, D. L., ed.), pp. 143–150, Alan R. Liss,
New York
Northrup, S. H., Boles, J. O. and Reynolds, J. C. (1988) Science 241, 67–70
Northrup, S. H., Luton, J. A., Boles, J. O. and Reynolds, J. C. (1988) J. Comp. Mol.
Des. 1, 291–311
Northrup, S. H. and Herbert, R. G. (1990) Int. J. Quant. Chem. 55–71
Kang, C. H., Brautigan, D. L., Osheroff, N. and Margoliash, E. (1978) J. Biol. Chem.
253, 6502–6510
Bechtold, R. and Bosshard, H. R. (1985) J. Biol. Chem. 260, 5191–5200
Corin, A. F., McLendon, G., Zhang, Q., Hake, R. A., Falvo, J., Lu, K. S., Ciccarelli,
R. B. and Holzschu, D. (1991) Biochemistry 30, 11585–11595
Corin, A. F., Hake, R. A., McLendon, G., Hazzard, J. T. and Tollin, G. (1993)
Biochemistry 32, 2756–2762
Millett, F., Miller, M. A., Geren, L. and Durham, B. (1995) J. Bioenerg. Biomembr. 27,
341–351
Pelletier, H. and Kraut, J. (1992) Science 258, 1748–1755
Nocek, J. M., Zhou, J. S., Deforest, S., Priyadarshy, S., Beratan, D. N., Onuchic, J. N.
and Hoffman, B. M. (1996) Chem. Rev. 96, 2459–2489
Gupta, R. K. and Yonetani, T. (1973) Biochim. Biophys. Acta 292, 502–508
Miura, R., Sugiyama, T., Akasaka, K. and Yamano, T. (1980) Biochem. Int. 1,
532–538
Boswell, A. P., McClune, G. J., Moore, G. R., Williams, R. J., Pettigrew, G. W.,
Inubishi, T., Yonetani, T. and Harris, D. E. (1980) Biochem. Soc. Trans. 8, 637–638
Eley, C. G. and Moore, G. R. (1983) Biochem. J. 215, 11–21
McLachlan, S. J., La Mar, G. N. and Sletten, E. (1986) J. Am. Chem. Soc. 108,
1285–1291
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

Page 11

449
Cytochrome c complexes with cytochrome b5and cytochrome c peroxidase
20Satterlee, J. D., Moench, S. J. and Erman, J. E. (1987) Biochim. Biophys. Acta 912,
87–97
Burch, A. M., Rigby, S. E., Funk, W. D., MacGillivray, R. T., Mauk, M. R., Mauk, A. G.
and Moore, G. R. (1990) Science 247, 831–833
Whitford, D., Concar, D. W., Veitch, N. C. and Williams, R. J. (1990) Eur. J. Biochem.
192, 715–721
Moore, G. R., Cox, M. C., Crowe, D., Osborne, M. J., Mauk, A. G. and Wilson, M. T.
(1995) in NMR of Paramagnetic Systems (La Mar, G. N., ed.), pp. 95–122, NATO
ASI C457, Kluwer, Dordrecht
Guiles, R. D., Sarma, S., DiGate, R. J., Banville, D., Basus, V. J., Kuntz, I. D. and
Waskell, L. (1996) Nature Struct. Biol. 3, 333–339
Wang, J. L., Larsen, R. W., Moench, S. J., Satterlee, J. D., Rousseau, D. L. and
Ondrias, M. R. (1996) Biochemistry 35, 453–463
Yi, Q., Erman, J. E. and Satterlee, J. D. (1994) Biochemistry 33, 12032–12041
Moench, S. J., Chroni, S., Lou, B. S., Erman, J. E. and Satterlee, J. D. (1992)
Biochemistry 31, 3661–3670
Jeng, M. F., Englander, S. W., Pardue, K., Rogalskyj, J. S. and McLendon, G. (1994)
Nature Struct. Biol. 1, 234–238
Sukits, S. F., Erman, J. E. and Satterlee, J. D. (1997) Biochemistry 36, 5251–5259
Ubbink, M. and Bendall, D. S. (1997) Biochemistry 36, 6326–6335
Dick, L. R., Sherry, A. D., Newkirk, M. M. and Gray, D. M. (1988) J. Biol. Chem.
263, 18864–18872
Jentoft, J. E., Gerken, T. A., Jentoft, N. and Dearborn, D. G. (1981) J. Biol. Chem.
256, 231–236
Gerken, T. A., Jentoft, J. E., Jentoft, N. and Dearborn, D. G. (1982) J. Biol. Chem.
257, 2894–2900
Gluck, M. and Sweeney, W. V. (1990) Biochim. Biophys. Acta 1038, 146–151
Zhang, M. and Vogel, H. J. (1993) J. Biol. Chem. 268, 22420–22428
Zhang, M., Thulin, E. and Vogel, H. J. (1994) J. Protein Chem. 13, 527–535
Goodin, D. B., Davidson, M. G., Roe, J. A., Mauk, A. G. and Smith, M. (1991)
Biochemistry 30, 4953–4962
Ferrer, J. C., Turano, P., Banci, L., Bertini, I., Morris, I. K., Smith, K. M., Smith, M.
and Mauk, A. G. (1994) Biochemistry 33, 7819–7829
Vitello, L. B., Huang, M. and Erman, J. E. (1990) Biochemistry 29, 4283–4288
Rafferty, S. P., Pearce, L. L., Barker, P. D., Guillemette, J. G., Kay, C. M., Smith, M.
and Mauk, A. G. (1990) Biochemistry 29, 9365–9369
Cutler, R. L., Pielak, G. J., Mauk, A. G. and Smith, M. (1987) Protein Eng. 1, 95–99
Ferrer, J. C., Guillemette, J. G., Bogumil, R., Inglis, S. C., Smith, M. and Mauk, A. G.
(1993) J. Am. Chem. Soc. 115, 7507–7508
Reference deleted
Funk, W. D., Lo, T. P., Mauk, M. R., Brayer, G. D., MacGillivray, R. T. and Mauk,
A. G. (1990) Biochemistry 29, 5500–5508
Mauk, M. R., Barker, P. D. and Mauk, A. G. (1991) Biochemistry 30, 9873–9881
Ozols, J. and Strittmatter, P. (1964) J. Biol. Chem. 239, 1018–1023
Brautigan, D. L., Ferguson-Miller, S. and Margoliash, E. (1978) Methods Enzymol.
53, 128–164
Margoliash, E. and Frohwirt, N. (1959) Biochem. J. 71, 570–572
Jentoft, N. and Dearborn, D. G. (1983) Methods Enzymol. 91, 570–577
Wallace, C. J. A. and Corte?sy, B. E. (1987) Eur. J. Biochem. 170, 293–298
Collings, B. A. and Douglas, D. J. (1996) J. Am. Chem. Soc. 118, 4488–4489
Barker, P. D., Mauk, M. R. and Mauk, A. G. (1991) Biochemistry 30, 2377–2383
Laskowski, Jr., M. and Finkenstadt, W. R. (1972) Methods Enzymol 26C, 193–277
Eltis, L. D., Herbert, R. G., Barker, P. D., Mauk, A. G. and Northrup, S. H. (1991)
Biochemistry 30, 3663–3674
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Received 3 December 1997/17 February 1998; accepted 4 March 1998
55Northrup, S. H., Thomasson, K. A., Miller, C. M., Barker, P. D., Eltis, L. D.,
Guillemette, J. G., Inglis, S. C. and Mauk, A. G. (1993) Biochemistry 32, 6613–6623
Keller, R. M. and Wuthrich, K. (1981) J. Mol. Biol. 183, 429–446
Feng, Y., Roder, H., Englander, S. W., Wand, A. J. and Di Stefano, D. L. (1989)
Biochemistry 28, 195–203
Maudsley, A. A., Muller, L. and Ernst, R. R. (1977) J. Magn. Reson. 28, 463–469
Forse?n, S. and Hoffman, R. A. (1963) J. Chem. Phys. 39, 2892–2901
Gupta, R. K., Koenig, S. H. and Redfield, A. G. (1972) J. Magn. Reson. 7, 66–73
Concar, D. W., Hill, H. A. O., Moore, G. R., Whitford, D. and Williams, R. J. (1986)
FEBS Lett. 206, 15–19
Jeener, J., Meier, B. H., Bachmann, P. and Ernst, R. R. (1979) J. Chem. Phys. 71,
4546–4553
Boyd, J., Moore, G. R. and Williams, G. (1984) J. Magn. Reson. 58, 511–516
Cox, M. C. (1993) Ph.D. thesis, University of East Anglia
Boswell, A. P., Moore, G. R., Williams, R. J., Harris, D. E., Wallace, C. J., Bocieck, S.
and Welti, D. (1983) Biochem. J. 213, 679–686
Falk, K. E., Jovall, P. A. and A ? ngstrom, J. (1981) Biochem. J. 193, 1021–1024
Moore, G. R. and Williams, R. J. (1980) Eur. J. Biochem. 103, 493–502
Eley, C. G., Moore, G. R., Williams, R. J., Neupert, W., Boon, P. J., Brinkhof, H. H.,
Nivard, R. J. and Tesser, G. I. (1982) Biochem. J. 205, 153–165
Burns, P. D. and La Mar, G. N. (1981) J. Biol. Chem. 256, 4934–4939
van Leeuwen, J. W. (1983) Biochim. Biophys. Acta 743, 408–421
Durham, B., Fairris, J. L., McLean, M., Millett, F., Scott, J. R., Sligar, S. G. and
Willie, A. (1995) J. Bioenerg. Biomembr. 27, 331–340
Hazzard, J. T., McLendon, G., Cusanovich, M. A. and Tollin, G. (1988) Biochem.
Biophys. Res. Commun. 151, 429–434
Baxter, C. S. and Byvoet, P. (1975) Biochem. Biophys. Res. Commun. 64, 514–518
Huque, M. E. and Vogel, H. J. (1993) J. Protein Chem. 12, 695–707
Osheroff, N., Borden, D., Koppenol, W. H. and Margoliash, E. (1980) J. Biol. Chem.
255, 1689–1697
Moore, G. R., Eley, C. G. S. and Williams, G. (1983) Adv. Inorg. Bioinorg. Mech. 3,
1–96
Mauk, M. R., Reid, L. S. and Mauk, A. G. (1982) Biochemistry 21, 1843–1846
Chen, Y., Reizer, J., Saier, Jr., M. H., Fairbrother, W. J. and Wright, P. E. (1993)
Biochemistry 32, 32–37
Jones, D. N., Bycroft, M., Lubienski, M. J. and Fersht, A. R. (1993) FEBS Lett. 331,
165–172
Clubb, R. T., Omichinski, J. G., Clore, G. M. and Gronenborn, A. M. (1994) FEBS
Lett. 338, 93–97
Gronenborn, A. M. and Clore, G. M. (1993) J. Mol. Biol. 233, 331–335
Osborne, M. J., Wallis, R., Leung, K.-Y., Williams, G., Lian, L.-Y., James, R.,
Kleanthous, C. and Moore, G. R. (1997) Biochem. J. 323, 823–831
Mauk, M. R., Mauk, A. G., Weber, P. C. and Matthew, J. B. (1986) Biochemistry 25,
7085–7091
Rodgers, K. K., Pochapsky, T. C. and Sligar, S. G. (1988) Science 240, 1657–1659
Stemp, E. D. and Hoffman, B. M. (1993) Biochemistry 32, 10848–10865
Zhou, J. S. and Hoffman, B. M. (1994) Science 265, 1693–1696
Zhou, J. S., Nocek, J. M., Devan, M. L. and Hoffman, B. M. (1995) Science 269,
204–207
Mauk, M. R., Ferrer, J. C. and Mauk, A. G. (1994) Biochemistry 33, 12609–12614
McLendon, G., Zhang, Q., Wallin, S. A., Miller, R. M., Billstone, V., Spears, K. G. and
Hoffman, B. M. (1993) J. Am. Chem. Soc. 115, 3665–3669
Northrup, S. H. and Erickson, H. P. (1992) Proc. Natl. Acad. Sci. U.S.A. 89,
3338–3342
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90