B cells are activated for switch recombination by signals from Th cells, but the site at which this first occurs in vivo has yet to be identified. By in situ hybridization of splenic sections using riboprobes specific for the Iγ1 switch transcript and Rad51 mRNA, we have visualized B cells that are newly activated for switch recombination and characterized the spatial and temporal patterns of Iγ1 and Rad51 mRNA expression. Within 2 days after immunization with (4-hydroxy-3-nitrophenyl)acetyl-chicken gamma-globulin, expression of Iγ1 switch transcripts and Rad51 mRNA was evident and was localized to B220+ B cells clustered within the T cell-rich periarteriolar lymphoid sheath (PALS) and surrounding follicles. By Ab staining, we have shown previously that cells switching from IgM to IgG expression can be visualized at 3 to 5 days postimmunization and colocalize to clusters of Rad51+ cells. Hybridization of adjacent sections with probes for Cμ and Cγ1 mRNA now shows that switching from μ to γ expression occurs within Rad51+Iγ1+ regions of the PALS and peaks between days 3 and 5. Colocalized expression of Iγ1 and Rad51 transcripts was observed from days 2 through 12 of the immune response. Iγ1 and Rad51 transcripts were down-regulated but still detectable at 12 days postimmunization, when they were evident in peanut agglutinin-positive germinal center B cells. Taken together, these observations show that B cells are first activated for switch recombination in the T cell-rich PALS.
Isotype switching in B lymphocytes is a regulated DNA recombination event that is initiated by signals from T cells following Ag activation (reviewed in Refs. 1–8). Switching joins an expressed heavy chain V (VDJ) region to a new downstream constant (CH) region and deletes the intervening DNA in the form of a circle (9, 10, 11). Recombination involves switch (S) regions, which are 2 to 10 kb of G-rich sequences located upstream of each CH region that participates in switching. Unlike other targeted and regulated recombination processes, switch recombination does not appear to depend upon either sequence specificity or on homology, making it of particular interest to understand the regulation and mechanism of switching.
Switch recombination in vivo occurs within the secondary lymphoid tissues, such as the spleen and lymph nodes. These tissues are comprised of red pulp, which contains large numbers of E, and white pulp, which contains mostly lymphocytes. In a naive animal, the histology of these tissues is relatively simple. The white pulp is sequestered into numerous follicles, which contain a network of follicular dendritic cells filled with surface IgM-expressing, recirculating B cells. Within each follicle is a central arteriole surrounded by a T cell-rich zone known as the periarteriolar lymphoid sheath (PALS)4 (12, 13). Nonrecirculating surface IgM+ B cells populate the marginal zones that surround the follicle, dividing white pulp from red. The histologic appearance of the spleen or lymph node changes dramatically when an animal is challenged by Ag. B blasts appear within the follicular dendritic cell network and proliferate rapidly, filling the follicle and displacing the recirculating B cells to the periphery, where they form the follicular mantle. Shortly afterward, the follicle becomes polarized, and the classical appearance of the germinal center (GC) becomes apparent. The rapidly dividing blast cells (centroblasts) cluster in the region adjacent to the PALS that is termed the dark zone. The nonproliferating progeny of these cells (centrocytes) fill the dendritic network and form the light zone of the GC.
T and B lymphocytes collaborate in the response to most protein Ags (reviewed in 14 ; following primary immunization, T cell-B cell interactions are first observed in the PALS, where B cells take up Ag, process it by fragmenting it into peptides, and present the resulting peptides to Th cells via class II molecules of the MHC (15). Recognition of the peptide-MHC complex by the TCR and its CD4 coreceptor then triggers the B cell response, which further depends upon functional interactions between CD40 molecules constitutively expressed by B cells and CD40 ligand expressed by activated Th cells (14, 16, 17, 18). B cell activation results in a burst of cell proliferation and the differentiation of many B lymphocytes into plasma cells secreting low-affinity Abs (19, 20, 21). It is selected progeny of these newly expanded T and B cell populations that are recruited from the PALS to the adjacent B cell-rich follicles, where they become founders of GCs, the specialized microenvironments in which somatic hypermutation occurs (Refs. 13, 20, 22, and 23; reviewed in Refs. 24 and 25).
Although the changes that occur in the spleen and lymph node during the primary response to Ag have been well characterized histologically, the site within these tissues at which B cells first become activated for switch recombination has not been defined. Previous reports using several different approaches have suggested the importance of the T cell-rich extrafollicular zones (19, 26, 27, 28). It has also been proposed that GCs provide the primary sites at which switch recombination occurs (22, 29, 30, 31).
Analysis of murine B cells that have been induced to switch in vitro by culture with LPS has shown that switching to a given CH region is preceded by production of a noncoding switch transcript from both regions targeted for recombination (reviewed in 8 . The switch transcript initiates at the IH (initiate) exon promoter upstream of the targeted CH region and proceeds through the IH exon, S region, and CH region (see Figure 1). Splicing of the primary transcript removes the repetitive S region sequences and results in the joining of the IH exon and CH exons. Analysis of mice carrying targeted deletions has shown that recombination to a given isotype depends upon switch transcription and splicing of the switch transcript (32, 33, 34, 35, 36, 37, 38). Extracellular signals delivered by cytokines and lymphokines target switch recombination to specific isotypes by inducing transcription at the corresponding IH exon promoter. The lymphokine IL-4, which is a potent activator of IgG1 and IgE expression, induces synthesis of Iγ1 switch transcripts in primary resting B cells cultured in vitro in the presence or absence of LPS (39, 40, 41, 42, 43, 44). Induction of switch transcripts is rapid and begins within 4 h of culture with IL-4 (41). As the production of a specific Iγ transcript is a prerequisite to recombination, Iγ expression should identify cells activated for switching in vivo.
In previous experiments, we have identified Rad51 as one early marker for switching B cells (45, 46). Rad51 is the highly conserved eukaryotic homologue of the prokaryotic recombination protein, RecA (47, 48). Mammalian Rad51 protein mediates homology recognition, initiates strand exchange reactions, and enhances the efficiency of ligation reactions in vitro (49, 50). We showed that induction of Rad51 protein expression correlates with switch recombination in vitro and in vivo. Immunofluorescence microscopy of LPS-activated, cultured primary B cells revealed large, brightly staining Rad51+ foci localized to the nucleus. Moreover, double-staining with anti-Rad51 and anti-IgM or anti-IgG Abs showed that the presence of Rad51+ foci correlated with the switch from IgM to IgG expression, and that foci were extinguished subsequent to switching. No Rad51 expression was evident in splenic sections from unimmunized mice. Splenic sections taken from days 2 to 7 postimmunization were shown to contain clusters of 200 to 600 cells that stained brightly with anti-Rad51 Abs. At early timepoints (days 2–3) after immunization, these clusters colocalized with areas of IgM expression, as visualized by staining with anti-IgM Abs. By day 5 postimmunization, IgM expression diminished; areas of Rad51 expression colocalized with cells that had completed switch recombination and expressed IgG. We further showed that the foci induced in primary B cells by LPS activation are distinct in size and number from the foci induced by agents that damage DNA (46). These observations not only correlated Rad51 expression with switch recombination but also demonstrated that the progression from IgM to IgG expression is accompanied by induction and then down-regulation of Rad51 protein. In the experiments that we conducted previously (45), we did not attempt to localize Rad51 expression or switching B cells to specific microenvironments within the spleen. If increased transcription of the Rad51 gene parallels the dramatic induction of Rad51 protein, in situ hybridization to Rad51 mRNA should provide another early marker that can be used in combination with Iγ transcripts to identify switching B cells.
We have now defined the spatial and temporal patterns of Iγ1 switch transcript and Rad51 mRNA expression in newly activated splenic B cells, using in situ hybridization to observe mRNA expression in splenic sections of C57BL/6 mice immunized with (4-hydroxy-3-nitrophenyl)acetyl (NP)-chicken gamma-globulin (CGG). By analysis of serial sections with several different probes, we have shown that cells expressing Iγ1 and Rad51 transcripts colocalize within the T cell-rich PALS. We have also correlated Iγ1 and Rad51 expression with the switch from Cμ to Cγ1, as assayed both by in situ hybridization, as reported here, and by staining with IgM- and IgG-specific Abs, as we reported previously (45). Therefore, our results show that B cells are first activated for switch recombination in the T cell-rich PALS.
We further report that cells expressing switch transcripts are still visible at late timepoints following primary immunization, when they are localized to GCs. Others have identified switch transcripts in GC B cells using different experimental approaches (28, 51, 52), and GC may consequently be a secondary site of switch recombination. A low level of Rad51 mRNA expression is evident in GC cells. We have reported previously that the peak of Rad51 staining of splenic sections occurs at about day 5, well before GC formation, and that GCs are not stained with Rad51 Abs (45). Taken together with our previous results, our current data suggest that expression of Rad51 protein is specifically down-regulated, or that its distribution is relocalized before GC development and somatic hypermutation.
Materials and Methods
Animals, immunization, and preparation of splenic sections
Female C57BL/6 mice (6 to 12 wk old) were immunized by i.p. injection of 100 μg alum-precipitated NP-CGG together with 2 × 109 heat-killed Bordetella pertussis cells (53). Mice were sacrificed on days 0, 2, 3, 5, 7, 9, and 12 following immunization; their spleens were removed, embedded in Tissue-Tek OCT (Baxter Scientific, Boston, MA), frozen on dry ice, and stored at −70°C. Cryostat sections (10 μm) were mounted onto ProbeOn Plus slides (Fisher Scientific, Pittsburgh, PA), fixed in 3% paraformaldehyde for 60 min, and dehydrated by sequential rinses (5 min each) in water and in 30%, 75%, and 95% ethanol. The air-dried sections were stored at −70°C until use.
Immunocytochemical staining of sections
Sections were rehydrated for 10 min in PBS, permeabilized by incubation in PBS containing 0.3% Triton X-100 for 30 min, blocked for 60 min in a blocking solution of PBS containing 10% goat serum (Life Technologies, Gaithersburg, MD), and incubated for 60 min with 100 μl of either the appropriate Abs or biotinylated peanut agglutinin (PNA) (Pierce, Rockford, IL) diluted 1/100 in blocking solution. All sections were rinsed for 10 min in fresh blocking solution.
PNA-staining, biotinylated anti-B220/CD45R Abs (PharMingen, San Diego, CA) and biotinylated anti-Thy-1.2 Abs (PharMingen) were detected using streptavidin-bound horseradish peroxidase. B220/CD45R is a B cell-specific form of the T200 glycoprotein family; the Thy-1 glycoprotein is expressed by Th cells located outside GCs. Sections were incubated for 60 min with 100 μl of streptavidin-bound peroxidase, prepared by incubating 20 μg/ml of streptavidin (Pierce) and 40 μg/ml of biotinylated horseradish peroxidase (Pierce) for 30 min in blocking solution. Following a 10-min rinse in PBS, sections were incubated with 1 ml of 3,3′-diaminobenzidine (DAB) substrate solution (Sigma, St. Louis, MO; 0.8 mg/ml DAB, 0.009% H2O2, and 100 mM Tris-HCl (pH 7.5)). Alkaline phosphatase-conjugated goat anti-mouse Abs specific for IgM, IgG1, and λ light chain (Southern Biotechnology Associates, Birmingham, AL) were detected by incubation with 1 ml of 4-nitroblue tetrazolium chloride (NBTC)/5-bromo-4-chloro-3-indolyl-phosphate solution (BCIP) (Boehringer Mannheim, Philadelphia, PA). Appropriate color development was visualized in <60 min, and the reactions were stopped by rinsing the sections in distilled water (DAB) or TE buffer (NBTC/BCIP) (TE is 10 mM Tris and 1 mM EDTA (pH 7.9)). Sections were air-dried and mounted in Aquamount (Fisher Scientific). Rabbit polyclonal anti-human Rad51 Abs (a generous gift of Dr. C. M. Radding, Yale University School of Medicine) were detected by a 60-min incubation with 100 μl of Texas red-conjugated goat anti-rabbit IgG Abs diluted 1/100 in blocking solution. FITC-conjugated goat anti-mouse IgM or IgG Abs were incubated with the relevant sections in the same way. Sections were rinsed for 10 min in PBS and 5 min in distilled water and left to air dry before mounting in antifade.
A probe specific for the murine λ light chain was generated using pMCλ linearized with either SpeI or NcoI as template plasmid. pMCλ, cloned in the pGEM-T vector (Promega, Madison, WI), carries a 432-bp region encompassing the 3′ portion of the λ leader sequence, Vλ, Jλ, and the 5′ portion of Cλ, which is produced by RT-PCR amplification using total RNA from hybridoma B1-8 as template (54) as well as primers 5′-TTCACTTATACTCTCTCTCCTGGCTCTC (λ leader sequence, codons −16 to −7) and 5′-GAGCTCTTCAGAGGAAGGTGGAAACA (Cλ sequence, codons 128–121). A probe specific for the murine Cμ C region was generated using plasmid pBPC3HP600 linearized with either HindIII or XbaI as template. pBPC3HP600, cloned in the pBluescribe vector (Stratagene, La Jolla, CA), contains a 600-bp HindIII/PstI fragment including a Cμ genomic sequence from exons 3 and 4. A probe specific for the murine Cγ C regions was generated using plasmid pSB650 linearized with HindIII as template. pSB650, cloned in the pSP65 vector (Promega), contains a 650-bp BamHI/SacI fragment of Cγ2b including sequences from the CH1 through CH3 domains that are highly conserved within the IgG subfamily. Most Abs produced in the C57BL/6 response to NP-CGG are IgG1 (53), and their transcripts are efficiently visualized by hybridization with this probe. A probe specific for the Iγ1 region was generated using plasmid pMCIγ1 linearized with either SpeI or NcoI as template. pMCIγ1, cloned in the pGEM-T vector, carries a 361-bp insert obtained by PCR using as template a 442-bp PCR product comprising a portion of the mouse Iγ1 exon spliced to a portion of the Cγ1 sequence (55) (kindly provided by Dr. W. Dunnick, University of Michigan, Ann Arbor, MI) and the Iγ1 region primers 5′-GACGGCTGCTTTCACAGCTT (nucleotides 35–54) and 5′-CTCTCAACCTGTAGTCCATGC (nucleotides 395–375). The entire nucleotide sequence of this Iγ1 probe is contained within the sequences of Iγ1 probes A and B that were used previously by Turaga et al. (56) and shown to be highly specific. A probe specific for Rad51 was generated using the plasmid pEG986 linearized with NotI or SpeI as template. pEG986 contains the entire 1.1 kb-mouse Rad51 cDNA (47) in the pCRII vector (Invitrogen, San Diego, CA) and was kindly provided by Drs. E. Golub and C. M. Radding (Yale University). In vitro transcription reactions to generate digoxigenin-labeled sense and antisense riboprobes were conducted as described in the Dig RNA-labeling kit using SP6, T7, or T3 RNA polymerases (Boehringer Mannheim).
In situ hybridization
In situ hybridization studies were performed using a modification of the method of Yang et al. (21); all solutions were prepared with water treated with diethylpyrocarbonate (Sigma). Frozen sections were thawed; rehydrated for 10 min in PBS; placed in 0.2 N HCl for 20 min; washed in water; and digested with 2 μg/ml of proteinase K (Boehringer Mannheim), 2 mM CaCl2, and 10 mM Tris-HCl (pH 7.4) at 37°C for 10 min. Sections were postfixed in 3% paraformaldehyde for 5 min; rinsed twice in 2× SSC; acetylated for 10 min in 0.1 M triethanolamine-HCl (pH 8.0) containing 0.25% acetic anhydride; rinsed in 2× SSC; incubated in 0.1 M Tris-HCl and 0.1 M glycine (pH 7.0) for 30 min; washed twice in 2× SSC; dehydrated via sequential washes in 70%, 80%, 90%, and 100% ethanol; and air-dried. Sections were prehybridized for 2 h with a solution containing 50% deionized formamide, 4× SSC, 50 mM sodium phosphate buffer (pH 6.5), 0.1% SDS, 1% glycine, 1× Denhardt’s solution, 200 μg/ml of yeast transfer RNA, and 500 μg/ml of heat-denatured herring sperm DNA (Sigma) at 56°C in a humid environment. Hybridization was performed at 56°C overnight with a solution containing 50% deionized formamide, 4× SSC, 20 mM sodium phosphate buffer (pH 6.5), 0.1% SDS, 1× Denhardt’s solution, 10% dextran sulfate, 500 μg/ml of heat-denatured herring sperm DNA, and 20 μg/ml of heat-denatured, digoxigenin-labeled probe. After hybridization, the slides were soaked for 15 min each in 1× SSC at 37°C and subsequently at 45°C, for 15 min in 0.5× SSC at 45°C, for 15 min in 0.5× SSC at 56°C, for 1 h in 50% deionized formamide and 0.2× SSC at 65°C, and subsequently for 5 min in 0.1× SSC at room temperature. Hybridized digoxigenin-labeled probe was visualized using the Dig wash and block buffer set, alkaline phosphatase-conjugated anti-digoxigenin Ab, and NBTC/BCIP substrate (Boehringer Mannheim). Slides were incubated for 5 min in washing buffer, for 1 h in blocking solution, for 1 h in blocking solution containing anti-digoxigenin Ab (diluted 1/5000), twice for 30 min each in washing buffer, for 5 min in detection buffer, and finally for 5 h (except where indicated) in detection buffer containing 0.45 mg/ml of NBTC and 0.175 mg/ml of BCIP. Color development was stopped by rinsing in TE buffer (10 mM Tris, pH 7.9, 1mM EDTA) and the slides were air-dried and mounted in Aquamount (Fisher Scientific). In situ hybridization experiments were routinely conducted in parallel using sense probes as a control. No hybridization was observed in any experiments using the sense probes.
Because a mature switch transcript contains sequences from the Iγ and Cγ exons, a Cγ probe can detect not only IgG mRNA but also switch transcripts. However, we found that by decreasing incubation time for color development from 5 to 2 h, we could detect intense signals from cells expressing high levels of Cγ mRNA and IgG protein but essentially no signal from cells expressing an Iγ1 switch transcript and IgM protein. This observation was confirmed by the demonstration that identical populations of cells in adjacent sections either hybridized with the Cγ probe after 2 h of development or were stained by anti-IgG1 Abs (e.g., Fig. 2). Turaga et al. (56) have observed previously the preferential detection of plasma cells in primary cultures of LPS/IL-4-stimulated B cells using a 35S-labeled Cγ1 riboprobe. Under our conditions for Cγ visualization, we obtained clear signals after brief incubation, probably because high levels of the IgG transcript are present in most newly activated cells. It is pertinent to point out that our in situ hybridization studies do not appear to give quantitative estimates of the absolute numbers of cells synthesizing Cγ-containing transcripts, because cells producing low levels of the Cγ transcript, such as memory cells, may not be visualized. However, within an individual experiment it is appropriate to compare the signal intensities achieved with each probe in different areas of the same section and in sections taken at different times postimmunization. We attempted to minimize variations in signal intensity by routinely developing slides for a period of 5 h when visualizing Iγ1 hybridization.
No expression of Iγ1 transcripts is evident before immunization
Splenic sections from C57BL/6 mice housed under clean conditions were analyzed before immunization. In accordance with previous reports (12, 13), we found that these sections contained characteristic zones of Thy-1.2-expressing T cells surrounded by follicular regions of B220+ B cells (Fig. 2,A, left). Expression of IgM transcripts by follicular B cells was visualized by Cμ hybridization (Fig. 2,A, right). The field shown in Figure 2 was taken at ×40 magnification, and dark dots correspond to individual cells that express high levels of Cμ mRNA. At day 0, an Iγ1-specific riboprobe did not hybridize to any cells (Fig. 2,A, right), nor were there clear signals from hybridization with either a Rad51-specific (Fig. 2 A, right) or a λ-specific (data not shown) riboprobe. The absence of IgG and λ light chain expression at day 0 was confirmed by staining with anti-λ and anti-IgG1 Abs; both produced a negligible staining pattern (data not shown).
Localized induction of Iγ1 transcripts precedes switch recombination
At 2 days after immunization with NP-CGG, both μ+ and λ+ cells were readily identified by the hybridization of splenic sections with specific riboprobes (Fig. 2,B). λ light chain expression is a specific marker for newly activated B cells in C57BL/6 mice immunized with NP-CGG, as this Ag induces a strain-specific response dominated by λ1 light chains; most murine B cells express κ light chains (19, 53, 57). In the consecutive serial sections shown (Fig. 2 B), adjacent PALS regions give the appearance of hourglass-shaped clusters of cells. Each PALS region surrounds a central blood vessel, which can be seen as a small circular region from which stain has been excluded.
Each panel in Figure 2 B was originally taken at ×40 magnification and corresponds to an area that includes ∼23,000 individual cells. We have chosen to present these panels at this level of magnification to give a more accurate impression of the whole spleen section and to show the consistency of observations across numerous PALS and follicular regions. In situ hybridization identifies individual positive cells as darkly stained dots against a background of cells that are negative for staining. By counting the numbers of darkly staining cells in the larger field, we can estimate the number of activated B cells at each day. There are ∼23,000 cells in a 1.6 × 1.1 mm field; at day 2, we estimate that a field this size contains ∼100 to 300 λ+ cells and approximately twice that number of μ+ cells. The number of the very darkly staining μ+ and λ+ cells appeared to correlate well with the number of newly activated cells previously reported by Jacob et al. (19).
It is clear from comparing the panels in Figure 2 B that the morphology of the PALS is preserved in consecutive sections. Moreover, both Cμ+ cells and λ+ cells were found localized predominantly within the PALS, as predicted from results from other laboratories (19, 28). Analogous results were obtained when sections were analyzed by staining with anti-λ and anti-IgM Abs instead of by hybridization (data not shown).
To determine whether expression of the Iγ1 switch transcript could be observed at day 2 postimmunization, a riboprobe specific for this transcript (referred to as the Iγ1 probe) was hybridized to splenic sections taken at this timepoint. Specificity of hybridization was assayed by hybridization of the complementary riboprobe (Iγ1 sense). Figure 2 B shows the typical pattern of hybridization observed with each probe, presenting consecutive sections from the series shown hybridized with the μ and λ probes as examples. As this example illustrates, there was clear hybridization by the Iγ1 probe. Moreover, hybridization was localized to regions of B cells that, in adjacent sections, displayed the characteristic phenotype of PALS-associated clusters. As with μ and λ hybridization, there were some very concentrated sites of Iγ1 staining that appeared to correspond to single cells. In contrast to the μ and λ signals, there was a considerable level of Iγ1 staining throughout the PALS. The probe for the complementary strand produced only faint staining, demonstrating the specificity of the Iγ1 staining pattern.
Hybridization with a Cγ probe (see Materials and Methods) allowed us to determine whether any switch recombination had occurred at day 2. A typical pattern of hybridization is shown in the fifth consecutive section from the series in Figure 2 B. A small number of cells hybridized to the Cγ probe, and a comparable number of IgG1+ cells were evident upon staining with anti-IgG1 Abs (data not shown). Therefore, switching has occurred in a small but significant fraction of newly activated cells at day 2.
It is clear from this series of panels that Iγ1 expression is specific to regions containing B lymphocytes that have migrated to the PALS following immunization and that express Cμ and λ mRNA. No hybridization was obtained using a riboprobe specific for the complementary strand of the switch transcript (Fig. 2 B), showing that the observed signal is specific for the transcript. These results suggest that Iγ switch transcription is initiated in newly activated B cells of the PALS soon after primary immunization.
In addition to the large clusters of cells that hybridized intensely with the Cμ, λ, and Iγ1 probes, there were also regions in the red pulp where hybridization was apparent. These regions may represent cells in the periphery of PALS clusters located above and/or below the plane of the sections shown, or they may derive from small numbers of cells that have migrated to the red pulp. Cells labeled within the red pulp also expressed λ mRNA and protein, identifying them as newly activated B cells rather than memory cells.
As noted above, a certain level of Iγ1 staining is visible throughout the PALS at day 2. While this may represent background that is not readily removed during the wash protocols, the fact that no comparable background is seen in hybridization with the μ, λ, or γ probes argues against this possibility. As shown below, we consider another possibility (see Discussion), namely that this staining may reflect a generalized activation of B cells and an initiation of Iγ1 switch transcript synthesis throughout the PALS at early timepoints after immunization.
Switch recombination occurs in PALS before formation of GCs
At day 5, hybridization with a Cγ-specific probe showed that considerable switch recombination had occurred. In the example shown (Fig. 2,C, left), Cγ+ cells can be seen tracking a blood vessel, which has been sliced along its length during sample preparation. In an adjacent serial section, the Iγ1-specific probe hybridized to cells clustered along this same arteriole. If the blood vessel had been sliced transversely, the staining would be visible as clusters of cells around a central hole. The staining of adjacent sections with PNA revealed no PNA+ GCs at this timepoint (Fig. 2,C, left), nor were there any diffuse regions of PNA staining that colocalized with regions of Cγ or Iγ1 expression. In additional serial sections, the regions of Cγ and Iγ1 hybridization were shown to contain both B220+ and Thy-1.2+ lymphocytes (data not shown), confirming that Cγ and Iγ1 expression was localized to the PALS. The specificity of Iγ1 and Cγ hybridization was confirmed by comparing Cγ hybridization with the pattern of staining by anti-IgG1 or anti-IgM Abs (Fig. 2 C, right). These results demonstrate that Iγ1 and Cγ signals are colocalized in regions within the PALS where expression of IgG protein is also evident.
Rad51 mRNA expression correlates spatially and temporally with Iγ1 switch transcript expression
The patterns of expression of Rad51, Iγ1, and Cγ transcripts were compared in experiments that analyzed hybridization to adjacent sections at several timepoints following immunization with NP-CGG. The Rad51 transcript was not expressed at day 0 (Fig. 2,A), but Rad51 expression was clearly evident by day 2. At this timepoint and through day 5, Rad51 specific hybridization colocalized with cells expressing Iγ1 switch transcripts (Fig. 3,A). Probes for both transcripts produced a similar hybridization pattern of intensely labeled individual cells surrounded by a more diffusely stained positive region. In contrast, no hybridization was obtained using a riboprobe specific for the complementary strand of Rad51 (data not shown). At day 3, some Cγ expression is evident; this expression increases dramatically by day 5, which is consistent with the considerable switch recombination occurring during this interval. At day 5, cells expressing Rad51 mRNA and Iγ1 transcripts were clustered in the PALS and follicles and colocalized to regions characterized by intense hybridization with the Cγ probe (Fig. 3,A, center). Previously, we have visualized the colocalization of Rad51 and IgG-expressing cells at day 5 by staining with specific Abs (45). At day 12, Rad51 expression is evident but appears to have diminished, as have Iγ1 and Cγ expression (Fig. 3 A, bottom). Calculations of the splenic area occupied by cells expressing Iγ1 transcripts at various timepoints following immunization indicated that Iγ1-specific hybridization increased to a peak at day 7 and then decreased, suggesting that Iγ1 expression is similarly down-regulated after switching is completed.
Iγ1 switch transcripts are present in GC B cells
Somatic hypermutation occurs in GCs, which are readily visualized by staining with PNA. Consistent with our previous findings (45), essentially no GCs were evident before about day 9. By day 12, GCs were very abundant and exhibited a characteristic pattern of staining. Figure 3,B shows an example of consecutive serial sections stained with PNA, anti-B220, and anti-Thy-1.2 Abs, respectively (Fig. 3 B, left). Staining revealed T cell-rich PALS zones located adjacent to the B cell-rich GCs. The black dots in these panels correspond to nonspecific deposition of the DAB4 color substrate solution. The PALS did not show significant levels of anti-B220 staining at day 12, perhaps because the B cell population in these regions was mostly comprised of plasma cells. A low level of B220 staining by plasma cells in vitro has been observed previously (58, 59).
Cells expressing Iγ1 switch transcripts were still evident at day 12. An example of a typical hybridization pattern is shown in Figure 3,B (top right), where intensely hybridizing individual Iγ1+ cells are scattered against a background of more diffuse staining localized to the GC and adjacent PALS. Hybridization of an adjacent section with a Cγ probe revealed large numbers of Cγ+ cells (Fig. 3,B, center right) that were predominantly located in the Thy-1.2+ T cell-rich PALS (Fig. 3,B, bottom left). Essentially no hybridization to a Cμ probe was observed (Fig. 3 B, bottom right), indicating that switching from μ had been completed by day 12.
Figure 3 C shows analysis of consecutive sections through two adjacent PNA+ GCs. Photographs were taken at ×100 magnification and illustrate the colocalization of Iγ1 and Cγ hybridization in consecutive serial sections. Intense signals are visible in the PALS region that separates the two GCs, and more diffuse staining is apparent within the two GCs. A clear Cγ signal is evident in the PALS between the two GCs.
We have identified and localized B cells expressing Iγ1 switch transcripts by in situ hybridization to splenic sections. Newly activated B cells expressing Iγ1 switch transcripts were first evident at 2 days postimmunization. At this timepoint, there was considerable expression of Cμ but not Cγ mRNA. Therefore, induction of Iγ1 switch transcripts is evident in splenic B cells before the occurrence of switch recombination, as expected from analysis of switching in primary cultured cells (39, 40, 41, 42, 43, 44). The rapid induction of switch transcripts that we have observed in vivo parallels the rapid induction of switch transcripts that occurs in activated primary B cells cultured in vitro (40, 41, 56).
The appearance of newly activated B cells corresponded both temporally and spatially with the development of PALS-associated clusters of Ag-specific B cells. As measured by the appearance of Cγ+ cells, active switch recombination began as early as day 2 and peaked at about day 5. During this interval, the number of Cγ+ cells increased dramatically. This picture of the kinetics of switch recombination in vivo during the NP-CGG response is consistent with kinetics determined using plaque assays to quantitate IgM- and IgG-secreting B cells (60). It is also consistent with the kinetic picture of switching in the NP response as visualized in splenic sections by Jacob et al. (19) and by our own laboratory (45). The possibility that switching occurs in the PALS is further supported by studies showing that mice which had been tolerized with soluble NP preimmunization developed significantly lower numbers of NP-specific GCs but showed no reduction in PALS-associated clusters of NP-specific cells or levels of switching (27). Likewise, lymphotoxin-α-deficient mice immunized with high doses of NP-OVA mounted a high affinity IgG1 response even though they failed to produce GCs (26).
The hypothesis that switch recombination occurs in GCs has been put forward by several laboratories (22, 29, 30, 31, 52). In some instances, this notion was based on experiments demonstrating that switch transcripts were present in cells bearing surface markers of GC B cells. Liu et al. (52) observed both switch transcripts (by RT-PCR) and switch circles (by PCR) in analysis of a FACS-sorted population with the surface phenotype of GC centrocytes. Islam et al. (51) used RT-PCR to analyze Iε and Cε mRNA levels in PBMCs from patients being treated for schistosomiasis and found Iε transcripts present at day 12. Toellner et al. (28) used RT-PCR to analyze switch transcript levels in splenic sections at various timepoints after immunization; while they found switch transcripts present at late timepoints, they observed no correlation between levels of switch transcripts and the size of GCs in adjacent sections and concluded that GCs were not the primary sites of switch recombination (28). Consistent with these other studies (28, 51, 52), we did find some level of Iγ1 expression in GCs. We cannot distinguish whether these transcripts are evident because they persist within B cells that have completed recombination in the PALS and migrated to the GCs or because Iγ transcription is newly activated in GC B cells. Switch transcripts may persist in activated B cells because they have a relatively long t1/2 for degradation. Alternatively, just as the presence of Rag1 and Rag2 transcripts may be evidence that V region replacement occurs in the GCs (61), the presence of Iγ1 switch transcripts may indicate that switch recombination is ongoing in the GC cells. As essentially no μ mRNA expression was evident at these late timepoints, if recombination is occurring it almost certainly involves other isotypes (for example, switching from one γ subclass to another).
We have shown previously that intense Rad51 staining can be observed in splenic sections during a brief window following primary immunization, and that staining correlates temporally and spatially with cells switching from IgM to IgG expression (45). While the evidence linking Rad51 to switch recombination is thus far only correlative, expression of Rad51 protein is clearly induced at early timepoints (days 2–5) following immunization and then down-regulated before GC formation and somatic hypermutation. Rad51 appears to function in both recombination and repair pathways. If hypermutation in the GC involves a DNA lesion (62, 63, 64), high levels of Rad51 similar to those present in switching B cells might even interfere with the mutagenic mechanism.
The number of μ+ and λ+ cells we observed at day 2 correlated with estimates others have made of the number of newly activated cells present at early timepoints after immunization with NP-CGG (19). However, we observed a surprisingly intense Iγ1 staining pattern at day 2, and significantly more cells appeared to be expressing Iγ1 transcripts than either λ or μ transcripts. This signal could represent background hybridization, but several observations argue against such an interpretation: the signal was specific to the Iγ1 probe; it was not evident in sections hybridized with its complement (Iγ1 sense); and there was little hybridization of the Cγ probe. Moreover, the Iγ1 hybridization signal at day 2 represented a clear increase from day 0, when almost no hybridization was evident. We suggest that there is another more interesting interpretation of this observation. Primary resting B cells can be stimulated to express Iγ1 switch transcripts and switch to γ1 by culture with IL-4 in the absence of LPS (40, 41, 42, 43). IL-4 is a rapid and potent inducer of Iγ1 transcripts in such cultures, suggesting that a significant fraction of resting B cells respond directly to this lymphokine. We hypothesize that, at early timepoints after immunization, a generalized induction of Iγ1 transcription may occur in vivo, as B cells respond to IL-4 secreted by activated Th cells. This early burst of Iγ1 transcription may not necessarily lead to productive switch recombination but would prime the cells to respond to additional signals necessary for commitment to switching. This hypothesis is also favored by the observation that the frequency of primary murine B cells expressing either germline γ1 or ε transcripts after 48 h in culture with LPS and IL-4 is higher than the number of cells that ultimately switch to the respective isotypes (65). In addition, evidence is accumulating for the requirement of additional factors to promote switching: isotype commitment can be influenced after IL-4-mediated germline transcription by signals from cell surface molecules (e.g., Lyb2 (66)), and IgE secretion can be suppressed by IFN-γ and IFN-α without altering germline Cε transcription (67). Additional signals required may depend upon physical contact with activated T cells and ligation of the CD40 molecule (68, 69).
Having established that Iγ1 is an early marker for imminent switch recombination in splenic B cells, it should be possible to identify additional genes that are specifically induced in B cells activated for switching. It is invaluable to be able to carry out such analysis in vivo, where critical cell-to-cell contacts and unique microenvironments are preserved.
We thank Drs. Efim Golub and Charles Radding (Yale University) for their gift of the plasmid pEG986 and Abs to Rad51. We also thank Dr. Wesley Dunnick (University of Michigan) for his gift of Iγ1 exon DNA.
This work was supported by National Institutes of Health Grants GM39799 and GM41712 (to N.M.).
Abbreviations used in this paper: PALS, periarteriolar lymphoid sheath; DAB, 3,3′-diaminobenzidine; GC, germinal center; PNA, peanut agglutinin; NP, (4-hydroxy-3-nitrophenyl)acetyl; CGG, chicken gamma-globulin; NBTC, 4-nitroblue tetrazolium chloride; BCIP, 5-bromo-4-chloro-3-indolyl-phosphate.