Phosphorylation of threonine 68 promotes oligomerization and autophosphorylation of the Chk2 protein kinase via the forkhead-associated domain.

Phosphorylation of Thr-68 by the ataxia telangiectasia-mutated is necessary for efficient activation of Chk2 when cells are exposed to ionizing radiation. By an unknown mechanism, this initial event promotes additional autophosphorylation events including modifications of Thr-383 and Thr-387, two amino acid residues located within the activation loop segment within the Chk2 catalytic domain. Chk2 and related kinases possess one or more Forkhead-associated (FHA) domains that are phosphopeptide-binding modules believed to be crucial for their checkpoint control activities. We show that the Chk2 FHA domain is dispensable for Thr-68 phosphorylation but necessary for efficient autophosphorylation in response to ionizing radiation. Phosphorylation of Thr-68 promotes oligomerization of Chk2 by serving as a specific ligand for the FHA domain of another Chk2 molecule. In addition, Chk2 phosphorylates its own FHA domain, and this modification reduces its affinity for Thr-68-phosphorylated Chk2. Thus, activation of Chk2 in irradiated cells may occur through oligomerization of Chk2 via binding of the Thr-68-phosphorylated region of one Chk2 to the FHA domain of another. Oligomerization of Chk2 may therefore increase the efficiency of trans-autophosphorylation resulting in the release of active Chk2 monomers that proceed to enforce checkpoint control in irradiated cells.

The maintenance of genomic integrity following DNA damage requires the coordinated actions of DNA repair and cell cycle checkpoint control. The Chk2/hCds1 protein kinase is activated by DNA damage and phosphorylates several known modulators of cell cycle control including the tumor suppressor proteins, p53 and BRCA1, and Cdc25A phosphatase (1)(2)(3)(4)(5)(6)(7)(8)(9). Mutations in the CHK2 gene have been identified in human hereditary and sporadic cancers suggesting that post-translational modifications mediated by Chk2 play important roles in altering the activities of these checkpoint control proteins (10 -14). Therefore, exploring the mechanisms by which Chk2 activity is regulated will enhance our understanding of the com-plex network of signaling pathways that serve to limit tumor development.
Chk2 is a direct target and major effector of the ATM 1 kinase, a key regulator of cell cycle checkpoint control in irradiated cells (for review see Ref. 15). In response to ionizing radiation (IR) and double-stranded DNA breaks, ATM phosphorylates Chk2 on Thr-68 within the amino-terminal SQ/TQrich domain, and this event is necessary for efficient activation of the Chk2 kinase (16 -18). Recent evidence suggests that Chk2 is phosphorylated on Thr-383 and Thr-387 in a phosphothreonine 68 (Thr(P)-68)-dependent manner (19). These residues are located within the activation loop segment of Chk2 and therefore may be Chk2 autophosphorylation sites that increase the specific activity of the enzyme when modified (19,20). This model is consistent with our observations that a catalytic inactive Chk2 fails to exhibit phosphorylationdependent mobility shifts upon SDS-PAGE even though clearly phosphorylated on Thr-68, suggesting that additional phosphorylation events are dependent upon a functional Chk2 catalytic domain and necessary for full activation (18). How Thr-68 phosphorylation specifically leads to Chk2 autophosphorylation and activation is currently unknown.
Additional clues as to how DNA damage-dependent Chk2 phosphorylation might be regulated are revealed through structural and biochemical studies of its two yeast homologues, Saccharomyces cerevisiae RAD53 and Schizosaccharomyces pombe Cds1. In addition to an amino-terminal SQ/TQ-rich domain, Chk2-related kinases like Cds1 and RAD53 contain one or two FHA (forkhead-associated) domains (for review see Ref. 15). Current structural and biochemical data indicate that FHA domains mediate phosphospecific protein-protein interactions and are thus considered important signaling moieties (21)(22)(23)(24). This is exemplified in RAD53, where the carboxylterminal FHA domain binds to DNA damage-induced phosphorylated forms of RAD9, and this interaction is required for DNA damage-induced checkpoint functions in budding yeast (21). This interaction may also be crucial for enhancing RAD53 activity in DNA-damaged cells because phosphorylated RAD9 serves as a scaffold for RAD53 binding via the RAD53 FHA domain and thereby facilitates RAD53 trans-autophosphorylation and activation (25). Therefore, based on the structural and functional similarities between Chk2 and RAD53, the Chk2 FHA domain may play an important role in the regulation of Chk2 activity in irradiated cells. The identification of mutations within the FHA domain of Chk2 in several variant Li-Fraumeni syndrome (LFS) families and sporadic cancers strengthens the concept that this domain is important for Chk2 function (10,11,26).
It has been reported previously that the Chk2 FHA domain is required for DNA damage-dependent Thr-68 phosphorylation and activation of Chk2 (19). Thus far, only the LFS-associated mutations (I157T or R145W) within the FHA domain have been characterized as to their ability to be phosphorylated on Thr-68 and activated by IR. Others have shown (27) that the I157T mutant is efficiently phosphorylated on Thr-68 and activated by IR suggesting that this mutation does not affect upstream signaling to Chk2. Unlike the I157T mutant Chk2, the R145W mutant is poorly Thr-68-phosphorylated and activated by IR consistent with this mutation abrogating Chk2 function (19,27). However, this mutant is also very unstable and may exist in a large, inactive complex that renders it inaccessible to upstream signaling cascades initiated by DNA damage (26,27). Therefore, it is difficult to interpret data obtained with the R145W mutant because it is unclear whether this mutation solely affects FHA domain function or induces a more drastic conformational change that alters the entire structure of Chk2.
We created a series of defined mutations within the FHA domain of Chk2 to further address whether it is necessary for the activation of Chk2 by ATM. Individual point mutations of theoretical phosphopeptide-binding residues within the Chk2 FHA domain did not alter the stability of the protein nor the ability of ATM to phosphorylate Thr-68. Instead, these mutations negatively affected autophosphorylation. We also demonstrate that the FHA domain of Chk2 binds specifically to the SQ/TQ-rich region of Chk2 in a Thr(P)-68-dependent manner. This association in turn may promote autophosphorylation at multiple sites including the FHA domain resulting in dissociation of this complex. Together, these data suggest that Thr-68 phosphorylation by ATM promotes association of two or more Chk2 molecules by enhancing the affinity of the SQ/TQ-rich domain of one Chk2 molecule for the Chk2 FHA domain of another. Oligomer formation may then facilitate trans-autophosphorylation and activation of Chk2 in irradiated cells.

MATERIALS AND METHODS
Antibodies-Rabbit polyclonal antibodies (Zymed Laboratories Inc.) specific to phospho-Thr-68 of Chk2 were raised against a peptide containing ETVST(-PO 4 )QELYS and purified using antigen peptide conjugated to Sepharose (Hartwell Center for Bioinformatics and Biotechnology at St. Jude Children's Research Hospital). Antibodies were further purified by passage through Thr-68 peptide (ETVSTQELYS)-Sepharose to deplete contaminating antibodies that are reactive with Chk2 not phosphorylated at Thr-68. Commercial antibodies used in this study included monoclonal anti-FLAG M5 (Sigma), monoclonal anti-GST and c-Myc (Roche Molecular Biochemicals), and polyclonal goat anti-Chk2 N-17 (Santa Cruz Biotechnology).
Cell Culture-293T/17 and HCT-15 (which contains a heterozygous R145W mutation within the FHA domain of Chk2 (10)) cells were obtained from the American Type Culture Collection and grown in Dulbecco's modified Eagle's media containing 10% fetal calf serum. For retrovirus production, 293T/17 cells were co-transfected with pBabe/ puro FLAG Chk2 and the amphotropic pEQPAM3 packaging construct using calcium phosphate. After 48 h, media were collected, and HCT-15 cells were exposed three times to retrovirus in the presence of 4 g/ml Polybrene. After 2 days, retrovirus-infected cells were selected in 10 g/ml puromycin and used for the further experimentation. Where indicated, cells were exposed to ionizing radiation from a 137 Cs source delivered at a dose rate of ϳ1.2 Gy/min.
Plasmid Constructs-The pET-15b containing full-length wild type (wt) and catalytic inactive (D368N) Chk2 (hsCds1) cDNAs were kindly provided by H. Piwnica-Worms (28). pGEX-Chk2-(1-80) has been described previously (18) and consists of the first 80 amino acids of Chk2 comprising the SQ/TQ-rich domain fused to GST. cDNA fragments containing the FHA domain (amino acids 60 -225) or catalytic inactive kinase domain (amino acids 222-543, D368N) of Chk2 were amplified by PCR and cloned into the pGEX4T-1 (Amersham Biosciences) and/or pET-28a (Novagen). Site-directed mutagenesis of pSG5-FLAG Chk2 (18) was performed by a two-step overlap PCR approach using oligonucleotide primers containing the desired mutation and primers corresponding to the 5Ј and 3Ј end of Chk2. FLAG-tagged wild type and mutant Chk2 cDNAs were excised from pSG5 and subcloned into the pBabe/puro retrovirus vector. All oligonucleotide synthesis and plasmid sequence analyses were performed by the Hartwell Center for Bioinformatics and Biotechnology at St. Jude Children's Research Hospital.
GST-FHA Binding Assay-Whole cell extracts were prepared from 293T/17 cells or HCT-15 cells expressing wild type or T68A mutant FLAG-tagged Chk2 in NTEN buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM NaF, 1 mM Na 3 VO 4 , and protease inhibitors). One mg of whole cell extract was incubated with 5 g of GST-FHA pre-bound to glutathione-agarose for 2 h at 4°C. Bound fractions were washed three times with NTEN buffer and then analyzed by SDS-PAGE.
Peptide Binding Assay-ETVST(-PO 4 )QELYS (Thr(P)-68) and ETVSTQELYS (Thr-68) peptides were synthesized and conjugated to Sepharose at a concentration of 15 mol of peptide/g (Hartwell Center for Bioinformatics and Biotechnology at St. Jude Children's Research Hospital). E. coli expressing GST, GST-FHA, or GST-FHA containing the N166A mutation (GST-N166A) was lysed in STE buffer containing 2% Triton X-100, protease inhibitors, and 1 mM dithiothreitol. Four mg of each lysate was incubated with 25 l of Thr-68 or Thr(P)-68-Sepharose at 4°C for 2 h. Mammalian whole cell extracts were prepared in NTEN buffer as above. One mg of each extract was incubated with peptide beads for 2 h at 4°C. The bound fraction of each extract was washed three times with STE (E. coli) or NTEN (mammalian) buffer and then analyzed by SDS-PAGE.
In Vitro Kinase Assay-Reactions were performed in kinase buffer (10 mM Hepes, pH 7.5, 10 mM MgCl 2 , 0.5 mM EGTA, 10 M ATP, 1 mM dithiothreitol, and 10 Ci of [␥-32 P]ATP) for 30 min at 30°C and contained 1 g of purified His-tagged wild type (wt) or catalytic inactive (kd) Chk2 plus 2 g of purified protein substrates. Kinase reactions were then subjected to 10% SDS-PAGE and transferred to nitrocellulose. Radiolabeled proteins were detected by PhosphorImager analysis (Molecular Dynamics). For binding assays using Chk2-phosphorylated GST-FHA, 10 g of GST-FHA bound to glutathione-agarose was incubated with wt or kd Chk2 in the same buffer without [␥-32 P]ATP for 30 min at 30°C. Bound GST-FHA was washed three times with NTEN buffer and once with NTEN buffer containing 500 mM LiCl. One-half of each sample was directly used for the GST-FHA binding assay, whereas the other half was eluted with 10 mM glutathionine and used for the peptide binding assay. Both assays were performed as described above.
Immunoprecipitation Assays-293T/17 cells were transiently cotransfected with SG5 FLAG-and Myc-tagged Chk2 constructs (18) using LipofectAMINE reagent (Invitrogen). Thirty six hours after transfection, cells were exposed to 0 or 10 Gy IR, harvested 30 min later, and then lysed in NTEN buffer. Extracts were clarified by centrifugation at 14,000 rpm for 20 min at 4°C, and soluble fractions were collected and diluted with an equal volume of 20 mM Tris-Cl, pH 8.0, and 1 mM EDTA. One mg of diluted whole cell extract was incubated with anti-FLAG (M2)-agarose (Sigma) or 10 g of anti-Myc antibody plus 25 l of protein A/G-agarose (Oncogene) for 2 h at 4°C. Immunoprecipitated proteins were washed with 0.5ϫ NTEN buffer and then analyzed by SDS-PAGE.

RESULTS AND DISCUSSION
The FHA Domain of Chk2 Is Necessary for Efficient Autophosphorylation in Response to IR but Not Necessary for Thr-68 Phosphorylation-A series of individual mutations within the FHA domain were generated to determine the role of this domain in IR-induced autophosphorylation of Chk2. The crystal structure of the first FHA domain of RAD53 bound to a phosphothreonine-containing peptide reveals that this complex consists of a multiple strand ␤-sandwich containing short ␣-helical loops that extend out and make contact with a phosphothreonine-containing peptide (24). These structural data were used to produce a modified sequence alignment of several FHA domains including Chk2 (24). Based on this information, we created mutations at conserved residues that are located within the ␤-sandwich (G116E, H143A, and R145W) or proposed ␣-helical loops that are expected to make contact with a phosphopeptide ligand (R117A, S140A, K141A, N166A, and I157T) (Fig. 1A). Mutations of residues believed to form the ␤-sandwich structure of the Chk2 FHA domain (G116E and H143A) appeared to destabilize the protein (data not shown). These results are similar to what has already been documented for the LFS-associated R145W (10, 26, 27) and suggest that mutations located within the ␤-sandwich are likely to destabi-  ). B, whole cell (WCE) were incubated with GST, GST-FHA, or GST-FHA containing the N166A mutation (GST-N166A) as above. Whole cell extracts or Chk2 bound to various GST fusion proteins were immunoblotted with anti-Chk2. The lower right panel shows the total amount of GST fusion protein subjected to each binding reaction as determined by fast green staining of nitrocellulose membrane. C, HCT-15 cells expressing wild type (wt) or T68A FLAG-tagged Chk2 protein were exposed to 0 or 6 Gy IR, harvested 30 min later, and used for the GST-FHA pulldown assay as in A. A fraction of each WCE was immunoblotted with either anti-FLAG or anti-Thr(P)-68 (top panels). Bound fractions were immunoblotted with either anti-FLAG or anti-GST (bottom panels).
Thr-68 Phosphorylation Promotes Chk2 Oligomerization lize Chk2 protein. Mutations within the ␣-helical loops did not affect protein stability, and these Chk2 mutants were therefore used for further investigation.
To determine whether mutations expected to disrupt phosphopeptide binding by the Chk2 FHA domain alter Chk2 activation by IR, we infected HCT-15 cells with retrovirus encoding various FLAG-tagged mutant cDNAs of Chk2. This cell line was chosen because endogenous Chk2 levels and activity are undetectable, and HCT-15 cells do not display normal Chk2 function (9,27). FLAG-tagged wild type Chk2 protein exhibited the characteristic phosphorylation-dependent mobility shift upon SDS-PAGE that correlates with kinase activation following DNA damage (1, 2) (Fig. 1B). Consistent with previous reports (16 -18, 27), the T68A mutant was defective in this response, and the LFS-associated I157T mutant was hyperphosphorylated just as efficiently as wild type Chk2. Mutation of the highly conserved Arg-117, Ser-140, or Asn-166 residue to alanine was the most disruptive to DNA damage-induced phosphorylation of Chk2, whereas mutation of the less conserved Lys-141 residue to alanine had the least impact. All mutants were phosphorylated on Thr-68 suggesting that the integrity of the FHA domain is not necessary for ATM-dependent phosphorylation of Chk2. These observations contrast those observed for the unstable R145W mutant (19,27). Interestingly, the catalytic inactive (kd) Chk2 protein containing the D368N mutation did not exhibit a mobility shift even though it showed strong Thr-68 phosphorylation after DNA damage. Therefore, changes in mobility of Chk2 protein during SDS-PAGE are caused by additional phosphorylation events that depend upon intrinsic kinase activity and an intact FHA domain but is independent of Thr-68 phosphorylation.
In budding yeast, the FHA2 domain of RAD53 interacts with RAD9 in both a DNA damage-and phosphorylation-dependent manner, and this interaction facilitates trans-autophosphorylation and activation of RAD53 (21,25). Like RAD9, mammalian BRCA1 and 53BP1 contain carboxyl-terminal tandem BRCT (BRCA1 carboxyl terminus) motifs, and it has been speculated that either protein could be a functional homologue of RAD9 (29). Of interest, both BRCA1 and 53BP1 can associate with and/or co-localize with Chk2 and therefore may facilitate Chk2 activation by serving as a scaffold for Chk2 transautophosphorylation (8,30). Cells expressing mutant BRCA1 (HCC1937 (31)) efficiently activated Chk2 in response to IR (data not shown) establishing that BRCA1 is not required for IR-induced activation of Chk2. Furthermore, our attempts to determine whether Chk2 and 53BP1 directly interact through co-immunoprecipitation studies were unsuccessful (53BP1 reagents were kindly provided by J. Chen). Therefore, other BRCA1-and 53BP1-independent mechanisms may exist that facilitate Chk2 autophosphorylation in irradiated cells.
Chk2 Specifically Interacts with Its Own FHA Domain following Modification by DNA Damage-We considered the possibility that autophosphorylation is facilitated through direct oligomerization of Chk2 which is dependent upon both Thr-68 phosphorylation and a functional FHA domain. Binding assays were performed using recombinant GST protein fused to the Chk2 FHA domain (GST-FHA) as bait and whole cell extracts for a source of Chk2 protein to determine whether Chk2 directly interacts with its own FHA domain. Chk2 protein present in whole cell extract prepared from irradiated cells specifically bound to GST-FHA immobilized on glutathione-agarose ( Fig. 2A). The interaction between Chk2 and GST-FHA was decreased if extract made from irradiated cells was pretreated with protein phosphatase consistent with the findings that protein interactions with FHA domains are phosphorylationdependent (23) (Fig. 2A). Therefore, binding of Chk2 from ir-radiated cells to GST-FHA required DNA damage-induced phosphorylation in addition to a functional FHA domain because mutating a highly conserved residue within the FHA domain of Chk2 (GST-N166A) abolished specific binding of Chk2 obtained from irradiated cells (Fig. 2B).
A Peptide Representing the Phosphothreonine 68 Region of Chk2 Is a Ligand for the Chk2 FHA Domain-Thr-68 phosphorylation by ATM is believed to be the first event in the pathway leading to full activation of Chk2 in irradiated cells. To determine whether Thr-68 phosphorylation was required for Chk2 binding to GST-FHA, we prepared whole cell extracts from HCT15 cells expressing FLAG-tagged wild type or T68A Chk2 protein and subjected them to the in vitro GST-FHA binding assay. T68A mutant Chk2 protein present in irradiated cells failed to interact with GST-FHA suggesting that phosphorylation of Thr-68 is necessary for this DNA damage-dependent interaction (Fig. 2C). To confirm that the FHA domain of Chk2 can specifically bind to Thr(P)-68 Chk2, additional binding assays were performed using extracts made from E. coli ex- were incubated with Thr-68 peptide or phosphothreonine 68 (P-T68) peptide conjugated to Sepharose. Bound fractions were washed, separated by SDS-PAGE, and transferred to nitrocellulose. The Thr(P)-68-peptide-Sepharose interacting proteins were visualized by fast green staining of nitrocellulose membrane (right panel). The two bands migrating faster than GST-FHA and GST-N166A represent degradation products that are reactive with the GST antibody. B, HCT-15 cells expressing different FLAG-tagged mutant Chk2 protein were incubated with Thr(P)-68 peptide-Sepharose. Whole cell extracts (WCE) or Thr(P)-68 peptide bound fractions were separated by SDS-PAGE and immunoblotted with anti-FLAG. C, whole cell extracts from 293T/17 cells exposed to 0 or 6 Gy IR were used for the peptide-Sepharose binding assay. Bound fractions were immunoblotted with anti-Chk2.

Thr-68 Phosphorylation Promotes Chk2 Oligomerization
pressing GST, GST-FHA, and GST-N166A and a 10-amino acid peptide representing the Thr-68 region of Chk2 conjugated to Sepharose. Only peptide containing Thr(P)-68 bound GST-FHA, and as expected, the N166A FHA domain mutant did not bind to Thr(P)-68 peptide-Sepharose (Fig. 3A). Therefore, a peptide consisting of the Thr(P)-68 region of Chk2 is a phosphospecific ligand for the Chk2 FHA domain, and the N166A mutation disrupts Chk2 FHA domain function without negatively affecting the stability of the protein.
Mutations within the FHA domain that alter efficient activation of Chk2 in irradiated cells (R117A, S140A, K141A, and N166A) significantly reduced the ability of Chk2 to bind to Thr(P)-68 peptide in vitro (Fig. 3B). Consistent with its ability to be efficiently autophosphorylated and activated following IR, the I157T mutant bound to Thr(P)-68 peptide as well as wild type (Fig. 3B). Therefore, the efficiency by which Chk2 is autophosphorylated in response to IR grossly correlates with whether Chk2 can bind to P-Thr-68 peptide via the FHA domain. To determine whether the interaction between Thr(P)-68 and Chk2 depends upon a modification induced by IR, 293T/17 cells were irradiated or mock-irradiated, and whole cell extracts were incubated with either Thr-68 or Thr(P)-68 peptide-Sepharose. Chk2 from control rather than irradiated cell extracts bound to Thr(P)-68 peptide-Sepharose suggesting that the affinity of the FHA domain is reduced after cells are exposed to IR (Fig. 3C). Thus, the interaction between Thr(P)-68 peptide and full-length Chk2 depends upon an intact FHA domain but is diminished when Chk2 is modified in DNAdamaged cells.
Chk2 Phosphorylates Its Own FHA Domain in Vitro Resulting in Reduced Affinity for Thr(P)-68 Peptide-DNA damagedependent phosphorylation of the Chk2 FHA domain may be one mechanism by which Chk2 binding to Thr(P)-68 peptide is reduced. To explore the possibility that Chk2 autophosphorylates within the FHA domain or other potential sites, we purified three different recombinant Chk2 fragments that together represent the entire Chk2 protein molecule and subjected each fragment to in vitro kinase assays using purified wt or kd Chk2 kinase. Consistent with Chk2 autophosphorylating the activation segment, recombinant wt Chk2 phosphorylated a catalytic inactive kinase domain fragment (Fig. 4A). The FHA domain was highly phosphorylated by wt Chk2, whereas the SQ/TQ-rich region of Chk2 does not appear to be a substrate for Chk2 phosphorylation in vitro (Fig. 4A). Phosphorylation of GST-FHA by wt Chk2 also reduced its ability to bind to Thr(P)- Thr-68 Phosphorylation Promotes Chk2 Oligomerization 68-Sepharose and to interact with endogenous Chk2 present in extract obtained from DNA-damaged cells (Fig 4, B and C). Therefore, autophosphorylation within the FHA domain of Chk2 may induce rapid dissociation of Chk2 oligomers. It is unclear whether the FHA domain of Chk2 is required for interactions with the proteins it targets for phosphorylation. If this proves to be the case, autophosphorylation within the FHA domain may selectively affect Thr(P)-68 Chk2 binding and not all phosphoprotein interactions.
Catalytic Inactive Chk2 Forms an Oligomer in Vivo-FLAGtagged and Myc-tagged catalytic inactive (kd) Chk2 proteins were co-expressed in 293T/17 cells to determine whether two or more Chk2 proteins associate with one another in vivo and whether Thr-68 phosphorylation and FHA domain integrity are required for this interaction. By using this approach, we have found that kd Chk2 forms a stable oligomer in 293T/17 cells based on the observation that both the FLAG and Myc antibodies can co-immunoprecipitate epitope-tagged Chk2 (Fig. 5). Despite DNA damage-induced Thr-68 phosphorylation of the N166A/kd double mutant, we were unable to detect oligomer formation by this mutant in irradiated cells suggesting that the integrity of the FHA domain is important for oligomer formation. Oligomer formation by the T68A/kd mutant is also deficient as compared with that exhibited by the kd Chk2 proteins (Fig. 5). Therefore, stable oligomer formation by catalytic inactive Chk2 requires both Thr-68 phosphorylation and an intact FHA domain.
Repeated attempts to co-immunoprecipitate two differentially tagged wild type Chk2 molecules using this approach have been unsuccessful (data not shown). The observation that only kd Chk2 formed oligomers suggests that autophosphorylation by Chk2 reduces the affinity of one Chk2 molecule for another. We have frequently found that transient overexpression of wild type Chk2 at high levels in 293T/17 cells is associated with Chk2 autophosphorylation and activation in the absence of DNA damage (data not shown). This phenomenon is analogous to what occurs when RAD53 is expressed at high concentrations in E. coli where it becomes hyperphosphorylated in the absence of RAD9 (25). Therefore, the inability to co-immunoprecipitate differentially tagged, wild type Chk2 proteins may be due to the fact that Chk2 is already activated and can no longer form stable oligomers once autophosphorylation occurs. This is consistent with our findings that recombinant GST-FHA phosphorylated by Chk2 in vitro exhibits reduced affinity for the Thr(P)-68 region of Chk2 and modified Chk2 expressed in irradiated cells (Fig. 4).
In addition to wild type Chk2 activated in a DNA damageindependent manner upon overexpression in 293T/17 cells, we have also observed high levels of constitutive Thr-68 phosphorylation of kd Chk2 under similar conditions (Fig. 5). High levels of Thr-68 phosphorylation may reflect an inability of a specific phosphatase to gain access to Thr-68-phosphorylated kd Chk2 because it is presumably shielded by the FHA domain. The fact that we have only been able to co-immunoprecipitate kd Chk2 suggests that once phosphorylated on Thr-68, oligomerization of kd Chk2 becomes very stable. This is most likely due to low autophosphorylation activity expressed by kd Chk2 that would in turn result in a lower dissociation rate of the oligomer complex.
Thus far, this is the first biologically relevant phosphorylationdependent ligand described for the FHA domain of Chk2. Together, these data support the model that phosphorylation of Thr-68 by ATM promotes binding of the SQ/TQ region of one Chk2 molecule to the FHA domain of another forming oligomers. Oligomerization may effectively increase the local concentration of Chk2 such that trans-autophosphorylation becomes more favorable (Fig. 6). Trans-autophosphorylation within the FHA domain may then cause dissociation of the Chk2 complex. Active Chk2 monomers are released and proceed to phosphorylate effector molecules important for mammalian checkpoint control.