Differential regulation of histone acetylation and generation of mutations in switch regions is associated with Ig class switching

Abstract

Class switch recombination (CSR) allows B cells to make effective protective antibodies. CSR involves the replacement of the μ constant region with one of the downstream constant regions by recombination between the donor and recipient switch (S) regions. Although histone H3 hyperacetylation in recipient S regions was recently reported to coincide with CSR, the relative histone H3 and H4 acetylation status of the donor and recipient S regions and the relationship between the generation of mutations and histone hyperacetylation in S regions have not been addressed. Here we report that histone H3 and H4 were constitutively hyperacetylated in the donor Sμ region before and after different mitogen and cytokine treatments. We observed an increased frequency of mutations in hyperacetylated Sγ DNA segments immunoprecipitated with anti-acetyl histone antibodies. Furthermore, time course experiments revealed that the pattern of association of RNA polymerase II with S regions was much like that of H3 hyperacetylation but not always like that of H4 hyperacetylation. Collectively, our data suggest that H3 and H4 histone hyperacetylation in different S regions is regulated differently, that RNA polymerase II distribution and H3 hyperacetylation reflect the transcriptional activity of a given S region, and that transcription, hyperacetylation, and mutation are not sufficient to guarantee CSR. These findings support the notion that there are additional modifications and/or factors involved in the complex process of CSR.

High-affinity IgG, IgA, and IgE antibodies protect higher organisms from infection by pathogenic organisms and other environmental threats. The generation of effective protective antibodies requires B cells to carry out somatic hypermutation (SHM) and class switch recombination (CSR), two related but quite different DNA transactions (1). SHM introduces many point mutations in the variable (V) regions of the Ig heavy and light chain genes that encode the antigen-binding site of the antibody molecule (2). In contrast to SHM, CSR is a region-specific recombination-deletion process that requires the generation of double-stranded DNA breaks (DSB) (3). These DSB are generated in the donor μ switch (S) region that is just upstream of the μ constant (C) region gene and in a downstream recipient γ, ε, or α S region (4, 5). Recombination then brings one of those downstream C regions into proximity to the V region. This allows the mutated heavy chain V region to be expressed with one of the C regions so that each antigen-binding site will be able to mediate different effector functions and be distributed throughout the body.

Despite the different outcomes of SHM and CSR, activation-induced cytidine deaminase (AID) is required for both processes (6) presumably because of its ability to deaminate deoxycytidine to deoxyuridine on single-stranded DNA (ssDNA) (7-11). Both SHM and CSR require transcription (12), suggesting that the ssDNA substrate for AID is created by the generation of transcription bubbles (7) and/or perhaps triplex RNA-DNA structures called R-loops in S regions (13, 14). Both the donor Sμ and downstream S regions contain GC-rich repeats rich in hot spots that can be targeted by AID (3, 4). AID-generated G·U mismatches are then resolved by DNA repair mechanisms that lead to mutations in the S regions (1, 3, 15). It has been proposed that the dUs in the S regions can be removed by uracil N-glycosylase (UNG), followed by apurinic/apyrimidinic endonuclease to generate single-stranded nicks (15, 16). A recent study suggests that UNG may play a scaffold role rather than an enzymatic role (17), indicating that more needs to be done to elucidate UNG's role in CSR. Closely spaced nicks on both strands can be converted into DSB (18, 19), or other enzymatic systems such as mismatch repair may create DSB that initiate recombination between the S regions (1).

Because AID could be highly mutagenic for any highly transcribed gene (20-22), there must be mechanisms to selectively target AID to the V region to produce SHM and to the S regions to initiate CSR. This is especially likely because SHM and CSR can occur independently, even though they are observed at the same time in B cell differentiation. In addition, the mistargeting of AID can result in mutations that cause the aberrant expression of protooncogenes and chromosomal rearrangements, leading to B cell malignancies (23). CSR requires transcription initiated at a promoter that is upstream of each of the S regions except δ. This transcription is sterile, in that the germ-line transcripts (GLT) are not translated into proteins (4, 24), and is presumably required to provide ssDNA either in the form of transcription bubbles (7) or R-loops with the nontranscribed strand being single stranded (13, 14). The production of GLTs precedes the generation of AID-induced mutations and of the DSB that are required for successful CSR (5, 6, 25).

It has been suggested that increased accessibility contributes to the selective targeting to Ig genes and that this could be accomplished through modifications of chromatin that have been implicated in targeting transcription, replication, and other repair processes (26-28). It has been well established that modifications of histones H3 and H4 are essential for stepwise V(D)J recombination and additional steps in B cell differentiation (29, 30). Recent studies have also implicated histone acetylation of H3 and H4 in targeting of V regions for SHM (31) and H3 acetylation of recipient S regions in targeting CSR (32). However, the details of how the chromatin of particular S regions is selectively modified by stimulation with different cytokines remain unclear, and the modification of the histones associated with the donor Sμ has not been reported. In addition, the role of H4 hyperacetylation relative to that of H3 in S regions has not been determined. Moreover, AID-dependent mutations in the S regions, which precede the recombination event, have not been directly linked to histone hyperacetylation. In particular, the relationship between sterile transcription, histone acetylation, and the targeting of AID to the S regions is not clear. In this study, we have attempted to answer some of these questions by stimulating primary splenic B cells in short-term culture with either lipopolysaccharide (LPS) to induce switching to IgG3 or LPS plus IL-4 to induce switching to IgG1 and by examining the changes in histone acetylation of the donor Sμ and recipient Sγ3 and Sγ1 by chromatin immunoprecipitation (ChIP). In addition, we compared the changes in histone acetylation with the density of RNA polymerase II (RNAP II) and examined the mutations of the hyperacetylated S regions in the immunoprecipitated hyperacetylated DNA.

Materials and Methods

ChIP. Splenic B cells were obtained from wild-type mice and were stimulated with LPS or LPS plus IL-4 as described in ref. 33. About 70 million cells in culture were directly crosslinked by addition of formaldehyde to a final concentration of 1% at room temperature for 4.5 min before the reaction was quenched with 2.5 M glycine on ice. Cells were washed once in PBS and then incubated on ice for 20 min in 25 ml of buffer containing 10 mM Tris (pH 8.0), 10 mM EDTA, 0.5 mM EGTA, and 0.25% Triton X-100. Cells were pelleted by centrifugation at 2,500 rpm for 5 min, resuspended in 25 ml of buffer containing 10 mM Tris (pH 8.0), 1 mM EDTA, 0.5 mM EGTA, and 200 mM NaCl, and shaken at room temperature for 20 min. Nuclei were collected by centrifugation at 2,500 rpm for 5 min and resuspended in 4 ml of buffer containing 1% Triton X-100, 4 mM EDTA, 40 mM Tris (pH 8), and 300 mM NaCl. Sodium butyrate (10 mM) (Upstate Biotechnology) and Complete protease inhibitors (Roche) were freshly added to each buffer. Chromatin was then sonicated to 200- to 500-bp fragments before being precleared as described in ref. 31. About 30 μg of precleared chromatin was incubated on a rotator overnight with 3 μg of each antibody (RNAP II, sc-899, Santa Cruz Biotechnology; Ac-H3, 06-599, Upstate Biotechnology; Ac-H4, 06-866, Upstate Biotechnology; normal rabbit IgG, 12-370, Upstate Biotechnology) in 1 ml of buffer containing 0.01% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris (pH 8), and 150 mM NaCl at 4°C. Then, 35 μl of 50% protein A/G Sepharose flurry was added and rotated at 4°C for 4 h. The beads were recovered, washed, reverse crosslinked, and purified as described in ref. 31. DNA samples were analyzed by PCR using Taq gold polymerase (Applied Biosystems) at 95°C for 30 sec, 60°C for 45 sec, and 72°C for 30 sec for 29, 30, or 33 cycles for Ac-H4, Ac-H3, and RNAPII ChIP, respectively. The rabbit IgG control was amplified for 33 cycles. ChIP primers for Sμ were 5′-AACTAGGCTGGCTTAACCGAGATG-3′ [forward (5198-5221)] and 5′-GTCCAGTGTAGGCAGTAGAGTTTA-3′ [reverse (5288-6265)]. ChIP primers for Sγ1 were 5′-GGAGGTCCAGTTGAGTGTCTTTAG-3′ [forward (8787-8810)] and 5′-TTGTTATCCCCCATCCTGTCACCT-3′ [reverse (8894-8871)]. ChIP primers for Sγ3 were 5′-CAGGCTGGGAAACTCTTG-3′ [forward (1887-1904)] and 5′-GGTCCCCACATCCTCACTTAT-3′ [reverse (2031-2011)].

Mutation Analysis of the Immunoprecipitated S Regions. Input DNA and DNA immunoprecipitated with anti-hyperacetylated H3 or H4 were used to amplify S regions with PfuTurbo polymerase (Stratagene) at 94°C for 30 sec, 60°C for 30 sec, and 72°C for 60 sec for 35 cycles. The primers used to amplify the Sμ region for sequencing were 5′-AGAAGGCCAGACTCATAAAG-3′ [forward (4973-4992)] and 5′-CTCACCCCAACACAGCGTAGC-3′ [reverse (5347-5327)]. The primers used to amplify the Sγ1 region were 5′-ACAGGGAAGCTATAGGAAAACCAG-3′ [forward (8138-8161)] and 5′-AGAATCCCCAACTACTACTTATCC-3′ [reverse (8558-8535)]. The primers used to amplify the Sγ3 region were 5′-TGGGGGAGCTGGGGTAGGTTC-3′ [forward (2046-2066)] and 5′-GCCAGGTCTCCATATTCCCACTTA-3′ [reverse (2421-2398)]. PCR products were cloned into pCR4-TOPO blunt vector (Invitrogen). DNA sequencing was performed at the Albert Einstein Cancer Center DNA sequencing facility.

Results

The Recruitment of RNAP II to the S Regions. As noted in the Introduction, the process of switching requires transcription initiated at a promoter that is 5′ to each of the downstream S regions. Transcription is usually associated with hyperacetylation of H3 and/or H4 (26), which could also help to make the ssDNA in the transcription bubbles or R-loops in specific S regions accessible to AID (7, 9, 13, 14). However, the presence of GLT does not always correlate with, and is not sufficient for, switching to a particular C region. For example, when mouse splenic B cells are stimulated with LPS plus IL-4, promoting switching to IgG1 but not IgG3, GLTs of both the Sγ3 and Sγ1 regions are observed (32). Because our goal is to determine how differential regulation of CSR is accomplished, we first examined the density of RNAP II because it reflects the relative number of transcription bubbles or R-loops in different S regions (34). To do this, we stimulated the mouse primary splenic B cells with either LPS or LPS plus IL-4 for 18, 48, 72, or 96 h. When assayed for surface Ig at 96 h, ≈10% of the LPS-stimulated B cells had switched to IgG3 and ≈30% of the B cells stimulated with LPS plus IL-4 had switched to IgG1 (data not shown). Unstimulated and stimulated splenic B cells at each time point were fixed with formaldehyde, and their chromatin was sheared to produce an average chromatin fragment of 200 to 500 bp. RNAP II was immunoprecipitated with an antibody that does not discriminate between different modifications of the C-terminal domain of RNAP II, so the ChIP results would reflect the number of RNAP II molecules associated with the examined DNA sequence (34). The immunoprecipitated DNA was amplified with primers specific for each of the S regions that generated ≈100-bp PCR fragments (see Materials and Methods) from regions that undergo high rates of recombination (33, 35-39). At each time point and each treatment, the ChIP was done with similar amounts of chromatin as demonstrated by the PCR products of the DNA before immunoprecipitation (Fig. 1, Input).

Fig. 1.

The kinetic association of RNAP II, acetylated histone H3 (H3-Ac), and H4 (H4-Ac) with the S regions upon stimulation. Primary mouse splenic B cells were purified and stimulated in vitro with LPS or LPS plus IL-4 for the indicated hours. ChIPs were performed with antibodies against RNAP II, H3-Ac, and H4-Ac and with normal rabbit IgG (RIgG) as a negative control. Each of the immunoprecipitated DNA was amplified with primers for the Sμ, Sγ1, and Sγ3 regions. m, no stimulation; L, LPS stimulation; L+I, LPS plus IL-4 stimulation.

In splenic B cells stimulated with LPS or LPS plus IL-4, there was an increase in the amount of RNAP II associated with Sμ between 18 and 48 h after stimulation (Fig. 1). This increase was maintained through 96 h in the cells stimulated with LPS but appeared to decrease at 96 h in cells stimulated with LPS plus IL-4 (Fig. 1). This was confirmed by comparing the semiquantitative PCRs at 48 h and 96 h (Fig. 2).

Fig. 2.

Semiquantitative PCR with 4-fold serial dilutions of ChIPed samples 48 (A) and 96 (B) h after the indicated treatments were amplified with primers for the Sμ, Sγ1, and Sγ3 regions. Other symbols are the same as in Fig. 1.

Treatment with LPS plus IL-4, which stimulates Sγ1 GLT and switching to IgG1, resulted in an increase in the amount of RNAP II associated with the Sγ1 region at 48 and 72 h (Figs. 1 and 2A). There was a decrease at 96 h (Figs. 1 and 2B) that resulted in similar amounts of RNAP II being associated with the Sγ1 region in LPS and LPS plus IL-4 treated cells at 96 h (Figs. 1 and 2B). Treatment of the cells with LPS, which does not cause the production of GLT from the Sγ1 region or switching to IgG1, resulted in little change in the amount of RNAP II associated with Sγ1 throughout the time course (Fig. 1).

LPS, which stimulates Sγ3 GLT and switching to IgG3, caused an increase in the amounts of RNAP II associated with the Sγ3 at 48 and 72 h that continued through 96 h (Fig. 1). However, LPS plus IL-4 also caused an increase in the amount of RNAP II associated with Sγ3 by 48 h, consistent with the reports that LPS plus IL-4 induces Sγ3 GLT, even though there was no switching to IgG3 (32, 52). The density of RNAP II associated with Sγ3 decreased at 72 and 96 h (Figs. 1 and 2B), also correlating with the decrease of Sγ3 GLT at the same time (51). Thus, the density of RNAP II, associated with the parts of the γ3 and γ1 S regions that frequently undergo recombination (33, 36, 41), was consistent with the sterile transcription of those S regions as reported in the literature (40, 51, 52) and probably provides a rough measure of the rate of transcription of recipient S regions (34).

Acetylation of Histone H3 Associated with the S Regions. To examine whether stimulation with LPS or LPS plus IL-4 triggers differential acetylation of histones associated with different S regions, ChIP was carried out with an antibody against acetylated H3. The antibody reacts with H3 acetylated in any of the lysines in its N-terminal tail (Upstate Biotechnology). Hyperacetylation of H3 often precedes, and is associated with, the increased transcription of individual genes (26), so it would not be surprising if the pattern of increase in acetylation mirrored the density of RNAP II at the same sites. In fact, the donor Sμ showed the same relative amounts of acetylation of H3 in the unstimulated cells as in the cells stimulated with LPS or LPS plus IL-4 for 18, 48, 72, and 96 h (Fig. 1). This finding suggested that H3 associated with Sμ was constitutively hyperacetylated and remained so even when less RNAP II was recruited to the Sμ region before 48 h and after 96 h.

Semiquantitative PCR at 96 h indicated that LPS induced more acetylation of H3 associated with Sγ3 than Sγ1, whereas LPS plus IL-4 induced more acetylated H3 at Sγ1 than at Sγ3 (Fig. 2B). This finding correlates very well with the induced switching specificity. The time course experiments revealed that LPS plus IL-4 caused a relative increase in the acetylation of H3 associated with Sγ1 at 48 and 72 h, mirroring the increase in RNAP II, whereas LPS alone did not stimulate a similar increase in acetylation of H3 associated with Sγ1 (Figs. 1 and 2A). The difference in the levels of acetylated H3 induced by LPS and LPS plus IL-4 persisted at 96 h even though the difference of the amounts of RNAP II decreased at that time (Figs. 1 and 2B). Stimulation with LPS alone caused an increase in the acetylation of H3 associated with Sγ3, and this hyperacetylation of H3 persisted at 96 h, consistent with persistent association of RNAP II with Sγ3 at that time (Fig. 1). Stimulation with LPS plus IL-4 also leads to a certain amount of H3 acetylation at Sγ3 at 48 and 72 h (Figs. 1 and 2A) with a decrease at 96 h (Fig. 2B). The same pattern was observed with the RNAP II ChIP.

Acetylation of Histone H4 Associated with the S Regions. As with H3, the acetylation of H4 associated with Sμ did not change with stimulation with either LPS or LPS plus IL-4 (Figs. 1 and 2). However, there was an increase in the acetylation of H4 associated with Sγ1 at 48 and 72 h after stimulation with both LPS and LPS plus IL-4. The increase after LPS stimulation did not correlate with either RNAP II or H3 acetylation at Sγ1. At 96 h, there was less acetylation of H4 after LPS than LPS plus IL-4 (Fig. 2B) at Sγ1, and this was similar to H3 acetylation (Figs. 1 and 2B). Acetylated H4 was associated with Sγ3 after both LPS and LPS plus IL-4 stimulations from 18 to 72 h but decreased at 96 h (Figs. 1 and 2B). This again suggested that acetylation of H4 precedes and persists after the increase in the recruitment of RNAP II and is roughly the same, regardless of whether recombination occurs.

The Frequency of Mutations in the Immunoprecipitated DNA. CSR is initiated by targeting AID to the S regions to generate G·U mismatches. Mutations observed in S regions are a reflection of this process because some of the G·U mismatches are fixed to generate mutations, whereas others are processed to generate DSB that result in recombination. Therefore, S-region mutations should reflect the process by which AID is being targeted to the S regions and initiating CSR. We therefore examined whether Sμ and Sγ segments precipitated in the H3 and H4 ChIP assay had accumulated mutations and whether the relative frequency of mutations correlated with the changes in H3 and/or H4 hyperacetylation. For each S region, we compared the mutation frequency in the immunoprecipitated S DNA segments to the input DNA from the same segments at 96 h after stimulation with LPS or LPS plus IL-4. The primers that were used for sequencing amplified longer regions (350-450 bp) than the regions amplified in ChIP assay (≈100 bp) to obtain enough mutations to analyze. The primers were designed such that the sequenced regions either contained or were next to the ChIPed regions (see Materials and Methods).

When cells were stimulated with LPS plus IL-4, the mutation frequency of input DNA at the Sγ1 locus was 3.1 × 10^-4 (Table 1). Compared to the input DNA, nearly twice as many of the DNA fragments (% of mutated sequences) immunoprecipitated with anti-acetyl H3 and H4 had mutations, and the mutation frequency of the Sγ1 DNA associated with acetylated H3 and H4 was two to three times higher [10.4 × 10^-4 (P < 0.01) and 6.8 × 10^-4 (P < 0.14), respectively] than the input DNA (Table 1). This finding is consistent with the fact that LPS plus IL-4 stimulates cells to switch to IgG1 and suggests that Sγ1 segments associated with hyperacetylated H3, and perhaps H4, are preferentially targeted for mutation and eventually for switching. When LPS was used, the mutation frequency of input DNA at the Sγ1 locus was 1.4 × 10^-4. There was minimal acetylation of Sγ1 with LPS stimulation, and the mutation frequency of the Sγ1 immunoprecipitated with anti-hyperacetylated H4 did not increase significantly (2.0 × 10^-4, P = 1.0) (Table 1). Because Sγ1 region had a very low level of H3 acetylation upon LPS treatment, and because we used low processive DNA polymerase PfuTurbo (see Materials and Methods) to amplify this region from the ChIPed DNA, we could not obtain enough PCR products to sequence.

Table 1. Switch mutation analyses using input and immunoprecipitated hyperacetylated Sγ1 and hyperacetylated Sγ3 DNA molecules.

Stimulus	DNA kinds	No. of sequences	No. of nucleotides sequenced	No. of mutations seen	Mutated sequences, %	Mutation frequency* (×10-4)	P value†
Hyperacetylated Sγ1
LPS	Input	32	13,728	2	6.2	1.4	N/A
LPS	H3‡	4§	1,716	0	0	0	N/A
LPS	H4‡	47	20,163	4	8.5	2.0	1.0
LPS + IL-4	Input	37	15,890	5	13.5	3.1	N/A
LPS + IL-4	H3‡	50	21,230	22	24.0	10.4	0.01
LPS + IL-4	H4‡	38	16,217	11	18.4	6.8	0.14
Hyperacetylated Sγ3
LPS	Input	29	9,314	4	10.3	4.3	N/A
LPS	H3‡	38	12,020	25	28.9	20.8	0.001
LPS	H4‡	36	10,779	14	25.0	13.0	0.04
LPS + IL-4	Input	40	12,678	2	5.0	1.6	N/A
LPS + IL-4	H3‡	43	12,347	27	39.5	21.9	<0.0001
LPS + IL-4	H4‡	41	12,155	14	22.0	11.5	0.002

N/A, nonapplicable.

^*The fidelity of Pfu Turbo was determined by amplifying the JH2-JH4 region with purified tail DNA from wild-type C57BL/6 mice. No mutation was seen in the 12,595 nucleotides sequenced

^†The P value was calculated (χ² test) by comparing the frequency of switch region mutations of hyperacetylated H3 or H4 DNA molecules with that of the corresponding input DNA

^‡H3 or H4 stands for the DNA segments that were derived from ChIP experiments with antibodies against hyperacetylated histone H3 or H4

^§Could not obtain enough PCR products to sequence (see text)

When we examined the Sγ3 locus (Table 1), stimulation with LPS resulted in a mutation frequency of the input DNA of 4.3 × 10^-4, which was higher than that with LPS plus IL-4 (1.6 × 10^-4) and consistent with the fact that LPS stimulation induces IgG3, but not IgG1, switching. Compared to the input, a higher percentage of the H3 and H4 hyperacetylated Sγ3 segments from LPS stimulated cells contained mutations, and they had a much higher frequency of mutations [20.8 × 10^-4 (P < 0.001) and 13.0 × 10^-4 (P < 0.04), respectively] (Table 1). Although this finding is consistent with the fact that the LPS-treated cells switch to IgG3, the Sγ3 region associated with hyperacetylated H3 and H4 from the LPS-plus-IL-4-treated cells also had a much higher mutation frequency than the input DNA [21.9 × 10^-4 (P < 0.001) and 11.5 × 10^-4 (P < 0.04), respectively] (Table 1), even though LPS plus IL-4 does not promote switching to IgG3. This finding is, however, consistent with acetylation results that showed hyperacetylation of the Sγ3 region in the LPS plus IL-4-treated cells. Although there was less RNAP II associated with Sγ3 at this time, the mutations represent an accumulation rather than a snapshot of what was occurring at 96 h. The presence of many mutations in hyperacetylated Sγ3 from the LPS plus IL-4-treated cells reveals a separation of the generation of mutations in S regions and the ultimate efficiency of CSR.

Because the Sμ in the unstimulated cells was already hyperacetylated, we examined the mutation frequency in the immunoprecipitated Sμ segments from those cells. Although a higher percentage of the acetylated DNA fragments had mutations and there was an increase in the frequency of mutation compared to the input DNA, the differences were not significant (Table 2). After stimulation with either LPS or LPS plus IL-4 and the induction of AID, the input Sμ DNA had a higher frequency of mutation than that from the unstimulated cells. However, compared to the input DNA segments, Sμ associated with hyperacetylated H3 and H4 did not have a higher frequency of mutations in cells stimulated with either LPS or LPS plus IL-4 (Table 2). Overall, both the input and immunoprecipitated Sμ region had frequencies of mutation that are similar to those reported in the literature (33, 36, 40, 41), suggesting that the Sμ region is constitutively acetylated and thus accessible to AID once AID is induced (Fig. 1).

Table 2. Switch mutation analyses using input and immunoprecipitated hyperacetylated Sμ DNA molecules.

Stimulus	DNA kinds	No. of sequences	No. of nucleotides sequenced	No. of mutations seen	Mutated sequences, %	Mutation frequency* (×10-4)	P value†
None	Input	86	28,724	1	1.2	0.3‡	N/A
None	H3§	86	28,724	4	4.6	1.4	0.37
None	H4§	79	26,386	4	5.1	1.5	0.32
LPS	Input	114	37,856	20	12.3	5.3‡	N/A
LPS	H3§	135	44,552	12	8.1	2.7	0.09
LPS	H4§	61	19,684	12	9.8	6.1	0.84
LPS + IL-4	Input	32	10,688	7	12.5	6.5‡	N/A
LPS + IL-4	H3§	78	24,872	20	22.8	8.0	0.64
LPS + IL-4	H4§	42	13,745	8	16.7	5.8	0.82

N/A, nonapplicable.

^*The fidelity of Pfu Turbo was determined by amplifying the JH2-JH4 region with purified tail DNA from wild-type C57BL/6 mice. No mutation was seen in the 12,595 nucleotides sequenced

^†The P value was calculated (χ² test) by comparing the frequency of switch region mutations of hyperacetylated H3 or H4 DNA molecules with that of the corresponding input DNA

^‡The input mutation frequency of no stimulation is significantly lower than that of LPS stimulation (P = 0.0002, Fisher's exact test) and that of LPS + IL-4 stimulation (P = 0.0013, Fisher's exact test)

^§H3 or H4 stands for the DNA segments that were derived from ChIP experiments with antibodies against hyperacetylated histone H3 or H4

Discussion

CSR from IgM to IgG, IgE, and IgA results from recombination between the donor Sμ region and one of the recipient γ, ε, or α S regions and is induced when surface receptors are engaged by different combinations of mitogens and cytokines (42, 43). The resulting signals lead to transcription of GLT (44). The transcription of the S region generates ssDNA that can serve as a substrate for AID. Because AID is required for CSR, it has been suggested that the resolution of the G·U mismatches created by AID leads to closely placed nicks on both strands that are converted into staggered DSB and/or to the recruitment of enzyme complexes like those responsible for mismatch repair and nonhomologous end joining to create and repair the DSB that are required for recombination (1, 3, 19, 45-47). A critical question is how AID and these various enzyme complexes are selectively recruited to certain S regions by particular combinations of mitogens and cytokines, whereas other S regions are spared from recombination.

To gain a better understanding of the regulation of the targeting of CSR to individual S regions in normal B cells, we have stimulated splenic B cells and used ChIP to compare the acetylation of H3 and H4 histones, the density of RNAP II at various times after stimulation, and the frequency of mutations in the Sμ,Sγ1, and Sγ3 regions that are associated with the hyperacetylated histones. We found that both H3 and H4 associated with the donor Sμ region were acetylated in the unstimulated splenic B cells, consistent with the fact that μ heavy chain mRNA as well as μ GLT are produced in resting B cells (48, 49). The relative amount of acetylated histones associated with the donor Sμ region was similar before and after stimulation (Fig. 1). The increase in RNAP II density did not coincide with a change in the degree of acetylation throughout the 96 h of stimulation with either LPS or LPS plus IL-4. Furthermore, there was no significant difference between the frequency of mutation in the input Sμ DNA and that in the DNA associated with acetylated H3 or H4 (Table 2). These results suggest that the Sμ region is constitutively hyperacetylated and potentially accessible to AID.

After stimulation, there was a statistically significant increase in the frequency of mutations in the Sμ compared to the cells before stimulation (P = 0.0002 and 0.0013 for LPS and LPS plus IL-4, respectively) (Table 2). This finding reflects the induction of AID after stimulation and possibly of more transcription bubbles whose ssDNA could be targeted by AID. Increases in the rate of transcription of the whole μ gene might have occurred upon stimulation (49, 50), because we saw an increase in the density of RNAP II associated with Sμ by the ChIP assay. Indeed, the majority of the mutations that we observed in Sμ,Sγ1, and Sγ3 regions in both the input and ChIPed DNA were at G or C bases (data not shown), suggesting that these mutations are caused by AID. We therefore favor the idea that the increase in RNAP II density after stimulation reflects the overall increase of transcription activity of Sμ region, either from the V promoter or the Iμ promoter (40, 49, 50), which may or may not be reflected by the steady state mRNA level (40, 48). Alternatively, RNAP II pausing at the S region and/or recruiting other factors during CSR could lead to the increased RNAP II density in the Sμ region upon stimulation. In fact, RNAP II has been reported to coimmunoprecipitate with AID (32).

We discovered that there was no significant increase in H3 acetylation or in RNAP II associated with Sγ1 in the cells treated with LPS alone, consistent with the fact that LPS stimulates switching to IgG3 but not to IgG1. However, in our studies, LPS alone caused an increase in the acetylation of H4 associated with Sγ1 from 48 to 72 h. This dissociation between H4 acetylation and transcription raises the possibility that the balance between acetylation of H4 and H3 plays a role in regulating the suppression of transcription.

Studies in the literature show that LPS alone triggers a progressive increase in the transcription of Sγ3 GLT during the course of stimulation (51). The studies reported here are completely consistent with this in that there was an increase of RNAP II and H3 associated with Sγ3 between 48 and 96 h after stimulation with LPS. There was also an increase in the acetylation of H4 associated with Sγ3 from 18 to 72 h. These changes were associated with a 3- to 5-fold increase in the frequency of mutations in the acetylated H3 and H4 Sγ3 DNA (Table 1), suggesting that hyperacetylated Sγ3 segments are favorably targeted by AID. Furthermore, this enrichment for mutations in the DNA associated with acetylated histones suggests that only a subset of the cells undergo changes in chromatin modifications that make the Sγ3 region accessible to AID.

LPS plus IL-4 stimulates switching to IgG1 and the expression of Sγ1 GLT peaks at 48 h (40). Consistent with this, there was an increase in RNAP II and in the acetylation of H3 and H4 associated with Sγ1 in the LPS-plus-IL-4-treated cells at 48 and 72 h. Although there was less RNAP II associated with Sγ1 at 96 h, the increased acetylation of H3 and H4 persisted, consistent with the presence of a hyperacetylated domain that was accessible even after transcription decreased. In addition, there was an ≈3-fold enrichment of mutations in the Sγ1 DNA immunoprecipitated with antibody to acetylated H3 (Table 1), suggesting that the changes observed in H3 acetylation occurred in ≈30% of the stimulated cells.

Despite the fact that LPS plus IL-4 does not stimulate switching to IgG3, it does stimulate the production of Sγ3 GLT (32, 51, 52). Sγ3 GLT peaks 48 h after LPS plus IL-4 treatment, then tapers off (51). This is consistent with our finding that association of RNAP II and of acetylated H3 and H4 with Sγ3 was increased after stimulation with LPS plus IL-4. In addition, the H3 and H4 hyperacetylated Sγ3 DNA segments from cells stimulated with LPS plus IL-4 also showed a much higher mutation frequency, compared with the input DNA. This finding reveals that AID was targeted to Sγ3 region even though there was no switching to IgG3 in these cells, suggesting that switching-specific factors and/or additional chromatin modifications are required for the completion of CSR. In fact, it was recently demonstrated that NF-κB subunit p50 is an IgG3-specific switching factor (53). Our results also suggest that IL-4 blocks switching to IgG3 after mutation in the hyperacetylated segments has occurred, revealing the complexity of the regulation of CSR.

In conclusion, our results showed that histone acetylation and RNAP II association are regulated differently between donor and recipient S regions, as well as between different recipient S regions upon different stimuli. We observed that hyperacetylated recipient S regions were preferentially targeted by AID for mutations, but this was not sufficient for switching. These findings suggest that differential regulation of histone acetylation and generation of mutations in S regions is necessary but not sufficient for the complex CSR process.