The Shaping of a B Cell Pool Maximally Responsive to Infections.

B cells develop in multiple waves throughout ontogeny beginning from extra-hematopoietic precursors in the yolk sack as early as embryonic day E7, then the fetal AGM region (splanchno-pleura), fetal liver and eventual the postnatal bone marrow ( 3 ). The earliest populations of fetal-derived B cells belong mainly to a population of B-1 cells , which differs in repertoire and function from the later, predominantly postnatal-developing B-2 cells ( 4 - 6 ). Some of the fetal-derived B-1 cells are maintained into adulthood, but whether the distinct waves of prenatal development result in B-1 cells of distinct functions is currently unknown. While transfer of the bacterial microbiota or of bacterial pathogens from mother to fetus is unlikely, given the strong maternal-fetal barriers ( 7 ), viral infections of a fetus can occur. Indeed, recent studies demonstrated a progressively more diverse repertoire of fetal B and T cells during development, with B cells preceding T cell development. B cells circulate in the fetal blood already by about week 12 of gestation in the human ( 8 ) and de novo fetal-derived IgM responses to viral infections have been measured in fetuses by week 24. Thus, the earliest developing fetal B cells can respond to external insults not prevented by the existing anatomical barriers. The bone marrow takes over as the site of continued hematopoiesis beginning shortly after birth. Following the release of B cells that are selected against strong self-recognition, from the bone marrow into the blood, these transitional B cells enter the spleen. In the spleen, they undergo another selection process that results in the further elimination of autoreactive B cells ( 9 ). Selected B cells then mature into either spleen-resident marginal zone (MZ) B cells, a process requiring NOTCH and ADAM10 signaling, or into mature follicular B (FOB) cells ( 10 - 12 ). The positioning of MZ B cells adjacent to the splenic marginal sinus, where arterial blood enters the white pulp, places these cells into immediate contact with any circulating pathogens or their components ( 13 ). FOB cells instead home to the lymphoid follicles. B cell migration to and their positioning within the follicles is guided by the chemokine receptor-ligand pair CXCR5/CXCL13, and the receptor EBI2 (GPR183) and its ligand, 7α,25-dihydroxycholesterol ( 14 ). The latter organizes the follicle into inner and outer zones. FOB cells and most MZ B cells derive from the same bone marrow precursors postnatally and are thus considered the later-developing “B-2” cells, although a subset of MZ B cells appears to develop during early fetal development and to be maintained by self-renewal ( 15 ). FOB cells recirculate via the blood between follicular regions of secondary lymphoid tissues, including the spleen, lymph nodes and Peyer’s Patches, the latter located in the small intestine and containing chronic germinal centers ( 16 ). While in circulation, they can also accumulate in the vasculature of a number of organs, where the interaction of B cells with the endothelium may contribute to the regulation of leukocyte composition of the tissue, but also imprint the transcriptional profile of the localized B cell populations ( 17 ). Alterations in the interaction of B cells with the endothelium can contribute also to the modulation of disease processes of the vasculature, such as observed in atherosclerosis ( 18 ). Remarkably little is known about the interaction of B cells with the endothelium at sites of infection-induced inflammation. However, B cells and perhaps plasmablasts and terminally-differentiated plasma cells can migrate to these tissues, where they secrete large quantities of antibodies ( 19 ). Interestingly, a subset of plasma cells with features of B-1-derived plasma cells ( B-1PC ) has been shown to also act as a source of IL-10 ( 20 , 21 ). Depending on the organ, formation of tertiary lymphoid structures that support B cell activation and differentiation can lead to local differentiation to memory B cells ( MBC ) or long-lived plasma cells ( LLPC ), as has been reported for the lung ( 22 , 23 ). Little is known about the role of IL-10 producing “ regulatory B cells ” in infections, which have been studied mainly in the context of inflammation and autoimmunity. We refer to recent reviews on these cells ( 24 - 26 ). Plasma cells are particularly numerous in the lamina propria of mucosal tissues that are exposed to and colonized by the microbiota, regulating the microbiota, mostly through secretion of dimeric IgA that is transported by epithelial cells expressing the polymeric Ig receptor to the luminal site of the mucosa ( 27 ). Indeed, the colonization by the microbiota is a pre-requisite for their presence ( 28 , 29 ). Interestingly, the skin contains predominantly IgM-secreting plasma cells, whose presence appears to be independent of microbial exposure ( 30 ) consistent with systemic natural IgM production ( 31 ). Enhanced plasma cell accumulation is seen also in instances of chronic tissue inflammation, such as in endometriosis, arthritis, myositis and granulomatous skin inflammation ( 30 , 32 ), as well as in classic granulomas forming in the lungs after infection with Mycobacterium tuberculosis ( 23 ), to name but a few. Thus, plasmablast and plasma cell responses provide a localized effector arm of humoral immunity. B cells are present also into the pleural and peritoneal cavities, accumulating there in a process that requires CXCL13 ( 33 ). B-1 cells are the dominant B cell subset, although conventional B cells are present also ( 4 ). B-1 cell accumulation in the cavities begins at about 2 weeks after birth following their initial expansion in the spleen. B-1 cells are highly sensitive to innate stimuli and function as sentinels. In response to infection-induced innate stimuli, such as Type I IFN or TLR stimulation, they rapidly migrate from the cavities to the regional secondary lymphoid tissues, where they differentiate into antibody or cytokine-secreting B-1 or B-1PC ( 34 - 37 ). The role of the body cavity B-2 cells, which are imprinted to express a BCR repertoire and transcriptional profiles distinct from that of FOB cells in secondary lymphoid tissues, has not been sufficiently explored ( 38 ). In addition to supporting the development of B cells from hematopoietic stem cells, the bone marrow harbors terminally-differentiated LLPC that develop from activated B cells in response to infections. Their migration to and retention in the bone marrow requires their expression of the chemokine receptor CXCR4 and secretion of the ligand, CXCL12 (SDF-1) by bone marrow stromal cells. The stromal compartment of the bone marrow provides plasma cells with a “survival niche”, in which LLPC secrete antibodies for extended periods of time ( 39 ). Another site of plasma cell accumulation is the splenic red pulp. In contrast to bone marrow LLPC, splenic plasma cells appear to be mostly short-lived (SLPC). Despite intensive research, it remains unknown whether the difference between LLPC and SLPC is due to the differences in the survival niches they occupy, or due to any cell-intrinsic differences ( 40 ). B cells are most numerous in secondary lymphoid tissues, but circulate continuously throughout the vasculature and even through tissues and are broadly distributed throughout the body. Their migration into secondary lymphoid tissues is controlled through chemokine and integrin-mediated processes, while their migration from lymph tissues into efferent lymphatics and blood is critically regulated by signaling through the sphingosine-1-phosphate receptor S1PR in processes similar between B and T cells ( 41 - 43 ). The presence of B cells in the afferent lymphatics, albeit at lower numbers than T cells, further suggests their continuous entry and exit also from solid tissues, even in the steady-state ( 44 , 45 ). Physiological triggers, such as microbiota colonization on mucosal surfaces, as well as pathological triggers of chronic inflammation result in the local accumulation of B cells and plasma cells. Overall, these findings indicate that B cells contribute to immunity against pathogens not only through the systemic secretion of antibodies, but also as active participants of local tissue-surveillance.

B cells express surface TLR 2 and 4 and endosomal TLRs TLR3, TLR7, TLR8 and TLR9. Endosomal TLR-engagement can enhance FOB cell responses by providing additional proliferation and differentiation signals that synergistically enhance BCR signaling in FOB cells following antigen internalization ( 160 ). This enhanced B cell responsiveness supports the robustness of pathogen-specific antibody responses to both viruses and bacteria ( 161 - 164 ), but it may also lead to antibody-mediated autoimmunity ( 165 , 166 ). Early studies showed a particular dependence of IgG2a/c generation on the expression of MyD88, which has now been linked to MyD88-dependent IFNγ-signaling, a cytokine long known to drive CSR to IgG2a/c ( 86 , 167 ). B cell-intrinsic absence of MyD88 was shown to reduce B cell proliferation and germinal center responses ( 160 , 168 ) and reduced terminal differentiation to antibody-producing plasmablasts and plasma cells ( 161 - 164 ). These are important observations that have obvious implication in the regulation of B cell responses to pathogens. Interestingly, even in the absence of BCR stimulation, TLR9 and TLR4-signaling induces phosphorylation of Syk ( 169 , 170 ), and TLR7 signaling phosphorylates Btk ( 171 ), two phosphatases critical for BCR-mediated stimulation, demonstrating a remarkable but also puzzling interconnectedness of TLR and BCR signaling. While FOB cells are not particularly sensitive to TLR4-stimulation, the B-1 and MZ B cell subsets show much stronger responses to TLR engagement, as briefly summarized below. Thus, non-cognate inflammatory signals created by the infection are critical modulators of cognate B cell responses, affecting their quality. The emerging data also underscore how features of B cell responses during infectious diseases parallel those observed in B cell-mediated autoimmunity. Rather than indicating that autoimmune diseases are triggered by infection, it suggests that PAMPS and DAMPS may initiate common signaling pathways. Understanding the regulation of pathogen-induced effective humoral responses can inform potential treatment strategies that modulate these pathways to either enhance immunity or suppress the unwanted outcomes in autoimmune diseases.

Natural infections occur mostly via the large mucosal surfaces of the gastrointestinal and respiratory tracts, less frequently through the reproductive tract or via breaches of the skin barrier. What follows infection is a rapid orchestration of local immune responses that initially are composed of mostly innate cells, but can involve tissue-resident memory B cells (MBC) and memory T cells when present due to a prior insult. The presence of local MBC cells has been associated with increased protection ( 46 , 47 ), as seen previously for tissue-resident T cells ( 48 ). In addition, and critical for the initiation of a strong immune response, dendritic cells and macrophages carry pathogen-derived antigens via the afferent lymphatics to the local “draining” lymph nodes. Pathogens, whole or in part, can also arrive in the lymph nodes as free antigens. Lymph nodes are designed to filter these antigens, preventing pathogens from spreading beyond the locally affected tissue area. When local lymph node filter functions are insufficient, antigen will enter the blood and reach the spleen. The white pulp of the spleen is structured similar to lymph nodes but filters blood-borne antigens/pathogens. “Conventional” or FOB cells are the most numerous B cell subset in lymph nodes and in the spleen. The phenotype of FOB cells is characterized by high expression of the IgD-BCR, varied expression of the IgM-BCR, intermediate expression of the complement receptor CD21 and expression of CD23 ( 49 ). B cells are antigen-presenting cells equipped with antigen-binding receptors, the B cell receptor (BCR). The unique structure of the lymph tissue follicles strongly supports B cell antigen-encounter. Antigen-delivery can occur through afferent lymphatics that shunt small antigens into the porous conduit system that emanates from the sinus and traverses the follicles ( 50 ). In lymph nodes, afferent lymphatics drain into the subcapsular sinus (SCS), where CD169+ F4/80- SCS macrophages are located at its floor, and onto which antigens larger than those traveling through the conduit system (> 70 kDa), or antigen-antibody complexes, can adhere ( 51 ). SCS macrophages are critical for displaying non-processed antigens to B cells in the follicular outer areas, rich in 7α,25-dihydroxycholesterol, that underlie the SCS floor, and into which the SCS macrophages stretch their antigen-covered processes ( 41 , 52 ). FOB cells do not have to be antigen-specific to acquire antigen. They can bind complement-tagged antigens and/or antigen-antibody complexes via their complement receptors and then transport and delivery the antigen to lymph tissue stromal cells termed follicular dendritic cells (FDC) ( 52 ). FDC in turn can then present the antigen to antigen-specific FOB cells located in the center of follicles, supporting the establishment and maintenance of germinal centers . Similarly, arteriolar blood-derived antigens pool in the splenic marginal zone sinus adjacent to the white pulp of the spleen. Here, antigen-uptake is facilitated in part by resident MZ B cells, which can shuffle between the marginal zone and the follicle, transporting non-processed antigens to FDC, which then present unprocessed antigens to FOB cells ( 53 - 55 ). FDCs are critical components of antigen-specific B cell responses, as their presence in the follicle and their ability to display antigen-antibody complexes for extended periods of time are required for successful initiation and maintenance of GC responses, considered hallmark T-dependent B cell responses ( 56 - 59 ). FDC are part of the stromal cell network of secondary lymph tissues and are derived from perivascular endothelial cell precursors ( 60 ). Interestingly, FDC precursors are present throughout the vasculature and upon signals from inflammatory cytokines such as lymphotoxin and TNF can differentiate to FDC, explaining the establishment of tertiary lymphoid tissues at local sites of chronic inflammation/infection ( 60 ). Some infections result in the dissolution of the normal lymphoid tissue organization and structure, disrupting the proper functioning of these organs, including the appropriate positioning of cells for antigen-capture, antigen-presentation, and antigen-specific CD4 T - B interactions. Following infection with Salmonella typhimurium, such changes were shown to depend on LPS-mediated stimulation via TLR4 which resulted in reduced production of CCL21, the chemokine responsible for the organization of the lymph tissue T cell zone ( 61 ). Similar disruption of the lymph node structures with dissolution of T- and B-cell zones was observed also in response to infection of mice with B. burgdorferi , although these alterations were independent of MyD88 and TRIF, and thus of TLR signaling ( 62 , 63 ), as well as following infection with Toxoplasma gondii ( 64 ). The frequency of antigen-specific FOB prior to a primary infection has been estimated to be in the order of 1 cell in 2 x 10 5 – 1 x 10 6 cells ( 65 ). The efficient and varied modes of antigen-delivery to secondary lymphoid B cell follicles outlined above are critical, as they enhance the likelihood that these rare antigen-specific B cells rapidly encounter and bind their “cognate” antigen. Important for the understanding of B cell response regulation, cognate antigen recognition initially occurs in the inflammatory milieu generated in response to an infection. For example, direct Type I IFN signaling was shown to affect B cells strongly early during viral infections, prior to cognate T-B interaction ( 66 - 68 ). This cytokine induced the upregulating of many IFN-regulated genes in B cells. It enhanced expression of CD69, which can trap B cells in the draining lymph nodes ( 69 ), caused the upregulation of the endosomal TLR 3 and 7, thereby enhancing responsiveness of B cells to internalized PAMP-containing antigens, and it enhanced surface expression of the co-stimulatory molecule CD86, driving enhanced activation and differentiation of B cells following cognate interaction with T cells ( 66 , 67 , 70 ). Lack of type I IFN dependent direct-signals were shown to reduce B cell activation, IgG production and plasma cell differentiation following influenza virus infection ( 66 - 68 ). Type I IFN, as well as other inflammatory stimuli also induce various cells, including epithelial cells, monocytes, dendritic cells, and neutrophils to secrete B cell activating factor (BAFF or Tnfsf13b) ( 71 , 72 ). BAFF is a critical B cell survival factor required for the maintenance of B cells beyond the transitional 1 stage of development ( 73 ). Enhanced secretion of BAFF by innate leukocytes was shown to support the development of neutralizing antigen-specific B cell responses during infection, as recently shown for infection with West Nile Virus, ( 74 ). These findings are significant, as they indicate that in vivo B cell activation is achieved through a diverse array of signals that include but are not limited to BCR signaling and co-stimulation by T cells. While T cell responses have long been appreciated to require polarizing signals for differentiation into appropriate effector CD4 T cell subsets (TH1, TH2 etc), much less is known about how these signals may drive distinct differentiation pathways in B cells. Yet, class-switch recombination (CSR) is determined by the cytokine milieu in which B cells are placed and polarization of B cells similar to those observed in CD4 T cells can similarly be achieved ( 75 ). This was further demonstrated by the identification of B cells expressing the TH-1 polarizing transcriptional regulator T-bet in various infections (see below). Recognition, i.e. binding of cognate antigens to the BCR of FOB cells triggers classic adaptive B cell responses by inducing a complex BCR signaling cascade that results in a) BCR signalosome reorganization and antigen-internalization, b) endosomal reorganization causing the loading of internalized antigen into MHCII c) transcriptional changes associated with enhanced B cell survival and B cell entry into the cell cycle for clonal expansion, d) relocation of FOB cells from follicles to the T-B border region, and e) the upregulation of MHCII, costimulatory molecules and chemokines productive interaction of B cells with CD4 T cells ( 76 ). At the same time, CD4 T cells are activated by antigen-presenting dendritic cells in the T cell zones of the same secondary lymphoid tissues. Their activation results in the upregulation of CXCR5, supporting their migration towards the B cell zone and into the T-B border, and the induction of CD40L, a critical costimulatory receptor engaging with CD40 on antigen-activated B cells. The presentation of peptides in the MHCII complex of B cells initiates T-B interaction when CD4 TCR bind to their cognate antigen. T and B cells may recognize distinct peptides derived from the same pathogen, due to the linked recognition of conjugate antigens by the B cell and then subsequently by CD4 T cells following presentation of another epitope of that protein within MHCII. Interestingly, a recent study on the glycoprotein gp120 of HIV demonstrated that CD4 T cells may not only recognize the peptide backbone of a protein, but may also engage with a MHCII-presented glycosylated peptide, where the nature of the glycosylation determined antigen-recognition ( 77 ). The data thus expand the universe of potential pathogen-derived antigens that can generate T-dependent B cell responses to include glycoproteins. Small viruses may engage and be internalized in toto by B cells through their BCR, which could lead to antigen-presentation of many distinct viral antigens to T cells, independent of the epitope-specificity of the B cell. Indeed, studies on the small, enveloped and negative strand RNA virus, influenza A, demonstrated that IgG responses to the surface hemagglutinin (HA)-spike protein were dependent on CD4 T cells, but that the support was similarly provided by CD4 T cells specific for the intraviral nuclear capsid or matrix protein ( 78 ). However, larger viruses as well as most bacteria and protozoa may not be taken up intact by B cells, unless they are the targets of pathogen infection. In a series of elegant studies, Sette and colleagues defined parameters of epitope recognition and antigen immune prevalence and immunodominance by CD4 and CD8 T cells by studying immunogenic peptides and antigens of vaccinia virus, a relatively large pathogen that contains over 200 open reading frames ( 79 ). They demonstrated that the generation of antibodies to a Vaccinia virus antigen correlated strongly with the presence of CD4 T cells of that same specificity ( 80 ). Furthermore, enhanced presence of antigen-specific CD4 T cells did not boost a specific B cell response, unless the CD4 T cells were specific for the same viral proteins than the B cells ( 80 ). The strong correlation between the specificity of B cells and CD4 T cells is further underscored by findings that B cells importantly expands the CD4 T cell compartment of cells with the same specificity (( 81 ) and references therein). Following antigen-uptake, FOB cells upregulate CCR7 and migrate to the T-B border zone, where they secrete CCL4, a chemoattractant for CD4 T cells ( 82 ). CD4 T cells migrate to the same region following their upregulation of CXCR5 in response to priming by dendritic cells in the T cell zone ( 83 ). The initial CD4 T-B interactions in the border region are critical. They induce the further expansion of B cells through cell-cell interaction via MHCII-TCR, CD86-CD28, ICOSL-ICOS and CD40-CD40L. The latter supports B cell survival and proliferation and induces expression of the enzyme activation-induced cytidine deaminase (AID), encoded by the gene aicda , and critical for CSR ( 84 ). Induction of AID further supported by secretion of cytokines such as IL-4 or IFN-γ provide early signals directing the isotype profile of the antibody response ( 85 , 86 ). Similarly, the interaction of CD4 T cells with B cells drives the polarization of the T cells towards a T follicular helper cell (T FH ) fate ( 87 ). Through continued engagement of CD4 T FH with B cells, a process that requires expression of “signaling lymphocyte activation molecule-associated protein” (SAP) ( 88 ), the further fate of the B cells is determined. In addition to harboring activated CD4 T cells and B cells, the T-B border zone also contains innate-like lymphocytes such as ILCs and invariant natural killer T cells (iNKT) that can interact with B cells. iNKT express an invariant α/β TCR that recognizes the canonical glycolipid α-galactosylceramide presented within the non-classical MHCI molecule CD1d expressed by B cells and other antigen-presenting cells ( 89 ). While mice only express one CD1 allotype (CD1d), humans harbor additional allotypes (a-e) that can present various glycolipids to a more diverse repertoire of NKT cells ( 90 ). Many bacterial pathogens express glycolipids. Studies on Streptococcus pneumoniae and B. burgdorferi have demonstrated the importance of iNKT cells in supporting B cell responses to these pathogens through antigen-specific FOB cells interacting with iNKT cells via antigen-presentation through the CD1d complex ( 91 , 92 ). These interactions appear very similar to interactions between classical peptide-restricted CD4 T cells, in that Bcl6+ iNKT cells required expression of CD40L for upregulation of AID and SAP, and supported B cell expansion and differentiation via secretion of IL-21, although the induced B cell responses appeared more short-lived ( 91 - 94 ). iNKT cells can also provide critical non-cognate support for B cell responses through secretion of cytokines, as has been shown following influenza virus infection, a virus that does not express glycolipids ( 95 ). Thus, although the original view of T-B interaction was restricted to that involving protein antigens, there is now support for T-dependent B cell responses for pathogen-derived glycan, lipids and protein antigens, greatly expanding the number of pathogen-antigens against which humoral immunity can be induced by natural infection or vaccination. Taken together, the initial clonal expansion of antigen-activated FOB cells occurs at the time of antigen encounter and is enhanced through costimulatory and cytokine signals provided at the T-B border zone ( 83 , 96 ). This process supplies an increased number of antigen-specific and often, but not always, class-switched B cells ( 97 - 99 ). The further fate of the B cell is determined by these signals and involves either their return to the follicles, where they initiate GC responses that give rise to MBC and LLPC ( 56 ), or they remain in the extrafollicular space, as antigen-experienced B cells/MBC or differentiate into plasmablasts. The latter congregate in so-called extrafollicular foci (EF) in the T-B border and the medullary cord area of lymph nodes or red pulp of the spleen ( 97 ). As outlined below, extrafollicular responses amplify the existing humoral effector B cells to eliminate acute infections and, I would argue are the main effector response to infections. GC responses, on the other hand, use the infections to alter the existing B cell repertoire through expansions of a more diverse and adapted set of B cell clones that can participate in EF responses following their differentiation into MBC, or provide protective antibodies directly following their terminal differentiation to LLPC ( Figure 1 ). Despite earlier work suggesting that MBC development depends on GC, more recent work, enabled by the use of fluorescent antigen-baits for identification of antigen-specific B cells, suggested GC-independent formation of non-switched IgM+ as well as class-switched IgG+ effectors/MBC ( 65 , 100 , 101 ). Although their specific developmental paths remain to be more firmly established, it appears likely that they derive following (T-dependent and T-independent) B cell activation at the T-B border. They may form from FOB cell clones that do not receive sufficiently strong BCR-signals, or qualitatively different signals than required for EF plasmablast formation. IgM+ “non-switched” MBC, which seem to separate into B cells that are IgM+ IgD− and those that have a naïve, IgD+ IgM+ phenotype ( 102 ), can respond to reinfection with rapid seeding of either EF or GC responses, thus in this regard behave similar to naïve B cells and thus are not a priori precluded from EF responses ( 103 ). The data are significant as they suggest that quiescent, antigen-experienced, and clonally-expanded non-switched B cell populations, including populations that by phenotype resemble naïve B cells, shape the repertoire of the circulating B cell pool, skewing it towards expanded numbers of B cells that have the capacity to be stimulated by pathogens. MBC’s apparent potential for self-renewal ( 103 ), further suggests that these cells remain in the peripheral circulating B cell pool long-term. By virtue of their increased frequencies alone, they can support more rapid responses to recall infections. The data may also explain how MBC responses can be found to pathogens an individual has no history of having encountered previously. By retaining a naïve-like (IgM+ IgD+, or IgM+ IgD−) state, infections with distinct pathogens that share epitopes, perhaps conformational in nature, can provide an initial wave of pathogen-specific antibodies of an appropriate quality, i.e. their isotype and glycan profile. The humoral response quality is then shaped by the inflammatory milieu of the newly encountered pathogen, rather than dependent the legacy of the previous infection that let to their initial induction and may support the development of cross-reactive humoral immune responses to broader classes of pathogens, as recently shown for cross-reactive B cell responses to LPS O-antigen in humans ( 104 ). Recent studies have shown that engagement of TLR9 following BCR-mediated antigen-stimulation in the presence of T cell help result in the development of a population of CD11c+ B cell subset that expresses the transcription factor T-bet ( 105 , 106 ). The first description of B cells with this phenotype was by Winslow and colleagues, who noted that a population of IgM-secreting extrafollicular CD11c+ plasmablasts emerged rapidly and independent of T cell-help following infection with the intracellular bacterium Ehrlichia muris ( 107 ), as well a population of CD11c+ CD19+ MBC that developed somewhat later in the response ( 108 ). Collectively, additional studies have revealed that this CD11c+ B cell subset harbors a transcriptional profile dominated by expression of the transcription factor T-bet ( 109 - 111 ), a well-known transcriptional regulator of CD4 TH1-differentiation ( 112 ). Plasmablasts and MBC with this phenotype emerge in situations of IFN-γ-driven infections of both humans and mice, such as following chronic infection with HIV ( 113 , 114 ) and repeat infections with Malaria ( 115 ), but also after acute influenza virus infection ( 86 ). While initial reports of MBC cells with this phenotype in chronic HIV-infected patients ( 114 ) as well as in the aging and in autoimmune diseases ( 116 ) suggested that these cells were “atypical” MBC that are showing signs of exhaustion and dysfunction, subsequent studies have demonstrated that these cells represent a population of B cells poised to develop into plasmablasts and that their emergence is related to the inflammatory milieu in which they develop. Elegant studies by Lund and colleagues recently defined the transcriptional control of these cells, demonstrating that B cell-intrinsic signaling via IFN-γ induces the induction of T-bet during influenza infection. T-bet expression then orchestrates transcriptional changes that suppresses the initial IFN-γ-driven inflammatory signaling profile towards one that enables these cells to respond to BCR-signaling with rapid upregulation of Blimp1+ and IRF4 and thus to initiate differentiation to plasma cells ( 86 ). (EF) are generated by antigen-specific B cells that undergo rapid rounds of proliferation followed by their differentiation into CD138+ Blimp-1-expressing plasmablasts that secrete either IgM or class-switched Ig. This response type is induced by all mature peripheral B cell populations: B-1 cells, MZ B cells and FOB cells, underscoring its significance for the survival of the host. While many EF responses require initial interaction of B cells with CD4 T cells, EF responses can form in response to T-independent antigens. Here, in addition to strong antigen-BCR interaction, TLR signals, and sufficient availability of BAFF and other cytokines supplied by innate immune cells, such as ILC, NK cells, or even neutrophils, can replace the need for T cell-derived co-stimulation of FOB cells. In addition, however, and as discussed below, splenic MZ B cells and B-1 cells respond more rapidly and more strongly to such signals compared to the FOB cells and may be responsible for an early wave of T-independent plasmablast responses. Class-switch recombination of EF-bound FOB cells is induced likely following CD40-CD40L interaction with CD4 T cells in the T-B border zone ( 97 - 99 ). During primary infections, EF-derived antibodies show only limited signs of affinity maturation. This may be due to the fact that continued engagement of these cells with CD4 T cells via CD40 is not required in the EF and thus that CD40-induced AID expression, responsible for both class-switch recombination and somatic affinity hypermutation, may not be sustained ( 84 , 97 ). Furthermore, recent studies showed that the transcriptional regulator Pax-5 supports AID expression, suggesting that the downregulation of Pax-5 during plasmablast differentiating further represses AID expression ( 85 ). This is in contrast to ongoing interactions of B cells with CD4 T FH in the GC, which continues to drive expression of AID, explaining the difference in the levels of affinity maturation seen between EF and GC B cell responses. Importantly, this does not mean that all EF responses are of low affinity. Indeed, development of EF responses requires a relatively high affinity BCR for cognate antigen binding ( 117 ), presumably because high affinity interactions induce strong IRF4 expression ( 118 ), a key transcriptional regulator of plasma cell development ( 119 ). For example, Gerhard and colleagues noted high-affinity early antibody responses during primary influenza A/Puerto Rico/8/34 vaccination of BALB/c mice to the Cb-site of HA1, encoded by germline encoded antibodies of the C12 idiotype. These responses did not generate MBC and did not participate in later antibody-responses ( 120 ). We subsequently identified B cells expressing this idiotype to form EF but not GC responses following primary influenza infections ( 121 ). Thus, when available in the existing repertoire, high-affinity B cell clones are favored for EF-development. This is also the case during recall responses, or during infections with related pathogens that induce cross-reactive responses and can explain the appearance of highly mutated plasmablasts in the blood of patients within a few days of acute infection during recall responses ( 122 , 123 ). Taken together, EF responses are rapidly induced and can provide critical, early and protective antibody responses of a variety of affinities. Given the need for strong BCR-signaling to initiate B cell differentiation ( 118 ), the notion that FOB-derived EF responses generate mostly low-affinity antibodies seems inconsistent with available evidence and it wrongfully diminishes the strong impact and importance of this B cell immune effector mechanisms for host survival. Such notion might have arisen from studies with model antigens in mice, in which the pre-immune repertoire simply does not contain clones able to mount a strong EF response. Indeed, during recall responses, extrafollicular or GC-derived MBC preferentially feed into the EF pathway for enhanced immune responses ( Figure 1 ). Based on these data, the EF response appears to be a central and critical effector response of the humoral immune system, not simply a short-lived, low-affinity response, replaced by GC. The outcome of EF development is the rapid generation of antibodies that critically support host-defenses ( Figure 2 ). Although current evidence suggests that the developing plasmablasts are relatively short-lived, with antibody-secretion usually occurring a few days prior to the establishment of GC B cell responses in the same lymph tissue ( 97 , 121 ), given the half-live of IgG in the order of 3-5 weeks, these responses nonetheless will outlast typical infections. Indeed, during acute infections with influenza virus, the kinetics of the EF response strongly correlates with that of virus-clearance, while the slower-developing GC do not form until the virus is already cleared ( 57 , 121 ). The early presence of antibodies can limit pathogen spread and thereby reduce the risk of overshooting cellular immune responses, which often engender higher levels of tissue destruction. The antibodies can also facilitate provision of co-stimulatory signals to FOB cells by forming complement-tagged antigen-IgG immune complexes that bind to CD21 ( 124 ). Via that same route, they can also help transport antigen to FDCs ( 51 , 52 ), thereby strongly supporting GC B cell responses. Tethering the immune complexes to the surface of FOB cells may also enhance even low-affinity BCR to engage with antigen. Co-ligation of BCR and immune complexes bound to the inhibitory FcγRIIb, the only IgG-FcR expressed by B cells, would also override any FcγRIIb-mediated inhibitory effects on the B cell response and instead provide support GC development ( 125 , 126 ). IgM-antigen complexes may fulfill similar functions as IgG-immune complexes, given that IgM strongly binds to complement ( 127 ). Indeed, the absence of secreted IgM has been associated with strong reductions in the ensuing IgG responses to infections and immunizations ( 127 ). FOB cells also express a Fc-receptor for IgM (FcμR) ( 128 ). Its absence was shown to phenocopy the reduced IgG responses to influenza virus infection seen in secreted IgM-deficient mice, suggesting that IgM-binding to the FcμR is important for IgM-enhancement of the subsequent IgG response ( 129 ). Thus, EF-derived antibody responses are critical for both, direct neutralization and opsonization of pathogens, as well as for optimal GC B cell responses. Few studies have documented mechanisms by which pathogens may exploit the EF response as an immune evasion strategy. One example is secondary dengue virus infections in individuals living in Dengue-endemic regions. Dengue repeat-infections induce very strong MBC-driven plasmablast responses, which in cases of heterotypic infections can be directed against non-neutralizing, cross-reactive epitopes. These non-neutralizing antibodies opsonize the virus, which perversely enhances disease severity in some patients, due to enhanced virus-infection of macrophages and other FcR-bearing cells ( 130 , 131 ). Host cell infection via IgG− but not IgA-mediated opsonization of plasmablast-derived antibodies was reported also for patients infected with Mycobacterium tuberculosis ( 132 ). Most recently, a study on infections with Plasmodium yoelii in mice suggested that rapidly proliferating plasmablasts can compete with GC B cells for nutrients, starving activated B cells and abrogating the establishment of GC responses. Interestingly, the inhibitory effects of EF plasmablasts on GC responses was overcome by supplementation of the drinking water with L-glutamate ( 133 ). These are remarkable observations providing mechanistic explanations for clinical observations on malnutrition and insufficient development of pathogen-specific immunity ( 134 ). Driving EF responses, while diminishing GC responses that provide MBC that can feed into future EF responses, as well as strong antibody-mediated protection via formation of LLPC makes evolutionary “sense” in that Plasmodium requires frequent reinfections of a host in Malaria-endemic areas. But it also suggests that strong EF responses may have a “cost” associated, that is usually balanced through their rapid involution. In response to infections or immunization the de novo formation of GC is delayed by a few days compared to that of extrafollicular responses, taking upwards of 10 - 14 days. Multiple excellent recent reviews have discussed the regulation of GC B cell responses to both, model antigens and infectious agents and the reader is referred to these for an in-depth discussion ( 56 , 135 - 137 ). Briefly, GC contain two histologically distinct compartments: the dark zone and the light zone. The dark zone encompasses a recently discovered subarea, “the grey zone” ( 138 ), in which rapidly proliferating B cells incorporate frequent point mutations into the hypervariable antigen-binding region of the Ig-locus. B cells with missense mutations undergo apoptosis and are eliminated by tangible body macrophages. Successfully mutated B cells will enter the major dark zone compartment, differentiate and re-express their BCR, and then move to the light zone which contains antigen-presenting FDC and CD4 T FH . Light zone B cells undergo selection, and only those that compete successfully for antigen-binding, and thus can present antigen for subsequent engagement with CD4 T FH , are selected for further rounds of proliferation. Continued competition for antigen and interaction with CD4 T cells selects for BCR with increasing binding-strengths to the selecting antigen. While much emphasis has been placed on the increasing affinity of individual GC-derived B cell clones, there is actually scant evidence that enhanced affinity is responsible for the development of protective B cell responses to pathogens ( 57 , 139 - 141 ). Importantly, however, and in contrast to the EF response that relies on the available B cell repertoire, the GC response actively shapes the B cell repertoire, driving diversity by selecting BCRs strongly binding to antigens from pathogens that could infect the host. While EF usually rapidly resolve within a few days after their formation, GC can persist for extended periods of time, even after the infection is cleared. Following influenza infection of mice in which the virus is cleared within 7-10 days, mediastinal lymph nodes contain GC for up to 5 months, although their total number diminishes as the lymph node involutes. This is consistent with the long retention of antigen on the surface of FDC, which act as the main antigen-presenting cell for B cells in the follicles ( 88 ). Chronic GC are also found in the mesenteric lymph nodes draining the gastrointestinal tract and in Peyer’s Patches of the small intestine, in response to microbial stimuli, suggesting that these structures are “repurposed” for differing antigens over time. Recent data show that the chronic Peyer’s Patch GC expand somatically mutated “public” clones, i.e. clones that occur in many individuals, further supporting the idea of these structures driving diversification of a repertoire of specificities to common antigenic structures ( 142 ). The long-lived GC are reminiscent of the processes of B cell diversification observed in the appendix of rabbit and sheep, among other species, that rely on gene conversion for generating a diversified repertoire upon interaction with microbiota-derived antigens ( 143 ). I argue that these findings further support the idea ( 57 ) that the primary goal of the GC responses cannot be solely the removal of an ongoing infection, but rather the use of antigens from the infection as a means to shape the repertoire of the B cell compartment on “important” antigens. Important antigens would be those that are derived from an intruding pathogen, or in the gastrointestinal tract, antigens that provide a “blueprint” of common bacterial structures, sampled through M cell-uptake from the luminal site for presentation to GC B cells within the Peyer’s patches. Consistent with this interpretation is the recent observation that GC-derived MBC do not usually return to the GC in a recall response ( 144 ). Thus, generation of MBC supports primarily a more effective and more diverse EF response upon re-challenge. Such repertoire reshaping overcomes the inherent limitations of a stochastic process of B cell development in many mammalian species, including humans and mice; a process by which most B cells are selected initially only for functionality of their BCR and against self-recognition, but not selected for “usefulness”. De novo GC responses establish in the follicles when B cells, activated in the T-B border return to these sites, where they are further stimulated by antigens trapped on FDC, as well as by CD4 T FH that are recruited by B cells to the GC, causing extensive further clonal expansion ( 88 ). Similar to the extrafollicular responses, GC responses result in the formation of antibody-secreting plasmablasts and plasma cells, as well as MBC. In contrast to the EF response, however, GC-derived plasmablasts migrate to the bone marrow, where they differentiate into LLPC and can remain for extended periods of time ( 145 , 146 ). The output of GC switches over time from an early wave of MBC to a wave of later-developing plasma cells ( 100 , 147 ). The fate-determinants of B cell development toward MBC or plasma cell fate have not been identified, but MBC fate may require lower BCR affinity compared to plasma cells, as they show lower rates of mutation, consistent with the earlier exit from the GC. Interestingly, compared to the LLPC, MBC may have an increased ability to respond to cross-reactive antigens, as demonstrated with virus escape mutants of influenza A and West Nile Virus ( 147 , 148 ). The impact of GC responses on host survival from an infection depends on the type of pathogen encountered. As discussed above, during acute infection with “hit-and-run” viruses like norovirus or influenza virus, where the virus is cleared within a few days of infection by the innate immune system as well as by the rapidly developing EF responses, GC responses develop too slowly to affect the course infection. However, GC responses become increasingly important when pathogen clearance cannot be achieved by these rapid immune responses. Already mentioned is the example of infections with P. yoelii in mice, where the abrogation of GC responses due to exuberant EF responses led to enhanced mortality ( 133 ). One of the most intriguing cases is chronic infection with HIV, where up to 50% of chronically-infected individuals will develop protective broadly neutralizing antibodies after many months to years after infection ( 149 , 150 ). What distinguishes these antibodies from those generated earlier in the same patients is their ability to bind to subdominant epitopes, thus epitopes that do not provide strong triggers of B cell activation, and/or epitopes that mimic self-antigens and against which B cell tolerance might have been induced ( 151 - 153 ). Only the continued presence of the antigen and ongoing GC responses would lead to sufficient BCR-diversification that eventually results in the emergence of such broadly protective B cell clones. During influenza infection broadly-neutralizing antibodies are also directed against subdominant epitopes, here the stalk region epitopes of the hemagglutinin molecules ( 154 ). Influenza infections might not be sufficiently long-lived to drive selection of such subdominant B cell clones. Instead, frequent repeat infections with the yearly emerging influenza variants would ensure maximal diversification of the response to influenza. Indeed, this interpretation is supported by recent studies demonstrating that individuals with high titers anti-influenza antibodies show the greatest degree of repertoire diversification to influenza ( 155 , 156 ). Given this repertoire broadening through repeat infections, however, the subdominant clones are unlikely strongly selected, as new dominant determinants are favored. Hence, strong broadly neutralizing antibody development to influenza and like viruses are unlikely to occur after natural infection. If it did, influenza infections would not likely be the continued threat to human health that they are today. Given the importance for GC responses in diversifying a suboptimal repertoire of specificities, it is not surprising to see that some pathogens that cause persistent infections suppress functional GC. One example is the rapid collapse of GC in mice infected with B. burgdorferi, and the resulting lack of affinity maturation and MBC and LLPC development, despite strong extrafollicular-derived antibody responses ( 157 , 158 ). Another is the suppression of GC responses following Salmonella infection ( 159 ). Taken together, GC responses are critical for shaping the B cell repertoire of a host, generating MBC that can induce rapid EF responses following recall infections, but also MBC that have a broadened repertoire and can engage in heterotypic immunity. The generation of LLPC provides strongly protective and neutralizing antibodies to prevent reinfections, as well as non-neutralizing antibodies that can rapidly complex antigens and thereby support de novo B cell responses as outlined above. Together extra and intra-follicular B cell immunity work together to provide both rapid responses to the infecting pathogen and better prepare the host for future pathogen encounter.

The first waves of B cells (B-1) emerge in early fetal development when there is limited exposure to microbial stimuli that could shape their repertoire. Instead B-1 cells, emerging from a distinct developmental path than that of B-2 cells ( 4 , 172 ), undergo positive selection for recognition of self-antigens ( 173 ). It is not surprising then that the majority of B-1 cells express BCR-signaling inhibitors, including CD5, Siglec G and CD148 strongly curtailing their ability to respond to these antigens ( 174 - 176 ), and consistent with findings that CD5+ B-1 cells do not respond to soluble anti-IgM with clonal expansion ( 174 , 177 ). The repertoire of B-1 cells dramatically changes over the course of the first 5 month of age, even in the absence of microbiota, leading to the emergence of a heavily skewed repertoire containing some large and highly dominant public B cell clones ( 178 , 179 ). The signals that drive these repertoire changes are unknown, but recent studies with TLR-deficient mice suggest that DAMPS may play a significant role ( 180 ). In the presence of microbiota, B-1 cells contribute strongly and continuously throughout life to the IgA-producing plasmablast pools in the intestinal tract ( 181 , 182 ). They also generate most of the circulating “natural” serum IgM that can bind numerous pathogens ( 183 , 184 ), critical for survival from infections ( 37 , 185 , 186 ). In addition to the direct anti-pathogen role that likely involves the activation of complement ( 187 ), the presence of IgM is required for maximal IgG responses to infections ( 129 , 185 , 188 , 189 ). The mechanisms of that are largely unexplored but seem to involve B cell-expressed FcμR ( 129 ). B-1 cells also respond to infections by migrating to secondary lymphoid tissues where they differentiate rapidly into IgM and cytokine-secreting cells providing a first wave of IgM ( 37 ), they may also secrete IL-10 ( 190 ) and their rapid induction of GM-CSF production has been associated with survival from experimentally-induced sepsis ( 36 ). Accumulation of body cavity B-1 cells in the lymph tissues depend on Type I IFN or MyD88-mediated integrin-activation ( 34 , 35 ). B-1 cells are exquisitely sensitive to TLR-mediated stimulation ( 191 ), which causes them to lose expression of CD5 and to proliferate to levels much stronger than seen by FOB cells ( 192 ). TLR-mediated stimulation of B-1, but not FOB, cells towards plasma cell differentiation was shown to depend on CMTM7 a surface-expressed protein that supports phosphorylation of p38 following TLR stimulation ( 119 , 193 ). In contrast to FOB cells, for which stimulation through TLR acts synergistically with BCR-mediated signals, BCR-engagement via anti-IgM before or immediately after TLR-stimulation inhibits TLR-induced B-1 cell proliferation ( 192 ). Thus, in vitro responses suggest that B-1 cells respond mainly through TLR not BCR stimuli, although TLR-induced loss of CD5 may alter their responsiveness to subsequent BCR stimulation ( 192 ). This raises questions about the specificity of the early B-1 cell response. In support of antigen-specific BCR-mediated B-1 cell activation, infections with Streptococcus pneumoniae ( 194 ), Borrelia hermsii ( 195 ), Salmonella typhimur ium ( 196 ) as well as Francisella tularensis ( 197 ) indicated B-1 cell clonal expansion, pathogen-binding, IgM secretion, and the formation of memory B-1 cells ( 37 ). In contrast, B-1 cell responses to influenza infection suggested a mainly innate-like response, as the number of influenza-binding IgM secreting B-1 cells was not higher after infection, than in non-infected mice ( 198 ). Furthermore, IgM responses were abrogated in mice in which B-1 cells lacked intrinsic TLR-signaling ( 180 , 192 , 199 ). Further work is required to determine the apparent differences between these infection models. Common to all B-1 cell responses to infections is the rapid induction of IgM-secreting B-1 plasmablasts, similar to those initiated by FOB during EF responses. Proliferating B-1 cells rapidly accumulated in the T-B border zone, where they differentiated into extrafollicular short-lived CD138+ plasmablasts ( 13 , 21 , 192 , 200 ) and MBC that accumulated in body cavities ( 195 , 201 ). Interestingly, work by Yang and colleagues suggested that B-1 plasma cell differentiation to recall responses of antigen-specific B-1-derived MBC required TLR4-mediated stimulation ( 202 ), further underscoring the importance of innate signals in quickly mobilizing this self-reactive broadly protective B cell population. Thus, fetal-derived B-1 cells differ significantly from FOB cells in their responsiveness to infections. Their unique distribution in the pleural and peritoneal cavities allow localized and rapid responses to infections that breach mucosal barriers. Their contribution to anti-pathogen immunity is two-fold: passive, through continued secretion of broadly-binding serum natural IgM and IgG and mucosal IgA, a repertoire that is shaped by TLR stimulation, and active, through rapid migration to secondary lymphoid tissues, where B-1 cells contribute early short-lived antibody responses and regulatory cytokines, such as GM-CSF and IL-10 ( 203 - 205 ). As discussed, splenic MZ B cells are critical for humoral immune response development, as they shuffle antigens from the marginal zone into the follicle for antigen-presentation by FDC in the spleen, while lymph nodes lack this B cell subset ( 51 , 52 ). Although MZ B cell precursors can develop during fetal life, the marginal zone organization seem to require some time to develop ( 12 ). MZ B cells share many traits with B-1 cells, including enhanced responsiveness to TLR-signaling compared to FOB cells ( 206 ). They also express high levels of CD1, consistent with a repertoire responsive to glycans and lipids. These traits together with their location in the splenic marginal zone enables their rapid responses to blood-borne pathogens or TLR-agonists such as LPS ( 13 , 206 ). It also explains their critical role for antibody production to glycans, including the polysaccharide capsule antigens of gram-positive pathogens, such as S. pneumoniae , explaining the sensitivity of newborns and splenectomized individuals to infections with gram-positive bacteria. Similar to B-1 cells, MZ B cell responses to pathogen-encounter is the rapid differentiation into EF plasmablasts that can be induced in a T-dependent or T-independent manner ( 13 , 206 ). Early studies already established a predilection of MZ B cells for generating non-switched IgM memory responses to model antigens, although participation in late GC responses was noted in one earlier study ( 207 ). Thus, all peripheral mature B cell populations, FOB, B-1 and MZ B cells are poised to respond rapidly to infections with formation of EF responses. In addition, B-1 cells are providing immune surveillance in the peritoneal and pleural body cavities and MZ B cells are strategically positioned to survey the circulation. B-1 and MZ B cell’s sensitivity to innate-like signals and their exclusion from GC responses, and their self-renewal capacity suggest that their repertoires fulfill critical specificity niche gaps that exist within the FOB cell pool, that must be maintained throughout life. Their rapid antibody responses after infection reduce freely available pathogens and PAMPS, thereby reducing the potential for overshooting cellular activation and cytokine responses at a critical window during early infection.