Distinction of the memory B cell response to cognate antigen versus bystander inflammatory signals

A comprehensive analysis of B_MEM responsiveness to antigen and bystander inflammatory stimuli was undertaken both in vitro and in vivo. The term “bystander” is defined herein as stimuli acting on B_MEM independent of BCR engagement. In these studies, we tracked B_MEM specific for the thymus-dependent protein antigen R-phycoerythrin (PE), as immunization of BALB/c mice with this fluorophore elicits a detectable population of PE-binding B_MEM at a high resolution (referred to as PE-B_MEM; Hayakawa et al., 1987). After immunization with PE adsorbed to alum i.p., PE-B_MEM are readily detectable in the spleen at 0.01–0.05% of total splenocytes beginning 21 d after immunization, with this population still present 60 d after priming (Fig. S1; Hayakawa et al., 1987; Schittek and Rajewsky, 1990; Benson et al., 2008). In nonimmunized mice, a minute naive PE-binding population is observed at a 10-fold lower frequency (Fig. S1). The PE-B_MEM in immunized mice are predominantly B220⁺PE⁺CD38⁺IgD⁻IgG1⁺CD80^hi and express a phenotype consistent with long-lived affinity-matured B_MEM (Fig. S1; Anderson et al., 2007). In all experiments (unless indicated otherwise), PE-B_MEM from PE/alum-immunized mice were analyzed between days 60 and 90 after primary immunization.

The response of PE-B_MEM to inflammatory stimuli was first analyzed by examining the response profile of B_MEM to a panel of TLR agonists in vitro. We also included two other sources of B_MEM in these in vitro analyses to first ensure that the results were reproducible in multiple systems and to test whether B_MEM activation occurs in the absence of BCR cross-linking. The second system used B_MEM raised against and specific for the hapten 4-hydroxy-3-nitrophenylacetyl (NP), with this system having been previously characterized (McHeyzer-Williams et al., 1991; Lalor et al., 1992). The third system used B_MEM from a transgenic mouse in which the polyclonal B_MEM population is readily identified. This latter source allowed for the isolation of B_MEM without the need for BCR engagement by antigen, as when PE-B_MEM and NP-B_MEM are purified using antigen binding to denote specificity, cross-linking the BCR is inevitable.

From PE-immunized mice, PE-B_MEM were first enriched by negative selection, and then electronically sorted based on a B220⁺PE⁺dump⁻CD38⁺ phenotype, with post-sort analysis indicating high levels of purity (Fig. S2 A).

NP-B_MEM were generated in a transgenic mouse strain where Cre recombinase was inserted in the Cγ1 locus, with this strain crossed with the Rosa26-EYFP reporter mouse (Casola et al., 2006). In this mouse, every post-GC B cell expresses YFP, with this strain affording the opportunity to track post-GC B cell subsets by YFP expression (Casola et al., 2006). Immunization of Cγ1-cre/Rosa26-EYFP mice with NP-keyhole limpet hemocyanin (KLH)/alum lead to the generation of NP-B_MEM (Fig. S2 B; Casola et al., 2006). After negative enrichment, NP-B_MEM were defined as B220⁺ NIP⁺IgG1/YFP⁺dump⁻CD38⁺, with post-sort analysis indicating purity (Fig. S2 B).

Polyclonal B_MEM can be clearly enumerated in Cγ1-cre/Rosa26-EYFP mice through expression of YFP, loss of IgD and IgM, and expression of other B_MEM markers. NP-immunized Cγ1-cre/Rosa26-EYFP mice were negatively enriched for B_MEM with polyclonal B_MEM defined as B220⁺IgG1/YFP⁺CD38⁺dump⁻CD80⁺ and sorted to purity (Fig. 1 C).

In parallel, NF B cells were defined as B220⁺CD23⁺IgD⁺CD38⁺ and were additionally sorted (Fig. S2 C).

The response of PE-B_MEM, NP-B_MEM, poly-B_MEM, and NF B cells to TLR agonists after 3 d of in vitro culture was analyzed (Fig. 1). PE-B_MEM and corresponding NF B cells and IL-4–only controls were stained with CFSE, whereas NP-B_MEM and poly-B_MEM with NF B cells and IL-4 only controls were stained with DDAO to track proliferation. After culture with the TLR4 agonist LPS, cells were analyzed for surface expression of B220 and CD138. The culture of all NF and B_MEM with IL-4 resulted in a uniform B220⁺ population after 3 d (unpublished data). Upon culture of NF B cells and B_MEM with LPS, NF B cells remained primarily B220⁺CD138⁻, whereas all three sources of B_MEM generated a B220^intCD138⁺ population of cells (∼21–30% of total) representing the phenotype of a ASC (Fig. 1, A and D). The ASCs derived from the three sources of B_MEM were of the IgG isotype and, with the ASCs generated from PE-B_MEM and NP-B_MEM, specific for either PE or NP, respectively (Fig. 1 G and not depicted). NF B cells generated a high frequency of IgM⁺ ASCs (Fig. 1 G). The overall degree of proliferation by the three sources of B_MEM was moderately less than NF B cells, although proliferation did occur, as indicated by comparison to the IL-4–only cultured cells (Fig. 1, B and E). The population of B220^intCD138⁺ cells generated from poly-B_MEM arose in the daughter peaks, showing that B_MEM undergo cell division while differentiating into B220^intCD138⁺ cells (Fig. 1 F).

The response of poly-B_MEM and NF B cells was analyzed to either Pam3CSK4 (TLR1/2 agonist), heat-killed Listeria monocytogenes (TLR2 agonist), poly(I:C) (TLR3 agonist), LPS (TLR4 agonist), flagellin from Salmonella typhimurium (TLR5 agonist), FSL-1 (TLR6 agonist), S-27609 (TLR7 agonist), ODN 1826 (TLR9 agonist), and IL-4. Both NF B cells and poly-B_MEM were responsive to agonists for TLR1/2, TLR4, TLR6, TLR7, and TLR9 (Fig. 1 I). Both poly-B_MEM and NF B responded poorly or did not respond to agonists for TLR2, TLR3, and TLR5 (unpublished data). For each TLR agonist that NF B cells or poly-B_MEM responded to, B_MEM generated a sizable population of B220^intCD138⁺ cells, whereas NF B cells maintained a B220⁺CD138⁻ phenotype (Fig. 1 H). NF B cells also proliferated to a greater degree than B_MEM (Fig. 1 I). Finally, it was observed that while engagement of CD40 drove proliferation of both B_MEM and NF B cells, B_MEM did not differentiate into B220^intCD138⁺ cells (Fig. 1, H–I). Collectively, these data show that B_MEM are sensitive to multiple TLR agonists in vitro and do not require BCR cross-linking to become activated and differentiate into ASCs. Furthermore, upon activation, B_MEM have an enhanced intrinsic capacity to differentiate into CD138⁺ ASCs compared with NF B cells. Lastly, CD40 signaling appears to drive B_MEM proliferation, but to block differentiation in vitro.

Having established that B_MEM are responsive to bystander stimuli in vitro, we used the PE-antigen system to compare B_MEM responsiveness toward cognate antigen versus bystander inflammation in vivo. We first performed a detailed analysis of the secondary humoral immune response to soluble PE (sPE) to define the behavior and kinetics of B_MEM activation in vivo. Unlike naive hosts, it is well-established that previously immunized mice are responsive to low-doses of antigen in the absence of adjuvant (McHeyzer-Williams et al., 2006). This heightened responsiveness results from the combined presence of antigen-reactive B_MEM and CD4⁺ memory T cells. In concordance with this, we observed that injection of naive mice with 1 µg sPE (or 10 µg; not depicted) failed to elicit PE-ASCs (Fig. 2, A and B), whereas PE-immunized mice produced high numbers of IgG⁺ PE-ASCs. A robust IgM ASC response was only elicited when naive or PE-immunized mice were injected with 1 µg of sPE along with the TLR9 agonist CpG ODN 1826 (CpG; Fig. 2 A). Thus, the IgG⁺ PE-specific humoral response generated after sPE injection is likely the direct consequence of B_MEM activation. These data demonstrate that a low level of sPE given in the absence of adjuvant is capable of activating PE-B_MEM to differentiate into IgG⁺ ASCs.

The in vivo proliferation of PE-B_MEM was measured by flow cytofluorimetric analysis of BrdU incorporation. PE-immunized mice were injected with sPE, and a 5-d BrdU pulse was performed starting on the day of sPE injection, followed by a chase period. BrdU incorporation by B_MEM was quantified at various times during and after the pulse period. After preenrichment, PE-B_MEM were detected by excluding CD4/CD8/IgD/IgM⁺ cells via a dump gate and gating them on B220⁺PE⁺ cells, with these cells analyzed for BrdU incorporation (Fig. S3 A). PE-immunized mice receiving BrdU, but no antigen, and immunized mice receiving neither antigen nor BrdU were included as negative controls (Fig. 2 C). 2 d after sPE rechallenge, B_MEM began to proliferate, with this response peaking by day 4 (Fig. 2 C and Fig. S3 A). In mice analyzed during the BrdU chase period (days 11 and 24 after sPE rechallenge), a proportion of B_MEM remained BrdU⁺ at a percentage similar to that observed at the peak of the response (35.5% ± SEM 4.6 and 38.4% ± SEM 5.1 versus 48.9% ± SEM 5.8, respectively), indicating that 71–77% of B_MEM proliferation occurs within the first 5 d after antigen rechallenge (Fig. 2 C and Fig. S3 A). A second PE-binding population was also observed in these experiments, with this population displaying a B220⁻PE⁺ phenotype (Fig. S3 A). When the BrdU incorporation frequencies by the B220⁻PE⁺ population in PBS versus sPE-immunized mice were compared, no differences were observed between the two groups over the 24-d kinetic analysis (unpublished data). This indicates that the B220⁻PE⁺ population does not respond to sPE, and as antigen-responsiveness is a crucial element of adaptive memory, we excluded these cells from further analysis. As BrdU labeling measures the accumulation of proliferation over the time of the BrdU “pulse” period, we also analyzed B_MEM expression of Ki-67 at various times after antigen challenge (Fig. 2 F). Ki-67 expression is an indicator of all stages of the cell cycle and provides a snapshot of the proliferative state of the B_MEM at a given point in time (Gerdes et al., 1984). PE-B_MEM are almost uniformly Ki-67⁻, which is indicative of their quiescent state (Fig. 2 F). By days 3 and 4 after sPE rechallenge, 76–77% of B_MEM cells were Ki-67⁺. At day 5, Ki-67 expression began to dissipate, and most PE-B_MEM exhibited decreased Ki-67 expression on day 7, with this decrease more pronounced by day 11. By day 24, PE-B_MEM were once again uniform in their lack of Ki-67 expression (Fig. 2 F). In correlating the expression pattern of Ki-67 by B_MEM (Fig. 2 F) with the 5-d BrdU pulse followed by a chase period (Fig. 2 C), the results indicate that the majority of B_MEM proliferation occurs between days 2 and 5 after sPE rechallenge. In addition, the presence of BrdU⁺ B_MEM 24 d after sPE injection provides direct evidence that B_MEM generate daughter B_MEM after secondary antigen encounter.

A rapid increase in the numbers of IgG⁺ PE-ASCs in both the spleen and BM was observed 3 d after sPE rechallenge (Fig. 2, B and D). This burst in ASC numbers peaked 4 d after challenge in the spleen and 5 d in the BM, after which ASC numbers gradually subsided over the next 20 d (Fig. 2 D). Corresponding to the increase of ASCs after rechallenge was an increase in anti-PE IgG serum titers increasing 10-fold over the next 20 d (Fig. 2 E). These data show that ASCs are rapidly generated within 3 d after antigen rechallenge and are present to substantial degrees in both the spleen and the BM. Soon thereafter, ASC numbers declined, indicating that short-lived ASCs are present in both organs.

The observation that ∼50% of B_MEM respond to sPE as indicated by BrdU incorporation (Fig. 2 C and Fig. S3 A) led us to question whether either the antigen dose or the absence of adjuvant during antigen administration was limiting the frequency of responding B_MEM. To test this, we varied the amount of sPE injected or added CpG during antigen rechallenge, with BrdU administration concurrently initiated. When analyzed 3 d later, it was observed that the dose of antigen positively correlated with the percentage of proliferating B_MEM (Fig. 2 G and Fig. S3 B). The addition of CpG had no further impact on the frequency of responding PE-B_MEM (Fig. 2 G and Fig. S3 C). However, the addition of CpG to sPE modestly increased the number of splenic IgG⁺ PE-ASCs elicited after 3 d when compared with sPE alone (Fig. 2 B). In sum, these data indicate that antigen can be a limiting factor in dictating the frequency of responding B_MEM. Notably, TLR9 agonists failed to enhance the frequency of activated B_MEM. This point indicates that CpGs alone may not play a role in driving in vivo B_MEM activation.

The rapid kinetics of the B_MEM response led us to question whether B_MEM form GCs after antigen encounter. Clonally expanding B_MEM were tracked to define their transitional stages en route to differentiating into ASCs. Under steady-state conditions, PE-B_MEM displayed a GL7⁻PNA⁻Ki-67⁻CD38⁺ phenotype as anticipated (Fig. 3 A; Lalor et al., 1992). As a positive control for GC markers, Peyer's patches were harvested and gated on B220⁺IgD⁻ cells, a population known to be primarily GC B cells, with these cells exhibiting a GL7⁺PNA⁺Ki-67^hiCD38⁻ phenotype (Butcher et al., 1982). When PE-B_MEM were harvested and analyzed 3 d after sPE challenge, they were GL7⁻PNA⁻Ki-67^hiCD38⁺, indicating these cells had responded to antigen because of Ki-67 expression, yet failed to express the GC markers GL7 and PNA (Fig. 3 A). It was plausible that the failure of antigen-responding B_MEM to gain a GC phenotype was caused by the solitary use of sPE, and that the induction of a GC phenotype requires the presence of an adjuvant. Therefore, CpG was added to sPE, and 3 d later, the phenotype of the responding B_MEM was analyzed. The addition of CpG to sPE again elicited GL7⁻PNA⁻Ki-67^hiCD38⁺ B_MEM (Fig. 3 A). It was considered possible that B_MEM gain a GC phenotype later than 3 d after antigen encounter, so the phenotype of B_MEM was analyzed at later points in time. The results indicate that PE-B_MEM maintain a GL7⁻PNA⁻CD38⁺ phenotype throughout the cell cycle (Fig. 3 B). These data define a proliferating state of B_MEM occurring shortly after in vivo antigen encounter, with these B_MEM expressing a GL7⁻PNA⁻Ki-67^hiCD38⁺ phenotype.

To investigate whether antigen-activated B_MEM generate GCs, GC structures were quantified in spleens where PE-B_MEM were either unchallenged or were responding to sPE or sPE + CpG by immunofluorescence (IF) microscopy. As a positive control for the presence of GCs, naive BALB/c were immunized with PE/alum and spleens were examined 10 d later at the height of the GC response. To ensure that PE-B_MEM in the sPE- and sPE + CpG–treated groups and the PE-GC B cells in the D10 PE/alum immunized groups were present and proliferating in these mice, FACS analysis was performed on half of the spleens, with BrdU incorporation by B220⁺PE⁺ cells analyzed (Fig. S4). Upon IF evaluation, in the D10 PE/alum immunized spleens, GL7⁺ clusters were observed and proximal to CD35⁺ follicular DCs, with these structures detected in ∼50% of the B cell follicles (Fig. 3, C and D). These GL7⁺ cell clusters were minimally present in the PBS-, sPE-, and sPE + CpG–treated groups (Fig. 3, C and D). Importantly, there was no significant increase in the number of GCs present in spleens where B_MEM were activated by sPE or sPE + CpG in comparison to PBS controls (Fig. 3 D). These data, coupled with the GL7⁻PNA⁻Ki-67^hiCD38⁺ phenotype displayed by B_MEM after in vivo antigen-dependent activation, demonstrate that GCs are not created by B_MEM after antigen rechallenge.

The in vivo B_MEM response to TLR agonists, CD40 agonists, and bystander T cell help was determined. PE-immunized mice were injected with PBS, sPE as a positive control, combinations of TLR agonists, agonistic αCD40, and/or αCD3. We treated mice with agonist doses at or greater than doses known to have an optimal biological effect in vivo (Ferran et al., 1990; Ahonen et al., 2004). In data where 50 µg of CpG was used, similar data were obtained using 100 µg (unpublished data). Furthermore, the efficacy of the TLR agonists was confirmed by observing an induction in the expression of CD86 by both NF B cells and CD11c⁺ DCs 24 h after TLR agonist injection in vivo (unpublished data). PE-B_MEM proliferation was tracked by BrdU incorporation over a period of 3 d (Fig. S5 A). NF B cells were included in this analysis and were identified as B220⁺IgD⁺ cells (Fig. S5 B). Injection with PBS alone elicited very low amounts of BrdU incorporation by both NF B cells and PE-B_MEM, whereas injection with sPE elicited only B_MEM proliferation (Figs. 4, A and B, and Fig. S5). It was observed that both B_MEM and NF B cells proliferate in response to injection with agonistic αCD40 antibody alone or coupled with LPS or CpG, with the mean proliferation of both subsets as measured by BrdU incorporation between 25 and 50% (Fig. 4, A and B, Fig. S5). After injection with LPS, CpG, or αCD3 alone or αCD3 and LPS or CpG, no proliferation was observed for either NF B cells or PE-B_MEM (Fig. 4, A and B, and Fig. S5). These data indicate that both NF B cells and PE-B_MEM proliferate in response to αCD40 signals in vivo, whereas TLR agonists or activation of T cells with αCD3 did not induce B_MEM proliferation.

To determine whether B_MEM differentiate into ASCs after signaling through CD40, PE-immunized mice were injected with PBS, sPE, CpG, αCD40, or both αCD40 and CpG. IgG⁺ PE-ASCs after agonist injection were quantified, and only in the sPE challenged mice was an increase in IgG⁺ PE-ASC numbers observed (unpublished data). The numbers of BM IgG⁺ PE-ASCs after agonist treatment were also analyzed, with serum harvested on the day of agonist injection and again when analyzed 14 d later. The only treatment group in which an increase in numbers of BM IgG⁺ PE-ASCs and serum PE-specific IgG was observed in the sPE-treated cohort (Fig. 4, C and D).

These data indicate that TLR4 and TLR9 agonists and the proinflammatory cytokine storm elicited by agonistic αCD3 treatment are incapable of activating B_MEM and driving their proliferation or differentiation in vivo. Engagement of CD40 (either alone or coupled with TLR4 and TLR9 agonists) drives both NF B cell and B_MEM proliferation, but is incapable of mediating their differentiation into ASCs.

The capacity of PE-B_MEM to proliferate and differentiate in response to either immunization with NP-KLH/alum administered with CpG or infection with vaccinia virus Western Reserve strain (VACV_WR) was analyzed. The levels of B_MEM proliferation and their differentiation into ASCs were measured 17 d later with this time chosen to allow for the full development of an adaptive immune response against either NP-KLH or VACV_WR. After immunization with VACV_WR or NP-KLH, PE-B_MEM did not incorporate any more BrdU (15.9% ± SEM 1.8 and 16.6% ± SEM 2.3, respectively) and then after injection with PBS alone (19.22% ± SEM 1.5; Fig. S6 C and Fig. 5 A). As anticipated, sPE induced robust PE-B_MEM cell proliferation, with the majority (73% ± SEM 2.8) of cells BrdU⁺ (Fig. S6 C and Fig. 5 A). The generation of IgG⁺ PE-ASCs was also measured by comparing the levels of PE⁺ IgG⁺ serum antibody titers present before and after immunization and by quantifying the number of IgG⁺ PE-ASCs present in the spleen and BM by ELISPOT. VACV_WR and NP-KLH immunizations did not impact the number of IgG⁺ PE-ASCs present in either the spleen or the BM 17 d after immunization (Fig. 5, B and C). There was also no difference in the levels of IgG⁺ PE-specific serum antibody titers after VACV_WR or NP-KLH immunization (Fig. 5 D). Consistent with our earlier observations (Fig. 2, D and E), sPE induced a substantial and significant increase in the number of IgG⁺ PE-ASCs in both the spleen and BM (Fig. 5, B and C) and led to an ∼30-fold increase in the levels of IgG⁺ PE-specific serum antibody titers (Fig. 5 D). Both cellular and humoral anti-VACV_WR immune responses were verified to have occurred by testing for the presence of IFN-γ⁺ VACV_WR-specific CD4 and CD8 T cells in spleens harvested from VACV_WR-infected mice, but not naive mice, and by observing for the presence of anti-VACV_WR IgG serum antibodies only in infected mice (Fig. S6, A and B).

These data collectively demonstrate that immunization with NP-KLH and VACV_WR infection does not induce PE-B_MEM proliferation or differentiation into ASCs. Thus, B_MEM are unresponsive to stimuli generated during a bystander immune response.

A detailed analysis of the in vivo secondary immune response of B_MEM in response to cognate antigen versus noncognate inflammatory bystander signals is presented. These data show that B_MEM respond robustly upon rechallenge with cognate antigen, but do not acquire the expression of GC markers or form GCs in response to cognate antigen. In striking contrast, the response of B_MEM to a battery of proinflammatory signals is remarkably silent. It has been proposed that serological memory is sustained by B_MEM responding to bystander inflammatory signals by undergoing proliferation and differentiation. However, the findings of this study indicate that B_MEM ignore these signals (TLR agonists, polyclonal T cell activation, aggressive vaccination, and acute viral infection), as they fail to drive B_MEM proliferation or differentiation in vivo.

Despite the preponderance of studies detailing the B cell dynamics of the primary humoral immune response, surprisingly few studies have detailed the events of the secondary humoral immune response. We have tracked B_MEM and demonstrate that cognate antigen drives the rapid clonal expansion of B_MEM and their terminal differentiation into ASCs, with these events occurring in the absence of GC formation. As has been widely reported, B_MEM rapidly differentiate to ASCs upon secondary antigen encounter (Höfer et al., 2006; Minges Wols et al., 2007), with this ASC response first detectable 3 d after antigen challenge in both the spleen and the BM and with the splenic ASC response an order of magnitude greater than that observed in the BM. This wave of ASCs dissipates after 11 d, and it is probable that the remaining ASCs in the BM 24 d after antigen challenge have been culled into the long-lived PC pool (Manz et al., 1997; Höfer et al., 2006; Minges Wols et al., 2007). These data detail the activation of B_MEM and define their proliferation as occurring between days 2 and 5 after antigen encounter. Proliferating B_MEM occupy a transitional state defined as antigen-binding B220⁺CD38⁺Ki-67⁺GL7⁻PNA⁻ before generating quiescent daughter B_MEM and/or differentiating into ASCs. This blasting state is unique because of the presence of high levels of CD38 expression by proliferating cells; CD38 is a phenotypic marker previously used to denote quiescent NF B cells and B_MEM and is lost upon B cell activation and entry into the GC (Oliver et al., 1997). Our finding that B_MEM cells do not form or enter a GC reaction during their differentiation and expansion is consistent with classical studies indicating that B_MEM might not undergo further SHM and isotype switching upon rechallenge with antigen (Askonas et al., 1970; Okumura et al., 1976; Siekevitz et al., 1987). In these studies, limiting numbers of hapten-specific “parental” B_MEM were adoptively transferred together with carrier-primed T cells into lightly irradiated hosts, followed by immunization of the hosts with soluble hapten-carrier antigen. Upon analyzing the isotype distribution and SHM exhibited by the parental B_MEM and their progeny generated in response to cognate antigen, no change in diversity in either measure was observed. Our data showing that B_MEM neither gain GC markers nor seed GCs are in concordance with these studies. It is apparent that B_MEM, at least in these studies and in our presented data, are not instructed to enhance the affinity or isotype distribution of their antigen receptors during secondary antigen encounter.

Our data indicate that quiescent IgG⁺ B_MEM are not activated by TLR agonists or bystander T cell help, and thus do not undergo self-renewal events or differentiate into ASCs in an antigen-independent manner. We should note that our analysis was limited to IgG⁺ PE-B_MEM and that our experimental system excluded testing whether IgM⁺ B_MEM are responsive to bystander inflammatory signals. There is a growing body of evidence both in favor of and in conflict with the hypothesis that serological memory is sustained by the activation of B_MEM by bystander inflammation. Evidence in favor of this theory includes in vitro observations in which human B_MEM were shown to have increased sensitivity and to selectively proliferate and differentiate into ASCs in response to bystander stimuli as compared with NF B cells (Bernasconi et al., 2002, 2003). In vivo studies in humans found that for single-pathogen specificities, serum antibody levels linearly correlated with the frequency of B_MEM present in a host, indicating that the ASC compartment was genealogically linked to and undergoing replenishment by the B_MEM compartment (Bernasconi et al., 2002). Mouse studies also supported this link between the B_MEM and ASC compartments as elimination of virus-specific B_MEM led to a slow decay of virus-specific ASC numbers over time in comparison to mice still containing B_MEM (Slifka et al., 1998). Lastly, it was observed that during an immune response against a single pathogen, an increase in both blood ASC numbers and serum antibody levels occurred for specificities to other pathogens (Bernasconi et al., 2002). This result was interpreted as evidence that host B_MEM were recipients of antigen-independent stimuli and responded by differentiating into ASCs.

Despite these data suggesting that B_MEM respond to bystander inflammation, more recent studies have put this hypothesis under scrutiny. In particular, further attempts to correlate antigen-specific B_MEM numbers with circulating ASCs in human subjects have been unsuccessful (Amanna et al., 2007). Mouse studies where long-lived ASC numbers were quantified after B_MEM depletion observed no decay in ASC numbers over time, contrary to initial reports (Ahuja et al., 2008; DiLillo et al., 2008). Lastly, a series of studies reported that in mice that were administered a TLR agonist or in humans that were given a vaccine, ASC numbers and serum antibody levels against specificities other than the immunizing antigen did not increase (Di Genova et al., 2006; Amanna et al., 2007; Xiang et al., 2007; Richard et al., 2008).

Our studies show that the in vitro sensitivities of B_MEM to activation are not representative of their sensitivities in vivo and cast doubt on the hypothesis that B_MEM can be activated by bystander inflammation. Our in vitro studies examining the response profiles of murine B_MEM to both antigen-dependent and -independent stimuli extended prior studies using human (Arpin et al., 1997; Bernasconi et al., 2002) and murine (Richard et al., 2008) B_MEM, as we observe that murine B_MEM are sensitive to TLR agonists and have an enhanced intrinsic ability to differentiate into ASCs upon activation with TLR agonists, regardless of whether the BCR is engaged by antigen. Given these data, and our knowledge of the robust response of B_MEM to cognate antigen in vivo, an exploration of the response of B_MEM to several nonspecific inflammatory signals in vivo demonstrated that B_MEM are in fact unresponsive to these stimuli. The first panel of signals tested was derived from activated T cells and included CD154 expression by these cells, as well as proinflammatory cytokines either alone or together with TLR agonists. We first analyzed the impact of agonistic αCD40 antibodies on B_MEM proliferation and differentiation. The injection of mice with αCD40 alone or with TLR4 or TLR9 agonists elicited substantial levels of B_MEM and NF B cell proliferation. Despite this proliferation, these B_MEM did not differentiate into ASCs. These data are in agreement with our in vitro observations, where αCD40 induced the proliferation of, yet failed to differentiate, poly-B_MEM into ASCs. These studies also extend previous in vitro studies where CD40 signaling triggered human B_MEM proliferation, but impeded their differentiation into ASCs (Arpin et al., 1995). Using a second approach, we activated T cells in vivo through the injection of agonistic αCD3 antibodies, with this treatment given either alone or with TLR agonists. Injection of this antibody is known to elicit rapid and transient activation of T cells, which release a proinflammatory cytokine storm that includes IL-2, IFN-γ, and TNF (Ferran et al., 1990; Scott et al., 1990). These stimuli did not induce either B_MEM proliferation or differentiation.

It is highly likely that the in vivo response of B_MEM to overt antibody-mediated CD40 signaling does not represent the hypothetical situation where B_MEM and activated T cells transiently engage in a CD40/CD40L synapse independent of an antigen bridge. As we found no B_MEM proliferation upon vaccination with an irrelevant thymus-dependent antigen or infection with vaccinia virus (a situation generating large numbers of activated CD4 T cells expression CD40L), it is probable that this interaction does not occur and is thus not a mechanism for propagating B_MEM renewal. Instead, this data should be interpreted as in vivo proof that CD40 signaling drives B_MEM proliferation but blocks their differentiation into ASCs, indicating that differentiation occurs upon cessation of CD40 signaling.

Immunization with NP-KLH, along with alum and CpG, and infection with vaccinia virus results in the acute induction of both innate and adaptive immunity. B_MEM ignored the stimuli associated with these responses and neither proliferated nor differentiated into ASCs during the multi-week period in which these immune responses were occurring. In summary, inflammatory signals associated with TLR signaling, T cell activation, and adaptive immune responses are incapable of driving B_MEM proliferation or differentiation into ASCs.

The possibility remains that B_MEM and long-lived PCs might survive for the lifetime of the host, with these populations not requiring self-renewal or replenishment and with serological memory not requiring maintenance. However, this hypothesis requires that B_MEM and PCs survive for perpetuity, and this quality appears unrealistic. From our presented work, we show that B_MEM do not clonally expand or differentiate in response to bystander inflammatory signals. It is possible that there are other signals capable of inducing in vivo B_MEM expansion and differentiation that are not included in this analysis. Alternatively, the maintenance of serological memory may occur via a stochastic rather than an induced process, with B_MEM not requiring an extrinsic signal to maintain serological memory such as those tested herein.

These studies were approved by the Institutional Animal Care and Use Committee of Dartmouth College (Lebanon, NH). BALB/c mice were purchased from the National Cancer Institute. RAG^−/− mice were purchased from The Jackson Laboratory. Cγ1-cre mice crossed with Rosa26-EYFP mice have been previously described and were provided by K. Rajewsky (Harvard Medical School, Boston MA) and S. Casola (Fondazione Italiana per la Ricerca sul Cancro Institute of Molecular Oncology Foundation, Milan, Italy; Casola et al., 2006). All animals were maintained in a pathogen-free facility at Dartmouth Medical School. For primary immunizations, 10 µg of R-PE (Cyanotech) or 100 µg of NP₂₈-KLH (Biosearch Technologies) adsorbed to prepared alum was injected i.p. in a volume of 200 µl. For secondary challenge, 10 µg of PE (unless indicated otherwise) in PBS in a volume of 200 µl was injected i.p. For VACV_WR infections, virus was propagated and titered on the 143B cell line as previously described (Fuse et al., 2008), with 2 × 10⁶ PFU injected i.p. For BrdU treatment, 0.8 mg/ml was added to the drinking water and changed daily except in Fig. 5 where BrdU was changed every other day.

Single-cell suspensions of lymphocytes were prepared as previously described (Benson et al., 2008). For enrichment of splenocytes for B_MEM before phenotyping and BrdU analysis, CD4⁺, CD8⁺, IgD⁺, and IgM⁺ cells were removed by negative depletion using a mix of biotinylated antibodies. Cells were removed using the EasySep Biotin Selection Kit for Mouse Cells (StemCell Technologies). For enrichment of B_MEM before cell sorting, IgD⁺ and IgM⁺ cells were first removed, and then CD43⁺ cells were removed in a two-step process using the EasySep Mouse B Cell Enrichment Kit.

B cells were cultured in RPMI media supplemented with 10% FBS (Atlanta Biologicals), Hepes, 50 µM β-ME, and penicillin/streptomycin/l-Glutamine. Sorted B cells were cultured in 96-well round-bottomed plates with 10,000–15,000 cells/well and were cultured at 37°C. For in vitro VACV_WR CD4 and CD8 recall, BALB/c splenocytes were infected for 2 h with a 10:1 ratio of virus to cells. Infected cells were then plated in a round-bottom 96-well plate with 5 × 10⁶ splenocytes harvested from either D17 VACV_WR-infected or noninfected mice added and cultured for 6 h in the presence of 10 µg brefeldin-A and 10 U/ml IL-2. Intracellular staining of IFN-γ was performed as previously described (Fuse et al., 2008). The following reagents and vendors were used in this study: CpG-ODN1826 (5′-TCCATGACGTTCCTGACGTT-3′) with phosphorothioate bases (Eurogentec S.A.), LPS 055:B5, BrdU (Sigma-Aldrich), Pam4CSK4, HKLM, poly(I:C), ST-FLA, and FSL1 (Invivogen), S-27609 (3M Pharmaceuticals), IL-4 (Peprotech), αCD40 clone FGK45, and αCD3 clone 145 2C11 antibodies (Klaus Lube; BioExpress).

Antibodies against the following antigens were used: B220 (clone 6B2), IgG1 (clone A85-1), IgD clone (11-26c), IgM (clone 11–41), CD4 (clone GK1.5), CD8 (clone 2.43), IFN-γ (clone XMG1.2), CD43 (clone 1B11), CD38 (clone 90), CD23 (clone B3B4), CD80 (clone 1610-A1), Ki-67 (clone B56), CD138 (clone 281–2), BrdU (clone PRB-1), CD35 (clone 8C12), GL7, and PNA. NIP-PE and PE-binding cells were detected by staining with NIP₂₀-PE (generated in-house) or with 2 µg/ml PE. For CFSE or DDAO dye dilution, cells were labeled with 5 µM CFSE or 10 µM DDAO in RPMI at 37°C for 10 min (Invitrogen). Intracellular BrdU (BD) and Ki-67 (eBioscience) staining was performed according to the manufacturers recommendations. Flow cytometry was performed on a refurbished FACSCAN running CellQuest software (BD), with data analysis performed using FlowJo (Tree Star, Inc.). Cell sorts were performed on a FACSAria (BD; Flow Cytometry Facility at Dartmouth Medical School).

For ELISpot analysis, cells were apportioned to PE, NP₃-BSA, or polyclonal anti-Ig–coated Multiscreen 96-well plates (Millipore) with twofold serial dilutions made before incubation for 7–18 h at 37°C. PCs were detected by HRP-conjugated or biotinylated anti–mouse IgM, IgG1, IgG2a, IgG2b, IgG3, IgE, or IgA polyclonal antibodies (SouthernBiotech.), with streptavidin-HRP used in a second step when necessary. ELISpots were developed as previously described (Benson et al., 2008).

For ELISA analysis, plates were coated with either 10 µg/ml PE or 4 × 10⁶ PFU of H₂O₂-inactivated VACV_WR (Hammarlund et al., 2008) overnight in PBS, blocked with PBS + 5% FBS, and washed, and then 1:2,000 diluted serum added at serial 1:2 dilutions. Serum from either a PE hyperimmunized mouse or from a VACV_WR mouse was included on each plate as a reference between plates and between experiments and used to generate a standard curve, with these serum allotted the value 8,000 arbitrary units. Antibody levels were detected with IgG-AP (SouthernBiotech) and developed with 1 mg/ml pNPP (Sigma-Aldrich) in 0.05 sodium carbonate buffer.

Spleens were presoaked in 15% sucrose/PBS for 30 min and snap-frozen in OCT compound (Sakura Finetek) in a dry ice/ethanol slurry and stored at −80°C. Tissues were sectioned onto slides using a cryostat with 8-µm-thick sections. After drying at room temperature, the tissue was fixed in ice-cold 1:1 acetone/methanol mix for 10 min and allowed to dry, after which sections were outlined with an ImmEdge Pen (Vector Laboratories). Sections were blocked with 10% heat-inactivated normal rat serum/PBS for 1 h, washed, and stained with IgD-Alexa Fluor 647, GL7-FITC, and CD35-biotin, followed by staining with SA-PerCP. After washing, slides were mounted with PermaFluor Aqueous Mounting Media (Thermo Fisher Scientific). Samples were analyzed on a LSM 510 META confocal microscope using LSM software (Carl Zeiss, Inc.).

Fig. S1 shows FACS gating scheme and phenotype of PE-B_MEM. Fig. S2 displays the FACS profiles and gating schemes used to sort-purify PE-B_MEM, NP-B_MEM, and NF B cells. Fig. S3 contains FACS profiles showing BrdU incorporation by PE-B_MEM and data corresponding to experiments presented in Fig. 2. Fig. S4 contains FACS plots showing BrdU incorporation by PE-B_MEM and D10 GC B cells, with these data corresponding to experiments presented in Fig. 3. Fig. S5 contains FACS plots showing BrdU incorporation by PE-B_MEM and NF B cells after in vivo administration of TLR agonists, CD40 agonists, and bystander T cell help. Data correspond to experiments presented in Fig. 4. Fig. S6 depicts the presence of VACV_WR-specific CD4 and CD8 cells and VACV_WR–specific serum IgG in infected hosts. FACS plots are also depicted showing levels of BrdU incorporation by PE-B_MEM after protein vaccination or viral infection, with data corresponding to experiments depicted in Fig. 5.

We thank Dr. Klaus Rajewsky and Dr. Stefano Casola for providing the Cγ1-cre mice. We thank Dr. Stacey Dillon and ZymoGenetics, Inc. (Seattle, WA) for providing some of the reagents used in these studies.

This work was supported by National Institutes of Health (NIH) grants RO1A1026296 and RO1A108896 to R.J. Noelle and NIH pre-doctoral training grant T32AI07363 to M.J. Benson.

The authors have no conflicting financial interests.