Tracking germinal center B cells expressing germ-line immunoglobulin γ1 transcripts by conditional gene targeting

Abstract

Germinal centers (GCs) represent the main sites for the generation of high-affinity, class-switched antibodies during T cell-dependent antibody responses. To study gene function specifically in GC B cells, we generated Cγ1-cre mice in which the expression of Cre recombinase is induced by transcription of the Ig γ1 constant region gene segment (Cγ1). In these mice, Cre-mediated recombination at the fas, Igβ, IgH, and Rosa26 loci occurred in GC B cells as early as 4 days after immunization with T cell-dependent antigens and involved >85% of GC B cells at the peak of the GC reaction. Less than 2% of IgM⁺ B cells showed Cre-mediated recombination. These cells carried few Ig somatic mutations, expressed germ-line Cγ1- and activation-induced cytidine deaminase-specific transcripts and likely include GC B cell founders and/or plasma cell precursors. Cre-mediated recombination involved most IgG1, but also a fraction of IgG3-, IgG2a-, IgG2b-, and IgA-expressing GC and post-GC B cells. This result indicates that a GC B cell can transcribe more than one downstream C_H gene before undergoing class switch recombination. The efficient induction of Cre expression in GC B cells makes the Cγ1-cre allele a powerful tool for the genetic analysis of these cells, as well as, in combination with a suitable marker for Cre-mediated recombination, the tracking of class-switched memory B and plasma cells in vivo. To expedite the genetic analysis of GC B cells, we have established Cγ1-cre F₁ embryonic stem cells, allowing further rounds of gene targeting and the cloning of compound mutants by tetraploid embryo complementation.

During T cell-dependent (TD) antibody responses, germinal centers (GCs) are formed in secondary lymphoid organs (1). GCs result from an orchestrated series of cellular interactions between antigen-specific B and CD4 T helper (Th) cells starting at the border between B and T cell areas and continuing at later stages in close proximity to the follicular dendritic cell (FDC) network within established GCs (2–5). During the GC reaction, rapidly proliferating B cells accumulate nontemplated point mutations (and less commonly deletions or duplications) within their rearranged Ig heavy (H) and light (L) chain variable region genes at a high rate (6). This process, which is initiated by activation-induced cytidine deaminase [AID; (7)], leads to the establishment of a pool of mutated B cells, expressing a repertoire of B cell receptor (BCR) specificities. From this pool, a subset of cells, namely those recognizing antigen on FDCs through their high-affinity BCRs, is selected into the compartment of long-lived resting memory-B and antibody-secreting plasma cells, whereas the majority of the remaining cells with low-affinity BCRs or lacking BCR expression because of crippling mutations in rearranged Ig genes die by apoptosis (8).

GCs also represent sites where B cells undergo Ig class switch recombination (CSR), leading to the replacement of the Cμ and C∂ constant regions with those of other isotypes. Initiated by AID, CSR is based on intrachromosomal recombination between DNA “switch” regions, one positioned upstream of Cμ and the other preceding one of the downstream C_H isotypes (9). This deletional event often occurs in both IgH loci and involves, in most cells, the same downstream C_H isotype (10, 11). CSR is triggered in antigen-engaged B cells in response to cytokine stimulation and crosslinking of coreceptors, such as CD40, resulting from the interaction with antigen-specific Th cells. Targeting of CSR to a specific C_H locus is preceded by the transcription from that locus of “preswitch,” spliced, noncoding polyadenylated RNAs (also called “germ-line” transcripts), in response to Th1- or Th2-type cytokines (9).

Despite extensive experimentation in the past, the nature of the signals that influence proliferation, differentiation, and selection of B cells within the GC are still largely unknown. Cre/loxP-mediated conditional gene targeting provides a unique experimental approach to tackle these biological questions in the context of the whole animal. The present study characterizes the Cγ1-cre knockin strain, a mouse mutant in which expression of Cre recombinase, induced upon onset of germ-line Cγ1 transcription, promotes conditional gene targeting in the majority of GC B cells generated in response to immunization with TD antigens. Driving Cre expression from the Cγ1 locus allowed us to track the fate of B cells undergoing sterile Cγ1 transcription and thus to investigate to what extent this process controls the targeting of CSR in vivo.

Results

Generation of Cγ1-cre Knockin Mice.

Cγ1-cre mice were generated from 129-derived ES cells in which an internal ribosome entry site (IRES) followed by the Cre-coding sequence was inserted into the 3′ region of the Cγ1 locus between the last membrane-coding exon and its polyadenylation sites (Fig. 1a). This approach allows for the expression from the Cγ1 locus of a bicistronic mRNA consisting of the Cγ1 and the Cre transcript, respectively. Results presented in this study were obtained from the analysis of Cγ1-cre animals on the C57BL/6J genetic background.

Fig. 1.

Knockin of the Cre transgene into the Cγ1 locus. (a) Strategy to insert the IRES-Cre cassette into the mouse Cγ1 locus. The targeted locus is depicted before and after removal by FLP recombinase of the frt-flanked selection marker. BamHI (B) sites within the targeted genomic region are indicated. (b) Southern blot analysis, using probe a, of BamHI-digested tail DNA from WT (lane 1) and Cγ1-cre heterozygous mice, before (lane 2) and after (lane 3) removal of the selection marker.

Induction of Cre-Mediated Recombination in IgG1⁺ Cγ1-cre B Cells in Vitro.

Purified splenic B cells from C57BL/6J controls and Cγ1-cre mice carrying a conditional rag-2 allele (12) as a reporter gene were stimulated in vitro with LPS and IL-4 to promote IgG1 CSR. B cells treated with LPS alone served as negative controls. The fraction and absolute number of IgG1⁺ B cells measured by flow cytometric analysis 4 days later was comparable between mutant and control cell cultures (data not shown). B cells from 4-day LPS and LPS plus IL-4 cultures were sorted based on surface IgM and IgG1 expression, respectively, and subjected to Southern blotting analysis. Southern blot quantification showed that 95% of IgG1-switched B cells had undergone Cre-mediated recombination in Cγ1-cre-bearing cells. In contrast, IgM⁺ B cells retained the loxP-flanked (“floxed”) rag-2 gene in >90% of the cells (Fig. 2a).

Fig. 2.

Cre-mediated recombination in B cell subsets of Cγ1-cre mice. (a) Status of the rag-2 allele. wt, wild type; fl, floxed; Δ, deleted) in IgM⁺ and IgG1⁺ B cells sorted from cultures of Cγ1-cre; Rag-2^fl/+ B cells. (b) Percentage of EYFP⁺ cells within gated splenic B cell subsets of Cγ1-cre/+; R26-EYFP mice. (c) Percentage of gated IgG1⁺EYFP⁺ within the GC B (IgG1^loCD38^lo; Upper) and memory B (IgG1⁺CD38⁺; Lower) cell subsets of Cγ1-cre/+; R26-EYFP and Cγ1-cre/+ control mice. EYFP gates were set based on controls. (d) Immunofluorescence analysis of SPL sections from day-10 SRBC-immunized Cγ1-cre/+; R26-EYFP (Upper) and R26-EYFP control animals (Lower) stained as indicated. Results shown in b–d are representative of at least four independent experiments. Experiments shown in a were repeated twice.

Cell-Type and Stage-Specific Expression of the Cγ1-cre Transgene.

To study the efficiency of Cre-mediated recombination in vivo at the single-cell level, Cγ1-cre mice were bred to the Rosa26 (R26)-EYFP reporter (R26-EYFP) strain (13). Flow cytometric analysis was performed to measure the fraction of EYFP⁺ cells within different B cell subsets isolated from spleen (SPL), lymph nodes (LNs), bone marrow (BM), and peritoneal cavity (PC) of the compound mutants. In BM, pro-B (IgM⁻B220^loCD43⁺) and pre-B (IgM⁻B220^loCD43⁻) cells lacked expression of the reporter gene. The same result was obtained when we analyzed transitional (IgM⁺B220^loAA41⁺) cells isolated from both BM and SPL (Fig. 2b and data not shown). Less than 2% of IgM⁺ follicular (CD23⁺CD21⁺CD38⁺) and marginal zone (CD23^loCD21^hiCD38^hi) B cells expressed the reporter gene (Fig. 2b) in unimmunized compound mutants. Similar fractions of EYFP-expressing cells were observed among non-GC B-2 cells (CD23⁺CD21⁺CD38⁺) in peripheral LNs (data not shown). Less than 1% of PC B-1 cells (B220^loCD19^hiCD43⁺IgM^hi) showed EYFP expression (data not shown). The largest fraction of EYFP-expressing B cells was within the population of PNA^hiFas^hiCD38^lo GC B cells in SPL and LN after i.p. immunization with TD antigens such as alum-precipitated (4-hydroxy-3-nitrophenyl) acetyl coupled to chicken gamma globulin (NP-CG) or sheep red blood cells (SRBC). Within the GC B cell fraction, 85–95% of these cells expressed the reporter gene 10–14 days after immunization (Fig. 2b). Furthermore, within the fraction of splenic IgG1⁺ B cells, 85–95% of the cells with a GC B cell phenotype (CD38^loIgG1^lo) and 70–80% of those identified as memory B cells (IgG1⁺CD38⁺) had undergone Cre-mediated recombination, respectively (Fig. 2c). Immunofluorescence analysis of splenic histological sections prepared from Cγ1-cre;R26-EYFP mice 10 days after SRBC immunization showed distinct areas expressing EYFP within CD21⁺ B cell follicles that were intensively stained with the GC marker peanut agglutinin (PNA; Fig. 2d). EYFP⁺ GC B cells expressed the proliferation marker Ki-67 and were in intimate contact with the follicular dendritic cell network expressing high levels of the complement receptor CD21 (Fig. 2d).

Efficiency of Cre-mediated recombination in GC B cells was further tested by breeding Cγ1-cre knockin animals to mice carrying conditional alleles for the death receptor fas (fas^fl) or the BCR signaling subunit Igβ (Igβ^fl-EGFP). In the latter case, Cre-mediated recombination is monitored by induction of EGFP expression from the Igβ locus (Fig. 6, which is published as supporting information on the PNAS web site). After immunization with NP-CG or SRBC, >85% of GC-B and IgG1⁺ memory B cells showed Cre-mediated recombination in both compound mutant strains (Fig. 7, which is published as supporting information on the PNAS web site, and data not shown).

Efficiency of Cre-mediated recombination population was only marginally improved in mice carrying two copies of the Cγ1-cre transgene (Fig. 8, which is published as supporting information on the PNAS web site). However, in these animals, the fraction of IgG1⁺ memory B cells as well as the IgG1 serum antibody titers were significantly reduced (Fig. 8 and data not shown), suggesting that the IRES-Cre cassette may interfere with expression of proper levels of IgG1 from the adjacent γ1 locus and thus the selection of GC B cells into the compartments of memory B and antibody-secreting cells.

Flow cytometric analyses failed to detect Cre-mediated recombination in peripheral CD4⁺ or CD8⁺ T cells, Gr1⁺ granulocytes, CD19⁻Mac1⁺ macrophages, CD11c⁺ dendritic cells, and NK1.1⁺ NK cells of Cγ1-cre mice (Fig. 9a, which is published as supporting information on the PNAS web site). Similar results were obtained when genomic DNA from kidney, brain, liver, and heart of Cγ1-cre mice were tested for Cre-mediated recombination by Southern blot or genomic PCR analyses (Fig. 9b). However, surprisingly, Cre-mediated recombination was observed in germ cells of 10–30% of Cγ1-cre mice older than 4 months (oocytes being more often subject to Cre-mediated recombination than spermatocytes), thus resulting in target gene deletion in the germ line of the corresponding offspring.

Early Onset of Cγ1-Cre Expression During the GC Reaction.

To determine the onset of Cre-mediated recombination during the GC reaction, Cγ1-cre;Igβ^fl-EGFP/+ mice were immunized with NP-CG and euthanized at different time points after immunization. Flow cytometric analysis revealed that, by 4 days after immunization, a significant fraction of GC B cells (25–50%) had already undergone Cre-mediated recombination. This fraction increased over time, reaching a maximum of 75–85% 12–14 days after immunization, after which it remained stable (Fig. 3a and data not shown).

Fig. 3.

Early induction of Cre-mediated recombination in GC responses of Cγ1-cre mice. (a) Percentage of EGFP⁺ cells among gated splenic PNA^hiFas^hi GC B cells at different time points after NP-CG immunization. Two representative Cγ1-cre; Igβ^fl-EGFP mice are shown. (b) NP- and allotype–specific serum IgG1 titers measured >28 days after NP-CG immunization of Cγ1-cre heterozygous (○) and CB6F1 (•) control mice; bars indicate geometric mean of IgG1 levels. (c) Percentage of boxed IgG1⁺ B cells in Cγ1-cre mice carrying either the B1-8i or B1-8f alleles. (d) Percentage of β-gal⁺ cells among gated B220⁺IgG1⁺ B cells of Cγ1-cre mice carrying the indicated B1-8 alleles. Two mice per group are shown. β-gal gates were defined based on control mice. Results shown are representative of at least three independent experiments

IgG1 Expression in Cγ1-cre B Cells Occurs Mainly from the Nontargeted IgH Allele.

To test whether the insertion of IRES-Cre cassette was detrimental for IgG1 expression from the same allele in vivo, we measured allele-specific IgG1 serum titers in heterozygous Cγ1-cre mice after immunization with NP-CG. This assay was possible because the Cγ1-cre transgene was originally inserted into an IgH locus of a allotype. The measurement of IgG1 serum titers >28 days after immunization revealed equivalent contribution of NP-specific IgG1a and -b antibody titers in C57BL/6J × BALB/c F₁ control mice. In contrast, heterozygous Cγ1-cre mice showed a significant and selective impairment in antigen-specific IgG1a titers deriving from the targeted Cγ1-locus (Fig. 3b). Thus, the impairment in the generation of IgG1-switched memory B and plasma cells in Cγ1-cre mice seems to be limited to those cells that express a functional IgH chain from the chromosome carrying the Cγ1-cre transgene.

The Cγ1-cre Transgene Promotes Efficient Cre-Mediated Recombination When Expressed from an IgH Locus Carrying D_HJ_H Rearrangements.

Because up to 90% of IgG1⁺ memory B cells undergo Cre-mediated recombination in heterozygous Cγ1-cre mice, we suspected that the majority of these cells expressed Cre from the nonfunctional IgH chromosome. To test this hypothesis, we bred Cγ1-cre transgenics to mice carrying a prerearranged V_H gene (B1-8), either flanked by loxP sites [B1-8f; (14)] or, as a control, lacking such sites [B1-8i; (15)]. A subset of the compound mutant mice also carried the R26-β-gal^fl-STOP reporter allele to further assess efficiency of Cre-mediated recombination. The majority of B cells in B1-8/Cγ-1cre double knockin mice express the prerearranged B1-8 IgH chain, and thus carry the second IgH chromosome in germ-line configuration or bearing a D_HJ_H joint, as a result of IgH allelic exclusion. After immunization with NP-CG, a normal sized compartment of IgG1⁺ memory B cells was found in B1-8i/Cγ1-cre; R26-β-gal^fl-STOP mice (Fig. 3c). Over 90% of these cells expressed β-gal, confirming efficient Cre-mediated recombination in memory B cells (Fig. 3d). In contrast, a significant reduction (>80%) in the fraction of surface IgG1⁺ B cells was observed in B1-8f/Cγ1-cre; R26-β-gal^fl-STOP mice as a consequence of efficient Cre-mediated deletion of the B1-8f allele (Fig. 3c). Furthermore, the remaining sIgG1⁺ B cells in these mice largely lacked expression of β-gal, possibly representing a selected pool of cells that had escaped Cre-mediated recombination to preserve BCR expression (Fig. 3d). Thus, Cγ1 and D_H/Iμ promoters, respectively, before and after completion of CSR to IgG1, can drive expression of sufficient Cre protein to induce efficient Cre-mediated recombination in Cγ1-cre B cells.

Identity of IgM⁺ B Cells Undergoing Cre-Mediated Recombination in Cγ1-cre Mice.

Flow cytometric analysis of SPL and LN cells from unimmunized Cγ1-cre; R26-EYFP mice identified <2% of IgM⁺ EYFP⁺ B cells with a pre-/post-GC phenotype (CD38⁺PNA^loFas^lo; Figs. 2b and 4a and data not shown). RT-PCR revealed expression of germ-line Iμ-Cμ and Iγ1-Cγ1 transcripts in these cells (Fig. 4b) (16), whereas circle Iγ1-Cμ switch transcripts, indicative of recent class-switching to IgG1 (17), were missing (Fig. 4b, lane 2). The latter transcripts, together with Iγ1-Cγ1 transcripts, were readily detected in sorted EYFP⁺CD38^lo GC B cells (Fig. 4b, lane 3), indicating that IgG1 CSR was completed in a subset of these cells (Fig. 4b). Onset of Cγ1 transcription in EYFP⁺CD38⁺PNA^loFas^lo non-GC B cells was associated with induction of AID expression (Fig. 4b, lane 2). Aid-specific transcripts were also detected in EYFP⁺ GC B cells (Fig. 4b, lane 3) whereas they were barely detectable in IgM⁺CD38⁺ EYFP⁻ B cells (Fig. 4b, lane 1).

Fig. 4.

Germ-line Cγ1 transcription induces Cre-mediated recombination in resting Cγ1-cre B cells. (a) EYFP⁺IgM⁺ B cells in Cγ1-cre; R26-EYFP mice are divided into two subsets according to Fas expression. (b) RT-PCR analysis to detect IgH germ-line (GLT), circle (CT)-switched and AID-specific transcripts in IgM⁺EYFP⁻ (lane 1), IgM⁺EYFP⁺Fas^lo (lane 2), and IgM^loEYFP⁺Fas^hi (lane 3) sorted B cells. cDNA from IgG1⁺ A20 B lymphoma cells (lane 4) and IgM⁺ c-MYC transformed pre-GC mouse B lymphoma cells were included as controls; arrows indicate specific PCR fragments. (c) Percentage of EGFP⁺ Cγ1-cre; Igβ^fl-EGFP/+ and Igβ^fl-EGFP/+ control B cells throughout a 3-day culture period in the presence of the indicated amounts of IL-4. Results shown are representative of at least three independent experiments

In humans, a fraction of IgM⁺ B cells consists of somatically mutated memory B cells (18). Thus, we tested whether the IgM⁺EYFP⁺ B cell population in Cγ1-cre; R26-EYFP mice included GC-derived memory B cells that had failed to successfully switch to IgG1. Sorted IgM⁺EYFP⁺CD38⁺ non-GC B cells were analyzed by genomic PCR for the occurrence of somatic mutations in the intronic region downstream of J_H4. Around 60% of the analyzed sequences were devoid of mutations. Most of the remaining ones carried just a few mutations (Table 1). In contrast, ≈70% of sequences amplified from sorted EYFP⁺ GC B cells were highly mutated (Table 1). Thus, IgM⁺ B cells that undergo Cre-mediated recombination in Cγ1-cre mice likely include, and possibly largely represent, a population of GC B cell founders expressing germ-line Cγ1 and Aid transcripts in response to Th2-type cytokines.

Table 1.

IgH somatic mutation analysis in EYFP⁺ B cells of Cγ1-cre; R26-EYFP mice

EYFP+ B cells	Unique rearrangements	Mutated sequences	Mutation range	Mutation frequency*
IgM+ CD38+	32	9	1–5	0.30
IgMlo CD38lo	15	12	1–18	0.78

*Calculated dividing the number of unique point mutations by the total number of sequenced nucleotides from PCR products containing point mutations.

IL-4 Promotes Cre-Mediated Recombination in Resting Cγ1-cre B Cells.

Because IL-4 induces expression of Cγ1 and Cε sterile transcripts (16, 19), we tested whether it was by itself sufficient to induce Cre-mediated recombination in Cγ1-cre resting B cells. Purified splenic B cells from Cγ1-cre; Igβ^fl-EGFP were cultured for 3 days in the presence of different concentrations of IL-4. After 48 h, Cre-induced EGFP expression was detectable in the IL-4-stimulated cultures, and the percentage of EGFP⁺ cells doubled over the following day in the absence of detectable cell proliferation, suggesting an increase in the fraction of B cells undergoing transcription from the Cγ1 promoter (Fig. 4c). Thus, Cγ1-germ-line transcription is sufficient to drive Cre-mediated recombination in Cγ1-cre B cells in the absence of stimulation through the BCR and/or coreceptors.

Cγ1-cre B Cells Undergoing Cre-Mediated Recombination Switch to Multiple Ig Isotypes in Vivo.

The high degree of Cre-mediated recombination (>85%) observed in GC B cells of Cγ1-cre mice suggested that Cre expression could have also occurred in cells switching to other isotypes (20). To test this hypothesis, we sorted CD38^loFas^hi GC B cells from Cγ1-cre; R26-EYFP mice based on EYFP expression. RT-PCR analysis indicated that, within this population, a fraction of cells had indeed switched to IgG3, IgG2a, and IgG2b, respectively, as revealed by the detection of specific IgH class-switched transcripts (Fig. 5a). Furthermore, flow cytometric analysis of Peyer’s patches cells from Cγ1-cre; R26-EYFP mice showed that up to 20% of IgA⁺ GC B cells, spontaneously arising in these sites, also expressed the reporter gene (Fig. 5b). Finally, immunofluorescence analysis of splenic sections of Cγ1-cre; R26-EYFP mice immunized 10 days earlier with SRBC identified a distinct subset of extra-follicular EYFP⁺ plasma cells, many of which expressed IgG2a and IgG2b antibodies (Fig. 5c).

Fig. 5.

Cγ1-cre B cells undergoing Cre-mediated recombination in vivo express multiple Ig isotypes. (a) Ig switch transcripts in sorted MLN GC B cells from two Cγ1-cre; R26-EYFP mice detected by RT-PCR. 2470.1, c-MYC-transformed, IgM⁺ primary B lymphoma cells and IgG1⁺ A20 mouse B lymphoma cells were included as controls. (b) EYFP expression in Peyer’s patches Fas^hiIgA⁺ GC B cells of Cγ1-cre; R26-EYFP mice. (c) Immunofluorescence analysis of SPL sections from Cγ1-cre; R26-EYFP mice 10 days after SRBC immunization, stained for EYFP (red) and each one of the indicated IgH isotypes (green). Arrows indicate coexpressing cells. (d) Percentage of EGFP⁺ cells 4 days after treating Cγ1-cre; Igβ^fl-EGFP/+ B cells with different stimuli promoting CSR. (e) Percentage of EGFP⁺ cells among isotype-switched B cells in Cγ1-cre; Igβ^fl-EGFP/+ mice (continuous black line). Stimulated Igβ^fl-EGFP/+ B cells were used as negative controls for EYFP expression (dotted line). Results shown in a–d are representative of at least two independent experiments.

To assess whether Cre-mediated recombination in non-IgG1 switched B cells resulted from basal transcription initiated from the Cγ1 promoter, purified Cγ1-cre; Igβ^fl-EGFP/+ B cells were induced to switch in vitro to specific IgH isotypes. Flow cytometric analysis of day-4 cultures indicated that >85% of B cells stimulated with CD40 plus IL-4 (or LPS plus IL-4) had undergone Cre-mediated recombination, regardless of whether they became IgG1-expressing cells (Fig. 5 d and e). In contrast, B cell stimulation with cytokines promoting CSR, respectively, to IgG3, IgG2a, and IgA failed to induce Cre-mediated recombination in the switched cells (Fig. 5 d and e). Thus, basal germ-line Cγ1 transcription does not account for Cre-mediated recombination observed in vivo in Cγ1-cre B cells expressing isotypes other than IgG1.

Cγ1-cre ES Cells Allowing the Rapid Production of Compound Mutant Mice Through Tetraploid Blastocyst Complementation.

To expedite the analysis of gene function in GC B cells, we established several independent ES cell lines from Cγ1-cre heterozygous mice, on the F₁ (C57BL/6J × BALB/c) genetic background. Such ES cells can be used for further rounds of genetic manipulation and ultimately injected into tetraploid blastocysts to obtain cloned mice, ready for analysis. One of the Cγ1-cre F₁ ES cell lines was targeted at the R26 locus with a conditional transgene encoding a constitutive active form of the signal transducer and activator of transcription (STAT)-3 protein (STAT-3c) (21) followed by an IRES-EGFP cassette (Fig. 10, which is published as supporting information on the PNAS web site). Correctly targeted Cγ1-cre; STAT-3c^fl-STOP ES cells were used to generate 18 cloned mice from two sessions of ES microinjection into tetraploid blastocysts. Upon SRBC immunization, induction of STAT-3c expression was readily detected in the majority of GC and IgG1 memory B cells of cloned mice by means of EGFP expression (Fig. 9 and data not shown).

Discussion

Germ-Line Cγ1 Transcription Induces Efficient Cre-Mediated Recombination in GC B Cells of Cγ1-cre Mice.

In Cγ1-cre mice, the insertion of the Cre coding sequence into the Cγ1 locus, leading to its expression from a bicistronic mRNA together with the Cγ1 transcript, induced efficient Cre-mediated recombination in GC B cells, in accordance with previous reports identifying the GC as the major site of IgG1 CSR (22). Within the GC B cell population, the fraction of cells undergoing Cre-mediated recombination increased over time, reaching an average of >85% at the peak of the GC reaction, when most cells carried high loads of Ig somatic mutations.

Approximately 25–50% of GC B cells in these animals were subject to Cre-mediated recombination as early as 4 days after immunization, suggesting that transcription from the Cγ1 locus (and thus Cγ1-cre expression) represents an early event in activated B cells driven into a TD antibody response, in accordance with previous reports (4, 23). Several pieces of evidence suggest that germ-line Cγ1 transcription was sufficient to promote efficient Cre-mediated recombination. First, recombination mediated by the Cγ1-cre transgene, using the Igβ^fl-EGFP reporter allele, was induced in vitro in resting, IgM⁺ B cells upon stimulation with IL-4, a specific inducer of sterile Cγ1 transcripts. Second, B1-8i/Cγ1-cre mice showed Cre-mediated recombination in >90% of GC B cells, although most of these cells expressed the B1-8H chain and thus had the second IgH allele (carrying the Cγ1-cre transgene) in germ-line configuration or at the most bearing D_HJ_H rearrangements. This result indicates that, in GC B cells of these mice, Cre is expressed from germ-line Cγ1 transcripts derived from the nonfunctional IgH allele (24). Third, in unimmunized Cγ1-cre mice carrying the EYFP reporter allele, we observed ≈2% of IgM⁺ EYFP⁺ B cells, lacking GC B cell markers. RT-PCR analysis showed expression in these cells of germ-line Cγ1 transcripts, whereas circular Iγ1-Cμ switch transcripts were not detected, indicating that transcription at the Cγ1 locus had just been initiated.

As descendents of GC B cells, up to 90% of IgG1⁺ memory B cells in Cγ1-cre mice had also undergone Cre-mediated recombination. These cells expressed IgG1 preferentially from the nontargeted locus. By using the Cγ1-cre system in combination with a reporter gene, IgG1⁺ memory B cells can be easily tracked in vivo, allowing one to monitor the homing and activation properties of these cells in steady-state conditions as well as during primary and secondary TD antibody responses.

To accelerate the analysis of gene function in GC B cells, we established Cγ1-cre F₁ ES cells allowing further rounds of genetic manipulations and ultimately the cloning of compound mutants by tetraploid blastocyst complementation. Through this approach, a large cohort of cloned Cγ1-cre; STAT-3c^fl-STOP compound mutants was readily obtained, allowing the rapid analysis of GC responses in these mice.

Transcription at Multiple C_H Loci Precedes CSR in Individual GC B Cells.

Expression of germ-line transcripts is an essential step preceding Ig CSR (16, 25). These transcripts, expressed from C_H promoters, are induced in response to stimulation by Th1- or Th2-type cytokines (9). In vitro induction of CSR in Cγ1-cre B cells followed the rules outlined above, as efficient Cre-mediated recombination was observed in cultures stimulated with CD40 (or LPS) plus IL-4, predominantly switching to IgG1. Over 85% of the B cells underwent Cre-mediated recombination under these conditions regardless of whether they became IgG1-expressing cells, further confirming the role of sterile Cγ1 transcription in driving Cre-mediated recombination. A subset of CD40 plus IL-4-stimulated B cells switched to IgE expression, most of which had undergone Cre-mediated recombination. This result suggests that induction of Cre expression occurred either in activated IgM⁺ B cells expressing both Cγ1 and Cε germ-line transcripts or in IgG1⁺ B cells, which subsequently switched to IgE (26, 27). Conversely, culture conditions promoting switching to IgG3, IgG2a, IgG2b, and IgA failed to induce Cre-mediated recombination by the Cγ1-cre transgene. However, germ-line transcription of Cγ1 did not always result in IgG1 or IgE CSR in vivo. Molecular, immunofluorescence, and flow cytometric analyses identified a subset of IgG3-, IgG2a-, IgG2b-, or IgA-switched GC B or extrafollicular plasma cells in Cγ1-cre mice that had undergone Cre-mediated recombination. This result suggests a scenario where at early stages of the GC reaction, possibly during the initial T cell–B cell interactions (28), both Th1- and Th2-type cytokines are secreted by antigen-specific and/or by-stander Th cells. This process then activates the simultaneous or sequential expression of germ-line transcripts from multiple C_H loci in individual GC B cells or their precursors. Ultimately, the concentration of individual cytokines determines the extent of germ-line transcription from the individual C_H loci and thus the frequency of switching to a particular isotype. In accordance with this scenario, <30% of Cγ1-cre B cells underwent Cre-mediated recombination in spontaneous GCs of gut-associated lymphoid tissues where TGF-β represents the predominant cytokine, promoting switching to IgA (29). Cre-mediated recombination in Cγ1-cre B cells switching to isotypes other than IgG1 may also result from CSR involving different C_H genes on the two IgH chromosomes (30) or sequential switching from Cγ1 to distal C_H isotypes (31), except in the case of cells expressing IgG3. The latter mechanism may contribute to the generation of IgA⁺ B cells because the predominant IgH allotype expressed by IgG1⁺ B cells in Cγ1-cre mice was specifically overrepresented within the fraction of IgA⁺ cells that had undergone Cre-mediated recombination (data not shown).

Methods

Generation of Cγ1-cre Mice.

The Cγ1-cre vector is a derivative of a plasmid previously used to target the mouse Cγ1 locus (32). The encephalomyocarditis (EMCV)-derived IRES DNA segment (gift from E. Wimmer) followed by the NLS-Cre coding sequence was inserted downstream of the termination codon in the last membrane-coding exon of IgG1. The targeting vector was electroporated into 129Sv-derived IB10 ES cells (gift from A. Berns). ES clones were screened by Southern blot analysis by using an external probe as described (32). Two independent targeted ES clones were injected into blastocysts to generate chimeric mice from which germ-line-transmitted animals were obtained. Removal of the frt-flanked selection marker was achieved breeding Cγ1-cre (NeoR⁺) mice to FLPe deleter mice.

Other Mouse Reagents.

Rag-2^fl, fas^fl, R26-β-gal^fl-STOP, R26-EYFP, and FLPe mice have been described (12, 13, 33–35). Igβ^fl-EGFP and R26-STAT3c^fl-STOP conditional mice are described in Supporting Text, which is published as supporting information on the PNAS web site. Mice were bred and maintained under specific pathogen-free conditions; mouse protocols were approved by the University of Cologne and the Harvard University Institutional Animal Care and Use Committee and by the CBR Institute for Biomedical Research.

Mouse Immunizations.

Eight- to 12-week-old mice were immunized i.p. with 100 μg of alum-precipitated (4-hydroxy-3-nitrophenyl) acetyl coupled to chicken gamma globulin (Biosearch) or with 1 × 10⁸ SRBC (Cedarlane Laboratories) in PBS.

Flow Cytometry and Cell Sorting.

Cell suspensions were stained with biotin or fluorescent-labeled mAbs as described (36). Additional antibodies included anti-IgG1 (X56), -IgG3 (R40-82), -IgE (R35-118), -IgG2a (R19-15) (BD Pharmingen), -IgA (Southern Biotechnology Associates), -AA4.1, and -CD38 (90) (eBioscience, San Diego). Biotin conjugates were visualized with PerCP or APC-streptavidin (BD Pharmingen). Flow cytometric determination of β-gal enzymatic activity was performed as described (37). All analyses were performed on a FACSCalibur (BD Biosciences), and results were analyzed with cellquest or flowjow software. Cell sorting was performed on a FACS Vantage or FACS Aria (BD Biosciences).

Immunohistochemistry and Immunofluorescence.

Paraffin-embedded sections were stained for hematoxylin/eosin (H&E) or immunostained as described (38). Briefly, antigen-retrieved, peroxidase-inhibited sections were blocked in 5% defatted milk powder, incubated overnight with a rabbit anti-EGFP/EYFP peptide Ab (BD Pharmingen) or rabbit Ig (Sigma), washed, counterstained with peroxidase-conjugated anti-rabbit polymer (DakoCytomation), washed, and tyramide amplified in Cy3 (PerkinElmer). Subsequently, peroxidase was quenched again, and biotin-conjugated anti-Ig isotype-specific antibodies (goat anti-IgG1, -IgG2a, -IgG2b, -IgM; Southern Biotechnology Associates) or biotin-PNA (Vector Laboratories) were added, washed, incubated with HRP-conjugated Avidin (DakoCytomation), washed, and tyramide-amplified in FITC. An FITC-conjugated goat anti-mouse CD21 (sc-7028; Santa Cruz Biotechnology) was counterstained with a peroxidase-conjugated goat anti-FITC (Roche Diagnostics), washed, and amplified with tyramide FITC. Slides, mounted with DAPI-containing mounting medium (Molecular Probes), were viewed on an E600-Nikon Microscope (Nikon). Images were edited for optimal color contrast with Adobe photoshop 7 and Adobe illustrator 10 (Adobe Systems, San Jose, CA).

In Vitro B Cell Cultures.

Splenic B cells purified by immunomagnetic depletion of CD43⁺ cells (Miltenyi Biotec, Auburn, CA) were cultured in B cell medium in the presence of stimuli including LPS (20 μg/ml, Escherichia coli, 055:B5; Sigma), anti-CD40 (3 μg/ml; IC10; eBioscience), IL-4 (100–200 units/ml; Preprotech), IFN-γ (100 ng/ml), and TGF-β (1 ng/ml; R & D Systems).

RT-PCR Analysis.

Total RNA was extracted from sorted cells by using the TRIzol reagent (Invitrogen). RNA from 1 × 10⁵ cell equivalents was subject to first strand cDNA synthesis by using random hexamers and Superscript RT II (Invitrogen). cDNA from 10⁴ cell equivalents was amplified by PCR. Intron-spanning primers and PCR conditions to amplify respectively Hprt, AID, circle, and isotype-switched IgH transcripts have been described (17, 36). PCR products were cloned into pGEM-T Easy (Promega) and sequenced.

IgH Somatic Mutation Analysis.

IgH V gene rearrangements from sorted B cells were PCR-amplified by using the Expand High fidelity PCR system (Roche Molecular Diagnostics) and primers as described (39). PCR-amplified IgH rearrangements carrying a J_H4 segment were cloned and sequenced.

Establishment of Cγ1-cre F₁ ES Cell Lines and Generation of Cloned Mice by Tetraploid Embryo Complementation.

Day 3.5 Cγ1-cre/+ (C57BL/6J × BALB/c) F₁ embryos were isolated and plated onto mitomycin-treated mouse embryonic fibroblasts, in ES cell medium (40) containing 1,000 units/ml leukocyte inhibiting factor, and 50 μM MEK1 inhibitor PD98059 (Cell Signaling Technology, Beverly, MA). After two successive rounds of trypsinization, cells were transferred onto a six-well plate, at which time point individual ES colonies were identified, isolated, and further expanded. ES clones displaying <30% of differentiated cells were subjected to a karyotype analysis and genotyped by PCR for the presence of the Y chromosome (41). Generation of cloned mice from one of the established Cγ1-cre/+ XY ES cell lines was performed as described (42).

Abbreviations

GC

germinal center

CSR

class switch recombination

AID

activation-induced cytidine deaminase

R26

Rosa 26

Th

T helper

IRES

internal ribosome entry site

SPL

spleen

LN

lymph node

SRBC

sheep red blood cell

PNA

peanut agglutinin

STAT

signal transducer and activator of transcription

NP-CG

(4-hydroxy-3-nitrophenyl) acetyl coupled to chicken gamma globulin

BCR

B cell receptor

TD

T cell-dependent.