Constraints contributed by chromatin looping limit recombination targeting during immunoglobulin class switch recombination

Abstract

Engagement of promoters with distal elements in long range looping interactions has been implicated in regulation of Ig class switch recombination (CSR). The principles determining the spatial and regulatory relationships among Igh transcriptional elements remain poorly defined. We examined the chromosome conformation of constant region (C_H) loci that are targeted for CSR in a cytokine dependent fashion in mature B lymphocytes. Germline transcription (GLT) of the γ1 and ε C_H loci is controlled by two transcription factors, IL4 inducible STAT6 and LPS activated NFκB. We showed that although STAT6 deficiency triggered loss of GLT, deletion of NFκB p50 abolished both GLT and γ1 locus:enhancer looping. Thus, chromatin looping between C_H loci and Igh enhancers is independent of GLT production and STAT6, whereas the establishment and maintenance of these chromatin contacts requires NFκB p50. Comparative analysis of the endogenous γ1 locus and a knock-in heterologous promoter in mice identified the promoter per se as the interactive looping element and showed that transcription elongation is dispensable for promoter/enhancer interactions. Interposition of the LPS responsive heterologous promoter between the LPS inducible γ3 and γ2b loci altered GLT expression and essentially abolished direct IgG2b switching while maintaining a sequential μ-> γ3-> γ2b format. Our study provides evidence that promoter/enhancer looping interactions can introduce negative constraints on distal promoters and affect their ability to engage in germline transcription and determine CSR targeting.

INTRODUCTION

Humoral immunity is mediated by Ig antigen receptors that are assembled from multiple V, D, J segments during early B cell development (1). In mature B cells, Ig class switch recombination (CSR) promotes diversification of constant region (C_H) effector function while retaining the original V(D)J rearrangement (2). The mouse Igh locus spans 2.8 Mb within which a 220 kb genomic region contains eight C_H genes, encoding μ, δ, γ3, γ1, γ2b, γ2a, ε, and α chains, each paired with repetitive switch (S) DNA (with the exception of Cδ). CSR is focused on S regions and involves an intra-chromosomal deletional rearrangement. Germline transcript (GLT) promoters (Prs), located upstream of I exon-S-C_H regions, focus CSR to specific S regions by differential transcription activation (2, 3). Activation induced deaminase (AID) initiates a series of events culminating in formation of double strand breaks (DSBs) at donor Sμ and a downstream acceptor S region to create S/S junctions and facilitate CSR.

Gene expression is regulated by combinations of regulatory elements that interact over hundreds of kilobases. Use of chromosome conformation capture (3C) and its derivatives has demonstrated in numerous genetic loci that distant chromosomal elements associate to form chromatin loops thereby providing a mechanism for Pr activation via long range enhancer function (4). The I-S-C_H region genes are embedded between the Eμ and 3’E〈 enhancers that are separated by 220 kb. Our 3C studies revealed that mature resting B cells engage in long range Eμ and 3’Eα chromatin interactions (5, 6). B cell activation leads to induced recruitment of the I-S-C_H loci to the Eμ:3’Eα complex that in turn facilitates GLT expression and S/S synapsis (6). Targeted deletion of DNase hypersensitive sites (hs) 3b,4, elements within 3’Eα, leads to loss of all GLT expression except for γ1 GLT which is reduced, impairment of CSR (7) and abrogation of Eμ:3’Eα and I-S-C_H loci:3’Eα looping interactions (6). Thus, CSR is dependent on three dimensional (3D) chromatin architecture mediated by long range intra-chromosomal interactions between distantly located transcriptional elements.

Given the importance of chromatin looping during CSR, several fundamental questions regarding the establishment and maintenance of DNA loop formation emerge: What is the relationship of transcription, transcription factors (TF), and specific transcriptional elements to the formation of DNA loops that promote or exclude GLT expression and S/S synapsis, preconditions for the CSR reaction? Additionally, it has been difficult to integrate the spatial relationships within the Igh locus with the preferential expression of some isotypes. Notably, IgG1 and IgE are both induced by CD40L and IL4 and require STAT6 and NFκB, but the γ1 locus is highly favored for CSR (8). We have addressed these questions by characterizing Igh chromatin topologies, GLT expression and CSR in the context of specific transcription factor deficiencies and GLT Pr substitutions in mice.

Here we report that long range interactions between I-S-C_H loci and Igh enhancers are independent of GLT production and STAT6, whereas the establishment and maintenance of these chromatin contacts requires NFκB p50. Replacement of the γ1 GLT Pr with the LPS responsive human metallothionein II_A (hMT) Pr (9) shows that the GLT Pr directly contacts the Igh enhancers and this looping is independent of productive transcription elongation. Strikingly, intercalation of the hMT Pr between the LPS inducible γ3 and γ2b loci constrains γ2b GLT expression and essentially abolishes direct μ->γ2b CSR whereas sequential μ->γ3->γ2b switching is retained, albeit at a reduced frequency. These findings demonstrate that specific long range contacts contribute spatial constraints that functionally impinge on gene expression, determine CSR targets and provide a mechanistic basis for direct IgG1- and sequential IgE switching.

MATERIALS AND METHODS

Mice, Cell Culture, Flow Cytometry and Statistics

Mice, C57BL/6 (WT), Stat6−/− and NFκB p50−/− (Nfkb1) on the C57BL/6 background were purchased from Jackson Laboratories. IgH^hMT/hMT mice (9) were kindly provided by C. O. Jacob (University of Southern California, CA) on the C57BL/6 background. All procedures involving mice were approved by the Institutional Animal Care Committee of the University of Illinois College of Medicine or the National Institute of Aging. Splenic B cells were sorted for CD43- resting B cells using CD43 magnetic microbeads (MACS, Miltenyi) according to the manufacturer’s instructions and cultured in 50 µg/ml LPS (Salmonella typhimurium, phenol extract, Westphal, Sigma-Aldrich), 10 ng/ml rIL-4 (R&D Systems). To prepare for flow cytometry, B cells activated for 4 days were washed in HBSS plus 2% FCS and stained with antibodies conjugated with fluorescein isothiocyanate (FITC-IgG3 -553403, FITC-IgG1 -553443; Pharmingen; FITC-IgG2b -406706; Biolegend) or with allophycocyanin-anti-mouse B220 (APC-B220 -103211; Biolegend). The flow cytometry analyses represented 5000–10,000 events and were gated for live lymphoid cells determined by forward and side scatter with CyanADP and Summit software (Becton Dickenson). P values were calculated by using two-tailed Student’s t test.

Real time quantitative RT-PCR, CT-PCR, and 3D DNA-FISH

Quantitative (q)-RT-PCR assays were performed as described (6) except that primers for 18S rRNA were used (10) to normalize samples. Quantitative circle transcript (CT) RT-PCR assays were carried out as described (11) with modifications. Primers γ2bF and γ3R were used for Iγ2b-Cγ3 CT assays (Table 1). Iγ2b-Cμ CTs were detected using the Iγ2bF and CμR.1 primers (12) (Table 1). Semi-quantitative RT-PCR for Iε−Cμ CTs was performed using primers IεF and CμR with Phire Hot Start II polymerase (Thermo Scientific) and an initial denaturation for 5 min at 98°C followed by 34 cycles of, 98°C for 5 sec, 60°C for 5 sec, 72°C for 8 sec, both in a 25 ul reaction on 5 fold serial diluted cDNA. 3D DNA FISH was carried out as described (13).

Table 1.

Primers for qRT-PCR, and CT analyses.

Assay	Primer1	Sequence	References
RT-PCR	HprtF	5’ GTTGGATACAGGCCAGACTTTGTTG 3’	(6)
“	HprtR	5’ TACTAGGCAGATGGCCACAGGACTA 3’	“
“	18SF	5’ TTGACGGAAGGGCACCACCAG 3’	(10)
“	18SR	5’ GCACCACCACCCACGGAATCG 3’	“
“	γ1F	5’ AGGAATGTGTTTGGCATGGAC 3’	(43>)
“	γ1R	5’ CACTGTCACTGGCTCAGGGAA 3’	“
“	γ3F2	5’ TGTCTGGAAGCTGGCAGGA 3’	(44)
“	γ3R	5’ GGCTCAGGGAAGTAGCCTTTG 3’	(45)
“	γ2bF	5’ TAAGCTGCGCACACCTACAGACAA 3’	“
“	γ2bR	5’ AGGGATCCAGAGTTCCAAGTCACA 3’	(44)
“	eF	5’ CAGAAGATGGCTTCGAATAAGAACA 3’	(46)
“	eR	5’ CGTTGAATGATGGAGGATGTGT 3’	“
“	AIDF	5’ CCATTTCAAAAATGTCCGCT 3’	(6)
“	AIDR	5’ CAGGTGACGCGGTAACACC 3’	“
DC-PCR	DC-ε F1	5’ CCGATTTGACCTACCAGATGCT 3’	(45>)
“	DC-ε F2	5’ TCTATGGGCATCAGGACCACTCC 3’	“
“	DC-μ R1	5’ TGAAGCCGTTTTGACCAGAAT 3’	“
“	DC-μ R2	5’ GGAGACCAATAATCAGAGGGAAGAA 3’	“
CT	Iγ2bF3	5’ CACTGGGCCTTTCCAGACCTA 3’	This paper
“	CμR.13	5’ TGGTGCTGGGCAGGAAGT 3’	(12)
“	IεF4	5’ GCGGCCCCTAGGTACTACCA 3’	(45)
“	CμR4	5’ AATGGTGCTGGGCAGGAAGT 3’	(11)

¹F and R indicate forward and reverse primers, respectively.

²γ3F and Iγ2bF are used for Iγ2b-Cγ3 CT.

³Iγ2bF and CμR.1 are used to detect Iγ2b-Cμ CT.

⁴IεF4 and CμR are used to assess Iε-Cμ CT.

Chromosome conformation capture (3C)

The 3C assay for the Igh locus was performed as described (6) and was optimized as follows. Cells were crosslinked with 1% formaldehyde for 8 minutes at room temperature and the reaction was quenched with glycine. All primers and probes for 3C product analyses were designed to avoid strain polymorphisms and are equally appropriate for both the C57Bl/6 and 129 mouse strains (Table 2). Several controls to confirm the efficacy of the 3C procedure were performed. Template concentration using the mb1 primers (Table 2) was determined as previously described (6). The efficiency of Hind III restriction site digestion in chromatin was monitored by real-time PCR analysis using primers spanning the restriction sites under study (Suppl. Table 1). To control for differences in amplification efficiency between primer sets we constructed a control template in which all possible 3C ligation products are present in equimolar concentration (Suppl. Table 2). DNA fragments that span restriction sites studied were PCR amplified, and mixed in equimolar amounts, then digested with Hind III, ligated and added to genomic DNA that had been digested and ligated, to serve as the 3C control template and which is used in a standard curve in quantitative 3C PCR analyses. To permit sample to sample comparisons the data were normalized using the interaction frequency between two fragments within the non-expressed Amylase I (Amy) gene, which are separated by 3.5 kb (Table 2). The equation used to calculate the relative crosslinking frequency between two Igh restriction fragments is: X_Igh = [S_Igh/S_Amy] Cell Type/[S_Igh/S_Amy] Control mix. S_Igh is the signal obtained using primer pairs for two different Igh restriction fragments and S_Amy is the signal obtained with primer pairs for the Amy I locus fragments. The crosslinking frequency for the two Amy1 fragments was arbitrarily set to a value of 1 to permit sample comparisons. A complete laboratory protocol for 3C is available upon request.

Table 2.

Primer and Probe Sequences Used in the Quantitative 3C assay

Probe1	Primer 2,3,4	Sequence	References
	mb1F	5’ CCACGCACTAGAGAGAGACTCAA 3’	(8)
	mb1R	5’ CCGCCTCACTTCCTGTTCAGCCG 3’	“
Amy		5’ TTGAATATGTACCGAGTACACATGGATGGTGCAT 3’	This paper
	T.AmyF	5’ GAGATCTTACGTAGGCACTTAGTGGTATAA 3’	“
	T.AmyR	5’ GCTTCCATGATACTCTATGTTCTTCCT 3’	“
A (Eμ)		5’ TGGCTTACCATTTGCGGTGCCTGGTTT 3’	(8)
	T.A-A*	5’ TCCACACAAAGACTCTGGACCTCT 3’	“
B (γ3)		5’ CTGACCCAGGAGCTG 3’	“
	T.B-B*	5’ CAGATCACAGGGTCCCAGGTT 3’	“
H (hs3b,4)		5’ TGACTCATCCACATCACCTTGCCTGTG 3’	This paper
H.1 (hs3b,4)		5'-CTCATCCACATCACCTTGCCTGTGTATTGTC-3'	(8)
	T.H-H*	5’ CTCCCACCAGCCAAGACAAT 3’	“
	T.H-H.1*	5'-ATAGGCCCTCCTCCCACCA-3'	“
	T.A	5’ GAAACCAGGCACCGCAAATG 3’	“
	T.B	5’ AGTAGATAGGACAGATGGAGCAGTTACA 3’	“
	T.C	5’ GTGATAATGAACTGAATCCCACATGTAC 3’	“
	T.D	5’ AGTACCCAGCATGTTCACATC 3’	“
	T.F	5’ AGGACCAAGGTTCACAGCCA 3’	“
	T.H	5’ GCCCCTAAGACCCTACTCTGCTA 3’	“

¹3C probes specific for unique HindIII fragments are denoted by a capital letter referring to the 3C fragments. The Amy probe is in the Amylase1 gene. The probes and anchor primers can be used in combination with any other 3C primers.

²F and R indicate forward and reverse primers, respectively.

³Anchor primers marked with the asterisk (^*) are used in combination with anchor probes.

⁴3C assay primers specific for each HindIII fragment are denoted by a capital letter referring to the 3C fragments.

Quantitative PCR for 3C Ligation Products

Quantitative PCR (qPCR) for 3C was used in combination with 5'FAM and 3'BHQ1 modified probes (IDT) to detect 3C products (Table 2). 3C primers were designed using Primer Express software (ABI) (Table 2). The qPCR protocol was run at 59C for 90 seconds to ensure maximum amplification and these conditions were used for all optimization and assay reactions. Primer and probe optimization were carried out according to the manufacturer’s recommendations, (http://www3.appliedbiosystems.com/cms/groups/mcb_support/documents/generaldocuments/cms_042996.pdf). Dilutions of control mix were assayed under optimized conditions to determine the linear range of amplification (64 to 0.03 pg of ligated mix per 50 ng of genomic DNA per µl). Finally, 3C chromatin template was analyzed in serial dilutions to determine the approximate minimum and maximum amount of DNA that yield a constant crosslinking frequency for a sample (10 to 200 ng). For all qPCR 3C reactions, 100 ng of chromatin were used. 3C assays for WT and NFκB p50−/− B cells were analyzed with the H probe and T.H-H* primer whereas WT, Stat6−/− and IgH^hMT/^hMT B cells were analyzed with the H.1 probe and T.H-H.1* primer (Table 2).

RESULTS

The γ1:3’Eα looping requires NFκB p50 and is independent of transcription

Given the centrality of Eμ:3’Eα looping to CSR (5, 6), we sought to better characterize the establishment and maintenance of Eμ:3’Eα interactions with regard to a requirement for transcription, transcription factors (TF), and specific transcriptional elements that promote or exclude GLT expression and S/S synapsis. To determine whether formation and/or maintenance of GLT Pr:enhancer contacts are dependent on transcription, we analyzed Stat6 or NFκB p50 deficient B cells for AID and GLT expression, CSR frequency and Igh looping interactions. The IL4 inducible STAT6 transcription factor, in collaboration with LPS inducible NFκB, bind to the promoters of GLT γ1 and ε, mediate their expression (14–16) and enhance AID transcription (17). As a control for B cell activation, AID transcripts are highly expressed upon induction with LPS alone or LPS+IL4, in both WT and Stat6−/− B cells (Fig. 1A, Table I). Typically, GLT γ3 and γ2b are stimulated by LPS, while GLT γ1 and ε also require IL4 (fig. 1A). Although GLT γ3 and γ2b expression is intact in Stat6 deficient B cells, the GLTs γ1 and ε fail to express, as expected (fig. 1A). LPS+IL4 represses γ3 and γ2b GLTs in WT but, not in Stat6 deficient B cells, demonstrating that STAT6 negatively regulates the expression of these GLTs either directly or indirectly (fig. 1A). We observe that Stat6 deficiency leads to impaired IgG1 switching in response to LPS+IL4 relative to WT, as expected. The findings are shown in a representative FACS analysis and summarized for multiple ex vivo B cell cultures from independent mice, (fig. 1B,C). In contrast, IgG3 switching was comparable in WT and Stat6 KO B cells stimulated with LPS alone (fig. 1B,C). Thus, the pattern of IgG1 and IgG3 switching in WT and Stat6−/− B cells is paralleled by the γ1 and γ3 GLT expression pattern (Fig. 1A–C).

Figure 1. NFκB but not transcription is required for γ1:3’Eα and Eμ:3’Eα looping.

Resting B cells from WT, Stat6−/− or NFκB p50−/− mice were stimulated with LPS or LPS and IL4^lo (10 ng/ml), for 40 hours (RT-PCR, 3C assays) or 4 days (FACS). A) Q-RT-PCR assays for GLT g3, g2b and g1 and AID were normalized to the 18S rRNA gene transcript using two to four samples from two to four independent experiments. B) FACS analyses of B cells stained with anti-IgG1 or anti-IgG3 in combination with anti-B220. Numbers indicate percentage of switched cells. C) Average percentage of IgG3 (top) or IgG1 (bottom) from FACS analyses of LPS or LPS+IL4^lo activated B cells. Each symbol represents a single mouse and the line indicates the average. D) Schematic for long range chromatin looping interactions for LPS activated WT B cells in which Eμ:3’Eα interacts with γ3 (a) or γ2b (b) loci. E) HindIII restriction fragments used in the 3C analysis are indicated (fragments A-D,G,H). 3C assays were anchored at Eμ (fragment A), hs1,2 (fragment G) or hs3b,4 (fragment H) were analyzed for interaction with I-S-C_H loci (fragments B, C, D). Two-tailed Students t test p values of 0.05 and 0.001 or less are indicated by * and **, respectively. F) 3C assays anchored at hs3b,4 (fragment H) interrogated interactions with the γ1 (C fragment) or Eμ (A fragment) as indicated. G) Q-RT-PCR assays for GLT g3, and g1 and AID from WT or NFkB p50−/− B cells activated with LPS or LPS+IL4 as indicated. Q-RT-PCR assays were normalized to the 18S rRNA gene transcript using two to four samples from two to four independent experiments.

B cell activation with LPS and LPS+IL4 leads to long range chromatin looping between I-S-C_H loci and the Eμ:3’Eα complex that in turn facilitates GLT expression and S/S synapsis (6) (fig. 1D). The differential expression of γ1 GLTs in STAT6 deficient and sufficient B cells allowed us to assess the requirement for transcription in establishing and maintaining γ1:3’Eα looping interactions using highly sensitive Taqman 3C assays. 3’Eα is composed of DNase HSs, termed hs3a, hs1,2, and hs3b,4 (18). Here, we evaluated the association of γ3, γ2b and γ1 loci with Eμ, 3’Eα hs1,2 and hs3b,4 in resting and activated WT B cells. LPS induction significantly increased Eμ:γ3 and Eμ:γ2b interactions, whereas LPS+IL4 treatment modestly reduced these interactions relative to LPS (fig. 1E, Table II), as previously observed (6). This reduction did not achieve statistical significance. Conversely, LPS+IL4 treatment significantly induced Eμ:γ1 (A:C) looping as compared to unstimulated B cells (fig. 1E). A similar trend of LPS and LPS+IL4 inducible contacts between I-S-C_H loci and hs1,2 and hs3,b4 was also found. However, the frequency of induced contacts between hs3b,4 with γ3, γ1, or γ2b was 4–8 fold higher than the equivalent interactions with Eμ (in all cases p<0.0005) and 1.5–3 fold higher than the equivalent interactions with hs1,2 (fragment G) (in all cases p<0. 01) indicating preferential association of I-S-C_H loci with hs3b,4 (fig. 1E). We conclude that the hs3b,4 element within 3’Eα is the primary point of contact with I-S-C_H loci, consistent with the critical requirement for hs3b,4 in mediating GLT expression (7). Based on the forgoing observations, 3C assays were focused to looping between hs3b,4 (fragment H), Eμ (fragment A) and γ1 (fragment C) loci.

Next we evaluated the frequency of the γ1:hs3b4 and Eμ:hs3b,4 looping interactions in Stat6 proficient and deficient B cells to assess the contribution of GLT expression to these chromatin contacts. In WT and Stat6−/− B cells, hs3b,4:γ1 (H-C) and hs3b,4:Eμ (H-A) contacts are comparably induced by LPS+IL4, indicating that these long range interactions are independent of both STAT6- and γ1 GLT expression (fig. 1F). Induction of hs3b,4:γ1 interactions by LPS+IL4 in Stat6 deficient B cells may occur through a STAT6 independent pathway involving the insulin receptor substrate family (19). In contrast, NFκB p50 deficient B cells do not express GLTs or AID, and fail to support hs3b,4:Eμ (H-A) (fig. 1 F,G) (17, 20). Although hs3b,4:γ1 (H-C) looping contacts are present in NFκB p50 deficient B cells these contacts are not IL4 inducible as compared to WT (fig. 1F). Evidence indicates that NFκB p65 (RelA) is required for inducible transcription elongation in some genes whereas, the role of NFκB p50 has been less clear (21). NFκB binding has also been detected in 3’Eα hs4 (22) and could be directly involved in long range looping. Together these studies show that 1) long range looping interactions between 3’Eα:γ1 and Eμ:3’Eα occur independent of GLT expression as observed in Stat6 deficient B cells, 2) that NFκB p50 is a major contributor to inducible γ1 GLT expression and hs3b,4:γ1 looping as well as the basic integrity of the Eμ:3’Eα complex as inferred in p50−/− B cells. NFκB has also been implicated in chromatin looping associated with Igκ gene expression (23).

Previous studies suggested AID deficiency led to a severe reduction of I-S-C_H:Eμ and Eμ:3’Eα interactions in stimulated B cells (6) raising questions regarding the relative contribution of transcriptional elements versus DNA damage to long range chromatin contacts. Due to breeding issues, the AID−/− mouse was re-derived by four backcrosses to C57Bl/6. We now find a smaller yet reproducible reduction of Eμ:hs3b,4 (A-H) interactions (p < 0.026) and hs3b,4:Eμ (H-A) (Suppl. fig. 1). We also find that the frequency of Eμ:γ2b (A-D) interactions is reduced and that hs3b,4:γ2b (H-D) contacts display a similar trend (Suppl. fig. 1). We conclude that induced Eμ:3’Eα and I-S-C_H:3’Eα chromatin interactions are modestly influenced by AID mediated DNA lesions. The earliest detectable switching occurs at 2.5 days post-B cell activation under our culture conditions (6). We isolate chromatin at 40 hours of B cell activation to capture B cell chromatin in a germline configuration prior to CSR. It is formally possible that low level μ->γ2b CSR occurs shortly after B cell activation. These early μ->γ2b CSR events would give the appearance of Eμ:γ2b (A-D) chromatin looping in 3C assays. However, this is not the case for Eμ:3’Eα (A-H and H-A) and 3’Eα:γ2b (H-D) interactions since these sites do not become contiguous post-switching and therefore are not prone to create false positives in the 3C assay. Therefore, the greatest contribution to Igh looping and S/S synapsis is via chromatin looping of GLT Prs and Igh enhancers with AID providing a more modest contribution. These findings also indicate that in p50−/− B cells the severe loss of Eμ:3’Eα looping, described above, is unlikely to be due to impaired AID expression.

GLT Pr identity determines chromatin contacts with Igh enhancers

Earlier studies showed that the enhancer elements, Eμ:3’Eα are in close spatial proximity in resting splenic B cells and this interaction is increased upon activation of CSR (6). Furthermore, downstream S regions gain proximity to the universal donor, Sμ, in a cytokine dependent fashion (6). However, it remained unclear whether the GLT Pr is responsible for bringing its associated S region into proximity with Sμ. To address this question we investigated whether the GLT Pr per se interacts with Eμ and 3’Eα. We studied IgH^hMT/hMT mice in which the LPS+IL4 reactive γ1 GLT Pr, is replaced by the LPS responsive human metallothionein II_A (hMT) Pr, referred to here as the γ1^hMT/hMT Pr (9). The γ1^hMT/hMT Pr lacks the Iγ1 splice donor (9) that is required for co-transcriptional pre-mRNA processing (24, 25). Unspliced or aberrantly spliced mRNA are prevented from leaving the nucleus by nuclear surveillance pathways, primarily mediated by the exosome (24). Despite impaired expression of γ1 GLTs, the γ1^hMT/hMT Pr is responsive to LPS as assessed in nuclear run-on assays (9) which detect induced transcription and are an indicator of elongation competent RNA polymerases (26). Activation of WT and IgH^hMT/hMT B cells with CSR inducers is confirmed by induction of AID transcripts and LPS+IL4 stimulation of ε GLTs (fig. 2A). It is unknown why AID expression is elevated in activated IgH^hMT/hMT B cells relative to WT cells. IgH^hMT/hMT B cells fail to express γ1 GLTs whereas WT B cells are competent in this regard (fig. 2A) (9). FACS analyses confirm an impaired IgG1 switching pattern in LPS+IL4 induced IgH^hMT/hMT B cells as compared to WT (fig. 2B,C). Furthermore, IgG3 switching in response to LPS activation is significantly reduced in IgH^hMT/hMT- versus WT B cells (fig. 2B,C). These findings indicate that substitution of the γ1 GLT Pr with a heterologous Pr has unanticipated collateral effects on the expression of adjacent I-S-C_H regions. This observation will be explored in the sections below.

Figure 2. The γ1^hMT/hMT locus perturbs γ2b and ε GLT expression.

Resting B cells from WT or IgH^{hMT/ hMT} mice were stimulated with LPS or LPS and IL4^lo (10 ng/ml) or IL4^hi (20 ng/ml) for 40 hours or as indicated. Two-tailed Students t test p values of ≤ 0.05 or 0.001 are indicated by (*) and (**), respectively. A) GLT g3, g2b, g1 and ε and AID expression were analyzed by qRT-PCR and signals were normalized to the 18S rRNA gene transcript using two to five samples from two to three independent experiments. B) FACS analyses of B cells stained with anti-IgG1 in combination with anti-B220. Numbers indicate percentage of switched cells. C) Average percentage of IgG3 (top) or IgG1 (bottom) from FACS analyses of LPS or LPS+IL4 activated B cells, as indicated. Each symbol represents a single mouse and the line indicates the average.

We tested the proposition that the GLT Pr is the primary regulatory element engaged in looping interactions with the Igh enhancers. In the WT, the LPS+IL4 reactive γ1 GLT Pr interacts with the Igh enhancers, while this contact profile is predicted to shift to LPS inducible looping for the γ1^hMT/hMT Pr (fig. 3A) (6). 3C studies for WT B cells show that LPS+IL4 induces abundant hs3b,4:γ1 (H-C) interactions (5.3 fold; p<0.007) relative to resting B cells while LPS stimulates a more muted induction (2.3 fold; p<0.016) (fig. 1E, 3B). A similar pattern albeit with lower frequency is found in WT for Eμ:γ1 (A-C), as expected (fig. 1E, 3C). In contrast, in IgH^hMT/hMT B cells, LPS preferentially increases hs3b,4:γ1^hMT/hMT (H-C) interactions (2.6 fold, p=0.02) but LPS+IL4 has less impact on long range contacts (p<0.013) relative to resting B cells (fig. 3B). A similar trend for Eμ:γ1^hMT/hMT (A-C) looping was observed but did not achieve statistical significance (fig. 3C). Importantly, equivalent LPS induced crosslinking frequencies were detected for hs3b,4:γ3 (H-B, B-H) and hs3b,4:γ2b (H-D) in WT and IgH^hMT/hMT B cells indicating that looping to flanking isotypes is unperturbed (fig. 3B,C). Thus, the shift from LPS+IL4 to LPS inducible looping is limited to the LPS responsive γ1^hMT/hMT Pr thereby identifying the Pr as the primary element engaging with the Igh enhancers. We further conclude that GLT Pr:enhancer interactions are modulated by different stimuli (LPS versus LPS+IL4), require specific transcription factors (NFκB p50), and that looping is independent of GLT expression.

Figure 3. Igh enhancers contact the γ1 GLT Pr in the absence of GLT production.

Resting WT or IgH^{hMT/ hMT} B cells were activated with LPS or LPS+IL4^lo for 40 hours then analyzed in 3C, 3D DNA FISH or ChIP assays as indicated. A) Schematic for long range looping interactions in LPS activated WT or IgH^hMT/hMT B cells in which Eμ:3’Eα interacts with γ3 (a), γ2b (b) or γ^hMT/hMT (c) loci. B–D) In 3C assays two to three 3C chromatin samples from at least two independent experiments were analyzed. B) Schematic showing HindIII restriction fragments (B,C,D) analyzed with anchor hs3b,4 (fragment H) (upper panel). 3C assays (lower panel). C) 3C HindIII restriction fragments (C,H) analyzed with anchor γ3 (fragment B) or Eμ (fragment A) (upper panel). 3C assays (lower panel). D) 3C HindIII restriction fragments (A,H) analyzed with anchor Eμ (fragment A) or hs3b,4Eμ (fragment H) (upper panel). 3C assays (lower panel). E) A 220 kb segment of the Igh locus (top). 3D DNA FISH probes h4 (green) (27), Eμ5 (red) (13) and BAC210H14 (blue) (Top panel). High resolution three color 3D-DNA FISH with WT splenic B cells. Probes were labeled with Alexa Fluor 594 (red) and 488 (green) and BAC RP23-201H14 was labeled with Alexa Fluor 697 (blue) and hybridized with fixed splenic B cells. Signals were visualized by epifluorescence microscopy and distances between probes were determined as described (13, 27). B cells were purified from two mice (Middle panel). Quantitation of FISH data is shown (Bottom panel). Distances between red and green FISH signals were divided into two categories (<0.2 µm, 0.2 – 0.5 µm) for ~100 nuclei. The probes h4 (green) and Eμ5 (red) correspond to hs3b,4 and Eμ, respectively. The percentage of Igh alleles in each group (y axis) was determined for each activation condition (x axis) and are displayed in different colors. P values of less than 0.0001 are indicated ***. F) ChIP assays were performed on nuclei from four to six samples derived from two independent experiments using anti-H3K9Ac or anti-H3K4me3 antibodies. All samples were analyzed in duplicate then averaged and SEM are shown. Histone modifications and primer pairs tested are indicated. The qPCR product concentration relative to 10% input and indicates the enrichment of a sequence following immunoprecipitation.

It is pertinent to ask whether the identity of the GLT Pr influences the abundance of Eμ:3’Eα contacts. This is because CSR requires a tripartite Eμ:3’Eα:GLT Pr interaction complex to provide synapsis between Sμ and the targeted downstream S region (fig. 3A) (6). In WT B cells, hs3b4:Eμ (H-A and A-H) interactions are of approximately equivalent frequency in response to LPS+IL4 and LPS (fig. 3D). In contrast, in IgH^hMT/hMT B cells, hs3b4:Eμ contacts are significantly (H-A, p<0.05) or modestly (A-H, p<0.067) impaired in response to LPS+IL4 versus LPS (fig. 3D). These findings strongly suggest that γ1 GLT Pr interactions with Eμ and 3’Eα contributes to the juxtaposition of tripartite Eμ:3’Eα:GLT Pr complex leading to synapsis of donor Sμ and downstream acceptor S regions (fig. 3A).

To independently assess the frequency of the prominent long range Eμ:3’Eα looping interactions in WT and IgH^hMT/hMT B cells we carried out high resolution 3D-FISH with 10 kb probes. The H14 BAC probe hybridizes outside the Igh locus and independently identifies chromosome 12. Previously described probes h4 (27) and Eμ5 (13), separated by about 200 kb, were employed to mark sequences close to Eμ and within 3’Eα, respectively (fig. 3E, top). This probe combination produced closely juxtaposed signals in LPS+IL4 activated B cells (0.1529 +/−0.0621 µm) but not in resting B cells (0.2719 +/− 0.0987 µm) (fig. 3E). We examined the proportion of Igh alleles in which two FISH signals are separated by different interprobe distances. The separation distances between FISH signals were decreased to < 0.2 µm in ~80% of Igh alleles originating in LPS+IL4 activated B cells as compared to ~20% of resting B cells (fig. 3E). In ~80% of resting B cells the signals were separated by 0.2–0.5 µm as compared to 20% in activated B cells. However, the separation distances detected for LPS+IL4 activated IgH^hMT/hMT B cells were similar to WT (0.1483 +/−0.0546 µm). Thus, unlike our 3C assays the 3D FISH studies did not detect the deficit of Eμ:3’Eα looping in IgH^hMT/hMT B cells. One explanation for this discrepancy is that the considerably higher formaldehyde concentration used in 3D FISH (4%) as compared to 3C assays (1%) may obscure the partial Eμ:3’Eα looping deficiency in IgH^hMT/hMT B cells, as suggested (28). Another explanation for these findings is that 3D-FISH reports more global chromatin interactions whereas 3C assays are capable of detecting chromatin contacts at the molecular level. Consequently, 3D-FISH detects a similar LPS+IL4 induced Igh locus conformation in both WT and IgH^hMT/hMT B cells, whereas 3C assays indicate that Eμ and 3'Eα are less frequently in molecular proximity in IgH^hMT/hMT B cells. These studies independently confirm that Eμ:3’Eα looping interactions in resting B cells are significantly induced by CSR stimuli.

Bromo- and chromo-domain factor binding to activating histone modifications at Pr proximal positions and at enhancers could in principle mediate Pr:enhancer looping (29). In transcribed I-S-C_H regions, trimethylated histone 3 lysine 4 (H3K4me3) and acetylated histone 3 lysine 9 (H3K9Ac) marks accumulate in S regions as a result of R-loop formation and RNA polymerase II (Pol II) pausing (30). H3K4me3 and H3K9Ac are enriched at the Sμ locus upon LPS and LPS+IL4 activation in WT and IgH^hMT/hMT B cells, whereas, these marks were severely depleted at the Sγ1^hMT/hMT locus relative to WT (Fig. 3F). We conclude that factor binding to S region epigenetic marks is dispensable for the formation and maintenance of γ1^hMT/hMT Pr:Igh enhancer looping.

Chromatin topology constrains the order of isotype switching

Intercalation of the γ1^hMT/hMT Pr between the γ3 and γ2b loci, offers an opportunity to assess the contribution of locus topology to differential isotype choice during CSR (fig. 3A,4A). We now ask whether interposition of the γ1^hMT/hMT Pr between the LPS responsive γ3 and γ2b loci will create a new topological constraint by means of LPS induced γ1^hMT/hMT Pr:3’Eα looping that now inhibits direct μ->γ2b CSR. Such a constraint may underpin sequential CSR, which occurs when the Sμ donor recombines with a proximal downstream I-S-C_H locus to form hybrid S/S regions (fig. 4A). The hybrid S/S regions can in turn serve as a donor substrate in a second round of CSR events (31) (fig. 4A). Thus, Sγ2b, located downstream of Sγ1, may become available for CSR only following the removal of the constraints contributed by γ3:3’Eα:Eμ and γ1^hMT/hMT:3’Eα:Eμ looping (fig. 3A).

Figure 4. GLT γ1^hMT/hMT Pr alters the balance between direct and sequential CSR.

Resting B cells from three WT and three IgH^hMT/hMT mice were independently stimulated with LPS or LPS and IL4^lo for 42 hours (DC-PCR, CT assays) or 4 days (FACS). A) Diagram depicting direct μ->γ2b (above the line) or sequential μ->γ3->γ2b (below the line) CSR. B) FACS analyses of B cells stained with anti-IgG2b in combination with anti-B220. Numbers indicate percentage of switched cells. C) Average percentage of IgG2b from FACS analyses of LPS or LPS+IL4^lo activated B cells, as indicated. Each symbol represents a single mouse and the line indicates the average. D) Circle transcript (CT) assays measure hybrid Ig2b-Cμ transcripts which arise from excised circular DNA generated by direct μ->γ2b CSR and are detected using a forward primer in Iγ2b together with a reverse primer in Cμ in qRT-PCR and normalized to the 18S rRNA gene transcript using 6 samples from three independent experiments. E) CT Ig2b-Cγ3 switching are detected using a forward primer in Iγ2b together with a reverse primer in Cγ3 indicating sequential CSR by qRT-PCR and signals were normalized to the 18S rRNA gene transcript using 6 samples from three independent experiments. F) Schematic of the DC-PCR assay with nested PCR primers and Eco RI (RI) sites indicated by the arrows and filled circles, respectively (upper panel) are compared to the nAChR gene used as a loading control. Representative gel images for the semi-quantitative DC-PCR assays (lower panel). PCR products were harvested in the second round at 26 (lanes 1,4,7,10), 29 (lanes 2,5,8,11), and 32 cycles (lanes 3,6,9,12). G) CT assays measure hybrid Iε-Cμ transcripts which are detected using a forward primer in Iε with a reverse primer in Cμ using semi-quantitative RT-PCR are compared to the Hprt loading control.

Because the relative strength of γ3 and γ2b GLT expression in WT and IgH^hMT/hMT B cells could be an indicator of constraints on I-S-C_H looping with 3’Eα, we assessed GLT expression levels using qRT-PCR. The γ3 GLT was expressed similarly in WT and IgH^hMT/hMT B cells upon treatment with LPS (fig. 2A). Furthermore, γ3 and γ2b GLT expression levels were reduced in response to LPS+IL4^lo as compared to LPS alone in both strains of mice, indicating appropriate IL4 dependent repression (fig. 2A). In contrast, the γ2b GLT was reduced (2.8-fold, p<0.021) in IgH^hMT/hMT B cells relative to WT, consistent with the notion that the γ1^hMT/hMT Pr element or its chromatin conformation interferes with γ2b GLT expression (fig. 2A). Next, the impact of the LPS reactive γ1^hMT/hMT Pr on IgG3 and IgG2b switching was directly tested. Following LPS induction, the frequency of IgG3⁺ IgH^hMT/hMT B cells is modestly though significantly lower (1.51-fold; p<0.002) relative to WT (fig. 2B,C). Similarly, FACS analyses show a reduced incidence (1.7-fold; p>0.016) of IgG2b+ IgH^hMT/hMT B cells relative to WT in response to LPS activation (fig. 4B,C). The elevated levels of AID in IgH^hMT/hMT B cells appear not to compensate for the diminished expression of γ2b GLTs (fig. 2A). Together these findings indicate that the γ1^hMT/hMT Pr identity impairs switching at both I-S-C_H loci flanking the γ1^hMT/hMT Pr.

It was of interest to determine whether direct or sequential IgG2b switching was differentially impacted by the presence of the γ1^hMT/hMT Pr. Circle transcript (CT) assays measure hybrid Ig2b-Cμ transcripts which arise from excised circular DNA generated by direct μ->γ2b CSR. These CTs are detected using a forward primer in Iγ2b together with a reverse primer in Cμ in qRT-PCR assays (Fig. 4D, Table I). Sequential IgG2b switching through the γ3 locus involves μ->γ3 (step 1) followed by γ3->γ2b (step 2) CSR and is detected in Iγ2b-Cγ3 CT assays (fig. 4E). Quantitative Iγ2b-Cμ CT assays show that direct μ->γ2b CSR was severely diminished 12.5-fold (p>0.02) in LPS activated IgH^hMT/hMT B cells relative to WT (fig. 4D). In contrast, sequential CSR through the γ3 locus is reduced approximately 3.1-fold (p>0.018) and is generally in accord with the lower incidence of IgG3 and IgG2b switching relative to WT (fig. 2B, 4B–C, 4E). We conclude that insertion of the γ1^hMT/hMT Pr between the γ3 and γ2b loci substantially impairs direct μ->γ2b CSR whereas sequential μ->γ3->γ2b format remains largely intact. Thus, LPS inducible γ1^hMT/hMT Pr:3’Eα contacts appear to create a chromatin architecture that interferes with the ability of the γ2b locus to directly access the Sμ donor substrate.

The mechanism underlying the strong propensity for IgG1 versus IgE switching despite very similar GLT Prs remains unclear. Simple genomic proximity of the endogenous γ1 locus to Sμ may favor CSR to that locus and simultaneously create a topological chromatin constraint restraining recombination with more distal loci. In circumstantial support of this notion, switching to IgE occurs predominately through a sequential mechanism and yields hybrid Sμ/Sγ1/Sε junctions (32–34). We reasoned that in response to LPS+IL4 γ1 GLT Pr:3’Eα looping may create a topological constraint that precludes direct μ->ε CSR. Replacement of this element with the γ1^hMT/hMT Pr that fails to interact with 3’Eα in response to LPS+IL4 will remove this constraint and permit direct IgE switching. Therefore, we examined direct and sequential IgE switching in WT and IgH^hMT/hMT B cells. FACS analysis of IgE switching is complicated by the expression of CD23/FcεRII receptors on activated B cells, which bind secreted IgE present in culture medium. To obviate this issue, we used semi-quantitative DC-PCR and CT assays to measure the relative frequency of total and direct μ->ε CSR, respectively. DC-PCR indicates that CSR is appropriately induced or repressed by LPS or LPS+IL4 and that total μ->ε CSR is very similar in WT and IgH^{hMT/ hMT} B cells (fig. 4F). CT assays show that Iε-Cμ transcripts are expressed at significantly higher levels in IgH^hMT/hMT B cells than in WT B cells when activated by LPS+IL4 (fig. 4G, compare lanes 1–3, 7–9). Thus, in contrast to WT, direct μ->ε CSR is dominant in IgH^hMT/hMT B cells, as when the γ1 Pr is deleted (35). Accordingly, the shift from sequential to direct IgE switching was recently reported in hMT mice based on analyses of hybrid Sμ/Sε junctions (36). These observations argue that in the WT context γ1 GLT Pr:3’Eα interactions preclude ε locus contacts with Sμ. We conclude that topological challenges contributed by the γ1 locus constrain switching to downstream isotypes.

DISCUSSSION

Our studies demonstrate that the establishment and maintenance of induced GLT Pr:enhancer chromatin interactions are contingent upon the integrity of the GLT promoter, and NF-κB activation by CSR stimuli. I-S-C_H:Igh enhancer interactions form in the absence of GLT production, and are independent of activating chromatin modifications associated with productive transcription elongation or STAT6. The dispensability of productive transcription elongation for GLT Pr:enhancer looping focuses attention to the early stages of transcription. RNA polymerase II (Pol II) transcription is comprised of two phases; initiation and elongation. Following recruitment of Pol II to promoters, the C-terminal domain (CTD) heptapeptides become phosphorylated at serine-5 (p-ser-5) permitting transcription initiation (26). During initiation, Pol II pauses ~40 bp downstream of the transcription start site prior to elongation (26). Following phosphorylation of CTD ser-2, the paused Pol II is released for elongation (26). Although the paradigm for inducible gene expression has been signal dependent Pol II recruitment and transcription initiation, several reports suggest that some genes (37) are regulated post-transcription initiation, in accord with genome-wide studies of Pol II occupancy (26). In some cases, ongoing transcription elongation is dispensable for sustaining preformed chromatin loops (38, 39). Our studies indicate that GLT Pr:hs3b,4 looping can form in the absence of transcription elongation but do not discriminate whether signal induced RNA Pol II engagement with the GLT Pr is necessary and sufficient for Pr:enhancer engagement or transcription initiation is also required. Notably, long range Igh locus looping interactions involving transcriptionally active Eμ and Pax5 activated intergenic repeat (PAIR) sites remain intact when transcription elongation is inhibited but are disrupted when transcription machines are disrupted (40). Accordingly, promoter centered chromatin looping provides a topological infrastructure for transcription regulation (41). Together these findings suggest that some feature(s) of the transcription apparatus contribute to the spatial organization of genetic elements in chromatin. However, it remains unclear precisely what those features are and how they may vary with the transcriptional requirements of specific loci.

We show here that DNA looping in the Igh locus is driven by association of transcriptional elements and has the functional property of influencing partner selection during CSR. Emerging evidence suggests extensive Pr/Pr and Pr/enhancer interactions exist in multigene complexes and these interactions can cooperatively regulate the transcriptional activity (41). Therefore it might be expected that co-regulated GLT Prs would be cooperatively expressed. Although IgG1 and IgE are induced by identical stimuli and require STAT6 and NFκB, the γ1 locus is highly favored for CSR (42). Our studies indicate that the LPS inducible hMT Pr engages in GLT Pr:enhancer looping and thereby creating chromatin constraints that impact on the downstream LPS responsive γ2b locus by reducing GLT expression and impairing direct IgG2b switching. We find that the potentially co-regulated GLT Prs are constrained by the 3D architecture of the multigenic Igh locus. These findings represent an example of signal induced Pr:enhancer interactions that constrain a similarly induced downstream locus, limit gene expression and determine recombination outcomes. Thus, chromatin looping per se contributes positively and negatively to the efficacy of gene expression.

Abbreviations

AID

activation induced cytidine deaminase

CTD

C-terminal domain

3C

chromosome conformation capture

CSR

class switch recombination

CT

circle transcript

(DC)-PCR

digestion circularization

DSBs

DNA double strand breaks

hs

DNase hypersensitive sites

GLT

germ line transcription

hMT

human metallothionein II_A

PAIR

Pax5 activated intergenic repeat

p-ser-5

phosphorylated at serine-5

Pr

promoter

q

quantitative

Pol II

RNA polymerase II

S

switch

TF

transcription factors