RIDDLE immunodeficiency syndrome is linked to defects in 53BP1-mediated DNA damage signaling

Grant S. Stewart*†, Tatjana Stankovic*, Philip J. Byrd*, Thomas Wechsler‡, Edward S. Miller*, Aarn Huissoon§, Mark T. Drayson¶, Stephen C. West‡, Stephen J. Elledge†ʈ, and A. Malcolm R. Taylor*

*Cancer Research UK, Institute for Cancer Studies, Birmingham University, Vincent Drive, Edgbaston, Birmingham B15 2TT, United Kingdom; ‡Cancer Research UK, Clare Hall Laboratories, London Research Institute, South Mimms, Hertfordshire EN6 3LD, United Kingdom; §Department of Immunology, Birmingham Heartlands Hospital, Birmingham, B9 5SS, United Kingdom; ¶Division of Immunity and Infection, Birmingham University Medical School, Vincent Drive, Edgbaston, Birmingham B15 2TT, United Kingdom; and ʈHoward Hughes Medical Institute, Department of Genetics, Harvard Partners Center for Genetics and Genomics, Harvard Medical School, 77 Avenue Louis Pasteur, Boston, MA 02115

Contributed by Stephen J. Elledge, September 6, 2007 (sent for review July 6, 2007) Cellular DNA double-strand break-repair pathways have evolved with an alternative constant region (e.g., ␣, ␥, ␧) to generate to protect the integrity of the genome from a continual barrage of different Ig isotypes e.g., IgA, IgG, and IgE (2). potentially detrimental insults. Inherited mutations in genes that During the process of CSR, it has been hypothesized that two control this process result in an inability to properly repair DNA closely positioned single-stranded DNA nicks result in the damage, ultimately leading to developmental defects and also production of a DNA DSB. Although the mechanism of DNA cancer predisposition. Here, we describe a patient with a previ- DSB generation during this process remains unclear, there is a ously undescribed syndrome, which we have termed RIDDLE syn- requirement for the presence of a DNA DSB to be produced for drome (radiosensitivity, immunodeficiency, dysmorphic features the proper resolution of the switch recombination process, as and learning difficulties), whose cells lack an ability to recruit loss of components of the DSBR repair pathway, such as H2AX, 53BP1 to sites of DNA double-strand breaks. As a consequence, DNA-PKcs, ATM, and Nbs1, may affect the ability to efficiently cells derived from this patient exhibit a hypersensitivity to ionizing switch from IgM to other Ig isotypes (2). radiation, cell cycle checkpoint abnormalities, and impaired end- More recently it has been demonstrated that loss of the DSB joining in the recombined switch regions. Sequencing of TP53BP1 response protein, 53BP1, in mice also results in a CSR defect (3, and other genes known to regulate ionizing radiation-induced 4). In support of this role, it has been shown that 53BP1 rapidly 53BP1 foci formation in this patient failed to detect any mutations. relocalizes to sites of DNA damage marked by ␥-H2AX after Therefore, these data indicate the existence of a DNA double- exposure of cells to IR, as well as to the Ig heavy-chain (IgH) loci strand break-repair protein that functions upstream of 53BP1 and that have been induced to undergo switch recombination (5). contributes to the normal development of the human immune Here, we describe the analysis of cells derived from a newly system. described human syndrome, termed RIDDLE syndrome whose clinical features are an increased radiosensitivity, immunodefi- DNA repair ͉ cell cycle checkpoint ͉ radiosensitivity ͉ BRCA1 ciency, dysmorphic features, and learning difficulties. Cells from this patient exhibit a different pattern of S␮–S␣ junctions in CSR NA damage recognition and repair is a fundamental process compared with normal and 53BP1-mediated cellular responses Drequired to maintain the fidelity of the genome. Disruption to DNA damage, without any genetic alterations in the TP53BP1 of this process is implicated in the development of human disease gene. The increased sensitivity to IR is likely a result of a failure and tumorigenesis. The impact of defective DNA double-strand to properly relocalize 53BP1, Rif1, and BRCA1 to sites of DNA break repair (DSBR) on human development can be observed in DSBs. In addition, these cells exhibit protracted ATM- patients with inherited disorders such as ataxia-telangiectasia dependent DNA damage signaling, characterized by elevated (A-T), Nijmegen breakage syndrome (NBS), DNA ligase IV numbers of ␥-H2AX, MDC1, and Nbs1 foci, hyperphosphory- deficiency syndrome (LiDS), and radiosensitive severe com- lation of Nbs1, RPA2, and H2AX and a prolonged G2/M bined immunodeficiency (RS-SCID) who have mutations in the checkpoint. These data indicate the existence of a previously DSBR genes ATM, NBS1, DNA Ligase IV, and Artemis and uncharacterized component of the DSB repair pathway that Cernunnos, respectively. These patients manifest a multitude of facilitates 53BP1 recruitment to sites of DNA damage as well as clinical features, which include a marked hypersensitivity to promoting efficient repair of DNA DSBs that occurs after agents that induce DNA double-strand breaks (DSBs), such as exposure to IR or during class switch recombination. ionizing radiation (IR), neurological abnormalities, mental re- tardation, growth defects, gonad dysfunction, and an elevated Results predisposition to the development of tumors. Because normal A clinical description of patient 15–9BI is given in supporting development of the immune system requires the introduction of information (SI) Text. DNA DSBs during antigen receptor gene assembly, defects in the repair of these specialized breaks can lead to profound immunodeficiency (1). This is most notable in LiDS and RS- Author contributions: G.S.S. designed research; G.S.S., T.S., P.J.B., T.W., E.S.M., S.C.W., and S.J.E. performed research; A.H. and M.T.D. contributed new reagents/analytic tools; G.S.S., SCID patients, although immune system abnormalities are also A.H., M.T.D., S.C.W., and S.J.E. analyzed data; and G.S.S. and A.M.R.T. wrote the paper. associated with A-T and NBS. The authors declare no conflict of interest. It is known that Artemis and DNA ligase IV, along with Freely available online through the PNAS open access option. XRCC4, Cernunnos/XLF, and the DNA-PK holoenzyme, com- Abbreviations: DSBs, DNA double strand breaks; DSBR, DSB repair; CSR, class switch prise the components of the nonhomologous DNA end-joining recombination; IR, ionizing radiation; IRIF, IR-induced foci; MEF, mouse embryo fibroblast. (NHEJ) machinery. The error-prone nature of this DNA repair †To whom correspondence may be addressed. E-mail: [email protected] or pathway is invaluable for creating genetic diversity when assem- [email protected]. bling immune system gene receptors. For example, class switch This article contains supporting information online at www.pnas.org/cgi/content/full/ recombination (CSR) results in the editing of a functional ␮-Ig 0708408104/DC1. heavy-chain gene, so that the ␮-constant region (C␮) is replaced © 2007 by The National Academy of Sciences of the USA

16910–16915 ͉ PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708408104 Downloaded by guest on September 26, 2021 Cells Derived from Patient 15–9BI Show Evidence of Normal SHM but Altered Patterns of S␮–S␣ Junctions in CSR. There was no evidence for a defect in CSR affecting all isotypes in patient 15–9BI. Nevertheless, the absence of any gross abnormalities in B and T cell number suggested that the low serum levels of IgG observed in patient 15–9BI could have been caused by a defect in a component of class switch recombination. To address this pos- sibility, genomic DNA was extracted from a peripheral blood sample of a healthy donor and patient 15–9BI and the S␮–S␣ and S␮–S␥ switch junctions amplified. In normal donors, CSR is characterized by generation of a range of IgA or IgG fragments, which exhibit little or no sequence homology across the switch junctions between S␮ and S␣ or S␥ sequences respectively. Mutations or insertions around the switch break points are observed. Although the number of fragments obtained from 10 PCRs run in parallel were not different between cells from the healthy donor and patient 15–9BI, sequencing across the switch junctions revealed that cells derived from the patient utilized significantly increased levels of microhomology across the break region compared with the control; 94% (15 of 16) of S␮–S␣ switch products amplified Fig. 1. Skin fibroblasts derived from patient 15–9BI are radiosensitive, as measured by colony formation assay. Normal, classical A-T and 15–9BI primary from patient 15–9BI exhibited microhomology of 4 bp or over fibroblasts were plated at low density and irradiated with the doses indicated, ␹2 Ͻ compared with only 17% (3 of 18) in the control; , P 0.001 and the number of surviving colonies were counted. (SI Table 1). A reduced frequency of insertions and mutations around the S␮–S␣ break points was observed in the patient 15–9BI (1 of 16) genetic defect in patient 15–9BI may lie within a previously compared with the control (11 of 18); ␹2, P ϭ 0.001. Analysis of uncharacterized DNA DSBR gene. a restricted number of S␮–S␥3 15–9BI switches revealed a defect similar in the patient to that observed for the S␮–S␣ switch Abnormal Relocalization of DSBR Proteins in RIDDLE Syndrome Cells sequences (data not shown). After IR Exposure. One important aspect of the cellular response Taken together, our analysis of the patient’s switch recombi- to DNA damage is the ability of the repair and checkpoint nation fragments suggested the presence of a defect similar, proteins to efficiently relocalize to sites of damage; this has been although less severe, to that observed in patients with hypomor- clearly demonstrated in mice that lack histone variant, H2AX, phic DNA ligase IV mutations (6). which fail to properly relocalize 53BP1, BRCA1, and Nbs1 to To determine whether the resolution of other developmen- repair foci, subsequently resulting in increased genomic insta- tally programmed DNA DSBs was also affected in the patient, bility and a hypersensitivity to agents that induce DNA DSBs (7, we addressed the status of somatic hypermutation. However, 8). To assess whether cells from patient 15–9BI exhibited any unlike the abnormal repair of switch junctions observed, we abnormalities relocalizing DNA DSBR proteins to sites of could not detect any gross differences in the profile of somatic damage, immunofluorescence was used to detect protein redis- hypermutation between controls and the patient (SI Fig. 5). tribution over time after exposure to IR. Unexpectedly, fibro- blasts from patient 15–9BI formed significantly increased num- Cells Derived from Patient 15–9BI Exhibit a Moderately Increased bers of MDC1 (Fig. 2), ␥-H2AX, Nbs1, and Rad51 (SI Fig. 7 A Hypersensitivity to IR. Given that many proteins involved in CSR are and B and SI Fig. 8) but not Fancd2 foci (data not shown) when also involved in repair of DNA DSBs that arise as a consequence compared with a control fibroblast cell line. These foci persisted of exposure to various DNA-damaging agents such as IR, we at 24 h after irradiation, when the majority of lesions in the examined the radiosensitivity of cells derived from the patient. Cells control fibroblast cell line had been repaired, as judged by from this individual exhibited a moderate but reproducible in- ␥-H2AX staining. creased hypersensitivity to IR as measured by both colony forma- Quite surprisingly, however, in stark contrast to the other tion and chromosomal radiosensitivity (SI Table 2 and Fig. 1). This DNA DSBR proteins analyzed, fibroblasts from patient 15–9BI indicated that the observed CSR defect exhibited by 15–9BI was completely failed to form any 53BP1 foci, even at late times after likely to be caused by an underlying inability to properly repair IR exposure (Fig. 3A). It has been reported that at early times DNA DSBs. However, no gross defect in NHEJ, as measured by (15–30 min) after irradiation, 53BP1 does not require H2AX to plasmid religation, could be detected in his cells (SI Fig. 6A), form DNA damage-induced foci but H2AX is required to indicating that the genetic abnormality in this patient does not stabilize 53BP1 once it has loaded on to damaged chromatin (9). reside in a core component of the end-joining machinery. A lack of 53BP1 IRIF at 15, 30, and 60 min after irradiation (Fig. 3A), indicates that the defect in 15–9BI is a global inability of Patient 15–9BI Cells Do Not Show Alterations in DNA DSB Repair 53BP1 to relocalize to damaged chromatin rather than being an Genes. To investigate the genetic defect resulting in the hyper- abnormality in a pathway dependent on H2AX-mediated sensitivity to IR, extracts of cells derived from patient 15–9BI recruitment. were immunoblotted to assess the expression levels of various To confirm this defect, we analyzed the ability of these cells proteins known to be involved in the cellular response to DNA to form Rif1 foci, a recently characterized checkpoint protein DSBs. None of the proteins analyzed in 15–9BI cells exhibited that requires both ATM and 53BP1 for its ability to form IRIF any alterations in expression level or size (SI Fig. 6B). Further- (10). Consistent with the 53BP1 relocalization defect in 15–9BI more, the ATM, hMRE11, NBS1, hRAD50, DNA ligase IV, cells, these cells also failed to form Rif1 IRIF (SI Fig. 9). XRCC4, Cernunnos/XLF, MCPH1 (BRIT1), and H2AX genes In addition to Rif1, 53BP1 has been reported to regulate

were sequenced by using genomic DNA from patient 15–9BI BRCA1 relocalization after the induction of DNA DSBs (11). CELL BIOLOGY cells and found to be wild type. These results indicate that the Therefore, we assessed whether the absence of 53BP1 IRIF

Stewart et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16911 Downloaded by guest on September 26, 2021 Fig. 2. The 15–9BI cells show enhanced MDC1 recruitment to DNA breaks after irradiation. Control and 15–9BI fibroblasts were mock-irradiated or exposed to 3 Gy of IR and fixed at the times indicated after irradiation. Cells were stained with an anti-MDC1 antibody and DAPI to visualize the nuclear DNA. The histogram indicates the number of cells with MDC1 foci at 24 h after irradiation. A minimum of 300 cells was counted, and this was repeated at least three times.

observed in cells from patient 15–9BI would also affect BRCA1 recruitment. Interestingly, although BRCA1 formed elevated numbers of IRIF at later times after IR exposure in a similar way Fig. 3. The 15–9BI cells fail to form 53BP1 IRIF after ␥-irradiation. (A) Control to ␥-H2AX and MDC1, 15–9BI cells exhibited defective BRCA1 and 15–9BI fibroblasts were mock-irradiated or exposed to 3 Gy of IR and fixed at the times indicated after irradiation. Cells were stained with an anti-53BP1 foci formation at the 30-min and 1-h time points (SI Fig. 10). antibody and DAPI to visualize the nuclear DNA. (B) Exogenous MDC1 and These data suggest that the underlying genetic defect in this 53BP1 do not correct the defective recruitment of 53BP1 to DNA breaks in cells patient results in a complete failure of his cells to properly target from patient 15–9BI. Control and 15–9BI fibroblasts were transfected with 53BP1 and 53BP1-dependent repair/checkpoint proteins, e.g., GFP-hMDC1 or GFP-m53BP1, irradiated 48 h after transfection with 5 Gy, and Rif1, to sites of DNA damage. In addition, it appears that this allowed to recover for 4 h. The cells were fixed and stained with either an ␥ defect also has a moderate affect on BRCA1 radiation-induced anti-53BP1 antibody to detect endogenous 53BP1 or an anti- -H2AX anti- body. DAPI was used to visualize the nuclear DNA. (C) The 15–9BI cells exhibit relocalization at early times during DSBR. normal levels of histone H4 dimethylation on lysine-20. Whole-cell extract was prepared from a normal or 15–9BI LCL and separated by SDS/PAGE. Western Reconstituting 15–9BI Cells with 53BP1 or MDC1 Does Not Correct the blots were carried out by using antibodies that recognized pan-histone H4 and IRIF Defect. In an attempt to further identify the genetic defect in histone H4 dimethylated on lysine-20. RIDDLE syndrome cells that results in the inability of these cells to form 53BP1 IRIF, the TP53BP1 gene was completely se- quenced at the genomic level, but no mutations were identified. foci after the induction of DSBs did not result from mutation of This analysis included sufficient amounts of intronic sequence the TP53BP1 gene. flanking each exon to exclude possible noncoding alterations It has been demonstrated that MDC1 is required for the that may affect splicing. To confirm the absence of any detect- efficient recruitment of 53BP1 to sites of DNA DSBs (12). The able mutations in the TP53BP1 gene, full-length TP53BP1 fused MDC1 gene was also sequenced at the genomic level from to GFP was transfected into control and 15–9BI cells. After IR 15–9BI cells and was also found to be wild type. Heterozygous exposure, the control cells formed robust GFP-m53BP1 con- polymorphic variants were identified in both the MDC1 and taining foci that colocalized with ␥-H2AX, whereas 15–9BI cells TP53BP1 genes, indicating that both alleles from patient 15–9BI remained unable to form GFP-m53BP1 IRIF (Fig. 3B). These were represented equally during the analysis. Again, control and data indicated that the inability of 15–9BI cells to form 53BP1 15–9BI cells were transfected with a full-length hMDC1-GFP

16912 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708408104 Stewart et al. Downloaded by guest on September 26, 2021 Fig. 4. After irradiation, cells from patient 15–9BI exhibit abnormal cell cycle checkpoint response and hyperactive ATM-dependent DNA damage signaling. (A) Control and 15–9BI fibroblasts were mock-treated or irradiated with 3 Gy of ␥-irradiation and then fixed at the times indicated. Cells were labeled with an anti-phospho histone H3 serine-10 antibody and propidium iodide and then subjected to FACS analysis. (B) Control and 15–9BI fibroblasts were labeled with 14[C]thymidine for 24 h, irradiated with the doses indicated, and then chased with 3[H]thymidine for 1 h. Cells were fixed and the 3[H]/14[C] calculated by using a scintillation counter. (C) Control and 15–9BI fibroblasts were mock-irradiated or exposed to 3 Gy of IR and fixed at the times indicated after irradiation. Whole-cell extract was prepared and subjected to SDS/PAGE/Western blot analysis using the antibodies indicated.

expression construct and the ability of this to correct endogenous Cells were fixed and stained with an antibody directed against 53BP1 foci formation after exposure to IR was assessed. Both phosphorylated serine-10 of histone H3, a well documented control and 15–9BI cells formed GFP-hMDC1 IRIF. However, marker of mitotic cells, and subjected to flow cytometric anal- this was unable to complement the 53BP1 IRIF defect observed ysis (16). in 15–9BI cells (Fig. 3B), thereby ruling out MDC1 as the A clear decrease in the percentage of mitotic cells was responsible gene. observed in 15–9BI cells within the first hour after irradiation, More recently, it has been shown that the tudor domain of which is indistinguishable from normal, indicating that these 53BP1 specifically binds dimethylated lysine-20 (K-20) on his- cells can clearly activate the G2/M checkpoint normally. Analysis tone H4 and that this interaction was responsible for the at later times after irradiation, however, revealed an inability of relocalization of 53BP1 to sites of damage (13). Therefore, it is 15–9BI cells to recover properly from the checkpoint, as indi- conceivable that reduced or absent histone H4 K-20 dimethy- cated by the lower percentage of cells entering mitosis at 8 and lation in 15–9BI cells could account for the lack of 53BP1 foci 24 h after irradiation (Fig. 4A). induced by IR exposure. Radioresistant DNA synthesis (RDS) is a characteristic phe- However, as shown in Fig. 3C, 15–9BI cells did not exhibit any notype exhibited by cells that cannot appropriately regulate the alterations in the level of histone H4 K-20 dimethylation, intraS phase checkpoint in the presence of DNA damage in- indicating that it was not loss of this histone modification duced by IR (17). To ascertain whether cells derived from resulting in the inability of 15–9BI cells to form 53BP1 IRIF. patient 15–9BI also displayed this trait, control and 15–9BI cells These data point to the existence of an additional factor, other were irradiated with 5 or 10 Gy of IR, incubated for 1 h, and then than MDC1, 53BP1, and PR-Set7, the histone H4 K-20 meth- DNA synthesis was determined by thymidine incorporation. In yltransferse, that can regulate the damage-induced accumulation a similar fashion to the 53BP1 knockout MEFs (15), 15–9BI cells of 53BP1 at sites of DNA breaks. exhibited a mild intraS phase checkpoint defect but only when irradiated with a high dose of IR (10 Gy) (Fig. 4B). This indicated Patient 15–9BI Cells Exhibit Aberrant DNA Damage-Induced Cell Cycle that the gene defective in this patient does not contribute Checkpoint Activation. It has been demonstrated by using siRNA significantly to the regulation of the DNA damage-activated knockdown, that 53BP1 functionally regulates both the intraS- intraS-phase checkpoint. and G2/M-phase DNA damage-activated cell cycle checkpoints after low-dose irradiation (11, 14). However, mouse embryo RIDDLE Syndrome Cells Show Hyperactivation of ATM-Dependent fibroblasts (MEFs) derived from the 53BP1 knockout mouse Signaling Pathways. It is known that ATM and ATM-dependent exhibited no dramatic abnormalities in regulation of either signaling pathways are primarily responsible for controlling the damage-activated checkpoints (15), although these discrepancies activation and maintenance of DNA damage-induced cell cycle may, in part, be attributed to differences in irradiation dose and checkpoints after exposure to IR (18). To investigate the status cell types used. of these pathways in 15–9BI cells with regard to the abnormal To determine whether 15–9BI cells exhibited any defects in checkpoint responses observed, a time course after exposure to the ability to activate the G2/M checkpoint, fibroblasts were 2 Gy of IR was carried out over a period of 24 h. Interestingly, CELL BIOLOGY irradiated with 3 Gy of IR and harvested over a period of 24 h. an antibody to phosphorylated serine-1981 of ATM, which is

Stewart et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16913 Downloaded by guest on September 26, 2021 believed to represent a marker of ATM activation (19), showed However it is unclear whether a lack of the histone H4- an increased signal in 15–9BI cells compared with a control cell dependent pathway completely abolishes all 53BP1 foci forma- line at the same time point, and was sustained over a 24-h time tion or whether at later times during DNA DSBR, the H2AX- period. The hyperresponsive activation of ATM was mirrored by dependent pathway is sufficient to take over. the phosphorylation status of various downstream ATM- Interestingly, the genetic abnormality in 15–9BI cells also dependent targets (Fig. 4C). In contrast however, the relocal- appears to affect BRCA1 foci formation albeit to a lesser extent ization of proteins recognized by the phospho-ATM serine-1981 than 53BP1. Consistent with these observations, 53BP1 has been antibody to IRIF (SI Fig. 11) was impaired in 15–9BI cells shown to regulate BRCA1 foci formation after IR exposure (11), compared with control cells. This observation suggested that the although this has not been assessed in any of the mouse 53BP1 gene product, which is defective in patient 15–9BI, contributes knockout models. Furthermore, from the BRCA1 foci kinetics to the regulation of ATM recruitment to DNA DSBs but not its in 15–9BI cells, it would appear that like 53BP1, BRCA1, also has activation, although it must be taken into account that phospho- different determinants for its relocalization at early versus late dependent antibodies, when used for immunofluorescence, are times after irradiation. likely to detect multiple epitopes in a number of different The checkpoint defects in cells derived from patient 15–9BI phospho-proteins. Interestingly however, this defect in the re- are subtle. Taken at face value, this implies that the mutant gene cruitment of phosphorylated ATM was not observed in the product does not play a significant role in the activation of DNA 53BP1 knockout MEFs, rather the recruitment was elevated damage-induced cell cycle checkpoints, although we cannot rule over and above that seen in the wild-type parental MEFs (SI Fig. out the possibility that this patient carries a hypomorphic mutant 12), which mirrored the ␥-H2AX/MDC1 response in these cells allele that allows activation of the cell cycle checkpoint after (SI Fig. 13 A and B). DNA damage to near normal levels. The sustained G2/M arrest In contrast to the other markers of DNA damage-induced observed in these cells may be accounted for by the hyperacti- signaling assessed in 15–9BI cells, the activation of the check- vation of ATM-dependent signaling and/or the persistence of point kinase Chk1 (20), as measured by phosphorylation on unrepaired DNA breaks. However, the possibility that this may serine-345, was not as efficient as that observed in the control also arise as a consequence of an inability to inhibit the check- cell line (Fig. 4C). This may, in part, account for the observed point signaling response even though the DNA damage has been mild intraS-phase checkpoint defect. repaired cannot be discounted. The mild RDS phenotype is Collectively, these data demonstrate that the factor required suggestive of a checkpoint defect, and this is consistent with the for 53BP1 relocalization after IR exposure may also play a role abnormal activation of Chk1 observed in these cells. However, in suppressing many signaling pathways controlled by ATM. the severity of the checkpoint defect observed maybe amelio- rated somewhat by the heightened MDC1-dependent DNA Discussion damage response, a pathway that is known to modulate Chk1 Here, we describe a patient with a newly described disorder that activation and the intraS-phase checkpoint (12). we have termed RIDDLE syndrome. This patient exhibits an It is tempting to speculate that cells with an inability to increased radiosensitivity, mild immunodeficiency, dysmorphic relocalize 53BP1 to sites of damage compensate by up-regulating features, and learning difficulties. Cells derived from this patient the MRN/MDC1 pathway. In support of this, it has been primarily fail to properly relocalize the DNA DSBR protein, suggested (21) by using 53BP1-directed siRNA, that this is 53BP1, to sites of DNA damage. The phenotype of cells derived indeed the case. Although it appears that 53BP1 and MDC1 from this patient recapitulated many of those previously docu- differ in their dependency for specific DNA damage-induced mented in the 53BP1 knockout mouse (3, 4, 15) although no histone modifications for their initial recruitment to breaks, mutations could be found in the TP53BP1 gene itself nor in any 53BP1 still requires H2AX phosphorylation to stabilize its additional genes that encode proteins, e.g., MDC1, H2AX, and binding to chromatin surrounding DNA breaks at later times PR-Set7 that are known to be involved in regulating 53BP1 during DNA repair. Therefore, in the absence of 53BP1, chro- relocalization to DNA breaks. This observation supports the matin/DNA breaks normally recognized by 53BP1 will eventu- existence of an as yet unknown protein that functions upstream ally switch, during later stages of repair, to an H2AX- of 53BP1 during the cellular response to DNA DSBs. dependency, thus elevating the cells requirement for the H2AX/ At present, the cause of the IgG deficiency and elevated level MDC1/Nbs1 pathway. of IgM is not known, although the DNA DSBR defect leading Taken together, although the gene mutated in RIDDLE to increased sensitivity to IR may also cause a subtle defect of syndrome has yet to be identified, our initial observations end-joining in class switch recombination. The normal level of characterizing cells derived from this patient serve to highlight somatic hypermutation may suggest that the gene responsible for several points, the most important being that a human disease the altered pattern of switch junctions in this patient lies associated with defects in 53BP1-mediated DNA damage sig- downstream of AID and UNG1 in this process. Moreover, the naling has not before been documented. Furthermore, a link lack of an obvious NHEJ repair defect implies that the deficient between the 53BP1 pathway and immune system development in factor is not part of the core NHEJ machinery. The observation humans has not been previously identified. Second, we demon- that cells from patient 15–9BI do not form any 53BP1 IRIF at strate the existence of an additional unknown biochemical event, early or late stages of DNA DSBR suggests that the defective other than the DNA damage-induced chromatin modification, gene product is required for both the histone dimethylation- which is required for 53BP1 relocalization to sites of DNA breaks mediated and the ␥-H2AX-mediated components of the signal- and which also has an impact on BRCA1 recruitment. Lastly, ing pathway responsible for 53BP1 relocalization to DNA breaks. cells with abnormalities in DNA damage-induced 53BP1 cellular Given that the levels of the dimethylated histone H4 do not signaling may partly compensate for this defect by up-regulating change after induction of DNA damage, it has been hypothesized the ␥-H2AX/MDC1/MRN pathway. that radiation-induced changes in the surrounding chromatin is the signal that allows the methylated lysine on histone H4 to Materials and Methods become visible, which can subsequently facilitate 53BP1 binding Cell Lines. Fibroblast cell lines were maintained in DMEM through its tudor domain (13). It is quite possible that the defect supplemented with 10% FCS, glutamine, and penicillin/ in patient 15–9BI resulting in loss of 53BP1 foci formation is due streptomycin (pen/strep). Lymphoblastoid cell lines (LCLs) were to a lack of an ability to remodel the nucleosome surrounding the routinely maintained in RPMI medium 1640 supplemented with DNA break that allows the histone methylation to be exposed. 10% FCS, glutamine, and pen/strep. The 53BP1 ϩ/ϩ and Ϫ/Ϫ

16914 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0708408104 Stewart et al. Downloaded by guest on September 26, 2021 (MEFs) were obtained from Junjie Chen and Irene Ward (Yale incubated for1hatroom temperature with the primary antibody University School of Medicine, New Haven, CT) and were also diluted in 2% FCS, washed three times in PBS, and then maintained in DMEM supplemented with 10% FCS, glutamine, incubated for an additional1hwiththesecondary antibody. and pen/strep. All experiments were carried out by using hTert- Slides were washed again three times in PBS and then sealed with immortalized skin fibroblasts derived from patient 15–9BI and a coverslip and Vectashield containing DAPI (Vector Labora- a normal donor under the same culture conditions, unless tories, Burlingame, CA). Primary antibodies used are described otherwise stated. Where indicated, phenotypic differences be- above. All secondary antibodies were either Alexa Fluor 488 or tween 15–9BI and normal control fibroblasts were verified in an Alexa Fluor 596 coupled and purchased from Invitrogen (Carls- LCL derived from patient 15–9BI. A variety of normal LCLs bad, CA). derived from laboratory donors were used as controls. Colony Formation and Radioresistant DNA Synthesis. These were Mutation Analysis. Candidate genes were sequenced from cDNA carried out as described (12). or genomic DNA isolated from patient 15–9BI cells. The TP53BP1 and MDC1 genes were sequenced at the genomic level Nonhomologous End-Joining Assay. Whole-cell extract was pre- and compared with reference sequences (AC018924.9 and pared from 6 liters of LCLs, and the assay was carried out as AB088099.1, respectively) deposited in the National Center for described (23). Biotechnology Information database. Transfections. Fibroblasts were transfected by using the Amaxa Immunoblot Analysis. Cell extracts were prepared as described nucleofector (Amaxa, Gaithersburg, MD) according to the (12). The generation of antibodies to MDC1 and ATM has been manufacturer’s instructions with 3 ␮g of plasmid DNA. Cells described (12, 22). The antibodies were used according to were seeded on to slides 24 h after transfection and harvested manufacturer’s instructions; anti-phospho-H2AX, phospho- after an additional 24 h. ATM, di-methyl histone H4 lysine-20 and H2A antibodies (Upstate Biotechnology, Lake Placid, NY), anti-BRCA1, Ku70, We thank Drs. Claudia Lukas (Institute of Cancer Biology and Centre and Chk1 antibodies (Santa Cruz Biotechnology, Santa Cruz, for Genotoxic Stress Research, Copenhagen, Denmark) and Yasuhisa CA), anti-53BP1, DNA-PK, and RPA1 antibodies (Merck, Not- Adachi (University of Edinburgh, Edinburgh, U.K.) for providing the tingham, U.K.), anti-phospho-53BP1, phospho-Nbs1, phospho- m53BP1-GFP expression construct, Drs. Phang-Lang Chen (University SQ/TQ, phospho-Chk1 and histone H4 antibodies (Cell Signal- of Texas, San Antonio, TX) and Matthew Weitzman (Salk Institute for ing Technology, Beverly, MA), anti-RPA2, Rad51, Rad50, Biological Studies, La Jolla, CA) for hMDC1-GFP expression con- Mre11, and Nbs1 antibodies (Novus, Littleton, CO), anti- structs, Dr. Elizabeth Blackburn (University of California, San Fran- phospho-SMC1 antibody (Bethyl, Montgomery, TX). Antibod- cisco, CA) for providing the Rif1 antibody, Drs. Irene Ward and Junjie Chen for the 53BP1 knockout MEFs, Dr. Zhenkun Lou (Mayo Clinic ies to XRCC4 and DNA ligase IV were a gift from Penny Jeggo College of Medicine, Rochester, MN) for providing the anti-mouse (University of Sussex, Brighton, U.K.). MDC1 antibody, and Dr. Jean-Pierre de Villartay’s laboratory for sequencing the Cernunnos gene in patient 15–9BI before it was pub- Immunofluorescence. Cells seeded on to glass slides were perme- lished. This work was supported by Cancer Research UK (G.S.S., P.J.B., abilized and then fixed in extraction buffer (12). Slides were and A.M.R.T.) and the Leukaemia Research Fund (T.S.).

1. Gennery AR (2006) Br Med Bull 77–78:71–85. 11. Wang B, Matsuoka S, Carpenter PB, Elledge SJ (2002) Science 298:1435–1438. 2. Chaudhuri J, Alt FW (2004) Nat Rev Immunol 4:541–552. 12. Stewart GS, Wang B, Bignell CR, Taylor AMR, Elledge SJ (2003) Nature 3. Manis JP, Morales JC, Xia Z, Kutok JL, Alt FW, Carpenter PB (2004) Nat 421:961–966. Immunol 5:481–487. 13. Botuyan MV, Lee J, Ward IM, Kim JE, Thompson JR, Chen J, Mer G (2006) 4. Ward IM, Reina-San-Martin B, Olaru A, Minn K, Tamada K, Lau JS, Cascalho Cell 127:1361–1373. M, Chen L, Nussenzweig A, Livak F, et al. (2004) J Cell Biol 165:459–464. 14. DiTullio RA, Jr, Mochan TA, Venere M, Bartkova J, Sehested M, Bartek J, 5. Petersen S, Casellas R, Reina-San-Martin B, Chen HT, Difilippantonio MJ, Halazonetis TD (2000) Nat Cell Biol 4:998–1002. Wilson PC, Hanitsch L, Celeste A, Muramatsu M, Pilch DR, et al. (2001) Nature 15. Ward IM, Minn K, van Deursen J, Chen J (2003) Mol Cell Biol 23:2556– 414:660–665. 2563. 6. Pan-Hammarstrom Q, Jones AM, Lahdesmaki A, Zhou W, Gatti RA, Ham- 16. Xu B, Kim ST, Lim DS, Kastan MB (2002) Mol Cell Biol 22:1049–1059. marstrom L, Gennery AR, Ehrenstein MR (2005) J Exp Med 201:189–194. 17. Bartek J, Lukas C, Lukas J (2004) Nat Rev Mol Cell Biol 5:792–804. 7. Bassing CH, Chua KF, Sekiguchi J, Suh H, Whitlow SR, Fleming JC, Monroe 18. Kurz EU, Lees-Miller SP (2004) DNA Repair (Amsterdam) 3:889–900. BC, Ciccone DN, Yan C, Vlasakova K, et al. (2002) Proc Natl Acad Sci USA 99:8173–8178. 19. Bakkenist CJ, Kastan MB (2003) Nature 421:499–506. 8. Celeste A, Petersen S, Romanienko PJ, Fernandez-Capetillo O, Chen HT, 20. Zhao H, Watkins JL, Piwnica-Worms H (2002) Proc Natl Acad Sci USA Sedelnikova OA, Reina-San-Martin B, Coppola V, Meffre E, Difilippantonio 99:14795–14800. MJ, et al. (2002) Science 296:922–927. 21. Mochan TA, Venere M, DiTullio RA, Jr, Halazonetis TD (2003) Cancer Res 9. Celeste A, Fernandez-Capetillo O, Kruhlak MJ, Pilch DR, Staudt DW, Lee A, 63:8586–8591. Bonner RF, Bonner WM, Nussenzweig A (2003) Nat Cell Biol 5:675–679. 22. Starczynski J, Simmons W, Flavell JR, Byrd PJ, Stewart GS, Kullar HS, Groom 10. Silverman J, Takai H, Buonomo SB, Eisenhaber F, de Lange T (2004) Genes A, Crocker J, Moss PA, Reynolds GM, et al. (2003) Am J Pathol 163:423–432. Dev 18:2108–2119. 23. Baumann P, West SC (1998) Proc Natl Acad Sci USA 95:14066–14070. CELL BIOLOGY

Stewart et al. PNAS ͉ October 23, 2007 ͉ vol. 104 ͉ no. 43 ͉ 16915 Downloaded by guest on September 26, 2021