Immunization and Infection Change the Number of Recombination Activating Gene (Rag)-Expressing B Cells in the Periphery by Altering Immature Lymphocyte Production

Recombination activating genes 1 and 2 (RAG1 and RAG2) encode a lymphocyte-specific enzyme that catalyzes V(D)J recombination 1,2,3. In the B lymphoid lineage, RAG expression is primarily restricted to developing B cells in the bone marrow 4,5. However, low level RAG expression is also found in B cells in peripheral lymphoid organs, and several laboratories have reported increases in RAG expression in peripheral B cells after immunization 6,7,8,9. Two hypotheses have been put forward to account for this increase: RAG expression could be reinduced in B cells undergoing antigen-activated immune responses 6,7; alternatively, RAG expression in the periphery might reflect accumulation of immature B cells containing residual RAG mRNA 6,7,10.

To examine regulation of RAG2 expression in vivo, we and others produced mice that carry RAG2 green fluorescent protein (RAG2-GFP) indicator genes 9,10,11. Although there were several differences in the techniques used to produce the indicator strains, the results of the experiments were similar: RAG2-GFP expression was found in spleen B cells, and in both cases it was the immature B cells that expressed the indicator 9,10. However, the source of the increased RAG expression observed after immunization was not determined 9,10.

Here we report experiments showing that transient increases in RAG expression in the spleens of immunized mice result from changes in the production of immature B cells and their export from the bone marrow.

RAG2-GFP (FVB/N or [FVB/N × C57BL/6 (B6)] F1) mice 10, TNF knockout (TNF^−/−) 12, type I IFN receptor knockout (IFNR^−/−) 13, and signal transducer and activator of transcription (STAT)1 knockout (STAT1^−/−) 14 mice and matched controls were bred and maintained under specific pathogen-free conditions. B6, RAG1 knockout (RAG1^−/−), and 129S6/SvEv mice were purchased from The Jackson Laboratory or Taconic Farms. Mice were 2–5 mo old and were always age and sex matched.

400 μl 10% aluminum potassium sulfate with or without 100 μg of (4-hydroxy-3-nitrophenyl) acetyl (NP) coupled to chicken gamma globulin (NP-CGG; Biosearch Technologies, Inc.) was precipitated by adjusting pH to 6.2 with 1 N potassium hydroxide; the alum precipitates were washed in PBS three times and injected intraperitoneally. CFA (Sigma-Aldrich) was mixed with an equal volume of PBS, and 200 μl was injected intraperitoneally.

Plasmodium yoelii (17X NL strain) was maintained by alternating cyclic passage of the parasites in Anopheles stephensi mosquitoes and BALB/c mice. Before infecting B6 mice, parasites were passed in this strain by injecting the mice with parasite-infected BALB/c red blood cells (PRBCs). For experimental infections in B6 mice, 1 × 10³ PRBCs obtained from B6 mice were injected intraperitoneally, and parasitemia was determined by microscopic examination of Giemsa-stained thin blood smears 15.

Adoptive transfer was performed as described 10. 2–4 × 10⁷ splenocytes from wild-type B6 or RAG2-GFP ([FVB/N × B6]F1) mice were transferred by intravenous injection into B6 RAG1^−/− recipients.

The following anti–mouse antibodies were used: biotin, PE, or allophycocyanin (APC) anti-B220 (RA3-6B2); biotin or FITC anti-CD43 (S7); biotin or PE anti–heat-stable antigen (HSA) (M1/69); biotin anti-IgM (R6-60.2); biotin anti-Igλ1λ2 (R26-46); biotin anti-Igκ (R5-250); GL7; biotin anti-CD4 (H129.19); APC anti-CD8 (Ly-2); PE anti-Fas (Jo2); biotin anti–Ly-6G (RB6-8C5); PE DX5; FITC annexin V; and FcBlock (2.4G2) (all from BD PharMingen). Biotinylated antibodies were visualized with streptavidin–Red 613 (GIBCO BRL) or streptavidin-PerCP (Becton Dickinson). GL7 was visualized using PE- or biotin-labeled mouse anti–rat IgM. For DNA staining, cells were suspended in 20 μM Hoechst 33342 (Molecular Probes)/2% fetal bovine serum/PBS and incubated at 37°C for 40 min. Data were acquired with a FACSCalibur™ or FACS Vantage™ (Becton Dickinson) and analyzed with CELLQuest™ software (Becton Dickinson). Nucleated lymphocytes were electronically gated on forward and side scatter; for DNA analysis, doublets were excluded by area and width of FL-5 signal pulses. For RNA analysis, spleen cells stained with anti-B220, anti-HSA, and anti-IgM antibodies and fractions of cells (1–2 × 10⁵) were sorted directly into TRIzol LS (GIBCO BRL).

RNA was extracted using TRIzol (GIBCO BRL), and cDNA was synthesized using Superscript II reverse transcriptase (GIBCO BRL). Reverse transcription (RT)-PCR reactions were performed on serial dilutions of cDNA template using HotStarTaq polymerase (Qiagen). PCR conditions and primers were as described 10. Specificity of PCR products was confirmed by Southern blotting as described 16. The oligonucleotide probe used for Igβ was 5′-GTGACCTGCCACTGAATTTCCAAG-3′.

GFP⁺ B cells in the spleens of RAG2-GFP mice display the surface features of immature B cells (B220^lowHSA^high 493⁺) 10, whereas GFP⁻ B cells are mature B cells (B220^high HSA^low493⁻) 10,17,18. Upon transfer to RAG^−/− recipients, GFP⁺ B cells become GFP⁻ and home to lymph nodes and spleen, suggesting that they mature and enter the long-lived B cell compartment 10. To determine the kinetics of loss of GFP expression, we transferred spleen cells from RAG2-GFP indicator mice into RAG1^−/− recipients (Fig. 1 A). After adoptive transfer the number of GFP^high B cells decreased with a half-life of 52 h (Fig. 1 A), and after 4 d these cells were no longer found in RAG^−/− recipients. Similar kinetics of loss of GFP expression were found when GFP⁺ spleen B cells were cultured in vitro (10; and not shown). Therefore, it is unlikely that GFP⁺ cells preferentially migrate out from spleen. We conclude that RAG2-GFP is only expressed for a short time after B cells leave the bone marrow, and that GFP⁺ B cells are not generated from spleen B cell precursors.

To confirm that loss of RAG2-GFP expression correlates with loss of endogenous RAG mRNA, we measured RAG1 mRNA by RT-PCR. Spleen cells were transferred into RAG1^−/− mice, and B220⁺HSA^high cells were purified after 7 d. RAG1 mRNA could not be detected in the transferred B cells, whereas immature spleen B cells from a wild-type control were positive for RAG1 mRNA expression (Fig. 1 B). We conclude that RAG2-GFP expression after transfer correlates with endogenous RAG mRNA expression and that RAG-expressing immature B cells are not of splenic origin.

Experiments in several laboratories have shown that B cell RAG expression increases in peripheral lymphoid organs after immunization 6,7,8,9. To determine whether immunization also increases RAG2-GFP expression, we injected indicator mice with NP-CGG in alum and performed time course experiments (Fig. 2 A). Although the number of immature GFP⁺ B cells in the spleen initially decreased (34 ± 15% of control [n = 5]; Fig. 2 A), by day 16 their number had increased above the starting value (199 ± 38% of control [n = 6]; Fig. 2 A). Both the timing and magnitude of this transient increase in RAG expression are consistent with previous observations 6,7,9.

The GFP⁺ B cells in the spleen on day 16 after immunization were heterogeneous and showed an increased proportion of less mature IgM^low B cells (Fig. 2 A). In contrast, there was no shift in the surface IgM expression in GFP⁻ B cells with immunization. To confirm that changes in RAG2-GFP expression after immunization reflect changes in endogenous RAG mRNA expression, we compared RAG1 mRNA levels in GFP⁺ and GFP⁻ B cells purified from spleens of immunized and control mice. Endogenous RAG1 mRNA was found in GFP⁺ but not in GFP⁻ cells, and the level of RAG1 expression per cell increased after immunization (Fig. 2 B). The relative increase in RAG1 expression in GFP⁺ B cells may reflect a relative increase in less mature IgM^low cells, since these are known to express higher levels of RAGs than IgM^high immature B cells (Fig. 2 A, middle panels [9, 10]). We conclude that the number of RAG-expressing immature B cells in the spleen increases after immunization with NP-CGG in alum. Finally, the increase in RAG2-GFP expression in spleen at day 16 was transient and the number of GFP⁺ B cells returned to normal by day 28 after immunization (Fig. 2 A). Interestingly, the number of immature B lineage cell in the bone marrow also varied during the response, but GFP intensity was not significantly altered by immunization (Fig. 2 A, and data not shown).

As noted above, the majority of the GFP⁺ B cells in unimmunized spleens are not derived from splenic precursors. To determine whether immunization induces spleen B cells to express RAG2-GFP, we performed adoptive transfer experiments. Spleen cells from RAG2-GFP indicator mice were transferred into RAG1^−/− hosts that were immunized on day 1 after transfer. Although the immunized recipients showed an increase in spleen GL7⁺ B cells consistent with an ongoing immune response, we found no increase in GFP⁺ B cells at day 16 after adoptive transfer of spleen cells (Fig. 2 C). We conclude that the GFP⁺ B cells found in the spleen at day 16 after immunization are not generated from splenic precursors.

To determine whether GFP⁺ B cells in the spleen at day 16 after immunization were dividing, we measured their DNA content with Hoechst dye. In seven independent experiments, only 1.18 ± 0.52% of the GFP⁺ B cells had S or G2/M phase DNA content. Most of these cycling cells were found in a GL7^lowB220^low but Fas⁻ subpopulation of GFP⁺ cells (Fig. 3A and Fig. B; see gate G2). Therefore, most of the GFP⁺ B cells in the spleens of immunized RAG2-GFP indicator mice are not germinal center (GC) cells, which are predominantly GL7^high and Fas⁺ (Fig. 3A and Fig. B; see gate G1) 6,19,20.

As noted above, B lymphopoiesis is initially suppressed by intraperitoneal injection with NP-CGG in alum. To determine which stages of B cell development are affected, we immunized B6 mice and counted B cell progenitors in the bone marrow (Fig. 4 A). We found that pro-B cell, pre-B cell, and immature B cell numbers are all reduced by 5–20-fold by day 4 after immunization but all of these precursors recover by day 16. In contrast, mature B cell numbers in the spleen change little with immunization (not shown). We conclude that immunization with NP-CGG in alum initially suppresses B lymphopoiesis.

To determine whether the effect of NP-CGG plus alum is antigen dependent, we injected RAG2-GFP indicator mice and controls with alum alone. Such mice were indistinguishable from mice injected with NP-CGG in alum. The number of GFP⁺ B cells in the spleen dropped to 36 ± 12% of control on day 7 (n = 5) and then increased to 189 ± 26% of control on day 16 after alum injection (n = 5). These changes coincided with initial suppression of B lymphopoiesis followed by recovery at day 16. Furthermore, CFA had similar effects on B lymphopoiesis (Table). Thus, altered lymphopoiesis in NP-CGG and alum–injected mice is antigen independent and can be induced by either alum or CFA.

In addition to suppressing B lymphopoiesis, adjuvant injection also suppressed thymopoiesis. 4 d after administration of alum or CFA, the number of CD4⁺CD8⁺ thymocytes decreased 7–50-fold (Fig. 4 B, and Table). The loss of CD4⁺CD8⁺ thymocytes was associated with an increase in annexin V staining indicative of increased programmed cell death (Fig. 4 B). In contrast, the number of myeloid cells in the bone marrow increased while the number of erythroid cells was not significantly altered (Fig. 4 C). We conclude that administration of alum or CFA initially suppresses both T and B lymphopoiesis, that this effect is cell type and developmental stage specific, and that it correlated with an increase in thymocyte apoptosis 21.

IFNs and TNF are known to suppress hematopoiesis during certain viral infections 22,23. For example, type I IFNs are essential for suppression of hemato-lymphopoiesis after lymphocytic choriomeningitis virus (LCMV) infection 22. To determine whether suppression of lymphopoiesis by adjuvants is mediated by IFNs or TNF, we injected type I IFNR^−/−, TNF^−/−, and STAT1^−/− mice with alum and examined B and T cell development on day 4 12,13,14 (Fig. 5). Although we found significant differences in the magnitude of the adjuvant-induced suppression of lymphopoiesis in the different control mouse strains, the absence of type I IFNR, STAT1, or TNF had no effect on adjuvant-mediated suppression of B or T lymphopoiesis. We conclude that neither type I IFNR, nor STAT1, nor TNF is essential for the suppression of lymphopoiesis by adjuvant.

Adjuvants are thought to impact the immune system by some of the same mechanisms as natural infections. To determine whether lymphopoiesis is also suppressed during acute infection, we exposed mice to P. yoelii. At the peak of parasitemia, day 14 after infection, the number of B220^low immature B lineage cells in the bone marrow was severely reduced (Fig. 6), but the number of developing B cells recovered and even increased by day 28 when the malaria parasites had been cleared 24. We conclude that the development of immature B lineage cells is suppressed during malaria infection, as it is after administration of adjuvants.

RAG2-GFP indicator mouse strains produced by two different technologies have been used to examine RAG expression in vivo 9,10,11. In one case, GFP was targeted into RAG2 to produce a locus that directs the synthesis of a RAG2-GFP fusion protein 9. The dynamic range of GFP fluorescence in this strain is about one log over background and to enhance detection analyses must be performed in homozygous mutant mice 9. The second indicator strain carries a 165-kb bacterial artificial chromosome transgene in which GFP was inserted into the RAG2 gene 10. In this transgenic indicator strain, GFP is not produced as a fusion protein and fluorescence levels are much higher, with a dynamic range of three to four logs. The half-life and precise specific activity of the targeted RAG2-GFP fusion protein are not known, but a clear disadvantage of the transgenic system is that the half-life of the GFP indicator in this strain is 2–3 d, probably much longer than that of the endogenous RAG2 protein. However, despite the differences between the two indicator strains, and the long half-life of the transgenic GFP, the results reported for the two indicator strains are very similar 9,10. In both cases, there were two peaks of RAG2-GFP indicator gene expression in developing B cells corresponding to two waves of RAG transcription during heavy and light chain gene recombination 4,5,9,10. In addition, both indicator strains showed GFP expression in immature B cells in the bone marrow and spleen 9,10. The amount of endogenous RAG mRNA found in immature B cells was inversely correlated with the levels of surface IgM expression and up to two orders of magnitude lower than that found in pro- or pre-B cells 9,10. Immature B cells with the highest levels of surface IgM had exceedingly low levels of RAG mRNA, below the dynamic range of detection in the targeted indicator strain, and the significance of this low level of RAG expression has yet to be determined.

Several groups have shown that immunization increases RAG expression in B cells in peripheral lymphoid organs, raising the possibility that RAG could be reinduced in mature B cells 6,7,8,9. In our initial experiments, we found that RAG expression was not reinduced in mature B cells after immunization and we were unable to induce RAG expression in vitro with LPS plus IL-4 10. Nevertheless, there is an increase in the number of RAG2-GFP–expressing B cells in spleens of mice injected with alum. Most of these cells are not of splenic origin, they are nondividing cells that carry the cell surface markers of immature B cells. Only a subset of the GFP⁺ B cells in the spleen is rapidly proliferating, and a small subfraction of these rapidly dividing cells expresses high levels of both GL7 and Fas, which are markers of GC B cells 19,20. Therefore, most GFP⁺ B cells found in the spleen on day 16 after injection with alum are immature, and only a very small number of these cells have the characteristics of GC cells. It has been proposed that RAG expression in peripheral B cells might contribute to further diversification of the antibody repertoire in GCs during immune reactions 6,7,8,25,26. Our results suggest that antibody gene replacement in GCs is a rare event. However, inflammatory or pathological processes that lead to accumulation of immature B cells might increase the frequency of such events 16,27.

What is the origin of the RAG2-GFP–expressing cells that accumulate in the spleen after immunization? Immature B cells are thought to be produced by the bone marrow at a steady rate of 1–2 × 10⁷/d, and homeostasis maintained by selecting a variable fraction of these cells into the long-lived B cell compartment 28,29,30,31,32,33. However, our experiments suggest that the increase in RAG2-GFP–expressing B cells found in the spleen after immunization is due to an increase in immature B cell production and export 12 d after initial suppression of B lymphopoiesis by adjuvant. This rebound effect resembles the increase in immature B cell production and emigration from the bone marrow found 12–14 d after sublethal irradiation 17 and differs from the augmentation of pre-B cell production observed 4 d after injection with sheep RBCs 34. We conclude that, like irradiation, adjuvants and infection affect lymphopoiesis and alter the rates of immature B cell production and export from the bone marrow.

Transient suppression of B and T lymphopoiesis has been reported during infection with LCMV and after IFN injection, but these agents produced a more general suppression of hematopoiesis not found with adjuvant injection 22,35. In LMCV infection, bone marrow suppression is mediated by type I IFNs 22, but neither type I IFNs, nor STAT1, nor TNF were essential for the effect produced by alum injection.

What is the significance of altering lymphopoiesis after administration of adjuvant or during infection? Alterations in lymphopoiesis could be limited to malaria and LCMV, and to administration of CFA or alum, and therefore have little physiological significance. However, the documented suppressive effects of inflammatory cytokines, IFNs, and TNF on lymphopoiesis 36,37,38,39 suggest that any infectious or inflammatory agent that significantly increases the production of these cytokines and possibly other such effectors may alter lymphopoiesis. Therefore, subsets of T and B cells recently exported from the bone marrow or thymus may resemble myeloid cells, which display well-characterized shifts during infection and inflammation. Such shifts in the composition of the lymphocyte populations might change the outcome of immune responses. In particular, immature B cells recently emigrating from the bone marrow are highly susceptible to tolerance and less likely to enter the memory pool 40,41,42,43,44,45. Thus, decreasing the number of such cells exported in acute infection may protect against tolerance and indirectly enhance B cell memory.

We thank Dr. Ruth S. Nussenzweig and members of the M.C. Nussenzweig laboratory for comments on the manuscript and helpful discussions, Dr. David E. Levy for providing type I IFNR^−/− and STAT1^−/− mice, and Dr. Michael W. Marino for TNF^−/− mice.

This work was supported by grants from the National Institutes of Health to M.C. Nussenzweig and to M. Tsuji (AI40656 and AI33890). M.C. Nussenzweig is an Investigator in the Howard Hughes Medical Institute.