Original antigenic sin – not so sinful after all

Abstract

“Original antigenic sin” predicts that antibody responses to secondary infections with escape mutants are dominated by specificities to the original pathogen. Using transgenic mice in which antibodies are tagged based on cellular origins and kinetics, Schiepers et al. support this prediction, and reveal accumulation of cross-reactive specificities chiefly among long-lived responses.

Early studies investigating antibody responses to influenza virus suggested that humans infected with one strain of influenza A virus and then re-infected with a drifted strain that carried several mutations in the hemagglutinin surface protein generated stronger recall serum antibody responses to the first virus than specific responses to the second [1]. The reduction in de novo antibody responses to the second virus was termed “original antigenic sin”, indicating the detrimental nature of this phenomenon as it might reduce protective immunity to secondary infections. While this observation has been the subject of considerable investigation and debate [1–3], insights into the underlying mechanisms have been lacking.

Schiepers et al [3] recently tackled these important questions and the mechanisms explaining it. Recall responses to booster vaccinations and to natural pathogen exposures following vaccinations are central tenants of vaccine-induced immunity. Thus, understanding when prior exposure to an antigen might result in reduced –and potentially suboptimal– responses is crucial for devising appropriate vaccination strategies.

Essential for exploring mechanisms of original antigenic sin, or as Schiepers et al. [3] prefer to call it “primary addiction”, is the examination of the cellular origins of antigen-specific serum antibodies. In mammals, B cells induce antibodies via at least two distinct response patterns, one resulting in Extrafollicular Foci (EF) and the other in Germinal Centers (GC) [4]. EF consist of rapidly proliferating B cell clusters that differentiate into short-lived plasmablasts and plasma cells in interfollicular and medullary cord areas of lymph tissues. Moreover, EFs provide a first wave of serum antibodies, including IgM, but also EF B cells also undergo immunoglobulin class-switch recombination to control initial pathogen spread [4]. With EF responses being active for only a few days to a week, and antibody half-lives between 1 day (IgM) to 4 weeks (IgG) in mice and humans, antibodies generated by EF responses appear and then disappear rapidly. Extrafollicular memory B cells (B_mem) may also be induced, many of which remain unswitched [5].

GCs are delayed by a few days to a couple of weeks compared to EF during many infections of mice and humans [4]. During GC induction, antigen-stimulated B cells together with CD4⁺ T cells return to the B cell follicle where re-stimulation by follicular dendritic cell-presented antigen induces rounds of GC B cell proliferation; during this time, the antigen-binding region of the immunoglobulin locus heavily mutates. The resulting mutated B cells compete for antigen-binding and T cell help in an iterative process that leads to the emergence of affinity matured B cells. B cells egressing early from GCs predominantly join the circulating pool of non-secreting Bmem, while those egressing later migrate to the bone marrow where they can be retained as antibody-secreting long-lived plasma cells (LLPC) [6]. Thus, EF and GC responses starkly differ in the quality and kinetics of their antibody responses, with the former providing immediate early protection and the latter contributing to the development of a persistent immune barrier that is shaped by prior pathogen exposure.

At the cellular level, EF and GC are easily distinguishable by flow cytometry. However, assigning specific serum antibodies to either the EF or GC response has been possible only when one or the other response was present. Schiepers and colleagues [3] have overcome this technical hurdle by generating transgenic mice in which the Igκ light chain locus, utilized by 90 – 95% of all B cells, includes a floxed Flag-tag and a second, Cre-ERT2-inducible Strep-tag, whereby a cell will switch to the Strep-tag when exposed to tamoxifen, because the Flag-tag is excised. By crossing these mice with mice expressing Cre under the control of a GC-specific promoter ((S1pr2);[7]), antibodies and cells marked by the Strep-tag are identified as being from GC origin at the time of tamoxifen treatment. Those expressing the Flag-tag, are generated from EF responses, or arise from GCs that are activated outside the window of tamoxifen injection. An additional DNA barcode enables BCR sequencing [3].

With these mice, the authors studied the effects of a primary antigen exposure upon recall responses using various antigens [3]. Homologous boost, using the same model of hapten-carrier antigen for primary and secondary immunizations revealed a strongly biased use (addiction) of primary GC-derived B cells for recall antibody responses, which were encoded entirely by Strep-labeled antibodies. Yet, Strep⁺ cells did not appear to form GCs. Instead, GC B cells induced by the boost were generated by Flag-tagged B cells. This is consistent with recent findings that GC-derived B_mem preferentially induce EF responses during a recall response [8,9]. However, the resulting increases in antibody titers from the Strep-tagged cells, i. e. the primary GC-derived responses, appeared to be long-lived – representing a response type that is typically associated with GCs and not EFs [3]. Therefore, this might suggest that either secondary EFs can give rise to LLPCs, or, as previously suggested [9], a small fraction of the original GC-derived B_mem might generate most of the observed increased antibody responses by forming GCs. Notably, the de novo generated Flag-tagged antibodies were induced but appeared to be limited to the generation of short-lived IgM [3]. Thus, de novo responses did not give rise to long-lived Flag-tagged antibody responses, despite the dominance of Flag⁺ cells in GCs. Thus, indicating suppression of antibody responses to new, non-cross-reactive epitopes at the level of LLPC generation.

The extent to which the Flag-tagged B cells (either de novo induced GC- or EF-derived) contributed to the antibody recall response was shown to depend on the antigen relatedness and amount of epitope overlap between the antigens used for prime and boost. Prime/boost with either two heavily drifted strains of influenza A virus, or two distinct SARS-COV-2 spike proteins, showed a significantly more modest suppression of GCs induced by naïve B cells or recall EF-derived B_mem responses compared with strong suppression seen when the same antigen was used for both prime and boost. When prime/boost antigens expressed larger numbers of unique epitopes, Flag-tagged antibodies appeared in the long-lived serum antibody pool. Thus, antibody suppression was modest.

The study indicates that the extent of de novo antibody response contributions triggered by infection with one pathogen is based on the degree of epitope sharing and antigen relatedness to a previously encountered one. This might be key for understanding the evolutionary pressures for the observed heavy reliance (addiction) on B cells induced during priming. With cross-reactive recall responses by B_mem boosting the LLPC pool and de novo B cell responses largely directed against non-shared epitopes and blocked from the LLPC pool, the phenomenon of “original antigenic sin” limits the LLPC pool to broadly useful specificities, while de novo responses generate specific and protective antibodies during an acute infection. Rather than a “sin”, this process may drive a humoral immune barrier that is trained to protect from infections with new but related pathogens, such as escape mutants.

Figure 1. Recall responses drive long-lived cross-reactive over short-lived unique antibody responses.

Work by Schiepers et al. explored the effects of primary antigen exposure (prime) on the antibody responses to repeat antigen encounters (boost). The study [3] showed suppression of long-term but not short-term antibody responses when prime and boost antigens were identical (top). When the antigens differed in a small number of epitopes, the cross-reactive response is boosted, but the unique response to the boost antigen is suppressed and only seen in a short-lived response (middle). In contrast, when there is little epitope sharing between the antigens, suppression is minimal and both the primary and unique secondary responses are induced and participate in the long-lived compartment due to germinal center response participation (bottom panel). Figure created with Biorender.