An XRCC4 mutant mouse, a model for human X4 syndrome, reveals interplays with Xlf, PAXX, and ATM in lymphoid development

We developed an Xrcc4M61R separation of function mouse line to overcome the embryonic lethality of Xrcc4-deficient mice. XRCC4M61R protein does not interact with Xlf, thus obliterating XRCC4-Xlf filament formation while preserving the ability to stabilize DNA ligase IV. X4M61R mice, which are DNA repair deficient, phenocopy the Nhej1-/- (known as Xlf -/-) setting with a minor impact on the development of the adaptive immune system. The core non-homologous end-joining (NHEJ) DNA repair factor XRCC4 is therefore not mandatory for V(D)J recombination aside from its role in stabilizing DNA ligase IV. In contrast, Xrcc4M61R mice crossed on Paxx-/-, Nhej1-/-, or Atm-/- backgrounds are severely immunocompromised, owing to aborted V(D)J recombination as in Xlf-Paxx and Xlf-Atm double Knock Out (DKO) settings. Furthermore, massive apoptosis of post-mitotic neurons causes embryonic lethality of Xrcc4M61R -Nhej1-/- double mutants. These in vivo results reveal new functional interplays between XRCC4 and PAXX, ATM and Xlf in mouse development and provide new insights into the understanding of the clinical manifestations of human XRCC4-deficient condition, in particular its absence of immune deficiency.


Introduction
Living organisms face DNA double-strand breaks (DSBs), the most toxic DNA lesions, from random or programmed (prDSBs) origins (Betermier et al., 2020), such as during the development of the adaptive immune system through V(D)J recombination. V(D)J recombination results in the somatic rearrangement of variable (V), diversity (D), and joining (J) elements of antigen receptor loci in T-and B-cell precursors (Jung et al., 2006). It is initiated by the domesticated transposase Recombination-Activating Genes 1 and 2 factors (RAG1/2), which introduce prDSBs at the border of V, D, and J elements within recombination signal sequences (RSS). The non-homologous end-joining (NHEJ) machinery is the sole DNA repair pathway to cope with these lymphoid-specific prDSBs. Briefly, the Roch the 'two-synapses' model knowing its structural and functional relationship with Xlf (Callebaut et al., 2006). XRCC4 is a keystone factor, playing two independent roles. First, X4 stabilizes Lig4 through its C-terminal coiled-coil domain (Grawunder et al., 1997). Indeed, Lig4 expression is abrogated in Xrcc4 Knock Out (KO) models both in vivo and in vitro (Gao et al., 1998). In addition, X4 and Xlf homodimers interact through their N-terminal globular head to form long polymeric 'filaments' (Reid et al., 2015;Ropars et al., 2011). X4-Xlf filaments generate a 'DNA repair synapse' to tether broken DNA ends (Brouwer et al., 2016;Reid et al., 2015). DNA-end synapsis is a central issue during NHEJ, and the recent development of in vitro single-molecule technologies has highlighted the dynamic formation of DNA end-to-end synapses (flexible/long range for DNA end tethering and close/short range for DNA ligation) in addition to the Xlf-X4 filament, in which the various core NHEJ DNA repair factors (Ku70/80, DNA-PKcs, Xlf, X4/L4, and PAXX) participate to various degrees, in particular the association of Xlf with both X4 and Ku (for a recent review, see Zhao et al., 2020). Several studies recently reported on the details of the structural assembly of these complexes using cryo-EM, thus improving our understanding on the composition of these complexes as well as the dynamics of the transition between various states during NHEJ-mediated DNA repair (Chaplin et al., 2021a;Chaplin et al., 2021b;Zhao et al., 2020). These studies support in particular the interaction of the L4X4 complex with that of Ku70/80 previously proposed by Costantini et al., 2007. The V(D)J recombination phenotype of Nhej1 KO mice would argue that the RAG2-mediated DNA tethering is also redundant with these DNA end-to-end synapses. Nevertheless, the intimate nature of DNA end joining during V(D) J recombination may not always strictly coincide with what we know for the repair of genotoxic DNA breaks, precisely because of the existence of the 'two-synapses' mechanism.
Since X4 is compulsory for Lig4 stabilization, Xrcc4-/-mice phenocopy Lig4-/-condition, with an embryonic lethality and SCID phenotype (Gao et al., 1998). The embryonic lethality is rescued on Trp53-/-background, but the SCID resulting from a complete block of lymphocyte development remains (Gao et al., 2000). Furthermore, Xrcc4-/-Trp53-/-DKO mice develop Pro-B cell lymphomas (Chen et al., 2016;Gao et al., 2000). To avoid disrupting the critical Lig4 stabilization function of X4, we engineered a Xrcc4 knock-in (KI) mouse model harboring the M61R missense mutation that abrogates X4-Xlf interaction (Ropars et al., 2011), while keeping the Lig4 interaction domain unperturbed. Xrcc4 M61R mice are viable attesting for the stabilization of functional Lig4, thus allowing the study of lymphocyte development. Introduced on Atm-/-, Paxx-/-, and Nhej1-/-backgrounds, the Xrcc4 M61R mutation allows to address the 'two-synapses' DNA repair model in V(D)J recombination, and to expand the picture of NHEJ apparatus in brain and lymphocyte development.

Results
Generation of X M61R mice X4 and Xlf interact through a hydrophobic interface, which is disrupted by the X4 M61R substitution, thus resulting in the loss of X4-Xlf filament (Ropars et al., 2011). We developed an Xrcc4 M61R KI mouse model through CRISPR/Cas9 ( Figure 1A). Homozygous Xrcc4 M61R/M61R mice (Xrcc4 M61R mice) were viable arguing against any harmful impact of the M61R substitution during embryonic development, as opposed to the embryonic lethality of Xrcc4 KOs.

X4 M61R stabilizes Lig4
RT-PCR analysis of X4 mRNA expression unveiled two isoforms in M61R mutation-bearing cells ( Figure 1B). The expression of the full-length transcript carrying the five mutations was reduced ( Figure 1B, upper band) and was accompanied by a shorter isoform ( Figure 1B, lower band). Sequencing this alternative transcript revealed the out-of-frame splicing-out of exon 3 (Figure 1figure supplement 1). The same X4 transcript missing exon 3 was described in a previous Xrcc4 KO mouse model (Gao et al., 2000;Gao et al., 1998). A subsequent attempt to generate Xrcc4 M61R KI mouse model using an ssODN template lacking the four silent mutations resulted in the same pattern of expression of the M61R allele (data not shown). At the protein level, X4 expression was barely detectable in spleen, brain, thymus, and mouse embryonic fibroblasts (MEFs) from Xrcc4 M61R mice as opposed to their homozygous WT counterparts ( Figure 1C and D). In summary, Xrcc4 M61R mice express two X4 transcripts, one lacking exon 3 and resulting in a complete loss of function, and a Figure 1. X4 and Lig4 expression and impaired non-homologous end-joining (NHEJ) in Xrcc4 M61R mice. (A) Schematic representation of CRISPR/ Cas9-driven homologous recombination strategy to generated Xrcc4 M61R knock-in mouse model. (B) 5′UTR to 3′UTR Xrcc4 RT-PCR in mouse thymocyte extracts from four littermates. The lower transcript represents the splicing out of exon 3 as described (Gao et al., 1998)  second, full length, harboring the M61R missense variant encoding a weakly expressed X4 protein lacking its ability to interact with Xlf. Of note, Lig4 protein was readily detectable by WB in protein lysates from Xrcc4 M61R mice spleen and thymus, attesting for its stabilization by the X4 M61R protein ( Figure 1E). We conclude that the Xrcc4 M61R mouse model represents a hypomorphic condition in which the X4 M61R protein, although weakly expressed, retains the capacity to stabilize Lig4 and thus ensures proper embryonic development with viable animals at birth.

DSB repair defect in X4 M61R cells
MEFs from Xrcc4 M61R mice presented a statistically significant decrease in cell viability compared to WT cells upon phleomycin, a DSB inducer, treatment, attesting for a profound DNA repair defect ( Figure 1F). Likewise, splenic T cells from Xrcc4 M61R mice showed an increased IR sensitivity with an almost complete loss of viability at 2 Gy ( Figure 1G). These experiments demonstrate that the X4 M61R mutant protein, which preserves the stabilization of Lig4, is nevertheless impaired in its capacity to achieve full repair of DSB introduced by random genotoxic agents.
These results indicate that B-and T-lymphocyte development is not severely affected in Xrcc4 M61R mice, despite the reduced expression of X4 M61R . The minor accumulation of DN3A thymocytes and immature B cells suggests a modest decreased V(D)J recombination efficiency, supported by the bias in TCRα usage. This is in sharp contrast with Xrcc4-/-and Lig4-/-mice, which present a complete arrest of lymphocyte maturation and a SCID phenotype. We conclude that when Lig4 is stabilized, the contribution of X4 through Xlf/X4 interaction, disrupted in the context of X4 M61R , is not critical for V(D) J recombination in vivo.
The online version of this article includes the following figure supplement(s) for figure 1:  We previously proposed a model of 'double DNA repair synapse' during V(D)J recombination according to which the X4-Xlf 'filament' and the RAG1/2 PCC, together with ATM and PAXX, provide two complementary means of DNA end synapsis (Abramowski et al., 2018;Betermier et al., 2020;Lescale et al., 2016a). To address the role of X4 in this model, we crossed Xrcc4 M61R onto Paxx-/-and Atm-/-mice (hereafter denoted as Xrcc4 M61R -Paxx and Xrcc4 M61R -Atm). Both double mutant mice were viable but exhibited severe growth retardation and facial dysmorphia as a consequence of either the X4 M61R mutation and/or the overall reduction in X4/L4 expression ( Figure 3-figure supplement 1).
We conclude that Xrcc4 M61R crossed on Paxx-/-or Atm-/-phenocopy the SCID phenotype of Nhej1-Paxx and Nhej1-Atm DKO conditions owing to impaired V(D)J recombination. Therefore, aside from its essential role in Lig4 stabilization, X4 does participate in DNA coding-end tethering through its interaction with Xlf during V(D)J recombination, and PAXX/ATM are compensatory factors for this later function in vivo, as they are for Xlf. Alternatively, one may consider that the Xlf function being impaired in the context of X4 M61R , PAXX/ATM complements this defect as already established.
Altogether, Xrcc4 M61R -Nhej1 embryonic development phenocopy Nhej1-Paxx DKO, with a profound defect in V(D)J recombination and the apoptosis of post-mitotic neurons leading to embryonic lethality.

Discussion
We created an Xrcc4 separation of function allele in mice, which harbor the M61R substitution previously described to abolish X4-Xlf interaction (Ropars et al., 2011). The X4 M61R protein, although weakly expressed, preserves the stabilization of Lig4 but severely compromises the NHEJ DNA repair function in affected mice. Nevertheless, Xrcc4 M61R mice are viable, arguing that the embryonic lethality of Xrcc4 KO mice is most likely the result of the de facto absence of Lig4 in this setting, thus phenocopying Lig4 KO condition. Interestingly, this also indicates that the formation of the X4-Xlf filament, abrogated by the M61R mutation, is dispensable for post-mitotic neuron viability, the loss of which causes lethality in Xrcc4 and Lig4 KO settings.
Immune development of Xrcc4 M61R mice phenocopied that of Nhej1-/-animals, with a slight decreased thymic cellularity, an increased thymocyte apoptosis, and skewed TCRα repertoire, which we previously linked to a suboptimal Tcra rearrangement efficiency (Berland et al., 2019;Roch et al., 2019;Vera et al., 2013). In addition to their TCRα repertoire bias, Xrcc4 M61R thymocytes also faced a developmental delay at DN3A both in fetal and adult thymus like Nhej1 KO condition, further attesting for a suboptimal V(D)J recombination activity in these mice. Therefore, although X4 is required for V(D)J recombination through Lig4 stabilization, it is not mandatory for coding-ends tethering during V(D)J recombination. These results confirm that X4 exerts a Lig4-independent function through its interaction with Xlf that is decisive for the repair of genotoxic-induced DSBs but compensated for by other DNA repair factors, and possibly RAG2 itself as shown for Xlf (Lescale et al., 2016a), during V(D)J recombination.   were severely immunocompromised, owing to aborted V(D)J recombination, as noticed in Nhej1-Paxx and Nhej1-Atm DKO. We conclude that PAXX and ATM are compensatory factors for X4 M61R during V(D)J recombination. This result further supports our previously proposed model of 'double DNA repair synapse' in V(D)J recombination, mediated by the X4-Xlf and RAG2-PAXX-ATM axes, respectively (Abramowski et al., 2018;Betermier et al., 2020;Lescale et al., 2016a). In addition to their defect in V(D)J recombination, Xrcc4 M61R -Nhej1 double homozygous mice were embryonic lethal owing to massive post-mitotic neuron apoptosis, recapitulating Xrcc4-/-setting. This observation may be the result of the lower expression of the X4 M61R protein, which could have additive/synthetic effects with the loss of Xlf interaction during NHEJ. Indeed, Nhej1 deficiency causes synthetic lethality with several other NHEJ-deficient conditions, such as Prkdc-/-, Mri-/-, H2ax-/-, and Paxx-/- (Abramowski et al., 2018;Hung et al., 2018;Xing et al., 2017;Zha et al., 2011). Therefore, Xlf can rescue very different NHEJ defects, through an unknown mechanism, perhaps in relation with recovery of replication fork stalling as shown in the Nhej1-H2ax DKO mice (Chen et al., 2019).
Altogether, the Xrcc4 M61R separation of function allele highlights for the first time at least two independent roles of X4 in NHEJ and establishes that X4 is not mandatory for coding ends tethering during V(D)J recombination. This study also unravels novel interplays between X4, PAXX, ATM, and Xlf during development of the brain and the immune system.
Several deleterious mutations in the XRCC4 gene have been reported in humans (see de Villartay, 2015 for review), most of which are associated with microcephalic primordial dwarfism (MPD), gonadal failure, early-onset metabolic syndrome, and cardiomyopathies. The DNA repair deficiency in these patients is manifest when tested in vitro retrospectively. Most surprisingly however, they do not present noticeable signs of immune dysfunction. Altogether, the clinical presentation of these patients was not evocative of an impaired NHEJ given the known impact of its deficiency on the development of the adaptive immune system. Indeed, X4 mutations were not identified in these patients through hypothesis-driven candidate gene sequencing but rather through unsupervised whole-exome sequencing. Our present study now provides some hints as to explain the absence of immunological features in X4-deficient patients; when Lig4 expression is spared to some extent by hypomorphic mutations, allowing birth, XRCC4 appears not critically required for the development of the adaptive immune system.

Cellular sensitivity to DSBs-inducing agents and thymocyte survival assay
For phleomycin sensitivity assay, 5000 MEF cells were seeded in triplicates and cultivated with increasing doses (0-300 ng/mL) of phleomycin. Living cells were counted by flow analysis with FACS LSR-Fortessa X-20 after 6 days of culture. Radiation sensitivity of T lymphocytes was performed as described (Abramowski et al., 2018). Ex vivo thymocyte survival assay was performed as described (Vera et al., 2013).

Quantitative real-time RT-PCR analysis
TaqMan PCR was performed on triplicates of 8 ng of reverse-transcribed RNA from freshly dissected total thymus as described (Vera et al., 2013).

TCRβ V(D)J recombination analysis
Tcrb rearrangements were analyzed by PCR on genomic DNA from total adult thymocytes or E18.5 fetal thymocytes as described (Abramowski et al., 2018).

E15.5 brain sections immunohistochemistry
Neuronal apoptosis was analyzed as previously described (Abramowski et al., 2018). E15.5 fetal heads were fixed overnight at 4 °C by immersion in 4% paraformaldehyde and embedded in paraffin with a Tissu-tek processor (VIP, Leica). 5 μm coronal sections were then obtained using a microtome (Leica RM2125RT) and mounted onto glass slides for histological analyses. After paraffin removal and citrate treatment, the brain sections were permeabilized with 0.5% Triton X-100 in phosphatebuffered saline (PBS) for 15 min and incubated for 2 hr with 7.5% fetal bovine serum and 7.5% goat serum in PBS. The sections were incubated with rabbit anti-CC3 (Cell Signaling 9661) overnight at 4 °C. After washing, the sections were incubated with goat anti-rabbit Alexa Fluor 488 or 594 conjugated secondary antibody (ThermoFisher) for 1 hr. After washing, nuclear staining was achieved by incubation with 4′-6-diamidino-2-phenylindole (DAPI) to quantify apoptosis induction by the detection of pyknotic nuclei (Roque et al., 2012). Slides were mounted under Fluoromount (Southern Biotechnologies Associates). Tissues were examined under a fluorescence microscope (50i, Nikon, Japan) with a 10× (NA = 0.3) objective in three channels (appearing red, green, and gray) as separate files. These images were then stacked with Photoshop software (Adobe).

Additional files
Supplementary files • Transparent reporting form
The following dataset was generated: