Breaking the Paradigm: Selection of Self-Reactive Natural Antibodies

The question of how the adaptive immune system can be, on the one hand, responsive to an apparently infinite number of exogenous Ags and, on the other hand, largely tolerized to all endogenous or “self ” Ags was the focus of many early studies in the burgeoning field of immunology. Collectively, they led to the current paradigm of B cell development and tolerance induction, which was confirmed and expanded through groundbreaking studies conducted in the late 1980s and early 1990s. Using mice genetically engineered to express transgenes encoding a BCR and its cognate, membrane-anchored self-antigen, researchers demonstrated unequivocally that self-reactive B cells are either rapidly deleted, undergo receptor editing, or are rendered Ag unresponsive (anergic) (1–3). Subsequent work identified and further defined the developmental checkpoints that guide B cell tolerance induction in adult mice and humans (4, 5).

The constitutive low-level presence of self-reactive, serum autoantibodies, noted more than 80 y ago (6), however, created an apparent paradox for the original models of B cell tolerance induction. These autoreactive Abs increase following tissue destruction and viral infections, and they are inducible by injection of tissue extracts, indicating that exclusion of self-reactivity among B cells is not complete. Indeed, the spontaneous secretion of self-reactive, “natural” Abs (nAbs) was shown to occur independent of obvious tissue destruction, immunization with foreign Ags, exposure to microbiota, or food-related Ags (6, 7). Importantly, rather than evoking an autoreactive pathological response, the presence of self-reactive natural IgM reduces the incidence of autoimmunity. nAbs were shown also to protect against harmful autoimmune responses (8). A classic example is the presence of venom-neutralizing serum autoantibodies in vipers (9). Furthermore, multiple studies have demonstrated early immune protection by broadly (self-)reactive natural IgM against viral, bacterial, and protozoal pathogens (10), indicating the evolutionary conservation of such broadly reactive nAbs. The Pillars of Immunology article by Kyoko Hayakawa, Randy Hardy, and colleagues, published in 1999 in Science (11), identified fundamental mechanisms leading to the development of these “useful,” self-reactive nAb-producing cells.

The authors had shown previously that nAb production in mice was associated with a small population of CD5⁺ B cells that accumulate in body cavities. Indeed, the discovery of these cells was first made by Dr. Hayakawa while a postdoc in the Herzenberg laboratory (12, 13). During the time of the Pillars article, a controversy had arisen as to whether CD5⁺ B cells constitute a lineage (B-1) arising from precursors during fetal and neonatal development that is distinct from that of B cells produced in the adult bone marrow (B-2). Alternatively, the B-1 cells might be derived through differential activation of follicular B cells, because induction of CD5 expression had been noted in vitro under certain culture conditions (14, 15), and its expression is a mark also of anergic B cells (16). Given the controversy at the time, it is perhaps not surprising that the authors chose to forgo integration of their findings in the context of the B-1/B-2 paradigm. Instead, they referred to this subset simply as “CD5⁺ B cells.”

The authors exploited their previous identification of a germline-encoded BCR expressed by some CD5⁺ B-1 cells, which they used to generate a BCR Igμ-H chain transgenic (TG) mouse. The TG BCR bound a carbohydrate epitope on the glycoprotein Thy-1, a receptor expressed by thymocytes. Thy-1 also is the target of previously identified nAbs, termed anti-thymocyte autoantibodies (ATAs) (17, 18). The Igμ TG mice showed increased circulating levels of serum ATAs compared with controls, with concentrations increasing over time. TG⁺ cells were identified by flow cytometry using the Igh allotype differences between the transgene-encoded BCR (Igμ-a) and the background host allotype (Igμ-b), which revealed that most spleen TG⁺ B cells were CD5⁻ and were unable to generate ATAs following in vitro LPS stimulation. In contrast, and consistent with the distribution of B-1 cells, the highest frequencies of TG⁺ CD5⁺ B cells accumulated in the body cavities. Stimulation of the latter with LPS induced differentiation into ATA-secreting cells. In support of these functional studies, the BCRs of hybridomas generated from CD5⁺ and CD5⁻ TG B cells showed pairing of the μTG with the canonical ATA-specific L chain Vκ1C-Jκ2 only in CD5⁺ cells, whereas the μTG⁺ CD5⁻ cells used a variety of L chains that did not confer ATA specificity. Thus, the data firmly established that B cells expressing a BCR encoding self-reactive nAbs could develop into CD5⁺ B-1 cells but not CD5⁻ B cells.

According to established paradigms of B cell development, the new data might suggest that the Thy-1–specific B cells escaped negative selection, as their precursors might have been sequestered away from Thy-1 Ag early in ontogeny. However, when Thy-1⁺ thymocytes were injected into the peritoneal cavity, neither deletion nor Ag-specific expansion of μTG⁺ CD5⁺ B cells was observed, which argued against a selection-centric explanation. More importantly, the authors examined how Thy-1–reactive B-1 cells develop in the absence of their cognate self-antigen, in essence, asking the flipside of questions posed in earlier studies of autoreactive bone marrow B cell development, where the introduction of a self-antigen caused elimination of the reactive B cells. By crossing the ATA μTG mice with Thy-1 knockouts, the authors found that serum ATA and ATA-encoding CD5⁺ μTG⁺ B cells were detectable only in Thy-1–expressing mice but not in Thy-1 knockout mice. Furthermore, they showed that ATA-encoding B cells developed after adoptive transfer of TG⁺ newborn liver cells (as a source of B cells) into SCID mice only when coadministering bone marrow from wild-type Thy-1 mice but not from Thy-1–deficient mice.

Collectively, the study described in this Pillars paper demonstrated that the development and/or expansion of B cells with BCRs encoding self-reactive nAb requires 1) their ability to be sequestered into the CD5⁺ B-1 cell pool, and 2) the presence of cognate self-antigen. The study thus defined fundamental differences in the developmental processes guiding CD5⁺ B-1 and B-2 cells, as well as their functions. Whereas B-2 cells underwent negative selection and/or receptor editing upon engagement with self-antigen, demonstrated by the lack of ATA-encoding specificities among B-2 cells, CD5⁺ B-1 cells required self-antigen recognition for positive selection of cells encoding a nAb. Given that serum ATA levels increased over time in wild-type mice but not in Thy-1 knockout mice, the presence of self-antigen might not simply help B-1 cell selection, but also their expansion, a distinction that could not be made in this initial study.

Given that B-1 and B-2 cells are produced at different times in ontogeny, these studies suggested a dichotomy for B cell development in the fetus/neonate versus the adult mouse. This reasoning ultimately led to the discovery of epigenetic regulators Let7/Lin28 as critical determinants of fetal versus adult hematopoiesis, which differentially drive B-1 and B-2 cell development in a cell-intrinsic manner (19, 20). The original Pillars paper also stimulated studies identifying B cell–extrinsic signals in fetal liver and bone marrow that differentially affect their development. Indeed, it was shown recently that the relatively low production of IL-7 in fetal liver compared with bone marrow promotes early L chain rearrangement, as IL-7 suppresses RAG expression in large pre-B cells. The resulting L chains in developing pro-B cells can then promote mature BCR assembly in cells with rearranged Igμ chains that are unable to assemble a pre-BCR in association with surrogate L chains. This process overcomes deletion of cells at the pre-B stage, and instead selects for a unique B-1 cell repertoire (21, 22), which is consistent with the presence of small numbers of CD5⁺ B-1 but not B-2 cells in surrogate L chain–deficient mice (23, 24). Self-antigen recognition, as presented in this Pillars article, presumably occurs at this point in development, supporting clonal expansion and driving expression of CD5 in a manner analogous to its induction on developing T cells, where CD5 serves to modulate TCR signaling strength (25).

The highlighted Pillars of Immunology article provides a context for more recent studies from the Rajewsky laboratory, who elegantly showed that B-2 cells were “turned into” B-1 cells by swapping out their BCR for one usually expressed on B-1 cells (26). At its fundamental level, these subsequent findings confirmed broader conclusions from the Pillars article, namely that BCR self-reactivity is a defining feature of B-1 but not B-2 cells, but they also provided a framework for understanding how the fetal/neonatal environment supports the development of B-1 cells in mice, and, as recently demonstrated, also in humans (27).