The PD-1 Regulatory Axis Inhibits T Cell–Independent B Cell Memory Generation and Reactivation

Humoral responses to pathogen-displayed polysaccharides are critical for protection against infection. These polysaccharides often behave as T cell–independent type 2 (TI-2) Ags, which by definition are able to elicit rapid Ab production in the absence of classical cognate T cell help or other strong innate stimuli but fail to do so in infants and in Xid mice lacking functional Btk (1). Pathogenic bacteria, including Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus influenzae, Salmonella typhi, and Bacillus anthracis, bear TI-2 Ags as capsule components (1–5). Fungi, parasites, and viruses also display TI-2 Ags (2, 6–10). Understanding the regulation of natural and vaccine-acquired humoral immunity to TI-2 Ags is critical for developing improved treatments and vaccines for TI-2 Ag-bearing pathogens.

A key aspect of TI-2 responses that is poorly understood relates to the pathways that regulate recall responses. Some TI Ags encountered on Gram-negative bacteria can induce plasmablast-like memory cells with the capacity for expansion and increased Ab production on pathogen re-exposure (11, 12); however, this does not readily occur with native polysaccharides. The refractoriness of polysaccharide-specific Ab responses to booster immunization with native polysaccharides has long been recognized (13–15). Although TI-2 Ags (pneumococcal polysaccharide [PPS], meningococcal polysaccharide, and haptenated Ficoll, etc.) were originally not thought to generate memory B cells (B_mems), several early studies using normal mice (16, 17) and more recent studies using adoptive transfers of transgenic (Tg) B cells expressing a high-affinity NP-specific BCR (V_HB1–8) have shown classical non–Ab-secreting B_mems develop after haptenated-Ficoll immunization (18, 19). Our work shows these Tg B_mems display a phenotype similar to T cell–dependent (TD) B_mems (20, 21), including CD80 and PDL2 expression (19). The extent to which these markers are also expressed by naturally occurring TI-2 memory cells is not known but is important to consider because each receptor interacts with potent immunoinhibitory molecules. Studies using V_HB1–8 Tg mice indicate Ag-specific IgG-dependent suppression of Ag-specific B_mems is a major mechanism contributing to failed boosting with TI-2 Ags (18, 19). However, the extent to which other regulatory pathways contribute to this regulation is not clear.

Overcoming impaired boosting to TI-2 Ags is a significant challenge that must be overcome to develop efficacious TI-2 Ag-based vaccines. Establishment of high titers, as well as functional B_mems that can respond to secondary immunization and/or infection, is necessary to achieve the highest level of protection. Protein conjugation to bacterial polysaccharides is one strategy that enables significant boosting in young children (except perhaps with serotype 3 PPS conjugates) (15). However, the cost of conjugate vaccines continues to increase as new polysaccharides are included, raising concerns about the feasibility of deploying them in developing countries (22). Concerns related to eliciting and maintaining high titers to an increasing number of distinct polysaccharides conjugated to the same protein have also not yet been addressed. Relative to young children, in adults, PPS conjugate vaccines do not boost as well, and overall Ab responses have been reported to be similar to those elicited to native PPSs (23). Finally, it is not clear whether the B_mems elicited by conjugate vaccines are adequately responsive to native polysaccharides, which may be encountered during infections. Hence alternative approaches to enhancing these vaccine responses may be required under some circumstances. Indeed, the use of B cell–activating adjuvants (24, 25), along with other immunomodulators, represents strategies to improve TI-2 Ag-specific B_mem formation and responsiveness to secondary Ag encounter in some individuals.

Our previous work demonstrated PD-1 suppresses primary Ab responses to TI-2 Ags (26, 27). However, the extent to which the PD-1:PD-1 ligand (PDL) pathway regulates the formation of TI-2 B_mems and their reactivation potential has not been investigated. In this study, we used the V_HB_1–8 Tg system, as well as normal B cells, to assess the role PD-1 and its ligands, PDL1 and PDL2, play in TI-2 B_mem formation and regulation. Our work demonstrates a key role for the PD-1:PDL regulatory axis in limiting B_mem formation largely because of interactions between B cell–expressed PD-1 and non–Ag-specific B cell–expressed PDL1. However, PD-1–mediated suppression of B_mem activation, division, and Ab production after boosting may also involve interactions between PD-1 upregulated on activated B_mem and B_mem-expressed PDL2, in addition to interactions with non–B_mem-expressed PDL1 and, to a lesser extent, non–B_mem-expressed PDL2. Thus, TI-2 B_mems may engage in a novel autoregulatory mechanism involving PD-1–PDL2 interactions. Inhibition of this pathway in the context of vaccination with adjuvanted polysaccharide vaccines yields high Ag-specific IgG production with significant boosting and enhanced protective capacity, demonstrating inhibition of TI-2 Ag boosting by Ag-specific IgG can be overcome by disrupting PD-1–mediated inhibition in combination with B cell–activating adjuvants. Thereby, our findings hold significance for developing alternative and improved polysaccharide vaccine formulations, especially for at-risk populations in which T cell help is impaired.

Wild-type (WT; CD45.2 and CD45.1 congenic) and V_HB1–8 Tg (B1–8hi IgH Tg knock-in; B6.129P2-Ptrpca Igh^tm1Mnz/J) mice on a C57BL/6 background were from Jackson Laboratory. PD-1^−/− (28), PDL1^−/− (29), and PDL2^−/− (30) mice were on a C57BL/6 background and obtained, as previously described, from Drs. Lieping Chen and Tahiro Shin (27). PD-1^−/− mice were crossed to CD45.1^+/+ V_HB1–8 Tg mice to yield CD45.2^+/+ PD-1^−/−V_HB1–8 Tg mice. Mice were housed under specific pathogen-free conditions, except during infection experiments. Mice were used at 2–4 mo of age and were age matched for experiments. All studies and procedures were approved by the Wake Forest Animal Care and Use Committee.

Mice were immunized with 1, 10, or 25 μg NP₄₀-AECM-Ficoll or TNP₆₅-AECM-Ficoll (Biosearch Technologies) in 200 μl PBS i.p. as indicated. Mice were immunized with vaccine-grade Pneumovax 23 (Merck, Whitehouse Station, NJ) containing 0.125 μg each of 23 PPSs or Prevnar-13 (Pfizer, formerly Wyeth Pharmaceuticals, New York, NY) containing ∼0.1 μg each of 13 PPSs, as previously described (19, 31). In some experiments, adjuvant containing 20 μg of Salmonella minnesota monophosphoryl lipid A (MPL), 20 μg synthetic cord factor (trehalose-6,6′-dicorynomycolate [TDCM]) in 0.5% squalene/0.05% Tween 80 (Sigma) was mixed with Ags before injections (24). In select experiments, mice were administered BrdU (0.8 mg/ml) in drinking water for 5 d after NP-Ficoll immunization or from weeks 5 to 6 postimmunization. NP-, TNP-, PPS3-, and Pneumovax-specific ELISAs were performed as previously described, with NP-specific IgM^a and IgG2a (selectively produced by Tg B cells) specifically measured in some experiments (19, 24). Standard curves were used to approximate Ag-specific Ab concentrations, as previously described (19, 24). PDL mAb blockade was performed by administering PDL1 (10F.9G2), PDL2 (TY25), or rat IgG control mAbs (all from BioXcell; InVivo mAb) i.p. on days 1 (200 μg), 3 (200 μg), and 5 (100 μg) postimmunization as previously described (26, 27). Immune mice were infected intranasally (i.n. 10⁷ CFUs) or i.p. with serotype 3 WU2 strain S. pneumoniae (10⁴ CFUs) and monitored every 12 h for signs of distress as previously described (31, 32). Ab levels were measured 14 d later. For passive serum transfer experiments, CD19^−/− mice were administered 0.5 μl pooled sera premixed with 200 CFUs WU2 in 100 μl PBS i.p.

Single-cell suspensions (2 × 10⁷/ml) in PBS containing 2% newborn calf serum were incubated with Fc Block (eBioscience) for 15 min, followed by staining with fluorochrome-conjugated Abs: CD19, PDL2, CD73, PD-1 (from BD Biosciences), and CD138 and CD86 or CD44 (from eBioscience); B220, CD80, CD11b, CD21/35, CD45.1, CD45.2 (BioLegend), pooled rat anti-mouse-IgG1, -IgG2b, -IgG2a (omitted specifically for Tg cell staining because of CD45 mAb reactivity), and -IgG3 Abs (Southern Biotech); and NP₄₀-allophycocyanin for 30 min at room temperature, followed by washing and fixing in 1.5% buffered formaldehyde (19). As we previously reported (19), we detect ∼90% of IgG-switched Tg B cells using staining for IgG1, IgG2b, and IgG3. BrdU-labeled DNA was detected as previously described (27). Fluorochrome-labeled isotype controls were used to determine background staining levels of NP-specific B cells. Adoptively transferred Tg B_mems were identified in recipient mice as CD45.1⁺ (or CD45.2⁺) NP-allophycocyanin⁺CD19⁺FSC^lowCD138^negCFSE^low cells. Cells were analyzed using a FACSCantoII or FortessaX20 cytometer (BD Biosciences) with forward scatter area/forward scatter height doublet exclusion. Data were analyzed using FlowJo analysis software (Tree Star).

V_HB1–8 Tg cells from single-cell splenocyte and peritoneal lavage preparations were subjected to CD43 negative selection (Dynal), CFSE labeled (1 μM CFDA-SE; InvivoGen), and transferred i.v. (5 × 10⁵) and i.p., respectively, into sex-matched WT CD45.2⁺ C57BL/6 recipients. PD-1^−/−V_HB1–8 Tg cells were transferred into CD45.1⁺ C57BL/6 WT recipients, unless otherwise indicated. In transfers of memory cells, CD43⁻ B cells were further depleted of any CD138⁺ cells and, in some cases, selected for IgG⁺ cells using biotinylated anti-mouse IgG1, IgG2a, IgG2b, and IgG3 mAbs (clones RMG1-1, RMG2a-62, RMG2b-1, and RMG3-1, respectively; BioLegend) in conjunction with Miltenyi streptavidin bead purification.

Peritoneal cavity (PerC) cells containing B_mems derived from V_HB1–8 Tg mice were harvested 2–3 mo postimmunization (i.p.) with 25 μg NP-Ficoll. In some experiments, B cells were selected using CD19⁺ beads (Miltenyi). Cells were CFSE labeled and cultured for 4 d in complete RPMI + 10% FCS (1 × 10⁶/ml) in media alone or with 5 ng/ml NP₄₀-Ficoll, with either 2 μg/ml rat IgG2b (LTF-2), rat anti-mouse PDL1 (10F.9G2), or rat anti-mouse PDL2 (TY25; all InVivo mAbs from BioXcell). On day 4, cells were stained in culture wells with fixable Live/Dead dye, NP-allophycocyanin, and fluorochrome-labeled mAbs to detect CD138, IgM, IgG, and CD45.1, as well as Countbright beads (Thermo Fisher) for enumeration. Cells were harvested from wells, washed, fixed in 1.5% buffered formaldehyde, and analyzed by flow cytometry.

Data are shown as means ± SEM. Differences between sample means were assessed using Student t test or one-way ANOVA with Tukey post hoc analysis. Survival analysis of Kaplan–Meier curves was performed using the log-rank test.

WT recipients of naive CFSE-labeled V_HB1–8 Tg (CD45.1⁺) CD80^negPLD2^neg B cells develop a CD19⁺CD138^negCFSE^loFSC^loB220^hiCD80⁺PDL2^+/− population after NP-Ficoll immunization (Fig. 1A), as we previously published (19). We assessed whether B_mem could be identified in normal non-Tg mice using these markers. Six weeks after immunization with 10 μg NP₄₀-Ficoll, WT mice harbored a population of endogenous NP-specific B cells bearing a CD19⁺CD138^negFSC^loCD80⁺PDL2^+/− phenotype (Fig. 1B). CD80⁺PDL2⁺ NP-specific B cells are scarce in naive WT mice but can be found for months after immunization (Supplemental Fig. 1A, 1B), consistent with what we have observed for adoptively transferred V_HB1–8 Tg B cells (19). Moreover, IgG⁺CD138^neg NP-specific B cells present in normal immune WT mice are largely CD80⁺PDL2^+/− (Fig. 1C), consistent with a B_mem phenotype. Non–Ag-specific IgG⁺CD138^neg B cells in these mice also express CD80, with a small fraction expressing PDL2 (Fig. 1D). Relative to non–class-switched NP-specific CD138^neg B cells, IgG⁺ NP-specific CD138^neg B cells express higher levels of CD80, PDL2, CD73, and CD21/35 but have comparable size (forward scatter) and B220 expression (Fig. 1E). Collectively, these data support that NP-Ficoll immunization elicits the formation of TI-2 B_mem in WT mice with a CD19⁺CD138^negFSC^loB220^hiCD80⁺PDL2^+/−CD73^+/−CD21/35^lo/hi phenotype, similar to that described for high-affinity V_HB1–8 Tg B_mems (19). Thus, V_HB1–8 Tg NP-specific B_mems adequately model the surface phenotype of normal NP-specific B_mems generated in response to NP-Ficoll, and these B cells may therefore be subject to similar modes of regulation by surface regulators (i.e., PDL2, CD80, CD73, etc.).

Given the capacity for PD-1 to regulate primary Ab responses to TI-2 Ags (26, 27), we assessed whether it also influenced the generation of TI-2 B_mems in non-Tg mice, defined as the expanded pool of Ag-specific B cells lacking CD138 and expressing IgG. As shown in (Fig. 2A, WT mice immunized with 10 μg NP₄₀-Ficoll generated a measurable pool of splenic CD138^negIgG⁺ NP-specific B cells that was significantly increased compared with that found in naive mice, as expected. However, frequencies and numbers of CD138^negIgG⁺ NP-specific splenic B cells were 2-fold higher in immunized PD-1^−/− mice. These cells had a CD80⁺PDL2^+/−FSC^lo phenotype, similar to WT mice. CD138^negIgG^neg NP-specific splenic B cell numbers in immune PD-1^−/− mice were not increased over WT numbers (Fig. 2B). Similar to findings obtained for spleen B cells, NP-specific CD138^negIgG⁺ peritoneal B cell numbers were also significantly increased in immunized PD-1^−/− mice relative to WT mice (3-fold), and CD138^negIgG^neg B cell numbers were also moderately (1.4-fold) increased over WT (Fig. 2C).

There were no differences in the frequencies or numbers of CD138^negIgG^neg or CD138^negIgG⁺ NP-specific B cells in naive PD-1^−/− versus WT mice (Fig. 2A–C), indicating the increase in IgG⁺ B_mems in PD-1^−/− mice was not due to an increase in naive precursors. Significantly increased BrdU labeling of NP-specific B cells in PD-1^−/− mice in the first 5 d postimmunization suggested increased IgG⁺ B_mem generation in PD-1^−/− mice was likely due to increased expansion early in the response (Fig. 2D). Consistent with this, BrdU labeling of B_mems between 5 and 6 wk postimmunization revealed little division (4–7% BrdU⁺) in IgG⁺ B_mems with no difference between WT and PD-1^−/− B_mems (Fig. 2E). However, there was a high level of BrdU labeling in the splenic CD138⁺ plasmablast/plasma cell pool at 6 wk, and a higher frequency of CD138⁺IgG⁺ cells was BrdU⁺ in PD-1^−/− mice. IgG⁺ B_mem remained significantly (3-fold) increased in PD-1^−/− versus WT mice at 6 wk (p = 0.002); however, we did not detect differences in the distribution of CD80⁺, PDL2⁺, or CD80⁺PDL2⁺ IgG⁺ B_mems relative to WT mice. Thus, PD-1 deficiency results in increased expansion of NP-specific IgG⁺ B cells in the early phase of the response to NP-Ficoll and this, as opposed to increased division in the established memory pool, explains increased NP-specific IgG⁺ B_mem cells in PD-1^−/− mice.

To determine whether B cell–intrinsic PD-1 expression regulates the generation of TI-2 B_mems, we generated V_HB1–8 Tg mice lacking PD-1 so that adoptive transfers could be used to perform rigorous B_mem analysis. We did not detect differences in the frequencies of NP-specific B cells among splenic and peritoneal B cell populations in naive V_HB1–8 Tg and PD-1^−/− V_HB1–8 Tg mice (Fig. 2F). We adoptively transferred CD43-depleted CFSE-labeled spleen B cells i.v. and CD43-depleted CFSE-labeled PerC B cells i.p. from CD45.1⁺V_HB1–8 Tg or CD45.2⁺PD-1^−/− V_HB1–8 Tg mice into recipient mice (CD45.2⁺ and CD45.1⁺ C57BL/6 WT mice, respectively). Recipient mice were immunized with 10 μg NP₄₀-Ficoll on day 1. Recipients of PD1^−/− Tg B cells generated significantly increased levels of NP-specific IgG (3-fold) relative to recipients of PD-1^+/+ Tg B cells (Fig. 2G), as well as increased IgM^a (p = 0.05) on day 10 (the peak of the response). Differences in IgG production were not observed when PD-1^−/− V_HB1–8 Tg B cells were transferred into muMT versus PD-1^−/− muMT mice (Supplemental Fig. 1C). Moreover, differences in Ab production were not observed when PD-1^−/− V_HB1–8 Tg B cells were transferred into WT versus PD-1^−/− mice (Supplemental Fig. 1D), supporting that PD-1 expression on Ag-specific B cells plays a major role in the inhibitory effect PD-1 has in regulating IgG responses to TI-2 Ags.

We next assessed whether B cell–intrinsic PD-1 expression regulated the formation of B_mems. As shown in (Fig. 2H, recipients of PD-1^−/− Tg B cells had significantly more NP-specific B cells in spleen, inguinal lymph node (LN), PerC, and bone marrow (BM), as well as significantly increased CD138⁺ Ab-secreting cells (ASCs) in spleen, LN, and BM 21 d postimmunization. In addition, recipients of PD-1^−/− Tg B cells had significantly more CD138^negCSFE^lo NP-specific B_mems. The frequencies of spleen and LN IgM⁺ B_mems were 2-fold higher, and BM frequencies were 25-fold higher, indicating PD-1 inhibits IgM⁺ B_mem generation (Supplemental Fig. 1E). The frequency of IgG⁺CD138^neg memory cells was also significantly increased in recipients of PD-1^−/− Tg B cells, with 3-fold higher frequencies in the spleen and 8- and 20-fold higher frequencies in the PerC and BM, respectively (Fig. 2I). At 5 wk postimmunization, PD-1^−/− Tg IgG⁺CD138^neg B cell frequencies remained significantly increased over PD-1–sufficient Tg B cells (Supplemental Fig. 1F); however, the frequencies of CD80⁺PDL2⁺ cells among and PD-1–deficient Tg IgM⁺ and IgG⁺ B_mems were similar (Supplemental Fig. 1G). Consistent with ELISA results, PD-1^−/− Tg B cells also gave rise to significantly increased CD138⁺IgG⁺ NP-specific ASC frequencies (Fig. 2I), although these frequencies were at least 10-fold lower than CD138^negIgG⁺ NP-specific B cell frequencies. Thus, B cell–intrinsic PD-1 expression inhibits primary NP-specific IgM and IgG production and the generation of both nonswitched and IgG⁺ B_mems in response to NP-Ficoll.

The failure to boost is a key characteristic of TI-2 Ab responses. Ag-specific IgG plays a role in suppressed boosting (18, 19), but it is probable that additional mechanisms are involved. PD-1 is upregulated on naive and memory V_HB1–8 Tg B cells after NP-Ficoll encounter in vivo (19), and NP-specific B cells in normal mice also upregulate PD-1 in response to both primary and secondary immunization (Supplemental Fig. 1H). Nonetheless, NP-specific B cells in immune mice fail to become properly activated, as evidenced by limited increases in CD86, CD44, size (FSC), and expansion relative to the primary responses made by naive B cells (Supplemental Fig. 1H, 1I). Given the upregulation of PD-1 on reactivated B_mem, we next investigated the extent to which PD-1 might suppress recall responses.

To examine whether B cell–intrinsic PD-1 expression suppresses secondary responses in a manner independent of its regulation of B_mem formation, we generated V_HB1–8 Tg and PD-1^−/− V_HB1–8 Tg B_mems in primary WT recipients (19) and then harvested, purified, and transferred CD43^negCD138^neg NP-specific memory spleen B cells into naive mice or mice previously immunized with 1 μg NP-Ficoll (day −35) (Fig. 3A). We compared responses of NP-specific B cells in naive and immune recipients 4 d postimmunization. V_HB1–8 Tg B_mems upregulated CD86 in recipient mice that were immunized for the first time, as did PD-1^−/− Tg B_mems, although PD-1^−/− B cells upregulated CD86 to significantly higher levels (Fig. 3B, gray bars). In contrast, V_HB1–8 Tg B_mems poorly upregulated CD86 in recipient mice that received secondary immunization after B_mem transfer (Fig. 3B, black bars). In fact, levels were not significantly increased over that for immune control recipient mice, which had been immunized only 35 d prior and were not boosted (Fig. 3B, white bars). However, PD-1^−/− Tg B_mems in immune recipient mice receiving secondary immunization upregulated CD86 to only a slightly lower degree than PD-1^−/− Tg B_mems in recipient mice that had been immunized for the first time (Fig. 3B, black bars versus gray bars). Four days postimmunization, a 4-fold increase in V_HB1–8 Tg and PD-1^−/− V_HB1–8 Tg NP-specific B cell frequencies was observed in naive recipients that were immunized at the time of transfer (Fig. 3C, gray bars). However, significant increases in NP-specific Tg B cells were not observed in recipients that were previously immunized and boosted, whereas significant increases were observed in boosted memory recipients of PD-1^−/− V_HB1–8 Tg NP-specific B_mems, albeit to a lesser extent than that found for naive recipients receiving Ag for the first time. Significant increases in PD-1^−/− Tg IgG⁺ NP-specific B cell frequencies relative to control mice that were not boosted at the time of transfer were also observed in recipients that were immunized for the first time or boosted, whereas these increases were diminished in recipients of PD-1–sufficient Tg B_mems (Fig. 3D). IgG Ab production from Tg B_mems was too low to distinguish from endogenous responses, but IgM^a responses were evident. IgM^a production was diminished in recipients that had been previously immunized and boosted relative to naive recipients, consistent with the effect of Ag-specific Ab in suppressing secondary responses (Fig. 3E). Nonetheless, PD-1^−/− Tg B cells generated significantly greater increases in NP-specific IgM^a over nonboosted recipients, indicating B cell–intrinsic PD-1 expression plays a major role in inhibiting secondary B_mem activation, expansion, and Ab production, even in the presence of Ag-specific Ab.

To specifically determine the extent to which PD-1 regulates the ability of IgG⁺ B_mems to respond to secondary Ag encounter in the absence of Ag-specific Ab, we again generated V_HB1–8 Tg and PD-1^−/− V_HB1–8 Tg B_mems, this time using intact V_HB1–8 Tg and PD-1^−/− V_HB1–8 Tg mice. On day 30, we positively selected IgG⁺ B cells from spleen and PerC of Tg mice and transferred equal numbers of NP-specific IgG⁺ B_mems into naive WT recipients (Fig. 3F). We immunized half of the recipients with NP-Ficoll (1 μg) and assessed cell expansion, differentiation, and Ab production 5 d later. As shown in (Fig. 3G, 5 d postimmunization, NP-specific IgG⁺ V_HB1–8 Tg B_mem frequencies increased 2-fold over frequencies in naive recipients, whereas PD-1^−/− V_HB1–8 Tg B_mems increased 20-fold. PD-1^−/− Tg B_mems also generated 6-fold greater IgG⁺CD138⁺ ASCs (Fig. 3H) and 3-fold greater NP-specific serum IgG (Fig. 3I) over V_HB1–8 Tg B_mems. As expected, NP-specific IgM^a was not detected above background (Fig. 3J), and recipients in all three groups produced comparable levels of NP-specific (endogenous) IgM (Fig. 3K). Collectively, these data demonstrate B cell–intrinsic PD-1 deficiency results in increased TI-2 IgG⁺ B_mem expansion, ASC differentiation, and IgG production.

Finally, we examined the extent to which PD-1 would regulate B_mem reactivation in immune recipients previously immunized with a higher (10 μg) dose of NP-Ficoll. V_HB1–8 Tg B_mems and PD-1^−/− V_HB1–8 Tg B_mems were generated in WT mice and then cotransferred into naive or previously immunized recipients, which were then immunized 1 d after transfer. IgG⁺ PD-1^−/− V_HB1–8 Tg B_mem frequencies were increased 55-fold 4 d postimmunization of naive mice versus 25-fold in previously immunized recipients, relative to B_mem frequencies in recipient mice that were not immunized after transfer (Supplemental Fig. 1J). By comparison, WT IgG⁺V_HB1–8 Tg B_mem frequencies increased 30-fold in naive versus 10-fold in previously immunized recipients. In recipients immunized for the first time, PD-1^−/− V_HB1–8 Tg IgG⁺CD138⁺ ASCs were increased 60-fold over WT Tg B_mems (Supplemental Fig. 1J). However, PD-1^−/− V_HB1–8 Tg IgG⁺CD138⁺ cells were decreased 30% in recipients that were boosted at the time of transfer. Thus, PD-1^−/− V_HB1–8 Tg B_mems exhibit significantly increased Ag-induced reactivation (i.e., expansion and ASC generation in IgG⁺ cells) relative to PD-1–sufficient V_HB1–8 Tg B_mems, but they are partially constrained in the memory environment.

To determine the extent to which PDLs, PDL1 and PDL2, regulate B_mem formation to TI-2 Ags, we assessed endogenous B_mem formation in PDL1^−/− and PDL2^−/− mice. Twenty-one days post–NP₄₀-Ficoll immunization, the frequency and number of IgG⁺ NP-specific CD138^neg B cells in spleens were significantly increased in PDL1^−/− (2-fold) relative to WT mice (Fig. 4A). The frequencies and numbers of unswitched CD138^neg cells were only ∼30% higher in immunized PDL1^−/− mice relative to WT mice (Fig. 4B). PDL2^−/− mice also had significantly increased frequencies (∼65% higher) of IgG⁺CD138^neg B cells relative to WT mice (Fig. 4A). Differences were not apparent in the PerC among groups of immune mice (Fig. 4A, 4B). Importantly, differences in NP-specific unswitched CD138^neg or IgG⁺CD138^neg B cells were not observed among WT, PDL1^−/−, and PDL2^−/− naive mice (Fig. 4A, 4B). Notably, CD80 expression was significantly lower on IgG⁺ B_mems from PDL1^−/− mice relative to WT and PDL2^−/− mice (Fig. 4C), a finding possibly related to the physical association between PDL1 and CD80 (33). Thus, PDL1 and, to a lesser extent, PDL2 significantly limit the generation of IgG⁺ TI-2 B_mems, and PDL1 significantly impacts the phenotype of TI-2 B_mem through promoting CD80 expression.

PDL1 is constitutively expressed by B cells, and PDL2 is upregulated on activated B cells, and both are significantly increased on B_mems in TD responses (34). This upregulation occurs on TI-2 Ag-specific B_mems along with other markers, including CD80 and CD73 (19), as well as IL-5Rα (CD125) (Supplemental Fig. 1K). To determine whether expression of PDL1 and/or PDL2 on cells other than Ag-specific B cells regulates TI-2 B_mem formation, we transferred naive CFSE-labeled V_HB1–8 Tg cells into WT, PDL1^−/−, and PDL2^−/− mice and assessed B_mem formation in response to NP-Ficoll immunization. PDL1^−/− recipients produced significantly more NP-specific IgG than WT mice, whereas PDL2^−/− recipients gave intermediate IgG responses (Fig. 4D). IgG⁺ B_mem generation was significantly increased (2-fold) in both PDL1^−/− and PDL2^−/− recipients relative to WT recipients (Fig. 4E). Notably, CD80 expression on V_HB1–8 Tg NP-specific B_mems was not different among WT, PDL1^−/−, and PDL2^−/− recipients (Fig. 4F), indicating B cell–expressed PDL1 supports CD80 expression on TI-2 B_mem. No difference in memory generation was observed when PD-1^−/− mice served as recipients (Supplemental Fig. 1L). Transfer of PD-1^−/− V_HB1–8 Tg B cells into WT, PDL1^−/−, and PDL2^−/− recipients also did not produce significant differences in B_mem generation, although B_mem frequencies and numbers were somewhat increased in PDL2^−/− recipients (Supplemental Fig. 1M). These data support that PDL1 and PDL2 limit generation of V_HB1–8 Tg TI-2 IgG⁺ B_mems because of their expression on a cell type other than V_HB1–8 Tg B cells (i.e., endogenous recipient cells) and interactions with PD-1 expressed on B cells. This result does not exclude the possibility that B cell–expressed PDL1 and PDL2 may also regulate these responses, either positively or negatively.

We next examined the extent to which PDL1 and PDL2 regulate B_mem reactivation. PDL1 and PDL2 are both expressed by TI-2 B_mems, with B-1b cells harboring the greatest frequency of PDL2-expressing B_mems (19). Using an in vitro reactivation assay of V_HB1–8 Tg memory B-1b cells, we assessed the effects of PDL1 and PLD2 mAb blockade on memory B-1b cells. As shown in (Fig. 5A, the NP-specific B_mem population was significantly increased in bulk cell cultures in the presence of NP₄₀-Ficoll and control IgG mAb relative to nonstimulated cultures. PD-1 and PDL1+PDL2 mAb blockade significantly increased the number of divided NP-specific B_mem cells (Fig. 5A). The number of NP-specific IgG⁺ and CD138⁺ cells also significantly increased (at least 3-fold) in response to PD-1, PDL1, PDL2, and PDL1+PDL2 mAb blockade relative to control mAb cultures (Fig. 5A). In cultures of purified CD19⁺ B cells containing V_HB1–8 Tg NP-specific B_mems, PDL1 and PDL2 blockade similarly increased the number of NP-specific IgG⁺ cells, CD138⁺ cells, and IgG⁺ CD138⁺ cells relative to control mAb (Fig. 5B). No effect was found for Abs alone in the absence of NP-Ficoll (Supplemental Fig. 1N), and we did not observe effects on PD1^−/− V_HB1–8 Tg B cells (Supplemental Fig. 1O). Thus, this in vitro data suggest PD-1, PDL1, and PDL2 negatively regulate the generation of IgG⁺ and CD138⁺ B cells from TI-2 B_mems after Ag reactivation, and that, along with B cell–expressed PD-1, B cell expression of PDL1 and PDL2 may participate in this inhibition.

Given our in vitro results, we next examined the extent to which PDL1 and PDL2 regulated B_mem reactivation and Ab production in vivo. We transferred naive V_HB1–8 Tg cells into naive WT recipients, immunized with NP-Ficoll, and 21 d later transferred V_HB1–8 Tg B_mems (enriched for IgM⁺PDL2⁺ memory B1b cells) (19) into new naive WT recipients. We then immunized recipient mice, which were given control, PDL1, PDL2, or PDL1+PDL2 blocking mAbs (depicted in (Fig. 5C). Given that the total IgG responses fell in the normal range of responses made by nonreconstituted mice, we measured NP-specific IgM^a and IgG2a levels, because these allotypes are produced by only Tg B cells. As shown in (Fig. 5D, PDL1 and PLD2 mAb blockade had variable effects on IgM^a secretion by B_mems, with NP-specific IgM^a levels for coblockade recipients declining more sharply than control mAb-treated mice. Recipients given PDL1 and PDL2 single blockade had lower IgM^a but significantly increased NP-specific IgG2a levels on day 7 relative to control mAb-treated mice (∼30-fold). However, PDL1 and PDL2 coblockade had the most notable effect on IgG2a and increased day 7 levels by 100-fold (Fig. 5D), with levels remaining significantly elevated over that of control mAb recipients out to day 21. We did not detect NP-specific IgG2a in WT recipients of Tg cells that had not been immunized. Thus, PDL1 and PDL2 both inhibit B_mem reactivation and Ab production in vivo, with the strongest increases in IgG production observed when both ligands are blocked.

To determine the extent to which PDL1 and PDL2 expression by Ag-specific B_mems versus other cell types controls B_mem reactivation and Ab production in vivo, we adoptively transferred V_HB1–8 Tg B_mems into PDL1^−/− and PDL2^−/− recipient mice using the same strategy employing blocking mAbs as described earlier. B_mems produced significantly more IgG2a in PDL1^−/− recipient mice given control mAb relative to WT control recipients (Fig. 5E). However, this was not further increased when PDL1^−/− recipient mice were given PDL1 blocking mAb; in fact, IgG2a levels declined. This result supports that Ag-specific B cell–extrinsic PDL1 expression plays a key role in negatively regulating TI-2 B_mem recall responses, whereas Ag-specific B cell PDL1 expression may have a more complex role in regulating responses, one of which may include regulating CD80 expression, which inhibits TI-2 B_mem recall (19). NP-specific IgG2a levels were also increased in PDL2^−/− mice given control mAb relative to WT control mAb-treated recipients (Fig. 5E), indicating non–Ag-specific B cell PDL2 expression also limits B_mem recall responses, either directly or through alterations in the PDL2^−/− environment (35). However, PDL2 blockade significantly increased NP-specific IgG2a responses in PDL2^−/− mice, indicating Ag-specific B_mem-intrinsic PDL2 expression plays an important role in suppressing B_mem IgG production. Collectively, these results support that PDL1-mediated inhibition of TI-2 B_mem reactivation primarily occurs through interactions between B_mem and other cell types expressing PDL1, whereas PDL2 expression on B_mems plays a key role in limiting B_mem reactivation.

By mechanisms that have not been elucidated, Ag-specific IgG suppresses B_mems from producing Abs in response to secondary Ag encounter (18, 19). Our results presented thus far indicate PD-1 negatively regulates the ability of B_mems to produce IgG in response to secondary Ag encounter when Ag-specific Ig is lacking (Fig. 3I), as well as IgM in weakly immune WT recipients of V_HB1–8 Tg IgM⁺ B_mems (Fig. 3E). We further examined the ability of PD-1^−/− mice to produce increased Ab after boosting using different TI-2 Ags. In response to low-dose TNP₆₅-Ficoll, some boosting was observed in WT mice (day 40 versus day 30), but PD-1^−/− mice generally produced greater increases in IgG after boosting (Fig. 6A). TNP₆₅-Ficoll generally produced modest boosting in PD1^−/− mice, albeit to a greater extent than in WT mice. Similarly, immunization of V_HB1–8 Tg mice with 25 μg NP-Ficoll followed by boosting with 100 μg NP-Ficoll failed to increase NP-specific IgM and IgG levels, whereas significant increases in both IgM and IgG were observed in PD-1^−/− V_HB1–8 Tg mice (Fig. 6B). In response to PPS3, PD-1^−/− mice generated more IgM after boosting relative to WT mice regardless of route and, in some cases, more IgG based on pairwise analysis, although increases were modest (Fig. 6C).

A failure of B_mems to differentiate into ASCs postinfection is a problem encountered in infections with encapsulated bacteria, such as Streptococcus pneumoniae. Consistent with this, Pneumovax23- and Prevnar13-immunized WT mice do not exhibit significant increases in PPS3-specific IgM and IgG after serotype 3 infection (Fig. 6D). However, we found PPS3-specific IgG levels significantly increase after S. pneumoniae infections in PD-1^−/− mice (1.5- to 3-fold). These results indicate PD-1 plays a role in inhibiting secondary responses to TI-2 Ags, including PPSs, even when encountered in the context of infection.

Based on the results showing PD-1^−/− mice exhibit PPS3-specific Ab boosting after pneumococcal infections, we hypothesized PD-1 deficiency and pathogen molecular pattern stimulation of TI-2-reactive B cells might enable greater enhancements in IgG production relative to either pathway alone. Our previous study demonstrated TLR4-based adjuvants enhance B cell responses to TI-2 Ags, including boosting, through TLR4 expressed on B cells (24). We therefore assessed responses to TNP-Ficoll in WT and PD-1^−/− mice when an MPL/TDCM-containing adjuvant was included. As shown in (Fig. 7A, both WT and PD-1^−/− mice generated significantly increased (3- to 4-fold) TNP-specific IgM responses when adjuvant was included. Adjuvant-supported IgM boosting was similar between WT and PD-1^−/− mice. Primary TNP-specific IgG responses were increased with adjuvant in WT and PD-1^−/− mice; however, responses were significantly higher (2-fold) in PD-1^−/− mice relative to WT mice (Fig. 7A). PD-1^−/− V_HB1–8 Tg B cells also produced significantly more (4-fold) NP-specific IgG in WT recipients that were immunized with adjuvant plus NP-Ficoll relative to those immunized with NP-Ficoll alone (Supplemental Fig. 1P), indicating the adjuvant effects on increasing IgG production were also observed when PD-1 deficiency was limited to Ag-specific B cells. Secondary TNP-specific IgG responses were moderately increased with adjuvant inclusion in both WT and PD-1^−/− mice, although responses did not differ, likely because of the high levels of primary Ag-specific IgG that suppress responses. Thus, these data demonstrate PD-1 deficiency results in enhanced adjuvant effects on primary IgG responses to haptenated Ficoll.

The MPL/TDCM-containing adjuvant increased IgM responses to PPSs within Pneumovax23 to a similar extent in both WT and PD-1^−/− mice (2- to 5-fold; (Fig. 7B). However, the adjuvant significantly augmented Pneumovax-specific IgG responses in PD-1^−/− mice over that found for WT mice. Similar results were obtained when PPS3-specific IgM and IgG responses were analyzed, although differences between adjuvant groups were more striking in the boost response, with PD-1^−/− mice producing as much as 8-fold higher PPS3-specific IgG relative to PD-1^−/− mice that did not receive adjuvant and ∼3-fold higher IgG levels relative to WT mice that had received adjuvant (Fig. 7C). We assessed the protective capacity of sera from these mice in a model of lethal serotype 3 S. pneumoniae challenge in CD19^−/− mice, which lack protective natural anti-pneumococcal Ab (36). Significantly increased survival was achieved using optimal (2 μl) quantities of pooled sera from Pneumovax + adjuvant-immunized PD-1^−/− (100%, n = 6) and WT (66%, n = 6) mice relative to untreated mice (0% survival; data not shown). However, at lower serum transfers (0.5 μl), only sera from Pneumovax + adjuvant-immunized PD-1^−/− mice were protective (Fig. 7D), demonstrating the significantly enhanced protective capacity that is achieved when PD-1 deficiency and adjuvant are combined in the context of Pneumovax immunization. Thus, a B cell–activating adjuvant enhances IgG responses to TI-2 Ags, including protective anti-PPS Ab levels, and does so in a more potent manner when PD-1 inhibition is relieved.

The lack of boosting to native polysaccharide-based vaccines after revaccination or infection is a barrier to establishing protection against infectious diseases. Our comprehensive study highlights the critical role the PD1-PDL regulatory pathway plays in negatively regulating recall responses to TI-2 Ags. First, we demonstrate PD-1 and its ligands, PDL1 and PDL2, inhibit the generation of TI-2 B_mem, with the most potent effects observed on IgG⁺ B_mems. This regulation is carried out through interactions between PD-1 upregulated on Ag-activated B cells and PDL1 and, to a lesser extent, PDL2 expressed on other cells. Further, we demonstrate TI-2 IgM⁺ and IgG⁺ B_mems upregulate PD-1 on secondary Ag encounter, and its interactions with either PDL2 expressed on B_mem or PDL1 and/or PDL2 on other cell types significantly decreases B_mem reactivation, expansion, class-switching, and Ab production. The potential for TI-2 B_mems to suppress their own reactivation through PD-1-PDL2 interactions reveals a novel form of auto-regulation not previously described. Finally, we demonstrate a B cell–activating adjuvant enables significant boosting of protective IgG responses to polysaccharide Ags when PD-1 inhibition is relieved. Collectively, our study (1) demonstrates a central role for B cell–intrinsic PD-1 in negative regulation of B_mem formation and reactivation in response to polysaccharide Ags, and (2) reveals combinatorial strategies aimed at relieving PD-1–mediated inhibition and further potentiating polysaccharide-specific B cell activation may improve protective responses to polysaccharide-based vaccines and the infections they are designed to prevent.

Evidence supporting the generation of B_mems in response to TI-2 Ags has been formally shown using adoptive transfers of NP-specific B cells from V_HB1–8 Tg mice (18, 19). Although the factors controlling TI-2 B_mem generation are poorly understood, our work demonstrates B cell–intrinsic PD-1, which is transiently upregulated on TI-2 Ag-activated B cells, limits establishment of B_mems, along with class-switching and IgG production in response to TI-2 Ags (26, 27). Consistent with this, the generation of switched B_mem is most affected by PD-1 regulation. In contrast with findings for TI-2 responses, PD-1 expressed by T cells (T follicular helper and regulatory cells) actually promotes B_mem generation in response to TD Ags, whereas B cell–expressed PD-1 appears to play a minor role (34,37,38). This can be explained by the critical role B cell Ag receptor signaling has in driving TI-2 Ag-dependent responses where cognate T cell help is lacking. PD-1 coengagement with the BCR suppresses major signaling processes required for division through recruitment of SHP-2 and/or SHP-1 phosphatases (26, 39) (K.M. Haas, unpublished observations). PD-1 may also impact bystander T cell help by suppressing expression of costimulatory molecules such as CD86 (26, 27, 40), as well as the expression of additional receptors required for survival, switching, and differentiation. Our work supports that interactions between Ag-specific B cell–intrinsic PD-1 and B cell–extrinsic PDL1 suppress B_mem generation, although this does not exclude a role for B cell–expressed PDL1. PDL2 expression is much more restricted; however, based on our results, PDL2 also limits B_mem generation. Notably, PDL2^−/− mice have significantly increased levels of IL-5 (35), which supports TI-2 Ag-induced IgG switching and Ab production (1). This could impact B_mems (which express higher levels of IL-5Rα) independently of PD-1, as could PDL2 interactions with its alternative ligand, repulsive guidance molecule b (41). Indeed, the trend for increased B_mem generation by PD-1^−/− V_HB1–8 Tg B cells in PDL2^−/− recipients relative to WT recipients raises this as a possibility that warrants further investigation.

PD-1 is also upregulated on B_mem encounter with TI-2 Ag and thus has the potential to downregulate Ag receptor-induced signals in B_mems on engagement with PDL1 and/or PDL2, both of which contribute to negative regulation of TI-2 B_mem reactivation. The expression of both PDL1 and PDL2 by TI-2 B_mem renders these cells capable of self-regulation because of homotypic interactions likely involving both Ag recognition and PD-1-PDL engagement. In vitro experiments with purified B cells support this possibility. Evidence for B_mem-intrinsic PDL2 expression in inhibiting B_mem activation was provided by both in vitro experiments and in vivo by using PDL2 ligand blocking mAb in PDL2^−/− mice reconstituted with PDL2⁺ B_mems. Although B cell–expressed PDL1 was found to reduce B_mem reactivation in vitro, PDL1 expression on non–Ag-specific B cells appeared to play a more dominant role in vivo, because V_HB1–8 Tg B_mems produced significantly higher levels of Ab in PDL1^−/− relative to WT recipient mice. In fact, PDL1 mAb blockade in PDL1^−/− recipient mice reduced Ab production by Ag-activated PDL1⁺ B_mems. It is possible this could have been because of the depleting effects of 10F.9G2 (a rat IgG2b mAb previously shown to have this effect) (42) on PDL1⁺ B_mems in the PDL1-deficient environment. PDL1 also has the potential to regulate reactivation via its association with CD80 expressed by B_mems. In the absence of PDL1, B_mems express low levels of CD80. The results of our previous study indicated an inhibitory role for CD80 in B_mem reactivation (19). Work on other cell types supports a role for PDL1-CD80 in cis interactions in diminishing CD80-CTLA4 and PDL1-PD-1 interactions in trans but enabling CD80 interactions with CD28 (33, 43). In the case of CD80⁺PDL2⁺ B_mems, PDL1 associated with CD80 on B_mems may shift PD-1-PDL1 interactions to PD-1-PDL2 interactions. PDL2 has a 3-fold higher affinity for PD-1 than PDL1 (44), and this may contribute to greater inhibition than that afforded by PDL1 interactions in the context of homotypic interactions. The extent to which interactions between B_mem-expressed PDL2 and B_mem-expressed PD-1 regulate B_mem reactivation remains to be fully established. Interactions between PD-1 on B_mems and ubiquitously expressed PDL1 provide an additional mechanism by which B_mem interactions with other cell types regulate reactivation. Our work thus far points to a strong inhibitory influence of TNF receptor ligand family members (CD80, PDL1, and PDL2) in TI-2 B_mem reactivation—regulation that is distinct from the roles these ligands have in regulating TD B_mems (21, 34, 45). However, our understanding of the complex interactions among these receptors and their ligands both in cis and in trans and their impact on the heterogenous TI-2 B_mem population is far from complete. Although the role(s) PD-1 and its ligands play in regulating B cell activation and function have not been fully elucidated, evidence thus far points to a role for PD-1 as a true checkpoint regulator that tempers T cell–independent B cell activation, and in particular IgG production and IgG⁺ B_mem generation, when cognate T cell help or danger signals are lacking.

Ag-specific IgG has been shown to also potently suppress TI-2 B_mem reactivation (18, 19). Although the mechanisms responsible have not been fully elucidated, epitope suppression and/or Ag clearance may be involved. In either case, TI-2 Ag crosslinking strength would be diminished and result in ineffective activation, as is supported by our finding of decreased blasting, CD86 upregulation, and expansion by B_mem in the memory environment. Nonetheless, when Ag-specific Ig levels decline, some persisting B_mems become responsive to Ags and produce Abs (19). Increasing Ag dose and/or providing additional B cell activation signals may help achieve the activation required to fully promote Ag-specific B_mem cell division, switching, and differentiation when Ag-specific Ig levels remain high. Consistent with this, priming with low-dose TI-2 Ag (haptenated Ficoll) in the presence of the MPL/TDCM-containing adjuvant enables greater boosting when a higher Ag dose is used in conjunction with adjuvant during secondary immunization (24). As evidenced by our study, overriding PD-1 inhibition may also lower the threshold for Ag receptor engagement in both primary and secondary responses. Thus, when combined with a B cell–activating adjuvant or infection bearing B cell–activating pathogen molecular patterns, the effects of ablating PD-1 inhibition may synergize to significantly increase TI-2 Ag-specific Ab. The increases in PPS-specific IgG achieved using this combinatorial approach were physiologically meaningful in a lethal type 3 S. pneumoniae challenge model, whereby sera from adjuvant-treated PD-1^−/− mice provided complete protection at limiting doses, but sera from WT mice were ineffective. In summary, our findings demonstrate a key role for PD-1 and its ligands in regulating the formation and functional reactivation of TI-2 Ag-specific B_mems and highlight new avenues to modulate protective Ab responses against TI-2 Ags relevant for human health.

K.M.H. has a patent filing related to PD-1 under consideration.