Functional Analysis of Histone Methyltransferase G9a in B and T Lymphocytes

G9a conditional knockout mice

Preparation of the G9a^cnd allele as well as mb1-Cre and Lck-Cre mice are published elsewhere (21, 22, 23). To generate mice with G9a-deficient B cells, we bred the G9a^cnd allele into the mb1-Cre background, which already had a G9a^Δ allele, to produce mb1-Cre/G9a^cnd/^Δ mice. The G9a^Δ allele arose from spurious Cre-mediated deletion of G9a^cnd in the germ cells of an mb1-Cre/G9a^wt/cnd mouse. All animal studies have been reviewed and approved by the Vanderbilt Institutional Review Board.

Immunoblotting

Immunoblotting analyses of global histone modifications and G9a were performed as described (7).

PCR and chromatin immunoprecipitation (ChIP) assays

For recombination, DNA extracts were prepared from primary lymphocytes (1000 cells/μl) as outlined (24). Primers and probes to detect rearrangements have been described for VαJα (25), VγJγ (26), VδJδ (27), VHDHJH (24), VκJκ (28), Dβ1Jβ1 (20), and Vβ12 and Vβ14 (29). Vβ16: primer, GGACCCAAAGTCTTACAGATCCC; Dβ2Jβ2, CACGATGTAACATTGTGGGGACTG and TCAGATCCCCACCAACCATGAGTG; probe, CCTTCACCAAAGTAGAGCTGC. VλJλ: primer, ATGAATTCACTGGTCTAATAGGTGGTACCA, CTAGGACAGTCAGTTTGGTTCC, and CTAGGACAGTGACCTTGGTTCC; probes, TGCTGTACCATAGAGCACAGAAAT (all VλJλ), GGTGTTCGGTGGAGG (Vλ1Jλ1), and TATGTTTTCGGCGGT (Vλ2Jλ2). For transcription assays, total RNA was extracted from lymphocytes using Trizol (Invitrogen) and converted to cDNA as described (20). Assays for RT-PCR have been reported for JH (30); Jκ (31); Blimp-1 (32); IgG1 germline transcripts (33); Vλ1, Vλ2, Jλ1, Jλ2, and β-actin (28). Activation-induced deaminase (AID): primer, GTGCCACCTCCTGCTCACTGG and TTCATGTAGCCCTTCCCAGGC; probe, GTGGCTGAGTTTCTGAGATGG. IgJ, primers, ATGAAGACCCACCTGCTTCTCTGG and TTGCTCTGGGTGGCAGTAACAACC; probe, ATGTGTACCCGAGTTACC. ChIP assays on B cell precursors were performed as described previously (34). Primers for Dβ ChIPs have been reported (20). JH4 ChIP: primer, GCTCAGTCACCGTCTCCTCAGG and GCCTCCAAAGTCCCTATCCCATC; probe, GGCTGAGAGAAGTTGGG. Jκ ChIP, primers, CGTTCGGTGCTGGGACCAAGC and GCCACGTCAACTGATAATGAGCCCTCTCC; probe, GACCAAGCTGGAGCTGAAACG. Vλ1 ChIP: primers, CCCCTGCAAACAGATGAGAAATCC and GCCTGCTGCTGACCAATATTG; probe, GAAAAGAATAGACCTGGTT. Vλ2 ChIP: CTGGCTCCTGCAAACAGGTA and GCCTGCTGCTGACCAATATTG; probe, GAAAATAATAGACTTGGTT. Jλ1 ChIP: primers, GATCTTTCAGTGATGTCACCACC and GCACCTCAAGTCTTGGAGAGAAC; probe, GGTGTTCGGTGGAGG. All amplification products were analyzed by Southern blotting.

Flow cytometry

Abs for cell surface markers were purchased from BD Biosciences except anti-Igλ-PE (Upstate Biotechnologies). Flow cytometry was performed on a FACSCalibur (BD Biosciences) and analyzed using CellQuest software.

Cell purification

MACS was used to purify splenic (CD43⁻) or bone marrow (B220⁺) B cells following the manufacturer’s protocol (Miltenyi Biotec). Splenic T cells were isolated using a mouse T cell Negative Isolation Kit (Dynal Biotech). Isolation of pro- and pre-B cell populations was performed as described (30).

Cell proliferation assays

Purified lymphocytes were seeded in 96-well plates (10⁶/ml) and cultured in RPMI supplemented with 10% FCS. Proliferation was induced in B cell cultures with anti-IgM (Jackson Immunoresearch Laboratories), LPS (Sigma), or anti-CD40 (BD Biosciences) and in T cells by precoating wells with anti-CD3 (2C11, 1 μg/ml). After 48 h (B cells) or 60 h (T cells), [³H]TdR was added to each well, cells were incubated for 12–18 h, and TdR incorporation was quantified using liquid scintillation counting.

Immunization

Eight-week-old mice were immunized by i.p. injection of keyhole limpet hemocyanin (KLH; Sigma-Aldrich) at a concentration of 100 μg/50 μl/mouse mixed 1:1 with CFA (Difco). Serum was collected preimmunization or 3 wk postinjection. Total or KLH-specific Ig levels were measured by ELISA (Southern Biotech) as described (35).

Ig class switching

CD43⁻ spleen cells (5 × 10⁵) were cultured for 6 days with 5 μg/ml LPS ± 5 ng/ml mouse IL-4 (Peprotech). IgG1 and IgE were detected by ELISA (Southern Biotech).

Lymphocyte-specific inactivation of G9a

Germline inactivation of the gene encoding G9a leads to early embryonic lethality in mice, precluding studies of lymphocyte development (7). To circumvent this obstacle, we generated embryonic stem (ES) cells with a G9a allele containing loxP sites that flank exons 26 and 27 (G9a^cnd), which encode for its enzymatic SET domain (23). Expression of Cre recombinase in cells containing the G9a^cnd mutation deletes floxed exons and produces a functionally null G9a allele (G9a^Δ; Ref. 23). The G9a^cnd allele was passed into the mouse germline as described (23).

To generate mice with G9a-deficient B cells, we bred G9a^cnd and mb1-Cre mice. The latter strain harbors a Cre expression cassette driven by the endogenous mb1 (Igα) promoter, which is expressed in early progenitor B cells (21). Southern blotting confirmed that G9a^cnd was efficiently converted to G9a^Δ (>90%) in bone marrow and splenic B cells (B220⁺) from the mb1-Cre genetic background (Fig. 1⇓A). In contrast, non-B cells (B220⁻) retained the G9a^cnd configuration, confirming cell type-specific deletion by mb1-Cre. To generate T cell-specific G9a knockouts, we crossed G9a^cnd mice with lck-Cre-transgenic animals (22). Southern blot analysis of G9a^cnd/lck-Cre offspring confirmed T lineage-specific conversion to G9a^Δ alleles (∼90%) in both thymus and peripheral lymphoid organs (data not shown). Western blot analysis of G9a protein expression revealed almost complete ablation of this HMT in B lineage cells from mb1-Cre crosses, whereas residual G9a was detected in thymocytes and resting T cells from lck-Cre crosses (Fig. 1⇓, B and C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Generation of G9a conditional knockout mice. A, Generation of the G9a^Δ allele in B cells. G9a^cnd mice were crossed with mb1-Cre animals and genomic DNA was isolated from thymus (T) or from the B220⁺ and B220⁻ cell fractions of the spleen (SP) and bone marrow (BM) of mice with the indicated genotypes. Tissue DNA was digested with HindIII and analyzed by Southern blotting. Sizes of the G9a^cnd (4.6 kb) and G9a^Δ (5.8 kb) alleles are indicated at the right. B, Loss of G9a protein and H3K9me2 in B cells. B220⁺ cells were purified from bone marrow or spleen of G9a^Δ/Δ and G9a^cnd/Δ mice. A portion of splenic B cells was cultured for 2 days in LPS-containing medium. Protein lysates were probed on immunoblots with a G9a-specific Ab. Immunoblot analysis of H3K9me was performed on nuclear protein extracts. The blot was sequentially probed with Abs specific for H3K9me2 and H3K9me3 as well as total H3 protein. C, Protein extracts from indicated lck-Cre/G9a^cnd/cnd cells were analyzed by immunoblotting as described in B. Control samples from either heterozygous or G9a-deficient ES cells are included. Total histone H4 levels serve as a protein loading control.

Prior studies established that G9a deficiency in ES cells produces a dramatic and specific decrease in global H3K9me2 (7). To determine whether this functional phenotype is recapitulated in G9a-deficient lymphocytes, we purified these cells and immunoblotted protein extracts for H3K9 modifications. Global levels of H3K9me2 are significantly decreased in resting and LPS-stimulated B cells from mb1-Cre/G9a^cnd mice when compared with control cells (≤20%; Fig. 1⇑B and data not shown). As expected, levels of H3K9me3 are largely unaffected by G9a deletion (7). In contrast, only a modest reduction in global H3K9me2 is observed in T cells from lck-Cre/G9a^cnd mice (Fig. 1⇑C), which correlates with persistent G9a expression. Notwithstanding, these data indicate that G9a is the major H3K9 dimethyltransferase in lymphocytes.

G9a-deficient lymphocytes develop normally

Given the global nature of the H3K9me2 mark on chromatin, we reasoned that G9a would regulate important aspects of gene expression programs that guide lymphocyte development, activation, and differentiation. Initially, we examined lymphocyte development using flow cytometric analyses of cell surface markers. Surprisingly, G9a-deficient mice generate a normal complement of mature B cells in their bone marrow and spleen despite a substantial defect in global H3K9me2 (Fig. 2⇓A). Likewise, G9a deletion by lck-Cre does not significantly alter thymocyte development or generation of splenic T cells (Fig. 2⇓B). Thus, expression of G9a and consequential H3K9me2 are largely dispensable for proper development of effector B and T cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Lymphocyte development in G9a conditional knockout mice. A, Bone marrow or splenic cells from wild-type (G9a^cnd/Δ mb1-cre⁻) or mb1-Cre crosses with G9a^cnd/Δ mice were analyzed by flow cytometry. Bone marrow cells were stained with anti-CD43-biotin, streptavidin-FITC, and anti-B220-PE (top). Spleen cells were stained with anti-IgM-FITC and anti-B220-PE (bottom). All data were gated for live lymphocyte populations. Representative FACS plots from independent analyses of at least eight mice for each genotype are shown. The percentages for each cell population are shown. No reproducible differences were observed in total cell numbers for spleen or bone marrow (BM) samples from any of the genotypes. B, T cell development was examined in thymocytes (top) from lck-Cre/G9a^cnd/cnd and wild-type animals, which were stained with CD4-FITC and CD8-PE (top). CD3-FITC and B220-PE stains revealed normal levels of B and T cell populations in the spleen samples from G9a-deficient animals (bottom).

Lymphocyte proliferation and humoral immune response

To test whether G9a plays an important role in lymphocyte activation programs, B cells were purified from control or G9a^Δ/Δ spleens and cultured with polyclonal agonists LPS or anti-IgM. As shown in Fig. 3⇓A, G9a-deficient B cells exhibit only a modest proliferation defect upon LPS stimulation. However, proliferation of G9a^Δ/Δ B lymphocytes is inhibited ∼4-fold after BCR cross-linking with anti-IgM. Decreased proliferation does not result from enhanced cell death in the induced cultures as monitored by annexin V-7-aminoactinomycin D staining (apoptotic fraction: wild type 31 ± 17%, G9a^Δ/Δ 36 ± 10% for LPS; wild type 53 ± 10%, G9a^Δ/Δ 60 ± 8% for anti-IgM). Furthermore, these effects are agonist specific because mutant cells proliferate normally upon CD40 cross-linking (data not shown). No significant defects are observed in the proliferative capacity of G9a-deficient T cells upon treatment with anti-CD3 Abs (data not shown). These data indicate that G9a is required for optimal activation of B cells following engagement of their Ag receptor.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Proliferative defects and primary immune responses. A, Purified splenic B cells from G9a^Δ/+ or G9a^Δ/Δ mice were stimulated for 2 days with indicated amounts of LPS or anti-IgM. Proliferation was determined by [³H]TdR incorporation after an 18-h pulse. Experiments from two mice of each genotype are shown and are representative of at least 15 independent experiments using both G9a^Δ/+ mb1-cre⁺ and G9a^cnd/Δ mb1-cre⁻ mice as wild-type controls. B, Natural Ig levels. Serum was collected from six G9a^Δ/+ or six G9a^Δ/Δ littermates at 8 wk of age. Steady-state levels of total serum Ig or specific Ig isotypes were determined by ELISA assays. Relative values for total Ig, IgM, Igκ, and Igλ are shown. No significant differences were observed in any IgH isotype (data not shown). Bars represent mean values of Ig concentration for each genotype. C, ELISA for IgH isotypes. Purified splenic B cells were stimulated with 5 ng/ml LPS ± 5 ng/ml IL-4. After 6 days, culture media were collected, and ELISA was used to determine IgG1 and IgE titers. Data are representative of two independent experiments. D, RNA expression analysis for B cell activation genes. Total RNA was prepared from wild-type or G9a-deficient B cells stimulated for 3.5 days with LPS alone (5 μg/ml) or LPS with 5 ng/ml IL-4. RNA was reverse transcribed, and the resultant cDNA was subject to PCR assays specific for Blimp-1, IgJ, AID, germline IgG1 transcripts, or β-actin. Titrations represent 3-fold dilutions. E, Plasma cell formation. G9a^Δ/Δ and control splenocytes were stimulated for 3.5 days in the presence of LPS and IL-4 as in D. The percentage of plasma cells in these cultures was determined by FACS using FITC-B220 and PE-CD138. The circled fractions highlight the plasma cell subset.

In addition to proliferation, the B cell activation program includes Ig isotype switching, affinity maturation, and terminal differentiation into Ig-secreting plasma cells. Isotype switching and affinity maturation require AID, whose expression is regulated by the transcription factor Blimp-1 (36). Importantly, Blimp-1 interacts with G9a to mediate gene-specific H3K9me2 and gene repression (37). To test whether these processes are perturbed upon G9a deletion, we measured steady-state levels of circulating Ig proteins. Steady-state levels of IgH isotypes in G9a^Δ/Δ mice are normal (Fig. 3⇑B and data not shown); however, a 50% reduction in circulating Igλ is observed in the mutant animals. Similar results for IgH isotypes (unchanged) and Igλ (reduced) are obtained following immunization with the T-dependent Ag KLH in CFA (data not shown).

To more rigorously test whether G9a regulates class switch recombination (CSR), purified B cells from control and G9a^Δ/Δ mice were cultured with the polyclonal agonist LPS for 6 days in the presence or absence of the TH2 cytokine IL-4. As shown in Fig. 3⇑C, G9a^Δ/Δ cells undergo class switching, but produce significantly less IgG1 and IgE than their wild-type counterparts. RT-PCR indicates that AID expression is unaffected by G9a deletion (Fig. 3⇑D). Likewise, no differences are observed in germline transcription of the IgG1 locus, a prerequisite for its CSR (Fig. 3⇑D). In contrast, expression of two plasma cell markers, Blimp-1 and J chain (IgJ), are ∼3-fold lower in G9a^Δ/Δ cells upon stimulation with LPS and IL-4 but not with LPS treatment alone (Fig. 3⇑D). Quantitative real-time PCR assays confirmed the reduction in Blimp-1 (3-fold) and IgJ mRNA expression (2.5-fold; data not shown). These findings suggest that G9a deficiency inhibits plasma cell differentiation, rather than CSR, leading to reduced secretion of IgG1 and IgE in LPS + IL-4 B cell cultures. Indeed, FACS indicated a 2-fold reduction of plasma cell differentiation in G9a^Δ/Δ cultures after LPS+IL-4 stimulation for 3 days (Fig. 3⇑E). Again, the paucity of plasma cells in these cultures could not be attributed to enhanced cell death because annexinV-7-aminoactinomycin D staining reveals similar numbers of apoptotic cells for the wild-type and G9a-deficient genotypes (data not shown). Together, these data indicate a specific role for G9a in processing the cues that lead activated B cells into their terminal differentiation program under certain stimulatory conditions.

G9a is dispensable for tissue-specific control of V(D)J recombination

Prior studies established correlations between H3K9me and tissue- or stage-specific repression of recombinase accessibility (18, 19). Moreover, G9a-mediated methylation of H3K9 inhibits recombination of chromosomal substrates (20). Our finding that lymphocyte development is unaffected in G9a-deficient mice suggested that it is dispensable for most aspects of Ag receptor gene assembly. However, these cell-based assays are insufficient to probe whether G9a is required for tissue-specific repression of Ig loci in thymocytes or TCR loci in developing B cells.

To test whether repression of TCR gene assembly is compromised in G9a-deficient B cells, we measured products of V(D)J recombination in DNA from bone marrow or splenic B cells. As shown in Fig. 4⇓A, repression of TCRα and TCRβ recombination is maintained in B cells from G9a-deficient mice. Similarly, the loss of G9a fails to induce rearrangement of TCRγ or TCRδ loci in B cells. Analogous assays were used to probe for loss of tissue-specific Ig gene assembly in G9a-deficient T cells. These sensitive assays fail to detect IgH or IgL gene rearrangements above background levels from contaminating B cells in thymocytes from G9a-deficient mice (Fig. 4⇓B). Although we observed no gross thymocyte developmental defects in G9a-deficient mice (Fig. 2⇑B), it remained possible that cellular selection processes are masking potential rearrangement defects at the endogenous TCRβ locus. To test this possibility, we measured D-to-J rearrangements of Dβ1 and Dβ2 gene segments, which indicate that the initial stage of TCRβ assembly is unaffected by the loss of G9a (Fig. 4⇓B). Likewise, rearrangement of distal (Vβ16), medial (Vβ12), and proximal (Vβ14) Vβ gene segments to Dβ1 remains largely normal upon deletion of G9a (Fig. 4⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Maintenance of tissue-specific V(D)J recombination in the absence of H3K9me2. A, TCR rearrangement assays in G9a-deficient B cells. DNA from purified B220⁺ cells, from bone marrow (BM) or spleen (SP), of two G9a^cnd/Δ or G9a^Δ/Δ mice was subjected to PCR assays for Vα→Jα, Dβ→Jβ, Vγ→Jγ, or Vδ→Jδ coding joins. Samples from a RAG-deficient pro-B cell line (63-12) or thymocytes served as negative and positive controls, respectively. Total DNA content for each sample was measured using PCR for Cλ. The linearity of each assay was confirmed using 1/3 serial dilutions of unsorted G9a^cnd/Δ thymus DNA. B, Rearrangement of Ig and TCR genes in G9a-deficient T cells. Thymocyte DNA from two G9a^Δ/+ or G9a^Δ/Δ mice was subjected to PCR assays for the indicated IgH, IgL, and TCR rearrangements. DNA from B220⁺ bone marrow cells or thymocytes was used as a positive control, and the wedge indicates 1/3 dilutions of these samples. C, ChIP assays for H3K9 modifications of the TCRβ or IgH loci in G9a-deficient precursor B cells. Chromatin from wild-type and G9a-deficient pro-B or a mixture of predominantly pre-B cells was immunoprecipitated with Abs specific for H3K9ac, H3K9me2, H3K9me3, or nonspecific IgG. Immunocomplexes were assayed by PCR for Dβ1 or JH4 sequences. Wedges indicate sequential 1/4 dilutions of each sample.

Our biochemical analyses of G9a-deficient B cells showed that global levels of H3K9me2 were significantly decreased when compared with control cells (Fig. 1⇑B). However, it remained possible that specific regions within the genome, including TCR loci, retain normal levels of H3K9me2, perhaps via the function of a redundant HMT. To address this possibility, we performed ChIP experiment to probe for modifications on chromatin isolated from pro-B (B220⁺/Rag2^−/−) or a mixture of bone marrow pro- and pre-B cells (B220⁺/IgM⁻/Rag2^−/−). Deletion of G9a has no effect on levels of H3K9me3 or acetylated H3K9 (H3K9ac) within the DβJβ cluster of these precursor B cells (Fig. 4⇑C). In sharp contrast, H3K9me2 is largely absent from chromatin associated with both the TCRβ and Ig regions in precursor cells from G9a-deficient mice. Together, these results indicate that H3K9me2 is dispensable for tissue-specific repression of recombinase accessibility at Ig and TCR loci.

Inhibition of Igλ gene assembly in G9a-deficient pre-B cells

In addition to tissue specificity, V(D)J recombination proceeds in a stage-specific manner during lymphocyte development. After IgH gene assembly in pro-B cells, V(D)J recombinase is targeted to the Igκ locus for Vκ→Jκ rearrangement in pre-B cells. Should a pre-B cell fail to make a productive join on either Igκ allele, recombinase is then retargeted to the Igλ locus. Consistent with ordered Ig gene assembly, histone modification patterns at Ig loci change during the pro-B to pre-B cell transition (18, 38).

One consequence of ordered IgL gene assembly is a preponderance of mouse B cells that express Igκ as their L chain (∼90% Igκ, 10% Igλ). In this regard, we noted further skewing of IgL isotypes in G9a^Δ/Δ serum before or after immunization (Fig. 3⇑B and data not shown). Upon further examination, we found that G9a-deficient mice possess fewer Igλ⁺ B cells in their spleens (Figs. 5⇓, A and B). This skewing of IgL usage is not simply due to selection in peripheral lymphoid organs because a similar exaggeration of Igκ:Igλ is observed in G9a-deficient bone marrow. Flow cytometry indicated that IgL isotype exclusion remains intact because Igκ⁺Igλ⁺ double-expressing cells are absent in G9a^Δ/Δ animals (Fig. 5⇓A).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Perturbation of Igλ gene assembly in G9a-deficient B cells. A, Skewed IgL usage in G9a-deficient B cells. FACS of bone marrow or spleen cells from mb1-Cre/G9a^cnd/Δ and control mice stained for surface expression of Igκ and Igλ light chains. FACS data are representative of eight independent mice of each genotype. B, Graphic representation of FACS data for Igλ:Igκ ratios. C, Schematic of the Igλ locus. Boxes represent coding sequences, triangles depict recombination signal sequences, and arrows indicate the predominant Vλ to Jλ rearrangement events observed in mouse B cells. Vλ1 and Vλ2 are separated from their corresponding Jλ segments by ∼20 and 60 kb, respectively. D, DNA extracts were prepared from pro-B, pre-B, or IgM⁺ bone marrow compartments of G9a^cnd/Δ and G9a^Δ/Δ mice. DNA was analyzed for relative levels of IgH (V_H558→J3), Igκ (all Vκ→Jκ1), and Igλ gene rearrangements. Distinct radiolabeled probes were used for general Vλ→Jλ rearrangements or specific VλJλ products as indicated at the right. The linearity of each assay was confirmed using 1/2 dilutions. DNA from the RAG-deficient pro-B cell 63-12 provides a negative control. Cλ serves as a loading control for total DNA content. E, Germline transcription of Ig gene segments. Total mRNA was prepared from pro-B cells (B220⁺ bone marrow on a RAG-deficient background) or a mixture of pro-B and pre-B cells (IgM⁻/B220⁺ bone marrow) from G9a^cnd/Δ or G9a^Δ/Δ mice. RNA samples were subjected to RT-PCR assays specific for JH, Jκ, or individual Vλ and Jλ transcripts as indicated at the left. cDNA from the RAG-deficient pro-T cell line P5424 was included as a negative control and assay linearity was confirmed using serial 1/2 dilutions. F, ChIP assays for H3K9 modifications of IgL loci in G9a-deficient precursor B cells. Chromatin from control and G9a-deficient pro- or pre-B cells was immunoprecipitated with Abs specific for H3K9ac, H3K9me2, H3K9me3, or nonspecific IgG. Immunocomplexes were assayed by PCR for Vλ1, Jλ1, or Jκ sequences. Wedges indicate sequential 1/4 dilutions of each sample.

To test whether redistribution of IgL usage results from perturbations in gene assembly, we measured levels of Vκ→Jκ and Vλ→Jλ joins in purified pro-B, pre-B, and mature B cells from G9a-deficient and control bone marrow. To confirm sorting purities, we measured levels of V_H→D_HJ_H and Vκ→Jκ rearrangements in each cell fraction (Fig. 5⇑D). As expected, V_H→D_HJ_H rearrangements are readily detected in pro- and pre-B cells from both genotypes, whereas Vκ→Jκ recombinations are restricted to pre-B cells. These data indicate that stage-specific restrictions on Igκ are maintained in the absence of G9a. Thus, premature activation of Igκ does not explain the paucity of Igλ-producing cells in mutant mice. In contrast, Vλ→Jλ rearrangement is consistently reduced ∼3-fold in G9a-deficient pre-B cells (Fig. 5⇑D).

To extend these findings, we examined the effects of G9a deficiency on recombination of specific VλJλ gene segments. The mouse Igλ locus consists of two cassettes with the first encoding a single Vλ gene segment, Vλ1, which rearranges to Jλ1 or Jλ3 (Fig. 5⇑C). A second cassette contains two V and two J segments (Vλ2, Vλx, Jλ2, and Jλ4). The predominant rearrangement in this cassette is Vλ2→Jλ2 because Vλx rarely rearranges and Jλ4 lacks a consensus recombination signal sequence (28). To determine the specificity of Vλ→Jλ rearrangement defects in G9a^Δ/Δ mice, we performed PCR assays that distinguish VλJλ joins in each cassette (Vλ1→Jλ1 or Vλ2→Jλ2). As shown in Fig. 5⇑D, Vλ1→Jλ1 and Vλ2→Jλ2 joins are both reduced ∼3-fold in G9a^Δ/Δ pre-B cells. These data indicate that G9a potentiates assembly of the entire Igλ locus during B cell development.

The recombination potential of a given gene segment correlates with its germline transcription (39, 40), indicating that shared cis-acting elements regulate chromatin accessibility to transcriptional and recombinational machinery. To investigate accessibility mechanisms at Igλ, we performed assays for germline transcription in pro-B cells isolated from Rag2^−/− animals or a mixture of pro-B and pre-B cells (B220⁺IgM⁻) from Rag2^+/+ mice. These assays revealed a consistent reduction in steady-state levels of germline Vλ1 and Vλ2 transcripts in precursors from G9a-deficient animals (Fig. 5⇑E). In contrast, germline transcripts of Jλ, JH, and Jκ segments are unaffected by the loss of G9a. These data suggest that reduced Igλ usage results from a Vλ-specific defect in chromatin accessibility, which inhibits Vλ→Jλ recombination.

To extend these findings, we performed ChIP analyses on the same cell populations. Consistent with immunoblot analyses of global methylation, H3K9me2 levels are significantly decreased at all IgL loci in G9a-deficient cells (Fig. 5⇑F). Loss of this histone mark alone does not impair V(D)J recombination because similar decreases in H3K9me2 are observed at IgH and Igκ gene segments, which rearranged at wild-type levels. ChIP analyses also revealed a 2- to 3-fold reduction of H3K9 acetylation in G9a^Δ/Δ pre-B cells at both the Vλ1 and Vλ2 segments (Fig. 5⇑F and data not shown). In contrast, H3K9 acetylation of Jλ1, Jλ2, and Jκ segments is comparable in wild-type and G9a-deficient pre-B cells (Fig. 5⇑F and data not shown). We conclude that Igλ gene assembly is inhibited in G9a-deficient pre-B cells due to altered patterns of epigenetic modifications at Vλ segments, leading to decreased chromatin accessibility.