Plasma cells negatively regulate the follicular helper T cell program

Abstract

B lymphocytes differentiate into antibody-secreting cells under the antigen-specific control of follicular helper T (T_FH) cells. Here, we demonstrate that isotype-switched plasma cells expressed MHCII, CD80 and CD86 and intracellular machinery required for antigen presentation. Antigen-specific plasma cells could access, process and present sufficient antigen in vivo to induce multiple T_H cell functions. Importantly, antigen-primed plasma cells failed to induce interleukin 21 or Bcl-6 in naïve T_H cells and actively shut down these key molecules in antigen-activated T_FH cells. Mice lacking plasma cells displayed altered T_FH activity, providing evidence for this negative feedback loop. Hence, antigen presentation by plasma cells defines a new layer of cognate regulation that limits the antigen-specific T_FH program controlling ongoing B cell immunity.

INTRODUCTION

Protein vaccination induces high-affinity B cell memory and persistent circulating antigen-specific antibodies for long-lasting humoral immunity^1,2. Follicular helper T (T_FH) cells have emerged as a new class of immune regulatory cells^3–5 specialized to control the stepwise development of antigen-specific B cell immunity. Within the first week after priming, antigen-specific T_FH cells emerge^6–8 to initiate antibody secretion, isotype switch and the germinal center (GC) reaction⁹. Within the GC, T_FH cells regulate high-affinity memory B cell development^10–12 and the production of long-lived plasma cells¹³. Upon antigen re-challenge, memory T_FH cells promote antigen-specific memory B cell expansion and the rapid induction of high-affinity plasma cells^7,14. Thus, antigen-specific T_FH function is central to multiple facets of B cell immunity, but how this cognate regulatory activity is controlled remains poorly understood.

Antigen-specific T_FH development and function emerges progressively with separable requirements for cognate control. Initial T_FH programming occurs upon first contact with peptide-MHC class II (pMHCII) expressing dendritic cells (DCs) in the T cell zones of draining lymphoid tissue¹⁵. Loss of CCR7 and expression of CXCR5 relocates T_FH cells to B cell zones⁶ and facilitates contact with antigen-primed pMHCII-expressing B cells^16–18. Bcl-6 [http://www.signaling-gateway.org/molecule/query;jsessionid=c6ca4b34229c15d89939cccc445b981f9b070d6997a2?afcsid=A000369] is expressed by the early pre-GC T_FH cells⁸ and is sufficient and necessary to induce this program in vivo^19–21. Long duration intercellular contact¹⁷ with signals promoted by SLAM receptors¹⁷ and ICOS engagement through ICOS-L on B cells is required for pre-GC T_FH cell entry into the GC environment^22,23. Ectopic ICOS expression by naïve T_H cells promotes excessive T_FH expansion, enlarged GCs and autoantibody production⁹. IL-21 [http://www.signaling-gateway.org/molecule/query?afcsid=A001258] and IL-21R [http://www.signaling-gateway.org/molecule/query?afcsid=A001259] interactions are central to effective B cell immunity^24,25. While T_FH cells can develop in the absence of IL-21R^10,11, the lack of IL-21 signaling in B cells substantially compromises the GC reaction and post-GC plasma cell production^10–12. PD-1 expressed by T_FH cells has also been implicated in PD-L2 mediated control of GC B cell survival and post-GC plasma cell production¹³. Hence, clonal expansion and the cognate delivery of specific T_FH cell functions also require a highly-regulated program of progressive development in vivo.

Antibody secreting cells (ASCs) are the critical effector B cell compartment produced following antigen-specific T_FH regulation in vivo. Plasma cells are defined as post-mitotic ASCs and represent the terminally differentiated component of the ASC compartment^1,2. Outside the GC, plasma cell development requires T_FH regulation and occurs rapidly during the first week after protein vaccination. These non-GC plasma cells do not somatically hypermutate^26,27 but can antibody class switch and typically have short half-lives of 3–5 days²⁸. In contrast, GC plasma cell development involves affinity maturation and antibody class switch resulting in long-lived compartments of high-affinity plasma cells^1,2,29. The transcription factor Blimp-1 [http://www.signaling-gateway.org/molecule/query?afcsid=A003268] initiates plasma cell development in both pathways³⁰ exerting its control largely through repression of Bcl-6 and Pax-5 (ref. 31) allowing de-repression of XBP-1 (ref. 32,33). Moreover, IL-21 abundantly produced by T_FH cells^8,10–12, potently induces Blimp-1 expression³⁴ and plasma cell development in a STAT3 and IRF4 dependent manner³⁵. XBP-1 controls the unfolded protein response and supports many facets of the cellular secretory apparatus³². Decreased surface expression of B cell receptor and down-regulation of the MHCII transactivator CIITA³⁶ suggest that plasma cells remain highly specialized for antibody secretion in vivo and have a diminished capacity for immune regulation and responsiveness.

Contrary to these expectations, we demonstrate continued high expression of MHCII, CD80, CD86 and the intracellular machinery for antigen presentation in antigen-specific isotype-switched plasma cells directly ex vivo. Importantly, after priming in vivo antigen-specific plasma cells expressed pMHCII complexes and were able to activate antigen-specific T_H cells. Antigen-pulsed plasma cells induced proliferation and effector cell differentiation in vitro from naive antigen-specific T_H cells but promoted Blimp-1 in favor of Bcl-6 and IL-21 induction in the T_H cell compartment. Furthermore, plasma cells shut down IL-21 production and decreased Bcl-6 expression in activated T_H cells in an antigen-specific manner. In support of this negative regulatory function, CXCR5⁺PD-1⁺ T_FH cells accumulated to exaggerated numbers in draining and distal lymphoid tissues following immunization of mice lacking B cell-expressed Blimp-1 that do not produce plasma cells in vivo. Large numbers of plasma cells appear interspersed with T_H cells at the T-B borders and throughout the lymphoid tissue during the second week after antigen priming with evidence for T_H cell contact with plasma cells in situ. Finally, we diminished Bcl-6 and IL-21 expression in antigen-specific T_FH cells in vivo through adoptive transfer of antigen-pulsed plasma cells. These data reveal an antigen presentation function for plasma cells during adaptive immunity that serves to limit ongoing antigen-specific T_FH function. Hence, we propose a new layer of negative regulation during adaptive immunity that is a functional sensor of plasma cell production that can refine the development of antigen-specific B cell memory.

Results

Antigen-specific plasma cells express MHCII, CD80 and CD86

The antigen-specific B cell response to nitrophenylacetyl (NP) coupled to keyhole limpet hemocyanin (KLH) as a protein carrier is regulated by T_FH cells and directly accessible by flow cytometry^14,37. Following NP-KLH immunization, antibody-secreting cells can be quantified using intracellular labeling with antigen, cell surface antigen binding and antigen-specific antibody secretion by ELISPOT (Supplementary Fig. 1). Therefore, antigen-specific ASCs (IgM⁻CD138⁺) with distinct developmental histories can be isolated for subsequent analysis of function (Fig. 1a). By day 5 after secondary immunization using the TLR4 agonist based Ribi adjuvant system, >90% of isotype-switched antibody-secreting cells did not incorporate BrdU over the previous 24 h in vivo (Fig. 1b). Thus, the majority of antibody-secreting cells (specific and non-specific) used in this study can be considered non-cycling, terminally differentiated plasma cells.

Figure 1. Isotype-switched plasma cells retain expression of MHCII and co-stimulation molecules.

a, NP-specific plasma cell numbers over time (PI⁻Dump⁻IgD⁻ and NP⁺CD138⁺) in the spleen of C57BL/6 mice after primary and secondary immunization with NP-KLH (adjuvant only for day 5 memory white square)(mean±sem; n≥4, * p<0.05).

b, CD138 versus IgM expression on NP-specific cells (gated as PI⁻Dump⁻IgD⁻ and NP⁺) at day 5 after secondary challenge (left). BrdU expression on NP-specific IgM⁻CD138⁺ plasma cells at day 5 memory (middle). Frequency of BrdU⁺ NP-specific IgM⁻CD138⁺ plasma cells at day 3 and day 5 memory compared to total IgM⁻CD138⁺ plasma cell compartment (right), 20h after single IP injection with BrdU (mean±sem, n≥3).

c–d, I-A^b (MHCII) (c), CD80 and CD86 (d) expression with summary mean fluorescence intensity (MFI) on NP-specific IgM⁻CD138⁺ plasma cells compared to CD4⁺ T_H cells (T), CD11c⁺ DC and naive B cells (B) (mean±sem, n≥3, ** p<0.01, *** p<0.001).

Within the antigen-specific compartment, IgM⁻CD138⁺ plasma cells expressed high amounts of surface MHCII protein (Fig. 1c). All recently formed plasma cells (non-GC at day 7 primary; post-GC at day 14 primary and memory-response at day 5 memory) expressed equivalent amounts of MHCII as compared to naive B cells and CD11c⁺ DCs. Long-lived plasma cells that persist in the spleen (> day14 memory) or bone marrow (BM)(> day14 memory; data not shown) expressed lower but readily demonstrable amounts of MHCII. Isotype-switched plasma cells also up-regulated the co-stimulatory molecules CD80 and CD86 and retained abundant expression at all developmental stages after antigen experience (Fig. 1d). These data suggested that antigen-experienced plasma cells could retain the capacity for cognate intercellular interaction with pMHCII-specific T_H cells.

The same expression pattern for MHCII, CD80 and CD86 was evident in switched plasma cells regardless of antigen specificity and across different mouse strains (Supplementary Fig. 2). In contrast, many IgM⁺ plasma cells had lost MHCII expression with the majority also negative for CD80 and CD86 (Supplementary Fig. 3). The mRNAs encoding MHCII, its transactivator CIITA, CD80 and CD86 were present at very low but detectable levels in the switched plasma cell compartment (Supplementary Fig. 4) as expected from the negative impact of Blimp-1 on CIITA during plasma cell development³⁶. Thus, isotype-switched plasma cells express key surface proteins associated with MHCII-restricted antigen presentation even in the presence of low mRNA encoding these proteins in vivo.

Plasma cells retain antigen presentation machinery

Next, we investigated the mechanism used by plasma cells to control MHCII-restricted antigen processing and presentation. Invariant chain protein (Ii; also called CD74) and CLIP-I-A^b complexes were readily detected intracellularly in antigen-specific IgM⁻CD138⁺ plasma cells in quantities similar to resting B cells (Fig. 2a). Peptide editing complexes H2-M and H2-O were also found by direct intracellular labeling for these proteins (Fig. 2b). The CD74, H2-M and H2-O proteins were detected at similar abundance in non-specific switched MHCII^hi and MHCII^int plasma cells representing recently-formed and long-lived plasma cell compartments, respectively (Supplementary Fig. 5). The abundance of mRNAs encoding CD74, H2-M and H2-O were low but detectable in this switched plasma cell compartment (Supplementary Fig. 5b–d). The mRNA for carboxypeptidases Cathepsin (Cts) B and CtsZ associated with early endosomes and the aminopeptidases AEP and CtsH associated with lysosomes were present in switched plasma cells (Fig. 2c). Other cathepsin protease mRNAs were expressed and present in low abundance (GILT, CtsS, CtsL, CtsC and CtsD; Supplementary Fig. 6). Therefore, in addition to cell surface MHCII, CD80 and CD86, there is abundant evidence for intracellular machinery involved in antigen processing and presentation in the specialized MHCII peptide-loading pathway in vivo.

Figure 2. Plasma cells express intracellular machinery for MHCII antigen processing and presentation.

a–b, Intracellular expression of CD74 (Ii, Invariant chain) and I-A^b-CLIP (a), H2-M and H2-O (b) with summary MFI on NP-specific IgM⁻CD138⁺ plasma cells (PC) in C57Bl/6 spleens at day 5 post secondary immunization with NP-KLH compared to naive B cells (B) and CD4⁺ T_H cells (T)(mean±sem; n=3, ** p<0.01, *** p<0.001).

c, mRNA expression analysis by quantitative PCR for the indicated proteases in CD4⁺ T_H cells, CD11c⁺ DC, naïve B cells and IgM⁻CD138⁺ plasma cells (PC). Estimates are presented in arbitrary units (AU) relative to naïve T cells set as 1 (mean±sem, n≥3).

Antigen-specific plasma cells present antigen in vivo

While the machinery for antigen processing is in place, it remained important to address the capacity of plasma cells to present specific antigenic peptides in vivo. Using hen egg lysozyme (HEL) as the immunogen, we visualized directly HEL-specific plasma cells in polyclonal B10.BR mice (H-2^k–restricted) 5 days after a secondary boost with soluble antigen (Fig. 3a). Direct labeling with Aw3.18 (ref. 38), an antibody specific for the dominant helper T cell epitope HEL_48–61 in I-A^k, revealed substantial amounts of specific pMHCII on HEL-specific plasma cells (Fig. 3b). There was negligible background antigen binding or pMHCII complex expression when animals were primed with irrelevant pigeon cytochrome c (PCC) antigen before the soluble HEL boost (Fig. 3b). Thus, antigen-specific memory-response plasma cells present detectable amounts of the dominant pMHCII complex 5 days after soluble protein rechallenge in vivo.

Figure 3. Antigen-specific plasma cells process and present antigen in vivo.

a, Day 5 after secondary challenge with soluble HEL in B10.BR mice primary immunized with HEL (left and middle) or PCC (right). Isotype-switched B cell compartment (Dump⁻IgD⁻IgM⁻ and CD19⁺/CD138⁺) (left) displaying HEL-binding and CD138 expression (middle), frequencies (insert), total cell numbers (right) for switched HEL-specific plasma cells (mean±sem; n=3, * p<0.05).

b, HEL binding versus peptide-MHCII (pMHCII) expression (left panels) with summary MFI for pMHCII (right) on isotype-switched plasma cells (gated as Dump⁻IgD⁻IgM⁻ and B220⁻CD138⁺ as described in a (mean±sem; n=3, * p<0.05).

c, CD44 and CFSE expression on CFSE-labeled 5C.C7αβ TCR transgenic T_H cells cultured for 4 days with isotype-switched plasma cells isolated 10 days after adjuvant only or PCC protein vaccination compared to addition of MCC peptide. Frequency of responders (inserts), total CD44^hiCFSE^lo T_H cells (right) (mean±sem, n>3, ** p<0.01).

d, CD44 and CFSE expression (left) on CFSE-labeled 3A9αβ HEL-specific TCR transgenic T_H cells 24h after transfer into B10.BR: immunization with peptide alone (left), transfer of IgM⁻CD138⁺ plasma cells pulsed with 10μM HEL peptide (middle) (mean±sem; n=3), immunization with HEL in adjuvant (n=1) (right). Frequencies of responders (inserts), total CD44^hiCFSE^lo T_H cells (right) (mean±sem; n=3, ** p<0.01).

Plasma cells drive specific T_H cell proliferation

Ten days after primary adjuvant-driven immunization, the vast majority of antigen-specific plasma cells express isotype-switched BCRs with point mutations and evidence for antigen-driven selection during the GC reaction²⁷. To test the capacity of post-GC plasma cells to present antigen, IgM⁻CD138⁺ plasma cells were isolated 10 days following immunization with PCC and co-cultured with naïve PCC-specific 5C.C7αβ transgenic T_H cells (Fig. 3c). The switched plasma cells from PCC immunized animals (but not adjuvant only controls) promoted substantial proliferation of naïve antigen-specific T_H cells without addition of exogenous antigen. The addition of exogenous peptide promoted no further CD44 up-regulation or CFSE dilution in vitro. Thus, antigen-specific post-GC plasma cells express sufficient pMHCII and adequate co-stimulation in vivo to drive naïve antigen-specific T_H cell proliferation in vitro.

Next, to assess whether plasma cells could activate naïve T_H cells in vivo, we adoptively transferred HEL_48–61-pulsed plasma cells into animals containing CFSE-labeled 3A9αβ HEL-specific naïve transgenic T_H cells (Fig. 3d). In this transfer, HEL-peptide–pulsed plasma cells, but not the same dose of soluble HEL peptide, promoted activation and proliferation of T_H cells. Thus, plasma cells likely gain physical access to naïve T_H cells in vivo by migrating through lymphatics to the T cell zones of the draining LNs. In this manner, pMHCII-expressing plasma cells can activate antigen-specific T_H cells in situ. Hence, antigen-specific plasma cells access protein antigen in vivo, process and present pMHCII complexes in sufficient abundance to promote naïve T_H cell activation and proliferation.

Plasma cells cannot induce pre-GC T_FH development

Based on the capacity of switched plasma cells as antigen presenting cells (APCs), it became important to compare and contrast this activity with conventional APCs. The same numbers of switched plasma cells and CD11c⁺ DCs pulsed with either peptide or protein antigen, induced similar activation and proliferation in naïve antigen-specific T_H cells (Fig. 4a and Supplementary Fig. 7). Similar results were seen across two antigen systems (PCC and HEL) using different naïve antigen-specific T_H cell populations (Supplementary Fig. 8). Even substantially decreasing antigen dose or APC to T_H cell ratios revealed very little difference between the different types of APC (Supplementary Fig. 9). Furthermore, activated antigen-specific T_H cells purified by flow cytometry for qPCR after culture (Supplementary Fig. 10) indicated that both types of APC induced robust and equivalent expression of mRNA (Fig. 4b) and protein (Supplementary Fig. 11) for a series of known T_H cell derived cytokines. However, in stark contrast, antigen-pulsed DCs induced high expression of IL-21 and Bcl-6 in activated T_H cells that were notably absent from the T_H cells in antigen-pulsed plasma cell cultures (Fig. 4c). Switched plasma cells uniquely induced Blimp-1, the transcriptional repressor of Bcl-6 and the T_FH program. Thus, unlike DCs, switched plasma cells were unable to drive naïve antigen-specific T_H cells towards the T_FH cell fate in vitro.

Figure 4. Plasma cells induce proliferation, multiple T_H cell functions and Blimp-1 but not Bcl-6 or IL-21 expression.

a, CD44 and CFSE expression on CFSE-labeled 5C.C7αβ TCR transgenic T_H cells activated for 4 days in vitro with IgM⁻CD138⁺ plasma cells or CD11c⁺ DC alone (0), MCC (peptide) or PCC (protein) (mean±sem frequency, n≥10). Frequencies of responders are displayed within inserts (left panels) and summary for total activated CD44^hiCFSE^lo T_H cell numbers is presented for each condition (right) (mean±sem, n≥10).

b–c, mRNA expression by RT-qPCR for the indicated molecules in CD4⁺ T_H cells activated in cultures using CD11c⁺ DC or IgM⁻CD138⁺ plasma cell as antigen-presenting cells in the presence of PCC protein. Levels of expression are compared to levels in naive CD4⁺ T_H cells directly ex vivo. Signals from all samples are normalized to β2m and compared to naive T_H cell values set at 1 (mean±sem, n≥3, * p<0.05).

Plasma cells decrease IL-21 and Bcl-6 in T_FH cells

Antigen-specific T_FH cells are required for the generation of isotype-switched plasma cells in vivo. In this context, we considered it unlikely that naïve T_H cells would be the primary target of switched plasma cell APC activity in vivo. Therefore, we assessed whether antigen-primed plasma cells had an impact on the development or function of previously differentiated antigen-specific T_FH cells. To this end, we established primary T_H cell cultures using DCs and protein to induce T_FH cells (as above) and then subsequently divided the activated antigen-specific T_H cells across secondary cultures with isotype-switched plasma cells or CD11c⁺ DCs, with or without antigen. CD11c⁺ DCs supported further accumulation of antigen-specific T_H cells regardless of antigen addition (Fig. 5a). In contrast, plasma cells supported no further accumulation of T_H cells in the absence of antigen and significantly decreased T_H cell numbers in the presence of antigen (Fig. 5a). The antigen-dependent decrease occurred without changes in the abundance of Bcl2, Bcl-X_L, Bax or Bak mRNA in the T_H cells suggesting no substantial changes by apoptosis in the culture (Supplementary Fig. 12a). Thus, in contrast to naïve T_H cell activation, switched plasma cells had a broad negative impact on previously activated effector T_H cells.

Figure 5. Switched plasma cells selectively inhibit IL-21 and Bcl-6 expression in T_FH cells.

a, Total CD44^hiCD62L^lo T_H cells after secondary culture of in vitro-activated CFSE-labeled 5C.C7αβ TCR transgenic T_H cells in medium alone (0) or with CD11c⁺ DC or IgM⁻CD138⁺ plasma cells (PC) as APC with or without PCC antigen (mean±sem, n≥3, statistics to “0” condition, *** p<0.001).

b, IL-21 and Bcl-6 mRNA expression by RT-qPCR of in vitro-activated T_H cells reactivated in the indicated secondary culture conditions. Signals are normalized to β2m and compared to naive T_H cell values set at 1 (mean±sem, n≥3, statistics to “0” condition, * p<0.05, *** p<0.001).

c, mRNA expression for the indicated molecules by RT-qPCR in 5C.C7αβ transgenic T_H cells activated in vivo, sorted and reactivated in the indicated culture conditions. The white bar indicates the mRNA level of in vivo-activated T_H cells before culture. Signals as in b (mean±sem, n≥3, statistics to “0” condition, ** p<0.01, *** p<0.001).

d, mRNA expression for the indicated molecules by RT-qPCR from non-transgenic, in vivo antigen-activated Vα11⁺Vβ3⁺ T_H cells day 7 post PCC immunization and reactivated in the indicated culture conditions. Signals as in c (mean±sem, n=4 from two separate experiments, ** p<0.01, *** p<0.001).

While IL-21 is produced across multiple effector T_H cell subtypes⁸ it has been a useful indicator of selective T_FH differentiation in vitro^20–23. Importantly, there was a striking and antigen-dependent decrease in IL-21 controlled by switched plasma cells (Fig. 5b). This functional deviation was selective for IL-21 without impact on the production of other cytokines, such as tumor necrosis factor (TNF) and IL-10 (Supplementary Fig. 12b) demonstrating antigen-specific down-regulation of a major T_FH function. Furthermore, Bcl-6 is a major inducer of the T_FH developmental program in vivo and in vitro^19–21. In the secondary cultures, plasma cells significantly reduced Bcl-6 expression in the antigen-specific T_H cells in an antigen-specific manner (Fig. 5b). We extended these results to in vivo antigen-activated 5C.C7αβ transgenic T_H cells that produced demonstrable amounts of IL-21 after priming with antigen. Within 24 h of short-term co-culture, in contrast to antigen-pulsed CD11c⁺ DC, switched plasma cells decrease IL-21 expression and diminished Bcl-6 expression (Fig. 5c). This effect was selective as IL-4 and the transcription factor GATA3 were unaffected and T-bet expression similarly impacted with DC and PC presentation (Supplementary Fig. 12c). These trends were recapitulated using non-transgenic polyclonal V_α11⁺V_β3⁺ antigen-specific T_H cells at day 7 after priming (Fig. 5d; Supplementary Fig. 12d). Therefore, antigen-specific plasma cells negatively regulate the function and developmental programming of antigen-specific T_FH cells in vitro.

T_FH cells accumulate in the absence of plasma cells

Towards understanding the role of plasma cell presentation in vivo, we immunized C56BL/6 mice lacking Blimp-1 in the B cell compartment. An antigen-specific primary and memory B cell response develops in these mice without antibody-secreting cells from all stages of the immune response³⁰. Interestingly, the primary GC reaction is also significantly enlarged in these mice for reasons that remain unresolved (30; Supplementary Fig. 13a). Here, we quantified the numbers of CXCR5⁺ PD-1⁺ T_FH cells within the activated CD44^hiCD62L^lo ICOS^hi subset of CD4⁺ T cells that accumulated after primary immunization (Fig. 6a and Supplementary Fig. 13b). In the absence of plasma cells, CXCR5⁺PD-1⁺ T_FH cells accumulate to significantly higher numbers in the draining lymph nodes (LN) and spleen at all time-points tested (D9, D11 and D15). Within the T_FH compartment, the brightest T_FH cells (CXCR5⁺⁺PD-1⁺⁺) were significantly higher (D11; P < 0.01) in the absence of plasma cells, suggesting a preferential accumulation of GC T_FH cells in these mice⁵.

Figure 6. Elevated T_FH accumulation in the absence of plasma cells in vivo.

a, CXCR5 versus PD-1 expression on lymph node (LN) and spleen (SP) activated CD4 T_H cells (gated as CD16/32⁻B220⁻CD8⁻⁻CD4⁺CD62L^loCD44^hi) from littermate (WT; left) and Prdm1^flox/floxCD19^Cre/+ (right; cKO Blimp-1) mice 11 days after vaccination with NP-KLH in adjuvant (mean±sem frequency; n=3). Summary of total PD-1⁺CXCR5⁺ and PD-1⁺⁺CXCR5⁺⁺ activated T_H cell numbers (mean±sem, n=3, * p<0.05).

b, CXCR5 versus PD-1 expression on activated spleen CD4 T_H cells from littermate (WT, left) and cKO Blimp-1 (right) 11 days after vaccination with NP-KLH in adjuvant (mean±sem frequency; n=3). Both groups of animals received 5.0 × 10⁷ unfractionated LN and spleen cell mixtures IP and base of tail from C57BL/6 animals day 4 following secondary NP-KLH immunization. Summary of total PD1⁺CXCR5⁺ activated T_H cell numbers (mean±sem, n=3, * p<0.05; ns, not significant).

c, CXCR5 versus PD-1 expression on activated splenic CD4 T_H cells (gated as in above) from littermate (WT; left) and Prdm1^flox/floxCD19^Cre/+ (right; cKO Blimp-1) mice 7 days after vaccination with NP-KLH in adjuvant (mean±sem frequency; n=3). Summary of total PD-1⁺CXCR5⁺ and PD-1⁺⁺CXCR5⁺⁺ activated T_H cell numbers (mean±sem, n=3, * p<0.05).

After local development in the GC, high-affinity post-GC plasma cells migrate to distal tissues that favor longevity and/or function of the secreting compartment. In the absence of plasma cells, CXCR5⁺PD-1⁺ T_FH cells accumulate to significantly higher numbers in the mesenteric LNs, consistent with the loss of a local regulatory function for plasma cells in this region (Supplementary Fig. 13c). Exaggerated numbers of CXCR5⁻PD-1⁺ cells accumulated within the CD44^hiCD62L^lo compartment of CD4⁺ T_H cells in the BM without plasma cells (Supplementary Fig. 13d). No change occurred within the Ly6C⁺ memory CD4⁺ T_H cell compartment, suggesting a selective impact on PD-1⁺ T_H cells within this typically plasma cell rich environment. These data suggested an additional function for plasma cells within favored survival niches and perhaps for extended periods after initial development within local GC environments.

Next, we attempted to reverse the T_FH accumulation in the day 11 immunized Blimp-1 cKO mice. We transferred unfractionated LN and spleen cells from C57BL/6 animals 4 days after a secondary immunization with the same antigen-adjuvant combination. Transferring this source of antigen-specific plasma cells restored T_FH cell numbers in the immunized Blimp-1 cKO model without impacting cell numbers within littermate controls in the spleen (Fig. 6b) and significantly reducing elevated T_FH cell numbers in the LNs (Supplementary Fig. 14a). At day 7 after priming, the majority of antigen-specific T_H cells are within the follicular B cell areas but not within the GCs. At this early time-point, there was significant elevation of CXCR5⁺PD-1⁺ T_FH cells in both spleen and LN of Blimp-1 cKO animals (Fig. 6c and Supplementary Fig. 14b,c). Thus, in the absence of plasma cell development, the accumulation of elevated T_FH cells is consistent with the loss of a plasma cell-controlled local negative regulatory loop in vivo.

Plasma cells negatively impact T_FH cells in vivo

Following PCC immunization, IgD⁻CD138⁺ plasma cells are found in large numbers interspersed with T cells outside GCs at T-B borders and throughout T cell zones and the medullary regions (Fig. 7a–e). Furthermore, we provide evidence for co-localization of CD138⁺ plasma cells with V_β3⁺ T cells that include the dominant antigen-specific responder T_H cell compartment^7,8 with evidence for cell contact at multiple points between V_β3⁺ T cells and CD138⁺ plasma cells at the T-B regions of these LNs (Fig. 7f–h). Adoptive transfer of plasma cells pulsed with MCC peptide two days before the peak of the local antigen-specific T_FH response induced a selective decrease in Bcl-6 and IL-21 expression and not IL-4 or GATA-3 (Fig. 7i) on a per cell basis within purified CD44^hiCD62L^loCXCR5⁺ V_α11⁺V_β3⁺ T_FH cells as previously described^7,8. There was no influence on the splenic T_FH response in the lymphoid tissues that did not directly drain the site of adoptive transfer (Supplementary Fig. 15). Thus, excessive numbers of plasma cells can negatively impact the local T_FH program within antigen-responsive T_FH cells in vivo.

Figure 7. Plasma cells co-localize with CD4⁺ T cells and negatively impact antigen specific T_FH program in vivo.

a–h, Confocal microscopy of draining LNs from PCC-immunized B10.BR mice at day 10 after priming. The approximate location of images is indicated on the diagram, medulla (M), with regions indicated for B cell follicle (B), germinal center (GC), T cell area (T) indicated in panels. Labeled antibodies for antigens are based on representative color used as depicted separately for panels a–e and f–h. White arrowheads in h highlight examples of CD138⁺ plasma cell contact with Vβ3⁺ T cells. Representative sections from one of three similar experiments are shown. Scale bar represents 100μm.

d, mRNA expression for the indicated molecules by RT-qPCR in T_FH cells sorted as Vα11⁺Vβ3⁺CD44^hiCD62L^lo and CXCR5⁺ from lymph nodes of B10.BR mice day 7 post PCC vaccination that had received (PC) or not (0) plasma cell transfer 2 days before analysis. Signals are normalized to β2m and compared to naive T_H cell values set at 1 (mean±sem, n=3 across duplicate wells, * p<0.05, ** p<0.01).

DISCUSSION

The current findings contest the existing paradigm that plasma cells are only specialized for antibody secretion. We demonstrate that plasma cells can participate in the cognate regulation of ongoing adaptive immunity. Isotype-switched plasma cells process and present antigen in vivo and can function as highly competent APCs for naïve T_H cells. However, we propose that antigen-specific plasma cells act at a later phase in the developing immune response to negatively regulate antigen-specific T_FH cell function. The impact of negative regulation may extend into the GC reaction with influence on distal immune response sites and plasma cell survival niches such as the BM. We propose a novel layer of negative regulation during adaptive immunity that senses antigen-specific plasma cell production to refine the ongoing development of high affinity B cell memory.

Plasma cells are clearly specialized to secrete antigen-specific antibody. Blimp-1 orchestrates plasma cell development by repressing existing B cell programs (Aicda, Bcl-6 and Pax-5), blocking cell cycle entry (c-Myc and E2f1) and initiating major changes in the secretory apparatus (J-chain) and unfolded protein response (XBP-1)³¹. In the current study, we provide clear evidence for the maintenance of proteins associated with MHCII antigen processing and presentation, despite a low abundance of mRNAs. How the MHCII machinery is maintained in the presence of low mRNA abundance and beyond the acute phase of the initial primary response remains unclear^42,43. Long-term protein stability is possible, but more likely the low mRNA amounts are sufficient for production of protein in plasma cells highly specialized for protein secretion³¹. There have been previous reports of MHCII and co-stimulatory molecules on plasma cells^39,40 and transformed myelomas⁴¹. However, the capacity of plasma cells to function as competent APCs was completely unexpected.

Post-mitotic cells that are highly-specialized for secretion still need to sense and respond to environmental cues to exert effective regulatory function in vivo. We provide clear evidence for continued expression of BCR on newly formed and long-lived plasma cells. The T_FH compartment is already sub-specialized for cognate pairing with antigen-experienced B cells^3–5,44. Multiple co-stimulatory and adhesion molecule pairs are expressed across these two cell types to create strong, long duration interactions upon cognate contact^17,18. First contact between T_FH and B cells leads to increased IL-21 and Bcl-6 expression^10–12. Clear evidence also exists for cognate contact between GC T_FH cells and GC B cells by real time imaging⁴⁵ and the sorting of T_FH-GC B cell conjugates⁴⁶. However, the developmental consequences of these cognate interactions for the GC T_FH cells remain unclear. IL-21 production by T_FH cells is central to affinity maturation, GC persistence, function and B cell fate determination^10,11. In the current study, we demonstrate the capacity of plasma cells to dampen effector T_FH cell production of IL-21 in a negative feedback loop. We propose that these actions comprise an acute regulatory activity to contract T_FH function, thereby constraining ongoing B cell immunity. Further, with evidence for continued BCR expression and MHCII presentation machinery in long-lived plasma cells, it is tempting to speculate a further role for these interactions in the long-term survival of plasma cells.

Commitment to the plasma cell fate requires T_FH contact. In the spleen, non-GC plasma cells first appear in T cell areas, migrate out through the bridging channels into the red pulp^1,2. In the lymph nodes, plasma cells can accumulate at the T-B borders and the medullary cords⁴⁷. The post-GC plasma cell commitment is also considered to be dependent on GC T_FH signals. Evidence for PD-1 involvement¹³, a direct role for IL-21^10,11 and expression of Blimp-1 within the GC itself⁴⁸ further support this position. Under normal physiological conditions, it is difficult to see plasma cell differentiation within the GC compartment. However, in the absence of the chemokine EBI-2, selected GC B cells are incapable of leaving the GC and produce more discernible compartments of putative plasma cells in situ⁴⁹. The rapid production of plasma cells upon antigen rechallenge is dependent on CD4⁺ T_FH cells, however the distribution of memory response plasma cells in lymph nodes has not been well characterized. Nevertheless, these trends indicate that plasma cells or their immediate precursors reside within the same microenvironments as T_FH subtypes to enable their selective control of the T_FH program in vivo.

Elevation of T_FH cells in the absence of plasma cells provides support for the existence of this negative regulatory loop in vivo. There were exaggerated numbers of all CXCR5⁺PD-1⁺ T_FH cells in these animals across multiple time-points after antigen priming. The GC T_FH cell compartment expresses the highest amounts of PD-1 and CXCR5 (ref. 5) and accumulates to higher numbers in the absence of plasma cells in vivo. It has also been suggested that these cells express the highest ICOS expression and more recently GL-7 expression⁵⁰ as a GC subset-specific marker. While B cell memory develops in the absence of plasma cells³⁰, its capacity for long-term immune protection is limited in the absence of plasma cells. These animals are also devoid of secreted antibodies that may influence the contraction of adaptive immunity. However, in the absence of FcγR on T_H cells these effects are unlikely to directly influence the T_FH program. Furthermore, elevated GC T_FH cells correlates with dysregulated GC selection and autoantibody production⁵. Thus, in the absence of plasma cells, we would predict that increased GC T_FH cells results in elevated IL-21 production that could alter normal GC physiology and memory B cell development and function.

Our studies into the mechanism of negative regulation by plasma cells indicate the requirement for antigen to dampen antigen-specific T_FH activity. These in vitro studies indicate cognate contact is necessary but do not report on the effector mechanism. Recently, PD-1 expression by T_FH cells and PD-L2 on GC B cells was shown to play a role in regulating selection in the GC and the numbers of long-lived plasma cells¹³. IL-21 itself plays in important role in these dynamic activities and the generation of high affinity memory B cells^10,11. Thus, following cognate contact, the delivery of secondary modifying signals may be cell-associated and secreted. It is also plausible that different sets of GC T_FH cells may impact different types of GC B cells. For example, pairing of IL-4-secreting GC T_FH cells with IgG1-expressing GC B cells⁴⁶ introduced the possibility of T_FH-B cell subset dedication. Similarly, the rules governing antigen-specific regulation by plasma cells in vivo focus around specific cognate TCR-pMHCII interactions and may be subsequently modified by cell-associated and secreted molecular determinants of cell fate.

Under the normal dynamics of protein vaccination, constraining T_FH numbers and reduced IL-21 production may progressively increase competitive pressure critical for GC survival signals. In this manner, we predict that the negative regulation of GC T_FH function could enhance affinity maturation within the GC and promote memory B cells with higher overall affinities. This cellular `product to inducer' negative regulatory mechanism may also increase GC selection thresholds to help guard against potential self-reactivity in the GC. In the memory response, a negative regulatory loop would act as an antigen-specific plasma cell sensing mechanism to limit the accelerated and explosive plasma cell production that accompanies antigen re-challenge. We suggest that sensing the antigen-specific plasma cell product of the T_FH regulatory pathway may also provide an efficient means to constrain autoreactivity and fine-tune ongoing antigen-specific B cell immunity in vivo.

Cognate control of B cells by T_FH cells is an important and currently active arena of research. We believe that the impact of B cell-controlled antigen presentation on antigen-specific T_FH cell development and function remains poorly understood. This cascade of bi-directional regulation is central to effective adaptive immunity and regulates the emergence of antigen-specific immune memory. The new mechanism of antigen-specific plasma cell control that we uncover in the current study is one important facet of this largely unexplored cognate immune regulatory mechanism.

METHODS

Mice

C57BL/6, B10.BR, B10.BR-Thy1.1, 5C.C7-TCRαβ, 3A9-TCRαβ and Prdm1^flox/floxCD19^cre/+ mice were maintained in specific pathogen-free conditions at The Scripps Research Institute (TSRI). Prdm1^flox/floxCD19^cre/+ mice were obtained from K. Calame (Columbia University) and backcrossed 9 generations onto C57BL/6. The Scripps Research Institutional Animal Care and Use Committee approved all experiments.

Immunization

Mice were immunized subcutaneously with 400 μg NP-KLH, PCC protein (Sigma) or HEL protein (Sigma) in monophosphoryl lipid A based adjuvant⁷ for primary immunization. Secondary challenge was done ≥ 6 weeks after primary immunization and was either identical to primary or 40 μg soluble HEL in PBS.

Flow cytometry

Cell suspensions as previously described⁷. For surface staining, 4 × 10⁸ cells per mL for 45 min at 4°C with labeled mAbs: CD3 (145.2C11), CD8 (53-6.7), CD11c (HL3), CD62L (MEL-14), CD80 (16-10A1), CD86 (GL1), CD138 (281-2), B220 (RA3-6B2), IgM (II/41), GL7 (GL7), Ly6C (AL21), CXCR5 (2G8) and I-A^k (10-3.6; all from BD Pharmingen); CD4 (GK1.5), CD11c (N418), CD278 (ICOS, C398.4A), Gr1 (RB6-8C5) and I-A^b (KH74; all from BioLegend); CD19 (6D5), CD38 (90), CD44 (IM7), PD1 (J43) and F4/80 (BM8; all from eBioscience); CD11b (M1/70.15), V_β3 (KJ25), IgD (11.26), Aw3.18 NP (4-hydroxy 3-nitrophenylacetyl, Biosearch) and HEL (Sigma) were conjugated in the laboratory. For double biotinylated mAbs, cross reactivity was reduced by blocking with excess free biotin (Perbio).

For intracellular staining, surface labeled cells were then permeabilized and fixed using Cytofix/Cytoperm (BD Pharmingen) then labeled with mAbs: IgG1 (A85-1), IgG2a (R19-15), IgG2b (R12-3), IgG3 (R40-82), IgA (R5-140), IgM (R6-60.2), CD74 (In-1; all from BD Pharmingen); H2M (2C3A), H2O (Ob1); and I-A^b-CLIP (15G4). Samples analyzed on FACS Vantage SE or FACSAria III flow cytometers (BD Biosciences) and data processed using FlowJo software (Tree Star). Exclusion criteria used are referred to as “Dump” in the figure legend and included antibodies to CD4, CD8, Gr1 and F4/80 (for PC, naive B cells); B220, CD8, CD11b (for CD4⁺ T_H cells) and CD19, CD3 (for CD11c⁺ DC).

BrdU Analysis

Mice received a single injection of 800μg BrdU (BD Biosciences) IP and cells prepared using BrdU flow kit (BD Pharmingen).

Ag-presentation assays

After in vivo priming with protein antigen PCC or adjuvant only, plasma cells were sorted from lymph nodes 10 days after immunization using PI⁻Dump⁻IgD⁻CD138⁺IgM⁻ and cultured for 4 days at 5 × 10³ cells per well with 5 × 10⁴ naïve PCC-specific T_H cells (PI⁻Dump⁻CD4⁺CD44⁻H CD62L⁺) sorted from 5C.C7-TCRαβ transgenic mice and labeled with 5 μM CFSE (Molecular Probes) for 5 min at 24°C. For in vivo antigen-presentation, 9.0 × 10⁵ isotype-switched plasma cells were sorted from lymph nodes 5 days after immunization with adjuvant only, pulsed for 1 h at 24°C with 10 μM HEL peptide and transferred subcutaneously into B10.BR mice that had previously received the equivalent of 5 × 10⁶ CFSE-labeled 3A9-TCR transgenic T_H cells from donor 3A9 spleens. T cell proliferation was analyzed 3 days later. For in vitro antigen-presentation assay to naïve T_H cells, 5 × 10⁴ CFSE-labeled naïve PCC-specific T_H cells were cultured with 5 × 10³ plasma cells or DC (PI⁻Dump⁻CD11c⁺), sorted from lymph nodes 5 days after immunization with adjuvant only, and with 10 μM MCC_88-103 peptide (Anaspec) or 10 μM PCC protein as indicated.

For antigen-presentation assay to activated T cells, PCC-specific naïve T cells were activated either in vitro for 4 days with DC in the presence of PCC, or in vivo for 7 days after transfer of 5 × 10⁵ 5CC7αβ transgenic T_H cells into B10BR immediately immunized with PCC-adjuvant or sorted antigen-activated V_α11⁺V_β3⁺ T_H cells from B10BR mice without transfer as previously described⁸. Subsequent activated T_H cells (PI⁻Dump⁻CD4⁺CD44^hiCD62L^lo) were sorted and cultured at a 10:1 ratio with plasma cells or DCs with or without PCC for 3 days or 24 h. All cultures were done in RPMI medium supplemented with 10% (V/V) FCS, 1× Penicillin-streptomycin-Lglutamine, 50 μM 2-mercaptoethanol (Invitrogen). For adoptive transfer into Blimp-1 cKO and littermate controls 1.0 × 10⁸ unfractionated spleen and LN mixtures from day 4 secondary NP-KLH in Ribi immunized C57BL/6 mice. Cells were distributed evenly IP and base of tail into recipient animals at day 9 after initial priming of both groups of mice. For antigen-presentation assay to antigen-specific T_FH cells in vivo, 3.0 × 10⁵ isotype-switched plasma cells were sorted from lymph nodes 5 days after immunization with adjuvant only, pulsed for 1 h at RT with 10 μM MCC peptide and transferred subcutaneously into B10.BR mice that had been immunized 5 days earlier with PCC in adjuvant. T_FH were sorted 2 days after for gene expression analysis.

ELISPOT assays.

Plasma cell function was measured by ELISPOT as described previously¹⁴ using NPBSA or anti-mouse IgM and IgG (Southern Biotechnology Associates).

mRNA estimation using RT-qPCR

For qPCR directly ex vivo, 1 × 10⁶ T_H cells (PI⁻Dump⁻CD4⁺), DCs (PI⁻Dump⁻CD11c⁺), naïve B cells (PI⁻Dump⁻B220⁺IgD⁺IgM⁺), or isotype-switched PC (PI⁻Dump⁻IgD⁻CD138⁺IgM⁻) were sorted either from lymph nodes of HEL-adjuvant-immunized B10.BR mice 1 day post-immunization (T cells, DCs) or from spleen of HEL-adjuvant-immunized B10.BR mice 5 days post-secondary immunization (B cells, PCs). For gene expression analysis following antigen-presentation assay, activated T_H cells (PI⁻Dump⁻CD4⁺CD44^hiCFSE^lo) were sorted from individual wells after culture. For gene expression analysis following plasma cell transfer in vivo, antigen-specific T_FH cells (PI⁻Dump⁻CD4⁺Vα11⁺Vβ3⁺CD44^hiCD62L^loCXCR5⁺) were sorted from B10.BR mice 7 days after PCC immunization. All samples were sorted for RNA extraction, cDNA synthesis and qPCR as previously described⁸. mRNA purified after in vitro Ag-presentation assay to naïve T_H cells (Fig. 4b & 4c) was amplified with two rounds of in vitro transcription as previously described⁸. The primers used are included in Supplementary Table 1. The results were analyzed with Rotor-Gene 6.0 software or StepOne software v2.0.

Confocal microscopy

Inguinal lymph nodes were embedded in Tissue-Tek OCT compound (Sakura) and snap-frozen on dry ice. 6-8 μm sections were fixed in cold acetone. Staining were performed in PBS containing 10% FCS, 0.1% sodium azide with labeled Abs: IgD (11.26), V_β3 (KJ25 labeled in the laboratory); CD90.2 (53-2.1), CD138 (281.2, both from BD biosciences), anti-rat IgG (vector) and rhodamin red-streptavidin (invitrogen). Confocal images were analyzed with an Olympus Fluoview 500 confocal microscope.

Statistical analysis

Mean values, standard error of the mean (s.e.m.) values, Student's t-test (unpaired) and Mann-Whitney test were calculated with GraphPad Prism (GraphPad Software).