|Home | About | Journals | Submit | Contact Us | Français|
B lymphocyte involvement in systemic lupus erythematosus has been recognized for several decades, mainly in the context of autoantibody production. Both mouse and human studies reveal that different types of antibody responses, as well as antibody-independent effector functions can be ascribed to distinct subpopulations (subsets) of circulating B cells. Characterizing human B cell subsets can advance the field of autoimmunity even further by establishing B cell signatures associated with disease severity, progression, and response-to-treatment. For this purpose, we have developed specialized B cell reagent panels for multiparameter flow cytometry, and combine their use with advanced bioinformatics strategies that together will likely be advantageous for improving the characterization, prognosis, and for possibly improving treatment regimens of chronic inflammatory diseases such as lupus.
Systemic lupus erythematosus (SLE) is a chronic inflammatory autoimmune disorder whose clinical complications include skin rashes, joint pain, and in more severe cases, kidney and nervous system pathology (1, 2). Due to the enormous heterogeneity of symptom types, severity, kinetics, and response-to-treatment among patients (3), establishing reliable biological indicators associated with different outcomes (biomarkers) is essential for improving diagnosis, for accuracy of prognosis, and for customizing therapies. Both T and B lymphocyte abnormalities are observed in SLE, including elevated levels of self-reactive antibodies that precede disease onset (4). The importance of B cells in SLE and the phenotypic diversity of human B cell subsets strongly suggest that phenotypic profiles of B cells could be such a biomarker (5). Herein, we review B cell phenotypic diversity in the context of SLE and also describe reliable methods developed by our group to measure B cell changes in human peripheral blood using high-parameter flow cytometry combined with advanced bioinformatics.
Antibodies (immunoglobulins, Ig) are expressed on the surface of and secreted by B cells. Each B cell clone expresses an antibody [B cell receptor (BCR), if surface membrane-bound] that is unique in its variable (V) region, and whose constant region is one of nine effector isotypes. Antibodies are designed by each B cell through DNA recombination of Ig V region gene segments, by somatic hypermutation of these assembled V region genes, and by isotype (class) switching, which changes the Ig constant region to modify its effector function.
During B cell development, Ig DNA rearrangements create a diverse BCR repertoire encoding a wide range of antigen-binding capabilities. Self-reactive B cells are usually deleted or inactivated at developmental checkpoints, which are defective in SLE patients (6–11). Common autoantibody reactivities include nuclear components and plasma membrane inner leaflet phospholipids (12). This reactivity may be attributed to poor clearance of apoptotic cells and debris in SLE patients (13,14), which can allow for exposure to such internal antigens (15,16). Thus, B cell tolerance might be broken by extended exposure of these antigens stimulating self-reactive B cells, which in turn promote inflammation through effector functions such as antigen presentation to T cells, cytokine secretion, and/or pathogenic Ig secretion.
Despite significant technical advances in cell sorting, molecular cloning, and re-expression of Ig-encoding cDNAs as recombinant monoclonal antibodies, the use of anti-idiotype reagents recognizing self-reactive BCRs allows highly efficient characterization of intact B cells by histology and by flow cytometry. One of these reagents is a rat anti-human monoclonal antibody called 9G4. 9G4 binds to the V region of BCRs or of soluble antibodies (denoted respectively, 9G4+ B cells or 9G4+ antibodies). Up to 10 % of healthy-human B cells are 9G4+, but are largely confined to the antigen-naïve, undifferentiated B cell compartments, maintaining very low-to-undetectable serum 9G4+ antibody levels (17). However, SLE patients often have high titers of 9G4+ antibody that correlates with disease activity (18–20), with 9G4+ IgG levels correlating more strongly with disease (especially nephritis) than 9G4+ IgM levels (20).
High-titer autoantibody in SLE has prompted interest in developing biologicals that target B cells (3). Inhibiting the B cell survival factor BAFF/Blys (B cell Activation Factor of the TNF Family/B Lymphocyte Stimulator) or depleting B cells can alleviate symptoms of lupus-associated disease in humans and mice (21–28). These outcomes strongly suggest that B cells actively contribute to disease progression, although the exact mechanisms are not known. Interestingly, both mouse and human data suggest that high-titer autoantibody alone does not account for the B cell-mediated effects (5, 22, 27, 29). In fact, B cells are important for memory T cell accumulation in autoimmune mice, suggesting mechanisms such as antigen presentation and/or cytokine production by the B cells (29).
However, the B cell subsets responsible for each specific function are incompletely understood. Thus, in subsequent sections, we discuss how B cell characterization could be further exploited for understanding this disease and for optimizing clinical benefit (5, 21, 30)
In adult bone marrow, nascent B cells begin expressing cell-surface BCR. In a “transitional” stage, the BCR+ B cell enters the peripheral circulation and differentiates into a mature-naïve cell that is competent for responding to BCR engagement by antigen. In addition to transitional and mature-naïve B cells, human blood contains 3–4 additional “core” B cell subsets with characteristics of antigen exposure (Ig somatic hypermutation, Ig class switching, and/or markers associated with Ig secretion). Such B cells that are neither naïve nor secreting antibody are generally regarded as “memory” in most nomenclatures (31) (Fig. 1b, c), which are discussed below. All of these basic core CD19+ populations can be identified using flow cytometry by various combinations of IgD BCR (more naïve) and CD27 (more differentiated) surface expression (Fig. 1c) (31). Further subsetting these core populations using additional markers is a highly advantageous way to characterize human B cells (31).
The CD10 marker on human BCR− B cell progenitors continues to be expressed on immature B cells that have acquired the first class of BCR, IgM, as well as on IgM+ transitional B cells coexpressing an IgD BCR with an identical V region. In healthy subjects, the frequency of autoreactive B cells is reduced when the BCR reaches the cell surface (6), and then again when CD10 is lost (7). At each step, more of these B cells are self-reactive in SLE patients compared with healthy subjects (7,8). Despite breaching these tolerance checkpoints, SLE patients have decreased numbers of naïve B cells contributing to overall reduced numbers of total CD19+ B cells (13, 22, 32, 33). The cause and significance of this reduction is poorly understood.
Given the critical involvement of the transitional-to-mature conversion in determining the SLE B cell repertoire, it is advantageous to define such cells and their subsets in clearer detail. This goal can be achieved using multiparameter flow cytometry that can be efficiently executed in the laboratory with less than 10 cc of blood, and also allows for isolating such subsets for functional analyses. Transitional and mature-naïve B cells are CD27−. In addition to CD10, which typically has a poor staining index, the earliest transitional populations coexpress high levels of CD38 and CD24. “T1” transitional cells express slightly higher CD24 and CD38 than “T2” cells, their apparent progeny; however, visualizing this distinction is difficult without overlaying a pro file of bone marrow, which lacks T2 B cells. Compared with T1/T2 transitional B cells, mature cells express lower levels of CD38, survive longer in culture, and divide in response to BCR engagement (34). Truly mature naïve B cells can extrude dyes such as Rhodamine 123 using the mitochondrial ABCB1 transporter, which is absent from both transitional and memory B cells (34). This characteristic is especially useful in separating true-mature (mature-naïve) B cells from an additional transitional population (T3). T3 cells express lower levels of CD24 and CD38 than T1/T2, but all three transitional populations fail to express active ABCB1 (35). Thus, without a dye-extrusion step, the T3 population is not distinguished from true-mature B cells. It remains to be definitively determined whether T3 is an obligate precursor to true-mature B cells. Nonetheless, such phenotypic distinctions may be clinically informative, as preferential early reconstitution of transitional B cells correlates with long-term remission in B cell-depleted SLE patients (once treatment is stopped) (30). Thus, thorough characterization of transitional and mature-naïve B cell subsets may contribute to defining more precise prognoses and perhaps predicting response to treatment.
Antigen engagement of the BCR on a mature-naïve B cell can stimulate activation and differentiation along any of several pathways that, ideally, help eliminate an invading organism and protect the host. One of these pathways is the germinal center (GC) reaction in secondary lymphoid tissue follicles. Typically, GC reactions require T cell help to promote B cell proliferation, BCR class switching from IgM/IgD to IgG, IgA, or IgE, and also Ig V region gene somatic hypermutation that changes the affinity of the encoded BCR for its cognate antigen (36). Somatic mutations in autoantibodies contribute to their self-reactivity (11,37,38). Antigen-driven selection results in clonal expansion of mainly the highest affinity B cells. In normal individuals with intact peripheral tolerance, these B cells are ideally only reactive with antigens from invading organisms. The surviving cells can differentiate into antibody-secreting plasma cells or into memory B cells (each discussed below) that are ready for a rapid response to a second encounter with an invading organism.
In healthy human tonsil, non-self-reactive B cells are evenly distributed among the B cell populations, but self-reactive 9G4+ B cells are confined to the naïve compartment and excluded from GCs (17, 39). Unlike in healthy humans, 9G4+ B cells in SLE patients colocalize with proliferating CD38+ GC structures (39). These observations strongly suggest that dysregulation of autoreactive B cells at GC differentiation stages contributes to SLE disease (40), and may explain the ability to detect IgG+ 9G4+ cells and 9G4+ CD138+ (plasma) cells in SLE peripheral blood, each of which are rare in healthy controls (39).
B cell memory (specific and rapid recall responses to previously encountered antigens) is often inferred from cells restimulated in vitro and from the ability to transfer such memory to naïve experimental animals (41–43). In humans, most of this activity is attributed to CD27+ B cells. Compared with CD27− (mostly mature-naïve and transitional) B cells, CD27+ B cells are slightly larger (34, 44), are more proliferative in vitro (45–47) and in vivo (48), more efficiently stimulate allogeneic CD4 T cell proliferation (46), and more readily differentiate into antibody-secreting cells (47). GC, memory, and in vitro-activated naïve B cells can all be CD27+. However, as GC are generally confined to lymphoid tissues, and few circulating CD27+ B cells show evidence of on-going proliferation (34), GC and recently activated B cells unlikely account for a significant portion of the CD27+ B cell pool, at least in healthy humans. Additionally, CD27+ B cells are barely detectable in natal cord blood, but proportionally increase with age and cumulative antigen exposure (44, 48–50). The few CD27+ B cells in the cord blood lack Ig V somatic hypermutations and have been attributed to a newly described innate-like human “B1” B cell equivalent (51).
Adult CD27+ B cells either (1) express an IgD/IgM BCR [IgM memory or nonswitched memory (NSM) B cells] or (2) have lost expression of these markers after the irreversible process of antibody class switching (switched memory B cells). CD27+ B cells are ~10–30% each IgM+ D+, IgG+, or IgA+ (34, 52–54). It is unknown whether these memory B cell subsets derive from common or from distinct differentiation pathways, or even if such pathways are common to all cells in a given pool. Thus, the switched memory B cell pool may contain cells that class-switched prior to and also those that class-switched after acquiring CD27. Similarly, the NSM pool may contain B cells that will remain NSM together with intermediates that will eventually class-switch. Preliminary analysis using high-parameter flow cytometry combined with automated analysis examining all parameters simultaneously (55) suggests that most NSM cells have more in common with other compartments than they have with each other (unpublished). However, most studies to-date have compared all NSM with all switched memory B cells using conventional manual gating methods. Briefly, both compartments have somatically hypermutated Ig V regions with characteristics of antigen selection (11, 36, 54, 56). However, reports disagree on whether NSM have a unique (33, 57, 58) or similar (59) Vh repertoire compared with IgD− CD27+ B cells. It is unknown whether sorting and/or sequencing strategies account for these different observations.
The NSM pool is thought to consist of adult B1 B cells, regulatory B cells, and splenic marginal zone-like B cells that provide carbohydrate-reactive antibody responses to encapsulated bacteria (38, 50, 51, 60, 61). Recent studies in mice suggest that at least some of the IgM memory pool functions to maintain very long-term memory against protein antigen (62, 63).
Mice and humans with defective GC have a paucity of memory B cells and profoundly decreased levels of class-switched serum antibody (43). These observations led to the dogma that class-switched humoral memory is GC-derived, whereas IgM memory is GC-independent. Indeed, development of NSM B cells with mutated Ig V regions does not require GC formation (64). However, some NSM have mutations in a noncoding region of the Bcl6 gene, which is characteristic of GC participation (59). Although antibody class switching is also characteristic of GC reactions, not all switched cells are necessarily GC-derived. A small proportion of CD27+ IgG+ cells express the IgG2+ subclass (34, 65), which can be induced independently of GC-type (i.e., T cell-derived) stimuli in culture (66).
Some pediatric SLE patients have more IgG3-expressing and fewer IgG2-expressing switched memory B cells compared with healthy controls (65). Given the contribution of IgG2 to anti-polysaccharide responses, this observation may suggest an over-representation of GC-derived memory against protein-type antigens. More notably, clinical relapse of B cell-depleted patients correlates with rapid reconstitution of CD27+ memory B cells (30). It is not clear if this effect is due to the functions of the memory B cells or due to pathological circumstances favoring memory over transitional B cell expansion. Interestingly still is the recurring observation that NSM are proportionally reduced in active SLE (32, 67, 68), correlating with higher autoantibody titers (67). Together with the stronger clinical correlation between serum 9G4+ IgG compared with 9G4+ IgM (20), this result may suggest that some of the IgM memory may play a protective role against pathology otherwise exacerbated by IgG memory. Increasing evidence supports such a dichotomy of protective and pathogenic B cells in cardiovascular disease (69–73).
A small proportion of class-switched B cells in healthy subjects do not express CD27 (52, 53). These B cells make up most of the CD19+ IgD− CD27− “double-negative” (DN) compartment, which is typically less than 5 % of the CD19+ pool. Most DN B cells in healthy subjects express either IgG or IgA (52). Some evidence suggests that the DN compartment in unchallenged individuals includes B cells that have further differentiated into memory B cells without acquiring CD27. Both CD27+ and CD27− pools contain B cells that can be readily induced to secrete Ig against tetanus toxin and influenza virus (34), suggesting prior exposure to vaccines. Other than CD27 expression, the DN and CD27+ switched memory B cell populations have similar surface phenotypes (34, 53) and have somatically mutated Ig V regions (52, 53). However, Ig V region mutations in DN B cells are less frequent than in switched CD27+ memory cells (52, 53). The DN compartment may thus be a mixture of both effector and memory B cells.
To varying degrees, SLE patients have increased frequencies of DN B cells (13, 22, 52) correlating with higher autoantibody titers and with a greater incidence of nephritis (52). The higher frequency of DN cells could result from B cells entering from other compartments, such as the naïve B cell pool becoming activated and losing IgD expression, or from activated cells entering blood from the tissues. Further investigation of the characteristics, origins, and functions of DN B cells in treated and untreated SLE patients, and in newly vaccinated and acutely infected controls will likely contribute to our understanding of B cells in this disease and in B cell activation in general.
Differentiation into an antibody-secreting cell (ASC) can occur downstream of the appropriate B cell-activating stimuli. ASC in human peripheral blood are typically CD27high CD38high, and nearly all express the intracellular Ki67 antigen, suggesting on-going proliferation (34). ASCs detected by Ig secretion assays or by flow cytometry are very rare in the blood of healthy, unchallenged individuals. However, a rapid and transient ASC response is readily detectable as early as 4 days after booster vaccination (55, 74– 76). During acute infection (especially viral), ASCs can be detected as early as day 2 after symptom onset (77). Specificity of the circulating ASCs is unique to antigens of recent exposure (94). Human and mouse studies have identified at least two “subsets” of blood ASC loosely referred to as plasmablasts and plasma cells (78, 79). True, terminally differentiated plasma cells might be enriched in the compartment expressing CD138, which is less than half of the ASC in the blood, but nearly all of the ASC in the bone marrow (78, 79). Compared with CD38high CD138− ASC, considered to be plasmablasts, CD138+ plasma cells are larger with more cytoplasm, lack surface Ig, and do not seem to be undergoing proliferation (34, 80). As with the various kinds of memory B cells, it is not entirely clear whether some or all plasmablasts and plasma cells have precursor–product relationships, even in elegant mouse reporter systems (78, 79). However, given evidence that long-lived specific antibody in the serum can be maintained without concurrently circulating memory B cells (42, 81), a model has arisen in which long-lived, CD138+ plasma cells in tissues (such as bone marrow, where they are most readily detectable) provide a long-lived Ig source, whereas plasmablasts in the circulation provide an immediate, but transient “boost” to existing serum Ig levels. This model does not favor nor exclude that plasmablasts and memory B cells (from either tissues or circulation) could each progress to the plasma cell stage upon migrating to the bone marrow.
ASC detected by functional secretion assays and by flow cytometry are proportionally increased in SLE patients. However, reports differ on whether this effect includes patients with quiescent and those with moderate disease (32, 67, 68, 82) or if it correlates with disease activity (22, 33). The differences are not accounted for by the choice of flow cytometry markers per se, and thus may instead include disparities in patient groups biologically, in clinical assessments, and even in technical strategies including inconsistent resolution and/or event-gating of these rare populations. Biologically, the contribution of antibodies compared with antibody-independent B cell functions might differ among patients or patient groups. B cell depletion and CD40L blockade can each reduce plasmablasts in SLE patients, correlating with symptom reduction (13, 22). However, in only some of these patients do serum autoantibody levels decrease. Thus, plasmablast expansion in SLE could possibly be more of a consequence of systemic inflammation, rather than a cause.
We have thus far discussed the tendency of SLE patients to have proportional increases in CD19+ IgD− CD27− DN B cells (which may become the largest switched memory subset) and in plasmablasts, but decreased proportions of NSM B cells. Previous studies and our own observations suggest that including additional cell-surface markers will provide additional advantages in further characterizing these populations in lupus. As previously reported (31), CD27+ resting memory cells in healthy subjects are predominantly B220− (Fig. 1d) as the expression of this marker is typically lost during germinal center differentiation. By contrast, B220 expression predominates in IgD− CD27− cells. This subset is characterized in SLE by the downregulation of CD24, a marker typically expressed by most PBL B cells in general and memory B cells in particular (see Fig. 1d for healthy subject example). Consequently, expanded fractions of B220+ CD24− cells within the IgD− CD27−compartment are prominent in lupus patients. Loss of CD21 and upregulation of CD95 have been independently associated with memory B cell activation (83, 84). Accordingly, an indication of memory B cell activation in SLE can be found in both SwMe and DN cells in which the CD21− and CD95+ fractions are greatly expanded in lupus compared to the healthy controls. By contrast, CXCR3 expression (suggesting migratory potential of activated cells to nonlymphoid, systemic inflamed tissues) is concentrated in the IgD− CD27+ memory subset. Therefore, the inclusion of CD21, CD95, and CXCR3 in the same reagent panel determines the coexpression of these informative functional markers. Consistent with a resting phenotype, most IgD− CD27+ memory cells are CD21+ CD95− in healthy controls (Fig. 1d). A significant fraction of IgD− CD27− cells, however, lack expression of CD21, a feature consistent with activation. Yet, they lack CD95 expression, indicating that these two markers are not necessarily correlated. Combined, CD95+ cells are relatively scarce in healthy memory cells. By contrast, CD95+ cells are greatly expanded in both memory subsets of SLE patients. Interestingly, the vast majority of CD95+ IgD− CD27− cells are CD21−. CD95+ cells within the lupus IgD− CD27+ memory are almost equally split between CD21+ and CD21−, illustrating again that the expression of these markers is not necessarily reciprocal. Furthermore, the CXCR3+ fractions of both IgD− CD27+ and IgD− CD27− memory subsets exhibit CD21/CD95 expression patterns that are similar to those of their respective total subsets and are not therefore, particularly enriched in activated cells, at least as defined by these markers.
Given the enormous heterogeneity in SLE disease presentation, and the contributions that B cells make to pathology, further defining subpopulations of these core subsets using the markers described above can establish “signatures” or patterns of changes in B cell populations that may be associated with different clinical outcomes. Such profiling to find B cell signatures could also help prognosis and custom-designed treatments. Essentially, the appropriate modality would ideally eliminate B cells harboring pathogenic properties, while it would enhance the numbers and/or activity of B cells with regulatory properties (5, 21). However, certain signatures may also be predictive of whether more classical therapies are or are not going to be useful in a given individual.
Here, we describe procedures for analyzing B lymphocyte subsets from human peripheral blood mononuclear cells (PBMC). We include methods for preparing, staining, and analysis by flow cytometry followed by bioinformatics analysis. As an example, we focus on our recently developed 12-color memory B cell staining panel (95) analyzed with a Becton-Dickson LSR II instrument. The protocol assumes basic knowledge of operating the instrument. Similar panels have been developed for transitional B cells and plasma cells, and these methods can be amended for other user-developed panels as well. For consistency in large studies, we provide detail on critical steps such as freezing and thawing, with recommended quality assurance/quality control to monitor and regulate consistency (Fig. 2).
Another integral element needed for consistent and reproducible flow results is a standardized strategy for flow data analysis. A rationalized gating strategy can significantly reduce user-to-user variability to ensure consistent interpretation of the data among all operators (e.g., Fig. 1). The following protocol assumes basic knowledge of FlowJo software. Tutorials on getting started with FlowJo are available at the TreeStar Web site (http://www.flowjo.com/home/basictutorial.html) and can also be arranged with TreeStar personnel.
Once primary analysis (gating) has been performed and crosschecked, the resulting data can be analyzed in numerous ways, depending on the experimental question. Considerations include whether to report the data as mean fluorescent intensity of a particular fluorescent parameter, the percentage of cells within a parent population, as a percentage of all B cells, or as the “absolute” cell number per volume of blood. For cell-number calculations, it is necessary to acquire a clinical blood count (CBC) from an aliquot of whole blood in which total white blood cells and lymphocytes are measured. These procedures can be requested from clinical staff or performed by the analyst using conventional or automated methods, such as a Sysmex XE-2100 instrument. Multiparameter flow analysis invariably generates large quantities of data that need to be managed efficiently in order to derive productive output. For large-scale studies with associated clinical information, it is recommended that the data be structured and housed in a relational database to support flexible interrogation of the data.
In some circumstances, many reagents in the panel are used for narrowing down a particular subset of interest, which is then analyzed for various characteristics and by standard statistical tests with conventional graphing techniques. In other circumstances, a more global pro filing approach can be used in which all of the data are considered together to obtain a system-wide view of immune cell populations. Profiling is useful for class-discovery in which natural groupings of samples can be sought (85, 86). Furthermore, it may be possible to relate those groupings to sets of samples with different disease states, disease activities, or clinical manifestations. Univariate approaches on individual subsets fail to reveal how collections of subsets and their relative distributions might contribute to such sample groupings.
The following represent a generic workflow that can be performed by several software packages or by technical computing platforms. This protocol assumes basic knowledge of Matlab software (Mathworks, Natick MA; http://www.mathworks.com/products/matlab/).
We thank John Jung and Ravi Misra for reading the manuscript and members of the Sanz lab for help and advice. Supported by NIH R01 AI049660-01A1 and U19 Autoimmunity Center of Excellence AI56390 to I.S.