Human memory B cells (MBCs) are generated and diversified in secondary lymphoid tissues throughout the organism. A paired immunoglobulin (Ig)-gene repertoire analysis of peripheral blood (PB) and splenic MBCs from infant, adult, and elderly humans revealed that throughout life, circulating MBCs are comprehensively archived in the spleen. Archive MBC clones are systematically preserved and uncoupled from class-switching. Clonality in the spleen increases steadily, but boosts at midlife, thereby outcompeting small clones. The splenic marginal zone (sMZ) represents a primed MBC compartment, generated from a stochastic exchange within the archive memory pool. This is supported by functional assays, showing that PB and splenic CD21+ MBCs acquire transient CD21high expression upon NOTCH2-stimulation. Our study provides insight that the human MBC system in PB and spleen is composed of three interwoven compartments: the dynamic relationship of circulating, archive, and its subset of primed (sMZ) memory changes with age, thereby contributing to immune aging.
Memory B cells (MBCs) are generated in secondary lymphoid tissues throughout the organism, can persist for a lifetime, and have the capacity to circulate via the bloodstream (Seifert and Küppers, 2016; Weisel and Shlomchik, 2017). MBCs can further adapt in secondary immune responses (Dogan et al., 2009; McHeyzer-Williams et al., 2015; Seifert et al., 2015). The dynamics of humoral immunity and MBC generation are well understood (De Silva and Klein, 2015; Höfer et al., 2006; Mesin et al., 2016; Seifert and Küppers, 2009). However, maintenance and age-related dynamics of the human MBC pool are barely investigated.
The spleen is the largest secondary lymphoid organ and filtration system of the peripheral blood (PB; Kraal, 1992; Steiniger and Barth, 2000). It contributes to B cell maturation and maintenance, and it serves as a reservoir for human MBCs (Cerutti et al., 2013; Dunn-Walters et al., 1995; Kruetzmann et al., 2003; Mamani-Matsuda et al., 2008; Mebius and Kraal, 2005; Tangye et al., 1998). The spleen fulfills a major role in the protection from blood-borne pathogens (Kruetzmann et al., 2003; Martin et al., 2001). This protective role is fully developed at around 2 yr of age and wanes in the elderly population. This renders both age groups, infants and elderly, susceptible to infections with polysaccharide-encapsulated bacteria, and unresponsive to many vaccination strategies (Gibson et al., 2009; Kruetzmann et al., 2003; Timens et al., 1989; Wasserstrom et al., 2008). The protective function of the spleen is essentially provided by specialized B lymphocytes, located in a histologically defined structure, the splenic marginal zone (sMZ), surrounding B cell follicles (Mebius and Kraal, 2005; Steiniger, 2015).
The human sMZ is a unique microenvironment that provides a plethora of stimuli to the residing B lymphocytes, including stromal cells providing stimulation by DLL1, B cell–activating factor (BAFF), a proliferation-inducing ligand (APRIL), and interleukins. Furthermore, B cell helper neutrophils and innate lymphoid cells contribute to the specialized microenvironment, together promoting sMZ B cell survival and proliferation, supporting plasma cell differentiation, Ig secretion, and class-switching (Magri et al., 2014; Puga et al., 2011; Sintes et al., 2017). The sMZ microenvironment imposes a priming effect onto the residing lymphocytes (Cerutti et al., 2013). Studies on murine or human sMZ B cells showed their high motility within the splenic microenvironment (Arnon et al., 2013; Cinamon et al., 2008), but also recirculation to the periphery (Weller et al., 2004). In mice, sMZ B cells are considered to represent a separate lineage with biased Ig heavy chain variable (IGHV) gene repertoire and distinct, Notch2-dependent generation (Cariappa et al., 2001; Hampel et al., 2011; Martin and Kearney, 2000; Tanigaki et al., 2002; Wen et al., 2005). Murine sMZ B cells are mostly Ig-unmutated and participate in the immune surveillance of the PB, and their precursors are selected according to their B cell receptor (BCR)–signaling strength into the sMZ niche (Cerutti et al., 2013; Martin and Kearney, 2002). Murine sMZ B cells develop from transitional B cell precursors, and show integrin-mediated sMZ homing and retention, follicular shuttling, and T cell–dependent (TD) and –independent (TI) responsiveness (Cerutti et al., 2013; Martin and Kearney, 2000; Pillai and Cariappa, 2009). Human sMZ B cells share several of the murine sMZ B cell characteristics, e.g., their NOTCH2-dependent generation and enhanced responsiveness (Cerutti et al., 2013; Descatoire et al., 2014; Kruetzmann et al., 2003; Magri et al., 2014; Puga et al., 2011; Scheeren et al., 2008; Song and Cerny, 2003; Wasserstrom et al., 2008).
Human sMZ B cells show a MBC phenotype, according to CD27 expression and IGHV-mutation status (Dono et al., 2000; Dunn-Walters et al., 1995; Steiniger et al., 2005; Steiniger et al., 2014; Tangye et al., 1998; Tierens et al., 1999). Moreover, human sMZ B cells show low CD23 but high CD21, CD35, and CD1c expression levels (Cerutti et al., 2013; Tangye et al., 1998; Weller et al., 2004). The human sMZ harbors two functionally distinct populations of CD27+CD21high B cells, one IgM-expressing (IgMhighIgD+/−) and one class-switched (IgG+ or IgA+) population (Colombo et al., 2013; Ettinger et al., 2007; Lettau et al., 2020; Steiniger, 2015; Tangye et al., 1998; Zhao et al., 2018). Besides the sMZ B cell subset, a substantial fraction of non-sMZ MBCs is detectable in the human spleen (Ettinger et al., 2007; Lettau et al., 2020; Tangye et al., 1998; Zhao et al., 2018). These non-sMZ MBCs consist of IgM-expressing but also class-switched MBCs in histologically defined niches (Lettau et al., 2020; Zhao et al., 2018), but likely also include MBCs circulating with the bloodstream and temporarily passing the spleen.
In humans, sMZ B cells can circulate (Bagnara et al., 2015; Weller et al., 2004), contributing to a B cell clone network spanning the whole organism, and showing clonal relation between PB, gut, bone marrow, lung, and spleen (Mandric et al., 2020; Meng et al., 2017). Analysis of human antigen-specific MBCs derived from PB and spleen from the same donor revealed uneven distribution of clonally related sequences with preferential localization in the spleen (Giesecke et al., 2014; Mamani-Matsuda et al., 2008; Meng et al., 2017). T cell– and antigen-independent diversification of the sMZ B cell repertoire occurs in infants (Weller et al., 2004; Weller et al., 2001; Weller et al., 2008; Willenbrock et al., 2005), whereas PB memory diversification occurs germinal center (GC)–dependent in adults (Budeus et al., 2015; Seifert and Küppers, 2009; Seifert et al., 2015).
A comprehensive understanding of the human splenic and particularly the sMZ B cell repertoire is limited. Our knowledge on the age-dependent dynamics of the human splenic B cell composition and homeostasis is similarly limited. Here, we present a phenotypical, molecular, and functional characterization of paired PB- and spleen-derived MBCs from healthy human subjects, covering eight decades of life. Our data suggest that throughout life, human MBCs segregate into three distinct compartments, which differ in tissue localization, diversity, and function. The PB MBC compartment (circulating memory) is generated in secondary lymphoid tissues (including the spleen) and shaped by immune responses throughout the organism. Virtually all circulating MBC clones are archived and expanded in the spleen (archive memory). From this archive compartment, MBCs are stochastically displayed and primed in the sMZ microenvironment (primed memory), which represents a functionally distinct subset of the splenic memory archive. The three MBC compartments are interwoven by selection, retention, and recirculation dynamics that change with age, thereby shaping human B cell immunity throughout life.
Circulating MBC, splenic non-marginal zone MBC, and sMZ B cell subsets differ in phenotype and responsiveness, and their relative abundance changes with age
The human MBC system is composed of distinct subsets and distributed throughout the organism. In this study, mature B cell subsets from PB and spleen were analyzed. Splenic B cells (prefix “s”) were distinguished into sMZ and non-sMZ B cell subsets, the latter from here on referred to as splenic IgM memory B cells (sMD27) and splenic class-switched B cells (sCSW). In detail, we distinguished naive B cells (sNaive and PB-Naive, CD23+IgDhighIgMlowCD27−CD21+), IgM MBCs (sMD27 and PB-MD27, IgM+IgD+/−CD27+CD23−CD21+), class-switched MBCs (sCSW and PB-CSW, IgG/IgA+IgM−IgD−CD27+CD23−CD21+), sMZ B cells (sMZ, IgMhighIgD+/−CD21highCD27+CD23−), and class-switched sMZ B cells (sMZ-CSW, IgG/IgA+IgM−IgD−CD21highCD27+CD23−; Fig. 1 A and Fig. S1, A and B). B cells with CD21high phenotype did not appear as a distinct population in PB (Fig. 1 A and Fig. S1, B and E).
The validity of our gating strategy was confirmed by phenotypical differences between the individual subsets, particularly in the surface expression of CD35, CD300a, and CD1c in addition to CD21 (Fig. 1 B), showing highest expression among sMZ and sMZ-CSW B cells. Circulating and splenic non-sMZ MBCs also showed significantly different expression of further surface molecules, including CD27, CD79b, CD148, and CD35 (Fig. 1 B and Fig. S2 A). We assessed the responsiveness and proliferation capacity of sMZ and non-sMZ B cell subsets upon TD and TI type I (TI-I) stimulation (Fig. 1 D). In line with previous studies (Ettinger et al., 2007; Magri et al., 2014; Puga et al., 2011; Sintes et al., 2017), sMZ B cells responded faster and more intensely than non-sMZ MBCs, with the majority of sMZ B cell subsets showing induction of CD25 and CD80 expression after 24 h, and having undergone cell division already on day 3 (Fig. 1, C and D). In contrast, PB-derived and splenic non-sMZ B cell subsets showed reduced expression of activation markers, the former responding with significantly reduced intensity compared with the latter (Fig. 1 C and Fig. S2 B). This tendency was also observed in their respective proliferation capacities, particularly between class-switched B cell subsets (Fig. 1, D and E). TI-I stimulation was highly efficient among splenic B cells, but PB B cells barely responded during 96 h of TI-I stimulation. We conclude that circulating splenic MBCs and sMZ B cells differ in phenotype and response intensity (speed and type of stimulus).
Next, we determined the composition of the splenic and PB B cell compartments from 141 individuals from 0 to 89 yr of age concerning subset quantity and diversity (Table S1). Similar to PB-Naive (Fig. S1, E and F), sNaive B cells are most frequent in infants (0–1 yr of age, up to 60% of total splenic B cells), but then steadily decrease to 15% in the elderly as compared with 40% PB-Naive B cells (Fig. 1 F). Conversely, MBCs in PB and spleen are barely detectable in neonates and infants, increase during childhood, and steadily accumulate with age. MBCs account for up to 60% in the spleen and 30% in the PB of the elderly (Fig. 1 F and Fig. S1, E and F). PB-CSW, sCSW, and sMZ-CSW B cells are nearly absent from infants and children <2 yr of age but increase to adult-like levels around 2–4 yr of age and show high frequency variability in the elderly (Fig. 1 F and Fig. S1, D and G). A distinct population of sMZ B cells was already detectable in infants and young children (n = 8; 2–24 mo; Fig. 1 F and Fig. S1 C). The average sMZ B cell frequency does not change with age, but the fraction of sMZ-CSW B cells increases from 10% in adults to an average of 20% in the elderly (Fig. 1 F and Fig. S1 C). Around 60 yr of age, the splenic B cell pool is dominated by class-switched MBC subsets (sMZ-CSW and sCSW; Fig. 1 F and Fig. S1, C–G).
We selected 17 donors, covering the age span from 0 to 77 yr, and performed deep-sequencing of IGHV gene rearrangements of sort-purified B cell subsets from spleen (sMZ, sMZ-CSW, sMD27, and sCSW) and paired PB samples (PB-MD27 and PB-CSW). We analyzed an average number of 485,000 splenic and 30,000 PB-derived B cells from each donor per subset and retrieved a total of 3,660,797 unique sequences (after quality control and collapsing identical sequences; Table S2). The average IGHV mutation frequency increases abruptly in the first years of life, and afterwards only moderately (Fig. 1 G). sMZ-CSW B cells consistently showed a slightly lower average IGHV mutation frequency compared with sCSW or PB-CSW B cells (Fig. 1 G).
Taken together, the age-related changes in the composition of the human splenic B cell compartment show the same tendencies as observed in the circulating compartment. However, in the spleen, these changes are more intense, leading to a dominance of MBCs, particularly class-switched MBCs, in the elderly.
Throughout life, circulating B cell memory is archived in the spleen
Clonal expansions and phylogenetic analysis of IGHV gene mutation patterns can be used to reconstruct MBC dynamics (Budeus et al., 2015; Seifert and Küppers, 2009; Wu et al., 2010). We investigated the clonal relation of PB and splenic B cells (sMZ and non-sMZ) to track their diversification and distribution in our IGHV gene deep-sequencing cohort (Fig. 1 G).
The clonal abundance (fraction of sequences in clones) increases strongly in the PB during early childhood but gradually extends to the spleen until similar levels are reached at around 40 yr of age (Fig. 2 A). In all individuals, the vast majority (75–99%) of the PB MBCs have clone members in the spleen (Fig. 2 B), indicating an extensive clonal overlap between PB and splenic MBCs throughout life. Conversely, the fraction of IGHV rearrangements that cannot be assigned to clones decreases with age. This results in an overall reduced MBC repertoire diversity, which is delayed in the spleen until midlife (Fig. 2 C). Likewise, the average clone size (counting collapsed sequences deviating by at least one mutation) in the spleen shows a mild but steady increase with age (Fig. 2 D).
The clonal abundance and diversity of PB and splenic MBCs shift at different rates with age, and changes occur earlier in PB. Of note, the number of clones shared between tissues does not change (Fig. 2 E), but the fraction of splenic sequences in clones increases with age, mostly independent of PB (Fig. 2 F). Finally, the number of splenic clone members increases evenly, but a shift is observed around midlife (Fig. 2 G), causing a marked loss of smaller clones and diversity in the spleen (Fig. 2 C). This shift is more intense for splenic clones for which no PB sequence was detected (Fig. 2, F and G).
Taken together, our data suggest that the human spleen (including sMZ and non-sMZ B cells) not only is a reservoir for MBCs but also serves as a comprehensive archive of circulating MBC clones. Moreover, splenic MBC clone sizes appear to increase independent of the periphery.
PB and splenic clone members segregate into distinct branches
The observation that virtually all MBCs from PB have clone members in the spleen indicates that from every MBC clone egressing from secondary lymphoid tissue into the PB, eventually at least a few clone members home to the spleen. Phylogenetic analysis of clone dendrograms allows tracking of the hierarchy and dynamic generation of clone members (Meng et al., 2017; Seifert and Küppers, 2009). Hence, we performed this type of analysis on clone dendrograms consisting of PB and spleen members (shared clones). The segregation of PB- and spleen-derived clone members in dendrograms is not random, but follows a specific pattern: in early childhood, the vast majority (up to 75%) of shared clones showed a consistent separation of PB clone members in one discrete branch (Fig. 3, A and C: 4, 10, 19, and 23 yr). The frequency of clones with discrete PB branches steadily decreases with age, but still accounts for 40% of shared clones in the elderly. This separation of PB clone members within distinct branches is in 98% of the cases not preceded by a common ancestor at the branching point, hampering the identification of subset hierarchies. However, the PB sequence clusters were located significantly further downstream in the dendrograms (Fig. 3 B), indicating that on average, PB and spleen clones frequently share the early steps in Ig diversification. Thus, the spleen-resident fraction of MBCs is phylogenetically overarching the PB fraction. This is not detectable in infants but increased in children and young adults (Fig. 3, A and C). Vice versa, shared clone dendrograms that are hierarchically dominated by PB-derived sequences are rare throughout life (<5% total), albeit most frequent among infant donors.
Finally, intermingling of PB and splenic clone members within a dendrogram increases with age (Fig. 3, A and C: 69 and 77 yr), supporting the idea of increased recirculation, i.e., iterative cycles of spleen egress and reentry during aging.
Taken together, the phylogenetic analysis of shared clones points to a nonrandom distribution of MBCs between PB and spleen throughout life, and PB subbranches are located significantly more downstream in dendrogram hierarchies.
Clonal expansion in the spleen is uncoupled from class-switching
To determine putatively distinct mechanisms of selection or diversification in the periphery and spleen, we quantified clonal expansion and relation between B cell subsets, and compared the frequency of class-switching and conserved mutations. In infants, expanded clones are scarce (Fig. 2), barely show Ig-mutations, and lack class-switched B cells (sMZ-CSW and sCSW; Fig. 1). Therefore, we focused on individuals >4 yr of age.
Throughout life, clonal relations within the spleen are predominantly detected between either IgM-expressing (sMZ and sMD27) or class-switched (sMZ-CSW and sCSW) B cell subsets, respectively (Fig. 4 A). About 35% of the clones in young individuals included both IgM and class-switched B cell subsets, and this fraction decreases substantially with age (Fig. 4 D). Among the shared clones between PB and spleen, the fraction of clones with IgM and class-switched members increases significantly with clone size (Fig. 4 B), as previously observed (Budeus et al., 2015). In contrast, the BCR isotype composition of the splenic MBC compartment is independent of clone size (Fig. 4 C).
Taken together, the isotype composition of clones in the circulating memory compartment is highly diverse, whereas the splenic memory compartment seems to be conserved.
The sMZ B cell repertoire is a stochastic selection of the total MBC compartment
We aimed at determining the distinctness of the human sMZ B cell repertoire. Our data show that sMZ B cells (CD21high) rarely belong to independent clones. In contrast, we observed a major clonal overlap of sMZ B cells and PB or splenic MBCs (CD21+) in all donors throughout age (Fig. 5 A). Moreover, the phylogenetic analysis did not reveal consistent hierarchical patterns, and larger clones frequently showed alternating CD21high and CD21+ members in consecutive positions (Fig. 5, B and C). Among unique MBC sequences (and also when all sequences were regarded, including clonally related sequences), the average IGHV gene usage (Fig. 5 D) and distribution of complementarity-determining region III (CDRIII) length (Fig. 5 E) were nearly identical between CD21high and CD21+ subsets.
Although our data lacked evidence for a specific selection or molecular distinctness of the human sMZ B cell subset, we calculated the enrichment of shared clonotypes (identical IGHV gene and at least 90% CDRIII amino acid identity in multiple donors) in our cohort. Such clonotypes were overall very rare, and most frequent in neonates and children <4 yr of age, in line with our previous observations (Budeus et al., 2020 Preprint). A marginal enrichment of IGHV-mutated clonotypes is detectable in young adults (Fig. 5, F and G).
According to our analysis, human sMZ B cells represent a stochastic sample of the total MBC compartment. To validate the hypothesis that the CD21high sMZ B cell phenotype can be induced in MBCs or even antigen-naive B cells, we made use of an established in vitro system for NOTCH2-dependent sMZ B cell development, which has previously been shown to affect the transcriptional program also (Descatoire et al., 2014). Under OP9-DLL1 stimulation in co-culture, PB-MD27 and PB-CSW B cells acquire a CD21high phenotype (Fig. 5, H and I). CD21high expression was more efficiently adopted by MBCs than naive B cells, especially by IgM-expressing MBCs (PB-MD27 and sMD27; Fig. 5 I). The CD21high phenotype is transient and dependent on the sustained presence of DLL1, as interruption of the co-culture or long-term cultivation of isolated sMZ B cells reverted their CD21high phenotype (not shown).
Taken together, human sMZ B cells show an extensive repertoire similarity to circulating and non-sMZ (CD21+) memory compartments.
The age-dependent dynamics of the human MBC repertoire are poorly investigated. Here, we present a phenotypical, molecular, and functional characterization of paired PB- and spleen-derived MBCs from healthy human subjects, covering eight decades of life.
In children >4 yr of age, the clonal relation of PB and splenic MBCs is already extensive. Even smallest MBC clones with only a single PB member nearly always had clone members in the spleen. This clonal relation of the PB and spleen B cell compartments is consistently detected in 14 donors analyzed. As only a small sample of the PB pool was tested, but was comprehensively covered by splenic relations, this suggests that members of nearly every MBC clone in the circulation may find their way into the spleen. Indeed, early in life, changes of clonal abundance and repertoire complexity are first detectable in PB and subsequently extend to the spleen. Thus, circulating MBCs egressing from GC-dependent maturation in secondary lymphoid organs, e.g., of the respiratory or gastrointestinal tract (Mamani-Matsuda et al., 2008; Mandric et al., 2020; Meng et al., 2017), channel into the splenic archive. Our study is not suitable to make reliable extrapolations to the size of individual clones (Friedensohn et al., 2018), as identical sequences were collapsed within each subset to reduce technical amplification bias. Moreover, we cannot quantify the degree to which primary diversification mechanisms contribute to dendrogram structures. We assume that early in life, where average mutation levels are low and class-switched MBCs are scarce, primary diversification might contribute to repertoire diversity (Scheeren et al., 2008; Weller et al., 2004; Weller et al., 2001; Weller et al., 2008). However, the strong increase of Ig mutation frequencies during late childhood, the frequent detection of consecutive (selected) mutations in clone dendrograms, and the existence of shared (IgM and class-switched) clone members argue for an antigen-dependent GC diversification (Budeus et al., 2015; Seifert and Küppers, 2009). Upon generation and tissue egress, MBCs are comprehensively collected in the spleen, where the blood flow is slowed down and the splenic architecture supports adhesion, residence, and survival of MBCs (Lu and Cyster, 2002; Puga et al., 2011). We anticipate that in this way, from birth on, circulating MBCs build a systematic MBC archive in the spleen. This archive not only is comprehensively composed of clones with relation to the PB, but also produces and hosts a memory compartment without detectable members in the circulation. The fraction of sequences belonging to splenic clones increases with age.
Archive memory is not simply a storage, but its members are sampled for priming and immune surveillance in the sMZ microenvironment. As mutation load, IGHV gene usage, CDRIII distribution, and dendrogram patterns do not differ significantly between sMZ and corresponding non-sMZ B cell subsets, this sampling process is apparently random. We assume that the intermingling of sMZ and non-sMZ B cells is a strong indication for a constant oscillation between both archive compartments. Moreover, our data suggest a different generation and dynamics of human sMZ B cells than in mice, where numerous studies support the idea of a distinct sMZ B cell lineage with unique characteristics (Cariappa et al., 2001; Casola et al., 2004; Cerutti et al., 2013; Martin and Kearney, 2000; Pillai and Cariappa, 2009; Rickert et al., 1995). It was reported that in young children, a human sMZ precursor with commitment to an sMZ B cell phenotype and gene expression upon Notch2 signaling contributes to sMZ B cell development (Descatoire et al., 2014). This is supported and extended by our functional studies, showing that virtually all mature B cell subsets develop a CD21high phenotype upon DLL1 stimulation. Our molecular and functional analyses suggest that the distinctness of human sMZ B cells is not intrinsically encoded but may be imposed from the sMZ microenvironment.
Despite their distinct origins, human and murine sMZ B cells seem to share similar immunological functions, as the priming effect of the sMZ microenvironment is evident in humans also (Cerutti et al., 2013; Puga et al., 2011; Tangye et al., 1998), and we confirmed the primed state of CD21high sMZ versus CD21+ splenic non-sMZ B cells and PB MBCs by their different responsiveness and proliferation capacity.
The phylogenetic analysis of shared clones between PB and spleen (sMZ and non-sMZ) indicates that the clonal relation is not entirely random, but often shows a highly ordered pattern. This pattern has two main features. First, circulating clone members often cluster in single, discrete branches of genealogic dendrograms, suggesting that the circulating MBCs adapted in secondary or tertiary GC reactions, and did not (yet?) reenter the splenic archive. Second, the PB branches originate further downstream of the dendrogram root, suggesting that splenic and PB clone members often share their primary diversification and/or antigen-dependent selection history. Discrete PB branches are less frequent in infants (0 and 1 yr of age), but dominant in childhood and young adults (2–20 yr of age), and progressively decrease with age. The highly diversified, discrete branches of PB clones support the idea of reactivated splenic MBCs, likely representing reactivated sMZ B cells that have previously been described to circulate via the bloodstream (Weller et al., 2004). Thus, the circulating MBC compartment is composed of newly generated MBCs (or new primary diversified B lymphocytes), which have the ability to home to the splenic archive, but presumably also (re)activated sMZ and non-sMZ B cells that egressed from there. We assume that human PB B cells include recirculating sMZ B cells, which are detectable by immunogenetic features, but not phenotypical CD21high expression.
Finally, an increasing alternation between PB- and spleen-derived sequences in the dendrograms is detectable with age, arguing for increased recirculation between periphery and spleen, and for characteristic dynamics of the MBC system in humans.
The comparative analysis of MBC clone abundance, clone size (counting collapsed sequences deviating by at least one mutation), and isotype composition identified two major features of splenic clone members. First, splenic MBCs expand continuously with age, and thereby, larger clones outcompete smaller ones. Second, splenic B cell memory appears conservative, as clonal expansion is uncoupled from class-switching. IgM MBC and class-switched members are preserved equally. This contrasts with the circulating MBC compartment, which is shaped by an increase in Ig diversification and class-switching, as reported previously (Budeus et al., 2015; Seifert and Küppers, 2009; Seifert and Küppers, 2016; Seifert et al., 2015; Tangye et al., 2003; Tangye and Tarlinton, 2009).
The dynamics of circulating and archive memory (including sMZ B cells) provide insight into essential mechanisms of human immune aging. The infant spleen is predominantly composed of naive B cells, showing little indication for IGHV gene selection, GC-dependent diversification, or expanded MBC clones, although a large population of sMZ B cells is present. Around midlife, archive B cell clonality substantially increases, causing accumulation and persistence of large MBC clones in the elderly. As this occurs at the expense of smaller clones, it is likely increasingly difficult to archive new MBC clones in old age. These changes coincide with the time of thymic involution and changes in lymphopoiesis in the bone marrow (Guerrettaz et al., 2008; Palmer et al., 2018; Pang et al., 2011). We assume that the differences in Ig repertoires of infants and the elderly are a major cause of the increased susceptibility to infections in these risk groups, compared with young adults (Di Sabatino et al., 2011; Siegrist and Aspinall, 2009; Timens et al., 1989). We can only speculate that the predominance of sMZ-CSW in elderly individuals affects responsiveness to polysaccharide vaccines, enabling an effective secondary response to previously encountered antigens but preventing the formation of new memory. By contrast, conjugated polysaccharide vaccines provide T cell help and trigger memory formation arising from naive B cells, albeit these decrease in the elderly. Thus, our findings might have direct implications for age-related immunization programs.
We propose here an updated view on the human MBC system, with a special emphasis on B cells in the spleen and their relation to PB MBCs. In this view, human B cell memory is composed of three compartments that are dynamically interwoven. Circulating MBCs carry traits of adaptive, evolving immune responses. Newly egressing MBCs are recruited into the spleen, where they contribute to a systematic MBC archive, which is homogenously expanded and conserved throughout age. A random but frequently exchanging sample of archive memory is recruited into the primed or preactivated sMZ compartment (primed memory), forming a fast-responding subset of the splenic archive. Splenic MBCs can reenter circulation for further adaptation in the periphery (or the spleen itself). Human MBC composition changes with age; the outgrowth of many large clones in the spleen starts at mid-life and dominates immune aging, presumably being a prerequisite for enhanced lymphomagenesis and susceptibility to infection in the elderly.
Materials and methods
Buffy coats were collected from PB donations of immunologically healthy human adults at the Institute for Transfusion Medicine at the Medical School Essen after informed consent was given according to the Declaration of Helsinki, as well as approval by the ethics committee of the Medical Faculty of the University of Duisburg-Essen, Germany. Splenic biopsies and paired PB were prepared on the same day from organ donors (Table S1). All samples were collected from leftover material from the Institute for Transfusion Medicine at the Medical School Essen, as approved by the ethics committee of the Medical Faculty of the University of Duisburg-Essen, Germany. All individuals died of noninfectious and nonmalignant causes, i.e., anoxic brain damage, cardiac arrest, cerebral trauma, bleeding, or infarction.
Mononuclear cell isolation, magnetic cell separation, flow cytometry, and cell sorting
Mononuclear cells from different tissues were isolated by Ficoll density gradient centrifugation followed by positive selection of CD19-expressing cells by magnetic cell separation (Miltenyi Biotec).
B lymphocytes were either analyzed on a CytoFLEX S flow cytometer (Beckman Coulter) using the CytExpert software V2.4 (Beckman Coulter) or FlowJo Software V10.6.2, or sort-purified on a FACSAriaFusion cell sorter or FACSAriaIII cell sorter (BD Biosciences) equipped with BD FACSDiva software (BD Biosciences). For IGHV gene rearrangement analysis by high-throughput sequencing, the B cell subsets were sorted according to the following phenotypes: MBCs from PB were sorted as PB-MD27 (CD20+IgM+IgD+CD27+CD23−CD21+) or PB-CSW (CD20+IgG+/IgA+CD27+CD23−CD21+) B cells. Splenic B cell populations were sorted as sMZ (CD20+IgM+IgD+/−CD27+CD23−CD21high), sMZ-CSW (CD20+IgG+/IgA+CD27+CD23−CD21high), sMD27 (CD20+IgM+IgD+/lowCD27+CD23−CD21+), and sCSW (CD20+IgG+/IgA+CD27+CD23−CD21+) B cells. For in vitro functional assays, B cell subsets were sorted according to the following phenotypes: PB-Naive and sNaive B cells were sorted for functional validations (CD20+IgMlowIgDhighCD27−CD23+CD5−CD21+), PB-MD27 and sMD27 (CD20+IgM+IgD+/−CD27+CD23−CD21+), PB-CSW and sCSW (CD20+IgM−IgD−CD27+CD23−CD21+), sMZ (CD20+IgM+IgD+/−CD27+CD23−CD21high), and sMZ-CSW (CD20+IgM−IgD−CD27+CD23−CD21high).
In vitro OP9 co-culture
OP9 co-culture assays were performed as described previously (Descatoire et al., 2014). Briefly, OP9-GFP or OP9-DLL1 stromal cells were seeded in 24-well plates at a density of 20,000 cells/well. After 24 h, sort-purified B cell subsets were added at a density of 50,000 cells/well, and after additional 5 d of incubation, cell surface expression of CD21 and CD27 was analyzed on a CytoFLEX S flow cytometer (Beckman Coulter) using the CytExpert software (Beckman Coulter).
In vitro functional assays
For functional studies, human splenic B cell subsets (sMZ, sMD27, sCSW, sMZ-CSW, and sNaive) and PB B cell subsets (PB-MD27, PB-CSW, and PB-Naive) were isolated using fluorescence-activated cell sorting (FACS) and cultured in RPMI 1640 medium supplemented with 20% fetal bovine serum (Pan Biotech), 100 U/ml penicillin, and 100 µg/ml streptomycin at 37°C and 5% CO2. All populations were left unstimulated or were activated by TD stimulation using 0.03 µg/µl anti-Ig (Jackson ImmunoResearch) and 1 µg/ml CD40-ligand-hemagglutinin with 5 ng/ml anti-hemagglutinin antibodies (R&D Systems). TI-I stimulation was mimicked by incubation with 2.5 µM cytosine-phosphate-guanine oligonucleotide type B (InvivoGen). Phenotypic analyses were performed directly after cell sorting without further activation, and after 24 h of incubation in the presence of TI-I or TD stimulation. In proliferation assays, cells were pulsed with eFluor670 (Invitrogen) at a concentration of 5 µM, stimulated with TD or TI-I stimulation, and analyzed after 48, 72, and 96 h of incubation.
IGHV gene rearrangement analysis by high-throughput sequencing
Rearranged IGHV genes of the sort-purified B cells were sequenced by next-generation sequencing. Genomic DNA was extracted from sorted B cell populations (Gentra Puregene Core Kit; Qiagen). Sequencing was performed using the LymphoTrackDx IGH FR1 assay (Invivoscribe Inc.) and the rapid run program with paired-end sequencing and two times 300 bp read length. Reads generated by MiSeq (Illumina) were included only when the average quality was ≥25. Ambiguities between forward and reverse reads were replaced by “N.” Identical sequences were classified as PCR duplicates and reduced to the longest detected sequence. Each N in a given sequence was replaced by the most frequent (majority vote) nucleotide at that position. Only in-frame sequences detected more than once were further processed, as a means to filter out PCR errors or sequencing artifacts present in single sequences. Sequences were considered clonally related when using the same IGHV gene and sharing at least 97% CDRIII sequence similarity and when sequences were derived from at least two B cell subsets. Clonally related sequences within single subsets (identified by intraclonal Ig diversity) were disregarded due to their low frequency and probability of artificial origin. All analyses were performed with R 3.6 (http://www.R-project.org). The IGHV sequencing data are accessible under the BioProject ID PRJNA681128.
Clonal abundance calculations
Clonal abundance (fraction of sequences belonging to clones) was calculated using a stochastically selected set of sequences with equal numbers of sequences from each sorted population (e.g., 40,000 sequences from sMZ from each donor, 4,800 sequences from PB-CSW from each donor). These reduced numbers were used to calculate clones to normalize for changes in clonal numbers due to different numbers of sorted, processed, and sequenced cells. All further analyses were performed with R 3.6, using several packages to visualize (ggplot2, ggpubr; circos plots: circlize; radar plots: fmsb; Gu et al., 2014; Kassambara, 2020; Nakazawa, 2019; Villanueva and Chen, 2019) and perform calculations, such as dose–response analyses (drc; Ritz et al., 2015).
Calculation and analyses of hierarchical trees
Hierarchical trees of all sequences in clones were calculated for each donor using IgTree (Barak et al., 2008). Tree analyses and visualizations were performed in R using the packages igraph (Csárdi and Nepusz, 2006) and Rgraphviz (Hansen et al., 2020). Subtree analyses were performed in trees with five or more sequences of the cell population analyzed (e.g., five or more sequences originating from PB).
Analyses of sequence characteristics
All evaluations were based on the international ImMunoGeneTics information system database (http://www.imgt.org/). Mutation frequencies were calculated based on the number of nucleotide exchanges in the IGHV region of each sequence in comparison with the most similar allelic variant present in the respective donor (determined from unmutated sequences). IGHV identification was performed with BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). CDRIII length distribution was calculated using the nucleotides covering the whole CDRIII as determined by the ImMunoGeneTics information system.
Clonotypes were analyzed using the amino acid sequence of the CDRIII and the information of the IGHV gene used. All sequences with at least 90% identity and the same IGHV gene regardless of their subset from at least two different donors were considered as a clonotype. Clonal expansions in clonotypes (sequences originating from the same donor) were excluded.
Online supplemental material
Fig. S1 shows the gating strategy for PB and splenic B cell subsets and shows representative analyses of the age-related changes in the IgM and CD21 expression among PB and splenic B cells of five individuals. It also shows the subset frequencies of PB B cell subsets and splenic CD21+ MBCs. Fig. S2 shows the steady-state expression of differentiation markers, and steady-state and up-regulation of activation markers after 24 h of TI-I or TD in vitro stimulation. Fig. S3 shows the clonal fraction between all sequences. It also shows all clonal relations among all B cell subsets. Table S1 shows the donor characteristics and usage in figures. Table S2 shows the cells included in BCR deep sequencing and number of sequences retrieved by next-generation sequencing.
We thank Klaus Lennartz for his valuable engineering support. We thank Ludger Klein-Hitpass and the staff members at the Biochip Laboratory at the Institute of Cell Biology in Essen. We thank the staff members at the flow cytometry and fluorescence core facility Imaging Center Essen.
This work was supported by the Deutsche Forschungsgemeinschaft through grants SE1885/2-1, SE1885/2-2, and SE1885/4-1 and the Deutsche Krebshilfe through grant 70112628.
Author contributions: A. Kibler designed and carried out most of the experiments and prepared the manuscript. B. Budeus developed and performed bioinformatical and statistical analysis. E. Homp, K. Bronischewski, V. Berg, L. Sellmann, F. Murke, A. Heinold, F.M. Heinemann, M. Lindemann, I. Bekeredjian-Ding, C. Kirschning, and P.A. Horn contributed experimental work and provided material. R. Küppers, B. Budeus, and M. Seifert developed the concept, designed experimental strategies, helped with data evaluation, and prepared the manuscript.
Disclosures: The authors declare no competing interests exist.