Expression Levels of B Cell Surface Immunoglobulin Regulate Efficiency of Allelic Exclusion and Size of Autoreactive B-1 Cell Compartment

Tg mouse lines carrying either H or L chain gene for the anti-RBC mAb (4C8 mAb) have been established previously 18. Double Tg (H×L) mice with H and L chain transgenes were obtained by mating H and L chain Tg mice. In this study we generated new lines of 4C8 mAb Tg mice that carried tandem joined H and L chain transgenes. To construct tandem joined transgenes, a 5.6-kb BamHI fragment of pMO-κ4C8 was first subcloned in the BamHI site of pSP73 Vector (Promega Corp.). Next, a 15.5-kb XhoI fragment of pMO-μ4C8 was subcloned between the SalI and XhoI sites. The pSP73 plasmid containing both H and L chain genes was designated as pSP73-κμ4C8. Tandem joined H and L chain Tg (H+L) mice were generated by injecting a 22.0-kb PvuI–XhoI fragment of pSP73-κμ4C8 (see Fig. 1 A). Heterozygous mice were mated to get homozygous mice in two representative lines, H+L5, with 4–12 copies, and H+L6, with 1–3 copies of the transgene. The presence of the transgenes and homozygosity of the Tg loci were screened by PCR and Southern blot analysis. Tg mice were maintained under conventional conditions in our animal facility.

Mice were typed by PCR analysis of tail DNA with set of primers to the H and L chain genes. PCR was carried out for 30 cycles consisting of denaturation (for 1 min at 94°C), annealing (for 1 min at 55°C), and extension (for 2 min at 72°C). The primer pairs for PCR were as follows: 5′Hc, 5′-CTACGCATTTAGTAGTGACTGG-3′; J_H3, 5′-TGCAGAGACAGTGACCAGAG-3′; IgL5′-2, 5′-CTGCAAGTCCAGTCAGAGCC-3′; IgL3′-2, 5′-CAGCACCGAACGTGAGAAAG-3′.

FITC- or Cy-Chrome^®–conjugated anti–mouse CD45R (B220), FITC- or PE-conjugated anti–mouse IgM^a (Igh-6^a), FITC- or PE-conjugated anti–mouse IgM^b (Igh-6^b), PE-conjugated anti–mouse CD11b (Mac-1 α chain), biotin-conjugated anti–mouse λ₁ and λ₂ L chain mAbs, and FITC- or Cy-Chrome^®–conjugated streptavidin (SA) were purchased from PharMingen. FITC- or biotin-conjugated F(ab′)² fragments of goat anti–mouse IgM were from ICN Pharmaceuticals (Cappel). PE-conjugated SA was from Dako. Anti-idiotype (Id) mAb (S54) against the transgene-encoded anti-RBC antibody (4C8) 18 was conjugated with N-hydroxysuccinimidobiotin (Pierce Chemical Co.) according to the directions provided by the manufacturer.

Single cell suspension from bone marrow was prepared by flashing femur bones with a staining buffer (PBS containing 2% FCS and 0.05% sodium azide), gently pipetted, and washed with the staining buffer. Single cell suspensions from bone marrow, spleen, and peritoneal cavity were prepared with the staining buffer and pretreated with 10% heat-inactivated normal rat serum. After 15 min of incubation, FITC- and/or biotin-conjugated mAbs at appropriate dilutions in the staining buffer were added directly. After 30 min of incubation, PE-conjugated mAbs and/or FITC- or PE-conjugated SA was added. Before and after each step of incubation, cells were washed with the staining buffer. Stained cells were then applied to FACSCalibur^® (Becton Dickinson). After excluding dead cells by propidium iodide staining, cells present in the lymphocyte gate defined by forward and side light scatters were analyzed.

To assess whether the levels of transgene expression are higher in homozygous than in heterozygous mice, we analyzed the bone marrow cells of H+L5 and H+L6 mice that carried tandem joined H and L chain transgenes cording for the 4C8 anti-RBC antibody (Fig. 1 A). In H×L mice, B220⁺IgM⁺ bone marrow cells almost completely disappeared 18. In contrast, both lines of H+L mice had significant numbers of B220⁺IgM⁺ cells in bone marrow (Fig. 1 B). The majority of these cells were stained both with antiallotypic mAb for IgM^a (Tg) and with antiidiotypic mAb (S54) for the anti-RBC antibody. We compared the fluorescence intensity of Id⁺ bone marrow cells in homozygous mice to that in heterozygous mice (Fig. 1 C). The mean fluorescence intensities (MFI) of Id⁺ cells in heterozygous and homozygous H+L5 mice are 481.70 and 717.42, respectively. Similarly, the MFI of Id⁺ cells in heterozygous and homozygous H+L6 mice are 997.76 and 1670.11, respectively. These results indicate that in both H+L5 and H+L6 lines the levels of transgene expression are clearly higher in homozygous than in heterozygous mice. We also examined surface Ig expression on the bone marrow cells of H3 Tg mice that had only H chain transgene. Fluorescence intensity of antiallotypic mAb for IgM^a (Tg) of bone marrow cells was also stronger in homozygous (MFI = 908.77) than in heterozygous (MFI = 319.25) mice. In a parallel experiment, anti-IgM^a mAb MFI of homozygous H+L5 and H+L6 were 173.07 and 250.72, respectively, indicating that homozygous H+L B cells had not saturated the capacity of surface Ig expression (Fig. 1 D). Taken together, homozygosity of Tg Ig loci enhances the level of transgene expression on B cell surface.

To test whether inhibition of endogenous H chain expression is stronger in homozygous than in heterozygous mice, we analyzed lymphocytes in bone marrow and spleen of H+L5, H+L6, and H3 mice by flow cytometry. We stained bone marrow and spleen cells with antiallotypic mAbs (which can clearly distinguish IgM of the control mice; Fig. 2A and Fig. B) for IgM^a (Tg) and IgM^b (endogenous). In all heterozygous Tg mice, the numbers of IgM^a+IgM^b+ cells (allelic inclusion) were small but significant, whereas these cells almost completely disappeared in all homozygous Tg mice. This conclusion is further confirmed by the finding that both bone marrow and spleen B cells with only endogenous H chains (IgM^a−IgM^b+) decreased in homozygous Tg mice as compared with heterozygous Tg mice (Fig. 2A and Fig. B). In all lines of Tg mice, total numbers of endogenous H chain–expressing (IgM^b+) spleen cells were lower in homozygous than in heterozygous mice (Fig. 2 C). These results indicate that inhibition of endogenous H chain expression (allelic exclusion) is stronger in homozygous than in heterozygous mice. Taken together, higher levels of Ig transgene expression appear to cause stronger inhibition of endogenous H chain expression, suggesting that a certain level of surface Ig expression is required for allelic exclusion of the H chain locus.

Next, we asked whether homozygosity of Tg loci also reduces endogenous L chain expression. Since the allotypic specificity for the Tg Lκ chain is not known, we estimated the degree of endogenous L chain expression by difference in expression of Tg H chain (IgM^a) and Id (the combined epitope of the H and L chains of the 4C8 anti-RBC mAb) that is recognized by the S54 mAb. In the H+L5 line, two populations of IgM^a+ cells were observed: one stained with both anti-IgM^a and anti-Id mAbs diagonally, and the other stained more weakly with anti-Id mAb than with anti-IgM^a mAb (Fig. 3a, Fig. B, and Fig. D, tops). As a control, IgM^a+ cells from H3 mice did not contain the population stained with the S54 mAb (Fig. 3B and Fig. D, bottoms). We presume that the cells that stained diagonally with both mAbs express the Tg H and L chains equally, and that the IgM^a+Id^low cells express the Tg H chain with both endogenous and Tg L chains. In support of this assumption, IgM^a+ bone marrow cells in heterozygous H+L6 mice, which consisted of almost exclusively IgM^a+Id⁺ cells and only few IgM^a+Id^low cells, contained <1.2% λ⁺ cells, which represent a fraction of B cells that express endogenous L chains (Fig. 3 A). In contrast, IgM^a+ bone marrow cells in heterozygous H+L5 mice, which consisted of ∼28% IgM^a+Id⁺ and ∼72% IgM^a+Id^low cells, contained >5.5% λ⁺ cells (Fig. 3 A). Furthermore, IgM^a+ spleen cells in heterozygous H+L5 mice, which consisted of ∼4% IgM^a+Id⁺ and ∼96% IgM^a+Id^low cells, contained >14% λ⁺ cells (Fig. 3 A). The total number of cells expressing endogenous L chains (IgM^a+Id^low) in bone marrow (Fig. 3B and Fig. C) and spleen (Fig. 3D and Fig. E) was less in homozygous than in heterozygous Tg lines, indicating that homozygosity of the Tg loci enhances the inhibition of endogenous L chain expression as compared with heterozygosity.

H×L mice show almost complete deletion of autoreactive B cells from bone marrow and spleen, whereas a normal level of B-1 cells is found in the peritoneal cavity 18. Bone marrow of both lines of H+L mice contained some autoreactive B cells (IgM^a+Id⁺), which have B-2 cell phenotypes (CD5⁻B220^highIgM^low), whereas autoreactive B cells in spleen almost completely disappeared in H+L mice (Fig. 3B and Fig. D). We examined whether homozygosity of the Ig transgene loci influences the size of the autoreactive B-1 cell compartment in the peritoneal cavities of H+L5 and H+L6 mice. Peritoneal cells of H+L mice had a limited number of B220⁺Id⁺ cells, almost all of which were IgM^a+ (Fig. 4A–C and Fig. E–G). Almost all Id⁺ cells in the peritoneal cavity were Mac1⁺ and therefore belong to the B-1 subset (Fig. 4D and Fig. H). Thus, IgM^a+Id⁺ cells are most likely to be autoreactive B-1 cells (Fig. 4C and Fig. G). These B220⁺Id⁺ cells in the peritoneal cavity were much more abundant in homozygous H+L mice than in heterozygous H+L mice, indicating that the size of the autoreactive B-1 cell subset is larger in homozygous than in heterozygous H+L mice. The relative size of the B-1 cell compartment is directly proportional to the B-1 cell number because total cell numbers in the peritoneal cavity did not vary between homozygous and heterozygous Tg mice. These results suggest that a higher expression level of the autoantibody facilitates the increase in autoreactive B-1 cells in the peritoneal cavity.

To test whether autoreactivity of Ig expressed on B cells is crucial to enlargement of the B-1 cell subset, we carried out similar studies on the peritoneal B cells in H3 Tg mice, the vast majority of which are not autoreactive due to the absence of the Tg L chain. The majority of peritoneal B cells expressed the Tg H chain (IgM^a+), in agreement with analysis of bone marrow and spleen cells (Fig. 4I and Fig. J). Indeed, B cells expressing endogenous H chains were less in homozygous than in heterozygous H3 mice (Fig. 4 J), probably because of more efficient allelic exclusion due to stronger expression of the transgene product on homozygous B cells (Fig. 1 C). Nonetheless, the percentage of IgM⁺Mac1⁺ cells in the peritoneal cavity did not increase in homozygous H3 mice as compared with those in heterozygous H3 mice (Fig. 4 K). These results indicate that increased surface Ig expression of non-auto-reactive B cells did not facilitate enlargement of the B-1 cell subset in the peritoneal cavity. Taken together, enhancement of autoreactive surface Ig expression appears to be crucial to increase the B-1 cell subset in the peritoneal cavity.

In this study, we generated new lines of the anti-RBC antibody Tg mice that carry tandem joined H and L chain transgenes, and compared B cell phenotypes between heterozygous and homozygous Tg mice to evaluate quantitatively the effect of surface Ig level on allelic exclusion and B-1 cell development. Higher levels of surface Tg Ig expression resulted in (a) stronger clonal deletion of B-2 cells from bone marrow and spleen; (b) reduction of B cells expressing endogenous H or L chains; and (c) enlargement of the autoreactive B-1 cell subset in the peritoneal cavity.

Several lines of evidence suggest that surface expression of the μ chain is critical for H chain allelic exclusion 2,4,5,6. In addition, pre-BCR, which is the H chain paired with the surrogate L chain, is suggested to mediate H chain allelic exclusion through downregulation of recombination-activating genes 42,43,44,45,46. Igα and Igβ, which associate with surface-expressed H chains, have also been shown to trigger the signals inducing allelic exclusion 47,48,49. Taken together, expression of the H chain as a pre-BCR on the cell surface may induce signals mediated by Igα and Igβ, resulting in H chain allelic exclusion. In this study we have demonstrated that homozygosity of the transgene Ig loci increases surface Ig expression on bone marrow B cells and causes stronger allelic exclusion, as compared with heterozygosity. Since allelic exclusion is all or none in each B cell, the present results suggest that there is a threshold of the pre-BCR signal intensity that induces allelic exclusion.

There are several other possibilities to explain our results. First, B cells with allelic inclusion may be negatively selected. However, Sonoda and Rajewsky 50 have shown that B cells with allelic inclusion can normally expand in the periphery using double Ig knock-in mice, indicating the absence of negative selection against allelic inclusion B cells. In addition, we have shown that endogenous only B cells (IgM^a−IgM^b+) are also reduced in homozygotes (Fig. 2). Second, reduction of allelic inclusion B cells could be due to selective expansion of higher surface Ig-expressing cells by a self-antigen. However, the self-antigen (RBC) can kill self-reactive B cells 19. In fact, Id⁺ B cells of H+L mice decreased in homozygotes as compared with heterozygotes in spleen as well as in bone marrow (Fig. 3B and Fig. D). In addition, we have shown that non–self-reactive (H chain alone expressing) Tg (H3) B cells also show suppression of endogenous Ig expression (Fig. 2). These results cannot be explained by positive selection of self-reactive Ig expressing B cells by the self-antigen. Third, excess Tg Ig expression on surface may inhibit detection of endogenous Ig expression. It is unlikely that endogenous and Tg Ig molecules compete for surface expression in H+L5 and H+L6 spleen cells. This is because the surface Ig levels of homozygous H3, H+L6, and H+L5 cells are 909, 251, and 173 (MFI using the anti-IgM^a Ab), respectively (Fig. 1 D), indicating that H+L5 and H+L6 B cells have not hit the ceiling of the surface Ig expression level. Sonoda and Rajewsky 50 have shown that there is no inhibitory mechanism for the H chain on the surface as long as it can associate with the L chain. It is inconceivable that the surface expression efficiency differs between H chains derived from the transgene and endogenous gene. The possibility that a lower detection efficiency by anti-IgM^b Ab staining in the presence of large amounts of IgM^a is also unlikely because the FACS^® profiles of H3 spleen cells expressing large amounts of IgM^a clearly show the presence of IgM^a IgM^b double-positive cells even in homozygotes (Fig. 2 B). Finally, the suppression efficiency of the endogenous locus may be increased in homozygotes simply because the frequency of silencing the transgene locus is reduced by doubling the number of the transgene locus in homozygotes. Assuming that this is the case, the efficiency of silencing of the transgene locus can be calculated from the endogenous H chain only cells shown in Fig. 2 A: H3, 0.06 / (0.06 + 0.11 + 6.50) = 0.009 in heterozygotes and 0.00144 = 0.01 / (0.01 + 0.03 + 6.9) in homozygotes. The value in homozygotes is ∼18 times of the expected value (0.009 × 0.009 = 0.000081) based on the simple statistics. Similar discrepancy was seen in the H+L5, H+L6, and H3 spleen cells (Fig. 2 B). In addition, homozygosity alters the number of Tg Id⁺ B cells in spleen and peritoneal cavity to the opposite direction. These considerations make the final possibility unlikely.

We have shown that H+L Tg mice have B cells expressing endogenous L chains together with the Tg H chain. These cells, expressing endogenous L chains in H+L mice, may be generated either by incomplete allelic exclusion or by receptor editing because the B cells which express the transgenes are self-reactive 51,52. Nemazee and colleagues 53,54,55 have suggested that interaction with autoantigens leads IgM^lowIgD⁻ bone marrow cells to undergo receptor editing but IgM^highIgD⁺ cells to undergo rapid apoptosis. On the other hand, Rusconi et al. 16 generated B cell hybridomas from anti-trinitrophenol antibody (anti-non-self) Tg mice that carried tandem joined H and L chain transgenes and demonstrated that the B cell hybridomas secreting the Tg antibody expressed the Tg L chain at about one-tenth of the level of coexpressed endogenous L chains. Endogenous L chain expression in their Tg mice is probably due to incomplete allelic exclusion because these B cells are unlikely to receive BCR stimulation by self-antigens to trigger receptor editing. Although we cannot exclude the possibility that receptor editing is involved in the appearance of B cells with endogenous L chains in our Tg mice, we think it less likely because of the following reason. If the expression of endogenous L chain is due to receptor editing induced by stimulation with the self-antigen, the mechanism to reduce the number of B cells with endogenous L chains by enhanced expression of self-reactive Ig should be clonal deletion. However, when B cells expressed more self-reactive Ig on surface, B cells with both Tg and endogenous L chains (IgM^a+Id^low) are more efficiently reduced than Tg only B cells (IgM^aId⁺), which are most likely eliminated by clonal deletion (Fig. 3). This observation is somewhat opposed to the expected efficiency of clonal deletion by the self-antigen because stronger BCR signaling will be induced in IgM^aId⁺ cells than IgM^aId^low cells.

Although it is still controversial whether B-1 cells belong to an ontogenetically different B cell lineage from conventional B cells 34,35,36,37,38,39,40, B-1 and B-2 cells clearly constitute different subsets of B cells. In this study we have shown that the size of the Tg B-1 cell compartment is larger in homozygous than in heterozygous H+L mice (Fig. 4). Since Tg B-1 cells in heterozygous and homozygous mice show the same antigen specificity, our results suggest that the level of surface Ig expression directly influences the size of the Tg B-1 cell compartment. Our findings are consistent with the previous observations that defects of BCR signaling cause reduction of the B1 cell 56,57, and loss of a BCR inhibitory molecule, SHP-1, increases the B1 cell number 58,59,60,61,62. It is important to note that increased levels of autoreactive BCR induced augmented clonal deletion of B-2 cells in bone marrow and spleen but expansion of B-1 cells in the peritoneal cavity (Fig. 4). Increased expression of nonautoreactive Ig (H3) enhanced neither clonal deletion of B-2 cells nor expansion of B-1 cells. These results suggest that there are at least three levels of BCR signaling that regulate self-reactive B1 and B2 cell differentiation. At a lower level, self-reactive B-2 cells can be stimulated to induce receptor editing or to become anergic 53,54,55,63. At an intermediate signaling level, B-2 cells are clonally deleted and B-0 40,41 and/or B-2 cells are induced to differentiate into B-1 cells, which migrate into the peritoneal cavity. At a strong signaling level, B-1 cells are also clonally deleted 19. It is tempting to speculate that at least a sizable fraction of peritoneal B-1 cells originate and expand from autoreactive B cells that are stimulated by self-antigens to a level strong enough to be activated but weak enough to avoid apoptosis.

We thank Ms. Y. Kobayashi and Ms. T. Taniuchi for their technical assistance, and Ms. Y. Takahashi for manuscript preparation.

This work was supported by grants from the COE program from the Ministry of Education, Science, Sports and Culture of Japan.