Insights into B-cell ontogeny inferred from human immunology

Due to ontogenetic changes in B-cell developmental lineages, the mature B-cell compartment constitutes by functionally different B-cell subsets that emerged from prenatal, early postnatal or adult precursors. While negative selection processes operate primarily within the framework of B-cell tolerance checkpoints during B-cell development, further differentiation into distinct B-cell subsets is additionally induced by positive selection. In addition to endogenous antigens, contact with microbial antigens is also involved in this selection process, with intestinal commensals having a significant influence on the development of a large layer within the B-cell compartment. The decisive threshold that triggers negative selection seems to be relaxed during fetal B-cell development, thereby allowing recruitment of polyreactive and also autoreactive B-cell clones into the mature naïve B-cell compartment. Almost all of the concepts on B-cell ontogeny are based on observations in laboratory mice that not only differ from humans in their developmental timeline but also in their composition of commensal microorganisms or rather a lack of exposure to these. In this review, we summarize conceptual findings on B-cell ontogeny and particularly describe key insights into the developing human B-cell compartment and immunoglobulin repertoire formation.


Introduction
The establishment of a B-cell compartment that effectively counteracts diverse challenges throughout lifetime requires the integration of partially divergent mechanisms into a complex system. First, in the formation of a primary immunoglobulin B-cell compartment of adults [3]. Almost all the concepts on B-cell ontogeny are based on observations in laboratory mice that not only differ from humans in their developmental timeline but also in their composition of commensal microorganisms or rather a lack of exposure to these [4]. Studies on the development and function of the human immune system have been facilitated by experimental refinements including systems-level high-throughput "omics" technology as well as the investigation of patients with genetically defined inborn errors of immunity [5][6][7][8]. In this review, we will summarize conceptual findings on B-cell ontogeny and describe key insights into the developing human B-cell compartment and Ig repertoire formation.

Development and selection of B-cell lineages in mice
The emergence of distinct B-cell lineages, which arise at different points during ontogeny, is a well-studied phenomenon in mice [9,10]. The consecutive appearance of these lineages is less construed as separated maturation steps of the immune system but rather represents intertwined developmental layers of functionally distinct B-cell subsets that build up the complex B-cell compartment in adults [3,10]. Indeed, recent findings from lineagetracing studies documented that a significant portion of B cells in adult mice is formed very early in life [11]. This suggests that in mice and potentially also in humans a substantial part of the B-cell compartment may be programmed in early life and microbial exposure at that period could instruct the trajectory of the developing B-cell system [12][13][14].
The first blood cells in mice emerge at embryonic day 7 (E7) in the yolk sac which are followed by erythro-myeloid progenitor cells at about E8.5 [15]. The first lymphoid-primed progenitor cells arise in a third wave of hematopoiesis in both the extraembryonic yolk sac and intraembryonic tissue at around E9 [9]. Concomitantly, the precursors of adult-repopulating hematopoietic stem cells (HSCs) can be detected in the intra-embryonic aorto gonad mesonephros region at E10 and long-term HSCs emerge at E11 [9]. Although hematopoietic progenitors from all these waves seed the fetal liver and may give rise to "fetal" innate-like B cells, long-term HSC-derived hematopoiesis that continually regenerates "adult" B cells is established in the bone marrow (BM) at E18 [9, 10,15] (Fig. 1).

B-1 cells: an early-life innate-like B-cell subset
In mice, B cells are generally divided into three main subsets deriving from separate lineages that emerge at different times during ontogeny and differentiate along distinct pathways [16]. B-1 cells are innate-like B cells that preferentially differentiate from embryonic pre-HSCs or potentially HSC-independent yolk sac progenitors during prenatal and particularly early postnatal life and express semi-invariant B-cell receptors (BCR) in germline configuration enriched in poly-and autoreactivity [17,18] (Fig. 1). They populate the peritoneal and pleural cavity, are excluded from germinal center (GC) reactions, and differentiate into plasma cells without T cell help. Most B-1 cells express the surface marker CD5 (B-1a cells) and are regarded to constitute the major source of naturally secreted IgM that provides immediate protection against invading pathogens and facilitates clearance of cellular debris and tissue homeostasis [19]. B-1a cells may develop prior to BM HSCs; however, it is not entirely clarified, whether B-1a cell development in un-manipulated mice also emerges independently from HSCs or whether it also arises from HSC [20][21][22]. The output of B-1a cells declines significantly after neonatal life and the adult B-1a compartment is primarily sustained by self-replenishment [18]. The immunoglobulin repertoire of B-1a cells continuously evolves by convergent selection of certain V(D)J-rearrangements as well as somatic hypermutation [23]. The notion that this pattern of diversification operates similarly in germ-free as well as specific pathogen-free mice suggests that selection is driven by endogenous and not microbial antigens [23].

Follicular and marginal zone B cells
In contrast, follicular (FO) B cells emerge later during ontogeny and express a highly diverse BCR repertoire that significantly differs from B-1a cells [23]. Unlike B-1a cells, naïve FO B cells rapidly die unless being transferred into the memory B-cell compartment and are constantly replenished from adult bone marrow (BM) HSPCs [9]. Follicular B2 cells recirculate between blood, spleen and lymph nodes, give rise to germinal center (GC) B cells in a T-dependent manner, undergo class switching and somatic hypermutation, and differentiate into memory B cells and plasma cells. Marginal zone (MZ) B cells, which together with FO B cells comprise B2 cells, reside within the splenic white pulp adjacent to the marginal sinus [24]. The BCR as well as Notch-2-mediated signals seem to be the decisive factors that instruct transitional B cells to further differentiate along the FO versus MZ B-cell developmental pathway [25]. MZ B cells resemble B-1a cells in their innate-like and T-independent rapid responses to bacterial pathogens and some of the MZ B-cell compartment seems to be of fetal and early postnatal life origin [25].

A developmental program controls positive B-cell selection
The formation of the preimmune BCR in developing B cells is shaped by constraints of negative selection at distinct tolerance checkpoints. However, several lines of evidence also support a role of active BCR signaling and antigen-dependent positive selection of distinct BCR specificities into mature B-cell subsets [26][27][28]. Although B-1a cell differentiation relies on strong BCR signaling and therefore favors selection of autoreactive BCRs, FO and especially MZ B-cell developments are disfavored by very strong BCR signals [25]. Indeed, BCRs reacting poorly or not at all with endogenous antigens support differentiation along the MZ B-cell trajectory and those reacting fairly well but not as strong seem to induce FO B-cell differentiation [29]. Thus, positive selection shapes the mature B-cell compartment and BCR signaling from autoreactive BCRs has an instructive role in lineage commitment of B-1a cells emerging during fetal and early neonatal life. These observations led to the hypothesis that active recruitment of autoreactive/polyreactive B-cell clones into the mature naïve B-cell compartment is a general feature of the fetal/neonatal B-cell developmental program, whereas the constraints by negative selection take over during adult B-cell development [28] (Fig. 1). In mice, this developmental program seems to be actively induced and under molecular control of the RNA-binding protein Lin28b [30,31]. It is suggested that Lin28b expression during early life B-cell development favors relaxed tolerance thresholds, thereby permitting the incorporation of distinct BCR specificities into the mature naïve B-cell pool [11]. Despite its potential risk to license polyreactive/autoreactive Bcell output during ontogeny, this molecular program, however, is probably a meaningful selection rather than a simple by-product of immaturity. First, the "pattern recognition" signature of polyreactive B cells convey an immediate layer of broad antimicrobial reactivity during neonatal life. Moreover, recent observations from studies in mice suggested that early life microbial exposures may induce incorporation of unique clonotypes into the memory B-cell repertoire that are not stimulated by immunization of adult mice and derive from the B-1 cell pool [14].
In summary, the B-cell compartment in mice is composed of diverse but intertwined layers of B cells that emerge from different developmental waves during ontogeny. A transcriptionally controlled change in BCR signaling thresholds during the transition from fetal to neonatal life may allow the incorporation of otherwise counterselected BCR specificities into the mature B-cell pool.

Fetal B-cell development in humans
The first appearance of the B-cell lineage in humans can be mapped to the fetal liver by appearance of B-cell precursors from seven postconceptional weeks (PCW) and the BM becomes enriched in B cells at mid-gestation [32,33]. Using single-cell RNA sequencing, a continuum of B-cell differentiation states could be detected in the human fetal liver with pre-B cells detected between seven and eight PCW and mature B cells only after nine PCW [34]. HSC and progenitor cells were present in yolk sac and FL and absent from nonlymphoid tissues (NLTs). However, in contrast to T-cell progenitors that exclusively appeared in the thymus, lineage-committed B-cell precursors could be detected in almost all prenatal organs (e.g. intestine, skin, thymus, and skin) [34,35]. Hence, B-cell hematopoiesis does not seem to be restricted to FL and BM and several fetal NLT may provide the environment that supports further B-cell differentiation. Whether this B-cell developmental program appearing in fetal NLT is waning later in life and may akin to murine B-1 cells mirror a wave of innate-like and self-replenishing B cells is not elucidated yet. However, the potential of several NLT to instruct specific maturation programs in these developing B cells may render the newborn widely prepared for microbial exposures and when establishing a commensal repertoire at surface barriers.
Several observations derived from the analysis of B cells obtained from human fetus and preterm infants are in line with the conceptual frameworks of ontogenetic changes in the constraints of B-cell selection observed in mice [28]. Although random V(D)J-recombination creates a diverse BCR repertoire in adults, neonatal human B cells rather display a restricted repertoire that is characterized by preferential usage of distinct V-and J-gene segments and lower complementarity-determining region (CDR)3 junctional diversity due to limited incorporation of Nnucleotides [36][37][38][39][40][41][42]. Similar to the transcriptional control of fetal/neonatal life B-cell hematopoiesis in mice [31], the use of a restricted "fetal/neonatal" versus a diverse "adult" BCR repertoire in developing B cells in humans seems to follow an intrinsically determined trajectory during ontogeny, which is not expedited by preterm birth or other environmental influences [31,43,44]. In healthy adults, B-cell tolerance checkpoints are operating that limit further differentiation of autoreactive B-cells clones generated in the BM [45]. A recent investigation elegantly demonstrated that these tolerance checkpoints also exist in human fetal B cells but are relaxed, thereby permitting the accumulation of polyreactive clones into the mature naïve B-cell pool [12]. These polyreactive B cells showed some degree of autoreactivity and importantly were four times more likely to bind commensal bacteria than were their adult counterparts [46]. Thus, altering the threshold of B-cell tolerance checkpoints in early human life favors the accumulation of innate-like B cells with restricted BCR repertoires that also bind to commensal bacteria. Interestingly, the IgG repertoire of former preterm neonates is subject to similar age-related changes without preferential selection of B-cell clones with "adult" repertoire features [44]. Despite extreme individual diversity of the Ig repertoire, shared clonotypes between individuals are already present in neonates and even persist in the adult repertoire [47,48]. These observations may support the idea that recruitment of early-life-origin B cells with distinct BCR specificities into the "adult" memory compartment may also occur in humans. Large-scale BCR analyses in healthy individuals prospectively monitored within birth cohorts hold immense potential to dissect distinct environmental exposers that shape the developing Ig repertoire from childhood to adults [49].

B-cell maturation and differentiation trajectories in humans A putative human B-1 cell subset
Despite fundamental similarities, murine and human B-cell subsets and their developmental trajectories differ, in some cases sig-nificantly. This appears most obvious when addressing the identity of B-1 cells in humans [50]. Neonatal B cells share some characteristic features with murine B-1 cells, but the precise identification of a distinct B-1 cell subset in humans is still controversially debated. CD5 is used in mice to delineate B-1a from B-2 cells but its expression in human B cells is not confined to a distinct early-life B-cell lineage [51]. Delineation of human B-1 cells has been advanced by reverse identification approaches starting from a set of functional criteria. By this, a subset of circulating CD20 + CD27 + CD43 + CD70 − B cells was identified in umbilical cord blood as well as adult peripheral blood that resembled murine B-1a cells [52]. These B cells spontaneously secreted IgM, were efficient stimulators of T cells, and displayed chronic intracellular signaling. The putative B-1 population consisted of both CD5 + and CD5 − subsets, with the latter subset preferentially producing antibodies against polysaccharide antigens. The gene expression program of these putative B-1 cells display partial overlap with that of plasmablasts, which raised the question whether this cell subset might rather resemble plasmablast precursors [53]. Although several other experimental approaches have verified the presence of CD20 + CD27 + CD43 + B cells, identification of this B-cell subset by flow cytometry needs to be executed carefully [54]. Indeed, T-cell/B-cell doublets or negligent staining/gating strategies may easily evoke false-positive signals and initially raised doubts about the existence of this putative B-1 cell subset humans [55]. Nevertheless, sc-RNA sequencing studies of human fetal tissue also revealed a distinct cluster whose transcriptomic pattern with increased expression of CD5, CD27, and SPN (CD43) resembled that of the putative B1 cells [35]. The frequency of these cells across most organs was highest in the early embryonic stages and thereafter declined, except for the thymus where they persisted. The putative B1 cells showed features of tonic signaling, an Ig repertoire distinct from mature naïve B cells with shorter CDR3 junctions as well as spontaneous secretion of IgM antibodies thereby resembling murine B1 cells. Prospectively, executing single cell omic analyses of human tissues will pave the way to dissect the phenotype of putative B-1 cells in humans. In particular, elucidating their tissue distribution, BCR specificities and functional relationship to other immune cells will improve our future understanding of this B-cell subset in the physiologic context as well as in disease settings.

Marginal zone B-cell trajectory
The developmental pathway, tissue distribution, and molecular Ig patterns of MZ B cells substantially differ between mice and humans [25,56]. Although murine MZ B cells are confined to the spleen, its counterpart in humans circulate between lymphoid tissues and seem to differentiate within gut-associated lymphatic tissue (GALT) [56,57]. In contrast to mice, the Ig genes of human MZ B cells are diversified by somatic hypermutation [58][59][60]. The finding that MZ B cells in patients with CD40/CD40L deficiency have Ig mutations suggests they may mature independent of T-cell help [61]. Human MZ B cells are characterized by a CD27 + IgD + IgM + phenotype ("unswitched/nonswitched memory B cells") with high expression of CD1c and CD21 and low CD23 expression and enriched in BCRs that are directed against polysaccharides derived from encapsulated bacteria [56]. In congruence with observations from mouse models suggesting that weak BCR signals promote MZ B-cell selection, human circulating MZ B cells are almost devoid of autoreactive B-cell clones [62]. Investigating the distribution of circulating MZ B cells within patients with mutations that impinge on TLR-signaling pathways suggested that MyD88-/IRAK4-but not UNC93-dependent pathways seem to be essential for MZ B-cell differentiation in humans [63,64]. Interestingly, defects in these pathways also impair a central tolerance checkpoint in human B cells permitting autoreactive clones to differentiate into transitional B cells [65]. Mouse as well as human MZ precursor differentiate into MZ B cells under influence of NOTCH2 signaling pathways. Indeed, Alagille patients with a NOTCH2 haploinsufficiency display a marked reduction of circulating CD27 + IgD + IgM + MZ B cells [66]. However, two distinct populations of MZ B cells have been described in human spleens recently, which are not clonally related and differ in their extent of NOTCH-induced genes as well as tissue distribution suggesting that different maturational pathways further diversify the MZ Bcell compartment [67]. The development of the splenic MZ B-cell compartments is also subject to age-related changes and is not fully established in infants [68].
A systems level approach using high dimensional analysis of blood as well as splenic and intestinal tissues revealed a differentiation pathway of MZ B cells that diverges at the level of transitional type 2 (T2) B cells and is characterized by high expression of IgM and CD45RB [69] (Fig. 2). These observations place IgM hi CD45RB + B cells within a MZ B-cell differentiation trajectory that emerges from T2 and extends to the naïve B-cell population [70][71][72] (Fig. 2). IgM hi CD45RB + T2 B cells display a gut homing phenotype and are selectively recruited into GALT where they get activated and introduce somatic hypermutation within their Ig genes [69,72,73]. Whether activation of T2 cells in the gut is directly succeeded by further differentiation into MZ B cells is not completely understood. The IgM hi CD45RB + naïve B cells in peripheral blood, however, harbor somatic hypermutation in their Igs, albeit at lower frequency as circulating CD27 + IgD + IgM + MZ B cells, and can therefore be regarded as circulating MZ precursor or "preMZ" B cells [74]. MZ B cells in GALT tissues are spatially and clonally segregated from IgA+ memory B cells but harbor clonal relationships with IgM-only CD27 + B cells [69]. The latter give rise to IgM plasma cells that account for a significant proportion of plasma cells in the human gut [75]. Interestingly, MZ B cells are clonally related to gut germinal center (GC) B cells, suggesting that this is the site where MZ B cells diversify the Ig repertoire [69]. Recent investigations into B-cell responses against pneumococcal polysaccharides in vaccinated humans demonstrated a mobilization of MZ and to a lesser extent switched-memory B-cell clones expressing BCRs that have been prediversified within mucosal immune responses against bacterial antigens [76].
Taken together, these observations strengthen the hypothesis of MZ B cells as an independent B-cell lineage in humans that confer protection against microbial polysaccharide antigens and diversify their preimmune Ig repertoire in GALT presumably by interaction with commensal microbes (Fig. 2). There is a strong bidirectional relationship between the immune system and the microbiome during ontogenetic priming in early life which probably also comprises the gut microbiome and the B-cell compartment [5]. Hence, the identification of microbiome signatures associated with improved antibody responses ("resilience") or lacking protection ("risk") and, consequently, the beneficial modification of the microbiome could be a potential strategy to facilitate establishment of protective B-cell memory.

Shaping of the preimmune Ig repertoire by negative and positive selection
After generation of a functional BCR, B cells leave the BM environment as transitional B cells that further differentiate into mature naïve B cells. In addition to the division into the MZ and follicular B-cell differentiation trajectories, which is influenced by BCR specificities, the composition of the preimmune Ig repertoire of naïve B cells is further shaped by distinct tolerance checkpoints (Fig. 2). Using unbiased expression cloning of the antibody repertoire from individual single sorted B cells, it has been shown that autoreactive B cells in humans are normally counterselected at two discrete checkpoints during their development [45]. First, a central checkpoint in the BM before development into transitional B cells removed most autoreactive B cells. Remaining autoreactive B cells are further counterselected at a second tolerance checkpoint in the periphery before they enter the mature naïve B-cell population. Analyses from patients with inborn errors of immunity revealed that the central tolerance checkpoint operates intrinsically in B cells by co-operation of BCR and Toll-like receptor (TLR) 7/9 signaling pathways [65,77,78]. In contrast, the peripheral B-cell tolerance checkpoint relies on B-cell extrinsic mechanism (CD40/CD40L-signaling, MHC-II) and is mediated by regulatory T cells [46,79,80]. The complete mature naïve B-cell pool that also contains the later identified subset of IgM hi CD45RB + pre-MZ B cells was included for the investigations of these B-cell tolerance checkpoints. The interesting question therefore arises as to whether the division into the IgM hi CD45RB + MZ and IgM +/low CD45RB − "follicular" B-cell trajectory acts before and independently of the peripheral B-cell tolerance checkpoint or whether it forms a functional unit with it. Furthermore, it is not known whether additional B-cell extrinsic mechanisms such as environmental/microbial factors may influence the peripheral B-cell tolerance checkpoints in humans. Interestingly, about 20% of mature naïve B cells express BCR that react with autoantigens [45]. This number is even increased in the compartment of CD27 + IgG + switched-memory B cells that arose from naive B cells within GC reactions [81]. In contrast, circulating CD27 + IgD + IgM + MZ B cells are devoid of autoreactive B-cell clones [62]. Furthermore, mature naïve B cells express a wide range of IgM levels, and autoreactivity within the naïve Bcell compartment correlates with downregulation of IgM but not IgD BCRs [27,82,83]. Here, IgD signaling in mature naïve B cells further shapes the composition of the preimmune Ig repertoire [84]. Hence, although autoreactive B-cell clones are counterselected from entering the IgM hi MZ B-cell trajectory, these clones still accumulate to some extent in the remaining naïve B-cell pool in healthy individuals. Encounters with self-antigens by polyreactive BCRs have been suggested to improve the developmental fitness of these B-cell clones, prime them for optimal vaccine responses, and may induce a natural memory B-cell compartment [85]. Differential signaling via IgM or IgD BCRs also seems to determine further differentiation fate decisions in B cells. At least in mice, IgM favored differentiation of naïve B cells into shortlived plasma cells, whereas activation through IgD excluded B cells from rapid extra-follicular plasma cell differentiation [29,86]. It was suggested that this mechanism may exclude autoreactive B cells that have downregulated IgM and are instructed to receive signals via the less responsive IgD from uncontrolled differentiation into antibody-secreting plasma cells [27]. In line with this, signaling through IgD may allow B cells to enter GC reactions where they may lose autoreactivity by means of somatic hypermutation [87]. This mechanism of "clonal redemption of autoreactivity" has also been demonstrated to operate in humans [88,89] (Fig. 2). In summary, a certain degree of autoreactivity is allowed in the naïve B-cell compartment, which might prevent creating holes in the Ig repertoire. It is tempting to speculate, whether the interaction with microbial commensals may shape the pool of remaining poly-and autoreactive B-cell clones during peripheral B-cell tolerance induction in humans.

Formation of a memory B-cell compartment
Humoral immune memory is mediated by long-lasting antibody titers that are maintained by plasma cells and memory B cells [90]. The cognate interplay of B cells and CD4 + T follicular helper (T FH ) cells within GCs of secondary lymphoid organs has emerged as an essential component for the induction of effective and longlasting B-cell responses [91,92]. Indeed, mutations in patients with inborn errors of immunity that alter the function of T FH cells impinge within aborted GC reactions on the induction of memory B cells and long-lived plasma cells that otherwise would constitute the cellular basis of long-lasting protection against re-infection [8]. The outcome of this B-cell response is shaped by the composition of the primary Ig repertoire of naïve B cells that fuels the response and the functional properties of T FH cells. Indeed, there are broad interindividual differences in the formation of the adult memory B-cell compartment and circulating T FH cell pool that may result from different developmental trajectories during childhood [93]. T FH cells from draining lymph nodes of vaccinated neonatal mice showed a Th-2 bias and preferentially differentiated toward short-lived T FH effector cells, which was suggested to limit the early life GC responses elicited by vaccines [94]. Interestingly, also T FH cells were absent in intestines of human fetuses, they emerged after birth and were abundant in infant tissues together with the presence of memory B cells and plasma cells [95]. These intestinal T FH cells produced IL-21 and TNF-α and were competent to induce the differentiation of IgG-or IgAexpressing B cells in vitro. These observations suggest that the GALT seems to be an important tissue environment that is competent to foster T FH cell-mediated differentiation of memory B cells in humans. Future investigations will provide indications of the extent to which the composition of the intestinal microbiome also influences the polarization of these T FH cells and thus the memory B-cell responses mediated by them. Investigations into the establishment of a memory B-cell compartment in humans have so far mainly focused on CD27 + IgD − IgM − switched-memory circulating in the blood, which rather seems to be a limited view on a much more complex system [96]. In addition, recent studies have demonstrated the existence of tissue-resident memory B cells in various human organs [97]. These cells are functionally and phenotypically distinct from circulating memory B-cell subsets and therefore cannot be classified using the categories established from blood analyses for subdividing the memory Bcell pool. Since tissue-resident memory B cells at least in mice seem to provide stronger and faster immune responses after antigen re-challenge than their circulating counterparts [98], future research on the ontogeny and establishment of B-cell memory in humans should particularly focus on origin, location, and function of this B-cell subset.

Conclusions
Due to ontogenetic changes in B-cell developmental lineages, the mature B-cell compartment is fed by functionally different B-cell subsets that emerged from prenatal, early postnatal, or adult precursors. Both negative and positive selection shape the composition of the mature B-cell pool and the (preimmune) Ig repertoires. In addition to endogenous antigens, contact with microbial antigens is also involved in this selection process, with intestinal commensals having a significant influence on the development of a large layer within the B-cell compartment. Future investigations should aim to determine the extent to which alterations of the intestinal microbiome have positive or detrimental influences on the balanced development of a tolerant B-cell compartment as well as the induction of a protective (tissue resident) B-cell memory.