Non-canonical B cells: characteristics of uncharacteristic B cells

Abstract

B lymphocytes were originally described as a cell type uniquely capable of secreting antibodies. The importance of T-cell help in antibody production was revealed soon after. Following these seminal findings, investigators made great strides in delineating steps in the conventional pathway B cells follow to produce high affinity antibodies. These studies revealed generalized, or canonical, features of B cells that include their developmental origin and paths to maturation, activation, and differentiation into antibody-producing and memory cells. However, along the way, examples of non-conventional B cell populations with unique origins, age-dependent development, tissue localization, and effector functions have been revealed. In this brief review, features of B-1a, B-1b, marginal zone, regulatory, killer, natural killer-like, age-associated, and atypical B cells are discussed. Emerging work on these non-canonical B cells and functions, along with the study of their significance for human health and disease, represents an exciting frontier in B cell biology.

Discovery and characteristics of conventional B cells

The discovery that B cells and T cells belonged to distinct lymphocyte lineages in chickens was reported in 1965 by Cooper et al.(1). Ten years later, work by this group performed in mice demonstrated that B cells originated from fetal liver and bone marrow (1). Over the next several decades, the foundational concepts of B cell biology grew exponentially as investigators detailed B cell surface phenotype, the processes of development and selection, antigen (Ag) receptor signaling, and T cell co-stimulation of B cell activation, germinal center formation, and antibody (Ab) production (2, 3). The majority of these studies centered on understanding aspects of conventional B cell biology. Conventional B cells are considered to be the predominant recirculating B cell population found within secondary lymphoid follicles that is continuously replenished from adult bone marrow progenitor cells. Conventional B cell responses are largely regarded to involve mature follicular B cells which participate in T cell-dependent germinal center reactions and yield high affinity Ab as well as memory cells which can be reactivated for rapid Ab production and further increases in affinity. However, over time, B cell populations which exhibit pathways of development, trafficking and tissue localization, activation requirements, and effector functions that are distinct from follicular B cells, have been identified. Examples of such non-canonical B cell populations are highlighted throughout this brief review.

B-1 cells: The original non-canonical B cell population

Characteristics of mouse B-1 cells

B-1 cells, an innate population of B cells with distinct characteristics from conventional B cells, were first proposed to represent a distinct lineage in the early 1990’s (4, 5). The B-1 lineage was defined based on differences in the capacity of fetal and adult progenitor populations in mice to reconstitute CD5-expressing B-1a cells, CD5-negative B-1b cells, and conventional (B-2) cells. In contrast to mouse B-2 cells, which readily develop from adult bone marrow progenitors and localize to secondary lymphoid tissues, B-1 cells develop in large part from fetal/early life-derived progenitors and are enriched in serosal cavities. Over the past 30 years, an intriguing and significant body work has provided evidence for the unique developmental and functional attributes of B-1 cells. However, the matter of whether B-1 cells derive from a separate lineage has remained controversial, despite the identification of progenitor cells and distinct developmental waves which give rise to B-1 versus B-2 cells during development (5-12). The identification of the Lin28b/let-7 signaling axis as a major pathway controlling the differences between fetal and adult hematopoiesis, with Lin28b promoting B-1 B cell development, in part through promoting positive selection and let-7 promoting B-2 development and negative selection (13-15), has provided new perspective on this long-held debate. The existence of a development switch influencing positive versus negative selection on self-Ags is consistent with the finding that B-1a cells can bypass the pre-BCR checkpoint (16) and often display autoreactive Ag receptors (17). Regarding the window of B-1 development, a recent time-stamping study performed in unperturbed mice (ie., without transplantation) indicated B-1a cells, and to a lesser extent, B-1b and marginal zone (MZ) B cells, largely arise during the early postnatal period and seed a self-renewing compartment (18) (Fig. 1). These cells, referred to as “early life origin” B cells make a significant contribution to the adult B cell compartment. During a 3 to 18 month chase period, time-stamped peritoneal B-1a cells showed no loss, B-1b cells showed 50% loss, and B-2 cells exhibited a complete loss (18). Time-stamped splenic B-1a, MZ, and follicular B cells, however, failed to persist. The results of this impactful and elegant study raise the possibility that peritoneal B-1a cells, and a fraction of B-1b cells, may in fact be representative of Ag-experienced memory cells shaped by early life exposure. This is supported by previous work demonstrating the influence of Ag, whether it be self or foreign, on the B-1a repertoire (17, 19, 20), as well as the ability of these cells to rapidly differentiate into Ab secreting cells (ASC) in response to toll-like receptor (TLR) agonists (21-24)— a characteristic of conventional memory cells (22).

Figure 1. Age-specific alterations in B cell populations.

B cells that originate in early life, including innate B-1a, B-1b, and marginal zone (MZ) B cell populations, are subjected to positive selection events during early development. Peak seeding of these fetal/neonatal progenitor-derived B cells occurs shortly after birth (blue shaded area). A developmental switch thereafter favors development and accumulation of follicular, or conventional B cells. MZ and B-1b cells continue to derive from adult progenitors, whereas B-1a cell development is much more limited. Fetal/neonatal innate B cells persist for an extended time due to a process of self-renewal, which may involve self-Ag-driven stimulation. Over time, selective pressures may alter the BCR repertoire of these innate cells. This may result in the loss of protective B cell clones with aging, as has been reported for B-1a cells. While MZ B cell numbers decline in aging, B-1b cells increase (green shading). During aging, a population of CD11c⁺T-bet⁺ age-associated (ABC) cells accumulates. These cells are also found at a higher frequency in states of autoimmunity and chronic viral infections. ABC cells share features of atypical CD27⁺CD11c⁺T-bet⁺ memory B cells that originate in response to infection and vaccination. The generation, persistence, and expansion of these populations may be affected by many other factors, including genetics, sex, diet, and other environmental exposures.

In addition to their unique development, anatomical localization, and capacity for self-renewal, CD5⁺ B-1a and CD5⁻ B-1b cells exhibit phenotypic and functional characteristics that are distinct from B-2 cells. The repertoire of B-1a cells is restricted, and consistent with their early development, non-templated (N)-nucleotide additions are infrequent in peritoneal B-1a cells (25-27), although N-nucleotides are increased in B-1a cells from aging mice, concomitant with a loss of protective germline-encoded Ab (27-29). The peritoneal B-1b repertoire is broader and has little overlap with B-1a cells, MZ B cells, or B-2 cells, and frequently contains N additions (25-27), providing further distinction between B-1a and B-1b cells. Along with CD11b expression, peritoneal and pleural B-1 cells typically express an IgM^hiIgD^loB220^lo/+ phenotype, with B-1a cells additionally expressing CD5 and often CD43 (Table I). Class-switched (predominantly IgA⁺) B-1 cells can also be identified.

Table I.

Phenotype and Functions of Non-canonical B cells

	Mouse	Human/Non-human primate	Primary Function	Refs.
B-1a	CD19hiIgMhiIgDloB220loCD11b+ CD43+CD23−CD21/35lo CD80+CD5+ (peritoneal)	CD20+CD27+CD43+CD70−CD38lo/int(blood) CD19hiIgMhiCD11b+CD80+/− CD21/35lo/+CD27+/−CD5+ (peritoneal)*	Ab secretion (natural and induced) Immunoregulation (IL-10)	(31, 36, 41, 46, 89, 91)
IRA	Derived from B-1a (see above)		Secrete GM-CSF, IL-3, Abs	(70, 71, 74)
B-1b	CD19hiIgMhiIgDloB220+/loCD11b+CD43+/−CD21/35lo/+CD5− (peritoneal)	CD19hiIgMhiCD11b+CD80+/− CD21/35lo/+CD27+/−CD5− (peritoneal) *	Ab secretion to TI Ags Some natural Ab secretion	(31)
MZ B	IgMhiIgDloB220+CD21/35hiCD1dhiCD23− (spleen)	IgMhiIgDloCD21hiCD1c+CD23lo CD27+	Ab secretion to TI Ags Shuttle Ags to follicle Immunity to mucosal Ag (hu)	(94, 96)
Breg	CD19+CD1dhiCD5+(B10) CD19+CD21/35hiCD23hiCD24hi CD1dhi	CD24hiCD27+(B10) CD19+CD24hiCD38hi PD-1hiCD5hiCD24+/ CD27hi/+CD38lo	IL-10, IL-35 secretion	(68, 105, 108-110, 114-117)
		CD19+IgM+CD38+CD1d+CD147+IgD−CD27−	Granzyme B+
	CD25hiCD69hi IgA+CD138+PDL1+/−		TGF-β secretion/PD-1 engagement
Killer B	CD19+CD5+FasLhi	FasLhi CD19+TRAIL/Apo-2L ligand+	Killing of Fas+ T cells, possibly tumor cells Killing of TRAIL-R1/2+ cells	(105, 108, 109, 118-120)
NKB	Lin−CD19+NK1.1+IgM+NKp46+#	CD3−CD20+NKG2A+NKp46+CD56+Ig+FasL+GzA+GzB+ NKGD2+	Secretion of IL-12, IL-18 (mouse) Secretion of IL-12, IL-18, TNF-α, IFN-γ, granzyme A and H, FasL+ (primate)	(121-125)
ABC/DN2	CD11c+Tbet+CD21/35−CD23−	CD11c+Tbet+CD21loCD27− IgD−CD27−CD11c+Tbet+CD21lo/−	Cytokine modulation Ag presentation Protective Ab/autoAb	(126, 135) (76, 138, 139)

Abbreviations: IRA, Innate Response Activator B cell; MZ, marginal zone B cell; Breg, Regulatory B cell; NKB, Natural killer-like B cell; ABC, Age Associated B cell or Atypical Memory B cell; DN2, double negative (IgD⁻CD27⁻) B cell, subset 2.

^*It is not clear whether CD5 can be used to distinguish peritoneal B-1a and B-1b cells in primates.

^#Controversial (see Kerdiles et al.)

In contrast to B-1a cells, B-1b cells express low to no levels of CD5 and variable levels of CD43 and B220 (30-32). Thus, peritoneal B-1b cells generally fall into two major subsets: CD43^lo/−B220^hi and CD43⁺B220^lo cells (Table I). This difference may be dictated by Ag receptor specificity (self versus non-self), and/or early life versus late bone marrow origin (Fig. 1). Indeed, the B-1b cells that develop in the absence of CD19 (which ablates B-1a development), predominantly express a CD43^lo/−B220^hi phenotype and although they are fully capable of producing IgM and IgG against foreign polysaccharide Ags (31, 32), CD19^−/− B-1b cells appear to lack reactivity with self-glycans (23). Similarly, B1-8^hi knockin mice expressing a transgenic heavy chain that when paired with lambda light chain confers specificity for the NP hapten largely develop B-1b cells expressing a CD43⁻B220^hi phenotype and lack B-1a cells (33). Anatomic location and activation state may also influence heterogeneity. TLR-activated B-1 cells are known to downregulate select cell surface proteins upon peritoneal exit (24, 33-38) as well as exhibit surface receptor modulation in culture (39, 40). Notably, tissue localization also impacts surface phenotype of naïve B-1 cells. For example, one week post adoptive transfer of sorted naïve B220^lo and B220^hi CD5⁻CD11b⁺ (B-1b) peritoneal B cells from B1-8^hi knockin mice into WT mice, naïve non-dividing B-1b cells adopt distinct surface phenotypes in circulation, serosal cavities, and secondary lymphoid organs, with only cells localized in blood and peritoneal cavity exhibiting a phenotype highly aligned with that at the time of transfer (33). Indeed, a significant portion of these B-1b cells adopt a CD21/35^hi phenotype in the spleen—a phenotype which overlaps with that of MZ B cells. Thus, changes in resting B-1 cell surface phenotype outside of the circulation and serosal cavities, along with activation-induced alterations in cell surface phenotype, complicates identification. Studies with traceable B cell populations are therefore important for delineating B-1 versus B-2 cell responses in vivo and for exploring relationships among B-1b, B-1a, and MZ B cells.

B-1 and conventional B cells are further differentiated based on responsiveness to innate stimuli which may induce Ab production in the absence of cognate T cell help. Their largely unmutated Ab repertoire along with rapid responsiveness to innate stimuli reflects the “innate-like” nature of B-1 cells. B-1 cells spontaneously secrete self and pathogen-reactive Abs that play a critical role in Ag clearance, immunosurveillance, and rapid early protection against infections (36, 41). Furthermore, B-1a cells are highly responsive to TLR agonists and rapidly secrete significant levels of Abs in the absence of known Ag receptor crosslinking, although Ag-specific responses can also be elicited (20, 22, 42-45). Nonetheless, a large fraction of B-1a cells show evidence of ongoing BCR signaling due to either tonic or self-Ag-induced signaling (20, 46, 47). Although TLR activation can induce autoAb production by B-1a cells (48, 49), these signals are critical for increasing protective Ab levels against invading pathogens (36). Along the same lines, TLR agonists can be leveraged to induce B-1a cells to secrete protective Abs against tumor associated carbohydrate Ags (23, 40).

B-1b cells make a significant contribution to the IgM and class-switched Ab response to T cell independent type 2 (TI-2) Ags, which display repeating epitopes capable of eliciting Ab production in the absence of MHC class II-restricted cognate T cell help (32, 33, 50-54). B-1b cells also produce Abs to other pathogen-derived TI Ags (45, 55-57) as well as to carbohydrate epitopes involved in tumor surveillance (58) and transplant rejection (34). Their ability to produce Ab in response to TI-2 Ags in the absence of cognate T cell help or pathogen-associated molecular pattern signals may in part be explained by their high levels of surface IgM and pre-primed state (33). Select TLR agonists also stimulate Ab production by B-1b cells (22) as well as significantly augment B-1b cell Ab production to TI-2 Ags (59). MZ B cells also contribute to these Ab responses, whereas the role of follicular B-2 cells appears to be more limited (33, 51, 52, 59). Following foreign Ag activation, naïve B-1b cells form ASC as well as long-lived quiescent memory cells that express common memory markers (ie., CD80, PDL2, CD73) and are capable of rapidly differentiating into ASC upon re-exposure to Ag (33, 52, 56).

Cytokine-producing regulatory B-1 cells and Innate Response Activator (IRA) B cells

In addition to producing Abs, B-1 cells spontaneously secrete IL-10, and thus function as regulatory cells to impact the outcome of diverse disease states (60-64). B-1a-produced IL-10 suppresses pathogenic T cell responses in models of colitis (61, 63), contributes to resolution of contact hypersensitivity (65), and dampens insulin resistance (66). Additional work suggests a role for B-1a cell-produced IL-10 in the therapeutic effects of amyloid fibril treatment in a mouse model of EAE(67). TLR activation further augments IL-10 secretion by B-1a, and to a lesser extent, B-1b cells, which may function to temper their expansion and differentiation in an autoregulatory fashion in the absence of addition signals (62, 63, 68).

Recently, a protective IL-27-producing B-1a cell population (i27-Breg) was demonstrated to traffic to the central nervous system and lymph nodes under conditions of neuroinflammation (69). A distinct B-1a-derived population referred to as innate response activator (IRA) B cells, has also been described (70, 71). TLR signals activate IRA B cells to produce granulocyte/macrophage colony-stimulating factor (GM-CSF) which functions in an autocrine fashion to stimulate IgM secretion by B-1a cells. In response to pneumococcal lung infection, GM-CSF-producing pleural B-1a cells (IRA) rapidly localize to lung parenchyma and secrete polyreactive IgM that provides early protection against infection (71). Notably, activated B-1 cells migrate to a variety of tissues where their local secretion of Ab can impact disease pathogenesis (40, 44, 71-73). In the cecal ligation puncture (CLP) model, GM-CSF-producing IRA B cells localize in the spleen and elicit protection against sepsis likely via a similar mechanism (70). Nonetheless, high levels of IL-3 produced by IRA B cells can actually contribute to unchecked inflammation and increased morbidity in this model (74). Subsequent work demonstrating B-1a cell-produced IL-10 contributes to the anti-inflammatory response and overall protection in the CLP model (75) highlights the need for additional work in order to decipher the precise signals that drive B-1a cells to produce distinct cytokines.

B-1-like cells in humans and non-human primates

Over the years, the terms “B1” and “B2” have been applied to human B cells. However, it should be noted that the B1/B2 lineage connotation is not adequately met for human B cells. Moreover, “B2” is sometimes used to describe follicular and/or naïve B cells in humans. A detailed review of these and other human B cell subsets can be found in Sanz et al.(76). Parallels have also been drawn between CD5⁺ B cells in humans and B-1a cells in mice, with some investigators referring to these cells as “B-1 cells” (77). However, CD5 is not a reliable marker to exclusively identify B-1a cells, as recently BCR-activated, anergic, and regulatory populations express CD5 (50, 78-81). Analysis of peritoneal B cells has been limited in humans as B cell yields are generally low (~1x10⁵; (82)), blood contamination is problematic, and obtaining peritoneal fluid from healthy individuals is difficult. However, as B-1 cells recirculate in mice, it is not unreasonable to suspect that a circulating B-1-like counterpart exists in human. Indeed, in 2011, Rothstein and colleagues described a CD20⁺CD27⁺CD43⁺CD70⁻ B cell population in human blood that harbored B-1-like cells expressing BCR self-reactivities and spontaneously secreted IgM (46). Although the report of these human B-1a-like cells was met with controversy (83-88), follow-up studies revealed unique transcriptional profiles and V_H usage for these cells using additional surface features (CD38^lo/int) to distinguish the population from CD38^hi plasmablasts (89). Thus far, studies of human patients with traceable mutations in adult HSC and xenograft mouse models show human blood B1 cells defined as CD19⁺CD20⁺CD27⁺CD43⁺CD38^lo/int readily originate from adult HSCs (89, 90), unlike mouse peritoneal B-1a cells. Whether an early life origin B-1a-like cell persists in human serosal cavities remains unknown.

Given the paucity of data on peritoneal B cells in humans, several years ago we investigated the functional and phenotypic attributes of peritoneal B cells in non-human primates (NHP) using controlled peritoneal washes that were free of blood contamination (91). In NHP, ~25-40% of peritoneal and omental B cells express CD11b, and these CD11b⁺ B cells express higher FSC, IgM, and CD19, and lower levels of CD21 relative to CD11b⁻ B cells. A fraction of CD11b⁺, but not CD11b⁻, peritoneal B cells express CD80 and CD5, constitutively active Stat3, and specificity for phosphorylcholine (91). These characteristics are highly aligned with those of murine B-1a cells (79, 92) (Table I). We also found Ag-specific B cell responses to TNP-Ficoll in NHP involved CD11b+ B cells (91) and showed strong similarities to TNP-specific B-1b cell responses in mice (33, 50, 51, 54). Further, we identified a TNP-specific CD11b⁺ memory (CD80⁺CD27⁺Ki67⁻)) B cell population in immune, but not naïve NHP, suggesting a B-1b-like TI-2 memory population develops in NHP, as occurs in mice (33).

In summary, data generated in humans and NHP support the existence of B-1a- and B-1b-like cell subsets. However, their relationship to mouse B-1 cell subsets requires additional investigation. The identification of Lin28b expression in human fetal HSC as well as Bhlhe41 transcriptional regulation in supporting the development and maintenance of B-1a B cells in mice (93) highlight further opportunities to explore putative B-1a and B-1b cell populations in humans.

Marginal zone (MZ) B cells

MZ B cells, like B-1a and B-1b cells, are considered innate B cells. They are predominantly localized in the MZ of the spleen, which demarcates the area between the white and red pulp. This provides MZ B cells as well as specialized macrophages access to blood borne Ags (94). As mentioned above, studies performed in mice indicate MZ B cells arise from both early life and adult bone marrow progenitors (18) (Fig. 1) and have a distinct repertoire relative to other subsets (25), with evidence of positive selection (95). Similar to B-1 cells, MZ B cells produce Ab responses to TI-1 and TI-2 Ags (33, 52, 59) and rapidly produce Ab in response to TLR agonists (22, 96).

Mouse and human MZ B cells share a common IgM^hiIgD^loCD21^hiCD23^lo CD1d^hi(mouse) CD1c⁺(human) phenotype (96, 97) (Table I). Human MZ B cells also express CD27 and show evidence of somatic hypermutation, which has led to the interpretation that these cells are memory cells (98). Although mouse MZ B cells generally lack mutations, it should be noted that they are capable of participating in T cell dependent germinal center (GC) reactions, accrue mutations, and form IgM memory to both TD and TI-2 Ags (33, 99). While mouse MZ B cells migrate into follicles carrying complement-tagged Ag as cargo (100), they are reported to have limited ability to leave the spleen (96, 97). However, in humans, MZ B cells are present in the spleen, subcapsular sinus of lymph nodes, mucosal tissue, and blood (96). A recent study suggested recirculating human MZ B cells acquire mutations during their exposures within gut associated lymphoid tissue but remain distinct from classical memory cells (101). This supports the existence of a spleen-gut axis whereby circulating MZ B cells and/or IgM⁺ memory B cells are “trained” on mucosal-derived Ags such that protection against invading mucosal organisms can be preemptively established in the periphery (102). Although this process has not been described for mouse MZ B cells, a recent report supports that some gut IgA⁺ and splenic IgM⁺ clones in mice are in fact, related (103). The heterogeneity within the human MZ B cell population has further complicated species comparisons. Interestingly, a recent report using multiparameter unsupervised analysis of human MZ B cells identified two distinct populations (MZB-1 and MZB-2), of which MZB-2 is more aligned with mouse MZ B cells (104). Further work is required to the determine the mechanisms by which these newly defined MZ B cell subpopulations contribute to host defense, malignancy, and autoimmunity.

Regulatory B cells

B cells are best known for their ability to secrete Abs. However, numerous reports of B cell populations capable of suppressing immune responses through either cytokine production or expression of immunoinhibitory ligands reveal there are critical alternative functional roles for B cells. IL-10-producing B cells were first described for CD5⁺ splenic and peritoneal B-1a cells populations (60). More recently, IL-10-producing B cells that have distinct phenotypic markers among mouse, as well as human B cell populations, have been described (68) (Table I). IL-10-producing B cells are important for maintaining tolerance and suppressing excessive activation to maintain homeostasis, but their suppressive effects can interfere with optimal immune responses, especially in the context of anti-tumor responses (68, 105). In mice, IL-10-producing B cells significantly impact disease progression and outcomes in models of autoimmunity, infection, cancer, and contact hypersensitivity/allergy, as well as diminish general humoral and cellular immune responses in healthy mice (68, 105, 106). The mechanism of suppression is often through inhibition of T cell responses and promotion of other regulatory populations (such as Treg), although effects on other cell types also occur. The lack of a singular phenotype or transcriptional signature to define IL-10-producing B cells makes study of these cells in healthy and disease states difficult, as the specific functional effects of IL-10 production (versus other factors) by defined populations must be determined. While this can be accomplished in mouse models, it may be challenging to interpret the functional effects of diverse IL-10-producing B cell populations in humans.

Two additional cytokines produced by regulatory B cells include IL-35 and TGF-beta. IL-35 is a member of the IL-12 family of cytokines (105). IL-35 is a heterodimer comprised of the p35 alpha (also used by IL-12) and EBi3 beta subunits (used by IL-27). Like IL-10, IL-35 has immunosuppressive properties and has been implicated in the maintenance of immune tolerance and suppression of T cell responses to tumors (105). IL-35-producing regulatory B cells were originally described as a plasma cell-secreting population that facilitated recovery from EAE but suppressed protection against Salmonella typhimurium infection (107). Subsequent work has demonstrated additional populations of mouse B cells that exert regulation via IL-35 production (105). Although less well studied, regulatory B cells are also found to produce TGF-beta, which promotes Treg development, invokes anergy in T cell populations (108, 109), and suppresses anti-tumor responses (105, 110). While suppressive Breg subsets have garnered significant interest, it is important to point out the importance of other B cell-produced cytokines. B cell subsets can be conditioned to produce IFNγ, IL-12 and TNFα (Be1); IL-2, IL-4, IL-6 (Be2); as well as lymphotoxin and IL-17A, all of which may impact host responses (111, 112).

In addition to producing cytokines, B cells produce other types of soluble immunosuppressive factors, including indoleamine 2,3 deoxygenase and adenosine (113). Human B cells have also been shown to secrete granzyme B which inhibits T cell function in the tumor microenvironment (114). Although granzyme B-secreting B cells elicit perforin-independent cytotoxicity when cultured with tumor cells in vitro (115), whether this occurs in vivo remains to be determined. Granzyme B expression has not been detected in mouse B cells. B cells can also potently regulate immune responses via expression of surface immunomodulators. A key example of this are PDL1^hi Breg which suppress T_FH, dampen autoimmunity, and limit CD8⁺ anti-tumor responses via PD-1 interactions (105, 116, 117). PD-L1^hi B cells have been reported to exist within the plasma cell pool, as well as within CD5⁺ populations. Thus, regulatory B cells secrete distinct soluble factors and utilize cell-cell contact to orchestrate control over immune responses (Table I).

Killer B cells and Natural Killer B cells

A specialized function of some Breg is the capacity to kill other cells (108). Fas ligand (FasL)^hi B cells induce apoptosis in activated T cells and thereby limit pathogenic responses in autoimmune diseases, but dampen protection in the context of infection (105, 109, 118). Importantly, FasL^hi B cells can also elicit killing of other B cells, epithelial cells, and tumor cells (105, 108, 119). Interestingly, FasL^hi B cells are predominantly found in the CD5⁺ B cell population, some of which have overlap with IL-10-producing Breg (108, 118). (Table I). A similar killing mechanism can be achieved by B cells expressing other membrane-bound TNF superfamily members, such as TRAIL ligand (120). However, as is the case with granzyme-expressing Bregs cells discussed above, the extent to which these membrane-bound TNF surface receptors enable B cells to kill cells in vivo remains unclear. Further, the impact co-expression of other regulatory functions has on these killing mechanisms has yet to be investigated.

In 2016, a population expressing markers of both B cells and NK cells with a restricted BCR repertoire, termed “natural-killer-like B cells” was reported to produce IL-18 and IL-12 early in infection (121). A follow-up study by another group provided data to argue against the existence of such a population, at least in mouse (122, 123). Nonetheless, two independent studies described a similar natural killer B cell (NKB) population in rhesus macaques, cynomolgus macaques, and humans (124, 125) (Table I). In addition to secretion of IL-18, TNF-alpha, and interferon gamma, primate NKB cells are reported to express high levels of granzyme A, H, and FasL, as well as possess cytolytic activity. Simian immunodeficiency virus infection is observed to drive expansion and activation of these cells in the gut, which may function to enhance inflammation. There are numerous questions that remain to be answered about this newly-described B cell population pertaining to its origin, regulation, effector functions, and relatedness to bona fide NK cells.

Age-associated B cells, Atypical B cells, and Double Negative B cells

Age-associated B cells (ABC) have been identified as a population which accumulates in aging mice and humans, with increased predominance in cases of autoimmunity (126) (Fig. 1, Table I). Age-associated B cells were originally described independently by two groups as splenic mouse B cells expressing a B220⁺CD19⁺CD21⁻CD23⁻CD43⁻(127) or a B220⁺CD19⁺CD21⁻CD11c⁺CD11b⁺CD80⁺Fas⁺ (128) phenotype. In the latter study, ABC were found to express plasmablast-associated markers (ie., CD138, Blimp1) and the transcriptional regulator T-bet (128). Many subsequent studies have relied on T-bet positivity along with CD11c expression to delineate ABC. There is some overlap in the originally described populations, but there is considerable heterogeneity in both the CD21⁻CD23⁻ and CD11b⁺CD11c⁺ populations. In both studies, a high level of responsiveness to TLR7 and 9 agonists was noted and TLR7 was reported to drive ABC formation in vivo (128).

The origins of ABC are not certain, but there are features which overlap with B cells involved in TI responses. Splenic CD11c⁺CD11b⁺ plasmablasts and T-bet⁺CD11b⁺CD11b⁺CD80⁺IgM⁺ memory B cells are generated in response to TI Ags encountered during Erhlichia muris infection (129). Similar to CD11b⁺CD21/35^−/−CD80⁺ B-1b IgM⁺ memory cells (33), these cells are functional in that they are capable of dividing, class-switching, and differentiating to ASC upon Ag reactivation (130). In the peritoneal cavity, cells responding to this infection display a similar phenotype (T-bet⁺CD11b⁺), but lack CD11c (131). Notably, CD43⁻ peritoneal B-1b and splenic CD11b⁺CD21/35⁻CD23⁻ cells accumulate with aging (30, 132, 133). Consistent with this, Ag-specific splenic CD11b⁺ B cell responses to type 3 pneumococcal polysaccharide (previously shown to be dominated by B-1b cells ref.(32)) are significantly increased in aged male mice (133). Although transcriptomic analysis revealed notable differences between splenic ABC and peritoneal B-1a cells in the Rubtsov et al. study (128), the extent to which the described ABC populations in aged mice derive from B-1b, splenic B-1a, MZ, and/or follicular B cell populations is not entirely clear.

Given the expression of memory markers, hyperresponsiveness to TLR agonists, and evidence of Ag-driven expansion, CD11c⁺T-bet⁺ ABC have been described as Ag-experienced memory B cells (134, 135). Consistent with the aforementioned studies with bacterial Ags, there is significant overlap between the ABC phenotype and that of a subset of memory B cells which have encountered Ag in the context of virus/parasite infection or vaccination, regardless of host age (135). These cells are referred to as “atypical B cells” or “atypical memory B cells” (also ABC). Whereas classical memory B cells in humans express a CD27^hiCD21⁺ phenotype, atypical B cells bearing a CD27^loCD21^loCD11c⁺ phenotype are transiently increased with acute infections or sustained at increased levels in chronic infections and autoimmunity (126, 135) (Table I). In particular, atypical class-switched IgD⁻CD27⁻ cells in humans are referred to as “double negative” (DN) B cells, with those having an ABC-like Tbet⁺CD11c⁺ phenotype (DN2) being the predominant population expanded in SLE (76, 136). Notably, DN2 are also expanded in patients suffering from severe COVID-19 infection(137). In contrast to CXCR5⁺Tbet⁻CD11c⁻ DN1 B cells, CXCR5⁻DN2 cells appear to be a precursor population to extrafollicular ASC poised for rapid autoreactive and/or disease-specific Ab secretion (76, 136). Reviews on emerging DN subsets can be found elsewhere (76, 138, 139).

Given the high degree of overlap between age-associated B and atypical memory B cell populations, they are considered by many to be a related, if not the same, population (135, 138, 140). However, the pathways giving rise to these cells appear diverse. Several studies suggest ABC are dependent on T cell help and GC reactions (126, 135), whereas others suggest they derive from GC-independent extrafollicular responses (135, 141). Indeed, the ABC-like DN2 population described in patients with SLE and severe SARS-CoV2 infection supports an extrafollicular origin (76, 136, 137). Interestingly, a recent study shows the IgD⁺CD21⁻CD23⁻ subpopulation of naïve splenic B cells in mice may serve as a progenitor population for the ABC-related ASC and memory populations that are induced in response to influenza infection via a T cell-independent, but TLR-dependent manner (142). Given the findings from this and other studies, it is likely that the ABC found with natural aging, infection/vaccination, and autoimmunity are generated through different mechanisms and may represent different stages of cellular activation and/or differentiation. This requires further investigation.

Along with the origins of ABC, our understanding of the functional roles ABC play in the immune system is far from complete. There is evidence of protective roles for ABC in infections and pathogenic roles in autoimmunity and chronic inflammation; however, the precise mechanisms by which ABC impact disease are not fully elucidated. Indeed, ABC have been described as: 1) exhausted cells, as they are refractive to anti-IgM stimulation, 2) cells at an intermediate or pre-primed state of activation, poised for Ab production against foreign or self-Ags, 3) cells with heightened Ag-presenting function, or 4) cells that promote inflammatory responses (ie., Th1/Th17 responses and inflammaging via IFN-γ or TNF-α production) or dampen responses (via IL-10 production) (126, 135). Strategies that enable depletion (143) and/or genetic deficiency of ABC, tracking of ABC development, activation, differentiation, and function at the Ag-specific B cell level, and integrative exploration of high dimensional datasets from human and mouse ABC populations in different contexts will undoubtedly produce answers regarding the origins, relatedness, function, and regulation of ABC cells.

Conclusions

Accumulating work highlights non-conventional B cells are key players in distinct facets of homeostasis, host defense, autoimmunity, cancer, allergy, and aging. For some non-canonical B cells, namely innate-like B-1a, B-1b, and MZ B cells, evidence of distinct developmental requirements, repertoire differences, responsiveness to innate versus Ag-receptor signals, and unique tissue localization clearly set them apart from follicular B cells. Whereas for other populations such as regulatory B cells, killer B cells, and ABCs, distinction is in some cases less clear, with functional characteristics and phenotypes being the primary features of divergence. Many questions remain about the signals that give rise to these non-conventional B cell populations, their functional significance in the immune system, and the extent to which murine non-conventional B cell populations model human populations. High dimensional multiparameter analyses which leverage single cell and spatial profiling technologies are beginning to yield unbiased perspectives on B cell development, BCR repertoire, phenotype, and stages of differentiation, in healthy and diseased states. These new insights will undoubtedly answer long-standing questions about these subpopulations as well as support generation of new hypotheses that may ultimately give rise to future therapies aimed at modulating effects of these and other non-canonical B cell populations.