Oct2 enhances antibody-secreting cell differentiation through regulation of IL-5 receptor α chain expression on activated B cells

Mice lacking the Oct2 transcription factor die perinatally, but reconstitution of immunodeficient Rag-1^−/− mice with oct2^−/− fetal liver has allowed the behavior of oct2-null lymphoid cells to be assessed. These studies highlighted the contribution that Oct2 makes to the differentiation of specialized mature B cell subsets (unpublished data) (24). Surprisingly, no evidence from studies on oct2 mutant mice supports the proposal that Oct2 regulates the transcription of Ig genes, either alone or in conjunction with its sometime partner, the coactivator OBF1/OCA-B/Bob.1 (25). Nevertheless, the serum of mice reconstituted with oct2^−/− lymphocytes contains significantly lower levels of Ig of all isotypes than control reconstituted mice, whether naive or immunized (23–25).

We therefore assessed the ability of oct2-null B cells to differentiate to ASCs in vitro in response to CD40L and cytokines. These conditions were chosen for two reasons. First, we have made detailed, quantitative studies of the behavior of normal B cells under these conditions, measuring rates of cell division, isotype switching, and ASC differentiation (1), and we can therefore recognize deviations from normal responses. Second, oct2^−/− B cells respond poorly to other polyclonal mitogens, such as LPS and BCR cross-linking antibodies (23, 24, 26), so these activators could not be used.

We found that oct2-null B cells behave indistinguishably from WT B cells in CD40L+IL-4 cultures, undergoing isotype switching to IgG1 and differentiating to Syndecan1⁺ ASCs in a cell division–linked manner (Fig. 1 A). Sorted Syndecan1⁺ cells were ASCs, as determined by ELISPOT, and Oct2-deficient ASCs secreted at least as much Ig as did controls (Fig. 1 B).

We previously demonstrated that when IL-5 was added to CD40L+IL-4 cultures, ASC differentiation was enhanced. The ASCs generated in such cultures were rapidly dividing plasmablasts that were no longer able to switch isotypes (1). This IL-5–dependent enhancement can be seen in the control cultures of Fig. 1 A but not in oct2^−/− B cell cultures (Fig. 1 C). The sensitivity of B cells to IL-5 is also reflected in the percentages of IgG1⁺ cells in cultures with increasing IL-5 concentrations; at high concentrations, switched cells are less frequent in control cultures because a higher proportion of cells have become ASCs, which do not switch (1). In oct2^−/− cultures, IL-5 had no effect on the proportion of switched cells, an indirect consequence of the lower frequency of ASC differentiation (Fig. 1 D). These data reveal both a new role for Oct2 in the cytokine-driven augmentation of ASC production and the absence of an essential role for Oct2 in the regulation of Ig production by preformed ASCs.

Most resting B cells express the β chain but not the α chain of the IL-5 receptor and therefore do not respond to the cytokine. Upon activation, IL-5Rα is induced, and the cells become responsive (Fig. 2 A) (27, 28). We therefore examined more closely the regulation of IL-5Rα expression on WT and mutant B cells. Resting B cells were activated for 2 d with anti-CD40 antibodies. This treatment did not significantly elevate IL-5Rα expression from resting cell levels (Fig. 2 A), indicating that activation alone is not sufficient for IL-5Rα induction. Weber et al. (29) reported that IL-4 addition enhanced IL-5Rα expression on LPS-treated B cells. We therefore looked for IL-5Rα induction in response to IL-4. This cytokine alone did not induce IL-5Rα, but the combination of αCD40 and IL-4 was a strong inducer (Fig. 2 A), and this induction required Stat6, the primary signal transducer of the IL-4R (Fig. 2 B). Oct2-deficient B cells failed to completely up-regulate IL-5Rα under these conditions, although expression of the receptor rose from resting cell levels (Fig. 2 C). Thus, Oct2 contributes to, but does not entirely control, IL-5Rα levels. Despite its role as an integrator of CD40 and IL-4 signals in the context of germinal center development, the coactivator OBF1 had no influence on IL-5Rα expression under these conditions (Fig. 2 C).

The observation that ASC differentiation is cell division dependent (1) prompted us to ask whether expression of the IL-5Rα chain was also division linked. We purified resting B cells, labeled them with the cell division tracking dye CFSE, and stained for CD125 expression after 48 h in culture with αCD40+IL-4. The results indicated that IL-5Rα is expressed even in undivided B cells (Fig. 2 D). Thus, although IL-5 acts to enhance ASC generation, it is unlikely to contribute significantly to the cell division dependence of this differentiation step.

We next examined the molecular consequences of IL-5 treatment on B cells. In these assays, the B cells were from a mouse expressing the GFP from the endogenous blimp1 promoter. Thus, all ASCs generated in vivo or in vitro express GFP (30). Addition of IL-5 to CD40L+IL-4 cultures increased the number of cells expressing blimp1/GFP, and the GFP⁺ cells were Syndecan1⁺ ASCs (Fig. 2 E and not depicted).

To assess the ability of IL-5 to directly activate transcription of the blimp1 gene, we prepared replicate cultures of WT B cells in LPS. 3 d later, cycloheximide (CHX) was added to both cultures at a concentration (50 μM) that we determined would inhibit protein synthesis while preserving cell viability. 1 h later, IL-5 was added to one culture only, and cells were harvested 5 h after cytokine addition. Quantitative PCR analysis showed that LPS alone is a strong inducer of blimp1 expression, as expected (Fig. 2 F). Addition of CHX did not influence blimp1 messenger RNA levels, but the further addition of IL-5 enhanced blimp1 expression approximately fourfold over LPS alone (Fig. 2 F). Blimp1 up-regulation was observed only in the IL-5⁺ CHX-treated and not the IL-5⁺ DMSO-treated sample. Similar results were obtained in TD culture conditions, using anti-CD40+IL-4 to drive differentiation (Fig. 2 F). These results suggest that the blimp1 gene is repressed by a labile repressor but that release from repression is alone insufficient to activate the gene. The addition of IL-5 rapidly up-regulates blimp1 transcription in the absence of new protein synthesis, most likely through activation and nuclear transport of its cognate signal transducer, Stat5 (31). There are two perfect consensus Stat5 binding sites in the blimp1 promoter, at positions −1325 and −931 relative to the start of transcription (32).

In addition to its effects on differentiation, IL-5 acts as a coactivator for B cell proliferation when in combination with anti-CD38 (33). This response is also absent in oct2^−/− B cells (Fig. 3 A). However, the proliferation of eosinophils, another IL-5–responsive cell type, was not dependent on Oct2. Oct2^−/− fetal liver stem cells cultured in IL-5 gave rise to normal numbers of eosinophil colonies of normal size (Fig. 3 B). Therefore, expression of the IL-5R is regulated differently in B cells and eosinophils.

As B cells required Oct2 to respond to IL-5 in all assays that we have performed, we assessed whether the IL-5R might require Oct2 for its expression. Although the IL-5Rβ chain was not affected in the mutants (not depicted), quantitative PCR analysis showed that the IL-5Rα chain was expressed at abnormally low levels in primary B cells from oct2^−/− mice compared with control B cells (Fig. 3 C). Indeed, the loss of Oct2 had an even more dramatic effect on expression of this new putative target gene than was seen for the bona fide Oct2 target, CD36 (Fig. 3 C) (34, 35).

To begin to dissect the mechanism of action of Oct2 on the IL-5Rα gene, we made use of OM1A, a cloned oct2^−/− B lymphoma cell line, and a cloned derivative that had been transduced with a retroviral construct expressing an Oct2–estrogen receptor (ER) fusion protein (36). In the absence of estradiol, Oct2-ER is cytoplasmic, and induction causes the Oct2-ER protein to moved quickly and quantitatively to the nucleus (unpublished data). The cells were induced with estradiol in the presence of CHX to ask whether Oct2 could directly activate transcription of the endogenous IL-5Rα gene. Interestingly, in contrast to the CD36 gene, estradiol⁺CHX modestly increased IL-5Rα expression in the absence of Oct2-ER, implying that IL-5Rα is normally repressed by a labile factor. Induction of Oct2 dramatically enhanced both CD36 and IL-5Rα transcription in the absence of new protein synthesis (Fig. 3 D), suggesting that Oct2 acts directly to activate the IL-5Rα gene.

Using 5′ rapid amplification of complementary DNA (cDNA) ends (RACE), we found that the majority of IL-5Rα transcripts in the spleen were initiated at or very near the A residue at position −4870 bp from the start of translation (or 166 bp upstream of the 5′ end of the IL-5Rα reference cDNA sequence, available from GenBank/EMBL/DDBJ under accession no. NM_008370; Fig. 4 A). The adjacent sequences did not contain a classical TATA box, but a 5-bp AT-rich sequence was found at approximately position −30 from the putative start of transcription. In their review of the features of eukaryotic promoters, Smale and Kadonaga (37) note that AT-rich sequences at approximately position −30 can profoundly influence promoter strength, even if they do not resemble the TATA consensus. In addition, there is a five out of seven match to the consensus initiator sequence (37) near the common 5′ end of the IL-5Rα RACE products. These features suggest that the preferred transcriptional start site for the IL-5Rα gene in splenocytes (of which B lymphocytes are the major cell type) lies in this region.

A search of the upstream sequence of the IL-5Rα gene using the ConSite transcription factor binding site search tool (38) revealed two nonconsensus octamer motifs (Fig. 4 A). Two methods were used to determine whether Oct2 is able to bind directly to these sites. First, chromatin immunoprecipitation (ChIP) was performed, using chromatin isolated from control WEHI231 or Oct2-deficient OM1A B lymphoma cell lines and a polyclonal anti-Oct2 antiserum. The results indicated that Oct2 bound specifically to chromatin in the IL-5Rα promoter in the vicinity of the two octamer motifs (5′ ChIP) and to the CD36 promoter, a known Oct2 target gene (Fig. 4 B). Two independent experiments are shown, and the relative enrichments (using WEHI231⁺ preimmune serum as the reference sample) differ between experiments because the second experiment had a lower level of nonspecific immunoprecipitation overall. EMSA was used to learn whether one or both of the octamer motifs were bound by Oct2. IL-5Rα octamer site 1 (the most 5′) was able to bind to a nuclear protein, forming a complex that had the same mobility as an Oct2–consensus octamer DNA complex (V_H octa) and that was specifically supershifted by a monoclonal anti-Oct2 antibody (Fig. 4 C). The more 3′ IL-5Rα octamer site 2 did not exhibit specific binding to Oct2. Collectively, these data indicate that Oct2 binds directly to the promoter of the IL-5Rα gene to activate its transcription specifically in B cells.

We crossed the oct2 mutant mice to a transgenic mouse strain expressing the IL-5Rα gene constitutively in B cells (39) to determine whether enforced expression of IL-5Rα could rescue some aspects of the oct2 mutant phenotype. Because oct2^−/− mice do not survive postnatally, all experimental mice were generated using adoptive transfer of fetal liver stem cells into Rag1^−/− recipients. The reconstituted mice were analyzed after at least 2 mo of reconstitution. We specifically assessed the ability of the transgene to complement the defective ASC differentiation of oct2^−/− B cells identified in vitro (Fig. 1). Purified, resting splenic B cells from mice reconstituted with control, oct2-null or oct2-null/IL-5Rα transgenic fetal liver stem cells were cultured in CD40L+IL-4 in the presence or absence of IL-5. After 4 d, cells were harvested and stained for Syndecan1 expression, and antibody in culture supernatants was measured by ELISA. IL-5 increased the proportion of Syndecan1⁺ cells in the control but not the oct2-deficient cultures. The IL-5Rα transgene largely restored the ability of oct2^−/− cells to differentiate in response to IL-5 in a concentration-dependent manner, generating a higher percentage of Syndecan1⁺ cells and a corresponding increase in secreted Ig in the cultures (Fig. 5 A and not depicted). These reconstitution studies indicate that a primary role of Oct2 in these TD cultures is to activate the IL-5Rα gene, enabling efficient differentiation of ASCs. We also observed a lower proportion of Syndecan1⁺ ASCs in the spleens and bone marrow of mice reconstituted with oct2^−/− B cells expressing the blimp1/gfp reporter allele compared with controls (Fig. 5 B and not depicted). This may reflect, in part, a deficiency of ASCs that would have been generated in vivo during TD responses to environmental antigens, mediated by IL-5 through Oct2 and IL-5Rα.

We found that the IL-5Rα transgene did not convincingly rescue any of the other defects of the B cell lineage of oct2 mutant mice in vivo. For example, the proportion of B cells in the spleen remained low (Fig. 5 B), and there was an abnormally high percentage of immature cells in the spleen of oct2-null mice, whether or not they bore the IL-5Rα transgene (Fig. 5 C). Similarly, the transgene failed to rescue the marginal zone (MZ) or B-1 cell populations of the oct2^−/− mice (Fig. 5, D and E). It was regularly noted that the B-1 compartment of control transgenic mice was larger than control nontransgenics, consistent with the reported activity of the transgene and the particular responsiveness of B-1 cells to IL-5 (6). The failure of the transgene to complement these defects highlights the selective role that IL-5 and its receptor play during B cell development, and emphasizes the fact that Oct2 regulates other genes important for normal B lymphocyte development and function, such as those mediating BCR signaling (23).

Recently, many groups have worked to elucidate the intrinsic factors that are required for the differentiation of antigen-activated B cells into ASCs (7). These studies have identified essential transcriptional regulators, such as Blimp1, Bach2, Bcl6, IRF4, Xbp-1, and Pax5, whose activities orchestrate the transition from the mature B cell and germinal center genetic programs to one dedicated to high level Ig synthesis and secretion. Less clear are the modes of action of extrinsic signals, as well as their associated downstream signaling pathways, in initiating or enhancing this important transition. In this study, we consider the T cell–derived cytokine IL-5 as one such extrinsic signal.

In the mouse B cell lineage, IL-5 seems to serve primarily as an enhancer of antibody. It supports cells of the B-1 cell lineage, which constitutively express IL-5Rα (27, 40). These cells are the source of most “natural” Ig in the serum and are poised to differentiate very rapidly to ASCs upon activation (41). Resting follicular B cells are unresponsive to IL-5. We show that two T cell signals, CD40L and IL-4, strongly synergize to condition splenic B cells to become IL-5 sensitive. This dramatic synergy would ensure that only those B cells activated in the context of T cell help could be accelerated toward ASC differentiation by IL-5. In follicular B cells, IL-5 has a clear and specific activity in enhancing ASC differentiation from follicular B cells (Fig. 1) (1). Finally, IL-5 has been shown to improve the survival of BM-derived plasma cells (42). We show that Oct2 is required for B cells to be able to effectively “read” this T cell signal. Oct2 binds directly to the promoter of the IL-5Rα gene and increases its transcription in splenic B cells. We have demonstrated a dependence on Oct2 during ASC differentiation in vitro in response to IL-5 stimulation, as well as a specific improvement of differentiation efficiency when IL-5Rα was provided to Oct2-deficent B cells as a transgene.

Overexpression of IL-5 has been shown to significantly increase antibody levels in vivo, predominantly from an expanded population of B-1 cells (40). Conversely, IL-5Rα–deficient or IL-5–depleted mice showed impaired B-1 cell development and reduced serum IgM, IgG3 and mucosal IgA (6, 39, 43). The IL-5Rα transgene used in this study supported increased percentages of peritoneal B-1 cells in control mice but no rescue of this population in the oct2^−/− mutants (Fig. 5 E). Furthermore, it was not possible to rescue the serum antibody defect in oct2-deficient animals by reconstitution with this single Oct2 target gene (Fig. 5 F), probably because serum Ig mainly comprises IgM from B-1 cells, and TD ASCs represent a minority of both cells and Ig in vivo. The contribution of other Oct2 target genes to antibody responses, through BCR signaling, B-1 and MZ B cell populations, and TLR4 signaling (important for both TI and TD humoral responses) (44), probably overshadows the effect of IL-5 in vivo.

It is not yet known whether Oct2 contributes to human TD ASCs through IL-5, or another cytokine, in an analogous way to the mouse. There is less evidence for a role of IL-5 on human B lymphocytes, although actions of IL-5 on human B cells have been reported (45, 46). Instead, IL-6 and/or IL-10 induce ASC differentiation of CD40L-activated human B cells in vitro (47–49). Both IL-6 and IL-10 signal through Stat3, as can IL-5 (50).

Our in vitro evidence (Fig. 2 E and not depicted) suggesting that IL-5 alone is unable to up-regulate blimp1 expression in mouse B cells indicates that the blimp1 gene is insulated against untimely induction. Here, a short pulse of IL-5 was only able to stimulate blimp1 expression after 3 d of culture. At earlier time points (24–48 h of culture), blimp1 expression was not influenced by IL-5 addition (unpublished data). These observations accord with those reported by Horikawa et al. (2) showing that IL-5 induces blimp1 in CD38-activated B cells slowly (over several days). These observations, the requirement for several cell divisions before ASC differentiation occurs in vitro (1), and the positive effect that an inhibitor of protein synthesis had on blimp1 expression (Fig. 2) are all consistent with the obligate loss of a blimp1 repressor over the first hours and days of activation. Once the repression is released, the blimp1 gene of a plasma cell precursor would become responsive to the positive influence of IL-5 signaling. Strong candidates for the blimp1 repressors are Bcl6 (8) and Bach2, which have been shown to bind directly to the blimp1 gene (51, 67), and Pax5 (9, 52, 53). Indeed, Horikawa et al. showed that bcl6 messenger RNA levels declined within hours of IL-5 addition in their culture system (2), and Kallies et al. showed that a reduction of Pax5 activity precedes blimp1 induction (17).

In vivo, signals that derepress the blimp1 gene are likely to come from T cells and their products and through the BCR, as has been proposed by Niu et al. (18). In one model, Bcl6 competes for binding on the blimp1 gene with the transcriptional activator Stat3, with which it shares a consensus DNA binding motif (54). A strong BCR signal leads to degradation of Bcl6, vacating the DNA binding site in favor of Stat3, which is activated in response to IL-6 and other cytokines. It is intriguing to note that Stat3 binds to the same consensus DNA binding motif as Stat5 (55). Stat5 is a major transducer of signals from the IL-5R (56). Thus, Bcl6 and Stat5 may also compete as negative and positive regulators of blimp1 expression.

Recent work has highlighted the importance of the combined influences of blimp1 and IRF4 during ASC differentiation (17, 57, 58). Thus a model for TD ASC differentiation in vivo (Fig. 6) may incorporate the combined effects of antigen receptor signaling (to relieve Bcl6 repression) and T cell signals: a CD40 signal to up-regulate IRF4 expression through the canonical NF-κB pathway and to further reduce Bcl6 expression (59), and cytokines such as IL-5, which boost the expression of the blimp1 gene (Fig. 2 F). Once blimp1 and IRF4 levels reach a critical threshold, a mutual reinforcement has been proposed (57, 58) that drives the ASC differentiation program forward.

Both Oct2 and OBF1 have been shown by genetic models to be entirely dispensable for Ig gene expression in B cells and plasma cells. However, these two transcription factors are required for some of the main “sensing” functions of B cells (e.g., Oct2 for LPS sensing through TLR4, and Oct2 and OBF1 for antigen signaling through the BCR and for cytokine responses; see Results; unpublished data). As a result, many aspects of peripheral B cell behavior are faulty in the mice. OBF1^−/− mice lack germinal centers and MZ B cells, and display a reduced mature B cell pool (60, 61). Oct2-null mice also lack MZ B cells and most mature B cells, as well as the B-1 cell population (unpublished data) (24). In the face of these many defects, it was difficult to discern their specific roles in ASC differentiation. Using quantitative in vitro analyses, novel modes of action for Oct2 and OBF1 have been discovered, whereby each is required for a different aspect of TD terminal ASC differentiation (see Results) (19). The influence of each factor on serum Ig is indirect and molecularly distinct, with OBF1 required for TD but not TI ASC differentiation, and with Oct2 enabling the IL-5 response. The latter defect was molecularly characterized and specifically corrected in the studies described in this paper. Such approaches strongly complement global gene expression studies to determine the physiological roles of tissue-restricted gene regulators such as Oct2 and OBF1.

All mice used in the study were on a C57BL/6 genetic background, including the blimp1-GFP reporter (30) and IL-5Rα transgenic mice (39). All animal studies were approved by the Animal Ethics Committee of the Royal Melbourne Hospital Research Foundation. Splenic B cells were prepared as previously described (19), using anti-CD45R/B220–coupled magnetic beads (Miltenyi Biotec). For CFSE labeling experiments, Percoll gradients were included to enrich for resting cells, as previously described (1). Soluble CD40L and recombinant IL-4 and IL-5, purified from baculovirus-infected Sf9 cells (both used at 1,000 U/ml unless otherwise noted in the figures), and LPS (from 10 μg/ml Escherichia coli 0111:B4; Sigma-Aldrich) were used as previously described (1, 19). Anti-CD38 (NIM-R5) and anti-CD40 (FGK-4.5) were purified from hybridomas provided by M. Howard (ChemoCentryx Inc., Mountain View, CA) and A. Rolink (University of Basel, Basel, Switzerland), and were used at 10 μg/ml. [³H]Thymidine incorporation was used as previously described (23) to measure proliferation in vitro.

Embryonic day 18 fetal liver cells were cultured at 25,000 cells per milliliter in 1-ml cultures of Dulbecco's modified Eagle's medium with 20% fetal calf serum and 0.3% agar. The reagents used to stimulate eosinophil colony formation were 0.1 ml of pokeweed mitogen–stimulated spleen-conditioned medium containing GM-CSF and IL-3, or 10 ng/ml IL-5, as a specific stimulus for eosinophil colony formation. Cultures were scored after 7 d. Cell counts ± SD were made on individual colonies in the stained cultures.

For the experiments in Fig. 2 F, the sensitivity of LPS-activated B cells to CHX was determined. B cells were cultured in LPS for 3 d, and then for 6 h with increasing concentrations of CHX, after which viability (using propidium iodide exclusion) and protein translational activity (using incorporation of 1 μCi per well of [³⁵S]-Express protein labeling mix; Perkin Elmer) were measured. At 50 μM CHX, translation was completely inhibited, but cell viability was equivalent to control cultures treated with DMSO only (not depicted). B cells were cultured for 3 d with the stimuli shown in Fig. 2 F, CHX or DMSO were added, and IL-5 was added to selected cultures 1 h later. Cells were harvested for RNA from all cultures 5 h after cytokine addition.

The oct2^−/− B lymphoma cell line OM1A (36) was infected with a retroviral vector (pMX-pie) (62) expressing GFP and oct2 fused to the hormone binding domain of the human ER from a single transcript. Puromycin resistant clones were selected for high GFP expression, and the β-estradiol–inducible nuclear translocation of the oct2-ER fusion protein was confirmed by subcellular fractionation and Western blotting (unpublished data). For the induction experiments in Fig. 3 D, a clone of OM1A–oct2-ER was cultured with 50 μM CHX (or DMSO as a vehicle control). 1 h later, IL-5 was added to the indicated cultures, and the cells were harvested for RNA after a further 5 h.

Anti-Syndecan1 (281-2), anti-CD125 (T21), and anti-B220 (RA3-6B2; all PE conjugated), and anti–thy-1.2 (30-H12), anti-IgM (R6-60.2), anti-IgD (11-26c.2a), and anti-CD23 (B3B4; all FITC conjugated) were purchased from BD Biosciences. Anti-CD21 was purified from a hybridoma supplied by K. Inoue (Osaka University Medical School, Osaka, Japan) and biotinylated. A. Strasser (Walter and Eliza Hall Institute, Melbourne, Australia) purified the NimR-5 (anti-CD38) antibody. CFSE labeling was performed as previously described (1).

In Fig. 1 B, 200 sorted Syndecan1⁺ cells from day 4 of a culture (CD40L+IL-4) of resting B cells were plated for ELISPOT assay, performed essentially as previously described (1) using anti-Igκ–coupled horseradish peroxidase as the developing reagent. For the ELISA assay, 6,000 sorted Syndecan1⁺ cells were plated in medium without stimuli, and the culture supernatants were harvested 6 h later. Measurement of Igκ, IgM, and IgG1 levels in culture supernatants, or Ig levels in serum, was performed as previously described (63).

First-strand cDNA was transcribed (SuperScript III First-Strand Synthesis System; Invitrogen) from total RNA (RNeasy Mini Kit; QIAGEN) using the manufacturers' protocols. Real-time PCR analysis was performed using an SYBR green system (Quantitect; QIAGEN), according to the manufacturer's instructions. The expression data were analyzed on a sequence detection system (ABI Prism 7900HT; Applied Biosystems), using relative quantification of gene expression. Expression was normalized using hydroxymethylbilane synthase (Hmbs) as a housekeeping gene. For the ChIP assays, the absolute quantification using standard curves was used to measure the products for PCR reactions.

Primers used for cDNA amplification were as follows: Hmbs, (forward) 5′-GACCTGGTTGTTCACTCCCTGAAG-3′ and (reverse) 5′-GACAACAGCATCACAAGGGTTTTC-3′; CD36, (forward) 5′-CAGCTCATACATTGCTGTTTATGCATG-3′ and (reverse) 5′-GGTACAATCACAGTGTTTTCTACGTGG-3′; oct2, (forward) 5′-GCATCGAGACGAATGTCCGCTTCG-3′ and (reverse) 5′-GTAGCTGGTCGGCTTTCCCGGGCT-3′; IL-5Rα, (forward) 5′-CAGTGGGAGAAACCACTTTCTGCC-3′ and (reverse) 5′-GAGATGCCATTCTACCAAGGACTTC-3′; and blimp1, the QuantiTect Primer Assay for Mm_Prdm1 (QIAGEN).

The transcriptional start site for the mouse IL-5Rα gene was mapped using a RACE cDNA amplification kit (SMART; BD Biosciences). Gene-specific nested primers were 5′ IL-5Rα–1 (5′-GCGTTAGCTCTGTCTCTGCTGATAACAGGGC-3′) and 5′ IL-5Rα–2 (5′-CTGCATACCGGGACCTGTCCAGTGAGCTTC-3′). A single prominent product was generated from the 5′ RACE PCR reaction, which was purified from an agarose gel, and cloned. At least one dozen clones were bidirectionally sequenced to ensure that microheterogeneities in sequence length were not overlooked.

ChIP was performed essentially as described by Ramirez-Carrozzi et al. (64). Specifically, chromatin from 1.2 × 10⁸ WEHI231 and OM1A cells was cross-linked using 1% formaldehyde for 10 min at room temperature. The cells were collected after two cold PBS washes, resuspended in 10 ml of cell lysis buffer (5 mM Pipes [pH 7], 85 mM KCl, 0.5% NP-40), and lysed for 10 min on ice. Nuclei were resuspended in 1 ml of nuclei lysis buffer (50 mM Tris-HCl [pH 7], 10 mM EDTA, 1% SDS) and sonicated with 30-s pulses using a sonifier (Branson) until the DNA was reduced in size to ∼500 bp. The purified chromatin (350 μg) was diluted twofold with ChIP dilution buffer (16.7 mM Tris-HCl [pH 7], 167 mM NaCl, 1.2 mM EDTA, 0.01% SDS, 1.1% Triton X-100) supplemented with 100 μg/ml of sonicated herring sperm DNA and protease inhibitors. After preclearing with 100 μl of protein G–sepharose for 2 h at 4°C, the diluted chromatin was immunoprecipitated overnight at 4°C using 5 μl of rabbit anti-Oct2 antiserum (26) or preimmune serum. Immune complexes were recovered by a 2-h incubation at 4°C with 50 μl of protein G–sepharose, washed once each with low-salt buffer (50 mM Hepes [pH 7], 0.15 M NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 0.1% deoxycholate), high-salt buffer (the same as low-salt buffer but with 0.5 M NaCl), and LiCl wash buffer (0.25 M LiCl, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl [pH 7]), and twice with TE (10 mM Tris-HCl [pH 8], 1 mM EDTA). The DNA was extracted with 500 μl of elution buffer (0.1 M NaHCO₃, 1% SDS), supplemented with 0.2 M NaCl, and incubated overnight at 65°C to reverse DNA cross-links. 10 mM EDTA, 40 mM Tris-HCl (pH 7), and 80 μg/ml proteinase K were added, and after a further 4–5 h at 45°C, the DNA was extracted with phenol/chloroform and precipitated. PCR to detect immunoprecipitated DNA used the following primers: CD36 ChIP, (forward) 5′-CCTTGTGAGCTTAACTAGCAGTGAAGTGTG-3′ and (reverse) 5′-CCAGTTGCTCCACACATTTCAGAAGGCAGC-3′; IL-5Rα 5′ ChIP, (forward) 5′-GACGGATGAGCTGGGTCAGGCTAGATAG-3′ and (reverse) 5′-CAATCTGAGAACTCTGCGAGCTTTAGCC-3′; and IL-5Rα 3′ ChIP, (forward) 5′-GGCTAAAGCTCGCAGAGTTCTCAGATTG-3′ and (reverse) 5′-GCGTTAGCTCTGTCTCTGCTGATAACAGGGC-3′.

Nuclear extracts were prepared (65) and EMSA was performed (26) as previously described. Probes were annealed, and double-stranded oligonucleotides were labeled using [³²P]dATP and Klenow DNA polymerase. Oligo sequences were as follows: IL-5Rα octa-1, (forward) 5′-GGGGTCACATATCAGATATCTTGCATATCAGATATC-3′ and (reverse) 5′-TTTGATATCTGATATGCAAGATATCTGATATGTGACCCC-3′; IL-5Rα octa-2, (forward) 5′-CTCTGCTAGTCACTCAGCATGAAAATCCTCAGACT-3′ and (reverse) 5′-TTTAGTCTGAGGATTTTCATGCTGAGTGACTAGCAGAG-3′; and V_H octa, (forward) 5′-CCTTGCCTCATGAGTATGCAAATCATGTGCGACTG-3′ and (reverse) 5′-TTTCAGTCGCACATGATTTGCATACTCATGAGGCAAGG-3′.

Day 13 embryos were recovered, fetal livers were dissected, and cell suspensions were prepared. After genotyping, the cells of interest were transplanted into irradiated (300 Rad) Rag1^−/− recipients, as previously described (23), at a ratio of three recipients per donor liver. Reconstitution was monitored by the appearance of mature lymphocytes in peripheral blood, and only stably reconstituted mice were included in the analyses presented in this paper.

We thank Dr. S. Nutt for critical reading of the manuscript, S. Mifsud and L. DiRago for technical assistance with colony assays, and J. Gilbert and E. Loza for animal care. We also thank Professor T. Speed for guidance on statistical analyses.

This work was supported by the Australian National Health and Research Council and the Cancer Council of Australia, and by a Grant-in-Aid for Scientific Research on Priority Areas from the Japanese Ministry of Education, Culture, Sports, Science and Technology.

The authors have no conflicting financial interests.