Chk2 Oligomerization Studied by Phosphopeptide Ligation

Chk2/CHEK2/hCds1 is a modular serine-threonine kinase involved in transducing DNA damage signals. Phosphorylation by ataxia telangiectasia-mutated kinase (ATM) promotes Chk2 self-association, autophosphorylation, and activation. Here we use expressed protein ligation to generate a Chk2 N-terminal regulatory region encompassing a fork-head-associated (FHA) domain, a stoichiometrically phosphorylated Thr-68 motif and intervening linker. Hydrodynamic analysis reveals that Thr-68 phosphorylation stabilizes weak FHA-FHA interactions that occur in the unphosphorylated species to form a high affinity dimer. Although clearly a prerequisite for Chk2 activation in vivo, we show that dimerization modulates potential phosphodependent interactions with effector proteins and substrates through either the pThr-68 site, or the canonical FHA phosphobinding surface with which it is tightly associated. We further show that the dimer-occluded pThr-68 motif is released by intra-dimer autophosphorylation of the FHA domain at the highly conserved Ser-140 position, a major pThr contact in all FHA-phosphopeptide complex structures, revealing a mechanism of Chk2 dimer dissociation following kinase domain activation.

Chk2 (also know as CHEK2 or hCds1) was originally discovered as a Saccharomyces cerevisiae Rad53-related kinase that is primarily activated by the ataxia telangiectasia-mutated kinase (ATM) 3 after DNA damage (1)(2)(3)(4)(5). Many substrates and interacting partners have since been identified (6) and through this increasingly diverse array of interactions, Chk2 appears to act not only as a regulator of DNA damage/stress-response signaling and cell cycle checkpoint activation, but also apoptosis, senescence, viral infectivity, and other pathways. In particular, its apparent role in regulating p53 (7)(8)(9), and the discovery of mutant Chk2 alleles in patients with wild-type p53 but who suffer from Li-Fraumeni syndrome, suggested that Chk2 may itself be a tumor suppressor (10). Although this remains controversial, recent studies have shown that Chk2 mutations may cause increased susceptibility to a wide variety of cancers, and it has emerged as a potentially important anti-cancer drug target (11,12).
The significance of Chk2 in the DNA damage response and checkpoint regulation has inspired a number of studies of its regulation and its interactions with target molecules. Chk2 is a member of a family of kinases that includes budding yeast Rad53p and Dun1p, fission yeast Cds1p and Drosophila DMNK. Overall, these molecules share a related architecture, and contain two main regions of sequence homology that comprise a C-terminal Ser/Thr kinase domain and an N-terminal (FHA) domain. Rad53 is an exception and is the only family member with a second, C-terminal FHA. It is well established that FHA domains mediate protein-protein interactions primarily through binding to short phosphothreonine-containing motifs in target proteins. The importance of the FHA domain in Chk2 function is exemplified by its absolute requirement for Chk2 activation and the identification of several point mutations in this region in various cancer cells types. In addition, the InsZ Chk2 transcript variant encoding only the FHA domain and N-terminal regulatory region may act as a dominant negative allele in late stage breast cancer (13).
Chk2 is a phosphoprotein in vivo, and its full activation requires phosphorylation on Thr-68. This is located within a partially conserved N-terminal region (14 -18) containing a series of Ser-Gln or Thr-Gln pairs (SCD, Ref. 19). While these represent preferred sites for PI3-kinase-related kinases such as ATM, ATM and Rad3-related (ATR), and DNA-dependent protein kinase (DNAPK), they may also be targeted by other kinases, such as Mps1, MRK, and Polo-like kinases (Plk) (20 -23). A single mutation of Thr-68 to alanine in the Chk2 SCD is sufficient to severely reduce ATM-dependent Chk2 activation after DNA damage (14 -16), and abolish interactions of IR-activated Chk2 with recombinant Chk2 FHA domain (24). In contrast, mutations at the six other potential PIKK sites have little or no effect alone or in combination (14 -16). A model for activation of Chk2 family kinases has emerged that involves self-association through intermolecular binding of the FHA domain to phosphorylated Thr-68 and possibly other sites in the SCD. This generates oligomers that are proficient for autophosphorylation of the kinase domain activation loop in trans (24 -28). Although structures of both the Chk2 FHA-phosphopeptide complex, and the Chk2 kinase domain have been determined individually (27,29), a structure of any full-length Chk2 orthologue remains elusive. Furthermore, efforts to tease out details of the activation mechanism have been complicated by the fact that Chk2 is highly phosphorylated at many sites (30 -32), of which only a subset have been characterized in terms of their contribution to Chk2 function and regulation in vivo. The experimental challenge is in generating Chk2 molecules with precisely defined phosphorylation states. A powerful alternative to enzymatic modification is expressed protein ligation (EPL) of recombinant proteins with synthetic peptides containing specific and stoichiometric post-translational modifications of many kinds. In the case of phosphorylation, chemical, and intein-mediated ligation has been previously used to great effect to generate semi-synthetic analogues of active TGF-␤ kinase (33) and phosphorylated substrate proteins, particularly R-smad MH2 domains (34). To this end, we have used EPL technology to generate a semi-synthetic molecule in which the Chk2 FHA domain is appended to its N-terminal regulatory region containing a unique pThr-68 phosphorylation site, as a model of ATM-phosphorylated full-length Chk2 protein.
Using this chemically defined species in combination with hydrodynamics, mass spectrometry, and in vitro binding assays, we investigate protein-protein interactions involved in Chk2 activation in vitro, their regulation by autophosphorylation and implications for Chk2 interactions with regulators and target molecules.
Isothermal Titration Calorimetry-ITC experiments were performed in 20 mM Tris, pH 8.0, 150 mM NaCl, and 1 mM ␤-mercaptoethanol at 20°C using a VP-ITC microcalorimeter (MicroCal Inc.). Chk2 FHA-(63-219) at a concentration of 60 M was titrated against 50 ϫ 5 l injections of Chk2 pThr-68 peptide (TVSTpQELYS) at a syringe concentration of 880 M. Following subtraction of the heat of dilution, data were analyzed assuming a single-site binding model using ORIGIN software.
Peptide Synthesis and Phosphopeptide Ligation-The peptide thioester with sequence H 2 N-LETVSpTQELY-SBn (pThr-68-SBn) was prepared using the sulfonamide safety-catch linker strategy (supplemental information). To produce a Chk2 FHA fragment appropriate for chemical ligation, residues 73-219 of Chk2 were PCR-amplified and cloned into pGEX-6P1. The 5Ј primer was designed so as to generate a mutation of Ser-73 to cysteine (S73C) and to place a cleavage site for Factor X a immediately upstream. Following purification of this fusion protein, cleavage with Factor X a yields the FHA domain fragment with a free N-terminal cysteine. The synthetic C-terminal thioester peptide at a concentration of 1 mM was incubated with 100 M purified Factor X a -digested FHA proteins overnight at room temperature in 200 mM phosphate buffer, pH 8.0, 150 mM NaCl, 10 mM TCEP, 122 mM 2-mercaptoethanesulphonic acid (MESNA) to form the ligated pThr-68 FHA product (pT68 FHA lig ).
Analytical Ultracentrifugation (AUC)-Sedimentation velocity experiments were performed with a Beckman Optima XL-A analytical ultracentrifuge equipped with absorbance optics. Chk2 FHA-(63-219) and pT68 FHA lig and were prepared with dilution buffer (20 mM Tris, pH 8.0, 150 mM NaCl, and 0.5 mM TCEP). The aluminum double sector centerpieces were filled with 400 l of the protein sample and 420 l the dilution buffer, respectively. Samples were centrifuged at a speed of 50,000 rpm and a temperature of 20°C using an An50-Ti rotor. Scans were acquired at a wavelength of 280 nm in time intervals of 600 s. The partial specific volume of 0.72 ml/g for FHA proteins was calculated based on the amino acid composition (36). Sedimentation velocity data were analyzed using SEDFIT (37).
Size Exclusion Chromatography Multiangle Laser Scattering (SEC-MALLS)-Molecular weights and molecular weight distributions were determined using on line multi-angle laser light scattering coupled with size exclusion chromatography (SEC-MALLS). Samples were applied to a Superdex 75 10/300 GL column equilibrated in 20 mM Tris-HCl, 150 mM NaCl 0.5 mM TCEP, pH 8.0 at a flow rate of 0.5 ml/min. The column was mounted on a Jasco HPLC controlled by the Chrompass software package. The scattered light intensity of the column eluent was recorded at sixteen angles (over the range 32-147°) using a DAWN-HELEOS laser photometer (Wyatt Technology Corp., Santa Barbara, CA). The protein concentration of the eluent was determined from the refractive index change (dn/ dc ϭ 0.186) using an OPTILAB-rEX differential refractometer equipped with a Peltier temperature-regulated flow cell, maintained at 25°C (Wyatt Technology Corp.). The wavelength of the laser in the DAWN-HELEOS, and the light source in the OPTILAB-rEX was 658 nm. The weight-averaged molecular mass of material contained in chromatographic peaks was determined using the ASTRA software version 5.1 (Wyatt Technology Corp.). Briefly, at 1-s intervals throughout the elution of peaks the scattered light intensities together with the corresponding protein concentration were used to construct Debye plots (KC/R versus sin 2 (/2)). The weight-averaged molecular mass was then calculated at each point in the chromatogram from the intercept of an individual plot. An overall average molecular weight and polydispersity term for each species was calculated by combining and averaging the data from the individual measurements.
Surface Plasmon Resonance Measurements-For phosphopeptide binding studies by surface plasmon resonance (SPR), a streptavidin sensor chip was used for immobilization of a biotinylated peptide, Biotin-T68pL (Biotin-T 59 LSSLET-VSTpQELYSIPE 76 ), derived from the N-terminal region of human Chk2. Steady state measurements were performed using a Biacore 2000 instrument. Samples were injected at a flow rate of 5 l/min for 12 min to ensure that steady state was established. The sensor chip surface was regenerated with a 12-min wash step. The Plk1 PBD protein was diluted in HBS-EP buffer as a series of analytes (78 nM, 156 nM, 312 nM, 625 nM, 1.25 M, 2.5 M, 5.0 M, 10.0 M). Responses (RU) from the plateau region of each sensorgram were taken. The steady state response for each concentration was obtained by subtracting responses from the ligand-immobilized flow cell with the correspondent blank flow cell. These were plotted against analyte concentration and the data fitted with non-linear single-site binding model in Equation 1, where Y is the specific binding at steady state, B is the maximum binding capacity of the surface, X is the analyte concentration, and K D is the equilibrium dissociation constant.
Competition experiments using the Plk1 PBD protein were measured using steady state methods. A series of analytes were prepared that contained a constant PBD concentration (50 nM) and increasing concentrations of competitor proteins/peptides (0 nM, 9.75 nM, 19.5 nM, 39 nM, 78 nM, 156 nM, 312.5 nM, 625 nM, 1.25 M, 2.5 M, 5 M, 10 M). The percentages of steady state signal remaining were calculated using the response from binding of 50 nM Plk1 PBD alone as 100%. These were plotted against logarithm of concentration of the competitor. Where appropriate, data were fitted with non-linear one-site competition model in Equation 2, where Y is percentage of specific binding, A 2 is the fit minimum, A 1 is the fit maximum, X is the logarithm of competitor concentration, and X 0 is the EC 50 .
Autophosphorylation Assays-Autophosphorylation/rephosphorylation assays (32) were carried out at a Chk2 concentration of 20 M in a buffer containing 50 mM HEPES pH 7.5, 10 mM MgCl 2 , 1 mM CaCl 2 , and 3 mM ATP at room temperature for 2 h. For FHA phosphorylation assays, ligated or unligated mutant or wild-type FHA fragments were included at equimolar concentration and incubated at room temperature for 4 h followed by overnight incubation at 4°C.
Phosphorylation Site Mapping-Identification of autophosphorylation sites was performed essentially as described (38). Briefly, excised SDS-PAGE protein bands were reduced with dithiothreitol and alkylated using iodoacetamide. The bands were dried and reswollen in a sufficient volume to cover of 2 ng/l chymotrypsin (sequencing grade Roche Diagnostic, Germany) in 5 mM ammonium bicarbonate. After overnight digestion at 32°C, the supernatant was acidified by the addition of a 1/10th volume of 4% trifluoroacetic acid. A phosphoprotein purification kit (Qiagen) was used to isolate the phosphorylated peptides, which were then fractionated with a Model 130A syringe pump HPLC system from Applied Biosystems using a 250-mm 5-mm Reliasil C18 BDX column (Column Engineering, Ontario). Peptide mass fingerprinting was performed using a Reflex III MALDI time-of-flight mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany), equipped with a nitrogen laser and a Scout-384 probe, to obtain positive ion mass spectra of digested protein with pulsed-ion extraction in reflectron mode. An accelerating voltage of 26 kV was used with detector bias gating set to 2 kV and a mass cutoff of m/z 650. Matrix surfaces were prepared using recrystallized ␣-cyano-4-hydroxycinammic acid and nitrocellulose. 0.4 l of digestion supernatant was deposited on the matrix surface and allowed to dry prior to desalting with water. Peptide mass fingerprints were searched against the non-redundant protein data base placed in the public domain by the National Centre for Biotechnology Information (NCBI) using the program MASCOT (39). Nanospray mass spectra were acquired on an LCQ "classic" quadrupole ion trap mass spectrometer (ThermoQuest, Austin, TX) equipped with a nanospray source (Protana, Odense, Denmark) operated at a spray voltage of 800 V and a capillary temperature of 150°C. Dried HPLC fractions were dissolved in 60% v/v methanol, 0.1% v/v formic acid, and 2 l transferred to an Econo12 nanospray needle (New Objective Inc, Cambridge, MA). Daughter ion spectra were typically acquired at a collision energy of 30% and an isolation width of 4 Da.

RESULTS
Chk2 FHA Binds to pThr-68 Peptides with Low Affinity-Of the two threonines present within TQ dipeptide motifs in the Chk2 Oligomerization Studied by Phosphopeptide Ligation DECEMBER 19, 2008 • VOLUME 283 • NUMBER 51 human Chk2 SCD (Thr-26 and Thr-68), Thr-68 appears to be the preferred site for DNA damage-dependent phosphorylation by ATM. Thr-68 is unique in the Chk2 SCD in that it resides within a sequence Thr 68 -Gln 69 -Glu 70 -Leu 71 that is conserved in all mammalian Chk2 orthologues (Fig. 1B). This sequence resembles the optimal Chk2 FHA-interacting motif identified by oriented peptide library screening, which showed a marked preference for medium hydrophobic residues (Ile/ Leu/Val) at the position three residues C-terminal to the phosphothreonine (pTϩ3Ј) (29). A peptide encompassing the pThr-68 motif has been reported to bind to the Chk2 FHA domain (29,40). We now show by ITC that it does so with an apparent K d of around 30 M (Fig. 1A) consistent with a potential role for pThr-68-FHA interactions in Chk2 self-association and activation. Many studies of FHA domain phosphodependent binding activity and specificity, including that of Chk2 (41) have demonstrated a profound binding preference for pThrcontaining motifs. Therefore it seems unlikely that Chk2 activation can be efficiently achieved through serine phosphorylation at SQ or other sites in the SCD, and this appears to be the case (14 -16). The remaining TQ pair in the human Chk2 SCD at Thr-26 is efficiently phosphorylated by ATM/Rad3-related (ATR) but not by ATM (16). More significantly, it is not conserved in mammalian Chk2 orthologues (Fig. 1C) and has a glutamine in the pTϩ3 position rendering it an unlikely FHA binding site. Similarly, the Schizosaccharomyces pombe Chk2 orthologue Cds1 contains two phosphorylated threonines in its SCD region but only one of these, Thr-11, is necessary and sufficient for activation (42). For these reasons we chose to focus on the role of Thr-68 phosphorylation in Chk2-Chk2 interactions.
A Semisynthetic Model of pThr-68/FHA-mediated Chk2-Chk2 Interactions-To circumvent problems of heterogeneous autophosphorylation of recombinant Chk2, and our currently incomplete understanding of the effects of secondary autophosphorylation on kinase activity and oligomerization, we used EPL technology to generate an uniquely Thr-68-phosphorylated fragment of Chk2 encompassing the FHA domain and intervening linker sequences ( Fig. 2A). Time course studies at two different peptide/protein ratios were used to monitor the efficiencies of the ligation reactions (Fig. 2B). A 10:1 peptide/ protein molar ratio produced greater than 80% ligation efficiency over a 20-h period, and these conditions were used in all subsequent experiments. For further purification and analysis, we employed gel-filtration chromatography of ligation products on Superdex 75 (Fig. 2C). Elution profiles showed the presence of well separated fast and slow migrating peaks consistent with self-association of the ligated product. The difference in retention time of the two species appeared to be directly attributable to Thr-68 phosphorylation as treatment with phosphatase converted the ligated product to a form with the expected molecular mass of the dephosphorylated product that migrated identically to unphosphorylated Chk2 FHA (63-219). The faster eluting protein was confirmed as the ligated product by mass spectrometry with no detectable contaminating unphosphorylated species.
Thr-68-phosphorylated Chk2 FHA Forms a Tight Dimer-The solution molecular weight and hydrodynamic properties

Chk2 Oligomerization Studied by Phosphopeptide Ligation
of non-phospho (Chk2 FHA-(63-219)) and the Thr-68phosphorylated (pT68 FHA lig ) protein generated by EPL were examined by a combination of AUC and light scattering methods. Initially, the solution molecular weight of unmodified Chk2 FHA-(63-219) was estimated using SEC-MALLS (Fig. 3A). These data report a weight-averaged molecular mass of 22.2 kDa, significantly greater than that expected for the protein monomer (18.4 kDa). This anomalously high solution molecular weight was further investigated using sedimentation velocity AUC. Typical sedimentation coefficient C(S) and molar mass C(M) distributions produced from sedimentation velocity experiments employing 18 M non-phosphorylated Chk2 FHA-(63-219) and 24 M pT68 FHA lig are shown in Fig. 3, C and D, respectively. Inspection of the C(S) function derived from sedimentation of Chk2 FHA-(63-219) reveals that around 70% of the material is contained within a peak with S 20,w of 1.9, M w ϭ 28.6 kDa, where the best fit weight-averaged frictional ratio (f/f o ) w for the whole distribution function has a value of 1.85. The remainder of the signal is distributed among several smaller peaks that have sedimentation coefficients ranging from 3 S to 10 S and are presumably the result of some aggregation in the sample. Because of the slight heterogeneity, the (f/f o ) w for the 1.9 S peak alone was determined by inspection of the contour plot of the two-dimensional distribution function of sedimentation coefficients and frictional ratios C(S,f/f o ) (data not shown). The most populated frictional ratio within the 1.9S peak determined by this method has a value of 1.7. Applying this frictional ratio, an apparent weight-averaged molecular weight of 24.8 kDa is then obtained from the C(S) distribution function (Fig. 3C) and a similar value of 25.0 kDa from the integration of the C(M) function (Fig. 3D). Thus, both the SEC-MALLS and sedimentation velocity measurements are consistent with weak, but significant FHA-FHA interactions in the absence of SCD phosphorylation.
In contrast to the non-phosphorylated fragment, AUC showed that at a loading concentration of 24 M the ligated FHA sediments as single species comprising greater than 95% of the signal in the C(S) distribution (Fig. 3C). The molecule has an S 20,w of 2.9 and an apparent mass of 35.9 kDa, within 2.5% of the expected dimeric mass of 36.8 kDa, with no observable monomeric species present. Furthermore, analysis of the selfassociation of the phosphorylated species by SEC-MALLS shows no observable dissociation at an on-column peak concentration of ϳ4.8 M (Fig. 3B). We, therefore, conclude that Thr-68-phosphorylated Chk2 forms a dimeric complex with high apparent affinity. This notion is supported by experiments designed to examine the accessibility of the pThr-68 motif in Chk2 dimers described below.
Chk2 Dimerization Impedes pThr-68-and FHA-dependent Interactions with Effectors-Several studies have suggested that pThr-68 can serve as a docking/recognition motif for  DECEMBER 19, 2008 • VOLUME 283 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 36023 phosphodependent binding modules present within Chk2 binding partners (43,44). We therefore wished to investigate the implications of the tight, pThr-68-dependent dimerization that we have described for downstream effector/substrate interactions. To do this we chose to use the PBD of human Plk1, which binds to the pThr-68 phosphopeptide with an affinity of ϳ400 nM (Fig. 4A), as a probe of pThr-68 motif accessibility in monomeric and dimeric Chk2. We first attempted to compare PBD/Chk2 binding directly by SPR titration but this proved impractical due to either artifactual dissociation of the pThr-68-dimerized FHA domain or disruption of the PBD binding surface during cross-linking to activated carboxymethyl dextran BIAcore chips. Instead, we developed a solution competition assay in which a series of mixtures containing a constant concentration of the PBD and increasing concentrations of competitor proteins, were injected over a Biacore streptavidin chip coated with the biotinylated pThr-68 peptide. Binding of competitor to the Plk1 PBD reduces its interaction with the surface immobilized phosphopeptide. Thus for the control titration (Fig.  4B), PBD-pThr-68 interactions are effectively competed by increasing concentrations of a peptide (MAGPMQSpTPLN-GAKK-Plktide) that we have previously shown to bind to the Plk1 PBD with a K d of ϳ300 nM (35). For Chk2-derived competitors (Fig. 4C), only the pThr-68 peptide and a ligated Chk2 FHA domain with an alanine substitution at Ser-140 (pT68 FHA lig S140A) showed a continuous concentrationdependent competition for PBD binding to the chip. This is significant because Ser-140 makes a crucial and highly conserved interaction with the pThr moiety (29). Its mutation to alanine abrogates phosphopeptide binding (see Fig. 6B below) and therefore pThr-68-dependent dimerization. Similar observations were made for competition experiments using the FHA domain of Mdc1, which has also been proposed to bind Chk2 in a pThr-68-dependent manner, in place of the Plk1 PBD (44) (supplemental Fig. S1). In contrast, the other competitors, including Chk2 Kin-(205-538) that contains the kinase domain and only a single phosphorylated Ser-516 (data not shown), show a background reduction in signal that remains at around 80%. Most importantly, dimerized pT68 FHA lig only reduces binding to control levels, despite the high affinity of the PBD for the isolated pThr-68 motif. This is consistent with our contention that tight dimerization of Chk2 effectively renders it unable to interact with downstream effectors/binding partners through either the pThr-68 motif or the FHA phosphobinding site.

Chk2 Oligomerization Studied by Phosphopeptide Ligation
FHA Domain Phosphorylation at a Highly Conserved pThr Contact Site-Modification of the Chk2 FHA domain by autophosphorylation may play a role in promoting the formation of Thr-68-phosphorylated Chk2 monomers after DNA damage (24). However, autophosphorylation site(s) that may promote dissociation of active Chk2 dimers have not been identified. With these observations in mind, we performed phosphosite mapping by MALDI-TOF mass spectrometry on recombinant Chk2 purified from E. coli. Similar studies of recombinant Chk2 autophosphorylation sites have revealed a number of modified residues spread throughout the molecule (30,31), several of which have been confirmed in vivo. Our efforts have thus far identified nine distinct proteolytic peptides that contain at least a single phosphorylation site. 4 Two such peptides were identified within the FHA domain, one containing Ser-120 that has been observed previously (30 -32) and another spanning residues 138 -144. In this case, the site of phosphorylation was confirmed by tandem mass spectrometry to be Ser-140 (Fig.  5A) that occurs within a sequence RTYS 140 resembling preferred Chk2 phosphorylation motifs (45).
To facilitate further studies of Ser-140 phosphorylation in vitro, we raised polyclonal antibodies against a pSer-140 peptide. As expected, the affinity-purified antibody clearly detects recombinant Chk2-(63-543) and full-length human Chk2 produced in insect cells in Western blots but fails to detect a kinase-dead version harboring a D347A mutation (Fig. 5B).  . Biacore binding and competition assays. A, steady-state SPR titration of Plk1 PBD binding to pThr-68 Chk2 peptide (left) and raw sensorgrams (right). Data were analyzed according to a simple, one-site binding model, which reports a K d of 380 nM. B, control assay for competition of a high affinity Plk1 PBD-binding peptide (Plktide) for binding of the PBD to immobilized Chk2 pThr-68 peptide. The assay scheme is represented schematically in the inset. The reduction in signal by competition is represented as the percentage of the total steady state binding response remaining at each concentration, where the steady state binding response of the sample without competitor was arbitrarily set at 100%. Data were analyzed as described under "Experimental Methods." C, increasing concentrations of competitors were injected with the Plk1 PBD at a constant concentration of 50 nM. Data are again plotted as %R eq /R eq comp against the logarithm of the concentrations of the competitor. The color scheme is as shown. Lines joining individual data points have been included for clarity and do not represent fitted curves. DECEMBER 19, 2008 • VOLUME 283 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 36025
Chk2-(63-543) are phosphorylated on Thr-68, a known site of low-level autophosphorylation in recombinant Chk2 prepara-tions, and on the T-loop (anti-pThr-387). Specificity of the anti-pS140 antisera was confirmed using fulllength Chk2 variants containing S120A or S140A mutations. To minimize background autophosphorylation of non-physiological sites, for example during folding of recombinant Chk2 in the bacterial host, we purified these proteins from strains that also express phosphatase from a compatible plasmid vector. 5 Under this regime, all three proteins are devoid of any detectable phosphorylation (Fig.  5B). However, following incubation with ATP/Mg 2ϩ , phosphorylation at Thr-68 and Thr-387 is restored in all three Chk2 variants. Most significantly, a strong signal is seen for the anti-pSer-140 antibody in only the wild-type and S120A proteins and not the S140A mutant, effectively eliminating pSer-120 as the cognate epitope. Because all three proteins show substantial levels of overall rephosphorylation by mass spectrometry (supplemental Fig. S2) we conclude that the anti-pSer-140 antisera are specific for this site.
FHA Phosphorylation as a Regulator of ATM-dependent Chk2 Dimerization-Of the two FHA domain autophosphorylation sites identified, the functional significance of Ser-120 phosphorylation is unclear. This residue is not conserved in mammalian Chk2 orthologues and the x-ray structure of the Chk2 FHA domain (29) shows that it is located in a solvent exposed region of the FHA ␤-sandwich, remote from the site of phosphopeptide interaction (Fig. 6A,  left). Consistent with these observations, we observe no significant effect of a S120A mutation on binding to surface-immobilized pThr-68 peptide (Fig. 6B). In contrast, Ser-140 occupies a highly conserved position at the N-terminal end of ␤5 immediately juxtaposed to the ligand-binding surface. Here, its side chain makes a direct hydrogenbonding contact to the pThr phos-  The upper panel shows that Ser-140 is highly conserved in FHA domains from yeast (Rad53 FHA1 and 2) and humans (Chk2, Ki67). In the FHA domains from the related Rad53 and Chk2 kinases, this serine occurs in a motif that contains basic residues in the Ϫ2 or Ϫ3 positions consistent with the expected kinase domain specificity of these enzymes. B, left, immunoblot of recombinant His-tagged baculovirus expressed Chk2, and E. coli-expressed Chk2-(63-543) and a kinase-dead version, with phosphospecific antibodies against previously characterized sites (pThr-68 and pThr-387) and the novel pSer-140 site. Right, full-length, unphosphorylated Chk2 wild-type (WT), S120A, and S140A mutants were incubated in the presence or absence of ATP and analyzed for rephosphorylation of Thr-68, Ser-140, and Thr-387 by immunoblotting with appropriate antibodies. phoryl oxygen, and additional stabilizing interactions with main-chain atoms from the ␤6-␤7 turn. Electrostatic and steric effects of phosphorylation of this residue would therefore be incompatible with phosphopeptide binding (Fig. 6A, right) consistent with the significance of this residue for pThr recognition and the loss of interaction observed for the S140A mutant (Figs. 4C and 6B). MALLS analysis also confirms that Ser-140 substitution abrogates tight homodimerization (Fig. 6C) but also suggests that activated pThr-68/pSer-140 monomers are able to interact with Thr-68 unphosphorylated Chk2 to form partially stabilized dimers (Fig.  6C).
Ser-140 Phosphorylation Is Dependent on pThr-68-Overall our model predicts that Ser-140 phosphorylation is most likely to occur following Thr-68-dependent Chk2 dimerization and activation through T-loop phosphorylation. We therefore wished to test whether Ser-140 could be phosphorylated in the context of a free FHA domain or if incorporation into active dimers through intermolecular pThr-68-mediated FHA interactions is necessary. Fig. 7A shows that while no Ser-140 phosphorylation is observed for unligated Chk2 FHA-(73-219), Chk2 pT68 FHA lig is efficiently Ser-140-phosphorylated. Thus, these data support the idea that Ser-140 phosphorylation only occurs within pThr-68-dimerized Chk2 and further suggest that each activated kinase domain within the dimer targets the FHA domain of its partner molecule (Fig.  7B).

DISCUSSION
The importance of Chk2 in a variety of DNA damage response and other cell cycle-dependent pathways has engendered considerable interest in the mechanisms of its activation and regulation. ATM phosphorylation of Thr-68 is prominent in most extant models of Chk2 regulation. This is largely founded on several independent observa- Right, modeling of phosphoserine 140 into the Chk2 complex structure highlights significant steric and electrostatic clashes with the pThr moiety of the bound phosphopeptide. B, SPR analysis of S120A and S140A mutations. Fulllength, unphosphorylated Chk2 wild-type (WT), S120A, and S140A mutants at three concentrations (3, 10, and 50 M) in HBS-EP buffer were flowed over a streptavidin sensor chip loaded with biotinylated pThr-68 peptide and the response histogram plotted as a function of FHA concentration (C) SEC-MALLS analysis shows that while Chk2-(63-219) (400 M loading) and the pThr-68 phosphorylated S140A mutant (200 M loading) are largely monomeric, a 1:1 mixture of these proteins (400 M loading) results in formation of a significantly stabilized heterodimer (green) with a lower overall retention time and a weight-averaged molecular weight of ϳ26 kDa, intermediate between that of the unphosphorylated and fully Thr-68-phosphorylated species. The asterisk in the lower panel represents the presence of the S140A mutation as a mimic of the effect of Ser-140 phosphorylation.
tions that Thr-68 phosphorylation is both necessary and sufficient for high-level Chk2 activation after DNA damage. The fact that it is uniquely conserved in mammalian Chk2 orthologues and that it resides in a motif that binds to the Chk2 FHA domain (Ref 40 and this study) further underpins its importance. In a landmark paper, Ahn and Prives (46) first proposed that the oligomeric Chk2 assemblies observed by several groups were likely to be dimers based on sucrose gradient sedimentation and gel filtration analysis of DNA-damaged cell extracts. However, these experiments are complicated by the non-homogeneity of Chk2 phosphorylation following DNA damage in vivo, and the fact that this approach is not ideal for analysis of equilibrium mixtures of multimeric interacting components. Given the central importance of FHA-mediated oligomeriza-tion in Chk2 regulation, we wished to further test current models of Chk2 self-association and regulation in vitro through a combined biochemical, biophysical, and chemical biology approach.
Hydrodynamic analysis using sedimentation velocity analytical ultracentrifugation and SEC-MALLS reveals significant but low affinity dimerization of the nonphosphorylated Chk2 FHA domain, consistent with DNA damage-independent activation observed previously during overexpression of Chk2 in 293T/17 cells (24). Our data also suggest that the unstable, non-pThr-68 dimers apparent in HCT116 cells after neocarzinostatin treatment (46) and irradiated 293 cells (25) and the residual activation observed in Thr-68 mutant Chk2 observed previously (14 -16) could arise from phosphoindependent FHA-FHA interactions. In contrast, the semi-synthetic pT68 FHA forms tight dimers that show little or no dissociation even at submicromolar concentrations. The apparent strength of the interaction has prevented direct determination of the dimer affinity because the experimental concentrations at which measurable dissociation might be observed are below the sensitivity of methods currently available to us. Even though the isolated pThr-68 peptide binds rather weakly to the Chk2 FHA (Fig. 1), an effective K d of ϳ1-2 nM would be predicted by assuming simple additivity for the cross-dimer interactions of the pThr-68 motifs. This high-affinity pThr-68-dependent association has clear implications for Chk2 function and regulation. On one hand, such a tight interaction may provide for damage-dependent release of Chk2, pre-localized through FHA binding to cognate sites within upstream regulators prior to Thr-68 phosphorylation. On the other, tight self-association that is presumably necessary for efficient Chk2 activation at low in vivo concentrations, will nonetheless obstruct subsequent pThr-68 or FHA-dependent binding to downstream effectors/ substrates such as Plk1 or Mdc1 (43,44). Although the functional significance of interactions with these molecules is unclear, direct binding of pThr-68 peptides to the Plk1 PBD and Mdc1 FHA has been demonstrated in vitro, (43,44)  derived from library selection studies (35,47). Nonetheless, our SPR assays show that the pThr-68-dimerized Chk2 FHA is unable to compete with either Plk1 PBD or the Mdc1 FHA for binding to immobilized pThr-68 peptide. Thus, interactions involving either direct binding to pThr-68 or to the phosphobinding surface on the FHA domain following Chk2 activation must be subject to additional regulatory mechanisms. It has been shown previously that Chk2 autophosphorylation of its FHA domain can disrupt dimerization, and that Chk2 kinase-dead dimers are more stable than wild-type kinase following DNA damage (24). A similar dependence on kinase activity for disassembly of Rad53 oligomers has been reported (28) indicating that FHA phosphorylation may be a conserved mechanism for regulation of Rad53/Chk2 family kinases. We have now been able to show autophosphorylation of Ser-140 that makes direct contact to the phosphothreonine residue in the Chk2 FHA complex structure (29). This residue is highly conserved across the FHA domain family and is present in all known Chk2/Rad53 family kinases supporting it as a strong candidate site for autophosphorylation-dependent control of dimerization. Antisera specific for the pSer-140 motif identify this modification in full-length human Chk2 produced in insect cells, which has been reported to be indistinguishable in terms of its phosphorylation status and oligomerization behavior from activated Chk2 immunoprecipitated from DNA-damaged HCT116 cells (46). Furthermore, rephosphorylation of phosphatase-treated recombinant Chk2 appears to robustly identify verified, physiologically significant in vivo phosphorylation sites (32), and our observation of Ser-140 rephosphorylation supports the contention that it is a functionally relevant site of Chk2 autoregulation.
In addition to FHA autophosphorylation, phosphodependent dimer stability also appears to be controlled by a p53-regulated Ser/Thr phosphatase, Wip1 through dephosphorylation of Thr-68 in Chk2 (48,49). So, while autophosphorylation directly disrupts the phosphobinding site and releases pThr-68 monomers, Wip1 activity potentially generates monomeric, active Chk2 with intact FHA phosphobinding activity but with unmodified Thr-68. Cells expressing kinasedead Chk2 accumulate abnormally high levels of pThr-68 dimeric forms (24). Therefore, some degree of weakening of the dimer by autophosphorylation at Ser-140 (this study) or other accessory sites may be required for full accessibility of phosphatases to pThr-68, as it is now appears to be for pThr 68-interacting partners. This also may explain the apparent requirement of Chk2 kinase activity for interaction with Wip1 (50). Because Wip1 contains no recognizable phosphobinding domain, Chk2 dimerization may occlude a phosphoindependent Wip1 interaction surface on the FHA domain, as proposed previously for Chk2 interactions with Brca1, Cdc25, and p53 (29). Such a mechanism may provide a safety-catch whereby rapid and stable dimerization protects pThr-68 against dephosphorylation until Chk2 becomes fully activated. Similarly, the fact that Ser-140 autophosphorylation only occurs within the context of pThr-68-dependent dimers may explain how the kinase domain gains transient access to the protected Ser-140 site by virtue of a local concentration effect. It also potentially ensures that monomerization does not short-circuit the Chk2 activation pathway. Finally, the observation that pThr-68/pSer-140 di-phosphorylated monomers can bind and potentially activate unphosphorylated Chk2 molecules, provides a means of amplifying DNA damage signals, and/or prolonging Chk2 signaling in the absence of ATM activity.
The results reported here, together with previous studies, present a rather complex picture in which Chk2 exists in a variety of oligomerization states arising from the combined effects of autophosphorylation, and the activities of SCD-specific kinases and phosphatases. Our data further suggest that such complexity may be necessary to allow for downstream interactions mediated by the FHA domain and/or phosphorylated SCD motifs that would otherwise be inaccessible in stable dimeric forms. Many questions remain concerning the potential role of scaffolding molecules in Chk2 activation, and the overall structural organization of the pre-activated dimer along with monomeric and dimeric activated forms. However, the application of EPL methods described here has provided new biochemical and biophysical insights into Chk2 self-association, and establishes this approach as a powerful means to further dissect the molecular basis of activation of Chk2 and its interactions with substrates, effectors, and regulators.