Dynamic O-GlcNAcylation governs long-range chromatin interactions in V(D)J recombination during early B-cell development

Abstract

V(D)J recombination secures the production of functional immunoglobulin (Ig) genes and antibody diversity during the early stages of B-cell development through long-distance interactions mediated by cis-regulatory elements and trans-acting factors. O-GlcNAcylation is a dynamic and reversible posttranslational modification of nuclear and cytoplasmic proteins that regulates various protein functions, including DNA-binding affinity and protein–protein interactions. However, the effects of O-GlcNAcylation on proteins involved in V(D)J recombination remain largely unknown. To elucidate this relationship, we downregulated O-GlcNAcylation in a mouse model by administering an O-GlcNAc inhibitor or restricting the consumption of a regular diet. Interestingly, the inhibition of O-GlcNAcylation in mice severely impaired Ig heavy-chain (IgH) gene rearrangement. We identified several factors crucial for V(D)J recombination, including YY1, CTCF, SMC1, and SMC3, as direct targets of O-GlcNAc modification. Importantly, O-GlcNAcylation regulates the physical interaction between SMC1 and SMC3 and the DNA-binding patterns of YY1 at the IgH gene locus. Moreover, O-GlcNAc inhibition downregulated DDX5 protein expression, affecting the functional association of CTCF with its DNA-binding sites at the IgH locus. Our results showed that locus contraction and long-range interactions throughout the IgH locus are disrupted in a manner dependent on the cellular O-GlcNAc level. In this study, we established that V(D)J recombination relies on the O-GlcNAc status of stage-specific proteins during early B-cell development and identified O-GlcNAc-dependent mechanisms as new regulatory components for the development of a diverse antibody repertoire.

Introduction

B lymphocytes, also called B cells, develop in the bone marrow, reside in secondary lymphoid organs such as the lymph nodes and spleen, and are crucial players in the adaptive immune system, particularly with respect to humoral immunity [1]. The humoral immune response produces antibodies against pathogens, including bacteria, viruses, and other foreign or harmful substances, providing long-term immunity [2]. Each B-cell expresses a unique B-cell receptor (BCR) on its surface, consisting of membrane-bound immunoglobulins (Igs) that serve as antigen-binding components. Upon antigen recognition by the BCR, B cells differentiate to secrete specific antibodies [3]. The human immune system can produce an extraordinarily diverse range of antibodies through several mechanisms, including gene rearrangement, somatic hypermutation, and class-switch recombination [4].

During B-cell development, genes encoding Ig heavy (H) and light (L) chains are activated by sequential gene rearrangement of the variable (V), diversity (D), joining (J), and constant (C) region gene segments. This process, known as V(D)J rearrangement, is mediated by recombinases encoded by the recombination-activating gene (RAG) and nonhomologous end-joining proteins [5, 6]. Ig genes constitute one of the largest multigene family loci, spanning approximately 3 megabases (MBs) on mouse chromosome 6 or 12 for the Igκ L chain or IgH chain, respectively [7]. Two key mechanisms, chromatin looping and locus contraction, are necessary for long-range interactions during V(D)J rearrangement [4]. Chromatin looping brings distant gene segments into spatial proximity by forming physical contacts between regulatory elements. Locus contraction is a higher-order chromatin organization process that brings gene segments closer within a confined nuclear space, allowing increased interaction probability. These spatial reorganizations contribute to the long-range interactions necessary for efficient V(D)J rearrangement and help generate a diverse repertoire of B cells capable of recognizing a wide range of antigens [8, 9].

Several key proteins facilitate long-range interactions between distant gene segments to enable the recombination process during V(D)J rearrangement [4]. Pax5 is a B-cell commitment factor that participates in B-cell development by positioning chromatin toward the recombination machinery [10]. Pax5 deletion results in a defective interaction between Pax5-activated intergenic repeats (PAIRs) in the distal V_H gene and the intronic enhancer Eµ site in the 3ʹ-proximal region of IgH, reducing IgH locus contraction [11]. CCCTC-binding factor (CTCF) is a highly conserved DNA architectural protein involved in chromatin loop formation. Many CTCF-binding elements are distributed throughout V(D)J recombination loci, and CTCF-mediated chromatin loops facilitate the accessibility and recombination of gene segments during V(D)J rearrangement [12, 13]. Cohesin is a protein complex with multiple subunits, including SMC1, SMC3, Rad21, and SA1/2, and is essential for sister chromatid cohesion during cell division [14]. The cohesin complex can also form loops via chromatin extrusion through its ring-like structure and contributes to the spatial organization of the V, D, and J gene segments, facilitating the recombination process [12, 15]. YY1, a transcription factor, also enables locus contraction of the IgH gene by bridging the intronic enhancer Eµ with the overall V_H gene loci [16].

O-GlcNAcylation is a dynamic and reversible posttranslational modification (PTM) in which a glycosyl group, N-acetylglucosamine (GlcNAc), is attached to serine or threonine residues of proteins. This modification plays a significant role in signal transduction, gene expression, and protein function [17, 18]. O-GlcNAcylation can impact protein–protein interactions by modulating the affinity or stability of protein complexes. This, in turn, enhances or disrupts protein–protein interactions, thereby influencing various cellular processes [19]. For example, O-GlcNAcylation at Ser430 stabilizes PGC1α but prevents its interaction with PPARγ, affecting the expression of gluconeogenic genes typically upregulated by the PGC1α–PPARγ complex [20]. SOX2, a transcription factor that regulates pluripotency in embryonic stem cells, decreases its interaction with PARP1 through O-GlcNAcylation at Ser248 [21]. O-GlcNAcylated p65, a major component of the NF-κB family, dissociates from the inhibitory protein IκBα, increasing NF-κB transcriptional activation [22]. Furthermore, O-GlcNAcylation affects the DNA-binding properties of proteins and regulates the binding affinity of transcription factors, thus modulating gene expression and chromatin organization [23]. Elevated glucose concentrations in the pancreatic beta cell line MIN6 increased O-GlcNAc modification, enhanced the DNA-binding activity of PDX-1 to the A3 box in the insulin promoter region and promoted insulin secretion [24]. Ser100 O-GlcNAcylation reduces HMGB1 binding to damaged DNA, resulting in error-prone processing during DNA repair [25]. Overall, O-GlcNAcylation is a regulatory mechanism that can fine-tune protein–protein interactions and DNA-binding properties, thereby controlling cellular functions and contributing to various physiological and pathological processes.

As described above, V(D)J recombination generates various combinations through looping structure formation and locus contraction, involving multiple protein interactions. However, the contribution of PTMs to these key proteins remains elusive, despite the long-recognized association of the DNA-binding ability and protein stability of RAG proteins with their phosphorylation status [26]. Therefore, we investigated the effect of O-GlcNAcylation, a type of PTM, on V(D)J recombination in B cells. We revealed a potential defect in V(D)J recombination via the inhibition of O-GlcNAcylation in mice during early B-cell development in the bone marrow. Our results demonstrated that proteins such as SMC1 and SMC3, which are known to participate in chromatin organization, including V(D)J recombination, undergo O-GlcNAcylation. Additionally, reduced O-GlcNAcylation disrupted the protein–protein interactions required for the formation of the cohesin complex. O-GlcNAcylation also regulated the ability of both YY1 and CTCF to bind to the IgH locus. Notably, we described, for the first time, the role of DDX5 in V(D)J recombination; the function of CTCF is modulated through its interaction with DDX5, whose protein stability is controlled by the O-GlcNAcylation status of B cells. Consequently, O-GlcNAcylation is involved in locus contraction and long-range interactions within the IgH gene. The maintenance of O-GlcNAc homeostasis is essential for diverse V(D)J recombinations and serves as a fundamental mechanism for ensuring antibody diversity.

Results

Inhibition of O-GlcNAcylation affects V(D)J recombination of the IgH gene in mouse B cells

V(D)J recombination provides antibody diversity by forming a rosette-like 3D chromatin structure at the Ig gene locus, facilitated by interactions between trans-acting factors bound to specific cis-acting regulatory sequences [4, 27]. However, the effects of posttranslational modification (PTM) on these proteins during V(D)J recombination remain unidentified, with no studies on O-GlcNAcylation. To elucidate the functional consequences of O-GlcNAc modification during V(D)J recombination of Ig genes, we first applied siRNA targeting OGT to early B cells (CD11b^–IgM^–B220⁺) from mouse bone marrow (BM) (Fig. 1A). V_H genes are classified into approximately 16 families on the basis of sequence similarity and are organized into distinct domains within the V_H region [28]. To analyze the IgH gene repertoire, we performed quantitative PCR using genomic DNA with degenerate primers specific to various V_H gene families and primers targeting the J_H1 region to assess recombined DNA. As a result, the knockdown of OGT reduced the cellular O-GlcNAcylation level (Fig. 1B–D), and under these conditions, we observed an approximately 33% decrease in the recombination of the J558 family, which is the most distant from the D_H-J_H recombinant gene, compared with normal levels (Fig. 1E). To verify these findings further in vivo, we cloned and sequenced V(D)J rearrangement fragments from the same early B cells of bone marrow (BM) from mice exposed to the OGT inhibitor OSMI-1 for 10 days (Fig. 1F). OSMI-1 administration also significantly decreased overall O-GlcNAc levels in total BM and early B cells (Fig. 1G, H). Notably, within the J558 family, the overall usage of V_H genes decreased in the presence of OSMI-1 (Fig. 1I). In addition, the expression of the V_H gene belonging to the Gam3.8 or 7183 families also slightly changed (Fig. 1I). Since the distal J558 family alone includes nearly half of the total V_H genes [28], our analyses of those three representative families suggested that low O-GlcNAcylation may have a significant negative impact on IgH gene recombination.

Fig. 1.

Inhibition of O-GlcNAcylation reduces the usage of highly utilized V_H genes during V(D)J rearrangement in early B cells. A Experimental design for targeting OGT with siRNA in early B (CD11b^–IgM^–B220⁺) cells isolated from mouse bone marrow. B Representative western blot showing changes in OGT protein and O-GlcNAc levels upon OGT knockdown in early B cells. C, D Quantification of the band intensities in (B). Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. E Quantitative PCR assay showing the relative gene usage of each V_H gene family in early B cells transfected with or without OGT-targeting siRNA. The data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against the constant region (C_μ) of IgH. F Experimental design for the administration of OSMI-1 (OGT inhibitor) to mice. G Representative western blot showing changes in O-GlcNAc levels in total bone marrow (BM) cells and early B (CD11b^–IgM^–B220⁺) cells from mice intraperitoneally injected with OSMI-1 (1 mg/day). H Quantification of the band intensities in (G). The data are presented as the mean ± S.D. (n = 7 total BM cells and n = 5 early B cells; two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. I The V_H-J_H1 rearrangement products for J558, Gam3.8, and 7183 of early B cells from mice injected with or without OSMI-1 were amplified from genomic DNA via PCR and subsequently cloned and inserted into the pCR^®2.1-TOPO vector. More than 50 independently determined V_H gene sequences from each group were identified via IgBLAST. Heatmap representation of the number of each V_H gene clones

To globally evaluate the sequence information utilized for rearrangement throughout the IgH gene locus, we conducted high-throughput genome-wide translocation sequencing-adapted repertoire sequencing (HTGTS-Rep-seq) [29] using genomic DNA from pro-B cells (IgM⁻B220⁺CD43⁺) extracted from mouse BM or resting B cells (CD11b⁻CD43⁻) from the spleen in the presence or absence of OSMI-1 injection. Consistent with previous analyses (Fig. 1), the J_H1 gene was used as bait to analyze the pattern of the V_H genes recombined with J_H1. As a result, under conditions of O-GlcNAc inhibition in mice, as confirmed in Fig. 1F–H, significant changes in the recombination of the IgH gene were observed in both pro-B and splenic resting B cells, such as a decrease in the overall frequency of V_H gene usage (Fig. 2A, B). In terms of the genes showing the highest usage in pro-B cells, IGHV1-50 was reduced by approximately 5-fold in pro-B cells because of the inhibited O-GlcNAcylation, and it was also decreased by 2-fold in splenic resting B cells. In the case of IGHV1-15 and IGHV1-19, their usage in pro-B cells was reduced by approximately 9-fold and 4-fold, respectively, and this reduction was maintained at approximately 2-fold in splenic resting B cells (Fig. 2A–C). These results suggest that the impaired recombination of specific V_H genes in pro-B cells caused by the abnormal reduction in O-GlcNAcylation may also extend to splenic resting B cells. However, owing to the lack of understanding of the effects of O-GlcNAcylation on other mechanisms that could influence the antibody repertoire beyond V(D)J recombination during B-cell development, further research is necessary. Despite the overall decrease in the recombination of highly utilized V_H genes in pro-B cells, the increase in the recombination of alternative V_H genes appeared to be limited (Fig. 2A). No clear compensatory mechanism was observed within chromosome 12, where the IgH gene is located (Fig. S1). In fact, in mouse pro-B cells, the ratio of J_H1 recombined within the V_H region accounted for only ~12% of the total recombination events, which decreased approximately 3-fold upon O-GlcNAc inhibition, with most of the recombination events shifting to non-V_H regions (Fig. 2D). In contrast, in splenic resting B cells, the ratio of J_H1 recombined within the V_H region was ~24%, which was 2-fold greater than that in pro-B cells, although this ratio also decreased by more than 30% under reduced O-GlcNAcylation conditions (Fig. 2D, E). The difference in the proportion of V_H recombination between pro-B and splenic resting B cells may suggest that during B-cell maturation, allelic exclusion ensures the selection of functional recombination within the V_H region, eliminating non-V_H recombined clones [30, 31].

Fig. 2.

Suppression of O-GlcNAcylation through OGT inhibitor administration in mice leads to an overall decrease in V_H gene usage during V(D)J rearrangement. A–E Results of HTGTS-Rep-seq using J_H1 as bait in pro-B cells (IgM^-B220⁺CD43⁺) of the bone marrow or resting B cells (CD11b^-CD43⁻) of the spleen from mice with or without OSMI-1 injection for 10 days. Bar plots showing changes in V_H genes recombined with D_H-J_H1 across the IgH locus depending on O-GlcNAc levels in pro-B cells (A) or splenic resting B cells (B). C Heatmap representation comparing the usage changes of the top 5 most frequently utilized V_H genes in pro-B cells, on the basis of normal O-GlcNAcylation levels, with the results from splenic resting B cells (the read counts for each gene are normalized to the total count). D, E Pie chart revealing the overall proportion of the V_H region that recombined with D_H-J_H1 in pro-B cells (D) or splenic resting B cells (E)

On the other hand, in the Igκ light-chain genes, the quantity of previously most commonly used Vκ genes in early B (CD11b^–IgM^–B220⁺) cells of mouse bone marrow (BM) decreased; however, no major defects were observed regarding position-specific changes or diversity because less frequently used nearby genes were replaced instead. (Fig. S2). These results clearly demonstrate that the normal and efficient operation of V(D)J recombination at the IgH gene locus highly depends on O-GlcNAc levels in B cells.

Co-administration of an OGA inhibitor improved V_H gene usage in diet-restricted mice

O-GlcNAcylation functions as a nutrient sensor, responding to various metabolites derived from carbohydrates, amino acids, fatty acids, or nucleotide. This process is essential for generating UDP-GlcNAc, the substrate required for O-GlcNAcylation [32]. Consequently, on the basis of the findings presented in Fig. 1, we hypothesized that O-GlcNAcylation might induce changes in V(D)J recombination in a diet-dependent manner in mice. As anticipated, overall O-GlcNAc levels in mouse bone marrow (BM) B cells exhibited dynamic changes depending on the diet (Fig. 3A–C). O-GlcNAcylation significantly increased with a high-fat diet (HFD) compared with a regular diet (RD) and markedly decreased under calorie restriction (CR; 60% calories compared with RD). Additionally, an evident decrease in O-GlcNAc levels was observed during a carbohydrate-deficient ketogenic diet (KD) within a HFD. Following two weeks of exposure to different diet conditions, we conducted quantitative PCR analysis using genomic DNA with the same degenerate primers mentioned earlier to assess the level of V_H-(D)J_H1 gene rearrangement in early B (CD11b^–IgM^–B220⁺) cells sorted from the BM of each mouse. Notably, the usage of V_H genes from the most frequently used families, including J558, Gam3.8, or 7183,  was reduced by nearly 50% under limited dietary conditions, such as CR or KD (Fig. 3D). Since dietary restriction is known to inhibit multiple metabolic pathways simultaneously, adversely affecting immune cell activation [33, 34], it was necessary to verify whether the observed reduction in V(D)J recombination under dietary restriction was truly due to O-GlcNAc suppression. To address this, we established a model by concomitantly administering the OGA inhibitor Thiamet G to CR diet-fed mice to restore O-GlcNAc levels (Fig. 3E‒G). As a result, the reduced V_H gene usage under dietary restriction tended to return to the original levels through the restoration of O-GlcNAcylation (Fig. 3H). Notably, the distal J558 family, which contains the greatest number of V_H genes, was significantly rescued, implying that the impairment in V(D)J recombination can be reversed by the restoration of O-GlcNAc metabolism (Fig. 3H). Collectively, our results demonstrate that suppressed V(D)J recombination can be restored by the reinduction of cellular O-GlcNAc levels, suggesting that the decreased V_H gene recombination during dietary restriction could be attributed to the inhibition of O-GlcNAcylation.

Fig. 3.

Decreased V gene usage during V(D)J recombination upon diet restriction is effectively rescued by co-administration of an OGA inhibitor to mice. A Experimental design for a mouse model subjected to various nutrient stresses. B Representative western blot showing changes in O-GlcNAc levels in early B (CD11b⁻IgM⁻B220⁺) cells from the bone marrow (BM) of each group in (A). C Quantification of the band intensities in (B). Data are presented as the mean ± S.D. (n = 4, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. D Quantitative PCR assay showing the relative gene usage of each V_H gene family in early B (CD11b^–IgM^–B220⁺) cells from mice fed various diets. The data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against the constant region (C_μ) of IgH. E Experimental design for the mouse model in which caloric restriction was applied in the presence or absence of Thiamet G (OGA inhibitor) injection. F Representative western blot showing changes in O-GlcNAc levels in total BM cells from the mice in each group in (E). G Quantification of the band intensities in (F). Data are presented as the mean ± S.D. (n = 4, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. H Quantitative PCR assay showing the relative gene usage of each V_H gene family in early B (CD11b⁻IgM⁻B220⁺) cells from mouse BM. Data are presented as the mean ± S.D. (n = 4, two-tailed Student’s t-test) normalized against the constant region (C_μ) of IgH

Inhibition of O-GlcNAcylation weakens the interaction between cohesin complex subunits

To elucidate the involvement of O-GlcNAcylation in V(D)J recombination, we identified O-GlcNAc-modified proteins expressed in early B (CD11b^–IgM^–B220⁺) cells of the BM, the site of V(D)J recombination. Owing to the relatively low abundance of O-GlcNAcylated proteins in total lysate, wheat germ agglutinin (WGA)-lectin, known for its affinity for GlcNAc, was employed to enrich O-GlcNAc-modified proteins [35]. Following enrichment via WGA-lectin, mass spectrometry (MS) analyses were conducted to identify the captured proteins (Fig. 4A). Gene ontology analysis via WebGestalt on the 88 common proteins derived from two independent MS analyses revealed associations with nucleic acid metabolism and DNA conformational changes (Fig. S3), suggesting that efficient V(D)J combinations may be induced by modifying the DNA structure through O-GlcNAc-targeted proteins expressed in early B cells. Combinatorial experiments, including immunoprecipitation (IP) and western blotting, were conducted to confirm whether these proteins were indeed modified by O-GlcNAcylation. Proteins such as Lamin B1, DDX5, SMC1, and SMC3, which were identified as the top 30 proteins in the two MS analyses (Fig. 4B), were detected in the protein pool collected via WGA-lectin (Figs. 4C and S4A). Importantly, among proteins such as RAG1, E2A, YY1, Pax-5, and CTCF, which were not derived from MS analyses but are well known to play crucial roles in V(D)J recombination [16, 36, 37], YY1 and CTCF were also included in the protein mixture pulled down by WGA-lectin (Fig. 4C). In a crossover experiment, after performing IP with an antibody specific to each protein captured by WGA-lectin, western blotting with the O-GlcNAc-recognizing antibodies CTD110.6 or RL2 confirmed the O-GlcNAc modification of SMC1, SMC3, YY1 (Fig. 4D; first and fifth blots), and CTCF (Fig. S4B). Notably, SMC1, SMC3, and YY1, which were detected via the CTD110.6 antibody, were identified simultaneously on the same membrane and served as negative controls.

Fig. 4.

The core heterodimer between SMC1 and SMC3 in the cohesin complex is dissociated upon the inhibition of O-GlcNAcylation in B cells. A Schematic representation of the experimental design and analysis strategy for the WGA-lectin-based pull-down assay. B List of the top 30 proteins representing the normalized spectrum counts from two independent LC‒MS analyses for protein mixtures from WGA-lectin-based pull-down. Spectrum counts for proteins captured by WGA beads were normalized against those of the control experiment with agarose beads alone. C Representative western blot confirming the presence of each target protein in protein complexes captured by WGA-lectin conjugated to agarose beads in early B (CD11b^–IgM^–B220⁺) cells. D, E Representative western blot following immunoprecipitation (IP) for each target protein to determine O-GlcNAc attachment via the CTD110.6 antibody or to verify the physical interaction between the proteins. Lysates used for IP were prepared from early B (CD11b⁻IgM⁻B220⁺) cells (D) and the PD36 pre-B-cell line (E) and were cultured with or without OSMI-1 treatment for 4 h and 24 h, respectively

Among the O-GlcNAc-modified proteins confirmed here in early B cells, SMC1 and SMC3, the major components of the cohesin complex, play critical roles in regulating V(D)J recombination through IgH loop extrusion [12, 15, 38]. Intriguingly, when O-GlcNAcylation was inhibited, no significant changes in SMC1 expression or immunoprecipitation (IP) efficiency were detected (Fig. 4D; second blot). However, the detection of O-GlcNAcylated SMC1 by the O-GlcNAc antibody was markedly reduced compared with that of SMC3 (Fig. 4D; first blot). Notably, under these inhibitory conditions, the binding affinity of SMC1 for both SMC3, its heterodimer partner within the cohesin complex, and CTCF, its interaction partner, was significantly diminished (Fig. 4D; third and fourth blots and Fig. 4E). In contrast, the interaction between SMC3 and CTCF remained unaffected (Fig. 4D; fourth blot). This suggests that limiting the level of protein O-GlcNAc modification may disassociate SMC1 from the cohesin complex; consequently, the ring-like shape of the complex breaks, reflecting the inability of cohesin-mediated IgH loop extrusion to form. These results indicate that maintaining O-GlcNAc homeostasis is essential for stabilizing the cohesin complex and, in line with the critical role of the cohesin complex in V(D)J recombination, that O-GlcNAcylation is centrally involved in the regulation of V(D)J recombination.

Downregulation of O-GlcNAcylation inhibits the DNA-binding affinity of YY1 for the IgH locus

YY1, a ubiquitously expressed protein, is involved in various essential cellular functions at all stages of B-cell differentiation, particularly mediating long-distance DNA interactions during V(D)J recombination in early B cells [16, 39]. Given that YY1 binding sites are abundant throughout the Ig gene loci [16] and that O-GlcNAcylation can regulate the DNA-binding affinity of its target proteins [23], we hypothesized that O-GlcNAcylation might modulate the DNA-binding properties of YY1 during V(D)J rearrangement. To investigate changes in the YY1 binding pattern at the IgH gene locus, chromatin immunoprecipitation followed by massively parallel DNA sequencing (ChIP-seq) was conducted. Two types of mouse BM B cells―pro-B cells (IgM^-B220⁺CD43⁺) isolated from mice treated with or without the OGT inhibitor OSMI-1 for 10 days and early B cells (CD11b^–IgM^–B220⁺) cultured in the presence or absence of OSMI-1 for 4 h―were utilized for these analyses to examine the effect of O-GlcNAcylation on the ability of the YY1 protein to bind to DNA. In two independent ChIP-seq analyses, a significant correlation was revealed between YY1 peaks from differently sorted B cells, irrespective of OSMI-1 application (Fig. S5A). Although the genome-wide distribution of YY1 binding showed no considerable changes with respect to the impact of O-GlcNAc inhibition (Fig. S5B), zooming in on IgH gene loci revealed a significant overall decrease in YY1 binding (Figs. 5A and S6A). A comparison of YY1 binding patterns in a location-specific manner for the IgH gene region revealed that YY1 was strongly bound to the intronic Eµ enhancer region among the cis-regulatory elements located at the 3ʹ IgH gene locus. However, this interaction was notably reduced under O-GlcNAc restriction (Figs. 5B and S6B). To determine the specific effect of O-GlcNAcylation on the DNA-binding ability of YY1, we mutated the O-GlcNAc site of YY1 (T236A), where threonine at the 236^th amino acid was substituted with alanine (Fig. 5C). According to the results of ChIP‒qPCR for endogenous YY1, strong binding of YY1 to the Eµ region of the IgH locus was observed not only in pro-B cells (IgM⁻B220⁺CD43⁺), where IgH gene recombination occurs but also in early B cells (CD11b^–IgM^–B220⁺), which contain a high proportion of pre-B cells, and in the pre-B-cell line PD36 (Fig. S7). This finding indicates that the Eµ region remains open across all tested cell types. Therefore, we analyzed changes in Eµ binding of FLAG-YY1 in the pre-B-cell line PD36 transiently transfected with plasmids expressing FLAG-tagged wild-type-YY1 (wt-YY1) or mutated-YY1 [mut-YY1 (T236A)]. The promoter region of ribosomal protein L30 (RPL30) was used as a positive control for YY1 binding, and ACTG1 was used as a negative control [40, 41]. As a result, the DNA-binding affinity of exogenously expressed wt-YY1 to the Eµ enhancer region was decreased upon O-GlcNAc inhibition by OSMI-1 treatment (Fig. 5D). Notably, the DNA-binding ability of mut-YY1 (T236A) resulted in only background signals at the level of an isotype-matched control antibody, regardless of the change in O-GlcNAc levels (Fig. 5D). However, the expression level of exogenously expressed FLAG-tagged YY1 was mostly unaffected by mutation of the O-GlcNAc modification site or by the OSMI-1 treatment-mediated reduction in the cellular O-GlcNAc level (Fig. 5C). Therefore, we conclude that the direct DNA-binding patterns of YY1 at the IgH locus during V(D)J recombination are highly dependent on its O-GlcNAc status in B cells.

Fig. 5.

The DNA binding affinity of YY1 and CTCF to the IgH locus is dependent on the cellular O-GlcNAc level. A, B Binding of YY1 and CTCF at the selected genomic loci by ChIP-seq in pro-B (IgM⁻B220⁺CD43⁺) cells from mice with or without OSMI-1 injection. The ChIP-seq data were visualized via the Integrative Genomic Viewer (IGV) browser track. The Y-axis represents the data scale for read counts averaged in each window. A Representative ChIP-seq tracks at the entire IgH locus. B Representative ChIP-seq tracks at specific cis-regulatory elements (CREs) of the IgH locus. Highlighted regions in blue indicate the different binding patterns of YY1 and CTCF according to each CRE (3’RR, Eμ, and IGCR) position, depending on O-GlcNAc level changes. C Schematic diagram for substituting threonine 236, the O-GlcNAc site of YY1, with alanine (upper), and representative western blot showing overexpressed FLAG-tagged YY1 and changes in O-GlcNAc levels in PD36 pre-B cells (lower). D ChIP‒qPCR assay showing changes in FLAG-tagged YY1 occupancy in the intronic Eμ enhancer region with or without OSMI-1 treatment and O-GlcNAc site mutation in PD36 pre-B cells. The promoter region of ribosomal protein L30 (RPL30) was used as a positive control for YY1 binding, and that of ACTG1 was used as a negative control. Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against the negative control (ACTG1)

O-GlcNAcylated DDX5 regulates the DNA-binding properties of CTCF at the IgH locus

CTCF was also identified within a protein complex pulled down by WGA-lectin (Fig. 4C) and was subsequently confirmed, through reciprocal IP experiments, to be directly modified by O-GlcNAcylation (Fig. S4B). Since the role of CTCF in V(D)J recombination through direct binding to the IgH locus is well defined [12, 13], we aimed to determine whether alterations in O-GlcNAcylation would also affect the ability of CTCF to bind to the IgH locus, similar to YY1. Two independent ChIP-seq analyses using the CTCF antibody in pro-B (IgM⁻B220⁺CD43⁺) and early B (CD11b⁻IgM⁻B220⁺) cells revealed a significant correlation between CTCF peaks (Fig. S5A). The genome-wide distribution of CTCF binding also showed no considerable change with respect to the impact of O-GlcNAc inhibition (Fig. S5B). However, upon reduced O-GlcNAcylation, ChIP-seq analysis of CTCF indicated a decrease in the overall distribution of CTCF binding to the IgH locus (Figs. 5A and S6A). In addition, there appears to be a more pronounced trend in which the absolute number of ChIP peaks due to CTCF decreases in the distal V_H region under suppressed O-GlcNAc levels. This finding suggests that altered CTCF-binding patterns may play a crucial role when distal V_H genes are limitedly utilized under O-GlcNAc restriction. On the other hand, SMC3 ChIP-seq showed a relatively lower peak intensity than YY1 or CTCF did, and the weakened binding upon low O-GlcNAcylation was more pronounced at the proximal region than at the distal region, differing from CTCF (Fig. S6A). The binding pattern of CTCF within the cis-regulatory elements located at the 3ʹ side of the IgH gene locus revealed that CTCF was strongly bound to the intergenic control region (IGCR) in the V_H-to-D_H intergenic region and the 3ʹ regulatory region (3ʹRR) at the 3ʹ boundary of the IgH locus in pro-B cells. Notably, CTCF binding to the 3ʹRR was reduced under O-GlcNAc restriction, similar to the V_H gene locus (Fig. 5B and Fig. S6B); however, it remained unaffected in the IGCR (Fig. 5B), suggesting that O-GlcNAcylation may not affect the insulator function of CTCF in pro-B cells.

According to the study by Tang et al., which defined the O-GlcNAcylation site of CTCF, mutation of this site led to an increase in the DNA-binding ability of CTCF [42]. This finding contrasts with our previous findings, which revealed a reduction in CTCF DNA binding under conditions of reduced O-GlcNAcylation. However, the experimental conditions differ between the studies. In our experiments, overall cellular O-GlcNAcylation levels were reduced, whereas the results from Tang et al. were derived solely from the direct mutation of the O-GlcNAcylation site on CTCF. On the basis of the findings from both studies, it is plausible that the intrinsic DNA-binding ability of CTCF may increase under O-GlcNAcylation inhibition. However, given the global effects induced by OGT inhibitor treatment, we hypothesized that the reduction in O-GlcNAcylation across various intracellular proteins may have impacted the access of CTCF to its target DNA via an unidentified cofactor. Although we observed a physical interaction between SMCs and CTCF (Fig. 4D; fourth blot), previous studies reported that the cohesin complex, which includes SMC1 and SMC3, has no significant effect on the DNA-binding ability of CTCF [43, 44]. Among the major proteins detected in the two MS analyses of proteins pulled down by WGA-lectin to identify potential protein targets of O-GlcNAc modification, DDX5 was the only protein (Fig. 4B, C), excluding SMCs, found to have considerable interaction with CTCF (Figs. 6A and S4B) [45]. Therefore, we determined whether O-GlcNAcylation modulates the binding occupancy of CTCF to the IgH locus via DDX5. Interestingly, the protein expression level of DDX5 was strongly decreased by the inhibition of O-GlcNAcylation, but no changes in transcription status were detected (Fig. 6B–D), indicating that O-GlcNAcylation affects the protein stability of DDX5 in B cells (Fig. S8). To investigate the DNA-binding ability of CTCF according to differences in DDX5 expression, a high-density CTCF-binding site, HS5, present in the downstream region of the IgH 3ʹRR, was utilized in a ChIP‒qPCR assay for CTCF. When the expression of DDX5 was suppressed by OSMI-1 treatment, transfection of siRNA against DDX5, or a combination of both (Fig. 6E–G), the enrichment of CTCF binding at the HS5 site was markedly reduced in the PD36 pre-B-cell line, with a synergistic effect observed when the two treatments were combined (Fig. 6H). The same result was observed when DDX5 expression was silenced in early B cells (CD11b⁻IgM⁻B220⁺) derived from mouse BM via siRNA transfection (Fig. 6I-J), which inhibited the DNA binding of CTCF (Fig. 6K). Therefore, our results suggest that the inhibition of O-GlcNAcylation strongly reduces the binding ability of CTCF to the IgH locus, which is, to some extent, attributed to the decreased protein expression of DDX5.

Fig. 6.

O-GlcNAcylation affects the DNA binding ability of CTCF via DDX5 protein stability. A Protein‒protein interaction network analysis was performed via STRING v12 (https://string-db.org/). CTCF, along with the top 30 proteins from the LC-MS results (Fig. 4B), was used as input. Proteins showing physical interactions with CTCF are illustrated in a red dashed box. B Representative western blot showing changes in the DDX5 protein level and O-GlcNAcylation in PD36 pre-B cells with or without OSMI-1 treatment for 24 h. C Quantification of the band intensities for O-GlcNAcylation in (B). Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. D Real-time RT‒qPCR assay for fold changes in the transcription level of DDX5 and quantification of the band intensities for DDX5 in (B) to determine changes in the protein level. Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. ns = not significant. E Representative western blot showing changes in the DDX5 protein level and O-GlcNAcylation level upon DDX5 knockdown in the presence or absence of OSMI-1 in PD36 pre-B cells. F, G Quantification of the band intensities for O-GlcNAcylation (F) and for DDX5 (G) in (E). Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. H ChIP‒qPCR assay using an α-CTCF antibody, showing changes in CTCF binding to the HS5 region in the 3’RR in PD36 pre-B cells. The quantitative data for the enrichment of CTCF with HS5 are shown relative to the input-DNA sample following normalization against the ACTG1 promoter region (n = 3; two-tailed Student’s t-test). The effects of DDX5 knockdown and OSMI-1 treatment on the binding of CTCF to HS5 were compared with those of scrambled siRNA (si-Con) controls in the absence of OSMI-1. I Representative western blot showing changes in the DDX5 protein level upon DDX5 knockdown in early B (CD11b⁻IgM⁻B220⁺) cells from mouse BM. J Quantification of the band intensities for DDX5 in (I). Data are presented as the mean ± S.D. (n = 3, two-tailed Student’s t-test) normalized against β-actin, which was used as a loading control. K ChIP‒qPCR assay using an α-CTCF antibody, which revealed changes in the binding of CTCF to the HS5 region in the 3’RR in early B cells. The quantitative data for the enrichment of CTCF with HS5 are shown relative to the input-DNA sample following normalization against the ACTG1 promoter region (n = 3; two-tailed Student’s t-test). The effects of DDX5 knockdown on CTCF binding to HS5 were compared with those of scrambled siRNA (si-Con) controls

O-GlcNAcylation controls IgH locus contraction and long-range chromatin looping interactions

Our results demonstrated that the DNA-binding abilities of CTCF and YY1, as well as the physical interaction between SMC1 and SMC3, were tightly regulated in an O-GlcNAcylation-dependent manner. These proteins are essential for the formation of the DNA loops required for long-range interactions during V(D)J rearrangement [15, 44, 46]. Therefore, we hypothesized that IgH locus contraction could be disrupted in pro-B cells (IgM⁻B220⁺CD43⁺) from mice administered an O-GlcNAc inhibitor, explaining the markedly altered V_H gene recombination. To investigate this, we performed three-dimensional (3D) DNA fluorescence in situ hybridization (FISH) using bacterial artificial chromosome (BAC)-derived probes corresponding to the 5ʹ and 3ʹ locations in the IgH locus (Fig. 7A). Representative confocal images revealed that pro-B-cell nuclei from normal mice presented a contracted IgH gene structure, in agreement with a previous report [8]. However, the corresponding samples from the mice injected with the OGT inhibitor OSMI-1 presented fewer contraction patterns (Fig. 7B). To quantify these results, we measured the center-to-center distance between two hybridization signals for more than one hundred alleles in these samples. Contraction significantly decreased throughout the IgH locus under low O-GlcNAcylation conditions (Fig. 7C). The average distance between the two signals was 0.43 μm for normal samples and 0.66 μm for O-GlcNAcylation-inhibited samples, representing a 35% decrease in contraction at the whole IgH locus upon O-GlcNAcylation inhibition in mice. The DNA-FISH results suggested a regulatory role for O-GlcNAcylation in long-distance chromosome interactions between the 5ʹ and 3ʹ locations for the IgH locus in mouse pro-B cells. To understand the interaction patterns globally at the IgH locus and identify more specific interconnecting loci, we employed circular chromosome conformation capture and sequencing (4C-seq) with a specific viewpoint, the Eμ enhancer, located in the J_H-C_H intron. The results revealed that long-range DNA interactions throughout the IgH locus were severely decreased in mouse pro-B cells (IgM^- B220⁺CD43⁺) upon O-GlcNAcylation inhibition (Figs. 7D and S9), which was consistent with the 3D DNA-FISH results. Indeed, while the interaction between V_H genes located at a distance from the DJ_H fragment and the Eμ enhancer decreased, the interaction in the proximal region actually increased. According to the ChIP-seq results, YY1 was more strongly bound to the Eµ enhancer than was CTCF, and the binding site occupancy of YY1 to Eµ decreased upon O-GlcNAc inhibition (Fig. 5B). Therefore, the reduction in YY1 binding may have played a leading role in the depletion of long-range interactions with Eµ in our 4C-seq experiment. In conclusion, our data clearly indicate that O-GlcNAcylation controls locus contraction as well as long-range interactions by regulating the DNA-binding properties of multiple factors, including YY1, CTCF, and the cohesin complex, contributing to diverse and efficient V(D)J recombinations.

Fig. 7.

Locus contraction and long-range interaction at the IgH gene locus are reduced upon O-GlcNAc inhibition in pro-B cells. A Map of the IgH locus in mouse chromosome 12 indicating the positions of color-coded BAC probes (Distal: RP23--218B2, Red; and Proximal: RP23--451B13, Green). The center-to-center distance between these two probes in naked DNA is approximately 2.3 Mb. B Representative 3D DNA FISH confocal optical sections of pro-B (IgM⁻B220⁺CD43⁺) cell nuclei from vehicle- or OSMI-1-injected mice. Nuclei outlined by white dashed lines, as identified by DAPI DNA staining (white). Scale bar = 1 μm. C Scatter plot representing the measured center-to-center distance between two hybridizing signals of the IgH locus BAC probes. Data from 135 and 120 pro-B-cell alleles from vehicle- and OSMI-1-injected mice, respectively, were collected from three independent experiments. D 4C-seq profile of the Eμ enhancer via enzyme cleavage by HindIII. Arc plot showing significant interactions between the viewpoint and entire V_H genes on chromosome 12 in mouse pro-B (IgM⁻B220⁺CD43⁺) cells cultured with or without OSMI-1. The positions of the V_H locus and Eμ enhancer are specified above each panel, along with an indication of the orientation for Bait1 (upper) or Bait2 (lower). In each panel, the light purple bar represents the location of the Eμ enhancer on the Arc plot. Each Arc plot was generated by analyzing combined FASTQ files from two biological replicates

Discussion

V(D)J recombination is initiated by the V(D)J recombinases RAG1 and RAG2 via the introduction of DNA double-strand breaks to immunoglobulin loci. Afterward, the breakage is repaired by the ubiquitously expressed NHEJ machinery [6]. For recombination factors, including RAG proteins, a specialized task involving chromatin remodeling is essential for switching from heterochromatin to euchromatin [47, 48]. Various proteins in these epigenetic processes control genomic instability and regulate recombination initiation in a defined order [49]. The function of such proteins is regulated in a PTM-dependent manner [50]. For example, the methylation of lysine residues in histone proteins is critical for their conversion to euchromatin. Deficiency of the histone methyltransferase Ezh2 inhibits the methylation of lysine 27 in histone H3, thereby inhibiting the recombination of distal V genes, including those in the J558 family [51]. In addition, despite having a similar V(D)J recombination mechanism, rearrangement of the immunoglobulin and T-cell repertoire loci occurs specifically in B and T cells, respectively, as lineage specificity is determined by the methylation of lysine 79 of histone H3 [52]. Furthermore, previous studies have reported that the stability and activity of RAG proteins are regulated by phosphorylation [26].

Several studies have shown that the epigenetic factors that regulate V(D)J recombination depend on PTMs; however, the role of PTMs in trans-acting factors involved in locus compaction and looping structure has not been fully explored. Diverse V genes participate in V(D)J recombination by spatially juxtaposing distant V genes close to (D)J-rearranged segments through locus contraction mediated by long-range looping interactions [53]. Although Pax5, YY1, CTCF, and the cohesin complex have been implicated as critical regulators of locus contraction, the effects of PTMs on these proteins remain elusive. In this study, we demonstrated that O-GlcNAcylation, a type of PTM, directly modifies the YY1, CTCF, and SMC1/SMC3 proteins within the cohesin complex. Functionally, the inhibition of O-GlcNAcylation in B cells reduces the overall binding of YY1 to the IgH locus, especially in the distal region of the V_H genes. Since O-GlcNAcylated YY1 can be dissociated from the retinoblastoma protein (Rb), leading to DNA binding [54], we also aimed to determine whether YY1 is regulated through its interaction with the Rb protein. However, detecting the Rb protein in pro-B (IgM^- B220⁺CD43⁺) or pre-B cells (IgM^- B220⁺CD43^-) was challenging (data not shown), indicating that other factors may exist in B cells. Regardless of these regulatory factors acting like Rb, the ability of YY1 to bind to IgH loci is tightly regulated in an O-GlcNAcylation site-dependent manner (Fig. 5). The limited interaction between Eμ and the distal V_H genes in the 4C assay is also presumed to be due to the decreased binding of YY1 to Eμ when O-GlcNAcylation is inhibited (Fig. 7).

Our results showed that SMC1 and SMC3, the central components of the cohesin complex, are O-GlcNAc-modified in B cells and that the physical interaction between these two proteins is considerably affected by O-GlcNAcylation inhibition. To the best of our knowledge, this is the first study describing the role of O-GlcNAc levels in modulating the interaction between SMC1 and SMC3. The cohesin complex plays an important role in daughter chromatid cohesion and chromosome segregation and chromosome-related processes such as gene expression and DNA repair [55], suggesting that O-GlcNAcylation may be involved in these various mechanisms through the stabilization of the cohesin complex. In B cells, the cohesin complex is essential for class switching recombination (CSR) and V(D)J recombination through its interaction with CTCF [56, 57]. We also confirmed that O-GlcNAcylation modulated the interconnection between SMC1 and SMC3 in CH12F3 cells, a representative model cell line for CSR research (data not shown), suggesting that O-GlcNAcylation may be related to the mechanisms regulating CSR. O-GlcNAcylation also directly modifies CTCF, which strongly interacts with other O-GlcNAcylated proteins, such as the cohesin complex and DDX5, in B cells. In particular, our results showed that the protein stability of DDX5 was regulated in an O-GlcNAcylation-dependent manner in B cells. Moreover, we describe for the first time that a deficiency in DDX5 affects the DNA-binding ability of CTCF in B cells. The RNA helicase DDX5 is involved in various RNA processing mechanisms, regulates gene expression, and influences epigenetic modifiers or chromatin-associated factors [58]. Therefore, DDX5 potentially regulates the chromatin binding of architectural proteins such as cohesin or CTCF. Our results successfully demonstrated that the functions of most major proteins, such as cohesin, CTCF, and YY1, which are known to be involved in controlling the dynamics of the IgH locus, are directly or indirectly regulated by O-GlcNAcylation (Fig. S10). Moreover, our results suggest that O-GlcNAcylation is an upstream regulatory metabolism that controls V(D)J recombination and that maintaining O-GlcNAc homeostasis in the body ensures the normal function of the humoral immune response by ensuring antibody diversity.

Even though the recombination mechanisms and related factors for heavy- and light-chain loci in Ig genes that constitute antibodies are very similar [6], O-GlcNAcylation inhibition affects the recombination of each chain locus in different ways. In particular, in the repertoire analysis of Igκ gene recombination, genes that were originally used were reduced; however, no significant change was observed in the overall diversity as adjacent genes replaced it. According to the ChIP-seq results (Fig. S11), the binding of CTCF or SMC3 to the Igκ gene locus was strongly reduced overall, as shown in the IgH region. However, for YY1, the positions of the ChIP-seq peaks were reorganized upon decreased O-GlcNAcylation, but no significant change in the absolute peak number was observed. These results suggest that YY1 may play a more central role in V‒J recombination of the Igκ gene than other factors do. Further studies are essential to fully understand the increased YY1 dependency or the immunological effects of the altered repertoire of the Igκ light-chain gene.

Our findings underscore the intricate relationship between V(D)J recombination and allelic exclusion, both of which play essential roles in determining the antibody repertoire of B cells. The observed decrease in normal recombination within the IgH locus points to a reduced capacity for antibody diversity, particularly among genes that typically undergo high-frequency recombination (Fig. 2A–C). This limited recombination suggests that certain V_H genes may become less available, potentially altering the breadth of the immune response. Through HTGTS-Rep-seq analyses, we observed that in pro-B cells, only 12% of D_H-J_H recombination events occurred within the V_H region. However, in splenic resting B cells, ~24% of the clones exhibited recombination strictly within the V_H region (Fig. 2D, E). This finding suggests that while early pro-B cells exhibit considerable flexibility in recombination targets, non-V_H recombined clones are likely nonfunctional and eliminated during B-cell maturation. Furthermore, as cells progress from the pro-B stage to maturity, there is strong selection for functional recombination within the V_H region. This selection process ensures the survival of only those clones capable of producing functional antibodies, playing a crucial role in preventing nonproductive or harmful recombination events [30, 31]. Our results partly reveal a mechanism by which allelic exclusion and the selection of functional V_H recombination are regulated during B-cell development. The flexible combinations of V(D)J rearrangement and selective pressures during maturation enable B cells to generate a broad yet functional antibody repertoire. This process not only eliminates nonfunctional alleles but also preserves the diversity essential for adaptive immunity.

Various intrinsic and extrinsic factors compromise the immune system. The organs in our body, including immune-related organs such as the thymus and BM, deteriorate with age. Increased secretion of stress hormones, such as cortisol, can also affect inflammation or white blood cell production, and abnormal cytokine secretion can occur with insufficient sleep [59]. Recently, many studies have focused on the effects of environmental factors such as fine particles on immune cell functions [60]. Nutritional status also affects immune regulation [61]; obesity and overnutrition can cause hyperinflammatory responses, affecting T-cell function. Conversely, nutritional deficiency can affect the differentiation of immune cells and reduce the number of antibodies produced. Here, we shed light on the effect of nutritional imbalance on ensuring antibody diversity, in contrast to previous studies on antibody productivity in terms of nutrition. As O-GlcNAcylation is a nutrient sensor, we confirmed that the level of O-GlcNAcylation was also severely reduced under nutritional deficiency. Under these conditions, there is a deficiency in V(D)J recombination, the fundamental mechanism of antibody diversity. Notably, when the O-GlcNAcylation level, which was reduced by diet restriction, was restored by cotreatment with an inhibitor specific to the OGA enzyme, V(D)J recombination was significantly reversed. Although various metabolic mechanisms can be adversely affected by nutrient deprivation, our results revealed that restoring O-GlcNAcylation alone could normalize V(D)J recombination, demonstrating a close correlation between the two mechanisms.

O-GlcNAcylation affects various aspects of B-cell development, activation, and function. The induction of O-GlcNAcylation in large pre-B cells during early B-cell development is essential for the rapid proliferation of functional pre-B-cell clones on the basis of the O-GlcNAc status of c-Myc [62]. NFATc1 and NF-κB are modified by O-GlcNAcylation, which strengthens B-cell activation [63]. O-GlcNAcylation regulates signaling pathways involved in BCR signaling, including the protein tyrosine kinases Syk and Lyn for B-cell activation and expansion [64] and Lsp1 to control the apoptosis of mature B cells [65]. Despite efforts to elucidate the effects of O-GlcNAcylation on the development and activation of B cells, specific mechanistic studies related to antibody production, which is the ultimate function of B cells, are limited. Therefore, we investigated the role of O-GlcNAcylation in antibody formation, especially in securing diversity. V(D)J recombination, the basis for ensuring antibody diversity, can be regulated according to the O-GlcNAc status of multiple proteins, such as YY1, CTCF, and cohesin. In particular, by revealing that CTCF function can be affected by DDX5, whose protein stability is regulated by O-GlcNAc levels, a significant function of DDX5 in V(D)J recombination has been suggested. In conclusion, we established that maintaining normal O-GlcNAcylation levels in B cells promotes IgH locus contraction, enabling long-range interactions and, thus, securing antibody diversity by regulating V(D)J recombination.

Materials and methods

Study design

In this study, we aimed to verify the effects of O-GlcNAcylation on V(D)J recombination during early B-cell development. We utilized a mouse model in which O-GlcNAc levels were decreased by the intraperitoneal administration of an OGT inhibitor, and changes in V(D)J recombination were analyzed via HTGTS-Rep-seq upon O-GlcNAcylation inhibition in BM B cells. B cells from another mouse model in which O-GlcNAcylation was inhibited through dietary restriction and O-GlcNAc levels were restored by injection of an OGA inhibitor were used for the V(D)J rearrangement assay. The O-GlcNAc-modified proteins were captured via WGA-lectin, which specifically binds to GlcNAc, and the proteins were analyzed via MS and western blotting. To confirm the effect of O-GlcNAcylation on protein function, Co-IP and ChIP-seq were performed, and long-distance chromatin interactions were analyzed via 3D DNA-FISH and 4C-seq methods.

Animal experimental models

All the mice were obtained from Koatech (Pyeongtaek, Korea) and kept under a specific pathogen-free facility at the Korea Research Institute of Bioscience and Biotechnology (KRIBB, Daejeon, Korea). All animal studies were performed following protocols approved by the Animal Experiments Ethics Committee at KRIBB (KRIBB-AEC-21215). The mice were maintained under a 12 h light‒dark cycle at 23 °C and were fed a standard diet and water ad libitum.

All the mice (6-week-old C57BL/6 males) were fed standard rodent chow ad libitum during the first 7 days to measure the amount of food intake per day. Since one mouse in the regular diet (RD) group consumed approximately 2.8 g per day, the comparable calorie restriction (CR) group was provided 1.5 g, which was 60% of the RD group’s consumption. A high-fat diet (Research diet, D12492) or ketogenic diet (ENVIGO, TD96355) was prepared according to the manufacturer’s instructions. In the case of coadministration of the OGA inhibitor Thiamet G (TG; MedChemExpress, HY-12588) in the CR group, 600 μg of TG dissolved in PBS was administered intraperitoneally to the mice for 14 days. During the 2 weeks of different diet methods, body weights, and food intake were measured at the same time each day.

One milligram of OSMI-1 (MedChemExpress, HY-119738), which was solubilized in 100 μL of buffer containing 5% DMSO and 7% Tween-80 (MilliporeSigma, P8074) in distilled water, was intraperitoneally injected daily into 6-week-old C57BL/6 male mice for 10 days.

Isolation of primary B cells

Total bone marrow and spleen cells separated from 6- to 8-week-old male C57BL/6 mice (Koatech, Korea) were treated with 1X red blood cell lysis solution (Miltenyi Biotec, 130–094–183) at room temperature (RT) for 10 min. After centrifugation, the pelleted cells were resuspended in sorting buffer [0.5% bovine serum albumin (BSA) and 2 mM ethylenediaminetetraacetic acid (EDTA) in Hank’s balanced salt solution (HBSS)] and forced through a 40 µm cell strainer (Corning Inc., 352235). Single-cell suspensions of total bone marrow and spleen cells were subsequently incubated with anti-CD16/32 (BD Biosciences, 553141) to block Fc receptors at 4 °C for 10 min. To eliminate unwanted cells in the bone marrow, first, the cells were stained with biotin-conjugated anti-CD11b (BioLegend, 101204) and biotin-conjugated anti-IgM (BioLegend, 406504) at 4 °C for 15 min. After being washed with sorting buffer, the cells were stained with streptavidin microbeads (Miltenyi Biotec, 130–048–101) in the same manner. Unlabeled cells were collected via negative selection on MACS LD columns (Miltenyi Biotec, 130–042–901), followed by two washes with sorting buffer. Negatively selected cells (CD11b⁻IgM⁻) were then stained with APC-conjugated anti-B220 (BD Biosciences, 553092) and FITC-conjugated anti-CD43 (BD Biosciences, 553270) at 4 °C for 15 min. After being stained and washed twice with sorting buffer, pro-B (B220⁺CD43⁺) and pre-B220⁺CD43^-) cells were sorted with a FACSAria cell sorter (BD Bioscience).

For the early B cells used in this study, negatively selected cells (CD11b^-IgM^-) were stained with B220 microbeads (Miltenyi Biotec, 130–049–501) at 4 °C for 15 min and then collected via positive selection on MACS LS columns (Miltenyi Biotec 130–042–401). Through these processes, CD11b⁻IgM⁻B220⁺ cells obtained from mouse bone marrow are referred to as early B cells. Fluorescence-activated cell sorting (FACS) analysis of these early B cells revealed that pro-B (IgM⁻B220⁺CD43⁺) and pre-B (IgM^-B220⁺CD43^-) were mixed at a ratio of approximately 1:4.

In the case of the resting B cells from the spleen used in this study, total splenic cells were stained with biotin-anti-CD11b (BioLegend, 101204) and biotin-anti-CD43 (Miltenyi Biotec, 130–101–954) at 4 °C for 20 min. After being washed with sorting buffer, the cells were stained with streptavidin microbeads (Miltenyi Biotec, 130–048–101) in the same manner. Unlabeled cells were collected via negative selection on MACS MS columns (Miltenyi Biotec, 130–042–201). Through these processes, CD11b^-CD43^- cells obtained from the mouse spleen are referred to as resting B cells.

Cell culture

Primary early B cells (CD11b⁻IgM⁻B220⁺) or pro-B cells (CD11b^–IgM^–B220⁺CD43⁺) from mice were maintained in RPMI 1640 (Corning) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Thermo Fisher Scientific), 2-mercaptoethanol (50 μM), nonessential amino acids (100 μM), sodium pyruvate (1 mM), glutaGRO (2 mM; Corning, 25015CI) and 1X antibiotic-antimycotic (Thermo Fisher Scientific). The Abelson virus-transformed mouse pre-B-cell line PD36 [62] was also grown in RPMI 1640 media supplemented with the same components as above, except that L-glutamine (Thermo Fisher Scientific) was used instead of glutaGRO. All the cells were cultured at 37 °C in a humidified incubator containing 5% CO₂. If necessary, the OGT inhibitor OSMI-1 dissolved in DMSO was added to the culture media at a final concentration of 20 μM. To investigate protein stability upon changes in O-GlcNAcylation, cycloheximide (CHX; Sigma, 239765), a protein synthesis inhibitor, was used. In brief, 8 × 10⁵ PD36 pre-B cells were incubated in the presence or absence of OSMI-1 (20 μM final), TG (2 μM final), or DMSO for 12 hours and then treated with CHX (50 μg/mL final) in a time-dependent manner for up to 12 h.

Western blotting

The cells were harvested and lysed in the appropriate volume of SDS lysis buffer [0.5% SDS, 50 mM Tris-HCl (pH 6.8), 1 mM EDTA, 1 mM DTT, 1X protease inhibitor cocktail (GenDepot, P3100-005)], incubated on ice for 5 min, diluted with ice-cold correction buffer [1.25% NP-40, 0.625% sodium deoxycholate, 62.5 mM Tris-HCl (pH 8.0), 2.25 mM EDTA, 187.5 mM NaCl, 1X protease inhibitor cocktail] at a ratio of 1:4, and then incubated on ice for 15 min. The cell lysate was put through a QiaShredder (Qiagen, 79654) twice to rupture genomic DNA and then centrifuged at 13,000 RPM for 10 min, after which only the supernatant was collected. Protein quantification was performed via a Pierce^TM BCA protein assay kit (Thermo Fisher Scientific, 23225), and 5–20 μg of protein was mixed with 5X SDS‒PAGE sample buffer (TransLab, TLP-102.1) and loaded into each lane of a traditional 4–12% Tris-glycine SDS‒PAGE protein gel. After electrophoresis, the proteins were transferred to nitrocellulose blotting membranes (Amersham, 10600002) via a power blot system (Thermo Fisher Scientific, PB0012). The membrane was blocked with TBS containing 5% BSA for 30 min at RT and then incubated overnight at 4 °C with the appropriate primary antibodies in TBS-T [20 mM Tris (pH 7.6), 137 mM NaCl, and 0.05% Tween 20] containing 5% BSA. After washing five times with TBS-T, the membrane was probed with a secondary antibody conjugated with HRP or IR (infrared) dye for 1 h at RT. The membrane was washed in the same manner as previously described. The protein bands were detected and visualized via an Odyssey Fc Imaging System and quantified via Image Studio software (LI-COR Bioscience). The following antibodies were used in this study: O-GlcNAc (RL2; ThermoFisher Scientific, MA1-072 or CTD110.6; Santa Cruz Biotechnology, SC-59623), O-GlcNAc transferase (OGT; Sigma, O6264), CTCF (Thermo Fisher Scientific, PA5-29691), CTCF (ABclonal, A19588), FLAG (Sigma, F1804), Pax-5 (Cell Signaling Technology, 8970), SMC1 (Thermo Fisher Scientific, PA5-29691), SMC3 (Thermo Fisher Scientific, PA5-29131), YY1 (Thermo Fisher Scientific, PA5-29171), RAG-1 (Santa Cruz Biotechnology, SC-377127), E2A (Santa Cruz Biotechnology, SC-133075), DDX5 (Cell Signaling Technology, 9877), Lamin B1 (Santa Cruz Biotechnology, SC-377000), β-actin (Santa Cruz Biotechnology, SC-47778) and β-actin-conjugated HRP (Cell Signaling Technology, 5125).

Analysis of the Ig gene repertoire

Total genomic DNA (gDNA) was extracted from early B cells (CD11b^–IgM^–B220⁺), which were sorted from mouse bone marrow, via a DNeasy^TM Blood & Tissue Kit (Qiagen, 69506) following the manufacturer’s protocol. The usage of different V_H gene families was determined quantitatively via a SYBR Green-based real-time PCR assay. Total gDNA was mixed with a final concentration of 100 nM of each primer set and iTaq^TM Universal SYBR® Green Supermix (Bio-Rad, 1725121), Fast SYBR^TM Green Master Mix (Thermo Fisher Scientific, 4385612) or TOPreal^TM SYBR Green qPCR Premix (Enzynomics, RT500), and qPCR was performed following the manufacturer’s protocols (5 min at 95 °C, 45 cycles of 15 s at 95 °C, 20 s at 60 °C and 30 s at 72 °C). Each PCR assay was carried out in duplicate or triplicate, and the sequences of primers used in the real-time reactions are listed in Table S1. For sequencing analysis, PCR products amplified via degenerate primers for V_HJ558, V_H3609, V_HGam3.8, V_H7183 or V_HQ52 with IGHJ1 or VκD with IGKJ1, which were approximately 300 bp in length, were gel purified and subcloned and inserted into the TOPO cloning vector via a TOPO^TM TA cloning kit (Thermo Fisher Scientific, K4575J10) following the manufacturer’s protocol. The sequences of the V_H or Vκ genes in each clone were identified via the IgBLAST program (National Center for Biotechnology Information, Bethesda, MD).

High-throughput genome-wide translocation sequencing-adapted repertoire sequencing (HTGTS-Rep-seq)

For HTGTS-Rep-seq, we modified the original linear amplification-mediated high-throughput genome-wide translocation sequencing (LAM-HTGTS) protocol previously published [66], as described by Lin et al. [29]. In brief, 250 ~ 400 ng of genomic DNA isolated from mouse pro-B (IgM⁻B220⁺CD43⁺) bone marrow sorted by FACS or 1–2 μg of resting B (CD11b^–CD43^–) spleen cells sorted by MACS were sonicated to ~1.5 kb with a Bioruptor® Pico Sonication Device System (Diagenode) and amplified via a J_H1-specific biotinylated primer (5’-/5BiosG/CTGCAGCATGCAGAGTGTG-3’). The biotinylated products were isolated with DynaBeads^TM MyOne^TM Streptavidin C1 (Thermo Fisher Scientific) and ligated to the bridge adapter under the following conditions: 1 h at 25 °C, 2 h at 22 °C, and 16 h at 16 °C. Nested PCR with ligation products as a template was performed via the following I7-Blue and I5-barcode-J_H1 primers under modified conditions: 5 min at 95 °C, 30 cycles of 1 min at 95 °C, 30 s at 56 °C and 1 min at 72 °C, and 6 min at 72 °C. The PCR products were cleaned via a QIAquick gel extraction kit (Qiagen, 28706). The final PCR was performed via the P5-I5 and P7-I7 primers under the following conditions: 3 min at 98 °C, 15 cycles of 10 s at 98 °C, 50 s at 72°C, and 10 min at 72 °C. The final PCR products were purified and size selected via AMPure XP beads (Beckman Coulter). The purified DNA fragments were then combined for library preparation, and sequencing was performed via the Illumina iSeq 100 platform. For HTGTS-Rep-seq, the sequences of primers used are listed in Table S2.

For the standard analysis of LAM-HTGTS, the custom bioinformatic pipeline by Hu et al. was primarily used [66], but an additional two filtering steps were employed. First, prior to the identification of break junctions, the raw reads were trimmed and filtered via TrimGalore (v0.6.7) with the options “--2color 20, -- length 150, -e 0.25” and fastp (v0.22.0) with the option “--trim_poly_g”. Second, after junction identification, only the reads with junctions detected within -10 to +10 bp from the expected cutting site were selected for final analyses. Subsequently, IgBLAST (v1.20.0) in the Change-O package (v1.3.0) [67] was used for the analysis of V_H repertoires [29], and the read counts of V_H genes recombined with D-J_H1 were normalized according to the total number of the finally selected reads.

WGA-lectin-based pull-down assay and immunoprecipitation

O-GlcNAc-attached proteins were isolated via a glycoprotein isolation kit, WGA (Thermo Fisher Scientific, 89805), following the manufacturer’s protocol. Early B cells (CD11b^–IgM^–B220⁺) isolated from mouse bone marrow were cultured with OSMI-1 (10 μM final) and DMSO for 4–6 h, after which protein lysates were prepared as described above. Three hundred micrograms of lysate were mixed with 75 μL of 5X binding/wash buffer, and 200 μL of WGA-lectin resin (50% slurry) was washed three times with 1X binding/wash buffer and then mixed with the lysate. Pierce^TM Control Agarose Resin (Thermo Fisher Scientific, 26150) was used as a control. The mixtures were then incubated for 30 min at room temperature with end-over-end mixing via a rotator. The resin captured with O-GlcNAcylated proteins by WGA-lectin was washed 4 times with 1X binding/wash buffer and eluted with 2X SDS‒PAGE sample buffer in RIPA buffer with heating while shaking at 95 °C for 10 min. Eluted samples were then loaded on SDS‒PAGE gels.

For immunoprecipitation, a total of 100–500 μg of protein extracted from early B cells (CD11b⁻IgM⁻B220⁺) or the pre-B-cell line PD36 was used. Approximately 2–5 μg of appropriate antibodies were mixed with 40 µL of SureBeads protein A magnetic beads (Bio-Rad, 161--4011) or Pierce^TM Protein A/G Magnetic Beads (Thermo Fisher Scientific; 88802) for 20 min at RT with end‒end mixing via a rotator. Each lysate was precleared with a previously prepared isotype-matched antibody-magnetic bead complex at 4 °C for 1 h. The precleared lysates were then incubated with the magnetized antibody-bead complex at 4 °C by overnight rotation, followed by five washes with PBS-T (0.1% Tween 20 in PBS). Proteins were eluted from the beads in the same manner as described above.

LC‒MS and proteome data analysis

Proteins were extracted from early B cells (CD11b^–IgM^–B220⁺) from mice, and 300 μg of protein was used for the WGA-lectin pull-down assay. The samples were eluted in 2X SDS‒PAGE sample buffer in RIPA buffer with heating while shaking at 95 °C for 10 min. The eluted samples were loaded onto SDS‒PAGE gels, which were subsequently sliced immediately after they were fully stacked. The gel pieces were reduced with 10 mM tris(2-carboxyethyl) phosphine (TCEP) in 25 mM ammonium bicarbonate (ABC) at 37 °C for 30 min. Alkylation was conducted with 55 mM iodoacetamide (IAA) in 25 mM ABC at room temperature for 30 min in the dark. The gel pieces were digested with trypsin in 25 mM ABC overnight at 37 °C. The samples were spun down briefly, and the supernatant was collected into a new tube. Gel pieces were added to 20 µL of 0.1% formic acid in 50% acetonitrile and incubated for 15 min. The supernatant was combined into the previous supernatant tube. The previous step was repeated, and the products were then combined into a supernatant tube. The supernatant was completely dried via a centrifugal vacuum concentrator. The digested peptides were desalted via C18 spin columns (Harvard Apparatus, Holliston, MA, USA), and the peptides were eluted with 80% acetonitrile in 0.1% formic acid in water. The prepared samples were resuspended in 0.1% formic acid in water and analyzed via a Q Exactive Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) along with an Ultimate 3000 system (Thermo Fisher Scientific, Waltham, MA, USA). We used a 2 cm × 75 μm ID trap column packed with 3 μm C18 resin and a 50 cm × 75 μm ID analytical column packed with 2 μm C18 resin to determine peptide hydrophobicity. The mobile phase solvents consisted of (A) 0.1% formic acid in water and (B) 0.1% formic acid in 90% acetonitrile, and the flow rate was fixed at 300 nL/min. The gradient of the mobile phase was as follows: 4% solvent B for 10 min, 4–20% solvent B for 20 min, 20–50% solvent B for 22 min, 50–96% solvent B for 0.1 min, holding at 96% solvent B for 7.9 min, 96–4% solvent B for 0.1 min, and 4% solvent B for 14.9 min. A data-dependent acquisition method was adopted, and the top 10 precursor peaks were selected and isolated for fragmentation. Ions were scanned at high resolution (70,000 in MS1, 17,500 in MS2 at m/z 400), and the MS scan range was 400–2000 m/z at both the MS1 and MS2 levels. The precursor ions were fragmented with an NCE (normalized collisional energy) of 27%. Dynamic exclusion was set to 30 s.

The Thermo MS/MS raw files of each analysis were searched via Proteome Discoverer™ software (ver. 2.5), and the Mus musculus database was downloaded from UniProt. The appropriate consensus workflow included a peptide-spectrum match (PSM) validation step and the SEQUEST HT process for detection as a database search algorithm. The search parameters were set as follows: 10 ppm tolerance of precursor ion masses, 0.02 Da fragment ion mass, and a maximum of two missed cleavages with the trypsin enzyme. The dynamic modifications of the peptide sequence were as follows: static carbamidomethylation of cysteine (+57.012 Da), variable modifications of methionine oxidation (+15.995 Da), acetylation of the protein N-term (+42.011 Da), and carbamylation of the protein in the N-term (+43.0006 Da). After searching, data results with a false discovery rate (FDR) of less than 1% were selected, and at least 6 additional peptide lengths were filtered out.

Chromatin immunoprecipitation (ChIP) assays

A total of 1 × 10⁶ sorted pro-B cells (IgM^-B220⁺CD43⁺) from the bone marrow of mice injected with the OGT inhibitor OSMI-1 were directly used for the ChIP experiment. In the case of culture with OSMI-1, sorted early B cells (CD11b⁻IgM⁻B220⁺) from mouse bone marrow or PD36 pre-B-cell lines were incubated with OSMI-1 (20 μM final) for 24 h and then used. ChIP experiments were conducted according to the manufacturer’s protocol via a ChIP assay kit (Sigma, 17-295). For each ChIP, antibodies against YY1 (Invitrogen, PA5-29171), CTCF (Invitrogen, PA5-17143), SMC3 (Invitrogen, PA5-29131) and FLAG (Sigma, F1804) were used, and normal rabbit IgG (Cell Signaling Technology, 2729) and normal mouse IgG (Santa Cruz Biotechnology, SC-2025) were used for preclearing. Real-time qPCR was performed as described above under the following conditions: 10 min at 95 °C, 45–55 cycles of 20 s at 95 °C, 60 s at 58–60°C and 10 min at 72 °C. The enrichment of target regions in the ChIP data was normalized to that of ACTG1. The sequences of primers used are listed in Table S1.

ChIP-sequencing

Library preparation and sequencing

The construction of the library was performed via the NEBNext® Ultra^TM DNA Library Prep Kit for Illumina (New England Biolabs, UK) according to the manufacturer’s instructions. Briefly, the chipped DNA was ligated with adapters. After purification, PCR was performed with adapter-ligated DNA and index primers for multiplex sequencing. The library was purified by using magnetic beads to remove all the reaction components. The size of the library was assessed via an Agilent 2100 bioanalyzer (Agilent Technologies). High-throughput sequencing was performed as single-end 75 sequences via the NextSeq 500 platform (Illumina, Inc., USA).

Data analysis

The reads were trimmed and aligned via Bowtie2 (Langmead and Salzberg, 2012). Bowtie2 indices were generated from either the genome assembly sequence or the representative transcript sequences for alignment to the genome and transcriptome. MACS (Model-based Analysis of ChIP-Seq) was used to identify peaks from the alignment file. Gene classification was based on searches performed via the DAVID (http://david.abcc.ncifcrf.gov/) and Medline databases (http://www.ncbi.nlm.nih.gov/). Normalized data were plotted along chromosomal coordinates (mm10) for visualization with IGV. For the correlation study, low-quality DNA reads and adapter sequences were filtered out via fastp-0.23.1 [68], duplicated reads were removed via samtools-1.17 [69], and ENCODE blacklist regions [70] were removed. Peaks in alignment files were identified via MACS2 [71].

Plasmid construction and transfection

The wild-type (wt) cDNA clone for mouse YY1 in the pCMV6-Entry vector was purchased from OriGene (Rockville, MR206531). A mutant clone (T236A) with a point mutation in specific amino acids of the wt-YY1 clone was generated via quick-change site-directed mutagenesis (COSMOGENETECH, Korea). Transient transfection of PD36 for overexpression of both wt-YY1 and mut-YY1 (T236A) was performed via the NEON^TM Transfection System (Thermo Fisher Scientific, MPK5000) following the manufacturer’s protocol and as previously described [62]. After transfection, the cells were incubated for 24 h and then treated with OSMI-1 (or DMSO) for 12 h. Transient transfection of the PD36 pre-B-cell line and primary early B cells (CD11b⁻IgM⁻B220⁺) from mouse BM for the knockdown of OGT or DDX5 was performed via the NEON^TM Transfection System (Thermo Fisher Scientific, MPK5000) following the manufacturer’s protocol and as previously described [62]. In brief, 3 × 10⁶ PD36 cells were transfected with 50 nM siRNA for DDX5 and then incubated with or without OSMI-1 (20 μM final) for 24 h. For primary early B cells (CD11b⁻IgM⁻B220⁺), 2.5 × 10⁶ cells were transfected with 50 nM OGT or DDX5 siRNA and then incubated with IL-7 (5 ng/mL final) for 24 h. The sequences of the siRNAs for OGT were 5′-GAA GAA AGU UCG UGG CAA A-3′ (sense) and 5′-UUU GCC ACG AAC UUU CUU C-3′ (antisense). The sequences of the siRNAs for DDX5 were 5′-GUA GAU ACA UAC AGA AGA ATT-3′ (sense) and 5′-UUC UUC UGU AUG UAU CUA CTT-3′ (antisense). The AccuTarget^TM Negative Control siRNA (Bioneer, SN-1003) was used as a scrambled siRNA.

DNA FISH and confocal analysis

Probes for 3D DNA FISH prepared from bacterial artificial chromosomes (BACs) were purchased from Empire Genomics (New York). BAC clone RP23-218B2 is labeled with 5-ROX (5-carboxy-X-rhodamine) fluorescence, and RP23-451B13 is labeled with fluorescein fluorescence. DNA FISH was performed as previously described [72]. Briefly, pro-B cells (IgM⁻B220⁺CD43⁺) isolated from mouse bone marrow were fixed on poly-L-lysine at RT for 10 min. Unattached cells were removed by washing three times with PBS, and then, the attached cells were treated with fixation/permeabilization buffer [20 mM KH₂PO₄, 130 mM NaCl, 20 mM KCl, 10 mM EGTA, 20 mM MgCl₂, 0.1% (v/v) Triton X-100, and 0.5% (v/v) glutaraldehyde (Sigma Aldrich, St. Louis, MO, grade 1, 70% aqueous)] for 30 min. After the slides were washed twice with PBS, they were incubated with 1 mg/ml sodium borohydride solution at RT for 15 min. The slides were rinsed three times with PBS and then sequentially incubated with PBS with 5% FBS for 30 min. Then, the slides were treated with RNase solution (100 μg/ml in PBS) for 1‒2 h in a humid chamber and incubated with 1 M NaOH for 2 min, followed by rinsing immediately in ice-cold PBS three times. The hybridization of the probes was performed at 37 °C overnight in a humid chamber. The slides were washed three times with 1X SSC at 37 °C for 30 min in the dark and then mounted in ProLong^TM Gold Antifade Mountant with the DNA stain DAPI (Thermo Fisher Scientific, P36931). Z stacks with 0.5 μm separation were analyzed via confocal microscopy via a ZEISS LSM800 confocal microscope, and the distance was measured via the ZEN system.

Circular chromosome conformation capture and sequencing (4C-seq analysis)

For 4 C analysis, we cultured primary pro-B cells (IgM^-B220⁺CD43⁺) following a previously published protocol [73]. In brief, 5–7.5 × 10⁵ pro-B cells incubated in the presence or absence of OSMI-1 (10–20 μM final) for 12 h were harvested and cross-linked in fixation buffer (2% formaldehyde and 10% FBS in PBS) for 10 min at RT with end-over-end mixing via a rotator. Immediately after 10 min of incubation, cold 1 M glycine buffer was added to a final concentration of 0.13 M, and then the cells were pelleted following a wash with cold PBS. The cells were lysed on ice with lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 0.5% IGEPAL) in the presence of a protease inhibitor (Sigma, 4693159001). Nuclei were resuspended in SDS (0.3% final) for 1 h at 37 °C and neutralized with Triton X-100 (2.5% final) for 1 h at 37 °C. First, DNA restriction was performed using HindIII (NEB, R0104T) in NEBuffer^TM r 2.1 overnight at 37 °C before heat inactivation for 20 min at 65°C. Ligation was performed for 4 h at 16 °C. Reverse crosslinking was performed overnight at 65 °C. Next, the DNA was treated with proteinase K and cleaned with phenol/chloroform. Second, DNA restriction was performed using DpnII (NEB, R0543M) in NEBuffer^TM DpnII overnight at 37 °C before heat inactivation for 20 min at 65 °C. Ligation was performed for 4 h at 16 °C. Next, the DNA was cleaned with phenol/chloroform. After this step, the DNA was quantified with a Qubit instrument. For library preparation, the first PCR step was performed to amplify the fragments. Two hundred nanograms of 4 C template were amplified for 16 cycles in 1 µL of PCR primer (20 pmol/µL) cocktail and 25 µL of 2X Phusion Master Mix (Thermo Fisher Scientific, F566S). The PCR settings included 2 min at 94 °C followed by 16 cycles of 10 s at 94 °C, 1 min at 62 °C, and 3 min at 68 °C. The samples were then held at 68 °C for 5 min before being lowered to 4 °C until they were collected. Amplified DNA was size selected via AMPure XP beads. The amount of template was quantified via the Qubit dsDNA BR Assay Kit. The second PCR step was performed via NEBNext Multiplex Oligos for Illumina (Dual Index Primers Set 1, NEB, E7600S) and the NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB, E7645S), following the manufacturer’s instructions. Amplified DNA was size selected via AMPure XP beads. The resulting libraries were quantified with a Qubit instrument, mixed in equal concentrations, and sequenced via an Illumina iSeq 100 instrument.

The raw reads were trimmed and mapped to the mm10 genome according to Krijger et al. [73]. The Basic4C-seq R package was subsequently used to count reads on restriction fragment ends and to calculate normalized reads per million (RPM) based on the reads with mapping quality (MapQ) ≥ 1. For the analysis of data generated by the nondirectional library, R1 and R2 fastq files were concatenated prior to data analysis.

Statistical analysis

Two-tailed Student’s t-tests in Microsoft Excel or GraphPad Prism 8 were used for all the data to determine the statistical significance. The animals/samples were randomly allocated into experimental groups. Detailed information on the statistical tests and replicates is provided in each figure legend. Significance is annotated as *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001.

Author contributions

Conceptualization: BCJ, JL, S-KP Methodology: BCJ, Y-JK, AKP, HH, T-DK, JL, S-KP Investigation: BCJ, Y-JK, AKP, M-RS, KMN, J Lee, DA, YP, HH. Visualization: BCJ, Y-JK, AKP, HH, JL, S-KP. Funding acquisition: Y-JK, AKP, HH, T-DK, JL, S-KP. Project administration: AKP, HH, T-DK, JL, S-KP. Supervision: JL, S-KP. Writing—original draft: S-KP. Writing—review & editing: BCJ, Y-JK, AKP, HH, T-DK, JL, S-KP.

Funding

This work was supported by a National Research Foundation (NRF) grant from Korea funded by the Ministry of Education (2021R1I1A2057945 to S-KP), the Ministry of Science & ICT (2020R1I1A3073845 to AKP, and 2021R1A2C1012477 to JL), the National Research Council of Science & Technology (NST) (GTL24021-000 to T-DK, GTL24021-400 to HH, and ZYM9382312 to Y-JK as the Postdoctoral Fellowship Program for Young Scientists), and the Korea Research Institute of Bioscience and Biotechnology (KRIBB) Research Initiative Program (KGM9942421).

Data availability

Data deposition: The sequencing and processed data reported in this paper have been deposited in the Sequence Read Archive (SRA) database, www.ncbi.nlm.nih.gov/sra (BioProject Accession: PRJNA1167816 for 4C-seq and HTGTS-Rep-seq; PRJNA1167709 for ChIP-seq). Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Sung-Kyun Park (skpark@kribb.re.kr).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41423-024-01236-9.