|
|
||||||||
Institute of Veterinary Pharmacology and Toxicology, University of Zürich, Zürich, Switzerland
1 Correspondence: Institute of Veterinary Pharmacology and Toxicology, University of Zürich, Winterthurerstrasse 260, CH-8057 Zürich, Switzerland. E-mail: naegelih{at}vetpharm.unizh.ch
| ABSTRACT |
|---|
|
|
|---|
Key Words: genome double strand break DNA synthesis eukaryote DNA-PK
| INTRODUCTION |
|---|
|
|
|---|
-rays) and chemicals with radiomimetic properties like the anticancer drug bleomycin or certain warfare agents (reviewed in ref 1
Higher eukaryotes have evolved different defense systems in response to DSBs. In many of these processes, the DNA-dependent protein kinase (DNA-PK) plays a central regulatory role. DNA-PK was first identified in HeLa cell extracts as a serine/threonine kinase that is activated by double-stranded DNA (3
, 4)
. Subsequent studies showed that this activity results from the assembly of a large catalytic subunit (DNA-PKCS) and a Ku70/Ku80 regulatory component with high affinity for DNA ends, which guides the kinase subunit to break sites (5
, 6)
. The biological significance of DNA-PK became evident with the finding that the syndrome of severe combined immunodeficiency (SCID) in mice is caused by a mutation in the DNA-PKCS gene (see, for example, ref 7
). DNA-PKCS mutations lead to SCID in Arabian horses (8)
and Jack Russell terriers (9)
. The syndrome of immunodeficiency in these spontaneous animal models arises from a failure in lymphocyte maturation due to defective V(D)J recombination. The other major phenotypic trait conferred by DNA-PKCS mutations is a severe hypersensitivity to ionizing radiation and radiomimetic chemicals (10)
.
DNA-PKCS and Ku are often considered to represent the subunits of a holoenzyme complex that binds to DSBs and stimulates repair of these detrimental lesions. However, DNA-PKCS and Ku are not constitutively associated and the trimeric factor is assembled only in contact with DNA (11)
, such that it seems intuitive to suggest that each subunit may contribute in a distinct manner to DNA repair or other cellular transactions. The different localization of DNA-PKCS and Ku observed during mitotic stages of human cell lines (12)
lends support to the hypothesis that individual subunits of the DNA-PK complex may carry out, in part, separate functions. Unlike the Ku heterodimer, DNA-PKCS is not essential to rejoin blunt-ended DSBs but becomes indispensable for the repair of more complicated lesions (13)
. DNA-PKCS interacts with several distinct DNA structures other than just DSBs (14
15
16
17
18)
; in fact, the critical binding substrate of DNA-PKCS during V(D)J recombination consists of closed hairpin intermediates rather than open DNA ends (19
, 20)
. In agreement with this versatility in the choice of DNA cofactors, the enzyme has been associated with multiple pathways that modulate stress responses and genome stability. For example, DNA-PKCS constitutes the key player in regulatory circuits that, after a genotoxic insult, can switch the cell fate to apoptotic death (21
, 22)
. Furthermore, DNA-PKCS is required for telomere homeostasis (16
, 23
24
25)
and has been implicated in the regulation of cell cycle progression (26
, 27)
. The purpose of this review is to relate these pleiotropic functions of DNA-PKCS with its biochemical properties and, in particular, with recent results describing its high degree of plasticity as a molecular sensor that recognizes a wide range of abnormal or unusual DNA structures.
| BACKGROUND |
|---|
|
|
|---|
Evolutionary appearance of DNA-PKCS
DNA-PKCS homologs have been identified in all vertebrates examined. No DNA-PKCS equivalents exist in the genome of many lower eukaryotes such as Saccharomyces cerevisiae, Caenorhabditis elegans, or Drosophila, although other members of the PIKK family are present in these organisms (29)
. It has been suggested that the evolutionary breakthrough of DNA-PKCS coincides with the advent of an adaptive immune system in vertebrates and the consequent new requirement for enzymatic factors that resolve hairpin intermediates during V(D)J recombination. However, DNA-PKCS orthologs have been found in Anopheles gambiae and Apis mellifera (30)
, indicating that the need for DNA-PK activity dates back to an evolutionary point before the divergence of arthropods to vertebrates. It is possible that DNA-PKCS arose as part of a more ancient innate immune system, where the function of the enzyme was to detect the characteristic structural features of foreign microbial DNA. Such an intriguing link between DNA-PKCS and the innate arm of immunity, suggesting an alternative evolutionary origin, is prompted by the observation that bacterial DNA or immunostimulatory oligodeoxyribonucleotides can activate DNA-PK, which in turn triggers a phosphorylation cascade that leads to NF-
B activation and production of proinflammatory cytokines. In DNA-PKCS/ macrophages challenged by oligodeoxynucleotides, however, this NF-
B response is abrogated and the release of inflammatory cytokines is reduced (31)
. In any case, the relatively late appearance of DNA-PKCS contrasts with the wide distribution of Ku homologs, which are not only found in all eukaryotes; even prokaryotic organisms possess primitive Ku representatives (6)
. This evolutionary divergence between DNA-PKCS and Ku reinforces the view that the two factors may display distinct activities other than their common involvement in DSB repair and V(D)J recombination.
Domain structure, activation, and autophosphorylation
The gene for human DNA-PKCS (PRKDC) maps to chromosome 8q11. Its complex organization (86 exons) may be a reflection not only of the size of the protein but of its diverse biological roles. The gene product consists of 4129 amino acids with a molecular mass of 470 kDa (28
, 32)
. The catalytic site is located in the C-terminal PIKK domain, which in DNA-PKCS consists of 380 amino acids (Fig. 1
). This PIKK motif is flanked on either side by conserved accessory regions (33)
. The FAT domain (so named because of the homology in this region between FRAP, ATM, and TRRAP) and the FATC domain (FAT at the extreme C terminus) are elements of unknown significance, but the FAT region of ATM is phosphorylated to release active ATM monomers from a dimeric precursor (34)
. In all PIKK family members, the FAT and FATC motifs occur in combination, suggesting that they fold together in a configuration that activates the catalytic site (33)
. DNA-PKCS associates with the Ku partners through a domain (amino acids 3002-3850) that is adjacent to the kinase motif (35)
. In the amino acid region 1503-1550, DNA-PKCS contains a leucine zipper that mediates interactions with C1D, a nuclear matrix-associated factor (15)
. Other partners that interact directly with DNA-PKCS include the Lyn tyrosine kinase (36)
, the c-abl tyrosine kinase (35)
, and KIP (kinase interacting protein), a factor homologous to calcineurin B (37)
.
|
DNA-PKCS has an intrinsic affinity for DNA. In low salt buffer (<100 mM NaCl), the catalytic subunit is activated by DNA even in the absence of Ku or any other regulatory subunit (38)
. When the ionic strength is increased to physiologic levels, however, a regulatory DNA targeting partner is required to recruit the quiescent DNA-PKCS to the nucleic acid cofactor (38
, 39)
. Other nuclear proteins can replace the Ku heterodimer as a molecular matchmaker between DNA-PKCS and its nucleic acid ligands (see below). Upon the addition of ATP to reactions containing DNA-PK and DNA, a wide palette of protein substrates become phosphorylated in vitro. Commonly accepted phosphorylation substrates include DNA-PKCS itself, both Ku antigens, Artemis, p53 protein, the 34 kDa subunit of replication protein A (RPA), histone H2AX, high mobility group (HMG) proteins, the carboxyl-terminal domain of RNA polymerase II, or heat shock protein 90 (20
, 39
, 40)
. The wide range of phosphorylation targets of DNA-PK has recently been extended to RNA binding proteins (41)
. Such a relaxed substrate specificity suggests that the choice of phosphorylation partners may be determined by colocalization in particular cellular foci rather than by characteristic protein sequences.
During incubations with ATP, DNA-PKCS gradually loses its activity due to an inhibitory feedback, involving autophosphorylation, which can be reversed by protein phosphatases (3)
. Different serine/threonine phosphorylation sites have been reported in the DNA-PKCS sequence (42)
. Six of these targets are clustered in a region of 38 amino acids (2609-2647) in the center of the protein (Fig. 1)
; two additional sites reside with the serine residues 2056 and 3205 (43)
. This self-modification reaction is thought to induce structural rearrangements that release DNA-PKCS protein from molecular contact points with the Ku subunits and the DNA substrate (44
, 45)
. The dynamic remodeling of DNA-PKCS by self-modification seems to be crucial for its proper function, because cells in which the central autophosphorylation cluster has been blocked by site-directed mutagenesis are highly radiosensitive and defective in DSB repair (46)
. A substitution of threonine 2609 is sufficient to confer radiation sensitivity and defective DSB repair (47)
, whereas mutation of threonines 2638 and 2647 result in hypersensitivity to ionizing radiation without affecting the DNA end-joining activity (48)
. This phenotypic variability suggests that distinct functions of DNA-PKCS in DNA repair and signal transduction may be uncoupled by replacing individual autophosphorylation sites in site-directed mutagenesis experiments.
| DNA REPAIR FUNCTIONS |
|---|
|
|
|---|
Role of DNA-PKCS in nonhomologous end-joining
Six core factors are required for NHEJ: Ku70, Ku80, DNA-PKCS, XRCC4 (X-ray repair cross-complementing 4), DNA ligase IV, and Artemis. The first step involves recognition of DSBs by the Ku heterodimer. Subsequently, the DNA-bound Ku proteins recruit DNA-PKCS and thereby translocate into the duplex by one helical turn, leaving DNA-PKCS near the DNA terminus (39)
. Electron microscopy studies showed conglomerates of juxtaposed DNA ends connected by two DNA-PKCS molecules (52)
, suggesting that the enzyme can form intermediate synaptic complexes in which the broken DNA fragments are brought together before the ligation takes place. In addition to keeping the DSBs in close proximity, DNA-PKCS may prevent exonucleolytic degradation of the ends (53)
, mediate the alignment of the DNA strands in search for sequence microhomologies (1
, 49)
, and serve as a landing platform for DNA polymerases and ligation factors. The final rejoining of DNA ends is driven by a dimeric factor that consists of DNA ligase IV and XRCC4 (13)
. Participation of DNA polymerase X family members to fill in short gaps prior to ligation is suggested by their colocalization with DNA-PKCS (54)
and direct interactions with Ku and the DNA ligase IV-XRCC4 complex (see, for example, ref 55
). Mutations in the catalytic domain of DNA-PKCS abrogate the ability of the enzyme to support all these different actions in the NHEJ pathway (56)
.
Role of DNA-PKCS in V(D)J recombination
The genetic recombination between V (variable), D (diversity), and J (joining) segments is required in lymphoid cells to form functional immunoglobulin and T cell receptor loci (reviewed in ref 57
). These rearrangements are initiated by the RAG1 and RAG2 (recombination activating gene 1 and 2) proteins, which recognize specific signal sequences and cleave the DNA duplex. The RAG1,2 complex operates by first producing a single nick with a 3'-hydroxyl and a 5'-phosphate end. Subsequently, it mediates a nucleophilic attack by the 3'-hydroxyl residue at the phosphodiester bond opposite to the nick in the complementary strand. The products of this transesterification reaction have two distinct structures: a linear fragment of excised DNA with blunt signal ends, and two covalently closed hairpins at each end of the coding regions that are to be joined. These DNA ends are processed by the NHEJ pathway (57
, 58)
. The lack of DNA-PKCS activity leads to accumulation of hairpin intermediates (59)
, implying that its role in V(D)J recombination is not limited to the final end-joining step but that DNA-PKCS is required for the cleavage of hairpin intermediates. In contrast, mutants lacking either Ku or the DNA ligase IV/XRCC4 complex show a general defect in the processing of blunt as well as hairpin ends.
The discovery of Artemis (named after the Greek goddess of the protection of children), which is mutated in a radiosensitive form of human SCID, clarified the role of DNA-PKCS in hairpin resolution. On its own, the Artemis protein is a 5' to 3' single-stranded DNA exonuclease. It can form a complex with DNA-PKCS in the absence of DNA and Ku (20)
. Presumably by phosphorylation, DNA-PKCS shifts the enzymatic properties of Artemis from exonucleolytic to endonucleolytic, thereby gaining DNA hairpin opening activity. Once cleaved by Artemis, the strands are amenable to further processing by the NHEJ machinery. Because a genetic defect in Artemis causes not only immunodeficiency but radiosensitivity, it became clear that this factor is needed for the repair of radiation-induced DSBs (60)
. In fact, ionizing radiation can break sugar and base rings or leave terminal 5' hydroxyl or 3' phosphate residues, none of which are compatible with direct ligation. It is thought that the resection of overhangs by the Artemis/DNA-PKCS complex converts such intractable termini to regular DNA ends that can be easily rejoined.
| SIGNALING FUNCTIONS OF DNA-PKCS: A MATTER OF CONTROVERSY |
|---|
|
|
|---|
catalytic fragment (63)
The link between DNA-PKCS and apoptosis has been resurrected in another report showing that the thymocytes of DNA-PKCS knockout mice are resistant to p53-dependent bax induction and apoptosis after whole-body treatment with ionizing radiation (66)
. A possible signaling function of DNA-PKCS was examined in more detail using mouse embryonic fibroblasts, which can be sensitized (by expression of the adenovirus E1A oncoprotein) to undergo apoptosis after irradiation. A different end point, cell cycle arrest at the G1/S transition, is observed in the absence of oncoprotein expression. Under conditions that lead to apoptosis, in E1A-sensitized cells DNA-PKCS forms a complex with p53 protein, resulting in p53 phosphorylation. However, when the G1/S arrest is promoted, this association between DNA-PKCS and p53 cannot be detected, indicating that only the apoptotic end point is dependent on DNA-PKCS activity (21
, 67)
. A similar involvement of DNA-PKCS as an upstream regulator of p53 has been demonstrated in cells that are committed to apoptosis after treatment with nucleoside analogs (68)
. Upon exposure to high concentrations (>1 µg/mL) of the antitumor agent cisplatin, DNA-PK generates a molecular death signal that is communicated to neighboring cells through gap junctions (22)
. Other studies have implicated DNA-PKCS in regulatory networks that protect from cell death induced by topoisomerase inhibitors (69)
or heat shock treatment (70)
.
Although it appears that DNA-PKCS is not signaling directly to the cell cycle checkpoint machinery, a role in controlling the progression of S phase is nevertheless suggested by its ability to phosphorylate the 34 kDa subunit of RPA. Two different observations indicate an involvement of RPA in replication arrest after exposure to genotoxic agents. First, RPA becomes hyperphosphorylated in response to DNA lesions occurring during S phase (71)
. Second, the damage-induced inhibition of replication is reversed by the addition of unphosphorylated RPA in an in vitro DNA synthesis assay (72)
. The requirement for DNA-PKCS in S phase arrest was tested directly by comparing the rates of DNA synthesis in DNA-PKCS-proficient M059K and DNA-PKCS-deficient M059J cells (73)
. Replication was down-regulated after UV irradiation in cells expressing DNA-PKCS, but not in the absence of DNA-PKCS. After UV irradiation, the RPA containing fraction from DNA-PKCS-proficient cells poorly supported DNA replication in vitro, whereas the replication activity of the matching DNA-PKCS-deficient control was not affected. These findings are consistent with the hypothesis that RPA constitutes the direct target for a DNA-PKCS-mediated control circuit that induces S phase arrest.
Transcription factor E2F-1 is responsible for the induction of numerous proteins that promote G1/S progression and thus represents another attractive regulatory target for cell cycle control. In mouse SCID cells, the transcriptional activity of E2F-1 is enhanced relative to normal counterparts, and the level of cyclin A, cyclin E, and b-Myb oncoprotein tends to be higher in SCID cells than in comparable controls (26)
. Differential regulation of pro- and antiapoptotic factors has been implicated in the higher staurosporine resistance of DNA-PKCS-deficient M059J cells relative to the DNA-PKCS-proficient M059K counterparts (74)
. DNA-PK participates in modulating transcription profiles by the phosphorylation of octamer transcription factor 1 (Oct1), leading, for example, to down-regulation of histone H2B expression (27)
. The basal activity of RNA polymerase I (75)
and RNA polymerase II (76
, 77)
can be modulated directly by the action of DNA-PK. Thus, DNA-PKCS may alter the function of key transcriptional processes governing cell cycle progression.
| ROLE OF DNA-PKCS IN TELOMERE HOMEOSTASIS |
|---|
|
|
|---|
| UNANSWERED QUESTIONS |
|---|
|
|
|---|
Another low-resolution study using electron crystallography suggests multiple interaction sites with nucleic acids, including two channels that could accommodate duplex DNA and several openings just large enough to permit the insertion of single-stranded DNA (85)
. On the basis of these findings, a dual model has been proposed for the interaction of DNA-PKCS with DNA termini. The first component consists of the binding of double-stranded DNA using one of the two channels; the second involves threading of single-stranded DNA from the substrate ends into the adjacent openings. Consistent with this dual binding model, DNA-PKCS activity is enhanced when unpaired single-stranded overhangs are added to blunt DNA ends (Fig. 2
A), thus indicating that DNA-PKCS is positioned near break sites where the DNA ends melt to form a transition fork between normal duplex and partially denatured configurations (82)
. These structural analyses have to be refined in future studies to provide more detailed insights into the molecular mechanism of DNA ligand recognition.
|
Interaction of DNA-PKCS with hairpin structures
Conflicting results have been published regarding the modality of binding to hairpin ends. An initial study (14)
showed that the DNA cofactors capable of activating DNA-PKCS include covalently closed "dumbbell" constructs consisting of a central duplex region and terminal single-stranded hairpin loops of four nucleotides (Fig. 2B
). However, such dumbbell substrates do not mimic the perfect hairpin intermediates generated by the RAG1,2 complex. Activation of DNA-PK by a homoduplex DNA molecule containing two hairpin ends (Fig. 2C
) was tested in the presence of a peptide substrate that serves as heterologous phosphorylation target. In this experimental setup, hairpin-ended DNA turned out to be an extremely poor cofactor for protein kinase activity, even though electrophoretic mobility shift assays demonstrated that the trimeric DNA-PK complex is able to assemble on such hairpin structures (58)
. These findings prompted the hypothesis that DNA-PK may be activated during V(D)J recombination by the adjacent open signal ends, which are generated as a by-product of RAG1,2 cleavage.
Direct activation of DNA-PK by a homoduplex DNA with perfect hairpin ends (Fig. 2C
) has been documented by Soubeyrand et al. (19)
. Unexpectedly, their artificial hairpin structure, consisting of 43 base pairs, induced rapid autophosphorylation of DNA-PKCS whereas phosphorylation of heterologous protein substrates was suppressed. The lack of heterologous phosphorylation may explain the apparently dissenting results of the earlier study by Smider et al. (58)
, who concluded that hairpin-ended DNA molecules constitute a binding substrate for DNA-PK but fail to stimulate its kinase activity. One limitation of these experiments performed with linear homoduplex substrates is the presence of two closely spaced hairpins, which may encourage the autophosphorylation of DNA-PKCS by promoting the simultaneous binding of two enzyme molecules on the same substrate. As a consequence, a more physiologic structure was prepared that contains only a single hairpin. The other end of the substrate was physically blocked by attachment to streptavidin beads (Fig. 2D
). In the presence of such immobilized oligonucleotides exposing only one hairpin end, autophosphorylation of DNA-PKCS was reduced while phosphorylation of a heterologous protein substrate was increased, confirming that a perfect hairpin structure, as it is generated in V(D)J recombination, can indeed trigger the DNA-PK reaction (19)
. Whether DNA-PKCS adopts a distinct protein configuration for its anchorage to hairpins compared with the complex with open DNA ends will be the subject of future structural studies.
Interaction of DNA-PKCS with other specialized DNA architectures
Beyond the interaction with DSBs and hairpin ends, DNA-PKCS displays an affinity for other specialized DNA configurations, including structured single-stranded DNA (19)
, long single-stranded gaps (14)
, sequences for nuclear matrix attachment, DNA kinks, telomeres, and 4-way junctions (see below), that interrupt the standard double helical conformation of DNA. The mechanism by which DNA-PKCS recognizes these structural irregularities has not been studied in detail, but its general DNA binding properties may be inferred from the way it interacts with DSBs (39)
. It has become clear, for example, that DNA-PKCS makes direct contacts with DNA, such that the catalytic subunit and its regulatory partners assume separate positions on the nucleic acid cofactor. Second, it appears that the DNA-PKCS monomer is endowed with multiple binding modules for the interaction with nucleic acids of variable helix complexity (85)
. Third, the folding of DNA-PKCS seems to be highly flexible as the enzyme undergoes substantial domain rearrangements upon binding to the DNA cofactor (86)
.
DNA-PKCS can be recruited to specific sites of action by partner proteins that bind to undamaged DNA in the absence of any DSBs or hairpin ends. Such a DNA damage-independent activation mechanism was revealed with the discovery that DNA-PKCS interacts with C1D, an 18 kDa matrix-associated nuclear protein. The nuclear matrix is thought to consist of a 3-dimensional protein scaffold within the nucleus that is required for the organization of DNA into higher order chromosomal domains. The interaction with C1D has been identified in a two-hybrid screen using the leucine zipper domain of DNA-PKCS as the bait (15)
. Additional studies demonstrated that C1D mediates the activation of DNA-PKCS on supercoiled plasmid DNA, without any requirement for DSBs and that C1D is an excellent heterologous target for its protein kinase activity. C1D protein is induced upon
-irradiation (15)
, and overexpression of C1D induces apoptosis (87)
. Further insights were obtained from gene targeting experiments in S. cerevisiae that demonstrate an involvement of C1D protein in NHEJ and HR (88)
. Taken together, these observations suggest that C1D may enforce a genotoxic stress response by facilitating recruitment of DNA-PKCS and other repair complexes to the nuclear matrix.
Another link to the nuclear matrix is provided by the preferential association of DNA-PKCS with plasmid DNA constructs containing matrix attachment region (MAR) elements (17)
. Furthermore, it has been reported that DNA-PK colocalizes with topoisomerase II and poly(ADP-ribose) polymerases at MAR sites in nuclear fractions of rat testis (89)
. These MAR sequences have been defined on the basis of their ability to connect chromosomal loops to the nuclear matrix (90)
. Because of their high AT content, MAR elements become stably base-unpaired under superhelical stress and undergo spontaneous bending with varying degrees of curvature (Fig. 2E
). The observation that HMG1 and HMG2 proteins promote the recruitment of DNA-PKCS to DNA and stimulate its kinase activity supports the conclusion that DNA-PKCS has an intrinsic affinity for bent DNA configurations (91)
. HMG proteins are abundant chromatin components that associate with the nuclear matrix and bind DNA without sequence specificity, but have a high preference toward kinked geometries. In addition, they are able to distort linear DNA and one prominent process in which HMG-induced bending has a biological role is V(D)J recombination. In fact, the recruitment of HMG proteins by the RAG1,2 complex results in site-specific DNA bending and is thought to ensure the correct spacing of signal sequences in obedience to the "12/23 rule" (49
, 57)
. The hypothesis that DNA-PKCS has a general affinity for unusual DNA architectures is further strengthened by the observation that the enzyme is activated by 350 base pair minicircles containing no DSBs (14)
. Presumably, these very short circular constructs impose a structural abnormality on the DNA double helix, making it a substrate for interactions with DNA-PKCS.
Another type of architecturally specialized DNA conformation is encountered at the telomeric ends of eukaryotic chromosomes, which culminate in a short single-stranded overhang. For most of its length, the telomeric DNA consists of tandem repeats of guanine-rich sequences that are susceptible to the formation of unusual quadruplex structures (Fig. 2F
) due to the propensity of guanine bases to associate with each other in a stable hydrogen-bonded arrangement (92)
. Localization of Ku at chromosomal ends is through its interaction with TRF1 (telomeric repeat binding factor 1), without direct binding to the telomeric DNA itself (78)
. The association of DNA-PKCS with telomere ends has been shown by chromatin immunoprecipitation studies (16)
. It is not yet clear whether DNA-PKCS interacts with the telomere DNA repeats or whether, as in the case of Ku, this localization is mediated by association with other protein subunits. However, the kinase activity of DNA-PKCS is required for telomere homeostasis, implying that the enzyme is stimulated through direct molecular contacts with the DNA at this particular site; it seems likely that DNA-PKCS is able to accommodate the narrow DNA turns encountered in quadruplex conformations, as these structural repeats have a strong resemblance to closed hairpin ends (compare Fig. 2C, F
). It is not known whether DNA-PKCS phosphorylates telomere binding proteins in order to modulate their function or, alternatively, if the kinase activity of DNA-PKCS is required to maintain its localization in competition with other factors associated with chromosome ends.
Interaction of DNA-PKCS with Holliday junctions
In view of the evidence for DNA-PKCS being a sensor of specialized DNA architectures, we constructed a series of photoreactive cross-linking probes to explore the range of structural DNA discontinuities that are recognized by DNA-PKCS (18)
. Surprisingly, this survey showed that DNA-PKCS associates preferentially with the crossover region of Holliday junctions, the universal HR intermediate (Fig. 3
). Direct comparisons revealed that DNA-PKCS binds to synthetic 4-way junctions with an affinity like that to open DNA ends. This interaction with 4-way junctions was enhanced when biotin-streptavidin complexes were assembled at the end of each double-helical stem of the 4-way junction molecule. Instead, binding of the Ku heterodimer was suppressed by obstruction of the DNA ends (Fig. 3)
. The association of purified DNA-PKCS with synthetic Holliday junctions was salt labile even in the presence of Ku proteins. However, salt resistance was conferred to the complex when the cross-linking probe was incubated with nuclear extracts of human cells, indicating that the binding of DNA-PKCS to 4-way junctions is stabilized in the context of an endogenous protein environment. Protein kinase assays, using p53 as a phosphorylation target, showed that DNA-PKCS adopts an active conformation in the complex with 4-way junctions (18)
.
|
The recognition of Holliday junctions by DNA-PKCS may simply reflect a general affinity for profoundly kinked or bent DNA structures, as outlined in the previous section. A structural similarity between 4-way junctions and the DNA at the entry and exit point of nucleosomes might be indicative of a constitutive interaction of DNA-PKCS with chromatin filaments. In a larger perspective, however, the remarkable affinity of this enzyme for 4-way crossings bears on the question of how vertebrate cells control the choice between the repair of DSBs by NHEJ or HR. In fact, there might be passive or active modes of competition between the two diverging pathways. With passive competition, the repair outcome may simply depend on whether NHEJ or HR proteins bind first to the ends of broken chromosomes. With active competition, the factors involved in NHEJ and HR may interact directly and influence each others function on the same DSB site. It has been suggested, for example, that NHEJ is the initial default pathway for DSB repair and, when NHEJ fails, HR may take over (93
, 94)
. Thus, the affinity of DNA-PKCS for the crossover region of Holliday junctions could anticipate an unexpected backup mechanism for the down-regulation of improper recombination events. It is conceivable that suppression of HR by DNA-PKCS can occur by phosphorylation of proteins involved in branch migration or Holliday junction resolution. Such an active regulatory role of DNA-PKCS appears to be important for the resetting of stalled or blocked replication forks, as indicated by the observation that spontaneous HR rates, which depend on DNA replication, are inhibited by DNA-PKCS more efficiently than strand exchanges after exogenously induced DSBs (50)
. A possible 4-way DNA substrate arising during DNA replication includes the chicken foot-like intermediate formed by backward migration of stalled replication forks. This transient 4-way intermediate is supposed to facilitate the replicative bypass of damaged templates before the correct replication fork is regenerated (2)
. The recognition of chicken foot intermediates by DNA-PKCS could prevent branch migration in the wrong direction or avoid the endonucleolytic resolution of this particular kind of Holliday junction. This hypothesis, although speculative, is supported by the rapid recruitment of DNA-PKCS to sites where replication forks run across a damaged template (95)
. Homologous strand exchanges between repetitive sequences give rise to gene amplification events and, hence, may generate another form of detrimental junction that is subject to resolution by DNA-PKCS. This potential function in safeguarding genome stability is suggested by the observation that cells lacking DNA-PKCS display increased gene amplification rates (51
, 96)
.
| PROSPECTS AND PREDICTIONS |
|---|
|
|
|---|
In conjunction with the Ku heterodimer, DNA-PKCS recognizes DSBs and coordinates the NHEJ machinery. The structural basis of Ku binding to DNA has been solved in detail. The two subunits, Ku70 and Ku80, fold into a rigid toroidal complex with an exquisite affinity for DNA ends (97)
. In contrast, the nucleic acid binding mechanism of DNA-PKCS is not yet understood, but it appears that this factor constitutes a more flexible sensor of abnormal or unusual DNA structures. DSBs may represent just one form of structural perturbation that attracts DNA-PKCS and stimulates its enzymatic activity. Besides DNA breaks, the enzyme is able to recognize many other DNA architectures that interrupt the standard Watson-Crick double helix, including hairpins, single-stranded loops, telomeric repeats, 4-way junctions, DNA bends, and the DNA of invasive microorganisms. In view of this extraordinary binding versatility, it is tempting to speculate that DNA-PKCS may act as a molecular "caliper" of specialized, unusual, or abnormal DNA structures that violate the canonical Watson-Crick geometry of the linear double helix.
Upon the specific recognition of its different DNA ligands, DNA-PKCS adopts a dual role in DNA repair and stress signal transduction and hence cooperates with other components of the genome surveillance network to determine the cell fate. Repair activities of DNA-PKCS induced by structural DNA discontinuities include NHEJ (to restore intact chromosomes), the dissolution of end-to-end fusions (to maintain proper telomere ends), as well as the resolution of detrimental 4-way junctions or other strand exchange intermediates that compromise genomic stability (Fig. 4
). Possible signaling end points of DNA-PKCS, again triggered by the recognition of structural DNA discontinuities, include the induction of apoptotic cell death, cell cycle arrest, or the stimulation of innate immune responses (Fig. 4)
, thus placing DNA-PKCS at the crossroad between diverse pathways that respond to stress stimuli. One fundamental issue that will continue to be the focus of considerable attention is the mechanism through which cells sense DNA structural perturbations in a manner that leads to differential activation of either DNA repair, inhibition of end-to-end fusions, cell cycle checkpoints, or apoptosis. Research on this unresolved but intriguing aspect is still feverishly ongoing. In this respect, the molecular elucidation of higher order complexes formed by DNA-PKCS with distinct DNA cofactors will be a major goal for future studies. It seems likely that the interaction of this large and versatile sensor with each type of DNA ligand leads to different conformational changes, which in turn may result in distinct protein foldings that specifically drive separate reaction pathways.
|
| ACKNOWLEDGMENTS |
|---|
Received for publication November 3, 2004. Accepted for publication December 22, 2004.
| REFERENCES |
|---|
|
|
|---|