Surface phenotypes of naïve and memory B cells in mouse and human tissues

Abstract

Memory B cells (MBC) protect the body from recurring infections. MBC differ from their naïve counterparts in many ways, but functional and surface marker differences are poorly characterized. In addition, though mice are the prevalent model for human immunology, information is limited concerning the nature of homology in the B cell compartments. To address this, we undertook an unbiased, large-scale screening of both human and murine MBC for their differential expression of surface markers. By correlating the expression of such markers with extensive panels of known markers in high dimensional flow cytometry we comprehensively identified numerous surface proteins that are differentially expressed between MBC and naïve B cells (NBC). The combination of these markers allows for the identification of MBC in humans and mice and provides insight into their functional differences. These results will greatly enhance our understanding of humoral immunity and can be used to improve immune monitoring.

One sentence summary:

Weisel and colleagues provide a Resource that phenotypically profiles naive and memory B cells, and provides a comparative analysis of memory B cells found in humans versus mice

Introduction

Modern research increasingly relies on technologies that allow broad, unbiased characterization, generating larger datasets that provide snapshots of cell states. In particular, gene expression has been cataloged, initially by microarray analysis using specific probes and later by RNA-sequencing at both the population and single cell level. Nucleic acids are particularly amenable to such analysis. However, expression of proteins, which carry out the bulk of biological function, is less easily quantified using highly parallel, untargeted assays. Yet, it is well established that protein expression—which largely dictates function—often diverges from mRNA expression ¹.

Immune system cells are distinct from most cell types in their ability to circulate and migrate throughout the body. Thus, their surface proteome, which mediates interactions and coordinates responses, is of particular interest. To this end, the immunological research community has been extensively defining and cataloging expression of surface proteins using collections of mAbs that recognize them. This has led to the “Clusters of Differentiation” or CD system of classification, which also seeks to unify nomenclature and align molecular identification across species². Tracking expression of these surface molecules has been useful for characterization, identification, purification and functional analysis. Nonetheless, comprehensive analysis of the expression of large numbers of surface molecules among specific cell types has not been generally available.

MBC are generated in response to pathogen exposure or vaccination, and they play key roles in protection from infection in both cases^{3, 4, 5, 6}. MBC may also be key players in mediating chronic autoimmunity^{7, 8, 9, 10}. MBC can be defined as progeny of cells that have responded to an Ag-specific stimulation and remain in the animal in a resting state after the initial exposure, oftentimes for prolonged periods. MBC have been characterized in transgenic mice that carry B cell receptors (BCRs) of a single antigen-specificity. In mice and humans it has been possible to detect and track Ag-specific MBC using labeled antigens using flow cytometry.^{3, 11, 12, 13} In most human studies, peripheral blood is the only available source for the study of MBC, although several studies have included spleen, other lymphoid tissues, gut and tonsil as additional sources^{14, 15, 16, 17, 18}. Expression of the surface marker CD27 is commonly used to identify MBCs in humans^{16, 19}, but it is clear that CD27 negative MBCs exist ^{20, 21, 22, 23}. Often, in both mice and humans, isotype-switched B cells are taken as surrogates for the MBC population, though it is clear that IgM MBC are prevalent in humans and mice, so such an approach risks incomplete characterization of MBC^{3, 14, 19, 24, 25, 26}.

Research on MBC, particularly in polyclonal murine models, has been hampered by their low numbers, difficulty in tracking their Ag-specificity and the lack of markers that define them unambiguously. Further, MBC are heterogeneous^{27, 28} and understanding their phenotypic diversity and the functional consequences of it are active and evolving areas of research. Hence, more insight into MBC heterogeneity will be helpful to advance our knowledge and understanding of humoral memory.

For these reasons, and in particular due to urgent needs for vaccines and therapies for autoimmunity, the biology of MBC in mice and humans has become a growing area of research focus³. Murine models are essential for testing basic biology, while human application is critical. The biology of murine and human MBCs is assumed to be related but the actual homology is very poorly defined, due to lack of comparative studies.

To address these issues, we undertook a comprehensive characterization of the surface proteome of human and murine MBC, taking advantage of the commercially available LEGENDScreen^™ assay technology. We were able to do this by leveraging our unique mouse systems in the Balb/cJ and C57BL/6 backgrounds to generate sufficient numbers of highly defined MBC ^{24, 25} as well as our access to an array of human tissues, including spleen, gut, lymph node, tonsil, peripheral blood and bone marrow from a large number of donors¹⁴. Mice are widely used as a model for human immune systems but there exists no comprehensive parallel map of B cell compartments in the two species. Here we present in-depth phenotyping of murine and human MBC and, uniquely, a direct comparison of MBC between these species, which should be invaluable in translating findings from one species to the other.

Results

Differentially expressed surface markers between NBC and MBC

We generated immune animals containing nitrophenol (NP)-specific MBC using a published strategy^{24, 25} (Fig. 1 and see methods). We used their splenocytes for the LEGENDScreen^™ assay, staining all of the cells with a panel of monoclonal Abs (mAbs) to enable gating on different known subpopulations. NBC were ascertained as non-antigen (Ag)-specific cells in the same wells after gating (Supplementary Fig. 1 and see methods).

Figure 1: Discovery of differentially regulated surface markers between memory and naive B cells by murine LEGENDScreen^™ analysis.

a, Overview of experimental procedures. We used an established murine transfer system that generates large numbers of MBC in response to hapten NP-carrier immunization ^{24, 25}. Splenocytes of 28 NP-immune transfer recipients (15 females and 13 males at 17 weeks post immunization) were harvested, pooled and subjected to flow cytometric staining with Abs listed in Supplementary Tables 5 and 6 “Murine Stain 1” to allow for the simultaneous identification of NBC and NP-specific MBC. Cells were then transferred to LEGENDScreen^™ plates with each well containing a single commercially available surface marker in PE (see Methods). Images are from BioRender software (https://app.biorender.com). b, Flow cytometric detection of NP-reactive MBC (red gate) and NBC (blue gate) and their expression of CD81-PE as an example. Gating strategy is in Supplementary Fig. 1.

To assess distribution of human B cell surface markers we used splenocytes of 3 individual human organ donors. These were barcoded with different amounts of violet proliferation dye (VPD, see Methods) allowing us to run three individuals simultaneously in each well (Fig. 2a). This experimental setup enabled us to analyze human splenocytes for their expression of 332 individual markers with MBCs gated as CD27⁺ (gating strategy in Supplementary Fig. 2), using CD27⁻ cells as a provisional NBC population (Fig. 2b)¹⁴.

Figure 2: Discovery of differentially regulated surface markers between memory and naive B cells by human LEGENDScreen^™ analysis.

a, Overview of experimental procedures of VPD barcoding and flow cytometric staining of splenocytes of 3 different donors with “Stain LS” (Supplementary Tables 3 and 4). Image is from BioRender software (https://app.biorender.com). b, Gating strategy for the flow cytometric detection of NBC (CD19⁺ CD27⁻) and MBC (CD19⁺ CD27⁺). Shown are histograms of the LEGENDScreen^™-encoded surface marker CD21 derived PE-signals as example data for NBC and MBC of each individual donor, as distinguished by their differential VPD-fluorescence. Arrows indicate subsequent gating of populations and numbers next to outlined areas indicate percentages of cells in gated populations. Gating strategy is in Supplementary Fig. 2.

We defined markers as expressed if the mean fluorescent intensity (MFI) of NBC (N), MBC (M) or both was at least 10x the MFI of the corresponding PE FMO. We categorized markers as higher on MBC than NBC (M>N, red in Fig. 3) if the ratio MFI MBC/ MFI NBC was > 1.1, while N>M was defined as a ratio <0.9 (blue in Fig. 3). Ratios <=1.1, >=0.9 were deemed N=M (not depicted in Fig. 3). The human categorization used the same criteria and included markers with MFI higher than 10 times the MFI of the PE FMO for at least two out of the three donors. The distribution of ratios of MFI MBC/MFI NBC for expressed markers is visualized in Extended Data Fig. 1.

Figure 3: Venn diagram of highly expressed surface markers on human and mouse naive and memory B cells.

Markers depicted showed an MFI greater than 10 times the respective FMO stain for NBC and MBC. Markers that reached this cut-off only on one of the latter populations are included. Markers in red are expressed higher on MBC (ratio M/N > 1,1), markers in blue are higher on NBC (ratio M/N < 0,9). In total there are 28 overlapping markers between both species. Blue (human) and red (mouse) asterisks depict markers that were detected by the master “gating” stain during LEGENDScreen^™ assays, rather than by the LEGENDScreen^™ PE-labeled antibody. M, MBC; N, NBC.

We summarize all of these results in Venn diagram format in Fig. 3, where the intersection lists the 28 Ab targets that were differentially expressed between NBC and MBC in the same fashion in humans and mice. Certain markers (29 for human and 16 for mice) could not be compared between the species since they were only present in either the human or murine LEGENDScreen^™ analysis and these are indicated. Interestingly, 6 markers were found to have opposite expression patterns between the 2 species. Finally, a substantial number of markers was differentially expressed in either humans (11) or mice (24) but not both species. In Supplementary Table 1 we list and categorize all markers of the LEGENDScreen^™ assays including markers which showed no or low expression to provide a complete dataset.

We depict LEGENDScreen^™ data analysis in Fig. 4 in a different fashion, summarizing the proportion of markers for each species in each category of regulation (Fig. 4a) as well as the overlap between species of marker that are M>N (Fig. 4b) or N>M (Fig. 4c). Notably, a greater number of surface proteins in both species was upregulated in MBC compared to NBC in both species (Fig. 3, and Fig. 4b), suggesting that MBC tend to acquire an added set of capabilities and specialization compared to NBC, rather than to shut down NBC-specific gene expression. With respect to M>N the concordance between human and mouse in markers that were evaluable (i.e. that were present on both human and mouse LEGENDScreen^™ plates) was remarkable; of 58 such markers total, 27 of them were increased in MBC compared to NBC in both species.

Figure 4: Summary of marker expression revealed by human and mouse LEGENDScreen^™ analysis.

a, Marker expression pattern of all analyzed human and mouse surface markers summarized in pie diagrams (also see Supplementary Table 1). The markers shown as “expressed” had an MFI greater than 10 times the respective FMO for NBC and/ or MBC. Markers in red have higher expression on MBC, markers in blue are higher on NBC. Markers in green show a ratio M/N between 0,9 and 1,1. Markers in grey didńt reach an MFI of 10 times FMO on any cells and were therefore categorized as “not or low expressed”. b, Venn diagram of markers that were more highly expressed on MBC and the overlap between species. c, Venn diagram as in (b) for markers more highly expressed on NBC. d, Heatmap of differentially expressed markers in humans and mice. Markers are sorted and grouped into four color categories based on the Venn Diagram in Fig. 3. Circle color shows the log2 MFI fold-change (FC) between MBC and NBC. Markers with more than 2 log2 fold-change values are set as 2. Circle size represents the overall level of expression as log2 fold change of MFI of the cell type with higher expression (either memory or naive) vs. FMO of that cell type. Heatmap was built using ggplot2 (version 3.2.1) in R (version 3.6.1).

In Fig. 4d, for each marker that was differentially expressed, we depict the degree of MFI fold change between MBC and NBC by color coding, allowing us to show the results of each of the three human samples we screened. This provides a sense of interindividual variation. At the same time, the plot uses circle size to represent the overall level of expression (by comparing MFI of stained cells with MFI of control unstained cells), enabling the visualization of more well-expressed markers that may have more practical value in future methods development. The markers in this figure are further grouped by how the differential expression pattern was shared between the species. This depiction provides a particularly useful list for investigators designing staining panels to distinguish NBC from MBC as well as for inferring function based on the particular annotated nature of the measured surface proteins.

As the LEGENDScreen^™ kit uses a predetermined amount of PE-conjugated Ab in each well, the Ab concentration may not have been optimal for the particular application of B cell staining. In addition, as a screen, it was only performed once. Thus, validation of key markers was necessarily performed in both murine and human systems. A total of 24 markers was individually stained on gated Ag-specific MBC in the murine system (Extended Data Fig. 2), which served to confirm their differential expression under more optimal conditions. As we identified these markers initially in Balb/cJ mice, we went on to validate them in the widely used C57BL/6 background (Supplementary Figs. 3 and 4). We did not observe differences in marker expression between these mouse strains, supporting the general relevance of these markers.

Similarly, for humans we confirmed the differential expression of 16 selected markers across three individual donors (Extended Data Fig. 3 and Supplementary Table 2).We show patterns of expression in histograms and visually represent the degree of separation that may be achieved in flow cytometry between NBC and MBC, recognizing that the extent of differences depends on both exact instruments and reagents being used. Histograms also reveal patterns and heterogeneity of expression of the target surface antigens among a population of B cells (either NBC or MBC). In this connection, while most of the mAbs demonstrate unimodal staining patterns, for several markers there are either broad peaks or even clear bimodal populations. In this category for mice are CD1d, CD43, CD59a and PlexinB2, among others; in humans this pattern pertains to CD45RB and the tissue residency marker CD69^{29, 30} (which we previously reported¹⁴), as well as CD21 and CD35. These broad or bimodal distributions are restricted to MBC, while NBC are almost always monodisperse, which suggests the existence of heterogeneity or subsets specifically among MBC. Defining the functional significance of such subsets is the subject of past and ongoing work by us and others^{3, 23, 25, 31, 32, 33}.

Combinations of new markers define MBCs in mice and humans

We next explored various combinations of markers in multicolor staining panels, with the goal of clearly distinguishing MBC from NBC; such panels would be of great general use both in mice and humans. Use of two markers, as opposed to just one, generally allows for greater separation of populations, with the goal of finding combinations of two markers that stain and distinguish MBC vs NBC in mice (Fig. 5). As the initial data of the LEGENDScreen^™ assay were derived from comparing MBC to endogenous NBC of transgenic animals we refined our BALB/cJ experimental system by mixing CD45 allotype-marked MBC and their direct naive precursors in a single flow cytometric stain. This system allowed the direct comparison of marker distribution (gating strategy in Supplementary Fig. 5 and see methods section).

Figure 5: Combinations of surface markers that allow for the identification of murine MBC.

Single cell suspensions of indicated tissues from 3 male adoptive transfer recipients (22 wks old; 13.5 wks post immunization) with defined CD45.2⁺ NP⁺ CD19⁺ MBC populations (Fig. 1) were mixed with cells isolated from respective tissues of 1 male naive CD45.1 B1–8i^+/− mice (14 wks old). This allowed for the simultaneous identification of MBC and their comparable naive counterparts in a single staining tube. Cells were stained for indicated surface markers with “Murine Stain 2” (Supplementary Tables 5 and 6), and MBC (red) were identified as viability fixable dye^neg, CD45.2⁺, NP⁺, CD19⁺ B cells and NBC (blue) were identified as viability fixable dye^neg, CD45.1⁺, CD19⁺ B cells. To avoid confounding by marginal zone B cells, which are only present in spleen, splenic NBC were gated as follicular B cells (CD21/35^low). Shown are contour plots of pairwise combinations of CD274/ CD205, CD11a/ CD81 or CD180/ CD267, which combinations can be used to distinguish MBC and NBCs across tissues. MLN, mesenteric lymph nodes; BM, Bone Marrow; LP, lamina propria; PP, Peyeŕs Patches.

As some markers appear to stain only a subset of MBC we excluded these from consideration, as they would not permit clean separation of all NBC. Markers that only stain a subset of MBC can be used to identify novel subsets of MBC, which will be the subject of future reports, as these data exceed the scope of this paper. Rather, we focused on combinations of makers that appeared to stain all or nearly all MBC.

Focusing on pan-expressed markers, in mice, we found that the combinations of the markers CD274/ CD205, CD11a/ CD81 and CD180/ CD267 can be used to distinguish between splenic NBC and MBC (Fig. 5). We also tested the utility of these staining combinations on MBC derived from other tissues; the quantification of MFIs achieved on NBC and MBC in each tissue is represented in Supplementary Fig. 6. While most of the individual markers included in the marker combinations performed similarly in all of the tissues, there were a few exceptions: CD274 (PD-L1) showed less differential staining in MLN while CD267 (TACI) failed to show differential staining in both lamina propria (LP) and Peyer’s Patches (PP). CD267 appeared to be dim in the particular experiment shown in Fig. 5 but had generally better separation as shown in Extended Data Fig. 2. Despite showing differential staining of MBC and NBC across tissues, most markers showed larger differences in spleen than in other tissues. To investigate whether this was in part related to cell processing techniques, which differ among tissues (see Methods), we compared staining MFI in conventionally prepared splenocytes to splenocytes isolated by enzymatic digestion, according to protocols used for LP, PP and lung. MFI on MBC of these tissues were also measured at the same time (Supplementary Fig. 7). These data confirmed the generally higher marker expression in spleen, and further showed that processing was not an explanation for this; in fact, in most cases, splenic MBC marker MFIs were higher in splenocytes prepared by digestion vs conventional cell dissociation. Among all the markers, CD267 was most sensitive to cell isolation methods.

To broaden the applicability of these antibody combinations, we directly immunized C57BL/6 mice i.p. with 100μg NP-KLH and assayed them on spleen, mesenteric LN, BM and lung tissues 28 days post immunization (Extended Data Fig. 4 left panel; gating strategy in Supplementary Fig. 8a). Additionally, we generated MBC in our OT-II adoptive transfer system in the C57BL/6 background (see Methods and Supplementary Fig. 3) and analyzed these marker combinations on splenocytes and mesenteric LN (Extended Data Fig. 4, right panel; gating strategy in Supplementary Fig. 8b). Altogether, the marker combinations CD274/ CD205, CD11a/ CD81 and CD180/ CD267 proved useful in the identification of MBC across tissues, murine strains, and direct and adoptive transfer immunization.

As humans can be immunologically heterogeneous, we sought to assess the variability and general applicability of several pan-memory markers elucidated by our LEGENDScreen^™ analysis across a larger number of donors and tissues. We focused on CD11a, CD24, CD54, CD180 and CD200 since they displayed differential expression on NBC and MBC across tissues (Fig. 6 and Supplementary Table 2). Cell staining with anti-CD200-BUV737 turned out weak (Fig. 6), as BUV737 is not a bright fluorophore in this setting; to better illustrate the potential of this marker to separate NBC and MBC, we stained cells from multiple tissues of D256 with anti-CD200 in BV605, demonstrating better separation of NBC and MBC.

Figure 6: Differentially regulated surface markers on naive and memory B cells across human tissues and donors.

Single cell suspensions of spleen (SP), blood (B), bone marrow (BM), lymph node (LN), intestinal tissue (Gut) and tonsil (T) were stained for flow cytometric analysis using “Stain 2”, “Stain 3”, “Stain 4” and “Stain 5” (Supplementary Tables 3 and 4). The left panel shows the summary of the differences in MFI between CD19⁺ CD27⁺ B and CD19⁺ CD27⁻ B cells (Δ MFI) for the depicted surface markers. For CD11a spleen (n=11 in red), blood (n=9 in blue), BM (n=3 in magenta), LN (n=3 in green), gut (n=10 in brown) and tonsil (n=2 in black) samples were analyzed. For CD24 spleen (n=5 in red), blood (n=3 in blue), BM (n=3 in magenta), LN (n=3 in green), gut (n=4 in brown) and tonsil (n=2 in black) samples were analyzed. For CD54 spleen (n=9 in red), blood (n=9 in blue), BM (n=3 in magenta), LN (n=3 in green), gut (n=9 in brown) and tonsil (n=2 in black) samples were analyzed. For CD180 spleen (n=5 in red), blood (n=3 in blue), BM (n=3 in magenta), LN (n=3 in green), gut (n=4 in brown) and tonsil (n=2 in black) samples were analyzed. For CD200 spleen (n=25 in red), blood (n=10 in blue), BM (n=10 in magenta), LN (n=9 in green), gut (n=9 in brown) and tonsil (n=2 in black) samples were analyzed. The right panel shows example histograms for depicted surface markers of CD19⁺ CD27⁻ B cells (blue) and CD19⁺ CD27⁺ B cells (red) across tissues of donor D256 or D260 for CD200-BUV737 respectively. The histograms of staining with anti-CD200-BV605 across tissues were obtained with cells from D256 and serve as example of CD200 expression revealed in a fluorochrome brighter than BUV737 (“Stain 3”; Supplementary Tables 3 and 4). Stars indicate significant differences in ΔMFI of indicated tissues compared to spleen using the unpaired two-tailed t-test with Welchś correction. * p< 0.05, ** p< 0.01, *** p< 0.001, **** p< 0.0001. Exact significant p-values for comparison between spleen and the indicated tissue for each marker are for CD11a: blood p=0.0022; bone marrow and gut p<0.0001; lymph node p=0.0013 and tonsil p=0.0006; for CD54: tonsil p=0.0485 and CD180: blood p=0.0249; lymph node p=0.0302, gut p=0.0230 and tonsil p=0.0110.

We further stained for the markers CD74, CD119, CD218a and CD370 (Extended Data Fig. 5). These results showed relative consistency among donors, albeit the degree of differential expression was somewhat variable, particularly for CD74 and CD370. Some of these markers did not perform as well in separating MBC from NBC in one or more tissues outside the spleen.

To combine these markers to enhance NBC and MBC separation, we focused on CD11a, CD200, CD54, CD81, CD24 and CD180. Combinations of CD200/CD11a, CD200/CD54, CD81/CD24, CD24/CD11a, CD180/CD54 and CD180/CD24 allowed for the separation of CD27⁺ B cells from CD27⁻ B cells over multiple tissues in individual donors (Fig. 7 and Supplementary Fig. 9).

Figure 7: Combinations of surface markers that allow for the identification of human memory B cells across tissues.

Human spleen, blood, bone marrow (BM), lymph node (LN), tonsil and gut single cell suspensions were stained for depicted markers (“Stain 3”, Supplementary Tables 3 and 4). Contour plots of combinations of CD200/ CD11a, CD200/ CD54, CD81/ CD24, CD24/ CD11a, CD180/ CD54 or CD180/ CD24 are shown for 3 individual donor tissues (D256, D290, D337). CD19⁺ CD27⁺ B cells are in red and CD19⁺ CD27⁻ B cells are in blue.

Markers that identify MBC among CD27 negative cells

Though CD27 can be a convenient memory marker of reference, not all MBC express CD27. As other markers have been lacking, typically these CD27⁻ MBC have been identified by their switched Ig isotype, an indicator of prior antigen exposure. In hopes of finding novel markers that could identify CD27⁻ MBC without staining for switched isotype, we selected two of our strongest markers, CD11a and CD200 (Figs. 6 and 7), for further investigation. If such markers can identify CD27⁻ MBC then they should differentially stain CD27⁻ B cells that are isotype switched. Indeed, among CD27⁻ B cells, IgG⁺ cells express the highest amounts of CD11a, at levels similar to those of CD27⁺ MBC (Extended Data Fig. 6 and Supplementary Fig. 10). Interestingly, for three of six donors IgA+ CD27⁻ cells were similar to IgG⁺ cells while for the others they displayed lower CD11a expression. Conversely, CD200, which is downregulated on CD27⁺ MBC, showed lower expression on IgG⁺ and IgA⁺ CD27⁻ B cells, again with variability among the donors. Thus, CD11a and CD200 can be used to identify switched and possibly unswitched CD27⁻ MBC (Extended Data Fig. 6 and Supplementary Fig. 10). Differences between donors and the observed broad or bimodal distribution of expression of both CD11a and CD200 on isotype switched CD27⁻ cells further suggest that there may be additional underlying heterogeneity in their origins.

Surface phenotype of MBC correlates with biological function

The comparative surface proteomics analysis described is operationally useful as a resource for investigations seeking to distinguish MBC from NBC or to further characterize the nature of B cell compartments in various settings. In addition, the suite of differentially expressed surface proteins can be informative for predicting responses and functions of NBC and MBC and how they may differ from each other. To provide some initial insight into this, we performed pathway enrichment analysis (see Methods for details) on the surface markers that are expressed on MBC to a greater extent than on NBC in each species. We then arrayed these into a network (Fig. 8), which depicts the pathway names and their relationships to each other. The human and murine pathways were similar, though not identical, and there were many parallel pathways identified, again pointing to the functional homology between murine and human MBC. Differences may have been due either to inherent biology or more likely to differences in the databases and annotated pathways used. As expected, B cell-specific pathways were identified; unexpectedly, in both species there was highly significant enrichment for “monocyte pathway”, although the exact proteins implicated related more to adhesion and migration. Indeed, pathways involving these latter two functions (e.g. KEGG “cell adhesion molecule” and “leukocyte transendothelial migration” pathways) were also highlighted in both species, indicative that MBC have distinctive cell-cell interaction and tissue penetration capabilities compared to NBC. Cell activation and proliferation pathways were also enriched among MBC, suggesting that they may be more poised toward both functions, or even partly engaged in both functions (despite being “resting” cells), compared to NBC.

Figure 8: Pathway enrichment analysis of differentially regulated markers between MBC and NBC revealed by LEGENDScreen^™ analysis.

The top 10 enriched pathways from MsigDB (canonical and Gene Ontology Biological Process (GOBP)) and CHEA (a database of transcription factor targets inferred from functional genomics) from enrichR are shown as part of a network. In total 30 pathways for human (pink nodes) and murine (gray nodes) are shown each. Significantly enriched (FDR P-values <0.2) pathways are indicated by bordered nodes. Canonical terms are shown as circles, GOBP terms are shown as triangles and CHEA as diamond nodes. CHEA annotations are aggregated at different levels of granularity (see Methods for details). For example, “IRF8 GC.B” corresponds to IRF8 targets in GC B-cells and “IRF8” corresponds to a merged set of IRF8 targets based on assays from any cell-type. Common enriched pathways between human and murine are show in the center of the network. Connections (edges) among pathways are shown if there are >33% shared genes between the connected nodes. Bolder edge lines represent higher fractions of genes shared between the pathway genesets. Enrichment was calculated based on one-sided Fisher’s Exact test (with all panel markers as background) followed by Storey’s Q-value false discovery rate (FDR) correction.

Discussion

Here we have used a new and comprehensive approach to measure expression of a large number of surface molecules on two closely related—yet crucially different—cell types: naive and memory B cells. We have done so in both mice and humans, providing a resource for work in these two important species, and at the same time allowing for an interesting comparison and contrast. We present a comprehensive analysis of which surface markers are meaningfully expressed by either NBC or MBC, which markers are differentially expressed, confirmatory analysis of key differentially expressed markers, dual marker staining schemes that effectively separate NBC and MBC using flow cytometry, analysis of marker expression in tissue-resident MBC, comparative analysis between mouse strains and humans; and, finally, pathway analysis that provides initial insight into functional differences between NBC and MBC in both mouse and human.

These datasets will provide immediate utility as a comprehensive resource for workers in a broad variety of immunological research across both mice and humans. One example includes study of vaccine and infection-induced immunity to pathogens. Generation of MBC is considered a key endpoint, but this is difficult to measure in many cases. Markers identified can be an important supplement to conventional CD27-based tracking in humans, especially since CD11a and CD200 can be used to identify CD27⁻ MBC. Subsets of MBC defined in related studies and amplified here may in the future provide correlates of protection or function in re-challenge ^{12, 23, 34}. In mice, to our knowledge, no marker that is expressed on the great majority of MBC, analogous to CD27 in humans, has been documented; here we discover and validate several such pan-MBC markers that will be useful for future investigations in murine systems.

In both humans and mice, these markers will also be useful in tracking and quantifying MBC and their subsets in conditions such as autoimmune or chronic inflammatory diseases. They will also be helpful for exploratory studies in contexts in which B cells are being appreciated as key players in predicting and possibly determining outcomes, such as in certain forms of cancer. In a related study, Glass et al.³⁴ recently presented a resource derived from a comprehensive CyTOF analysis mostly focused on peripheral blood lymphocytes, with a small sampling of lymph node, bone marrow and tonsil; spleen and gut were not examined, and the study was restricted to human.

Glass et al.³⁴, separated the peripheral blood MBC compartment into switched and non-switched cells for their analysis. Comparing median expression of non-switched MBC versus NBC Glass et al.³⁴, and we found CD79b, CD1c, CD48, CD298, CD29, CD24, CD45RB, CD27, CD97, CD206, CD183, CD43, CD63, CD54, CD84 and CD81 upregulated on MBC versus NBC. CD98 and CD147 were not present in the human LEGENDScreen^™ assay and CD82 didn’t make our cutoff for stronger expression in MBC compared to NBC. Comparing median expression of switched MBC to NBC Glass et al., and this manuscript both found CD298, CD29, CD49d, CD45RB, CD27, CD97, CD206, CD43, CD63, CD54, CD95 and CD62L were more highly expressed on MBC. Overall, both studies find a similar expression profile of MBC compared to NBCs. Differences in expression could be due to sensitivity of the utilized approach or differential expression on splenic versus peripheral blood lymphocytes. Our study greatly extends that of Glass et al.³⁴ by encompassing many more tissues and donors, as well as covering both murine and human MBC populations to enable direct inter-species comparison.

The suite of cell surface proteins that we have documented as expressed on either NBC, MBC or both, also provide important new information about functionality. While inevitably, the positive expression of many of these proteins has been documented individually or in small groups, our data provides a single resource that can be mined for functional pathways or particular receptors that connote function.

As an initial effort in this direction, we have analyzed pathways that are differentially over-represented in MBC vs NBC at the level of surface protein expression. These pathways suggest that MBC have functionally altered migration and adhesion characteristics. The identification that MBC differentially express molecules involved in activation and proliferation is not unexpected from direct functional data with such readouts^{3, 33, 35, 36}, or from gene expression data^{37, 38, 39, 40}; rather, our analysis confirms and extends the concept that MBC are more activatable or poised to proliferate by virtue of expression of set of surface proteins. The most striking differences have to do with pathways involving adhesion and migration, the markers for which also make up most of the monocyte pathway signature. These include multiple integrins, selectins, cell adhesion molecules (such as PECAM1), and CD44. These results predict very different cell-cell interaction and migratory properties for MBC ⁴¹, as well as the potential for tissue residency^{14, 42, 43, 44, 45}.

While in general in this work we have focused mainly on pan-expressed markers, the raw data shows bi- or multi-modal expression for a number of markers. These expression patterns imply phenotypic heterogeneity, the biological significance of which in many cases has yet to be determined. In the human system we recently focused on subsets defined by expression of CD45RB and CD69 ¹⁴, which are found in different locations in the body (for example with CD45RB+/CD69+ cells predominating in the gut)^{17, 46}.

We are currently working to characterize specific functional differences among these subsets and with respect to NBC. In mice, we defined phenotypic heterogeneity using CD80 and PD-L2 expression, as well as CD73^{27, 39}, and have shown both unique origins and unique functions for these subsets^{25, 47}; such subset heterogeneity has also been documented in multiple MBC systems in mice by other groups ^{45, 48,11,49}. Others have focused on CD73 as a functional memory marker ^{48, 50, 51, 52}. Our current comprehensive analysis has the potential to reveal substantially more phenotypic heterogeneity, the significance of which is currently unknown. Due to the requirements of running multiple markers together and on different types of MBC, to better define additional subsets will require substantially more work, which is ongoing.

A further aspect of our work is the evaluation of differentially expressed markers in various tissues. Accumulating research has demonstrated the presence of various types of tissue-resident MBC in both humans and mice^{14, 42, 43, 44, 45}. We found that a number of markers change their differential expression pattern when tested on MBC isolated from certain tissues. This was particularly apparent in the human studies, where some markers such as CD74 and CD370 were more expressed only in MBC from spleen and not other sites. In mice, the combination of CD180 and CD267 provided good separation in some tissues, but was notably less effective in spleen, lung and BM. This could be due to natural resident MBC in these tissues or of marginal zone B cells in spleen, which could have memory-like properties. CD267 in mice was differentially expressed in all tissues except gut, and it showed the most different expression in spleen. We recently reported that gene expression patterns in MBC resident in human gut are distinctive, even when comparing parallel subsets in both gut and spleen¹⁴. One could imagine that gut resident NBC also differ from their splenic counterparts. If gut resident NBCs have a more “memory-like” phenotype compared to splenic NBC (which we did not experimentally test), differences to gut resident MBCs would appear smaller. The presence of a tissue-specific signature of resident cells has been reported for both tissue-resident memory T cells as well as macrophages, and seems a universal feature of the immune system ^{29, 53, 54, 55}. As such, a comprehensive view of MBC will require more extensive testing of MBC isolated from various tissues, which remains a technical challenge.

A cardinal feature of this study is the profiling and comparison of both mice and humans in a parallel fashion. We found remarkable correspondence for the majority of markers that could be evaluated, when comparing spleen, our reference tissue. The shared markers and pathways must reflect a shared fundamental biology that defines the MBC as different from the NBC. Nonetheless, as has been known for some time, there are many markers with discordant expression between the species; our study provides a comprehensive assessment of what is, and is not, translatable between mouse and human. Data presented here will thus form a basis to analyze and understand how long-lived humoral immunity develops and is maintained. Further work may use these markers to further delineate phenotypic subsets that ultimately can be correlated with specific ontogenies and functions across both mouse and human.

Methods:

Mice, immunization and generation of murine MBC

All mice were maintained under specific-pathogen-free conditions and all animal experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee. C57BL/6 WT (CD45.2), CD45.1 congenic, RAG2^−/− and B1–8i^+/− mice were purchased from The Jackson Laboratory. OT-II^tg+ RAG2^−/− mice which carry a T-cell receptor specific for chicken ovalbumin⁵⁶ were ordered from Taconic. Use of B1–8i^+/− and B1–8i^+/− Jk^−/− genetically targeted BALB/cJ mice was as described^24,25. 6- to 12-week-old AM14^tg+ Vk8R^+/− CD45.1/2 Balb/cJ mice carry an irrelevant BCR specificity and cannot mount an endogenous immune response to NP immunization^24,56. These mice were adoptively transferred with 2 × 10⁵ NP reactive B cells from B1–8i^+/− Jk^−/− genetically targeted BALB/cJ mice and were immunized i.p. with 50 μg of NP-CGG precipitated in alum 24h later to induce MBC formation as recently described²⁴. Mice were analyzed 8–12 weeks later. To generate MBC in the C57BL/6 system mice were immunized i.p. with either 50μg NP-CGG or 100μg NP-KLH (Keyhole limpet hemocyanin) in Alum. To generate MBC in the OT-II adoptive transfer system we bred C57BL/6 OT-II^tg+ RAG2^−/− CD45.1/2 mice as recipients in our own facility. These mice were co-transferred intravenously with 2 × 10⁵ bead purified NP reactive B cells from C57BL/6 B18^+/− CD45.1 and 1 × 10⁷ bead purified CD4 T cells from C57BL/6 CD45.1 congenic mice. Recipients were i.p. immunized 24h later with 50ug NP-CGG and mice were analyzed 28 days later.

Acquisition of tissues from human organ donors

Human tissues were obtained from deceased organ donors at the time of organ acquisition for clinical transplantation through an approved research protocol and MTA with LiveOnNY, the organ procurement organization for the New York metropolitan area, as previously^14,57,58. The information on donors used in this study is presented in Supplementary Table 2. All donors were previously healthy, were free of cancer, and were serologically negative for Hepatitis B, C, and HIV. Isolation of tissues from organ donors does not qualify as “human subjects” research, as confirmed by the Columbia University internal review board.

Antigens, antibodies and detection reagents for murine analysis

For murine analysis: Chicken γ-globulin (CGG; Sigma-Aldrich) was haptenated with nitrophenyl (NP)-hydroxysuccinimide ester (Cambridge Research Biochemicals). NP-KLH was purchased from Biosearch Technologies. Allophycocyanin, Phycoerythrin were haptenated with nitro-iodo-phenyl (NIP) –hydroxysuccinimide ester. The haptenation ratios of NP or NIP to proteins were determined by spectrometry. NP₃₃CGG was used for immunizations. The following reagents were prepared and/ or conjugated in our laboratory: NIP_6.6-BSA-Alexa Fluor 700, NIP-APC, CD45.2 (BioLegend) Alexa Fluor 488, CD180 (BioLegend) Alexa Fluor 532, PNA (Vector Labs) biotin, PNA (Vector Labs) Alexa Fluor 488, CD38 (Biolegend) Alexa Fluor 594. Antibodies used and staining panels are in Supplementary Tables 5 and 6 respectively.

Antibodies and detection reagents for human analysis

The following reagents were conjugated in our laboratory: CD180 Alexa Fluor 700 and CD81 Alexa Fluor 594. Antibodies used and staining panels are in Supplementary Tables 3 and 4 respectively.

Murine cell preparation

Murine spleens were disrupted either in HBSS supplemented with 45U/ml DNAseI (Sigma) and 80U/ml CollagenaseD (Roche) using the OctoMACS™ Separator (Miltenyi Biotec) or by crushing between frosted glass slides in PBS/ 2% FCS/2mM EDTA. Red blood cell depletion of single cells was mediated by incubation in ACK (Lonza) solution as directed. Lymph nodes were crushed between frosted glass slides in PBS/ 2%FCS/ 2mM EDTA.

Magnetic bead based cell enrichment

Murine splenic single suspensions were incubated for 2min in ACK lysis buffer (Lonza) to deplete red blood cells. Cells were washed and adjusted to 2 × 10⁸ cells/ml in Stem Cell Buffer (SCB; 1xPBS, 2% FCS, 2mM EDTA) and incubated with 50μl/ml rat serum and 10μl/ml anti-CD16/CD32 antibodies (clone: 2.4G2) to block Fc-receptors. Double concentrated biotinylated antibody cocktails were added to achieve a final concentration of 1 × 10⁸ cells/ml. All antibodies were grown, purified, biotinylated and titered in our own laboratory. For CD19⁺ B cell enrichment the following biotinylated antibodies were utilized: anti-CD43 (clone S7), anti-CD11b (clone M1/70), aCD49b (clone DX5), anti-GR-1 (clone RB6–8C5), anti-CD8 (clone TIB105), anti-CD11c (clone N418) and anti-Ter119 (clone Ter119). For CD4 T-cell enrichment following biotinylated antibodies were utilized: anti-CD19 (clone 1D3), anti-CD11b (clone M1/70), anti-CD49b (clone DX5), anti-B220 (clone GK1.5) and anti-CD8 (clone TIB105). Cells were incubated for 15min at 4°C and then washed twice in SCB. Cells were resuspended at a concentration of 1 × 10⁸ cells/ml and exposed to 90μl Streptavidin Particles Plus (BD) per ml. After 4min incubation at 4°C cells were moved into magnets for 3min and negative fractions were poured off, washed twice, resuspended and counted. Cells were transferred i.v. in transfer buffer (for 50ml: 48ml PBS, 0.5ml 1M HEPES, 0.25ml Pen/Strep, 1.25ml ACDA) Purification led in general to over 95% purity of target cell populations⁵⁹.

Processing of murine Lamina propria (LP) and Peyer’s Patches (PP)

Intestines were flushed with PBS after harvesting and Peyer’s Patches were separated. Fat was removed and the intestine was then cut longitudinally, washed in PBS and further cut into smaller pieces. Tissue pieces were incubated in pre-warmed RPMI medium supplemented with 3% FCS, 5mM EDTA and 2mM DTT at 37°C for 20 min in bacterial shaker at 200 RPM. Tissue was then strained into PBS with 5mM EDTA and vigorously shaken to separate the intestinal epithelial lymphocyte and lamina propria layer. Small intestine pieces were rinsed with cold PBS to remove any residual EDTA and DTT. Small intestine pieces and PP were then minced finely and incubated horizontally at 37°C at 200 RPM for 15 minutes (LP) or 30 minutes (PP) in a bacterial shaker in RPMI supplemented with 800U/mL Collagenase D for (LP) or 400U/mL (PP) and 20 Kunitz U/mL DNAseI (LP) or 40 Kunitz U/mL (PP). Samples were then filtered and EDTA was added to stop enzyme activity. Samples were centrifuged and then subjected to flow cytometric procedures.

Processing of murine lung tissue

Lungs were harvested and disrupted by enzymatic digestion in HBSS buffer with 100U/mL Collagenase D and 50KU/mL DNAseI utilizing the gentleMACS Octo Dissociator (program m_lung_01_02; Miltenyi Biotech). Tissues were then further incubated at 37°C and subjected to another run in the gentleMACS Octo Dissociator with the program program m_lung_02_01. EDTA was added to a final concentration of 5mM, cells were filtered and washed with RPMI media supplemented with 2.5% FCS. Red blood cells were lysed with ACK (Lonza). Cells were washed and then subjected to flow cytometric procedures.

Flow cytometry

Flow cytometric analysis was performed essentially as described^56,57 and data acquired on an LSRII or Fortessa instrument (BD Biosciences). In some experiments as indicated, the Cytek^™ Aurora Cytometer (Cytek Biosciences) was used to acquire data. Prior to antibody staining murine cells were incubated with anti-CD16/CD32 Abs in staining buffer (SB; 1xPBS, 2% FCS, 2mM EDTA,) for 5min to block Fc receptors. Cells were exposed to Zombie NIR (BioLegend) or Ghost BV510 (Tonbo Biosciences) to mark dead cells. Cells were incubated with indicated antibodies then washed and fixed in 1% PFA before data acquisition.

Murine MBC were identified in adoptive transfer recipients as CD45.2⁺ CD19⁺ NIP⁺ live singlets. Endogenous NIP⁻ CD19⁺ live singlets of adoptive transfer recipients served as naive B cells in the LEGENDScreen^™ assay (gating strategy in Supplementary Fig. 1). For validation of surface markers (Extended Data Fig. 2) and data presented in Fig. 5 naive cells of CD45.1 congenic Balb/cJ mice were mixed into the same flow tube together with cells of adoptive transfer recipients. This allowed for the simultaneous detection of MBC (CD45.2⁺), their naive counterparts (CD45.1⁺) and endogenous B cells of transfer recipients (CD45.1⁺ and CD45. 2⁺) by their CD45 allotype mark (gating strategy in Supplementary Fig. 5). Cryopreserved human single cell suspensions of spleen (SP), blood (B), bone marrow (BM), lymph node (LN), Tonsil (T) and intestinal tissue (Gut) were thawed and stained as previously described⁵⁷. Used antibody staining cocktails are listed in Supplementary Table 3. Analysis of stains “LS”, “Stain 1” and “Stain 2” were performed on a BD LSRII with subsequent data analysis in FlowJo v.9 (Becton Dickinson), “Stain 3”, “Stain 4” and “Stain 5” were acquired on a Cytek^™ Aurora Cytometer (Cytek Biosciences) and data analysis was performed in FlowJo v.10 (Becton Dickinson).

Murine LEGENDScreen^™ assay

For the LEGENDScreen^™ (BioLegend, Cat. 700005) assay, spleens of 28 adoptive transfer recipients (see above) were harvested and processed as described in the flow cytometry section. Single cell suspensions were exposed to Ghost BV510 reagent to mark dead cells and then stained against antigens indicated in Fig. 1 with stain “Murine Stain 1” (Supplementary Table 6). Cells were washed in staining buffer and 4 × 10⁶ splenocytes were transferred into each well of LEGENDScreen^™ plates, which were reconstituted according to the manufacturer’s instructions. Plates were incubated for 45min on ice. Cells were washed and fixed in 1% PFA until acquisition on a BD LSRII. Raw data have been deposited and are publically available at http://flowrepository.org with the accession ID: FR-FCM-Z4LQ.

Human LEGENDScreen^™ assay

Cryopreserved splenocytes from three donors (D182, D185 and D186) were thawed as previously described⁵⁷. For violet proliferation dye (VPD) “barcoding”, cells from D185 were exposed to 6μM VPD, cells from D182 to 20μM VPD and cells from D186 to 0μM VPD in VPD-staining buffer (VPD-SB; HBSS w/o Ca/Mg, 0.25% FCS, 2mM EDTA).VPD labeling was quenched after 12min by adding 3 volumes pre-warmed RPMI media supplemented with 20% FCS for 1min. Cells were washed and then subjected to Fc receptor block by exposure to 10% human plasma in staining buffer for 5 min on ice. After a washing step in staining buffer, equal numbers of cells of individual donors were pooled and 1.5 × 10⁶ total cells were transferred into each well of the plates of the LEGENDScreen^™ Human Cell Screening Kit (BioLegend Cat. 700001), which contained 25μl antibody solution with one antibody specificity per well. The plates were prepared according to the manufactureŕs instructions and stored at 4°C. After incubation for 30min on ice, plates were washed with staining buffer and pellets were resuspended in a staining cocktail containing antibodies against CD19-APC-Vio770, CD27-BUV737, CD38-BUV395, CD138-APC, CD11c-FITC and IgD-PE-Cy7 for 30min on ice. Cells were washed and then subjected to 15 minutes viability staining at RT using Ghost BV510 reagent (Tonbo; Cat. 13–0870). Cells were washed and fixed in 1% PFA for 30min before analysis on a BD LSRII. Data analysis was performed in FlowJo v.9 (Treestar). Raw data have been deposited and are publically available http://flowrepository.org with the accession ID: FR-FCM-Z4LS.

Pathway enrichment analysis

Pathway enrichment was performed using Fisher Exact test (with all panel markers as background) followed by Storey’s Q-value false discovery rate (FDR) correction. Different gene sets for Canonical and Gene Ontology Biological Process were taken from mSigDB (“https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp”) database and CHEA genes set from EnrichR (https://maayanlab.cloud/Enrichr/). CHEA genesets are annotated with transcription factor (TF), cell-type, and organism describing the corresponding ChIP-X experiment. We created two sets of merged annotations corresponding to TF, cell-type pairs (organisms merged) and TFs (cell-type and organism merged). Only those pathways are tested where at least 5 genes are matching from background gene list. Network plot was build using Cytoscape (version 3.5.1).

Statistical Analysis

Graphpad Prism v.8 and v.9 software was used for statistical analysis of flow cytometric data. For group comparisons unpaired two-tailed t-test with Welchś correction was used as described in figure legends. P values are shown as * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, ns = not significant. Error bars are mean ± SD.

Pathway enrichment was performed using Fisher Exact test (with all panel markers as background) followed by Storey’s Q-value false discovery rate (FDR) correction using R (version 3.6.1).

Extended Data

Extended Data Fig. 1. Differentially expressed surface markers on human and murine naive and memory B cells, in support of Fig. 3.

Differentially regulated surface markers as in Fig. 3 are sorted based on the ratio of the mean fluorescence intensity (MFI) of MBC to NBC. a, MFI ratio of MBC to NBC in humans (mean of three donors 182, 185, and 186). b, MFI ratio MBC to NBC in mice. Red and blue dots depict higher expression on MBC or NBC, respectively. Plot was built using ggplot2 (version 3.2.1) in R (version 3.6.1).

Extended Data Fig. 2. Validation of differentially regulated surface markers on murine naive and memory B cells, in support of Figs. 3 and 4.

Histograms of flow cytometric expression of 24 depicted surface markers on MBC (red) and NBC (blue). MBC were identified as CD45.2⁺ NIP⁺ CD19⁺ live singlets on splenocytes of transfer recipients (7 females, 39 wks old, 31 wks post immunization plus 3 males, 34 wks old, 23 wks post immunization) and splenic B cells of CD45.1 naive B1–8i^+/− Balb/cJ mice (3 males 11 wks old plus 2 males, 9 weeks old) were mixed into the staining to serve as NBC, identified by their CD45 allotype mark. Cells were stained with “Murine Stain 3” (Supplementary Tables 5 and 6). The fluorescence minus one control FMO for the PE-channel is shown in the bottom ri ht histo ram.

Extended Data Fig. 3. Validation of differentially regulated surface markers on human naive and memory B cells in support of Figs. 3 and 4.

Cryopreserved human splenocytes were stained with either “Stain 2”, “Stain 3”, “Stain 4” or “Stain 5” (Supplementary Tables 3 and 4) for flow cytometric analysis. Shown are histograms of the expression of 16 depicted surface markers on MBC (CD19⁺ CD2J⁺, red) and NBC (CD19⁺ CD27⁻, blue). All markers were validated on 3 individual donor spleens listed in the tables below the histo rams.

Extended Data Fig.4. Combinations of surface markers that allow for the identification of murine MBC and NBC in the C57BLl6 background in support of Fig. 5.

Direct immunization (left panel): Single cell suspensions of indicated tissues from 3 male C57BLl6 wt (CD45.2, 12 wks old, 4 wks post immunization) mice were analyzed at day 28 post i.p. immunization with 100l-1g NP-KLH. 5 × 10⁶ cells were mixed with 1 × 10⁶ cells of corresponding tissues of 1 naive male C57BLl6 CD45.1 congenic mouse (8 wks old) to allow for the simultaneous identification of MBC and their comparable naive counterparts in a single staining tube. MBC and NBC were identified as described in Supplementary Fig. 8a and displayed data are concatenated of 3 individual samples. OTII adoptive transfer system (right panel): Single cell suspensions of indicated tissues from 5 individual OTII adoptive transfer recipients (males, 12 wks old, 4 wks post immunization) were analyzed at day 28 post i.p. immunization with 50l-1g NP-CGG. 5 × 10⁶ cells were mixed with 1 × 10⁶ cells of corresponding tissues of 1 male naive C57BLl6 wt (CD45.2) mouse (7 wks old) to allow for the simultaneous identification of MBC and their comparable naive counterparts in a single staining tube. Cells were stained with “Murine Stain 5” (Supplementary Tables 5 and 6). MBC (red) and NBC (blue) were identified as described in Supplementary Fig. 8b and displayed data are concatenated of 5 individual samples. Shown are contour plots of pairwise combinations of CD2051 CD274, CD811 CD11 a and CD2671 CD180 as in Fig. 5, which can be used to distinguish MBC and NBCs across tissues. MLN, mesenteric lymph nodes; BM, Bone Marrow.

Extended Data Fig. 5. Differentially regulated surface markers on naive versus memory B cells across human tissues, in support of Fig. 6.

Single cell suspensions of spleen (SP), blood (B), bone marrow (BM), lymph node (LN), intestinal tissue (Gut) and tonsil (T) were stained for flow cytometric analysis using “Stain 2” and “Stain 4” (Supplementary Tables 3 and 4). The left panel shows the summary of the differences in MFI between CD19⁺ CD27⁺ B cells and CD19⁺ CD27⁻ B cells (Δ MFI) for the depicted surface markers. For CD74 and CD119 spleen (n=23 in red), blood (n=7 in blue), BM (n=9 in magenta), LN (n=6 in green), gut (n=3 in brown) and tonsil (n=2 in black) samples were analyzed. For CD218a spleen (n=21 in red), blood (n=10 in blue), BM (n=10 in magenta), LN (n=8 in green), gut (n=8 in brown) and tonsil (n=2 in black) samples were analyzed. For CD370 spleen (n=21 in red), blood (n=10 in blue), BM (n=10 in magenta), LN (n=8 in green), gut (n=8 in brown) and tonsil (n=2 in black) samples were analyzed. The right panel shows example histograms for depicted surface markers of CD19⁺ CD27⁻ B cells (blue) and CD19⁺ CD27⁺ B cells (red) across tissues of donor D260. Stars indicate significant differences in ΔMFI of indicated tissues compared to spleen using the unpaired two-tailed t-test with Welch’s correction. *** p< 0.001, **** p< 0.0001. Exact significant p-values for comparison between spleen and the indicated tissue for each marker are for CD74: all tissues p<0.0001; for CD119 blood, bone marrow and gut p<0.0001, lymph node p=0.0006 and tonsil p=0.1112; for CD218a: blood and bone marrow p<0.0001, gut p=0.0001, tonsil p=0.1047 and CD370: blood, lymph node, gut and tonsil p=0.0002 and bone marrow p=0.0003.

Extended Data Fig. 6. Expression of surface markers CD11a and CD200 on human splenic MBC and NBC separated by Ig isotype in support of Fig. 7.

Splenic single cell suspensions were stained for depicted markers (“Stain 5”, Supplementary Tables 3 and 4). Overlayed histograms for expression of CD11a (upper panel) or CD200 (lower panel) of either total CD27⁻ and CD27⁺ (first row), or specific Ig isotypes for CD27⁻ overlayed with total CD27⁺ (row 2 CD27⁻ IgM/D; row 3 CD27⁻ IgG; row 4 CD27⁻ IgA) are shown for 6 individual donors (D192, D215, D228, D333, D365, D388). CD19⁺ CD27⁺ MBC are in red and CD19⁺ CD27⁻ NBC are in blue. The last rows of each panel show a summary of the CD27⁻ isotypes analyzed (IgM/D green; IgG blue; IgA orange) in direct comparison with the total CD27⁻ B cells (red).

Data availability

All data are available upon request and relevant data are available in the Source Data. Flow cytometry files of LegendScreen^™ assays are deposited and are publically available at http://flowrepository.org with the accession ID: FR-FCM-Z4LQ (murine LegendScreen^™) and with the accession ID: FR-FCM-Z4LS (human LegendScreen^™).