YY1 controls Eμ-3′RR DNA loop formation and immunoglobulin heavy chain class switch recombination

Immunoglobulin heavy chain (IgH) class switch recombination (CSR) is a crucial immune function that produces immunoglobulin isotypes with distinct effector functions.¹  CSR is initiated by the introduction of double-stranded DNA breaks by activation-induced DNA deaminase (AID) at switch sequences nearby Cμ (Sμ) and targeted C_H exons (Sγ1, Sγ2a, Sγ2b, etc). Repair of these breaks results in linkage of the VDJ exon with the switched C_H region. The 200-kb region from the VDJ exon to the terminal Cα exon forms a loop involving interactions between sequences nearby the Eμ intronic enhancer and the 3′ regulatory region (3′RR) located at the end of the IgH locus.^2-4  Although enhancers associated with the 3′RR are essential for CSR,^5,6  the role of Eμ in this process is more complex.^7-9  Eμ deletion impairs, but does not abolish CSR, suggesting either the V_H promoter 3 kb upstream from Eμ contacts the 3′RR or the V_H promoter may substitute for Eμ activity on Eμ-deficient alleles. Consistent with prior convention, we refer to these interactions as the Eμ-3′RR loop.

Although the Eμ-3′RR DNA loop has been described, its necessity for CSR is uncertain, and the proteins that control DNA interactions are poorly understood. Previously, we found that conditional ablation of transcription factor YY1 causes a significant drop in CSR.¹⁰  YY1 physically interacts with AID and regulates its nuclear accumulation.¹⁰  We proposed that YY1 controls CSR, at least in part, by regulating the amount of nuclear AID. However, YY1 is also known to impact long-distance DNA contacts in a number of systems.^11-16  Given the importance of YY1 in CSR and long-distance DNA loops, we tested the importance of YY1 for Eμ-3′RR DNA loops.

Isolation of splenic follicular B cells and treatment with recombinant TAT-CRE enzyme to knock-out YY1 have been described.¹⁰ 

Virus production of various YY1 constructs and transduction of splenic B cells are described in the supplemental Methods.

Position-specific 10-kb probes were generated by polymerase chain reaction using BAC templates with primers as described.^14,15  fluorescent in situ hybridization (FISH) was performed as described,¹⁵  with additional details in the supplemental Methods.

Chromosome conformation capture (3C) analyses are described in supplemental Methods.

Microarray methods and analyses are described in the supplemental Methods.

The 3′ region of the IgH locus involved in CSR is shown in Figure 1A. The DNA loop formed between the Eμ and 3′RR enhancer regions in activated splenic B cells can be detected by 3D FISH using probes that hybridize nearby the Eμ and the 3′RR enhancers (FISH probes 1 and 2, respectively; Figure 1A).¹⁴  To determine the importance of YY1 for Eμ-3′RR loop formation, YY1 was ablated ex vivo with recombinant TAT-CRE (supplemental Figure 1A-B). YY1 deletion resulted in a three- to eightfold drop in immunoglobulin G1 CSR similar to our previous work¹⁰  (supplemental Figure 1C). Strikingly, ablation of YY1 also dramatically reduced Eμ-3′RR DNA loop formation (Figure 1B; supplemental Table 1).

Chromosome conformation capture (3C) assays can also identify the Eμ-3′RR long-distance loop.^2,3  Using the Eμ region as anchor, we probed the IgH 3′ region by 3C assay in the presence and absence of YY1. Initial 3′RR contacts were observed at the Eα region as previously observed²  (Figure 1C Eμ-Eα peak; supplemental Figure 2 with expanded scale). Deletion of YY1 caused a 50% drop in DNA contacts (supplemental Figure 2). Scanning further, we observed the dramatic Eμ-3′RR loop that makes Eμ contacts within the hs3b/hs4 region of the 3′RR enhancer (Figure 1C Eμ-3′Rra-2 peak). Deletion of YY1 caused a striking fivefold drop in this loop (Figure 1C). This loss was specific as there was little impact on DNA loops at the Igκ locus using the Igκ E3′ enhancer as anchor (supplemental Figure 3). Thus, our 3C experiments also showed the Eμ-3′RR DNA loop is YY1 dependent.

YY1 contains a number of domains with distinct functions (Figure 2A). We tested the ability of YY1 mutants that ablate 1 or more functions to rescue the Eμ-3′RR DNA loop after YY1 ablation. Mock-treated samples showed 80% of cells with Eμ-3′RR loops, whereas DNA loops were lost in TAT-CRE–treated cells due to YY1 deletion (Figure 2B-D). Transduction with empty vector failed to rescue the Eμ-3′RR DNA loop (Figure 2B-D). On the contrary, wild-type YY1 rescued the Eμ-3′RR DNA loop, whereas the 1-200 mutant failed in this rescue (Figure 2B-D). Of particular importance, the 201-414 mutant lacking the activation domain but containing the REPO and DNA binding domains completely restored the Eμ-3′RR DNA loop (Figure 2B-D; supplemental Table 1).

In parallel, we explored the ability of YY1 mutants to rescue CSR. Treatment with TAT-CRE resulted in a fourfold drop in CSR (Figure 2E). Empty vector failed to rescue CSR, whereas vector expressing wild-type YY1 and the YY1ΔREPO mutant fully restored CSR (Figure 2E). The YY1 1-200 and YY1 288-414 mutants failed to restore CSR, but the YY1 201-414 mutant, which rescued the Eμ-3′RR DNA loop, completely restored CSR (Figure 2E; supplemental Figure 4B-H). We conclude that the C-terminal half of YY1 is sufficient for rescuing both the Eμ-3′RR DNA loop and the CSR.

The ability of the YY1 201-414 mutant (lacking the transactivation domain) to rescue the Eμ-3′RR DNA loop as well as CSR argued that DNA looping and CSR do not require YY1 transactivation function. RNA transcript studies showed YY1 deletion decreased expression of genes involved in nucleotide binding, glycolysis, protein folding, lipid biosynthesis, and mitochondrial function (supplemental Figure 5A), and caused increased expression of genes involved in chromatin organization, proteolysis, translation, apoptosis, and non–membrane-bound organelles (supplemental Figure 5A).

Nearly all genes implicated in long-distance DNA interactions (44/48 genes) or CSR (42/47 genes) were not differentially expressed (supplemental Figure 5B blue and green dots, respectively; supplemental Tables 2 and 3). Gene Set Enrichment Analysis demonstrated that DNA looping and CSR genes were not enriched in our dataset. Therefore, YY1 does not likely transcriptionally regulate genes that control DNA looping or CSR.

Our studies define a critical role for YY1 controlling the Eμ-3′RR long-distance DNA loop and strongly argue for its role in CSR. We propose that YY1 plays a largely structural role in regulating long-distance DNA loops. We previously showed that YY1 physically interacts with components of the PcG, condensin, and cohesin complexes,^17-19  each of which can function in long-distance DNA interactions.^20-22  DNA-bound YY1 may provide a platform for recruitment of these complexes (directly or indirectly) to mediate long-distance DNA interactions. Deletion of the Eμ YY1 binding region minimally impacts loops and CSR, suggesting other binding sites may compensate for this loss, or YY1 bound to the 3′RR contacts proteins bound near the VDJ promoter or switch sequences. YY1 levels increase after splenic B-cell activation,²³  and highest expression levels are in germinal center B cells.²⁴  YY1 may inducibly interact with proteins to nucleate recruitment to DNA. Proteomic and chromatin immunoprecipitation sequencing experiments may identify key YY1-interacting proteins, and whether they are recruited to specific sites within the immunoglobulin loci. Our findings here provide foundational insight into a critical immune function.

The authors thank Yang Shi (Harvard University) for yy1^f/f mice and Daniel Beiting (University of Pennsylvania) for assistance and advice on transcript experiments. BAC RP23-201H14 was kindly provided by Cornelis Murre (University of California, San Diego).

This work was supported by National Institute of Allergy and Infectious Diseases, National Institutes of Health (NIH) R01 grant AI079002, National Institute of General Medical Sciences, NIH R01 grant GM111384, and US Department of Defense grant W81XWH-14-1-0171 (M.L.A.), and in part by the Intramural Research Program of the NIH, National Institute on Aging (R.S.).

Contribution: P.M., T.G., V.J., A. Basu, V.S., M.L.A., and R.S. designed the experiments; P.M., T.G., A. Basu, C.T.B., V.S., F.G., M.L.A., and R.S. evaluated data; P.M., T.G., V.J., V.S., F.G., A. Basu, and A. Banerjee performed experiments; M.L.A. and R.S. wrote the manuscript; evaluation and edits were performed by all the authors.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current affiliation for P.M. is Diabetes and Obesity Center, Department of Cardiovascular Medicine, University of Louisville, Louisville, KY.

The current affiliation for A. Basu is Penn State University, Brandywine, Media, PA.

Correspondence: Michael L. Atchison, University of Pennsylvania, School of Veterinary Medicine, 3800 Spruce St, Philadelphia, PA 19104; e-mail: atchison@vet.upenn.edu.