LPS was used to induce switching of B cells to IgG3 and, in the presence of TGF-β, to IgG2b and IgA. Switching to all three isotypes increased with division number according to a consistent relationship that was independent of time in culture. The mode of activation altered the relationship with division, as CD40 ligand increased switching to IgA and decreased switching to IgG2b and IgG3 when measured per division. This division-linked switching behavior could be described by Gaussian probability distributions centered around a mean division number. The divisions at which switching to IgG3 and IgG2b occurred overlapped, raising the possibility that the two switching mechanisms were linked. However, when IgG3+ and IgG3 B cells were sorted and placed back in culture, they switched to IgG2b at an equivalent rate, indicating that alternative switching decisions were made independently within a single cell. As a consequence, isotype switching could be predicted at the population level by standard probability laws. Therefore, division number provides a framework for a stochastic description of differentiation that may be widely applicable.

Immunoglobulin isotype switching by B lymphocytes is a biologically important feature of the humoral immune response. In mice, conventional B cells possess the ability to switch from IgM and IgD expression to any of six alternative heavy chain genes after activation. The mechanism of switching depends upon the initial expression of sterile germline transcripts of the target Ig heavy chain gene followed by deletional recombination, which leads to splicing of the Ag-recognizing VDJ unit adjacent to the selected heavy chain gene (1, 2, 3, 4, 5). B cells display an exquisite ability to control and target heavy chain switching. Cytokines such as IL-4, IFN-γ, and TGF-β (6, 7) play a crucial role in directing the switching machinery. The mode of B cell activation, whether T-independent, as invoked by LPS, or T-dependent after exposure to intact T cells or CD40 ligand (CD40L)3 also affects the outcome of cytokine stimulation, including the efficiency and direction of isotype switching (7, 8, 9). It remains of great interest to understand the “logic” of Ig isotypes, why they are induced by some stimuli and not others, and how they might be controlled to facilitate regulation of the immune system, including the class of Ab response elicited during natural infection or after vaccination.

Methods for inducing each of the murine isotypes in vitro have been developed. Switching to IgG1 and IgE with expression of γ1 and ε germline transcripts occurs as a result of activation of B cells with the T-independent stimulus LPS in the presence of IL-4 (10, 11) or with the T cell-dependent combination of CD40L and IL-4 (12). Addition of IFN-γ to LPS-stimulated cultures induces switching to IgG2a and the production of γ2a germline transcripts (13, 14). When TGF-β is added to the same culture system, the rate of switching to IgG2b and the steady-state levels of γ2b germline transcripts increase (15). Furthermore, several groups have demonstrated that TGF-β can induce LPS-stimulated membrane IgA B cells to secrete IgA (16, 17). One isotype to which switching does not appear to require a cytokine is IgG3. Thus, stimulation of B cells with LPS is sufficient on its own to induce switching to this isotype, with production of γ3 germline transcripts (18, 19, 20).

Further complexity in isotype switching results from the potential for sequential switching. It has been shown by analysis of DNA switch circles, and from the sequence of recombined switch regions, that the switch to IgE and to IgA sometimes proceeds through IgG intermediates, whereas on other occasions it is made directly from IgM (21, 22, 23, 24, 25, 26, 27). Sequential switching also takes place between IgG isoforms as shown, for example, in the case of IgG2b, where switching can occur via an IgG3 intermediate (28). The relative importance of direct vs sequential switching for the production of a given isotype remains unclear. For instance, although a high proportion of IgE+ cells switch via IgG1 (21), interference with the switch to IgG1 does not affect the rate of switching to IgE (1, 29). The mechanism of regulation underlying this diverse range of pathways therefore requires further investigation.

An unexpected feature that is shared by all isotype switching in vitro is its inefficiency. Under identical culture conditions involving activation of purified resting B cells in the presence of, for example, IL-4, only 20–50% and 2–25% of cells in culture switch to IgG1 and IgE, respectively (21, 29, 30, 31). Recently, we provided an explanation for this phenomenon. B cells undergoing isotype switching increase in frequency by a measurable amount per division (32). However, B cell proliferation in vitro is highly asynchronous, as the cells that are recovered from culture encompass a wide range of division numbers (33). Thus, even though the cells are following a simple division-based rule, the net result appears complex due to variation in the division history of individual cells. These studies have also established the principle that control of the division-based rate of switching (expressed as a probability per division) and the rate of proliferation (expressed in units of time) are independent events that can be distinguished and studied separately (32, 33). This is particularly important when studying the effect of individual cytokines, alone or in combination, as they frequently alter the rates of both switching and proliferation (33).

In this study, the relationship between division number and the isotypes IgG3, IgG2b, and IgA has been examined by culturing B cells in the presence of LPS, CD40L, and TGF-β.

Male and female CBA/H mice were obtained from the Animal Resources Centre (Canning Vale, Australia). Mice were maintained under specific pathogen-free conditions in the Centenary Institute animal facility and used between 8 and 12 wk of age.

Percoll was obtained from Pharmacia Biotech AB (Uppsala, Sweden). 5- (and 6-)carboxyfluorescein diacetate succinimidyl ester (CFSE) was obtained from Molecular Probes (Eugene, OR). LPS from Salmonella typhosa was obtained from Sigma (St. Louis, MO) and human recombinant TGF-β1 was acquired from Boehringer Mannheim (Mannheim, Germany). CD40L prepared from the Sf9 insect cell line transfected with baculovirus vector containing the CD40L gene (34) was a generous gift of Dr. M. R. Kehry (Boehringer Ingleheim, Ridgefield, CT). Biotinylated goat anti-mouse IgG2b, IgG3, and IgA were purchased from Southern Biotechnology Associates (Birmingham, AL). Biotinylated rat anti-mouse IgG2a was obtained from Serotec (Oxford, U.K.). R1E4 (anti-mouse IgE) and AMS 15.1 (anti-mouse IgD) were conjugated to PE using succinimidyl 4-(p-maleimidophenyl) butyrate (Pierce, Rockford, IL). RS3.1 (anti-mouse IgM) biotin or allophycocyanin conjugate and goat anti-mouse IgG1 PE conjugate were supplied as a gift by Dr. Paul Lalor (Walter and Eliza Hall Institute, Melbourne, Victoria, Australia). Streptavidin (SA)-tricolor was purchased from Caltag (Burlingame, CA), and SA-Texas Red was obtained from Molecular Probes.

Small resting B cells were prepared as described previously (35). Briefly, single-cell suspensions were prepared by teasing mouse spleens through a stainless steel mesh. After lysing RBCs with hypotonic ammonium chloride solution, adherent cells were depleted by incubation on a plastic tissue culture dish. T cells were depleted by complement lysis using a mixture of CD4-specific (RL172), CD8-specific (31 M), and Thy-1-specific (30H12) mAbs. Small dense B cells were further purified by centrifugation on a Percoll density gradient. The B cells recovered from the 65/75% Percoll gradient interface were considered to be resting B cells and were ∼95% B220+ and <1% CD4+ and CD8+ as determined by flow cytometry. Small resting B cells were cultured in B cell medium (BCM) RPMI 1640 medium (Life Technologies, Grand Island, NY) supplemented with 2 mM l-glutamine, 0.1 mM nonessential amino acids, 10 mM HEPES (pH 7.4), 100 μg/ml streptomycin, 100 U/ml penicillin, 5 × 10−5 M 2-ME (all from Sigma), 1 mM sodium pyruvate (Life Technologies), and 10% heat-inactivated FCS (Life Technologies). Typically, cells were cultured with various combinations of cytokines and mitogens in 24- or 96-well flat-bottom plates (Falcon, Becton Dickinson, San Jose, CA) at a density of 2.5 × 105 cells/ml and incubated at 37°C in a humidified atmosphere of 5% CO2.

Resting B cells were labeled with CFSE according to the original method described elsewhere (36). Briefly, B cells were washed in PBS containing 0.1% BSA (PBS/BSA) and resuspended at 1 × 107 cells/ml. CFSE stock (5 mM in DMSO) was diluted 1/10 in PBS/BSA, and 30 μl was added to a 1-ml cell suspension (final concentration of 15 μM). Cells were then incubated in a water bath at 37°C for 10 min. Ice-cold BCM was added to quench unlabeled CFSE, and cells were spun down and resuspended in BCM.

At various timepoints, cells were harvested and washed twice with PBS/BSA containing 0.1% sodium azide and aliquoted to 96-well V-bottom plates (Falcon, Becton Dickinson) at ∼106 cells/well. All samples in the plate were separated by one well to avoid cross-contamination between samples. Abs were diluted in PBS/BSA/azide with 1% normal rat serum to block nonspecific binding. Cells were then gently vortexed to ensure mixing and incubated on ice with biotin or PE Ab conjugates for 40 min, followed by an additional 30 min of incubation with SA-tricolor. Between incubations, cells were washed twice with 200 μl of PBS/BSA/azide. After a final wash, the cells were transferred to round-bottom FACS tubes (Falcon, Becton Dickinson) for flow cytometry analysis. In some experiments, cells were fixed and permeabilized for intracellular staining, as this method enhances the sensitivity of staining. For this procedure, cells were harvested, pelleted, and fixed in 500 μl of 2% paraformaldehyde for 10 min at room temperature. After fixation, 500 μl of PBS/BSA/azide and 1 ml of PBS/0.2% Tween 20 (ICN, Irvine, CA) were added to yield a final concentration of 0.5% paraformaldehyde with 0.1% Tween 20; the cells were incubated overnight at room temperature in the dark. After permeabilization, cells were washed twice in PBS/BSA/azide and stored at 4°C until stained as described above. Analysis was conducted on a FACScan flow cytometer (Becton Dickinson) using CellQuest software (Becton Dickinson). Cell sorting was conducted on a FACStar+ flow cytometer (Becton Dickinson).

To enable determination of the cell numbers in each sample after culture, a known number of Calibrite beads (unlabeled, Becton Dickinson) were added to the cell samples before harvesting (typically, 104 beads per sample). The cells with beads were then harvested and run simultaneously on flow cytometry. Bead and live cell events were discriminated using side and forward scatter parameters acquired during flow cytometric analysis. The number of live cells in each sample was calculated from the ratio of live cells to bead events and the known total bead numbers per sample. The proportion of cells in each division was determined by setting gates around each peak using CellQuest analysis software as described previously (32, 33).

Small CFSE-labeled B cells were cultured with 50 μg/ml LPS for 4 days before being stained for each murine Ig isotype. As shown in Fig. 1,B, cells divided asynchronously and were spread over a range of divisions, as noted previously after CD40L stimulation (33). Contour plots of CFSE vs Ig confirmed that LPS induced a proportion of cells to switch to IgG3 and caused a smaller fraction of cells to express IgG2b (18, 20) (Fig. 1). However, in the absence of exogenous cytokines, LPS stimulation failed to induce switching to the other four murine isotypes (Fig. 1). In addition, IgM and IgD were shown to display different division-based patterns of loss that resembled those observed after CD40L stimulation (33). IgM remained at a high level before diminishing, presumably as a result of switching to downstream isotypes. In contrast, IgD was lost progressively from dividing cells, irrespective of whether or not they underwent isotype switching.

FIGURE 1.

Ig isotype expression by B cells after LPS stimulation. CFSE-labeled B cells were stimulated with 50 μg/ml LPS, harvested after 4 days, and stained for each isotype as described in Materials and Methods. The top left histogram illustrates the distribution of CFSE intensity above which the predicted position of each division is shown. The panels present contour plots of Ig isotype expression of CFSE-labeled B cells stimulated with LPS. The percentage of Ig-switched cells is given in the top right hand corner of the appropriate panels.

FIGURE 1.

Ig isotype expression by B cells after LPS stimulation. CFSE-labeled B cells were stimulated with 50 μg/ml LPS, harvested after 4 days, and stained for each isotype as described in Materials and Methods. The top left histogram illustrates the distribution of CFSE intensity above which the predicted position of each division is shown. The panels present contour plots of Ig isotype expression of CFSE-labeled B cells stimulated with LPS. The percentage of Ig-switched cells is given in the top right hand corner of the appropriate panels.

Close modal

LPS-stimulated CFSE+ cells were harvested on consecutive days from equivalent cultures, stained for IgG3, and examined by flow cytometry. The resulting CFSE histograms were analyzed as described previously (32, 33) with a view toward calculating both the proportion of total viable cells found in each division, and the proportion of IgG3+ cells in each division. Fig. 2 shows the proportion of cells in each division at each time for two separate experiments and indicates that B cells progressed on average ∼1.5 divisions per day. However, when the proportion of IgG3+ cells in each division was plotted, the results were remarkably consistent, revealing a reproducible, time-independent, division-linked relationship for switching, as was observed for other isotypes (32, 33).

FIGURE 2.

Switching to IgG3 is division-related and independent of time. CFSE-labeled B cells were stimulated with 50 μg/ml LPS, harvested after 3, 4, and 5 days of culture, and stained for IgG3. A shows the percentage of total viable cells found in each division for two experiments (expt.1, open symbols; expt. 2, closed symbols) on days 3 (squares), 4 (diamonds), and 5 (circles). B shows the percentage of IgG3+ cells in each division for both experiments.

FIGURE 2.

Switching to IgG3 is division-related and independent of time. CFSE-labeled B cells were stimulated with 50 μg/ml LPS, harvested after 3, 4, and 5 days of culture, and stained for IgG3. A shows the percentage of total viable cells found in each division for two experiments (expt.1, open symbols; expt. 2, closed symbols) on days 3 (squares), 4 (diamonds), and 5 (circles). B shows the percentage of IgG3+ cells in each division for both experiments.

Close modal

The cytokine TGF-β has been reported to promote switching to IgG2b and IgA and inhibit proliferation (37). If the TGF-β-induced switching to these isotypes is division-related, its effects on proliferation will complicate the interpretation of any influence it may exert on the switching process per se, as has been shown for IgG1 and IgE (33). For this reason, it was important to separate the two potential actions of TGF-β. The proportion of live cells in each division was calculated through division slicing based on CFSE intensity. Pulsing LPS-stimulated B cells with TGF-β caused a potent inhibition of proliferation (Fig. 3) that was dependent upon the time of addition of the cytokine. When added in high concentrations (0.5 ng/ml) at the start of culture, cell proliferation was diminished to very low levels. However, as shown previously (38), a delay in addition of the cytokine by 24 or 48 h permitted significant numbers of cells to progress to later divisions. Therefore, for additional experiments, TGF-β was added at 24 h, to allow sufficient proliferation to occur so that its effects on isotype switching could be monitored.

FIGURE 3.

Effect of TGF-β on LPS-stimulated B cell proliferation. CFSE-labeled B cells were incubated with 50 μg/ml LPS, and various concentrations of TGF-β were added either at the start of culture (top), after 24 h (middle), or after 48 h (bottom). Cells were harvested at 4 days after the start of culture, and the number of live cells in each division was determined as described in Materials and Methods.

FIGURE 3.

Effect of TGF-β on LPS-stimulated B cell proliferation. CFSE-labeled B cells were incubated with 50 μg/ml LPS, and various concentrations of TGF-β were added either at the start of culture (top), after 24 h (middle), or after 48 h (bottom). Cells were harvested at 4 days after the start of culture, and the number of live cells in each division was determined as described in Materials and Methods.

Close modal

The effect of TGF-β on switching to IgG2b was examined by stimulating B cells with LPS in the presence of 0.5 ng/ml TGF-β (added after 24 h). After harvesting of the cultures on day 5, IgG2b+ cells were found to comprise 20% of the total cell population (Fig. 4). When the proportion of IgG2b+ cells per division was calculated, switching to this isotype exhibited a division-related profile. Moreover, there was a low level of switching in the absence of exogenous TGF-β that also appeared to be division-dependent. According to Snapper et al. (39), this low level of “endogenous” IgG2b switching is due to secretion of small concentrations of TGF-β into culture by activated B cells.

FIGURE 4.

TGF-β induces division-linked switching to IgG2b by LPS-stimulated B cells. CFSE-labeled B cells were stimulated with 50 μg/ml LPS in the presence or absence of 0.5 ng/ml TGF-β (added 24 h after the initiation of culture). Cells were harvested after 5 days and stained with anti-IgG2b. A, Contour plots show expression of IgG2b vs CFSE intensity. The percentage of IgG2b+ cells in the boxed region is indicated. B, The percentage of IgG2b+ cells in each division is shown.

FIGURE 4.

TGF-β induces division-linked switching to IgG2b by LPS-stimulated B cells. CFSE-labeled B cells were stimulated with 50 μg/ml LPS in the presence or absence of 0.5 ng/ml TGF-β (added 24 h after the initiation of culture). Cells were harvested after 5 days and stained with anti-IgG2b. A, Contour plots show expression of IgG2b vs CFSE intensity. The percentage of IgG2b+ cells in the boxed region is indicated. B, The percentage of IgG2b+ cells in each division is shown.

Close modal

To examine whether the switch to IgG2b proceeded independently of time in culture, a time course experiment was performed on four separate occasions (Fig. 5). Although the proportion of cells found in each division varied between experiments, suggesting some sensitivity to culture conditions (Fig. 5,A), the pattern of division-related switching to IgG2b was clearly time-independent and was remarkably consistent between the four experiments (Fig. 5 B).

FIGURE 5.

Switching to IgG2b is division-related and independent of time. In four separate experiments, CFSE-labeled cells were stimulated with LPS (50 μg/ml), and a saturating dose of TGF-β (0.5 or 1 ng/ml) was added after 24 h. The cells were harvested after 3, 4, and 5 days. The percentage of the total live cells in each division was determined. The upper panel shows an overlay of the results of three of the experiments indicated by either open symbols, shaded symbols, or filled symbols. Squares indicate cells harvested at day 3, diamonds indicate cells harvested at day 4, and circles indicated cells harvested at day 5. The percentage of IgG2b+ cells in each division was also calculated. The lower panel shows the mean ± SE of IgG2b expression taken from all four experiments harvested on days 3 (squares), 4 (diamonds), and 5 (circles).

FIGURE 5.

Switching to IgG2b is division-related and independent of time. In four separate experiments, CFSE-labeled cells were stimulated with LPS (50 μg/ml), and a saturating dose of TGF-β (0.5 or 1 ng/ml) was added after 24 h. The cells were harvested after 3, 4, and 5 days. The percentage of the total live cells in each division was determined. The upper panel shows an overlay of the results of three of the experiments indicated by either open symbols, shaded symbols, or filled symbols. Squares indicate cells harvested at day 3, diamonds indicate cells harvested at day 4, and circles indicated cells harvested at day 5. The percentage of IgG2b+ cells in each division was also calculated. The lower panel shows the mean ± SE of IgG2b expression taken from all four experiments harvested on days 3 (squares), 4 (diamonds), and 5 (circles).

Close modal

Next, the effect of varying the dose of TGF-β on switching to IgG2b was examined (Fig. 6). As the concentration of TGF-β was reduced from 0.5 to 0.005 ng/ml, a progressive decrease in division-associated switching occurred, with 0.005 ng/ml equivalent to the background level. Furthermore, the reduced rate of division-linked switching seen at lower concentrations was again time-independent. By comparing the dose response curves for induction of switching to IgG2b and for inhibition of proliferation, the two responses were shown to occur over a similar concentration range (Fig. 6 C).

FIGURE 6.

The TGF-β dose range for inhibition of proliferation and promotion of switching to IgG2b are similar. A, CFSE-labeled B cells were stimulated with 50 μg/ml LPS, and TGF-β was added at the concentrations indicated after 24 h. Contour plots show IgG2b expression vs CFSE intensity for cells harvested after 5 days. The proportion of IgG2b+ cells is given for each plot in the top right corner. B, The percentage of IgG2b+ cells in each division is presented for cells harvested after 3, 4, and 5 days. C, Dose response curves for proliferation and switching were compared. Data represent two experiments (distinguished by circles and squares). The percent inhibition of proliferation (closed symbols) was calculated from the total live cell number with and without TGF-β. To represent switching ability, the proportion of switched cells in division 7 for a given concentration of TGF-β was calculated as a percentage of the proportion of switched cells in division 7 saturating concentrations of TGF-β. The switching values are normalized against the levels of switching seen with stimulation with LPS alone.

FIGURE 6.

The TGF-β dose range for inhibition of proliferation and promotion of switching to IgG2b are similar. A, CFSE-labeled B cells were stimulated with 50 μg/ml LPS, and TGF-β was added at the concentrations indicated after 24 h. Contour plots show IgG2b expression vs CFSE intensity for cells harvested after 5 days. The proportion of IgG2b+ cells is given for each plot in the top right corner. B, The percentage of IgG2b+ cells in each division is presented for cells harvested after 3, 4, and 5 days. C, Dose response curves for proliferation and switching were compared. Data represent two experiments (distinguished by circles and squares). The percent inhibition of proliferation (closed symbols) was calculated from the total live cell number with and without TGF-β. To represent switching ability, the proportion of switched cells in division 7 for a given concentration of TGF-β was calculated as a percentage of the proportion of switched cells in division 7 saturating concentrations of TGF-β. The switching values are normalized against the levels of switching seen with stimulation with LPS alone.

Close modal

In addition to promoting switching to IgG2b, TGF-β is known to enhance switching to IgA (16, 17). When cells were stimulated with LPS, <2% switched to IgA even in the presence of TGF-β (Fig. 7,A). This low percentage resulted from a low rate of switching during later divisions (Fig. 7,B). As IgA is a T cell-dependent isotype, cells stimulated with CD40L and TGF-β were also examined. Under these conditions, TGF-β inhibited CD40L-induced proliferation (data not shown) and once again resulted in expression of IgA by only a small percentage of the total cells (Fig. 7,A). However, when plotted per division, it was apparent that a high proportion of the small number of cells reaching the later divisions had switched to IgA (Fig. 7 B), revealing a significant difference in the division-based switching rate to IgA between LPS- and CD40L-activated B cells. Furthermore, the high rate of division-linked switching to IgA induced by CD40L and TGF-β occurred in a time-independent manner, as had been observed for all other isotypes. In contrast to LPS, CD40L stimulation alone failed to induce switching to IgG3 or IgG2b, and induced only low levels of IgG2b expression in the presence of TGF-β (data not shown).

FIGURE 7.

Switching to IgA is division-related. CFSE-labeled B cells were stimulated with LPS (50 μg/ml; upper panels) or CD40L (1/500; lower panels) for 3, 4, or 5 days. TGF-β (0.5 ng/ml) was added to all cultures after 24 h. The cells were harvested and stained for IgA expression. A, Contour plots show the expression of IgA vs CFSE intensity for day 5; the percentage of IgA+ cells is given at the upper right. The percentage of IgA+ cells in each division was calculated and is plotted in B for days 3 (squares), 4 (diamonds), and 5 (circles).

FIGURE 7.

Switching to IgA is division-related. CFSE-labeled B cells were stimulated with LPS (50 μg/ml; upper panels) or CD40L (1/500; lower panels) for 3, 4, or 5 days. TGF-β (0.5 ng/ml) was added to all cultures after 24 h. The cells were harvested and stained for IgA expression. A, Contour plots show the expression of IgA vs CFSE intensity for day 5; the percentage of IgA+ cells is given at the upper right. The percentage of IgA+ cells in each division was calculated and is plotted in B for days 3 (squares), 4 (diamonds), and 5 (circles).

Close modal

The above results revealed that TGF-β induced switching to IgG2b over the same range of divisions as IgG3 switching occurred in the absence of this cytokine (compare Fig. 2 with Fig. 5). These observations provided an opportunity to resolve the question of how individual B cells would behave when confronted with alternative switching influences.

Initially, the effect of TGF-β on LPS-induced switching to IgG3 was examined. Fig. 8,A shows that TGF-β caused a decrease in the total percentage of cells expressing IgG3. When examined on a division basis, this reduction was clearly shown to be due to a decrease in the division-related appearance of IgG3+ cells in the later divisions (Fig. 8 B). The number of double-positive cells expressing both IgG3 and IgG2b in the cultures was insignificant (data not shown), thereby raising the possibility that the inhibitory effect of TGF-β on the switch to IgG3 occurred as an indirect result of some cells switching to IgG2b. Because the γ2b constant region lies downstream of γ3, a B cell that has switched to IgG3 is still capable of switching to IgG2b, whereas an IgG2b+ cell is incapable of switching to IgG3. Cell sorting was used to demonstrate that IgG3+ cells could switch to IgG2b when placed back in culture (data not shown).

FIGURE 8.

TGF-β weakly inhibits division-linked switching to IgG3. A, B cells were stimulated with 50 μg/ml LPS in the presence or absence of TGF-β (0.5 ng/ml added after 24 h) and harvested after 5 days. Contours of IgG3 expression plotted against CFSE intensity are shown, with the percentage of IgG3+ cells given adjacent to the gated cells. B, The percentage of IgG3+ cells in each division was also calculated for cells cultured with LPS (○) or LPS plus TGF-β (□). These data represent the mean of two experiments harvested at three timepoints each: days 3, 4, and 5. Therefore, the data points show the mean and SD of three to six separate determinations. ♦ indicates the predicted effect of TGF-β on levels of IgG3+ if the probability of switching to IgG2b is independent of switching to IgG3 but the switch to IgG2b is dominant. The predicted value is calculated from the percentages here for LPS alone per division (IgG3lps), the division-based map of switching to IgG2b in the presence of TGF-β (Fig. 5 B: IgG2bTGF), and the endogenous level of IgG2b induced in the absence of TGF-β (IgG2blps), according to the following formula: ([IgG3lps] × [100 − IgG2bTGF])/(100 − IgG2blps).

FIGURE 8.

TGF-β weakly inhibits division-linked switching to IgG3. A, B cells were stimulated with 50 μg/ml LPS in the presence or absence of TGF-β (0.5 ng/ml added after 24 h) and harvested after 5 days. Contours of IgG3 expression plotted against CFSE intensity are shown, with the percentage of IgG3+ cells given adjacent to the gated cells. B, The percentage of IgG3+ cells in each division was also calculated for cells cultured with LPS (○) or LPS plus TGF-β (□). These data represent the mean of two experiments harvested at three timepoints each: days 3, 4, and 5. Therefore, the data points show the mean and SD of three to six separate determinations. ♦ indicates the predicted effect of TGF-β on levels of IgG3+ if the probability of switching to IgG2b is independent of switching to IgG3 but the switch to IgG2b is dominant. The predicted value is calculated from the percentages here for LPS alone per division (IgG3lps), the division-based map of switching to IgG2b in the presence of TGF-β (Fig. 5 B: IgG2bTGF), and the endogenous level of IgG2b induced in the absence of TGF-β (IgG2blps), according to the following formula: ([IgG3lps] × [100 − IgG2bTGF])/(100 − IgG2blps).

Close modal

The above demonstration of sequential switching allowed the division-based association between IgG3 and IgG2b to be explored further. If the internal switching machinery for these two isotypes is shared within individual B cells, IgG3+ cells might be expected to switch to IgG2b at a greater frequency than unswitched cells taken from the same divisions. Alternatively, the two isotypes may be regulated independently, in which case prior switching to IgG3 would not affect switching to IgG2b. To distinguish between these alternative predictions, B cells were stimulated with LPS and TGF-β for several days before sorting IgM/IgG3+ and IgM+/IgG3 populations from divisions 4–8 (Fig. 9). The sorted cells were recultured with LPS and TGF-β and examined for IgG2b expression after 24 h. As shown in Fig. 9 C, both IgG3+/IgM and IgM+/IgG3 cells switched to IgG2b at similar frequency (∼13%). In a second experiment, the sorted cells were recultured in the absence of TGF-β. Again, a similar proportion of the IgG3+/IgM and the IgM+/IgG3 cells switched to IgG2b (data not shown).

FIGURE 9.

IgG3+/IgM and IgG3/IgM+ cells switch to IgG2b at the same rate. CFSE-labeled B cells were cultured with 50 μg/ml LPS and TGF-β (0.5 ng/ml added after 24 h). The cells were harvested after 3.5 days and stained for IgG3 and IgM. Cells were then sorted from divisions 4–8 based on CFSE intensity (A) into either IgG3+/IgM or IgG3/IgM+ populations (B). C, Cultures of 4 × 104 sorted cells were supplemented with LPS and TGF-β and incubated for an additional 24 h before being stained for IgG2b expression. The contour plots show the expression of IgG2b for both sorted populations plotted against CFSE intensity; the proportion of IgG2b+ is indicated at the upper right.

FIGURE 9.

IgG3+/IgM and IgG3/IgM+ cells switch to IgG2b at the same rate. CFSE-labeled B cells were cultured with 50 μg/ml LPS and TGF-β (0.5 ng/ml added after 24 h). The cells were harvested after 3.5 days and stained for IgG3 and IgM. Cells were then sorted from divisions 4–8 based on CFSE intensity (A) into either IgG3+/IgM or IgG3/IgM+ populations (B). C, Cultures of 4 × 104 sorted cells were supplemented with LPS and TGF-β and incubated for an additional 24 h before being stained for IgG2b expression. The contour plots show the expression of IgG2b for both sorted populations plotted against CFSE intensity; the proportion of IgG2b+ is indicated at the upper right.

Close modal

Recent studies in our laboratory have highlighted the crucial role played by division number in selection of the appropriate class of response by both T and B cells to a particular antigenic stimulus (32, 33, 40). For B cells, the outcome is switching from IgM to a downstream isotype. In our study, isotype switching to the T-independent isotypes IgG3 and IgG2b and the T-dependent, secretory isotype IgA was shown to display the same features as reported previously for IgG1, IgE, and IgG2a (32, 33, 41) (i.e., they are division number-related and independent of time in culture). Thus, division-based regulation can be regarded as a general mode of control for Ab responses of all classes.

Division-related control of the B cell response is influenced by both the type of activation stimulus and the cytokine milieu in which the B cells reside. As shown here, T-independent (LPS) and T-dependent (CD40L) stimuli directed B cells to follow alternate division-linked switching pathways (Fig. 7), thereby providing a mechanistic framework for earlier demonstrations of the importance of activation regime in determining the outcome of B cell isotype switching (7). For example, LPS on its own induced B cells to switch to IgG3 and in the presence of TGF-β to IgG2b, whereas little to no switching occurred under similar culture conditions when CD40L was added. By contrast, CD40L provoked a higher rate of division-based switching to IgA than did LPS. Thus, the mode of activation must have some influence over the as yet unknown molecular mechanisms responsible for increasing the probability of isotype switching with progressive cell division.

The internal cellular processes responsible for influencing isotype switching are separate from the mechanisms that dictate the timing of division. In other words the fastest and slowest dividing cells display an identical division-based isotype switching frequency. As a result of this time-independent behavior, division-based switching can be represented as a probability distribution which serves as a “map” of isotype transition with division number as the sole relevant variable. The construction of a probability map is illustrated in Fig. 10,A for both IgG3 and IgG2b induced by LPS and LPS plus TGF-β, respectively. When plotted as a probability of switching per division, the distribution obtained for each isotype is approximated by a normal distribution (Fig. 10 B).

FIGURE 10.

Description of isotype switching using division-based probability distributions. A, IgG3 (LPS only) and IgG2b (LPS plus TGF-β) switching data from different timepoints and experiments were averaged to prepare a division-based map for each. Data present the mean and SE of three to six points. The dashed line presents the proportion of cells that became positive in that division (i.e., the proportion in the division minus the proportion in the previous division). A normal distribution is fitted. B, The two probability distributions for IgG3 and IgG2b are overlaid. Note that the area under the curves is not one, indicating that not all cells are destined to switch. Also shown is the effect of reducing the TGF-β concentration, which appears to delay and increase the spread of the probability curve.

FIGURE 10.

Description of isotype switching using division-based probability distributions. A, IgG3 (LPS only) and IgG2b (LPS plus TGF-β) switching data from different timepoints and experiments were averaged to prepare a division-based map for each. Data present the mean and SE of three to six points. The dashed line presents the proportion of cells that became positive in that division (i.e., the proportion in the division minus the proportion in the previous division). A normal distribution is fitted. B, The two probability distributions for IgG3 and IgG2b are overlaid. Note that the area under the curves is not one, indicating that not all cells are destined to switch. Also shown is the effect of reducing the TGF-β concentration, which appears to delay and increase the spread of the probability curve.

Close modal

The division-based switching maps for the two T-independent isotypes allow us to draw an important conclusion from the experiments described here. When TGF-β was introduced into an LPS culture, the IgG2b map covered a range of divisions that was similar to those for IgG3, raising the question of how cells respond when two switch decisions intersect. Because the gene for the IgG2b constant region is downstream of that for IgG3, switching in this sequential direction is possible, whereas the converse, namely switching from IgG2b to IgG3, is not, suggesting that IgG2b should be the dominant switch outcome. Here sequential switching from IgG3 to IgG2b was shown to occur by sorting IgG3+ cells after 4–8 divisions and placing them back in culture, after which they continued to divide and switched to IgG2b. These data are consistent with previous reports that extrachromosomal DNA from LPS-stimulated splenocytes contains switch circles with segments of both the γ3 and γ2b switch regions arranged in a manner consistent with a sequential switch (28). To assess whether there was an enhanced probability of IgG3+ cells undergoing switching to IgG2b, IgM+/IgG3 cells were also sorted from the same divisions and placed back in culture. Strikingly, these cells not only continued to proliferate in an identical manner to IgG3+ cells but switched to IgG2b at an identical frequency. The equal rate of switching to IgG2b suggests that an intermediate switch to IgG3 is neither obligatory nor detrimental for switching to IgG2b, and that the probabilities of switching to IgG2b and IgG3 are independent. In other words, cells switching to IgG3 will continue to be subject to the division-related possibility of gaining access to γ2b for recombination. As a consequence, isotype switching can properly be described as a stochastic process, whereby it is difficult, if not impossible, to predict the behavior of a single cell. However, the behavior of the population of cells as a whole is accurately described by probabilities as illustrated in Fig. 8 B.

This probabilistic view of isotype switching sheds some light on the question of how B cells can switch to downstream isotypes by different routes, for example from IgM directly to IgE, or via an IgG1 intermediate (21, 22). If switching to each isotype exhibits an independent division-linked probability under a given set of activation conditions, multiple allowable sequential switch pathways will be achieved at a predictable frequency, as observed here for IgG2b and IgG3. This model can explain why preventing the switch to IgG1 by gene targeting has no influence on the ability of B cells to switch to IgE (1, 29), despite the high frequency of IgE+ cells that usually switch via a γ1 intermediate (21). The independent division-linked control of switching to IgE may not be altered in any way by prior switching to another isotype. Therefore, it will be of considerable interest to further explore this prediction through experiments examining the probabilities underlying other sequential switch pathways.

The independent control of isotype switching and division rate also provides an insight into the broader functioning of B cells during an immune response. For example, IgG3 is induced by LPS, a component of Gram-negative bacteria (42, 43). This isotype, like IgM, is an excellent complement-fixing Ab, and is capable of protecting against bacterial infection (44). Furthermore, IgG3 possesses self-aggregating properties that may increase the avidity of interaction with multivalent epitopes present on bacteria (45). The mechanism whereby selection of this isotype operates can be illustrated by reference to LPS, which, in addition to being a nonspecific B cell mitogen, induces an Ag-specific Ab response (46). In this situation, Ag specificity is thought to depend upon the ability of B cells specific for LPS to focus the mitogen to the cell surface, thereby increasing the effective level of stimulation (47). The consequence is that a range of high- and low-affinity B cells are likely to be activated early in the response to LPS, resulting in rapid division and IgM secretion. As Ab accumulates, however, it will progressively block the ability of the lower-affinity B cells to concentrate LPS at the cell surface, leaving only B cells of higher affinity to continue dividing and to switch to IgG3. Such a scenario illustrates how the link between division number and isotype switching leads to the selection of isotypes with the capacity to function more effectively as a bivalent Ab.

A similar role for sustained Ag-driven B cell proliferation can be postulated for the regulation of IgA, which is typically produced at sites draining mucosal surfaces (48). Thus, stimulation of B cells with LPS or CD40L in the presence of TGF-β yielded only a small percentage of cells expressing IgA (49, 50). However, the percentage of cells within the Peyer’s patches expressing IgA is much higher (51), which led McIntyre et al. to pose the question of how this high proportion of IgA+ cells arose if TGF-β induces such a low rate of switching (52). According to the results presented here, the explanation is likely to reflect sustained stimulation of B cells by continual exposure to gut-derived Ags. Under these conditions, reactive B cells would be driven into the later divisions despite the antiproliferative effects of TGF-β, thereby ensuring a high rate of switching. The cytokine TGF-β produced at sites of chronic Ag exposure should therefore act both as a dampener of clonal expansion and as a facilitator of the switch to the secretory isotype.

We thank Tony Basten for his persistent and valuable reviews of the manuscript, Paul Lalor for Abs, Marilyn Kehry and Brian Castle for the CD40L, and Joseph Webster for assistance with cell sorting.

1

This work was supported by grants from the National Health and Medical Research Council and the Medical Foundation of the University of Sydney.

3

Abbreviations used in this paper: BCM, B cell medium; CD40L, CD40 ligand; CFSE, 5- (and 6-)carboxyfluorescein diacetate succinimidyl ester; SA, streptavidin.

1
Jung, S., K. Rajewsky, A. Radbruch.
1993
. Shutdown of class switch recombination by deletion of a switch region control element.
Science
259
:
984
2
Lorenz, M., S. Jung, A. Radbruch.
1995
. Switch transcripts in immunoglobulin class switching.
Science
267
:
1825
3
Bottaro, A., R. Lansford, L. Xu, J. Zhang, P. Rothman, F. W. Alt.
1994
. S region transcription per se promotes basal IgE class switch recombination but additional factors regulate the efficiency of the process.
EMBO J.
13
:
665
4
Stavnezer, J..
1996
. Antibody class switching.
Adv. Immunol.
61
:
79
5
Stavnezer, J..
1996
. Immunoglobulin class switching.
Curr. Opin. Immunol.
8
:
199
6
Coffman, R. L., D. A. Lebman, P. Rothman.
1993
. Mechanism and regulation of immunoglobulin isotype switching.
Adv. Immunol.
54
:
229
7
Snapper, C. M., J. J. Mond.
1993
. Toward a comprehensive view of immunoglobulin class switching.
Immunol. Today
14
:
15
8
Snapper, C. M., T. M. McIntyre, R. Mandler, L. M. Pecanha, F. D. Finkelman, A. Lees, J. J. Mond.
1992
. Induction of IgG3 secretion by interferon γ: a model for T cell-independent class switching in response to T cell-independent type 2 antigens.
J. Exp. Med.
175
:
1367
9
Snapper, C. M., L. M. Pecanha, A. D. Levine, J. J. Mond.
1991
. IgE class switching is critically dependent upon the nature of the B cell activator, in addition to the presence of IL-4.
J. Immunol.
147
:
1163
10
Stavnezer, J., G. Radcliffe, Y. Lin, J. Nietupski, L. Berggren, R. Sitia, E. Severinson.
1988
. Immunoglobulin heavy-chain switching may be directed by prior induction of transcripts from constant-region genes.
Proc. Natl. Acad. Sci. USA
85
:
7704
11
Snapper, C. M., W. E. Paul.
1987
. Interferon-γ and B cell stimulatory factor-1 reciprocally regulate Ig isotype production.
Science
236
:
944
12
Maliszewski, C. R., K. Grabstein, W. C. Fanslow, R. Armitage, M. K. Spriggs, T. A. Sato.
1993
. Recombinant CD40 ligand stimulation of murine B cell growth and differentiation: cooperative effects of cytokines.
Eur. J. Immunol.
23
:
1044
13
Snapper, C. M., C. Peschel, W. E. Paul.
1988
. IFN-γ stimulates IgG2a secretion by murine B cells stimulated with bacterial lipopolysaccharide.
J. Immunol.
140
:
2121
14
Severinson, E., C. Fernandez, J. Stavnezer.
1990
. Induction of germ-line immunoglobulin heavy chain transcripts by mitogens and interleukins prior to switch recombination.
Eur. J. Immunol.
20
:
1079
15
McIntyre, T. M., D. R. Klinman, P. Rothman, M. Lugo, J. R. Dasch, J. J. Mond, C. M. Snapper.
1993
. Transforming growth factor β1 selectivity stimulates immunoglobulin G2b secretion by lipopolysaccharide-activated murine B cells.
J. Exp. Med.
177
:
1031
16
Coffman, R. L., D. A. Lebman, B. Shrader.
1989
. Transforming growth factor β specifically enhances IgA production by lipopolysaccharide-stimulated murine B lymphocytes.
J. Exp. Med.
170
:
1039
17
Sonoda, E., R. Matsumoto, Y. Hitoshi, T. Ishii, M. Sugimoto, S. Araki, A. Tominaga, N. Yamaguchi, K. Takatsu.
1989
. Transforming growth factor β induces IgA production and acts additively with interleukin 5 for IgA production.
J. Exp. Med.
170
:
1415
18
Bergstedt-Lindqvist, S., C. Fernandez, E. Severinson.
1982
. A synergistic polyclonal response to dextran sulphate and lipopolysaccharide: immunoglobulin secretion and cell requirements.
Scand. J. Immunol.
15
:
439
19
Rothman, P., S. Lutzker, B. Gorham, V. Stewart, R. Coffman, F. W. Alt.
1990
. Structure and expression of germline immunoglobulin γ3 heavy chain transcripts: implications for mitogen- and lymphokine-directed class-switching.
Int. Immunol.
2
:
621
20
Severinson-Gronowicz, E., C. Doss, F. Assisi, E. S. Vitetta, R. L. Coffman, S. Strober.
1979
. Surface Ig isotypes on cells responding to lipopolysaccharide by IgM and IgG secretion.
J. Immunol.
123
:
2049
21
Siebenkotten, G., C. Esser, M. Wabl, A. Radbruch.
1992
. The murine IgG1/IgE class switch program.
Eur. J. Immunol.
22
:
1827
22
Baskin, G., K. B. Islam, B. Evengard, L. Emtestam, C. I. Smith.
1997
. Direct and sequential switching from μ to ε in patients with Schistosoma mansoni infection and atopic dermatitis.
Eur. J. Immunol.
27
:
130
23
Mills, F. C., M. P. Mitchell, N. Harindranath, E. E. Max.
1995
. Human Ig S γ regions and their participation in sequential switching to IgE.
J. Immunol.
155
:
3021
24
van der Stoep, N., W. Korver, T. Logtenberg.
1994
. In vivo and in vitro IgE isotype switching in human B lymphocytes: evidence for a predominantly direct IgM to IgE class switch program.
Eur. J. Immunol.
24
:
1307
25
Zhang, K., F. C. Mills, A. Saxon.
1994
. Switch circles from IL-4-directed ε class switching from human B lymphocytes: evidence for direct, sequential, and multiple step switch from μ to ε Ig heavy chain gene.
J. Immunol.
152
:
3427
26
Brinkman, V., S. Muller, C. H. Heusser.
1992
. T cell-dependent differentiation of human B cells: direct switch from IgM to IgE, and sequential switch from IgM via IgG to IgA production.
Mol. Immunol.
29
:
1159
27
Iwasato, T., H. Arakawa, A. Shimizu, T. Honjo, H. Yamagishi.
1992
. Biased distribution of recombination sites within S regions upon immunoglobulin class switch recombination induced by transforming growth factor β and lipopolysaccharide.
J. Exp. Med.
175
:
1539
28
Matsuoka, M., K. Yoshida, T. Maeda, S. Usuda, H. Sakano.
1990
. Switch circular DNA formed in cytokine-treated mouse splenocytes: evidence for intramolecular DNA deletion in immunoglobulin class switching.
Cell
62
:
135
29
Jung, S., G. Siebenkotten, A. Radbruch.
1994
. Frequency of immunoglobulin E class switching is autonomously determined and independent of prior switching to other classes.
J. Exp. Med.
179
:
2023
30
Mandler, R., F. D. Finkelman, A. D. Levine, C. M. Snapper.
1993
. IL-4 induction of IgE class switching by lipopolysaccharide-activated murine B cells occurs predominantly through sequential switching.
J. Immunol.
150
:
407
31
Snapper, C. M., F. D. Finkelman, D. Stefany, D. H. Conrad, W. E. Paul.
1988
. IL-4 induces co-expression of intrinsic membrane IgG1 and IgE by murine B cells stimulated with lipopolysaccharide.
J. Immunol.
141
:
489
32
Hodgkin, P. D., J. H. Lee, A. B. Lyons.
1996
. B cell differentiation and isotype switching is related to division cycle number.
J. Exp. Med.
184
:
277
33
Hasbold, J., A. B. Lyons, M. R. Kehry, P. D. Hodgkin.
1998
. Cell division number regulates IgG1 and IgE switching of B cells following stimulation by CD40 ligand and IL-4.
Eur. J. Immunol.
28
:
1040
34
Kehry, M. R., B. Castle.
1994
. Regulation of CD40 ligand expression and use of recombinant CD40 ligand for studying B cell growth and differentiation.
Semin. Immunol.
6
:
287
35
Hodgkin, P. D., M. R. Kehry.
1996
. Methods for polyclonal B lymphocyte activation to proliferation and Ig secretion in vitro. L. A. Herzenberg, and D. M. Weir, and L. A. Herzenberg, and C. Blackwell, eds. In
Weir’s Handbook of Experimental Immunology
Vol. 89
: Blackwell Science, Cambridge, MA.
36
Lyons, A. B., C. R. Parish.
1994
. Determination of lymphocyte division by flow cytometry.
J. Immunol. Methods
171
:
131
37
Stavnezer, J..
1995
. Regulation of antibody production and class switching by TGF-β.
J. Immunol.
155
:
1647
38
Cross, D., J. C. Cambier.
1990
. Transforming growth factor β1 has differential effects on B cell proliferation and activation of antigen expression.
J. Immunol.
144
:
432
39
Snapper, C. M., W. Waegell, H. Beernink, J. R. Dasch.
1993
. Transforming growth factor-β1 is required for secretion of IgG of all subclasses by LPS-activated murine B cells in vitro.
J. Immunol.
151
:
4625
40
Gett, A. V., P. D. Hodgkin.
1998
. Cell division regulates the T cell cytokine repertoire, revealing a mechanism underlying immune class regulation.
Proc. Natl. Acad. Sci. USA
95
:
9488
41
Hasbold, J., J. S. Hong, M. R. Kehry, and P. D. Hodgkin. 1999. Integrating signals from IFN-γ and IL-4 by B cells: positive and negative effects on CD40L-induced proliferation, survival, and division-linked isotype switching to IgG1, IgE, and IgG2a. J. Immunol. In Press.
42
Perlmutter, G. M., D. Hansburg, D. E. Briles, R. A. Nicolotti, J. M. Davie.
1978
. Subclass restriction of murine anti-carbohydrate antibodies.
J. Immunol.
121
:
566
43
Slack, J., G. P. Der-Balian, M. Nahm, J. M. Davie.
1980
. Subclass restriction of murine antibodies: the IgG plaque-forming cell response to thymus-independent type 1 and type 2 antigens in normal mice and mice expressing an X-linked immunodeficiency.
J. Exp. Med.
151
:
853
44
Briles, D. E., J. L. Claflin, K. Schroer, C. Forman.
1981
. Mouse IgG3 antibodies are highly protective against infection with Streptococcus pneumoniae.
Nature
294
:
88
45
Greenspan, N. S., J. N. Cooper.
1992
. Intermolecular cooperativity: a clue to why mice have IgG3?.
Immunol. Today
13
:
164
46
Coutinho, A., E. Gronowicz, W. Bullock, G. Möller.
1974
. Mechanism of thymus-independent immunocyte triggering.
J. Exp. Med.
139
:
74
47
Coutinho, A., G. Möller.
1975
. Thymus-independent B-cell induction and paralysis.
Adv. Immunol.
21
:
113
48
Mestecky, J., J. R. McGhee.
1987
. Immunoglobulin A (IgA): molecular and cellular interactions involved in IgA biosynthesis and immune responses.
Adv. Immunol.
40
:
153
49
Lebman, D. A., F. D. Lee, R. L. Coffman.
1990
. Mechanism for transforming growth factor β and IL-2 enhancement of IgA expression in lipopolysaccharide-stimulated B cell cultures.
J. Immunol.
144
:
952
50
Ehrhardt, R. O., W. Strober, G. R. Harriman.
1992
. Effect of transforming growth factor (TGF)-β1 on IgA isotype expression: TGF-β1 induces a small increase in sIgA+ B cells regardless of the method of B cell activation.
J. Immunol.
148
:
3830
51
Butcher, E. C., R. V. Rouse, R. L. Coffman, C. N. Nottenburg, R. R. Hardy, I. L. Weissman.
1982
. Surface phenotype of Peyer’s patch germinal center cells: implications for the role of germinal centers in B cell differentiation.
J. Immunol.
129
:
2968
52
McIntyre, T. M., M. R. Kehry, C. M. Snapper.
1995
. Novel in vitro model for high-rate IgA class switching.
J. Immunol.
154
:
3156