Structure 14, 983–992, June 2006 ª2006 Elsevier Ltd All rights reservedDOI 10.1016/j.str.2006.04.001
The Rad51/RadA N-Terminal Domain Activates
Nucleoprotein Filament ATPase Activity
Vitold E. Galkin,1Yan Wu,2Xiao-Ping Zhang,3
Xinguo Qian,2Yujiong He,2Xiong Yu,1
Wolf-Dietrich Heyer,3Yu Luo,2
and Edward H. Egelman1,*
1Department of Biochemistry and Molecular Genetics
University of Virginia
Charlottesville, Virginia 22908
2Department of Biochemistry
University of Saskatchewan
A3 Health Sciences Building
107 Wiggins Road
Canada S7N 5E5
3Section of Microbiology
Section of Molecular and Cellular Biology
Center for Genetics and Development
University of California, Davis
Davis, California 95616
Proteins in the RecA/RadA/Rad51 family form helical
filaments on DNA that function in homologous recom-
bination. While these proteins all have the same highly
have an N-terminal domain that shows no homology
with the C-terminal domain found in RecA. Both the
Rad51 N-terminal and RecA C-terminal domains have
has been established. We show that RadA filaments
tion with respect to the ATPase and that activation
involves a large rotation of the subunit aided by the
N-terminal domain. The G103E mutation within the
yeast Rad51 N-terminal domain inactivates the fila-
ment by failing to make proper contacts between the
N-terminal domain and the core. These results show
that the N-terminal domains play a regulatory role in
filament activation and highlight the modular architec-
ture of the recombination proteins.
ago (Clark and Margulies, 1965) and has served as
a model system for understanding homologous genetic
recombination. RecA is involved in recombination, DNA
repair, activation of the SOS response (Courcelle et al.,
2001), cleavage of phage repressors, and mutagenic
translesion synthesis (Schlacher et al., 2005), among
other activities. A great deal has been learned about
how RecA functions from in vitro DNA strand-exchange
reactions, where this protein can catalyze the transfer
of a DNA strand between two homologous molecules
(Kowalczykowski and Eggleston, 1994). The active form
of RecA in this reaction is a helical filament formed on
DNA. Interest in bacterial RecA was enhanced by the
discovery that eukaryotic cells encode a homologous
protein, Rad51 (Shinohara et al., 1992; Aboussekhra
et al., 1992), that forms similar helical nucleoprotein
filaments on DNA as those formed by RecA (Ogawa
et al., 1993). A second eukaryotic RecA-like protein, the
meiosis-specific Dmc1, also forms such nucleoprotein
filaments (Sehorn et al., 2004). With the discovery that
archaea encode a RadA protein (Sandler et al., 1996)
and that RadA-DNA filaments are also similar to RecA-,
Rad51-, and Dmc1-DNA filaments (Yang et al., 2001), it
is apparent that all of the major kingdoms of life encode
a recombination system that is based upon a similar
A detailed mechanistic picture of how RecA-like fila-
ments function in recombination and other reactions is
still lacking despite much effort invested in this area.
The first crystal structure of RecA (Story et al., 1992) pro-
vided a high-resolution picture of a RecA filament, but in
the absence of DNA. This crystal structure is now under-
stood to be a compressed, inactive state (VanLoock
et al., 2003). More recent crystal structures of archaeal
RadA (Wu et al., 2004) and yeast Rad51 (Conway et al.,
of RecA (Story et al., 1992; Xing and Bell, 2004a, 2004b;
Rajan and Bell, 2004; Datta et al., 2000, 2003a, 2003b),
RadA (Shin et al., 2003; Wu et al., 2004), Rad51 (Conway
among different proteins from different species. In fact,
this same nucleotide binding core was also seen in the
F1-ATPase, and the rms deviation for the superposition
of 120 C-a residues within this core between bacterial
RecA and the bovine F1-ATPase is less than 2 A˚(Abra-
hams et al., 1994). Despite the high degree of structural
conservation of the cores, the modular architecture of
the RecA protein (and filament) is different from that of
RadA, Rad51, and Dmc1. RecA has a C-terminal domain
that has no homology, either at the level of sequence or
structure, with the conserved N-terminal domains con-
tained within RadA, Rad51, and Dmc1. Nevertheless,
there are a number of parallels between these two differ-
entdomains. Bothhave beenshown tobind DNA (Aihara
et al., 1997, 1999). Approximately 25 C-terminal residues
ably due to disorder, while approximately 15 N-terminal
residues in human Rad51 (Aihara et al., 1999) and 79 res-
way et al., 2004) have been suggested to be disordered.
Crystal structures and electron microscopy have shown
that the relation between both the RadA/Rad51/Dmc1
minal domain, on the other, and the relatively invariant
nucleotide binding core can be quite variable (Rajan
and Bell, 2004; Shin et al., 2003; Yang et al., 2001; Van-
Loock et al., 2003; Kinebuchi et al., 2004).
We have used archaeal RadA and yeast Rad51 to look
at structure/function relations involving the N-terminal
domain. We have done this by creating a RadA protein
missing 62 N-terminal residues (RadA-D62), examining
the structure of this protein by X-ray crystallography
and electron microscopy, and relating this to the bio-
chemical activity of the fragment. We have also used
electron microscopy to examine the structure of nucleo-
tant protein. Residue 103 is in the N-terminal domain
of yeast Rad51 (which corresponds to the highly con-
served G23 in MvRadA), and a previous study showed
that the G103E mutant inactivated the protein with re-
spect to both ATPase and DNA strand exchange activity
(Zhang et al., 2005). The studies of both RadA-D62
and RAD51-G103E lead to a common conclusion: the
N-terminal domain in these proteins plays a crucial role
in coordinating the ATPase activity of the catalytic core,
which involves rotations of the core within the filaments.
Crystal Structure of RadA-D62
The RadA fragment lacking the N-terminal domain,
RadA-D62, crystallized in the P61space group, forming
a filament in the crystal with six subunits per turn of a 91
A˚pitch right-handed helix. This helical crystallographic
arrangement is similar to that seen for the full-length
protein, except that the helical pitch of the full-length
protein was w107 A˚(Wu et al., 2004). The RadA-D62
structuresuperimposed quitewellon thecorresponding
core of the full-length RadA protein (Figure 1A), except
for a shift by several Angstroms in the polymerization
motif (Pellegrini et al., 2002) containing Phe64 and the
adjacent hinge region. This shift reflects the change in
helical pitch between the two structures and shows
that deformation of this hinge region is involved in
the variable pitch observed for all RecA/RadA/Rad51
The ATP binding pocket is formed by the interface
between two protomers in the filament, as previously
shown for the full-length RadA (Wu et al., 2004) and
yeast Rad51 (Conway et al., 2004). A nonhydrolyzable
AMP-PNP molecule can be seen bound within this
pocket (Figure 1B). The structure thus shows that a heli-
cal filament of RadA can be formed, at least in a crystal
that properly binds ATP, even in the absence of the
In Vitro Activities of RadA-D62
An in vitro assay revealed that RadA-D62 had a DNA-ac-
tivated ATPase activity that was very similar to that of
activity within RecA-like proteins requires filament for-
mation, a mechanistic consequence of the ATP binding
and hydrolysis site being located between two subunits
in the active form of the helix (Figure 1B), this provides
a prima facie suggestion that RadA-D62 must be form-
ing filaments on DNA. This has been confirmed by elec-
tron microscopy (see below). In contrast to the ATPase
activity that is very similar to the full-length protein,
RadA-D62 did not exhibit any DNA strand-exchange
activity (Figure 2C). This suggests that the N-terminal
domain is required for strand exchange but not for fila-
Structure of RadA-D62 Filaments
Electron microscopy confirms that the RadA-D62 pro-
tein forms filaments on DNA in the presence of ATP
(Figure 3C). In the absence of either DNA or ATP, such
filaments are not observed (Table 1). While helical stria-
tions can be seen in the extended filaments (with a pitch
of w100 A˚) formed by the full-length protein in the pres-
ence of ADP-AlF (Figure 3A) or the more compressed fil-
aments (with a pitch of w70 A˚) formed by the full-length
protein in the presence of AMP-PNP (Figure 3B), most
of the RadA-D62 filaments appear rather featureless
(Figure 3C). In a crystal, in the absence of DNA, the
full-length protein forms a rather extended filament
(107 A˚pitch) with AMP-PNP that is in an active confor-
mation (Wu et al., 2004), while the filaments observed
by EM of the full-length protein on DNA in the presence
of AMP-PNP are much more compressed.
Figure 1. Crystal Structure of the RadA-D62 Fragment
the polymerization motif around Phe-64 and a short helical hinge.
(B) The ATPase site of the RadA-D62 filament resembles that of the previously determined full-length RadA structure. The AMP-PNP (in cyan) is
found between two adjacent protomers. The yellow protomer contributes the triphosphate-wrapping P loop, base-stacking Arg-158, and
the catalytic Glu-151 and Gln-257. The gray protomer contacts the ATP analog through the ATP cap (residues Asp-302 to Asp-308) and the
C-terminal elbow of the L2 region (residues His-280 to Arg-285). The magnesium ion is shown in red.
The iterative helical real space reconstruction (IHRSR)
method (Egelman, 2000) was used to analyze both the
RadA-D62 filaments and the control filaments formed
bythe full-length protein. Imageanalysis revealed aheli-
cal periodicity of w50 A˚for the relatively featureless
RadA-D62 filaments. However, this periodicity was de-
termined to arise from a two-start helix having a pitch
of w100 A˚and not the one-start helix that is dominant
in all other RecA-like filaments. Three-dimensional re-
constructions can be compared for the RadA-D62 two-
start filaments (Figure 4D), the one-start helix formed
by the full-length protein in the presence of ADP-AlF
with a pitch of 99 A˚(Figure 4B), or in the presence of
AMP-PNP with a pitch of 71 A˚(Figure 4C) and a low-res-
of the full-length protein in the presence of AMP-PNP
with a helical pitch of w107 A˚(Figure 4A). In both the
full-length protein filaments (Figures 4B and 4C), there
is less density seen for the N-terminal domain than in
the crystal filament. This can be explained by partial dis-
order of this domain since we have previously shown
that in some states of a Sulfolobus solfataricus RadA fil-
ament the N-terminal domain was not seen at all after
averaging and three-dimensional reconstruction due to
a more extreme disorder (Yang et al., 2001).
(Figure 4B) appears to be shifted with respect to the po-
sition of this domain in the compressed filament
(Figure 4C). This N-terminal domain also makes con-
tacts with the ATP binding core of an adjacent subunit
in the extended state (red arrow, Figure 4B), but such
Fitting the crystal structure of the subunit into both
reconstructions yields a simple conclusion: there is a
rotation of the subunit by w30º between these two
states (Figure 4E). This is similar to the rotation that has
of a compressed RecA filament (Story et al., 1992) and
fits of atomic models to EM reconstructions of the
extended RecA-DNA filaments (VanLoock et al., 2003).
The two-start RadA-D62 filaments have approxi-
mately eight subunits per turn on each 100 A˚pitch heli-
cal strand, with a stagger by half a subunit between the
two strands. Thus, the two strands are not related by
a 2-fold symmetry. Approximately 40% of the segments
analyzed from the two-start filaments were used for this
in the extendedstate
Figure 2. ATPase and DNA Strand Exchange Activities
(A) ATPase activity of full-length MvRadA.
(B) ATPase activity of RadA-D62. There were no significant differences between the two proteins’ DNA-dependent ATPase activities.
(C) The DNA strand-exchange reaction leads to the accumulation of a heteroduplex band (hdDNA) on the ethidium bromide-stained gel. The full-
length RadA is active in the presence of all tested nucleoside triphosphates.
(D) The truncated RadA-D62 is inactive.
Figure 3. Electron Micrographs of RadA Filaments
Negatively stained RadAfilaments formed bythe full-length protein (RadA-S2G) (A and B) and filaments formedby the truncated RadA-D62frag-
ment (C). A mixture of filaments ([A], black arrow) and rings ([A], white arrow) were found in the presence of ADP-AlF, while when AMP-PNP was
the deep helical groove, if one compares them with either RadA-S2G-ADP-AlF extended filaments (A) or RadA-AMP-PNP compressed ones (B).
These filaments (A) are shown to be two extended filaments that are coiled about each other. The scale bar is 500 A˚.
The Rad51/RadA N-Terminal Domain
reconstruction. The remaining segments appeared to
suffer from disorder, which was consistent with a slip-
page of one of the strands with respect to the other. In
these disordered segments, the relationship between
the two strands did not appear to be fixed. Fitting the
crystal structure of RadA-D62 into the reconstruction
of the two-start helix shows that the subunit orientation
is more similar to that of the extended state than it is
to the compressed one (Figure 4F). The relationship
between residues implicated in ATP hydrolysis on two
adjacent subunits is shown in Figure 5 for the different
These results show that RadA-D62 can form filaments
on DNA, in a novel two-start configuration, that are in an
active conformation for hydrolyzing ATP. We first ob-
served these filaments in the presence of dsDNA and
wondered whether the two strands might arise from
the two strands of DNA. We found that these filaments
formed equally well on ssDNA (which were the ones
analyzed), so that there is no necessary dependence
upon dsDNA being present. The apparent slippage of
the two strands with respect to each other provides
a mechanism for how subunits might rotate in this
structure during the ATPase cycle. We have also shown,
by using the modeling, why the compressed filaments
are in an inactive state due to the ATP binding site being
rotated out of the pocket between adjacent subunits.
The Yeast Rad51-G103E Mutation
We have used electron microscopy and three-dimen-
sional reconstruction to look at the structural basis for
the inactivation of both ATPase and DNA strand-ex-
change activity by the G103E mutation in yeast Rad51
(Zhang et al., 2005). A striking difference with the wild-
type protein is that in the presence of either the slowly
hydrolysable ATP analog ATP-g-S (Figure 6A) or the
nonhydrolysable analog AMP-PNP (Figure 6B), the
wild-type Rad51 forms filaments on DNA, while the
G103E mutation fails to form filaments on either dsDNA
or ssDNA with these ATP analogs (Table 1). In the pres-
ence of ATP and aluminum fluoride, the G103E mutant
readily forms filaments (Figure 6C) that appear very sim-
ilar to the filaments formed by the wild-type protein
under the same conditions (Figure 6D).
Three-dimensional reconstructions of both the wild-
type protein filament (Figure 7B) and the G103E mutant
Table 1. Complexes Examined by EM
ATP Cofactor/ComplexAMP-PNPATP-g-SATP-AlF ADP-AlF
Figure 4. Three-Dimensional Reconstructions of RadA-S2G Filaments
Filaments formed in the presence of ADP-AlF (B) and AMP-PNP (C) show up as one-start helices, while RadA-D62 produces a two-start helix in
thepresenceofATP-AlF(D).Filamentin(B)has apitchofw99A˚,whilefilamentin (C)is inthe compressed formhaving apitch ofw71A˚.A crystal
structure of the ‘‘core’’ part (residues 63–322) of MvRadA was used to explore the orientation of protomers in these reconstructions, and this
tal’’ filament (A) and RadA-S2G-ADP-AlF filament (B), except N terminus in (B) was not fully visualized due to partial disorder. Protomers within
are shown in (E) correspond to those in the centers of (B) and (C), and the rotation is thus about an axis that is perpendicular to the filament axis.
The orientation of the protomers in the two-start helix produced by RadA-D62 ([F], yellow ribbon) was close to the orientation found in Rad-S2G-
ADP-AlF complex ([F], red ribbon).
filament (Figure 7C) show that the N-terminal domain is
seen in the wild-type filament but missing in the G103E
mutant. We have used a second preparation of both
the G103E mutant and the wild-type protein to generate
new EM images and three-dimensional reconstructions
to independently confirm these observations. The result
with the wild-type protein was surprising because the
first three-dimensional reconstruction failed to visualize
an N-terminal domain (Ogawa et al., 1993; Yu et al.,
2001), presumably due to large disorder, while this
the N-terminal domain in the original studies of yeast
Rad51 (Ogawa et al., 1993) was due to the inactivity of
those formed by human Rad51. Also the absence of the
N terminus in the reconstruction of the Rad51-G103E
mutant was surprising because it had been hypothe-
sized that E103 makes a new contact with a positively
Figure 5. RadA Residues Involved in ATP Hydrolysis
The positions of these residues (Wu et al., 2004) in extended RadA-ADP-AlF filaments (A), compressed RadA-AMP-PNP filaments (B), and the
two-stranded RadA-D62-ATP-AlF filaments (C). Residues Arg-158 (cyan) and Pro-307 (blue), which tether the base, and Lys-111 (cyan) and
Glu-151 (cyan), needed for catalysis, are shown as ball-and-stick models. The distances between the Ca atoms of Pro-307 and Arg-158 are:
RadA-ADP-AlF, 9.4 A˚; RadA-AMP-PNP, 19.0 A˚; RadA-D62-ATP-AlF, 12.2 A˚.
Figure 6. Electron Micrographs of Yeast
The wild-type protein forms filaments on
dsDNA in the presence of the ATP analogs
AMP-PNP (A) or ATP-g-S (B), while the
G103E mutant failed to polymerize under
these conditions. In the presence of ATP
and aluminum fluoride, both the wild-type
protein (C) and the G103E mutant (D) formed
long filaments on dsDNA. The scale bar is
The Rad51/RadA N-Terminal Domain
charged patch on the helical domain of the ATPase core
formed by R260, H302, and K305 that could lock the N
terminus to the core (Zhang et al., 2005). It is unclear
whether the N terminus in Rad51-G103E is disorderd or
present in multiple states that preclude its visualization.
can be expected to be weakened in the G103E mutant
We have fit the crystal structure of yeast Rad51 (Con-
wayetal.,2004)intoour EMreconstructions. Thecrystal
of yeast Rad51 contained a helical filament with a pitch
of 130 A˚, but it can be seen that the crystal filament
(Figure 7A) matches quite well the reconstruction of the
extended Rad51-DNA filament that has a pitch of 99 A˚
(Figure 7B). The filament formed by RAD51-G103E has
an ATPase core that appears to be in a similar orienta-
tion to that in the active, extended wild-type Rad51 fila-
ment. The question is then why does the G103E mutant
fail to hydrolyze ATP? A likely explanation is that each
subunit in this filament might hydrolyze ATP, but turn-
over is prevented by the absence of a fixed N-terminal
have shown (above) that the N-terminal domain makes
a contact with the ATPase core of an adjacent subunit
in the extended state, which we suggest plays an impor-
tant role in the ATPase cycle. Consistent with this
prediction, an initial burst of ATPase activity is seen
with the G103E mutant, followed by a decline in activity,
while the wild-type protein shows steady state ATPase
activity (Figure 8).
We have used archaeal RadA and yeast Rad51 to exam-
ine the role of the N-terminal domain in filament forma-
tion, ATP hydrolysis, and DNA strand exchange. Dele-
tion of the N-terminal domain in RadA, and the G103E
mutation in RAD51’s N-terminal domain, inactivate the
proteins for DNA strand exchange. We can show that
ate conditions, but these conditions are more restrictive
for G103E than for the wild-type Rad51. In the presence
of the ATP analogs ATP-g-S or AMP-PNP, the wild-type
Rad51 forms filaments on DNA, while the G103E mutant
fails to polymerize. Since we know that residue 103 is far
from the site of nucleotide binding in the filament (Con-
way et al., 2004), there cannot be a direct interaction be-
tween the nucleotide and the glutamic acid that has
been substituted at residue 103. An interaction occurs
between the N-terminal domain of one subunit and the
nucleotide binding core of an adjacent subunit, which
we have directly visualized in both the wild-type yeast
Rad51 filament reconstruction (Figure 6B) and for the
wild-type RadA filament (Figure 4B). These interactions
between the N-terminal domain of one subunit and the
ATP binding core of an adjacent subunit have also
been visualized in high-resolution crystal structures of
the extended filaments formed by both RadA (Wu
et al., 2004) and Rad51 (Conway et al., 2004). We have
shown that the G103E mutation leads to disorder in
the N-terminal domain, and we suggest that for yeast
Figure 7. The Yeast Rad51 Filament
Three-dimensional reconstructions of yeast Rad51-DNA filaments (B and C) and a crystal structure (Conway et al., 2004) of the protein (A).
The wild-type Rad51-DNA filament (B) is quite similar to the crystal filament (A), while the G103E mutant forms a filament on DNA in which
the N-terminal domain is not visualized after averaging (C).
Figure 8. ATPase Activity of Wild-Type and G103E Mutant Yeast
The initial activityseen with the G103E mutant, followed bya plateau
showing no further activity, is consistent with a single hydrolytic
event with no subsequent turnover.
and the core of an adjacent subunit is needed for fila-
ment formation in the presence of the analogs ATP-g-S
or AMP-PNP. For archaeal RadA, we have found that
in the absence of the N-terminal domain (RadA-D62)
compressed filaments can be formed on DNA in the
presence of AMP-PNP.
In the presence of ATP, the Rad51-G103E protein can
form filaments on DNA, as can the RadA-D62 fragment
that is missing an N-terminal domain. We suggest, how-
ever, that normal hydrolysis of ATP by this filament
requires a proper coordination between the N-terminal
domain of one subunit and the ATP binding core of an
adjacent subunit. Why, then, is the filament formed by
RadA-D62, completely lacking an N-terminal domain,
active in ATP hydrolysis, while the RAD51-G103E fila-
ment is not? We have shown that the predominant fila-
ment formed on DNA by RadA-D62 in the presence of
ATP is not the normal one-start helix with wsix subunits
per turn but rather a two-start helix, with each strand
having weight subunits per turn. Within this two-
stranded filament, each nucleotide binding core makes
contact with four adjacent subunits, in contrast to
in the normal one-start helix. The ability of these strands
to slip with respect to each other is probably crucial for
the turnover in the ATPase cycle, as subunits rotate be-
tween the ‘‘inactive’’ and ‘‘active’’ orientations. We sug-
gest that the new contacts that appear in the two-start
structure can allow for the proper coordination and hy-
drolysis of ATP in the absence of an N-terminal domain.
Not surprisingly the two-start helical filament is inactive
in DNA strand exchange.
The flexibility of the N-terminal domain in the archaeal
RadA and eukaryotic Rad51/Dmc1 family of proteins is
now firmly established. A crystal structure of a hepta-
meric ring formed by one archaeal RadA (PfRad51) con-
tained only one N-terminal domain, as the other six
N-terminal domains were disordered in the crystal (Shin
et al., 2003). Fitting the PfRad51 crystal subunit into an
EM reconstruction of the Sulfolobus solfataricus RadA-
DNA filament (Yang et al., 2001) required a rotation of
the N-terminal domain by w60º and a translation by
w20 A˚(Shin et al., 2003). A crystal structure of an octa-
meric ring formed by the human Dmc1 protein did not
contain any N-terminal domains, as they were all disor-
dered (Kinebuchi et al., 2004). Electron microscopic
observations have suggested an even greater degree
of flexibility and disorder (Yu et al., 2001). In the crystal
structures, movements of the N-terminal domain by
w2–4 A˚would be enough to weaken or eliminate inter-
pretable electron density. For the N-terminal domain to
vanish at w20 A˚resolution, as shown previously in EM
reconstructions for certain states of the Sulfolobus
solfataricus RadA-DNA filament (Yang et al., 2001) and
as shown in this paper for yeast Rad51-G103E-DNA fila-
ments, the N-terminal domain must be making move-
ments with amplitudes on the order of 20 A˚or greater.
Given these observations, it is interesting to consider
that the BRC3 repeat within BRCA2 has been shown to
bind to the N-terminal domain of human Rad51 protein
within Rad51-DNA filaments (Galkin et al., 2005). Muta-
tions in BRCA2 have been linked with an increased
risk of cancer, and BRCA2 has been shown to recruit
Rad51 to sites of DNA damage (Powell and Kachnic,
2003). Given that the G103Emutation inyeast Rad51 de-
stabilizes theN-terminal domain andeliminates boththe
in vitro ATPase and DNA strand-exchange activities, it is
tempting to speculate that the binding of BRCA2 to the
human Rad51 N-terminal domain is the basis for the nu-
cleation and regulation of Rad51-DNA filaments at the
site of DNA damage. While there are many caveats
that are needed in extrapolating from yeast Rad51 to
human Rad51, including the additional residues that
are present at the N terminus of the yeast protein, this
is a testable model that should serve as the basis for
further experiments both in vitro and in vivo.
RadA Protein Preparation and Crystallization
Full-length RadA from M. voltae was cloned into pET19 as reported
(Reich et al., 2001). The ORF was amplified by PCR by using pfu
polymerase (Fermentas) with a pair of primers (Integrate DNA Tech-
nologies): RADAS2GNcoIFor (CATGccatgg GTGATAATTT AACTG
ATTTG CC) and RADAXhoIRev (CCGctcgagT TAATCTTGAA TACCT
TTTTC AG). The coding sequence for amino acid sequence 63–322
of M. voltae RadA was amplified with primers (IDTDNA) D62RADAN-
deIFor (GAGATTTATG TcatatgGGTT TTAAAAGTGG TATTG) and
RADAXhoIRev. The coding sequences were inserted into pET28a
(Novagen) by double digestion and ligation. The resultant plasmids
were used to transform BL21-Codon-Plus-RIL(DE3) cells (Strata-
gene). Transformed cells were grown at 37ºC in the presence of 35
mg/l each of kanamycin and chloramphenicol. IPTG was added to
0.25 mM when cell density reached an OD600between 0.5 and 0.8.
The cells were harvested 4 hr after IPTG induction and then soni-
cated. Both proteins were found in the supernatant after centrifuga-
tion. The recombinant full-length protein carrying a S2G mutation
was purified as described (Wu et al., 2004). The D62RadA protein
fused with an N-terminal hexa-histidyl tag was purified by Ni-affinity
chromatography (Novagen manual). The fusion tag was removed by
over-night thrombin digestion at 4ºC, and the resultant protein with
four extra N-terminal residues (Gly-Ser-His-Met) was further purified
by gel filtration with an isocratic solution of 0.5 M NaCl and 30 mM
of Tris-HCl buffer (pH 7.9). Both purified proteins were concentrated
to w30 mg/ml by ultrafiltration.
The hanging drop method was employed to crystallize the
D62RadA protein at a room temperature of 21ºC. The optimal well
solution had a composition of 2 mM AMP-PNP (Sigma-Aldrich),
0.05 M MgCl2, 0.2 M NaCl, and 20% PEG 400 (Sigma) and 0.05 M
Tris-HCl (pH 7.5). The crystals grew to a maximum size of 0.05 mm 3
0.05 mm 3 0.2 mm in a week. Harvested crystals were gradually
transferred to stabilization solutions composed of the well solution
supplemented by 5%, 10%, 15%, 20%, and 25% glycerol and then
flash cooled to 100 K in a nitrogen stream generated by an Oxford
CryoSystem device. The 0.4º oscillation images were acquired
and processed with a Brukers Proteum-R system as described
(Wu et al., 2004). The counterpart in the previously solved MvRadA
model (PDB entry 1T4G)was employedas the search model to solve
the structure by the molecular replacement method implemented
in AMoRe (Navaza, 2001). Each model was iteratively rebuilt with
XtalView (McRee, 1999) and refined by CNS (Brunger et al., 1998).
in Table 2. The molecular figures were generated with Molscript
(Kraulis, 1991) and PyMOL (DeLano Scientific). The coordinates and
RadA DNA Strand-Exchange Assay
The DNA substrates were chosen from a published study (Mazin
et al.,2000). Threeoligonucleotides (#1, 63-nt, ACAGCACCAG ATTC
A GGA; #45, 31-nt, ACAGCACCAG ATTCAGCAAT TAAGCTCTAA G;
tained from Intergrated DNA Technologies. Equal molarities of com-
plementary oligonucleotides #45 and #55 were heated at 95ºC for
The Rad51/RadA N-Terminal Domain
5 min and then slowly cooled down to generatethe dsDNA substrate.
The solution for DNA strand-exchange reaction was composed of
3 mM ATP or an analogous nucleoside triphosphate, 10 mM MgCl2,
100 mM KCl, 50 mM HEPES-Tris buffer (pH 7.4), 20 mM RadA, and
1 mM oligonucleotides. The 63 nt ssDNA substrate (oligonucleotide
#1) was preincubated at 37ºC with RadA for 1 min before adding
the dsDNA substrate. The reaction was stopped at 30 min by adding
EDTA to a concentration of 20 mM and trypsin to a concentration of
1 mg/ml. After 10 min of trypsin digestion, a 10 ml sample was mixed
with 5 ml of a loading buffer composed of 30% glycerol and 0.1%
bromophenol blue and then loaded onto a 17.5% acrylamide gel.
The SDS-PAGE was developed, stained with ethidium bromide,
and visualized with an UV illuminator. An optional ATP regeneration
system was used as specified, which was composed of 6 mM of
phosphoenolpyruvate and 0.01 unit/ml of pyruvate kinase (New
RadA ATPase Assay
nium molybdate, and 1.0 M HCl was used to monitor the release of
inorganic phosphate from ATP hydrolysis (Itaya and Ui, 1966). Ab-
sorbance at 620 nm was recorded for quantification. The reaction
solutions for the DNA-dependent ATPase assay contained 3 mM
RadA, 18 mM ssDNA (in nucleotides) or dsDNA (in base pairs),
5 mM ATP, 0.05 M Tris-HEPES buffer (pH 7.4), 10 mM MgCl2,
100 mM of KCl, and 0.1% v/v 2-mercaptoethanol. Solutions without
RadA were taken as absorbance references. A 36-nt oligonucleo-
tides poly-(dT)36 (Integrated DNA Technologies) was used as
the ssDNA substrate in this assay. Full-length fX174 DNA (New
England Biolabs) linearized by Pst I digestion was used as the
Rad51 ATPase Assay
The wild-type and G103E mutant Rad51 proteins were prepared as
incubated at 30ºC as described (Seitz et al., 1998). The reaction
buffer contained 33 mM Tris-HCl (pH 7.5), 13 mM MgCl2, 1.8 mM
DTT, 1 mM ATP, and 90 mg/ml BSA (Zhang et al., 2005). In a 50 ml re-
action, Rad51 and poly(dT) ssDNA were 2 mM and 24 mM, respec-
tively (stoichiometry 1:12). The reaction was started by adding
2.5 ml [g-32P]ATP solution (containing 10 mM Tris-HCl [pH 7.5],
9.6 mM ATP, and 0.385 mCi/ml [g-32P]ATP). At the time points indi-
cated, 2.5 ml reaction volume was withdrawn and mixed with 1.25 ml
stop solution (contains 6.7 mM ATP, 6.7 mM ADP, and 33.3 mM
EDTA). One microliter of the final mixture was spotted on cellulose
PEI-F TLC plate (J.T.Baker). The free radioactive phosphate was
separated from nonhydrolyzed radioactive ATP by running the TLC
was dried and exposed to a phosphoimager plate and quantified
with Storm 840 and Imagequant 5.2 software (GE Healthcare).
Complex Formation and Electron Microscopy
All complexes were formed in 25 mM Triethanolamine-HCl (Fisher)
buffer (pH 7.2).
Preparation of mvRadA-S2G or mvRadA-D62 with dsDNA
and ADP and Aluminum Fluoride Complexes
Incubation at 37ºC for 30 min, with mvRadAS2G (or mvRadAD62)
concentration of 3 mm, protein to calf thymus dsDNA (Sigma) ratio
of 40:1 (w/w), ADP (Sigma) 1.25 mM, magnesium acetate (Sigma)
2 mM, NaF (Aldrich) 1.25 mM, and Al(NO3)3(Aldrich) 1.25 mM.
Preparation of mvRadA-S2G or mvRadA-D62 with dsDNA
and AMP-PNP Complexes
Incubation was at 37ºC for 10 min, with mvRadA concentration of 3
mm, protein to thymus dsDNA (Sigma) ratio of 40:1 (w/w), AMPPNP
(Sigma) 2.5 mM, and magnesium acetate (Sigma) 2 mM.
Incubation was at 37ºC for 30 min, with mvRadA-D62 concentration
of 3 mm, protein to M13 ssDNA (Sigma) ratio of 80:1 (w/w), ATP
(Sigma) 1.25 mM, magnesium acetate (Sigma) 10 mM, KCl
(J.T.Baker) 100 mM, NaF (Aldrich) 1.25 mM, and Al(NO3)3(Aldrich)
Preparation of Rad51 or Rad51-G103E with dsDNA and ATP
and Aluminum Fluoride
Incubation was at 25ºC for 15 min, with Rad51 concentration of 1.5
mm, protein to thymus dsDNA (Sigma) ratio of 40:1 (w/w), ATP
(Sigma) 1.25 mM, magnesium acetate (Sigma) 10 mM. After this
initial incubation, NaF (Aldrich) and Al(NO3)3(Aldrich) were added
to a final concentration of 1.25 mM and incubated at 25ºC for an
additional 30 min.
or ATPgS Complexes
Incubation was at 25ºC for 15 min, with Rad51 (or Rad51-G103E)
concentration of 1.5 mm, protein to thymus dsDNA (Sigma) ratio of
40:1 (w/w),AMP-PNP (Sigma) or ATPgS (Boehringer) 1.25 mM, mag-
nesium acetate (Sigma) 10 mM.
Samples were applied to carbon-coated grids and stained with
2% uranyl acetate (w/v). Images were recorded on film with a Tecnai
12 electron microscope operating at 80 keV with a nominal magnifi-
cation of 30,0003. Negatives were scanned with a Nikon Coolscan
8000 densitometer at a raster of 4.2 A˚/pixel.
We extracted 945 segments, each 60 3 110 pixels. From each seg-
ment, five overlapping boxes, each 60 3 60 pixels, were extracted to
form a set of 4725 images. The IHRSR method (Egelman, 2000) was
used to make an overall reconstruction of the set. To evaluate the
quality of the raw images, each set of five overlapping segments
was cross-correlated with projections of the reconstruction. Only
sets having at least four segments with the same polarity were re-
tained to create a new set of 2880 images. To sort these segments
by pitch, we created a set of 11 model volumes having pitch values
from 75–125 A˚with a step size of 5 A˚, and these were cross corre-
lated with the images. By this method, 2051 filament segments
having a pitch from 90–100 A˚were used for the final reconstruction.
After 60 iterations, the structure converged to a symmetry of 57.6º
rotation per subunit and an axial rise of 15.8 A˚per subunit.
Similar reconstruction procedures were used for these as described
above. The 3340 image segments with the best polarity were cross
correlatedwithprojections ofmodelvolumes havingapitch ofeither
64, 71, or 79 A˚. Segments assigned to the 71 A˚pitch class (n = 1864)
an axial rise of 9.9 A˚per subunit.
Table 2. X-Ray Crystallographic Data and Structure Refinement
X-Ray Crystallographic Data (Cu Ka
Radiation, Wavelength = 1.5418 A˚;
Space Group P61)
Unit cell dimensions (A˚)
a, b = 67.8; c = 91.1
Crystal Structure Refinement
Reflection with F > 0
R factor/free Rc
237 amino acids,
1 AMP-PNP, 2 Mg2+
0.0075 A˚/1.29ºRmsd: bond/angle
aValues in parentheses refer to values in the highest resolution shell.
bRsym=PjIh2<I>hj=PIh, where <I>his average over symmetry
Fobs. The free R factor is calculated
with a randomly selected 5% of the reflections set aside throughout
equivalents, and h is reflection index.
Theappearanceofthefilamentsformed bythisfragmentinthe pres-
ence of ATP-AlF was quite different from that of the full length
MvRadA, and analysis of diffraction patterns showed that the layer
line at w1/50 A˚contained a Bessel order n = 2. By using the IHRSR
procedure, a stable solution was found for a symmetry of 157º per
subunit, with an axial rise of 6.2 A˚, which corresponds to a two-
stranded helical structure with weight subunits per turn of each
each other by exactly half the rise per subunit along one strand. In-
dications existedthat the relationship between the strands was vari-
able within the data set. To improve the reconstruction, we created
modelvolumeshaving differentshifts androtationsbetweenthetwo
strands, and these were cross correlated with 3855 images.Only the
1235 segments that yielded the best correlation with the symmetri-
cal two-stranded structure described (a 180º rotation between the
strands, with an axial shift of one-half of the subunit rise) were
used for the final reconstruction. This yielded a symmetry of 157.2º
and an axial rise of 6.3 A˚.
We collected 5635 segments (each 60 3 60 pixels) and sorted these
by pitch by using the procedure described for MvRadA-ADP-AlF-
dsDNA (see above). Segments having a pitch of 90–100 A˚(n = 3379)
were used for the final reconstruction, which yielded a symmetry of
56.2º with an axial rise of 15.5 A˚.
segments collected and 3379 images having a pitch of 90–100 A˚
included in the final reconstruction. The symmetry converged to
56.1º with an axial rise of 15.5 A˚.
We thank the Saskatchewan Structural Sciences Centre for access
to its X-ray facility. This work was supported by Natural Sciences
Research Foundation, and Canadian Institutes of Health Research
(Y.L.) and National Institutes of Health GM58015 (W.D.H.) and
GM35269 (E.H.E.). X.P.Z. is a Susan G. Komen Breast Cancer
Foundation postdoctoral fellow (PDF403213). Y.L. is a CIHR New
Received: February 24, 2006
Revised: April 8, 2006
Accepted: April 10, 2006
Published: June 13, 2006
Aboussekhra, A., Chanet, R., Adjiri, A., and Fabre, F. (1992). Semi-
dominant suppressors of Srs2 helicase mutations of Saccharomy-
ces cerevisiae map in the RAD51 gene, whose sequence predicts
a protein with similarities to procaryotic RecA proteins. Mol. Cell.
Biol. 12, 3224–3234.
Abrahams, J.P., Leslie, A.G., Lutter, R., and Walker, J.E. (1994).
Structure at 2.8 A˚resolution of F1-ATPase from bovine heart mito-
chondria. Nature 370, 621–628.
Aihara, H., Ito, Y., Kurumizaka, H., Terada, T., Yokoyama, S., and
Shibata, T. (1997). An interaction between a specified surface of
the C-terminal domain of RecA protein and double-stranded DNA
for homologous pairing. J. Mol. Biol. 274, 213–221.
Aihara, H., Ito, Y., Kurumizaka, H., Yokoyama, S., and Shibata, T.
(1999). The N-terminal domain of the human Rad51 protein binds
DNA: structure and a DNA binding surface as revealed by NMR.
J. Mol. Biol. 290, 495–504.
Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,
Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M.,
Pannu, N.S., et al. (1998). Crystallography & NMR system: a new
software suite for macromolecular structure determination. Acta
Crystallogr. D Biol. Crystallogr. 54, 905–921.
of recombination-deficient mutants of Escherichia coli K12. Proc.
Natl. Acad. Sci. USA 53, 451–459.
Conway, A.B., Lynch, T.W., Zhang, Y., Fortin, G.S., Fung, C.W., Sy-
mington, L.S., and Rice, P.A. (2004). Crystal structure of a Rad51
filament. Nat. Struct. Mol. Biol. 11, 791–796.
Courcelle, J., Khodursky, A., Peter, B., Brown, P.O., and Hanawalt,
P.C. (2001). Comparative gene expression profiles following UV
exposure in wild-type and SOS-deficient Escherichia coli. Genetics
Datta, S., Prabu, M.M., Vaze, M.B., Ganesh, N., Chandra, N.R.,
Muniyappa, K., and Vijayan, M. (2000). Crystal structures of Myco-
bacterium tuberculosis RecA and its complex with ADP-AlF(4):
implications for decreased ATPase activity and molecular aggrega-
tion. Nucleic Acids Res. 28, 4964–4973.
Datta, S., Ganesh, N., Chandra, N.R., Muniyappa, K., and Vijayan, M.
(2003a). Structural studies on MtRecA-nucleotide complexes:
insights into DNA and nucleotide binding and the structural signa-
ture of NTP recognition. Proteins 50, 474–485.
RecA and its nucleotide complexes. J. Bacteriol. 185, 4280–4284.
Egelman, E.H. (2000). A robust algorithm for the reconstruction of
helical filaments using single-particle methods. Ultramicroscopy
Galkin, V.E., Esashi, F., Yu, X., Yang, S., West, S.C., and Egelman,
E.H. (2005). BRCA2 BRC motifs bind RAD51-DNA filaments. Proc.
Natl. Acad. Sci. USA 102, 8537–8542.
Itaya, K., and Ui, M. (1966). A new micromethod for the colorimetric
determination of inorganic phosphate. Clin. Chim. Acta 14, 361–366.
Kinebuchi, T., Kagawa, W., Enomoto, R., Tanaka, K., Miyagawa, K.,
Shibata, T., Kurumizaka, H., and Yokoyama, S. (2004). Structural ba-
sis for octameric ring formation and DNA interaction of the human
homologous-pairing protein Dmc1. Mol. Cell 14, 363–374.
Kowalczykowski, S.C., and Eggleston, A.K. (1994). Homologous
pairing and DNA strand-exchange proteins. Annu. Rev. Biochem.
Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both de-
tailedand schematic plots of protein structures. J. Appl. Crystallogr.
Mazin,A.V.,Zaitseva,E.,Sung, P.,andKowalczykowski, S.C.(2000).
Tailed duplex DNA is the preferred substrate for Rad51 protein-
mediated homologous pairing. EMBO J. 19, 1148–1156.
McRee, D.E. (1999). XtalView/Xfit—a versatile program for manipu-
lating atomic coordinates and electron density. J. Struct. Biol. 125,
Navaza, J. (2001). Implementation of molecular replacement in
AMoRe. Acta Crystallogr. D Biol. Crystallogr. 57, 1367–1372.
Ogawa,T., Yu, X., Shinohara, A., and Egelman, E.H. (1993). Similarity
of the yeast RAD51 filament to the bacterial RecA filament. Science
Pellegrini, L., Yu, D.S., Lo, T., Anand, S., Lee, M., Blundell, T.L., and
structure of a RAD51-BRCA2 complex. Nature 420, 287–293.
Powell, S.N., and Kachnic, L.A. (2003). Roles of BRCA1 and BRCA2
in homologous recombination,DNA replication fidelity and the cellu-
lar response to ionizing radiation. Oncogene 22, 5784–5791.
Rajan, R., and Bell, C.E. (2004). Crystal structure of RecA from Dein-
ococcus radiodurans: insights into the structural basis of extreme
radioresistance. J. Mol. Biol. 344, 951–963.
Reich, C.I., McNeil, L.K., Brace, J.L., Brucker, J.K., and Olsen, G.J.
(2001). Archaeal RecA homologues: different response to DNA-
damaging agents in mesophilic and thermophilic Archaea. Extremo-
philes 5, 265–275.
Sandler, S.J., Satin, L.H., Samra, H.S., and Clark, A.J. (1996). recA-
like genes from three archaean species with putative protein
products similar to Rad51 and Dmc1 proteins of the yeast Saccha-
romyces cerevisiae. Nucleic Acids Res. 24, 2125–2132.
Schlacher, K., Leslie, K., Wyman, C., Woodgate, R., Cox, M.M., and
Goodman, M.F. (2005). DNA polymerase V and RecA protein, a min-
imal mutasome. Mol. Cell 17, 561–572.
The Rad51/RadA N-Terminal Domain
Sehorn, M.G., Sigurdsson, S., Bussen, W., Unger, V.M., and Sung, P. Download full-text
(2004). Human meiotic recombinase Dmc1 promotes ATP-depen-
dent homologous DNA strand exchange. Nature 429, 433–437.
kowski, S.C. (1998). RadA protein is an archaeal RecA protein
homolog that catalyzes DNA strand exchange. Genes Dev. 12,
Shin, D.S., Pellegrini, L., Daniels, D.S., Yelent, B., Craig, L.,Bates, D.,
Yu, D.S., Shivji, M.K., Hitomi, C., Arvai, A.S., et al. (2003). Full-length
archaeal Rad51 structure and mutants: mechanisms for RAD51
assembly and control by BRCA2. EMBO J. 22, 4566–4576.
Shinohara, A., Ogawa, H., and Ogawa, T. (1992). Rad51 protein in-
volved in repair and recombination in S. cerevisiae is a RecA-like
protein. Cell 69, 457–470.
Story, R.M., Weber, I.T., and Steitz, T.A. (1992). The structure of the
E. coli recA protein monomer and polymer. Nature 355, 318–325.
VanLoock, M.S., Yu, X., Yang, S., Lai, A.L., Low, C., Campbell, M.J.,
and Egelman, E.H. (2003). ATP-mediated conformational changes in
the RecA filament. Structure 11, 187–196.
Wu, Y., He, Y., Moya, I.A., Qian, X., and Luo, Y. (2004). Crystal struc-
ture of archaeal recombinase RADA: a snapshot of its extended
conformation. Mol. Cell 15, 423–435.
Xing, X., and Bell, C.E. (2004a). Crystal structures of Escherichia coli
RecA in complex with MgADP and MnAMP-PNP. Biochemistry 43,
Xing, X., and Bell, C.E. (2004b). Crystal structures of Escherichia coli
RecA in a compressed helical filament. J. Mol. Biol. 342, 1471–1485.
Yang, S., Yu, X., Seitz, E.M., Kowalczykowski, S.C., and Egelman,
E.H. (2001). Archaeal RadA protein binds DNA as both helical fila-
ments and octameric rings. J. Mol. Biol. 314, 1077–1085.
Yu, X., Jacobs, S.A., West, S.C., Ogawa, T., and Egelman, E.H.
(2001). Domain structure and dynamics in the helical filaments
formed by RecA and Rad51 on DNA. Proc. Natl. Acad. Sci. USA
Zhang, X.P., Lee, K.I., Solinger, J.A., Kiianitsa, K., and Heyer, W.D.
(2005). Gly-103 in the N-terminal domain of Saccharomyces cerevi-
siae Rad51 protein is critical for DNA binding. J. Biol. Chem. 280,
The coordinates and structure factors for RadA-D62 have been en-
tered into the Protein Data Bank with the accession number 2GDJ.