Antibody modulation of B cell responses – incorporating positive and negative feedback

Summary

Antibodies are powerful modulators of ongoing and future B cell responses. While the concept of antibody feedback has been appreciated for over a century, the topic has seen a surge in interest due to the evidence that the broadening of antibody responses to SARS-CoV-2 after a third mRNA vaccination is a consequence of antibody feedback. Moreover, the discovery that slow antigen delivery can lead to more robust humoral immunity has put a spotlight on the capacity for early antibodies to augment B cell responses. Here we review the mechanisms whereby antibody feedback shapes B cell responses, integrating findings in humans and in mouse models. We consider the major influence of epitope masking and the diverse actions of complement and Fc receptors and provide a framework for conceptualizing the ways antigen-specific antibody may influence B cell responses to any form of antigen, in conditions as diverse as infectious disease, autoimmunity and cancer.

eTOC Blurb

Antibodies modulate ongoing and future B cell responses. Cyster and Wilson review the various mechanisms whereby antibody feedback shapes B cell responses and present a framework for conceptualizing the ways antigen-specific antibody may influence immunity in conditions as diverse as infectious disease, autoimmunity and cancer.

Introduction

The effectiveness of an antibody response in protecting from a pathogen is influenced by its magnitude, isotype, affinity, breadth and duration. Many processes impact on these parameters and important among them is antibody feedback. This is the process where antigen-specific antibody produced during a B cell response has actions that influence, positively or negatively, ongoing or subsequent (secondary or tertiary, etc) B cell responses to the same or closely related antigens. By promoting antigen opsonization and improved capture and display, antibodies can positively feedback on B cell and T cell antigen encounter in the early phases of responses. By promoting antigen stability, they can extend the period over which intact antigen can be encountered by B cells. Antibody enhancement of antibody responses can be very strong (1000-fold) ¹. Negative feedback effects of humoral responses on the ability to mount a secondary response can also be strong and were identified in the early 1900’s in the context of anti-toxin responses ². By clearing antigen, blocking epitopes and engaging inhibitory Fc receptors (FcRs), antibodies can have a negative feedback effect that may limit the magnitude of the response. A challenge in understanding and predicting effects of antibody feedback is the multitude of ways that antibodies can act and how these can differ based on the properties of the antigen, the properties of the antibody (isotype, affinity, glycosylation status) and the inflammatory milieu of the response. Moreover, whether a feedback effect is “positive” or “negative” can depend, for example, on whether the epitope-specific or antigen-specific antibody response is being measured, or whether a high affinity epitope-focused or low affinity multi-epitope response is desired.

We have structured the review to first describe core features of a B cell response to provide context for considering the stages where feedback can act. We then discuss the important concept of epitope masking and how it can inhibit antibody responses against the specific epitope while augmenting responses to other (often subdominant) epitopes and thereby help diversify the response. We next discuss the influences of complement (C’) coating of antigens, a process often promoted by antibody, on the B cell response via engagement of receptors on B cells or on follicular dendritic cells (FDCs). We consider how IgG can have feedback actions via FcγRs on myeloid cells, B cells and FDCs. The less understood contribution of IgM- and IgA-binding FcRs to antibody feedback is briefly considered. We then discuss how feedback may differentially affect Bmem versus naive B cell responses. The feedback actions of passively transferred antibody in newborn and mother are noted. Finally, we consider the implications of antibody feedback on human vaccination responses.

B cell response basics

Most B cell responses are initiated in lymphoid tissues – lymph nodes, spleen, tonsils, Peyer’s patches – where B and T cells colocalize and antigens are filtered and trapped ³. Naïve B cells spend the majority of their one-day transit time through a lymphoid tissue inside lymphoid follicles surveying for antigens. There are several modes of B cell encounter with antigen ⁴: for small, soluble antigens, encounter may occur in fluid phase; particulate antigens entering lymph nodes can be transiently displayed on the surface of subcapsular sinus macrophages; for antigens that have been coated with active fragments of complement (C3b and C4b or their breakdown products), or IgG, they can be captured and displayed by specialized stromal cells termed FDCs (Figure 1). A large body of work has established that FDCs can retain and display intact (opsonized) antigen for weeks or months ^5–7. As well as depending on the unique cell biological properties of FDCs ⁵, this antigen stability was suggested to reflect the B cell follicle being an antigen sanctuary with low protease and phagocyte activity ⁴. Recent work has provided strong evidence that lymphoid follicles have unusually low extracellular protease activity ⁸.

Figure 1. Types of antibody feedback on the B cell response.

The different mechanisms of feedback have been classified as “Positive” and “Negative” for simplicity. However, whether an effect is positive or negative can depend on whether the epitope-specific or antigen-specific response is being considered, or whether antibody response affinity versus breadth is being considered. For example, epitope masking has a negative effect on B cells recognizing the same epitope as the secreted antibody but may have a positive effect on the response of B cells recognizing other epitopes in the antigen. FDC capture of opsonized antigen can increase the ability of B cells to participate in the response (“positive” feedback) but by increasing antigen availability and avidity it may, at least transiently, support responses that are of lower affinity. The antibody isotype mediating a given type of feedback is shown above each arrow. Effector molecules (C’, receptors) involved in each type of feedback are shown below each arrow. The antibody diagram is shown as IgG for simplicity. Receptors present on FDCs (CR1 and lower amounts of CR2 and FcγRIIb) are not shown for clarity. The B cell is only shown in the cases where an additional receptor (CR2 or FcγRIIb) is engaged in cis with the BCR. In all cases the feedback impacts the extent of BCR engagement occurring in responding B cells, except antigen uptake by DCs that impacts on TCR engagement in helper T cells. Although not shown, in some cases FcγR⁺ myeloid cells may also make opsonized antigen available to B cells. Several additional less studied feedback mechanisms that are discussed in the text are not depicted.

After encountering a multivalent antigen, especially if it contains ligands for innate receptors, B cells undergo T cell-independent proliferation outside of follicles and differentiate into plasma cells (PCs) and B memory (Bmem) cells ^3,9. For antigens containing protein, antigen-presenting B cells can engage with cognate CD4 T cells at the follicle-T zone border and receive helper signals. Some B cells receive signals to return to the follicle and form a germinal center (GC) that is situated over the FDC network. Here GC B cells can again encounter antigen and present MHCII-peptide complexes to co-localizing T follicular helper (Tfh) cells, thereby receiving CD40L and other cytokine signals that support high activation-induced deaminase (AID) expression and induction of immunoglobulin (Ig) gene somatic hypermutation (SHM). Receipt of B cell receptor (BCR) and Tfh cell-derived signals is needed for positive selection in GCs, with the collective strength of these signals determining GC B cell fate ¹⁰. As well as supporting iterative rounds of B cell SHM and selection over a period of weeks to months, GCs give rise to long-lived PCs and Bmem cells. GC B cell positive selection depends on re-encounter with antigen, and this is mostly thought to involve capturing opsonized antigen displayed on FDCs in the GC light zone (LZ). GC B cells move between receiving positive selection signals in the LZ and undergoing proliferation and SHM in the dark zone (DZ), with substantial exchange between zones occurring every 12 hours ¹⁰. When access to antigen is blocked, FDCs are ablated, or the ability to present antigen to T cells is inhibited, GCs are rapidly shutdown ^11–14.

Upon secondary exposure to an antigen Bmem cells robustly differentiate into PCs. Secondary (and tertiary, etc.) responses to protein-containing antigens are also associated with induction of GC responses, though estimates of the extent of Bmem cell participation in GCs varies across studies and Bmem subsets ^15–18. An important determinant of B cell participation in recall GCs appears to be antibody feedback, as we will discuss in later sections of this review. The types of antibody feedback on the B cell response are summarized in Figure 1; we will consider each mechanism in turn.

Antibody-mediated feedback by epitope masking – diversifying the response

A feature of antibody feedback that has been a center of attention for many years is the ability of antibodies to mask epitopes and thereby inhibit B cell responses to those epitopes (Figure 1). If all (or select) epitopes in a soluble or particulate antigen have already been bound and occupied by high affinity antibodies, there is no further opportunity for cognate B cells to bind and be recruited into the response (Figure 2). Such masking effects have been observed in response to model protein antigens, to foreign RBCs, to viruses, to malaria sporozoites, to vaccine antigens, and to allergens ^1,19–22.

Figure 2. Epitope masking drives diversity of the B cell response.

Accumulation of serum antibody in the current or from past immune responses can sterically block B cells from binding to epitopes. This blocking effect can inhibit B cell activation to shared epitopes. This blocking will in turn prioritize targeting of additional, less accessible, or subdominant epitopes as responses progress over time, or will block memory B cell responses upon re-exposure to the same or similar pathogens. In recent years epitope masking has been shown to be particularly important for immunity to evolving viruses such as influenza, HIV, and SARS-CoV-2.

When considering how masking will influence the outcome of a B cell response it is important to recognize that most antigens have dominant epitopes – ones that are readily accessible to B cells and/or for which there is a high precursor frequency – and subdominant epitopes. The latter tend to be less accessible on the antigen, and the frequency of B cell precursors that can recognize these regions may be lower ²³. It is also necessary to consider that the amount of B cell response that occurs is proportional to the amount of T cell help available. B cells that present more MHCII-peptide complexes tend to receive more T cell help and thus contribute more dominantly to the response ^24–26. When dominant epitopes are masked by antibody, B cells recognizing subdominant epitopes can likely present the highest amounts of MHCII-peptide, allowing them to win the competition for T cell help and contribute to the response. Thus, this type of antibody feedback can lead to a broadening of the antibody response. It should be noted that the classification of masking in Figure 1 as a “negative” feedback mechanism is only accurate at the level of B cells recognizing the same epitope as the masking antibody. For B cells recognizing non-overlapping epitopes, the feedback is more likely to be “positive”. In the next paragraphs of this section, we review recent studies that have provided important insights regarding the effects of epitope masking on B cell responses. Subsequent sections will further explain how masking can feedback positively on responses against non-masked epitopes.

Influenza vaccines contain immunogens from the most recent circulating variants of H1N1, H3N2 and B type influenza strains, any of which may be updated to reflect viral evolution in the previous year, but rarely with all three strains updated. It has been appreciated for some time that people vaccinated in two consecutive years in which only one of the three included strains was updated (i.e., H3N2) because of viral evolution will preferentially activate B cells to the newly introduced variant strain; in this example to the H3N2 variant with a low or absent response to the H1 and B strains that were identical in both influenza seasons ²⁷. Pre-existing serum antibody titers inversely correlate with the influenza subtype targeted after vaccination, suggesting a role for antibody-feedback ^27,28. A recent study showed that pre-existing serum reactivity to an epitope involving even a single dominant amino acid (K163) on influenza hemagglutinin (HA) predicts suppression of future responses ²⁹. Correlations with serum titers in these studies was suggested to drive preferential targeting of newly introduced epitopes by masking previously encountered epitopes.

Studies on patients receiving an attenuated malaria sporozoite vaccine showed that a third vaccine dose did not increase the magnitude of the immunodominant anti-circumsporozoite (CSP) response ¹⁹. Experiments in mice led to similar findings and antibody transfer and antigen-specific B cell tracking experiments provided evidence that the antibody plateau occurred due to blocking effects mediated by antibody induced by the first two immunizations. Moreover, the amount of antibody sufficient to block the anti-CSP B cell response was lower than required for prevention of infection ¹⁹. This suggests that while a high level of epitope occupancy was needed, full masking of the epitope was not necessary for blocking the epitope-specific B cell response. Importantly, this work showed that the third immunization led to an increase in antibodies against subdominant epitopes, thus a broadening of the response ¹⁹. Conversely, the work indicates that attempts to boost responses to conserved but subdominant epitopes in complex antigens may be inhibited by pre-existing antibody titers that do not fully occupy the epitope.

People who received two doses of the mRNA vaccine encoding the original Wuhan strain spike protein of SARS-CoV-2 showed limited amounts of antibody that could bind and neutralize the Omicron variant. However, after receiving a third dose of the same vaccine, most people and non-human primates (NHPs) showed a marked (20-30x) increase in neutralizing antibody titer against Omicron ^30–33. Interestingly, on boost with omicron variants after two Wuhan exposures, all neutralizing serum antibody activity could be absorbed with the Wuhan strain, showing that only cross-reacting antibodies were being boosted to protective levels ³⁴. Thus it appears that serum antibody after several boosts with the Wuhan strain masked dominant epitopes similarly to boosting with a variant strain mutated at these epitopes, causing retargeting to sub-dominant cross-reactive epitopes by memory B cells. Recent work found that the frequency of Omicron binding Bmem cells increased several months after the second dose of the Wuhan mRNA vaccine ²¹. Using mouse models, serum transfer from Wuhan receptor binding domain (RBD)-immune mice to mice newly immunized with Wuhan RBD led to an increase in Omicron RBD-binding GC B cells and Bmem cells ²¹. These data suggested that antibody feedback, likely by masking, could broaden the RBD-specific B cell response. Supporting this interpretation, ablation of PCs for two weeks following Wuhan RBD immunization to limit antibody feedback led to a reduction in Omicron reactive GC B cells and PCs 2-4 weeks later ²¹.

Another recent study used several Ig transgenic mouse lines specific for HIV-1 envelope (Env) or SARS-CoV-2 RBD to show that pre-existing antibody from endogenous priming could inhibit naive B cell participation in GCs, and this inhibition was largely epitope specific and more efficacious with higher affinity antibody ²⁰. However, the effect on the endogenous response varied between the HIV-1 (epitope focused) and SARS-CoV-2 immunogens such that strong inhibition of GC participation of Ig transgenic cells was only seen with the former. This was suggested to reflect the lower extent of competition for epitopes in the case of the SARS-CoV-2 immunization. When passive antibody transfer was performed, the inhibition was antibody dose-dependent and could be overcome by providing repeated doses of antigen, most likely because this caused sufficient antibody depletion to ensure not all the immunogen was masked ²⁰.

A new approach to study antibody feedback involves protein fate-mapping. Mice were generated with an epitope tagged Ig light chain (Lc) that could be switched to a different epitope tag using Cre ³⁵. This work showed that naive B cells had a very limited ability to contribute to the secondary antibody response to an immunogen. However, when the secondary exposure was with a variant form of the primary immunogen (whether it was influenza HA or SARS-CoV-2 spike), the extent of contribution to the boosted response by newly recruited B cells increased in proportion to the antigenic distance between the boosting and primary immunogens. Importantly, the epitopes recognized by antibody from the newly recruited B cells were distinct from those recognized by the boosted Bmem cells. Antibody deriving from the newly recruited B cells had greater neutralization potency against the boosting (variant) virus than the primary or Bmem-derived antibody ³⁵. Considering negative antibody feedback as a possible contributing mechanism, the results of these fate-mapping experiments suggest that the relative inability of naive cells to contribute to plasma recall responses could be explained through epitope-masking by pre-existing antibodies. For reasons that we will speculate on in later sections, this negative feedback seems to affect naive cells more than Bmem cells.

Passive immunization with antigen-specific antibodies has been thought to alter the evolution of the endogenous antibody response (³⁶ and references therein) though direct evidence for such an effect has been limited until recently. Treatment of HIV-1 infected people with a broadly neutralizing antibody (bNAb) to HIV-1 led, over a several month period, to an endogenous response with greater breadth and potency ³⁷. Whether this reflected the effects of masking or other actions of the bNAb was unclear. In a study in humans who had received two mAbs against SAR-CoV2 and were later mRNA vaccinated, the induced antibody responses were only slightly reduced but the Bmem compartment was altered to being primarily low affinity IgM and having low numbers of SHMs. Moreover, the Bmem cells showed a shift in recognition to epitopes non-overlapping with the mAbs ³⁸. Similar findings were made in a single RBD mAb treatment study, in this case where infected individuals were treated with mAb and later vaccinated ³⁹. The authors propose that the pre-existing antibodies act both to lower the affinity threshold for B cells to enter the response, perhaps by increasing antigen retention and display to follicular B cells as discussed further below, and through direct masking of epitopes.

As well as masking the identical epitope recognized by B cells with antibodies of the same specificity as the secreted antibody, instances exist where antibody binding can block B cells of other specificities by steric interference. This has been demonstrated in haptenated SRBC immunization studies where it was found that with highly but not lowly haptenated SRBC, IgG to the hapten could suppress the IgM antibody response against SRBC antigens ^40,41. This does not seem to be a prevalent mechanism, however, as an increasing number of studies, such as those discussed above, indicate that responses to non-masked epitopes tend to be augmented by antibodies against other regions of the antigen.

Another form of masking that may be influential but is less studied involves antibodies shaping antigen processing and T cell epitope generation after IC uptake by antigen-presenting cells ^42,43. Antibody binding to a protein surface that overlaps with a T cell epitope can reduce the efficiency of proteolysis and MHCII-peptide generation. The extent that this type of masking can by inhibitory of antigen presentation and thus T cell help for a B cell response depends on antigen size and the number of T cell epitopes it contains ^42,44. It might be anticipated that the more completely coated an antigen surface is with antibody, the stronger this feedback effect will be.

Efforts have been made to generate mathematical models of the B cell response that incorporate the effects of antibody feedback. Some such models can predict that high antigen availability and epitope masking at the time of the second GC response favors induction of Bmem to subdominant sites ^45,46. However, these models are not confirmed to predict other aspects of the response accurately, such as the affinity threshold for selection of GC B cells. Thus, although modeling can predict a range of possibilities for the GC response and possible influences of antibody feedback, our mechanistic understanding of the process is still insufficient for accurate modelling.

In summary, epitope masking is increasingly being recognized to have a major influence on the evolution of the B cell response and this can involve promoting shifts in the response away from dominant and towards subdominant epitopes. However, while some central findings regarding antibody feedback seem to be sufficiently explained by an affinity-based physical epitope masking effect, most studies indicate that antibodies also influence the B cell response through changing antigen availability. Many factors can impact on how antibodies affect antigen availability as well as it’s B cell stimulatory properties, as we discuss in the following sections.

Feedback through complement-mediated antigen display on FDCs

Antibodies have an important feed-forward effect on antigen deposition on FDCs. This occurs most prominently through their ability to promote activation of the C’ cascade, causing C’ fragment deposition on the antigens and thereby facilitating antigen binding to complement receptors (CR) on FDCs. While some complex antigens may engage the C’ pathway from early in a primary response due to low affinity binding by natural (often polyreactive) IgM, by display of sugars for mannose binding lectin (MBL) attachment, or due to an inability to prevent coating by the alternative pathway of C’ ⁴⁷, some antigens may escape these pathways, for example as was seen for HIV gp120 ⁴⁸. As soon as antigen-specific antibody begins being generated, it can augment opsonization of the antigen and thus the efficiency of FDC deposition (Figure 3). While IgM is especially potent at activating C’, binding of multiple IgG molecules to an antigen can also strongly activate C’ ¹. Of the IgG subclasses, IgG2a, IgG2b and IgG3 in mouse and IgG1 and IgG3 in human are most effective in activating the classical C’ pathway ^49,50. The action of newly generated antibody in augmenting antigen deposition on FDCs is inferred to be an important feature of how slow (or serial) antigen delivery leads to more robust GC and antibody responses ^51–54. That is, by becoming complexed with antigen-specific IgM and IgG generated several days after initial vaccination, the persisting antigen can have increased delivery to, and deposition on, FDCs. Striking imaging studies in NHPs (Rhesus Macaques) have shown that this can involve hundreds of follicles in a single draining LN ^48,52,55.

Figure 3. Immune complex and C’ fixation affect B cell responses.

(A) Antibody both fixes C’ and cross-links antigen to form immune complexes (ICs) that are strewn with C3b, C4b, and C3d complexes. (B) Through CR1 and CR2 interactions, IC is transported by B cells and deposited on FDC networks in GCs. Interaction with CR1/2 also increases antigen persistence and half-life on FDCs to allow a persistent source of whole antigen for GC B cell uptake and presentation to TFH cells, driving affinity maturation. (C) Activation of B cells specific to the antigen in IC is amplified by increased BCR and CR1/2 cross-linkage. (D) C’ mediated direct lysis of enveloped virus (virolysis), bacteria, or infected cells through membrane attack complex (MAC) formation can decrease antigen available.

FDCs highly express complement receptor-1 (CR1 or CD35) and CR2 (CD21), which are alternatively spliced products of the same gene in mice and encoded by separate genes in humans ⁴⁷. CR1/2 deficient mice show strong defects in the response to unadjuvanted phage, SRBCs, and viral like particles ^56–60. Bone marrow chimera approaches provided evidence supporting the role of CR1/2 on FDCs for mounting antibody responses ^61–64. The center of the FDC network expresses higher amounts of CR1/2 than the periphery, and this helps concentrate opsonized antigens to the FDC network center ⁶⁵. This focusing may help achieve more complete protection of antigen from extrafollicular proteases that are low within lymphoid follicles compared to surrounding regions of the lymphoid tissue ⁸. Thus, the high IC binding capacity of FDCs and their non-degradative properites, and the low protease activity of follicles likely cooperate to ensure the sustained presentation of intact antigens within this microenvironment.

As well as augmenting binding to FDCs, C’ opsonization of antigens facilitates their delivery to these specialized antigen-display cells and this is thus another site for “positive” antibody feedback. Naïve B cells express CR1/2 and have been visualized capturing immune complexes (ICs) from subcapsular sinus (SCS) macrophages in lymph nodes in a CR1/2 dependent manner during their surveillance of follicles and then delivering them to FDCs ⁶⁶. The hand-off of ICs from non-cognate B cells to FDCs likely occurs due to the much higher CR1/2 expression on FDCs enabling stronger binding. FDCs express CR1 more abundantly than CR2 whereas for B cells, the reverse is the case ⁶⁵. Although both CR1 and CR2 can bind C3d/g, the predominant forms of C’ that coat opsonized antigens ⁴⁷, whether CR1 has properties that facilitate IC capture from B cells (such as its greater length than CR2) needs exploration. Interestingly, human FDCs express a distinct CR1 splice variant compared to B cells ⁶⁷ but whether it has a specialized role is not known. In the spleen, marginal zone (MZ) B cells express higher amounts of CR1/2 than follicular B cells and they undergo a constitutive shuttling behavior between the blood-exposed MZ and the FDC-containing follicle, enabling rapid delivery of systemic ICs to splenic follicles ^68,69.

Beyond increasing the magnitude and durability of B cell responses, antibody-mediated augmentation of antigen deposition on FDCs may have complex effects on GC B cell selection. For example, an antibody-induced surge in antigen deposition on FDCs might reduce selection stringency in an ongoing GC response, possibly allowing lower affinity B cells to be selected. Due to their multivalent nature, antibodies will often mulitmierize antigens and deposition on the FDC surface may further increase antigen valency. Antigen valency has long been recognized to strongly influence the strength of B cell triggering by antigen (⁷⁰ and referenes therein). A recent study that compared HIV immunogens with a range of valencies showed that higher valency antigens were able to recruit lower affinity B cells into the response ⁷¹. By increasing the valency of the antigen, antibody and C’ mediated opsonization may allow selection of lower affinity B cells and thus help diversify the antibody response ³⁸.

Feedback through complement-mediated costimulation of B cell activation

As well as improving deposition on FDCs, C3d/g coating is expected to increase the engagement of the CR2 signaling complex on cognate B cells (Figure 3). Co-crosslinking of the BCR with the CR2/CD19/CD81 complex can increase the sensitivity of the B cell to antigen by over 1000-fold ^47,72. B cell intrinsic expression of CR2 contributes significantly to the magnitude of the B cell response to several types of antigen ^56,73. However, whether this contribution becomes more marked during antibody feedback (where C3d/g coating of the immunogen has likely increased) has not been directly tested. As noted in the preceding section, FDCs are thought to extract ICs from the surface of non-cognate B cells due to higher CR1/2 expression. CR1 and CR2 are long, flexible molecules ⁷⁴, though the extent that accessible C3 (and C4) fragments within ICs become fully occupied by binding to FDC CR1/2 is not known. Thus, once an IC is bound to an FDC it is unclear how accessible the C’-fragment coating will be to the CR1/2 on cognate B cells. It might be speculated that due to the concentration effect of binding the antigen via two receptor classes (BCRs and CRs) in the same membrane, CR1/2 on naive cognate B cells may compete successfully with FDC CR1/2 for binding C’ fragments and delivering a costimulatory signal to the B cell. While CR1/2-mediated costimulation can strongly augment activation of cognate follicular B cells, it is unclear whether this is the case for GC B cells. In GC B cells, Cr2 transcript expression is downregulated (Immgen.org). CR1/2 downregulation may be important for ensuring the amount of BCR signaling and antigen capture in GC B cells is determined by BCR affinity rather then the extent of antigen opsonization by C’.

The roles of CR3 (CD11b/CD18 or Mac1) and CR4 (CD11c/CD18) in antibody feedback on B cell responses are not well understood. It is notable that subsets of memory B cells express these C’ binding integrins ^75,76 and it seems possible that they contribute positive signals to cognate CD11b+ and CD11c+ B cells during encounters with opsonized antigens.

Feedback through FcγRs on myeloid cells

Another layer of IgG mediated feedback is through engagement of FcγRs. This can have both positive and negative effects on the B cell response (Figure 4). The activating FcγRs (Fcγr1, Fcγr3, Fcγr4 in mouse, all of which associate with the ITAM-containing Fcrγ and depend on this subunit for signaling) are well expressed on macrophages and granulocytes where they have an important role in promoting phagocytosis and clearance of antibody coated particles (ADCP) ^77–79. IgG binding antigens expressed on cell surfaces such as budding viral particles can activate antibody-dependent cellular cytotoxicity (ADCC) through Fc-FcγR interactions, classically by natural killer (NK) cells, to remove the infected cells ⁸⁰. ADCC can also be mediated by macrophages, neutrophils, and eosinophils ⁸¹. By depleting antigen or reducing its production, activating FcγR activity may often have a negative impact on the antibody response. In contrast to epitope masking that can diminish the response to the bound epitope while augmenting the response to non-masked epitopes, FcγR-mediated antigen clearance diminishes availability of the entire antigen and thus is expected to diminish the B cell response to all epitopes on the antigen.

Figure 4. FcγR and antibody isotype diversity orchestrates immune responses.

The balance of FcγR types on various leukocytes, the accumulation of IgG subclasses in serum and IgG Fc glycosylation status can dictate which and how strongly cells are activated (i.e., human FcγRIIIa binds IgG3>IgG1 but not well to IgG2, and IgG3 binding is dramatically improved by Fcγ fucosylation). (A) This interplay can affect APC activation status which in turn drives CD4 T cell responses and B cell activation. (B) FcγR can also mediate clearance of antigen or of infected cells by antibody dependent cellular phagocytosis (ADCP) or antibody-dependent cellular cytotoxicity (ADCC). (C) Antigen half-life is believed to be increased through Fcγ interaction with FcRN that increases in the acidic environment of endosomes, allowing IC recycling to the cell surface. (D) FcγRIIb interactions with BCR cross-linkage directly inhibit B cell signaling capacity and can be differentially inhibiting with FcγRIIb variants.

Acting in opposition to antigen clearance, increased antigen uptake by myeloid cells due to FcγR expression may strongly augment antigen presentation and helper T cell activation under some conditions ^42,77. For example, IgG-mediated enhancement of the antibody response to soluble multivalent antigen was abrogated in mice lacking activating FcγR function ^82,83. However, the in vivo evidence that such antibody response augmentation occurs by increased MHCII-peptide presentation is limited (especially for Tfh cell induction) and often hard to deconvolute from other ways that activating FcγR signaling could increase B cell responses, such as by promoting upregulation of cytokine (e.g., IL10, IL12) expression by myeloid cells ⁸⁴. Moreover, in some cases binding of IgG increases soluble antigen uptake by myeloid cells in a C’-dependent manner ⁸⁵. Thus, antibody feedback to increased antigen uptake and MHCII-peptide presentation can occur secondarily to activation of the C’ cascade. Another complexity in interpreting FcγR-deficient mouse studies is that as well as losing the signaling and internalization properties, the cells lose the ability to bind IgG. Thus, in some cases, the FcγR-deficiency studies may be best explained by a loss of receptors that serve to transiently retain and display ICs on cell surfaces.

The relative influence of FcγR-dependent pathways on B cell responses may depend on features of the antigen and the inflammatory nature of the response (with inflammatory cytokines often upregulating FcγRs), requiring careful consideration for each antigen and context. Inflammatory conditions may also influence the glycosyltransferases expressed by PCs and Bmem and thus the extent of fucosylation and sialylation of the secreted antibodies ^86–88. These modifications can influence the strength of IgG binding to FcγRs and lectins, and thus impact on the feedback process ^88–90. Adding to the complexity, several Fc receptor-like (FCRL) molecules have been suggested to bind ICs but delineating their roles during the immune response has been challenging due to divergence in this family of receptors between rodents and humans ⁹¹.

IgG feedback has the potential to increase antigen half-life, at least in the case of small soluble antigens, by enabling FcRN (neonatal IgG Fc receptor) mediated recycling by endothelial and epithelial cells ^78,92. Such an enhancement in antigen durability has parallels to the function of lymphoid follicles and FDCs in augmenting the chances of B cell encounter with intact antigen. FcRN is also expressed by some DCs, and a role has been proposed for the receptor in antigen cross-presentation ⁸⁴. The influence of FcRN on IgG feedback effects on B cell responses needs more exploration.

Feedback through FcγR-mediated inhibition of B cell activation

While activating FcγRs tend to dominate the response of macrophages, neutrophils and NK cells to IgG, the inhibitory FcγRIIb is the dominant FcγR on B cells and is also important on DCs. Co-engagement of the BCR and FcγRIIb leads to tyrosine phosphorylation of the FcγRIIb immunoreceptor tyrosine-based inhibitory motif (ITIM) followed by recruitment of SH2 domain-containing inositol phosphatase-1 (SHIP1) and dampening of BCR signaling ⁹³. This negative feedback signal has a strong influence on the magnitude of the antibody response to multivalent soluble antigen ^82,83,94. Using a model wherein FcγRIIb was deleted in B cells, the primary and secondary IgG response to a multivalent soluble antigen was augmented ⁹⁵. The feedback actions via FcγRIIb are likely most influential for the B cell subsets expressing the highest amounts of this receptor, including activated, MZ, and GC B cells ⁹³. Indeed, elevated spontaneous IgG3 production in B cell-specific FcγRIIb-deficient mice reflected loss of feedback inhibition of BCR signaling in MZ B cells ⁹⁶. IgG mediated feedback of the response to foreign RBCs was variable across studies, with one showing an elevated response in Fcgr2b^−/− mice ⁹⁴ and others not ⁹⁷. FcγRIIb is not thought to be sufficient to block a B cell response but rather to exert a ceiling on its magnitude ¹.

As well as restraining the BCR, FcγRIIb engagement in the absence of BCR co-crosslinking and without recruitment of SHIP1 can lead, possibly via c-Abl dependent signaling, to the induction of B cell or PC apoptosis ^98,99. While this action of FcγRIIb has been well studied in vitro, the extent that this occurs in vivo, where there are many trophic factors that support B cell and PC survival, is less well established. That said, there is substantial evidence that defects in FcγRIIb expression or function due to coding or promoter polymorphisms are associated with autoantibody production, including studies showing a requirement for the receptor in B cells ⁹³. Some work suggests that the prevalence of disease associated low function FcγRIIb variants in the human population may be a consequence of selective pressure for stronger antibody response against pathogens such as malaria ^100,101. it remains unclear how FcγRIIb restrains autoreactive B cells, but it may help ensure that when autoreactive IgG is made, its abundance is kept in check, thereby limiting the pathological impact of the antibody.

Feedback through FcγR on FDCs

Like CR1/2, GC-associated FDCs express FcγRIIb ^65,102,103. In these specialized stromal cells, a negative signaling role for the receptor has not been demonstrated. Instead, the receptor is presumed to help present opsonized antigens. Dependence on this receptor might be most prominent when C’ deposition on the antigen is low. Based on a partial hematopoietic chimerism approach, FDC FcγRIIb was suggested to be required for maximal splenic IgG responses to ICs of soluble multivalent antigens ^102,104. However, these studies did not exclude a contribution of FcγRIIb in radioresistant myeloid cells. Moreover, other work has not found a notable deficiency in IC trapping when FDCs lack FcγRIIb ^65,105, consistent with the dominant activity of CR1/2 in FDC IC trapping. One study suggested reduced IC capture by primary follicle FDCs in FcγRIIb-deficient mice ¹⁰⁶ which is surprising because FcγRIIb is minimally expressed by naive FDCs, being strongly induced in FDCs during GC responses ^6,102. The selective upregulation of FcγRIIb on GC FDCs could be to increase the capacity of the GC FDC network to present ICs. It may also be the case that FDC FcγRIIb competes with FcγRIIb on the B cells for binding IgG Fc regions. That is, in cases where FDC FcγRIIb expression is high, the extent of negative feedback signaling in the GC B cell may be reduced. This concept could be explored by testing whether the influence of FDC FcγRIIb on the GC response to IgG-ICs is lost in mice that lack FcγRIIb on B cells. A role for FcγRIIb in positive signaling in FDCs to upregulate molecules such as VCAM1 has also been reported ¹⁰⁵ but a definitive demonstration of this activity awaits in vivo FcγRIIb mutagenesis studies where this undefined signaling activity is selectively disrupted.

The outcome of IgG binding on antigen deposition on FDCs is likely to vary with the properties of the antigen, the IgG subclass, and the location. For example, while IgM was very effective at promoting C’ dependent viral like particle deposition on FDCs, IgG diminished deposition and this appeared to be due to promoting phagocytic uptake ⁶². Moreover, further studies are needed to delineate the influence of FcγRIIb variants on receptor function in FDCs versus B cells and myeloid cells.

Mature B cells express CD23 (FcεRII) and GC FDCs in some lymphoid tissues (including lymph nodes in mouse and tonsils in humans) upregulate this low affinity IgE receptor. IgE is typically present in very low abundance in circulation and while it has the capacity to have feedback effects on B cell responses ¹, the extent this occurs physiologically needs more study. CD23 was suggested to also bind sialylated forms of IgG but this activity has not been confirmed in recent biochemical studies ⁹⁰.

Potential contributions of IgM and IgA FcRs

There are two IgM binding Fc receptors in addition to the epithelial polymeric Ig receptor, FcμR and FcαμR. FcμR (TOSO) is most strongly expressed by B cells in mouse (and additionally by T cells and NK cells in humans) ¹⁰⁷. FcμR binds the IgM pentamer with high affinity. Studies in several gene deficient mouse lines yielded discrepant findings, but a common observation is elevated pre-immune IgM and autoantibody (IgM and IgG) levels ¹⁰⁷. FcμR can mediate internalization of IgM and it has been suggested from studies on the influenza response in mice lacking FcμR in B cells that the receptor may support uptake of IgM-virus ICs by B cells leading to augmented helper T cell induction and improved virus-specific responses ¹⁰⁸. Another study found that FcμR was preferentially upregulated on bystander (non-antigen specific) Bmem cells in the lung following influenza infection ¹⁰⁹. Chimeras lacking FcμR in the B cell compartment showed a reduced influenza-specific IgA response and it was suggested that noncognate FcμR⁺ Bmem cells display ICs to colocalizing antigen-specific Bmem cells to enhance the humoral memory response ¹⁰⁹.

FcαμR is expressed by B cells, being highest on MZ B cells, and by FDCs in mouse and human ¹¹⁰. An unexpected finding in FcαμR-deficient mice was a prolonged retention of IgM-ICs on FDCs in the spleen ^110,111. Thus, in contrast to the well defined property of FDCs in stably maintaining ICs, these data suggested that FcαμR can promote IgM-IC degradation by FDCs. In accord with this counter-intuitive observation, FcαμR negatively regulated T-independent B cell responses ¹¹⁰. The cell biological basis for FDCs supporting endocytic recycling of intact ICs ⁶⁴, versus possibly promoting their FcαμR-dependent degradation is unknown. Moreover, the logic for degrading instead of maintaining IgM-ICs is not obvious. Adding to the complexity is evidence that the form of the IgM pentamer may be a determinant in whether or not an IgM-IC engages FcαμR. An under-appreciated feature of IgM pentamers is their association with the serum protein apoptosis inhibitor of macrophage (AIM/CD5L) a circulating member of the scavenger receptor cysteine-rich (SCRC) superfamily ^112–114. AIM association blocked IgM from binding to the FcαμR and promoted longer retention of IgM-ICs on FDCs, perhaps due to reduced internalization and degradation ¹¹¹. Further work is needed to define the conditions affecting the proportion of IgM that is complexed with AIM, and to determine how broadly FcαμR influences the durability of IgM-IC display on FDCs.

IgA is the most abundant antibody isotype at mucosal surfaces. Dimeric IgA is transported into mucosal secretions by the polymeric Ig receptor (pIgR), whereas monomeric IgA is present in circulation. In humans but not mice there is a monomeric IgA receptor, FcαR, an Fcrγ-associating activating receptor that is expressed by neutrophils and various macrophages and that can support ADCP, providing a possible route for IgA-mediated negative feedback on the B cell response ¹¹⁵. IgA receptor(s) may also be involved in “positive” feedback on mucosal antibody responses since IgA increases uptake of some antigens into mucosal lymphoid tissues via M cells ¹¹⁵.

As well as the differences in FcRs engaged by IgG, IgM and IgA, the various IgG subclasses differ in FcγR engagement strength ^50,78,81. Moreover, IgA, human IgG2 and IgG4 and mouse IgG1 have little ability to activate C’ ^49,50. Systemic responses can sometimes be dominanted by such antibody types ¹¹⁶. Thus, the antibody feedback that occurs during an infection or vaccination may vary based on the antibody isotype and subclass dominating the response.

Antibody feedback on naïve versus memory B cell and GC versus PC responses

An important question we introduced earlier is whether antibody feedback affects Bmem cell and naive B cell responses differentially. Supporting this possibility, antibody from a prior immunization strongly prevented participation of epitope-specific naïve B cells in GC responses while not preventing Bmem cells from becoming PCs ¹¹⁷. The basis for this selective inhibitory effect on naive but not Bmem cells remains incompletely defined but may in part be explained by some Bmem cells having a higher affinity BCR and thus being more competitive with soluble antibody for antigen. It may also be that some types of Bmem cells need less antigen engagement to enter into a response than naïve B cells. For example, IgG⁺ Bmem cells may be more sensitive to antigen due to differential signaling properties of the IgG isotype ^118,119. Additonally, studies of lung Bmem cells have provided evidence that they can differentiate into PCs in response to subthreshold amounts of antigen when provided in the context of innate signals ¹²⁰.

Another factor contributing to differential effects of antibody feedback on recall PC and GC responses may be the different kinetics and locations of these responses. Bmem cells can differentiate into PCs within 2-3 days of antigen exposure while GCs take several days to be launched and can involve selection events operating over weeks. Spatially, Bmem cells can encounter antigen in a variety of locations while GC B cells generally depend on the presence of antigen on FDCs. Given that antigen binding to FDCs is promoted by opsonization, the more antibody and C’ coating of the antigen, the more it is expected to dominate in the FDC network ^7,54,65. Moreover, during the time required for launching a recall GC response, there is likely to be an increase in antibody titer due to PCs that differentiate from Bmem cells. Thus, the same epitope that was available for induction of recall PCs may become more completely masked by newly secreted antibody. When deposited in the FDC network, this epitope will be less able to support participation of specific B cells in the GC response.

As well as being present in lymphoid tissues, Bmem cells can be resident at peripheral sites such as in the lung parenchyma following influenza infection ^109,120,121 The antigen reaching naive (follicular) B cells in lymphoid tissues will typically have been exposed to lymph and/or blood providing opportunity for coating by systemic antibody. The antigen reaching tissue resident Bmem cells will have been less exposed to systemic antibody and thus may be less heavily coated. Moreover, the types of FcR- and CR-expressing cells in non-lymphoid and lymphoid tissues will be quite different, and this will influence the extent that opsonized antigen is cleared versus presented. Further study is needed to understand the ways antibody feedback affects tissue resident B cell responses differently from responses in lymphoid tissues.

Antibody feedback in newborn and mother

An important feature of antibody feedback is that it can act trans-generationally. Thus, passively transferred IgG from the mother can diminish the ability of infants to mount antibody responses against vaccines ¹²². It has been suggested that high affinity maternal antibody binding to immunodominant epitopes can cause the fetal response to focus on non-immunodominant epitopes, possibly leading to inferior responses ¹²². With improvements in the efficacy of some adult vaccines, it has been proposed that there is a need to extend the period over which immunizations are given to the infant to ensure exposure of infant B cells to unmasked immunodominant epitopes. Maternal IgG and IgA can also dampen T and B cell responses against the microbiota, in this case in a beneficial manner. In the absence of maternal antibodies in a mouse model, a compensatory T-dependent antibody response to the microbiota occurred, altering the microbiome ¹²³.

A long running therapeutic use of antibody feedback has been to prevent Rh disease (or hemolytic disease) of the newborn. Mothers who lack the Rhesus antigen (Rh−) but who carry an Rh+ fetus are given an injection of anti-Rh IgG within a day or so of giving birth. This prevents the mother from mounting an antibody response against the Rh+ infant RBCs that enter the mother’s circulation ¹²⁴. Perhaps not surprisingly given the complex layers of feedback we have summarized above, the mechanism of protection is incompletely understood as it does not appear to require complete masking of epitopes on the fetal RBC antigen. Studies in mouse models have also not supported a dominant role for IgG mediated antigen (fetal RBC) clearance. A mechanism that has been suggested is trogocytosis, where the IgG promotes removal of antigen from the membrane by FcγR+ cells ^124–126. Additionally, steric interference with optimal BCR engagement may have a role. That is, the strength with which the BCR can be triggered may depend on an appropriate spatial occupancy of BCRs at the B cell-RBC interface ⁷⁰; if some of the antigens on the RBC surface are occupied with IgG this may disrupt BCR clustering even without (or in addition to) involving FcγRIIb negative signaling.

Understanding antibody feedback in humans to improve vaccines

Most conclusive mechanistic insights on antibody feedback have been observed or tested using critical mouse models. An important consideration is the lifespan of mice, 1-2 years, versus over seven decades for humans. This distinction will require unique adaptations of immunity where mice might be challenged only once per pathogen and thus have more clearly distinct innate like B cell subsets including B1 and MZ cells ¹²⁷ and a reduced tendency to “re-train” memory B cells with clear naive-bias in GCs or what the authors termed “naïve-addiction” ^16,35. Humans will likely encounter the same or evolved variants of pathogens repeatedly over a lifetime, and so appear to retrain memory B cells more readily ¹²⁸ and have less defined innate-like B cell subsets. This distinction should be considered as one can envision unappreciated differences in key feedback mechanisms that could alter assumptions based on studies in mice alone.

The concept that existing immunity could bias or limit current responses, particularly against evolving viruses such as influenza or now for SARS-CoV-2, has been referred to as “original antigenic sin ¹²⁹.” While this concept was initially centered on the uniqueness of a first exposure, it is likely a composite of exposures over a lifetime beginning in childhood that bias memory to certain strains and epitopes of evolving pathogens, referred to as “antigenic seniority ¹³⁰.” A more recent adaptation of this concept referred to as “immune imprinting” ¹³¹ has been used to express that biases based on existing immunity can have both negative and positive impacts on protection. The underlying factors driving immune memory bias are critical to understand or to exploit to improve vaccines against evolving pathogens. The complexity of various antibody feedback mechanisms that accumulate to a particular pathogen, especially those that are repeatedly encountered, will also shape B cell memory and influence future immune responses. While the presence of B and T cell memory is always predicted to enhance responses (good or bad), the wild card is serum antibody levels that can, as detailed above, drive increased IC and FcR or C’ interactions and increase immune responses in moderation. Conversely, if serum antibody is in abundance, responses can be inhibited by antigen masking, clearance, or FcgRIIB–mediated feedback inhibition. Importantly these antibody-mediated imprinting effects are driven by the persistence or waning of serum antibodies over time or by the degree of cross-reactivity of antibodies to variant pathogen epitopes. The context of past exposures can also alter serum antibody-mediated effector functions and C’ or FcR interactions due to the spectrum of Ig classes or glycosylation status as noted above. For example, people who have been previously infected, those receiving a live-attenuated vaccine, an mRNA vaccine to a single protein, or an adjuvanted vaccine would have distinct routes of exposure, and predicted serum antibody compliments that could variably dictate future immunity and vaccine outcomes. Thus, appropriate prime-boost strategies and timing are critical to elicit a desired response that competes with and complements existing feedback. A pressing example illustrating these issues is the current efforts to develop vaccines that universally protect against all influenza variants.

Antibody-feedback mechanisms are central to the biases of immunity, particularly regarding which epitopes are targeted. Immunodominant epitopes likely become selected by pathogen evolution, particularly for viral receptors, such that dominant epitopes tend to surround receptor binding domains where antibodies can inhibit infection. However, these same receptor surfaces can also readily mutate to evade immunity without affecting pathogen fitness as has been appreciated for many years for influenza ¹³², HIV ¹³³, and now based on extensive recent literature for SARS-CoV-2 antibody responses. Protective epitopes that mediate active processes affecting viral fitness are often sensitive to mutation and become conserved across viral variants. However, likely through viral evolution, these epitopes frequently become immunologically subdominant and effectively shielded from selective pressure to escape immune recognition. Viral variants with dominance of key conserved epitopes would be removed by natural selection. By way of illustration, the contact residues of the influenza receptor binding site mediate attachment to sialic acids on cells and is readily targeted by antibodies. Conversely, the HA stalk is critical for viral entry and highly conserved, but it is targeted at log reduced frequencies than more malleable HA head epitopes ^{132,134–136}. These subdominant conserved epitopes are a coveted target for producing universally protective vaccines against mutating pathogens. For example, booster immunization with highly divergent influenza strains can drive broadly protective memory and antibody responses in people on first ^{134,135,137–140} but not subsequent exposures ^136,137. Once the dominant head epitopes are targeted for an influenza strain, the HA stalk again becomes subdominant. Pre-existing serum antibody levels to the HA stalk correlates with reduced boost against this epitope ¹³⁷. Because HA is a single molecule with head and stalk regions linked, antigen clearance by pre-existing antibody or Fc inhibition of reactive B cells would be to the entire molecule and so do not fit with this observation when modelled; however masking of particular epitopes does and is predicted to be a dominant factor in this case ¹⁴¹. Targeting subdominant epitopes such as the HA stalk may be exasperated by the relative inaccessibility ¹³⁶ or relative paucity of stalk versus head epitopes that is predicted to make the stalk more easily masked ¹⁴¹. As serum antibody levels wane, as most responses do, cross-protection will decline but these epitopes will nonetheless remain masked. As detailed above, the dichotomy of having antibody levels high enough to obscure targeting of key epitopes but insufficient to provide protection is evident for malaria vaccines against circumsporozoite protein ¹⁹. Only easily accessed immunodominant or de novo (drifted) epitopes will be efficiently targeted on future exposures, redirecting immunity back to epitopes susceptible to escape by antigenic drift, as has been observed ^136,137. The conundrum of how to overcome pre-existing immune biases in a durable fashion has long been a hurdle to producing universal influenza vaccines ¹⁴².

As introduced in the epitope masking section above, in the case of SARS-COV-2 immunity, likely due to serum antibody masking, there is a shift in epitope targeting from the dominant epitopes prone to mutational drift to subdominant epitopes conserved across variants that is predicted to provide broader protection ^30–34. Such responses are dependent on antibody titers that will wane to intial SARS-CoV-2 variants, and so responses are predicted to again predominantly target dominant epitopes on future variants similar to influenza. Other key infectious diseases with evident impacts from antibody feedback in conjunction with pathogen adaptations to exploit these tendencies include malaria and HIV. For malaria, the repetitive nature of the key protective epitopes in the circumsporozoite protein provide various challenges in part from binding of competing antibodies that ultimately limit the ability to boost durable protective antibody titers, as reviewed in detail by Wahl and Wardemann ¹⁴³. In the case of HIV, viral evolution is massive even within a single individual both coupled to and because of immune subversion, making a broadly cross-protective vaccine even more daunting. There is evidence that similar to SARS-CoV-2 responses, repeat boosts with variant HIV envelope proteins can mask dominant epitopes and diversify the response ¹⁴⁴. Again, an outcome of increased breadth of protection would require careful consideration of booster timing and antigen design to exploit antibody feedback. In total, the landscape of epitopes targeted becomes a moving target for rapidly evolving pathogens with dramatic impact from antibody feedback mechanisms.

Concluding Remarks

Through studies in mouse models and increasingly in NHPs and in humans, a sophisticated understanding is emerging of how antibody feedback shapes the B cell response. Epitope masking can reduce and even inhibit B cell responses but is also a major factor in broadening responses and thereby providing some amount of pre-emptive protection against rapidly evolving pathogens. While challenging given the variability in immune responses in the human population, it may be valuable to incorporate this knowledge into vaccination strategies, for example, in deciding when during the antibody response decay curve to give booster immunizations. Similarly, knowledge about the amounts and kinetics of newly formed antibodies that are needed to augment the initial vaccine response through opsonization and follicular deposition may inform about the optimal delivery duration of an immunogen. The masking capability of strong IgG responses may also be a central mechanism of allergen-specific immunotherapy. With the growing prevalence of allergies, learning how to more effectively utilize antibody feedback to block IgE responses may be a fruitful area for further investigation. In the cancer field, as the importance of anti-tumor antibody responses becomes better understood, the influence of antibody feedback in augmenting or antagonizing these responses will need interrogation. The presence of tertiary lymphoid structures with opsonized antigen-capturing FDCs is increasingly recognized as a positive correlate of improved outcome in several solid tumor types. Moreover, epitope spreading is an important concept in both cancer immunity and in autoimmunity and the role of antibody feedback in this process needs deeper investigation.

The range of mechanisms by which antibodies can influence B cell responses points to the challenge of being able to precisely predict how to engage these feedback mechanisms in a desired manner therapeutically. We are of the view that a lot of additional data are needed before an algorithm could be developed to accurately predict the types of antibody feedback that will occur during a new infection or vaccination. Questions that must be considered include: How much and what classes of early antibodies are induced? How much antigen is available to be opsonized? How efficiently are intact forms of the (opsonized) antigen captured on FDCs for pro-longed display? To what extent does the opsonization cause clearance of the antigen by myeloid cells versus making it more stimulatory (via CR2) or inhibitory (via FcγRIIb) for B cells? How much (acute and chronic) inflammation is occurring and to what extent is this upregulating FcγR expression? Are the IgM FcRs upregulated and influencing IgM-IC display? Are glycosyltransferase expression levels in B cells and PCs being modulated? For each of the molecular systems that have influences on antibody feedback, there are likely to be variants that alter the activity of the pathway. Understanding how human genetic variability in pathways associated with antibody feedback influences the success of humoral immune responses remains a fascinating area for future discovery.