 |
Introduction |
The DNA-dependent protein kinase (DNA-PK) is a
nuclear serine/threonine protein kinase that is activated
upon association with DNA. Biochemical and genetic data have revealed
DNA-PK to be composed of a large catalytic subunit, termed DNA-PKcs,
and a regulatory factor termed Ku. In recent years, mammalian DNA-PK has been shown to be a crucial component of both the DNA double-strand break (DSB) repair machinery and the V(D)J recombination
apparatus. In addition, recent work has implicated DNA-PK components
in a variety of other processes, including the modulation of chromatin structure and telomere maintenance.
Our DNA is constantly under attack from reactive oxygen
intermediates
by-products of the oxidative metabolism we have evolved for energy supply. Reactive oxygen species are capable of producing DNA
single-strand breaks and, where two of these are generated in close
proximity, DNA double-strand breaks (DSBs). In addition, single- and
double-strand breaks can be induced when a DNA replication fork
encounters a damaged template, and are generated by exogenous agents
such as ionizing radiation (IR) and certain anti-cancer drugs (e.g.,
bleomycin). DSBs also occur as intermediates in site-specific V(D)J recombination, a process that is critical for the
generation of a functional vertebrate immune system. If DNA DSBs are
left unrepaired or are repaired inaccurately, mutations
and/or chromosomal aberrations are induced, which in turn
may lead to cell death or, in extreme cases, cancer. To combat the
serious threats posed by DNA DSBs, eukaryotic cells have evolved
several mechanisms to mediate their repair. In higher eukaryotes, the
predominant of these mechanisms is DNA nonhomologous end-joining
(NHEJ), also known as illegitimate recombination. DNA-PK plays a key
role in this pathway.
Early studies on DNA-PK focused primarily on its biochemistry, and led
researchers to speculate on its function as a modulator of
transcription. However, this viewpoint took a dramatic change when it
was shown that DNA-PK is activated most potently by DNA DSBs,
suggesting that it might play a role in recognizing DNA damage. This
observation stimulated investigations into the potential role of
DNA-PK in DNA repair and led to the identification of cell lines that
are radiosensitive due to mutations in DNA-PK components. At around
the same time, DNA-PKcs was found to be mutated in cells derived from
the radiosensitive and V(D)J recombination deficient severe
combined immune-deficient (SCID) mouse. The subsequent cloning of the
DNA-PKcs cDNA revealed this very large polypeptide to be similar in
sequence, over its kinase domain, to an expanding family of proteins
involved in controlling cell cycle progression and maintaining genomic
stability. Here, we review what is currently known about DNA-PK, with
emphasis on recent biochemical analyses that have begun to shed light
into how DNA-PK and associated proteins function at the molecular
level. We also focus on the results of ablating the function of DNA-PK
subunits in both yeast and mice, which have suggested other
physiological roles for DNA-PK and its components.
| |
DNA-PK: a historical perspective |
A conspicuous property of DNA-PK is that its catalytic activity is
triggered upon association with DNA. The ability of double-stranded DNA
(dsDNA) to stimulate a protein kinase was first observed in mouse
spleen cell nuclear extracts by Ohtsuki et al. (1980)
, although whether
this activity corresponded to DNA-PK is uncertain. Later, a
DNA-activated protein kinase was discovered serendipitously by Walker
et al. (1985) when studying RNA activation of protein phosphorylationan RNA sample contaminated with DNA led to the finding
that rabbit reticulocyte, human cell, Xenopus oocyte, and
several murine and amphibian cell extracts are capable of phosphorylating endogenous proteins in a dsDNA-dependent manner. Independently, a DNA-dependent kinase was discovered in HeLa cell extracts as an activity that was capable of phosphorylating SV40 virus
large T antigen, transcription factor Sp1, and a variety of other
DNA-binding proteins (Carter et al. 1988; Jackson et al. 1990).
Intriguingly, these two latter studies noted that the enzyme functions
efficiently in the presence of linear but not supercoiled DNA. Through
exchange of materials, the three groups concluded that they had
identified the same kinase, henceforth termed DNA-PK (Carter et al.
1990; Jackson et al. 1990; Lees-Miller et al. 1990). Initial work
indicated that DNA-PK activity copurified with a ~350-kD
polypeptide, originally termed `p350'. Lees-Miller et al. (1990) did
note, however, that this protein did not always copurify precisely with
DNA-PK catalytic activity and found that a high resolution
chromatographic step resulted in a p350 preparation dramatically
reduced in catalytic activity. This suggested either that p350 did not
correspond to DNA-PK or that p350 required an additional
polypeptide(s) to function. The latter possibility was shown to be
correct through subsequent biochemical studies, which revealed that
DNA-PK comprises a large catalytic subunit (p350; now termed
DNA-PKcs) and a DNA-targeting component corresponding to the
nonspecific DNA end-binding protein Ku (Dvir et al. 1992; Gottlieb and
Jackson 1993).
| |
DNA-PKcs is a member of the PI 3-kinase family |
Cloning of the DNA-PKcs cDNA revealed that it corresponds to a
~470-kD polypeptide, the amino-terminal ~3500 amino acid residues of which does not appear to have significant homology to other characterized proteins (Hartley et al. 1995). More significantly, however, the carboxy-terminal ~500 residues of DNA-PKcs comprises a
catalytic domain that falls into the phosphatidylinositol 3 (PI
3)-kinase family (Hartley et al. 1995; Poltoratsky et al. 1995).
Although this suggested initially that DNA-PK might be capable of
phosphorylating inositol phospholipids, like certain well-characterized
members of the PI 3-kinase family (Toker and Cantley 1997), the
available evidence indicates that DNA-PK has protein but not lipid
kinase activity (Hartley et al. 1995; Smith et al. 1999). At a similar
time to the cloning of the DNA-PKcs cDNA, the genes and cDNAs for a
range of other large PI 3-kinase-like (PIKL) proteins were identified
and cloned (for reviews, see Zakian 1995; Jackson 1996). These proteins
have been shown to be involved in controlling transcription, the cell
cycle and/or genome stability in organisms ranging from
yeast to man. DNA-PKcs, however, appears to be restricted to higher
eukaryotes; clear homologs have been identified in mouse (Araki et al.
1997), horse (Shin et al. 1997), and Xenopus laevis (Labhart
1997), but is not present in the genome of Saccharomyces
cerevisiae and has not been identified in the genomic sequences
available for Caenorhabditis elegans.
Besides DNA-PKcs, probably the best characterized member of the PIKL
family is ATM, the protein deficient in the human neurodegenerative and
cancer predisposition condition ataxia-telangiectasia (A-T; for review
see Lavin and Shiloh 1997). ATM has been linked intimately to the
detection and signaling of DNA damage (for review, see Rotman and
Shiloh 1998). ATM homologs also exist in S. cerevisiae (Tel1p;
Greenwell et al. 1995; Morrow et al. 1995) and Schizosaccharomyces pombe (Naito, et al. 1998) and are involved in genome surveillance and in controlling telomeric function. Another PIKL protein involved in
genome surveillance is human AT-related
(ATR)/FRAP-related (FRP) (Cimprich et al. 1996), together
with its homologs in S. cerevisiae (Mec1p; Kato and Ogawa
1994; Weinert et al. 1994), S. pombe (Rad3; Jimenez et al.
1992; Seaton et al. 1992), and Drosophila (mei-41;
Hari et al. 1995). Human FRAP (Kunz et al. 1993; Brown et al. 1994;
Chiu et al. 1994; Sabatini et al. 1994) and its S. cerevisiae
homologs Tor1p and Tor2p (Heitman et al. 1991; Helliwell et al. 1994),
are also in the PIKL family. These proteins control mRNA translation in
response to nutrient supply and certain other environmental stimuli,
including polypeptide growth factors (Brunn et al. 1997). As with
DNA-PKcs, the available data support the proposal that the above PIKL
family members are serine/threonine protein kinases but
not lipid kinases. A final, apparently more distantly related, member
of the PIKL family has been identified recently, and has been named
Tra1p in S. cerevisiae and TRRAP in humans (McMahon et al.
1998; Saleh et al. 1998). This ~450 kD polypeptide is associated
with the SAGA histone acetyltransferase complex that functions in
transcriptional control (Apone et al. 1998; Grant et al. 1998;
Natarajan et al. 1998). Intriguingly, TRRAP/Tra1p (and
its homolog in C. elegans) appears to have lost its ability to
function as a kinase and most probably acts as a scaffold for proteins
involved in chromatin remodeling.
Outside the kinase domain, DNA-PKcs has little or no similarity with
other proteins and, besides the presence of a putative leucine zipper
motif, which is required for interactions with the high-affinity DNA
binding protein C1D (Yavuzer et al. 1998), it has no clear features
that might hint at its molecular functions. Nevertheless, the
carboxy-terminal quarter of DNA-PKcs has been reported to interact
with Ku (Jin et al. 1997). Last year, a significant step forward in
understanding the architecture of DNA-PKcs was brought about by
cryoelectron microscopy imaging of DNA-PKcs to a resolution of ~20
Å (Chiu et al. 1998). This suggested that DNA-PKcs has an open,
pseudo two-fold symmetric structure with a gap separating a
crown-shaped top from a rounded bottom. The hollow nature of the
DNA-PKcs interior suggests it may interact with DNA via
internalization of double-stranded ends through tunnels and cavities
identified in the structure. Although the size of the interior of the
protein is large, it is thought not big enough to simultaneously
internalize the Ku heterodimer and DNA. Therefore, the Ku binding
site(s) is most probably located on an exterior surface. Another more
recent study has reported the structure of DNA-PKcs at 22 Å resolution by electron crystallography (Leuther et al. 1999). This
structure shows DNA-PKcs to be similar to other dsDNA-binding
proteins, with it possessing an open channel and an enclosed cavity
with three openings large enough to accommodate ssDNA. Biochemical
analyses based on this knowledge have suggested that activation of the
kinase requires interactions with both double- and single-stranded DNA
(Leuther et al. 1999). It will be of great interest to conduct more
detailed structural determinations of DNA-PKcs and the
DNA-PKcs/Ku complex to learn more about their molecular architectures.
| |
Analyses of the Ku subunits |
Ku was first identified as an autoimmune antigen in patients with
scleroderma-polymyositis overlap syndrome (Mimori et al. 1981). The
initial and detailed characterization of this protein revealed it to be
a highly abundant nonspecific DNA-binding protein comprising two
tightly-associated subunits of ~70 and 83 kD (Ku70 and Ku80,
respectively; Ku80 is sometimes referred to as Ku86). The cloning of
cDNAs and genes for Ku subunits from a variety of species has now taken
place and has revealed that both Ku70 and Ku80 exist in organisms
ranging from yeast to man (for review, see Dynan and Yoo 1998).
Although sequence analyses have so far told us relatively little about
the structural organization of the Ku heterodimer, it has recently
become apparent that homology exists between Ku70 and Ku80, suggesting
that they arose through the duplication and subsequent divergence of a
single polypeptide, which presumably functioned as a homodimer (Dynan
and Yoo 1998; D. Gell and S.P. Jackson, unpubl.). Each Ku subunit has
been reported to possess a leucine-zipper motif, but whether or not
these have roles in the overall function or conformation of the
proteins is still open to question (Wu and Lieber 1996). Also
identified in the sequences of both Ku70 and Ku80 are weak ATP binding
site homologies, which may be essential for the proposed ATPase and helicase functions of Ku (Cao et al. 1994; Tuteja et al. 1994; see
below). However, the significance of the ATP-binding motifs is unclear,
as mutation of these so far have not been found to affect Ku function
in vivo (Jin and Weaver 1997; Singleton et al. 1997). Numerous yeast
two-hybrid and biochemical studies have been conducted to map the
regions of the Ku subunits that make contact with one another and with
DNA (Wu and Lieber 1996; Jin and Weaver 1997; Osipovich et al. 1997;
Cary et al. 1998; Wang et al. 1998). Although both conflicting and
corroborating results have arisen from these studies, it seems apparent
that a carboxy-terminal region of ~150 amino acid residues in both
Ku70 and Ku80 are essential for dimerization, and that larger regions
of both proteins are required for effective interactions with DNA ends.
Still to be identified, however, are the precise residues that mediate
the extremely strong Ku70-Ku80 interaction, interactions with DNA, and
the weaker but critically important interaction with DNA-PKcs.
| |
DNA-binding properties of Ku |
The Ku heterodimer was first described as having the apparently
unique ability to bind with high selectivity to free dsDNA ends (Mimori
and Hardin 1986). The early studies on DNA binding by Ku revealed that
it takes place with high affinity and that end binding is independent
of the structure or sequence of the end (Mimori and Hardin 1986;
Paillard and Strauss 1991; Falzon et al. 1993). Studies have also shown
that Ku can bind to a variety of DNA structures; however, it does not
bind to closed circular DNA (Dynan and Yoo 1998). Both Ku70 and Ku80
make contact with DNA, although it appears that Ku70 makes the more
intimate interactions, with a carboxy-terminal stretch of 73-amino acid
residues in Ku70 having been shown to make contact with DNA by
Southwestern blot analysis (see Dynan and Woo 1998). Importantly, and
consistent with genetic data indicating that Ku70 and Ku80 are
functionally dependent on each other (see later section), neither
subunit alone can bind DNA effectively (Griffith et al. 1992; Ono et
al. 1994; Wu and Lieber 1996; Ochem et al. 1997). Ku has been reported
on many occassions to bind DNA in a sequence-specific manner. It now
appears that in most cases this was artifactual and resulted from the
high abundance of Ku and its unique DNA end-binding properties (see
Dynan and Woo 1998 for an illuminating summary of this subject). Nevertheless, at least for the case of Ku binding to the NRE-1 element
in the long terminal repeat of the mouse mammary tumor virus, it
appears that this example of sequence specific binding by Ku does have
some functional basis (Giffin et al. 1996).
In addition to being able to interact specifically with DNA ends, Ku
also has the ability to translocate along DNA molecules in an
ATP-independent manner (DeVries et al. 1989; Paillard and Strauss
1991). This and related properties have led to models in which the
binding of Ku to DNA is likened to the sliding of a bead on a piece of
string-DNA ends are not required for binding per se but are required
for Ku to bind to or dissociate from the DNA template. Consistent with
such a model, Ku can generate footprints at internal sites as well as
the termini of linear DNA molecules (Mimori and Hardin 1986; DeVries et
al. 1989). In addition, atomic force microscopy studies have revealed
the existence of internal as well as DNA end-bound DNA-PK complexes,
and have shown that Ku can juxtapose two DNA ends via a DNA looping
mechanism (Cary et al. 1997). Finally in this regard, electrophoretic
mobility shift assays have been employed to show that Ku can actually
transit directly from one linear DNA molecule to another if the termini of the two DNAs are capable of base pairing (Bliss and Lane 1997). That
the above DNA end-alignment activities may be of functional significance is suggested by the fact that Ku can stimulate DNA end
ligation by eukaryotic DNA ligases in vitro (Ramsden and Gellert 1998).
Other, possibly interrelated functions that have been reported for Ku
are helicase and ATPase activities. Tuteja et al. (1994) have shown
that Ku preparations possess weakly processive helicase function.
Furthermore, it has been reported that the Ku70 subunit contains the
ATPase activity and is able to perform helicase function independently
of Ku80 (Ochem et al. 1997). Cao et al. (1994) have also reported that
Ku possesses weak (as compared to other known enzymes) but significant
ATPase function and have shown that this is stimulated by
DNA-PK-mediated phosphorylation. Perhaps relevant to the above, Ku70
and Ku80 do have some homology with classical Walker box ATP-binding
motifs, although, as noted previously, mutation of these sites has not
been found to impair Ku function in vivo.
| |
Mechanism of DNA-PK activation |
The original papers establishing Ku as a DNA-PK component revealed
that it dramatically stimulates DNA-PKcs kinase function towards a
variety of DNA-bound targets (Dvir et al. 1992, 1993; Gottlieb and
Jackson 1993). The DNA-PKcs/Ku complex has been shown to
phosphorylate proteins most effectively when it is bound to the same
DNA molecule as DNA-PK itself, indicating that part of the activation
produced by DNA is through the juxtaposition of DNA-PK and its target
(Gottlieb and Jackson 1993). However, DNA also stimulates the ability
of DNA-PK to phosphorylate non-DNA-binding peptide substrates,
implying that binding to DNA must directly or indirectly induce an
activating conformational change in DNA-PKcs. Recent studies with ATP
noncompetitive inhibitors of DNA-PK indicate that such a
conformational change is unlikely to correspond to an unmasking of the
ATP-binding site (Izzard et al. 1999). Protein-DNA cross-linking
studies have revealed that DNA-PKcs makes intimate contacts with the
DNA, suggesting that DNA might directly induce conformational
alterations (Lees-Miller et al. 1990; Gottlieb and Jackson 1993).
Consistent with this idea, DNA-PKcs activity can be stimulated to some
degree in vitro by DNA in the absence of Ku (Yaneva et al. 1997;
Hammarsten and Chu 1998; West et al. 1998). Taken together with data
revealing that Ku helps to target DNA-PKcs to DNA (Dvir et al. 1992;
Gottlieb and Jackson 1993; Suwa et al. 1994; Chan et al. 1996; Cary et
al. 1997; Yaneva et al. 1997; Hammarsten and Chu 1998; West et al.
1998), the available data lead to a model in which Ku recruits
DNA-PKcs to DNA, which in turn facilitates DNA-PKcs-DNA
interactions that release the catalytic potential of the DNA-PK
complex. Such a model is also consistent with physiological data
revealing that Ku and DNA-PKcs function, at least in a large part, as
an interdependent two-component system (see below). Clearly, many
questions regarding the mechanism of DNA-PK activation remain to be
answered, including how DNA-PKcs and Ku bind to one another and to
DNA, and perhaps most importantly, how only very specific types of DNA
structure mediate DNA-PK activation.
| |
Regulation of DNA-PK |
As well as DNA-PK being modulated through its interactions with Ku
and DNA, it has become apparent that DNA-PK activity and function are
likely to also be regulated by a variety of other mechanisms. One area
that has received particular attention has been the ability of other
proteins to influence the activity of DNA-PKcs and/or
the DNA-PKcs/Ku complex. For example, the high-affinity DNA binding protein C1D was identified through the yeast two-hybrid approach as interacting with the putative leucine zipper region of
DNA-PKcs (Yavuzer et al. 1998). Notably, DNA-bound C1D can trigger
DNA-PK activation in a DNA end-independent manner, possibly via
inducing alterations in the structure of the DNA double helix. Although
the physiological function for C1D is not yet clear, the observations
that it behaves as a component of the nuclear matrix and is induced in
response to IR suggests that it could play a role in targeting DNA-PK
to specific nuclear regions in response to genotoxic insult. Such a
model would be consistent with reports that Ku80-deficient
xrs5 cells are reported to have nuclear envelope and nuclear
matrix alterations compared to their wild-type controls (Yasui et al.
1991; Korte and Yasui 1993). Perhaps related to the activation of
DNA-PK by C1D, high mobility group (HMG) proteins 1 and 2 have also
been shown to be capable of stimulating DNA-PK activation in vitro,
hinting to the possibility that DNA-PK activation in vivo is
influenced by chromatin context (Watanabe et al. 1994; Yumoto et al.
1998). Another protein-protein interaction implicated in regulating
DNA-PK function is that between DNA-PKcs and the Lyn tyrosine kinase,
which is capable of disrupting the DNA-PKcs/ Ku complex
in vitro (Kumar et al. 1998). Heat shock transcription factor 1 (HSF1)
can also stimulate DNA-PK activity in vitro through a mechanism that
involves interactions between HSF1 and Ku and weaker interactions
between HSF1 and DNA-PKcs (Peterson et al. 1995a; Huang et al. 1997).
These findings suggest that HSF1 could cooperate with Ku and DNA-PKcs,
possibly through stabilizing interactions between the DNA-PK
holoenzyme and DNA. Although the physiological relevance of this
interaction is currently unclear, it will certainly be of interest to
determine whether DNA-PK-deficient animals or cells display
hypersensitivity to heat or altered heat shock responses.
Another example of a protein-protein interaction modulating DNA-PK
activity has been revealed through studies on the human Ku80
autoantigen-related protein (KARP-1; Myung et al. 1997, 1998). This
protein, which apparently is only present in primates, is expressed
from the Ku80 locus and corresponds to a 9-kD amino-terminally extended
derivative of Ku80. The KARP-1-specific domain encodes heptad repeats
of leucine residues flanked by a basic region, and expression of
dominant-negative derivatives of KARP-1 in cells has been reported to
result in a derepression of DNA-PK activity and hypersensitivity to
IR. These observations, together with the observation that KARP-1 is
induced upon irradiation of cells in a p53- and ATM-dependent manner
(Myung et al. 1998), suggests that KARP-1 might be involved in latter
stages of DNA DSB repair, or may be required to repair lesions that are
refractory to the actions of the DNA-PK complex containing the
constitutively-expressed form of Ku.
Consistent with its proposed role as a primary DNA damage sensor and
not as an inducible downstream effector of DNA damage signaling,
DNA-PK is present at relatively high levels (up to 1% of HeLa cell
nuclear protein) and its levels do not appear to be regulated strongly
by DNA-damaging agents (Lee et al. 1997). However, rodents have much
lower levels of DNA-PKcs, Ku, and DNA-PK activity than primate cells
(Anderson and Lees-Miller 1992; Finnie et al. 1995). Indeed, it is
notable that DNA-PK levels in a variety of species correlate well with
the species' life-span, suggesting that elevated DNA-PK levels in
longer-lived organisms is a mechanism to enhance genomic stability.
Despite DNA-PK being constitutively present, its activity does appear
to be modulated throughout the cell cycle (Lee et al. 1997).
Furthermore, elevated levels or activity of Ku have been reported in
response to cellular growth state (Cai et al. 1994) possibly reflecting
the increased need for efficient DNA DSB repair in rapidly dividing
cells. DNA-PK activity is found to be induced to some extent when
lymphoid cells are induced to undergo site-specific V(D)J
recombination or switch recombination (Grawunder et al. 1996); two
processes that rely on DNA-PK function (e.g., see Rolink et al. 1996;
Casellas et al. 1998; Manis et al. 1998). However, as this result was
obtained from cells in culture, it remains unclear if this activation
of DNA-PK occurs in the whole animal. In addition, decreased vitamin D
receptor expression has been found to reduce DNA-PKcs mRNA levels, although the reason for this is currently obscure (Dabrowski et al. 1998).
The recent cloning and analysis of the 5'-untranslated region of
the DNA-PKcs gene has revealed it to have no TATA or CCAAT box
sequences and to contain potential binding sites for the ubiquitous transcription factor Sp1features associated with `housekeeping genes' (Connelly et al. 1998). It will be of interest to study the
potential transcriptional induction of DNA-PK components in response
to the aforementioned stimuli. Another potential mode for DNA-PK
regulation, suggested from the cloning of its cDNA, is at the level of
pre-mRNA splicing. The initial DNA-PKcs cDNA sequence lacked an exon
of 93 base pairs that was found in subsequent studies (Hartley et al.
1995; Poltoratsky et al. 1995; Connelly et al. 1996). Although it is
possible that this reflects aberrant rather than differential splicing
in the initial clone, the fact that the differentially spliced exon
encodes a part of the catalytic domain suggests that its differential
usage could have an important regulatory function.
DNA-PK activity also appears to be regulated by post-translational
modification. For example, autophosphorylation of DNA-PKcs in vitro
has been shown to induce its dissociation from Ku and result in
inhibition of DNA-PK catalytic function (Chan and Lees-Miller 1996).
Another DNA-PKcs-phosphorylation event that has been reported to
dissociate the DNA-PKcs-Ku complex is that mediated by the c-Abl
proto-oncogene product (Jin et al. 1997; Kharbanda et al. 1997).
Because c-Abl is itself induced in response to IR in a manner that is
reported to be DNA-PK dependent (Kharbanda et al. 1997), this suggests
the existence of an autoregulatory negative feed-back loop that might
lead to repression of DNA-PK activity after the appropriate DNA damage
signaling and/or repair pathways have been initiated.
Another post-translational modification that has been found recently to
affect DNA-PK activity in vitro is ADP-ribosylation mediated by the
DNA repair-associated enzyme poly-ADP(ribosyl) transferase (PARP),
which can stimulate the ability of DNA-PK to phosphorylate some
protein substrates (Ruscetti et al. 1998). Although the physiological
relevance of this is currently unclear, the fact that PARP activity is
also induced by DNA damage suggests that it may be a way of
coordinating the activities of the DNA-PK and PARP systems.
Perhaps the most striking characterized examples of regulating DNA-PK
activity are its inactivation during programmed cell death and upon
viral infection. During apoptosis, DNA-PKcs is specifically
cleaved by caspase-3 or a caspase-3-like protease with
subsequent loss of its kinase potential (Casciolarosen et al. 1995; Han
et al. 1996; LeRomancer et al. 1996; Song et al. 1996). This DNA-PK
inactivation is probably to prevent signaling from or repair of the
degraded genomic DNA that is produced during the latter steps of the
apoptotic pathway. Furthermore, as apoptosis is an energy-dependent
process, DNA-PK inactivation might prevent the massive activation of
this highly abundant kinase by fragmented genomic DNA and the
subsequent depletion of ATP reserves. Interestingly, although neither
Ku subunit is a target for the apoptotic proteases, a loss of Ku
protein has been observed in apoptotic lymphocytes and in myeloid cells
destined to undergo apoptosis (Ajmani et al. 1995). A negative
regulation of DNA-PK function by protein kinase C 
in apoptotic
cells has also been proposed (Bharti et al. 1998). In a different
scenario, DNA-PKcs has been found to be degraded, apparently via a
proteosomal mechanism, during herpes simplex virus (HSV) type 1 infection of mammalian cells (Lees-Miller et al. 1996; Parkinson et al.
1999). Although other possibilities exist, this suggests that
inhibiting DNA-PK function aids virus replication and/or
packaging. In line with this idea, at low titers, HSV has been reported
to replicate more efficiently in a DNA-PKcs null cell line compared to
the DNA-PK positive cells (Parkinson et al. 1999).
| |
Manipulation of DNA-PK activity by exogenous agents |
Ku has the ability to bind to artificially derived RNA molecules
with an affinity as great as that observed for dsDNA (Yoo and Dynan
1998). Binding to these RNAs does not result in DNA-PK activation,
meaning that they have the potential to be developed into agents that
can regulate DNA-PK function in vitro or in vivo. The sequence
similarity between DNA-PKcs and PI 3-kinases has also provided a route
to identify DNA-PK inhibitors. Thus, Wortmannin, a classical PI
3-kinase inhibitor, was found to inhibit DNA-PK with an IC50 of
~250 nM, around two orders of magnitude higher than is
required for PI 3-kinase inhibition (Hartley et al. 1995). A more
detailed analysis has shown that, similar to PI 3-kinases, DNA-PK is
inhibited by Wortmannin in a noncompetitive manner, and that Wortmannin
is able to bind covalently to the kinase active site (Izzard et al.
1999). Wortmannin is capable of binding to DNA-PKcs in vivo (Izzard et
al. 1999) and has been shown to radiosensitize mammalian cells (Price
and Youmell 1996; Boulton et al. 1996; Rosenzweig et al. 1997; Hosoi et
al. 1998). Other PI 3-kinase inhibitors, such as LY294002, quercitin,
quercitrin, and rutin, have also been shown to inhibit DNA-PK
activity, with LY294002 being demonstrated to radiosensitize cells
(Rosenzweig et al. 1997; Izzard et al. 1999). Another compound that has
potential as a DNA-PK inhibitor is OK1035 (Take et al. 1995, 1996).
Although somewhat nonspecific in nature, these compounds represent
useful tools for elucidating the functions of DNA-PK in DNA repair and other processes, both in vivo (see above references) and in vitro (Gu
et al. 1996, 1998; Baumann and West 1998). In addition, they might
serve as starting points for the identification of more specific
inhibitors of DNA-PK and/or the other related PI
3-kinases for academic research or for the development of
pharmacologically active therapeutic agents.
| |
Role of DNA-PK and associated factors in DNA NHEJ |
The involvement of DNA-PK in DSB repair became evident from
analyses involving a specific series of mutant rodent cell lines. Early
studies into these cells found them to be hypersensitive to IR and
radiomimetic agents with little or no cross sensitivity to other types
of DNA damaging agent, and showed them to be defective in the repair of
chromosomal DNA DSBs (for review, see Zdzienicka 1995). Subsequent cell
fusion studies allowed these cells to be placed into three distinct
complementation groups, termed IR4, IR5, and IR7, and the human genes
complementing them were preassigned to the XRCC
nomenclature (X-ray
cross-complementing; XRCC4 is the
gene that complements cells of IR4, XRCC5 complements IR5, and
XRCC7 complements IR7; Thompson and Jeggo 1995; Zdzienicka 1995).
In late 1994 and early 1995, a series of reports were published showing
that cells of IR5 lack Ku DNA end-binding activity and can be
complemented by the gene for Ku80 (Getts and Stamato 1994; Rathmell and
Chu 1994; Smider et al. 1994; Taccioli et al. 1994; Boubnov et al.
1995; Finnie et al. 1995). Prompted by the above findings, several
groups then established that cells of IR7 also lack DNA-PK activity;
in this case the defect being complemented by the gene for DNA-PKcs
(Blunt et al. 1995; Kirchgessner et al. 1995; Peterson et al. 1995b).
Consistent with this, a radiosensitive human cell line (MO59J) was
found to be defective in DNA-PKcs expression and DNA-PK activity
(Lees-Miller et al. 1995). Subsequent work showed cells of IR5 and IR7
to harbor inactivating mutations in the genes for Ku80 and DNA-PKcs,
respectively, and revealed that inactivation of Ku80 leads to a
dramatic destabilization of both itself and Ku70 (Errami et al. 1996,
1998a,b; Blunt et al. 1996; Danska et al. 1996; Araki et al. 1997;
Singleton et al. 1997; Peterson et al. 1997; Fukumura et al. 1998;
Priestley et al. 1998). It was therefore concluded that mutations in
Ku80 or DNA-PKcs lead to IR hypersensitivity, that XRCC5 and
XRCC7 encode Ku80 and DNA-PKcs, respectively, and that
DNA-PK is a crucial component of the mammalian DNA DSB repair
apparatus. None of the original rodent cell lines was defective in
Ku70, but it was assumed that cells lacking it would have a similar
phenotype to those in IR4-7. This has been confirmed by targeted
disruption of the gene for Ku70 in mouse cells, allowing such cells to
be designated IR6 and the gene for Ku70 to be designated XRCC6
(Gu et al. 1997). In other work, the XRCC4 gene product was
cloned, and IR4 cells were shown to be deleted for this gene (Li et al.
1995). Although the sequence of the 334-amino acid XRCC4 protein did
not initially yield insights into its mode of action, subsequent
studies revealed that it forms a tight and specific association with
DNA ligase IV (Critchlow et al. 1997; Grawunder et al. 1997), a protein
which has itself been revealed recently to function in DNA NHEJ (Barnes et al. 1998; Frank et al. 1998; Grawunder et al. 1998)
Under most circumstances, the predominant mechanism for DNA DSB repair
in mammalian cells is that of NHEJ, and it is in this process that the
IR4-7 mutants are defective. Characteristically, NHEJ does not need
extensive homologies between the recombining DNA molecules nor does it
require an undamaged DNA partner; it is therefore distinct from the
well-characterized homologous recombination pathway of DSB repair (for
reviews, see Chu 1997; Critchlow and Jackson 1998; Kanaar et al. 1998).
In contrast to the situation in mammalian cells, S. cerevisiae
mainly repairs DSBs by homologous recombinationthis requires genes in
the RAD52 epistasis group (Kanaar et al. 1998). Nevertheless,
recent work has shown yeast to have a NHEJ pathway that is highly
related to that in higher eukaryotes. Thus, yeast possesses homologs of
Ku70 and Ku80 (Yku70p and Yku80p, respectively; also termed Hdf1p and
Hdf2p, respectively) that exist in a heterodimeric complex.
Furthermore, loss of either yeast Ku subunit leads to IR
hypersenstivity in rad52 mutant backgrounds and causes NHEJ
defects as ascertained by an in vivo plasmid repair assay (Boulton and
Jackson 1996a,b; Feldmann et al. 1996; Mages et al. 1996; Milne et al.
1996; Siede et al. 1996). Yeast also contains homologs of XRCC4 and DNA
ligase IV (Lif1p and Lig4p or Dnl4p, respectively), and these have been
shown to function in the Ku pathway of NHEJ (Schar et al. 1997; Teo and
Jackson 1997; Wilson et al. 1997; Herrmann et al. 1998; Ramos et
al. 1998). No clear homolog of DNA-PKcs exists in the S. cerevisiae genome however, suggesting either that its functions are
not conserved or are mediated by other proteins.
Recently, several other yeast NHEJ components have been identified,
including the nuclease complex containing Rad50p, Mre11p, and Xrs2p
(Milne et al. 1996; Boulton and Jackson 1998). Human homologs of Rad50p
and Mre11p have also been identified and have been linked to DNA DSB
repair by way of the fact that they become targeted to sites of
IR-induced damage in vivo (Maser et al. 1997; Nelms et al. 1998).
Moreover, it has been established that human Rad50 and Mre11 exist in a
complex with NBS1 (also called Nibrin), a protein whose deficiency
leads to the rare human genetic disorder called Nijmegen breakage
syndrome (NBS) (Carney et al. 1998; Matsuura et al. 1998b; Varon et al.
1998). NBS is characterized by chromosomal instability, developmental
abnormalities, and cancer predisposition (for review, see Featherstone
and Jackson 1998). In addition, defects in NBS1 are reported to lead to
impaired induction of p53 in response to IR, thus providing an exciting
potential linkage between DNA DSB rejoining and DNA damage signaling
(Jongmans et al. 1997; Matsuura et al. 1998b). Three other yeast
proteins shown to be required for efficient NHEJ are the
heterochromatin components Sir2p, Sir3p, and Sir4p (Tsukamoto et al.
1997; Boulton and Jackson 1998). An interaction between yKu70p and
Sir4p in the yeast two-hybrid system (Tsukamoto et al. 1997), has
suggested that Ku recruits the Sir protein complex to DNA DSBs in vivo
and that this might contribute to NHEJ, either by preventing access to
nucleases or by facilitating the juxtaposition of the two DNA termini
via chromatin condensation. Although such models are attractive, recent
work from Rine and colleagues has cast doubt on a direct involvement of
the Sir proteins in NHEJ and has suggested that their effects are, at
least in part, mediated by influencing the mating type status of the
yeast cell (Åström et al. 1999).
There are a variety of other ways in which Ku and DNA-PKcs might
function in DNA NHEJ (for more extensive discussions, see Chu 1997;
Critchlow and Jackson 1998; Kanaar et al. 1998). Most obviously
perhaps, the fact that Ku can bind with great avidity to DNA ends
suggests that it directly recognizes DNA DSBs in vivo and, in higher
eukaryotes, recruits DNA-PKcs to such sites. Once bound, Ku or the
DNA-PK complex might then protect the ends from nucleolytic
degradation, as suggested by the ability of Ku-containing complexes to
exclude nucleases in vitro (Mimori and Hardin 1986; DeVries et al.
1989; Gottlieb and Jackson 1993). Consistent with this idea, linear DNA
transfected into IR5 cells is more susceptible to end degradation than
in control cells (Liang and Jasin 1996), and the rare NHEJ products
that are generated in mammalian or yeast cells lacking Ku have
generally suffered large deletions of terminal sequences before
ligation has taken place (for example, see Taccioli et al. 1993;
Boulton and Jackson 1996b). Nevertheless, and as discussed later,
V(D)J recombination intermediates are relatively stable in the
absence of Ku or DNA-PKcs, indicating that in this case at least, Ku
or DNA-PKcs are not required for DNA end stabilization (Zhu et al.
1996). Another way that Ku/DNA-PKcs could potentiate end
ligation is by tethering two DNA ends together. Indeed, Ku is able to
promote interactions between two DNA termini (Cary et al. 1997) and can
enhance end ligation by eukaryotic DNA ligases in vitro (Ramsden and
Gellert 1998). Notably, in situations in which NHEJ cannot occur simply
by the direct ligation of mutually complementary 5'- or
3'-overhanging termini, repair products tend to suffer short
deletions and become rejoined at sites of short direct repeats of
microhomology. It is tempting to speculate that the weak helicase
functions of Ku could play a role in dissociating the two strands of
the DNA ends to allow such microhomology alignments to be produced,
although it should again be noted that no clear phenotype has yet been
described for Ku proteins mutated in the proposed helicase motifs (Jin
and Weaver 1997; Singleton et al. 1997). Furthermore, it is clear that
microhomology directed repair does occur in Ku80 knockout cells (Bogue
et al. 1997). Therefore, if Ku does play a role in this pathway, it
cannot be essential.
Once positioned at the DNA DSB, Ku and DNA-PKcs might then recruit
other NHEJ factors, or, in the case of the DNA-PK holoenzyme, might
regulate the activities of these components by phosphorylation. Suggestive that this may indeed be the case are the observations that
DNA-PK is able to phosphorylate XRCC4 in vitro and that the XRCC4/ligase IV complex can interact directly or
indirectly with DNA-PK in crude nuclear extracts (Critchlow et al.
1997; Leber et al. 1998). Another complex that might be recruited
and/or activated by Ku or DNA-PKcs is that containing
Rad50, Mre11, and NBS1 (Rad50p, Mre11p, and Xrs2p in yeast), whose
nuclease activities may be critical in `tidying up' damaged DNA
termini before they can be ligated together (Furuse et al. 1998; Paull
and Gellert 1998; Trujillo et al. 1998). Recent data have indicated
that the endonuclease function of Mre11p is not needed for NHEJ of
restriction enzyme-generated DNA DSBs (Moreau et al. 1999), but whether
this is also the case for the repair of noncomplementary ends or
IR-induced damage has not been reported. Finally in regard to the
mechanism of Ku and DNA-PKcs action in NHEJ, it is possible that they
act to dissociate repair factors from the DNA after their job is
complete, or help to remove from the DNA other proteins, such as
recombination factors that might block the repair process (Zhu et al.
1996). In this context, it is noteworthy that DNA-PK is capable of
repressing transcription by RNA polymerase I in vitro (Kuhn et al.
1995; Labhart 1995). Also, autophosphorylation leads to the
dissociation of DNA-PKcs from DNA-bound Ku (Chan and Lees-Miller
1996). If such an autophosphorylation were to take place in
trans, it could provide a mechanism for removing DNA-PKcs
only when two DNA ends are brought into close proximity. In addition to
possibly triggering repair-associated functions of Ku, this could
provide an elegant mechanism for coupling repair to inactivation of
DNA-PK-mediated DNA damage signaling (discussed below).
To help us in understanding the molecular details of DNA NHEJ, the
establishment of biochemical systems that accurately reflect this
process will be of importance. A potentially major step towards this
goal is the recent development of a mammalian cell-based system that
appears to be dependent on DNA-PKcs, Ku, and the ligase IV/XRCC4 complex and also requires other,
as-yet-uncharacterized, components (Baumann and West 1998). A recently
developed Xenopus NHEJ system also appears to be dependent on
functional DNA-PK catalytic activity (Gu et al. 1996; 1998). In
addition to allowing the identification and characterization of NHEJ
components, such systems may be of great use in studying the control of
NHEJ during the cell cycle and in analyzing interfaces between NHEJ and
other processes, such as transcription and the control of chromatin structure.
| |
Role of DNA-PKcs and Ku in DNA-damage signaling |
Despite DNA-PK having been identified for >10 years, the nature
of its physiological target(s) is still uncertain. Numerous in vitro
substrates have been identified, which include a variety of
transcription factors, most notably p53 (discussed in more detail
below), the RNA polymerase II large subunit carboxy-terminal domain
(CTD) and chromatin components (for review see Anderson and Lees-Miller
1992). However, whether any or all of these are functional DNA-PK
targets in vivo is not yet clear. The 34-kD subunit of replication
protein A (RPA) is also a good candidate for a DNA-PK target (Brush et
al. 1994) although conflicting reports exist in the literature as to
whether it is indeed phosphorylated by DNA-PK in vivo (Boubnov and
Weaver 1995; Fried et al. 1996). Other potential DNA-PK targets
include components of the DNA NHEJ machinery, such as XRCC4, which is
an effective substrate for DNA-PK in vitro and is known to be a
phosphoprotein in vivo (Critchlow et al. 1997; Leber et al. 1998).
Furthermore, as discussed above, both Ku subunits and DNA-PKcs are
subject to DNA-PK phosphorylation in vitro.
Key to elucidating the functional consequences of DNA-PK-mediated
phosphorylation events will be the identification and subsequent mutagenesis of phosphorylation sites. Such studies will be facilitated by DNA-PK target consensus sequences having been defined. The available data indicate that DNA-PK has a marked preference for Ser or
Thr residues preceding (and to a lesser extent preceded by) a Gln
residue, and that phosphorylation is potentiated by adjacent acidic
residues and tends to be inhibited by basic residues (Anderson and
Lees-Miller 1992; Bannister et al. 1993). Unfortunately from a
predictive standpoint, DNA-PK does not recognize all sequences conforming to this consensus, probably due to conformational
constraints. It is also clear that not all DNA-PK target sites conform
to this consensus.
Being a DNA damage-activated protein kinase, DNA-PK is inherently well
suited to functioning in DNA damage signaling. Thus, DNA-PK activation
in response to IR or other agents could trigger signaling pathways that
result in apoptosis or cell cycle checkpoint arrestsuch responses
being designed to prevent the proliferation of potentially mutated
cells or to allow DNA repair to occur prior to DNA replication or
mitosis (for review, see Elledge 1996). Consistent with this idea, an
efficient target for DNA-PK in vitro is p53, a factor whose levels and
activity are induced in response to DNA damage and which plays key
roles in triggering apoptosis or cell cycle arrest in such
circumstances (for review, see Ko and Prives 1996). Furthermore,
DNA-PK phosphorylates serine 15 and serine 37 of p53 in vitro
(Lees-Miller et al. 1990, 1992), and this has been reported to
destabilize interactions between p53 and Mdm2, a protein that
negatively regulates p53 in vivo by targeting it for ubiquitin-mediated
proteolysis (Haupt et al. 1997; Kubbutat et al. 1997; Shieh et al
1997). Initial analyses of cells defective in DNA-PK components have,
however, shown them to have intact DNA damage checkpoints and to be
capable of mediating p53 stabilization in response to IR (Bogue et al.
1996; Fried et al. 1996; Guidos et al. 1996; Huang et al. 1996; Nacht
et al. 1996; Candeias et al. 1997; Rathmell et al. 1997; Shieh et al. 1997). Nevertheless, recent work using certain DNA-PKcs defective cell
lines has indicated that DNA-PK activation in response to DNA damage
is necessary, but not sufficient, for the activation of sequence
specific DNA binding by p53 by an as yet undefined mechanism (Woo et
al. 1998). However, new data now exist questioning the interpretation
of this study (M. Hubank and G.Wahl pers. comm.). The above studies
indicate that DNA-PK either does not regulate p53 through serine 15 phosphorylation or does so in a redundant manner with other proteins.
In this regard, it has been shown recently that the DNA-PK related
proteins ATM and ATR are also capable of phosphorylating p53 on serine
15 in vitro (Banin et al. 1998; Canman et al. 1998; Khanna et al.
1998), and serine p53 phosphorylation in response to DNA-damaging
agents is debilitated in cells impaired or lacking ATM or ATR
function (Kastan et al. 1992; Tibbets et al. 1999). Thus, it may be
that ATM, ATR, and possibly DNA-PK signal different but partially
overlapping types of DNA damage to a common p53 effector pathway (Fig.
1). Notably, DNA-PK has been demonstrated to be
capable of phosphorylating Mdm2 in vitro (Mayo et al. 1997), suggesting
that this might be an alternative mechanism by which DNA-PK, ATM, or
ATR regulate p53 activity.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 1.
A hypothetical model whereby signaling to p53 by the
DNA-PK, ATM and ATR systems leads to phosphorylation at the amino
terminus of p53. These phosphorylation events result in the disruption
of the p53/MDM2 interaction, and hence lead to p53
protein stabilization and to p53-dependent downstream events. The three
DNA damage responsive systems may also signal to other effectors. The
arrow from DNA-PK to Mdm2 indicates another potential mechanism by
which DNA-PK and its relatives might modulate p53 activity or
stability.
|
|
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
Introduction
DNA-PK: a historical...
/
