![[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}
|
|
|
|
|
||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Vol. 15, No. 17, pp. 2177-2196, September 1, 2001
Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, North Carolina 27710, USA
The genomes of eukaryotic cells are under
continuous assault by environmental agents (e.g., UV
light and reactive chemicals) as well as the byproducts of normal
intracellular metabolism (e.g., reactive oxygen intermediates and
inaccurately replicated DNA). Whatever the origin, genetic damage
threatens cell survival, and, in metazoans, leads to organ failure,
immunodeficiency, cancer, and other pathologic sequelae. To ensure that
cells pass accurate copies of their genomes on to the next generation,
evolution has overlaid the core cell-cycle machinery with a series of
surveillance pathways termed cell-cycle checkpoints. The overall
function of these checkpoints is to detect damaged or abnormally
structured DNA, and to coordinate cell-cycle progression with DNA
repair. Typically, cell-cycle checkpoint activation slows or arrests
cell-cycle progression, thereby allowing time for appropriate repair
mechanisms to correct genetic lesions before they are passed on to the
next generation of daughter cells. In certain cell types, such as
thymocytes, checkpoint proteins link DNA strand breaks to apoptotic
cell death via induction of p53. Hence, loss of either of two
biochemically connected checkpoint kinases, ATM or Chk2, paradoxically
increases the resistance of immature (CD4+CD8+) T
cells to ionizing radiation (IR)-induced apoptosis (Xu and Baltimore
1996 In a broader context, cell-cycle checkpoints can be envisioned as
signal transduction pathways that link the pace of key cell-cycle phase
transitions to the timely and accurate completion of prior, contingent
events. It is important to recognize that checkpoint surveillance
functions are not confined solely to the happenings within the
nucleus-extranuclear parameters, such as growth factor availability
and cell mass accumulation, also govern the pace of the cell cycle
(Stocker and Hafen 2000
Introduction
; Hirao et al. 2000).
). However, for the purposes of this review we
will focus exclusively on the subset of checkpoints that monitor the
status and structure of chromosomal DNA during cell-cycle progression
(Fig. 1). These checkpoints contain, as their most proximal signaling elements, sensor proteins that scan chromatin for partially replicated DNA, DNA strand breaks, or other
abnormalities, and translate these DNA-derived stimuli into biochemical
signals that modulate the functions of specific downstream target proteins.
View larger version (36K):
[in a new window]
Figure 1.
A generic cell-cycle checkpoint signaling pathway.
(ROI) Reactive oxygen intermediate.
Despite the recent explosion of information regarding the molecular
components of cell-cycle checkpoints in eukaryotic cells, we still have
only a skeletal understanding of both the identities of the DNA damage
sensors and the mechanisms through which they initiate and terminate
the activation of checkpoints. However, members of the Rad group of
checkpoint proteins, which include Rad17, Rad1, Rad9, Rad26, and Hus1
(nomenclature based on the Schizosaccharomyces pombe gene
products) are widely expressed in all eukaryotic cells, and are prime
suspects in the lineup of candidate DNA damage sensors (Green et al.
2000; O'Connell et al. 2000). Three of these Rad proteins, Rad1, Rad9,
and Hus1, exhibit structural similarity to the proliferating cell
nuclear antigen (PCNA), and accumulating evidence supports the idea
that this similarity may extend to function as well (Thelen et al. 1999; Burtelow et al. 2000). During DNA replication, PCNA forms a
homotrimeric complex that encircles DNA, creating a "sliding clamp"
that tethers DNA polymerase 
to the DNA strand. Rad1, Rad9, and Hus1
are also found as a heterotrimeric complex in intact cells, and it has
been postulated that the Rad1-Rad9-Hus1 complex encircles DNA at or
near sites of damage to form a checkpoint sliding clamp (CSC)
(O'Connell et al. 2000), which could serve as a nucleus for the
recruitment of the checkpoint signaling machinery to broken or
abnormally structured DNA. The analogy between PCNA and the
Rad1-Rad9-Hus1 complex extends even further. The loading of the PCNA
clamp onto DNA is controlled by the clamp loading complex, replication
factor C (RFC). Interestingly, yet another member of the Rad family,
Rad17, bears homology to the RFC subunits and, in fact, associates with
RFC subunits to form a putative checkpoint clamp loading complex (CLC)
that governs the interaction of the Rad1-Rad9-Hus1 CSC with damaged
DNA (Green et al. 2000; O'Connell et al. 2000). Although this model is
fascinating, rigorous biochemical evaluations of the interplay between
the CLC and CSC complexes, and the interactions of both complexes with
damaged chromatin, are needed before the model can be accepted without reservation.
Moving downstream of the sensor apparatus, we find additional analogies
between checkpoint pathways and standard signal transduction cascades,
in that both rely heavily on protein phosphorylation for signal
transmission and amplification. Cell-cycle checkpoint kinases belong
largely, if not entirely, to the serine-threonine kinase family, and
the proteins they target for modification range from more downstream
members of the checkpoint pathway itself (e.g., additional protein
kinases or noncatalytic scaffolding proteins) to distal elements that
mediate cell-cycle arrest and DNA repair responses (e.g., the Cdc25C
phosphatase or type 2A histones) (Rogakou et al. 1999; Downs et al.
2000; O'Connell et al. 2000; Paull et al. 2000).
During the very earliest stages of checkpoint activation, DNA damage
sensors relay information, via a still-elusive mechanism, to members of
a recently defined family of phosphoinositide 3-kinase related kinases
(PIKKs; Tibbetts and Abraham 2000). In mammalian cells, two PIKK family
members, ATM (ataxia-telangiectasia
mutated) and ATR (ATM and Rad
3-related), play critical roles in early signal transmission through
cell-cycle checkpoints. Homologs of ATM and ATR are present in all
eukaryotic cell types examined to date, including budding and fission
yeast. The present review focuses on the biochemistry and function of
the mammalian checkpoint kinases, ATM and ATR, with only brief
references to the precedent literature from yeast model systems. For
more information regarding the yeast ATM/ATR homologs, or a more global
overview of cell-cycle checkpoints, the reader is referred to several
recent reviews (Elledge 1996; Weinert 1997; Lowndes and Murguia 2000;
Tibbetts and Abraham 2000; Zhou and Elledge 2000).
| |
The PIKK family |
|---|
As mentioned earlier, ATM and ATR belong to a structurally unique
family of protein serine-threonine kinases whose catalytic domains
share a clear evolutionary relationship with those of mammalian and
yeast phosphoinositide 3-kinases (Hunter 1995; Zakian 1995; Tibbetts
and Abraham 2000). The molecular cloning of PIKK-encoding cDNAs began
with the isolation of the products of the target of rapamycin genes
(TOR1 and TOR2) from Saccharomyces
cerevisiae (Cafferkey et al. 1994; Helliwell et al. 1994), followed
by the cloning of the mammalian ortholog, termed mTOR (also
called FRAP or RAFT1) (Brown et al. 1994; Sabatini et
al. 1994; Sabers et al. 1995). As the TOR acronym indicates, these PIKK
family members are the protein targets of the potent antifungal and
immunosuppressive agent, rapamycin (for reviews, see Abraham and
Wiederrecht 1996; Gingras et al. 2001). Shortly thereafter, a flurry of
reports described a series of yeast, fly, worm, and mammalian proteins that also expressed the characteristic phosphoinositide 3-kinase-like catalytic domain (Zakian 1995; Tibbetts and Abraham 2000). One of the
cloned cDNAs, termed ATM, attracted widespread attention, because mutations in this gene underlie the heritable chromosomal instability disorder, ataxia-telangiectasia (A-T) (Savitsky et al.
1995). On the heels of these initial papers, the ATR cDNA was
uncovered during a search of an EST database for additional gene
products bearing the mTOR/FRAP-like catalytic domain (Cimprich et al. 1996).
Recent studies in the worm, Caenorhabditis elegans, and in
human cells have identified yet another large (~390 kD) protein kinase whose catalytic domain indicates membership in the PIKK family.
The C. elegans gene, SMG-1, encodes a putative
protein kinase that plays an essential role in the elimination of RNA species containing premature termination codons, a process termed nonsense-mediated mRNA decay (NMD; Maquat 2000). Subsequently, a human
PIKK family member bearing regional sequence homology to SMG-1 was
cloned independently by at least two research teams (Denning et al.
2001; K.M. Brumbaugh, D.M. Otterness, and R.T. Abraham, in prep.).
Although the human cDNA has been designated hSMG-1 by one group
(Denning et al. 2001), we suggest that this name remain provisional
until a role for this PIKK family member in NMD is unequivocally shown.
Our own results indicate that hSMG1 is found in both the nucleus and
cytoplasm, and, like ATM and ATR, exhibits phosphotransferase activity
toward a variety of proteins containing Ser-Gln motifs under in vitro
assay conditions. At this early stage, it seems quite likely that the
putative hSmg1 homolog will join ATM and ATR as a stress-responsive
protein kinase, and that, in addition to its predicted role in NMD,
this intriguing addition to the PIKK family will offer further insights
into stress-induced signaling in mammalian cells.
In addition to the sequence homology in the catalytic domains, the PIKK
family members exhibit a similar overall structural organization (see
Fig. 2 for examples). When compared with
other kinases (protein or lipid), the PIKKs stand out immediately as very large polypeptides, with molecular masses ranging from ~300 kD
to >500 kD. The catalytic domains (~300 amino acids) of the PIKK
family members are located near their carboxyl termini, and are flanked
by two loosely conserved domains termed FAT
(FRAP/ATM/TRRAP) and FATC (the
"C" indicates carboxy-terminal) (Bosotti et al. 2000). Although
the FAT/FATC domains contain no identifiable catalytic sequences, the
fact that these domains are always expressed in pairs has raised the
still untested hypothesis that they interact in an intramolecular
fashion, and thereby regulate the conformation of the interposed
kinase domain (Bosotti et al. 2000). Despite the sequence similarity
to phosphoinositide kinases, the catalytic domains of the PIKKs
appear to transfer phosphate exclusively to protein rather than lipid
substrates.
|
The current members of the PIKK family can be grouped into six
subfamilies on the basis of both sequence homology and function (Durocher and Jackson 2001) (Table 1). The
mammalian members of five of the six subfamilies are known to
phosphorylate protein substrates on serine or threonine residues. In
mammalian cells, ATM and ATR are thought to share responsibilities as
the apical protein kinases in all of known cell-cycle checkpoints, with
the possible exception of the mitotic spindle checkpoint, which is activated by treatment with nocodazole, an inhibitor of microtubule polymerization. The catalytic subunit of DNA-dependent protein kinase
(DNA-PKcs) functions as a heterotrimeric complex containing two Ku subunits (Ku-70 and Ku-80), and makes its own crucial
contribution to genome maintenance by overseeing the nonhomologous
end-joining (NHEJ) pathway of DNA double-strand break (dsb) repair
(Smith and Jackson 1999). On the other hand, mTOR plays no identifiable role in genome surveillance; rather, this PIKK coordinates G1 phase progression with the supply of nutrients and growth factors (Gingras et al. 2001). As discussed previously, a very recent addition
to this list is hSMG-1, which has no yeast counterparts, but, based on
sequence homology to the Caenorhabditis elegans SMG-1 protein,
is predicted to function in NMD (Denning 2001). Our group has cloned a
partially overlapping open reading frame, and we have provisionally
termed the encoded polypeptide ATX (K.M. Brumbaugh, D.M. Otterness, and
R.T. Abraham, in prep.). Finally, the members of the TRAPP subfamily
express catalytic domains that have sustained disabling mutations
during eukaryotic evolution (Grant et al. 1998). The latter proteins
may function as molecular scaffolds during eukaryotic gene
transcription (McMahon et al. 1998).
|
| |
Biochemistry of the ATM and ATR kinases |
|---|
Before launching into a discussion of the roles of ATM and ATR in
checkpoint signaling, it seems appropriate to review the current state
of knowledge regarding their protein kinase activities, as many
laboratories are actively pursuing the identification of in vivo
substrates for the ATM and ATR catalytic domains. The first PIKK family
member to be characterized as a protein kinase was DNA-PK (Gottlieb and
Jackson 1993; Smith et al. 1999). Like most protein kinases, partially
purified preparations of DNA-PK displayed Mg2+-dependent,
serine-threonine kinase activity in the presence of ATP and substrate
(e.g, the transcription factor, SP1; Gottlieb and Jackson 1993; Hartley
et al. 1995). Consistent with its role in signaling during DNA dsb
repair, the in vitro kinase activity of DNA-PK was optimal when both
enzyme and substrate were bound to the same DNA fragment (Gottlieb and
Jackson 1993). In addition, the preferred target sequence for
phosphorylation by DNA-PK was serine or threonine followed by glutamine
at the +1 position. Hence DNA-PK is commonly identified as an
"S/T-Q-directed kinase", based on the single-letter amino acid code
for the consensus phosphorylation site.
The availability of ATM- and ATR-specific antibodies allowed
investigators to determine whether these two PIKKs also expressed protein kinase activity, at least in the test tube. Immune complex kinase assays quickly revealed that, like DNA-PK, ATM and ATR displayed
S/T-Q-directed kinase activities under in vitro conditions, and that a
physiologically relevant substrate for both protein kinases was p53
(Banin et al. 1998; Canman et al. 1998; Tibbetts et al. 1999; Ziv et
al. 2000). The major in vitro phosphorylation site for ATM and ATR was
Ser 15, which resides in the sequence context, LSQE
(phosphorylation site is underlined). Interestingly, all of the PIKKs
characterized to date, mammalian or otherwise, function as
S/T-Q-directed kinases, with the notable exception of the yeast and
mammalian TOR proteins. In the case of mTOR, phosphorylation of the
best-characterized protein substrate, PHAS-I (also termed 4E-BP1)
occurs mainly at two threonine residues (Thr 36 and Thr 45) nested
within duplicated STTPGG sequences (Brunn et al. 1997a,b;
Gingras et al. 1999, 2001). However, mTOR fails to phosphorylate Ser 15 in p53, or S/T-Q sites in other proteins that are known targets for
ATM/ATR (R.T. Abraham, unpubl.). The reader should not be confused by
ATM and ATR kinase assay protocols that use PHAS-I as the substrate
(e.g., Banin et al. 1998; Yang and Kastan 2000). The PHAS-I polypeptide
is rich in serine and threonine residues, and, in fact, contains three
S/T-Q sites, one of which (Ser 111) is phosphorylated by ATM or ATR in
vitro (Yang and Kastan 2000; R.T. Abraham, unpubl.). The potential
relevance of Ser 111 phosphorylation by ATM to PHAS-I function (Yang
and Kastan 2000) is discussed in a later section of this review.
Nonetheless, the major mTOR phosphorylation sites, Thr 36 and Thr 45, are not targeted by either ATM or ATR in immune complex kinase assays.
Two parameters related to the protein kinase assays for ATM and ATR are
worthy of mention, in that both raise some cautionary notes concerning
the extent to which the activities determined in vitro reflect the
actual situation in intact cells. First, the protein kinase activity of
ATM, but not of ATR or DNA-PK, is quite sensitive to inhibition by the
nonionic detergents (e.g., Triton X-100, NP-40) commonly used to
prepare whole cell or nuclear extracts (Gottlieb and Jackson 1993;
Sarkaria et al. 1998, 1999; Yavuzer et al. 1998). The inhibitory effect
of certain nonionic detergents on ATM kinase activity is not without
precedent, in that a similar result was obtained during the initial
characterization of the catalytic activity of mTOR (Brunn et al.
1997a). In both cases, the detergent problem was circumvented through
substitution of Triton X-100/NP-40 with a less stringent detergent
(e.g., Tween-20) in the cell lysis buffer (Brunn et al. 1997a; Banin et
al. 1998; Sarkaria et al. 1998; Tibbetts et al. 1999; Ziv et. al.
2000). Although the explanation remains unclear, the inhibitory effects of nonionic detergents on ATM kinase activity may reflect either the
disruption of intramolecular interactions that are required to maintain
the native conformation of the catalytic domain, or the loss of
regulatory protein or lipid partners during the extraction and
purification process.
The protein kinase activities of ATM and ATR, as measured in immune
complex kinase assays, also display unusually strong dependencies on
the presence of millimolar concentrations of Mn2+ in the
kinase assay buffer. Under test-tube conditions, it appears that the
presence of the Mn2+-ATP complex is virtually essential for
the phosphorylation of various substrates by ATM and ATR (Kim et al.
1999; Chan et al. 2000; R.T. Abraham, unpubl.). Again, this property of
extreme Mn2+ dependence is shared with mTOR (Brunn et al.
1997a). The catalytic activities of many conventional protein kinases
(e.g., Src family protein tyrosine kinases) are stimulated by addition
of Mn2+ in lieu of Mg2+ as the ATP-binding cofactor
in the kinase reaction buffer (Cooper and King 1986). However, the fact
that ATM and ATR display almost undetectable phosphotransferase
activities in the presence of the physiologic phosphate donor,
Mg2+-ATP, does raise questions about their actual activities
in intact cells. As a cofactor for phosphate transfer from ATP to
substrate, Mn2+ is intrinsically far more active than is
Mg2+ (Schieven and Martin 1988). Perhaps the low levels of
phosphorylating activity observed in the presence of
Mg2+-ATP complexes accurately reflects the fact that ATM and
ATR turn over their respective protein substrates at relatively low
rates in intact cells. This scenario meshes nicely with the idea that substrate phosphorylation by ATM/ATR, in the context of a DNA-bound damage recognition complex, places a higher premium on spatially restricted substrate modification than on signal amplification through
rapid phosphorylation of many molecules of the same target protein. An
alternative and equally plausible explanation is that the processes of
extraction and purification compromise the normally Mg2+-dependent kinase activity of ATM/ATR to the point where
significant activity in vitro can only be visualized in the presence of
Mn2+-ATP complexes.
Despite these uncertainties, recent studies have shed some meaningful
light on the preferred target sequences for phosphorylation by ATM and
ATR. One approach was based on the earlier finding that ATM
phosphorylated the physiologically relevant substrate, p53, exclusively
within the S-Q motif beginning at Ser 15 (Siliciano et al. 1997; Banin
et al. 1998). Kastan and coworkers generated a panel of glutathione
S-transferase (GST)-p53 fusion proteins in which the residues
surrounding the ATM/ATR target site, Ser 15, were systematically varied
(Kim et al. 1999). The general conclusion from these experiments was
that hydrophobic or acidic residues surrounding the targeted Ser-Gln
motif favored phosphorylation of the Ser residue by immunopurified ATM,
whereas positively charged amino acids were inhibitory. Furthermore,
ATM exhibited a strong preference for Ser over Thr as the
phosphoacceptor site, at least in the context of the p53 amino-terminal
sequence. A second group of investigators used an iterative peptide
library screening approach to define an optimal sequence context for
phosphorylation by ATM (O'Neill et al. 2000). These efforts identified
a consensus phosphorylation site
[(M/F)-(Q/P)-L-S-Q-(E/Q)] that was in reasonable
agreement with that defined by Kastan and coworkers (Kim et al. 1999).
Remarkably, the relatively unbiased peptide selection strategy zeroed
in on a core L-S-Q-E target sequence, which was identical to that
surrounding the Ser 15 site in p53.
The available evidence suggests that the consensus sequence for
phosphorylation by ATR overlaps extensively with that defined for ATM
(Kim et al. 1999). The similarity in terms of substrate preference
contrasts sharply with the differential activities of these proteins
during DNA damage responses (see below for discussion). Once again,
these results suggest that proximity, rather than sequence context,
plays a pivotal role in the selection of those substrates that undergo
phosphorylation by ATM versus ATR in response to genotoxic stress. In
general, the specific activity of ATR toward most substrates in vitro
is significantly lower than that displayed by ATM (Canman et al. 1998;
R.S. Tibbetts and R.T. Abraham, unpubl.). One notable exception is the
human checkpoint protein, hRad17, which contains two S-Q sites that are
phosphorylated by both ATM and ATR in immune complex kinase assays. A
peptide containing these sites was phosphorylated at a higher rate by
ATR than by ATM (Kim et al. 1999), and we have found that a GST-hRad17
fusion protein is a significantly better substrate for ATR than for ATM (Bao et al. 2001; R.S. Tibbetts and R.T. Abraham, unpubl.).
The identification of favorable sequence contexts for phosphorylation
by ATM/ATR prompted in silico searches for candidate physiological
substrates for these protein kinases (Kim et al. 1999; O'Neill et al.
2000). These database mining efforts yielded a rich harvest of
checkpoint/DNA repair proteins, including p53, hRad17, hChk1, the
Nijmegan breakage syndrome protein (NBS1, also termed p95 or nibrin),
BRCA1, and BRCA2. Many of these early candidates have now been
established as substrates for either ATM only, or both ATM and ATR. As
will become apparent from the remaining sections of this review,
substrate identification has become a very hot topic in the ATM/ATR
field, and the search has now expanded to include novel suspects in
various checkpoint signaling pathways, based, in part, on the presence
of conserved S-Q target motifs in the primary sequences of these proteins.
The final, and possibly the most controversial, aspect of ATM/ATR
biochemistry concerns the mechanism whereby DNA damage triggers substrate phosphorylation by these protein kinases. The standard paradigm for protein kinase signaling suggested that the stimulus (i.e., genotoxic stress) would somehow shift the ATM/ATR kinase domains
from low-activity to high-activity states. This prediction held true
for ATM, as the protein kinase activity of the immunoprecipitated enzyme increases several fold within 1 h after cellular exposure to IR
or radiomimetic agents (Banin et al. 1998; Canman et al. 1998). On the
other hand, our group has failed to detect any increases in the in
vitro kinase activity of ATR after treatment of cells with various
genotoxic agents, including IR and UV light (R.S. Tibbetts and R.T.
Abraham, unpubl.). However, immunofluorescence microscopy revealed that
ATR responded to DNA damage by undergoing a dramatic shift in
intranuclear localization, from diffuse to focal in nature (Tibbetts et
al. 2000; see below for discussion). Under identical experimental
conditions, ATM did not enter into nuclear foci in cells damaged with
IR or other DNA-damaging agents. Collectively, these results hint that
ATM and ATR respond to DNA damage in fundamentally different fashions:
One checkpoint kinase (ATM) becomes catalytically active, whereas the
other (ATR) redistributes into DNA damage-induced nuclear foci, where
it presumably gains access to its substrates.
As is usually the case with "simple" models, the bifurcating DNA
damage response mechanism proposed earlier will undergo substantial embellishment during the next few years. Although immunoprecipitation of ATR from whole cell extracts provides no evidence for activation of
the ATR kinase domain by genotoxic agents, it is possible that enzyme
activation in fact occurs, but is restricted to the subpopulation of
ATR molecules that migrates into nuclear foci. In this situation, the
increase in activity may be lost in the background noise generated by
immunoprecipitation of the entire extractable pool of ATR from the
damaged cells. Biochemical strategies directed toward selective recovery of focally localized ATR might simultaneously enrich for the
activated form of this protein kinase. Conversely, recent findings from
Shiloh and coworkers suggest that ATM might also undergo relocalization
to nuclear foci in response to radiomimetic agents (Andejecko et al.
2001). These ATM-containing complexes are colocalized with the
phosphorylated form of histone H2AX, which suggests that they are
generated in close proximity to DNA dsbs (Downs et al. 2000; Paull et
al. 2000). Thus, future experiments may lead us to conclude that the
early responses of ATM and ATR to DNA damage are not as mechanistically
distinct as they now appear.
A major area of uncertainty surrounds the mechanism whereby IR
treatment of intact cells leads to an increase in the protein kinase
activity of ATM. An obvious possibility is that reactive oxygen
intermediates or DNA dsbs produced during IR exposure triggers a
posttranslational modification (e.g., phosphorylation) of ATM that
activates the kinase domain. Interestingly, a candidate
autophosphorylation site (Ser 440) has been identified in the
amino-terminal region of ATM, but whether this site is actually
phosphorylated in response to IR remains unknown (Kim et al. 1999). An
alternative model, which has garnered some experimental support, is
based on the DNA-PK paradigm. Binding of the DNA-PK holoenzyme complex
to free DNA ends stimulates the activity of the DNA-PKcs
subunit (Smith et al. 1999). The familial relationships among ATM,
ATR, and DNA-PK prompted speculation that ATM and ATR activities might
also be regulated, in a more or less direct fashion, through
interactions with DNA. Subsequent studies revealed that biochemically
purified ATM could be captured from solution by immobilized DNA
fragments (for review, see Durocher and Jackson 2001). The association
of ATM with free DNA ends was also visualized directly by atomic force
microscopy (Smith et al. 1999). Similarly, studies in the Xenopus model system indicate that ATR binds to DNA, although specific DNA end-binding activity was not investigated (Guo et al.
2000). Nonetheless, these results are consistent with the notion that
ATM and ATR are capable of interacting directly with DNA.
A far more contentious issue is whether single- or double-stranded DNA
stimulates the rates of substrate phosphorylation by ATM and ATR in
vitro. Unfortunately, several groups have recently addressed this
model, and diametrically opposite results were obtained
some
investigators observe DNA-dependent activation, and others do not (Guo
et al. 2000; Durocher and Jackson 2001). In certain, but not all, cases
(Guo et al. 2000), the apparent stimulatory effect of DNA on ATM/ATR
kinase activity may stem from the use of DNA-binding substrates, such
as full-length p53 or replication protein A. The concern here is that
the effect of DNA is entirely indirect, that is, the DNA-bound form of
the substrate is in a more favorable conformation for phosphorylation by the soluble ATM or ATR kinase. As discussed earlier, it might be
argued that the presence of Mn2+ in the kinase assay buffers
alleviates a biologically relevant dependency of ATM/ATR activity on
DNA. However, Kastan and coworkers have reported that the complete
replacement of Mn2+ with Mg2+ in the assay buffer
fails to uncover any stimulatory effect of DNA on ATM/ATR kinase
activity (Kim et al. 1999). The discrepant outcomes in these assays
might be attributable to the different preparations of ATM and ATR
(e.g., biochemically purified vs. immunoprecipitated protein) used by
the different laboratories. Perhaps some groups copurify ATM and ATR with
sheared chromosomal DNA, which causes the resulting protein kinase activities
to appear refractory to stimulation by exogenously added DNA in vitro.
The strongest case for DNA-dependent activation of ATR comes from
recent studies of Xatr derived from Xenopus egg extracts (Guo
et al. 2000; Hekmat-Nejad et al. 2000). An intriguing enzyme purification scheme was used in one of these studies, beginning with
the capture of a subpopulation of the total pool of Xatr molecules on
DNA cellulose, followed by digestion of the DNA with DNase I (Guo et
al. 2000). The resolubilized Xatr was then immunoprecipitated with
anti-Xatr antibodies for immune complex kinase assays. Remarkably, the
specific kinase activity of the DNA-binding subpopulation of Xatr
molecules was 10- to 20-fold higher than that of Xatr immunoprecipitated directly from the egg cytosol. These results suggest
either that binding to DNA-cellulose triggers an increase in Xatr
kinase activity, or that the DNA-cellulose selectively binds to a
preactivated form of Xatr.
In summary, although an increasingly compelling body of evidence argues
that ATM and ATR are capable of associating directly or indirectly with
DNA, the impact of DNA binding on their catalytic activities remains a
matter of debate. The most parsimonious model would suggest that, like
the DNA-PK holoenzyme (Smith et al. 1999), ATM and ATR are attracted to
sites of DNA damage through constitutive or inducible associations with
DNA-binding regulatory subunits. Members of the Rad family of
checkpoint proteins are presently the best candidates for the putative
Ku analogs in the ATM/ATR pathway. Recent studies have identified Rad26
and LCD1/PIE1/DDC2 as potential regulatory subunits of the ATM/ATR
homologs expressed in fission and budding yeast, respectively (Edwards
et al. 1999; Paciotti et al. 2000; Rouse and Jackson 2000; Wakayama et
al. 2001), and many laboratories are undoubtedly engaged in the search for the mammalian Rad26 homolog. Although there is no reason to expect
that Rad26 itself would tether ATM/ATR to damaged DNA, it may place
these protein kinases in a DNA-responsive conformation. A second
potential DNA-targeting subunit is Rad17 (S. pombe
nomenclature). In fission yeast, rad17+ associates with
chromatin throughout the yeast cell cycle, and this basal level of
DNA-bound rad17+ is either increased or decreased in response
to genotoxic stress, with the direction of change determined by the
type of DNA damage incurred by these cells (Griffiths et al. 2000; Kai
et al. 2001). In addition to its predicted role as a checkpoint clamp
loader for the Rad1-Rad9-Hus1 complex (Thelen et al. 1999; Venclovas and Thelen 2000), our group has shown that treatment of human cells
with genotoxic agents triggers rapid associations of ATM and ATR with
hRad17 (Bao et al. 2001). Thus, it is conceivable that chromatin-bound
Rad17 might regulate the trafficking of ATM and ATR onto and off of
damaged DNA in vertebrate cells. The next few years should provide a
much clearer view of both the recruitment of ATM/ATR to sites of DNA
damage, and of the proximal events that couple DNA damage recognition
to enhanced phosphorylation of substrates by both of these protein kinases.
| |
The ATM/ATR-associated signaling machinery: complex complexes |
|---|
Like other members of the PIKK family, ATM and ATR contain very
large amino-terminal domains, the functions of which are largely unknown. However, it has long been speculated that the very extended amino-termini of the PIKKs also play central roles in scaffolding these
protein kinases into macromolecular signaling complexes. Indeed, gel
filtration analyses indicate that both ATM and ATR are constitutive
residents of very high molecular weight (MR, >2 × 106 D) protein complexes in mammalian cells (Wright
et al. 1998; Shiloh 2001). The identification of the components of
these complexes is an area of burgeoning interest, with the expected
payoff being novel insights into both the upstream regulators and
downstream targets of ATM and ATR in cell-cycle checkpoint pathways.
An intriguing first step toward the elucidation of the ATM- and
ATR-associated signaling complexes comes from mass spectrometric analysis of anti-BRCA1 immunoprecipitates prepared from mammalian cell
extracts (Wang et al. 2000). The breast cancer susceptibility protein,
BRCA1, is a critical component of the checkpoint signaling and DNA
repair machinery, and is a direct target for phosphorylation by ATM and
ATR in cells exposed to genotoxic stress (Cortez et al. 1999; Scully
and Livingston 2000; Scully et al. 2000; Tibbetts et al. 2000). The
sequencing of BRCA1-associated polypeptides yielded a remarkably broad
array of proteins with clear connections to checkpoints, DNA repair,
and human chromosomal instability syndromes, an outcome that prompted
the authors to coin the acronym BASC
(BRCA1-associated genome
surveillance complex) as a global descriptor
for this complex (Wang et al. 2000). In addition to ATM, the members of
the BASC include the mismatch repair proteins, MLH1, MSH2, and MSH6,
the Bloom's syndrome helicase (BLM), and the Mre11-Rad50-NBS1
complex. The latter complex plays important roles in the
recombinational repair of DNA dsbs, and the loss of one of the
components of this complex, NBS1, gives rise to the Nijmegen breakage
syndrome, a chromosomal instability disorder that bears several
similarities to A-T (Shiloh 1997). Two independent studies of
BRCA1-associated proteins added components of the chromatin remodeling
apparatus and the Fanconi anemia-related protein FANCD2 to an expanding
list of BASC components (Bochar et al. 2000; Garcia-Higuera et al. 2001).
The identification of the BASC represents the start of a challenging
but much needed effort to understand the afferent inputs that regulate
ATM and ATR function, and the intermolecular interactions that control
the presentation of appropriate substrates to these checkpoint kinases.
It is important to recognize that the BASC is likely not a single
entity, but rather a dynamic collection of protein complexes whose
compositions change with the type of DNA damage, location relative to
the damaged site, and time after initiating genetic insult. If BRCA1 is
indeed a central scaffold for complex assembly, then it will be
important to determine whether site-specific phosphorylation of BRCA1
by ATM/ATR (Cortez et al. 1999; Tibbetts et al. 2000), hChk2 (J.S. Lee
et al. 2000), and other protein kinases controls the migration of
specific checkpoint and repair proteins into and out of the BASC.
A related area that requires additional investigation concerns the
relationship between the biochemically defined complexes containing ATR
and BRCA1, and the DNA damage-induced nuclear foci that have been
observed by immunofluorescence microscopy (Scully et al. 1997; Tibbetts
et al. 2000). Many of these foci react with both ATR- and
BRCA1-specific antibodies, suggesting that the two proteins colocalize
at sites of DNA damage. The colocalization of a checkpoint kinase with
its substrate seems quite logical; however, this simple model is
complicated by the recognition that, to be detected as
immunofluorescent nodes, DNA damage-induced foci must contain hundreds
of ATR and BRCA1 polypeptides. If we assume that nuclear foci mark
individual sites of DNA damage, then, in lieu of trivial explanations,
such as fixation-induced structural artifacts, it becomes important to
understand why each DNA lesion triggers the coalescence of so many ATR
and BRCA1 polypeptides. In keeping with the potentially dynamic state
of the BASC at various stages of the DNA damage response, it is
plausible that a single damaged site stimulates the generation of a
dynamic, interactive series of ATR-BRCA1 complexes, each comprising a
subset of the BASC components described earlier.
| |
Cell-cycle checkpoint functions of ATM and ATR |
|---|
Investigations of the checkpoint signaling functions of ATM and ATR
have largely followed a time-tested strategy for protein kinases, which
begins with the identification of substrates and proceeds through
analyses of the functional consequences of substrate phosphorylation.
Before the cloning of the ATM gene, the dedicated efforts of
many laboratories documented that ATM-deficient cells displayed
significant defects in the G1, S, and G2
checkpoints. These observations prompted intensive searches for
ATM substrates among the numerous proteins that function in each of
these checkpoints. Most of these ATM targets have also been tested, in
an empirical fashion, as substrates for ATR. Although the empirical
approach has been rewarding, the inevitable outcome has been that the
list of documented ATR substrates overlaps largely, if not entirely, with that linked to ATM. Despite the seeming overlap, the emerging picture suggests that the checkpoint signaling functions of ATM and ATR
are far from redundanta conclusion that should become apparent as we
march through the G1, S, and G2 checkpoints in the
following sections.
| |
The G1 checkpoint |
|---|
At the heart of the G1 checkpoint lies the series of
events leading to the accumulation of the tumor suppressor protein,
p53. Although p53 exerts a pervasive influence on checkpoint functions during the mammalian cell cycle, the G1 checkpoint represents the only case in which loss of p53 leads to total checkpoint abrogation (Ko and Prives 1996; Giaccia and Kastan 1998; North and Hainaut 2000).
DNA damage induced by most, if not all, forms of genotoxic stress
induces a rapid increase in the level of p53, a response that is
mediated primarily through an increase in protein stability. In
addition to triggering the accumulation of p53, genotoxic stress induces posttranslational modifications that regulate the
transcriptional activating functions of this protein. With respect to
the G1 checkpoint, a key target for transcriptional
activation by p53 is the cyclin-dependent kinase inhibitor, p21 (also
termed WAF1 or CIP1). The p53-dependent increase in p21 expression
suppresses cyclin E- and cyclin A-associated cdk2 activities, and
thereby prevents G1-to-S phase progression. In addition to
p21, the activated form of p53 stimulates the expression of a large
panel of genes, which, depending on the cellular context and type of
initiating insult, may modulate intracellular redox status, or induce
the host cell to undergo apoptosis (Yu et al. 1999).
An intricate web of protein kinases and phosphatases, as well as
histone acetylases and ubiquitin-conjugating enzymes, regulates the
accumulation and transcriptional-activating functions of p53 (Ko and
Prives 1996; Giaccia and Kastan 1998). A link between ATM and p53 was
predicted on the basis of earlier studies, which demonstrated that A-T
cells exhibited a delayed and reduced level of p53 protein induction
following exposure to IR (Kastan et al. 1992; Lu and Lane 1993). The
subsequent cloning of ATM allowed several groups to test the
straightforward hypothesis that ATM was a direct effector of p53
phosphorylation in IR-damaged cells (Banin et al. 1998; Canman et al.
1998). These studies pinpointed a single serine residue (Ser 15) in the
amino-terminal region of p53 as a phosphorylation site for the ATM
kinase in vitro. Moreover, phosphorylation of Ser 15 was rapidly
induced in IR-treated cells, and this response was ATM dependent, as
IR-induced Ser 15 phosphorylation was significantly, but not
completely, suppressed in A-T cells (Siliciano et al. 1997; Banin et
al. 1998; Canman et al. 1998). The residual phosphorylation at Ser 15 in A-T cells hinted that ATM was not the only IR-regulated Ser 15 kinase, and this suspicion was confirmed with the observation that
UV-induced Ser 15 phosphorylation was virtually unimpaired in A-T
cells. Subsequent studies showed that ATR was also capable of
phosphorylating p53 at Ser 15 in immune complex kinase assays
(Hall-Jackson et al. 1999; Lakin et al. 1999; Tibbetts et al. 1999). In
cells rendered functionally deficient for ATR by overexpression of a
kinase-inactive ATRKI mutant (Cliby et al. 1998), the early
phase (0-2 h) of IR-induced p53 phosphorylation was not impaired,
which is consistent with the idea that ATM serves as the major Ser 15 kinase during the acute response to IR (Tibbetts et al. 1999). However,
the ATRKI-overexpressing cells did show a significant defect
in their ability to maintain phosphorylation of Ser 15 at later times,
which suggests that the maintenance phase of p53 phosphorylation was
more highly dependent on ATR. On the other hand, treatment of human
fibroblasts with UV light triggered a Ser 15 phosphorylation response
that was largely independent of ATM expression, and was strongly
reduced, at all times postirradiation, by overexpression of
ATRKI (Tibbetts et al. 1999).
This pattern of dual regulation of substrate phosphorylation by ATM and ATR in cells exposed to different forms of genotoxic stress has become a recurrent theme in the checkpoint-signaling field. Of the PIKK family members, ATM represents the primary responder to IR- or radiomimetic agent-induced DNA damage. In the absence of ATM, or in normal cells that incur a high level of IR-induced DNA damage, ATR serves mainly as a backup kinase for ATM. On the other hand, ATR takes on the front-line signaling responsibilities when cells are challenged with other forms of genotoxic stress, such as UV light exposure or treatment with agents that interfere with DNA replication (aphidicolin, hydroxyurea; HU).
As is frequently the case, deciphering the functional consequences of
p53 phosphorylation by ATM and ATR proved far more challenging than
identification of the phosphorylation. The location of Ser 15 at the
p53 amino terminus suggested that modification of this residue might
trigger the dissociation of p53 from MDM2, a protein that targets p53
for ubiquitination, nuclear export, and proteosomal degradation
(Freedman et al. 1999; Juven-Gershon and Oren 1999). Therefore, if the
model were correct, ATM/ATR-dependent phosphorylation of Ser 15 would
free p53 from its destabilizing binding partner, thereby favoring p53
accumulation. It turns out, however, that Ser 15 phosphorylation is not
sufficient to disrupt the p53-MDM2 interaction; rather, this
modification stimulates the transactivating function of p53 by
enhancing the binding of this protein to the transcriptional
coactivator, p300 (Dumaz and Meek 1999). However, these results do not
rule out the possibility that phosphorylation of p53 at Ser 15 sets
this protein up for a secondary modification that does modify the
binding of MDM2 to p53, thereby inhibiting p53 degradation. Indeed,
Dumaz et al. (1999) have shown that Ser 15 phosphorylation greatly
enhances the subsequent phosphorylation of p53 at Ser 18 by casein
kinase I, at least under test-tube assay conditions with purified
proteins. The presence of phosphates at Ser 15 and Ser 18 reduces the
avidity of full-length p53 for MDM2 by approximately threefold. Further
studies are required to determine whether, and under what conditions,
the tandem modification of p53 by ATM/ATR and casein kinase I
contributes to p53 accumulation in intact cells.
Recent findings have reinforced the notion that ATM enhances p53
accumulation by triggering the release of this protein from MDM2. One
mechanism for p53 stabilization involves an intermediate protein
kinase, hChk2 (also named hCds1), which relays ATM-dependent signals to
p53 and many other downstream target proteins in IR-damaged cells. ATM
activates hChk2 by phosphorylating an amino terminal Thr residue (Thr
68) (Ahn et al. 2000; Melchionna et al. 2000), and hChk2, in turn,
phosphorylates yet another amino-terminal Ser residue (Ser 20) in p53
(Chehab et al. 2000; Hirao et al. 2000; Shieh et al. 2000). Unlike the
Ser 15 modification mentioned earlier, phosphorylation at Ser 20 interferes directly with the binding of p53 to MDM2, thereby favoring
p53 accumulation in response to IR-induced DNA damage. The
physiological relevance of hChk2 in the regulation of p53 is supported
by the finding that loss-of-function mutations in hChk2 can give rise
to a variant form of Li-Fraumeni syndrome, a heritable, cancer-prone
disorder typically associated with germ-line mutations in p53 (Bell et
al. 1999).
It now seems likely that ATM also targets the p53-MDM2 interaction via
direct modification of the partner protein, MDM2 (Maya et al. 2001). In
this report, the authors followed a circuitous but rewarding line of
investigation, which began with the identification of anti-MDM2
antibody epitopes that governed the ability of MDM2 to target p53 for
degradation. These efforts uncovered a carboxy-terminal motif that
contained the now-familiar S-Q target sequence, and Ser 395 in this
sequence was identified as a phosphorylation site for ATM, both in
vitro and in intact cells. It appears that Ser 395 phosphorylation
favors stabilization of p53 by interfering with the shuttling activity
of MDM2, which normally exports p53 out of the nucleus for proteosomal
degradation in the cytoplasm.
These findings demonstrate that ATM establishes multiple regulatory contacts with p53 during the activation of the G1 checkpoint by DNA dsbs (Fig. 3). In contrast, the potential role of ATR in the activation of the G1 checkpoint by any form of genotoxic stress remains relatively obscure. It is conceivable that a parallel ATR-Chk1 pathway also drives p53 accumulation in cells that have incurred IR or UV light-induced DNA damage. Definitive answers to this and many other questions await the development of an ATR-deficient cell line that retains an intact G1 checkpoint. For the time being, the real action with respect to ATR begins with the entry of cells into S phase.
|
| |
The S-phase checkpoint |
|---|
During DNA replication, mammalian cells must be on high alert for
DNA structural abnormalities, such as strand breaks or base modifications that interfere with the accurate copying of the genome.
In addition to the usual array of environmental insults, the process of
DNA replication itself adds intrinsic risks, such as base
misincorporation errors and stalled replication forks, which demand an
immediate response from the checkpoint machinery if genome integrity is
to be preserved. Fortunately, DNA damage detected during S phase is apt
to be repaired precisely via homologous recombination mechanisms
involving sister chromatids (Johnson and Jasin 2000). Indeed, studies
in bacteria suggest that S phase cells rely heavily on homologous
recombination to restart stalled replication forks, even in the absence
of genotoxic agents (Cox 1999; Cox et al. 2000). Nonhomologous repair
mechanisms also play very prominent roles in DNA dsb repair during all
phases of the cell cycle; however, these mechanisms are inherently less
precise, and therefore confer an increased risk that inaccurately
repaired DNA will be carried forward into M phase. Although the
G2 checkpoint should, in principle, catch any cells that have
exited S phase with damaged DNA, these cells may have already missed their
best opportunity to perform error-free repair via homologous recombination.
Given this state of affairs, it comes as no surprise that the S-phase
checkpoint is considerably more multifaceted than this blanket
descriptor implies (Fig. 4). The canonical
checkpoint defect displayed by A-T cells is radioresistant DNA
synthesis (RDS; Painter and Young 1980). In normal cells, exposure to
IR provokes a rapid but reversible decrease in DNA synthesis, which reflects decreases in the rates of both replication origin firing and
DNA strand elongation (Painter and Young 1980). In the absence of ATM,
the IR-induced decrease in DNA synthesis is dampened significantly, giving rise to the RDS phenotype. Treatment of cells with wortmannin, at drug concentrations that abrogate ATM kinase activity, also induces
RDS in normal (ATM-proficient) cells (Sarkaria et al. 1998). Although
loss of ATM function is causally related to RDS, the downstream
effectors in this S-phase pathway checkpoint pathway have largely
eluded identification, that is, until very recently. An intriguing
report now shows that the surveillance functions of the ATM-hChk2
pathway are not confined to G1 phase (Falck et al. 2001). IR
exposure during S phase activates the same pathway, except that, in
this setting, an important outcome is the degradation of Cdc25A, a
protein tyrosine phosphatase that activates cyclin A·cdk2 complexes
as cells transit from G1 to S phase. The results obtained by
Falck et al. (2001) show that hChk2 phosphorylates Ser 123 in Cdc25A,
and that this modification targets Cdc25A for ubiquitin-dependent
degradation. The down-regulation of Cdc25A interferes with the timely
activation of cyclin A·cdk2, which is a requisite event for the
firing of early origins of replication during S phase (Donaldson and
Blow 1999; Takisawa et al. 2000). Genetic manipulations that disrupt
any step in the pathway from hChk2 to cyclin A·cdk2 also give rise to
the RDS phenotype. Remarkably, mutant hChk2 alleles associated with a
variant form of Li-Fraumeni syndrome (Bell et al. 1999) failed to bind
and/or phosphorylate Cdc25A (Falck et al. 2001), which implies that
genetic lesions in the RDS pathway promote genome instability and
cancer development.
|
As is usually the case in signal transduction research, the simple
linear pathway outlined earlier almost certainly oversimplifies the
actual situation in vivo. Overexpression of the previously described
ATRKI mutant in SV40-transformed (but ATM-positive) human
fibroblasts also induces RDS (Cliby et al. 1998). With the caveat in
mind that overexpressed ATRKI may nonspecifically
cross-inhibit ATM function, these results suggest either that ATR also
resides upstream of hChk2, or that ATR regulates a parallel pathway
that contributes to the suppression of DNA synthesis in IR-damaged
cells. By analogy to the p53 phosphorylation mechanism discussed
earlier, it will be interesting to determine whether the
down-regulation of Cdc25A by UV light (Mailand et al. 2000) or HU (if
this occurs) is mediated predominantly through an ATR-dependent
pathway, perhaps via an alternative avenue involving the hChk1 kinase.
A second participant in the DNA damage-induced S-phase checkpoint is
NBS1 (also termed nibrin), the product of the gene mutated in the human
chromosomal instability disorder, NBS (Shiloh 1997; Carney 1999;
Petrini 1999). The clinical features of NBS show considerable but
not complete overlap with those displayed by A-T patients.
Interestingly, NBS cells also display the RDS phenotype, which suggests
that the NBS1 protein is an upstream regulator of Cdc25A stability in
IR-treated cells. The NBS1 protein is found in a complex with two other
genome maintenance proteins, Mre11 and Rad50, which play important
roles in the recombinational repair of DNA dsbs. Treatment of cells
with IR induces the rapid formation of nuclear foci containing the
NBS1-Mre11-Rad50 complex, and a technologically elegant study offered
strong evidence that the foci assemble in close proximity to DNA dsbs
(Nelms et al. 1998). The appearance of these foci is dependent on the
expression of NBS1; however, the absence of major DNA repair defects in
cells from NBS patients indicates that the Mre11-Rad50 complex carries out its repair functions quite capably in NBS1-deficient cells (Petrini
1999). The overlapping clinical phenotypes of A-T and NBS prompted
speculation that the two proteins might be functionally interconnected.
This prediction was validated by reports that ATM phosphorylates NBS1
on up to three serine residues (Ser 343, Ser 397, and Ser 615), and
that Ser 
Ala substitutions at any one of these sites generated a
mutant NBS1 protein that failed to complement the checkpoint defects in
NBS cells (Gatei et al. 2000b; Lim et al. 2000; Wu et al. 2000; Zhao et
al. 2000). At present, we do not understand the exact roles of these
phosphorylation events in the function of NBS1. Nonetheless, the
functional linkage between ATM and the NBS1-Mre11-Rad50 complex
becomes even more compelling with the discovery that hypomorphic Mre11
alleles give rise to an A-T-like disorder in humans (Stewart et al. 1999).
The connection between S-phase checkpoint proteins and human disease
does not end with the NBS1-MRE11-Rad50 complex. Recent findings
position ATM and ATR as critical upstream modulators of the breast
cancer susceptibility protein, BRCA1, which was discussed previously as
the central component of the BASC (Wang et al. 2000). Accumulating
evidence suggests that BRCA1 is a critically important caretaker of the
replicating genome in vertebrate cells. Like the NBS1 complex, BRCA1
participates in both checkpoint and repair pathways in DNA-damaged
cells (Scully and Livingston 2000; Scully et al. 2000; Xu et al. 2001).
Studies of the DNA damage responses in BRCA1-deficient cells are
focusing increasing attention on the role of BRCA1 in promoting
high-fidelity DNA repair through homologous recombination between
sister chromatids (Moynahan et al. 1999; Scully et al. 2000).
Several groups have documented that BRCA1 is phosphorylated in vitro on
multiple sites by the ATM and ATR kinases (Cortez et al. 1999; Chen
2000; Gatei et al. 2000a; Tibbetts et al. 2000). The carboxy-terminal
region of BRCA1 contains numerous S-Q motifs (total of 14), and, of
these potential ATM/ATR target sequences, 10 are localized in an ~300
amino acid stretch (residues 1250-1550) termed the SQ cluster domain
(SCD) (Cortez et al. 1999). Cortez et al. (1999) observed that the
phosphorylation of BRCA1 in IR-damaged cells was significantly impaired
in the absence of ATM, and demonstrated that at least five of the
predicted S-Q target sites in the SCD were phosphorylated by ATM in
vitro. Three of these sites, Ser 1387, Ser 1423, and Ser 1524, were
identified as major IR-induced phosphorylation sites in intact cells
(Cortez et al. 1999; Gatei et al. 2000a). Finally, a BRCA1 double
mutant containing Ala substitutions at Ser 1423 and Ser 1524 showed a
modest impairment in terms of its ability to correct the radiosensitive
phenotype of a BRCA1-deficient breast cancer cell line (HCC1937). The
finding that ATM is an upstream regulator of BRCA1 is consistent with
the highly publicized, but hotly debated, finding that human
ATM heterozygotes are at increased risk for the development of
breast cancer (Meyn 1999). As is the case for p53, ATM makes indirect
as well as direct connections to BRCA1. Once again, the indirect
pathway proceeds through the ATM-regulated checkpoint kinase, hChk2
(J.S. Lee et al. 2000). Phosphorylation of BRCA1 at a single site (Ser
988) by hChk2 terminates an IR-stimulated interaction between these
two proteins in cell nuclei, and is also required for the
radioprotective effect of ectopically expressed BRCA1 in the HCC1937
cell line mentioned earlier.
The earlier studies clearly hinted that at least one additional
BRCA1-directed kinase was activated by DNA-damaging agents, because
significant BRCA1 phosphorylation was observed in cells that lacked
either ATM or DNA-PK (Scully et al. 1997; Cortez et al. 1999; Tibbetts
et al. 2000). ATR was the obvious suspect, and subsequent studies
showed that ATR phosphorylated the carboxyl terminus of BRCA1 at six
Ser/Thr residues in immune complex kinase assays (Tibbetts et al.
2000). The ATR phosphorylation sites partially overlapped with those
modified in vitro by ATM. Studies with phospho-Ser 1423-specific
antibodies indicate that ATM and ATR share responsibilities for the
modification of BRCA1 during IR-, UV-, and replication inhibitor-induced genotoxic stress. As is the case with p53, BRCA1 phosphorylation induced by IR was strongly dependent on ATM, whereas the phosphorylation of this protein in UV- or HU-damaged cells was more
heavily reliant on ATR (Tibbetts et al. 2000). Given the complexity of
the BRCA1 phosphorylation response, and the pleiotropic functions of
this tumor suppressor in checkpoint signaling, BASC assembly, protein
ubiquitination, and homologous recombination (Lorick et al. 1999;
Scully et al. 2000; Wang et al. 2000), a complete understanding of the
interplay between the checkpoint kinases and BRCA1 could take quite
some time.
BRCA1 is the archetypal member of a family of checkpoint repair
proteins that contain the conserved BRCA1 carboxy-terminal (BRCT)
structural motif (Callebaut and Mornon 1997). In addition to BRCA1,
another member of this BRCT domain-containing family has recently been
identified as a substrate for ATM, and likely ATR as well. The p53
binding protein 1 (53BP1) is a nuclear protein that rapidly localizes
to sites of DNA damage in cells treated with IR and other DNA-damaging
agents (Schultz et al. 2000; Rappold et al. 2001). This protein
contains numerous Ser/Thr-Gln sites, and is phosphorylated by ATM in
vitro (Rappold et al. 2001). In intact cells, the phosphorylation of
53BP1 increases rapidly after IR exposure, and this response is
partially dependent on the expression of ATM, with ATR acting as the
presumptive p53BP1 kinase in ATM-deficient cells (Rappold et al. 2001).
Although the function of p53BP1 is unclear, it has been speculated that
this protein is a homolog of the S. cerevisiae Rad9p and
S. pombe Crb2 proteins, which play important roles in multiple
DNA damage-induced checkpoints in these microorganisms (Elledge 1996).
If 53BP1 is indeed the vertebrate counterpart of yeast Rad9p/Crb2, this
protein could be an ATM/ATR substrate of paramount relevance to
checkpoint signaling throughout the cell cycle. The results of DNA damage
response experiments with 53BP1-deficient cells are eagerly anticipated.
To digress a bit, the importance of understanding the molecular
machinery that drives the various S-phase checkpoint pathways cannot be
overstated. Proliferating cells must traverse S phase, and, in spite of
the magnificent accuracy of the DNA replication apparatus, the sheer
magnitude of the task of DNA replication dictates that errors will
inevitably occur with each pass through S phase. If uncorrected, these
errors, and the resultant loss of replication fidelity, may not only
lead to the nucleotide sequence alterations and gross chromosomal
rearrangements traditionally associated with cellular transformation,
but may also favor the aberrant expansion and contraction of repetitive
DNA sequences. Compelling results implicate instability of triplet
repeat DNA sequences in the pathogenesis of neurodegenerative diseases
(Kroutil and Kunkel 1998). The strong connections between DNA
replication fidelity and human disease warrant an intensive effort to
define both the checkpoint and repair mechanisms that ensure the
accuracy with which the genome is duplicated during S phase.
| |
The G2 checkpoint |
|---|
Top
Introduction
The PIKK family
