Genomic Instability, Defective Spermatogenesis, Immunodeficiency, and Cancer in a Mouse Model of the RIDDLE Syndrome

Eukaryotic cells have evolved to use complex pathways for DNA damage signaling and repair to maintain genomic integrity. RNF168 is a novel E3 ligase that functions downstream of ATM,γ-H2A.X, MDC1, and RNF8. It has been shown to ubiquitylate histone H2A and to facilitate the recruitment of other DNA damage response proteins, including 53BP1, to sites of DNA break. In addition, RNF168 mutations have been causally linked to the human RIDDLE syndrome. In this study, we report that Rnf168−/− mice are immunodeficient and exhibit increased radiosensitivity. Rnf168−/− males suffer from impaired spermatogenesis in an age-dependent manner. Interestingly, in contrast to H2a.x−/−, Mdc1−/−, and Rnf8−/− cells, transient recruitment of 53bp1 to DNA double-strand breaks was abolished in Rnf168−/− cells. Remarkably, similar to 53bp1 inactivation, but different from H2a.x deficiency, inactivation of Rnf168 impairs long-range V(D)J recombination in thymocytes and results in long insertions at the class-switch junctions of B-cells. Loss of Rnf168 increases genomic instability and synergizes with p53 inactivation in promoting tumorigenesis. Our data reveal the important physiological functions of Rnf168 and support its role in both γ-H2a.x-Mdc1-Rnf8-dependent and -independent signaling pathways of DNA double-strand breaks. These results highlight a central role for RNF168 in the hierarchical network of DNA break signaling that maintains genomic integrity and suppresses cancer development in mammals.


Introduction
DNA damage checkpoint signaling and DNA repair pathways are key elements of the DNA damage response (DDR) and are critical for the maintenance of genomic integrity [1][2][3]. Mammalian cells constantly experience DNA damage as a result of exogenous exposure to ionizing radiation (IR), ultraviolet light (UV), chemical compounds, and radical oxygen species, as well as endogenous insults due to DNA replication errors. In addition, double-strand DNA breaks (DSBs) are also programmed to occur during immune-receptor rearrangements and meiosis.
Mutations of genes involved in DNA damage signaling or repair can lead to many diseases including neurodegenerative diseases, immunodeficiency and cancer, underlining the importance of these processes [1,4,5]. Among the various types of DNA damage, DSBs are the most serious and require elaborated networks of proteins to signal and repair the damage. In mammalian cells, DSBs are initially recognized by the Mre11/Rad50/NBS1 (MRN) complex that induces the activation and recruitment of the ataxia-telangiectasia-mutated (ATM) kinase to the break sites [6][7][8]. At the flanking sites of DSBs, H2A.X, a variant of the histone H2A, is rapidly phosphorylated at the serine 139 residue (c-H2A.X). Phosphorylation of H2A.X is mediated by activated ATM, which itself is phosphorylated at serine 1981 (phospho-ATM-S1981), or alternatively by two other phosphoinositide 3-kinase like kinases (PIKKs), namely the ataxia telangiectasia and Rad3 related (ATR) and the DNA-dependent protein kinase catalytic subunit (DNA-PKcs). Active ATM also phosphorylates a number of other proteins including the structural maintenance of chromosomes 1 (SMC1), the Nijmegen breakage syndrome protein 1 (NBS1), the checkpoint kinase 2 (Chk2), the breast cancer 1-early onset (BRCA1) and the mediator of DNA damage checkpoint protein-1 (MDC1). Subsequent to its phosphorylation, MDC1 binds to c-H2A.X via its tandem BRCA1 C-terminal (BRCT) domains, and recruits additional active ATM to sites flanking the DSBs [7,9,10]. MDC1 also associates with MRN complex through its Ser-Asp-Thr (SDT) repeats and the fork-head associated (FHA) domain of NBS1. Furthermore, through its C-terminus, NBS1 also interacts with ATM [11][12][13][14][15]. As such, MDC1 bridges the interaction of MRN to c-H2A.X and ATM. Enrichment of ATM-MDC1 and ATM-MRN at the break sites further amplify the phosphorylation of H2A.X and triggers the recruitment of other DNA damage response proteins to the DSB flanking regions [16]. The three conserved T-Q-X-F clusters between 698 to 800 amino acids of MDC1 are also phosphorylated by ATM. The threoninephosphorylated MDC1 has been shown to interact with the FHA domain of RING finger protein 8 (RNF8), thus recruiting this E3 ligase to the sites of damage. Subsequently, RNF8, along with the E2 ubiquitin-conjugating enzyme UBC13, mediate lysine 63 (K63)-linked ubiquitylation of the histone H2A at the flanking regions of DSBs [17][18][19]. Such ubiquitylated form of H2A interacts with the MIU2 (motif interacting with ubiquitin 2) domain of RNF168 and recruits this E3 ligase to DSB sites, allowing it to further ubiquitylate surrounding H2A [20][21][22]. This likely amplifies the modification of chromatin structure at regions adjacent to DSBs, and facilitates the recruitment of tumor suppressor p53 binding protein 1 (53BP1). In addition, RAP80 selectively binds to the K63-polyubiquitin chains on H2A via its tandem ubiquitin interaction motifs (UIMs) [23], and acts as a bridge to recruit BRCA1 to the regions of DSBs. In sum, DSBs signaling is a highly regulated process, in which RNF168 plays a major role through its contribution to the recruitment of various downstream DDR proteins.
RNF168 is mutated in RIDDLE syndrome, which is characterized by radiosensitivity, immunodeficiency, dysmorphic features and learning difficulties [24]. RNF168 contains a RING finger domain and two MIU domains. The RING finger domain of RNF168 is critical for its ubiquitin E3 ligase activity, whereas its MIU2 domain mediates its binding to ubiquitylated H2A [20][21][22]. Knockdown of RNF168 in human cells significantly impaired the formation of 53BP1, RAP80 and BRCA1 ionizing radiation induced foci (IRIF). While recent studies have revealed a role for RNF168 in ubiquitylating H2A during DSB signaling [20][21][22], the full physiological functions of RNF168 are not understood.
Here, we report the generation of a mouse model for RNF168 deficiency. Rnf168 2/2 mice are viable. Consistent with the clinical features of RIDDLE syndrome, Rnf168 2/2 mice are immunodeficient, and their cells show increased radiosensitivity. Similar to 53bp1 deficiency, long-range V(D)J recombination of T-cell receptor (TCR)d is also impaired in Rnf168 2/2 thymocytes, while aberrant long insertions have been observed at class switch junctions of Rnf168 2/2 B-lymphocytes. Moreover, in contrast to the transient recruitment of 53bp1 to DSBs in H2a.x-, Mdc1-and Rnf8-deficient mouse embryonic fibroblasts (MEFs), 53bp1 recruitment to DSB sites is completely abolished in Rnf168-deficient MEFs. Our data also demonstrate novel roles of Rnf168 in spermatogenesis, maintenance of genomic integrity and cancer, and therefore, further highlight the other important physiological functions of this molecule.

Rnf168 is dispensable for embryonic and postnatal development
To investigate the physiological role of Rnf168, we generated Rnf168 deficient mice using two different gene trap embryonic stem cell (ES) clones (156B6 and 405F11). Gene trap of Rnf168 in the two ES clones was confirmed by Southern blot, PCR and sequencing of the genomic site of the retrovirus integration ( Figure S1A and S1B). RT-PCR revealed the disruption of full length Rnf168 transcript in the two Rnf168 gene trap ES clones ( Figure S1C). The loss of Rnf168 protein was also confirmed in Rnf168 2/2 MEFs and splenocytes by immunoblotting ( Figure S1D and S1E).
Intercrossing of Rnf168 +/2 mice obtained with 156B6 and 405F11 ES clones indicated that Rnf168 2/2 mice were viable and born at the expected Mendelian ratio (Table S1). Rnf168 2/2 mice did not exhibit any gross developmental defects and were indistinguishable from their wildtype (WT) littermates. Collectively, these data indicate that the E3 ligase Rnf168 is not required for embryonic and postnatal development.

Rnf168 deficiency leads to increased radiation sensitivity
Mutation of the RNF168 gene is associated with RIDDLE syndrome [22,24]. To determine whether Rnf168 deficiency in mice confers increased sensitivity to DNA damage, primary lymphocytes isolated from peripheral lymph nodes (LN) of Rnf168 2/2 mice and WT littermates were subjected to various doses of IR or UV. Lymphocyte viability was examined using 7aminoactinomycin D (7AAD) and flow cytometry. These studies demonstrate a significant increase in the sensitivity of Rnf168 2/2 lymphocytes to IR and UV treatment ( Figure 1A).
DNA damage signaling plays essential roles in the activation of checkpoints and cell cycle progression [25]. Examination of the effect of Rnf168 inactivation on cell cycle progression using G1/S synchronized MEFs showed no difference between Rnf168 2/2 and WT MEFs ( Figure S2A).

Author Summary
The repair of DNA damage is fundamental as illustrated by the many human syndromes, immunodeficiencies, and cancers associated with defects in DNA damage signaling and repair. RIDDLE syndrome, caused by mutations of the human RNF168, is a novel hereditary disease clinically characterized by radiosensitivity, immunodeficiency, dysmorphic features, and learning difficulties. RNF168 is an E3 ligase that modifies histones and chromatin structure at sites of DNA breaks. In this study, we show that Rnf168 deficiency in mice leads to increased radiosensitivity, immunodeficiency, and defective spermatogenesis. Additionally, dual inactivation of Rnf168 and p53 leads to increased cancer risk. Collectively these data demonstrate important and broad physiological functions for Rnf168.
Rnf168 promotes the recruitment of 53bp1 and Brca1 to DNA breaks Through ubiquitylation of histone H2A, RNF168 plays an important role in the recruitment of 53BP1 to the sites of DNA breaks downstream of H2A.X, MDC1 and RNF8 [20][21][22]. Although H2a.x, Mdc1 and Rnf8 are each essential for the stable 'retention' of 53bp1 foci at DSB sites, studies of H2a.x 2/2 , Mdc1 2/2 and Rnf8 2/2 MEFs demonstrate that these cells can still transiently recruit 53bp1 to form IRIF [29]. In addition, a partial accumulation of 53bp1 at DSBs has been also observed in activated Rnf8 2/2 Bcells post-irradiation [30]. To investigate the effect of Rnf168 deficiency on the recruitment of 53bp1 to sites of DSBs, Rnf168 2/2 , Rnf8 2/2 and WT MEFs were subjected to 5 Gy of IR, followed by a time course analysis of 53bp1 foci formation by immunofluorescence microscopy. In contrast to H2a.x 2/2 , Mdc1 2/2 and Rnf8 2/2 MEFs [5], no 53bp1 IRIF were observed in Rnf168 2/2 MEFs post-IR (Figure 2A and 2B and Figure S3A). Transfection of Rnf168 2/2 MEFs with GFP-tagged RNF168 expression constructs fully restored formation of 53bp1 IRIF ( Figure 2C). In addition, radiosensitivity of Rnf168 2/2 MEFs was also rescued by expression of Rnf168 ( Figure 2D). Rnf168 deficiency also significantly impaired Brca1 recruitment to sites of DSBs ( Figure 2E and Figure S3B). These data indicate an important role for Rnf168 in the recruitment of both 53bp1 and Brca1 to the DNA break sites.

Rnf168 is dispensable for the activation of Atm signaling pathway
To examine whether Rnf168 deficiency affects the activation of Atm pathway, WT and Rnf168 2/2 thymocytes were irradiated ex vivo. One hour post-IR, the protein levels of total Atm, as well as the expression and phosphorylation levels of its downstream targets, were assessed by immunoblot (IB) analysis. The protein levels of Atm were similar between Rnf168 2/2 and WT thymocytes in response to IR ( Figure 3A). Phosphorylation levels of the Atm substrates Smc1 (serine 966), Nbs1 (serine 343), Chk2 (threonine 68) and p53 (serine 15), were similar or slightly higher in Rnf168 2/2 thymocytes compared to WT controls ( Figure 3A and 3B). IR also induced a slightly higher phosphorylation level of Chk2 (threonine 68) and p53 (serine 15) in Rnf168 2/2 MEFs compared to WT controls ( Figure 3C). We also observed increased c-H2a.x and Mdc1 foci formation in untreated and irradiated Rnf168 2/2 MEFs compared to WT controls ( Figure 3D and 3E, Figure S3C and S3D). In addition, c-H2a.xand Mdc1-IRIF remained visible longer in irradiated Rnf168 2/2 MEFs compared to WT controls, suggesting defective DSB repair in the absence of Rnf168. As both Atm and DNA-PK can phosphorylate H2a.x at serine 139 in response to IR [31], we examined the effects of inhibition of DNA-PK on c-H2a.x IRIF in WT and Rnf168 2/2 MEFs. The levels of c-H2a.x IRIF in irradiated WT and Rnf168 2/2 MEFs were not affected by DNA-PK inhibitors ( Figure 3F). Collectively, these data indicate that Rnf168 is dispensable for the activation of ATM-Chk2-p53 pathway and that Atm signaling pathway is unaffected in Rnf168 2/2 cells.

Age-dependent requirement for Rnf168 in spermatogenesis
H2a.x, Mdc1 and Rnf8 are required for DDR upstream to Rnf168, and are essential for male fertility [10,27,30,[32][33][34][35], whereas 53bp1 homozygous mutant males are fertile [28]. We therefore investigated the fertility of Rnf168 2/2 mice. Rnf168 2/2 males and females were fertile at 8 weeks of age, albeit their intercrosses produced smaller litters (6.0 pups60.6) compared to WT littermates (9.8 pups60.7, p,0.05) ( Figure 4A). In contrast, 12-month-old Rnf168 2/2 males showed either a significantly reduced fertility compared to WT littermates or were completely infertile ( Figure 4B). Testicular sizes were comparable between 8-week-old Rnf168 2/2 and WT littermates, whereas at 12 months of age, Rnf168 2/2 males showed reduced testicular size compared to WT littermates ( Figure 4C). Histological examination of testes from 8-week-old Rnf168 2/2 males indicated no abnormalities compared to WT littermates ( Figure 4D). However, testes from aged Rnf168 2/2 mice displayed signs of testicular degeneration and atrophy as evidenced by the reduced number or lack of spermatids in seminiferous tubules, the increased vacuolization of germ cells and the prominent Leydig cells compared to WT littermates ( Figure 4D). The amount of sperm was also drastically decreased in the epididymis of 12-month-old Rnf168 2/2 males compared to controls ( Figure 4E). All together, these data indicate that Rnf168 is required for spermatogenesis in an age-dependent manner.

Inactivation of Rnf168 impairs immunoglobulin class switch recombination and results in immunodeficiency
The proportion of B and T lymphocytes in one RIDDLE syndrome patient was reported to be within the normal range; however, this patient was immunodeficient and exhibited low levels of serum immunoglobulin G (IgG) [24]. To investigate the effect of Rnf168 deficiency on lymphocyte maturation, we first examined bone marrow (BM) from 6-8-week-old Rnf168 2/2 mice and their WT littermates. BM cellularity and the number of Proand Pre-B-cells were not affected by Rnf168 deletion ( Figure S4A). Similarly, examination of the number of splenocytes, LN cells and their subpopulations also showed no significant difference between Rnf168 2/2 mice and WT controls ( Figure S4B and S4C).
The repair of programmed DSBs is essential for class switch recombination (CSR) of immunoglobulins [36,37]. Failure to initiate, signal, or repair these programmed DSBs will lead to defective CSR and immunodeficiency. To assess the role of Rnf168 in Ig heavy chain (Igh) class switching in vivo, we evaluated the concentrations of serum Ig isotypes in Rnf168 2/2 mice and WT littermates. Total serum IgG concentrations were significantly lower in 6-8-week-old Rnf168 2/2 mice and in 9-12-month-old Rnf168 2/2 mice compared to age-matched WT littermates ( Figure 5A). To further examine the in vivo role of Rnf168 in CSR, we determined the portion of IgG1-and IgG3-expressing B-cells in Peyer's patches from Rnf168 2/2 mice and WT littermates. B-cells in the Peyer's patches constantly encounter the gut flora, and, therefore, they exhibit elevated levels of CSR [38]. Peyer's patches from Rnf168 2/2 mice showed decreased representation of IgG1-and IgG3-expressing B-cells compared to WT littermates (p,0.05; Figure 5B and 5C).
We further examined the role Rnf168 plays in CSR for various Igh isotypes using purified splenic B-cells from Rnf168 2/2 mice and WT littermates. When B-cells were activated in culture with LPS6IL4, Rnf168-deficient B-cells displayed a significantly reduced secretion of IgG1, IgG2a and IgG3 compared to WT controls ( Figure 5D). We also stimulated in vitro purified splenic B-cells with anti-CD40 or LPS in combination with IL-4, and examined the levels of CSR to IgG1. To analyze the efficiency of CSR in each cell division, we used FACS analysis to assess the expression of surface IgG1 on CFSE-labeled stimulated B-cells ( Figure 5E and 5F, Figure S4D and S4E). The portion of B-cells expressing surface IgG1 post activation was significantly reduced in Rnf168 2/2 mice (13.261.1%) compared to WT controls (28.360.8%, p,0.05) ( Figure 5E). Examination of the CFSE dilution profiles indicated similar proliferation levels of mature B-cells from WT and Rnf168 2/2 mice ( Figure 5F and Figure S4E). In addition, the efficiency of CSR in each cell division was decreased in activated Rnf168 2/2 B-cells compared to WT B-cells ( Figure 5F). These data indicate that CSR defects in Rnf168 2/2 B-cells were directly due to the loss of Rnf168 function, rather than a result of proliferation defects of the mutant B-cells.
The RING finger domain of RNF168 is critical for its E3 ligase activity, while its MIU2 domain is important for its recruitment to DSB sites [20,22]. To further explore the role of Rnf168 in CSR, we generated deletion or point mutant Rnf168 constructs using the pMSCV retroviral vector. Purified B-cells from WT or Rnf168 2/2 mice were activated with LPS and IL-4 and then retrovirally transduced with the various Rnf168-expression constructs. These (B) Representative micrographs of MEFs stained with anti-53bp1 antibody and DAPI are shown. Rnf168 2/2 , Rnf8 2/2 and WT MEFs were untreated or exposed to 5 Gy of IR and fixed at the indicated time post IR. Three independent experiments were performed. Bars, 20 mm. (C) Recovery of 53bp1 IRIF formation in Rnf168 2/2 MEFs complemented with full length RNF168. Immortalized Rnf168 2/2 MEFs were mock-transfected or transfected with GFP-tagged RNF168 expression vectors and cultured for 24 hours. Cells were fixed at 1 hour post-IR (5 Gy) and processed for immunofluorescence staining with anti-53bp1 antibodies. Three independent experiments were performed. Bars, 20 mm. (D) Clonogenic assay was performed to examine radiosensitivity of Rnf168 2/2 MEFs complemented with exogenous Rnf168 (left panel). Expression level of Rnf168 is shown for the MEFs used for the clonogenic assay (right panel). Data shown is representative of three independent experiments and is presented as the mean 6 SEM. *p,0.05. (E) Representative micrographs of MEFs stained with anti-Brca1 antibody and DAPI. Rnf168 2/2 and WT MEFs were either untreated or exposed to 5 Gy of IR and fixed at the indicated time after IR. Bars, 2 mm. doi:10.1371/journal.pgen.1001381.g002 cells were then examined for functional rescue of the impaired CSR by WT or mutants Rnf168 ( Figure 5G, 5H and Figure S4F). Rnf168 2/2 B-cells, infected with pMSCV expressing either WT Rnf168 or Rnf168 with its MIU1 domain deleted (DMIU1), rescued IgG1 CSR to the level of WT B-cells. However, Rnf168 2/2 B-cells infected with retroviruses expressing Rnf168 with mutated RING finger domain (C21S; cysteine 21 residue to serine), deleted MIU2 domain (DMIU2) or deleted MIU1 and MIU2 domains (DMIU1/ 2), showed no rescue of their CSR defects.
We next examined the switch recombination directly by using a quantitative digestion-circularization PCR (DC-PCR) assay [39]. These analyses demonstrated lower frequency of Sm-Sc1 switch recombination in activated Rnf168 2/2 B-cells compared to WT Bcells ( Figure 5I and Figure S4G). x antibody and DAPI. Rnf168 2/2 and WT MEFs were untreated or exposed to 5 Gy of IR and fixed at the indicated time after IR. Bars, 20 mm. Three independent experiments were performed. (E) Representative micrographs of MEFs stained with anti-Mdc1 antibody and DAPI. Rnf168 2/2 and WT MEFs were untreated or exposed to 5 Gy of IR and fixed at the indicated times post IR. Three independent experiments were performed. Bars, 20 mm. (F) Representative micrographs of MEFs treated with DNA-PK inhibitors and stained with anti c-H2a.x antibody and DAPI. Rnf168 2/2 and WT MEFs were treated with DNA-PK inhibitors (NU7026 or NU7441) and cultured for 1 hour. MEFs were left untreated or exposed to 5 Gy of IR and fixed 2 hours post-IR. Three independent experiments were performed. Bars, 20 mm. doi:10.1371/journal.pgen.1001381.g003 The DNA sequences of CSR junctions can offer additional insights into the mechanistic defects of the joining process [40]. For instance, B-cells with homozygous mutations in Xrcc4 or DNA Ligase IV undergo impaired CSR and form CSR junctions with increased sequence microhomology, indicating the use of an alternative joining pathway in these mutant cells [41,42]. Previous studies reported that B-cells deficient for H2a.x or 53bp1 show no significant differences in the extent of microhomology at the switch junctions [40,[43][44][45]. To examine the effects of Rnf168 deficiency on the nucleotide composition of the switch junction regions, we cloned and sequenced the Sm-Sc1 junctions from Rnf168 2/2 and WT B-cells stimulated in vitro with LPS and IL-4. Such analysis revealed no significant differences in the extent of donor/acceptor homology ( Figure 6A and 6B), in the frequency of mutations ( Figure 6C) nor in the average length of overlaps (Table S2). Interestingly, in contrast to WT controls, 5.1% of the examined   CSR junctions in Rnf168 2/2 B-cells were found to harbor long insertions ( Figure 6D). Similar long insertions of nucleotides were observed in 2 out of 40 (5%) CSR junctions of 53bp1 2/2 B-cells but not in Atm 2/2 , H2a.x 2/2 , Nbs1 2/2 or WT B-cells [40,43,44,46]. These data therefore suggest that Rnf168 contributes to the 53bp1 function in the synapsis of DNA ends.
Collectively, these data demonstrate that Rnf168 is required for in vitro and in vivo CSR, and its inactivation in mutant mice leads to immunodeficiency, which parallels the symptoms of the RIDDLE syndrome. Furthermore, we demonstrate that the RING finger and MIU2 domains of Rnf168 are indispensable for efficient CSR. Finally, our data also indicate that, similar to 53bp1 2/2 B-cells, Rnf168 2/2 B-cells display increased frequency of long nucleotide insertions at the CSR junctions.
FACS analysis of thymocytes from Rnf168 2/2 mice showed increased frequency of DN thymocytes compared to WT littermates ( Figure S5A). Examination of DN subpopulations indicated a significantly impaired transition of Rnf168 2/2 early thymocytes at the DNIII stage ( Figure 7A, left panel). In addition, total number of DNIII thymocytes in Rnf168 2/2 mice (1.5610 6 62.8610 5 ) was significantly increased compared to WT littermates (1.1610 6 61.9610 5 , p = 0.025) ( Figure 7A, right panel). Rnf168 2/2 thymocytes also displayed a mild decrease in their expression of TCRb compared to WT controls ( Figure 7B) and the frequency of TCRcd + thymocytes in Rnf168 2/2 mice was significantly reduced compared to WT controls ( Figure S5B; WT 0.2460.21%, Rnf168 2/2 0.2060.01%, p,0.005). Productive rearrangement of the TCRb locus takes place during the DNIII to DNIV transition and leads to pre-TCR expression [49][50][51]. Only cells expressing functional pre-TCR undergo exponential expansion, a process referred to as bselection [49][50][51]. To assess the proliferation level of each DN thymocyte subpopulations, we examined in vivo thymocyte BrdU uptake in WT and Rnf168 2/2 mice. The number of BrdU + cells was not affected in Rnf168 2/2 mice compared to WT littermates ( Figure 7C). Therefore, the increased number of DNIII thymocytes, the decreased expression of TCRb and the reduced number of TCRcd cells in Rnf168 2/2 thymocytes were not due to proliferative defects. However, it remains possible that developmental defects can also contribute to these differences.
Loss of function of 53bp1 results in impaired Tcr locus integrity due to dysfunctional long-range V(D)J rearrangement [47]. To examine whether Rnf168 facilitates joining of distal gene segments during V(D)J recombination, we performed quantitative PCR assays of the partial (Dd2-Jd1 and Dd1-Dd2) and complete (Vd5-Dd1 and Vd4-Dd1) rearrangements. We observed that short-range rearrangements were more abundant in Rnf168 2/2 thymocytes compared to WT controls ( Figure 7D, left panel, Figure 7E, and Figure S5C). On the other hand, complete Vd-to-DdJd recombination was significantly reduced in Rnf168 2/2 thymocytes compared to WT controls ( Figure 7D, right panel, Figure 7E and Figure S5C).
These data support a role for Rnf168 in early thymocyte development and indicate that long-range V(D)J recombination is impaired in the absence of Rnf168.

Discussion
The RNF168 gene is located on human chromosome 3q29, and encodes a novel E3 ligase that plays an important role in the signaling of DSBs [20][21][22]. Interestingly, mutations of RNF168 have recently been identified as the genetic defects leading to RIDDLE syndrome in humans [22,24]. In order to investigate the in vivo functions of this E3 ligase, we have generated a mouse model for Rnf168 mutation.
Similar to mice mutant for its upstream DSBs signaling proteins, including H2a.x, Mdc1 and Rnf8, Rnf168 2/2 mice are viable [10,27,30,34,35]. In contrast to Rnf8 2/2 mice that display reduced number of lymphocytes [30,34], no defect in T-or B-cell numbers was observed in Rnf168 2/2 mice. The normal number of lymphocytes in Rnf168 2/2 mice is in accordance with the lack of lymphopenia in the patient with RIDDLE syndrome [24].
Initial recruitment of 53bp1 to DSB sites is not affected in the absence of H2a.x, Mdc1 or Rnf8 [10,29,30,[63][64][65]. However, inactivation of Rnf168 in MEFs completely abolishes both the transient and retained 53bp1 IRIFs. These data highlight the importance of Rnf168 for the initial recruitment of 53bp1 to DSB sites, and indicate that this function of Rnf168 is independent of H2a.x, Mdc1 and Rnf8.
Rnf168 2/2 mice showed reduced serum immunoglobulin levels, although this defect was milder compared to Rnf8 2/2 mice [30]. The impaired B-cell development observed in Rnf8 2/2 mice [30,34], but not Rnf168 2/2 mice, might contribute to the more pronounced reduction in non-IgM classes in Rnf8 2/2 mice compared to Rnf168 2/2 mice. Although Rnf168 2/2 mice showed a normal IgA levels in serum, they demonstrated impaired CSR for various IgG isotypes, thus reproducing the CSR defect associated with RIDDLE syndrome [24]. Defective CSR in Rnf168 2/2 mice is also consistent with impaired CSR following the knockdown of Rnf168 in the B-cell line CH12F3-2 [66].
Recent studies have demonstrated the functions of 53BP1 in DNA damage signaling and repair [40,45,47,[67][68][69][70]. 53BP1 promotes and maintains synapsis during V(D)J recombination [40,47]. Prior to recombination, 53BP1 constitutively associates with chromatin, supports long-range tethering of recombination signal sequence synapsis, and prevents extensive DNA end resection [47,69,70]. 53BP1 is also important for early T-cell development and for CSR [47]. In this study, we show that inactivation of Rnf168 impairs long-range V(D)J recombination in both early T-cell development and in B-cell class switch junction, suggesting that Rnf168 is required for proper synapsis of distal DSBs. Impaired signaling of DSBs often results in male infertility as observed in H2a.x 2/2 [32], Mdc1 2/2 [10] and Rnf8 2/2 mice [30,34,35]. Rnf8 deficient males exhibit complete or partial infertility, impaired ubiquitylation of H2A in the XY body [34], and failure of global nucleosome removal; however, they are proficient in meiotic sex chromosome inactivation [35]. In contrast to H2a.x 2/2 , Mdc1 2/2 and Rnf8 2/2 males, 53bp1 2/2 males are fertile, and 53bp1 is not required for meiotic recombination during spermatogenesis [28,33]. While fertility of young Rnf168 2/2 males was normal, these males became infertile with age, and elderly mice displayed testicular degeneration and atrophy. These data indicate that Rnf168 plays important roles in spermatogenesis in an age-dependent manner.
In this study, we demonstrate that Rnf168 deficiency also leads to increased radiosensitivity and genomic instability. However, similar to H2a.x 2/2 mice, tumor susceptibility was not increased in Rnf168 2/2 mice. We also demonstrate that Rnf168 and p53 collaborate in suppressing cancer since Rnf168 2/2 p53 2/2 mice exhibited shorter life spans and tumor latency compared to p53 2/2 littermates. These data indicate that p53 is critical for preventing tumorigenesis in Rnf168 2/2 mice. Therefore, not only is Rnf168 important for maintaining genomic stability, but it also collaborates with p53 to suppress cancer development.
While facial dysmorphism and short stature that are associated with RIDDLE syndrome were not observed in Rnf168 2/2 mice, these mutants displayed increased radiosensitivity, impaired CSR and immunodeficiency similar to the clinical features of this disease. Importantly, our data further highlight novel roles of Rnf168 in spermatogenesis, genomic integrity and cancer.

Generation of Rnf168-deficient mice
For the generation of Rnf168-deficient mice, two gene-trap ES clones, 156B6 and 405F11, were obtained from The Center for Modeling Human Disease (Toronto, Canada). In both ES clones, Rnf168 gene was disrupted by the integration of the gene trap vectors between exons 5 and 6 of Rnf168. Both ES cell clones were successfully used to generate Rnf168 mutant mice. Southern blotting and PCR analysis confirmed the disruption of Rnf168 locus and were used to genotype the animals. The following primers were used for PCR genotyping: Rnf168 mutant allele (forward 59-ATCGCCTTC-TATCGCCTTCT-39 and reverse 59-GCAGAAGACTCCGAA-CCTTG-39), Rnf168 WT allele (forward 59-GCCCAAGTC-TGGCTCATTTA-39 and reverse 59-GCAGAAGACTCC-GAACCTTG-39). Southern blot analysis was performed using standard procedures. Rnf168 probe was generated by PCR using the following primers: 59-TATTCCTGCTGCTGCTGCTA-39, and 59-CTCAAACCTCTTGCCCTCAG-39. Rnf168 2/2 mice were generated by intercrossing Rnf168 +/2 mice obtained from each genetrap clone. All mice in this study were in a mixed 129/J6C57BL/6 genetic background, and were maintained in a specific-pathogen free environment.

Cell culture and generation of MEFs
MEFs were generated from day 12.5 embryos using standard procedures. MEFs were cultured in DMEM (Gibco Invitrogen Corporation) supplemented with 10% FCS. Splenocytes, thymocytes and lymphocytes were cultured under the same conditions in RPMI1640 (Gibco) with 10% FCS.

Clonogenic assay
Mouse Rnf168 cDNA amplified by PCR from WT MEF total cDNA was subcloned into pBABE-puro retroviral vector. Virus supernatant was collected 48 hour post transfection of phoenix cells with pBABE-puro-Rnf168. 3T3 immortalized MEFs were infected with Mock or Rnf168 retroviruses and were subjected to puromycin selection. Infected cells (1610 3 ) were seeded to 6 cm dishes, irradiated (2, 4 and 6 Gy) and cultured for 11 days. Number of colonies was counted with crystal violet staining.

Cell cycle and checkpoints analysis
MEFs (passage 2-5) were synchronized using aphidicolin (Merck-Calbiochem) and either left untreated or treated with 10 Gy of IR. Subsequently, BrdU (Roche) was added to the cultures at various time points. MEFs were harvested, fixed using 70% ethanol and stained with FITC conjugated anti-BrdU (eBioscience) and PI. G2/M checkpoint was assessed by phosphohistone-H3 staining. Primary MEFs were either left untreated or treated with 2 Gy of IR. MEFs were harvested 1 hour post IR, fixed using 70% ethanol and stained with phospho-histone-H3 antibody (Cell signaling Technology) and PI.

Purification of splenic B-cells and in vitro CSR
B-cells were purified from spleen by negative selection using Dynal Mouse B-cell Negative Isolation kit (Dynal, Invitrogen). All B-cell experiments were performed using B-cells with a purity of 95.760.31%. B-cells (1610 6 ) were stimulated with mouse anti-CD40 antibody (10 mg/ml, eBioscience) or LPS (20 mg/ml, Sigma) plus recombinant mouse IL-4 (1000 U/ml, eBioscience) for 4 days, and switching of Ig to IgG1 and IgG2a isotypes was examined. LPS (15 mg/ml) was used to induce IgG2b and IgG3 switching.

MSCV infection of primary B-cells
Mutants for Rnf168 cDNA were generated by PCR. Wild-type and mutant Rnf168 cDNAs were cloned into the MSCV-IRES-GFP vector. The ecotropic retroviruses expressing WT or mutated Rnf168 were generated by transient transfection of Phoenix cells with the appropriate retrovirus constructs. Purified splenic B-cells were activated with LPS and infected with virus supernatants containing polybrene (Sigma). Infected B-cells were cultured with LPS plus IL-4 for 4 days, and stained with anti-IgG1 for FACS analysis.

Analysis of CSR by FACS
Purified B-cells were labeled with CFSE (Molecular Probes, Invitrogen) and cultured to induce IgG1 class switching for 4 days. Cells were stained with anti-IgG1 antibody and analyzed by FACS using the CellQuest software (BD Biosciences) or FlowJo analysis software (Tree star).

Analysis of Sm-Sc1 switch recombination junctions
Genomic DNA isolated from stimulated B-cells was amplified by PCR using Pfu ultra polymerase (Stratagene) and the following primers (5m3, 59-AATGGATACCTCAGTGGTTTTTAATGG-TGGGTTTA-39 and c1-R, 59-CAATTAGCTCCTGCTCTTC-TGTGG-39). PCR products (500-1000 bp) were cloned into pCR2.1 using the TOPO TA cloning kit (Invitrogen). Sequence analysis of the cloned PCR products was performed using Sequencher software (GeneCode) and NCBI-BLAST. ELISA B-cells (1610 6 ) were stimulated with mouse anti-CD40 antibody or LPS (20 mg/ml) plus IL-4 (1000 U/ml) for 4 days and the levels of IgG1, IgG2a, and IgM isotypes in the culture supernatants were determined. Isotype switching to IgG2b and IgG3 was similarly examined using supernatants from B-cells stimulated with LPS (15 mg/ml). The levels of immunoglobulin in cell culture supernatants and serum from young (6-8 weeks) and old (9-12 months) mice were determined using SBA Clonotyping System/HRP and Mouse Immunoglobulin Isotype Panel (SouthernBiotech).
Genomic DNA was extracted from total thymocytes and TCRd rearrangement was assessed using the Applied Biosystems 7900HT Fast Real Time PCR System (Applied Biosystems), Power SYBR Green PCR Master Mix (Applied Biosystems) and specific primers as described previously [47].

BrdU uptake experiments
Rnf168 2/2 and WT littermates received two intraperitoneal injections of BrdU (1 mg each) with 2 hours interval. Thymocytes collected from these mice, 2 hours after the second injection, were examined for BrdU incorporation using a BrdU Flow kit (Becton Dickinson) and cytofluorometry.
Cells were stained with DAPI for 5 min and then mounted with Mowiol mount solution (Calbiochem). The slides were observed under a Leica DMIRB fluorescence microscope (Germany) equipped with digital camera (Leica DC 300RF). Images were acquired under 1006 magnification using Leica Image Manager software. Foci-positive cells were quantified by manual counting.

Histology
Paraffin sections of testes, epididymides, tumors and organs were stained with hematoxylin-eosin for histological analyses. The slides were observed under a Leica DM4000B microscope (Germany) equipped with digital camera (Leica DC 300RF). Images were acquired using Leica Image Manager software.

Cytogenetic analysis
Purified B-cells were activated with LPS for 48 hours, and either left untreated or irradiated as indicated. Cells were collected following 4 hours of colcemid treatment and processed by standard cytogenetic procedures. Number of chromosomes and gross chromosomal rearrangements were determined in 60 metaphase cells per sample from each genotype. The slides were observed under a Leica DMIRB fluorescence microscope (Germany) equipped with digital camera (Leica DC 300RF). Images were acquired under 1006 magnification using Leica Image Manager software.

Multicolor fluorescence in situ hybridization (mFISH)
Mouse mFISH probe kit was obtained from MetaSystems GmbH, Germany. In mFISH, all 21 chromosomes are each painted in a different color using combinatorial labeling. The stained metaphases were captured using the Axio Imager Z1 microscope (Carl Zeiss) with filter sets for FITC, Cy3.5, Texas Red, Cy5, Aqua, and DAPI. Images were captured, processed, and analyzed using ISIS mBAND/mFISH imaging software (MetaSystems). Detailed experimental procedures were outlined in earlier publications [74,75].

Statistical analysis
Data are presented as the mean 6 SEM. The statistical significance of experimental data (p-values; Values#0.05) was determined using the Wilcoxon test. Log-rank (Mantel-Cox) test was used for comparisons of survival curves.

Ethics statement
All experiments were performed in compliance with Ontario Cancer Institute animal care committee guidelines.  Figure S3 Quantification of the effect of Rnf168 inactivation on IRIF for DDR proteins. (A) Quantitative analyses of 53bp1 nuclear foci are shown. Rnf168 2/2 , Rnf8 2/2 and WT MEFs were either untreated or exposed to 5 Gy of IR and fixed at the indicated times after IR. Three independent experiments were performed. (B) Quantitative analyses of the formation of Brca1 nuclear foci are shown. Rnf168 2/2 and WT MEFs were either untreated or exposed to 5 Gy of IR and were fixed at the indicated times after IR. Three independent experiments were performed. (C) Quantitative analyses of the formation of c-H2a.x nuclear foci 6 hours post-IR. Rnf168 2/2 and WT MEFs were untreated or exposed to 5 Gy of IR. Three independent experiments were performed. (D) Quantitative analyses of the formation of Mdc1 nuclear foci. Rnf168 2/2 and WT MEFs were untreated or exposed to 5 Gy of IR and cells were fixed at the indicated times post-IR. Three independent experiments were performed. The data are presented as the mean 6 SEM.