Dr B. Heyman, Department of Genetics and Pathology, Rudbeck Laboratory, Uppsala University, SE-751 85 Uppsala, Sweden. E-mail: email@example.com
Antibodies administered in vivo together with the antigen they are specific for can regulate the immune response to that antigen. This phenomenon is called antibody-mediated feedback regulation and has been known for over 100 years. Both passively administered and actively produced antibodies exert immunoregulatory functions. Feedback regulation can be either positive or negative, resulting in >1000-fold enhancement or >99% suppression of the specific antibody response. Usually, the response to the entire antigen is up- or downregulated, regardless of which epitope the regulating antibody recognizes. IgG of all isotypes can suppress responses to large particulate antigens like erythrocytes, a phenomenon used clinically in Rhesus prophylaxis. IgG suppression works in mice lacking the known Fc-γ receptors (FcγR) and a likely mechanism of action is epitope masking. IgG1, IgG2a and IgG2b administered together with soluble protein antigens will enhance antibody and CD4+ T-cell responses via activating FcγR, probably via increased antigen presentation by dendritic cells. IgG3 as well as IgM also enhance antibody responses but their effects are dependent on their ability to activate complement. A possible mechanism is increased B-cell activation caused by immune complexes co-crosslinking the B-cell receptor with the complement-receptor 2/CD19 receptor complex, known to lower the threshold for B-cell activation. IgE-antibodies enhance antibody and CD4+ T-cell responses to small soluble proteins. This effect is entirely dependent on the low-affinity receptor for IgE, CD23, the mechanism probably being increased antigen presentation by CD23+ B cells.
It is remarkably difficult to generate an antibody response to small soluble protein antigens without using adjuvants, a fact referred to by Charlie Janeway as ‘the immunologist's dirty little secret’. Therefore, a significant part of our knowledge about antibody responses stems from studies of mice immunized with antigens in adjuvants. In contrast, antibody feedback regulation is studied in systems where antibodies and antigens are administered in physiological salt solutions either as pre-formed complexes or within hours of each other. In this situation, antibodies can function as natural adjuvants, and potent primary and secondary antibody responses as well as T-cell responses are generated (reviewed in [1, 2]). Antibodies produced early in humoral responses, IgM and IgG3, as well as later on, IgG1, IgG2a, IgG2b and IgE, can enhance antibody responses albeit by different mechanisms. In other situations, antibodies can exert negative effects on antibody responses. An interesting finding is the dual effects of IgG1, IgG2a and IgG2b administered with soluble protein antigens. Their enhancing effects seem to be mediated via activating Fc-γ receptors (FcγR) but the enhancement is simultaneously negatively regulated via the inhibitory FcγRIIB. Another example is the ability of IgG to inhibit responses to erythrocytes which takes place in the absence of the known FcγR, noteworthy also in the absence of FcγRIIB. This effect is probably caused by antibodies masking the antigen and preventing its recognition by B cells.
Antibody-mediated feedback regulation has been known for over a century  and has been extensively reviewed elsewhere [1, 2]. The purpose of the present review is to give a comprehensive summary of the four major feedback pathways operating when antibodies and external antigens form immune complexes in vivo.
IgG-mediated suppression of antibody responses
The most frequently studied antibody feedback pathway is IgG-mediated suppression of responses to erythrocytes [4, 5], although IgG can also suppress responses to proteins administered in adjuvants . All murine IgG subclasses can suppress anti-erythrocyte responses [7, 8] but the efficiency of suppression correlates with affinity of the IgG antibodies [7, 9]. Both primary and secondary antibody responses are suppressed whereas T-cell priming appears to occur normally . Remarkably, IgG administered several days after erythrocytes can suppress an ongoing antibody response [11, 12] implying that the antigen needs to be recognized by cells of the immune system not only during the initiation of an antibody response but also during the following days.
Three different mechanisms for how IgG exerts its suppressive capacity have been proposed:
1IgG may by masking of antigenic epitopes prevent specific B cells from recognizing and binding to the antigen.
2IgG/erythrocyte complexes may be captured by FcγR+ phagocytes and eliminated more rapidly than erythrocytes alone. This would remove antigen from areas where it can be recognized by the immune system.
3IgG/erythrocyte complexes may co-crosslink the B-cell receptor (BCR) with the inhibitory FcγRIIB, also expressed on the B-cell surface. This would lead to negative regulation of the B cell, mediated by immunoreceptor tyrosine-based inhibitory motifs in FcγRIIB acting on immunoreceptor tyrosine-based activation motifs (ITAM) in the BCR . Such co-crosslinking inhibits B-cell activation in vitro [14, 15].
The first mechanism, epitope masking, would function without the IgG(Fc)-portion whereas the two latter mechanisms would require intact IgG(Fc)-portions. Studies testing whether F(ab′)2 fragments were suppressive or not have given conflicting results (for discussion, see [10, 16]). In an alternative approach, FcγR-deficient mice were used to assay for FcγR dependence and it was shown that IgG suppresses equally well in wild-type mice as in mice lacking all known FcγR, including FcγRIIB [10, 12]. These observations together with previous findings that complement activation is not required for the suppressive ability of IgG , suggest that an Fc-independent mechanism is operative. Although direct evidence is difficult to obtain, epitope masking by IgG seems to be a likely explanation (Fig. 1). This is in agreement with the fact that not only IgG, but also IgM [4, 7, 18] and IgE [10, 19] can in certain situations suppress anti-erythrocyte responses. Should epitope masking be the mechanism, a more appropriate term for IgG-mediated suppression may be IgG-mediated prevention of an antibody response.
The ability of IgG antibodies to suppress responses to erythrocytes has been used clinically since the 1960s . Rhesus-negative women, carrying Rhesus-positive fetuses, can be sensitized against fetal erythrocytes acquired via transplacental haemorrhage. This sometimes causes haemolytic disease of the newborn as maternal IgG antibodies specific for fetal Rhesus-positive erythrocytes can pass the placenta and destroy fetal erythrocytes. To prevent this, IgG anti-Rhesus D (RhD) is routinely administered to Rhesus-negative women after delivery of a Rhesus-positive baby. The passively administered IgG antibodies prevent an active immune response against RhD and this treatment has dramatically decreased the incidence of haemolytic disease of the newborn . It seems reasonable to assume that IgG-mediated suppression of responses to sheep erythrocytes (SRBC) in mice reflects the IgG-mediated suppression of responses to RhD-positive erythrocytes in Rh-negative women. However, a few discrepancies are worth noting (discussed in ). In the human system, doses of IgG anti-RhD which are too low to mask all D epitopes still result in efficient inhibition of the anti-RhD response . Moreover, IgG specific for another blood group antigen, Kell, was in one study shown to suppress the anti-D response to D+K+ erythrocytes in D−K− humans . One possible explanation for the discrepancies between IgG-mediated suppression in the murine and human systems could be that the considerably lower responses to allogeneic erythrocytes in humans is easier to suppress than the vigorous responses to sheep erythrocytes in mice. Therefore, also an incomplete epitope masking may be sufficient to lower the immunogenicity of the RhD-positive erythrocytes. Alternatively, antibodies may target an erythrocyte for destruction via mechanisms that are not dependent on FcγR or complement. There is to date no evidence to support that Rh prophylaxis requires FcγRIIB in vivo, although this, based on in vitro studies, is sometimes assumed to be the mechanism. Although IgG-mediated suppression of antibody responses to sheep erythrocytes in vitro is dependent on FcγRIIB, suppression in vivo works equally well in the absence of this receptor . Therefore, one has to be careful when interpreting the in vivo relevance of findings made in vitro.
Although IgG antibodies are used with great success to suppress anti-RhD responses in humans and easily suppress >99% of a vigorous anti-SRBC response in mice, the biological role of this feedback pathway is probably not to completely suppress antibody responses. More likely, the masking of certain epitopes by high affinity IgG antibodies emerging relatively early in an immune response may give the immune system the opportunity of responding also to other epitopes and thereby ensure a broader range of antibody specificities.
IgG1-, IgG2a- and IgG2b-mediated enhancement of antibody responses
Although IgG1, IgG2a and IgG2b suppress responses to particulate antigens, they are able to enhance responses when administered together with soluble protein antigens such as ovalbumin (OVA), bovine serum albumin and keyhole limpet haemocyanine (KLH). This dual effect is indeed dependent on the type of antigen and not on differences in the IgG antibodies used: one and the same monoclonal trinitrophenyl (TNP)-specific IgG suppressed SRBC-specific responses when administered with SRBC-TNP but enhanced KLH-specific responses when administered with KLH-TNP [25, 26]. These IgG isotypes enhance primary as well as memory responses [27–29] and immunization with IgG1 anti-nitrophenyl (NP)/NP-KLH complexes increases somatic mutations in NP-specific germinal centre B cells .
IgG1-, IgG2a- and IgG2b-mediated enhancement is dependent on the presence of activating FcγR and these must be expressed on bone marrow-derived cells [29, 31]. In vitro and ex vivo, antigens complexed to IgG are captured by FcγR-expressing antigen-presenting cells (APCs) and presented more efficiently to CD4+ T cells than antigen alone [32–36]. To test whether such a mechanism also explains IgG-mediated enhancement of antibody responses, we have used a system where expansion of antigen-specific CD4+ T cells can be studied directly in vivo . CD4+ T cells from DO11.10 mice, transgenic for a T-cell receptor (TCR) recognizing an OVA-peptide together with MHC-II Ad, were transferred to syngeneic wild-type BALB/c mice. After immunization with IgG2a anti-TNP/OVA-TNP intravenously without adjuvants, the expansion of OVA-specific T cells was followed using a monoclonal antibody specific for the transgenic TCR. Specific T cells expanded rapidly and peaked 3 days after immunization  (Fig. 2). The enhanced T-cell response was followed by an enhanced OVA-specific antibody response and both T- and B-cell responses were dependent on activating FcγR . This observation is compatible with the idea that complexes of IgG and antigen are captured by FcγR+ APC and that they enhance immune responses in vivo via increased antigen presentation to specific T helper cells (Fig. 3). The biological role of IgG-mediated enhancement is most likely to help in the induction of an efficient memory response, as antigen-specific IgG would often be present in the circulation when the immune system encounters a booster dose of an immunogen.
FcγRIIB is the only inhibitory FcγR and is expressed on most cells of the immune system except NK and T cells. It exerts its negative regulation on ITAM-bearing receptors such as the BCR, FcγRI, FcγRIII, FcγRIV and FcɛRI . This gives FcγRIIB a central role in immunoregulation and it has been shown to negatively regulate a number of immune responses such as B-cell activation, BCR-mediated antigen presentation, antigen presentation by DC, maturation of DC and release of mediators (reviewed in ). FcγRIIB-deficient mice have increased susceptibility to many autoimmune diseases  and have increased anaphylactic reactions .
The role of the receptor in regulation of in vivo antibody responses has been studied using FcγRIIB-deficient mice. As IgG-mediated enhancement is severely impaired in mice lacking activating FcγR, but which have normal levels of FcγRIIB , FcγRIIB is not required for induction of enhancement. However, this receptor has a negative effect on responses to IgG-complexed soluble antigens, demonstrated by an enhanced IgG-mediated enhancement in FcγRIIB-deficient mice [31, 38]. The response in FcγRIIB-deficient mice to IgG-complexed antigen was sometimes more than 100-fold higher than in wild-type mice [31, 38]. Notably, FcγRIIB does not completely turn off antibody production but ‘permits’ IgG to enhance antibody responses, as demonstrated by the enhancement seen in wild-type mice. This illustrates the intricate balance between inhibitory and activating effects mediated by IgG and FcγR. The outcome of regulatory effects of IgG antibodies is dependent on factors such as the expression level of activating and inhibitory FcγR, which are often co-expressed on the same cell. It also depends on the IgG subclass composition, as different subclasses have different ratios of activating-to-inhibitory receptor binding .
IgM- and IgG3-mediated enhancement of antibody responses
It has been known since the 1970s that complement factor C3 is important for the capacity of animals to mount normal antibody responses to suboptimal doses of antigen . With the availability of gene-targeted mice, it became clear that deficiencies in classical [45, 46] but not alternative  pathway components resulted in impaired antibody responses, suggesting that antibodies were involved. Mice where complement receptor 1 and 2 (CR1/CR2) was blocked by monoclonal antibodies or deleted by gene targeting have a similar phenotype with regard to antibody responses as animals lacking classical pathway components [48–51]. CR1/CR2 are derived by alternative splicing of the same gene. In mice, they are expressed on B cells and follicular dendritic cells (FDC) implying that one or both of these cells are the effector cells.
IgM is an efficient complement activator and when IgM is passively administered together with large antigens like erythrocytes [5, 52], malaria parasites  and KLH  it enhances the antibody responses. IgM upregulates both primary [5, 52–54] and memory responses  and increases affinity maturation . The enhancing effect is only seen with suboptimal doses of antigen  and in order to enhance, IgM must be administered within a few hours of the antigen. The ability of specific IgM to enhance antibody responses requires a functioning complement system. IgM cannot enhance in mice lacking C3 owing to treatment with cobra venom factor  or in mice lacking CR1/CR2 . Mutated or monomeric IgM molecules which no longer activate complement also loose their enhancing effects [52, 54].
We recently found that IgG3 also can enhance antibody responses . IgG3 enhances in mice lacking all activating FcγR  and in mice selectively lacking FcγRI , which is suggested to be the IgG3-binding FcγR . Unlike IgE  and IgG2a , IgG3 did not induce proliferation of specific T cells in vivo  (Fig. 2). Enhancement was severely impaired in mice lacking CR1/CR2 and in animals depleted of C3 by treatment with cobra venom factor .
Thus, both IgG3 and IgM depend on the complement system for their enhancing capacity.
Both isotypes are relatively T independent and both are produced early in an immune response. One single IgM molecule can activate complement , provided it can change conformation. IgG3 has the capacity to self-associate into multivalent complexes, giving it a high functional affinity and probably also increasing the likelihood of C1q binding [63, 64]. These factors may explain why IgM and IgG3 utilize the complement system in feedback enhancement and, speculatively, also why IgM can only enhance responses to large antigens (which are required to enable the huge IgM molecule to bind with its five arms and change conformation). IgE and IgG1 are unable to activate complement, whereas IgG2a and IgG2b require the rather unlikely event that two IgG molecules bind close enough on the antigen to cooperate in C1q-binding. Therefore, it seems logical that these isotypes use Fc-receptors for their feedback-enhancing effects.
It is not understood why complement has such a dramatic impact on antibody responses, but several, not mutually exclusive, mechanisms are possible:
1Antibody/antigen/complement complexes co-crosslink the BCR and the CR2/CD19 complex, known to lower the threshold for B-cell activation  (Fig. 4).
2Antibody/antigen/complement complexes are captured by FDC in the spleen and lymph nodes, increasing the effective concentration of antigen [54, 66–68].
3Antibody/antigen/complement complexes are endocytosed efficiently via CR1/CR2 on B cells and presented to T helper cells which in turn help B cells to produce antibodies. In vitro, B cells can take up antigen via CR1/CR2 and present to T cells [69–72], but in vivo studies in adoptive transfer/hapten-carrier systems do not support a role for complement in T helper-cell activation [45, 73].
Regardless of which mechanism(s) that cause antibody/complement-mediated enhancement, it remains to be explained how a primary antibody response can be dependent on classical (antibody induced) complement activation: in naive mice very low amounts of specific IgM or IgG3 is available to form complexes with the antigen. One possibility is that capture of antigen by natural IgM and IgG3 suffices to start complement activation in primary antibody responses. Once activation of specific B cells has taken place, these will secrete antigen-specific IgM and IgG3, further enhancing the positive feedback loop. In line with this, mice lacking secretory IgM had impaired antibody responses [74–76] which could be reconstituted by transfusion of IgM from naive mice .
IgE-mediated enhancement of antibody responses
In the 1990s, in vitro experiments showed that IgE-complexed antigens, binding to the low-affinity receptor for IgE (CD23) on the surface of B cells, could be internalized and presented efficiently to T helper cells [77, 78]. These observations led to the hypothesis that a physiological role of IgE may be to capture antigens and enhance T helper-cell activation . We tested this idea in an in vivo system where mice were immunized with monoclonal IgE anti-TNP together with various protein antigens coupled to TNP and found that IgE indeed upregulated carrier-specific antibody responses . IgE enhances responses to small soluble protein antigens but not to erythrocytes or large proteins such as KLH [80, 81]. All isotypes, including IgE, are upregulated and the IgG response peaks as early as 6 days after priming [81, 82]. The enhancing effect of IgE is mediated by CD23, evidenced by lack of enhancement in mice where CD23 is blocked by a monoclonal antibody [80, 81] or in CD23-deficient mice [61, 83, 84]. Lack of involvement of the high-affinity IgE-receptor (FcɛRI) was confirmed by the normal IgE-mediated enhancement in FcRγ-chain-deficient mice, which do not express FcɛRI .
IgE-mediated enhancement works normally in IL-4-deficient mice . As IL-4 is required for expression of the CD23b isoform , this observation suggests that enhancement is mediated via the constitutively expressed CD23a isoform. In mice, CD23a is expressed only on B cells and FDC. Adoptive transfer experiments demonstrated that FDC are not involved, leaving the B cells as the likely effector cells . Therefore, should the IgE-enhanced antibody response in vivo be caused by enhanced presentation of antigenic peptides to T helper cells, in analogy with in vitro findings [77, 78], it would imply that B cells can activate naive T cells. This is usually considered to be performed by DC , and to investigate whether IgE/antigen complexes do indeed activate specific T cells in vivo, we used the DO11.10 experimental system described above. IgE anti-TNP administered together with OVA-TNP induced a marked proliferation of OVA-specific CD4+ T cells peaking 3 days after immunization  (Fig. 2). Enhancement of T-cell responses required the presence of CD23 which had to be expressed on B cells. However, once such cells were present (in chimaeric mice), also CD23− B cells produced enhanced levels of specific antibody after immunization with IgE/antigen . These observations are compatible with the antigen presentation hypothesis suggesting that the enhanced antibody levels are a result of increased T-cell help to specific B cells (Fig. 5). IgE-mediated enhancement, as well as all other feedback regulatory pathways, is specific for the antigen within the immune complex. This means that although all CD23+ B cells regardless of BCR specificity, take up and present IgE/antigen to specific T cells, only B cells which are able to recognize the antigen via their BCR receive cognate T-cell help leading to antibody production. This resembles previous in vitro findings, demonstrating that unspecific B cells, via CR2, take up and present immune complexes containing complement factors, but that only antigen-specific B cells are stimulated to antibody production . Thus, when IgE is present, CD23+ B cells in vivo seem to function like DC (which also lack specificity for the antigen they present) but induce specific T cells which then provide help to B-cells.
It has been suggested that IgE-mediated enhancement of antibody responses may create a vicious circle in atopic patients who already have high levels of allergen-specific IgE ready to form complexes with allergen . The biological role of IgE-mediated enhancement of antibody responses is not clear. However, the magnitude of T-cell proliferation observed after immunization with IgE and 20 μg of OVA-TNP  is similar to what was seen after immunization with 2 mg of uncomplexed OVA together with LPS . This suggests that capture of low doses of antigen by IgE is an efficient way of priming CD4+ T cells and this may be important in situations where little antigen is available. Although serum IgE levels are generally low, local concentrations may be high. As CD23 binds IgE also in the absence of antigen, B cells may be pre-loaded with IgE, ready to capture antigen and induce efficient CD4+ T-cell responses.
Work in the authors’ laboratory was supported by Agnes and Mac Rudberg's Foundation; Ellen, Walter and Lennart Hesselman's Foundation; Emil and Ragna Börjesson's Foundation; Hans von Kantzow's Foundation; Ankarstrand's Foundation; King Gustaf V:s 80 Years Foundation; Lilly and Ragnar Åkerhams's Foundation; The Swedish Research Council; Ollie and Elof Ericsson's Foundation; The ‘Network for Inflammation research’ funded by the Swedish Foundation for Strategic Research; The Swedish Society for Medical Research; and Uppsala University.