Absence of XRCC4 and its paralogs in human cells reveal differences in outcomes for DNA repair and V(D)J recombination

Highlights

• Generation of viable human somatic cells lacking all 3 XRCC4 paralogs.

• Demonstration that the absence of PAXX has little impact on C-NHEJ in human cells.

• Demonstration that XRCC4-deficient cells rely almost exclusively on microhomology-mediated DSB repair.

• Demonstration that the absence of XRCC4 has a more deleterious impact on C-NHEJ and V(D)J recombination than either the loss of XLF or PAXX.

• Demonstration that the absence of XRCC4 has no effect on the frequency of gene targeting.

Abstract

The repair of DNA double-stranded breaks (DSBs) is an essential function performed by the Classical Non-Homologous End-Joining (C-NHEJ) pathway in higher eukaryotes. C-NHEJ, in fact, does double duty as it is also required for the repair of the intermediates formed during lymphoid B- and T-cell recombination. Consequently, the failure to properly repair DSBs leads to both genomic instability and immunodeficiency. A critical DSB protein required for C-NHEJ is the DNA Ligase IV (LIGIV) accessory factor, X-Ray Cross Complementing 4 (XRCC4). XRCC4 is believed to stabilize LIGIV, participate in LIGIV activation, and to help tether the broken DSB ends together. XRCC4′s role in these processes has been muddied by the identification of two additional XRCC4 paralogs, XRCC4-Like Factor (XLF), and Paralog of XRCC4 and XLF (PAXX). The roles that these paralogs play in C-NHEJ is partially understood, but, in turn, has itself been obscured by species-specific differences observed in the absence of one or the other paralogs. In order to investigate the role(s) that XRCC4 may play, with or without XLF and/or PAXX, in lymphoid variable(diversity)joining [V(D)J] recombination as well as in DNA DSB repair in human somatic cells, we utilized gene targeting to inactivate the XRCC4 gene in both parental and XLF⁻ HCT116 cells and then inactivated PAXX in those same cell lines. The loss of XRCC4 expression by itself led, as anticipated, to increased sensitivity to DNA damaging agents as well as an increased dependence on microhomology-mediated DNA repair whether in the context of DSB repair or during V(D)J recombination. The additional loss of XLF in these cell lines sensitized the cells even more whereas the presence or absence of PAXX was scarcely negligible. These studies demonstrate that, of the three LIG4 accessory factor paralogs, the absence of XRCC4 influences DNA repair and recombination the most in human cells.

Introduction

The accurate repair of DNA DSBs is absolutely necessary for cellular survival and genomic stability. Albeit rarer than most other types of DNA damage, DSBs can be caused not only by exposure to exogenous agents such as ionizing radiation (IR) or chemotherapeutic compounds, but they are also spontaneously created by endogenous processes [1] including errors of DNA replication [2] as well as during V(D)J recombination and class switch recombination, the latter two of which are required for lymphogenesis [3].

DSBs are especially pathologic because of their ability to induce chromosomal translocations [4]. In order to protect the integrity of the genome from such occurrences, mammalian cells utilize two main pathways to repair DSBs: homology dependent recombination (HDR) and NHEJ. The repair of DSBs by the HDR pathway is predominately error-free, but requires a non-damaged template from which to enact repair and thus is usually only active in the late S or G2 phases of the cell cycle when an undamaged sister chromatid may be available as a donor template [5]. Because most of the cells in a human being are, however, non-cycling and thus never in S or G2 of the cell cycle, NHEJ is per force the main pathway used for the repair of DSBs [6]. NHEJ consists of at least two genetically and mechanistically distinct pathways: C-NHEJ, the predominant pathway, and Alternative-EJ (aEJ), which appears primarily to be a back-up pathway [7]. The key proteins involved in C-NHEJ include the Ku (Ku70 and Ku86) heterodimer, the catalytic subunit of the DNA-dependent protein kinase complex (DNA-PK_cs/PRKDC), Artemis, XRCC4, XLF/Cernunnos (hereafter XLF), and LIGIV. For simple DNA DSBs, the DNA ends can likely be rejoined by LIGIV alone [8,9]. However, if either one or both of the ends requires processing first, then the entire complement of C-NHEJ factors is utilized in a three-step process involving: i) recognition of the DSB, ii) processing of the DNA ends, and ultimately iii) DNA ligation [6]. The first step consists of detection of the DSB by the Ku (Ku70/Ku86) heterodimer. Once bound to the ends, Ku recruits DNA-PK_cs leading to the formation of the DNA-PK holoenzyme that, upon autophosphorylation [10], undergoes a conformational change that allows for the recruitment of the DNA end-processing enzymes including Artemis, polynucleotide kinase/phosphatase, and/or DNA polymerases. Finally the XRCC4:LIGIV complex ligates the processed broken ends back together.

In mammalian cells, in addition to DNA DSB repair, C-NHEJ is also required for V(D)J recombination. V(D)J recombination is a process used by the immune system to generate functional B-cell (immunoglobulin) and T-cell receptors [3,11]. Briefly, in lymphoid progenitor cells, variable (V), diversity (D) and joining (J) segments are excised from three loci on chromosomes 2, 14 and 22 and enzymatically recombined to generate genes capable of encoding functional immunoglobulin or T-cell receptor proteins. V(D)J recombination is a site-specific process that is facilitated by recombination signal sequences (RSS), which are comprised of conserved heptamer and nonomer sequences separated by non-conserved spacer sequences of either 12 or 23 bp. During V(D)J recombination the proteins encoded by the recombination activating genes-1 and -2 (RAG1 and RAG2) form the RAG complex that introduces nicks between individual V, D, or J coding sequences at the heptamer junction of the 12- and 23- signals. This RAG-mediated cleavage produces two blunt 5’-phosphorylated signal ends and two covalently-sealed (hairpin) coding ends. Subsequent joining of the signal ends and coding ends by C-NHEJ forms both signal and coding joints. Although C-NHEJ is required for V(D)J recombination, individual factors in the pathway can preferentially influence the formation of either coding or signal joints. For example, both DNA-PK_cs and Artemis are required significantly more for coding, rather than signal, joint formation due to their hairpin opening activities [12]. In addition, ablating XRCC4′s affinity for XLF results in a reduction of coding joint but not signal joint formation [13].

XRCC4 was first characterized almost 25 years ago after it was identified as the factor that complemented the DSB repair deficiencies in a hamster IR hypersensitive mutant cell line, XR-1 [14,15]. Although XRCC4 has no apparent enzymatic activity, it is required for the stabilization and activation of LIGIV [16,17]. A complete loss-of-function of XRCC4 appears to be exceedingly rare and has only been reported for a handful of patients afflicted with microcephalic primordial dwarfism [18,19]. Relevantly, these patients (in contrast to, say, LIGIV patients) do not have a recognizable immunological phenotype indicating that genetic redundancy for XRCC4′s V(D)J recombination activity must exist in vivo. Importantly, polymorphisms within the XRCC4 gene have been identified as causing susceptibility to a bevy of cancers including bladder [20], breast [21], prostate, hepatocellular carcinoma, lymphoma and multiple myeloma [22]. XLF/Cernunous was identified through its association with patients exhibiting developmental anomalies, such as microcephaly and (unlike XRCC4) immunodeficiency, as well as through its interaction with XRCC4 [23,24]. XLF and XRCC4 share similar structural features including an N-terminal head domain and a C-terminal coiled-coiled domain that is required for homodimerization [[25], [26], [27]]. Importantly, cells lacking XLF exhibit impaired V(D)J recombination using either plasmid substrates [23,24,28] or chromosomal loci [29,30].

XRCC4 and XLF, besides homotypically interacting, can also interact with each other to form a filamentous complex that extends along DNA. These filaments are thought to bridge separate DNA molecules independently of LIGIV. It has been suggested that these filaments enhance the ligation of DSBs by forming a scaffold that assists in synapsis of the broken ends [[31], [32], [33], [34], [35], [36]]. This modestly well-understood (albeit hypothetical) mechanism was significantly complicated by the discovery of a third XRCC4-like paralog, PAXX [[37], [38], [39]]. PAXX appears to interact more with Ku than with either of its paralogs [37,38], but the role of this protein in C-NHEJ and how functionally redundant it is with either XRCC4 or XLF — especially in human cells — is still unclear although it seems likely that it is not via filament formation [40].

In order to get a better understanding of how XRCC4 and its paralogs function in NHEJ, and to examine how their loss affects DSB repair in general, we used recombinant adeno-associated virus (rAAV)-mediated gene targeting to create both XRCC4^−/− and XRCC4^−/−:XLF^−/− human HCT116 cell lines. Here we show that the absence of XRCC4 leads to several pronounced phenotypes including increased sensitivity to the DNA-damaging agents etoposide and IR, a decrease in overall DSB repair with a concomitant increase in the use of aEJ. Moreover, we show that cells lacking XRCC4, XLF or both factors, are not only unable to complete V(D)J recombination efficiently, but result in disparate recombination configurations. Finally, we utilized clustered regularly interspersed palindromic repeats (CRISPR)/CRISPR-associated 9 (Cas9)-mediated gene editing to generate HCT116 cells lacking all three XRCC4 paralogs (XRCC4^−/−:XLF^−/−:PAXX^−/−). These cells were viable and demonstrated that, in human somatic cells, the presence or absence of PAXX seems to have no detectable effect for C-NHEJ.