Report

Different B Cell Populations Mediate Early and Late Memory During an Endogenous Immune Response

See allHide authors and affiliations

Science  04 Mar 2011:
Vol. 331, Issue 6021, pp. 1203-1207
DOI: 10.1126/science.1201730

Abstract

Memory B cells formed in response to microbial antigens provide immunity to later infections; however, the inability to detect rare endogenous antigen-specific cells limits current understanding of this process. Using an antigen-based technique to enrich these cells, we found that immunization with a model protein generated B memory cells that expressed isotype-switched immunoglobulins (swIg) or retained IgM. The more numerous IgM+ cells were longer lived than the swIg+ cells. However, swIg+ memory cells dominated the secondary response because of the capacity to become activated in the presence of neutralizing serum immunoglobulin. Thus, we propose that memory relies on swIg+ cells until they disappear and serum immunoglobulin falls to a low level, in which case memory resides with durable IgM+ reserves.

Memory B cells are generated during the primary immune response to foreign antigens. This process is initiated when naïve B cells expressing surface immunoglobulin (Ig) bind the antigen in secondary lymphoid organs, receive signals from helper T cells, and proliferate (1). This proliferation produces short-lived immunoglobulin-secreting plasmablasts and germinal center cells, many of which switch their immunoglobulin constant region from IgM to IgG, IgA, or IgE and acquire somatic mutations in the variable region (13). Cells that acquire immunoglobulin mutations that improve antigen binding gain a survival advantage and emerge from the germinal center reaction as long-lived surface-switched immunoglobulin (swIg)+ memory cells, or surface Ig plasma cells that maintain serum immunoglobulin levels (4). After subsequent exposure to antigen, the memory cells proliferate rapidly and generate plasmablasts, which boost the amount of antigen-specific immunoglobulin in the serum to aid in antigen clearance (1, 4). There is, however, evidence for the existence of IgM+ memory B cells that have or have not passed through germinal centers or undergone somatic mutation (5).

Recently, genetic labeling of B cells that expressed activation-induced cytidine deaminase (AID), which is required for isotype switching and somatic mutation (6), suggested that IgM+ memory cells make up part of the memory B cell pool in mice (7). Whether these cells were antigen-specific was not addressed. Thus, the relative contribution of IgM+ B cells—especially those that may not express AID—to the antigen-specific memory pool remains unclear.

We sought to gain a comprehensive view of all memory B cells in normal mice by tracing the fate of antigen-specific precursors throughout the primary immune response on the basis of antigen-specificity alone without the complications related to the use of immunoglobulin transgenic mice (810). Phycoerythrin (PE) and allophycocyanin were chosen as model foreign antigens because their fluorescent properties allowed direct flow cytometric detection of B cells expressing complementary immunoglobulin (11, 12). Naïve PE-specific B cells could not, however, be detected in a conventional 106-cell sample of the 2 × 108 spleen and lymph node cells from a mouse that had never been exposed to PE (Fig. 1, A and B). To solve this problem, antigen-specific B cells from the entire spleen and lymph node cell sample were enriched with magnetic beads (13). Naïve PE-specific B cells, mainly of the CD43 CD21 CD23+ B2 phenotype (fig. S1A) were detected among the cells in a sample that bound to a magnetic column after staining with PE and magnetic beads coated with antibody against PE (Fig. 1C). The unbound cells generated a PE-specific antibody response when transferred into B cell–deficient hosts that was only 20% that of unfractionated spleen and lymph nodes, which suggested that about 80% of the naïve PE-specific B cell population was captured by the enrichment procedure (fig. S1B). The PE-specific B cells that were missed may have had immunoglobulin that bound PE with very low affinity. The enrichment approach revealed that naïve B6 mice contained ~20,000 PE-specific B cells (Fig. 1D) in the spleen and lymph nodes. In contrast, naïve mice contained only 4000 B cells specific for allophycocyanin (Fig. 1D), which indicated that preimmune populations specific for different antigens vary in size. PE-specific cells were not detected in PE-enriched samples from MD4 transgenic mice (14) that contain only monoclonal hen egg lysozyme-specific B cells (Fig. 1E), which demonstrated the specificity of the enrichment method.

Fig. 1

Detection of PE-specific B cells. (A) B cells were identified by flow cytometry in spleen and lymph node samples as cells that did not bind a cocktail of antibodies specific for CD4, CD8, CD11c, Gr1, or F480 (non-B cells) and expressed immunoglobulin heavy and light chains (H+L, both intracellular and extracellular). The cells with large amounts of immunoglobulin were plasmablasts. (B) Representative flow cytometric analysis of unenriched spleen and lymph node B cells from a naïve B6 mouse after staining with PE. (C) Representative flow cytometric analysis of spleen and lymph node B cells from a naïve B6 mouse after staining with PE and enrichment with anti-PE magnetic beads. (D) Number of PE- or allophycocyanin (APC)–specific B cells in individual naïve B6 mice. The bar indicates the mean. (E and F) Representative flow cytometric analysis of spleen and lymph node B cells from (E) a naïve MD4 Rag1–/– mouse or (F) a B6 mouse, 24 days after subcutaneous injection of 15 μg PE and CFA, that were stained with PE and enriched with anti-PE magnetic beads. The PE gate was drawn to exclude cells that bind antigen-antibody complexes [arrow in (F)]. (G) Combined data from multiple experiments showing the numbers of PE-specific B cells in the spleen and lymph nodes at the indicated times after subcutaneous injection of 15 μg PE and CFA (filled circles) or CFA alone (open circles) shown over the first 32 (left) or 450 (right) days. The mean and SEM (n = 3 to 6) are shown for all time points except the day 61 time point for PE and CFA-injected mice (where n = 2, and the mean is shown without error bars) and the day 52 and day 72 time points for mice injected with CFA alone (n = 1) (10).

PE-specific B cells increased dramatically in the draining lymph nodes of B6 mice after subcutaneous injection of PE in complete Freund’s adjuvant (CFA) but not after CFA alone (Fig. 1, F and G). Note that our gating strategy excluded the B220, PElow non-B cells (Fig. 1F) that were capable of capturing secreted immunoglobulins (15) (fig. S1C). The PE-specific B cells peaked at ~106 cells by day 13 then declined to a stable population of ~150,000 cells (Fig. 1G).

CD38, GL7, and immunoglobulin isotype switching were monitored to assess the cell types present in the PE-specific B cell population (16, 17). PE-specific B cells in naïve mice expressed IgM and CD38 but not GL7 and contained low amounts of intracellular immunoglobulin, as expected for naïve cells (Fig. 2, A to C). In contrast, the PE-specific population present 8 days after PE and CFA injection contained subsets that expressed IgM or swIg (Fig. 2D), which could be divided further into intracellular Ighigh plasmablasts (Fig. 2, E and G), CD38 GL7+ germinal center cells, and CD38+ GL7 naïve or memory cells (Fig. 2, F and H).

Fig. 2

Detection of PE-specific plasmablasts, germinal center cells, and memory B cells. PE-specific cells enriched from the spleen and lymph nodes of (A to C) a naïve B6 mouse or (D to H) a B6 mouse on day 8 after subcutaneous injection of PE and CFA, identified as shown in Fig. 1. (A and D) Flow cytometry–based gating strategy used to identify cells expressing surface IgM and/or surface and intracellular IgG1, IgG2c, IgG2b, IgG3, IgA, and IgE. (B, E, and G) Flow cytometry–based gating strategy used to identify Iglow nonplasmablasts (left) or Ighigh plasmablasts (PBs) (right) within the IgM+ or swIg+ populations. (C, F, and H) Flow cytometry–based gating strategy used to identify CD38+ GL7 naïve and/or memory (N/M) or CD38 GL7+ germinal center (GC) cells within the non-PB populations. (I to K) Combined data from multiple experiments showing the numbers of IgM+ (open symbols, solid line) or swIg+ (filled symbols, solid line) PE-binding plasmablasts (I), germinal center cells (J), or memory cells (K) in the spleen and lymph nodes after subcutaneous injection of PE and CFA on day 0. The mean and SEM (n = 3 to 6) are shown for all time points except days 5, 7, 24, 61, and 74, where n = 2, and the mean is shown without error bars. The mean titer and SEM (n = 4 to 10) of total PE-specific immunoglobulin in the serum are shown [dashed line in (I)] (10).

These populations underwent dramatic changes over time. PE-specific plasmablasts, both IgM+ and swIg+, peaked at ~20,000 cells between days 10 and 13, as PE-specific immunoglobulin appeared in the serum, and then declined rapidly by day 30 (Fig. 2I). IgM+ and swIg+ PE-specific germinal center cells peaked on day 13 at 400,000 cells, and declined slowly out to day 100 (Fig. 2J), probably because of the slow release of PE from the CFA emulsion (18).

CD38+ GL7 PE-specific swIg+ memory cells were first detected on day 6, peaked on day 30 at ~60,000 cells, and declined to very low numbers by day 450 (Fig. 2K). By day 7, CD38+ GL7 PE-specific IgM+ cells began to increase over naïve levels, peaked on day 10 at ~120,000 cells (Fig. 2K), and remained remarkably stable until at least day 450. To confirm that the CD38+ GL7 PE-specific IgM+ cells in primed mice were genuine memory cells, we labeled IgM+ B cells from CD45.1+ naïve mice with carboxyfluorescein succinimidyl ester (CFSE) and transferred them into naïve CD45.2+ hosts to track their division history. Nine days after PE and CFA injection, 75% of the CD45.1+ CD38+ IgM+ PE-specific B cells (Fig. 3, A and B) had lost CFSE (Fig. 3C) but were no longer blasts (Fig. 3D). Thus, both populations fit the definition of memory cells (19), having responded to antigenic stimulation, retained CD38 expression, and become quiescent. The difference in the life-span of the populations could not be explained by differential expression of the survival-promoting tumor necrosis factor receptor superfamily member 13c (BAFFR) (20) (fig. S2A). swIg+ cells, however, had higher amounts of tumor necrosis factor receptor superfamily member 13b (TACI), which may inhibit B cell survival (21).

Fig. 3

Properties of the PE-specific IgM+ memory B cell population. (A to D) PE-specific cells enriched from B6 recipients of CD45.1+ CFSE-labeled B cells, 9 days after subcutaneous injection of PE and CFA or CFA alone, identified as in Fig. 1. (A) Flow cytometry–based gating strategy used to identify CD45.1+ donor PE-specific B cells that did (right) or did not (left) express CD38. (B) Flow cytometry–based gating strategy used to identify CD38+,CD45.1+ PE-specific B cells that expressed IgM. (C) Flow cytometric analysis of CFSE in CD45.1+ PE-specific CD38+ IgM+ cells from mice injected with CFA alone (filled) or PE and CFA (black line). The frequency of CFSElow cells is indicated for PE and CFA-injected mice. (D) Representative flow cytometric analysis of forward light scatter of PE-specific IgM+ CD38+ naïve cells from mice injected with CFA alone (filled), or PE-specific CD38+ IgM+ memory cells (black line), or CD38 IgM+ cells from mice injected with PE and CFA (dashed line). (E) The number of mutations resulting in amino acid substitutions, in the CDR1 and two variable region heavy chain genes for each of 28 PE-binding IgM+ or IgG1+ clones from mice injected 25 days earlier with PE and CFA. Statistical significance was established with an unpaired Student’s t test. (F) Combined data from two independent experiments showing the ratio of PE fluorescence intensity divided by total immunoglobulin fluorescence intensity for naïve (gray filled square) IgM+ memory (open squares), swIg+ memory (black filled squares), IgM+ germinal center (open triangles), or swIg+ germinal center (black filled triangles) cells at the indicated times after PE and CFA injection. The mean and SEM (n = 3 to 4) are shown for all time points except day 0 and day 61, where n = 2, and the mean is shown without error bars (10).

Both swIg+ and IgM+ PE-specific memory cells were T cell–dependent (fig. S3). The populations differed, however, in that IgM+ memory cells retained IgD, expressed lower amounts of memory cell surface markers (22) (fig. S2B), and contained fewer mutations in complementarity-determining regions (CDRs) 1 and 2 of the heavy chain (Fig. 3E, and tables S1 and S2) than swIg+ memory cells. In accordance with the latter finding, swIg+, but not the IgM+ memory cell population, underwent affinity maturation, manifested by increased PE binding capacity, until the germinal center reaction peaked (Fig. 3F). Thus, the IgM+ memory cells showed less evidence of germinal center selection, affinity maturation, and differentiation than swIg+ memory cells.

Immunized mice were then challenged with an intraperitoneal injection of PE to assess memory B cell function. Mice that were primed with PE 320 days earlier and contained 100,000 PE-specific IgM+ and 2000 swIg+ memory cells produced 25,000 swIg+ plasmablasts but very few IgM+ plasmablasts or germinal center cells of either type (Fig. 4A). The number of swIg+ memory cells increased 150-fold over the prechallenge level, whereas the number of IgM+ memory cells increased less than twofold (Fig. 4A). Thus, antigen restimulation caused swIg+ memory cells to generate plasmablasts and more memory cells without new germinal center cells, whereas IgM+ memory cells responded poorly.

Fig. 4

Functional capabilities of memory B cells. (A to D) PE-specific cells enriched from the spleen and lymph nodes of B6 mice and identified as in Figs. 1 and 2. (A to C) Closed bars represent the numbers of plasmablasts on day 4 and germinal center cells or memory cells on day 14 after intraperitoneal injection of PE and CFA into (A) day 320 memory, (B) day 450 memory, or (C) naïve mice. White bars represent numbers before the intraperitoneal challenge injection. The mean and SEM (n = 3 to 4) are shown for all groups except for the day 320 memory cells after challenge, where n = 2, and the mean is shown without error bars (range = 102,678). (D) Numbers of PE-binding germinal centers present on day 12 to 14 in mice left untreated or injected with serum from mice immunized with PE or control antigens (chicken ovalbumin or allophycocyanin) on days 3 to 6 after antigen injection. Statistical significance was established with an unpaired Student’s t test. (E and F) Flow cytometric analysis of PE-specific cells of donor or recipient origin in the spleen and lymph nodes of CD45.1+ recipients of IgM+ (E) or swIg+ (F) CD45.2+ memory cells before and 10 to 14 days after PE and CFA challenge. Gates used to identify IgM+, swIg+, and germinal center cells are shown. Scatter plots show the percentage of the indicated cell types present in the PE-specific populations of donor or recipient origin in individual mice 10 to 14 days after PE and CFA challenge. (G) PE-specific cells derived from transferred IgM+ or swIg+ memory cells in individual recipient mice 5 days after PE and CFA challenge. Plasmablasts were identified as Ighigh cells. Statistical significance was established with an unpaired Student’s t test (10).

One possibility, however, was that IgM+ memory cells switched immunoglobulin isotype after challenge and contributed to the swIg+ progeny. This possibility was difficult to assess as long as swIg+ memory cells were present at the time of challenge. Therefore, the secondary response was tested in mice that were primed with PE 450 days earlier and contained 100,000 PE-specific IgM+ and scarcely any swIg+ memory cells. These mice generated very few swIg+ cells of any kind after challenge, which indicated that the IgM+ memory cells did not undergo isotype switching. The IgM+ memory cells increased only twofold after challenge (Fig. 4B), in contrast to the robust primary response of naïve IgM+ cells to intraperitoneal injection of PE, which generated many IgM+ and swIg+ germinal center and memory cells (Fig. 4C).

Several lines of evidence suggested that the poor secondary response of IgM+ memory cells was related to anti-PE immunoglobulin present during challenge. Injection of hyperimmune serum containing anti-PE immunoglobulin inhibited the generation of germinal center cells from naïve cells (Fig. 4D). In addition, purified IgM+ memory cells underwent isotype switching and germinal center cell formation when transferred into naïve recipient mice and then challenged with PE (Fig. 4E). In contrast, purified swIg+ memory cells responded to PE in naïve recipient mice, as they did in immune mice, by producing memory cells but very few germinal center cells (Fig. 4F). Notably, the presence of swIg+ but not IgM+ memory cells inhibited germinal center formation by the naïve cells of the adoptive recipients (Fig. 4, E and F). This inhibitory effect correlated with rapid production of plasmablasts by swIg+ but not IgM+ (Fig. 4G) memory cells after challenge. Thus, IgM+ memory cells were not intrinsically hyporesponsive in immune hosts but were functionally inhibited by antigen-specific immunogloulins produced either before challenge by plasma cells or after challenge by memory cell–derived plasmablasts. Inhibitory FcγRIIb (23) was probably not involved, because IgM+ and swIg+ memory cells expressed equal amounts of this receptor (fig. S2A).

PE- and allophycocyanin-specific naïve B cells accounted for about 1:5000 and 1:25,000 of all B cells in mice. These high frequencies are likely related to the presence of multiple epitopes on these large multimeric proteins (24). It will be of interest to use the enrichment approach to enumerate naïve B cells specific for monomeric antigens, although antigen mulitmerization may be required (25).

Naïve PE-specific B cells generated IgM+ memory cells after immunization with PE and the adjuvants CFA (Fig. 2K), lipopolysaccharide, or alum (fig. S4), which suggested that this is a general feature of the primary immune response. These memory cells had few mutations in their IgM molecules, which indicated inefficient selection in germinal centers. It is possible that these poorly mutated IgM molecules had a high enough natural affinity for PE to trigger memory cell differentiation before extensive somatic mutation could occur.

The remarkable stability of the IgM+ memory cells compared with swIg+ memory cells was not related to selective enrichment of IgM+ cells (fig. S5A), migration of swIg+ cells to bone marrow (fig. S5B), or homeostatic proliferation (fig. S5C). The instability of swIg+ memory cells may be related to inhibitory signals through TACI (fig. S2A) (21) or deleterious off-target mutations induced by AID (26). Despite being shorter-lived and outnumbered by IgM+ memory cells, swIg+ memory cells dominated the secondary response because of a capacity to be activated in the presence of high-affinity neutralizing serum immunoglobulin. However, even swIg+ memory cells could not produce germinal center cells, perhaps because their plasmablast progeny secreted enough immunoglobulin to clear the antigen very quickly. The failure to be activated efficiently in the face of immunoglobulin from swIg+ memory cells or plasma cells suggests that IgM+ memory cells do not contribute to the secondary response until these molecules decline. Serum immunoglobulin induced by certain subunit vaccines has been reported to decrease over time in humans (27), which suggests that IgM+ cells could become the reservoirs of humoral immune memory for these vaccines. Because of their lower affinity and ability to produce germinal center cells, IgM+ memory cells may also be useful for responding to antigenic variants produced by mutating pathogens.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1201730/DC1

Materials and Methods

Figs. S1 to S5

Tables S1 and S2

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank J. Walter and R. Speier for expert technical assistance and all members of the Jenkins lab for helpful discussions. This work was supported by grants from the NIH (R01 AI036914 and R37 AI027998 to M.K.J., F32 AI091033 and T32 CA009138 to J.T.), and the Intramural Research Program of the NIH, National Institute on Aging (R.W.M. and P.J.G.).
View Abstract

Navigate This Article