ATM phosphorylates the FATC domain of DNA-PKcs at threonine 4102 to promote non-homologous end joining

Abstract Ataxia-telangiectasia mutated (ATM) drives the DNA damage response via modulation of multiple signal transduction and DNA repair pathways. Previously, ATM activity was implicated in promoting the non-homologous end joining (NHEJ) pathway to repair a subset of DNA double-stranded breaks (DSBs), but how ATM performs this function is still unclear. In this study, we identified that ATM phosphorylates the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), a core NHEJ factor, at its extreme C-terminus at threonine 4102 (T4102) in response to DSBs. Ablating phosphorylation at T4102 attenuates DNA-PKcs kinase activity and this destabilizes the interaction between DNA-PKcs and the Ku-DNA complex, resulting in decreased assembly and stabilization of the NHEJ machinery at DSBs. Phosphorylation at T4102 promotes NHEJ, radioresistance, and increases genomic stability following DSB induction. Collectively, these findings establish a key role for ATM in NHEJ-dependent repair of DSBs through positive regulation of DNA-PKcs.


INTRODUCTION
DNA double-stranded breaks (DSBs) are cytotoxic DNA lesions that pose an immediate threat to genome stability, and failure to properly repair them can lead to cell death, chromosomal aberrations, or carcinogenesis ( 1 ). Complex mechanisms collecti v el y termed the DN A damage response (DDR) hav e e volv ed in cells to manage DSBs. These mechanisms include DNA damage recognition, activation of signaling cascades and cell cycle checkpoints, chromatin remodeling, transcription regulation, and repair of the DSB. Two members of the phospha tid ylinositol-3-kinase-like kinase (PIKK) famil y, DN A-PK cs (DN A-dependent protein kinase catalytic subunit) and ATM (ataxia telangiectasia-phosphorylates many components of the NHEJ pathway, such as XLF, Artemis, and DNA-PK cs , but there is limited knowledge of the functionality of these phosphorylation e v ents ( 20 ). ATM plays at least one important additional role in NHEJ that has not been elucidated, but it is redundant with the core NHEJ factor XLF and thus not normall y a pparent ( 21 ). Finall y, it has been reported that ATM is dispensable for NHEJ but promotes repair fidelity via an undefined mechanism ( 22 ).
Here, we uncover a novel function for ATM in NHEJ via phosphorylating DNA-PK cs . The data show that DNA-PK cs is phosphorylated by A TM in its FA TC domain at threonine 4102 in response to DSBs. Ablating phosphoryla tion a t T4102 decreased DNA-PK cs kinase activity and results in destabilization of the interaction between DNA-PK cs and the Ku-DNA complex. Phosphorylation of DNA-PK cs at threonine 4102 results in stabilization of the NHEJ machinery at DSBs, leading to increased NHEJ efficiency and genomic instability. Our results establish a model in which phosphorylation of the FATC domain of DNA-PK cs by ATM promotes efficient NHEJ.

Irradiation
Cells were irradiated with ␥ -rays generated by a Mark 1 137 Cs irradiator (J.L. Shepherd and Associates) with a dose of 10 Gy, unless otherwise indicated in the figure.

Generation of antibody against phosphorylated DNA-PK cs at T4102
Anti-pT4102 polyclonal antibodies were generated by immunizing New Zealand white rabbits with KLH (keyhole limpet hemacyanin)-conjugated phosphopeptide GLSEET[PO 3 ]QVKCLMC. The phospho-specific antibodies were first passed through the corresponding unphosphoryla ted peptide-conjuga ted Sepharose CL-4B column (Pierce) to deplete IgGs that were not phosphospecific. The flow through IgGs were then affinity-purified using a phospho-peptide column. Eluted anti-pT4102 polyclonal antibodies were verified with phospho and non-phospho peptides via dot plot analysis.

Immunoprecipitation (IP) assays
To detect phosphorylation of DNA-PK cs at T4102, V3 cells stab ly e xpr essing YFP-tagged wild-type DNA-PK cs wer e irradiated with a dose of 10 Gy and allowed to recover for 30 min. The cells were then washed twice with cold PBS, harvested, and lysed using IP lysis buffer (50 mM Tris-HCl pH 7.4, 500 mM NaCl, 0.5% NP-40, and 10% glycerol with 1 × protease and phosphatase inhibitor cocktails (Thermo Fisher)). The lysates were sonicated on ice and then cleared of cellular debris by centrifuging at 20 000 × g for 30 min. 2 mg of total protein was incubated with the DNA-PK cs monoclonal antibody  and Protein A / G beads (Thermo Fisher) overnight at 4 • C with mixing. The following day the beads were washed three times with lysis buffer and then twice with washing buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% NP-40 and 10% glycerol). Following the final wash, the beads were divided two parts, with one resuspended in 1 × SDS sample buffer and other treated with lambda protein phosphatase ( PP) (New England Biosciences) at 30 • C for 30 min to remove phosphorylation e v ents, the samples were washed, and finally resuspended in 1 × SDS sample buffer. The samples were resolved via SDS-PAGE and phosphoryla tion a t T4102 was assessed via immunoblotting.
To investigate pr otein-pr otein interactions, CHO V3 cells  and V3 cells complemented with WT, T4102A, or T4102D were irradiated with a dose of 10 Gy and allowed to recover for 10 min. Subsequently, the irradiated cells were washed three times in cold PBS, harvested, and lysed using IP Lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 0.4% NP-40, 0.6% Triton X-100, 1 × protease inhibitor cocktail, 1 × phosphatase inhibitor cocktail 2, 20 U / ml Benzonase (Novagen)) on ice. The lysates were sonicated on ice and then cleared of cellular debris by centrifuging at 20 000 × g for 30 min. 2 mg of total protein was incubated with 2 g anti-GFP antibody, and 30 l of Protein A / G magnetic agarose beads (Thermo Fisher) overnight with spinning at 4 • C. The beads were washed with IP washing buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.2% Triton X-100) for 5 times, and boiled in 1 × SDS sample buffer. The samples wer e r esolved via SDS-PAGE and immunoblotting was performed for the proteins indicated in the figures.

In vitro phosphorylation assay
Purification of ATM from HT1080 cells stab ly e xpressing FLAG-tagged ATM and DNA-PK cs from CHO V3 cells stab ly e xpressing YFP-tagged DNA-PK cs was performed as previously described with some modifications ( 23 ). To activate the ATM protein, the cells were irradiated with a dose of 10 Gy and allowed to recover for 30 min. Next, the cells were washed three times with cold PBS, harvested, and lysed using Purification Lysis Buffer (50 mM Tris-HCl pH 7.4, 500 mM NaCl, 0.5% NP-40, and 10% glycerol with 1 × protease and phosphatase inhibitor cocktails (Thermo Fisher)). The lysates were sonicated on ice and then cleared of cellular debris by centrifuging at 20 000 × g for 30 min. 2 mg of total proteins was incubated with 30 l M2 FLAG magnetic beads (Sigma-Aldrich) for ATM and 2 g DNA-PK cs antibody (25-4) and 30 l Protein A / G beads for DNA-PK cs . After an overnight incuba tion a t 4 • C , the beads were washed twice with Purification Lysis Buffer, twice with Wash Buffer 1 (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% NP-40 and 10% Glycerol) and finally twice with Wash Buffer 2 (20 mM Tris-HCl pH 7.4, 50 mM KCl, and 10% Glycerol). For ATM protein, half of beads were resuspended in 1 × SDS sample buffer and the sample was resolved via SDS-PAGE, and the other half was for the in vitro phosphorylation assay. To remove phosphorylations associated with the purified DNA-PK cs protein, the beads were further equilibrated in 1 × PP reaction buffer (New England Biosciences) and treated with PP at room temperature for 30 min. The sample was washed using the protocol described abov e. One-thir d of the beads containing DNA-PK cs was resolved via SDS-PAGE, and the rest were utilized for the ATM-mediated in vitro phosphorylation assay. The SDS-PAGE gel that resolved purified YFP-DNA-PK cs and FLAG-ATM was Coomaissie Blue stained to show purification of each protein.
In vitro phosphorylation of DNA-PK cs by ATM was conducted by mixing beads containing YFP-DNA-PK cs with beads containing FLAG-tagged ATM or control beads. The bead mixtures were placed in Kinase Reaction Buffer (20 mM HEPES 7.4, 50 mM KCl, 2 mM MgCl 2 , 2 mM ATP, 1 mM DTT and 5% glycerol) and incuba ted a t 37 • C for 1 h with mixing. The reactions were terminated by adding 1 × SDS sample buffer and boiling the samples at 95 • C for 5 min. The phosphorylation of DNA-PK cs at T4102 was then assessed by immunoblotting with the pT4102 antibody.

In vitro kinase assay
DNA-PK cs kinase activity was assessed by examining phosphorylation of H2AX peptide in vitro , using a modified version of our published protocol ( 24 ). Briefly, the reaction was conducted in a 10 l mixture containing 1 × kinase buffer (25 mM Tris-HCl, pH 7.9, 5 mM MgCl 2 , 1 mM DTT, 25 mM KCl and 10% glycerol), 0.1 g sonicated herring DNA, 20 nM Ku70 / 80, 0.1 g biotin-labeled H2AX (biotin-AVGKKASQASQEY) and 2.5 g of nuclear extract from V3, V3 + DNA-PK cs WT, V3 + DNA-PK cs T4102A, or DNA-PK cs T4102D cell lines. After a 30 min incuba tion a t 30 • C , the reactions were terminated by the addition of 1 l 0.5 M EDTA, and the biotinylated-H2AX peptide was captured using a SAM2 Biotin Capture Membrane (Promega) and the membrane w as w ashed following the manufacturer's suggested protocol. H2AX phosphorylation was detected using an anti-H2AX pS139 antibody via immunoblotting and signal was quantified using ImageJ (1.53e). Background phosphorylation of H2AX was observed in the DNA-PK cs null V3 cell line and this was subtracted from the other samples' readouts. 100% kinase activity was normalized using the V3 cells expressing WT human DNA-PK cs . The r eported r esults ar e deri v ed from three independent experiments.

Fluorescent immunostaining and microscopy
IR-induced 53BP1 foci kinetics were monitored in G1 cells as previously described with modifications ( 25 , 26 ). Briefly, CHO V3 cells and V3 cells complemented with WT, T4102A or T4102D were seeded on 'PTFE' Printed Slides (Electr on Micr oscopy Sciences) and two days later the cells were mock treated or irradiate with a dose of 2 Gy. At different time points after IR (0.5, 1, 3 or 7 h), the cells were washed twice with cold PBS and fixed with 4% paraformaldehyde (in PBS) for 20 min at room temperature, washed fiv e times with PBS, and incubated in 0.5% Triton X-100 on ice for 10 min. Cells were washed fiv e times with 1 × PBS and incubated in blocking solution (5% goat serum (Jackson Immuno Research) in 1 × PBS) for 1h. The blocking solution was replaced with the 53BP1 (ab175933, Abcam) and Cyclin A2 (ab16726, Abcam) primary antibodies (1:1000 dilution for both antibodies) diluted in 5% normal goat serum in 1 × PBS and the cells were incubated at 4 • C ov ernight. The ne xt day the cells were washed fiv e times with Wash Buffer (1% BSA in 1 × PBS). Next, the cells were incubated with anti-rabbit IgG conjugated with Alexa Fluor 488 (Molecular Probes) and anti-mouse IgG conjugated with Texas Red (Molecular Probes) (1:1000 dilution for both antibodies) secondary antibodies in 1% BSA, 2.5% goat serum in 1 × PBS for 1 h in the dark, followed by fiv e washes.
After the last wash, the cells were mounted in VectaShield Antifade mounting medium containing 4 ,6-diamidino-2phenylindole (DAPI). Images were acquired using a Zeiss AxioImager fluorescence microscope utilizing a 63 × oil objecti v e lens. The 53BP1 foci were only counted in the cells with no Cyclin A staining.

Laser micro-irradiation and real-time recruitment
Real-time recruitment of fluorescent tagged DNA-PK cs WT, T4102A, T4102D, Ku80, XRCC4, XLF and PNKP in response to DSB induction was examined following laser micro-irradiation with a Carl Zeiss Axiovert 200M microscope with a Plan-Apochromat 63 ×/ NA 1.40 oil immersion objecti v e (Carl Zeiss) as pre viously described ( 24 , 26 ). The following cell lines were used in this study to measur e r ecruitment of the proteins to laser-generated DSBs. For DNA-PK cs , CHO V3 cells stab ly e xpressing YFPtagged DNA-PK cs WT, T4102A, or T4102D. For Ku80, XRCC4, XLF and PNKP recruitment, GFP-tagged Ku80, XRCC4, XLF or PNKP was transiently expressed in CHO V3 cells complemented with FLAG-tagged DNA-PK cs WT, T4102A or T4102D. The YFP / GFP or FLAG tag was attached the N-terminus in all proteins. The cells were seeded on a 35 mm glass-bottomed dish (Mattek) and incubated with 10 M BrdU. 24 h later, the medium was replaced with CO 2 -independent medium and placed in a chamber on the microscope that was set at 37 • C. To generate laser-induced DSBs, a 365-nm pulsed nitrogen laser (Spectra-Physics, Catalog #VSL337NDS2, purchased in May 2020) was set at 80% of maximum power output and micro-irradiation was performed using the pulsed nitrogen laser. Time-lapse images were taken using an AxioCam HRm camera (Carl Zeiss). Carl Zeiss Axiovision software (v4.91) was used to measur e fluor escence intensities of the micro-irradiated and control areas, and the resulting intensity of irradiated area was normalized to non-irradiated control area to obtain the alteration of the interested proteins as described previously ( 24 , 26 ).

Subcellular fractionation
The accumulation of DNA damage response proteins to chromatin following IR-induced DNA damage was examined as previously described with some modifications ( 27 ). Briefly, the V3 cells expressing either DNA-PK cs WT or T4102A wer e mock-tr ea ted or irradia ted with 10 Gy, allowed to recover for 10 or 30 min before being harvested. The harvested cells were incubated with CSK100 buffer containing 10 mM PIPES pH 6.8, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 , 1 mM EGTA, 0.2% Triton X-100 and protease inhibitor and phosphatase inhibitor cocktails (Thermo Fisher) for 40 min at 4 • C. The cells were then pelleted by centrifugation at 5000 × g and the resulting pellets were lysed with buffer 1 containing 50 mM HEPES pH 7.5, 50 mM NaCl, 0.05% SDS, 2 mM MgCl 2 , 10% glycerol, 0.1% Triton X-100, 1 × protease inhibitor and phosphatase inhibitor cocktails, and 10 Units of RNase-free DNase I. The chr omatin pr oteins were included in the supernatant after centrifuga tion a t 18 000 × g. The protein concentration of each sample was measured using a Pierce BCA Protein Assay kit (Thermo Fisher). 20 g of each fraction was resolved via SDS-PAGE, and then transferred to a PVDF membrane for immunoblotting.

NHEJ assay
CHO cells were seeded in triplica te a t a density of 40 000 cells per well in 12-well plates. The following day, cells were transfected with reporter plasmids using Lipofectamine 3000 (ThermoFisher) according to the manufacturer protocol. For each cell line, one well was transfected with undamaged plasmid cocktail, one well was transfected with NHEJ plasmid cocktail, and one well was left untransfected. After 24 h, cells were trypsinized and analyzed by flow cytometry using an Attune NxT flow cytometer. Compensation and gating were established by running untransfected cells along with cells transfected with individual undamaged plasmids expressing wild type fluorescent proteins (single color controls) as described previously ( 28 ). Each experiment was conducted three times on separate days.

Colony formation assay
Cell survival curves were obtained by measuring the colonyforming abilities of irradiated cell populations as previously described ( 29 ). CHO V3 cells and V3 cells complemented with YFP-tagged DNA-PK cs WT, T4102A or T4102D were mock treated or irradiated at doses of 1, 2, 4 or 6 Gy, or incuba ted with dif ferent concentra tion of etoposide or Camptothecin as indicated in the figure legend, and then plated on 60-mm plastic Petri dishes. After 8 days, cells were fixed with 100% ethanol and stained with 0.1% crystal violet in a 100% ethanol solution. Colonies were scored and the mean value for triplicate culture dishes was determined. Cell survival was normalized to plating efficiency of untreated controls for each cell type.

Chromosome aberration assay
To investigate genome stability following irradiation in CHO V3 cells and V3 cells complemented with WT, T4102A or T4102D, the cells were irradiated with a dose of 2 Gy and then allowed to recover in normal cell culture conditions for 24 hrs. Next, the cells were incubated with Colcemid at concentration of 0.1 g / ml for 4 h and then harvested by trypsinzation. After washing with PBS at room temperature, the cells were then incubated with warmed (37 • C) 75 mM KCl. Samples were processed and chromosomal abnormalities were scored as previously described ( 25 ).

RESULTS
ATM phosphorylates DNA-PK cs in its FATC domain at threonine 4102 in response to DNA double-stranded breaks DNA-PK cs is composed of HEAT (Huntington-elongation factor 3A-PP2A subunit-TOR) repeats, which are differentiated into a unique N-terminal domain (N-HEAT, amino acids 1-892) and a central unit called the M-HEAT domain (amino acids 893-2801) and a C-terminal region that contains the kinase domain (amino acids 3565-4100), which is flanked N-terminally by the FA T (FRAP, A TM, TRRAP) domain and C-terminally by the F ATC (F AT C-terminal) domain ( 30 ). The FATC domain comprises the extreme Cterminus of DNA-PK cs and is a highly conserved domain of a pproximatel y 30 amino acids ( 31 ). Early work showed that the FATC domain is indispensable for DNA-PK cs activity, as deletion or mutagenesis in this domain deleteriously affects the function of DNA-PK cs ( 32 ). This drove us to postula te tha t modula tion of DNA-PK cs may be regulated by a post-transla tional modifica tion in the FATC domain. Mass spectrometry analysis of an in vitro auto phosphorylation assay previously identified that DNA-PK cs is phosphorylated in the FATC domain at threonine T4102 (pT4102) ( 33 ). Howe v er the biological relevance of this phosphorylation e v ent has not been pre viously inv estigated, propelling us to determine if this site regulates the function of DNA-PK cs . Alignment of DNA-PK cs orthologues identified that T4102 is highly conserved in most primates, rodents, and birds ( Figure 1 A and Supplementary Figure S1), but this site is less conserved in other vertebrates (Supplementary Table S1) ( 34 ). Analysis of the DNA-PK cs structure shows that T4102 is surface exposed and thus a viable phosphorylation target (Figure 1 B) ( 7 , 35 ). T4102 is positioned at the beginning of the DNA-PK cs FATC domain and proximal to the FAT and kinase domains, and in the context of the NHEJ long-range synaptic complex it is not near the DNA-PK cs dimer interface, DNA or other core NHEJ proteins. A model of phosphorylated T4102 shows that the phosphoryl group will clash with S4099 and repel E4093 within the kinase domain (Figure 1 B). Conformational changes within DNA-PK cs will ther efor e be required to accommodate phosphorylated T4102, which may be facilitated by se v eral other charged or polar residues from the FAT, kinase, and FATC sub-domains in this region.
To examine if DNA-PK cs is phosphorylated at this amino acid after DNA damage in cells, a phospho-specific antibody to T4102 was generated and validated using dot blot analysis (Supplementary Figure S2A). DNA-PK cs is phosphoryla ted a t T4102 (pT4012) following exposure to ionizing radiation (IR) and the signal was lost when the (ov ervie w) and 7OTP (zoomed view) ( 7 , 35 ). T4102 residue is highlighted in magenta. DNA-PK cs domains are highlighted thusly: FATC domain (green), kinase domain (yellow), FAT domain (blue) and N-terminal HEAT repeat domains (gray). Ku70 (light green) and Ku80 (light orange) are highlighted and other NHEJ proteins are in white. Zoomed view compares T4102 with modeled phosphorylated T4102, with nearby polar and charged residues shown as sticks, ( C ) DNA-PK cs is phosphorylated at T4102 (pT4102) after ionizing radiation (IR) and this signal is lost when treated with lambda protein phosphatase ( PP). YFP-tagged DNA-PK cs was immunoprecipitated from CHO V3 cells stably expressing YFP-DNA-PK cs using an anti-DNA-PK cs antibody (25-4) 20 min following treatment with 10 Gy of ␥ -rays. Half of the beads containing DNA-PK cs wer e tr ea ted with PP to dephosphoryla te the protein. ( D ) Ablating the T4102 phosphorylation site via alanine substitution (T4102A) results in loss of IR-induced phosphorylation at T4102. CHO V3 cells expressing YFP-tagged DNA-PKcs WT or T4102 were irradiated with a dose of 10 Gy of ␥ -rays, allowed to recover for 20 min, and immunoblotting was performed to examine the phosphorylation of DNA-PK cs at T4102. ( E ) Tracking IR-induced dose-dependent pT4102. CHO V3 cells stably expressing YFP-DNA-PK cs wer e tr eated with the indicated doses of IR, allowed to recover for 20 min, and then phosphorylation of DNA-PKcs at T4102 and H2AX at S139 were assessed via immunoblotting. ( F ) Time course of pT4102 following treatment with 10 Gy of ␥ -rays. CHO V3 cells expressing YFP-tagged DNA-PKcs wer e tr eated with a dose of 10 Gy of IR and allowed to recover for the times indicated in the figure. Phosphoryla tion of DNA-PK cs a t S2056 and T4102, and KAP1 at S824 were assessed via immunoblotting. ( G ) Phosphorylation of DNA-PK cs at T4102 following treatment with various DNA dama ging a gents. CHO V3 cells stab ly e xpr essing YFP-DNA-PKcs wer e tr ea ted for 1 h with 200 ng / ml neocarzinosta tin (NCS), 1 M etoposide (ETO), 1 M camptothecin (CPT), 0.5 g / ml mitomycin C (MMC) or 50 g / ml methyl methanesulfonate (MMS). Cells were then harvested and immunoblotting was performed to assess phosphorylation of DNA-PK cs at S2056 and T4102, KAP1 at S824, and H2AX at S139. ( H ) Cell cycle-dependent phosphorylation of DNA-PK cs at T4102 following IR. Cells were treated with a double thymidine blocked and then released and collected at selected times to enrich cells in G1 and S / G2 phases of the cell cycle, irradiated with a dose of 10 Gy, and harvested 30 min la ter. Phosphoryla tion of DNA-PK cs at T4102, KAP1 at S824, and ATM at S1981 were assessed via immunoblotting. ( I ) Inhibition of ATM, but not ATR or DNA-PK cs , blocks IR-induced DNA-PK cs phosphoryla tion a t T4102. CHO V3 cells stably expr essing YFP-DNA-PK cs wer e pr etr eated with DMSO, 5 M DNA-PKcs inhibitor NU7441 (NU), ATM inhibitor KU55933 (KU) or ATR inhibitor VE821 (VE) for 2 h and then irradiated with a dose of 10 Gy and allowed to recover for 30 min. Cells were then harvested and immunoblotting was performed to assess phosphorylation of DNA-PK cs at S2056 and T4102 and H2AX at S139. ( J ) IR-induced phosphoryla tion of DNA-PK cs a t T4102 is lost in the A TM-deficient cell line A T5BIVA. A TM-deficient A T5BIVA cells and the cells stab ly e xpressing ATM were irradiated with a dose of 10Gy of ␥ -rays and allowed to recover for 0.5 or 1 h. Cells were then harvested and immunoblotting was performed to assess phosphorylation of DNA-PK cs at S2056 and T4102, KAP1 at S824, and A TM at S1981. ( K ) A TM phosphorylates DNA-PK cs in vitro . FLAGtagged ATM was isolated from irradiated HT1080 cells stably expressing FLAG-tagged ATM using anti-FLAG M2 sepharose. DNA-PKcs was isolated from untreated CHO V3 cells stably expressing YFP-tagged DNA-PKcs using an anti-DNA-PKcs antibody . The two samples were mixed together in a kinase reaction buffer for 1 hr. The reaction was terminated and immunoblotting was performed to assess phosphorylation of DNA-PKcs at T4102 and ATM at S1981. sample was treated with lambda phosphatase, supporting that the antibody recognizes a phosphorylation e v ent (Figure 1 C). Next, the specificity of the phospho-antibody was assessed by complementing the DNA-PK cs deficient Chinese Hamster Ovary (CHO) cell line V3 with YFPtagged wild-type DNA-PK cs (WT) or DNA-PK cs in which the phosphorylation site at T4102 was ablated via alanine substitution (T4102A) (Supplementary Figure S2B). The signal was significantly decreased in cells expressing the T4102A mutant protein, indicating the antibody specifically r ecognizes pT4102 (Figur e 1 D). Furthermor e, pT4102 occurs following IR exposure in a dosage-dependent manner (Figure 1 E), with observation of the pT4102 signal starting at 5 minutes, peaking at 30-60 min, and still detectable 24 h post-IR (Figure 1 F). This phenotype is conserved in human cells, as pT4102 initiates 5 min and peaks 60 min post-IR in the human cell line U2OS (Supplementary Figure  S2C), and IR-induced pT4102 is also observed in HeLa cells (Supplementary Figure S2D). In addition to IR, treatment with the radiomimetic agent neocarzinostatin (NCS) and the topoisomerase 2 inhibitor etoposide (ETO) induced robust pT4102 signal (Figure 1 G). Modest phosphorylation of DNA-PK cs at T4102 was observed after treatment with the topoisomerase 1 inhibitor camptothecin (CPT), but limited to no phosphorylation occurred following treatment with the DNA cross-linking agent mitomycin (MMC) and DNA alkylating agent methyl methanesulfonate (MMS) (Figure 1 G). As IR, NCS and ETO directly generate DSBs and CPT-induced DNA damage can be processed to form DSBs, the data suggests that DSBs induce phosphorylation of DNA-PK cs at T4102. We then assessed if IR-induced phosphoryla tion of DNA-PK cs a t T4102 is cell cycle specific. Cells were synchronized via a double thymidine block and then released to allow examination of IR-induced phosphorylation of DNA-PK cs at T4102 in G1 and S / G2 phases of the cell cycle. We observed that pT4102 occurs in both G1 and S / G2 phases, but phosphorylation is more prominent in G1 phase of the cell cycle (Figure 1 H). We next aimed to identify the kinase responsible for phosphorylating DNA-PK cs at T4102 in response to DNA damage. As phosphoryla tion a t this site is induced by DSBs, we focused on the DNA damage-responsi v e kinases, including DNA-PK cs , A TM, and A TR. Pr etr eatment of cells with the inhibitors NU7441, KU55933 and VE-821 to block DNA-PK cs , ATM and ATR acti vity, respecti v ely, shows that pT4102 is significantly attenuated in cells pr etr eated with KU55933 but not NU7441 or VE-821 (Figure 1 I). This da ta indica tes tha t ATM phosphoryla tes DNA-PK cs a t T4102 in response to DNA damage. To verify this result, we assessed pT4102 in the ATM deficient cell line AT5 and AT5 cells complemented with ATM ( 36 ). IR-induced phosphorylation of DNA-PK cs at T4102 is almost completely lost in the ATM-deficient cell line and this was rescued by complementing AT5 cells with wild-type ATM (Figure 1 J). Finally, we examined the ability of ATM to phosphorylate DNA-PK cs at T4102 in vitro . DNA-PK cs was isolated from cells and incubated with purified ATM in the presence of ATP (Supplementary Figure S3). ATM autophosphorylates at S1981 in this system and it phosphorylates DNA-PK cs at threonine 4102 as monitored by the pT4102 antibody (Figure 1 K). Collecti v ely, the data illustrates that ATM phosphoryla tes DNA-PK cs a t T4012 in response to DNA damage.

Ablating the T4102 phosphorylation site attenuates DNA-PK cs activity and destabilizes the interaction between DNA-PK cs and the Ku-DNA complex
As T4102 lies in the FATC domain, we next assessed if phosphorylation of this site modulates DNA-PK cs kinase activity. We complemented V3 cells with YFP-tagged WT, T4102A, and the DN A-PK cs m utation in which T4102 was mutated to aspartic acid to mimic phosphorylation at this amino acid (T4102D) (Supplementary Figure 2B). Subsequently, we examined the kinase activities of these proteins in vitro using H2AX peptide as a substrate ( 24 ). We observed that the T4102A mutant protein had significantly reduced ( ∼ 40% decrease) activity compared to WT and T4102D proteins in vitro (Figure 2 A). Next, we assessed DNA-PK cs -media ted phosphoryla tion in response to DSBs in vivo . Previously, we and others identified that at early time points (3-15 minutes) post-IR treatment that DNA-PK cs phosphoryla tes H2AX a t S139 and KAP1 a t S824 ( 24 , 37 ). Ther efor e, we examined DNA-PK cs autophosphorylation and phosphorylation of H2AX and KAP1 in V3 cells complemented with DNA-PK cs WT and the kinase defecti v e T4102A in response to IR-generated DSBs at these times points. We found that IR-induced autophosphorylation of DNA-PK cs at S2056 is decreased in T4102A cells compared to WT cells (Figure 2 B). Furthermore, phosphorylation of H2AX and KAP1 are decreased in T4102A cells at early time points post-IR treatment (Figure 2 B). This data illustra tes tha t abla ting the T4102 phosphoryla tion sites a ttenuates DNA-PK cs activity in vitro and in vivo . Next, we determined if the decrease in DNA-PK cs -mediated phosphorylation e v ents was due to a decrease in the DNA-PK cs -Ku interaction. We found that T4102A has decreased interaction with Ku following exposure to IR, compared to WT and T4102D (Figure 2 C). Consistent with previous results, no strong interaction between DNA-PK cs WT, T4102A, or T4102D and Ku was detected in the absence of DNA damage (Supplementary Figure S4A) (38)(39)(40). The disruption of the DNA-PK cs -Ku interaction was further examined via monitoring the dynamics of DNA-PK cs at laser-generated DSBs. We found that blocking T4102 phosphorylation results in decreased recruitment of DNA-PK cs to DSBs compared to WT and the T4102D proteins (Figure 2 D). Moreover, kinetic analysis shows that the T4102A mutant prema turely dissocia tes from laser-genera ted DSBs compared to the WT and T4102D proteins (Figure 2 E). As ATM media tes the phosphoryla tion of DNA-PK cs a t T4102, we examined if blocking ATM kinase activity affects the dynamics of DNA-PK cs at laser-generated DSBs. We found that inhibition of ATM kinase activity using the small molecule KU55933 did not alter the initial recruitment of DNA-PK cs to laser-induced DSBs (10 s after DSB induction), but significantl y reduced accum ulation / retention of DN A-PK cs at the DNA damage site (Supplementary Figure S4B). Collecti v ely, the da ta illustra tes tha t blocking phosphoryla tion of DNA-PK cs at T4102 results in a decrease in DNA-PK cs kinase activity, resulting in destabilization the DNA-PK cs -Ku complex. Ablating the T4102 phosphorylation site attenuates DNA-PK cs activity and destabilizes the interaction between DNA-PK cs and the Ku-DNA complex. ( A ) Measurement of DNA-PK cs in vitro kinase activity. Nuclear extracts from the CHO V3 cells and those stab ly e xpressing DNA-PK cs WT, T4102A, and T4102D were examined for their ability to phosphorylate a biotin-tagged H2AX peptide. H2AX phosphorylation was observed in the CHO V3 cells (DNA-PK cs null) cell line and this was subtracted from the other samples' readouts. The 100% kinase activity was normalized using the DNA-PK cs WT cell lysate results. The data ar e pr esented as the mean ± SD from three individual experiments. ( B ) Blocking DNA-PK cs phosphorylation at T4102 suppresses IR-induced DNA-PK cs -mediated phosphorylation e v ents. CHO V3 cells stably expressing YFP-tagged DNA-PKcs WT and T4102A were irradiated using a dose of 10 Gy of ␥ -rays and then allowed to recover at times indicated in the figure. Phosphorylation of DNA-PKcs at S2056, ATM a t S1981, KAP1 a t S824, and H2AX a t S139 were assessed via immunoblotting. ( C )T4102 phosphorylation promotes stabilization of the interaction between DNA-PK cs and the Ku70 / 80 heterodimer after IR. CHO V3 cells and V3 cells stab ly e xpressing YFP-tagged DNA-PK cs WT, T4102A and T4102D wer e tr eated with a dose of 10 Gy of IR and allowed to r ecover for 10 min and w hole cell l ysa tes were genera ted. DNA-PK cs was pulled down from the lysates using ChromoTek GFP-Trap ® Agarose beads and immunoblotting was performed to verify immunoprecipitation of DNA-PK cs and the ability of Ku70 and Ku80 to interact with DNA-PKcs was e xamined. ( D ) Ab la ting T4102 phosphoryla tion a ttenua tes the initial recruitment (up to 5 min) of DNA-PK cs to laser-induced DSBs. Relati v e fluorescent intensity of YFP-tagged wild type DNA-PK cs WT, T4102A, and T4102D are presented as mean ± SEM and significance was assessed via a student's t-test. The bar indicates 5 m. ( E ) T4102 phosphorylation promotes accumulation / retention (up to 80 min) of DNA-PK cs at laser-induced DSBs. Relati v e fluorescent intensity of YFP-tagged wild type DNA-PK cs WT, T4102A and T4102D are presented as mean ± SEM and significance was assessed via a student's t-test. The bar indicates 5 m.

Stabilization of the NHEJ machinery at DSBs is promoted by phosphorylation of DNA-PK cs at T4102
As DNA-PK cs has been implicated in promoting the longrange synaptic complex and T4102 promotes stabilization of the DNA-PK cs -Ku complex, we next assessed if phosphoryla tion a t this site af fects the d ynamics of the NHEJ machinery at DSBs. First, the recruitment of the core NHEJ factors Ku80, XRCC4 and XLF to laser-generated DSBs was examined in V3 cells complemented with FLAG-tagged DNA-PK cs WT, T4102A and T4102D ( Supplementary Figure S5A). The initial dynamics of Ku80 is similar in all three cell lines, illustra ting tha t T4102 phosphoryla tion does not regulate the ability of the Ku heterodimer to bind to DSBs or its initial dynamics at DSBs (Figure 3 A). We observed that GFP-tagged XRCC4 and XLF are quickly recruited to laser-induced DSBs in WT, T4102A, and T4102D cells (Figure 3 B, C). Howe v er, XRCC4 and XLF signal wanes in the T4102A complemented cells compared to the WT and T4102D cells (Figure 3 B, C), suggesting that stability of the NHEJ machinery is decreased when phosphorylation of DNA-PK cs at T4102 is block ed. Moreo ver, we extended this to the DNA end processing factor PNKP. Similar to XRCC4 and XLF, PNKP r ecruitment / r etention at laser-generated DSBs was decreased in T4102A complemented cells compared to WT and T4102D (Figure 3 D). Finally, we observed that IR-induced recruitment of XRCC4, LIG4, and DNA-PK cs to the chromatin faction was significantly decreased in the T4102A cells compared to WT cells, supporting the observation that blocking T4102 phosphoryla tion af fects the d ynamics of NHEJ factors a t DSBs (Supplementary Figure S5B). Together, the data show that ATM-media ted phosphoryla tion of DNA-PK cs a t T4102 promotes stabilization of the NHEJ machinery at DSBs.

Phosphorylation of DNA-PK cs at T4102 is important for NHEJ-mediated DSB repair
As blocking phosphorylation of DNA-PK cs at T4102 destabilizes the NHEJ machinery at DSBs, we next examined if blocking phosphorylation at this site affects NHEJ. First, we assessed IR-induced 53BP1 focus formation and resolution in G1 cells, which was used as an indirect marker for DSB repair and NHEJ. To make 53BP1 foci enumerable, we used a dose of 2 Gy of IR, which also induced phosphorylation of DNA-PK cs at T4102 (Supplementary Figure S6A). As shown in Figure 4 A, we found at 30 minutes post-IR, the number of 53BP1 foci is the same in V3, WT, T4102A, and T4102D cells. Howe v er, at 1-, 3-and 7-hours post-IR, 53BP1 focus resolution was a ttenua ted in V3 and T4102A cells compared to WT and T410D cells. 53BP1 focus resolution is more a ttenua ted in the DNA-PK cs -deficient V3 cells compared to the T4102A cells, suggesting that T4102 phosphorylation does not result in a complete abolishment of NHEJ. Next, NHEJ was monitored via a FM-HCR assay ( 28 ). We observed that the loss of DNA-PK cs (V3) results in almost complete loss of NHEJ, which was rescued by expression of WT DNA-PK cs (Figure 4 B). The T4102A mutant complemented cells have a modest ( ∼20% decrease), but significant, decrease in NHEJ repair efficiency compared to WT, while the T4102D complemented cells have a significant increase ( ∼50%) increase in NHEJ efficiency compared to WT cells (Figure 4 B), indica ting tha t phosphorylation of T4102 promotes NHEJ. As NHEJ is decreased in T4102A cells, we determined if blocking this phosphorylation e v ent r esults in incr eased sensitivity to DNA dama ging a gents , including radiation, etoposide , and CPT. We observed that T4102A cells ar e mor e sensiti v e to radiation ( Collecti v ely, the findings in this study show that ATM phosphorylates the FATC domain of DNA-PK cs at T4102 in response to DSBs. Moreover, phosphorylation at T4102 promotes stabilization of the NHEJ machinery at DSBs, and ultimately, NHEJ and genome stability. This study provides further insight into how ATM functions in the response to DSBs by promoting NHEJ.

DISCUSSION
DN A-PK cs is heavil y phosphorylated in response to DSBs ( 41 ). The best characterized DNA-PK cs phosphorylation sites include T2609, S2612, T2620, S2624, T2638 and T2647 (collecti v ely called the T2609 or ABCDE cluster) and S2023, S2029, S2041, S2051 and S2056 (termed the S2056 or PQR cluster) and phosphorylation of both clusters is required for NHEJ and modulation of DSB repair pathway choice. The S2056 / PQR and T2609 / ABCDE phosphorylation clusters have opposing roles in protecting DSBs, with S2056 / PQR protecting the DNA ends and T2609 / ABCDE promoting DNA end processing ( 9 ). Phosphorylation of the T2609 / ABCDE cluster functions by destabilizing the binding of DNA-PK cs to the DNA-Ku complex, triggering its release from DSBs to allow access of DNA ends for processing, including the terminal ligation step of NHEJ ( 40 ). The kinase activity of DNA-PK cs is modulated by phosphoryla tion a t T3950 and S56 / S72, whereas phosphoryla tion at T946 / S1004 inhibits NHEJ without affecting the enzymatic activity of DNA-PK cs ( 42 ). Here, we add to the importance of DNA-PK cs phosphorylation in modulating its functionality, as we show that in response to DSBs, ATM phosphorylates DNA-PK cs in the FATC domain of the protein at T4102. This phosphorylation e v ent promotes stabilization of the DNA-PK cs -Ku interaction and the stabilization of the NHEJ machinery at DSBs, both of which stimulate NHEJ.
A body of evidence shows that crosstalk occurs between DNA-PK cs and ATM, and that they may cooperati v ely initiate DSB repair signaling and regulate DSB repair ( 20 , 32 ). First, combined deficiency of DNA-PK cs and ATM leads to synthetic lethality in mice ( 43 , 44 ). Second, DNA-PK cs and ATM phosphorylate many common targets r equir ed for DDR signaling and DSB repair, including H2AX, KAP1 and components of the NHEJ pathway, such as XLF and Artemis ( 2 ). Third, DNA-PKcs and ATM phosphorylate each other to modulate specific activities. For example, ATM phosphorylates DNA-PK cs at S3205 and at the T2609 / ABCDE cluster in response to DNA  . Phosphorylation of DNA-PK cs at T4102 is important for NHEJ-mediated DSB repair. ( A ) IR-induced 53BP1 foci resolution in G1 cells is a ttenua ted in T4102A cells compared to WT and T4102D cells. CHO V3 cells and V3 cells complemented with FLAG-tagged DNA-PK cs WT, T4102A or T4102D were irradiated with 2 Gy of ␥ -rays and 53BP1 foci formation and resolution was assessed 0.5, 1, 3 and 7 h post-IR. 53BP1 foci at each time point were calculated in over 50 Cyclin A-negative cells and the data are presented as Mean ± SEM. Student's t -test (two-sided) was performed to assess statistical significance (n.s., not significant; ** P < 0.01; *** P < 0.001; **** P < 0.0001). ( B ) DNA-PK cs phosphorylation at T4102 promotes NHEJ. NHEJ reporter and control plasmids were transfected into CHO V3 cells and V3 cells stably expressing YFP-tagged DNA-PK cs WT, T4102A and T4102D and NHEJ efficiency was calculated using flow cytometry analysis as described in the Materials and Methods. The data ar e pr esented as mean ± SD with P -value from three independent repeats. * P < 0.05. (C-E) T4102A complemented cells are sensiti v e to DNA damaging agents including ionizing radiation ( C ), etoposide ( D ) and camptothecin (CPT) ( E ), compared to WT and T4102D cells. Clonogenic survival assays were performed to compare the radiation sensitivities of CHO V3 cells and V3 cells complemented with WT, T4102A or T4102D. Cells were irradiated at the indicated doses or incubated with etoposide or CPT at the indicated concentrations, and plated for analysis of survival and colony-forming ability. Two-sided Student's t -test were presented for comparison between V3 cells expressing YFP-tagged DNA-PK cs WT and T4102A. * P < 0.05; ** P < 0.01. (F, G) Blocking T4102 phosphorylation increases IR-induced chromosome and chromatin breaks ( F ) and chromosome fusions ( G ). Metaphase spreads were performed following treatment with 2 Gy of IR in CHO V3 cells and V3 cells stably expressing YFP-tagged DNA-PK cs WT, T4102A and T4102D. Chromosome and chromatin breaks as well as inter / intra-chromosome fusions were enumerated in at least 50 cells. Data are presented as mean ± SD and significance was assessed via a two-sided Student's t -test. n.s., not significant; **** P < 0.0001. damage ( 42 , 45 ). ATM-mediated phosphorylation of the T2609 / ABCDE cluster is essential for NHEJ and allows the freeing of DNA ends for processing by factors including the endonuclease Artemis and the dissociation of DNA-PK cs to allow DNA end ligation (46)(47)(48). DNA-PK cs also phosphorylates the T2609 / ABCDE cluster in trans ( 8 , 47 , 49 , 50 ), suggesting that phosphorylation of this cluster by ATM or DNA-PK cs may be context dependent. Moreover, ATM phosphorylation of DNA-PK cs overcomes DNA-PK cs -Ku inhibition of resection in vitro ( 51 ). Lastl y, DN A-PKcs inhibits ATM activity upon DNA damage via phosphorylation at multiple sites to regula te ATM-media ted signaling ( 52 ). This study adds to the interplay between DNA-PK cs and ATM and supports another positi v e role in ATM promoting NHEJ via phosphorylation of DNA-PK cs .
A significant number of single molecule studies have produced data showing the makeup of the NHEJ machinery at DNA ends. In particular, two synaptic complexes, the longrange and short-range comple xes, hav e been identified and these complex es ar e belie v ed to protect the DSB ends and then position them for ligation, respecti v ely ( 6 , 7 ). Although powerful, these studies provide limited insight into the dynamics of the NHEJ machinery at DSBs or how DNA end processing enzymes are recruited. Here, we provide evidence tha t phosphoryla tion of DNA-PK cs in its FATC domain a t T4102 promotes stabilization of DN A-PK cs at DSBs, w hich supports assembly and maintenance of the NHEJ core complex at the DNA damage site. We found that ablating the T4102 site a ttenua tes ov erall DNA-PK cs kinase acti vity and tha t this correla tes with decreased interaction with the Ku-DNA complex and stabilization of the NHEJ machinery. We hypothesize that the decrease in the NHEJ core complex stability at the DNA damage site is due to the attenuated DNA-PK cs kinase activity in the T4102A mutant. This is supported by previous data showing that the transition from the NHEJ long-range to the short-range complex and recruitment of factors r equir ed for DNA end processing are modulated by DNA-PK cs kinase activity ( 6-9 , 30 ). It is also possible that T4102 phosphorylation of DNA-PK cs induces long-range conformational changes in DNA-PK cs that are important for mediating DNA-PK csmediated pr otein-pr otein interactions with the core NHEJ factors. The structural predictions suggest that conformational changes within DNA-PK cs will be r equir ed to accommoda te phosphoryla ted T4102, which may be facilitated by residues from the FAT, kinase, or FATC sub-domains in this region. The modeling does not provide evidence that phosphoryla tion a t T4102 will directly affect the interaction between DNA-PK cs with the Ku heterodimer or another core NHEJ factor, but we hypothesize that this phosphorylation e v ent ma y induce an allosteric conf ormational change that allows the stabilization of the long-range complex. Furthermore, ablating the T4102 phosphorylation site attenuates the recruitment of the DNA end processing factor PNKP to DSBs, suggesting that DNA-PK cs activity and / or a conformational change upon DNA-PK cs phosphorylation at T4102 may promote the processing of unligatable DSBs. We postula te tha t this allows ATM protein to ensure the complete repair of DSBs by stabilizing the NHEJ complexes to favor the processing and correct joining of DNA ends.

DA T A A V AILABILITY
All study materials will be made available to other r esear chers. Please contact Anthony J. Davis (anthon y.davis@utsouthwestern.edu) f or reagents.