Ataxia‐telangiectasia‐mutated dependent phosphorylation of Artemis in response to DNA damage

Abstract

Artemis plays a crucial role in the hairpin‐opening step of antigen receptor VDJ gene recombination in the presence of catalytic subunit of deoxyribonucleic acid (DNA)‐dependent protein kinase (DNA‐PKcs). A defect in Artemis causes human radiosensitive‐severe combined immunodeficiency. Cells from Artemis‐deficient patients and mice display increased chromosomal instability, but the precise function of this factor in the response to DNA damage remains to be elucidate. In this study, we show that Artemis is hyperphosphorylated in an Ataxia‐telangiectasia‐mutated (ATM)‐ and Nijmegen breakage syndrome 1 (Nbs1)‐dependent manner in response to ionizing radiation (IR), and that S645 is an SQ/TQ site that contributes to retarded mobility of Artemis upon IR. The hyperphosphorylation of Artemis is markedly reduced in ATM‐ and Nbs1‐null cells. Reintroduction of wild‐type ATM or Nbs1 reconstituted Artemis hyperphosphorylation in ATM‐ or Nbs1‐deficient cells, respectively. In support of this functional link, hyperphosphorylated Artemis was found to physically associate with the Mre11/Rad50/Nbs1 complex in an ATM‐dependent manner in response to IR‐induced DNA double strand breaks (DSB). Since deficiency of either DNA‐Pkcs or ATM leads to defective repair of IR‐induced DSB, our finding places Artemis at the signaling crossroads downstream of DNA‐PKcs and ATM in IR‐induced DSB repair. (Cancer Sci 2005; 96: 134–141)

Artemis was originally identified by positional cloning as a causative gene for human radiation sensitive‐severe combined immunodeficiency (SCID). ⁽¹⁾ Artemis‐deficient patients are completely devoid of circulating T and B lymphocytes due to a defect in the coding joint formation of VDJ recombination process. Purified Artemis possesses single strand‐specific exonuclease activity and in the presence of catalytic subunit of deoxyribonucleic acid (DNA)‐dependent protein kinase (DNA‐PKcs), acquires endonucleolytic activity on overhangs as well as hairpins. ⁽²⁾ The cells from both Artemis‐deficient patients and mice display increased sensitivity to DNA double strand breaks (DSB) and increased spontaneous chromosomal instability, including telomere fusions and chromosomal rearrangements. This indicates that Artemis is also required for maintenance of genomic integrity and protection against lymphomagenesis. ³ , ⁴ , ⁵ , ⁶ , ⁷

DNA DSB in mammalian cells induce a series of responses that result in cell cycle arrest, DNA repair, gene transcription and cell death. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² Four members of the phosphatidylinositol 3‐kinase‐like protein kinase (PIKK) family, DNA‐PKcs, Ataxia‐telangiectasia‐mutated (ATM), Ataxia‐telangiectasia‐mutated and Rad3‐related (ATR), and human suppressor with morphogenetic effect on genitalia (hSMG‐1) are critical regulators of this process. Among PIKK family proteins, DNA‐PKcs is the best known member, and is critical for the non‐homologous end‐joining (NHEJ) pathway of DSB repair. ⁽¹³⁾ The DNA‐PK holoenzyme contains a large (450‐kDa) catalytic subunit (DNA‐PKcs), and two accessory proteins, Ku70 and Ku80, which form a heterodimer. DNA‐PKcs possess an intrinsic DNA end‐binding activity that is greatly stimulated by the Ku70/Ku80 heterodimer. Reduced activity of DNA‐PKcs in mice leads to a profound defect in the adaptive immune system, SCID. Therefore, one of the critical roles of DNA‐PKcs is to promote the rejoining of DSB produced in VDJ recombination, ¹⁴ , ¹⁵ , ¹⁶ for which process phosphorylation of Artemis has recently been shown to be essential. ² , ³ It has also been noted that DNA‐PKcs‐deficient mice demonstrate extreme hypersensitivity to ionizing radiation (IR), ⁽¹⁷⁾ for which the mechanism is not known, but could be attributed to a defect in the DSB repair system as represented by VDJ recombination.

ATM is another member of the PIKK family. ATM is the causative gene for Ataxia telangiectasia (A‐T), which is characterized by immunodeficiency, hypersensitivity to genotoxic agents, defects in cell cycle checkpoints, chromosome instability and cancer predisposition. ⁽¹⁸⁾ Cells from A‐T patients or ATM‐deficient mice are defective in activating IR‐induced G1/S, intra‐S and G2/M checkpoints, ⁽¹⁹⁾ and are hypersensitive to IR. ATM has been shown to phosphorylate a number of substrates involved in cell cycle checkpoints, DNA repair or apoptosis, such as p53, Nijmegen breakage syndrome 1 (Nbs1), Chk2, BRCA1, MDM2, SMC1, FANCD2, MDC1 and H2AX. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ Both ATM and DNA‐PKcs are SQ/TQ‐directed protein kinases with structural similarities, though these kinases are not completely redundant in their biological functions. ⁽²⁶⁾

Although the nucleolytic activity of Artemis in the presence of DNA‐PKcs is critical in the process of NHEJ in VDJ recombination, it is open to question how these activities are involved in the repair process for other DSB, such as those caused by IR. In the present study, we strove to identify the upstream signaling molecule(s) of Artemis in response to IR‐induced DNA damage. Our study demonstrates that Artemis plays an important role downstream of ATM in the response to DNA damage.

Materials and Methods

Cell lines and culture.  Epstein‐Barr virus‐transformed lymphoblastoid cell lines (EBV‐LCL) were grown in RPMI 1640 medium with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 U/mL streptomycin (Gibco BRL). LCL‐wt1, LCL‐wt2, and LCL‐wt3 were from three healthy individuals; LCL‐AT65, LCL‐AT52, and LCL‐AT15 were from A‐T patients; ⁽²⁷⁾ LCL‐NBS (GM07078A) was from a Nijmegen breakage syndrome (NBS) patient. Primary fibroblasts from an A‐T patient and a normal individual were cultured in DME with 20% FBS. A‐T fibroblasts GM05849C, stably transfected with human full‐length ATM cDNA (pEBS‐YZ5) or empty vector (pEBS7), were cultured in DME supplemented with 10% FBS and 200 µg/mL Hygromycin B (Wako, Japan). NBS fibroblasts, stably transfected with wild type human Nbs1 or empty vector, were cultured in DME with 10% FBS, 50 µg/mL Gentamycin (Sigma, USA) and 250 µg/mL Hygromycin B. 293T and HCT‐15 cells (Chk2 wild‐type and deficient cells, respectively), M059K and M059J cells (DNA‐PK‐proficient and ‐deficient cells) were grown in DME with 10% FBS. Cells were irradiated at room temperature using a SOFTEX irradiator (1.5 G/min [Softex, Japan]).

Reverse transcriptase‐polymerase chain reaction and plasmid construction.  Total ribonucleic acid (RNA) was isolated from normal fibroblasts using an SV Total RNA Isolation System kit (Promega, USA). cDNA was prepared from mRNA using random primer and AMV reverse transcriptase (Life Sciences, USA) according to the manufacturer's instructions. Polymerase chain reaction (PCR) products of the Artemis open reading frame were ligated to the BglII/XhoI sites of the plasmid pCMV‐TagI (Stratagene, La Jolla, CA, USA) with an N‐terminal Flag epitope tag, and further subcloned into BamHI/XhoI sites of the pcDNA3.1 vector (Invitrogen, USA) for optimal expression of the cDNA in mammalian cells. The cDNA inserts of the pcDNA3.1‐Artemis plasmid were verified by DNA sequencing. Details of the primers are available on request.

Site‐directed mutagenesis.  Mutants of Artemis (T91A, T251A, S516A, S534A, S538A, S553A, S562A, S645A) were generated from Flag‐tagged pcDNA3.1‐Artemis plasmid, using a QuickChange site‐directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. The identity of the mutated nucleotides was confirmed by DNA sequencing.

Western blotting and antibodies.  Ionizing‐treated or untreated cells were washed with ice‐cold phosphate‐buffered saline with 0.1 mM Na₃VO₄, pelleted, and lyzed in lysis buffer (1% Tween20, 0.3% NP‐40, 150 mM NaCl, 50 mM Tris‐HCl pH 7.5, 20 mM NaF, and 10 mM Na₃VO₄), with 1 mM PMSF, 5 mM ethylenediamine tetraacetic acid (EDTA), and 1/100X protease inhibitor cocktail (Sigma). After sonication and centrifugation, the supernatants were fractionated in 10% sodium dodecylsulfate‐polyacrylamide gel, transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA) and immunoblotted using the antibodies indicated. The polyclonal antibody against human Artemis was purchased from Orbigen. Anti‐ Nbs1 and antip53 antibodies were obtained from Oncogene Science‐Calbiochem. The hMre11 polyclonal antibody was from Novus Biologicals; the DNA‐PKcs monoclonal antibody was from Upstate Biotechnology; Flag/M2 and Flag/M5 monoclonal antibodies were from Sigma; polyclonal antiphospho‐SQ/TQ ATM/ATR substrate antibody and antiphospho‐S15p53 antibody were from Cell Signaling Technology.

Transfection and Immunoprecipitation.  293T cells were transiently transfected with the pcDNA3‐WT‐Artemis plasmid, pcDNA3‐Artemis mutant plasmid, or with empty vector using an Effectene transfection kit (Qiagen Inc., Valencia, CA, USA) according to the manufacturer's protocol. Thirty hours after transfection, the cells were either IR‐treated or mock‐treated. Following 2 h of recovery at 37°C, the cells were lyzed for 30 min in ice‐cold buffer containing 0.5% NP‐40, 150 mM NaCl, 50 mM Tris‐HCl (pH 7.5), 25 mM NaF and 1 mM Na₃VO₄, and supplemented with protease inhibitors: 1 mM PMSF, 5 mM EDTA, and 1/100X protease inhibitor cocktail (Sigma). The supernatants were precleared with Sepharose protein G beads at 4°C for 30 min, immunoprecipitated with 2 µg of indicated antibody or control sera at 4°C for 3 h, and then incubated with Sepharose protein G beads for an additional hour. The immunoprecipitates were resolved in SDS‐PAGE and analyzed by immunoblotting.

Phosphatase treatment of immune complexes.  Phosphatase treatment was performed as described. ⁽²⁸⁾ Briefly, Artemis immunoprecipitates were washed twice with basic phosphatase buffer (50 mM Tris‐HCl, pH 7.5, 5 mM DTT, 2 mM MnCl₂) and then incubated with 100 units of λ‐PPase (New England Biolabs, Beverly, MA, USA) for 30 min at 30°C. The reaction was stopped by addition of ice‐cold lysis buffer containing 10 mM Na₃VO₄ and 10 mM NaF. Phosphatase‐treated immune complexes were washed twice with lysis buffer and were subjected to immunoblotting.

Results

Artemis is phosphorylated in response to DNA damage.  In response to ionizing irradiation, several proteins are phosphorylated and show migration shifts in SDS‐PAGE. We evaluated the possibility that migration of Artemis was retarded by phosphorylation in response to IR. We examined the mobility shift of Artemis after treatment of cells with low (3 Gy) (Fig. 1a) and high doses (10 Gy) (Fig. 1b) of IR, which might activate different PIKK family members. ⁽²⁹⁾ After IR treatment, EBV‐LCL from a normal individual were harvested at the time points indicated and the cell lysates were immunoblotted with anti‐Artemis antibody. The antibody specifically recognized Artemis protein with a relative molecular mass of 97 kDa on SDS‐PAGE in all cell lines examined (Fig. 1f). Treatment of cells with 3 Gy irradiation resulted in a slower migration of Artemis protein, which appeared at 30 min, became most prominent at 2 h, and then gradually decreased at 24 h (Fig. 1a). Because of this, we chose the time point of 2 h post‐irradiation in the following experiments. In response to 10 Gy, the mobility shift of Artemis was more marked, particularly at 30 min, and involved most of the Artemis protein pool (Fig. 1b). This suggested that the alteration of mobility of Artemis depended on the extent of DNA damage. We then examined the mobility shift of Artemis at 2 h in response to doses of IR ranging from 0.5 Gy to 20 Gy. Decreased mobility of Artemis was observed even after 0.5 Gy of IR (Fig. 1c). We also examined the migration of Artemis after IR in several tumor‐derived cell lines (Fig. 1d and data not shown). In these cells, retardation of Artemis in the gel after IR was similar to that seen in EBV‐LCL, suggesting that this phenomenon is not restricted to specific cell types. To determine whether the mobility shift was due to phosphorylation of Artemis, immunoprecipitates from irradiated and unirradiated cells were mock‐treated or treated with phosphatase. Artemis migrated faster on SDS‐PAGE after treatment with phosphatase (Fig. 1e), demonstrating that the observed mobility shift in response to IR is due to phosphorylation. Notably, phosphatase treatment also increased the migration of Artemis immunoprecipitated from unirradiated cells, suggesting that this factor is constitutively phosphorylated, but hyperphosphorylated following IR.

Figure 1.

Phosphorylation of Artemis in response to ionizing radiation (IR). Epstein‐Barr virus transformed lymphoblastoid cell lines (EBV‐LCL) of a normal control (LCL‐wt1) were irradiated at (a) 3 Gy or (b) 10Gy, and harvested at the time points indicated. Whole cell lysates were fractionated on 10% sodium dodecysulfate–polyacrylamide gel electrophoresis (SDS‐PAGE) and analyzed by immunoblotting with anti‐Artemis antibody. The top arrow indicates the slower mobility band of Artemis. EBV‐LCL of (c) normal control (LCL‐wt1) and (d) 293T cell lines were irradiated at the indicated doses and incubated for 2 h, and whole cell lysates analyzed by Western blotting. (e) Normal lymphoblasts (LCL‐wt1) were treated with either 0 Gy (–) or 3 Gy (+) irradiation. Whole cells were lyzed at 2 h post‐ionizing radiation, and Artemis proteins were immunoprecipitated. The immunoprecipitates were either untreated (–) or treated (+) with lambda phosphatase (λ‐PPase) and then analyzed by Western blotting for Artemis. (f) The specificity of the anti‐Artemis antibody. The whole cell lysates of fibroblasts from three Artemis deficient patients (lane 1, 2 and 3), and a normal individual (lane 4) were immunoblotted with an anti‐Artemis antibody. Molecular weight is indicated by arrows.

DNA‐PKcs is not responsible for IR‐induced Artemis hyperphosphorylation.  The fact that Artemis is phosphorylated by DNA‐PKcs during VDJ recombination, prompted us to investigate whether DNA‐PKcs is also involved in the phosphorylation of Artemis in response to IR‐induced DSB. DNA‐PKcs‐proficient M059K cells and ‐deficient M059J cells were mock‐treated or treated with 3 Gy or 10 Gy of IR, incubated for 2 h, and the lysates were analyzed for Artemis. We observed that Artemis was phosphorylated in DNA‐PKcs‐deficient (M059J) and ‐proficient (M059K) cells to the same extent following low (3 Gy) or high dose (10 Gy) of IR. These data suggest that DNA‐PKcs is not required for IR‐induced hyperphosphorylation of Artemis as determined by the mobility shift assay (Fig. 2). However, because this assay is not the only way to identify phosphorylation, our findings do not completely rule out the possibility that Artemis is phosphorylated by DNA‐PKcs in response to DNA damage.

Figure 2.

DNA‐dependent protein kinase (DNA‐PKcs) is not involved in ionizing radiation (IR)‐induced Artemis phosphorylation as determined by mobility shift assay. M059K (DNA‐PK‐proficient) cells, M059J (DNA‐PK‐deficient) cells, and Ataxia‐telangiectasia fibroblasts were irradiated as indicated, harvested at 2 h post‐IR, and the cell lysates were assayed for phosphorylation of Artemis. Absence of DNA‐PKcs in M059J cells was confirmed by anti‐DNA‐PKcs immunoblotting (data not shown). Endogenous hMre11 is shown as a loading control.

Hyperphosphorylation of Artemis in response to IR‐induced DSB is ATM‐dependent.  To further investigate upstream signaling pathways for Artemis phosphorylation, we evaluated the mobility shifts of Artemis in WT ATM‐cell lines and ATM‐deficient cell lines. Three EBV‐LCL from normal individuals (LCL‐wt1, wt2, wt3) and three EBV‐LCL from patients with A‐T (LCL‐AT65, AT15, AT52), ⁽²⁷⁾ were mock irradiated or irradiated with 3 Gy or 10 Gy, incubated for 2 h, and cell lysates were analyzed by immunoblotting. Although Artemis in normal cells was efficiently hyperphosphorylated, its hyperphosphorylation in ATM‐deficient lymphoblasts was greatly attenuated, particularly at the lower IR dose (3 Gy) (Fig. 3a). We also examined the phosphorylation of Artemis in primary fibroblasts from normal individuals and A‐T patients (Fig. 3b), and obtained the same result as that was observed in the EBV‐LCL from the A‐T patients. These indicate that ATM is required for the hyperphosphorylation of Artemis in response to a low dose of IR. To further demonstrate the ATM‐dependent hyperphosphorylation of Artemis in the same cell type, we analyzed Artemis in the A‐T fibroblast strain GM05849C transfected with mock or wild‐type ATM cDNA. In this experiment, the low radiation dose was 4 Gy because this better demonstrated phosphorylation of Artemis in these cells. Introduction of ATM cDNA into GM05849C cells restored the mobility shift of Artemis after irradiation (Fig. 3c). These results provide strong evidence that the hyperphosphorylation of Artemis indeed depends on ATM in response to IR‐induced DSB. At 10 Gy, however, we detected hyperphosphorylation of Artemis in A‐T lymphoblasts and A‐T fibroblasts, which is consistent with previous studies that indicate the existence of ATM‐independent DNA damage response pathways at high IR doses. ⁽¹²⁾ We also tested whether genotoxic agents other than irradiation induce phosphorylation of Artemis. We treated a normal EBV‐LCL with ultraviolet (UV) or hydroxyurea, and detected similarly retarded migration of Artemis. Because this phosphorylation was also seen in UV‐ or hydroxyurea‐treated AT cells (Fig. 3d), it appears that Artemis hyperphosphorylation is ATM‐dependent in response to DNA lesions induced by IR, but ATM‐independent in response to UV‐ or hydroxyurea‐induced DNA lesions. These results further suggest that other upstream kinases can trigger hyperphosphorylation of Artemis following diverse forms of DNA damage.

Figure 3.

Phosphorylation of Artemis in response to ionizing radiation is Ataxia‐telangiectasia‐mutated (ATM)‐dependent. (a) Normal (LCL‐wt1, LCL‐wt2, LCL‐wt3) and Ataxia‐telangiectasia (A‐T) (LCL‐AT65, LCL‐AT15, LCL‐AT52) lymphoblasts were either mock‐irradiated or irradiated at 3 Gy or 10 Gy and harvested 2 h later. Whole cell lysates were analyzed for Artemis phosphorylation by immunoblotting with anti‐Artemis antibody. (b) Primary fibroblasts from a normal control and an A‐T patient were unirradiated or irradiated at 3 Gy or 10 Gy and harvested 2 h later, and whole cell lysates were assayed as above. (c) A‐T fibroblasts, GM05849C, were stably transfected with either empty vector (pEBS7) or full‐length ATM cDNA vector (pEBS‐YZ5). The cells were mock‐irradiated or irradiated at 4 Gy or 10 Gy, and analyzed for Artemis phosphorylation. Specificity of anti‐ATM antibody and absence of ATM in A‐T fibroblast was confirmed by anti‐ATM Western blot (data not shown). (d) Phosphorylation of Artemis upon ultraviolet (UV)‐induced DNA damage and replication block (hydroxyurea [HU] treatment) in normal (LCL‐wt1) and A‐T (LCL‐AT65) EBV‐LCL. LCL‐wt1 and LCL‐AT65 cells were treated with UV as indicated and harvested at 1 h point (upper panel), or the cells were treated with HU as indicated and harvested 24 h later (lower panel). Whole cell lysates were analyzed for Artemis phosphorylation.

Artemis is a downstream substrate of ATM in response to DNA damage.  Our finding of ATM‐dependent Artemis phosphorylation led us to investigate whether Artemis is a direct substrate of ATM in response to DNA damage. 293T cells were transiently transfected with Flag‐tagged WT Artemis cDNA or empty vector, and mock‐irradiated or irradiated at 3 Gy or 10 Gy. The Artemis protein was then immunoprecipitated with α‐Flag/M2 antibody and the phosphorylation of Artemis was detected by the polyclonal ATM/ATR substrate α‐phospho‐SQ/TQ antibody. ⁽³⁰⁾ (Fig. 4a). We observed SQ/TQ phosphorylated band of Artemis at 3 Gy and 10 Gy of IR, and the higher dose of IR increased the phosphorylation of SQ/TQ sites on both Artemis and the positive control protein p53. SQ/TQ phosphorylation of Artemis by ATM was also confirmed by in vitro SQ/TQ phospho‐specific immunoblot assay by using in vitro translated Artemis as a substrate (data not shown). These observations indicate that Artemis is phosphorylated on SQ/TQ motif(s) in response to IR and likely serves as the downstream substrate of ATM.

Figure 4.

Phosphorylation of Artemis at SQ/TQ sites. (a) 293T cells were transiently transfected with Flag‐tagged Artemis cDNA or empty vector. After 36 h, cells were treated with ionizing radiation as indicated, incubated for 2 h, then the Artemis protein was immunoprecipitated with α‐Flag/M2 antibody. The phosphorylation of Artemis was detected by α‐phospho‐SQ/TQ antibody. The amount of expressed Artemis in each lane was checked by anti‐Artemis blotting. Ser‐15 phosphorylation of endogenous p53 from the same cells was used as a positive control. (b) 293T cells were transiently transfected with Flag‐tagged Artemis‐WT cDNA or Artemis mutants of T91A, S538A, S553A, S562A and S645A. Thirty‐six hours post‐transfection, cells were mock irradiated or irradiated at 3 Gy or 10 Gy, incubated for 2 h, and cell lysates were analyzed by immunoblotting with α‐Flag/M5 antibody. Slower migration of Artemis was also detected in T251A, S516A, and S534A mutants (data not shown).

Ser645 is a SQ/TQ site responsible for hyperphosphorylation of Artemis upon IR‐induced DSB.  Artemis has one ATM/DNA‐PK SQ/TQ consensus site at high stringency (S538), four sites at medium stringency (T91, S553, S562, and S645), and several low stringency sites (S362, S516, and S534). To determine the phosphorylation site(s) of Artemis by ATM, we generated eight different Artemis mutants, T91A, T251A, S516A, S534A, S538A, S553A, S562A and S645A. Each mutant was expressed in 293T cells; the cells were mock irradiated or irradiated at 3 Gy or 10 Gy, and examined for the mobility shift of the mutant Artemis following IR. We observed that mutations at residues of T91, T251, S516, S534, S538, S553 or S562 showed the mobility shift of Artemis comparable to that of 293T cells with WT‐Artemis (Fig. 4b and data not shown). However, the Serine to Alanine mutation at amino acid 645 canceled the molecular weight change of Artemis following IR (Fig. 4b). This was not due to the aberrant localization of the S645A molecule, since majority of the wild‐type, as well as the mutant Artemis proteins, were homogeneously distributed in the nuclei at 36 h after transfection (data not shown). Our finding indicates that the phosphorylation of S645 contributes to the mobility shift of Artemis in response to DNA damage. It is plausible that other SQ/TQ site(s) is (are) also phosphorylated upon γ‐irradiation, although the modification did not lead to the slower migration in this assay.

Hyperphosphorylation of Artemis is impaired in Nbs1‐deficient but not in Chk2‐deficient cells.  Although our present data show a link between Artemis and ATM, it is not clear how Artemis relates to other identified ATM substrates. Given that Nbs1 functions in the ATM signaling pathway and that Nbs1 is required for the phosphorylation of other ATM substrates, ²⁴ , ²⁵ , ³¹ we examined the dependence of Artemis hyperphosphorylation on Nbs1. EBV‐LCL from an NBS patient (LCL‐NBS), an A‐T patient (LCL‐AT65) and a normal individual (LCL‐wt1), were irradiated with 3 Gy or 10 Gy, or left unirradiated. Two hours after treatment, whole cell lysates were assayed by Western blotting for the altered migration of Artemis. We observed that in contrast to wild‐type LCL, the Nbs1‐deficient LCL showed significantly decreased, but not completely abolished, hyperphosphorylation of Artemis (Fig. 5a). The difference was more evident at 3 Gy than 10 Gy. To further determine Nbs1 dependence of Artemis hyperphosphorylation in the same cell background, we examined Artemis in NBS fibroblasts transfected with mock or wild‐type Nbs1 cDNA. The expression of wild‐type Nbs1 in the NBS fibroblasts reinstated the hyperphosphorylation of Artemis to the same extent as seen in normal fibroblasts following irradiation. In contrast, the cells transfected with mock vector showed only marginal increase in the mobility shift of Artemis (Fig. 5b). The correlation between Nbs1 expression and effective hyperphosphorylation of Artemis indicates that Nbs1 protein is required for efficient phosphorylation of Artemis in response to IR.

Figure 5.

Artemis phosphorylation is dependent on Nijmegen breakage syndrome 1 (Nbs1), but not Chk2. (a) Artemis phosphorylation in wild‐type (LCL‐wt1), Nbs1‐deficient (LCL‐NBS), and Ataxia‐telangiectasia (A‐T) (LCL‐AT65) lymphoblasts. Cells were mock irradiated or irradiated at 3 Gy or 10 Gy as indicated, and harvested 2 h post ionizing radiation (IR). Cell lysates were analyzed for Artemis phosphorylation by Western blotting. (b) NBS fibroblasts GM24SV were stably transfected with either full‐length Nbs1 cDNA or empty vector. The cells were unirradiated or irradiated as indicated, and the whole cell lysates were analyzed for Artemis phosphorylation as in Figure 5a. Expression of Nbs1 in each sample is shown in the bottom column. (c) 293T (Chk2‐proficient) cells, HCT‐15 (Chk2‐deficient) cells, and A‐T fibroblasts were treated with IR as indicated, and the whole cell lysates were analyzed 2 h post IR by anti‐Artemis antibody.

To determine whether Chk2, a crucial link between ATM and p53, ³² , ³³ , ³⁴ is involved in the phosphorylation of Artemis, we examined both Chk2‐deficient HCT‐15 cells and 293T cells that were treated as indicated in Figure 5c. We observed no significant difference in migration of Artemis in Chk2‐deficient and ‐proficient cells following either the low dose or high dose of IR (Fig. 5c). Our observation suggests that Chk2 is not involved in IR‐induced hyperphosphorylation of Artemis as determined by mobility shift experiment.

Hyperphosphorylated Artemis is physically associated with Mre11/Rad50/Nbs1 complexes in an ATM‐dependent manner in response to IR‐induced DSB.  The finding that hyperphosphorylation of Artemis depends on ATM and Nbs1 in response to IR‐induced DSB, prompted us to investigate whether these proteins are associated with each other in the same complex in the absence or the presence of DNA damage. 293T cells were transiently transfected with Flag‐tagged Artemis cDNA, and mock‐irradiated or irradiated at 3 Gy or 10 Gy. The cells were harvested and the endogenous Nbs1 proteins were immunoprecipitated and probed with anti‐Flag, anti‐Nbs1, anti‐hMre11, or anti‐DNA‐PKcs antibody. Nbs1 was found associated with hMre11, but not with DNA‐PKcs. Notably, overexpression of Artemis by transfection of Flag‐tagged Artemis cDNA revealed Artemis was co‐immunoprecipitated with Nbs1, and increasing doses of IR increased the amount of Artemis in the Nbs1‐immunoprecipitates (Fig. 6a). The reciprocal experiment confirmed the presence of the Nbs1/Mre11 in anti‐Flag immunoprecipitates in Flag‐Artemis overexpressing 293T cells (data not shown). DNaseI treatment did not affect Artemis–Nbs1/Mre11 interaction, suggesting that the association is not mediated through DNA (data not shown). The interaction between endogenous Artemis and Nbs1 was examined to confirm this observation in a more physiological setting. 293T cells were mock irradiated or irradiated, and endogenous Nbs1 protein was immunoprecipitated. We observed that endogenous Nbs1 was predominantly associated with hyperphosphorylated form of endogenous Artemis, and the association was more prominently IR‐dependent, however, DNA‐PKcs was not included in this complex (Fig. 6b). To further examine the effect of ATM on the interaction between Artemis and the Mre11/Rad50/Nbs1 (MRN) complex, GM05849C (ATM‐deficient) cells were mock irradiated or irradiated at 3 Gy or 10Gy, then endogenous Nbs1 protein was immunoprecipitated. Endogenous Artemis was not detected in the MRN complex in ATM‐deficient cells following IR treatment, which was in contrast to the result in ATM‐proficient 293T cells (Fig. 6b). This indicates that the interaction of Artemis with the MRN complex is ATM dependent.

Figure 6.

Hyperphosphorylated Artemis physically associates with Mre11/Rad50/Nbs1 (MRN) complexes in an Ataxia‐telangiectasia‐mutated (ATM)‐dependent manner following ionizing radiation. (a) 293T cells were transiently transfected with Flag‐tagged Artemis. After 30 h, cells were mock irradiated or irradiated as indicated. The cells were harvested 2 h later and the Nijmegen breakage syndrome 1 (Nbs1) immunoprecipitates were subjected to immunoblotting with anti‐Artemis, anti‐Nbs1, anti‐Mre11, and anti‐DNA‐dependent protein kinase (DNA‐PKcs) antibodies. (b) 293T cells were unirradiated or irradiated as indicated, incubated for 2 h, and the endogenous Nbs1 proteins were immunoprecipitated. The endogenous Artemis, Nbs1, Mre11 and DNA‐PKcs proteins in the immunoprecipitates were detected by immunoblotting. Control serum (immunoglobulin G) was used as negative control. In the right column, GM05849C cells (ATM‐deficient SV40‐transformed fibroblasts) were mock irradiated or irradiated at 3 Gy or 10 Gy, the endogenous Nbs1 protein was immunoprecipitated, and co‐immunoprecipitated endogenous Artemis, Nbs1, Mre11 proteins were detected by immunoblotting.

Discussion

In the present study, we demonstrated that Artemis is hyperphosphorylated in response to low dose IR‐induced DSB in an ATM‐dependent manner. Our study has also revealed for the first time that the hyperphosphorylated Artemis physically associate with the MRN complex in an ATM‐dependent fashion. ATM is the central regulator in responses to IR, and phosphorylates a number of target proteins involved in various cellular responses, such as p53, Chk2, Mdm2, Nbs1, SMC1, FANCD2, BLM, MDC1 and BRCA1. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ Our present study adds Artemis, the essential factor for hairpin opening in VDJ recombination, to the increasing list of DSB‐dependent signaling molecules downstream of ATM.

Artemis has several potential ATM/DNA‐PK consensus phosphorylation sites. Through our study, S645 was mapped as a phosphorylation site that contributes to retarded mobility of Artemis upon DNA damage stimuli. Functional relevance of S645 phosphorylation is not yet clear, since mutation at this SQ site did not significantly affect cellular survival to γ‐irradiation (data not shown). Although not contributing to the slower migration, involvement of other potential SQ/TQ site(s) upon DNA damage stimuli is likely, as antiphospho‐SQ/TQ ATM/ATR substrate immunoblot has demonstrated that S645A mutant can still be SQ/TQ‐phosphorylated (data not shown). Determination of phosphorylation sites by ATM, DNA‐PK, and VDJ recombination signal, and investigation of potential difference in the roles of each phosphoprotein will give us information on how Artemis functions are regulated.

While preparing our article, two papers on Artemis phosphorylation were released: Zhang et al. reported that Artemis was phosphorylated by ATM and ATR in response to IR and UV‐induced DNA damage, and that Artemis was a checkpoint protein involved in DNA damage‐induced G2/M cell cycle arrest. ⁽³⁵⁾ Their data also suggested phosphorylation of Artemis by DNA‐PKcs may contribute to IR‐induced mobility shift of Artemis, while our data did not support the contribution of DNA‐PKcs in the mobility shift. It awaits further study to pinpoint the site(s) of phosphorylation by DNA‐PKcs, ATM and ATR for providing answer to the question. Poinsignon et al. reported that C‐terminal SQ sites were involved in IR‐induced Artemis phosphorylation and that in vitro‐generated Artemis mutants, which show impaired IR‐induced phosphorylation, still display an activity to complement the V(D)J recombination defect and the increased radiosensitivity of Artemis‐deficient cells. ⁽³⁶⁾ These accumulating data still do not clarify physiological relevance of phosphorylated form of Artemis. Detailed study on stable cell lines expressing different Artemis mutant would be essential to delve into physiological roles of these sites.

The hyperphosphorylation of Artemis in response to IR was greatly decreased in Nbs1‐deficient cells. Reconstitution with Nbs1 cDNA restored efficient phosphorylation of Artemis in Nbs1 fibroblasts, which demonstrates that Nbs1 is critical for efficient phosphorylation of Artemis in response to IR. Nbs1 forms a complex with hMre11 and Rad50 (MRN complex), which is recruited to the site of DNA damage to form radiation‐induced nuclear foci. Nbs1 is phosphorylated by ATM in response to IR. However, it is required for the phosphorylation of other ATM substrates, such as Chk2 and SMC1, serving as a signal amplifier or a platform for in the phosphorylation of other ATM substrates. ²⁴ , ²⁵ , ³¹ Therefore, it is hypothesized that Nbs1 might function as one of the essential molecules for the hyperphosphorylation and activation of Artemis.

Hyperphosphorylated Artemis formed complexes with MRN after DNA damage. This finding suggests that Artemis‐bound and ‐unbound MRN complexes exert distinct functions, as has been proposed for MRN complex that is either downstream or upstream of ATM. ³⁷ , ³⁸ Considering that Artemis is recognized for its nucleolytic activity, ⁽²⁾ and only the hyperphosphorylated form binds to MRN, we favor the hypothesis that the Artemis/MRN complex functions downstream of ATM at the site of DNA damage to form damage‐induced nuclear foci. Our study places Artemis at the signaling crossroads downstream of DNA‐PKcs in VDJ gene recombination and ATM in IR‐induced DSB repair. Deficiency of either DNA‐Pkcs or ATM leads to a defect in terms of repair function of IR‐induced DSB, ⁽¹⁰⁾ and neither of these can substitute for the function of the other in this process. Although ATM and DNA‐PKcs similarly phosphorylate SQ/TQ motifs, they are not redundant in view of the hyperphosphorylation and activation of Artemis in response to IR‐induced DSB. Further study is needed to elucidate how these two pathways are interconnected with Artemis in nucleolytic processing of damaged DNA ends. In spite of these questions that still need to be addressed, our study places Artemis at the signaling crossroads downstream of DNA‐PKcs and ATM in IR‐induced DSB repair and VDJ gene recombination.