Tài liệu Báo cáo khoa học: Aggregative organization enhances the DNA end-joining process that is mediated by DNA-dependent protein kinase pdf

Aggregative organization enhances the DNA end-joining
process that is mediated by DNA-dependent protein kinase
Masahiko Takahagi and Kouichi Tatsumi
Research Center for Radiation Safety, National Institute of Radiological Sciences, Chiba, Japan
DNA double-strand breaks (DSBs) are a serious threat
to the genetic integrity of organisms, causing cell death
if not repaired. The repair mechanism for DSBs resides
not only in catalytic processes but also in the associ-
ation with chromatin structures [1,2], although the
details in the higher-order context remain obscure.
Some evidence has implicated structural alterations in
the vicinity of DSB sites. DSBs can form nuclear foci
linked to phosphorylated histone H2AX (c-H2AX) [3],
which is responsible for the redistribution of repair
factors to DSB sites [4], although it is dispensable
for initial damage recognition [5]. Approximately 2000
c-H2AX molecules accumulate per focus in a normal
human cell, suggesting reorganization of chromosomal
DNA over a region of Mbp order [6]. c-H2AX-associ-
ated foci are morphologically dynamic; the DSB-con-
taining chromosome domains can be mobile, and in a
subpopulation of damaged cells, they can juxtapose
via an adhesion process irrespective of DNA repair
processes [7].
In addition to these large-scale responses to DSBs,
real-time analysis of the temporo-spatial distribution
of DNA repair factors in situ in living cells has been
providing us with striking information. For instance,
even following exposure to ionizing radiation (IR), the
DNA end-binding factor Ku moves rapidly throughout
the nucleus but associates transiently with filamentous

nuclear substrates [8]. A checkpoint regulator NBS1,
the product of the Nijmegen breakage syndrome gene,
shuttles rapidly between DSB sites and the flanking
chromatin [9]. These findings indicate that the action
of DSB-interactive proteins within the nuclear micro-
environment must be coupled with the mobile state of
those proteins.
Keywords
coacervate; DNA aggregation; DNA end-
joining; DNA-PK; S ⁄ MAR
Correspondence
K. Tatsumi, Research Center for Radiation
Safety, National Institute of Radiological
Sciences, 9–1, Anagawa 4, Inage-ku,
Chiba 263–8555, Japan
Fax: +81 43 255 6497
Tel: +81 43 206 3087
E-mail:
(Received 7 April 2006, accepted 11 May
2006)
doi:10.1111/j.1742-4658.2006.05317.x
The occurrence of DNA double-strand breaks in the nucleus provokes in
its structural organization a large-scale alteration whose molecular basis is
still mostly unclear. Here, we show that double-strand breaks trigger pref-
erential assembly of nucleoproteins in human cellular fractions and that
they mediate the separation of large protein–DNA aggregates from aque-
ous solution. The interaction among the aggregative nucleoproteins pre-
sents a dynamic condition that allows the effective interaction of
nucleoproteins with external molecules like free ATP and facilitates intrin-
sic DNA end-joining activity. This aggregative organization is functionally

coacervate-like. The key component is DNA-dependent protein kinase
(DNA-PK), which can be characterized as a DNA-specific aggregation fac-
tor as well as a nuclear scaffold ⁄ matrix-interactive factor. In the context of
aggregation, the kinase activity of DNA-PK is essential for efficient DNA
end-joining. The massive and functional concentration of nucleoproteins
on DNA in vitro may represent a possible status of nuclear dynamics
in vivo, which probably includes the DNA-PK-dependent response to mul-
tiple double-strand breaks.
Abbreviations
DSB, double-strand break; DNA-PK, DNA-dependent protein kinase; EJ, end-joining; c-H2AX, phosphorylated histone H2AX; HMGB, high
mobility group box; IR, ionizing radiation; LD, linear duplex; NC, nicked circular; NF, nuclear fraction; SC, supercoiled; S ⁄ MAR,
scaffold ⁄ matrix attached region; ss, single strand; SSC, single-strand circular.
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3063
One hallmark of the initial response to DSBs is the
rapid activation of ATM (product of the ataxia telan-
giectasia mutated gene), the protein kinase that is
mutated in the human hereditary disease ataxia telan-
giectasia: induced chromatin alterations relay its inter-
molecular autophosphorylation and dimer dissociation
[10]. Over 50% of the ATM molecules in primary
fibroblast-like cells are activated within 5 min after
exposure to 0.5 Gy of IR. This activation is triggered
even by treatment with hypotonic conditions or chem-
ical chromatin modifiers. Nevertheless, detectable
DSBs are not induced by such treatments. It seems
likely therefore that chromatin alterations, at a certain
distance from DSB sites, initiate the ATM activation.
Such distant effects of IR have been similarly recog-
nized as a rapid change of the topological constraints
on chromosomal DNA in the nuclei [11–13].

The magnitude of these nuclear changes appears to
be relevant to the anchoring of DNA loop domains
on the nuclear scaffold or matrix, where the nuclear
scaffold ⁄ matrix attached region (S ⁄ MAR) DNA
sequences are thought to mediate the higher-order net-
working of S ⁄ MAR-binding proteins. This property
has been biochemically defined as the preferential
aggregation of those proteins onto specific DNA. The
potential for aggregation has been characterized
for major nucleoproteins including histone H1 [14],
topoisomerase II [15], lamin B1 [16] and SAF-A
(hnRNP U) [17].
It is known that the aggregation of DNA is com-
patible with the functions of proteins under in vitro
conditions. Polyamine-mediated aggregation involves
the suitable arrangement of DNA helices, allowing
the efficient catenation of the circular duplexes by
Escherichia coli topoisomerases, the free exchange of
DNA between neighboring aggregates by the enzyme
[18], and the activation of transcription by E. coli
RNA polymerase [19]. The E. coli recombination fac-
tor RecA promotes coaggregation between single- and
double-strand DNA, and the diffusible nature of com-
ponents within the nucleoprotein networks encourages
inefficient homologous pairing [20,21]. These examples
illustrate that an aggregative organization underlies
the interplay among proteins and DNA and that it
serves with a compact but fluid state for catalytic
reactions.
Thus, in view of the large-scale nuclear responses in

human cells, it is proposed that aberrant DNA struc-
tures, particularly DSBs, are recognized and processed
through interactions with repair factors and S ⁄ MAR-
binding proteins in an aggregative context. We exam-
ined under in vitro conditions whether DSBs mediate
the aggregative assembly of nucleoproteins that assures
reliable repair reactions. Here we show that an exten-
sive aggregation of human nucleoproteins, including
DNA repair kinase DNA-PK, is mobilized onto DNA
with ends in a cell-free system, and that it produces
the characteristic ‘coacervate-like organization’, which
greatly stimulates DNA end-joining.
Results and Discussion
Aberrant types of DNA coaggregate with
nucleoproteins
To know whether the assembly of nucleoproteins with
DSBs is aggregative, we examined the precipitable
property of the complex. The assay system adopted
not only linear duplex as a model substrate of DSBs
but also other forms of DNA including supercoiled,
nicked circular and single-strand circular DNA as
competitors. The reaction mixtures containing nuclear
fractions (Fig. 1A) from the human lymphoblastoid
cells, WI-L2-NS, were centrifuged after incubation
with DNA in solution. A visibly gelated protein–DNA
complex precipitated at the bottom of the test tubes.
After being separated from the supernatant and depro-
teinized, the precipitate was subjected to gel electro-
phoresis for size-based analysis of DNA. Except for
the supercoiled form, all aberrant types were preferen-

tial targets for aggregation, which occurred depending
on the concentration of protein added (Fig. 1B, upper
panel, lanes 3–7). Under the physiological ionic
strength (0.15 m NaCl), the formation of stable aggre-
gates required a large amount of extracts. To avoid
this problem and the complexity between substrates,
the aggregation of a linear DNA was examined under
a lower ionic condition (0.05 m NaCl), and the extent
of the aggregation was compared to residual molecules
in the supernatant (Fig. 1C, upper panel). The increas-
ing amounts of extracts quantitatively precipitated the
DNA, ending up with saturation at a full extent (lanes
7 and 14). Considering the biological significance of
the aggregation, we expected that the aggregates would
be structurally fluid and anisotropic like the polyam-
ine-induced DNA aggregation and so would possess
the feature of coacervate [22], defined as a dynamic
phase which appears by self-assembly of colloidal bio-
polymers in dilute solutions [23].
Nucleoprotein aggregation enhances intrinsic
DNA end-joining activity
With regard to the biochemical aspects of the aggrega-
tion, we examined whether or not the assembly is rela-
ted to DNA repair processes. We adopted as a simple
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3064 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
model system the rejoining reaction of DNA ends
(end-joining; EJ), which has been characterized as the
primary repair mechanism of DSBs [1]. EJ reaction
was carried out under a condition for full aggregation

of linear DNA (0.1 lg) (Fig. 1C, lanes 7 and 14). After
a 2.7 kbp linearized-plasmid DNA was coaggregated
with human nuclear fractions, an EJ reaction was
initiated by the addition of ATP as a cofactor into the
surrounding buffer solution (Fig. 2A). At low ionic
strength (50 mm NaCl), the intermolecular joining of
DNA ends was greatly enhanced. This stimulation was
also noted at higher ionic strength (150 mm NaCl),
although the yield of EJ products was lower in accord-
ance with the efficiency of DNA aggregation (unpub-
lished work). Even when the precipitable aggregate
was detached from the bottoms of the tubes, the EJ
A
Human Nuclear Extract
DEAE Sepharose
Denatured DNA Cellulose
Mono Q
0.2 M / FT 0.5 M
1.0 M NaCl
Active fraction
Nuclear fraction
0.5
M /FT
0.1 M / FT 0.5 M
1.0 M NaCl
1.0 M NaCl
B
SC
NC
SSC

LD
-
Nuclear
fraction
*
Mono Q
fraction
DNA affinity
fraction
Input DNA
Substrate
N
oD
N
A
1234567
*
C

ppt. sup.
LD
Nuclear
fraction
LD
DNA affinity
fraction
LD
Mono Q
fraction
Inpu

t
1 2 3 4 5 6 7 8 9101112131415
DNA affinity fraction
Mono Q fraction
Fig. 1. Purification of DNA aggregation activity from human cells. (A) Scheme for the purification of DNA aggregation activity. (B) DNA aggre-
gation activity in chromatographic fractions. Nucleoproteins from WI-L2-NS cells were sequentially separated through DEAE sepharose (nuc-
lear fraction), denatured DNA cellulose (DNA affinity fraction) and Mono Q columns (Mono Q fraction). Input DNA contained four forms of
uX174 phage DNA, including 0.05 lg of HaeIII-digested linear duplex (LD), 0.05 lg of single-stranded circular (SSC) and 0.05 lg of super-
coiled (SC) DNA containing nicked (NC) forms. Input DNA was incubated with each fraction increasing in amount by two-fold at physiological
ionic strength (150 m
M NaCl). Either 240 lg of nuclear fraction, 70 lg of DNA affinity fraction or 70 lg of Mono Q fraction was used as a
maximal protein concentration (lane 7). After centrifugation of reaction mixtures, precipitated DNA was analyzed by 1% agarose gel electro-
phoresis. DNA was visualized by staining with ethidium bromide. Results are represented by negative images. Asterisks indicate nucleic
acids derived from cells. (C) Coaggregation of linear DNA by separated fractions. Reactions of fractions with EcoRI-digested pUC18 DNA
(0.1 lg) as an input substrate (LD) were conducted in the presence of 50 m
M NaCl. After centrifugation of reaction mixtures, both centrifugal
precipitates (ppt.) and its supernatant (sup.) were resolved on an agarose gel. Maximal aggregation of DNA was achieved by using either
5.4 lg of nuclear fraction, 1.0 lg of DNA affinity fraction or 0.7 lg of Mono Q bound fraction (lanes 7 and 14).
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3065
activity was not enhanced (supplementary Fig. S1).
The activation can therefore be attributed to the aggre-
gative organization of nucleoproteins but not to con-
tact with the plastic surface. A similar activity was
detectable in nuclear fractions from several other
human lymphoblastoid cell lines, HeLa S3 cells and
placenta tissue (unpublished work), implying that the
promotion of EJ by aggregation is common to many
types of human cells. It is very likely that the resultant
organization of nuclear components from dilute aque-

ous solutions involves either preferential assembly of
DSB repair factors containing DNA ligase activity,
valid synapsis of DNA ends, or sequestration from
inhibitory factors. Because the aggregates with efficient
EJ capacity are selective in the intermolecular mode
of EJ and are readily accessible to external ATP
molecules, the internal structure must be fluid and
must be an open system. Thus, we argue that this
organization of nucleoproteins on linear DNA qualifies
as coacervation, which has been implicated in the
mechanism for the condensed process of prebiotic
charged polymers and has been established as a condi-
tion interactive with the external environment during
their continuous synthesis and breakdown [23, and ref-
erences therein].
DNA aggregation activity is coupled with
EJ activity
To further confirm the relationship between protein–
DNA aggregation and EJ activity, we fractionated
the flow-through of DEAE-Sepharose (the nuclear
fraction) by sequential chromatography with dena-
tured-DNA coupled affinity columns and Mono Q
anion-exchange columns (Fig. 1A). Both the aggrega-
tion activity and EJ activity were enriched by a series
of fractionations. As is the case with nuclear frac-
tions, the DNA affinity- and Mono Q bound
fractions coaggregated with linear, nicked and single-
strand DNA with a similar preference (Fig. 1B). On
the other hand, the substrate specificity of EJ was
distinct among fractions (Fig. 2B). For both DNA

affinity- and Mono Q bound fractions, 5¢-overhangs
were the most reactive form, and 3¢-overhangs were
also relatively reactive. The reactivity of blunt ends
was much lower in DNA affinity fractions, but was
as high as that of 3¢-overhangs in Mono Q fractions
(compare lanes 24 and 29). The active fraction from
the Mono Q column was also devoid of ATP-inde-
pendent EJ activity, which was present in DNA affin-
ity fractions (lanes 5, 10 and 15). These results
indicate that EJ and aggregation activities can be clo-
sely associated but are functionally independent. The
difference in protein composition among fractions
may account for their distinct reactivity with end-
structures in EJ (Fig. 3A, compare lanes 3 and 4).
Apparently, our purification scheme is different from
the established one for major end-joining activity [24].
In addition, immunoblot analysis showed that the spe-
cific ligase complex, including DNA ligase IV and
XRCC4 proteins, fully passed through the DNA affin-
ity-column, so they were not present in the Mono Q-
active fraction (supplementary Fig. S2). We reason
that the EJ activity enriched here seems to depend on
some other ligase activities. The candidate could be
DNA ligase III, which has recently been demonstrated
A
B
In aggregate In solution
Multimer
Trimer
Dimer

Monomer
0 5 10 30 60 120 0 5 10 30 60 120 Time (min)
Input
0 43 57 68 72 73 0 16 20 25 30 35 EJ products (%)
Fraction
Aggregation
5'
-
overhang 3'-overhang blunt
T4 DNA ligase

1 23456789101112131415
Multimer
Trimer
Dimer
Monomer
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
DNA affinity
fraction
Multimer
Trimer
Dimer
Monomer
Mono Q
fraction
ATP
+
-

+
+
+
+
+
+
+
+
+
++
++
+
+
+
+
+
++
++
+
+
+
+
Fig. 2. End-joining activity is associated with nucleoprotein-DNA
coaggregation. Linearized pUC18 DNA of an input substrate (Mono-
mer) was treated with separated fractions under conditions for its
maximum aggregation as shown in Fig. 1C. EJ reaction was initi-
ated by the addition of ATP onto the aggregates. (A) Nucleopro-
tein–DNA coaggregation enhances intrinsic EJ activities in nuclear
fraction. EJ reactions were monitored as a function of incubation
time and were compared between in aggregate and in solution.

The EJ occurred predominantly in an intermolecular manner, produ-
cing dimer, trimer and multimers. (B) Preference of DNA end-struc-
ture in the aggregation-coupled EJ. Three types of linearized pUC18
(0.1 lg), 5¢-overhangs, 3¢-overhangs and blunt ends, were incubated
with DNA affinity (1.0 lg) and Mono Q bound (0.7 lg) fractions.
Under identical conditions except for centrifugal aggregation, T4
DNA ligase promoted not only intermolecular but also intramolecu-
lar ligation, parts of which were resolved with faster mobility than
the monomer (lanes 2, 7, 17 and 22).
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3066 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
in the context of an alternative EJ pathway [25,26].
The characterization of ligase III is underway in the
aggregation system.
Identification of proteins aggregative with
aberrant DNA
The aggregative nucleoproteins were collected under
conditions that allow maximal precipitation of input
DNA including linear, nicked and single-strand (ss)
forms (Fig. 3A, lanes 7–9), and were resolved on an
SDS ⁄ PAGE. For identification, the primary structure
of major polypeptides was determined by peptide map-
ping and partial protein sequencing. When Mono Q
bound fractions were subjected to assembly with
DNA, all aggregates contained proteins of DNA-PKcs,
180 k, 170 k, SAF-A, 98 k, nucleolin, Ku80, Ku70,
histone H1 and 28 k proteins, regardless of the type of
damage (Fig. 3A, lanes 7–9). In the aggregation of lin-
ear and nicked DNA, and high mobility group box
(HMGB)1 and HMGB2 proteins were additional ele-

ments (lanes 7 and 8). ssDNA induced a noticeable
accumulation of 95 k and nucleolinD proteins in addi-
tion to DNA-PKcs, Ku80 and Ku70 (lane 9). On the
other hand, a relaxed form of closed circular double-
strand DNA as an undamaged substrate limited the
composition and quantity of aggregative proteins (lane
6). It should be noted that proteins that specifically
aggregate with linear DNA were similar to those that
aggregate with nicked DNA, and that some of them
also assemble with ssDNA. Interestingly, the compo-
nents of DNA-dependent protein kinase, DNA-PKcs,
Ku80 and Ku70, were the common factors to aggre-
gate with preference to damaged types. By contrast,
histone H1 was detectable in the aggregates with all
kinds of DNA examined. Also, these results indicate
that centrifugal precipitation was a uniquely effective
means to collect and classify the factors aggregative
with aberrant DNA.
Characterization of DNA aggregation-promoting
factors
We expected that the candidates that coaggregate with
aberrant DNA would be relatively abundant factors.
Major proteins in the Mono Q bound fraction, inclu-
ding DNA-PKcs, SAF-A, nucleolin, nucleolinD,
Ku (as a Ku70 ⁄ Ku80 heterodimer), HMGB1 and
HMGB2, have been purified by column chromatogra-
phy (Fig. 3B). In contrast with experiments using par-
tially purified fractions, none of the isolated proteins
coprecipitated with DNA at physiological ionic
strength. However, as the NaCl concentration in reac-

tion mixtures was decreased from 150 mm to 50 mm,
three proteins, DNA-PKcs, SAF-A and nucleolin,
began to interact selectively with specific targets in
an aggregative manner (Fig. 4). Nucleolin assembled
(kDa)
100
75
50
35
25
150
225
DNA-PKcs
SAF-A
Ku80
Ku70
Nucleolin
Nucleolin∆
HMGB2
HMGB1
170 k
180 k
Histone H1
98 k
NF
Marker
Mono Q
DNA a
f
fin

i
t
y
n
o
DN
A
relaxed
L
D
NC
SSC
Separated
fractions
Aggregative
proteins
123456789
28 k
DNA
-PKcs
Ku70/Ku80
SAF-A
Nucleolin
Nucl
eo
l
i
n∆
Histone H1
Marker

Marker
H
M
GB2
HM
G
B1
A
B
Fig. 3. Identification and isolation of the aggregative nucleoproteins.
(A) Nucleoproteins in chromatographic fractions and in aggregates
with DNA. Proteins were analyzed by SDS ⁄ PAGE in a 10% gel and
stained with Coomassie brilliant blue. The profile of fractionated pro-
teins included nuclear fraction (NF; lane 2), DNA affinity fraction
(lane 3) and Mono Q fraction (lane 4). For aggregation of proteins,
the Mono Q bound fraction (7.0 lg) was incubated with no DNA
(lane 5) or aberrant types of DNA (1.0 lg), including relaxed closed
circular (lane 6), NC, nicked form (lane 8) and SSC, single-strand cir-
cular (lane 9) derived from uX174, and linear duplex (LD) of EcoRI-
digested pUC18 (lane 7). Several proteins were identified by amino
acid sequencing. Relatively abundant proteins are marked with
arrows. (B) Major nucleoproteins isolated from Mono Q fraction. Iso-
lated preparations of DNA-PKcs (2.0 lg), Ku70 ⁄ Ku80 (2.0 lg), SAF-A
(0.4 lg), nucleolin (0.8 lg), nucleolinD (0.4 lg) and histone H1 (2 lg)
were resolved by a 10% SDS ⁄ PAGE. Purified HMGB1 (2 lg) and
HMGB2 (2 lg) were analyzed on a 12% SDS ⁄ PAGE. The sizes of
marker proteins are 94, 67, 43, 30 and 20 k.
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3067
preferentially on ssDNA but not on any variations

of duplex DNA. This interaction required Mg
2+
as a
cofactor. Interestingly, nucleolinD, whose N-terminal
138 residues of 709 amino acids are spontaneously
truncated in most human lymphoblastoid cell lines
(unpublished work), was deficient in aggregation activ-
ity, implying that the truncated region may be indis-
pensable for aggregation. Likewise, DNA-PKcs
promoted predominant aggregation of ssDNA when
incubated at relatively low concentrations. At higher
ratios of protein to DNA, DNA-PKcs started to inter-
act with nicked and linear DNA but not with super-
coiled DNA. Again, Mg
2+
moderately facilitated
protein–DNA interaction. SAF-A coaggregated with
all damaged types of DNA in a Mg
2+
-dependent man-
ner. This preference is conditional since supercoiled
DNA was aggregative in the absence of Mg
2+
. These
results suggest that the DNA repair factor DNA-PKcs
is an aggregation factor that responds to aberrant
DNA structures in a manner similar to that of
S ⁄ MAR-binding proteins, including nucleolin and
SAF-A. None of the isolated Ku, HMGB1 or
HMGB2 coaggregated with any DNA, although in

Mono Q fractions they are aggregative dependent on
linear and nicked DNAs (Fig. 3A, lanes 7 and 8). Con-
versely, histone H1 aggregated with all substrates
including supercoiled DNA, a finding consistent with
the fact that it was accompanied in protein analysis by
all forms (Fig. 3A, lanes 6–9). The basic polypeptide is
thought to interact electrostatically with negative-
charged DNAs in a nonspecific manner as well as
multivalent cations of polyamines [27]. This way may
be distinct from that of nonbasic factors of
DNA-PKcs, SAF-A and nucleolin, which were stably
concentrated on anion-exchange resins (Experimental
procedures). With regard to the action mode, the role
of the N-terminal portion of nucleolin is suggestive.
This region includes the basic and repeated octapep-
tides that are similar to histone H1 [28] and acidic
stretches that are able to be associated with histone
H1 [29]. The truncation of the former motif and a part
of the latter one abolished the ssDNA-specific aggrega-
tion activity (Fig. 4) but remained a valid binding
capacity to ssDNA (unpublished work). Therefore, the
aggregation is likely to occur not only by DNA bind-
ing but also through an aggregation-promoting pro-
cess, probably DNA-mediated intra- or intermolecular
association. We presume that the structural and func-
tional distribution on relatively large proteins may be
involved in the aggregation of aberrant DNAs.
DNA-PK is the key factor for aggregation-coupled
EJ
The functional interplay between DNA-PKcs and Ku

in the aggregates is noteworthy because they function
in a common pathway, and the protein kinase activity
of DNA-PKcs is greatly stimulated by Ku during
assembly at DNA ends [30]. We examined whether the
aggregation activity of DNA-PKcs is regulated by Ku.
While DNA-PKcs interacts exclusively with nicks and
ssDNA, the addition of increasing amounts of Ku sti-
mulated progressive aggregation of linear DNA with-
out size-dependency in the range from 200 to 1350

Nucleolin
+Mg -Mg
Nucleolin∆
+Mg -Mg
DNA-PKcs
SAF-A
HMGB1
HMGB2
Histone H1
Input
Input
SC
NC
SSC
LD
Ku70/Ku80
Fig. 4. Selective aggregation of different
forms of DNA by purified proteins. Condi-
tions were similar to those used in Fig. 1C.
Protein-mediated aggregation of DNA sub-

strates (0.15 lg) that contained linear duplex
(LD), nicked form (NC), single-strand circular
(SSC) and supercoiled (SC) were examined
in the absence or presence of 10 m
M Mg
2+
.
The two-fold increasing concentrations of
nucleolin (0.12 lg), nucleolinD (0.12 lg),
DNA-PKcs (0.3 lg), SAF-A (0.15 lg), Ku
(0.3 lg), HMGB1 (0.12 lg), HMGB2
(0.12 lg), and histone H1 (0.12 lg) proteins
were subjected to experiments.
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3068 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
base pairs (Fig. 5A, upper panel). By contrast, Ku
failed to promote the assembly of SAF-A on linear
DNA under conditions similar to those optimal for
selective aggregation of nicks and ssDNA (Fig. 5A,
lower panel). The effect of Ku on aggregating activities
of other purified factors was negative as well. These
data indicate that through protein–protein interaction,
Ku operates to specifically promote the aggregation of
DNA-PKcs on DNA ends.
The previous gel-shift study indicated that highly
purified DNA-PKcs binds to linear DNA in the
absence of Ku and predominantly forms a large com-
plex without any mobility shift in the well of the gel
[31]. The stability of the complex against salt is also
enhanced by the presence of Ku. Interestingly, the

complex formation was easily inhibited by ssDNA,
including poly(dT) and hairpin-ended DNA, although
these are unable to activate DNA-PK kinase effi-
ciently [31,32]. In addition, since such an immobile
complex in the well was found under conditions for
aggregation between DNA-PKcs and DNA, we con-
clude that the complex in this study has aspects of an
aggregate. Electron microscopic imaging has consis-
tently shown that DNA-PK is able to link multiple
linear DNA fragments into a large complex [33]. The
autophosphorylation of DNA-PK causes the disrup-
tion of such a complex formation, which is biochemi-
cally dynamic. If so, our centrifugal manipulation
appears to conduct higher-order accumulation of pre-
existing similar complexes in solution and could gen-
erate a preferential condition for the intermolecular
DNA end-joining.
The autophosphorylation of DNA-PKcs was detect-
able in aggregates with linear DNA (Fig. 5B, lane 1)
as well as in reactions in solution. Ku also enhanced
the autophosphorylation of DNA-PKcs in addition
to its own phosphorylation (Fig. 5B, lane 2). This
biochemical evidence supports the view that the DNA-
PKcs-mediated aggregates stay in a catalytically com-
petent state.
A further concern is whether the protein kinase
activity of DNA-PKcs is really associated with the EJ
reaction in the context of protein–DNA coaggregative
networks. To answer this question, we examined the
effect of the DNA-PK inhibitor wortmannin [34].

This inhibitor covalently binds to the catalytic domain
on DNA-PKcs [35], thereby leading to an abortion of
the EJ process in solution through an irreversible
A
B
Ku70/Ku80
DNA-PKcs
SAF-A
-
Input
N
o
p
rote
i
n
1234567
C
Wortmannin
Multimer
Trimer
Dimer
Monomer
Aggregate
DNA-PKcs
ATP
sup.
after
incubation
1234567891011

(kDa)
100
75
50
35
25
150
225
DNA-PKcs
Ku80
Ku70
12 345
DNA-PKcs
Ku70/Ku80
Mono Q fraction
Wortmannin
Fig. 5. Activity of DNA-PKcs in the aggre-
gates. (A) Ku facilitates DNA-PKcs-mediated
DNA aggregation. Ku (0.15, 0.3, 0.6, 1.2 lg)
was added to the mixture of DNA-PKcs
(1.2 lg) or SAF-A (0.6 lg) with DNA
(0.15 lg) including LD, NC, SSC and SC.
The effect of Ku on DNA-PKcs-DNA
coaggregation is indicated by a bracket
(upper panel). (B) Aggregation is compatible
with the kinase activity of DNA-PKcs. The
combination of DNA-PKcs (1.6 lg) and Ku
(1.0 lg) or Mono Q fraction (1.4 lg) was
coaggregated with linearized pUC18 in the
absence or presence of wortmannin.

Phosphorylation was initiated by the addition
of c-P
32
-labeled ATP to the aggregates, and
the radioactive proteins were resolved on an
SDS ⁄ PAGE. (C) The kinase activity of DNA-
PKcs regulates the aggregation-coupled
EJ. The effect of the addition of purified
DNA-PKcs (0.2, 0.4, 0.8, 1.6 lg) to wort-
mannin-inactivated aggregates was also
tested (lanes 6–9). The aggregate and the
supernatant (sup.) were analyzed separately
(lanes 10 and 11). The conditions in lanes 4,
5 and 9 are parallel to those in lanes 3–5 in
Fig. 5B, respectively.
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3069
inactivation of kinase activity [36,37]. Preincubation
with the inhibitor did not alter aggregation activity in
Mono Q bound fractions (Fig. 5C, compare lanes 2
and 3). However, the resultant aggregates yielded only
a slight amount of EJ product in the presence of
ATP (Fig. 5C, lane 5). The subsequent addition of
purified DNA-PKcs to the inactivated aggregate
restored EJ activity to a considerable degree, depend-
ing on the amount of DNA-PKcs added (Fig. 5C,
lanes 6–9). Here DNA-PK activity in the aggregates
showed a wortmannin-sensitivity and was restored by
exogenous DNA-PKcs (Fig. 5B, lanes 3–5). Since the
aggregate stuck to the bottom of the test tube during

reactions (Fig. 5C, compare lanes 10 and 11), DNA-
PKcs molecules supplied to the supernatant must be
interactive with the aggregate and must be accessible
to its critical sites. Thus, the aggregate has the quality
of an open system under continuous interaction with
the surrounding environment. Taken together, we can
describe the DNA end-directed aggregation as a coac-
ervate-like organization, which serves as an activated
state to enable DNA-PK-dependent end-joining. This
term, coacervate-like organisation, means a molecular
network that has properties of naturally occurring
coacervate but is obtainable through centrifugal mani-
pulation in an in vitro system.
Additionally, when a DNA-PKcs-enriched subfrac-
tion, which was isolated from other aggregative pro-
teins including Ku through a Mono Q column (refer
to Experimental procedures), a significant portion of
EJ activity was co-eluted (supplementary Fig. S3). The
activity was stimulated by aggregation and was wort-
mannin-sensitive in the presence of purified Ku. The
evidence from this reconstruction system emphasizes
that DNA-PKcs and Ku are key elements for the
aggregation-coupled EJ.
Aggregative organization as a large-scale
response to DNA damage
To our surprise, the DNA repair factors DNA-PKcs
and Ku themselves participate in aggregation, sup-
porting the tight link of aggregative organization
with repair processes. Under reconstituted condi-
tions in vitro, nick (single-strand break) and ssDNA

(unpaired form) lesions, which could release the super-
helical torsion of DNA loops are also mediators for
aggregation, and are even predominant substrates for
the aggregative action of DNA-PKcs. Experiments
have shown that cellular protein fractions can target
those lesions with affinity similar to that of DNA ends.
We have also noted that the aggregates prepared from
one type of DNA lesion assemble more readily with
DNA of other types of lesions diffused in external
solution under physiological conditions (unpublished
work). This predisposition implies the possibility that
in vivo, instead of centrifugal force, the pre-existing
aggregates may develop into damage-mediated ones. It
is presumable that such nucleation of aggregates takes
place at AT-rich S ⁄ MAR DNA of an anchor of chro-
matin loops that tend to transform into ssDNA by
unwinding [38] and are potential sites for aggregation
of S ⁄ MAR-binding proteins. In this view, the direct
association between DNA-PK and S ⁄ MAR-DNA or
-binding proteins is a considerable point. An experi-
mental observation seems to be the case that in an
in vitro system, the binding of DNA-PK to DNA ends
is prerequisite for the secondary assembly with
S ⁄ MAR DNA concomitant with its interactive factors
including SAF-A [39]. It should be also noted that
more frequent nucleation may originate from reversible
association between nucleosomal units of chromatin in
equilibrium of ionic conditions [40,41]. Probably, the
extent of the aggregation would be regulated by DNA
damage-responsive chromatin factors, and one of the

candidates may be linker histone H1 of an aggregative
protein. IR provokes ATM-dependent dephosphoryla-
tion of histone H1 [42]. A recent study demonstrates
that whereas histone H1 is an inhibitory factor for EJ
reaction, its phosphorylation by DNA-PK results in
the promotion of EJ even in the presence of nucleo-
somes [43]. We presume that in addition to the nuc-
leosome component H2AX, the change in the
phosphorylated state of histone H1 would regulate the
aggregative nature of the array of nucleosomes.
Concerning DNA-PKcs, the results from the previ-
ous in vitro experiments can be applied to the action
mode according to a simple stoichiometry: one mole-
cule of DNA-PKcs is recruited to each Ku hetero-
dimer-bound DNA end to form an active kinase
complex, and synapsis of the complexes promotes
the juxtaposition between the opposing DNA ends,
prior to their processing and ligation [44]. However,
this scheme cannot fully explain the evidence for the
in vivo modification and localization of DNA-PKcs.
An exposure of cells to IR induces the autophospho-
rylation of many DNA-PKcs molecules at DSB sites
[45], which are coupled with DSB-induced c-H2AX
nuclear foci and are likely to be concomitant with
the accumulation of activated ATM molecules [46].
Such accumulation of the phosphorylated DNA-PKcs
into the foci may be parallel to its quantitative
requirement for in vitro aggregation; the conditional
ratio of DNA-PKcs in aggregates is calculated as
being >10 molecules per a single end of 2.7 kbp lin-

ear DNA. Very likely, the analysis of components of
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3070 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
Mono Q fraction coaggregated with linear DNA
(Fig. 3A, lane 7) indicates that one DNA end
attracts a similarly excessive number of molecules of
nucleoproteins. Our estimation, however, remains ten-
tative until their distribution on individual DNA is
determined.
Interestingly, rodent cells that are defective in Ku80
or DNA-PKcs are incomplete in the repair of chroma-
tin loops containing multiple DSBs, which were
induced by high doses (> 50 Gy) of X-ray or c-ray
[47,48]. DNA-PK components are likely to be involved
in two distinct processes: fast kinetics for single DSB
and a slow one for multiple DSBs in loops. The way
that deals with multiple DSBs may be comparable to
our present study. Presumably, multiple damaged sites
localized within a chromatin region may elicit another
pathway dependent on DNA-PK. Alternatively, an
experiment using a radiation, which was irradiated at
0.5 Gy-equivalent dose per one track, demonstrated an
interesting behavior of multiple DSBs [7]. The intro-
duction of a-particle track onto the cellular nucleus
provokes either rapid mobility or adhesion of the
DSB-induced c-H2AX nuclear foci. This appears to be
representative of the interaction between multiple
DSBs on different chromosomes, which is likely to pre-
dispose to an irregular rejoining. Given high fluidity
and self-organization, one speculation is that the struc-

tural essence of the nuclear foci could be illustrated
by the coacervate-like nature of the DSB-directed
aggregates.
Concluding remarks
The present results demonstrate that the aggregative
organization composed of human nucleoproteins and
aberrant DNA is structurally flexible and biochemi-
cally active, suggesting that the coacervate-like phase
transition from random diffusion to ordered network
could be an architectural strategy in DNA damage-
response. The organization feasible for catalytic repair
reactions has several remarkable characteristics: (i) the
recognition and incorporation of multiple damage
sites in a noncatalytic fashion; (ii) the concentration of
factors diluted in solution; (iii) the acceleration of
reactions; and (iv) the involvement of DNA aggrega-
tion-promoting factors containing S ⁄ MAR-binding
proteins. Practically, the aggregation-based technique
is versatile for extraction of DNA-interactive factors
and for their higher-order assembly, and will contrib-
ute to a thorough understanding of the mechanism for
huge accumulation, high mobility, transient interaction
and activity-regulation of proteins in the nuclear envi-
ronment.
Experimental procedures
Cells
The human lymphoblastoid cells, WI-L2-NS (CRL8155),
were provided from the Health Science Research Resources
Bank (HSRRB), Osaka, Japan (formerly the Japanese
Cancer Research Resources Bank). The cells were cultured

in RPMI1640 medium supplemented with 10% fetal calf
serum, and were incubated at 37 °C in a humidified atmo-
sphere with 5% CO
2
.
DNA substrates
DNA substrates for assays were ss circular, supercoiled,
nicked, and HaeIII-digested linear DNA derived from
uX174 phage (New England BioLabs, Beverly, MA, USA).
The relaxed closed circular form was made by sealing
nicked DNA with T4 DNA ligase (TaKaRa, Shiga, Japan).
The linearized forms of pUC18 plasmid DNA were pre-
pared by digestion with the restriction enzymes EcoRI, PstI
and ScaI (TaKaRa) as alternative substrates. DNA concen-
trations were determined from absorbance at 260 nm.
Cellular fractions and proteins
Nuclear extracts were prepared as described previously with
the following modifications [49]. Cells (4 · 10
9
cells at a
density of 1 · 10
6
cellsÆmL
)1
) were treated by swelling in
three packed volumes of A buffer [10 mm Hepes–KOH,
pH 7.9, 1.5 mm MgCl
2
,10mm KCl, 0.5 mm dithiothreitol,
and 0.5 mm phenylmethanesulfonyl fluoride] on ice for

15 min. The swollen cells were homogenized by 20 strokes
of a B-pestle in a Dounce homogenizer (Wheaton, Millville,
NJ, USA). The nuclei collected by centrifugation at 2500 g
for 10 min were rinsed and suspended in B buffer (20 mm
Hepes–KOH, pH 7.9, 0.5 m NaCl, 25% glycerol, 1.5 mm
MgCl
2
, 0.2 mm EDTA, 0.5 mm dithiothreitol, 0.5 mm phe-
nylmethanesulfonyl fluoride). Following holding on ice for
30 min, the nuclear extract was obtained as the supernatant
after centrifugation at 2500 g.
For chromatographic procedures, the nuclear extract was
filtered through a 0.20 lm pore membrane (Minisart plus
filter; Sartorius, Go
¨
ttingen, Germany) and then applied to
a column with DEAE Sepharose resin (Amersham Bio-
sciences, Piscataway, NJ, USA) in R buffer (20 mm
Tris ⁄ HCl, pH 7.5, 0.5 mm EDTA, 1 mm 2-mercaptoetha-
nol) with 0.5 m NaCl. The flow-through fraction was
referred to as the nuclear fraction. The fraction was diluted
to 0.2 m NaCl in R buffer and loaded onto a column with
denatured DNA cellulose resin (Amersham Biosciences)
that had been pre-equilibrated with 0.2 m NaCl in R buffer.
The proteins adsorbed on the DNA affinity column were
eluted with 0.5 m NaCl in R buffer. For more fraction-
ation, the DNA affinity fraction was diluted to 0.1 m NaCl
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3071
with R buffer and was applied to a Mono Q HR 5 ⁄ 5 col-

umn (Amersham Biosciences). The proteins bound in R
buffer including 0.1 m NaCl were eluted with 0.5 m NaCl
for assays. Alternatively, for purification of proteins, the
Mono Q-bound proteins were developed with a 0.1–1.0 m
NaCl linear gradient in R buffer. This procedure generated
distinct peaks, each of which was individually recovered
and applied to a Mini S PC 3.2 ⁄ 3 column (Amersham Bio-
sciences) in R buffer containing 0.1 m NaCl (Fig. S3). Each
of the adsorbates, including DNA-dependent protein kinase
catalytic subunit (DNA-PKcs), nucleolin, an N-terminal
truncated nucleolin (nucleolinD), high mobility group box
(HMGB) 1 and HMGB2, was eluted from the column with
a 0.1–1.0 m NaCl gradient in R buffer. Also, the Mono S
unbound proteins, including SAF-A and Ku, were recov-
ered in the flow-through fractions. Finally, these proteins
were loaded onto a Mini Q PC 3.2 ⁄ 3 column (Amersham
Biosciences), and were isolated as a single peak fraction by
a 0.1–1.0 m NaCl linear gradient in R buffer. Histone H1,
which exists in the DNA affinity fraction, was separated as
a major component stepwise elution from 0.2 to 0.4 m
NaCl on a tRNA-affinity column, which was prepared by
coupling bovine liver-derived tRNA (Sigma, St Louis, MO,
USA) to HiTrap NHS-activated HP (Amersham Bioscienc-
es). Histone H1 was finally purified by size exclusion chro-
matography on a Superdex 200 PC 3.2 ⁄ 30 column
(Amersham Biosciences) in R buffer with 0.5 m NaCl.
Aliquots of fractions and isolated proteins were stored at
)80 °C. Before use, they were concentrated and buffer-
exchanged on centrifugal filter devices (Microcon YM-10,
Millipore, Billerica, MA, USA), and were then separated

from nonspecific aggregates by centrifugation at 8000 g. The
concentration of proteins was determined by the Nano-
Orange Protein Quantitation kit (Molecular Probes, Eugene,
OR, USA) using bovine serum albumin as a standard.
Protein identification
The N-terminal structures of proteins, including nucleolin,
Ku80, nucleolinD, HMGB1 and HMGB2, were determined
with an Applied Biosystems 492 model sequencer according
to the instructions. For proteins that might be modified in
the N-terminus, including DNA-PKcs, SAF-A, Ku70 and
histone H1, following treatment with sequencing grade
trypsin (Promega, Madison, WI, USA), the fragmental
polypeptides were isolated through reverse-phase column
chromatography (lRPC C2 ⁄ C18 column, Amersham Bio-
sciences) and were identified.
Nucleoprotein–DNA coaggregation
Analysis of large nucleoprotein–DNA complexes was per-
formed basically as described previously [14,15]. Reactions
(20 lL) were carried out in R buffer containing 150 mm or
50 mm NaCl, and 0.05% digitonin (Wako Pure Chemical,
Osaka, Japan). Cellular fractions or isolated proteins were
incubated with DNA substrates in the presence of 10 mm
MgCl
2
for 10 min at 20 °C. The mixtures were then centri-
fuged at 8000 g for 10 min at 4 °C. The supernatants were
removed or recovered, and the precipitable aggregates, after
addition of the reaction buffer (100 lL), were again centri-
fuged. After one more rinse, the aggregates were deprotei-
nized by incubation with 0.5% SDS and 0.5 mgÆmL

)1
proteinase K for 15 min at 65 °C and subsequently for 1 h
at 37 °C for DNA analysis, and were analyzed by 1% ag-
arose gel electrophoresis followed by staining with ethidium
bromide. For protein analysis, the aggregates were heat-
denatured in loading buffer with SDS and were subjected
to 10% SDS–polyacrylamide gel electrophoresis followed
by staining with Coomassie brilliant blue.
DNA end-joining
EJ reactions (20 lL) of linearized pUC18 DNA catalyzed
by cellular fractions or by their resultant induced aggre-
gates were performed in R buffer containing 10 mm MgCl
2
,
1mm ATP, 50 mm NaCl, and 0.05% digitonin at 20 °C for
an hour or the indicated times. After the reaction was
stopped by the addition of 0.5% SDS and 0.4 mgÆmL
)1
proteinase K, products were analyzed by 1% agarose gel
electrophoresis in TAE buffer (40 mm Tris-acetate, pH 7.8,
1mm EDTA). DNA was stained with ethidium bromide.
Visualization and quantification of products were conduc-
ted with an EX molecular imager device (Bio-Rad, Her-
cules, CA, USA).
Effect of a protein kinase inhibitor on EJ
Mono Q bound fractions (1.4 lg) were pretreated with
1 lm of a DNA-PK kinase inhibitor wortmannin (Sigma)
for 10 min at 20 °C. Using the inactivated fractions, the
protein–DNA aggregative complex was prepared by centrif-
ugation as described above. The aggregates were separated

from supernatants and repeatedly rinsed with reaction buf-
fer, resulting in the removal of excess inhibitor. For EJ, the
aggregates were incubated with 1 mm of ATP for 1 h at
20 °C and terminated by incubation with 0.5% SDS and
0.5 mgÆmL
)1
proteinase K. Additionally, after the wort-
mannin-treated aggregates were incubated with increasing
amounts of purified DNA-PKcs for 10 min at 20 °C, EJ
reactions were pursued in the presence of ATP, and its
complementary effect was examined.
DNA-PK activity
Protein phosphorylation was examined under the condi-
tions similar to the EJ reaction. A combination of DNA-
PKcs (1.6 lg) and Ku70 ⁄ Ku80 heterodimer (1.0 lg) or
mono Q bound fractions (1.4 l g) was used as a source of
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3072 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
DNA-PK kinase. After centrifugal aggregation with linear
DNA (0.1 lg), DNA-PK-mediated protein phosphorylari-
on, including its autophosphorylation, were initiated by
the addition of 1 mm ATP containing its c-
32
P-labeled
derivative. Reactions were carried out for 20 min at 20 °C
and stopped by incubation with SDS ⁄ PAGE loading buf-
fer. Also, the inhibitory effect of wortmannin on DNA-PK
activity and its complementation by active DNA-PKcs were
examined under aggregative conditions. The proteins were
separated on a 4–20% SDS ⁄ PAGE and the radioactive

ones were visualized on an imaging plate with BAS 1800II
(Fujifilm, Tokyo, Japan).
Immunoblotting
Proteins were resolved on a 4–12% gradient polyacrylamide
gel, and were transferred onto polyvinylidene difluoride
membrane. The proteins reacted with first antibodies were
labeled by ECL plus western blotting kit (Amersham Bio-
sciences) and were detected on FP-3000B film (Fujifilm).
Antibodies used were as follows; rabbit anti-ligase IV IgG
(Oxford Biotechnology, Kidlington, UK) and rabbit anti-
XRCC4 antibody (Serotec, Kidlington, UK). Recombinant
products of DNA ligase IV and XRCC4 (Trevigen,
Gaithersburg, MD, USA) were used as positive controls.
Acknowledgements
We thank I. Furuno-Fukushi and Y. Hoki for techni-
cal assistance and B. S. Strauss for reading the manu-
script. This work was supported by the Funds for
Cross-over Research on Underlying Technology of
Nuclear Energy and the Special Coordination Funds
for Promoting Science and Technology from the Min-
istry of Education, Culture, Sports, Science and Tech-
nology of Japan.
References
1 Critchlow SE & Jackson SP (1998) DNA end-joining:
from yeast to man. Trends Biochem Sci 23, 394–398.
2 Peterson CL & Cote J (2004) Cellular machineries for
chromosomal DNA repair. Genes Dev 18, 602–616.
3 Rogakou EP, Pilch DR, Orr AH, Ivanova VS & Bonner
WM (1998) DNA double-stranded breaks induce his-
tone H2AX phosphorylation on serine 139. J Biol Chem

273, 5858–5868.
4 Paull TT, Rogakou EP, Yamazaki V, Kirchgessner CU,
Gellert M & Bonner WM (2000) A critical role for his-
tone H2AX in recruitment of repair factors to nuclear
foci after DNA damage. Curr Biol 10, 886–895.
5 Celeste A, Fernandez-Capetillo O, Kruhlak MJ, Pilch
DR, Staudt DW, Lee A, Bonner RF, Bonner WM &
Nussenzweig A (2003) Histone H2AX phosphorylation
is dispensable for the initial recognition of DNA breaks.
Nat Cell Biol 5, 675–679.
6 Sedelnikova OA, Pilch DR, Redon C & Bonner WM
(2003) Histone H2AX in DNA damage and repair.
Cancer Biol Ther 2, 233–235.
7 Aten JA, Stap J, Krawczyk PM, van Oven CH, Hoebe
RA, Essers J & Kanaar R (2004) Dynamics of DNA
double-strand breaks revealed by clustering of damaged
chromosome domains. Science 303, 92–95.
8 Rodgers W, Jordan SJ & Capra JD (2002) Transient
association of Ku with nuclear substrates characterized
using fluorescence photobleaching. J Immunol 168,
2348–2355.
9 Lukas C, Falck J, Bartkova J, Bartek J & Lukas J
(2003) Distinct spatiotemporal dynamics of mammalian
checkpoint regulators induced by DNA damage. Nat
Cell Biol 5, 255–260.
10 Bakkenist CJ & Kastan MB (2003) DNA damage acti-
vates ATM through intermolecular autophosphorylation
and dimer dissociation. Nature 421, 499–506.
11 Roti Roti JL & Wright WD (1987) Visualization of
DNA loops in nucleoids from HeLa cells: assays for

DNA damage and repair. Cytometry 8, 461–467.
12 Malyapa RS, Wright WD & Roti Roti JL (1996) DNA
supercoiling changes and nucleoid protein composition
in a group of L5178Y cells of varying radiosensitivity.
Radiat Res 145, 239–242.
13 Malyapa RS, Wright WD, Taylor YC & Roti Roti JL
(1996) DNA supercoiling changes and nuclear matrix-
associated proteins: possible role in oncogene-mediated
radioresistance. Int J Radiat Oncol Biol Phys 35, 963–973.
14 Izaurralde E, Kas E & Laemmli UK (1989) Highly pre-
ferential nucleation of histone H1 assembly on scaffold-
associated regions. J Mol Biol 210, 573–585.
15 Adachi Y, Kas E & Laemmli UK (1989) Preferential,
cooperative binding of DNA topoisomerase II to scaf-
fold-associated regions. EMBO J 8, 3997–4006.
16 Luderus ME, de Graaf A, Mattia E, den Blaauwen JL,
Grande MA, de Jong L & van Driel R (1992) Binding of
matrix attachment regions to lamin B1. Cell 70, 949–959.
17 Romig H, Fackelmayer FO, Renz A, Ramsperger U &
Richter A (1992) Characterization of SAF-A, a novel
nuclear DNA binding protein from HeLa cells with high
affinity for nuclear matrix ⁄ scaffold attachment DNA
elements. EMBO J 11, 3431–3440.
18 Krasnow MA & Cozzarelli NR (1982) Catenation of
DNA rings by topoisomerases. Mechanism of control
by spermidine. J Biol Chem 257, 2687–2693.
19 Baeza I, Gariglio P, Rangel LM, Chavez P, Cervantes
L, Arguello C, Wong C & Montanez C (1987) Electron
microscopy and biochemical properties of polyamine-
compacted DNA. Biochemistry 26, 6387–6392.

20 Tsang SS, Chow SA & Radding CM (1985) Networks
of DNA and RecA protein are intermediates in homolo-
gous pairing. Biochemistry 24, 3226–3232.
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3073
21 Gonda DK & Radding CM (1986) The mechanism of
the search for homology promoted by recA protein.
Facilitated diffusion within nucleoprotein networks.
J Biol Chem 261, 13087–13096.
22 Pelta J, Livolant F & Sikorav JL (1996) DNA aggrega-
tion induced by polyamines and cobalthexamine. J Biol
Chem 271, 5656–5662.
23 Oparin AI (1965) The origin of life and the origin of
enzymes. Adv Enzymol Relat Areas Mol Biol 27, 347–380.
24 Huang J & Dynan WS (2002) Reconstitution of the
mammalian DNA double-strand break end-joining reac-
tion reveals a requirement for an Mre11 ⁄ Rad50 ⁄ NBS1-
containing fraction. Nucleic Acids Res 30, 667–674.
25 Audebert M, Salles B & Calsou P (2004) Involvement
of poly(ADP-ribose) polymerase-1 and XRCC1 ⁄ DNA
ligase III in an alternative route for DNA double-strand
breaks rejoining. J Biol Chem 279, 55117–55126.
26 Wang H, Rosidi B, Perrault R, Wang M, Zhang L,
Windhofer F & Iliakis G (2005) DNA ligase III as a
candidate component of backup pathways of nonhomo-
logous end joining. Cancer Res 65, 4020–4030.
27 Bloomfield VA (1996) DNA condensation. Curr Opin
Struct Biol 6, 334–341.
28 Erard M, Lakhdar-Ghazal F & Amalric F (1990)
Repeat peptide motifs which contain beta-turns and

modulate DNA condensation in chromatin. Eur J
Biochem 191, 19–26.
29 Erard MS, Belenguer P, Caizergues-Ferrer M, Pantaloni
A & Amalric F (1988) A major nucleolar protein,
nucleolin, induces chromatin decondensation by binding
to histone H1. Eur J Biochem 175, 525–530.
30 Smith GC & Jackson SP (1999) The DNA-dependent
protein kinase. Genes Dev 13, 916–934.
31 Hammarsten O & Chu G (1998) DNA-dependent pro-
tein kinase: DNA binding and activation in the absence
of Ku. Proc Natl Acad Sci USA 95, 525–530.
32 Hammarsten O, DeFazio LG & Chu G (2000) Activa-
tion of DNA-dependent protein kinase by single-
stranded DNA ends. J Biol Chem 275, 1541–1550.
33 Merkle D, Douglas P, Moorhead GB, Leonenko Z, Yu
Y, Cramb D, Bazett-Jones DP & Lees-Miller SP (2002)
The DNA-dependent protein kinase interacts with DNA
to form a protein-DNA complex that is disrupted by
phosphorylation. Biochemistry 41, 12706–12714.
34 Sarkaria JN, Tibbetts RS, Busby EC, Kennedy AP,
Hill DE & Abraham RT (1998) Inhibition of phos-
phoinositide 3-kinase related kinases by the radiosen-
sitizing agent wortmannin. Cancer Res 58, 4375–
4382.
35 Wymann MP, Bulgarelli-Leva G, Zvelebil MJ, Pirola L,
Vanhaesebroeck B, Waterfield MD & Panayotou G
(1996) Wortmannin inactivates phosphoinositide
3-kinase by covalent modification of Lys-802, a residue
involved in the phosphate transfer reaction. Mol Cell
Biol 16, 1722–1733.

36 Baumann P & West SC (1998) DNA end-joining cata-
lyzed by human cell-free extracts. Proc Natl Acad Sci
USA 95, 14066–14070.
37 Calsou P, Frit P, Humbert O, Muller C, Chen DJ &
Salles B (1999) The DNA-dependent protein kinase
catalytic activity regulates DNA end processing by
means of Ku entry into DNA. J Biol Chem 274,
7848–7856.
38 Bode J, Kohwi Y, Dickinson L, Joh T, Klehr D, Mielke
C & Kohwi-Shigematsu T (1992) Biological significance
of unwinding capability of nuclear matrix-associating
DNAs. Science 255, 195–197.
39 Mauldin SK, Getts RC, Liu W & Stamato TD (2002)
DNA-PK-dependent binding of DNA ends to plasmids
containing nuclear matrix attachment region DNA
sequences: evidence for assembly of a repair complex.
Nucleic Acids Res 30, 4075–4087.
40 Schwarz PM & Hansen JC (1994) Formation and stabi-
lity of higher order chromatin structures: contributions
of the histone octamer. J Biol Chem 269, 16284–16289.
41 Schwarz PM, Felthauser A, Fletcher TM & Hansen JC
(1996) Reversible oligonucleosome self-association:
dependence on divalent cations and core histone tail
domains. Biochemistry 35, 4009–4015.
42 Guo CY, Wang Y, Brautigan DL & Larner JM (1999)
Histone H1 dephosphorylation is mediated through a
radiation-induced signal transduction pathway depen-
dent on ATM. J Biol Chem 274, 18715–18720.
43 Kysela B, Chovanec M & Jeggo PA (2005) Phosphory-
lation of linker histones by DNA-dependent protein

kinase is required for DNA ligase IV-dependent ligation
in the presence of histone H1. Proc Natl Acad Sci USA
102, 1877–1882.
44 DeFazio LG, Stansel RM, Griffith JD & Chu G (2002)
Synapsis of DNA ends by DNA-dependent protein
kinase. EMBO J 21, 3192–3200.
45 Chan DW, Chen BP, Prithivirajsingh S, Kurimasa A,
Story MD, Qin J & Chen DJ (2002) Autophosphoryla-
tion of the DNA-dependent protein kinase catalytic sub-
unit is required for rejoining of DNA double-strand
breaks. Genes Dev 16, 2333–2338.
46 Stiff T, O’Driscoll M, Rief N, Iwabuchi K, Lobrich M
& Jeggo PA (2004) ATM and DNA-PK function redun-
dantly to phosphorylate H2AX after exposure to ioniz-
ing radiation. Cancer Res 64, 2390–2396.
47 Johnston PJ & Bryant PE (1994) A component of
DNA double-strand break repair is dependent on the
spatial orientation of the lesions within the higher-
order structures of chromatin. Int J Radiat Biol 66,
531–536.
48 Johnston PJ, MacPhail SH, Stamato TD, Kirchgessner
CU & Olive PL (1998) Higher-order chromatin struc-
ture-dependent repair of DNA double-strand breaks:
involvement of the V(D)J recombination double-strand
break repair pathway. Radiat Res 149, 455–462.
Aggregation-coupled DNA end-joining M. Takahagi and K. Tatsumi
3074 FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS
49 Dignam JD, Lebovitz RM & Roeder RG (1983) Accu-
rate transcription initiation by RNA polymerase II in a
soluble extract from isolated mammalian nuclei. Nucleic

Acids Res 11, 1475–1489.
Supplementary material
The following supplementary material is available
online:
Fig. S1. EJ process occurs within the aggregates. An
aggregative body consisting of the nuclear fractions
and linear DNA was detached from the internal sur-
face of a test tube by repeated pipetting. The EJ activ-
ity in the free aggregates was compared with the
originally fixed one (compare lanes 4 and 5). In the
course of this experiment, irrespective of the presence
of the supernatant from the initial centrifugation, the
yield of EJ products did not appreciably change (com-
pare lanes 1 and 2). The replacement of the superna-
tant with a reaction buffer resulted in a slight increase
in the production of the multimers (compare lanes 2
and 5). Experiments to separate the aggregate and the
supernatant (sup.) after 2 h indicated that both input
substrates and the EJ products remained stable in the
aggregates (compare lanes 2 and 3 and lanes 5 and 6).
Fig. S2. The aggregation-coupled EJ activity is
independent of DNA ligase IV ⁄ XRCC4. Chromato-
graphically separated fractions were analyzed by
immunoblotting using antibodies for DNA ligase IV
and XRCC4. Their antigens were detectable in nuclear
fraction (NF) (lane 2), but not in DNA affinity- and
Mono Q-fractions that EJ activity are enriched (lanes
3 and 4). On the other hand, the ligase components
were found in a Heparin column fraction, which was
separated from a flow-through fraction of the DNA

affinity column (lane 5). Therefore, the EJ activity is
thought to depend on a factor other than DNA ligase
IV ⁄ XRCC4 complex. Recombinant proteins were used
to check the antibodies (lane 1).
Fig. S3. The aggregative EJ activity was enriched along
with DNA-PKcs. (A) The fractions containing major
proteins were separated as distinct peaks through
Mono Q column. (B) Major proteins in fractions were
resolved on 10% SDS ⁄ PAGE. In this panel, the separ-
ation of DNA-PKcs (lane 2) and Ku70 ⁄ Ku80 (lane 3)
was confirmed. (C) EJ activity in Mono Q fractions
was tested. DNA-PKcs-enriched fraction (lane 2;
marked with asterisk) contained the highest rejoining
activity of pUC18-linearlized DNA (lane 2). (D) The
aggregation-coupled EJ in the DNA-PKcs-enriched
fraction was characterized. The DNA-PKcs-fraction
not only coaggregated with linear DNA, but also
showed an aggregation-enhanced EJ. This EJ activity
was strongly inhibited by wortmannin in the presence
of purified Ku70 ⁄ Ku80 proteins, implicating that
DNA-PK and a ligase activity are closely interactive in
aggregates. It should be noted that when the DNA-
PKcs-enriched fraction was further fractionated
through a Mini S column (Experimental procedures), a
major peak fraction composed of DNA-PKcs retained
a significant level of EJ activity, which was aggregative
with DNA-PKcs (data not shown). The additional step
using a Mini Q column allowed isolating DNA-PKcs
from the concomitant EJ activity.
This material is available as part of the online article

from
M. Takahagi and K. Tatsumi Aggregation-coupled DNA end-joining
FEBS Journal 273 (2006) 3063–3075 ª 2006 The Authors Journal compilation ª 2006 FEBS 3075