V(D)J recombination process and the Pre-B to immature B-cells transition are altered in Fanca−/− mice

B-lymphocytes in the bone marrow (BM) must generate a functional B-cell receptor and overcome the negative selection induced by reactivity with autoantigens. Two rounds of DNA recombination are required for the production of functional immunoglobulin heavy (Ig-HCs) and light (LCs) chains necessary for the continuation of B-lymphocyte development in the BM. Both rounds depend on the joint action of recombination activating gene-1 (RAG-1) and RAG-2 endonucleases with the DNA non-homologous end-joining pathway. Loss of the FANC gene leads to the chromosome breakage and cancer predisposition syndrome Fanconi anemia. Because the FANC proteins are involved in certain aspects of the recombination process, we sought to determine the impact of the FANC pathway on the Ig diversification process using Fanca−/− mice. In this work we demonstrated that Fanca−/− animals have a mild B-cell differentiation defect characterized by a specific alteration of the IgM− to IgM+ transition of the B220low B-cell population. Pre-B cells from Fanca−/− mice show evidence of impaired kLC rearrangement at the level of the Vk-Jk junction. Furthermore, Fanca−/− mice showed a skewed Vκ gene usage during formation of the LCs Vk-Jk junctions. Therefore, the Fanca protein appears as a yet unidentified factor involved in the primary diversification of Ig.

To cope with the enormous variety of pathogens and to recognize non-self molecules, B-cells have evolved controlled genetic processes at their immunoglobulin (Ig) loci known as Ig diversification. Primary diversification occurs during early B-cell development in the bone marrow (BM) via the assembly of a complete IgM antigen receptor exposed on the B-cell surface (BCR) by a site-specific recombination process called V(D)J recombination. Mature B-cells that express a functional IgM migrate from the BM to the periphery, where antigen-dependent secondary diversification occurs following two activation-induced cytidine deaminase-dependent processes known as somatic hypermutation and class switch recombination (CSR) 1 .
To produce the Ig heavy chain (HC), V(D)J recombination starts in the BM at the pro-B cell stage by the D-to-J H rearrangement followed by the V H -to-DJ H rearrangement. Productive HC rearrangement leads to IgM-HC expression. After the assembly of the IgM-HC with a surrogate light chain (LC) and CD79a and b proteins, the IgM-HC is exposed on the cell surface as the precursor-B cell receptor (pre-BCR). Signals from the pre-BCR orchestrate the proliferation and subsequent developmental transition to the small pre-B-cell stage, where Igκ or Igλ LC VJ recombination is initiated 2,3 . Successful pairing of a productive LC with an IgM-HC results in the expression of a BCR at the cell surface and progression to immature B cells, which are checked for autoreactivity before leaving the BM 4 . V(D)J recombination depends on the action of the lymphoid-specific RAG-1 and RAG-2 endonucleases that initiate DNA cleavage at defined recombination signal sequences (RSS) that flank the V, D, and J gene segments. The RAG complex mediates the formation of two hairpinned extremities, called coding ends (CEs), cutting-off a DNA segment creating an one-ended blunted DSB at each extremity, at the signal end 5,6 . The signal ends of a DNA fragment are sealed by non-homologous end joining (NHEJ), and the formed circle is displaced. The two hairpins are opened by the endonuclease Artemis in association with activated DNA-PKcs 7 and joined by the complex formed by XRCC4, LIG4 and XLF/Cernunnos. Due to the intrinsic error prone property of NHEJ, the obtained coding joints frequently lose and/or gain nucleotides 8 . While the loss of nucleotides is a consequence of the 5′ and 3′ overhang modification by Artemis 7 , a nucleotide gain results from the template-independent activity of the terminal deoxynucleotidyl transferase (TdT) DNA polymerase (N nucleotide additions) or from the activities of DNA polymerases operating on the hairpin that has been opened asymmetrically (P nucleotide additions) [9][10][11][12][13] . The junction of regions from the V, D and J segments encodes the CDR3, the major determinant of the antigen binding site specificity. V(D)J is a highly regulated process that ensures the development of a normal immune system and prevents potential oncogenic events such as translocations, during the sealing step of the CE.
Fanconi anemia (FA) is a rare inherited disorder characterized by chromosome breakage, cancer predisposition and BM failure 14,15 . The syndrome is genetically heterogeneous, and twenty FANC genes (named A to U) have been identified to date 14,16,17 . The major and most robust role of the FANC pathway is its involvement in the DNA damage response. Following DNA damage or replicative stress, eight upstream FANC proteins (FANCA, FANCB, FANCC, FANCE, FANCF, FANCG, FANCL, and FANCM) assemble into the "FANCcore complex", which, together with FANCT/UBE2T, is necessary for the monoubiquitination and nuclear foci formation of both FANCD2 and FANCI. The monoubiquitinated FANCD2/FANCI heterodimer functionally and/or biochemically interacts with the downstream FANC proteins FANCD1/BRCA2, FANCN/PALB2, FANCJ/BRIP1, FANCO/ RAD51C, FANCP/SLX4, FANCQ/XPF, FANCR/RAD51, FANCS/BRCA1 and FANCU/XRCC2 to eliminate DNA lesions and/or to rescue stalled replication forks 14,18,19 . Accumulating evidence indicates that FA proteins function to coordinate DNA double-strand breakage repair activity by regulating homologous recombination and/or NHEJ [20][21][22] . Studies using model organisms, cells and cellular extracts have revealed that altered DNA end joining activities occur in the absence of a FANC protein 21,23,24 . Recently, we have shown that Fanca is required during CSR to stabilize duplexes between pairs of short microhomology regions located at DNA ends 25 . Because DNA end-joining activities are a prerequisite for primary immunoglobulin diversification, we reasoned that the FANC pathway could be involved in these mechanisms. To address this question, we analysed B-cell development and V(D)J recombination in Fanca −/− mouse-derived B-cells.

Results
Impaired IgM − to IgM + transition in bone marrow lymphoid cells isolated from Fanca −/− mice. To determine whether Fanca is involved in the mechanism of V(D)J recombination in B cells, we first analysed lymphoid tissue development in the BM and spleen of 8-week old Fanca −/− and WT mice. Flow cytometry analysis of the BM showed no quantitative difference in the proportion of total B cells (B220 + ), pro-B cells (B220 + IgM − CD43 high ), and pre-B cells (B220 + IgM − CD43 low ) between Fanca −/− and WT mice (Fig. 1A,B).
However, we observed a significantly increased ratio of IgM − (IgM − B220 low ) to immature B cells (IgM + B220 low ) in Fanca −/− mice (3.5 in Fanca −/− mice vs. 2.9 in WT, Fig. 1C,D), suggesting that the transition from IgM − to immature B cells was somewhat defective. Indeed, the Fanca −/− mice had a mild accumulation (15.9% vs 15.2% of total lymphocytes) in BM B cells at the IgM − stage and consequently showed a reduction in the proportion of immature B cells (Table S1). However, the proportion of mature, recirculating B cells (IgM + B220 high ) was unchanged in the absence of Fanca (Table S1), indicating that subsequent maturation steps in the IgM + compartment are unaffected.
Consistent with the absence of differences in mature B cells compartment in the BM, an analysis of total B cells in the spleen showed that the percentage of B cells was also similar between WT and Fanca −/− mice (Fig. 1E,F). Impaired P-addition process during HC rearrangement in Fanca −/− mice. To obtain a better characterization of the transition from the pro-B (the phase where HC rearrangement occurs) to the pre-B stage in Fanca deficient cells, we addressed V(D)J rearrangement efficiency by a PCR-based assay using V H and J H4 consensus primers to amplify HC-rearranged junctions in DNA isolated from Fanca +/+ and Fanca −/− BM B220 + IgM − cells. The rearranged junctions were analysed for length, size range and junctional diversity of CDR3. Our data showed that the average length, size range and length distribution of CDR3 of rearranged VDJ H4 genes as well as the proportion of in-frame (vs. out-of-frame sequences were similar between Fanca −/− and WT mice ( Table 1 and Fig. 2A). A determination of the length of each V, D, and J segment contributing to the CDR3 region showed no considerable difference between WT and the Fanca −/− mice (Table S2). Additionally, we found that the average number of N-additions, whether estimated for both the V H -D and D-J H junctions (total) or for either the V H -D or the D-J H junction alone appeared to be similar between the two groups of mice (Fig. 2B). On the contrary, P-additions at the V H -D or D-J H junctions differ significantly between Fanca −/− and WT mice (Fig. 2C). Surprisingly, whereas the ratio of P-additions at D-J H vs V H -D junctions was similar in WT cells (0.86+ /− 0.2), a significant disequilibrium was observed in Fanca −/− cells (ratio of 3.11+ /− 0.8). In other words, in Fanca −/− B cells, we observed 3 times more sequences with P-additions at D-J H junctions than at V H -D.
Consequently, even if our data demonstrated that Fanca −/− mice are competent in the transition from pro-B to pre-B stage, they uncover a still undetermined role for Fanca during the early step of V(D)J rearrangement of the HCs. Fanca −/− mice harbour a defect in the pre-B to immature B-cell transition (Fig. 1C,D), a step that requires a successful LC rearrangement, we hypothesized that Fanca could be specifically involved in the regulation of VJ recombination at the LC locus. To test this hypothesis, we first assessed the Igκ rearrangement efficiency in WT and Fanca −/− BM IgM − B cells using PCR. A degenerate Vκ primer (V d κ Vκ D) that binds to ∼ 90% of Vκ gene segments was used together with a primer downstream of Jκ 5 (Jκ ; Fig. 3A). A genomic sequence within the Scientific RepoRts | 6:36906 | DOI: 10.1038/srep36906 murine DLG5 gene was used to normalize the DNA input. The intensities of PCR bands for Vκ -Jκ rearrangements were comparable in IgM − B cells from WT and Fanca −/− mice (Fig. 3B) indicating that a Fanca deficiency does not detectably affect the Igκ LC recombination step. We next sequenced and analysed the Vκ -Jκ 1 (proximal) and Vκ -Jκ 4 (distal) rearranged junctions from BM-sorted B220 + IgM − cells that had not yet produced a functional B cell receptor. The Vκ -Jκ 1 and Vκ -Jκ 4 junctions represented the primary and secondary rearrangements during Igκ LC recombination, respectively. A V d κ primer and a downstream primer Jκ 1 or Jκ 4 were used to amplify the Vκ -Jκ 1 or Vκ -Jκ 4 junctions, respectively (Fig. 3A). Rearranged junctions were analysed for length, size range and junction diversity of CDR3. Our results showed that the average length and size range of CDR3 of both the Vκ -Jκ 1 and the Vκ -Jκ 4 rearrangements in BM B220 + IgM − cells were similar between Fanca −/− and WT mice (Table 1).  Nevertheless for the Vκ -Jκ 1 rearrangement, we observed that the ratio of in-frame vs. out-of-frame sequences was 0.47 for WT mice and 0.92 for Fanca −/− mice. In other words, 32% of the analysed CDR3 sequences from WT animals in IgM − B220 + B-cells are in-frame compared with 48% in Fanca −/− mice, whereas no difference was noticed for the Vκ -Jκ 4 rearrangements (Fig. 3C). The increased proportion of in-frame Vκ -Jκ 1 sequences in Fanca −/− mice could simply be a reflection of the accumulation of the IgM − B-cells that we previously observed. However, only the Vκ -Jκ 1 rearrangements appear unbalanced, suggesting a specific role of Fanca in their joining. A noticeable effect of a Fanca deficiency is evident on the histogram in Fig. 3D, which shows the observed CDR3 sizes. The canonical length of 27 nucleotides was observed in 40% of the Vκ -Jκ 1 CDR3 from Fanca −/− mice compared with 24% retrieved from WT. On the other hand, the distribution of CDR3 lengths in Vκ -Jκ 4 rearrangements was similar between two groups of mice (Fig. 3E). Further analysis of the Igκ rearranged junctions, indicated that even if the global frequency of sequences showing nucleotides addition was similar (less than 15%) for the Vκ -Jκ 1 rearrangement between WT and Fanca −/− mice, the average of N addition was significantly higher in Fanca −/− mice and the proportion of P-additions vs. N-additions was clearly reversed (Fig. 3F). Again as a supplementary clue of specificity, the relative proportion of P-additions vs. N-additions in the Vκ -Jκ 4 junctions was similar between the WT and Fanca −/− mice (Fig. 3G).
Because N-nucleotide additions depend on TdT activity, the observed excess of N-additions in the Vκ -Jκ 1 rearrangements in Fanca −/− mice suggests that the absence of Fanca stimulates TdT action or expression/stabilization. Accordingly, even if Fanca-deficient pro-B and pre-B cells showed similar levels of TdT mRNA (Fig. 4A), pro-B cell population from Fanca −/− mice express significantly more protein that their Fanca-proficient littermates, as determined by flow cytometry ( Figure S2) and shown in Fig. 4B. Our analysis indicates that Fanca-deficient pre-B cells present a residual level of TdT expression slightly more elevated than in Fanca-proficient cells. Although the differences between the two genotypes is not statistically significant, it could affect N-additions in the Vκ -Jκ 1 rearrangements in pre-B cells.  Table 1). Collectively, these data indicate that, during Vκ -Jκ 1 recombination in IgM − B-cells, Fanca loss-of-function specifically results in the accumulation of both in-frame rearrangements and N-nucleotide additions.  [26][27][28] (Fig. 5A). Remarkably however, with respect to in-frame Vκ -Jκ 1 rearrangements, we noticed the presence of a higher than expected proportion of Vκ 1, 2, 8, 21 and 23-family junction in Fanca −/− mice. Furthermore, the Vκ 1 family was the most significantly increased in Fanca −/− mice. Interestingly the Vκ 8 Vκ 21 Vκ 23 gene families are located less than 1.0 Mb from Jκ 1, whereas Vκ 1 is located more than 2 Mb away (Fig. 5C). A similar analysis of Vκ gene usage in  Vκ -Jκ 4 rearrangements failed to show differences with respect to both distance and Vκ family usage between WT and Fanca −/− mice ( Figure S3) further supporting the specificity of the previous observation.
Altogether, our findings indicate that the absence of Fanca specifically leads to altered Vκ gene usage in the in-frame Vκ -Jκ 1 rearrangements.

Discussion
In this study, we used Fanca −/− mice to investigate a potential function for the FANC pathway in V(D)J recombination. We found that the absence of Fanca leads two subtle but consistent molecular abnormalities during the process of both HC and LC formation. Whereas the first observed molecular alteration occurred during HC formation with no impact in the pro-B to pre-B transition, the second is associated to a defect in pre-B to immature B-cell transition. However, because of the selection process to which B-cells are subjected to become fully competent, the observed alterations seems have only a modest impact on Ig diversity and functionality.
During the process of HC formation, two rounds of rearrangement follow one to another to allow, first, the junction of a D sequence with a J H sequence and, second, the joining of a V H sequence with the rearranged DJ H sequence. Following the RAG-mediated hairpin formation at CE sequences, rearrangement proceeds thank to the opening of each hairpin, by the joint action of DNA-PK and Artemis 7 , followed by the remodelling of the opened extremities and their joining. The remodelling of the open extremities eventually leads to N and P nucleotide additions at coding joins. P-additions are the consequence of asymmetric opening of hairpin loops that form at gene ends during the HC rearrangement process. The extension of the ss extremities created by the hairpin opening, thus creating a palindrome of 0 to 4 nucleotides at the end. P nucleotides have been associated with V H , J H and D genes. Unexpectedly, in this work we noticed that the frequency of P-nucleotides addition during the DJ H rearrangement was significantly more elevated in Fanca −/− than in WT cells, whereas the opposite was observed for the V H D rearrangement. Thus, whereas in WT cells the frequency of P-additions is similar at DJ H and V H D, in Fanca −/− cells we found 3 times more P-additions at DJ H than at V H D junctions. An elevated frequency of  Table 1). sequences with P-additions, but at both DJ H and V H D junctions, was previously reported in X-linked anhidrotic ectodermal dysplasia with hyper-IgM syndrome (HED-ID), a rare pathology due to a genetic determined deficiency in NF-kB activation 29 . In HED-ID exacerbated P-nucleotide additions have been directly associated to an altered exonucleolytic processing of the coding ends.
Furthermore, it has been robustly ascertain that cells with a loss-of-function of one FANC protein, in addition to an increase of NF-kB activity 30,31 , generally associate an exacerbated use or activity of the NHEJ pathway [20][21][22] . Indeed, inappropriate recruitment of DNA-PKcs at the site of DSBs was observed in the FANCD2-, FANCC-and FANCA-deficient cell 20,22 . Since the number of nucleotides found at the coding ends is the result of the balance between the nucleotide addition and the level of exonuclease activity that occurred before germline ends joining, we hypothesized that the observed differences in P additions is due to alterations in the action of Artemis or the DNA PKcs-Artemis complex during the opening of the hairpin leading to longer ssDNA extremities, a defect not sufficiently compensated by exonucleolytic activities. Indeed, in addition to the endonucleolytic activity of the DNA PKcs-Artemis complex, Artemis alone display an exonucleolytic activity 7 . Although It is not well known when Artemis is free from DNA-PKcs during the process of V(D)J recombination, one can speculate that in the absence of FANCA and/or in the presence of an increase of NF-kB activity, its activity is selectively reduced during DJ H joing and thus leaving more P nucleotides. Nevertheless, we cannot exclude the possibility that V H DJ H length selection process may also contribute to the length of P and N nucleotide in both D-J H and V H -DJ H ends.
Looking at LC formation, we observed that the rearranged CDR3 show an excess of N-nucleotide additions. (and the recalled parallel deficit in P-nucleotide additions). It is important to note that the structure of the DNA ends created during V(D)J recombination differs markedly from other classical DNA double-strand breaks. Indeed, the V(D)J CEs are protected by a closed hairpin structure. The CE configuration is subject to a concomitant end-processing and sealing process. Furthermore, the end-joining step is accompanied by considerable DNA-end modifications, which contribute to the diversification of the pre-immune repertoire. Following the opening of the hairpin mainly mediated by the endonucleolytic activity of Artemis, the extremities are processed by various polymerases, including the lymphoid-specific enzyme TdT. For B cells, TdT is a key factor in the expansion of the diversity of their repertoire by means of the addition of N-nucleotides onto the ends of the newly formed P-nucleotides. TdT adds an average of two to five nucleotides per N region, and G:C pairs are added more often than A:T pairs 32 . However, the number and sequence of both P-and N-nucleotides that are added vary for each junction. B cells express TdT during the pro-B stage of development, as the heavy chain is produced. N nucleotides are added to both the V-D junctions and the D-J junctions of Ig HC. Once a functional HC is made, the expression of TdT is down-regulated 33,34 . The amount of TdT expressed in vivo correlates with the degree of N region diversity in the antigen receptor. Finally, it is accepted that N nucleotide addition contributes significantly to the diversity of the HC, whereas few LC rearrangements include the addition of N nucleotides. We observed that, while the levels of TdT mRNA in both pro-B and pre-B cells was similar in WT and Fanca −/− mice, the protein level was significantly higher in pro-B cells from Fanca −/− mice compared with WT animals (Figs 4B and S2). Noteworthy, although not significant differences are observed in the aggregate data analysis (Fig. 4B), TdT protein level in Fanca deficient pre-B cells is often slightly more important than in their Fanca-proficient counterpart supporting the possibility that the observed increase in nucleotide N-additions in Fanca −/− mice at this stage is a consequence of this residual higher expression inherited from the higher expression in the earlier pro-B stage. Interestingly, several published works have demonstrated a requirement of the FANC pathway for either the optimal activity or the stabilization of some TLS polymerases 25,35,36 , which suggests a broad, direct or indirect, role for the FANC pathway in the turnover and the management of the DNA polymerase functions that are not involved in bulky DNA replication/repair. We propose that Fanca participates by unknown mechanism to the regulation of TdT turnover in mice pro-B cells. In the absence of Fanca, the half-life of TdT increases and, as a consequence, at the pre-B stage the cells still will display an elevated residual amount of the protein potentially responsible for the increase of N addition during the Vκ -Jκ 1 rearrangement.
We also observed that the frequency of in-frame Vk-Jk1 sequences retrieved from B cells was more important in Fanca −/− mice than in their WT littermates. Additionally, the Vk sequences chosen to join with the Jk1 sequences were generally skewed towards the more proximal Vk sequences (i.e., within less than 1.0 Mb), such as the Vk8, 21 and 23 sequences. The Vk1 sequence is a notable exception: it was more than 2.0 Mb away but was more frequently used (Fig. 5). The first rearrangements at the κ LC locus most often employ the Jκ 1 segment and the closest Vκ segments 37 . Once an in-frame and productive Vκ -Jκ exon has been successfully created, the B-cells express a κ LC that is verified for functionality by pairing with a μ HC, thus forming a BCR that becomes an immature (sIgM + ) B-cell 4 . Our observations suggest that Fanca may be involved in the regulation of the expression and/or pairing of the rearranged LC with a μ HC. Indeed, although in-frame, a high proportion of the Fanca −/− -derived small pre-B cells bearing rearranged LC failed to fully progress to the immature B-cell compartment. The higher frequency of in-frame Vκ -proximal-Jκ 1 junctions (Fig. 3C) supports the hypothesis that even though they are in-frame, the LCs are not expressed or are unable to pair with a μ HC in Fanca-deficient cells. Interestingly, Vk1, which is a distal Vk sequence located more than 2.0 Mb from Jk1, is the most overrepresented between the Vk-Jk1 in-frame retrieved sequences. Interestingly, Vk1 is the only Vk family member that possesses potential NF-κ B binding sites in their intronic regions 38 . In the last few years several studies including our own unpublished observations revealed some as yet poorly understood transcriptional activities of the FANC pathway that could both increase or inhibit the transcription of several genes 39,40 . Interestingly, it was demonstrated that FANCD2 protein can inhibits NF-κ B-dependent transcription through a specific association between monoubiquitinated FANCD2 and a NF-κ B consensus-binding site 39 . Moreover, an aberrant activation of NF-κ B-dependent transcriptional activity has been observed in FA cells 30,31 . Taken together these observations, we speculate that Fanca deficiecy in pre-B cells leads to enhanced NF-κ B mediated transcription activity which in turn deregulate the Vk1 expression and managing. A constitutive expression of this particular rearrangements VJ exons can explain both an accumulation of in frame Vκ -Jκ 1 rearrangement and an increase load of N addition at this Scientific RepoRts | 6:36906 | DOI: 10.1038/srep36906 specific rearrangement. In an alternative but not exclusive manner, the observed abnormality could be associated to the higher than normal NHEJ activity associated to the loss-of-function of the FANC pathway. In particular, we have reported that in absence of FANCcore complex or FANCD2 proteins, 53BP1 accumulate strongly and stay longer than in WT cells to DSBs 22 . Thus, an altered 53BP1 accumulation during the Vk-Jk recombination process could greatly favour the joining of proximal DSB, as observed here. Alternatively, 53BP1 and FANCA could be indirectly involved in this DNA end joining process by facilitating chromosomal accessibility or influencing chromatin organization. Although the exact role of 53BP1 in the absence of FANCA in DNA DSB repair remains to be determined, the two proteins appear to function in coordinating certain aspects of DNA end joining during the light chain rearrangement.
Recently we have shown that the Fanca (and likely the FANC pathway) plays a role, alone or in cooperation with other factors, during the antigen-dependent diversification phase of the Ig genes 25 . In this work we showed that Fanca not only plays a role in the nucleotide addition at the CE via the regulation of TdT protein expression/ stabilization but also point out to the role in the expression of the LC Vk1 family. Nevertheless, because of the huge repertoire of V, (D), and J segments and the selection process to which B cells are subjected to become fully competent, the observed alterations have, at least in mice, only a modest impact on Ig diversity and functionality.

Materials and Methods
Mice. Fanca −/− mice were described previously 25 Fanca +/− mice were backcrossed with WT FVB/N mice (> ten generations). As Fanca −/− mice show severely reduced fertility, WT and Fanca −/− mice used for analysis correspond to siblings derived from crossbreeding of heterozygous mice. The project was officially approved by the Animal Experimentation Ethics Committee of the Gustave Roussy Institute (IGR) and registered under no. 26 by the IGR Department of Research and conducted in accordance with French laws and regulations.
PCR assay for κLC rearrangement. Genomic DNA was isolated from sorted BM B220 + IgM − cells.
DNA was analysed by PCR with different cycles and by Southern blots for Vκ -Jκ rearrangement as previously described 41 .
Sequence analysis of Ig gene rearrangements. Genomic DNA was extracted from sorted BM B220 + IgM − cells. V H DJ H4 , Vκ -Jκ 1 and Vκ -Jκ 4 rearrangements were amplified by PCR as previously described 42 . PCR products were subsequently cloned into the Zero Blunt vector (Invitrogen) and sequenced. All V(D)J recombined products were analysed with the IgBLAST webserver (NCBI).