The agonists of TLR4 and 9 are sufficient to activate memory B cells to differentiate into plasma cells in vitro but not in vivo

Summary

Memory B cells can persist for a lifetime and be reactivated to yield high affinity, isotype switched plasma cells. The generation of memory B cells by antigen immunization requires adjuvants that generally contain Toll-like receptor (TLR) agonists. However, requirements for memory B cell activation and the role of TLRs in this activation are not well understood. Here we analyzed the response of memory B cells from immunized mice to TLR9 and 4 agonists CpG ODN and LPS. Mouse memory B cells express both TLR9 and TLR4 and respond to both CpG ODN and LPS in vitro by differentiating into high affinity IgG secreting plasma cells. In contrast, neither CpG ODN nor LPS alone is sufficient to activate memory B cells in vivo. Antigen is required for the clonal expansion of antigen-specific memory B cells, the differentiation of memory B cells to high affinity IgG secreting plasma cells, and the recall of high affinity antibody responses. The antigen-specific B cells that have not yet undergone isotype switching showed a relatively higher expression of TLR4 than memory B cells, which was reflected in a heightened response to LPS, but in both cases yielded mostly low affinity IgM secreting plasma cells. Thus, although memory B cells are sensitive to TLR agonists in vitro, TLR agonists alone appear to have little affect on B cell memory in vivo.

INTRODUCTION

Immunological memory is a hallmark of adaptive immunity. The humoral branch of immunological memory consists of memory B cells, which are the precursors of high affinity antibody secreting cells (ASCs), and long-lived plasma cells, which maintain serum antibody levels independent of antigenic stimuli. Recent studies have begun to characterize these two important humoral memory components (see reviews (1–4)). In the mouse system, to date no distinct surface markers identify memory B cells. In general, memory B cells exhibit the characteristics of post germinal center B cells. They express B cell antigen receptors (BCRs) that have undergone isotype switching, somatic hypermutation, and affinity maturation, and thus bind to antigens with high affinities. IgM-expressing memory B cells with somatic hypermutation have been reported in human (5, 6), however, whether these cells are IgM-expressing memory B cells or marginal zone B cells remains controversial (7). Both long-lived plasma cells and memory B cells persist up to the lifetime of an individual as demonstrated by persistent antibody levels and the ability to mount rapid and robust antibody response to antigen challenge years after immunization. Memory B cells undergo rapid clonal expansion and differentiation to mount high affinity antibody responses upon exposure to antigens. Long-lived plasma cells are terminally differentiated and continue secreting antibodies without antigenic stimulation in the bone marrow that provides the necessary environment for their longevity (4, 8, 9).

The requirements for the generation and maintenance of memory B cells has been studied in some detail (10). The generation of both memory B cells and long-lived plasma cells requires germinal center reaction, although germinal center-independent generation of memory B cells has also been reported (11, 12). Both antigen and the interaction of B cells with T_H cells in the context of antigenic peptide-MHC class II complex and through CD40-CD40L are essential for the generation of germinal centers and memory B cells (13, 14). In contrast, the maintenance of the B cell memory seems to be independent of antigen and T_H cells (15–17). Crotty et al. reported that small pox vaccine-specific memory B cells in human were detected 60 years post immunization and 30 years after small pox was eradicated worldwide (15). Maruyama et al. showed that antigen-specific memory B cells persisted in transgenic mice where memory B cells switched their antibody specificity away from the immunizing antigen (17). Furthermore, mouse memory B cells have been shown to persist in the absence of T_H (16) or follicular dendritic cells (18).

Although the requirements for the generation of B cell memory are becoming better understood, the requirements for the activation of memory B cells are still not clear. The rapid and robust responses of memory B cells to antigenic challenge suggest a low signaling threshold for memory B cell activation. Bernasconi et al. reported that Toll like receptor (TLR) agonists polyclonally activated human memory B cells to proliferate and differentiate into plasma cells in vitro, and that tetanus toxoid immunization increased serum antibodies to unrelated antigens, suggesting that TLRs play a role in the polyclonal activation and long-term maintenance of human memory B cells in the absence of antigen in vivo (19).

The ability of common microbial products to increase the efficacy of immunization has long been observed (20). The discovery of TLRs reveals that some of the major components of commonly used adjuvant contribute to immune responses through TLRs. TLRs are responsible for the initiation of innate immune responses and the maturation of dendritic cells that activate T cells, triggering adaptive responses (21, 22). The importance of TLRs in humoral immune responses was demonstrated using MyD88-knockout mice, where T_H cell activation and T-dependent antibody responses were reduced or completely abolished (23). In addition to their role in activating T_H cells, TLR agonists can directly act on B cells. Mouse naïve B cells express TLRs, including TLR2, 4, 7, and 9, and proliferate and differentiate into plasma cells in response to TLR agonists (24–27). In contrast, human naïve B cells do not express TLRs, but are induced to express TLRs in response to antigen stimulation through the BCR (19, 28). Human memory B cells constitutively express TLRs, including TLR2, 6, 7, 9 and 10, and proliferate and differentiate into plasma cells in response to TLR agonists alone in vitro. Using MyD88-knockout mice and LPS as adjuvant, Pasare and Medzhitov (29) demonstrated that TLR signaling in B cells was required for optimal antibody responses to T-dependent antigens. In contradiction, using Myd88^−/−-/Trif^Lps2/Lps2 mice, Gavin et al. (30) showed that both T cell-dependent and independent antigens induced comparable humoral immune responses in the absence of TLR signaling. Thus, the exact role of TLR in the activation humoral memory responses requires further examination.

In this study, we examined the role of TLR9 and TLR4 in the activation of memory B cells in vitro and in vivo using NP-KLH immunized mice as a model. We show that TLR4 and 9 agonists alone promote the differentiation of memory B cells into high affinity IgG ASCs in vitro, however, TLR agonists alone are not sufficient to effectively activate humoral memory responses in vivo.

MATERIALS and METHODS

Immunization

To generate humoral memory responses, C57BL/6 mice of 6–8 weeks old (Charles River Laboratories, Inc. Wilmington, MA) were immunized twice, 28 days apart, with 400 µg/mouse NP₁₉-KLH (NP, 4-hydroxy-3-nitrophenylacetyl; KLH, keyhole limpet hemocyanin) (Biosearch Technologies, Novato, CA) and Ribi adjuvant (MPL + TDM Adjuvant System, Sigma, St. Louis, MO). To activate humoral memory responses in vivo, the immunized mice were rested for more than 6 weeks and boosted for the third time using 400 µg/mouse NP₁₉-KLH and/or one of the following adjuvants: Ribi adjuvant, LPS (50 µg/mouse) from E. coli (Sigma) in oil in water (PBS) emulsion, or class B CpG containing phosphorothioated oligodeoxynucleotide (CpG ODN) for mouse (50 µg/mouse) (5’-TCCATGACGTTCCTGACGTT-3’, Operon Biotechnologies, Inc., Huntsville, AL) in oil in water (PBS) emulsion.

B cell isolation

Single cell suspension of splenocytes was treated with ACK lysis buffer to remove erythrocytes. After washing, cells were incubated with magnetic beads coated with anti-mouse CD4, CD8, CD11b, and CD11c antibodies (Miltenyi Biotec, Auburn, CA), and cells binding to the beads were removed by AutoMACS (Miltenyi Biotec).

Flow cytometry analysis

Purified B cells were first incubated with anti-mouse CD16/CD32 mAb (BD Bioscience, San Diego, CA) to block Fcγ receptors, followed by FITC-anti-mouse IgD, FITC-anti-mouse IgM (Southern Biotech, Birmingham AL), PE-anti-mouse CD138, PerCP-Cy5.5-anti-mouse B220 antibodies (BD Bioscience) and NP₁₉-APC at 4°C in FACS buffer (1% FBS, 20 mM EDTA, 0.02% NaN₃ in PBS). To label the surface TLR4, cells were incubated with PE-anti-mouse TLR4 antibody (BD Bioscience) or anti-mouse TLR4 antibody (IMGENEX, San Diego, CA) followed by a pacific blue-conjugated secondary antibody. To label TLR9, cells that were labeled with FITC-anti-IgD, FITC-anti-IgM, PerCP-Cy5.5-anti-B220 antibodies and NP₁₉-APC were washed, fixed with 4% paraformaldehyde, and permeabilized with a permeabilization buffer (0.1% saponin, 10 mM HEPES, 10 mM glycine in DMEM). Cells were incubated with anti-mouse TLR9 antibody (IMGENEX) in the permeabilization buffer, followed by a pacific blue- or Alexa Fluor 750-conjugated secondary antibody. After washing with the permeabilization buffer and PBS, cells were fixed with 1% paraformaldehyde and analyzed using CyAN™ flow cytometer (Dako, Carpinteria, CA).

Cell sorting

Purified B cells were first incubated with anti-mouse CD16/CD32 mAb to block Fcγ receptors, followed by FITC-anti-mouse IgD, FITC-anti-mouse IgM, PE-anti-mouse CD138, PerCP-Cy5.5-anti-mouse B220 antibodies and NP₁₉-APC in 2% FBS/PBS. After washing, cells were sorted using FACS Aria (BD Bioscience) into four different populations: B220⁻ IgD⁻ IgM⁻ CD138⁺ NP⁻ as plasma cells, B220⁺ IgD⁻ IgM⁻ CD138⁻ NP⁺ as memory B cells, B220⁺ IgD⁺ IgM⁺ CD138⁻ NP⁺ as NP-binding unswitched B cells, and B220⁺ IgD⁺ IgM⁺ CD138⁻ NP⁻ as non-NP-binding unswitched B cells.

B cell in vitro stimulation

Splenic B cells isolateed from mice 5 days or 6 weeks after the second immunization and sorted plasma and B cell subpopulations were incubated at 5×10⁵ cells/ml for 5 days with graded concentrations of CpG ODN, LPS, or Gardiquimod (InvivoGen, San Diego, CA) at 37°C, 5%CO₂ before ELISPOT analyses.

ELISPOT analysis

Assay plates (Millipore, Billerica, MA) were coated with 10 µg/ml of NP₃-BSA, NP₃₀-BSA (Biosearch Biotechnologies), or anti-mouse IgG+M antibody (Jackson ImmunoResearch, West Grove, PA) over night at 4°C and blocked with 10% FBS in PBS for 2 h at 37°C. Cells were serially diluted into wells and incubated at 37°C 5% CO₂ for 5 h. Plates were washed with PBS and PBS containing 0.05% Tween 20. Antibody secreted by cells were detected with biotin-labeled anti-mouse IgM antibody (Jackson ImmunoResearch) or a mixture of biotin-labeled anti-mouse IgG₁, IgG_2a, IgG_2b, and IgG₃ antibodies (Southern Biotech), followed by HRP-conjugated streptavidin (KPL, Gaithersburg, MD), and visualized by HRP substrate True Blue (KPL). Plates were scanned by Cellular Technologies Ltd (CTL, Cleveland, OH) and spots were counted with software provided by CTL.

ELISA analysis

ELISA plates (Nalge Nunc International, Rochester, NY) were coated with 40 ng/ml of NP₅-BSA, NP₃₀-BSA (Biosearch Biotechnologies), or 5 µg/ml of anti-mouse IgG+M antibody (Jackson ImmunoResearch, West Grove, PA) in 50 mM NaHCO₃ (pH 9.6) over night at 4°C and blocked with PBS containing 0.3% milk, 1% FBS, and 0.1% Tween 20 for 1 h at 37°C. Mouse serum was serially diluted into wells and incubated at room temperature for 2 h. After washing, the plates were incubated with a mixture of equal amount of HRP-conjugated anti-mouse IgG₁, IgG_2a, IgG_2b, and IgG₃ antibodies (Southern Biotech), and visualized by 2,2’-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (Sigma) and 0.03% hydrogen peroxide in phosphate-citrate buffer, pH 5.0. The reaction was stopped after 5 min with 0.5% SDS. Absorbance at 405 nm was measured using a plate reader (Wallac Victor, PerkinElmer). Titers were determined as highest serum dilution that give a value that was ≥3 standard deviation above the average reading for secondary controls.

RESULTS

B cell responses to TLR4, TLR7 and TLR9 agonists in vitro

The ability of splenic B cells from NP-KLH immunized mice to differentiate into NP-specific IgG and IgM high affinity antibody secreting cells (ASCs) when cultured with TLR4, TLR7, and TLR9 agonists was determined. Mice were immunized with 400 µg/mouse NP₁₉-KLH plus Ribi adjuvant, a LPS-based adjuvant, twice, four weeks apart. B cells were purified from spleens of NP-KLH immunized mice 5 days after the second immunization and cultured with graded concentrations of CpG ODN or LPS for 5 days (Fig. 1A–D). The number of NP-specific antibody ASCs was determined by ELISPOT using NP₃-BSA as the coating antigen to capture only high affinity antibodies and NP₃₀-BSA to capture both high and low affinity NP-specific antibodies. In the absence of CpG ODN or LPS in the culture medium, the splenic B cells did not differentiate into a significant number of NP-specific ASCs (Fig. 1A–D). When cultured with CpG ODN or LPS, splenic B cells differentiated into NP-specific, IgG ASCs, a significant portion of which exhibited high affinity for NP (Fig. 1A–B). CpG ODN induced relatively fewer NP-specific IgM ASCs (Fig. 1C) than LPS (Fig. 1D). However, the majority of these ASCs showed low affinity for NP (Fig. 1C–D). The numbers of NP-specific ASCs generated in the B cell cultures peaked at 0.25 µg/ml of CpG ODN and 0.6 µg/ml LPS (Fig. 1A–D), suggesting that the stimulatory effect of TLR agonists CpG ODN and LPS was saturable. To test the presence of the precursor B cells for high affinity IgG ASCs in mice long after the immunization, splenic B cells from mice 6 weeks after the second immunization were cultured with graded concentrations of CpG ODN, LPS, or Gardiquimod for 5 days. Similar to the splenic B cells isolated at 5 days (Fig. 1A–D), when cultured with TLR4, 7, or 9 agonist, splenic B cells isolated at 6 weeks post the second immunization differentiated into NP-specific IgG ASCs, most of which exhibited high affinity (Fig. 1E–G). TLR4 agonist induced a greater number of NP-specific IgM ASCs than TLR7 and 9 agonists, however, a significant portion of those exhibited low affinity (Fig. 1H–J). These results are consistent with previous reports that showed that TLR agonists can stimulate the differentiation of B cells into plasma cells (25, 27). Our results further showed that TLR4, 7, and 9 agonists induced similar numbers of high affinity IgG ASCs from splenic B cells from 5 days and 6 weeks post immunization, but TLR4 agonist predominantly activated splenic B cells to differentiate into low affinity IgM ASCs.

Figure 1. Differentiation of B cells to antigen-specific ASCs in response to TLR4, 7, and 9 agonists in vitro.

Splenic B cells were isolated from mice 5 days (A–D) or 6 weeks (E–J) after the second immunization and cultured with different concentrations of CpG ODN (A, C, E, and H), LPS (B, D, F, and I), or Gardiquimod (G and J) for 5 days. The numbers of NP-specific IgG (A–B and E–G) and IgM (C–D and H–J) ASCs were determined by ELISPOT using plates coated either with NP₃₀-BSA for all NP-specific ASCs (Total) or NP₅-BSA for high affinity NP-specific ASCs (High affinity). Shown are the representative results of three independent experiments.

B cells with memory phenotype differentiate into high affinity IgG secreting cells in response to TLR4 and 9 agonists in vitro

Based on the general properties of memory B cells that have been described so far (1, 3), we defined antigen-specific memory B cells as isotype-switched, antigen-binding B cells (B220⁺ IgD⁻ IgM⁻ CD138⁻ NP⁺). Using flow cytometry, NP-specific memory B cells in the spleen were undetectable in nonimmune mice (frequency of 0.03%) (Fig. 2). Five days following the second immunization, the frequency of NP-specific memory B cells in the spleen reached ~ 0.5% of splenic B cells (Fig. 2). Six weeks after the second immunization, the frequency of NP-specific memory B cells in the spleen declined to ~0.1%, which was still higher than that of nonimmune mice. The frequency of NP-specific memory B cells remained at this level for several months (data not shown). The frequency of NP-binding, unswitched B cells (B220⁺ IgD⁺/IgM⁺ CD138⁻ NP⁺) was 1.1% in nonimmune mice, increased to 6.2% following immunization, and decreased to 1.6%, near nonimmune levels by 6 weeks following the second immunization (Fig. 2) in contrast to memory B cells. The relatively high frequency of these B cells in nonimmune mice as compared to the predicted frequency suggests that they may be very low affinity cells. These data show that immunization induced rapid increases in the frequencies of NP-specific B cells with the memory phenotype in the spleen. While the frequencies of memory B cells decreased with the time after the second immunization, a significant number of memory B cells persisted in the spleen for several months.

Figure 2. The frequencies of NP-specific memory B cells in the spleen of immunized mice.

Splenic B cells were isolated from unimmunized mice (nonimmune) and mice 5 days and 6 weeks post the second immunization with NP-KLH and Ribi adjuvant. Cells were stained with FITC-anti-IgD, FITC-anti-IgM, PE-anti-CD138, PerCP-Cy5.5-anti-B220 antibodies, and APC-NP₁₉ and analyzed using a flow cytometer. Shown are representative data from 8 independent experiments.

B cells from NP-KLH-immunized mice 5 days after the second immunization were sorted into four different populations (Fig. 3A): (I) B220⁻ IgD/IgM⁻ NP⁻ CD138⁺ as plasma cells, (II) B220⁺ IgD/IgM⁻ NP⁺ CD138⁻ as NP-specific memory B cells, (III) B220⁺ IgD/IgM⁺ NP⁺ CD138⁻ as NP-binding unswitched B cells, and (IV) B220⁺ IgD/IgM⁺ NP⁻ CD138⁻ as unswitched B cells that were not specific for NP. Population III, NP-binding unswitched B cells, might consist of antigen experienced, unswitched B cells and NP-binding marginal zone B cells. Population IV, unswitched B cells that were not specific for NP, may contain naïve B cells and marginal zone B cells specific for unrelated antigens. Before the in vitro culture, only population I, plasma cells, contained NP-specific ASC (Fig. 3B and 3E), however, these did not survive in the culture (Fig. 3C–D and 3F–G). After culture in vitro with CpG ODN and LPS, B cells in population II, memory B cells, differentiated exclusively into NP-specific IgG ASCs, a large portion of which were high affinity (Fig. 3C–D and 3F–G). In contrast, populations III and IV containing unswitched B cells generated mostly low affinity IgM ASCs in response to CpG ODN and LPS (Fig. 3C–D and 3F–G). As expected, the NP-binding population of unswitched B cells generated a higher number of NP-specific IgM ASCs as compared to the non-NP-binding population (Fig. 3F and 3G). Response of the unswitched B populations to LPS was nearly ten-fold greater than to CpG ODN (Fig. 3F and 3G). These results demonstrate that TLR agonists can directly activate memory B cells in vitro to differentiate into high affinity, isotype switched plasma cells and unswitched B cells to differentiate into low affinity, IgM ASCs.

Figure 3. B cells with the memory phenotype differentiated into antigen-specific, high affinity IgG ASCs in culture with CpG ODN and LPS.

(A) Splenic B cells from NP-KLH immunized mice were sorted into four populations: B220⁻ IgM⁻ IgD⁻ NP⁻ CD138⁺ (I, plasma), B220⁺ IgM⁻ IgD⁻ NP⁺ CD138⁻ (II, memory), B220⁺IgM⁺IgD⁺ NP⁺ CD138⁻ (III, NP-binding, unswitched), and B220⁺ IgM⁺ IgD⁺ NP⁻ CD138⁻ (IV, unspecific, unswitched). Naïve B cells (V) were isolated from the spleen of unimmunized mice. The frequencies of high affinity and total NP-specific IgG (B–D) and IgM (E–F) ASCs were determined by ELISPOT before (B and E) and after culture in CpG ODN (2 µg/ml) (C and F) or LPS (5 µg/ml) (D and G) for 5 days. Shown are representative results from three independent experiments.

Differential expression of TLR4 and TLR9 in populations of B cells

The differential responses of the B cell populations to TLR4 and 9 agonists shown in Fig. 3 suggest different expression levels of TLR9 and TLR4. To test this hypothesis, the levels of expression of TLR4 and TLR9 in the different populations of splenic B cells isolated from NP-KLH immunized mice 5 days after the second immunization were determined by flow cytometry. The analyses showed higher TLR9 expression in NP-binding B cell populations (II and III) compared to naïve B cells (population V) from nonimmune mice (Fig. 4, left panels). In contrast, the TLR4 surface levels of naïve (population V) and unswitched B cells (populations III and IV) were significantly higher than that of NP-specific memory B cells (population II) (Fig. 4, right panels). These differences in expression levels provided an explanation for the equivalent response of memory B cells to CpG ODN and LPS and the hyper-response of NP-binding unswitched B cells to LPS.

Figure 4. Expression levels of TLR9 and TLR4 in different subpopulations of splenic B cells.

Splenic B cells were isolated from unimmunized mice (nonimmune) and mice that were immunized with NP-KLH plus Ribi adjuvant. The splenic B cells were preincubated with the FcγR blocking antibody, followed by PerCP-Cy5-anti-B220, FITC-anti-IgD, FITC-anti-IgM antibodies, APC-NP₁₉, and PE-anti-TLR4-MD2 antibodies for staining TLR4. For staining TLR9, the cells were washed, fixed, permeabilized, and stained with anti-TLR9 antibody and a pacific blue-conjugated secondary antibody. The cells were analyzed by flow cytometry. B220^high B cell populations of nonimmune mice (V), NP-specific memory B cells (II), NP-binding, unswitched B cells (III), and non-specific, unswitched B cells (IV) of immunized mice were gated, and their TLR4 (right panels)- and TLR9 (left panels)-staining levels are shown. Shown are representative results from 5 independent experiments.

The activation of B cell memory responses by TLR4 and 9 agonists in vivo

To determine if TLR4 and 9 agonists can stimulate memory B cells in vivo, mice were immunized with NP₁₉-KLH plus Ribi adjuvant twice, 4 weeks apart, to establish a pool of NP-specific memory B cells. Six weeks later, the mice were challenged with NP₁₉-KLH, Ribi adjuvant, CpG ODN, LPS alone, or NP-KLH plus either Ribi adjuvant, CpG ODN, or LPS. Five days after the challenge, high and low affinity serum titers of NP-specific IgG and the number of high affinity, NP-specific IgG ASCs in spleens were determined. A certain level of high affinity serum titer (Fig. 5A) and a small number of high affinity, NP-specific IgG ASCs (Fig. 5B) existed in the spleens before the boost, indicating the persistence of NP-specific IgG ASCs in the spleen even 6 weeks after an immunization. The challenge with antigen alone significantly increased (~10 fold) the serum titer of high affinity NP-specific IgG and the number of high affinity, NP-specific IgG ASCs (Fig. 5A and 5B). However, the challenge with adjuvant alone, including Ribi adjuvant, CpG ODN and LPS, did not significantly change either the serum titer of high affinity NP-specific IgG or the number of high affinity NP-specific IgG ASCs, in comparison with those before the challenge (Fig. 5A and 5B). In contrast, challenging with antigen plus adjuvant dramatically increased the number of the high affinity IgG ASCs, with ~56, ~46 and ~16 fold increases for antigen plus Ribi, LPS and CpG ODN adjuvant, respectively (Fig. 5B). LPS or Ribi appeared more effective than CpG ODN as the adjuvant under this specific immunization condition. At 5 days post the third immunization, the serum titers of mice challenged with CpG or LPS plus antigen were significantly higher than those that did not received the third immunization and those that were challenged with CpG or LPS alone, but lower than those that were challenged with antigen alone (Fig. 5A). This did not completely mirror what was observed for the numbers of NP-specific ASCs (Fig 5B) and may reflect different kinetics and timing for B cell differentiation and for NP-specific plasma cells to exit from spleens in mice challenged with antigen alone and antigen plus adjuvant.

Figure 5. Clonal expansion and differentiation of memory B cells in response to immunization of antigen or TLR agonists alone or in combination.

C57BL/6 mice were twice immunized with NP-KLH plus Ribi adjuvant 4 weeks apart and rested for 6 weeks. The mice were immunized for the third time with NP-KLH (400 µg/ml), Ribi adjuvant, CpG ODN (50 µg/mouse) or LPS (50 µg/mouse) alone or in combination. Controls included nonimmune mice and mice that did not receive the third immunization (None). (A) Serum was collected from one day before and five days after the challenge and analyzed for NP-specific high affinity and total IgG titers. Shown are average titers (±SE, n = 3~25). The serum titers of mice challenged with NP-KLH plus the adjuvant were significantly higher than those of mice that did not receive the third immunization or received the adjuvant alone for the third immunization. ***, p ≤ 0.001. *, p ≤ 0.05. (B–C) B cells and plasma cells were enriched and frequencies of NP-specific IgG ASCs were determined by ELISPOT before (B) and after (C) cultured with CpG ODN (2 µg/ml) in vitro for 5 days. (D) The percentage of NP-specific memory B cells among the splenic B cells was determined by flow cytometry, as described in Figure 2. Shown are the representative results of four independent experiments.

Taking advantage of the ability of memory B cells to specifically differentiate into high affinity IgG ASCs in culture with CpG ODN, we estimated the relative frequencies of NP-specific memory B cells in spleen after the different challenges. A significant number of NP-specific, high affinity IgG ASCs was detected in the 5-days culture of B cells from mice rested for 6 weeks without the challenge (Fig. 5C), indicating the persistence of memory B cells in the spleen long after an immunization. The number of high affinity IgG ASCs differentiating from B cells from mice challenged with antigen or adjuvant alone was similar to or lower than that of B cells from mice that were not challenged (Fig. 5C). The number of high affinity IgG ASCs differentiating from B cells from mice challenged with antigen plus adjuvant was 5 to 20 fold higher than those of mice without the challenge (Fig. 5C). Among different adjuvants, antigen plus either Ribi or LPS generated the largest increase. To further confirm this result, we determined the frequency of NP-specific memory B cells by flow cytometry. The challenge with antigen plus adjuvant, but not antigen or adjuvant alone, induced significant increases (5 to 20 fold) in the numbers of NP-specific memory B cells (Fig. 5D). Ribi or LPS as adjuvants appeared to be more effective than CpG ODN as an adjuvant in expanding the NP-specific memory B cell pool (Fig. 5D).

DISCUSSION

Humoral memory responses consist of two major components, long-lived plasma cells that secrete high affinity antibody and memory B cells, the precursors of long-lived plasma cells. In this study, we examined the role of TLR agonists in activation of humoral memory responses in vitro and in vivo by tracking these two major components of the humoral memory responses. We found that TLR4, 7 and 9 agonists could directly stimulate the differentiation of memory B cells into high affinity plasma cells in vitro, however, TLR agonist alone was not sufficient to stimulate the clonal expansion and differentiation of memory B cells and specific antibody recall responses in vivo.

Our study showed that when cultured with TLR4, 7, and 9 agonists, the splenic B cells from NP-KLH immunized mice, either 5 days or 6 weeks post the second immunization, differentiated into NP-specific ASCs with most of IgG ASCs as high affinity and most of IgM ASCs as low affinity. Using sorted memory B cells, we further showed that among different populations of B cells cultured with TLR4 or 9 agonist, NP-specific memory B cells exclusively differentiated into high affinity, isotype switched ASCs, and NP-binding unswitched B cells mainly differentiated into low affinity IgM ASCs. This demonstrates that TLR agonists can directly act on memory B cells in vitro and polyclonally stimulate the differentiation of memory B cells into high affinity, isotype switched plasma cells in the absence of T cells. In contrast, TLR agonists stimulated unswitched B cells in vitro primarily into low affinity, unswitched ASCs. Specific differentiation of sorted memory B cells into high affinity IgG ASCs indicates that they are the precursors of high affinity plasma cells and confirms that B cells sorted based on the phenotypic characteristics consist of mainly antigen-specific memory B cells. Furthermore, the differentiation specificity in the in vitro culture with TLR agonists provides us a good tool to determine the frequency of memory B cells in a mixture of multi-subpopulations of B cells. The finding that splenic B cells isolated from mice 5 days and 6 weeks post the immunization responded to TLR4 and 9 agonists in a similar manner suggests the persistence of antigen-specific memory B cells in mice.

The experimental definition of memory B cells is still controversial, especially in the murine system. McHeyzer-William et al. (31, 32) established a multi-color flow cytometry method to follow NP-specific memory B cells in NP-KLH-immunized mice based on their phenotypic characteristics of post germinal center B cells (B220⁺IgD⁻IgG₁⁺NP⁺) and demonstrated that B cells with such phenotypes exhibited a high frequency of point mutations in their variable regions. Anderson et al. recently suggested that CD80 and CD35 expression levels could be used to define B cell subtypes that have or have not undergone somatic hypermutation (33). Slifka and Ahmed (34) established an in vitro culture system to determine the frequency of acute lymphocytic choriomeningitis virus-specific memory B cells using the combination of limiting dilution and ELISPOT. Taking advantage of the ability of CpG ODN to stimulate the differentiation of memory B cells into high affinity IgG ASCs in vitro, we determined the frequency of memory B cells using this in vitro culture system, in addition to flow cytometry. Both flow cytometry analysis and the in vitro culture system showed dramatic increases in the number of antigen specific memory B cells 5 days after the second immunization and maintenance of a small, but significant number of memory B cells 6 weeks after the second immunization.

In addition to NP-specific, isotype switched B cells, a population of NP-binding, unswitched B cells were found in the immunized mice. This population of B cells was expanded in response to the second immunization, but returned to the basal level of unimmunized mice by 6 weeks. When cultured with TLR4 and 9 agonists, they differentiated into NP-specific IgM ASCs, most of which exhibited low affinity for the antigen. While the exact nature of the NP-specific unswitched B cells is unclear, it may contain IgM-expressing memory B cells. The existence of IgM-expressing memory B cells is still controversial. The discovery of a subpopulation of CD27⁺ IgM⁺, somatic mutated B cells in human peripheral blood (5) suggests the presence of IgM expressing memory B cells. Weller et al. showed that this population of B cells were circulating spleen marginal zone B cells that mutated their immunoglobulin receptors during ontogeny, prior to their differentiation into T-independent antigen-responsive cells (7). The role of this population of B cells in humoral memory responses remains to be further defined.

Our studies present a first comparison of B cell responses to TLR4 and 9 agonists in vitro and in vivo. Our results showed that TLR4 and 9 agonists were equally effective in stimulating memory B cells to differentiate, but TLR4 agonist was much more efficient than TLR9 agonist in stimulating unswitched B cells from immunized mice to differentiate in vitro. This differential response can be explained by different expression levels of TLR4 and 9 in different subpopulations of B cells. We found that the immunization increased the expression levels of both TLR4 and TLR9 in mouse B cells, and the expression level of TLR9 was maintained and the expression of TLR4 down regulated when activated B cells differentiated into memory B cells. Our finding of an increase in TLR9 expression level by immunization is similar to the report of Bernasconi et al. that human memory B cells, but not naïve B cells, constitutively express TLR9 (28). The biological significance of this differential expression of TLR9 and TLR4 among different B cell subpopulations remains to be explored. The higher expression level of TLR4 in naïve and activated, unswitched B cells than memory B cells predicts that LPS preferentially induces primary humoral responses.

The direct stimulatory effect of TLR9 and TLR4 agonists on memory B cells in vitro showed in this study and previous studies (19, 28) suggests their potential role in polyclonal activation of memory B cells in vivo. Our finding that TLR9 or TLR4 agonist alone failed to increase the frequencies of either antigen-specific high affinity IgG ASCs or antigen-specific memory B cells as well as the concentration of antigen-specific IgG in serum does not support the polyclonal activation hypothesis. Our result that antigen in combination with TLR9 or TLR4 agonist was more effective than antigen alone for inducing clonal expansion and plasma cell differentiation suggests that the activation of memory B cells in vivo still requires more than one signal, antigen triggered BCR-mediated signal in combination with TLR-triggered signals or T cell help. The findings that on day 5 post the third immunization, antigen alone induced a higher serum titer but a lower frequency of high affinity NP-specific IgG ACS in spleens than antigen plus CpG or LPS may reflect differences in the kinetics of B cell differentiation and plasma cell migrating from the spleen to the bone marrow in response to different stimuli. Some other factors may limit the direct role of TLRs on B cells, such as the competition between B cells and dendritic cells to interact with TLR agonists. TLR agonists are very effective in the induction of dendritic cell maturation. Dendritic cell-activated T_H cells provide memory B cells with a second stimulation signal. Above all, requirement of both antigen and TLR agonists for the activation of memory B cells in vivo ensures the specificity of humoral memory responses.

The necessity of TLRs in humoral memory responses is controversial. Disruption TLR signaling by knocking out Myd88 and/or Trif genes either did or did not significantly inhibit antibody responses against T-dependent antigens, depending on the specific experimental system (23, 29, 30). The study presented here reveals a direct stimulatory role of TLR on memory B cell differentiation in vitro and the requirement of both antigen and TLR agonist for effective activation of humoral memory responses in vivo. These results strongly support a critical role for TLRs in recalling humoral memory responses.