Leaky severe combined immunodeficiency in mice lacking non-homologous end joining factors XLF and MRI

Non-homologous end-joining (NHEJ) is a DNA repair pathway required to detect, process, and ligate DNA double-stranded breaks (DSBs) throughout the cell cycle. The NHEJ pathway is necessary for V(D)J recombination in developing B and T lymphocytes. During NHEJ, Ku70 and Ku80 form a heterodimer that recognizes DSBs and promotes recruitment and function of downstream factors PAXX, MRI, DNA-PKcs, Artemis, XLF, XRCC4, and LIG4. Mutations in several known NHEJ genes result in severe combined immunodeficiency (SCID). Inactivation of Mri, Paxx or Xlf in mice results in normal or mild phenotype, while combined inactivation of Xlf/Mri, Xlf/Paxx, or Xlf/Dna-pkcs leads to late embryonic lethality. Here, we describe three new mouse models. We demonstrate that deletion of Trp53 rescues embryonic lethality in mice with combined deficiencies of Xlf and Mri. Furthermore, Xlf-/-Mri-/-Trp53+/- and Xlf-/-Paxx-/-Trp53+/- mice possess reduced body weight, severely reduced mature lymphocyte counts, and accumulation of progenitor B cells. We also report that combined inactivation of Mri/Paxx results in live-born mice with modest phenotype, and combined inactivation of Mri/Dna-pkcs results in embryonic lethality. Therefore, we conclude that XLF is functionally redundant with MRI and PAXX during lymphocyte development in vivo. Moreover, Mri genetically interacts with Dna-pkcs and Paxx.


INTRODUCTION
Non-homologous end-joining (NHEJ) is a DNA repair pathway that recognizes, processes and ligates DNA double-stranded breaks (DSB) throughout the cell cycle. NHEJ is required for lymphocyte development; in particular, to repair DSBs induced by the recombination activating genes (RAG) 1 and 2 in developing B and T lymphocytes, and by activation-induced cytidine deaminase (AID) in mature B cells [1]. NHEJ is initiated when Ku70 and Ku80 (Ku) are recruited to the DSB sites. Ku, together with DNA-dependent protein kinase, catalytic subunit (DNA-PKcs), forms the DNA-PK holoenzyme [2]. Subsequently, the nuclease Artemis is recruited to the DSB sites to process DNA hairpins and overhangs [3]. Finally, DNA ligase IV (LIG4), X-AGING ray repair cross-complementing protein 4 (XRCC4) and XRCC4-like factor (XLF) mediate DNA end ligation. The NHEJ complex is stabilized by a paralogue of XRCC4 and XLF (PAXX) and a modulator of retroviral infection (MRI/CYREN) [4,5].
In this study, we rescue synthetic lethality from Xlf and Mri by inactivating one or two alleles of Trp53. We also show that both Xlf -/-Mri -/-Trp53 +/and Xlf -/-Paxx -/-Trp53 +/mice possess a leaky SCID phenotype with severely reduced mature B and T lymphocyte counts in the spleen, low mature T cell counts in the thymus, and accumulated progenitor B cells in the bone marrow. Finally, we demonstrate that MRI is functionally redundant with DNA-PKcs and PAXX.

Leaky SCID in mice lacking XLF and PAXX
Combined inactivation of XLF and PAXX has been shown to result in embryonic lethality in mice [4,14,15,20]. To determine the impact of XLF and PAXX on B and T cell development in vivo, we rescued the synthetic lethality by inactivating one allele of Trp53, as described previously [20]. We did not detect any direct influence of altered Trp53 genotype on lymphocyte development (Supplementary Tables 1-9 Table 9). Therefore, we conclude that B cell development is blocked at the pro-B cell stage of Xlf -/-Mri -/-Trp53 +/and Xlf -/-Paxx -/-Trp53 +/mice.

Paxx -/-Mri -/mice possess a modest phenotype
Both PAXX and MRI are NHEJ factors that are functionally redundant with XLF in mice. Combined inactivation of Paxx and Xlf [4,14,15,20], or Mri and Xlf ( [5]; this study) results in synthetic lethality in mice, AGING as well as in abrogated V(D)J recombination in vAbl pre-B cells [4,5,14,15,27]. To determine if Paxx genetically interacts with Mri, we intercrossed mice that are heterozygous or null for both genes (such as Paxx -/-Mri +/and Paxx +/-Mri +/-). We found that resulting Paxx -/-Mri -/mice are live-born, fertile, and are similar in size to WT littermates (17 g, p>0.9999) ( Figure 3A and 3B). Specifically, we observe that Paxx -/-Mri -/mice have normal thymocyte and splenocyte counts. Furthermore, Paxx -/-Mri -/mice underwent normal T cell development that was indistinguishable from the WT, Paxx -/-, and Mri -/controls ( Figures 1H, 1I and 3C). However, Paxx -/-Mri -/mice had reduced CD19+ B cell counts ( Figure  1G) when were compared to WT, Paxx -/and Mri -/controls (p<0.0025). Moreover, CD19+ B cell counts were similar in Paxx -/-Mri -/and Xlf -/mice (p>0.9270), suggesting that combined depletion of PAXX and MRI has modest phenotype similar to the one in XLFdeficient mice. CSR to IgG1 was performed in order to determine if DNA repair-dependent immunoglobulin production is affected in mature B cells lacking PAXX and MRI [16,18]. Paxx inactivation did not affect Ig switch to IgG1 in MRI-deficient B cells ( Figure 3D and 3E). The quantity of IgG1+ cells after CSR stimulation was similar between Paxx -/-Mri -/and Mri -/naïve B cells (p>0.73). From this, we can conclude that there is a genetic interaction between Paxx and Mri in vivo, and it is only detected in B cells.
Our findings have demonstrated that mice lacking XLF, MRI and p53, although live-born, possess a leaky SCID phenotype. Xlf -/-Mri -/-Trp53 +/mice have a clear fraction of mature B cells in the spleens (CD19+) and bone marrow (B220+CD43-IgM+) (Figures 1 and 2), as well as clear fractions of double-and single-positive T cells in the thymus (CD4+CD8+, CD4+, CD8+) and singlepositive T cells in the spleen (CD4+ and CD8+) ( Figure  1). However, the cell fractions from these mice are noticeably smaller than those of WT or single-deficient mice. Strikingly, we were able to identify one Xlf -/-Mri -/-Trp53 +/+ mouse at day P30 post-birth. This mouse resembled Xlf -/-Mri -/-Trp53 +/mice of similar age with respect of B and T cell development (Supplementary  Table 10), although this mouse was generally sicker than its littermates and had to be euthanized. Similarly, one live-born Xlf -/-Paxx -/mouse was reported by Balmus et al. 2016 [15], indicating that, exceptionally, embryonic lethality in NHEJ ligation-deficient mice can be overcome, likely due to activity of alternative endjoining. Previously, in 2018, Hung et al. [5] reported that combined inactivation of Xlf and Mri in vAbl pre-B cells results in a severe block in V(D)J recombination and accumulation of unrepaired DSBs in vitro, although it was unclear whether this combined inactivation would lead to a deficiency in B lymphocytes when translated to a mouse model [5]. Similarly, double deficient vAbl pre-B cells lacking Xlf and Paxx are also unable to sustain V(D)J recombination. Importantly, the lack of a progenitor T cell model system left the question of T cell development in Xlf -/-Mri -/and Xlf -/-Paxx -/mice completely unexplored.
Previously, we showed that mice lacking XLF, PAXX and p53 were live-born and had nearly no B and T cells, reduced size of spleen and hardly detectable thymus [20] ( Figure 5). Consistent with this model, a conditional AGING knockout mouse model, which results in doubledeficiency of XLF/PAXX in early hematopoietic progenitor cells, was also able to overcome the embryonic lethality of Xlf -/-Paxx -/mice [33]. With this model, impairment of V(D)J recombination in Xlf -/-Paxx -/cells, as well as the resulting depletion of mature B cells and lack of a visible thymus could also be observed in vivo [33]. Our new data provide evidence that Xlf -/-Paxx -/-Trp53 +/and Xlf -/-Paxx -/-Trp53 -/mice possess a very small number of mature B cells in the spleen and bone marrow, as well as very minor fractions of single positive T cells in thymus and spleen (Figures 2, 5 and Supplementary Figure 1). Therefore, both mature B and T cells are present in mice lacking XLF/PAXX and XLF/MRI. This can be explained by incomplete blockage in NHEJ and V(D)J recombination, in which the process is dramatically reduced but still possible. We also detected more mature T cells than B cells in these double-deficient mice. Potential explanations include longer lifespan of T cells, which accumulate over time following low efficiency of V(D)J recombination, while B cells are eliminated faster from the pool due to the different physiology [34,35]. It is also possible that the T cells we detected are a resultant subpopulation that is descendent from the few cells that were able to bypass V(D)J recombination [12]. In this case, the repertoire of T cells based on T cell receptor in mice lacking XLF/PAXX and XLF/MRI would be significantly lower than in control mice, even if normalized to the total cell count. Due to the small presence of mature B and T cells in Xlf -/-Mri -/-Trp53 +/-, Xlf -/-Paxx -/-Trp53 +/and Xlf -/-Paxx -/-Trp53 -/mice, we categorize the observed immunodeficient phenotypes as "leaky SCID". Previously, leaky SCID has been described in mice lacking other NHEJ factors, such as Ku70 -/- [6], Artemis -/- [3], Lig4 -/-Trp53 -/- [10,30], Xrcc4 -/-Trp53 -/- [9,31], Xlf -/-Atm -/- [19] and Xlf -/-Rag2 c/c [23].
In addition to XLF/MRI and XLF/PAXX deficient mice, inactivation of one or two alleles of Trp53 also rescues the embryonic lethality of Xrcc4 -/- [9,31], Lig4 -/- [10,30] and Xlf -/-Dna-pkcs -/- [20] mice. We propose a model ( Figure 5), when single deficiency for DNA-PKcs, PAXX or MRI results in no or modest ). The χ 2 was 1.8 and its corresponding probability was between 25 and 50%. *Expected distribution assuming lethality. AGING phenotypes, and DSBs are efficiently repaired. Combined inactivation of Xlf/Dna-pkcs, Xlf/Paxx and Xlf/Mri results in inefficient DSB ligation, accumulation of DNA breaks, activation of ATMdependent DDR, checkpoint protein CHK2, stabilization of p53 and massive apoptosis. This results in embryonic lethality in mice. Furthermore, inactivation of Trp53 results in Xlf/Dna-pkcs/Trp53, Xlf/Paxx/Trp53 and Xlf/Mri/Trp53 triple-deficient mice. While DNA breaks in these mice are not repaired, ATM-dependent DDR response and activation of CHK proteins takes place. However, without p53, apoptosis is not activated, allowing survival of mice ( Figure 5). Moreover, we propose that inactivation of Atm will also rescue embryonic lethality of Xlf/Paxx and Xlf/Mri mice due to the mechanisms proposed above. However, inactivation of Atm will not rescue embryonic lethality of Xlf/Dna-pkcs mice, due to synthetic lethality between Atm and Dna-pkcs.

AGING
It is important to note that altered Trp53 expression is not always sufficient to rescue embryonic lethality in mice; for example, PLK1-interacting checkpoint helicase (PICH)-deficient mice possess developmental defects in the presence or absence of p53 [36], and ATR mutants (Seckel syndrome) are not completely rescued from embryonic lethality with the inactivation of Trp53 [37]. Embryonic lethality of XLF/PAXX and XLF/MRI double-deficient mice can be explained by the presence of Ku70/Ku80 heterodimer at the DSBs sites, which blocks DNA repair by alternative end-joining pathway(s), leading to massive apoptosis and cell cycle arrest [38]. Previously, it was shown that embryonic lethality of LIG4-deficient [39] and XLF/DNA-PKcs double-deficient mice [25] could be rescued by inactivating Ku70 or Ku80 genes. Similarly, we propose that inactivation of either Ku70 or Ku80 gene will rescue the embryonic lethality of XLF/PAXX and XLF/MRI double-deficient mice and will result in mice indistinguishable from Ku70-or Ku80-deficient controls ( Figure 5).
Recent studies have shown that Xlf genetically interacts with Rag2 [23] and DDR factors, such as Atm, 53bp1, H2ax, and Mdc1 [17,[19][20][21][22]38]. Xlf -/-Rag2 c/c mice almost completely lack mature B cells and have significantly fewer mature T cells than single deficient controls [23]. Xlf -/-Atm -/and Xlf -/-53bp1 -/mice are liveborn and exhibit reduced body weight, increased genomic instability, and severe lymphocytopenia as a result of V(D)J recombination impairment in developing B and T cells [1,17,19,22]. Xlf -/-H2ax -/and Xlf -/-Mdc1 -/-, on the other hand, are embryonic lethal [19][20][21]. There are several possible explanations for the functional redundancy observed between DNA repair genes. For instance, the two factors could have identical (e.g., if both proteins are involved in ligation or DNA end tethering) or complementary (e.g., if one protein stimulates ligation while the other is required for DNA end tethering) functions. To date, XLF has been shown to genetically interact with multiple DNA repair factors [1,4,5,14,15,19,20,24,25], and this list is likely to grow [38,40]. However, no clear genetic interaction has been shown between Xlf and Artemis or Xrcc4 in the context of mouse development and V(D)J recombination [24], meaning that it remains difficult to predict genetic interactions without developing and characterizing genetic models.
We found that mice with combined inactivation of Paxx and Mri (Paxx -/-Mri -/-) are live-born, fertile, and undergo almost normal B and T cell development (Figure 3), where only the number of splenic B cells is affected, giving rise to a modest phenotype. Moreover, inactivation of Paxx did not affect the CSR efficiency in in vitro stimulated MRI-deficient B cells (Figure 3), thereby confirming our observations in vitro. It has been also shown that combined inactivation of Paxx and Mri genes in vAbl pre-B cells lead to similar V(D)J recombination efficiency to single-deficient Mri -/-, Paxx -/and WT controls [5]. Thus, we conclude that there is a genetic interaction between Paxx and Mri, which results in a modest phenotype.
Lastly, we found that combined inactivation of Mri and Dna-pkcs (Mri -/-Dna-pkcs -/-) leads to embryonic lethality, and that E14.5 Mri -/-Dna-pkcs -/murine embryos were about 40% smaller than single-deficient siblings (Figure 4). DNA-PKcs is associated with the Nterminus of the MRI and Ku heterodimer in the process of recognizing DSBs [5], which may account for genetic interaction between Mri and Dna-pkcs. Thus, inactivation of Trp53, Ku70 or Ku80 may be a viable method to rescue synthetic lethality from Mri -/-Dnapkcs -/mice.

Mice
All experiments involving mice were performed according to the protocols approved by the Comparative Medicine Core Facility (CoMed) at the Norwegian University of Science and Technology (NTNU, AGING Trondheim, Norway). Xlf +/- [11] and Dna-pkcs +/- [2] mice were imported from the laboratory of Professor Frederick W. Alt at Harvard Medical School. Trp53 +/mice [32] were imported from Jackson Laboratories. Paxx +/- [16] and Mri +/- [18] mice were generated by the Oksenych group and described previously.

Lymphocyte development
Lymphocyte populations were analyzed by flow cytometry [16,18,19,22]. In summary, cells were isolated from the spleen, thymus, and femur of 5-7week-old mice and treated with red blood cell lysis buffer Hybri-Max TM (Sigma Aldrich, St. Louis, MO, USA; #R7757). The cells were resuspended in PBS (Thermo Scientific, Basingstoke, UK; #BR0014G) containing 5% Fetal bovine serum, FCS (Sigma Life Science, St. Louis, Missouri, United States; #F7524), and counted using a Countess™ II Automated Cell Counter (Invitrogen, Carlsbad, CA, United States; #A27977). Then, the cell suspension was diluted with PBS to get a final cell concentration of 2.5 x 10 7 cells/mL. Finally, surface markers were labeled with fluorochrome-conjugated antibodies and the cell populations were analyzed using flow cytometry.

Supplementary Table 4. Summary of splenic CD8+ T cells.
WT