Non-Homologous End Joining Factors XLF, PAXX and DNA-PKcs Maintain the Neural Stem and Progenitor Cell Population

Non-homologous end-joining (NHEJ) is a major DNA repair pathway in mammalian cells that recognizes, processes and fixes DNA damage throughout the cell cycle and is specifically important for homeostasis of post-mitotic neurons and developing lymphocytes. Neuronal apoptosis increases in the mice lacking NHEJ factors Ku70 and Ku80. Inactivation of other NHEJ genes, either Xrcc4 or Lig4, leads to massive neuronal apoptosis in the central nervous system (CNS) that correlates with embryonic lethality in mice. Inactivation of either Paxx, Mri or Dna-pkcs NHEJ gene results in normal CNS development due to compensatory effects of Xlf. Combined inactivation of Xlf/Paxx, Xlf/Mri and Xlf/Dna-pkcs, however, results in late embryonic lethality and high levels of apoptosis in CNS. To determine the impact of NHEJ factors on the early stages of neurodevelopment, we isolated neural stem and progenitor cells from mouse embryos and investigated proliferation, self-renewal and differentiation capacity of these cells lacking either Xlf, Paxx, Dna-pkcs, Xlf/Paxx or Xlf/Dna-pkcs. We found that XRCC4-like factor (XLF), DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and paralogue of XRCC4 and XLF (PAXX) maintain the neural stem and progenitor cell populations and neurodevelopment in mammals, which is particularly evident in the double knockout models.


Introduction
Double-strand DNA breaks (DSBs) are common DNA damage events that threaten the stability of our genome. DSBs can be repaired by homologous recombination (HR), classical non-homologous end-joining (C-NHEJ, or simply NHEJ) and alternative end-joining (A-EJ, also known as backup end joining, or microhomology-mediated end joining) [1][2][3][4]. HR is only available during S/G2 cell cycle phases of the cell cycle when the sister chromatid is accessible and then used as a template. C-NHEJ acts throughout the entire cell cycle, sealing directly the broken ends and is the predominant repair pathway in mammalian cells [3,5]. A-EJ is often microhomology-mediated and more obvious in the absence of classical NHEJ [6].
C-NHEJ involves recognition of the DSBs by Ku70/Ku80 heterodimer (Ku), which in turn recruits DNA-dependent protein kinase catalytic subunit (DNA-PKcs) to form a DNA-PK holoenzyme complex that protects free DNA ends. Assembly of DNA-PK triggers the autophosphorylation of DNA-PKcs, as well as DNA-PKcs-dependent phosphorylation of multiple other DNA repair factors [3]. Ku facilitates recruitment of NHEJ proteins, such as X-ray repair cross-complementing factor 4 (XRCC4)-like factor (XLF), a paralogue of XRCC4 and XLF (PAXX), and a modulator of retrovirus infection (MRI). Ligation of the All experimental procedures involving mice were performed according to the protocols approved by the Comparative Medicine Core Facility at Norwegian University of Science and Technology (NTNU, Trondheim, Norway). Dna-pkcs +/− [13], Xlf +/− [14], and Paxx +/− [18] mouse models were previously described. Dna-pkcs +/− and Xlf +/− mice were imported from Professor Frederick Alt lab (Harvard Medical School and Boston Children's Hospital, Boston, MA, USA). Paxx +/− mice were generated by Oksenych group (NTNU, Trondheim, Norway).

Mouse Genotyping
A conventional polymerase chain reaction (PCR) was used to determine the mouse genotypes. DNA was isolated from ear punches by incubating overnight at 56 • C with 2% proteinase K in DNA lysis solution, containing 10 mM pH = 9.0 Tris, 1 M KCl, 0.4% NP-40 and 0.1% Tween 20. Next, the samples were heat-treated for 30 Table S5) in a final volume of 25 µL. The PCR product was revealed in a 0.7% agarose gel.

Neural Stem and Progenitor Cell Cultures
NSPCs were cultured as free-floating aggregates, also known as neurospheres [29,30]. Briefly, murine embryos were collected at embryonic day E15.5, the brains were isolated and the cerebellums were removed. Remaining brain parts were mechanically disrupted in proliferation medium, containing Dulbecco Modified Eagle Medium/Nutrient Mixture F12 (DMEM/F12), supplemented with 1% penicillin/streptomycin, 2% B27 without vitamin A (Thermo Fischer Scientific, Waltham, MA, USA), 10 ng/mL epidermal growth factor (EGF) and 20 ng/mL basic fibroblast growth factor (bFGF). The neurospheres were formed and incubated at 37 • C, 5% CO 2 and 95% humidity. The neurospheres were dissociated every seventh day using 0.25% of trypsin in ethylenediaminetetraacetic acid (EDTA), as previously described in Castaneda-Zegarra et al. (2019) and Wang et al. (2010) [29,30]. For more details, see also Supplementary Materials and Methods. The neurospheres from passages 3 to 10 were used in all the experiments.

Proliferation Assay
NSPCs' proliferation rates were analyzed using PrestoBlue™ Cell Viability Assay (Thermo Fisher Scientific, Waltham, MA, USA; A13261) following the manufacturer's protocol and as described in Xing and Oksenych (2019) [31]. Briefly, 0.25% trypsin-EDTA was used to obtain single NSPCs, which were then plated onto a 96-well-suspension plate at 8000 cells per well in proliferation medium and incubated at 37 • C, 5% CO 2 and 95% humidity. At day 3, PrestoBlue™ was added to final concentration of 10% in proliferation medium in each well, and the cells were incubated for 2 h at 37 • C, 5% CO 2 and 95% humidity before measuring the fluorescence intensity using FLUOstar Omega system (BMG Labtech, Ortenberg, Germany), 570 nm. The fluorescence intensity indicates the proportion of live cells. The NSPC proliferation assay was carried out on 6 replicates per clone, in 3 independent experiments.

Self-Renewal Capacity Assay
For self-renewal assay, we followed the protocol described earlier [29]. Briefly, the capacity of neural stem cells to maintain their multipotency ex vivo was assessed by determining the number and two-dimensional size of neurospheres. Dissociated single NSPCs were plated onto 6-well suspension plates containing proliferation medium (day 0). At day 8, images of the entire wells were captured using the EVOS microscope (Invitrogen, Carlsbad, CA, USA). The pictures were analyzed using the ImageJ software (NIH, Bethesda, MD, USA) to obtain the total number of neurospheres per well and size of spheres (pixels, px).

Differentiation Assay
Differentiation was induced in dissociated NSPCs, as described previously [29,30]. Briefly, 25,000 single NSPCs were cultured onto 48-well plates pre-coated with 30 µg/mL poly-D-lysine and 2 µg/mL laminin, with differentiation medium containing NeuroBasal A medium (Thermo Fischer Scientific, USA) supplemented with 1% penicillin/streptomycin, 2% B27, 1% GlutaMAX and 10 ng/mL bFGF (day 0) (also see Supplementary Materials and Methods). On day 5, the differentiated cells were fixed with 4% paraformaldehyde for 15 min at room temperature. Furthermore, immunostaining was performed using antibodies recognizing either the neuron-specific β-III tubulin (Tuj1) or the glial fibrillary acidic protein (GFAP) proteins, to determine neurons and astrocytes respectively after differentiation [29,30]. Briefly, the cells were permeablized with 0.1% Triton X-100 for 30 min, washed 3 times with phosphate-buffered saline (PBS) (Oxoid Limited, Basingstoke, UK), and blocked with 1:2 dilution of blocking solution containing 10% BSA (Sigma, USA), 10% goat serum (Invitrogen, USA) and 0.1% Triton X-100 (Sigma, USA) for an hour, and washed with PBS. Then, the cells were incubated with the indicated primary antibodies in 10% blocking solution for one hour at room temperature and washed with PBS. Next, the cells were incubated for one hour with the secondary fluorescent marker-conjugated antibodies at room temperature and counterstained with 1 µg/mL of 4 6-diamidino-2phenylindole (DAPI, Molecular Probes, Eugene, OR, USA). Images were collected using the EVOS microscope. Positively-stained cells were counted using ImageJ software and presented as a proportion of total cells normalized to WT control.

Western Blot
Western blots were performed using antibodies against XLF, PAXX, DNA-PKcs, and β-actin (Supplementary Materials and Methods) [18,31,32]. Neurospheres were collected and lysed with radioimmunoprecipitation assay buffer (RIPA, Sigma, USA) containing cOmplete™ EDTA-free Protease Inhibitor Cocktail (Roche, Branchburg, NJ, USA) and 1 mM phenylmethanesulfonyl fluoride (PMSF, Sigma, USA). Protein concentrations were determined by Bradford assay (Biorad, Hercules, CA, USA). Further, 40 µg of protein from each clone was analyzed by the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred to the membranes using XCell II™ Blot Module (ThermoFisher Scientific, Waltham, MA, USA) at 4 • C. Then, the membranes were blocked with 5% milk in PBS with 10% Tween 20 (PBST) for one hour at room temperature. Primary antibodies were incubated overnight at 4 • C, rinsed with PBST 3 times for 5 min and incubated with the secondary antibodies for one hour at room temperature. The blot was washed and incubated with SuperSignal™ West Femto (Thermo Fischer Scientific, USA) to reveal the proteins with ChemiDoc™ Touch Imaging System (BioRad, USA).

Statistical Analysis
To analyze the data, we pulled together two clones per genotype, representing an independent mouse embryo each, and performed three independent experiments with every clone. All the data shown were normalized to WT average levels. To find statistical differences among the genotypes, Kruskal-Wallis test with Dunn's multiple comparisons test, as a non-parametric alternative of one-way ANOVA, was used. The statistical analyses were performed using GraphPad Prism 7.03 software (GraphPad Prism, La Jolla, CA, USA) [29,30].
Detailed statistical analysis for the relative neurosphere count per well (self-renewal), Figure 2C. WT vs. By analyzing the neurosphere cultures, we observed that the average proliferation rates of Xlf −/− Paxx −/− and Xlf −/− Dna-pkcs −/− double knockout neurospheres were reduced when compared to WT and single-deficient Xlf −/− , Dna-pkcs −/− or Paxx −/− neurospheres ( Figure 2B). To quantify the self-renewal capacity of neurospheres, we plated 10,000 NSPCs and counted the formed neurospheres at day 8 in culture ( Figure 2C). Inactivation of Xlf resulted in 20% reduction and inactivation of Paxx resulted in a 40% reduction of neurosphere count when compared to WT controls. Combined inactivation of Xlf and Paxx resulted in about 80% reduction of neurosphere count ( Figure 2C), further highlighting the severe neurological phenotype of Xlf −/− Paxx −/− mice observed in vivo [16,17,19]. Surprisingly, inactivation of Dna-pkcs resulted in a higher number of viable neurospheres, although of smaller size. Combined inactivation of Xlf and Dna-pkcs resulted in neurosphere count similar to WT controls. We concluded that inactivation of Xlf and Paxx affected self-renewal capacity and viability of NSPCs ( Figure 2C).
To determine neurosphere growth rate, we used an alternative quantification based on the image size in pixels (px) ( Figure 2D). Inactivation of Xlf, Dna-pkcs, or both Xlf/Dnapkcs, resulted in neurospheres with 30% to 50% reduction in size when compared to WT controls. Inactivation of Paxx did not affect the size of neurospheres in WT and Xlf -deficient backgrounds ( Figure 2D). We concluded that both XLF and DNA-PKcs support growth of NSPCs in neurospheres.

Impact of XLF, PAXX, and DNA-PKcs on Differentiation Capacity of Neural Stem and Progenitor Cells
To determine whether XLF, PAXX, and DNA-PKcs affect neural differentiation capacity, single NSPCs (25,000 cells) were plated on pre-coated 48-well plates and cultured with differentiation medium for 5 days. Neuronal and glial lineages were identified by immunolabeling using markers for early neurons (Tuj1), and astrocytes (GFAP). Inactivation of Xlf, Paxx or Dna-pkcs, and combined inactivation of Xlf /Paxx did not affect early neuronal differentiation based on average proportions of Tuj1-positive cells ( Figure 3A). Combined inactivation of Xlf and Dna-pkcs, however, resulted in two-fold reduced neurodifferentiation capacity of NSPCs ( Figure 3A,C). The proportion of GFAP-positive glial lineage cells increased, although not significantly, when NSPCs were lacking either XLF or PAXX, or both XLF and PAXX ( Figure 3B,D).
Detailed statistical analysis for the neuron differentiation, Figure 3C.  Overall, XLF possesses functional redundancy with PAXX during the NSPC self-renewal, and with DNA-PKcs during cell growth and neuronal differentiation (Figures 2 and 3). Overall, XLF possesses functional redundancy with PAXX during the NSPC selfrenewal, and with DNA-PKcs during cell growth and neuronal differentiation (Figures 2 and 3).
In particular, we found that proliferation of Xlf −/− , Xlf −/− Paxx −/− and Xlf −/− Dnapkcs −/− NSPCs is significantly lower than that in WT cells; proliferation of Xlf −/− Dnapkcs −/− NSPCs is lower than that in Paxx −/− cells ( Figure 2B). However, the difference between the proliferation rates in XLF/PAXX or XLF/DNA-PKcs double-deficient NSPCs is not different from single deficient controls, and proliferation rates of PAXX-or DNA-PKcs-deficient NSPCs are not different from WT controls, suggesting that the effect is XLF-dependent and it is more prominent when two factors are inactivated (either XLF and PAXX or XLF and DNA-PKcs, Figure 2B). It was previously shown that the expression and activity of DNA-PKcs are increased in the NSPC murine cells when compared to mouse embryonic fibroblasts [26], suggesting that DNA-PKcs can be essential for homeostasis of NSPCs. Expression and activity of NHEJ vary in different cell types and require closer attention in future research [33]. Our data further highlighted this observation, and our findings suggest that the DNA-PKcs is required for NSPC proliferation and self-renewal capacities, although its role is partially compensated by XLF ( Figure 2B,C).
Self-renewal capacity was significantly affected in NSPCs lacking XLF, PAXX or both XLF/PAXX, based on the cell number. However, inactivation of XLF or both XLF/DNA-PKcs did not change self-renewal of NSPCs when compared to WT controls ( Figure 2C). Size of neurospheres is significantly reduced when XLF, DNA-PKcs or both XLF/DNA-PKcs are inactivated ( Figure 2D). Moreover, lack of DNA-PKcs resulted in neurospheres of smaller size when compared to XLF-or PAXX-deficient samples ( Figure 2D).
Finally, differentiation of NSPCs towards neurons or astrocytes was not significantly affected when the cells were lacking either XLF, DNA-PKcs, or PAXX ( Figure 3). Nevertheless, we observed a trend towards reduced Tuj1+ neurons in cell populations lacking both XLF and DNA-PKcs ( Figure 3A) when compared to WT and single-deficient controls. In addition, we observed a potential trend towards increased GFAP+ astrocytes in cell populations lacking either XLF, or PAXX, or both XLF and PAXX, when compared to WT controls ( Figure 3B).
Overall, our data suggest that PAXX is required for self-renewal of NSPCs, while DNA-PKcs is essential for cell proliferation and growth of neurospheres ( Figure 2B-D). None of the NHEJ proteins studied (XLF, DNA-PKcs, PAXX) was essential for the NSPC differentiation to neurons or astrocytes. Further analyses of early neurodevelopment in vivo and in vitro will help to reveal new insights regarding the role of NHEJ factors in neurodevelopment. Double-and multiple-knockout genetic models will facilitate these studies unraveling functional redundancy between the DNA repair factors.
Here, for the first time, we directly addressed the roles of NHEJ factors XLF, PAXX and DNA-PKcs in the processes of proliferation, self-renewal and differentiation of NSPCs. The roles of XLF and PAXX in neurodevelopment were previously characterized using murine embryonic brains, and the mouse embryos lacking both XLF and PAXX possessed increased neuronal apoptosis when compared to single-deficient and WT controls [16,17,19]. However, no previous studies involved NSPCs lacking XLF, PAXX or both XLF and PAXX. Furthermore, combined inactivation of XLF and DNA-PKcs resulted in ablated NHEJ and perinatal lethality of XLF/DNA-PKcs double deficient mice [27]. It was later suggested that the mice lacking both XLF and DNA-PKcs possess, among others, neurological defects [1,3]; however, no direct tests of this option were previously published. NSPC models were used to demonstrate that NHEJ factor XRCC4 as well as DNA damage response factor ataxia telangiectasia-mutated (ATM) are required for neuronal development, e.g., likely to repair DSBs, including ones in the long neural genes [34][35][36][37]. The DSB repair and DNA damage signaling were later more generally suggested to be required for healthy brain development, using macrocephaly and autism spectrum disorder-derived NSPCs [38]. Finally, it was recently demonstrated that regional regulator Filia is required for neurodevelopment. In this study, both murine hippocampus and the NSPCs' research models were used [39].
Although mice lacking XLF possess normal CNS development [14,15], human patients with mutations in Cernunnos/XLF gene suffer from neurological defects, in addition to immunodeficiency [40,41]. The difference between human and murine phenotypes might be related to the fact that multiple NHEJ and DNA damage response factors, e.g., ATM and H2AX [42], MDC1 [43], 53BP1 [44,45], DNA-PKcs [27], PAXX [2,16,17,19,20], MRI [2,21] and RAG2 [46], partially compensate for the lack of XLF in mice. In other words, XLF compensates for the lack of multiple factors, including DNA-PKcs and PAXX. Our recent observations revealed that DNA-PKcs and PAXX are likely in the same sub-pathway of NHEJ because Dna-pkcs −/− Paxx −/− mice and human cells do not possess any additional phenotype when compared to the Dna-pkcs −/− or Paxx −/− mice and cells [20,31]. In particular, human HAP1 cell lines lacking both DNA-PKcs/PAXX possess the same levels of genomic instability and sensitivity to DNA damage-inducing agents etoposide, doxorubicin and bleomycin as DNA-PKcs-deficient ones [20,31]. Moreover, mice lacking both DNA-PKcs and PAXX are live-born, fertile and do not show any additional phenotype when compared to immunodeficient Dna-pkcs −/− knockout mice [20].
An important question challenging current research is: what is the mechanism underlying genetic interaction between Xlf and other NHEJ and DNA damage response (DDR) factors in cells? As one option, it was suggested that XLF can be complementary to the second factor having an alternative function (e.g., structural vs. enzymatic functions in XLF/ATM or XLF/DNA-PKcs double mutants). Alternatively, the proteins can be completely redundant key structural components of NHEJ machinery (in XLF/DNA-PKcs, XLF/MRI, XLF/PAXX, XLF/MDC1, XLF/53BP1, and XLF/RAG double mutants) [1,3,4]. Several models have been proposed to explain the DSB repair via NHEJ. A two-stage model, where DNA ends are first tied in a complex [47]. More specifically, first, Ku70-Ku80-DNA-PKcs (DNA-PK holoenzyme) form the initial complex, where DNA ends are sufficiently distant. Next, a larger complex is formed by DNA-PK, XLF, LIG4 and XRCC4 [47]. More recently, the XLF was proposed to be a flexible connector in this model, by interacting with both Ku70/Ku80 and XRCC4/LIG4/DSB [48].
Another model suggests that there are two major structural complexes formed during the NHEJ. Ku70-Ku80-XRCC4-LIG4 form the flexible synaptic complex (FS). Next, DSBs are brought together through interaction of XRCC4, giving rise to two Ku-XRCC4-Lig4-DNA complexes. Both XLF and PAXX are required for transition from the flexible synaptic to the second synaptic complex, or close synapsis (CS). Here, XLF stabilizes the close synapsis to a greater extent. In this model, it was suggested that DNA-PKcs is not involved in the FS and CS formation [49,50]. Moreover, the structural roles of MRI [21] and PAXX [19] were shown in the NHEJ, explaining the functional redundancy of these factors with XLF.

Conclusions
XLF is functionally redundant with PAXX during the neuronal stem and progenitor cells self-renewal and proliferation, and with DNA-PKcs during cell growth and neuronal differentiation. The NHEJ factors DNA-PKcs, PAXX and XLF are required for efficient early-stage development of neuronal stem and progenitor cells in mice. Additional NHEJ factors, such as MRI/Cyren, Ku70, Ku80, XRCC4 and LIG4, as well as multiple ATMdependent DDR factors might have similar functions in neurodevelopment. Future studies will directly address the roles of NHEJ factors, including XLF, DNA-PKcs, PAXX and MRI, in learning, memory and mood regulations.