Biochem. J. (2012) 444, 395–404 (Printed in Great Britain)
395
doi:10.1042/BJ20111742
DNA binding by the plant-specific NAC transcription factors in crystal
and solution: a firm link to WRKY and GCM transcription factors
Ditte H. WELNER*, Søren LINDEMOSE†, J. Günter GROSSMANN‡, Niels Erik MØLLEGAARD§, Addie N. OLSEN†,
Charlotte HELGSTRAND*, Karen SKRIVER† and Leila LO LEGGIO*1
*Biophysical Chemistry Group, Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen Ø, Denmark, †Section for Biomolecular Sciences,
Department of Biology, University of Copenhagen, Ole Maaløes Vej 5, 2200 Copenhagen N, Denmark, ‡Molecular Biophysics Group, Institute of Integrative Biology, Faculty of Health
and Life Sciences, The University of Liverpool, Crown Street, Liverpool L69 7ZB, U.K., and §Department of Cellular and Molecular Medicine, Panum Institute, Faculty of Health
Sciences, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen N, Denmark
NAC (NAM/ATAF/CUC) plant transcription factors regulate
essential processes in development, stress responses and nutrient
distribution in important crop and model plants (rice, Populus,
Arabidopsis), which makes them highly relevant in the context
of crop optimization and bioenergy production. The structure of
the DNA-binding NAC domain of ANAC019 has previously
been determined by X-ray crystallography, revealing a dimeric
and predominantly β-fold structure, but the mode of binding
to cognate DNA has remained elusive. In the present study,
information from low resolution X-ray structures and small
angle X-ray scattering on complexes with oligonucleotides,
mutagenesis and (DNase I and uranyl photo-) footprinting,
is combined to form a structural view of DNA-binding,
and for the first time provide experimental evidence for the
speculated relationship between plant-specific NAC proteins,
WRKY transcription factors and the mammalian GCM (Glial
cell missing) transcription factors, which all use a β-strand motif
for DNA-binding. The structure shows that the NAC domain
inserts the edge of its core β-sheet into the major groove, while
leaving the DNA largely undistorted. The structure of the NAC–
DNA complex and a new crystal form of the unbound NAC also
indicate limited flexibility of the NAC dimer arrangement, which
could be important in recognizing suboptimal binding sites.
INTRODUCTION
NAC proteins consist of a conserved N-terminal region (NAC
domain) with DNA-binding and oligomerization abilities, and
a diverse C-terminal region which functions as a transcription
regulatory domain. Initial biophysical analysis suggested that the
NAC domain was dominated by β-strand structures [11]. This was
later confirmed by our determination of the first crystal structure
of a NAC domain, that of A. thaliana ANAC019, revealing a
novel dimeric transcription factor fold consisting of mainly βsheets and containing no well-known DNA-binding motif [12].
We have also shown by EMSA (electrophoretic mobility-shift
assay) that the ANAC019 NAC domain binds DNA as a dimer
and characterized the dimer interface [12] as well as identified
a consensus DNA sequence (CGT[GA]) to which ANAC019
and other relatively distant NAC TFs bind [13–16]. Despite the
considerable interest in these proteins since the publication of
the structure [e.g. 2,3,7–10,17], no experimental information
on their molecular interactions with DNA has been published,
besides mutational studies [13].
In the present study, interaction of the ANAC019 NAC domain
with DNA is probed structurally by crystallography and in
solution, providing for the first time experimental evidence for
a DNA-binding mode similar to that of the plant-specific WRKY
TF family [18,19] and the mammalian family of GCM (Glial
cell missing) TFs [20]. The elucidation of the interaction between
ANAC019 and a cognate dsON (double-stranded oligonucleotide)
is thus a key in consolidating the relationship between these three
TF families, which have previously been noted to have structural
similarities [5,21].
Transcription is a fundamental biological process, and a large
part of the molecular machinery is conserved among kingdoms.
However some TF (transcription factor) families, such as NAC
(NAM/ATAF/CUC) proteins, have only been identified in plants.
NAC proteins constitute a highly prolific group of plant-specific
TFs, with representatives in monocots, dicots, conifers and
mosses. Many plants of commercial and scientific interest, such
as Arabidopsis thaliana [1], tobacco [2] and rice [3] have more
than 100 different NAC proteins. Early findings indicated an
important role of NAC proteins in plant development [4]. This has
later been confirmed, and a variety of other important processes
such as biotic and abiotic stress responses and senescence have
been associated with NAC proteins, as reviewed in [5]. A recent
example is the work of Carvallo et al. [6] that suggested that NAC
proteins are part of an evolutionarily conserved cold response,
since NAC019 was up-regulated in response to cold in distantly
related plant species.
Research has also suggested a useful role for NAC proteins in
bioenergy production from lignocellulosic material, since a subset
of NAC proteins, now termed the Secondary Wall NACs, were
shown to influence cell wall thickness [7,8]. Applications in crop
plant engineering have also been demonstrated, e.g. improvement
of rice drought tolerance and yield on expression of a NAC protein
in the roots [9], and increase of wheat protein, zinc and iron
content by replacement of a non-functional NAC gene allele in
modern wheat with a functional ancestral counterpart [10].
Key words: crystal, DNA-binding, small angle X-ray scattering
(SAXS), transcription factor, uranyl photo-probing.
Abbreviations used: DIG, digoxigenin; dsON, double-stranded oligonucleotide; EMSA, electrophoretic mobility-shift assay; GCM, Glial cell missing;
MR, molecular replacement; NAC, NAM/ATAF/CUC; NACBS, NAC-binding site; PEG, poly(ethylene glycol); SAXS, small-angle X-ray scattering; SEC,
size-exclusion chromatography; SNAC1, stress-responsive NAC 1; TF, transcription factor.
Co-ordinates and structure factors have been deposited into PDB with accession codes 3SWP (Native_complex), 3SWM (Au_complex) and 4DUL
(Form IV).
1
To whom correspondence should be addressed (email [email protected]).
c The Authors Journal compilation c 2012 Biochemical Society
396
Table 1
D. H. Welner and others
Data, refinement and validation statistics
Data were collected at 100 K. The complex data were processed with XDS [25], and the form IV data with Denzo/Scalepack [24]. R int is the redundancy independent R -factor on the intensities [47] for
the complex data or the conventional R sym for form IV. Ramachandran statistics are calculated with Rampage [48]. Parentheses denote the values for the highest resolution shell. n.a., not applicable;
RMSD, root mean square deviation.
Parameter
Native_complex
Au_complex
Form IV
Data collection wavelength
Space group
Unit cell
Resolution
R int
I /σ
Completeness
Redundancy
R -factor
R -free
RMSD bonds
RMSD angles
PDB code
Protein atoms
DNA atoms
Gold atoms
Ramachandran statistics (favoured, allowed, outliers)
RMSD of DNA atoms from ideal B-DNA
1.038 Å
P 21 2 1 2 1
a = 69.1 Å, b = 105.5 Å, c = 175.2 Å
20–4.40 Å (4.53–4.40 Å)
10 % (63 %)
17.1 (4.0)
99.4 % (100.0 %)
8.7 (9.0)
26.0%
34.9%
0.015 Å
2.484 ◦
3SWP
4639 with average B = 245.3 Å2
1066 with average B = 330.3 Å2
n.a.
97.0 %, 2.8 %, 0.2%
1.514 Å
1.038 Å
P 21 21 21
a = 68.5 Å, b = 109.1 Å, c = 173.9 Å
20–4.45 Å (4.58–4.45 Å)
11 % (64 %)
10.9 (2.6)
99.4 % (99.6 %)
3.8 (3.9)
25.0%
30.9%
0.015 Å
2.285 ◦
3SWM
4639 with average B = 236.6 Å2
1066 with average B = 275.0 Å2
9 with average B = 415.6 Å2
96.6 %, 3.2 %, 0.2%
1.304 Å
1.087 Å
P 21
a = 40.1 Å, b = 69.1 Å, c = 75.9 Å, β = 97.4 ◦
30–3.00 Å (3.11–3.00 Å)
5.3 % (39.2 %)
25.2 (2.3)
98.9 % (91.3 %)
3.7 (3.2)
25.7%
34.2%
0.012 Å
1.820 ◦
4DUL
2097 with average B = 99.0 Å2
n.a.
n.a.
90.4 %, 7 %, 2.6%
n.a.
EXPERIMENTAL
Protein and oligonucleotide samples
The NAC domain of ANAC019 (At1g52890, residues 1–168,
NCBI Accession number NP_175697) with an N-terminal
His6 tag was expressed from vector pET-15b, purified as
described previously [13] and stored in 20 mM Tris/HCl
(pH 7.5), which is the buffer used for all experiments in the
present study unless otherwise stated. The 26 bp oligo2 (5 gtcttgcgtgttggaacacgcaacag-3 ) was designed with 5 overhangs
to facilitate oligomerization and crystal growth [22], and contains
two palindromically oriented NACBSs (NAC-binding sites; as
defined in [13], sequence TTGCGTG, in bold) separated by
a 6 bp spacing. ssONs (single-stranded oligonucleotides) were
purchased from TAG Copenhagen and annealed to a final
concentration of 250 μM dsON in 10 mM Tris/HCl (pH 7.5)
and 50 mM NaCl. The sequence of the 30 bp blunt-ended
dsON used in EMSA was similar to the sequence of oligo2
(5 -cagtcttgcgtgttggaacacgcaacagtc-3 ; NABCS in bold).
Crystallization and data collection
A new crystal form of the free ANAC019 NAC domain (form
IV) was obtained under similar conditions to the ones reported
previously [23], using 12.5 % PEG [poly(ethylene glycol)] 4000,
0.1 M malic acid/imidazole buffer (pH 7.0) and 5 % glycerol, and
a new batch of protein (4.0–5.8 mg/ml). Data were collected at
beamline 911-2, MAXLAB (Lund, Sweden) and processed with
Denzo/Scalepack [24] (statistics in Table 1).
For the complexed structures, NAC dimer and oligo2 were
mixed in a 1:1 molar ratio. After extensive screening (including
different crystallization conditions and a set of ten dsON), the
conditions leading to crystallization were identified. Crystals
grew within two weeks in manually set up 24-well VDXTM
trays (Hampton Research) using 0.01 M MgSO4 ·7H2 0, 0.05 M
Mes (pH 6.0) and 0.5 % PEG 400 as main precipitant. The
hanging drop was composed of equal amounts (2 μl) of
reservoir solution and protein–DNA mixture (with protein at a
1 mg/ml concentration). Crystals routinely diffracted to 8–9 Å (1
Å = 0.1 nm), and two were found to diffract to ∼ 4.45 Å. One
c The Authors Journal compilation c 2012 Biochemical Society
of them (Native_complex) was grown in the presence of 0.3 M
glycyl–glycyl–glycine as an additive. A dataset (Native_complex)
was collected at beamline I911-2 of MAXLAB at cryogenic
conditions using 22 % glycerol as a cryoprotectant. Additives did
not consistently improve diffraction quality, thus the differences
in diffraction power must be due to variation in individual
crystal quality and handling. Dehydration, annealing and other
postcrystallization improvement techniques did not improve on
the resolution limit. Later in the study, attempts were made
to obtain derivatives with gold-, platinum-, mercury- and leadcontaining compounds. Data were collected at beamline ID14-4
of the ESRF (Grenoble, France) for a trimethyl lead acetate and
two potassium dicyanoaurate derivative crystals, but finally only
one (Au_complex, crystal was soaked for 20 h in 5 mM potassium
dicyanoaurate) was deemed of sufficient quality for structure
determination. Data collection statistics for processing with XDS
[25] for the two complex data sets are reported in Table 1.
Structure determination
The structure of the free form IV was determined by MR
(molecular replacement) using Phaser within the PHENIX suite
[26,27] and an edited NAC monomer (PDB code 1UT7) as
the search model. The MR solution was deemed correct, as it
showed a similar dimer interface as the one identified in the high
resolution crystal structure. Refinement was carried out initially
in Phenix.refine [27] and later in Refmac 5 [28] using jelly-body
restraints, NCS (non-crystallographic symmetry) restraints and
group TLS (Translation/Libration/Screw) refinement, alternated
with manual rebuilding in Coot [29]. This strategy resulted in
acceptable geometry and R-factors (R-factor of 25.7 % and Rfree of 34.2 %) for a model comprising 75 % of the amino
acids in the dimer. The B molecule makes fewer crystal contacts
and was particularly difficult to build in some regions. Further
details on structure determination and validation are given in
the Supplementary online data (at http://www.BiochemJ.org/bj/
444/bj4440395add.htm) and the refinement statistics are given in
Table 1.
The Native_complex data were phased with MR using
Phaser v1.3.1 [26]. The structure is shown in Figure 1 and
Supplementary Movie S1 (at http://www.BiochemJ.org/bj/444/
DNA binding by NAC plant transcription factors
Figure 1
397
Crystal structure of the NAC–DNA complex (Native_complex)
Binding sites according to the assigned DNA sequence are in yellow.
bj4440395add.htm). There were several high-ranking solutions,
which only differed by translation of the DNA along its major axis.
Thus oligo2 packs in the crystal structure as long fibres in a headto-tail arrangement, basically as expected from the design strategy.
Based on MR alone, it was difficult to assign unambiguously the
correct sequence to the DNA. Since a model lacking the bases
would be problematic both in terms of crystallographic refinement
and SAXS (small-angle X-ray scattering) profile simulation,
potentially leading to wrong refinement and conclusions on the
basis of the χ-values, the sequence was instead assigned by
combining the information of protected phosphates in uranyl
photo-probing (Figure 2A) with the crystal structure, to obtain the
best possible match between phosphate protection in solution and
reduction in phosphate solvent accessibility on complex formation
with the AB dimer, as derived from the crystal structure. The
match is also shown in Figure 2(A). All of the nucleotide bases
for oligo2 are thus included in the final model, which is highly
plausible in view of the biochemical data obtained in solution.
Refinement was carried out in phenix.refine [27] and minor
manual rebuilding and inspection of maps were done in Coot v0.6
[29]. The CCP4 package suite v6.1.3 [30] was used for various
tasks in phasing, refinement, validation and analysis. Many
refinement protocols were tested. The best compromise between
R-factors and geometry was obtained using the protocol described
in further detail in the Supplementary online data, which resulted
in R-factors of 26.0 % and 34.9 % for Native_complex, and
25.0 % and 30.9 % for Au_complex. Validation is summarized
in the results section, with further details in the Supplementary
online data. The statistics for both models are shown in Table 1.
Figures showing structures and electron density were prepared
with PyMOL v1.3 (http://www.pymol.org). Analysis of surface
accessibility on the basis of the crystal structures was carried out
with AreaImol from the CCP4 package [30,31], using the default
probe radius of 1.4 Å. Surface area differences upon NAC–DNA
complex formation were calculated per protein amino acid residue
(to identify residues listed in Table 2) or DNA phosphate group
(for the phosphate accessibility analysis in Figure 2A).
Uranyl photo-cleavage analysis and DNase I footprinting
The uranyl photo-cleavage analysis and DNase I footprinting were
performed as described previously using a sample volume of 30 μl
Figure 2 Backbone phosphate accessibility of ANAC019–DNA complex
according to uranyl photo-probing assay and the complex crystal structure
(A) Model of phosphate contacts in the ANAC019–DNA complex. The model represented by
arrows is built from four independent uranyl photo-footprinting experiments on both strands
of the oligonucleotide. The data reflect phosphate interactions in the presence of 200 ng of
ANAC019. The palindromic core NACBS is underlined and the length of the arrows denotes the
relative protection level (i.e. semi-quantitative footprints). Black bold letters indicate nucleotides
whose phosphate group loses more surface area than the threshold value on complex formation
with the AB dimer in the Native_complex crystal structure. The threshold value was chosen
as 40 % of lost accessible area compared with the phosphate losing most surface area on
complex formation, according to the AreaImol analysis, which was set to 100 %. Open bold
letters represent nucleotides whose phosphate groups were judged as likely to be protected by
residues in the loops/termini missing in the crystal structure, on the basis of their approximate
spatial orientation. (B) Differential cleavage plots. The differential cleavage plots compare
the susceptibility of the individual DNA nucleotides to uranyl photo-cleavage in the absence
and presence of 200 ng of ANAC019. Negative values (threshold = 0.1) correspond to a
protein-protected phosphate, and positive values represent enhanced cleavage (more accessible
phosphates upon ANAC019 binding). The vertical scale is in units of ln(f a ) − ln(f c ), where f a is
the fractional cleavage at any bond in the presence of 200 ng of ANAC019 and f c is the fractional
cleavage of the same band in the control (0 ng). The NACBSs are underlined and black boxes
indicate phosphate footprints on both strands.
and up to 200 ng ANAC019 per sample [32–34]. In brief, the
purified [32 P]-labelled DNA fragments were subjected to either
DNase I or uranyl photo-cleavage in binding buffer (see [13]). The
concentration of uranyl nitrate was 1 mM and the concentration
of ANAC019 is indicated in the text. Since uranyl photo-cleavage
is absent above pH 7, the pH of the buffers were adjusted to
6.0–6.5. The samples were incubated for 30 min at 23 ◦ C and
then irradiated in open tubes placed just below a 40W/03 Phillips
fluorescent light tube with maximum emission at 420 nm. After
25 min of irradiation, sodium acetate at pH 4.5 was added to a final
concentration of 0.2 M together with 2.5 vol. of 96 % ethanol on
c The Authors Journal compilation c 2012 Biochemical Society
398
Table 2
D. H. Welner and others
DNA-binding residues of ANAC019
Overview of the residues close to DNA as detected by an analysis of surface and contact areas upon DNA binding performed with AreaImol [30,31]. For simplicity, the residues have been placed in
six groups according the secondary structure element they are part of. The sequence of each group is displayed, with non-conserved residues in regular type and conserved in bold. Parentheses
indicate less likely, but possible, interactions.
Group number
Chain A
Chain B
Chain C
Chain D
Sequence
Location in NAC
1
86–88
86–87
–
86–87
P86 NR
2
3
4
5
6
94–101
113–118, 120
123–125, 129–130
133, 135, 138
160, 162–163
95–101
114–116, 118, 120
123–125, 129–131
133, 135, 138
160, 162–163
–
–
–
–
–
95–101
115, 117–118, 120
123–125, 129
133
162
Y94 WKATGTD
I113 KKALVFY
K123 APKGTKTNW
I133 MHEYR
Y160 KKQ
Poorly defined loop
between β1 and β2
β3
β4
Loop between β4 and β5
β5
C-terminal
ice for 15 min, and the mixtures were centrifuged at 15 000 g for
20 min. The dried pellet was dissolved in 6 μl formamide, 90 mM
Tris/borate and 1 mM EDTA (pH 8.3), containing xylene cyanol
and Bromophenol Blue, and the samples were heated at 90 ◦ C
before loading on to a 8 or 10 % denaturing polyacrylamide gel
(19:1 acrylamide/methylenebisacrylamide). The autoradiograms
were obtained by overnight exposure using intensifying screens.
A Molecular Dynamics STORM 860 PhosphorImager was used
to collect data from the phosphor storage screens and baseline corrected scans were obtained by using the software SAFA
package [35]. Differential cleavage plots were calculated from the
expression ln(f a ) − ln(f c ) representing the differential cleavage
at each bond relative to the control (where f a is the fractional
cleavage at any bond in the presence of the protein and f c
is the fractional cleavage of the same bond in the control).
Using this expression, positive values indicate enhanced cleavage
(hypersensitivity), whereas negative values indicate cleavage
inhibition (footprints).
DNA-interaction
region
Probable function in
DNA binding
Backbone
Increasing affinity
Major groove
(Backbone)
Backbone
(Backbone)
Backbone
Specific recognition
None/increasing affinity
None/Increasing affinity
None
Increasing affinity
ctcctaaaggtactgcaaacaattggatca-3 ; and K162A, 5 -gggttctatgtcgaatatacaaggcgcaatcaagtgcacaaaaacaa-3 .
EMSA with DIG (digoxigenin) labelling (Roche) was
performed as described previously [12], except for the
DNA sequence which was as described under protein and
oligonucleotide samples.
In order to determine the K d value of the interaction between
wt (wild-type) ANAC019 and oligo2, three independent EMSAs
(n = 3) with full titrations of ANAC019 were carried out with
32
P-labelled oligo2 instead of DIG-labelling. The EMSAs were
performed as described in [13], but the DNA concentration was
kept at 75 pM, which is roughly 1000-fold lower than the expected
K d value (results not shown). Autoradiograms were obtained
by exposure using intensifying screens. A Molecular Dynamics
STORM 860 PhosphorImager was used to collect data from
the phosphor storage screens and ImageQuant software version
5.2 (Molecular Dynamics) was used to calculate the fractional
saturation. The data was used to establish a binding isotherm,
which were fitted to the ‘one site binding’ (hyperbola) equation:
Y = (Bmax ×X)/(K d + X) using GraphPad Prism version 5.00.
SAXS analysis
The free ANAC019 domain alone was investigated by SAXS in
order to confirm that the previously determined X-ray structure
(PDB code 1UT4) was compatible with the solution structure.
Data for a concentration series (1–8 mg/ml) in 20 mM Tris/HCl
(pH 7.5) were measured at beamline 2.1, SRS (Daresbury,
England). Additional information on data analysis and modelling
are given in the Supplementary online data.
ANAC019(1–168) (1 mg/ml) and oligo2 (250 μM) were mixed
in equimolar amounts and incubated at room temperature (23 ◦ C)
for 15 min. The sample was then fractionated on a pre-packed
24 ml Superdex 75 column (GE Healthcare) with a flow rate of
0.5 ml/min with 20 mM Tris/HCl (pH 7.5) as the running buffer.
The resulting peak fractions were analysed separately with SAXS
at beamline 711, MAXLAB. Non-peak fractions were used as
background subtraction. Exposure time was 20 s. A concentration
series was recorded on each fraction after concentration with
a Microcon Ultracel YM-10. The data were analysed with the
programs Raw [36] and the ATSAS program package [37].
Site-directed mutagenesis and EMSA
ANAC019(1–168), produced as described above, and mutated
variants thereof, were used for EMSA. Site-directed mutagenesis
was performed with QuikChangeTM II Site-Directed Mutagenesis
kit (Stratagene). The mutagenic primers were the following (sense
strand): R85A, 5 -aatatccaaacgggtcagcacctaaccgggttgcc-3 ;
R88A, 5 -gggtcaagacctaacgcggttgccggatcgg-3 ; K129A, 5 -aaag
c The Authors Journal compilation c 2012 Biochemical Society
Dimer angle calculations
Dimer closure was quantified by defining a vector for each NAC
monomer going through the dimerization interface as well as the
protein core and then calculating the angle between the vector pair
of the dimer. The vector was constructed through the Cα-positions
of residues 18 (part of the dimerization β-sheet) and 135 (situated
in the centre of the core β-sheet). The angle between the vectors
was calculated using the basic formula: cos(α) = a·b/|a|x|b|.
RESULTS AND DISCUSSION
Crystal structure of the NAC–DNA complex
Lack of detailed information on the DNA-binding mode prompted
us to initiate co-crystallization of the ANAC019 NAC domain and
dsONs containing a palindrome of the 7 bp ANAC019-binding
site (NACBS) identified previously [13]. Although the literature
indicates that some NAC TFs target single NACBSs in vivo
[13,38], palindromic NACBSs bind with higher affinity [13] and
were thus chosen. After extensive optimization of crystals, a native
data set to 4.40 Å (Native_complex) and one derivatized with gold
(Au_complex) to 4.45 Å were collected for a complex with oligo2
(data collection statistics are shown in Table 1).
The MR solutions obtained with Native_complex and
Au_complex are very similar and show a ternary complex
consisting of two NAC dimers (chains AB and CD) bound to
DNA binding by NAC plant transcription factors
399
one DNA duplex constituted by chains EF (Figure 1). Combined
with the low resolution, the unusual stochiometry prompted
thorough validation. In short, crystal packing, a biologically
sensible DNA–protein interface, omit maps and similar gold sites
as those originally used to solve the NAC structure alone all
confirmed the MR solution (more details in Supplementary online
data and Supplementary Figure S1 at http://www.BiochemJ.
org/bj/444/bj4440395add.htm). The DNA sequence was assigned
to agree with the pattern of phosphate protection in the uranyl
photo-probing assay described below. The agreement is shown in
Figure 2(A), which shows the phosphate protection in solution
as well the phosphates losing the largest accessible surface
on complexation with the AB dimer in the Native_complex
crystal structure, according to the sequence as assigned in the
deposited PDB file. Additionally, phosphates potentially protected
by protein residues missing in the crystallographic model are
marked, on the basis of the vicinity with missing loops or extended
termini. As it is not possible to completely exclude a possible
small register error, in our analysis we refrain from discussing
interactions with individual bases. As a final validation that the
overall structure is fundamentally correct, the structures could
be refined as well as expected with the limited resolution, to
Rwork = 26.0 % and Rfree = 34.9 % for the Native_complex model
(Table 1).
Each monomer inserts into adjacent major grooves of the DNA
as is common for TFs (Figure 1), and a change in dimer opening
with respect to the previously determined free structure [12] can
be observed, as will be discussed below.
DNase I footprinting and uranyl photo-cleavage analysis of
ANAC019–DNA interactions
DNase I and uranyl photo-footprinting analysis were used to
map the binding in solution. Using binding conditions described
previously [13], the DNase I results clearly demonstrate protection
of the palindromic NACBSs in oligo2 by ANAC019 in a
concentration-dependent manner (Supplementary Figure S2 at
http://www.BiochemJ.org/bj/444/bj4440395add.htm).
Uranyl photo-cleavage can be used to study phosphate contacts
involved in protein– and ligand–DNA complexes [32,33]. The
technique exploits the fact that the uranyl-(VI) cation (UO2 2 + )
forms strong complexes with accessible phosphates of the DNA
backbone and upon irradiation is excited, which initiates a
complex photo-oxidation process eventually leading to strand
break. Binding of a protein to DNA protects phosphates near the
binding site from oxidation. Using this method, a dsON model of
specific protein–phosphate contacts was built using full titrations
(0, 10–200 ng) of ANAC019 (Figure 2A). On the basis of this
model, the DNA sequence in the crystal structure was assigned.
The differential cleavage plot (Figure 2B) shows a nearly perfect
symmetrical pattern when both strands are compared. We thus
conclude that the two monomers in the dimer interact similarly
with the backbone phosphates in each of the individual NACBS
in the palindrome, and no additional binding site was detected
that could support a ternary complex in solution.
Solution structures of the free NAC domain and the
NAC–DNA complex
SAXS gives low resolution information on the structure of
macromolecules and macromolecular complexes in solution.
The data obtained (see Figure 3A, the Supplementary online
data and Supplementary Table S2 at http://www.BiochemJ.org/
bj/444/bj4440395add.htm for further details) show that the free
Figure 3
SAXS analysis
(A) SAXS data for the free ANAC019 NAC domain in solution is shown alongside the CRYSOL
theoretical scattering curve for one of the published crystal structures (PDB code 1UT7) and for
a ‘completed’ model where loops and termini missing in the crystal structure have been built
and the termini perturbed using elNémo [49] (more details in the Supplementary online data
at http://www.BiochemJ.org/bj/444/bj4440395add.htm). χ-Values for the fits are shown on the
Figure. (B) Ab initio DAMFILT model (P1) based on the SAXS data shown in (A) and superposed
with the NAC dimer from the crystal structure. (C) Ab initio DAMFILT models based on SAXS
data (shown in D) from SEC fractions B1 and B3 compared with the Native_complex crystal
structure. The ternary complex is shown in cartoon with the AB dimer in magenta, the CD dimer
in green and the DNA in black. It is superposed using DAMSUP with the DAMFILT models
from B1 (blue mesh) and B3 (yellow spheres). The ab initio reconstructions are consistent
with fraction B1 containing DNA and fraction B3 containing binary complex. (D) SAXS data
from SEC fractions B1–B3 are shown together with the theoretical scattering curves for selected
models derived from the complex crystal structure. For fraction B2 a fit for a model obtained
with OLIGOMER containing 82 % free DNA and 18 % binary complex is additionally shown.
χ-Values for the fits are shown on the Figure. Residual curves highlight the goodness of the
fits at different values of S.
NAC dimer in solution is primarily symmetrical within the
resolution of SAXS (Figure 3B). Evidence for this comes
from ab initio shape modelling imposing 2-fold symmetry,
giving χ-values very similar to the ones obtained without
symmetry imposed (see the Supplementary online data). The
newly published crystal structure of the SNAC1 (stress-responsive
NAC 1) NAC domain [39] also supports the notion that NAC
domains in general form symmetrical dimers, since in the SNAC1
structure the two monomers in the NAC dimer are related by
crystallographic symmetry and are thus identical, although local
asymmetry has been noted previously for the ANAC019 NAC
domain crystal structure [12]. The differences noted in this crystal
structure are beyond the resolution of the SAXS technique. The
c The Authors Journal compilation c 2012 Biochemical Society
400
D. H. Welner and others
SAXS data for free NAC are reasonably well represented by the
known crystal structures of the free ANAC019 NAC domain
(the fit to the structure with PDB code 1UT7 is shown as example
in Figure 3B and Supplementary Table S2). The fit is further
improved by modelling loops and termini missing in the crystal
structure, as can also be seen in Figure 3(B) and Supplementary
Table S2 (see the Supplementary online data for details on how
this complete model was generated). These loops and termini
most likely adopt a variety of conformations in solution.
In order to isolate the complex, SEC (size-exclusion
chromatography) on a NAC/DNA mixture was carried out and
yielded an asymmetric peak (eluting at 10.2 ml, results not
shown), which we interpreted as containing one or more complex
species as well as free DNA. SAXS data could be collected from
three individual peak fractions, designated B1 to B3, where B1
represents the higher-elution-volume fraction. The experimental
curves were compared with different models: (i) oligo2; (ii) the
expected binary complex of one NAC dimer to one dsON created
by removing one dimer from the Native_complex structure;
and (iii) the ternary complex seen in the crystal structure. Free
protein species were excluded, since they elute at higher volume
(∼ 11.5 ml). CRYSOL analysis [40] showed that although the
curves from fractions B1 and B2 fit the free dsON model better
(χ-values of 1.3 and 1.7 respectively), fraction B3 yields an
improved fit using the binary complex as model (χ-value = 1.1)
(Figure 3D). Analysis of the scattering curves with OLIGOMER
[41] enhances the fit to the B2 data by adding 18 % binary complex
to 82 % free oligo2 (Figure 3D), but did not detect the presence of
ternary complex in any of the fractions. Ab initio modelling with
DAMMIN (Figure 3C) supported the CRYSOL/OLIGOMER
results, in that the model from fraction B1 fitted well with free
oligo2 and the model from fraction B3 fitted well with the binary
complex, but could not accommodate the ternary complex. The
relevant calculated and experimental Rg -values are provided in
Supplementary Table S2.
That NAC binds DNA as a dimer has been shown previously
[13]. The DNase I footprinting and uranyl photo-cleavage
data in the present study are fully consistent with symmetric
binding of one NAC dimer to one oligo2 in solution. Additional
higher-molecular-mass bands are observed in EMSA when high
amounts (100–500 ng) of NAC domain are used (e.g. in [13]),
which we interpret as non-specific binding. In the crystal the
high concentrations and dsON arrangement might also favour
non-specific binding leading to an artifactual ternary complex.
Recognition of non-optimal DNA sequences may have relevance
in vivo where many target sequences have single NACBSs,
which must entail non-specific recognition by one of the NAC
monomers.
ANAC019 residues interacting with DNA
As it was not known a priori which of the NAC dimers in the
crystal best represents the interactions with DNA in solution,
the interactions of all monomers with DNA were analysed. First,
an analysis of the surface and contact areas was performed.
The interface between DNA and protein is of comparable size
2
for chains A, B and D (695 +
− 80 Å ). The contact area for chain B
is shown in Figure 4(A), and as illustrated in Figure 4(B) overlaps
well, as expected, with a large positively charged surface patch.
The contact area of chain C, however, is only 129 Å2 , indicating
that chain C is only loosely associated with DNA.
The residues pinpointed by AreaImol were grouped into six
clusters on the basis of their secondary structure (Table 2 and
Figure 5A). The interacting groups are similar for monomers
c The Authors Journal compilation c 2012 Biochemical Society
Figure 4
Important areas of the ANAC019 NAC domain
For simplicity only one NAC monomer is shown (chain B). (A) Surface residues in the NAC–DNA
interface as defined by AreaImol are highlighted in blue. (B) The electrostatic surface generated
with PyMOL. (C) Conserved residues are highlighted in green and residues that are similar in
more than half of the ten aligned NAC domain sequences are coloured yellow. This panel is
based on the sequence alignment of Jensen et al. [16]. (D) Red spheres indicate the residues
found not to be essential for DNA binding in [13]. In the present study we further tested single
mutants of Arg85 (not modelled in crystal structures) and Arg88 (green) and found that Arg88 is
necessary for DNA binding. (E) Important residues of the ANAC019 NAC domain. Conserved
residues are in black boxes and residues that are similar in more than half of the sequences are
in grey boxes, according to the alignment of [16]. Above, blue bars indicate residues at the DNA
interface in any of the four NAC monomers, see Table 2. Stars indicate residues that have been
mutated and found by EMSA to be essential for DNA binding (green) or not (red) in accordance
with (D). An open circle indicates an arginine or a lysine residue close enough to DNA to be an
interaction partner (see Table 3).
A, B and D. On the basis of the fact that uranyl photo-probing
shows a symmetric interaction between NAC and the investigated
DNA sequence, dimer AB probably best represents the interaction
with oligo2 in solution, and thus our further analysis focuses
on this dimer. Group 2 protrudes into the DNA helix and
interacts with the sugar/base region of DNA. These residues
are probably responsible for specific recognition. Their sequence
(WKATGTD) is highly conserved, except for the second threonine
residue [16], and has been suggested to be reminiscent of the
WRKYGQK motif [42], which is responsible for DNA binding
in WRKY TFs [19,21] (see below). Groups 1, 4 and 6 are part
of loop regions and each contains one or more arginine or lysine
residues with adequate distances to DNA to allow interaction with
the backbone (see Table 3 for details). These residues are likely to
contribute to the overall affinity of DNA binding. The direct role of
groups 3 and 5 in DNA binding is arguable, since they are part of
the central β-sheet and do not contain potential phosphate-binding
residues. In the remaining groups, Arg88 , Lys96 , Lys123 , Lys129
and Lys162 can potentially interact with backbone phosphates and
are all conserved (Figures 4C and 4E), except for Lys123 , which
is either lysine or arginine, and in one instance proline [VOZ2
(vascular plant one-zinc finger protein 2)] [16]. Residues 79–85
and 144–151 as well as part of the N- and C-terminal could not
be modelled in the crystal structure and were not included in the
DNA binding by NAC plant transcription factors
Table 3
401
Arginine and lysine residues in the ANAC019 NAC domain
The shortest distance (in Å) to a DNA backbone atom in the Native_complex structure was
measured from the Cβ-atoms of all lysine and arginine residues. The maximum distance from
Cβ to Nζ of lysines is approximately 4.8 Å, and that from Cβ to NH2 of arginine residues is
approximately 6.0 Å. Giving a bonding length of approximately 3 Å, the maximum distance from
Cβ to a DNA backbone atom must be 7.8 Å for lysine and 9.0 Å for arginine residues in order for
them to be probable candidates for DNA-interacting residues. Those that meet this criterium are
underlined. This approach allows identification of potentially interacting residues independent
of the accuracy of side-chain-atom positions. Conserved residues are in bold, according to the
multiple sequence alignment of [16]. Residues not modelled in the crystal structure are not
taken into consideration.
Distances (Å)
Residue name
Chain A
Chain B
Arg19
Arg34
Lys35
Lys53
Lys62
Lys68
Arg76
Arg88
Lys96
Lys102
Arg110
Lys114
Lys115
Lys123
Lys126
Lys129
Arg138
Arg158
Lys161
Lys162
15.4
25.1
25.7
15.4
20.6
11.6
12.5
6.0
6.9
9.1
13.5
9.1
9.7
6.4
8.3
5.2
10.8
10.0
9.2
4.8
15.4
24.1
23.6
16.0
22.0
12.3
12.8
7.1
7.5
8.5
15.0
8.9
9.1
7.5
10.0
4.9
11.0
10.1
9.0
4.6
analysis, but some of these residues could potentially interact in
the complex, for example Arg85 .
The K d value of the interaction between the ANAC019 NAC
domain and oligo2 is estimated in the present study by EMSA to be
24.2 +
− 6.2 nM. We have previously reported that a K123A/K126A
variant of the NAC domain was not affected in its DNA binding
as measured by EMSA, whereas the R85A/R88A variant was
impaired [13]. To further investigate this and other potentially
interacting residues, the single mutants R85A, R88A, K129A
and K162A were made and analysed by EMSA. The ANAC019
residues that have been mutated to date are mapped on to the
structure of the complex in Figure 4(D).
The R88A mutant showed significant impairment in DNA
binding (Supplementary Figure S3 at http://www.BiochemJ.
org/bj/444/bj4440395add.htm). On the other hand, the other
single mutations had no detectable effect (see for example
R85A, also shown in Supplementary Figure S3), indicating K d
values similar to the native or changes below the detection level
of EMSA. Arg85 is absent from all of the currently available
structures of ANAC019 NAC domains, including the complexes
with DNA, but is included in the recently reported structure
of SNAC1 NAC domain [39]. Though the resolution of this
structure (2.6 Å) is lower than the highest resolution obtained
for the ANAC019 NAC domain (1.9 Å for PDB code 1UT7),
some additional residues could be traced in the missing loops,
including Arg85 . If the structure of SNAC1 (PDB code 3ULX) is
superposed on to the AB dimer in the Native_complex structure,
Arg88 is in a conformation ideally positioned for interaction with
one of the DNA phosphate groups, whereas Arg85 is pointing away,
offering a possible structural explanation for the mutagenesis
results. Although the implication of Lys129 and Lys162 in binding
Figure 5
DNA interactions and dimer flexibility
(A) The residues contacting the DNA according to the contact surface analysis are shown,
together with a cartoon representation of the DNA and a transparent depiction of the NAC
surface. Groups 3 and 5 are shown in green, to illustrate the fact that these residues are part of
the core β-sheet. Groups 1 and 6 (blue) as well as 4 (red) contact the DNA backbone and are
shown in sticks. The interesting group 2, which protrudes into the major groove, is depicted
in yellow sticks. (B) Dimer arrangement in different crystal structures of NAC, representing the
open, semi-closed and closed dimer arrangements. The B monomers of a previously solved free
ANAC019 NAC structure (open arrangement, PDB code 1UT7, in green, N-terminal fragment
connected with a broken line), the AB dimer (semi-closed arrangement, from Native_complex, in
yellow) and CD dimer (closed arrangement, from Native_complex, in red) were superimposed,
showing clearly that A monomers are not aligned with each other. (C) Superimposition of the
DNA-binding strands of GCM (MRNTNNHN, PDB code 1ODH) (cyan) and NAC (WKATGTDK)
(green). The DNA from the GCM structure is shown in grey together with the DNA from the NAC
complex structure as a black ribbon. (D) Superimposition of the DNA-binding strands of WRKY
(WRKYGQK, PDB code 2LEX) (cyan) and NAC (WKATGTDK) (green). The DNA from the
WRKY structure is shown in grey together with the DNA from the NAC complex structure
as a black ribbon.
is strongly suggested by the structures in complex with DNA
reported in the present study, it could be that removal of a single
one of these side chains does not change the affinity sufficiently
for EMSA detection. In conclusion our mutagenesis studies show
that, out of the residues mutated so far, Arg88 is the most important
for DNA affinity.
NAC dimer flexibility
A certain amount of flexibility in the dimer arrangement is
evidenced by comparison (Figure 5B) of the structures of
complexes shown in the present study, the previously solved
native structures (PDB codes 1UT7 and 1UT4), and the newly
determined native structure to 3.0 Å resolution (form IV, the
Experimental section and the Supplementary online data for
details). The degree of closure in the dimer was quantified as
an angle between the domains (see the Experimental section for
details) and values for the different crystal structures of ANAC019
and for comparison also of the newly determined SNAC1 structure
[39] are tabulated in Table 4. The values fall roughly into three
groups, representing what we refer to as open, semi-closed and
closed dimer arrangements, with angles of approximately 120 ◦ ,
110 ◦ and 100 ◦ between monomers.
In order to see how well the various dimer arrangements are
represented in solution, three models were made and compared
with the SAXS data for the free NAC domain. In all cases
c The Authors Journal compilation c 2012 Biochemical Society
402
D. H. Welner and others
Table 4 Angles between dimers in crystal structures of NAC domains,
calculated as described in the Experimental section
Crystal structure
Angle
Dimer arrangement
Free ANAC019 (PDB code 1UT4)
Free ANAC019 (PDB code 1UT7)
Free ANAC019 (Form IV)
Native_complex AB dimer
Au_complex AB dimer
Free SNAC1 [39] (PDB code 3ULX, crystallographic dimer)
Native_complex CD dimer
Au_complex CD dimer
120 ◦
119 ◦
108 ◦
107 ◦
107 ◦
97 ◦
102 ◦
100 ◦
Open
Open
Semi-closed
Semi-closed
Semi-closed
Closed
Closed
Closed
the models were based on NAC monomers from the highest
resolution free ANAC019 crystal structure (PDB code 1UT7),
completed to account for missing loops and termini as described
in the Supplementary online data, and for which the CRYSOL
fit is shown in Figure 3(A). The CRYSOL fit of the three
dimer models were compared: an open arrangement (119 ◦ angle),
representing the original free ANAC019 crystal structures; a semiclosed arrangement (107 ◦ angle), representing the form IV free
ANAC019 crystal structure and the AB dimers in the crystal
complexes with DNA; and a closed arrangement (100 ◦ angle),
representing the free SNAC1 structure and the CD dimers in the
complexes. As one approaches more closed dimer arrangements,
both the fit to the experimental Rg - and the χ-values became worse,
as shown in Supplementary Table S2. Our conclusion is that in
solution the free NAC dimer is mostly in an open conformation.
To a minor extent, more closed conformations may also exist in
solution, and lead to form IV and the SNAC1 crystals. From the
evidence so far, it would appear that on the DNA, NAC proteins
take more closed conformations, but it cannot be excluded that
NAC can also bind DNA in the open dimer arrangement. A
possible residue involved in mediating the transition from open
to closed conformation is Tyr21 . Tyr21 in form IV of ANAC019
and the corresponding His21 in SNAC1 are interestingly more
similar than in the original free ANAC019 crystal structures.
However, especially since the resolution of both form IV and
SNAC1 structures is limited, this suggestion should be confirmed
by further experimental investigations.
The recognition of single NACBSs in vivo by the homodimeric
NAC domains [38] would entail the binding of one of the
monomers to a suboptimal nucleotide sequence. The ability
of the dimer to open and close slightly would provide a
molecular mechanism by which the NAC dimer binds to a highaffinity site with one monomer, whereas the other monomer can
then sample the nearby bases for the second highest-affinity
binding site available in non-palindromic sequences. Furthermore
this flexibility explains how NAC can recognize palindromic
sequences in vitro which are separated by a variable number
of bps: 5 bp as described in [15]; 6 bp in oligo2; and 7 and
8 bp in variants of oligo2, although a drop in affinity is observed
(S. Lindemose and K. Skriver, unpublished work).
DNA conformation
An ideal B-DNA model of oligo2 was used as a search model
in MR, and this form is largely conserved after refinement. DNA
bending is not uncommon with TFs [43] and the related GCM
TF (see section below) bends DNA 30 ◦ [44]. However, the very
recent report of DNA complex of a WRKY domain, which is
also structurally related to NAC domains (see below), shows
only slight distortion from B-DNA conformation [19]. We see
no evidence of DNA bending in the complex crystal structures,
c The Authors Journal compilation c 2012 Biochemical Society
but one could imagine a scenario where one NAC dimer bends
DNA in one direction and the other NAC dimer reverses this,
so that the DNA still can pack in straight fibres in the crystal.
Slight distortions like this would be undetected at this resolution
or could be lost in the crystal if the lattice favours an unbent DNA
conformation.
In the uranyl photo-probing experiments some phosphates
exhibit hypersensitivity towards uranyl cleavage (positive values
in Figure 2B). This is most pronounced for the 5 -GCGTGTTG-3
sequence (and 5 -GCGTGTTC-3 on the lower strand). This may
indicate a local distortion in the DNA structure upon ANAC019
binding which will elicit a change in the electrostatic potential of
the (minor) groove, thus creating phosphates more accessible to
uranyl photo-cleavage [31]. However, the exact nature and extent
of the structural changes cannot be deduced.
Similarity to the WRKY and GCM TFs
An analogy between NAC, WRKY and GCM transcription factors
has been suggested previously [13,21,42]. The present structure
confirms this relationship and shows that the DNA-interaction
mode is indeed analogous to that of GCM [44] and WRKY, for
which very recently an NMR structure in complex with DNA
has been published [19]. Although there are clear differences,
for example GCM TFs are monomeric, the three proteins have
a central β-sheet with very similar topology. When the sheets
are superimposed, the DNA-binding strand of NAC aligns with
the DNA-binding strands of GCM and WRKY, with a resulting
similar orientation of the bound dsONs (Figures 5C and 5D).
A glycine residue in NAC (Gly99 ) and WRKY has been noted
to induce an unusual curvature of the DNA-binding strand [21].
From the superimpositions (Figures 5C and 5D) it is clear that this
is a point of divergence in the Cα traces of the three strands, which
otherwise align well. The importance of this glycine is underlined
by the dysfunction of the sog1-1 (suppressor of gamma response)
NAC mutant protein caused by a change of the glycine residue
into an arginine [45].
The present study highlights the similarity in the DNA-binding
mode of NAC, WRKY and GCM. The similarities, previously
only suggested, but now firmly established, together with similar
topologies of the central β-sheet have both previously led to the
proposal of a common origin of the three different TF families
[19,42]. However, since similarities are not apparent overall at the
primary structure level, other explanations, such as mechanistic
convergence of DNA-binding by the β-strand, are also possible.
Concluding remarks
Despite great interest in plant stress and associated NAC TFs,
there is very little molecular description of the interaction with
DNA. We therefore undertook the study of the interaction of
ANAC019, a NAC TF whose structure we have previously solved
[12], with DNA. We show in the present study the first crystal
structure of a NAC–DNA complex and complement it with
solution studies. The structure shows that ANAC019 interacts
with DNA via the curved outer β-strand of its core β-sheet which
encompasses the conserved WKAT sequence, and establishes
firmly the relationship with WRKY and GCM transcription
factors. The structural similarity among these TFs, coupled with
differences in DNA sequence preference, could inspire protein
engineering aimed at modulating target gene specificity. Further,
we show indications of NAC dimer flexibility as a possible
mechanism for adapting to different spacings between binding
sites, adding NAC to the list of TFs for which dimer flexibility may
play a role in biological function, a classic example being λ phage
DNA binding by NAC plant transcription factors
Cro [46]. The structure provides the first framework to understand
the interactions of NAC TFs with DNA at the molecular level.
This can guide the study of the molecular mechanisms of NAC
TF cellular function with resulting increased ability to exploit the
NAC proteins in technological applications.
AUTHOR CONTRIBUTION
The planning of experiments, overall analysis and writing of the paper were performed by
Ditte Welner, Leila Lo Leggio and Karen Skriver; Søren Lindemose and Charlotte Helgstrand
additionally contributed to writing the paper. All authors contributed to experiments and/or
analysis of the experimental data: Søren Lindemose and Niels Erik Møllegaard mostly
for the DNase I footprinting and uranyl photo-cleavage; Ditte Welner and Leila Lo Leggio
mostly for the crystallography and SAXS with additional contributions from Charlotte
Helgstrand (crystallography) and J. Günter Grossmann (SAXS); and Addie Olsen and
Karen Skriver mostly for the mutagenesis.
ACKNOWLEDGEMENTS
We thank Heidi Ernst for helpful discussions, Charlotte O’Shea, Jens-Christian Poulsen
and Dorthe Boelskifte for technical help and the synchrotrons (MAXLAB and ESRF) for
beamtime.
FUNDING
This work was supported by the Danish Agency for Science Technology and Innovation
[grant numbers 10-084503 (to L.L.L.), 274-07-0173 (to K.S.) and 10-093596 (to S.L.).
Travel to synchrotrons was supported by the DANSCATT programme, funded by the Danish
Agency for Science Technology and Innovation [grant number 09-074479] and the ELISA
programme, funded by the European Community’s Seventh Framework Programme [grant
number 226716].
REFERENCES
1 Ooka, H., Satoh, K., Doi, K., Nagata, T., Otomo, Y., Murakami, K., Matsubara, K., Osato,
N., Kawai, J., Carninci, P. et al. (2003) Comprehensive analysis of NAC family genes in
Oryza sativa and Arabidopsis thaliana . DNA Res. 10, 239–247
2 Rushton, P. J., Bokowiec, M. T., Han, S., Zhang, H., Brannock, J. F., Chen, X., Laudeman,
T. W. and Timko, M. P. (2008) Tobacco transcription factors: novel insights into
transcriptional regulation in the Solanaceae . Plant Physiol. 147, 280–295
3 Nuruzzaman, N., Manimekalai, R., Sharoni, A. M., Satoh, K., Kondoh, H., Ooka, H. and
Kikuchi, S. (2010) Genome-wide analysis of NAC transcription factor family in rice. Gene
465, 30–44
4 Aida, M., Vernoux, T., Furutani, M., Traas, J. and Tasaka, M. (2002) Roles of
PIN-FORMED1 and MONOPTEROS in pattern formation of the apical region of the
Arabidopsis embryo. Development 129, 477–486
5 Olsen, A. N., Ernst, H. A., Lo Leggio, L. and Skriver, K. (2005) NAC transcription factors:
structurally distinct, functionally diverse. Trends Plant Sci. 10, 79–87
6 Carvallo, M. A., Pino, M. T., Jeknic, Z., Zou, C., Doherty, C. J., Shiu, S. H., Chen, T. H. and
Thomashow, M. F. (2011) A comparison of the low temperature transcriptomes and CBF
regulons of three plant species that differ in freezing tolerance: Solanum commersonii ,
Solanum tuberosum , and Arabidopsis thaliana . J. Exp. Bot. 62, 3807–3819
7 Hu, R. B., Qi, G., Kong, Y., Kong, D., Gao, Q. and Zhou, G. (2010) Comprehensive
analysis of NAC domain transcription factor gene family in Populus trichocarpa . BMC
Plant Biol. 10, 145
8 Shen, H., Yin, Y. B., Chen, F., Xu, Y. and Dixon, R. A. (2009) A bioinformatic analysis of
NAC genes for plant cell wall development in relation to lignocellulosic bioenergy
production. Bioenerg. Res. 2, 217–232
9 Jeong, J. S., Kim, Y. S., Baek, K. H., Jung, H., Ha, S.-H., Choi, Y. D., Kim, M., Reuzeau, C.
and Kim, J.-K. (2010) Root–specific expression of OsNAC10 improves drought tolerance
and grain yield in rice under field drought conditions. Plant Physiol. 153, 185–197
10 Uauy, C., Distelfeld, A., Fahima, T., Blechl, A. and Dubcovsky, J. (2006) A NAC gene
regulating senescence improves grain protein, ainc, and iron content in wheat. Science
314, 1298–1301
11 Greve, K., La Cour, T., Jensen, M. K., Poulsen, F. M. and Skriver, K. (2003) Interactions
between plant RING-H2 and plant-specific NAC (NAM/ATAF1/2/CUC2) proteins:
RING-H2 molecular specificity and cellular localization. Biochem. J. 371, 97–108
12 Ernst, H. A., Olsen, A. N., Skriver, K., Larsen, S. and Lo Leggio, L. (2004) Structure of the
conserved domain of ANAC, a member of the NAC family of transcription factors. EMBO
Rep. 5, 297–303
403
13 Olsen, A. N., Ernst, H. A., Lo Leggio, L. and Skriver, K. (2005) DNA-binding specificity
and molecular functions of NAC transcription factors. Plant Sci. 169, 785–797
14 Mitsuda, N., Hisabori, T., Takeyasu, K. and Sato, M. H. (2004) VOZ: isolation and
chracterization of novel vascular plant transcription factors with a one-zinc finger from
Arabidopsis thaliana . Plant Cell Physiol. 45, 845–854
15 Xue, G.-P. (2005) A CELD-fusion method for rapid determination of the DNA-binding
sequence specificity of novel plant DNA-binding proteins. Plant J. 41, 638–649
16 Jensen, M. K., Kjaersgaard, T., Nielsen, M. M., Galberg, P., Petersen, K., O’Shea, C. and
Skriver, K. (2010) The Arabidopsis thaliana NAC transcription factor family:
structure–function relationships and determinants of ANAC019 stress signaling.
Biochem. J. 426, 183–196
17 Kunieda, T., Mitsuda, N., Ohme-Takagi, M., Takeda, S., Aida, M., Tasaka, M., Kondo, M.,
Nishimura, M. and Hara-Nishimura, I. (2008) NAC family proteins NARS1/NAC2 and
NARS2/NAM in the outer integument regulate embryogenesis in Arabidopsis . Plant Cell
20, 2631–2642
18 Rushton, P. J., Somssich, I. E., Ringler, P. and Shen, Q. J. (2010) WRKY transcription
factors. Trends Plant Sci. 15, 247–258
19 Yamasaki, K., Kigawa, T., Watanabe, S., Inoue, M., Yamasaki, T., Seki, M., Shinozaki, K.
and Yokohama, S. (2012) Structural basis for sequence-specific DNA recognition by an
Arabidopsis WRKY transcription factor. J. Biol. Chem. 287, 7683–7691
20 Wegner, M. and Riethmacher, D. (2001) Chronicles of switch hunt: gcm genes in
development. Trends Genet. 17, 286–290
21 Yamasaki, K., Kigawa, T., Inoue, M., Watanabe, S., Tateno, M., Seki, M., Shinozaki, K. and
Yokoyama, S. (2005) Solution structure of an Arabidopsis WRKY DNA binding domain.
Plant Cell 17, 944–956
22 Tan, S., Hunziker, Y., Pellegrini, L. and Richmond, T. J. (2000) Crystallization of the yeast
MATa2/MCM1/DNA ternary complex: general methods and principles for protein/DNA
cocrystallization. J. Mol. Biol. 297, 947–959
23 Olsen, A. N., Ernst, H. A., Lo Leggio, L., Johansson, E., Larsen, S. and Skriver, K. (2004)
Preliminary crystallographic analysis of the NAC domain of ANAC, a member of the
plant-specific NAC transcription factor family. Acta Crystallogr. Sect. D Biol. Crystallogr.
D60, 112–115
24 Otwinowski, Z. and Minor, W. (1997) Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol. 276, 307–326
25 Kabsch, W. (1993) Automatic processing of rotation diffraction data from crystals of
initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800
26 McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. and
Read, R. J. (2007) Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674
27 Adams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R., McCoy, A. J., Moriarty,
N. W., Read, R. J., Sacchettini, J. C., Sauter, N. K. and Terwilliger, T. C. (2002) PHENIX:
building new software for automated crystallographic structuredetermination. Acta
Crystallogr. Sect. D Biol. Crystallogr. D58, 1948–1954
28 Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A.,
Winn, M. D., Long, F. and Vagin, A. A. (2011) Acta Crystallogr. Sect. D Biol. Crystallogr.
D67, 355–367
29 Emsley, P., Lohkamp, B., Scott, W. and Cowtan, K. (2010) Features and development of
Coot. Acta Crystallogr. Sect. D Biol. Crystallogr. D66, 486–501
30 Collaborative Computational Project Number 4 (1994) The CCP4 suite: programs for
protein crystallography. Acta Crystallogr. Sect. D Biol. Crystallogr. D50, 760–763
31 Potterton, E., Briggs, P., Turkenburg, M. and Dodson, E. (2003) A graphical user interface
to the CCP4 program suite. Acta Crystallogr. Sect. D Biol. Crystallogr. D59, 1131–1137
32 Lindemose, S., Nielsen, P. E. and Møllegaard, N. E. (2005) Polyamines preferential interact
with bent adenine tracts in double-stranded DNA. Nucleic Acids Res. 33, 1790–1803
33 Lindemose, S., Nielsen, P. E. and Møllegaard, N. E. (2008) Dissecting direct and indirect
readout of cAMP receptor protein DNA binding using an inosine and 2,6-diaminopurine
in vitro selection system. Nucleic Acids Res. 36, 4797–4807
34 Lindemose, S., Nielsen, P. E., Hansen, M. and Møllegaard, N. E. (2011) A DNA minor
groove electronegative potential genome map based on photo-chemical probing. Nucleic
Acids Res. 39, 6269–6276
35 Das, R., Laederach, A., Pearlman, S. M., Herschlag, D. and Altman, R. B. (2005) SAFA:
semi-automated footprinting analysis software for high-throughput quantification of
nucleic acid footprinting experiments. RNA 11, 344–354
36 Nielsen, S. S., Toft, K. N., Snakenborg, D., Jeppesen, M. G., Jakobsen, J. K., Vestergaard,
B., Kutter, J. P. and Arleth, L. (2009) BioXTAS RAW, a software program for
high-throughput automated small-angle X-ray scattering data reduction and preliminary
analysis. J. App. Crystallogr. 42, 959
37 Konarev, P. V., Petoukhov, M. V., Volkov, V. V. and Svergun, D. I. (2006) ATSAS 2.1, a
program package for small-angle scattering data analysis. J. Appl. Crystallogr. 39,
277–286
38 Tran, L. S., Nakashima, K., Sakuma, Y., Simpson, S. D., Fujita, Y., Maruyama, K., Fujita,
M., Seki, M., Shinozaki, K. and Yamaguchi-Shinozaki, K. (2004) Isolation and functional
analysis of Arabidopsis stress-inducible NAC transcription factors that bind to a
drought-responsive cis -element in the early responsive to dehydration stress 1 promoter.
Plant Cell 16, 2481–2498
c The Authors Journal compilation c 2012 Biochemical Society
404
D. H. Welner and others
39 Chen, Q., Wang, Q., Xiong, L. and Lou, Z. (2011) A structural view of the conserved
domain of rice stress-responsive NAC1. Protein Cell 2, 55–63
40 Svergun, D., Barberato, C. and Koch, M. H. J. (1995) CRYSOL: a program to evaluate
X-ray solution scattering of biological macromolecules from Aatomic coordinates. J.
Appl. Crystallogr. 28, 768–773
41 Konarev, P. V., Volkov, V. V., Sokolova, A. V., Koch, H. J. and Svergun, D. I. (2003)
PRIMUS: a Windows PC-based system for small-angle scattering data analysis. J. Appl.
Crystallogr. 36, 1277–1282
42 Yamasaki, K., Kigawa, T., Inoue, M., Watanabe, S., Tateno, M., Seki, M., Shinozaki, K. and
Yokoyama, S. (2008) Structures and evolutionary origins of plant-specific transcription
factor DNA-binding domains. Plant Physiol. Biochem. 46, 394–401
43 Locasale, J. W., Napoli, A. A., Chen, S., Berman, H. M. and Lawson, C. L. (2009)
Signatures of protein–DNA recognition in free DNA binding sites. J. Mol. Biol. 286,
1054–1065
44 Cohen, S. X., Moulin, M., Hashemolhosseini, S., Kilian, K., Wegner, M. and Müller, C. W.
(2003) Structure of the GCM domain–DNA complex: a DNA-binding domain with a novel
fold and mode of target site recognition. EMBO J. 22, 1835–1845
Received 14 October 2011/27 March 2012; accepted 29 March 2012
Published as BJ Immediate Publication 29 March 2012, doi:10.1042/BJ20111742
c The Authors Journal compilation c 2012 Biochemical Society
45 Yoshiyama, K., Conklin, P. A., Huefner, N. D. and Britt, A. B. (2009) Suppressor
of gamma response 1 (SOG1) encodes a putative transcription factor governing
multiple responses to DNA damage. Proc. Natl. Acad. Sci. U.S.A. 106,
12843–12848
46 Ahlstrom, L. S. and Miyashita, O. (2011) Molecular simulation uncovers the
conformational space of the λ Cro dimer in solution. Biophys. J. 101,
2516–2524
47 Diederichs, K. and Karplus, P. A. (1997) Improved R-factors for diffraction data analysis in
macromolecular crystallography. Nat. Struct. Biol. 4, 269–275
48 Lovell, S. C., Davis, I. W., Arendall, III, W. B., de Bakker, P. I. W., Word, J. M.,
Prisant, M. G., Richardson, J. S. and Richardson, D. C. (2003) Structure validation
by Cα geometry: phi, psi and Cβ deviation. Proteins: Struct. Funct. Genet. 50,
437–450
49 Suhre, K. and Sanejouand, Y. H. (2004) ElNemo: a normal mode web-server for protein
movement analysis and the generation of templates for molecular replacement. Nucleic
Acids Res. 32, W610–W614
Biochem. J. (2012) 444, 395–404 (Printed in Great Britain)
doi:10.1042/BJ20111742
SUPPLEMENTARY ONLINE DATA
DNA binding by the plant-specific NAC transcription factors in crystal
and solution: a firm link to WRKY and GCM transcription factors
Ditte H. WELNER*, Søren LINDEMOSE†, J. Günter GROSSMANN‡, Niels Erik MØLLEGAARD§, Addie N. OLSEN†,
Charlotte HELGSTRAND*, Karen SKRIVER† and Leila LO LEGGIO*1
*Biophysical Chemistry Group, Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen Ø, Denmark, †Section for Biomolecular Sciences,
Department of Biology, University of Copenhagen, Ole Maaløes Vej 5, 2200 Copenhagen N, Denmark, ‡Molecular Biophysics Group, Institute of Integrative Biology, Faculty of Health
and Life Sciences, The University of Liverpool, Crown Street, Liverpool L69 7ZB, U.K., and §Department of Cellular and Molecular Medicine, Panum Institute, Faculty of Health
Sciences, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen N, Denmark
METHODS
Table S1
Structure determination of a new crystal form of the ANAC019 NAC
domain
The reported statistics are from Phaser [3]. A final Z-score above 8 indicates that Phaser has
definitely solved the structure according to the PhaserWiki (http://www.phaser.cimr.cam.ac.uk).
Furthermore, LLG should increase for each added component, which is also the case here. N/A,
not applicable.
The structure was determined by MR, which gave solutions with
two monomers in the asymmetric unit with CNS [1], MOLREP
[2] or Phaser [3] within the PHENIX suite [4] [Z and LLG (log
likelihood gain) scores after placing of the first monomer were
13.7 and 331 respectively, Z and LLG scores after placing the
second monomer were 19.6 and 622 respectively]. Despite high
initial R-factors (around 50 %), poor density for the B molecule
especially and difficulties in refinement, the MR solution was
deemed correct, as it showed a similar region forming the dimer
interface to the one previously observed in the high resolution
crystal structure, although the respective orientation of the two
monomers is not identical.
After testing several refinement protocols in various programs,
the structure was initially refined with Phenix.refine [4] alternated
with manual rebuilding in Coot [5]. The final refinement strategy
included positional refinement of only few rebuilt residues at each
round, and TLS (Translation/Libration/Screw) group refinement
for chains A and B separately. However, using this strategy the
structure could only be refined to an R-factor of 32.3 % and Rfree of 37.0 % with RMSD (root mean square density) in bond
lengths of 0.021 Å and RMSD in bond angles of 3.215◦ and
poor Ramachandran geometry. The model was at this point rather
incomplete, with only 210 built residues out of 336 amino acid
residues in total. The B molecule was particularly incomplete with
only 83 out of 168 amino acids built because of the corresponding
density being of very poor quality. The poor quality density
could be partly accounted for by fewer crystal contacts for the
B molecule than the A molecule.
Subsequently, the so-called jelly-body refinement in Refmac5
[6] was tried as an alternative refinement strategy at low resolution.
This proved more successful. Jelly-body restraints were combined
with automatic detection of local NCS (non-crystallographic
symmetry) and group TLS refinement with one group per protein
chain. The refinement proved much more robust than previously
observed and also led to improvement of the electron density,
so that 40 additional residues could be built. The final model
is about 75 % complete and contains 250 out of 336 total
residues, including 111 out 168 amino acids for monomer B. For
comparison, 297 residues were built for the deposited free NAC
structure at 1.9 Å resolution (PDB code 1UT7), 147 residues for
the A and 150 for the B chain. The R-factor and R-free were
Molecular replacement statistics
Search model
Step
Z-score
Log-likelihood gain
Dimer 1
Rotation
Translation
Rigid Body refinement
Rotation
Translation
Rigid Body refinement
Rotation
Translation
Rigid Body refinement
4.8
6.1
N/A
4.4
3.3
N/A
5.8
8.8
N/A
46.7
42.4
64.3
105.2
137.7
265.4
339.6
271.4
333.2
Dimer 2
DNA
improved to 25.7 % and 34.2 % respectively. The model also has
improved geometry, with RMSD in bond length of 0.012 Å and
RMSD in bond angles of 1.820◦ and 90.4 % residues in favoured
regions of the Ramachandran according to Rampage [7]. The
model statistics are shown in Table 1 of the main text.
Crystal structure determination of NAC–DNA complexes
The structure of Native_complex was solved with MR. A trimmed
version (N-terminal removed) of the ANAC019 structure (PDB
code 1UT4) and a B-DNA model of oligo2 (3DNA software [8])
were used as consecutive search models. The MR solution is a
ternary complex consisting of two ANAC019 dimers (AB and
CD) bound to one DNA double helix constituted by chains EF.
The MR statistics can be found in Table S1.
Attempts were made to supplement the MR data with
experimental phase information from the heavy-atom derivatives.
This approach was hindered by poor diffracting quality and lack of
isomorphism of derivatives (e.g. see the cell dimensions in Table 1
of the main text), but finally a gold-derivative data set of similar
quality as the native was collected with the intention of using it in
MAD (multi-wavelength anomalous diffraction) or SAD (singlewavelength anomalous diffraction). The resulting experimental
maps were of very low quality not suited for building. An MR
solution could be obtained using a similar procedure and giving
similar results as for the native. This solution was combined
with the weak SAD information using MRSAD (molecular
Co-ordinates and structure factors have been deposited into PDB with accession codes 3SWP (Native_complex), 3SWM (Au_complex) and 4DUL
(Form IV).
1
To whom correspondence should be addressed (email [email protected]).
c The Authors Journal compilation c 2012 Biochemical Society
D. H. Welner and others
Figure S1
Validation of MR solution
(A) Crystal packing and DNA fibre formation. The overall packing of the crystals looks good, with
the DNA polymerizing as B-DNA fibers throughout the crystal. (B) Omit map at 3σ generated
by removing 5 nucleotides (grey) from the model. This does not allow for an identification of
the individual bases, but does confirm the position of the DNA. (C) Gold sites in structures
of ANAC019 and ANAC019 in complex with DNA. Chain A of NAC alone (PDB code 1UT7, blue)
and of the MR solution (grey) is superimposed. Due to the higher resolution of the NAC structure
(1.9 Å), the site can be modelled in more detail than in the complex structure. Consequently,
it is modelled as two sites with low occupancy (yellow spheres) and the co-ordinating Cys33
(sticks) is modelled in double conformation. The gold site found in the complex structure by
MRSAD is shown as an orange sphere.
replacement SAD) procedure number 1 in phenix.autosol v1.6289 [4], which uses the MR solution as a partial model and
writes out the best combined phases. It identified nine gold
sites, but phase combination did not improve the appearance of
electron density maps. The derivative was however fundamental
in confirming the MR solution and was refined in parallel, using
the native MR solution combined with the nine gold sites as a
starting model.
The MR model showed sensible packing, with the DNA
forming continuous fibres throughout the crystal (Figure S1A) and
interacting with the expected face of the NAC dimer, i.e. opposite
the dimer interface. Omit maps constructed from models lacking
a number of phosphates in the DNA backbone and/or a number of
nucleotides also confirmed the position of the DNA duplex (Figure
S1B). Gold binding in Au_complex is observed at the same sites
as in the gold derivative used in structure determination of the
protein alone [9] (Figure S1C) and difference density shows up at
these sites when the gold-derivative dataset is phased with a MR
model not containing gold atoms. The oligonucleotide sequence,
which could not be assigned based on the crystal structure alone,
was assigned to agree with the phosphate protection information
from uranyl photo-probing as described in the main text. Despite
the possibility of a register error at this resolution, inclusion of the
bases is important for the crystallographic refinement and SAXS
simulations.
Several refinement strategies were tested. Use of experimental
phase restraints and simulated annealing strategies did not
improve refinement. The final refinement strategy chosen was
as follows: (i) MR was followed by rigid body refinement with
each NAC monomer as a rigid body and the dsON (both strands)
as an additional rigid body. For the Native_complex the R- and Rfree values went from 52.0 % and 52.4 % to 46.1 % and 47.9 %
respectively, whereas for Au_complex they went from 51.4 %
and 51.7 % to 44.4 % and 43.0 % respectively; (ii) positional
c The Authors Journal compilation c 2012 Biochemical Society
Figure S2
DNase I footprinting of the ANAC019–DNA complex
Typical autoradiograph showing the DNase I digestion of an oligonucleotide containing the
palindromic NACBS [13] in the absence (control) and presence of 10, 20, 50, 100 and
200 ng of ANAC019 (lanes 1–6). The orientation of the DNA is indicated with 5 and 3 . S
indicates the Maxam–Gilbert A/G markers. The palindromic NACBS contains the 20 bp sequence
5 -ttgcgtgttggaacacgcaa-3 . However, due to the 3 -labelling (see Experimental section of main
text), the footprint presented displays the interaction on the lower strand in the 3 to 5 direction
(i.e. the complement strand). Brackets indicate the boundary of the palindromic NACBS and
the DNase I protected area respectively. Due to the physical size of the DNase I protein itself
(32 kDa), the actual ANAC019 footprint is overestimated by 5–10 bp in both the 5 - and
3 -end.
refinement was carried out with NCS restraints between AC
and BD NAC monomers, including TLS refinement with each
protein chain as a group and the dsON as an additional group.
In Au_complex, individual isotropic B-factors were refined for
the heavy atoms. In order to maintain reasonable Ramachandran
statistics, the highest resolution ANAC019 structure available
(PDB code 1UT7) was used as an additional restraint. For the
Native_complex the R- and R-free values went down to 25.9 %
and 34.9 % respectively, whereas for Au_complex they went
down to 25.3 % and 31.9 % respectively; and (iii) finally, the Nterminal residues 8–11 of chain A, which were missing in the MR
model, could be built. For the Native_complex the R- and R-free
values went down to 26.0 % and 34.9 % respectively, whereas for
Au_complex they went down to 25.0 % and 30.9 % respectively.
SAXS analysis on the free NAC domain
Solution scattering profiles for a concentration series from 1
to 8 mg/ml of the domain were measured in 20 mM Tris/HCl
(pH 7.5), at two camera lengths of 1 m and 4.25 m at
beamline 2.1, SRS (Daresbury, England). Concentration- and
time-dependent X-ray-induced aggregation affects the low-angle
DNA binding by NAC plant transcription factors
Figure S3
DNA-binding ability of mutant NAC domains
Binding of wild-type ANAC019 NAC domain to a 30 bp olugonucleotide containing 2 NACBS
[13] (as detailed in Experimental section of the main text) was compared with that of the R85A
and R88A mutants in an EMSA. The concentrations of protein used are specified above each
lane.
scattering region. The Rg could nevertheless be extrapolated
to 27.3 Å at infinite dilution. The data were processed with
XOTOKO (available from the SRS) and further analysed in the
ATSAS program package [10].
Ab initio modelling was carried out with GASBOR version 2.2i
(reciprocal space) with unknown expected particle anisometry,
with either no symmetry imposed or 2-fold symmetry imposed.
The models were then averaged with DAMAVER and filtered
using DAMFILT. Ab initio modelling showed a very similar shape
compared with the known X-ray structure (Figure 3 of the main
text). Imposing 2-fold symmetry in the ab initio modelling did
not result in a significant increase in the χ-value (average χvalue = 1.06 for nine models in P2, average χ-value = 1.03 for
seven models in P1). We conclude that within the resolution limits
of SAXS, the NAC dimer in solution is symmetrical.
The NAC dimer scattering profile is consistent with the X-ray
crystallographic structures previously determined for free NAC,
as shown from the CRYSOL fit in Figure 3(A) of the main text
and Table S2. However, a discrepancy is found at very low angles,
which is also reflected in the difference between experimental and
calculated Rg -value. A value much closer to the experimental Rg value could be obtained if one of the crystallographic models
for free NAC was completed by adding the loops missing in the
crystallographic model (from a database of loop structures found
in O [11] and by adding the missing parts of the termini in an
extended conformation; see Table S2); however, this resulted in
a worse χ-value. A further improvement could be obtained by
perturbing the termini with the elNémo server [12], where some
of the resulting models fit better both with respect to experimental
Rg - and χ-values (see example of fit in Figure 3A of the main text
and Table S2). Subsequently the SAXS data for the free NAC
domain were used to test complete models with different degrees
of dimer opening, as described in the main text.
Table S2 Experimental and theoretical R g -values derived from SAXS data and selected structural models derived from crystallographically determined
structures
For comparison, the theoretical R g -value calculated from the ternary complex in the crystal structure was 32.3 Å. For the free NAC sample, the experimental R g was calculated from extrapolation to
infinite dilution from a concentration series. χ-Values for the fit of theoretical scattering profiles calculated in CRYSOL [14] for a particular model and experimental data are also given.
Experimental sample
Experimental R g -value
from Guinier analysis (Å)
Fraction B1
Fraction B2
24.6 +
− 3.9
27.9 +
− 1.2
Fraction B3
Free NAC
29.5 +
− 1.8
27.3
Structural model
Theoretical R g -value
calculated from model (Å)
χ-Value
DNA (from Native_complex)
DNA (from Native_complex)
OLIGOMER model (82 % DNA and 18 % binary complex from Native_complex)
Binary complex (DNA and AB dimer from Native_complex)
Free NAC (PDB code 1UT7)
Full model including loops and termini missing in the crystal structure
Example of best perturbed full model, dimer angle 118◦ (open dimer)
As above, but dimer angle of 107◦ (semi-closed dimer)
As above, but dimer angle of 100◦ (closed dimer)
26.6
26.6
–
31.2
26.2
26.9
27.4
26.9
26.5
1.3
1.7
1.2
1.1
1.9
2.2
1.9
2.5
3.1
c The Authors Journal compilation c 2012 Biochemical Society
D. H. Welner and others
REFERENCES
1 Brünger, A. T., Adams, P. D., Clore, GM., DeLano, W. L., Gros, P., Grosse-Kunstleve,
R. W., Jiang, J.-S., Kuszewski, J., Nilges, N., Pannu, N. S. et al. (1998) Crystallography
and NMR system. Acta Crystallogr. Sect. D Biol. Crystallogr. D54, 905–921
2 Vagin, A. and Teplyakov, A. (1997) MOLREP: an automated program for molecular
replacement. J. Appl. Crystallogr. 30, 1022–1025
3 McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. and
Read, R. J. (2007) Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674
4 Adams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R., McCoy, A. J., Moriarty,
N. W., Read, R. J., Sacchettini, J. C., Sauter, N. K. and Terwilliger, T. C. (2002) PHENIX:
building new software for automated crystallographic structure determination. Acta
Crystallogr. Sect. D Biol. Crystallogr. D58, 1948–1954
5 Emsley, P., Lohkamp, B., Scott, W. and Cowtan, K. (2010) Features and development of
Coot. Acta Crystallogr. Sect. D Biol. Crystallogr. D66, 486–501
6 Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A.,
Winn, M. D., Long, F. and Vagin, A. A. (2011) REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr. Sect. D Biol. Crystallogr. D67, 355–367
7 Lovell, S. C., Davis, I. W., Arendall, III, W. B., de Bakker, P. I. W., Word, J. M., Prisant,
M. G., Richardson, J. S. and Richardson, D. C. (2003) Structure validation by Cα
geometry: phi, psi and Cβ deviation. Proteins: Struct. Funct. Genet. 50, 437–450
Received 14 October 2011/27 March 2012; accepted 29 March 2012
Published as BJ Immediate Publication 29 March 2012, doi:10.1042/BJ20111742
c The Authors Journal compilation c 2012 Biochemical Society
8 Lu, X.-L. and Olson, W. K. (2003) 3DNA: a software package for the analysis, rebuilding
and visualization of three-dimensional nucleic acid structures. Nucleic Acids Res. 31,
5108–5121
9 Ernst, H. A., Olsen, A. N., Skriver, K., Larsen, S. and Lo Leggio, L. (2004) Structure of the
conserved domain of ANAC, a member of the NAC family of transcription factors. EMBO
Rep. 5, 297–303
10 Konarev, P. V., Petoukhov, M. V., Volkov, V. V. and Svergun, D. I. (2006) ATSAS 2.1, a
program package for small-angle scattering data analysis. J. Appl. Crystallogr. 39,
277–286
11 Jones, T. A., Zou, J. Y., Cowan, S. W. and Kjeldgaard, M. (1991) Improved methods for
building protein models in electron density maps and the location of errors in these
models. Acta Crystallogr. Sect. A Found. Crystallogr. A47, 110–119
12 Suhre, K. and Sanejouand, Y. H. (2004) ElNemo: a normal mode web-server for protein
movement analysis and the generation of templates for molecular replacement. Nucleic
Acids Res. 32, W610–W614
13 Olsen, A. N., Ernst, H. A., Lo Leggio, L. and Skriver, K. (2005) DNA-binding specificity
and molecular functions of NAC transcription factors. Plant Sci. 169,
785–797
14 Svergun, D., Barberato, C. and Koch, M. H. J. (1995) CRYSOL: a program to evaluate
X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl.
Crystallogr. 28, 768–773