Artemis deficiency confers a DNA double-strand break repair defect and Artemis phosphorylation status is altered by DNA damage and cell cycle progression

DNA Repair 4 (2005) 556–570

Artemis deficiency confers a DNA double-strand break repairdefect and Artemis phosphorylation status is altered by

DNA damage and cell cycle progression

Junhua Wanga, 1, Janice M. Plutha, Priscilla K. Coopera, Morton J. Cowanb,David J. Chena, 1, Steven M. Yannonea, ∗

a Life Sciences Division, Department of Molecular Biology, Lawrence Berkeley National Laboratory,Mail Stop 74-157, 1 Cyclotron Road, Berkeley, CA 94720, USA

b BMT Division, UCSF Children’s Hospital, San Francisco, CA 94143-1278, USA

Received 26 October 2004; accepted 4 January 2005

Abstract

is-deficientc (NHEJ).H elevateds ere DSBr and G2/Mb like kinase-d ersensitivet d suggestm e that ofh with cellc all subseto served inA©

K

1

ci

m7

andaysl offoruseouss ofer-l theV(D)J

the

1d

Mutations in the Artemis gene are causative in a subset of human severe combined immunodeficiencies (SCIDs) and Artemells exhibit radiation sensitivity and defective V(D)J recombination, implicating Artemis function in non-homologous end joiningere we show that Artemis-deficient cells from Athabascan-speaking Native American SCID patients (SCIDA) display significantlyensitivity to ionizing radiation (IR) but only a very subtle defect in DNA double-strand (DSB) break repair in contrast to the sevepair defect of NHEJ-deficient cells. Primary human SCIDA fibroblasts accumulate and exhibit persistent arrest at both the G1/Soundaries in response to IR, consistent with the presence of persistent DNA damage. Artemis protein is phosphorylated in a PI3-ependent manner after either IR or a number of other DNA damaging treatments including etoposide, but SCIDA cells are not hyp

o treatment with etoposide. Inhibitor studies with various DNA damaging agents establish multiple phosphorylation states anultiple kinases function in Artemis phosphorylation. We observe that Artemis phosphorylation occurs rapidly after irradiation likistone H2AX. However, unlike H2AX, Artemis de-phosphorylation is uncoupled from overall DNA repair and correlates insteadycle progression to or through mitosis. Our results implicate a direct and non-redundant function of Artemis in the repair of a smf DNA double-strand breaks, possibly those with hairpin termini, which may account for the pronounced radiation sensitivity obrtemis-deficient cells.2005 Elsevier B.V. All rights reserved.

eywords:Artemis; DNA repair; NHEJ; V(D)J recombination; SCID; DNA double-strand breaks

. Introduction

DNA double-strand breaks (DSBs) are a potentially toxicellular event, highly mutagenic when misrepaired and lethalf unrepaired. Eukaryotic cells repair this type of damage by

∗ Corresponding author. Tel.: +1 510 495 2867; fax: +1 510 486 6816.E-mail address:[email protected] (S.M. Yannone).

1 Current address: Division of Molecular Radiation Biology, Depart-ent of Radiation Oncology, UT Southwestern Medical Center, Dallas, TX5390-9187, USA.

two primary pathways, homologous recombination (HR)non-homologous end joining (NHEJ). While both pathwcontribute significantly to the repair of DSBs and survivacells, NHEJ is thought to be the predominant pathwayDSB repair in human somatic cells. NHEJ and HR eacha discrete set of proteins, with HR requiring a homologDNA template to facilitate repair. The fundamental stepNHEJ involve recognition of a DSB, processing of DNA tmini into ligatable ends, and a final ligation step to reseabreak. Some of these basic events are also essential forrecombination, a related but distinct process specific to

568-7864/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.dnarep.2005.02.001

J. Wang et al. / DNA Repair 4 (2005) 556–570 557

lymphoid cell lineages. V(D)J recombination is initiated bythe recombinase activating genes (RAG1 and RAG2) whichcoordinately recognize and cleave at signal sequences withinthe immunoglobulin (IgG) and T cell receptor (TCR) genesthereby excising the intervening region of DNA. This liber-ated signal sequence and coding sequence ends are each thenrejoined by a process involving the NHEJ machinery to rendera signal joint (SJ) and a coding joint (CJ) respectively. Defi-ciency in this process results in failure of B and T lymphocytematuration but not of natural killer (NK) cells, a conditiontermed T-B-NK+ severe combined immunodeficiency (T-B-NK+ SCID). The majority of human T-B-NK+ SCID patientshave been ascribed to mutations in the RAG1 or RAG2 genes[1–3]. However, some T-B-NK+ SCID patients have normalRAG and NHEJ genes, and the underlying genetic flaw fortheir disorders had until recently remained undefined. Onesuch group of patients, termed SCID-Athabascan (SCIDA),are Athabascan-speaking Navajo and Apache Native Amer-icans which have a very high incidence of T-B-NK+ SCIDdue to a founder mutation in these isolated populations[4–6].Other similar groups of T-B-NK+ SCID patients of variousethnic origins have been identified in France. These Europeanpatients exhibit increased radiosensitivity and have been clas-sified as radiation-sensitive SCID (RS-SCID)[7,8]. Recently,mutations in the gene encoding Artemis were identified ascausative in both RS-SCID and SCIDA[9,10]. Among theR aveb on 8iT ffec-t ypec rm-i gies[

herc K),a vea D)Jr festsa ncy( se( inep nit,D o-n thep ofp lingp fD iesoD toh hee ionr arlyd esses[

The other essential NHEJ proteins are ligase IV andXRCC4, which exist as a hetero-complex[19–22]. Cells de-ficient in XRCC4 exhibit severe IR sensitivity and have sig-nificantly reduced levels of ligase IV, suggesting that ligaseIV may require XRCC4 for its stability[23,24]. Despite be-ing one of four identified mammalian DNA ligases, the ligaseIV/XRCC4 (L4X) enzyme is specifically required for NHEJand V(D)J recombination, and other mammalian ligases can-not functionally substitute in these processes (reviewed in[25]). This specificity is likely mediated by particular affini-ties between L4X and the components of DNA-PK[26,27].Therefore, L4X is considered to be indispensable for the finalDNA resealing steps in both NHEJ and V(D)J recombina-tion.

Recently, GST-fused Artemis protein was shown to inter-act with DNA-PK and to be phosphorylated by DNA-PK invitro. Artemis also was shown to exhibit inherent exonucle-ase activity and endonuclease activity when in the presenceof DNA-PKcs and ATP[28]. Based on its interactions withDNA-PK and its role in V(D)J recombination, it was con-cluded that Artemis is essential for NHEJ[28]. However,aside from inferences from the radiation sensitivity of RS-SCID cells and the reports of DNA-PK/Artemis interactions,the function(s) of Artemis in response to DNA damage re-main largely undefined.

To better understand the role that Artemis plays in the cel-l emisa ellsl ion,at portt fol-l cler sionw airedD e-l airm EJ-d e de-f toneH beb ed inr neticso rallD hesefi re-q cedD

2

2

s nd

S-SCID patients, eight different Artemis mutations heen identified, while a single nonsense mutation in ex

s responsible for the defect in all SCIDA individuals[9,10].he mutations causing both RS-SCID and SCIDA are e

ively null, and the defective V(D)J recombination phenotan be complemented by wild-type Artemis cDNA, confing that an Artemis deficiency underlies these patholo9–11].

To date Ku70, Ku80, and DNA-PKcs, which togetonstitute the DNA-dependent protein kinase (DNA-Pnd the DNA ligase IV/XRCC4 complex of proteins hall been identified as essential for both NHEJ and V(ecombination. Deficiency in any of these proteins manis profoundly defective DSB repair and immunodeficiereviewed in [12]). The DNA-dependent protein kinaDNA-PK) is a highly abundant nuclear serine/threonrotein kinase consisting of a 460-kDa catalytic subuNA-PKcs, and a heterodimeric DNA binding compent, Ku70/80. The catalytic subunit is a member ofhosphatidylinositol 3-kinase-like kinase (PIKK) familyroteins, which also includes the DNA damage signaroteins ATM and ATR[13,14]. The Ku component oNA-PK recognizes and tightly binds nearly all varietf DNA termini regardless of sequence context[15].NA-PKcs assembles with the Ku/DNA complex inolo-DNA-PK, resulting in kinase activation. While txact role(s) of DNA-PK in NHEJ and V(D)J recombinatemains unclear, the effects of DNA-PK deficiencies cleemonstrate essential roles in these two related proc

16–18].

ular response to DNA damage, we have examined Artnd its modification in vivo as well as the behavior of c

acking Artemis. While this manuscript was in preparatnother study of Artemis function appeared[29]. In con-

rast to the conclusions of that concurrent study, we rehat primary SCIDA cells accumulate in G1/S and G2/Mowing irradiation and are profoundly delayed in cell cyesumption. This extended block to cell cycle progresould be consistent with a persistent presence of unrepNA damage. Surprisingly in view of this cell cycle d

ay and their significant radiosensitivity, SCIDA cells repost IR-induced DSBs with normal kinetics, unlike NHefective cells. Here we observe a small but detectabl

ect in overall DSB repair by the sensitive measure of his2AX phosphorylation. Furthermore, we show Artemis toasally phosphorylated and rapidly hyper-phosphorylatesponse to agents that cause DNA damage and the kif Artemis de-phosphorylation do not correlate with oveSB repair but instead coincide with mitosis. Based on tndings, we propose that Artemis function is uniquelyuired for repair of a small but critical subset of IR-induSBs.

. Materials and methods

.1. Cell lines

Fibroblast cell lines were cultured in�-MEM mediumupplemented with 10% FCS, penicillin (100 U/ml), a

558 J. Wang et al. / DNA Repair 4 (2005) 556–570

streptomycin (100�g/ml) at 37◦C in a humidified in-cubator (5% CO2) unless otherwise noted. Early pas-sage fibroblast lines established from skin biopsies fromSCIDA patients 03-A2 (03) and 07-A1 (07) and a wild-type unrelated Navajo adult (AK and AU) were immor-talized by introduction of SV40 large T antigen as de-scribed previously[10]. Early passage wild-type (AK)and primary fibroblast lines (04) and (05) from two dif-ferent SCIDA patients were immortalized by transfec-tion of hTERT cDNA using methods previously described[30]. Lymphoblast cell lines JHP (wild-type) and KT(ATM−/−) were a kind gift from Lavin and cowork-ers [31] and propagated in RPMI 1640 medium (Invitro-gen) supplemented with 10% fetal bovine serum and an-tibiotics. T-Rex-293 (Invitrogen) human kidney cells ex-pressing the tetracycline repressor were cultured in�-MEM supplemented with 10% tetracycline-free FBS and5 mg/ml of blasticidin. T-Rex-293 cells were transfectedwith pcDNA4/TO/Myc-HisA (Invitrogen) carrying wild-type Artemis cDNA, incubated for 48 h and then placedunder zeocin selection (250�g/ml). Clones were isolated,induced for Artemis expression with 1�g/ml doxycyclinefor 24 h and Artemis expression evaluated by immuno-blotting.

2.2. Fraction of activity released (FAR) — double strandb

ap s thef at-i m-b tiallya dis with4 singa ir-r es at3t on of∼ m-p witht .T1f ft ltra-P cew ara-ts aneo frac-t andt tilla-t

2.3. Immunofluorescence

Cells were seeded at∼2× 104 cells/well in each well of afour-well chamber slide. Cells were cultured for 24 h prior toirradiation then incubated for various times for repair. At thenoted times after damage, cells were fixed for 20 min at 20◦Cwith ice-cold methanol, permeabilized for 5 min with 0.5%Triton X-100, and blocked with 10% serum specific to thesecondary antibody species used. Fixed cells were incubatedat room temperature with primary antibodies for 1 h, thenwith fluorochrome-conjugated secondary antibodies for 1 hand then washed 5 times in PBS/1% BSA. The last wash con-tained the DNA counterstain 4′,6-diamidino-2-phenylindole(DAPI) at 0.1�g/ml. Slides were mounted in Vectashield andviewed using epifluorescence. Images were captured with aCCD camera. Paired and two-sample unequal variance, two-tailed t-test analysis was carried out on 100 data points/cellline.

2.4. Flow cytometry and survival assays

For flow cytometry, cells were pulse labeled for 30 minwith 10�g/ml bromodeoxyuridine (BrdU) (Sigma) unlessotherwise noted. Cell cycle distribution and BrdU incorpo-ration were analyzed with a Beckman-Coulter EPICS XL-MCL flow cytometer using XL Data Acquisition softwarea ckp . Af-t t then cells( edfg perfi 1 h,w batef ystalv

2

fulllt ntopp them en).P ethylL en-W st-e ita-t 3Tc in-s atedb o-B ase ith

reak repair assay

Overall capacity for DSB repair was quantified byulsed field gel electrophoresis assay that determine

raction of DNA released from the well (FAR) and migrng into the gel, which is directly proportional to the nuer of DSBs. The FAR assay was carried out essens described previously[32]. Briefly, cells were culture

n 25 cm2 flasks and labeled with [14C]thymidine (Amer-ham) prior to irradiation. Flasks were irradiated on ice0 Gy X-rays filtered through a 0.5 mm copper plate uPantak® generator operating at 320 kV/10 mA. After

adiation, repair samples were incubated for noted tim7◦C in a humidified atmosphere with 5% CO2. Cells were

rypsinized and resuspended in PBS at a concentrati107 cells/ml. An equal volume of 1.5% low melting teerature agarose (Type VII, Sigma) in PBS was mixed

he cell suspensions and formed into 200�l agarose plugshe plugs were immersed in lysis solution (0.4 M Na2EDTA,% sarkosyl, 1 mg/ml Proteinase K, pH 8.0) for 1 h at 0◦C

ollowed by at least 24 h at 50◦C. Approximately 30% ohe sample plug was loaded into wells of a 0.8% Uure agarose (Bio-Rad), 0.5× TBE gel and sealed in plaith agarose. The gels were run on a CHEF-DR II app

us (Bio-Rad) at 14◦C in 0.5× TBE for 68 h at 50 V with awitch time changing linearly from 60 to 3600 s. Each lf the gel was then cut into 15 equal slices and the

ion of the radioactivity released (FAR) from the plugshat retained in the plugs was measured by liquid scinion.

nd WinMDI© 2.8 software. A standard thymidine blorocedure was used to block cells at the G1/S transition

er release from the block, samples were harvested aoted times. For survival assays, varying number of5× 102 to 2× 103) were platedin triplicate, and incubator 12–15 h. Dishes were irradiated using a Pantak® X-rayenerator operating at 320 kV/10 mA with 0.5 mm copltration, or treated with 30 (g/ml etoposide (Sigma) forashed, overlaid with fresh medium, and allowed to incu

or 10–15 days. Resulting colonies were stained with criolet, scored, and surviving fraction calculated.

.5. Protein expression, purification and antibodies

Artemis expression constructs were derived fromength Artemis cDNA as described previously[10]. A C-erminal∼600 bp fragment of Artemis was subcloned iET-15b (Novagen) vector and expressed inE. coli. Artemisrotein from these cells was then purified according toanufacturers directions (pET System Manual, Novagurified protein was used to generate rabbit antiserum (Baboratories) and anti-Artemis IgY chicken antibodies (Gay Biotech). Antibody specificity was verified by We

rn blots of wild-type and SCIDA cells, immunoprecipion and MS/MS, and by use of Artemis inducible 29ells. Full length Artemis protein was purified from SF9ect cells infected with recombinant baculovirus genery sub-cloning Artemis cDNA into pFASTBAC-HT (GibcRL). Briefly, the amino-poly-histidine tagged Artemis wxtracted and purified using IMAC chromatography w


manufacturers protocols (Ni-NTA, Pharmacia). Artemis frac-tions were then dialyzed into 50 mM HEPES, pH 7.5, 10%glycerol, 2 mM EDTA, 1 mM dithiothreitol, 0.01% NonidetP-40, 20�g/ml phenylmethylsulfonyl fluoride, 1 ug/ml apro-tinin, pepstatin A, and leupeptin (HCB) containing 0.1 MNaCl, loaded onto mono-Q column (Pharmacia) and elutedwith a 100–500 mM NaCl linear gradient of HCB. Aliquotsof Artemis containing fractions were snap frozen and storedat−70◦C.

2.6. Kinase and phosphatase reactions

In vitro kinase reactions were carried out as describedpreviously[30]. Briefly, ∼1�g of DNA-PK, 3�g Artemis,5�g/ml of sheared salmon sperm DNA and 100�M ATPwere incubated for 30 min at 25◦C. Kinase reactions andpurified proteins were separated on a 4–12% gradient SDS-PAGE (Novex), and visualized by coomassie staining. Celllysates or immuno-precipitates were treated with 1�l (400 U)of lambda protein phosphatase (�-Ppase, NEB) incubated in1× reaction buffer, and 2 mM MnCl2 for 1 h at 30◦C.

2.7. Immunoprecipitation and Western analyses

Endogenous Artemis was immunoprecipitated from ex-tracts containing approximately 1× 108 lymphoblastoid cells( PBSa1 ica-tc -b dsw1 GEab u-p y-sa e re-s dingo h ei-to y(

3

3d

efi-c theA adi-a ofS ith

�-radiation exposed SV40-transformed fibroblasts from twodifferent SCIDA patients and an unrelated normal individual.As expected, we find SCIDA cells to be significantly moresensitive to IR than wild-type controls, exhibiting a radia-tion sensitivity similar to that reported for Artemis-deficientRS-SCID patient cells and that reported for the DNA-PKcs-deficient cells M059J (Fig. 1(A)) [7,33].

To investigate the DSB repair capacity of SCIDA cellswe measured cellular DSB repair using an assay that quanti-fies genomic fragmentation. The fraction of activity released(FAR) assay measures the proportion of genomic DNA mi-grating into pulsed field electrophoresis gels at various timesafter severe IR damage, with the FAR decreasing proportion-ally as DNA repair proceeds. We initially performed FAR as-says on SV40-transformed wild-type and SCIDA fibroblastsafter irradiation with 40 Gy (Fig. 1(B)). These data reveal theDSB repair capacity of SCIDA cells to be essentially indis-tinguishable from wild-type cells despite their pronouncedradiosensitivity. To confirm these results and to ensure thatthe altered p53 status in SV40-transformed cells did not influ-ence our results, we generated hTERT-immortalized SCIDAcell lines originating from two different SCIDA patients. Wethen compared the DSB repair capacity of these cells to anhTERT-immortalized wild-type cell line and to the DNA-PKcs-deficient human cell line, M059J (Fig. 1(C)) [34].These data also show that the SCIDA mutation in Artemisd mentw ells[ id-n pro-f lls(

3S

DSBr oseso ir ca-p werI tonev ofD atedH tlyp tionc dS ia-t dardi m ir-r andp d for1am ia-tr but

JHP and KT). Cells were harvested, washed withnd extracted by vortexing in 4 ml RIPA buffer (1× PBS,% NP-40, 0.5% deoxyholate 0.1% SDS) and brief son

ion. Cleared extracts were incubated with 20�l �-Artemishicken IgY for 2 h at 4◦C, followed by an overnight incuation with 20�l Ultralink Protein A/G beads. The beaere washed 3 times with 1 ml of RIPA, boiled in 40�l× Laemmli buffer and proteins resolved by 6% SDS-PAnd immunoblotted with a 1:1000 dilution of�-Artemis rab-it polyclonal antibody. All buffers contained aprotinin, leeptin, and pepstatin A at 1�g/ml. For direct Western analis, cells were lysed with 300–500�l of 1× Laemmli buffernd briefly sonicated. Denatured protein lysates werolved by appropriate SDS-PAGE gels of 6–15% depenn the mass of the target protein, and immunoblotted wit

her�-Artemis rabbit polyclonal antibody,�-ATM-2C1 mon-clonal antibody (GeneTex),�-Ku70 monoclonal antibodNeomarkers) or�-phospho-H2AX-Ser139 (Upstate).

. Results

.1. SCIDA cells are radiosensitive but not grosslyefective in DSB repair

Both RS-SCID and SCIDA share common immunodiency phenotypes but arise from different mutations inrtemis gene and therefore may differ with respect to rtion sensitivity[8–10]. To determine the radiosensitivityCIDA cells, we carried out colony formation assays w

oes not cause a severe defect in DSB repair, in agreeith previous reports on Artemis-deficient RS-SCID c

35] and with a recent report on RNAi Artemis-depleted key cells[29]. This is contrasted in these assays by the

oundly repair defective DNA-PKcs-deficient M059J ceFig. 1(C)).

.2. �H2AX foci studies reveal a slight repair defect inCIDA cells

While the FAR assay has been reliably used to detectepair defects in many types of cells, it requires high df radiation, 40 Gy in this case. To characterize the repaacity of SCIDA cells using a more sensitive assay and lo

R doses, we scored the in situ modification of the hisariant H2AX, which is phosphorylated proximal to sitesNA double-strand breaks. The number of phosphoryl2AX (�H2AX) foci in a nucleus is reported to be direcroportional to the number of DSBs, and de-phosphorylaoincides with DSB repair[36]. Asynchronous wild-type anCIDA primary fibroblasts were treated with 2 Gy of rad

ion 24 h after seeding, then fixed and stained using stanmmunofluorescence techniques 6 or 24 h post-IR or shaadiation. The number of foci/nucleus was scored blindlylotted and the average number of foci/nucleus calculate00 nuclei per cell line at each time point (Fig. 1(D)). Thesenalyses reveal that SCIDA cells have an average of∼2-foldore�H2AX foci/nucleus than controls at 6 h after irrad

ion. By 24 h after irradiation the�H2AX foci/nucleus wereeduced in all cell lines, reflecting continued DNA repair,


Fig. 1. SCIDA fibroblasts are sensitive to ionizing radiation and have a slight DSB repair defect. (A) Clonogenic cell survival of SV40 immortalized fibroblaststreated with increasing doses of�-irradiation. Each point represents the mean of three plates (± S.D.). (B) SV40 immortalized SCIDA cells have a nearlynormal DSB repair defect as assessed by the fraction of activity released (FAR) assay. Asynchronous SV40 immortalized wild-type (AK) and SCIDA (03,07) fibroblasts were cultured with14C-thymidine for 24 h, irradiated (40 Gy), harvested at noted times post-IR, embedded in agarose, lysed and run on PFGE.Lanes were excised and DNA remaining in the wells (repaired or unbroken) and that eluted out (broken) quantified by scintillation counting, and unirradiatedcontrol FAR values from the same cell line were subtracted and the resulting values plotted. (C) hTERT-immortalized SCIDA cells also show nearly normalDSB repair compared to DNA-PK deficient M059J cells. Asynchronous hTERT-immortalized wild-type (AK) and SCIDA (04, 05) fibroblasts and the humanDNA-PKcs-deficient glioma cell line, M059J, were assayed as described above. (D) SCIDA cells have a small defect in DSB repair revealed by�H2AX fociscoring. SCIDA and wild-type hTERT-immortalized fibroblasts were seeded onto chamber slides, irradiated with 2 Gy of X-rays or mock irradiated, culturedfor 6 or 24 h, fixed and stained for�H2AX, scored blindly, and the number of foci/nucleus plotted for SCIDA cell lines (04 and 05) and a wild-type AK cellline at both 6 and 24 h after irradiation. The 100 nuclei are plotted for each cell line at each time point and the mean foci/nucleus values are annotated abovethe respective data.


the average number of foci/nucleus remained∼2-fold higherin SCIDA cells (Fig. 1(D)). The number of foci/nucleus inSCIDA cells is not significantly different from wild-type inunirradiated controls (p> 0.001), but all lines show elevatedaverages relative to the earlier time point (6 h), likely due toincreased numbers of cells in S-phase at this later time point.Notably, the differences observed between SCIDA and wild-type cells at both 6 and 24 h after irradiation were highly sta-tistically significant (p< 0.0001). These results establish thatArtemis deficiency confers a relatively small but detectabledefect in DSB repair capacity.

3.3. Artemis-deficient SCIDA cells accumulate at G1/Sand G2/M boundaries

Normal cells halt cell cycle progression at checkpointspreceding both the S and G2 phases of the cell cycle in re-sponse to DNA damage. Defects in cell cycle checkpointscan contribute to radiosensitivity independent of defects inDNA repair. Because SCIDA cells exhibit a small repair de-fect relative to the severity of their IR sensitivity, we inves-tigated whether SCIDA cells can arrest cell cycle progres-sion in response to DNA damage. Primary fibroblast celllines were synchronized into G0/G1 by growth in mediumcontaining 0.2% serum for 72 h prior to irradiation (greaterthan 95% for all lines tested, data not shown). After IR, cellsw and1 e-

cause BrdU is added to G0/G1 cells after irradiation, onlycells passing the G1/S checkpoint and entering S-phase willincorporate BrdU. We observe roughly fifteen percent of thewild-type cells passed the G1/S checkpoint and synthesizedDNA within 48 h after a 6 Gy dose. In contrast, less than5% of the SCIDA cells progressed into S-phase after dam-age (Fig. 2(A)). It is interesting to note that in the SCIDAbut not wild-type cells, the small percentage of cells notsynchronized into G0/G1 phase prior to irradiation (<5%,in all lines) could fully account for all the BrdU positiveSCIDA cells. If true, this would suggest that SCIDA cellsirradiated in G0/G1 are completely incapable of passing theG1/S boundary and entering S-phase after this dose. In anycase, these data indicate that SCIDA cells are significantlyimpaired and/or delayed in cell cycle resumption after DNAdamage inflicted in G0/G1, consistent with the observed DSBrepair defect.

To test whether SCIDA cells can arrest cell cycle afterdamage in S-phase, we irradiated asynchronously growingBrdU pulse labeled primary fibroblast cells and analyzed cellcycle progression by flow cytometry at various times. Thisassay can therefore be used to follow the cell cycle progres-sion of cells that were in S-phase at the time of irradiation.Untreated control and SCIDA cell lines show nearly indistin-guishable cell cycle kinetics (Fig. 2(B)). IR treatment signif-icantly delays both mutant and wild-type cell cycle progres-sa n the

F and a ells able toi roblas teda en har ld-typec nchron ) primarfiac

ere overlaid with fresh medium containing 15% serum0�g/ml BrdU and incubated for up to additional 96 h. B

ig. 2. SCIDA cells accumulate in G1/S and G2/M in response to IRncorporate BrdU at various time points after IR treatment. Primary fibnd overlaid with fresh medium containing 15% serum and 30�M BrdU, thells, cells progress through the cell cycle at comparable rates. Asy
broblast cells were BrdU pulse-labeled for 30 min prior to collection at the innd analyzed by flow cytometry, the BrdU-labeled cells are plotted based onycle arrest after IR treatment. Cells were BrdU-pulse labeled, irradiated (6 G
ion relative to unirradiated controls, as expected (Fig. 2(B)nd (C)). However, there are pronounced differences i

re impaired in resuming cell cycle. (A) The percentage of G0/G1 cts were synchronized into G0/G1 by serum starvation for 72 h, irradia(6 Gy),vested and analyzed at the noted times. (B) Untreated SCIDA and wious populations of SCIDA (04, 05) and unrelated wild-type (AK, AUy
dicated times. Cells were fixed and stained with propidium iodide and anti-BrdUPI fluorescence. (C) SCIDA cells accumulate in G2/M and exhibit persistent celly), harvested and analyzed as above.


extent of the delays between SCIDA and controls. As early as6 h post-irradiation a large fraction of labeled wild-type cellshave completed S-phase, whereas a significant proportion oflabeled SCIDA cells remain in S-phase. At 24 h post-IR, bothcell types have completed S-phase but only wild-type cellshave completed the cell cycle and begun to divide. The con-trast between SCIDA and wild-type cells is most evident at48 h post-IR when∼95% of the labeled SCIDA cells persist atthe G2/M boundary whereas∼40% of the labeled wild-typecells have cycled into G1. These studies reveal that SCIDAcells have an impaired or delayed ability to resume cell cycleafter DNA damage, similar to the extended arrest observedat the G1/S boundary. Both results are consistent with a re-quirement for Artemis in repair of IR-induced DNA damage.Moreover, these data reveal that Artemis-deficiency alterscellular responses to DNA damage encountered in G0/G1 aswell as in S-phase.

3.4. Endogenous Artemis protein ishyper-phosphorylated in response to IR

We next investigated Artemis protein and its modifica-tion in vivo to gain further insights into its function(s) inresponse to DNA damage. Immunoprecipitation followedby Western analysis was used to detect Artemis proteinin irradiated and control lymphoblast cell extracts. We ob-s oweds top od top hos-p emist pro-t va-t ellsw rtemisml rtemisi ivoa izingr

3P

s-p beaw wec m-p entsv fA esi on-s IR(

To further investigate the kinase(s) responsible for Artemisphosphorylation in vivo, we used the PIKK inhibitors wort-mannin and caffeine. The relative sensitivity of the threeknown PIKKs to the fungal toxin wortmannin has been es-tablished as, DNA-PK≈ ATM > ATR [39,40]. Nearly com-plete inhibition of DNA-PK and ATM kinase activity isobserved at approximately 30�M wortmannin, while ATRis significantly more resistant, requiring over 100�M forhalf maximal inhibition. To eliminate the immunoprecipi-tation step required to observe Artemis from lymphoblas-toid cells, in these experiments we used human 293 kid-ney cells inducibly over-expressing Artemis from an inte-grated cDNA vector under tetracycline control (293-A) anddirectly probed cellular extracts. 293-A cells were irradi-ated in the presence of increasing amounts of wortman-nin, and direct Western blot analysis was used to detectArtemis (Fig. 3(C)). Artemis hyper-phosphorylation was notsignificantly inhibited at 10�M, but the majority was in-hibited at 25�M, and near complete inhibition was ob-served with the addition of 50�M wortmannin. These dataindicate that in vivo IR-induced hyper-phosphorylation ofArtemis is PIKK dependent and are consistent with a re-quirement for DNA-PK and/or ATM activity. It is interest-ing to note that 100�M wortmannin, which significantlyinhibits all three PIKKs, does not perturb basal phospho-rylation of Artemis in irradiated cells (Fig. 3(C)). This ob-s misi ep-a NA-P tingA as-s A-P anA ivo(

3p

nset ellsw ),t N2),a Wec senceo ibitP sA nctfN ish notah ande entr o-s to

erved that endogenous Artemis from irradiated cells shignificantly impeded migration on SDS-PAGE relativerotein from control cells (Fig. 3(A), lanes 1 and 2). Tetermine if the protein’s altered mobility was duehosphorylation, we treated immunoprecipitates with phatase prior to electrophoresis and found that Art

hen migrated more rapidly than unirradiated controlein (Fig. 3(A), lanes 3 and 4). To confirm this obserion, we treated immunoprecipitates from unirradiated cith phosphatase and found that phosphatase-treated Aigrated more rapidly than untreated protein (Fig. 3(A),

anes 3–5). These results establish that endogenous As constitutively phosphorylated at a basal level in vnd becomes hyper-phosphorylated in response to ionadiation.

.5. IR-induced Artemis hyper-phosphorylation isI3-like kinase dependent

DNA-PK, ATM, and ATR are three different phohatidylinositol 3-kinase-like kinases (PIKKs) known toctivated in response to DNA damage[37]. To determinehether ATM is required for Artemis phosphorylationompared the mobility of Artemis from irradiated lyhoblastoid cells from ataxia telangiectasia (AT) patis. normal individuals[31,38]. We found the mobility ortemis to be reduced in both wild-type and AT cell lin

n response to IR, indicating that ATM is not solely respible for Artemis hyper-phosphorylation in response toFig. 3(B)).

ervation indicates that basal phosphorylation of Artes independent of PIKK activity and/or temporally srated from DNA damage. To determine whether DK is capable of directly recognizing and phosphorylartemis, we carried out an in vitro DNA-PK kinaseay with purified recombinant proteins. We find that DNK hyper-phosphorylates Artemis in vitro resulting inrtemis mobility change akin to that observed in v

Fig. 3(D)).

.6. Different DNA damage classes cause varioushosphorylate states of Artemis

We next tested Artemis phosphorylation in respoo various types of DNA damage. We treated 293-A cith highly toxic doses of X-rays, ultraviolet light (UV

he interstrand crosslinking agent nitrogen mustard (Hnd the type II topoisomerase inhibitor etoposide.arried out these experiments in the presence or abf caffeine, a radiosensitizing compound known to inhIKK activities [41]. Treatment of cells with UV inducertemis hyper-phosphorylation but confers mobility disti

rom that of Artemis from irradiated cells (Fig. 3(E)).otably, 10 mM caffeine nearly completely blocks Artemyper-phosphorylation in response to UV but doesffect IR-induced phosphorylation (Fig. 3(E)). Artemisyper-phosphorylation was also induced by both HN2toposide, albeit to differing extents, with HN2 treatmesulting in a smaller shift in Artemis mobility than etopide (Fig. 3(F)). Artemis phosphorylation in response


HN2 is not inhibited by caffeine, while the response toetoposide treatment is partially inhibited (Fig. 3(F)). Thesedata together show that multiple hyper-phosphorylationstates of Artemis exist in vivo and suggest that differentPIKKs are likely to participate in Artemis phosphoryla-tion in response to various DNA damage classes and/orlevels.

3.7. Hyper-phosphorylation of Artemis is an early eventin response to DNA damage

We next investigated Artemis hyper-phosphorylationkinetics with respect to DSBs, using�H2AX as a tem-poral benchmark for DSB induction and repair. 293-Acells expressing recombinant Artemis were exposed to a

FbcSpAApAsoii

ig. 3. Artemis is basally phosphorylated, and hyper-phosphorylated in resasally phosphorylated (compare lanes 1 and 3) and hyper-phosphorylatedells was enriched by immunoprecipitated using anti-Artemis chicken antiboDS-PAGE, and probed with anti-Artemis rabbit serum. The three differenthosphorylation occurs in ATM deficient cells. Endogenous Artemis was immT lymphoblasts, resolved on SDS-PAGE, transferred to nitrocellulose and prrtemis expressing 293-A cells were treated with the indicated doses of worhosphatase where noted, resolved on SDS-PAGE and probed for Artemis. (rtemis (lane 1), purified Ku (lane 2) and an Artemis/DNA-PK kinase reactiotaining. (E) Different hyper-phosphorylation forms Artemis are evident afterf ultraviolet and X-ray radiations in the presence of noted doses of caffeine (

nduce different phosphorylation states of Artemis. Western analysis of 293-An the presence of noted doses of caffeine (P denotes phosphatase treatmen

ponse to IR in a PI3-like kinase-dependent manner. (A) Endogenous Artemis is(lane 2) in response to IR. Endogenous Artemis from wild-type lymphoblastoid

dies, incubated with�-phosphatase where indicated, resolved on high-resolutionstates of the protein observed are indicated by a, b, and c. (B) Artemis hyper-

unoprecipitated from extracts of both irradiated (10 Gy) and control wild-type andobed for Artemis. (C) Artemis hyper-phosphorylation is inhibited by wortmannin.tmannin for 30 min prior to irradiation (10 Gy), cell extracts were treatedwith �-D) DNA-PK can directly hyper-phosphorylate Artemis in vitro. Purified full-lengthn (lane 3) resolved by 4–12% gradient SDS-PAGE and visualized by coomassieUV and IR treatment. Western analysis of 293-A cell extracts after noted dosesP denotes phosphatase treatment). (F) Nitrogen mustard and etoposidetreatmentscells after 30 min treatments with noted doses of nitrogen mustard andetoposidet).


Fig. 4. Artemis hyper-phosphorylation occurs rapidly in response to IR, and persists longer than the DSB marker�H2AX. (A) Artemis expressing 293-A cellswere treated with 10 Gy of IR and harvested at the noted times, cell extracts were resolved on high-resolution SDS-PAGE and probed for Artemis and�H2AX.(C denotes unirradiated controls). (B) Artemis hyper-phosphorylation persists for 4 days after IR. Western analysis was performed on IR treated cells at thenoted times as described above. (C) Artemis hyper-phosphorylation persists for over 24 h after a minimal saturating dose of 2 Gy. Artemis expressing 293-Acells were treated with 2 Gy and analyzed as described above. (D) Hyper-phosphorylation of endogenous Artemis persists for at least 12 h. Endogenous Artemisfrom wild-type lymphoblastoid cells was enriched by immunoprecipitated with anti-Artemis chicken antibodies, resolved on high-resolution SDS-PAGE, andanalyzed by Western blot with anti-Artemis rabbit serum (2P denotes phosphatase treatment of 2 h time point).

10 Gy dose of IR, harvested at various times followingdamage, and analyzed by Western blot for both Artemishyper-phosphorylation and�H2AX. Hyper-phosphorylationof Artemis occurred very rapidly after DNA damage,with near complete hyper-phosphorylation evident by15–30 min, in close synchrony with the appearance of�H2AX (Fig. 4(A)). Consistent with previous reports,�H2AX returns to near basal levels by 6 h post-IR as visu-alized by Western blot[40]. Surprisingly, Artemis remainedpersistently hyper-phosphorylated, with virtually no basallyphosphorylated protein detectable out to 24 h after DNAdamage (Fig. 4(A)). We performed similar studies out to 96 hpost-irradiation and found that Artemis remained almostentirely in the hyper-phosphorylated form up to 48 h afterIR with appreciable amounts of de-phosphorylated proteinonly evident after 72 h (Fig. 4(B)). Moreover, a completereturn to a basally phosphorylated state was not observedwithin the course of this experiment, with the majority ofprotein remaining hyper-phosphorylated even 4 days afterirradiation. These results establish that like�H2AX, Artemishyper-phosphorylation occurs rapidly in response to DNAdamage, but unlike�H2AX, Artemis hyper-phosphorylationpersists for long periods following irradiation.

3.8. Artemis phosphorylation is uncoupled from them

mentsi therA af-t leteA pec ism isp ost-

IR (Fig. 4(C)), indicating that persistent Artemis hyper-phosphorylation is not specific to extreme levels of DNAdamage. To ensure that hyper-phosphorylation was not anartifact of Artemis over-expression in 293-A cells, we de-termined whether endogenously expressed protein exhibitedsimilar phosphorylation kinetics. We observed nearly com-plete hyper-phosphorylation of endogenous Artemis in lym-phoblastoid cells for up to 12 h following 2 Gy IR (Fig. 4(D)),after which the apoptotic response of these cells precludedreliable protein analysis. These results indicate that persis-tent IR-induced Artemis hyper-phosphorylation is a normalresponse observed with both endogenous and recombinantArtemis protein, and persists well beyond the time requiredfor the vast majority of DSB repair (Fig. 1).

3.9. Artemis de-phosphorylation coincides with cellcycle progression to G2/M

To further investigate the kinetics of Artemis de-phosphorylation, 293-A cells were irradiated, harvested at2 h after damage to evaluate the initial degree of hyper-phosphorylation, and at 48 h after IR to measure the amountof hyper-phosphorylated protein remaining (Fig. 5(A)). Fol-lowing a 0.5 Gy dose,∼50% of the protein is hyper-phosphorylated 2 h after IR, with the majority of the pro-t 8 h.A per-p irelyi pro-p pro-t eent tet statei selyp

ajority of DSB repair events

Because the 10 Gy dose used in the preceding experis lethal to the vast majority of cells, we tested whertemis would return to a basally phosphorylated state

er a lower dose of 2 Gy. This dose elicits nearly comprtemis hyper-phosphorylation, yet a majority of wild-tyells will survive this amount of radiation. Following thinimal saturating dose of 2 Gy, the majority of Artemersists in the hyper-phosphorylated form out to 24 h p

ein returning to a basally phosphorylated state by 4fter a 2 Gy dose, the protein is nearly completely hyhosphorylated at 2 h, and by 48 h Artemis is almost ent

n the basal form. At doses exceeding 2 Gy we observe aortional decrease in the fraction of de-phosphorylated

ein after 48 h, and at 8 Gy there is little difference betwhe 2 and 48 h time points (Fig. 5(A)). These data indicahat the return of Artemis to a basally phosphorylateds regulated by the amount of DNA damage and inverroportional to IR doses.


Fig. 5. Artemis de-phosphorylation coincides with mitosis following DNA damage. (A) The rate at which Artemis is de-phosphorylated corresponds to theinitial dose of IR. Artemis expressing 293-A cells were irradiated at various doses, harvested at the noted times, and cell extracts were resolved on high-resolutionSDS-PAGE and analyzed by Western blot using rabbit anti-Artemis antibodies. (B) The levels of Artemis hyper-phosphorylation and G2/M accumulationbothincrease in a dose-dependent manner and to a similar degree following IR. Samples from each time point were divided and analyzed by both Western blot andflow cytometry. The percentage of BrdU positive cells in G2/M (gray bars), and the percentage of hyper-phosphorylated Artemis protein (crosshatchedbars)were quantified and plotted. (C) Artemis de-phosphorylation correlates with mitosis. Artemis expressing 293-A cells were synchronized by double thymidineblock, treated with IR (2 Gy), stimulated to resume cell cycle by addition of fresh medium, and harvested at the noted times. Cell extracts were then resolvedon high-resolution SDS-PAGE and analyzed by Western blot using Artemis and phospho-H3 antibodies. Cell cycle distributions were also evaluated based onDNA content by flow cytometry at each time point.

Because Artemis de-phosphorylation occurred between24 and 48 h after moderate DNA damage and is delayedbeyond 96 h after high doses of IR, we considered that de-phosphorylation may be linked to cell cycle progression. Toinvestigate this possibility, we compared cell cycle statusand Artemis phosphorylation status at 48 h after IR. Cellswere pulse labeled with BrdU and analyzed both by West-ern blot (Fig. 5(A)) and flow cytometry as in the experimentsof Fig. 2(C). The percent of hyper-phosphorylated Artemiswas quantified and plotted together with the correspondingpercent of BrdU positive cells arrested in G2/M (Fig. 5(B)).We observe a clear correlation between the percentage of

hyper-phosphorylated Artemis and the percentage of cellsthat persist in G2/M. This coincident trend is suggestive of alink between cell cycle status and Artemis phosphorylationstatus.

To further investigate this relationship we used a doublethymidine block to synchronize Artemis-expressing 293-Acells at the G1/S boundary. The synchronized cultures wereexposed to a 2 Gy dose of IR then stimulated to resume cell cy-cle progression by removal of thymidine and addition of freshmedium containing 15% serum. We then evaluated Artemisphosphorylation status and cell cycle distributions at vari-ous times after irradiation (Fig. 5(C)). Consistent with our


earlier observations (Fig. 4(C)), asynchronous cultures showonly a minor amount of Artemis returning to a basally phos-phorylated state 24 h after a 2 Gy dose (Fig. 5(C)). In con-trast, the synchronized cell cultures show nearly half of theArtemis has returned to a basally phosphorylated state by20 h following IR, indicating that cell cycle status influences

Artemis de-phosphorylation kinetics (Fig. 5(C)). Moreover,basally phosphorylated Artemis is observed as early as 12 hpost-irradiation in the synchronous cultures and increasesin proportion thereafter. To more specifically determine thecell cycle status of these cultures, samples were also probedfor phosphorylation of histone H3 (phospho-H3), which oc-

Fdc(iToi4t

ig. 6. SCIDA cells are more sensitive to IR than etoposide. (A) Clonogenic coses. Asynchronous fibroblasts were plated in triplicate, irradiated and incuells are less sensitive to etoposide than IR. Cells were treated for 1 h withC) �H2AX and Artemis hyper-phosphorylation induction in response to a rrradiation with noted doses, cell extracts were resolved on high-resolution So ensure normalized loading Ku protein levels were also determined. (D) Lef etoposide. Cells were treated with etoposide for 1 h then harvested and an

n SCIDA relative to wild-type cells. Cells were labeled and analyzed as inFig. 28 h after treatment at the noted doses. (F) Etoposide treatment does not ind

he noted concentrations of etoposide for 1 h then labeled and analyzed as i

ell survival assays of hTERT-immortalized fibroblasts at increasing�-irradiationbated for 10–14 days, and resulting colonies were stained and scored.(B) SCIDAnoted doses of etoposide, washed with PBS, then processed as described above.ange of IR doses. Artemis expressing 293-A cells were harvested 30 min afterDS-PAGE and analyzed by Western blot using Artemis, and�H2AX antibodies.vels of�H2AX and Artemis hyper-phosphorylation in response to various dosesalyzed described above. (E) IR causes significantly increased G2/M accumulation

the percentage of BrdU labeled cells in the G2/M peak was determined atuce G2/M accumulation in SCIDA cells. Cells were labeled as above, treatedwith

n (E).


curs specifically during mitosis[42,43]. Significantly, the ini-tial appearance of basally phosphorylated Artemis at 12 hpost-IR coincides with the initial increase in phospho-H3levels and the accumulation of synchronous cells in G2/M(Fig. 5(C)). These data show that Artemis de-phosphorylationoccurs concurrently with mitosis and that de-phosphorylationof Artemis is dictated by cell cycle progression and not viceversa.

3.10. Artemis deficiency confers differential sensitivityto IR and etoposide

While Artemis and DNA-PK deficiencies hinder the re-pair of IR-induced DSBs to different extents (Fig. 1), it ispossible that Artemis is required for repair of a small sub-set of DSBs critical for survival. We reasoned that if thiswere the case, an agent that produced a more homogeneouspopulation of DSBs could either exacerbate or amelioratethe hypersensitivity of SCIDA cells seen in response to IR,depending on the Artemis requirement for repair. The drugetoposide introduces a relatively homogeneous populationof enzymatically induced DSBs and while more toxic to S-phase cells, induces breaks throughout the cell cycle[44]. Incontrast, IR introduces a broad spectrum of different typesof DNA lesions including a variety of base damages, single-strand breaks (SSBs) and DSBs and has the unique character-i ions[

et b-l eter-m ns-fS ntI Rs ild-t ntlyh ents( d IRa butm am-a SBs[o pec-t per-pi e( la-t topo-s )t ivala

nsi-t per-p esti-g and

cell cycle progression, we measured the percentage of cellsaccumulating at the G2/M boundary after treatment witheither IR or etoposide. Normal or SCIDA fibroblasts werepulse labeled with BrdU as described for the experiments ofFig. 2(C), then either immediately irradiated or treated for 1 hwith etoposide. Following treatment, fresh medium contain-ing 15% serum was added and cells were cultured for an ad-ditional 48 h. We then analyzed the BrdU-containing cells todetermine the fate of cells that were in S-phase when exposedto the respective DNA damaging agents. Consistent with ourprevious observations (Fig. 2), SCIDA cells showed a dra-matic increase in the percentage of G2/M cells relative to nor-mal cells following irradiation in S-phase with doses greaterthan 1 Gy (Fig. 6(E)). Clear increases in G2/M populationswere also observed with increasing etoposide doses, but theG2/M accumulation in SCIDA cells was not significantly dif-ferent than that observed in wild-type cells (Fig. 6(F)). Anal-ysis of the entire cell population showed a similar distinctionbetween IR and etoposide treatments (data not shown). Thus,while �H2AX and Artemis hyper-phosphorylation levels in-dicate similar numbers of DSBs, SCIDA cells exhibit greatlyincreased cell death and cell cycle blockage following IRtreatment relative to etoposide treatment. Therefore, the na-ture of the DNA damage dictates the requirement for Artemisfunction irrespective of Artemis hyper-phosphorylation in-duction.

4

cellc l ast s inr inese un-c thatA ions IDa t hu-m rc singbD thatS Bsa viousr vea n of� asm lowd ire-m cedD withA re-p arlyd ndX air.

stic of producing clusters of these different types of les45,46].

To gain insight into Artemis function(s) in DNA repair, wreated hTERT-immortalized wild-type and SCIDA fibroasts with various doses of IR and etoposide and then d

ined survival by colony formation. Similar to SV40 traormed SCIDA fibroblasts (Fig. 1(A)), hTERT-immortalizedCIDA cells from different patients exhibited significa

R sensitivity (Fig. 6(A)), establishing that SCIDA cell Iensitivity is independent of p53 status. Using the wype survival as a reference, SCIDA cells show significaigher sensitivity to IR treatments than etoposide treatmFig. 6(A) and (B)). The chosen doses of etoposide anre comparable in terms of wild-type cellular survival,ay not be comparable with respect to levels of DNA dge. Because H2AX phosphorylation is proportional to D

36,47], we measured the levels of�H2AX as an indicationf the relative number of DSBs induced after the res

ive IR and etoposide treatments. Complete Artemis hyhosphorylation and a detectable increase in�H2AX is ev-

dent following 2 Gy of IR, and a 2.5�M dose of etoposidFig. 6(C) and (D)). Using H2AX and Artemis phosphoryion as a means to estimate “equivalent DSB” doses of eide and IR (approximately 2.5�M and 2 Gy, respectivelyhe difference between wild-type and SCIDA cell survfter IR exposure is even more dramatic (Fig. 6(A) and (B)).

Artemis-deficient cells are not particularly hyperseive to etoposide treatment but Artemis protein is as hyhosphorylated in response to etoposide. To further invate the relationship between Artemis phosphorylation

. Discussion

In this study, we have investigated DSB repair andycle arrest functions in Artemis-deficient cells as welhe nature and kinetics of Artemis protein modificationesponse to DNA damage. We have utilized SCIDA cell lstablished from individuals deficient for Artemis butharacterized for cellular radiation sensitivity. We showrtemis-deficient SCIDA cells exhibit pronounced radiatensitivity similar to that described previously for RS-SCnd comparable to that reported for DNA-PKcs-deficienan cell line M059J (Fig. 1(C)) [34,35]. The DSB repai

haracteristics of SCIDA cells were also investigated uoth the FAR assay and the surrogate DSB marker�H2AX.espite, their high-degree of radiosensitivity, we findCIDA cells have essentially normal rejoining of total DSs measured by the FAR assay, in agreement with a preeport for RS-SCID cells[35]. However, the more sensitinalysis of DSB rejoining made possible by quantitatioH2AX foci reveals that SCIDA cells have about twiceany unrepaired DSBs than controls after a relativelyose of IR (2 Gy). This finding is consistent with a requent for Artemis in repair of a small subset of IR-induSBs. Thus, the observed radiosensitivity associatedrtemis deficiency may be attributable directly to a DNAair defect. Nonetheless, Artemis function in NHEJ is cleistinct from that of DNA-PKcs, Ku70, Ku80, ligase IV aRCC4, all of which exhibit severely defective DSB rep


We show here that endogenous Artemis protein existsin a basally phosphorylated state and is rapidly hyper-phosphorylated after both ionizing radiation and other DNAdamaging treatments. The complete hyper-phosphorylationof Artemis observed thirty minutes after UV irradiation sug-gests that DSBs are not required for this response. Addition-ally, our studies reveal that various DNA damaging agentsresult in different Artemis phosphorylation states that are in-hibited to differing extents by caffeine, further suggestingthat multiple PIKKs likely play a role in the modification ofArtemis under various conditions. Similar observations as tothe involvement of multiple kinases in Artemis phosphory-lation were published elsewhere during the final preparationof this manuscript[29]. Unexpectedly, we find that Artemisde-phosphorylation kinetics are extremely slow, distinct fromthose of H2AX and DNA-PK[36,48], and delayed beyondthe vast majority of DSB repair events (compareFig. 4withFig. 1(B) and (C)). Instead, Artemis de-phosphorylation af-ter IR is coincident with cell cycle progression to or throughmitosis. Together, these observations raise the question as towhether Artemis-deficient cells have defects similar to that ofAT cells, possibly cell cycle checkpoint defects that may con-tribute to their IR sensitivity[31,38]. While our findings arein agreement with the aforementioned study with respect toArtemis phosphorylation, Zhang et al. conclude that Artemisis necessary for maintenance of the G2/M checkpoint[29].I ans-f eret de-f cellc tirelyc seto mis-d ath.F t IRd emish ersis-t 2/Ma orre-l al-t sig-n tiono et hro-n n isd andn f thec , thet di-m fromm tran-s mi-t ationa rt ands rmits

cell cycle progression, which in turn results in Artemisde-phosphorylation.

Our studies show that the nature of the DNA damage dic-tates the Artemis requirement for survival as well as for cellcycle progression through G2/M. In contemplating the pos-sible characteristics of Artemis substrates in vivo, severalpoints are worth consideration. Artemis functions in V(D)Jrecombination as an endonuclease providing a unique hair-pin opening activity. There is a notable lack of redundancyfor this function as evidenced by the failure to resolve hairpinintermediates of V(D)J recombination and the consequent B-T-NK+ SCID phenotype in Artemis-deficient patients. Fur-thermore, Artemis-deficient cells are defective only in repair-ing the hairpin-terminated coding joints in V(D)J recombina-tion but retain the ability to rejoin blunt-ended signal joints[4,8]. These phenotypes indicate that hairpin DNA terminiare in vivo substrates of Artemis and raise the possibilitythat Artemis may have similar substrate specificities in bothV(D)J recombination and NHEJ, namely DSBs with hairpintermini. DSBs in close proximity to inverted repeats havelong been proposed to form such termini, and this character-istic is thought to be responsible for the particularly unstablenature of inverted repeats in the human genome[51–53]. Onepossibility is that Artemis functions specifically in the repairof hairpin termini arising among the spectrum of DSBs in-duced by ionizing radiation. The fact that Artemis-deficientc bye t ter-m ni,a . Theh n ofb nt ofh stentw ula-t ity,a lasso otalD ob-s ells[ re-d ility[ tedi uallyc ageo NAd

theh tiont en-d ish per-s ichm ellc tioni sidew er-

n contrast to those findings, which were made with trormed 293 kidney cells depleted for Artemis, we show hhat primary human SCIDA fibroblasts with a geneticect in Artemis accumulate at both the G1/S and G2/Mycle boundaries after IR damage. These data are enonsistent with Artemis functioning in the repair of a subf DSBs, and the persistence of these breaks in Arteeficient cells causing G2/M accumulation and cell deurther consistent with this interpretation, we find thaoses exceeding that required for the saturation of Artyper-phosphorylation (above 2 Gy), cause increased p

ence of the hyper-phosphorylated form of Artemis and Gccumulation. Because Artemis de-phosphorylation c

ates with G2/M accumulation, we also considered theernative possibility that Artemis phosphorylation mayal cell cycle arrest. However, the fact that synchronizaf 293-A cells prior to IR significantly diminishes the tim

aken for Artemis de-phosphorylation relative to unsyncized controls indicates that Artemis de-phosphorylatioictated by cell cycle progression to or through G2/Mot the converse. Thus, by changing only the fraction oell cycle that must be traversed prior to reaching mitosisime required for Artemis de-phosphorylation is sharplyinished. Many proteins are de-phosphorylated at exititosis and PP1 phosphatase activity is required for this

ition. This increased phosphatase activity at the exit fromosis may also be responsible for Artemis de-phosphorylnd resetting Artemis to an inactive state[49,50]. Togethe

hese data indicate that the repair of DNA damageubsequent satisfaction of cell cycle checkpoints pe

ells show little increase in sensitivity to DSBs inducedtoposide relative to DSBs induced by IR indicates thaini requiring Artemis function, possibly hairpin termirise more abundantly after IR than etoposide treatmentseterogeneity of IR-induced damage and the formatiolocked termini may lead to a more frequent developmeairpin termini at DSBs. These notions are entirely consiith the observed phenotypes of G1/S and G2/M accum

ion, significant IR sensitivity, lack of etoposide sensitivnd a small defect in overall DSB rejoining, as this cf DSBs would represent only a small fraction of the tSBs. Notably, prior to the discovery of Artemis and theerved genomic instability in Artemis-deficient murine c54], an NHEJ-interacting hairpin opening activity was picted to resolve hairpin-DSBs to prevent genomic instab

55]. However, aside from the hairpin specificity exhibin V(D)J recombination, these other phenotypes are eqonsistent with Artemis function on clustered DNA damr other relatively rare substrates unique to IR-induced Damage.

While our studies have not defined a function foryper-phosphorylation of Artemis, Artemis phosphoryla

hrough interaction with DNA-PK is known to activate theonuclease activity of Artemis[28]. The prolonged Artemyper-phosphorylation we observe after IR may reflectistent activation of Artemis endonuclease activity, whay have utility in repairing complex DSBs and allowing c

ycle resumption. However, Artemis hyper-phosphorylas efficiently induced by UV, nitrogen mustard, and etopohile Artemis-deficient cells have a notable lack of hyp


sensitivity to etoposide, suggesting that activation of Artemisoccurs in response various forms of DNA damage not just thesubset whose repair requires Artemis function. The activationof Artemis by phosphorylation in response to apparently ir-relevant types of DNA damage may reflect a failsafe mecha-nism to deal with hairpin DSBs and/or other complex lesionsarising from DNA replication, repair, or other processes in-volving a damaged genome. Although we cannot formallyrule out the possibility that Artemis phosphorylation has mul-tiple functions, our data indicate that it is the nature of theDNA damage that is the determining factor in survival andG2/M accumulation of Artemis-defective cells, irrespectiveof ability to induce Artemis hyper-phosphorylation.

The hairpin opening activity of Artemis[28], slight DSBrepair defect, substantial radiation sensitivity, G2/M accu-mulation after IR, and the genomic instability in Artemis-deficient murine cells[54], are all consistent with Artemisuniquely functioning in the processing of a small subset ofDNA breaks. Therefore, we propose that the radiation sen-sitivity observed in Artemis-deficient cells is principally dueto failed resolution of a small subset of complex DSBs, likelythose at which hairpin termini arise.

Acknowledgements

dw udyC jornR ionr UCB hisw ento ught d byN C),a

R

.G.on-A.cco,Vali-K.AG

pre-

lo,. Im-

d-erio,tive

.J.ase in

Athabascan-speaking Native Americans is located on chromosome10p, Am. J. Human Genet. 62 (1998) 136–144.

[5] S. Murphy, A. Hayward, G. Troup, E.J. Devor, T. Coons, Geneenrichment in an American Indian population: an excess of severecombined immunodeficiency disease, Lancet 2 (1980) 502–505.

[6] J.F. Jones, C.K. Ritenbaugh, M.A. Spence, A. Hayward, Severe com-bined immunodeficiency among the Navajo. I. Characterization ofphenotypes, epidemiology, and population genetics, Human Biol. 63(1991) 669–682.

[7] M. Cavazzana-Calvo, F. Le Deist, G. De Saint Basile, D. Pa-padopoulo, J.P. De Villartay, A. Fischer, Increased radiosensitivityof granulocyte macrophage colony-forming units and skin fibroblastsin human autosomal recessive severe combined immunodeficiency,J. Clin. Invest. 91 (1993) 1214–1218.

[8] N. Nicolas, D. Moshous, M. Cavazzana-Calvo, D. Papadopoulo, R.de Chasseval, F. Le Deist, A. Fischer, J.P. de Villartay, A humansevere combined immunodeficiency (SCID) condition with increasedsensitivity to ionizing radiations and impaired V(D)J rearrangementsdefines a new DNA recombination/repair deficiency, J. Exp. Med.188 (1998) 627–634.

[9] D. Moshous, I. Callebaut, R. de Chasseval, B. Corneo, M.Cavazzana-Calvo, F. Le Deist, I. Tezcan, O. Sanal, Y. Bertrand, N.Philippe, A. Fischer, J.P. de Villartay, Artemis, a novel DNA double-strand break repair/V(D)J recombination protein, is mutated in hu-man severe combined immune deficiency, Cell 105 (2001) 177–186.

[10] L. Li, D. Moshous, Y. Zhou, J. Wang, G. Xie, E. Salido, D. Hu, J.P.de Villartay, M.J. Cowan, A founder mutation in Artemis, an SNM1-like protein, causes SCID in Athabascan-speaking Native Americans,J. Immunol. 168 (2002) 6323–6329.

[11] D. Moshous, L. Li, R. Chasseval, N. Philippe, N. Jabado, M.J.Cowan, A. Fischer, J.P. de Villartay, A new gene involved in DNA

d on88.

[ an,

[ enes

[ ions

[ dentRes.

[ , C.D.J.

T186

[ on-pro-esis,

[ ng,oth-lyticim-

6.[ and

callyiol.

[ uble-IV,

[ nn,r-97)

We thank Dr. Martin Lavin for providing the AT anild-type lymphoblast cells and are grateful to Drs. Jampisi, Albert Davalos, Sandeep Burma, and Bydberg for providing helpful comments and discuss

egarding this work. We also thank Hector Nolla at theerkeley flow cytometry facility for his assistance. Tork is supported by the Office of Science, US Departmf Energy, under contract no. DE-AC03-76SF00098 thro

he Low Dose Radiation Research Program (PKC) anIH grants AI28339 (MJC), CA86936 and CA50519 (DJnd PO1-CA92584 (PKC and DJC).

eferences

[1] A. Villa, C. Sobacchi, L.D. Notarangelo, F. Bozzi, M. Abinun, TAbrahamsen, P.D. Arkwright, M. Baniyash, E.G. Brooks, M.E. Cley, P. Cortes, M. Duse, A. Fasth, A.M. Filipovich, A.J. Infante,Jones, E. Mazzolari, S.M. Muller, S. Pasic, G. Rechavi, M.G. SaS. Santagata, M.L. Schroeder, R. Seger, D. Strina, A. Ugazio, J.aho, M. Vihinen, L.B. Vogler, H. Ochs, P. Vezzoni, W. Friedrich,Schwarz, V(D)J recombination defects in lymphocytes due to Rmutations: severe immunodeficiency with a spectrum of clinicalsentations, Blood 97 (2001) 81–88.

[2] A. Villa, S. Santagata, F. Bozzi, L. Imberti, L.D. NotarangeOmenn syndrome: a disorder of Rag1 and Rag2 genes, J. Clinmunol. 19 (1999) 87–97.

[3] K. Schwarz, G.H. Gauss, L. Ludwig, U. Pannicke, Z. Li, D. Linner, W. Friedrich, R.A. Seger, T.E. Hansen-Hagge, S. DesidM.R. Lieber, C.R. Bartram, RAG mutations in human B cell-negaSCID, Science 274 (1996) 97–99.

[4] L. Li, D. Drayna, D. Hu, A. Hayward, S. Gahagan, H. Pabst, MCowan, The gene for severe combined immunodeficiency dise

double-strand break repair and V(D)J recombination is locatehuman chromosome 10p, Human Mol. Genet. 9 (2000) 583–5

12] S.E. Critchlow, S.P. Jackson, DNA end-joining: from yeast to mTrends Biochem. Sci. 23 (1998) 394–398.

13] G.C. Smith, S.P. Jackson, The DNA-dependent protein kinase, GDevelop. 13 (1999) 916–934.

14] V.A. Zakian, ATM-related genes: what do they tell us about functof the human gene? Cell 82 (1995) 685–687.

15] W.S. Dynan, S. Yoo, Interaction of Ku protein and DNA-depenprotein kinase catalytic subunit with nucleic acids, Nucl. Acids26 (1998) 1551–1559.

16] H. Ouyang, A. Nussenzweig, A. Kurimasa, V.C. Soares, X. LiCordon-Cardo, W. Li, N. Cheong, M. Nussenzweig, G. Iliakis,Chen, G.C. Li, Ku70 is required for DNA repair but not forcell antigen receptor gene recombination in vivo, J. Exp. Med.(1997) 921–929.

17] A. Kurimasa, H. Ouyang, L.J. Dong, S. Wang, X. Li, C. CordCardo, D.J. Chen, G.C. Li, Catalytic subunit of DNA-dependenttein kinase: impact on lymphocyte development and tumorigenProc. Natl. Acad. Sci. U.S.A. 96 (1999) 1403–1408.

18] G.E. Taccioli, A.G. Amatucci, H.J. Beamish, D. Gell, X.H. XiaM.I. Torres Arzayus, A. Priestley, S.P. Jackson, A. Marshak Rstein, P.A. Jeggo, V.L. Herrera, Targeted disruption of the catasubunit of the DNA-PK gene in mice confers severe combinedmunodeficiency and radiosensitivity, Immunity 9 (1998) 355–36

19] K.J. Lee, J. Huang, Y. Takeda, W.S. Dynan, DNA ligase IVXRCC4 form a stable mixed tetramer that functions synergistiwith other repair factors in a cell-free end-joining system, J. BChem. 275 (2000) 34787–34796.

20] S.E. Critchlow, R.P. Bowater, S.P. Jackson, Mammalian DNA dostrand break repair protein XRCC4 interacts with DNA ligaseCurr. Biol. 7 (1997) 588–598.

21] U. Grawunder, M. Wilm, X. Wu, P. Kulesza, T.E. Wilson, M. MaM.R. Lieber, Activity of DNA ligase IV stimulated by complex fomation with XRCC4 protein in mammalian cells, Nature 388 (19492–495.


[22] B.L. Sibanda, S.E. Critchlow, J. Begun, X.Y. Pei, S.P. Jackson, T.L.Blundell, L. Pellegrini, Crystal structure of an Xrcc4-DNA ligase IVcomplex, Nat. Struct. Biol. 8 (2001) 1015–1019.

[23] T.D. Stamato, R. Weinstein, A. Giaccia, L. Mackenzie, Isolationof cell cycle-dependent gamma ray-sensitive Chinese hamster ovarycell, Somatic Cell Genet. 9 (1983) 165–173.

[24] M. Bryans, M.C. Valenzano, T.D. Stamato, Absence of DNA lig-ase IV protein in XR-1 cells: evidence for stabilization by XRCC4,Mutat. Res. 433 (1999) 53–58.

[25] A.E. Tomkinson, Z.B. Mackey, Structure and function of mammalianDNA ligases, Mutat. Res. 407 (1998) 1–9.

[26] H.-L. Hsu, S.M. Yannone, D.J. Chen, Defining interactions betweenDNA-PK and ligase IV/XRCC4, DNA Repair 1 (2002) 225–235.

[27] P. Calsou, C. Delteil, P. Frit, J. Drouet, B. Salles, Coordinated assem-bly of Ku and p460 subunits of the DNA-dependent protein kinaseon DNA ends is necessary for XRCC4-ligase IV recruitment, J. Mol.Biol. 326 (2003) 93–103.

[28] Y. Ma, U. Pannicke, K. Schwarz, M.R. Lieber, Hairpin opening andoverhang processing by an Artemis/DNA-dependent protein kinasecomplex in nonhomologous end joining and V(D)J recombination,Cell 108 (2002) 781–794.

[29] X. Zhang, J. Succi, Z. Feng, S. Prithivirajsingh, M.D. Story, R.J.Legerski, Artemis is a phosphorylation target of ATM and ATR andis involved in the G2/M DNA damage checkpoint response, Mol.Cell Biol. 24 (2004) 9207–9220.

[30] S.M. Yannone, S. Roy, D.W. Chan, M.B. Murphy, S. Huang, J.Campisi, D.J. Chen, Werner syndrome protein is regulated and phos-phorylated by DNA-dependent protein kinase, J. Biol. Chem. 276(2001) 38242–38248.

[31] J. Yan, K.K. Khanna, M.F. Lavin, Defective radiation signal trans-duction in ataxia-telangiectasia cells, Int. J. Radiat. Biol. 76 (2000)

[ isons incells. 153

[ on of125–

[ r,anther-

[ F. Letablepro-

mun-.

[ le-oses,

[ rs of13

[ pleBiol.

[39] J.N. Sarkaria, R.S. Tibbetts, E.C. Busby, A.P. Kennedy, D.E. Hill,R.T. Abraham, Inhibition of phosphoinositide 3-kinase related ki-nases by the radiosensitizing agent wortmannin, Cancer Res. 58(1998) 4375–4382.

[40] S. Burma, B.P. Chen, M. Murphy, A. Kurimasa, D.J. Chen, ATMphosphorylates histone H2AX in response to DNA double-strandbreaks, J. Biol. Chem. 276 (2001) 42462–42467.

[41] J.N. Sarkaria, E.C. Busby, R.S. Tibbetts, P. Roos, Y. Taya, L.M.Karnitz, R.T. Abraham, Inhibition of ATM and ATR kinase activi-ties by the radiosensitizing agent, caffeine, Cancer Res. 59 (1999)4375–4382.

[42] L.C. Mahadevan, A.C. Willis, M.J. Barratt, Rapid histone H3 phos-phorylation in response to growth factors, phorbol esters, okadaicacid, and protein synthesis inhibitors, Cell 65 (1991) 775–783.

[43] K. Ajiro, K. Yoda, K. Utsumi, Y. Nishikawa, Alteration of cellcycle-dependent histone phosphorylations by okadaic acid. Induc-tion of mitosis-specific H3 phosphorylation and chromatin conden-sation in mammalian interphase cells, J. Biol. Chem. 271 (1996)13197–13201.

[44] E. Estey, R.C. Adlakha, W.N. Hittelman, L.A. Zwelling, Cell cy-cle stage dependent variations in drug-induced topoisomerase IImediated DNA cleavage and cytotoxicity, Biochemistry 26 (1987)4338–4344.

[45] J.F. Ward, Biochemistry of DNA lesions, Radiat. Res. Suppl. 8(1985) S103–S111.

[46] B. Rydberg, L. Heilbronn, W.R. Holley, M. Lobrich, C. Zeitlin,A. Chatterjee, P.K. Cooper, Spatial distribution and yield of DNAdouble-strand breaks induced by 3–7 MeV helium ions in humanfibroblasts, Radiat. Res. 158 (2002) 32–42.

[47] O.A. Sedelnikova, E.P. Rogakou, I.G. Panyutin, W.M. Bonner, Quan-titative detection of (125)IdU-induced DNA double-strand breaks

.[ ory,

pro-ble-

[ einl. Rev.

[ ol of

[ , E.yotic

[ enko,the

[ urka,theome

[ J.uchi,is-

[ cific999)

1025–1035.32] B. Fouladi, C.A. Waldren, B. Rydberg, P.K. Cooper, Compar

of repair of DNA double-strand breaks in identical sequenceprimary human fibroblast and immortal hamster–human hybridharboring a single copy of human chromosome 11, Radiat. Res(2000) 795–804.

33] B.S. Hoppe, R.B. Jensen, C.U. Kirchgessner, Complementatithe radiosensitive M059J cell line, Radiat. Res. 153 (2000)130.

34] M.J. Allalunis-Turner, G.M. Barron, R.S. Day III, K.D. DobleR. Mirzayans, Isolation of two cell lines from a human malignglioma specimen differing in sensitivity to radiation and chemotapeutic drugs, Radiat. Res. 134 (1993) 349–354.

35] N. Nicolas, N.J. Finnie, M. Cavazzana-Calvo, D. Papadopoulo,Deist, A. Fischer, S.P. Jackson, J.P. de Villartay, Lack of detecdefect in DNA double-strand break repair and DNA-dependenttein kinase activity in radiosensitive human severe combined imodeficiency fibroblasts, Eur. J. Immunol. 26 (1996) 1118–1122

36] K. Rothkamm, M. Lobrich, Evidence for a lack of DNA doubstrand break repair in human cells exposed to very low X-ray dProc. Natl. Acad. Sci. U.S.A. 100 (2003) 5057–5062.

37] D. Durocher, S.P. Jackson, DNA-PK, ATM and ATR as sensoDNA damage: variations on a theme? Curr. Opin. Cell Biol.(2001) 225–231.

38] H. Beamish, R. Williams, P. Chen, M.F. Lavin, Defect in multicell cycle checkpoints in ataxia-telangiectasia postirradiation, J.Chem. 271 (1996) 20486–20493.

with gamma-H2AX antibody, Radiat. Res. 158 (2002) 486–49248] D.W. Chan, B.P. Chen, S. Prithivirajsingh, A. Kurimasa, M.D. St

J. Qin, D.J. Chen, Autophosphorylation of the DNA-dependenttein kinase catalytic subunit is required for rejoining of DNA doustrand breaks, Genes Develop. 16 (2002) 2333–2338.

49] H. Ceulemans, M. Bollen, Functional diversity of protphosphatase-1, a cellular economizer and reset button, Physio84 (2004) 1–39.

50] N. Berndt, Protein dephosphorylation and the intracellular contrthe cell number, Front. Biosci. 4 (1999) D22–D42.

51] D.A. Gordenin, K.S. Lobachev, N.P. Degtyareva, A.L. MalkovaPerkins, M.A. Resnick, Inverted DNA repeats: a source of eukargenomic instability, Mol. Cell Biol. 13 (1993) 5315–5322.

52] E. Akgun, J. Zahn, S. Baumes, G. Brown, F. Liang, P.J. RomaniS. Lewis, M. Jasin, Palindrome resolution and recombination inmammalian germ line, Mol. Cell Biol. 17 (1997) 5559–5570.

53] J.E. Stenger, K.S. Lobachev, D. Gordenin, T.A. Darden, J. JM.A. Resnick, Biased distribution of inverted and direct Alus inhuman genome: implications for insertion, exclusion, and genstability, Genome Res. 11 (2001) 12–27.

54] S. Rooney, F.W. Alt, D. Lombard, S. Whitlow, M. Eckersdorff,Fleming, S. Fugmann, D.O. Ferguson, D.G. Schatz, J. SekigDefective DNA repair and increased genomic instability in Artemdeficient murine cells, J. Exp. Med. 197 (2003) 553–565.

55] S.M. Lewis, Palindromy is eliminated through a structure-sperecombination process in rodent cells, Nucl. Acids Res. 27 (12521–2528.

Artemis deficiency confers a DNA double-strand break repair defect and Artemis phosphorylation status is altered by DNA damage and cell cycle progression

Documents