CCCTC-Binding Factor Locks Premature IgH Germline Transcription and Restrains Class Switch Recombination

In response to antigenic stimulation B cells undergo class switch recombination (CSR) at the immunoglobulin heavy chain (IgH) to replace the primary IgM/IgD isotypes by IgG, IgE, or IgA. CSR is initiated by activation-induced cytidine deaminase (AID) through the deamination of cytosine residues at the switch (S) regions of IgH. B cell stimulation promotes germline transcription (GLT) of specific S regions, a necessary event prior to CSR because it facilitates AID access to S regions. Here, we show that CCCTC-binding factor (CTCF)-deficient mice are severely impaired in the generation of germinal center B cells and plasma cells after immunization in vivo, most likely due to impaired cell survival. Importantly, we find that CTCF-deficient B cells have an increased rate of CSR under various stimulation conditions in vitro. This effect is not secondary to altered cell proliferation or AID expression in CTCF-deficient cells. Instead, we find that CTCF-deficient B cells harbor an increased mutation frequency at switch regions, probably reflecting an increased accessibility of AID to IgH in the absence of CTCF. Moreover, CTCF deficiency triggers premature GLT of S regions in naïve B cells. Our results indicate that CTCF restricts CSR by enforcing GLT silencing and limiting AID access to IgH.

inTrODUcTiOn B cells remodel their immunoglobulin genes first as part of their differentiation program in the bone marrow, and then during the immune response, in the context of germinal centers (1). In the bone marrow, the heavy (IgH) and the light chain (IgL) immunoglobulin genes are subject to V(D)J recombination, a site specific recombination reaction triggered by RAG recombinases (2,3). Subsequently, B cells that have been activated by antigen engage in the germinal center reaction where immunoglobulin genes are further diversified by somatic hypermutation (SHM) and class switch recombination (CSR), both of which are initiated by the activation-induced cytidine deaminase (AID) enzyme (4)(5)(6). SHM introduces (mostly) point mutations in the antigen recognizing, variable region of the IgH and IgL genes and antibody variants with higher affinity for antigen are selected through iterative rounds of mutation, proliferation, and cell death (7). CSR is a region-specific recombination reaction whereby the primary μ constant region (Cμ) is replaced by a downstream constant region (Cγ, Cε, or Cα) in the IgH gene, such that the primary IgM/IgD isotypes switch to IgG, IgE, or IgA (8,9). Highly repetitive sequences called switch (S) regions preceding each constant region are targeted by AID, which triggers the deamination of cytosines followed by the generation of double strand breaks and the recombination reaction. CSR to a particular isotype is determined by stimulation-driven cues that activate specific S region I promoters, inducing their germline transcription (GLT) (10). GLT of both the donor and acceptor S regions is an absolute requirement for CSR, presumably by promoting transcription-coupled AID recruitment (11).
Both V(D)J recombination and CSR are tightly developmentally regulated, are intimately linked to transcription, and involve the formation of long-range DNA loops that facilitate the interaction between distant cis regulatory regions. CCCTCbinding factor (CTCF) is an architectural protein that regulates the genome function by mediating interactions between distant DNA sequences (12,13) and is known to act as a transcriptional insulator through the generation of chromatin loops (14)(15)(16). CTCF regulates bone marrow B cell differentiation at different levels. First, CTCF-deficient pre-B cells show impaired proliferation and differentiation (17). Further, CTCF mediates interactions between distal and proximal regions of the IgH locus, regulating its transcription (18), and CTCF-binding elements (CBEs) are critical to inhibit rearrangement of proximal VH genes and promoting rearrangement of distal VH genes (19,20). Finally, IgH employs CBE-based subdomains to regulate RAG on-and off-target activity (21). Regarding the role of CTCF during the germinal center reaction, we have recently shown that CTCF is an essential transcriptional regulator that allows high proliferation rate of germinal center B cells and represses the expression of Blimp-1, thus preventing plasma cell differentiation (22). However, the role of CTCF in CSR is not completely understood. The IgH locus is flanked by a 3′ regulatory region (3′RR), located 3′ of the most downstream CH, the Cα gene (23,24). The proximal 3′RR contains several B cell-specific enhancers (hs3a, hs1,2, hs3b, and hs4) that are required for IgH GLT, class switching to all isotypes and high levels of Ig transcription in plasma cells (25)(26)(27). The distal 3′RR enhancer region (containing hs5, 6, 7, and 38 enhancers) is densely riveted with CBEs, and deletion of hs5-7 (harboring seven of the nine CBEs at the distal 3′RR) affected IgH locus compaction and VH usage during V(D)J recombination, but did not affect CSR (28). In contrast, a CBE has been very recently identified within the Cα gene whose deletion promotes isotype-specific GLT in developing and resting B cells and alters CSR (29).
Here, we have directly assessed the role of CTCF in CSR using a conditional mouse model for CTCF depletion in B cells. We find that CTCF-deficient B cells show increased rate of CSR to various isotypes and under various stimulation conditions. This effect was not related to any measurable alterations of cell cycle or proliferation and was not a consequence of increased levels of AID expression. Instead, we find that AID activity is quantitatively increased on S regions. Finally, we show that GLT is markedly increased from activation-inducible, acceptor S regions in CTCF-deficient naïve B cells, indicating that CTCF restricts the transcription of activation-inducible S regions and limits AID accessibility and CSR.

Mice and immunization
CCCTC-binding factor-deficient mice were obtained by breeding CTCF fl/fl mice (30)  T-dependent immunization was induced in 6-11 weeks CTCF fl/fl ; CD19-Cre ki/+ or CTCF fl/+ ; CD19-Cre ki/+ mice by intravenously injection of 10 8 sheep red blood cells resuspended in 100 µl of sterile PBS. Immunization response was analyzed in spleen 7 days after sheep red blood cell injection. Number of animals per group to detect biologically significant effect sizes was calculated using appropriate statistical sample size formula and indicated in the biometrical planning section of the animal license submitted to the governing authority.

next-generation sequencing
Analysis of Sμ mutations was performed as previously described (32). Briefly, DNA was extracted from LPS/IL-4 stimulated B cells from CTCF fl/fl ; CD19-Cre ki/+ (n = 3) or CTCF fl/+ ; CD19-Cre ki/+ (n = 3) mice. Sμ fragment was PCR amplified using the oligonucleotides (forward) 5′-AAT GGA TAC CTC AGT GGT TTT TAA TGG TGG GTT TA-3′ and (reverse) 5′-GCG GCC CGG CTC ATT CCA GTT CAT TAC AG-3′. Amplification reactions were carried out with 2.5 U of Pfu Ultra (Stratagene) in a 50-μl reaction with the following profile: 94°C for 5 min followed by 25 cycles at 94°C for 10 s, 60°C for 30 s, and 72°C for 1 min. PCR products from four independent reactions per mouse were pooled and NGS sequencing was performed by the Genomics Unit at CNIC. Briefly, Libraries were prepared using NEBNext Ultra DNA Library Prep (New England Biolabs) following manufacturer's instructions and sequenced on a HiSeq2500 (Illumina).

statistics
Statistical analyses were performed with GraphPad Prism (version 6.01 for Windows, GraphPad Software, San Diego, CA, USA) using two-tailed Student's t-test for all parameters conforming to normal distributions (according to Shapiro-Wilk normality test). Variance similarity was assessed with F test. p ≤ 0.05 was considered statistically significant. Error bars in figures represent SD.

resUlTs
To address the function of CTCF during CSR we generated a conditional mouse model to deplete CTCF late during B cell differentiation. CTCF fl/fl mice (30) were bred to CD19-Cre ki/+ , where the Cre recombinase is progressively expressed in the bone marrow and allows full depletion of floxed alleles at late stages of B cell differentiation (31). CTCF fl/fl ; CD19-Cre ki/+ (CTCF-deficient mice, hereafter CTCF fl/fl ) and CTCF fl/+ ; CD19-Cre ki/+ littermates (control mice, hereafter CTCF fl/+ ) were used in all experiments, since we have previously shown that CTCF is not haploinsufficient in mature B cells (22). We first analyzed the effect of CTCF depletion in Peyer's patches, where germinal centers are constitutively generated in response to antigens of the gut microbiota, and found that CTCF fl/fl Peyer's patches are almost completely devoid of germinal center B cells ( Figure 1A). To analyze the role of CTCF in de novo germinal center formation, we immunized CTCF fl/fl and CTCF fl/+ mice with sheep red blood cells and analyzed spleen populations by flow cytometry 7 days after immunization. We found that both Fas + GL7 + germinal center B cells and CD138 + plasma cells were virtually absent in CTCF fl/fl mice (Figure 1B), further indicating that CTCF is absolutely required for germinal center formation, and concomitantly, for plasma cell generation. In vitro stimulation of spleen B cells in the presence of LPS and IL-4 revealed that CTCF fl/fl cells had a major B cell survival defect as measured by Annexin V staining ( Figure 1C) and by viability stain exclusion ( Figure 1D). Thus, these results suggest that CTCF is required for germinal center formation at least in part by enabling B cell survival.
We have previously generated a conditional mouse model where CTCF depletion is induced only upon AID expression [CTCF fl/fl ; AID-Cre TG/+ mice (22)], i.e., delayed in comparison with the model presented here, where naïve B cells are already devoid of CTCF ( Figure 1E). Consistently with the findings reported here, in CTCF fl/fl ; AID-Cre TG/+ mice CTCF is also essential for the germinal center reaction (22). In contrast, B cells from the AID-Cre and the CD19-Cre models widely differ in their response to in vitro stimulation: while CTCF depletion in CTCF fl/fl ; AID-Cre TG/+ B cells did not lead to any measurable defect upon CSR stimulation by LPS + IL-4 (22), CTCF fl/fl ; CD19-Cre ki/+ B cells display a clear survival defect. Besides the functional implications of this disparity (discussed below), the CTCF fl/fl ; CD19-Cre ki/+ model opens the possibility to study the function of CTCF in CSR, which could not be accurately addressed in CTCF fl/fl ; AID-Cre TG/+ mice.
To assess the role of CTCF in CSR we stimulated spleen B cells from CTCF fl/fl and CTCF fl/+ mice in vitro in the presence of LPS and IL-4 and measured CSR to IgG1 by flow cytometry analysis. We found that CTCF-deficient cells consistently showed increased rate of CSR at all analyzed time points (Figures 2A,B). To determine whether the enhancement of CSR in CTCF-deficient cells was restricted to the IgG1 isotype or to the specific conditions of LPS + IL-4 stimulation, we stimulated CTCF fl/fl and CTCF fl/+ B cells with anti-CD40 + IL-4, LPS, or TGFβ + anti-CD40 + IL-4, which preferentially trigger the activation of Iγ1, Iγ3, and Iγ2b S region promoters and the switch to IgG1, IgG3, and IgG2b, respectively. All three stimulations resulted in a notable increase in CSR in CTCF fl/fl cells, indicating that CTCF deficiency broadly enhances the rate of CSR, regardless of the activation stimulus or the particular S region being used in the recombination reaction ( Figure 2C). Thus, our results indicate that CTCF restricts the extent of CSR during B cell activation.
Class switch recombination is tightly associated with proliferation (9,33), such that B cells that have undergone more cell divisions accumulate a bigger fraction of isotype-switched cells. To investigate whether the increased rate of CSR could be secondary to an alteration in cell proliferation, we first performed cell cycle analysis of CTCF fl/fl and CTCF fl/+ B cells stimulated with LPS + IL-4. We could not detect any alteration in the proportion of cells in G1, S, or G2/M as measured by BrdU incorporation (Figure 3A). The rate of cell proliferation analyzed by CFSE dilution and the proliferation index were also indistinguishable between CTCF fl/fl and CTCF fl/+ LPS + IL-4 stimulated B cells (Figures 3B,C). Further, we measured the rate of CSR to IgG1 as a function of the number of cell divisions, and we found that CTCF fl/fl cells display an enhanced CSR frequency compared to CFCF fl/+ cells, regardless of the number of divisions they have undergone ( Figure 3D). Together, these observations indicate that the increased CSR rate observed in CTCF-deficient B cells is not a collateral effect of cell proliferation alterations and instead it suggests that it could be intrinsically related to the molecular mechanism of CSR.
Activation-induced cytidine deaminase initiates CSR by deaminating cytosines on the IgH S regions, and AID levels roughly correlate with the efficiency of CSR (34). To address whether CTCF regulates the expression of AID during B cell activation, CTCF fl/fl and CTCF fl/+ B cells were stimulated with LPS + IL-4, and AID levels were assessed by qRT-PCR and western blot analysis. We found that CTCF deficiency did not significantly affect the expression of AID at the mRNA or protein level (Figures 4A,B).
A fraction of AID-induced deaminations on S regions are processed into DNA double strand breaks, while others give rise to mutations that can be used to track AID activity (33,(35)(36)(37). Thus, we next compared the mutation frequency in Sμ of CTCF fl/fl and CTCF fl/+ B cells stimulated with LPS + IL-4 by PCR-Seq (32). We found that Sμ regions of CTCF fl/fl B cells accumulated a significantly higher frequency of mutations than CTCF fl/+ B cells (Figure 4C). In both CTCF-proficient and -deficient B cells, mutations clustered preferentially at WRCY/ RGYW AID mutational hotspots (38) (Figure 4C) and extended throughout the length of the sequenced region ( Figure 4D; Table  S1 in Supplementary Material). These findings indicate that AID-induced mutation frequency at S regions is quantitatively increased in CTCF fl/fl B cells, suggesting that CTCF deficiency could enhance AID activity by facilitating its access to S regions at IgH. In agreement with this idea, we found that the density of CTCF-binding to the different CBEs interspersed at the IgH locus was altered in LPS/IL-4 stimulated B cells as compared with naïve B cells (39) (Figure 4E). Specifically, we found that several CBEs, such as the 5′hs1R1 site (29,39), and peaks 3 and 7 at the 3′RR ( Figure 4E) were more densely bound by CTCF in resting than in activated B cells, while we found no difference or the reverse trend for other CBEs. This analysis indicates that during B cell activation, binding of CTCF to the IgH is rearranged, suggesting that CTCF could be an important factor involved in chromatin remodeling of IgH during B cell activation.
To gain further insights into the role in CTCF in regulating IgH accessibility, we measured GLT at donor and acceptor S regions in CTCF-proficient and -deficient B cells. GLT is absolutely required for CSR and is widely considered an indicator of S region accessibility (10,11). While Sμ is transcribed constitutively, GLT at acceptor S regions is induced only upon appropriate stimulation of I promoters. We measured GLT at S regions in naïve and stimulated B cells and found that Sμ transcription was not affected in CTCF fl/fl B cells compared to CTCF fl/+ B cells, regardless of their resting or activated state ( Figure 5A). In contrast, GLT expression of Sγ1 acceptor region was significantly increased in naïve B cells, but not in activated B cells ( Figure 5B). Accordingly, GLT-γ3 and GLT-γ2b also showed a clear increase in CTCF fl/fl naïve B cells when compared to CTCF fl/+ B cells ( Figure 5C). Together, these results show that CTCF prevents GLT of S regions prior to B cell activation, and that CTCF deficiency unleashes transcription from the I promoters in naïve B cells.

DiscUssiOn
In this work, we have assessed the role of CTCF in CSR making use of a conditional mouse model for CTCF depletion in mature B cells (CTCF fl/fl ; CD19-Cre ki/+ ). We previously explored the function of CTCF in germinal center B cells using a different mouse model (CTCF fl/fl ; AID-Cre TG/+ ), which promotes CTCF depletion only after B cell activation and induction of AID expression (22). Thus, in the case of CTCF fl/fl ; CD19-Cre ki/+ mice, the levels of CTCF are dramatically reduced already in naïve B cells (Figure 1E), while in CTCF fl/fl ; AID-Cre TG/+ mice CTCF depletion is delayed and dependent on B cell activation and thus naïve B cells have normal CTCF expression (22). Interestingly, these mouse models have revealed important functional differences upon B cell activation. CTCF fl/fl ; AID-Cre TG/+ B cells stimulated in the presence of LPS and IL-4 do not have a detectable phenotype, while CTCF fl/fl ; CD19-Cre ki/+ B cells show a compromised cell viability and an increase in CSR. This phenotype is supported by the finding that under these conditions, . Thus, on one hand, these alternative genetic models have unveiled distinct contributions of CTCF at early and late stages of B cell activation, which will be further investigated. On the other hand, the CTCF fl/fl ; CD19-Cre ki/+ model has established an important role of CTCF in the regulation of CSR, the focus of this study.
We have found here that CTCF depletion in CTCF fl/fl ; CD19-Cre ki/+ B cells increases the rate of CSR to all isotypes tested, and it does so without significantly altering B cell proliferation rate. This is an interesting observation because it is a first hint that, regardless of the potentially broad functional consequences of CTCF depletion, CTCF fl/fl ; CD19-Cre ki/+ B cells may have an intrinsic CSR defect. In search for this defect we first assessed AID levels and found that CTCF has not a noticeable role in the regulation of AID expression. In contrast, we found that the mutation frequency at Sμ is significantly increased in the CTCF fl/fl ; CD19-Cre ki/+ B cells, suggesting that the accessibility of AID to the switch regions is enhanced in the absence of CTCF. In agreement with this hypothesis, we find that B cell activation  results in a redistribution of CTCF-binding along the IgH locus, and that GLT is increased in CTCF-deficient naïve B cells.
Although we cannot rule out that CTCF could also restrict CSR by the transcriptional silencing of factors, other than AID, important to trigger GLT or to process AID-induced lesions, our data strongly suggest that CTCF restricts GLT and CSR by its interaction with CBEs found in the IgH locus. In this regard, our findings are reminiscent of recent data that made use of a complementary approach (29). Specifically, Khamlichi and coworkers generated mice with a germline deletion of a DNAseI hypersensitivity site along with a CBE (5′hs1R1) and found activation of GLT at various S regions in naïve B cells and isotype-specific alterations of CSR after B cell activation. These data suggest that 5′hs1R1 contains a CBE acting as a CTCF insulator that precludes premature activation of I promoters in S regions (29). Likewise, our data strongly suggest that CTCF binds to CBEs at IgH, preventing GLT prior to B cell stimulation. Dissecting the contribution of specific CBEs and CTCF-mediated chromatin loops to particular CSR phenotypic effects is an essential issue remaining. First, it will be important to precisely define the CTCF-dependent loops taking place before and after B cell activation by 3C-as previously described for CSR synapsis (40). This should be complemented with functional analysis performing individual and combined deletion of CBEs; this will prove a very challenging task, given the complex array of CBEs at the constant region of IgH-including at least the 5′hs1R1 CBE and nine CBEs downstream of the 3′RR.
The finding that CTCF depletion promotes increase of GLT in resting but not activated B cells is intriguing. We speculate that CTCF contributes to establish a chromatin architecture that keeps the I promoters silent prior to CSR. Unleashing this architectural constraint in naïve B cells by depleting CTCF may facilitate the early loading or pre-poising of CSR cofactors that could allow faster or better loading of AID and additional CSR components at the time of activation. Although this hypothesis would require experimental testing, it agrees with previous observations made upon depletion of the 5′hs1R1 CBE (29).
In sum, our results are compatible with a model where CTCF maintains the IgH locus in a locked architectural conformation in naïve B cells that blocks GLT and AID access to S regions, thus preventing premature CSR.

acKnOWleDgMenTs
The authors thank all members of the B Cell Biology Laboratory, J Méndez and VG de Yébenes for critical reading of the manuscript, F Álvarez-Prado for help with sequence analysis, F Sanchez-Cabo for advise on statistics analysis, and N Galjart and K Rajewsky for kindly providing the CTCF fl/+ and the CD19-Cre ki/+ mice, respectively. AP-G was a fellow of the research training program (FPU-AP2009-1732) funded by the Ministerio de Educación, Cultura y Deporte; EM-Z is a fellow of the research training program (FPI) funded by the Ministerio de Economía y Competitividad (BES-2014-069525); AR is supported by Centro Nacional de Investigaciones Cardiovaculares (CNIC). This work was funded with the following grants to AR: SAF2013-42767-R and SAF2016-75511-R (Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016 Programa Estatal de I+D+i Orientada a los Retos de la Sociedad Retos Investigación: Proyectos I+D+i 2016, Ministerio de Economía, Industria y Competitividad) and co-funding by Fondo Europeo de Desarrollo Regional (FEDER) and the European Research Council Starting Grant program (BCLYM-207844). The CNIC is supported by the Ministry of Economy, Industry and Competitiveness (MEIC) and the Pro CNIC Foundation and is a Severo Ochoa Centre of Excellence (MEIC award SEV-2015-0505).