Reciprocal patterns of methylation of H3K36 and H3K27 on proximal vs. distal IgV_H genes are modulated by IL-7 and Pax5

Abstract

The usage of >100 functional murine Ig heavy chain V_H genes, when rearranged to D_HJ_H genes, generates a diverse antibody repertoire. The V_H locus encompasses 2.5 Mb, and rearrangement of V_H genes in the D_H-distal half of the locus are controlled very differently from the V_H genes in the proximal end of the locus. The rearrangement of distal but not proximal V_H genes is impaired in mice deficient in the cytokine IL-7 or its receptor, in the transcription factor Pax5, or in Ezh2, a histone methyltransferase for Lys-27 of histone H3 (H3K27). The relative role of IL-7, Pax5, and Ezh2 in regulating distal vs. proximal V_H rearrangement is not clear. Here, we show by ChIP and ChIP-on-chip that the active histone modification H3K36me2 is most highly associated with distal V_H segments and the repressive histone modification H3K27me3 is exclusively present on proximal V_H segments. We observed an absence of H3K27me3 in fetal pro-B cells, which predominantly rearrange proximal V_H genes. Absence of IL-7 signaling reduces H3K36me2, and overexpression of IL-7 increases H3K36me2. In contrast, the major effect of the absence of Pax5 is the reduction in H3K27me3. Our data indicate that the cytokine IL-7 and the transcription factor Pax5 influence the rearrangement of the two regions of the V_H locus by differentially modulating two reciprocal histone modifications during B lymphocyte development.

During B lymphocyte development, the diversity of Ig genes is created by the assembly of a unique set of variable (V), diversity (D), and joining (J) gene segments in each precursor B cell. V(D)J recombination is highly regulated in both a lineage-specific and developmental stage-specific manner (1). In pro-B cells, successful D_H–J_H rearrangement is followed by rearrangement of one V_H gene to the joined D_HJ_H segment. This latter step is particularly complex, because the murine V_H locus contains >100 functional genes and spans 2.5 Mb (1).

The rearrangement of the V_HJ558 and V_H3609 family genes, which comprise the D_H-distal half of V_H locus, is controlled very differently than the rearrangement of V_H genes in the D_H-proximal end of the locus, particularly the most proximal V_H7183 family. This differential control is demonstrated by the greatly impaired rearrangement of distal, but not proximal, V_H genes in mice deficient in the cytokine IL-7 or its receptor, in the transcription factors Pax5 or YY1, or in mice deficient in Ezh2, a histone methyltransferase that catalyzes the methylation of Lys-27 of histone H3 (H3K27) (2–6). Rearrangement of proximal V_H genes is also favored early in ontogeny (7, 8). Pro-B cells from IL-7Rα^−/− mice are deficient in distal V_HJ558 germ-line transcription, suggesting lack of accessibility of the distal V_H genes (3). Consistent with the role of IL-7 signaling affecting distal V_HJ558 rearrangement, culture of adult pro-B cells from RAG^−/− mice with IL-7 can induce histone acetylation on V_HJ558 genes, but not proximal V_H genes (9). Pro-B cells from Pax5^−/−, YY1^−/−, and Ezh2^−/− mice also have a defect in distal V_H rearrangement, but germ-line transcription and histone acetylation are normal (4–6), suggesting a different mode of repression of distal V_H gene rearrangement. A critical step in facilitating distal V_H gene rearrangement, compaction of the V_H locus, is impaired in Pax5^−/− and YY1^−/− pro-B cells (6, 10). In addition, Pax5 has been shown to be necessary for the loss of the repressive H3K9 methylation on V_H genes in pro-B cells (11). However, the mechanism by which Ezh2 represses distal V_H rearrangement is unknown.

Changes in the structure of the chromatin surrounding the V_H gene segments are likely to regulate the accessibility of the RSS to undergo RAG recombinase binding and subsequent recombination (12). These chromatin changes can involve histone posttranslational modifications (PTMs) and ATP-dependent nucleosomal remodeling. The phenotype of Ezh2-deficient mice suggested that methylation of H3K27 may play a crucial role in controlling distal V_H gene rearrangement (5). We therefore compared the methylation status of distal and proximal V_H genes to determine whether they differed. Here, we show by ChIP and ChIP-on-chip that the repressive H3K27me PTM is observed exclusively on the proximal V_H genes. In addition, we analyzed other histone lysine methylations to determine whether the chromatin structure of the distal V_H genes displays any other different characteristics from that of the proximal V_H genes. We describe that H3K36me, a PTM associated with active genes (13), is present on V_H genes. We demonstrate that H3K36me2 is greatly enriched on distal V_H genes and is also present to a lesser extent in B-lineage cells on proximal V_H genes, thus displaying the reciprocal pattern of expression to H3K27me3. Absence of IL-7 signaling reduces H3K36me2, and overexpression of IL-7 increases H3K36me2. In contrast, the major effect of the absence of Pax5 is the reduction in H3K27me3. We propose a model in which IL-7 signaling promotes two positive PTMs, H3K36 methylation and acetylation of H3. Subsequently, Pax5 is necessary for the methylation of H3K27, possibly by recruiting Ezh2 to V_H genes via Pax5 binding sites within the V_H locus. Thus, together the cytokine IL-7 and the transcription factor Pax5 are essential to set the epigenetic profile that permits full utilization of the distal and proximal V_H genes to create the diverse antibody repertoire.

Results

Histone PTMs Associated with Distal and Proximal V_H Genes.

PTMs of the amino-terminal tails of histones play important roles in biological processes through their effects on chromatin structure. Histone methylation of Lys-4, -36, and 79 on histone H3 are associated with gene activation, whereas methylation of Lys-9 or -27 on H3 or Lys-20 on H4 are generally associated with gene repression (13). We have screened these six sites of histone lysine methylation on D_H-proximal vs. D_H-distal V_H segments in various stages of B and T cell development and from multilineage progenitor (MLP) cells and ES cells. ChIP data revealed two PTMs that showed very distinct patterns of expression on the proximal V_H7183 genes vs. the distal V_HJ558/3609 genes. The repressive H3K27me3 is exclusively present on the proximal V_H7183 family genes in all B and T lineage progenitors and on MLP (Fig. 1 A and B). The H3K27me2 modification has a similar pattern to H3K27me3 [supporting information (SI) Fig. S1]. Conversely, we observed that H3K36me2 is associated with distal V_H genes in all B and T lineage cells and in MLP. It is also enriched to a lesser extent on proximal V_H7183 genes exclusively in B lineage cells. H3K36me3 has a similar pattern to H3K36me2 (data not shown). However, in ES cells, H3K36me2 is absent on all V_H genes, and H3K27me3 is low on all V_H genes (Fig. 1B). Thus, H3K36me and H3K27me display reciprocal patterns of methylation on the distal vs. proximal V_H genes and are thus candidates for controlling distal vs. proximal V_H gene rearrangement.

Fig. 1.

The reciprocal pattern of histone H3K27me3 and H3K36me2 on Ig V_H genes. (A) Genomic organization of the murine IgH locus, showing position of the different V_H families and the location of the genes analyzed in this work. (B) ChIP assays were performed by using antibodies reactive with H3K27me3 or H3K36me2 on pro-B cells from 4- to 6-week-old RAG1^−/− mice; pro-B, pre-B, immature B, CD4⁻CD8⁻ T (DN T), CD4⁺CD8⁺ T (DP T) thymocytes from 4- to 6-week-old BALB/c mice; MLP cells from 4- to 6-week-old μMT mice, and an ES cell line. The names of the genes assayed are shown at the bottom of the graphs. Distal V_H genes are black bars, the middle V_HS107.1 gene is a gray bar, and proximal V_H7183 genes are white bars. V_HJ558 and V_H7183 are family-specific primers, and other primer sets are specific for individual genes. Analysis was performed by real-time PCR. Data are presented as relative to the positive control of the Neuregulin gene (Neuregulin = 1) for H3K27me3 and CAD gene (CAD = 1) for H3K36me2. Results represent the mean ± SD. (C) ChIP-on-chip analysis was performed on pro-B cells from RAG^−/− mice with anti-H3K27me3 and anti-H3K36me2 antibodies. The bars represent the distribution of the statistically significant regions of these modifications on Ig V_H locus as determined by MAT software (40). The locations of the D_HJ_H, the V_H7183 and the interspersed V_HJ558 and V_H3609 families are marked on the top of the graph.

ChIP-on-Chip Analysis of the IgH Locus.

To obtain a global view of H3K27me3 and H3K36me2 patterns on the entire V_H locus, we performed ChIP-on-chip analysis with RAG^−/− pro-B cells (Fig. 1C). This procedure clearly revealed that H3K27me3 is exclusively present on the proximal V_H7183 genes, whereas H3K36me2 forms a gradient that is highest at the distal end of the V_H locus. Unlike other PTMs, these modifications are more highly expressed on V_H genes than on J_H genes, suggesting their importance in differential V_H gene accessibility. The inverse pattern of expression of H3K27me3 and H3K36me2 demonstrates the distinct epigenetic character of distal vs. proximal V_H genes that may directly affect the accessibility of the V_H genes.

Fetal Pro-B Cells Lack H3K27 Methylation.

Preferential rearrangement of proximal V_H genes in Ezh2^−/− mice suggests that H3K27 methylation on proximal V_H genes is essential for the rearrangement of distal V_H genes. We hypothesize that H3K27me on proximal V_H genes will decrease the rearrangement potential of these proximal V_H genes, and that this inhibition will provide more opportunity for rearranging distal V_H genes. Determining the pattern of H3K27me3 on fetal cells is an ideal test of this hypothesis, because the V_H7183 family is greatly favored in rearrangements early in ontogeny (7, 8). In contrast, rearrangement in the adult BM is skewed more to the distal V_H genes. If this rearrangement pattern is regulated by H3K27 methylation, we predict that the level of H3K27 methylation would be greatly reduced in fetal B lineage cells.

To test this hypothesis, we analyzed fetal liver and adult bone marrow pro-B cells from μMT mice. These mice have a complete block of B cell development at the pro-B cell stage (14). ChIP analysis of CD19⁺ cells from μMT BM and fetal liver and CD19⁺ fetal pro-B cells from BALB/c mice revealed a dramatic decrease of the H3K27me3 modification on fetal pro-B cells and also a slight decrease of the permissive H3K36me2 modification on the distal V_H region (Fig. 2A). Furthermore, we performed ChIP with Ezh2 antibody on fetal liver and adult BM pro-B cells, and the pattern of Ezh2 binding was very similar to that of H3K27 methylation (Fig. 2B). These data suggest that the presence of methylation on H3K27 on proximal V_H genes catalyzed by Ezh2 is necessary for the rearrangement of distal V_H genes. In addition, it demonstrates a profound difference between the regulation of V(D)J rearrangement in fetal vs. adult pro-B cells.

Fig. 2.

The patterns of histones H3K27me3 and H3K36me2 and Ezh2 in fetal liver are different from adult bone marrow. (A) ChIP assays were performed by using antibodies reactive with H3K27me3 or H3K36me2 on pro-/pre-B cells (CD19⁺IgM⁻) from BALB/c fetal liver, pro-B cells (B220⁺CD19⁺) from μMT fetal liver, and pro-B cells (B220⁺CD19⁺) from 3- to 4-week-old μMT bone marrow. (B) ChIP assays were performed by using Ezh2 antibody on pro-B cells from μMT fetal liver and 3- to 4-week-old μMT bone marrow. Data are presented as relative to the positive control of the Neuregulin gene (Neuregulin = 1).

IL-7 Signaling Modulates the Extent of Methylation of H3K36.

IL-7 signaling plays many roles in B cell development. B cell progenitors in adult IL-7^−/− and IL-7Rα^−/− mice >7 weeks of age are blocked before the CD19⁺ pro-B cell stage, but the defect is much less severe in fetal mice, with B cell development progressively waning during the first 2 months of life (15–17). In IL-7Rα^−/− mice, rearrangements to proximal V_H genes are present, but distal V_HJ558 rearrangements are diminished (3). To determine whether IL-7 signaling affects the patterns of H3K27me3 or H3K36me2, we performed ChIP analysis on CD19⁺ pro-B cells from young adult BM (≈3–4 weeks old) or fetal liver from IL-7^−/− mice. Pro-B cells from IL-7^−/− fetal liver cells showed a similar pattern to that of the WT fetal pro-B cells (Fig. S2A), but the patterns of modification in the IL-7^−/− adult BM pro-B cells were quite different from age-matched μMT pro-B cells (Fig. 3A). We observed that the extent of H3K36me2 decreased on all V_H genes in adult pro-B cells from IL-7^−/− mice (Fig. 3A), and conversely, IL-7 overexpression in adult pro-B cells from IL-7 transgenic (IL-7 Tg) mice (18) resulted in an increase in H3K36me2 throughout the V_H locus. The effect of IL-7 on H3K27me3 was less clear. There was no obvious difference of H3K27me3 pattern in pro-B cells from IL-7 Tg mice (Fig. 3B), suggesting that IL-7 did not influence H3K27me3 levels. However, the IL-7^−/− BM pro-B cells had higher levels of H3K27me3 on the proximal V_H genes and also showed methylation of H3K27 even on distal V_H genes. This discrepancy between the transgenic and knockout mice might be caused by a more indirect effect of the IL-7 deficiency on H3K27me3 levels, possibly because of the developmental blockade in these mice, or the influence of other chromatin modifications. Nonetheless, the reciprocal phenotype of the IL-7 Tg and IL-7^−/− pro-B cells data clearly demonstrates that one role of IL-7 signaling is to increase the level of H3K36 methylation throughout the V_H locus.

Fig. 3.

IL-7 signaling modulates the extent of H3K27me3 and H3K36me2 on V_H genes in B lineage cells. (A) ChIP assays were performed on μMT or IL-7^−/− pro-B cells (B220⁺CD19⁺) from 3- to 4-week-old bone marrow. (B) ChIP assays were performed on pro-B cells (B220⁺CD43⁺CD19⁺IgM⁻) from 4- to 7-week-old bone marrow of IL-7 transgenic (Tg) mice or their littermates (B6).

Lymphocyte progenitors progress through a series of developmental steps after the hematopoietic stem cell stage (19). The IL-7 receptor is absent on MLPs but begins expression in common lymphoid progenitors. If the initial origin of H3K27me3 and H3K36me2 is determined by IL-7 signaling, we should detect deficiency of these two modifications in MLP and IL-7^−/− T cells. However, we found similar patterns of H3K27me3 and H3K36me2 in MLP as in B and T lineage cells (Fig. 1B), and both modifications are the same in double positive (DP) T cells from WT, IL-7^−/−, or IL-7 Tg mice (Fig. S2B). This result suggests that the origins of these two modifications are independent of IL-7 signaling and that they occurred before the MLP stage. Thus, the maintenance of these modifications on V_H genes appears to depend on IL-7 signaling in B cells but is under different regulation in T cells.

Pax5^−/− Pro-B Cells Display Reduced H3K27 Methylation.

The transcription factor Pax5 is essential for B cell development (20). In Pax5^−/− mice, B cell development is completely blocked at the pro-B stage (21, 22). Proximal V_H gene rearrangements occur, but Pax5^−/− pro-B cells show greatly impaired rearrangement of the distal V_H genes (4, 10). ChIP analysis on B220⁺c-Kit⁺ pro-B cells from Pax5^−/− mice compared to littermate controls showed greatly reduced H3K27me3 on V_H genes with only a slight decrease of H3K36me2 (Fig. 4).

Fig. 4.

Pax5 modulates H3K27me3 on V_H genes. ChIP assays were performed on B220⁺c-Kit⁺ pro-B cells from Pax5^−/− mice and their littermates (WT).

Interestingly, DP T cells from Pax5^−/− mice also showed a significant reduction in the level of H3K27me3 on V_H genes compared with littermate controls (Fig. S3), although Pax5 is strictly expressed in B lineage cells and is essential for commitment of progenitors to the B lineage cells (20). Because Pax5^−/− pro-B cells have been shown to have the capability to differentiate into T cells in vivo and in vitro (23, 24), one explanation of this result is that Pax5^−/− pro-B cells are diverted to differentiate into the T lineage in vivo, thus resulting in Pax5^−/− thymocytes that express B lineage patterns for these two histone modifications.

The low level of H3K27me3 in fetal liver or high level in IL-7^−/− pro-B cells is not caused by alterations in Ezh2 expression (Fig. S4). Pax5 expression is higher in IL-7^−/− adult pro-B cells than in μMT adult pro-B cells and is reduced in all fetal pro-B cells, which corresponds to the high level of H3K27me3 in IL-7^−/− BM and the low level in fetal pro-B cells (Fig. S4). Two methyltransferases, SYMD2 and NSD1, have been described that methylate H3K36 (25). We found that IL-7 promotes SYMD2 expression, and the expression pattern of SYMD2 mirrors the pattern of H3K36me2 in each cell population, suggesting that SYMD2 is the methyltransferase that methylates H3K36 in lymphoid progenitors. In contrast, NSD1 shows the inverse expression pattern to SYMD2 (Fig. S4).

Discussion

Although it is clear that the distal and proximal V_H genes are under distinct regulation, the basis for this differential regulation is not clear. No large differences in the epigenetic character of the chromatin associated with the distal vs. proximal V_H genes have been described, although the differences in rearrangement patterns in various knockout mice suggest that the chromatin structure is likely to be quite different. Here, we describe two histone PTMs, H3K27me3 and H3K36me2, which are unique in that they show equal or higher enrichment on a subset of V_H genes than on D or J genes. Most importantly, they show reciprocal patterns of expression on proximal vs. distal V_H genes.

The H3K27me3 modification is of particular interest, because conditional deletion of Ezh2 results in greatly impaired rearrangement of distal V_H genes, whereas rearrangement of proximal genes is normal. We demonstrate here that the H3K27 is methylated exclusively on the proximal V_H genes, and the ChIP-on chip analysis of the entire IgH locus clearly shows the reciprocal relationship of these H3K27me3 and H3K36me2 on V_H genes. Importantly, Pax5^−/− pro-B cells have much lower levels of H3K27me3, suggesting a relationship between Pax5 and H3K27 methylation. The fact that H3K27me3 is observed only on proximal V_H genes and that the Ezh2 conditional deletion results in the inability to rearrange distal V_H genes strongly argues that the repression of proximal V_H genes by H3K27 methylation is essential for proper rearrangement of distal V_H genes.

Culture of pro-B cells from RAG^−/− mice in IL-7 has been shown to induce histone acetylation of distal V_H genes, and germ-line transcripts from the distal V_HJ558 genes are not present in the absence of IL-7 or its transcriptional mediator STAT5 (9, 26). Here, we show that IL-7 also induces H3K36me2, and this PTM is present at a much higher level on V_H genes than acetylation. However, unlike the preferential induction of histone acetylation on distal V_H genes, IL-7 produces a generalized increase of H3K36me2 on all V_H genes. Thus, one important role of IL-7 is to induce accessibility of the V_H genes through chromatin modifications. Pax5, YY1, and Ezh2 deficiencies are manifest at a subsequent step in V(D)J rearrangement, because the distal V_H genes in these mice are accessible as measured by enrichment with acetylated histones and normal levels of V_HJ558 germ-line transcription (4–6). Ezh2^−/− pro-B cells even display increased germ-line transcription of proximal V_H genes, which would be consistent with the depletion of the repressive H3K27me3 mark. Importantly, the compaction of the V_H locus, bringing the distal V_H genes closer to the DJ_H region, does not occur in either Pax5- or YY1-deficient mice (6, 10), thus placing compaction at a later stage than histone acetylation and germ-line transcription of distal V_H genes.

Together, these results revealed that distal vs. proximal V_H genes have different epigenetic profiles, with H3K27me3 and H3K36me2 displaying an inverse pattern on distal vs. proximal V_H genes. Because Ezh2^−/− pro-B cells rearrange proximal V_H genes at normal frequency but cannot rearrange the distal V_HJ558 genes, and because we have shown that H3K27me3 is exclusively present on proximal V_H7183 genes, we propose that adding the repressive H3K27me on proximal V_H genes is essential to allow distal V_H genes to rearrange. The reciprocal patterns of H3K36me and H3K27me suggest that these modifications may act in concert to promote rearrangement of all V_H gene families. The decrease of H3K27me3 on proximal V_H genes may explain the low frequency of rearrangement of distal V_H genes in fetal life. The impaired rearrangement of distal V_H genes in IL-7Rα^−/− cells is likely caused by the absence of germ-line transcripts and the accompanying paucity of histone acetylation from the distal V_H genes. In addition, sufficient levels of H3K36me2, which is directly regulated by IL-7 signaling, may also be necessary for the rearrangement of distal V_H genes. Our data suggest that IL-7 signaling and Pax5 affect V_H gene rearrangement by modulating these histone modifications during early B lymphocyte development.

We present a model to show how histone modifications may regulate the recombination of IgV_H genes (Fig. S5). At the MLP stage, or even earlier, while the IgH locus is positioned at the nuclear periphery (27), we propose that the distal V_H locus is marked with H3K36me by a complex likely containing SYMD2, and the proximal V_H region is marked with H3K27me by Ezh2 in polycomb repressive complex 2 (28, 29). In pro-B cells, the IgV_H locus moves from the nuclear periphery to the center of the nucleus where it subsequently contracts and forms loops to bring distal V_H genes close to the DJ_H region (10, 27, 30). Both Pax5 and YY1 have been shown to be essential for the contraction process and the rearrangement of distal V_H genes, and Pax5 binding sites are located throughout the V_H locus, with the highest affinity sites being the ones in the proximal V_H genes (6, 31, 32). In the proximal V_H–D_HJ_H loop, we propose that Ezh2 may be recruited by Pax5 to maintain the appropriate level of H3K27me. YY1 may also be involved in this complex because it has been shown to associate with Ezh2 (33). Because it has been shown that prior methylation of H3K36 and acetylation of H3 precludes binding of Ezh2 complexes (34), we propose that the high level of H3K36me in the distal V_H loops prevents Ezh2 binding to any Pax5(YY1) complexes so H3K27 cannot be methylated on distal V_H genes. In Pax5^−/− pro-B cells, we hypothesize that Ezh2 cannot be recruited to V_H locus. Together, cooperation of H3K27me, H3K36me, and DNA looping and locus contraction contribute to the usage of Ig V_H genes throughout the entire V_H locus to create a diverse antibody repertoire.

Materials and Methods

Mice and Cell Line.

Mice were bred and maintained in animal facilities at The Scripps Research Institute (TSRI), and the studies were approved by the TSRI Institutional Animal Care and Use Committee. μMT (14), IL-7^−/− (2), and IL-7 transgenic (18) mice are on the C57BL/6 background. RAG1^−/− mice (35) are on the BALB/c background. Pax5^−/− mice (21) were made by M. Busslinger (Research Institute of Molecular Pathology, Vienna, Austria) and given to us from J. Hagman's colony (National Jewish Medical and Research Center, Denver, CO). The R1 murine embryonic stem cells were derived from 129 strain and cultured as described (36).

Purification of MLP Cells.

Bone marrow cells were isolated from 4- to 6-week-old μMT mice by crushing the bones with a mortar and pestle. The cell suspension was filtered through nylon mesh and cotton plug to remove debris, and the cells were washed and resuspended in magnetic cell sorting (MACS) buffer. Lineage negative cells (Lin⁻) were obtained by using the Lineage Cell Depletion kit (Miltenyi Biotec) according to Miltenyi's protocol. Using the kit, bone marrow cells were incubated with a mixture of antibodies against CD5, B220, CD11b, Gr-1, 7–4, and Ter-119 and thus were depleted of T cells, B cells, monocytes, macrophages, granulocytes, and erythrocytes and their committed precursors after MACS column purification. Collected Lin⁻ cells were stained with the following antibodies: AA4-allophycocyanin (APC), CD43-FITC, and IL-7Rα-phycoerythrin (PE). MLP cells (AA4^high CD43⁺IL-7Rα⁻) were sorted on a BD FACSAria cell sorter.

Purification of Pro-B, Pre-B, and Immature B Cells from Bone Marrow.

Bone marrow cells were isolated from 4- to 7-week-old WT or IL-7 Tg mice as described above. B220⁺ cells were enriched on MACS anti-PE beads (Miltenyi Biotec) after staining with B220-PE antibody. After four-color stain (anti B220-PE, CD43-FITC, CD19-Pacific blue, and IgM-APC) of the B220⁺ cells, we sorted for: pro-B, fractions B and C (B220⁺CD43⁺CD19⁺IgM⁻); pre-B, fraction D (B220⁺CD43⁻CD19⁺IgM⁻); and immature B, fraction E (B220^modCD43⁻CD19⁺IgM⁺), as defined by Hardy and colleagues (37).

Purification of CD19⁺ Cells.

Bone marrow cells were isolated from 4- to 6-week-old RAG1^−/− and BALB/c mice. Because B cell production in the bone marrow decreases with age in IL-7^−/− mice, we used 3- to 4-week-old mice from IL-7^−/− and age-matched B6 or μMT mice. Fetal liver cells were isolated from 16- to 18-day-old fetal pups from WT (BALB/c), μMT, or IL-7^−/− mice. CD19⁺ cells were enriched on MACS anti-mouse CD19 antibody-coated microbeads (Miltenyi Biotec). Then B220⁺CD19⁺ pro-B cells from μMT or IL-7^−/− mice and CD19⁺ IgM⁻ cells from WT fetal liver were sorted on a BD FACSAria cell sorter.

Purification of Double Negative (DN) and DP Thymocytes.

DN (CD4⁻CD8⁻) thymocytes were purified from 6-week-old WT mice by using MACS anti-CD4 (L3T4) microBeads (Miltenyi Biotec), collecting the negative fraction, followed by MACS CD8a (Ly-2) MicroBeads (Miltenyi Biotec), collecting the negative fraction. For DP T cells, we sorted CD4⁺CD8⁺ cells from 4- to 6-week-old WT mice, 4-to 7-week-old IL-7 Tg mice, 3- to 4-week-old IL-7^−/− mice, and 15- to 16-day-old Pax5^−/− mice and their littermates by staining thymocytes with anti-CD4-APC and CD8a-PE.

Purification of Early Pro-B Cells from Pax5^−/− and Littermates.

Because Pax5^−/− mice die at ≈3–3.5 weeks of age, bone marrow cells were isolated from 15-to 16-day-old Pax5^−/− mice or littermate controls. B220⁺ cells were enriched on MACS anti-PE beads (Miltenyi Biotec) after staining with B220-PE antibody. The B220⁺c-Kit⁺ cells were then purified on a BD FACSAria cell sorter.

ChIP and Real-Time PCR.

Formaldehyde cross-linking and ChIP were performed as described (38) with the exception of using a QIAquick PCR Purification Kit (Qiagen) to purify ChIP DNA after reversal of the cross-links. We used antibodies against H3K27me2 (Upstate; 07-452), H3K27me3 (Upstate; 07-449), H3K36me2 (Upstate; 07-369), and Ezh2 (Lake Placid Biologicals; AR-0163). Real-time PCR was performed by using the Quantitec SYBR PCR Kit (Qiagen) and the ABI Prism 7900 Sequence Detection System (Applied Biosystems). The sequences of the primers used for PCR analysis have been described: carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase (CAD) (39), V_HJ558.47 (9), V_H3609.4 (V_H3609a) (40), and V_HS107.1 (V1), family-specific V_HJ558, family-specific V_H7183, V_H 81X-specific primers, and neuregulin (38). Data are presented relative to Neuregulin (H3K27me and Ezh2) or CAD (H3K36me).

ChIP-on-Chip.

Amplified immunoprecipitated DNA was generated from individual ChIP samples by PCR according to Affymetrix's protocol. Sample fragment, labeling, hybridization, and data extraction were performed by the DNA Array Core Facility at TSRI. GeneChip Mouse Tiling 2.0 R Array E (Affymetrix) was used for hybridizations. Analysis of the Affymetrix hybridizations was performed by using the MAT (model-based analysis of tiling array) algorithm (41). The analysis was based on the February 2006 assembly of the mouse genome.

RT-PCR.

RNA was harvested from sorted cells by using the AllPrep DNA/RNA Mini Kit (Qiagen). cDNA was synthesized by using the QuantiTect Reverse Transcription kit (Qiagen), and mRNA levels were assayed by real-time PCR by using the Quantitec SYBR PCR Kit (Qiagen). The primer sequences were: GAPDH 5′, ATGGTGAAGGTCGGTGTGAAC; GAPDH 3′, GCCTTGACTGTGCCGTTGAAT; EZH2 5′, GTAGACACTCCTCCAAGAAAGAA; EZH2 3′, GATGGTCACAGGGTTGATAGTT; SYMD2 5′, CTTCGCCCAGGTGAACTGTAAT; Pax5 5′, TGCCCATCAAGGTGTCAGGC; Pax5 3′, GATGCCACTGATGGAGTATG; SYMD2 3′, CTGCCAGGGTCCCTTTGTAGGT; NSD1 5′, CAGACAGCCCAAGTTTCGTAGT; and NSD1 3′, GAGCCATTGGTATCAGAAGAACAG.