Content uploaded by Peter H J Keizers
Author content
All content in this area was uploaded by Peter H J Keizers on Aug 09, 2018
Content may be subject to copyright.
pubs.acs.org/Biochemistry Published on Web 07/13/2010 r2010 American Chemical Society
6846 Biochemistry 2010, 49, 6846–6855
DOI: 10.1021/bi100598f
A Solution Model of the Complex Formed by Adrenodoxin and Adrenodoxin Reductase
Determined by Paramagnetic NMR Spectroscopy
†
Peter H. J. Keizers,
‡,
)
Berna Mersinli,
§,
)
Wolfgang Reinle,
§
Julia Donauer,
§
Yoshitaka Hiruma,
‡
Frank Hannemann,
§
Mark Overhand,
‡
Rita Bernhardt,*
,§
and Marcellus Ubbink*
,‡
‡
Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA Leiden,
The Netherlands, and
§
Universit€
at des Saarlandes, FR 8.8 Biochemie, Campus B2.2, 66041 Saarbr€
ucken, Germany.
)
These authors
contributed equally to this work.
Received April 19, 2010; Revised Manuscript Received July 13, 2010
ABSTRACT:
Lanthanide tags offer the opportunity to retrieve long-range distance information from NMR
experiments that can be used to guide protein docking. To determine whether sufficient restraints can be
retrieved for proteins with low solubility and availability, Ln tags were applied in the study of the 65 kDa
membrane-associated protein complex formed by the electron carrier adrenodoxin and its electron donor,
adrenodoxin reductase. The reductase is only monomeric at low concentration, and the paramagnetic
iron-sulfur cluster of adrenodoxin broadens many of the resonances of nuclei in the interface. Guided by the
paramagnetic restraints obtained using two Ln-tag attachment sites, protein docking yields a cluster of
solutions with an rmsd of 3.2 A
˚. The mean structure is close to the crystal structure of the cross-linked
complex, with an rmsd of 4.0 A
˚. It is concluded that with the application of Ln tags paramagnetic NMR
restraints for structure determination can be retrieved even for difficult, low-concentration protein complexes.
A basic understanding of how macromolecules such as pro-
teins bind other macromolecules is a cornerstone in structural
and systems biology. NMR spectroscopy has been used to deter-
mine the structure of small complexes, and only recently, much
larger complexes have been studied. Nevertheless, improvement
of the methods remains necessary, especially for the cases of
challenging proteins, such as membrane-associated proteins (1-4).
Other proteins that can be troublesome in NMR experiments are
those containing metal cofactors. The nuclei surrounding the
endogenous paramagnetic metal are usually invisible in NMR
spectroscopy due to enhanced nuclear relaxation. Substitution of
the metal for a diamagnetic analogue can resolve this problem (5),
but this procedure is often accompanied with large loss of protein
(6). In electron transfer proteins generally the metal binding site is
close to the interface with the partner proteins to facilitate the elec-
tron transfer. Therefore, the interaction sites in electron transfer
proteins are often difficult to study by conventional NMR. Inter-
estingly, paramagnetism offers at the same time excellent oppor-
tunities to obtain long-range distance information (5,7). With an
artificial paramagnetic Ln probe it is possible to retrieve distance
restraints from far beyond the interaction site, avoiding the
problem of the broadened resonances in the interface. Another
important advantage of the application of a Ln tag is that the
most useful paramagnetic effects, the pseudocontact shift, and the
paramagnetic relaxation enhancement can be readily obtained
from sensitive heteronuclear correlation experiments requiring
only a low protein concentration. A prerequisite for reliable
restraints is to have a rigid paramagnetic tag.
In this study, the complex of adrenodoxin reductase (AdR)
1
and adrenodoxin (Adx) is investigated. These proteins are compo-
nents of the steroid hormone biosynthesis redox chain in the
adrenal mitochondria of vertebrates, containing also the CYP11
family of cytochrome P450 enzymes. CYP11A1 (P450scc) con-
verts cholesterol to pregnenolone, which is the initial and rate-
limiting step of steroid hormones biosynthesis (8). Enzymes of the
CYP11B subfamily are responsible for the formation of cortisol
(CYP11B1) as well as aldosterone (CYP11B2) (9,10). These
cytochromes require two electrons to catalyze oxygen cleavage
and substrate hydroxylation, which are obtained from NADPH
and delivered via AdR and Adx. AdR is localized at the matrix
side of the inner mitochondrial membrane and membrane-bound
by ionic interactions (9,11,12). It is a 51 kDa, NADPH-depen-
dent, electron donor carrying FAD as a prosthetic group (13).
The ferredoxin Adx is a 14 kDa soluble protein, consisting of 128
amino acids. Via its Fe
2
S
2
cluster, Adx carries electrons from
AdR to several cytochrome P450 enzymes (10). The crystal struc-
ture has been determined of the 4-108 truncated Adx as well as
the full-length protein, which demonstrated the presence of a
flexible C-terminus (14,15). The crystal structure of NADPH-
bound AdR was also solved, followed by the structure of a 1:1
cross-linked complex of AdR and full-length Adx (16,17).
Here, we use the AdR-Adx complex as a test case to deter-
mine whether Ln tags can provide sufficient restraints to establish
†
This research has been funded by the Volkswagenstiftung, Grant I/
80 854, The Netherlands Organisation for Scientific Research (NWO),
VENI Grant 700.58.405 (P.H.J.K.) and VICI Grant 700.58.441 (M.U.),
and the Access to Research Infrastructures activity in the sixth Frame-
work Programme of the EC (Contract RII3-026145, EU-NMR).
*To whom correspondence should be addressed. R.B.: telephone, 0049
6813023005; fax, 0049 6813024739; e-mail, ritabern@mx.uni-saarland.de.
M.U.: telephone, 0031 715274628; fax, 0031 715275856; e-mail, m.ubbink@
chem.leidenuniv.nl
1
Abbreviations: Adx, adrenodoxin; AdR, adrenodoxin reductase;
CLaNP-5, caged lanthanide NMR probe 5; 7-DHC, 7-dehydrocholes-
terol; 7-DHP, 7-dehydropregnenolone; drms, distance root mean
square; rmsd, root-mean-square deviation; R
2para
, paramagnetically
enhanced transverse nuclear relaxation rate.
Article Biochemistry, Vol. 49, No. 32, 2010 6847
the orientation of proteins in a complex in the case of membrane-
associated proteins with poor availability and including a Fe
2
S
2
cluster in the interface. We demonstrate that with two Ln-tag
positions sufficient information is obtained to yield a reliable
model of the complex. This is the first structural model of the
AdR-Adx complex in solution.
MATERIALS AND METHODS
Chemicals.
Caged lanthanide NMR probe 5 (CLaNP-5) was
synthesized and chelated to various Ln ions (Ln-CLaNP-5,
Figure 1) as described (18). Superdex 75, Superdex 200, and Source
30Q columns were from GE Healthcare (Munich, Germany). All
other chemicals were reagent grade commercial products and
were used without further purification.
Mutagenesis and Expression of AdR Mutants.
AdR
mutants A, C364S/S425C/K429C, B, C364S/C145S/Q232C/
K236C, C, C364S/M352C/V357C, and D, C364S/K395C/H398C
were produced in cloning vector pMOSBlue, using the Quik-
Change site-directed mutagenesis kit (Stratagene, La Jolla, CA)
and subcloned into pBar1607, using the NdeI/Hind III sites (19).
Amplification of AdR cDNA in pMOSBlue was performed using
forward and reverse primers designed according to the manufac-
turers’ guidelines. All mutations were verified by DNA sequencing.
Protein Production.
The genes for the AdR variants were
heterologously coexpressed in Escherichia coli C43DE3 harboring
pBAR1607, using the chaperone-containing plasmid pGro12
(19-21). Purification was performed as described, and protein
concentration was determined using ε
450
=11.3mM
-1
cm
-1
(19,22). UV/vis spectra were recorded at room temperature on a
Shimadzu double-beam spectrophotometer UV2101PC using a 50
mM potassium phosphate buffer, pH 7.4, containing 0.1 mM
dithioerythritol, 0.1 mM EDTA, and 0.7 M KCl. Protein fractions
with an A
270
/A
450
ratio between 7.6 and 8.0 were collected, yielding
4 mg/L culture of wild type and 1.5-2 mg/L culture of the mutants.
CD spectra of the AdR forms were obtained on a Jasco 715
spectropolarimeter at room temperature. Reaction mixtures con-
tained 20 μM AdR in HEPES buffer, pH 7.4, containing 0.05%
Tween 20.
2
H
15
N-enriched wild type and (4-108) truncated Adx
were produced using E. coli BL21 and the pKKHC plasmid (23).
The cells were grown in M9 minimal medium containing
15
NH
4
Cl
and either glycerol-d
3
for Adx wild type or acetate-d
3
for Adx
(4-108). The D
2
Ocontentwasgraduallyincreasedupto99%.
Purification was performed as described using Source-30Q anion-
exchange and Superdex G75 gel filtration (24). Non-isotope-
enriched Adx was produced as described (25). Adx was quantified
using ε
415
=9.8mM
-1
cm
-1
, and protein with an A
414
/A
276
ratio
>0.9 was isolated (26). Yields were 1.4 mg/L for wild type and 2-3
mg/L for Adx (4-108). CYP11A1 was produced as described; the
protein concentration was determined by carbon monoxide dif-
ference spectroscopy using ε
450-490
=91mM
-1
cm
-1
(27,28).
Ln-CLaNP-5 Attachment to AdR.
Asolutionof0.2mM
AdR in 20 mM degassed sodium phosphate buffer, pH 7.4,
containing 0.1 M NaCl, was first treated with 5 mM freshly
prepared dithiothreitol for 1 h on ice. Superdex 200 gel filtration
was used to remove the surplus of dithiothreitol. The resulting
AdR solution was mixed with 3 mol equiv of Ln-CLaNP-5 and
incubated for 1 h on ice. Surplus of Ln-CLaNP-5 and protein
dimer was separated using Superdex 200 gel filtration. Protein
concentration was determined by UV/visspectroscopy. The yield
after the entire labeling procedure, including purification, was
about 20-25%. AdR has been labeled with Eu-CLaNP-5, Gd-
CLaNP-5, Tm-CLaNP-5, and Lu-CLaNP-5 for fluorescence,
paramagnetic relaxation enhancement, pseudocontact shift, and
diamagnetic control measurements, respectively.
Adx Reduction.
Reduction of full-length Adx was deter-
mined by UV/vis spectroscopy. Samples contained 50 μMfull-
length Adx, 5 μMAdR,and50μM NADPH. Quantitative Adx
reduction was determined indirectly by measuring subsequent
reduction of cytochrome cby Adx. The reduction of Adx by AdR
is rate limiting in this experiment (23,29). Reaction mixtures
contained 1 μM full-length Adx or Adx (4-108), 200 μMcytoch-
rome c,and0.05-1.0 μM AdR. The reaction was initiated by
adding 200 μM NADPH. Cytochrome creduction was moni-
tored at 550 nm, using ε
550
=20mM
-1
cm
-1
(30). All samples
contained 50 mM potassium phosphate buffer, pH 7.4. The
assays were performed on a Shimadzu double-beam spectro-
photometer UV2101PC at room temperature.
Substrate Conversion.
The conversion of 7-DHC to 7-DHP
was performed using a reconstituted assay system (31), contain-
ing 1 μMAdR,1μMAdx,1μM CYP11A1, and 200 μM7-DHC
in 50 mM HEPES-Tween 20 buffer. An NADPH-regenerating
system (5 μM glucose 6-phosphate, 1 U glucose-6-phosphate
dehydrogenase, and 1 μMMgCl
2
) was added, and the reaction
was initiated by adding 100 μM NADPH. After 10 min incuba-
tion at 37 °C the reaction was stopped by adding one sample
volume of chloroform. The steroids were extracted twice with
chloroform, dried, and resuspended in acetonitrile. Aliquots were
analyzed by HPLC, using a Nucleodur 100-3 C18 RP column
(Macherey-Nagel) with an isocratic mixture of acetonotrile-
2-propanol (30:1) and cholecalciferol as internal standard. Peak
areas of product and standard at 280 nm were determined using
the JASCO Borwin software.
Protein NMR.
Typically, NMR samples comprised 250-300
μL in a Shigemi tube, containing 20-50 μM
15
N
2
H Adx with 1.2
equiv of Ln-CLaNP-5-attached AdR in 20 mM sodium phos-
phate, pH 7.4, 6% D
2
O (v/v), and 100 mM NaCl. [
15
N,
1
H]-
TROSY spectra (32) were recorded at 285K on a Bruker Avance
DMX 600 spectrometer equipped with a TCI-Z-GRAD cryop-
robe. Data were processed in Azara (www.bio.cam.ac.uk/azara)
and analyzed in Ansig for Windows (33). The assignments of the
resonances were based on previous work (34). Chemical shift
perturbations (Δδ
avg
) were defined as described by Grzesiek
et al. (35). The agreement between observed and back-calculated
NMR parameters is expressed in a quality factor Q(eq 1), defined
as the ratio of the rmsd between observed and calculated data and
the rms of the sum of the observed and the calculated data (36).
Q¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
i
fΔδ
obs
PC
ðiÞ-Δδ
sim
PC
ðiÞg
2
P
i
fΔδ
obs
PC
ðiÞþΔδ
sim
PC
ðiÞg
2
v
u
u
u
u
tð1Þ
F
IGURE
1:
Structure of the Ln tag: caged lanthanide NMR probe 5
(CLaNP-5). The tag has been chelated to the Ln ions Eu
3þ
,Gd
3þ
,
Lu
3þ
,andTm
3þ
.
6848 Biochemistry, Vol. 49, No. 32, 2010 Keizers et al.
Paramagnetic NMR.
Pseudocontact shifts and paramag-
netic relaxation enhancements were determined as described
before (18). For the paramagnetic relaxation enhancements,
the paramagnetic over diamagnetic intensity ratios were normal-
ized using the resonances for residue T21, a residue with no
significant broadening, surrounded by other residues with no
significant broadening and located at the other side of the protein
relative to the residues for which the strongest effects were
observed. From the R
2para
, the Gd ion-to-amideproton distances
were calculated using the simplified Solomon and Bloembergen
equation (37). For 24 amides the resonances were assigned and
sufficiently resolved to yield reliable paramagnetic relaxation
enhancements. The obtained paramagnetic relaxation enhance-
ments were divided in two classes of restraints. The first class
(13 restraints) was formed by the amide resonances with an
intensity ratio up to 0.83. For these residues the distances as
determined from the paramagnetic relaxation enhancement
were used ((4A
˚). The second class (11 restraints) consisted of
the amide resonances with an intensity ratio of more than 0.83.
These residues where considered to be too far from Gd for
accurate distance determination and to these were assigned a
lower distance limit of 44 (-4, þ100) A
˚. As AdR is a membrane-
associated protein, it could only be concentrated to limited
extent. At concentrations higher than 50 μM, significant aggre-
gation occurred, even in presence of Adx. For this reason, for the
paramagnetic relaxation enhancement analysis, the method
according to Battiste and Wagner was used instead of the one
of Iwahara et al. (38,39). The latter method has been reported to
be less prone to artifacts, but it is less sensitive. Moreover, in our
hands the one-point method yields reliable distances and has
successfully been employed to determine structures of protein
complexes before (18,40).
Protein Docking.
To determine the solution model of the
complex formed by Adx and AdR, the proteins were docked
using the software XPLOR-NIH (41). The structure of AdR was
taken from the crystal structure of the cross-linked complex
(PDB entry 1E6E) and that of Adx from the crystal structure of
the free protein (PDB entry 1AYF). For inclusion of the
pseudocontact shift restraints, the PARArestraints module was
employed, and the Prodrug server was used to create FAD
parameter and topology files (42,43). Chemical shift perturba-
tions were treated as NOE-type distance restraints. Only the Adx
residues having an accessible surface area (ASA) of more than
40% were taken into account, as determined with the software
NACCESS 2.1.1 (44). A total of 16 amides with a Δδ
avg
of >0.03
ppm were given a distance of 11 (-8/þ15) A
˚to any amide
nitrogen atom of AdR, applying the sixth power distance
averaging option. The Ln-CLaNP-5 molecules and their accom-
panying χ-tensors were represented by a Ln ion and an axis
system, made out of three pseudoatoms, representing the χ-tensor
direction. The Ln ions were placed at 8 A
˚from the introduced
Cys CRatomsin the direction of the side chains, with the z-axis of
the χ-tensor initially pointing away from the backbone, as
described before (18). The magnitude of the χ-tensor of Tm-
CLaNP-5 was obtained previously and amounts 55.3 and 6.9
10
-32
m
3
for the axial and rhombic components, respectively (18).
The script of the docking procedure is supplied in the Supporting
Information. In short, the protein complex formed by AdR and
Adx was energy minimized guided by the energy terms for the
chemical shift perturbations, paramagnetic relaxation enhance-
ments, and pseudocontact shifts. Adx started at a random posi-
tion 60 (10 A
˚away from AdR, with a random rotation of 15°.
The energy minimization started with a rotation of the χ-tensors,
using only the pseudocontact shift energy term and keeping
proteins and Ln ions fixed. Subsequently, the proteins were
allowed to translate and rotate, guided by all the restraints, with
the Ln ions and their χ-tensors fixed relative to AdR. The
proteins were kept rigid, and only the backbone and Cβand
Cγatoms were considered in the repel function used to avoid
steric clashes. A total of 400 0.04 ps steps of restricted MD was
performed at 300 K, and subsequently the χ-tensor orientations
were optimized once more based on the pseudocontact shift
restraints, with proteins and Ln ions fixed. This iterative mini-
mization cycle was repeated 50000 times in a typical run. When
during the minimization the total energy did not decrease for six
cycles, Adx was placed at a random position 60 (10 A
˚away
from AdR, with a random rotation of 15°, and minimization was
continued. Structures of complexes were saved and pseudocon-
tact shifts and paramagnetic relaxation enhancements back-
calculated, when the total energy was below an energy cutoff.
Only the lowest energy structure was saved per approach from a
given starting position. The force constants used to scale the
chemical shift perturbations was set to 50, the force for pseudo-
contact shifts was set to 250, and for paramagnetic relaxation
enhancements it was set to 100. These were varied to yield similar
energy contributions, and it was verified that both the para-
magnetic relaxation enhancement and pseudocontact shift energy
terms played a role in ranking the top lowest energy structures.
Protein structures are visualized using Pymol (De Lano Scientific,
Palo Alto, CA).
RESULTS AND DISCUSSION
Mapping the AdR Binding Site of Adx.
AdR wild type was
titrated into a solution of truncated
2
H
15
N-enriched Adx (4-108)
from 0.3 to 1.2 mol equiv. After each addition, the TROSY NMR
spectrum was recorded (Figure S1 of the Supporting Informa-
tion). The disappearance and reappearance of shifted cross-peaks
are indicative of binding in the slow exchange regime, in line with
the results found before in surface plasmon resonance measure-
ments, in which a dissociation rate constant of 0.0038 s
-1
was
determined (25). Due to the presence of the Fe
2
S
2
cluster in Adx
many resonances are broadened beyond detection. Nevertheless,
several shifted cross-peaks could be assigned by comparison (6),
and chemical shift perturbations (Δδ
avg
) were determined (Figure 2).
Large shifts (Δδ
avg
> 0.15) were observed for residues 31, 40, 42,
43, 59, 60, 72, 76, 79, 85, and 96. From the binding map, it is clear
that Adx binds with its cofactor binding site toward AdR, as
inferred from the surrounding ring of large chemical shift pertur-
bations. This binding orientation is similar to the one observed in
the crystal structure of the complex and in line with the role of
Adx as an electron transfer protein (14). The stoichiometry of
binding cannot be revealed from a single set of chemical shift
perturbations from a protein complex in slow exchange, but the
binding map is consistent with a well-defined binding site in a 1:1
complex.
TROSY spectra were also acquired of the full-length
2
H
15
N-
enriched Adx in the absence and presence of AdR. The spectral
changes observed resembled those for the truncated Adx, indi-
cating a similar interaction (Figure S2 of the Supporting Informa-
tion). Large shifts (Δδ
avg
> 0.12) were observed for residues 14,
42, 43, 59, 75, and 96. The additional cross-peaks of the flexible
C-terminus remained unassigned. The 4-108 truncated form of
Adx was selected to study protein complex formation in more
Article Biochemistry, Vol. 49, No. 32, 2010 6849
detail, because Adx (4-108) is more stable than the full-length
protein (45).
Paramagnetic AdR.
To acquire intermolecular distance
restraints to steer the docking of AdR and Adx, AdR was
equipped with paramagnetic Ln tags. The dipolar interaction
between the unpaired electrons of the paramagnetic Ln ions and
the observed nuclei of Adx causes broadening and shifts of the
NMR resonances in a distance-dependent manner (46). CLaNP-
5 is a Ln tag that is attached via two Cys residues on the protein
surface (Figure 1). It shows minimal mobility relative to the
attached protein, and therefore, the paramagnetic effects are
large and the Ln ion position is easily modeled (18). By attaching
Ln-CLaNP-5 at two positions on the AdR surface, one at a time,
and adding this tagged AdR to
15
N
2
H isotope-enriched Adx,
interprotein distance information is obtained from the paramag-
netic changes in the NMR spectra of the protein complex. To
retrieve pseudocontact shifts, AdR was tagged with CLaNP-5
chelated to Tm
3þ
, for paramagnetic relaxation enhancement Gd-
CLaNP-5 was used, and the diamagnetic Lu-CLaNP-5 was taken
as a control. In the selection of the AdR variants (see below), Eu-
CLaNP-5 was used for its luminescent properties (18).
The addition of Ln-CLaNP-5 to wild type AdR led to exten-
sive loss of the FAD cofactor and large amounts of aggregated
protein. The cause of this was thought to be binding of the tag to
endogenous Cys residues, causing unfolding of the protein.
According to the AdR crystal structure, only the thiol of C145
is somewhat surface accessible. Nevertheless, to prevent the
aggregation, all five endogenous Cys residues were mutated to
Ser, but this AdR variant did not yield correctly folded protein.
Then, single Cys to Ser mutants were made and tested for
diminished binding of Eu-CLaNP-5, as compared to wild type.
The protein stability was monitored by gel filtration chromatog-
raphy. AdR variant C364S showed highly increased stability
toward tag treatment, and only a limited amount of labeling was
observedby correlating the Ln-tag fluorescence with flavin-based
protein absorbance (data not shown). C364 is in the NADPH
binding site. Probably the Ln tag can enter this site and
subsequently attach to C364, which leads to loss of the FAD
cofactor and aggregation of the protein. Therefore, the C364S
mutant was used as the starting point to make double Cys surface
mutants, required for CLaNP-5 attachment.
Positions for the Ln tag on AdR were selected 25-35 A
˚away
from the expected binding site of Adx, in rigid parts of the
protein. Cys mutations were designed preferably replacing one or
two Lys or Arg residues, to reduce the charge change upon
attachment of the CLaNP-5, which has a 3þoverall charge. It
was decided to make four double Cys mutants indicated with A,
B, C, and D (see Materials and Methods section), more than
required for obtaining docking restraints. Without optimization,
mutants A and B displayed excellent labeling efficiencies, whereas
for C and D this was not the case. Although optimization of the
labeling procedure would have been possible, it was deemed not
practical or necessary, because the study could be completed
using only mutants A and B.
AdR Mutant Characterization.
AdR mutants A and B were
produced and compared to the wild type enzyme. The three
proteins displayed the same characteristic UV/vis spectra indi-
cative of a correctly incorporated FAD (19) (Figure S3 of the
Supporting Information). Furthermore, both AdR mutants also
showed the same CD spectra as the wild type, confirming that
there are no severe structure differences introduced in the FAD
region. The AdR mutants were capable of reducing Adx, as
indicated by the changes in the Adx spectra (Figure S3 of the
Supporting Information). To quantify Adx reduction by the
mutants, the nonphysiological acceptor cytochrome cwas
used (23,29). Both mutants displayed somewhat higher K
m
values and slightly lower V
max
values than the wild-type enzyme
for the reduction of both full-length and truncated Adx, but no
severe loss of activity was observed (Table 1).
The AdR mutants were also used to convert 7-dehydrocho-
lesterol (7-DHC) to 7-dehydropregnenolone (7-DHP) in a re-
constituted CYP11A1 system. During substrate conversion both
F
IGURE
2:
Chemical shift perturbation (Δδ
avg
)analysisof
2
H
15
N
Adx (4-108) after binding of AdR. (A) The obtained Δδ
avg
of Adx
plotted per residue. (B) Chemical shift perturbations mapped on the
surface of the protein (structure from PDB entry 1E6E).Residues are
color coded according to their Δδ
avg
: larger than 0.15 ppm are
colored red, from 0.10 to 0.15 ppm orange, from 0.05 to 0.10 ppm
yellow, and less than 0.05 ppm blue. These levels are indicated by
dashed lines in (A). The residues for which information was lacking
are colored light gray and those broadened beyond detection due to
the Fe
2
S
2
cluster are colored dark gray.
6850 Biochemistry, Vol. 49, No. 32, 2010 Keizers et al.
AdR mutants showed only slight differences compared to
wild type AdR for both Adx forms (Table 2). Similar to what
was described before, product formation was found to be 60%
higher with the truncated Adx as compared to the full-length
protein (47).
Intermolecular Distance Restraints.
AdR mutants A and B
were labeled with Lu-CLaNP-5 and Tm-CLaNP-5, to C425/
C429 and C232/C236, respectively, and added to
2
H
15
N-enriched
Adx (4-108). [
1
H,
15
N]-TROSY spectra were acquired for all
complexes (Figure S4 of the Supporting Information), and
pseudocontact shifts were determined from the chemical shift
difference between the paramagnetic and the diamagnetic sam-
ples. The two AdR variants labeled with the diamagnetic tag
caused the same changes in the TROSY spectrum of Adx
(4-108) as did wild type AdR, indicating that labeling with
Ln-CLaNP-5 did not affect complex formation between AdR
and Adx. The paramagnetic tags caused pseudocontact shifts
ranging from 0.1 to less than -1 ppm (Figure 3). No residual
diamagnetic peaks were observed in the paramagnetic samples,
indicative of complete labeling. Tm-CLaNP-5 is a strong align-
ment agent, giving rise to HN residual dipolar couplings of up to
30 Hz at 14.1 T (18). Consequently, an error is introduced of half
the HN residual dipolar coupling in pseudocontact shifts ob-
tained from the TROSY cross-peaks, which may be up to 0.025
ppm for the proton. This value is small relative to the observed
pseudocontact shift and was neglected.
To obtain paramagnetic relaxation enhancement based re-
straints, AdR mutant B was labeled with Gd-CLaNP-5. Para-
magnetic relaxation enhancements were determined from the
intensity decreases in the TROSY cross-peaks of Adx (Figure S5
of the Supporting Information). The intensity ratios of the
paramagnetic and diamagnetic signals are mapped on the crystal
structure of Adx in Figure 4. Adx is facing the Ln tag with the
small helix consisting of residues 61-64 and the loop containing
residue 86.
The Structure of the AdR-Adx Complex.
The solution
structure of the complex was modeled by docking Adx on AdR,
guided by the experimentally obtained chemical shift perturba-
tions, pseudocontact shifts, and paramagnetic relaxation en-
hancements as intermolecular restraints. The starting structures
of AdR and Adx were taken from the crystallized complex and
the free protein, respectively (PDB entries 1E6E and 1AYF),
and treated as rigid bodies. AdR was equipped with Ln ions and
accompanying initial χ-tensors according to a protocol described
before (18). During the docking, the orientation of the χ-tensors
was optimized in an iterative fashion. The details of the protocol
are given in the Materials and Methods section. The lowest
energy complex structure is shown in Figure 5A, overlaid with
1E6E, with the AdR molecules superimposed. The location and
orientation of Adx on AdR are very similar for the solution
model and crystal complex. In both structures the Adx molecules
are equally far from the Ln ions. For instance, the distance from
the Ln ion in AdR mutant A, attached via C425 and C429, to the
CRof Adx N37 is 30.8 and 31.8 A
˚in the crystal and the best
docking solution, respectively. Similarly, the distance from the
other Ln ion, attached via C232 and C236, in AdR mutant B, is
23.3 and 23.6 A
˚to the CRof D86 of the crystal structure and the
best docking solution, respectively.
Table 1: Activity of AdR Wild Type and Mutants
full-length Adx Adx (4-108)
AdR K
m
(μM)
a
V
max
(nmol of Cyt cred/min)
a
K
m
(μM)
a
V
max
(nmol of Cyt cred/min)
a
wild type 0.4 (0.2 (4.6 (0.6) 10
2
0.6 (0.2 (4.8 (0.7) 10
2
mutant A 1.3 (0.5 (3.1 (0.9) 10
2
1.7 (0.5 (3.8 (0.6) 10
2
mutant B 0.8 (0.1 (3.7 (0.3) 10
2
0.8 (0.1 (4.3 (0.3) 10
2
a
The assay conditions are reported in the Materials and Methods section.
Table 2: CYP11A1-Dependent Conversion of 7-Dehydrocholesterol to
7-Dehydropregnenolone
k
cat
(s
-1
10
-3
)
a
AdR full-length Adx Adx (4-108)
wild type 13.4 (0.3 22.8 (0.1
mutant A 12.4 (0.1 20.5 (0.7
mutant B 13.3 (0.1 22.0 (0.3
a
The assay conditions are reported in the Materials and Methods
section.
F
IGURE
3:
Pseudocontact shifts (PCSs) of
2
H
15
N Adx (4-108) in
complex with Tm-CLaNP-5 AdR mutants A and B (A and B,
respectively) plotted per residue. The pseudocontact shifts were
obtained as described in the Materials and Methods section.
Article Biochemistry, Vol. 49, No. 32, 2010 6851
Relative to the crystal structure, in the lowest energy solution
Adx is slightly rotated around the core β-sheets consisting of
residues 56-58 and 88-90. The CRatoms of H56 are only 0.7 A
˚
apart and the CRatoms of L90 1.8 A
˚. The loop of Adx containing
the Fe
2
S
2
cluster is tilted to a small extent in the lowest energy
solution. This tilt causes the distance between the closest iron
atom of the Fe
2
S
2
cluster and the FAD isoalloxazine ring N3
atom to increase from 10.9 to 15.0 A
˚. This is still within the
critical electron transfer distance of 16 A
˚(48).
The total energy of the best docking solution is 19% lower than
that of the crystal structure after χ-tensor optimization, 13.3 kcal/
mol compared to 15.8 kcal/mol. This is reflected in the Q-value,
that measures the agreement between back-predicted and experi-
mental pseudocontact shifts and paramagnetic relaxation en-
hancement based distances (Figure 5B-D). Q, averaged over the
pseudocontact shifts and paramagnetic relaxation enhancement
classes of restraints, equals 0.110 and 0.102 for the crystal
structure and solution model, respectively. The lower Q-value
of the solution model can be attributed to a large extent to the Ln
tag attached to residues 425 and 429 (AdR mutant B). The
Q-values for the pseudocontact shifts originating from this tag
are 0.23 and 0.21 for the crystal structure and solution model,
respectively.
The rmsd between the lowest energy solution from the docking
and the crystal structure is 6.4 A
˚, with a distance root mean
square (drms) of 2.5 A
˚. The difference between the rmsd and the
drms is a consequence of the fact that the structures are rotated
rather than translated (49).
Single Minimum.
For the ensemble of low-energy structures
obtained during the docking calculations the restraint energy is
plotted against the rmsd relative to the lowest energy structure in
Figure 6A. As shown in the figure, the docking appears to
converge to a single minimum. All low-energy structures were
similar, with Adx bound in the groove between the FAD domain
and the NADPH domain of AdR and the Adx Fe
2
S
2
cluster
facing the FAD isoalloxazine ring. The centers of mass of Adx of
the top ten of low-energy solutions are displayed in Figure 6B.
These structures can all be found on a straight line equidistant
from the Ln ions, in the groove between the two AdR domains.
The farther the structure is from the minimum, the higher the
total energy. The crystal structure is found within the cloud. The
top ten low-energy structures form a cloud with an rmsd from the
mean of 3.2 A
˚.
As can be seen in Figure 6C,D, during the optimization the
χ-tensors rotate mostly in the xy-plane whereas the z-axis is
similar in all low-energy solutions; only inversion is observed.
The fact that the x-andy-axes display some freedom of movement
can be explained by the relatively low rhombicity of the χ-tensor,
making the pseudocontact shifts less sensitive to changes in the
x-andy-axes. The largest deviations of the average χ-tensors are
found for the structures with the highest total energies.
The outcome of the docking procedure is of course sensitive to
the relative weights given to the different restraints. The weights
for the paramagnetic relaxation enhancements and pseudocon-
tact shifts were chosen such that the energy terms for both wereof
similar magnitude and that both energy terms contribute to the
ranking of the lowest energy structures. This was based on the
assumption that both classes of restraints are equally important.
A possible source of error for the calculation of atomic distances
from paramagnetic relaxation enhancements is τ
c
.Basedupon
the crystallized complex, τ
c
was estimated by HYDRONMR (50)
to be 52.0 ns. The influence of τ
c
was verified by varying it from
43.0 to 57.0 ns and comparing the paramagnetic relaxation
enhancement based amide to Ln distances with the distances
from the best docking structure. The lowest Q-factor (considering
only paramagnetic relaxation enhancement derived distances)
was obtained for a value of 47.0 ns (Figure S6 of the Supporting
Information). However, the influence of τ
c
remains within the
error in the Q-factor at least in the range from 43 to 53 ns, and
simulations performed with τ
c
set to 47 or 49 ns gave similar
results.
Dependence of the Docking on the Adx Starting Con-
formation.
This study aims to provide a methodology for
F
IGURE
4:
Analysis of paramagnetic relaxation enhancement of the
2
H
15
N Adx (4-108) resonances by bound AdR mutant B with Gd-
CLaNP-5 attached. (A) The observed R
2para
plotted per residue.
(B) The observed R
2para
mapped on the surface of the protein, with
R
2para
>80s
-1
in red, 40 < R
2para
<80s
-1
in orange, 10 < R
2para
<
40 s
-1
in yellow, and R
2para
<10s
-1
in blue. These levels are
indicated by dashed lines in (A). The residues for which information
was lacking are colored light gray, and those broadened beyond
detection due to the Fe
2
S
2
cluster are colored dark gray.
6852 Biochemistry, Vol. 49, No. 32, 2010 Keizers et al.
finding a structure of a protein complex based on any given
structures of the free components, so the effect of different
starting structures should be small. The docking calculations
were repeated with different Adx structures, taken from crystal
structures of Adx in complex with AdR (PDB entry 1E6E) and as
a free oxidized protein (PDB entry 1CJE). For any of the three
Adx structures, a single minimum was found in the docking
calculations. The lowest energy solutions are very similar to the
one obtained using 1AYF-based Adx (Figure S7 of the Support-
ing Information). The backbone rmsd of Adx in the three
complexes, aligned by AdR, is only 0.9 A
˚, so it can be concluded
that the same minimum energy structure of the complex is found
independent of the starting structure, inaccord with the similarity
between the three Adx structures (backbone rmsd to the mean
is 0.3 A
˚).
The top ten lowest energy solutions, using the Adx structure
from 1CJE and 1E6E, form an ensemble with a backbone rmsd of
2.5 and 2.1 A
˚, respectively. Apparently, the accuracy of the
calculations is determined by the experimental restraints, whereas
the precision is dependent on the starting structures. The best
solution of the calculation starting with Adx from the crystallized
complex (1E6E) also has a somewhat lower energy compared to
the other two, 12.0 versus 13.3 and 13.7 kcal/mol for 1AYF and
1CJE, respectively, and consequently,also the Q-value for the fits
of back-calculated to the experimental restraints is a little lower,
0.098 versus 0.102 and 0.102. These findings show that side-chain
orientations have little influence on the orientation determined
for Adx in the complex but result in a slightly better fit if already
optimized for binding. This limited influence is a consequence of
the “soft docking” used in the protocol, with only the backbone
atoms and Cβand Cγtaken into account for the repel term in the
docking energy.
Methods for Protein Complex Structure Determination.
The combination of chemical shift perturbation, pseudocontact
shift, and paramagnetic relaxation enhancement restraints was
successfully used to obtain a model of the complex formed by
AdR and Adx in solution. We have also investigated whether all
restraints were required to obtain a reliable solution.
Several software packages such as Bigger and Haddock (51,52)
are available for the prediction of structures of protein complexes
based on a minimal set of experimental restraints, often giving
excellent results. Haddock was employed (53), to dock Adx on
AdR, using the experimental chemical shift perturbations as re-
straints. In the best docking solution, Adx was placed in between
the FAD and NADPH domains of AdR, but on the opposite
side of the isoalloxazine ring (Figure S8 of the Supporting
F
IGURE
5:
Structural comparison of the docking solution and the crystal structure. (A) The lowest energy docking solution of the AdR-Adx
complex superimposed on the crystal structure of the complex (PDB entry 1E6E). The structures are aligned by AdR, which is shown in surface
representation. The Ln ion positions are indicated by pink spheres, with the axis frame representing the χ-tensor of Tm in yellow for the x-and
y-axes and in blue for the z-axes. The positions of Adx as determined by NMR and in the crystal structure are shown as green and red cartoons,
respectively. The Fe
2
S
2
clusters are shown as spheres in CPK colors. (B, C) Plots of the pseudocontact shifts (PCSs) back-calculated from the
lowest energydocking solution (b) and the crystalstructure (3) against the experimental data are shown in (B) and (C) for AdR mutants A and B,
respectively. (D) The Gd-to-H
N
distances back-calculated from the lowest energy docking solution (b) and the crystal structure (3) are plotted
against the experimental paramagnetic relaxation enhancement (PRE) based distances. The solid lines represent the ideal correlation, and dashed
lines in (D) represent the 3 A
˚error margins used in the analysis. Q-values arefor (B) 0.210 (b) and 0.233 (3), for (C) 0.065 (b) and 0.067 (3), and
for (D) 0.031 (b) and 0.030 (3).
Article Biochemistry, Vol. 49, No. 32, 2010 6853
Information). In the second best solution Adx was placed at the
same place as in the crystallized complex, but rotated. Similar
results were obtained when the docking was performed in
XPLOR, using only the chemical shift perturbation restraints
(Figure S9 of the Supporting Information). Many of the solutions
were found with Adx in the groove between the NADPH domain
and the FAD domain. Like with Haddock, in the lowest energy
structure the Fe
2
S
2
cluster of Adx was facing the back of the FAD
in AdR. Therefore, to determine the solution structure of the
AdR-Adx complex, additional information about the binding
site on AdR is required.
Including the chemical shift perturbation and pseudocontact
shift but not the paramagnetic relaxation enhancement restraints
yielded a clear minimum as well with Adx facing the FAD
isoalloxazine ring with its Fe
2
S
2
cluster. There are only small
differences with the model obtained when the paramagnetic
relaxation enhancements are included. In the best structure
obtained without the paramagnetic relaxation enhancement
restraints, the Fe
2
S
2
cluster is rotated only 0.5 A
˚further away
from the FAD. The ten lowest energy structures form a cloud
of similar structures with an rmsd of 2.4 A
˚from the mean
(Figure S10 of the Supporting Information). The crystal structure
has a backbone rmsd of 3.1 A
˚with the mean of this ensemble. So
with only chemical shift perturbation and pseudocontact shift
restraints originating from two Ln positions it is possible to
retrieve a model of the complex with reasonable resolution.
Apparently, the addition of the paramagnetic relaxation en-
hancement restraints to the docking does not significantly change
the outcome of the docking. The lowest energy models obtained
using increasing sets of restraints are shown in Figure 7.
In principle, residual dipolar couplings could also aid in
structure determination. Ln-CLaNP-5 is a strong alignment
agent (54,55), potentially yielding the residual dipolar couplings
that can be used to refine protein structure (56). Unfortunately,
to deconvolute the residual dipolar coupling and pseudocontact
shift contributions to the paramagnetic shift in the TROSY
spectra, the antiTROSY spectrum is required as well. For the
low-concentration samples of the AdR-Adx complex, the sensi-
tivity required for the anti-TROSY spectra could not be achieved.
Solution versus Crystal Structure.
Thefactthatthebest
docking solution is highly similar to the crystallized complex
answers the longstanding question about the physiological
relevance of the cross-linked crystal structure. In this study, we
show that the crystallized complex structure fulfills the experi-
mental restraints found in solution by NMR very well (Figure 5).
In the crystal structure, Adx D39 and AdR K27 are chemically
F
IGURE
6:
Population analysis of the low-energy solutions from the
docking of Adx on AdR. (A) The energies of the docking solutions
are plotted against their rmsd relative to the lowest energy structure.
(B) The top ten lowest energy docking solutions of the Adx-AdR
complex are shown. AdR is displayed in surface representation.
Centers of mass of Adx are displayed as spheres, with the lowest
energy solution in green, the crystal structure in red, and the others in
brown. The Ln ion positions are indicated by pink spheres, with the
axis frame representing the χ-tensor of Tm in yellow for the x-and
y-axes and in blue for the z-axes. (C, D) Detailed views of the
χ-tensors of the Tm ions attached to mutants A and B, respectively.
F
IGURE
7:
The structures of the lowest energy solutions of the
Adx-AdR complex using increasingly more experimental restraints.
Adx structures are shown as cartoons, in blue using chemical shift
perturbations only, in yellow using chemical shift perturbations and
pseudocontact shifts, and in green using chemicalshift perturbations,
pseudocontact shifts, and paramagnetic relaxation enhancements,
and Adx from the crystal complex is in red. AdR is displayed as a gray
surface-filled structure and the Ln ion positions are indicated by pink
spheres, with an axis frame representing the χ-tensor of Tm in yellow
for the x-andy-axes and in blue for the z-axes.
6854 Biochemistry, Vol. 49, No. 32, 2010 Keizers et al.
cross-linked. Although this artificial covalent binding may have
had an effect on the crystallized complex, the distances between
Adx and AdR in the solution model are not very different from
those in the crystallized complex. For instance, the distance
between D39 and K27 decreases from 8.9 to 7.6 A
˚(CRto CR);
therefore, a cross-link between these residues as in the crystal
does not seem to be physically impossible.
CONCLUSIONS
We have established a model for the 65 kDa protein complex
formed in solution by AdR and Adx and found that thisstructure
is very similar to the crystallized complex of chemically cross-
linked AdR and Adx. The solution model was obtained using
TROSY NMR experiments in combination with perdeuteration
of the observed protein on the basis of paramagnetic restraints,
delivered by two Ln tags. Because of the strength and rigidity of
the Ln tag, significant long-range distance restraints up to 56 A
˚
could be obtained, even though the amount of AdR was limited.
Furthermore, the presence of the iron-sulfur cluster in Adx
makes the interface residues invisible in NMR experiments, so
only nuclei of Adx far from the interface could be studied.
Therefore, the paramagnetic NMR approach, using rigid Ln
tags, seems well suited as a general tool to determine structures of
difficult, large and poorly available targets.
ACKNOWLEDGMENT
We thank Dr. Hans Wienk and the NMR facility of the
Bijvoet Center in Utrecht for technical assistance and use of the
spectrometers.
SUPPORTING INFORMATION AVAILABLE
NMR spectra, wild-type Adx binding map, mutant AdR
validation, τ
c
optimization, structures of the AdR-Adx complex,
and the XPLOR-NIH docking script. This material is available
free of charge via the Internet at http://pubs.acs.org.
REFERENCES
1. Fern
andez, C., and W€
uthrich, K. (2003) NMR solution structure
determination of membrane proteins reconstituted in detergent
micelles. FEBS Lett. 555, 144–150.
2. Grzesiek, S., and Sass, H. J. (2009) From biomolecular structure to
functional understanding: new NMR developments narrow the gap.
Curr. Opin. Struct. Biol. 19, 585–595.
3. Kay, L. E. (2005) NMR studies of protein structure and dynamics.
J. Magn. Reson. 173, 193–207.
4. Hiller, S., and Wagner, G. (2009) The role of solution NMR in the
structure determinations of VDAC-1 and other membrane proteins.
Curr. Opin. Struct. Biol. 19, 396–401.
5. Bertini, I., Luchinat, C., Parigi, G., and Pierattelli, R. (2005) NMR
spectroscopy of paramagnetic metalloproteins. ChemBioChem 6,
1536–1549.
6. Xu, X., Reinle, W., Hannemann, F., Konarev, P. V., Svergun, D. I.,
Bernhardt, R., and Ubbink, M. (2008) Dynamics in a pure encounter
complex of two proteins studied by solution scattering and paramag-
netic NMR spectroscopy. J. Am. Chem. Soc. 130, 6395–6403.
7. Clore, G. M., Tang, C., and Iwahara, J. (2007) Elucidating transient
macromolecular interactions using paramagnetic relaxation enhance-
ment. Curr. Opin. Struct. Biol. 17, 603–616.
8. Lambeth, J. D. (1990) Enzymology of mitochondrial side-chain
cleavage by cytochrome P-450scc, Vol. 3, Akademie-Verlag, Berlin.
9. Bernhardt, R. (1996) Cytochrome P450: structure, function, and
generation of reactive oxygen species. Rev. Physiol., Biochem., Phar-
macol. 127, 137–221.
10. Bernhardt, R. (2006) Cytochromes P450 as versatile biocatalysts.
J. Biotechnol. 124, 128–145.
11. Ishimura, K., and Fujita, H. (1997) Light and electron microscopic
immunohistochemistry of the localization of adrenal steroidogenic
enzymes. Microsc. Res. Tech. 36, 445–453.
12. Hanukoglu, I. J. (1992) Steroid Biochem. Mol. Biol. 43, 779–804.
13. Ziegler, G. A., Vonrhein, C., Hanukoglu, I., and Schulz, G. E. (1999)
The structure of adrenodoxin reductase of mitochondrial P450 sys-
tems: electron transfer for steroid biosynthesis. J. Mol. Biol. 289, 981–
990.
14. Muller, A., Muller, J. J., Muller, Y. A., Uhlmann, H., Bernhardt, R.,
and Heinemann, U. (1998) New aspects of electron transfer revealed
by the crystal structure of a truncated bovine adrenodoxin, Adx-
(4-108). Structure 6, 269–280.
15. Pikuleva, I. A., Tesh, K., Waterman, M. R., and Kim, Y. (2000) The
tertiary structure of full-length bovine adrenodoxin suggests func-
tional dimers. Arch. Biochem. Biophys. 373, 44–55.
16. Ziegler, G. A., and Schulz, G. E. (2000) Crystal structures of
adrenodoxin reductase in complex with NADP
þ
and NADPH sug-
gesting a mechanism for the electron transfer of an enzyme family.
Biochemistry 39, 10986–10995.
17. Muller, J. J., Lapko, A., Bourenkov, G., Ruckpaul, K., and Heine-
mann, U. (2001) Adrenodoxin reductase-adrenodoxin complex struc-
ture suggests electron transfer path in steroid biosynthesis. J. Biol.
Chem. 276, 2786–2789.
18. Keizers, P. H. J., Saragliadis, A., Hiruma, Y., Overhand, M., and
Ubbink, M. (2008) Design, synthesis, and evaluation of a lanthanide
chelating protein probe: CLaNP-5 yields predictable paramagnetic
effects independent of environment. J. Am. Chem. Soc. 130, 14802–
14812.
19. Sagara, Y., Wada, A., Takata, Y., Waterman, M. R., Sekimizu, K.,
and Horiuchi, T. (1993) Direct expression of adrenodoxin reductase in
Escherichia coli and the functional characterization. Biol. Pharm. Bull.
16, 627–630.
20. Miroux, B., and Walker, J. E. (1996) Over-production of proteins in
Escherichia coli: mutant hosts that allow synthesis of some membrane
proteins and globular proteins at high levels. J. Mol. Biol. 260, 289–
298.
21. Nishihara, K., Kanemori, M., Kitagawa, M., Yanagi, H., and Yura,
T. (1998) Chaperone coexpression plasmids: differential and syner-
gistic roles of DnaK-DnaJ-GrpE and GroEL-GroES in assisting
folding of an allergen of Japanese cedar pollen, Cryj2, in Escherichia
coli.Appl. Environ. Microbiol. 64, 1694–1699.
22. Hiwatashi, A., Ichikawa, Y., Maruya, N., Yamano, T., and Aki, K.
(1976) Properties of crystalline reduced nicotinamide adenine dinu-
cleotide phosphate-adrenodoxin reductase from bovine adrenocorti-
cal mitochonria. I. Physicochemical properties of holo- and apo-
NADPH-adrenodoxin reductase and interaction between non-heme
iron proteins and the reductase. Biochemistry 15, 3082–3090.
23. Uhlmann, H., Kraft, R., and Bernhardt, R. (1994) C-terminal region
of adrenodoxin affects its structural integrity and determines differ-
ences in its electron transfer function to cytochrome P-450. J. Biol.
Chem. 269, 22557–22564.
24. Worrall, J. A., Reinle, W., Bernhardt, R., and Ubbink, M. (2003)
Transient protein interactions studied by NMR spectroscopy: the case
of cytochrome C and adrenodoxin. Biochemistry 42, 7068–7076.
25. Schiffler, B., Zollner, A., and Bernhardt, R. (2004) Stripping down the
mitochondrial cholesterol hydroxylase system, a kinetics study.
J. Biol. Chem. 279, 34269–34276.
26. Huang, J. J., and Kimura, T. (1973) Studies on adrenal steroid
hydroxylases. Oxidation-reduction properties of adrenal iron-sulfur
protein (adrenodoxin). Biochemistry 12, 406–409.
27. Pikuleva, I. A., Mackman, R. L., Kagawa, N., Waterman, M. R., and
Ortiz de Montellano, P. R. (1995) Active-site topology of bovine
cholesterol side-chain cleavage cytochrome P450 (P450scc) and evi-
dence for interaction of tyrosine 94 with the side chain of cholesterol.
Arch. Biochem. Biophys. 322, 189–197.
28. Omura, T., and Sato, R. (1964) The carbon monoxide-binding
pigment of liver microsomes. II. Solubilization, purification and
properties. J. Biol. Chem. 239, 2379–2385.
29. Lambeth, J. D., Seybert, D. W., and Kamin, H. (1979) Ionic effects on
adrenal steroidogenic electron transport. The role of adrenodoxin as
an electron shuttle. J. Biol. Chem. 254, 7255–7264.
30. Massey, V. (1959) The microestimation of succinate and the extinc-
tion coefficient of cytochrome c.Biochim. Biophys. Acta 34, 255–256.
31. Sugano, S., Miura, R., and Morishima, N. (1996) Identification of
intermediates in the conversion of cholesterol to pregnenolone with
a reconstituted cytochrome p-450scc system: accumulation of the
intermediate modulated by the adrenodoxin level. J. Biochem. 120,
780–787.
32. Pervushin, K., Riek, R., Wider, G., and W€
uthrich, K. (1997) Atte-
nuated T2 relaxation by mutual cancellation of dipole-dipole coupling
and chemical shift anisotropy indicates an avenue to NMR structures
of very large biological macromolecules in solution. Proc. Natl. Acad.
Sci. U.S.A. 94, 12366–12371.
Article Biochemistry, Vol. 49, No. 32, 2010 6855
33. Helgstrand, M., Kraulis, P., Allard, P., and Hard, T. (2000) Ansig for
Windows: an interactive computer program for semiautomatic as-
signment of protein NMR spectra. J. Biomol. NMR 18, 329–336.
34. Xu, X., Kim, S. K., Schurmann, P., Hirasawa, M., Tripathy, J. N.,
Smith, J., Knaff, D. B., and Ubbink, M. (2006) Ferredoxin/ferredox-
in-thioredoxin reductase complex: complete NMR mapping of the
interaction site on ferredoxin by gallium substitution. FEBS Lett. 580,
6714–6720.
35. Grzesiek, S., Bax, A., Clore, G. M., Gronenborn, A. M., Hu, J.,
Kaufman, J., Palmer, I., Stahl, S. J., and Wingfield, P. T. (1996) The
solution structure of HIV-1 Nef reveals an unexpected fold and
permits delineation of the binding surface for the SH3 domain of
Hck tyrosine protein kinase. Nat. Struct. Biol. 3, 340–345.
36. Bashir, Q., Volkov, A. N., Ullmann, G. M., and Ubbink, M. (2010)
Visualization of the encounter ensemble of the transient electron
transfer complex of cytochrome cand cytochrome cperoxidase.
J. Am. Chem. Soc. 132, 241–247.
37. Solomon, I., and Bloembergen, N. (1956) Nuclear magnetic interac-
tions in the HF molecule. J. Chem. Phys. 25, 261–266.
38. Battiste, J. L., and Wagner, G. (2000) Utilization of site-directed spin
labeling and high-resolution heteronuclear nuclear magnetic reso-
nance for global fold determination of large proteins with limited
nuclear overhauser effect data. Biochemistry 39, 5355–5365.
39. Iwahara, J., Tang, C., and Clore, G. M. (2007) Practical aspects of
(1)H transverse paramagnetic relaxation enhancement measurements
on macromolecules. J. Magn. Reson. 184, 185–195.
40. Vlasie, M. D., Fernandez-Busnadiego, R., Prudencio, M., and Ub-
bink, M. (2008) Conformation of pseudoazurin in the 152 kDa
electron transfer complex with nitrite reductase determined by para-
magnetic NMR. J. Mol. Biol. 375, 1405–1415.
41. Schwieters, C. D., Kuszewski, J. J., Tjandra, N., and Clore, G. M.
(2003) The Xplor-NIH NMR molecular structure determination
package. J. Magn. Reson. 160, 65–73.
42. Banci, L., Bertini, I., Cavallaro, G., Giachetti, A., Luchinat, C., and
Parigi, G. (2004) Paramagnetism-based restraints for Xplor-NIH.
J. Biomol. NMR 28, 249–261.
43. van Aalten, D. M., Bywater, R., Findlay, J. B., Hendlich, M., Hooft,
R. W., and Vriend, G. (1996) PRODRG, a program for generating
molecular topologies and unique molecular descriptors from coordi-
nates of small molecules. J. Comput.-Aided Mol. Des. 10, 255–262.
44. Gerstein, M. (1992) A resolution-sensitive procedure for comparing
protein surfaces and its application to the comparison of antigen-
combining sites. Acta Crystallogr., Sect. A 48, 271–276.
45. Burova, T. V., Beckert, V., Uhlmann, H., Ristau, O., Bernhardt, R.,
and Pfeil, W. (1996) Conformational stability of adrenodoxin mutant
proteins. Protein Sci. 5, 1890–1897.
46. Otting, G. (2008) Prospects for lanthanides in structural biology by
NMR. J. Biomol. NMR 42, 1–9.
47. Schiffler, B., Kiefer, M., Wilken, A., Hannemann, F., Adolph, H. W.,
and Bernhardt, R. (2001) The interaction of bovine adrenodoxin with
CYP11A1 (cytochrome P450scc) and CYP11B1 (cytochrome P45011beta).
Acceleration of reduction and substrate conversion by site-directed
mutagenesis of adrenodoxin. J. Biol. Chem. 276, 36225–36232.
48. Page, C. C., Moser, C. C., Chen, X., and Dutton, P. L. (1999) Natural
engineering principles of electron tunnelling in biological oxidation-
reduction. Nature 402, 47–52.
49. Kim, Y. C., Tang, C., Clore, G. M., and Hummer, G. (2008) Replica
exchange simulations of transient encounter complexes in protein-
protein association. Proc. Natl. Acad. Sci. U.S.A. 105, 12855–12860.
50. Garcia de la Torre, J., Huertas, M. L., and Carrasco, B. (2000)
HYDRONMR: prediction of NMR relaxation of globular proteins
from atomic-level structures and hydrodynamic calculations. J. Magn.
Reson. 147, 138–146.
51. Palma, P. N., Krippahl, L., Wampler, J. E., and Moura, J. J. G. (2000)
BiGGER: a new (soft) docking algorithm for predicting protein
interactions. Proteins 39, 372–384.
52. Dominguez, C., Boelens, R., and Bonvin, A. M. (2003) HADDOCK:
a protein-protein docking approach based on biochemical or biophy-
sical information. J. Am. Chem. Soc. 125, 1731–1737.
53. de Vries, S. J., van Dijk, M., and Bonvin, A. M. (2010) The HADDOCK
web server for data-driven biomolecular docking, Nat. Protoc. (in press).
54. Keizers, P. H. J., Desreux, J. F., Overhand, M., and Ubbink, M.
(2007) Increased paramagnetic effect of a lanthanide protein prob e by
two-point attachment. J. Am. Chem. Soc. 129, 9292–9293.
55. Xu, X., Keizers, P. H., Reinle, W., Hannemann, F., Bernhardt, R.,
and Ubbink, M. (2009) Intermolecular dynamics studied by para-
magnetic tagging. J. Biomol. NMR 43, 247–254.
56. Prestegard, J. H., and Kishore, K. I. (2001) Partial alignment of
biomolecules: an aid to NMR characterization. Curr. Opin. Chem.
Biol. 5, 584–590.