Efficient Culture of Human Naive and Memory B Cells for Use as APCs

B cells are key to adaptive immunity and are now recognized for their multifunctionality; B cells not only produce Abs, they also present Ags to T cells (1), secrete cytokines (2), and regulate other immunocytes (3). Ag presentation by B cells is involved, to a significant extent, in immunoprotection and the pathogenesis of autoimmune diseases (1, 4, 5). The effects of Ag presentation by B cells on T cells depend on the activation state of B cells. Studies show that CD154- or mitogen-activated B cells function as effective APCs to induce T cell activation (6, 7), whereas resting B cells are tolerogenic (8).

The Ag-presentation function of B cells has long been known (9, 10), and B cells are recognized as professional APCs along with dendritic cells (DCs), macrophages, and thymic epithelial cells (11). Ag-presenting B cells participate in the initiation and continuation of autoimmune diseases, such as systemic lupus erythematosus (12, 13), rheumatoid arthritis (14, 15), type 1 diabetes (16), and multiple sclerosis (5), in humans and mice. Beyond the scope of autoimmunity, B cells serving as APCs are characteristic of atherosclerosis (17), insulin resistance (18), allergy (19), allorejection (20), infection, and even immune responses elicited by vaccination (21).

On the whole, professional APCs initiate adaptive immune cellular responses by processing and presenting Ags to T cells, as well as by providing costimulatory signals necessary for the activation of T cells. These functional properties of APCs were applied in the clinical assessment of T cell responses in vitro; for example, to evaluate the efficacy of vaccination (22), to identify the causal allergens for patients (23), and to predict the compatibility of allografts (24). Generally, autologous APCs are loaded with target Ags and are cocultured with T cells; T cell proliferation or function is then measured (25, 26). To develop effective vaccines that target T cells, epitope mapping of the vaccine Ags is inevitable (22). This is because T cell responses are generally focused on only a few epitopes among the many present on microbial pathogens (27). With ample epitope candidates and multiple rounds of screening, a thorough mapping of T cell epitopes requires large numbers of APCs (22, 28, 29).

Indeed, the availability of autologous APCs is often problematic in studies of human T cell responses (22). Although tetramers of MHC molecules conjugated with peptides provides an alternative option for measuring T cell responses to specific Ags (30), in practice, only limited numbers of Ags can be assessed using tetramers (31), restricting the application of tetramers in large-scale evaluations of candidate epitopes. For this reason, autologous APCs are still the primary choice in T cell epitope discovery. To overcome the low numbers of APCs in the circulating blood, usually the rate-limiting step for mapping human T cell epitopes, leukapheresis is often required to obtain adequate numbers of APCs from a patient’s blood (28, 29). Alternatively, APCs can be expanded in vitro. The low numbers of circulating DCs and macrophages in blood and their limited capacity for proliferation in vitro limit their applications (32–34). In contrast, B cells are more abundant in circulating blood and are easier to expand in vitro compared with DCs and macrophages (35–37). To that end, B cells offer a useful and a potentially more convenient source of APCs; however, current methods for B cell culture still do not generate sufficient cell numbers (35–37).

In this study, we adapted the culture methods established by Luo et al. (38) to expand in vitro the numbers of naive and memory human B cells. This culture method efficiently induces the activation, proliferation, and differentiation of unselected or Ag-binding B cells. Significantly, the culture-derived (CD) B cells express high levels of accessory molecules necessary for effective APC function (MHC class II [MHCII], CD80, and CD86) and effectively present alloantigens and microbial Ags to human T cells. Expansion of Ag-specific human memory B cells in CD cultures results in the generation of Ag-specific APC activity that is significantly more efficient for the cognate Ags than for unrelated Ags of comparable mass. Using CD cultures, we are able to characterize, globally, the TCR repertoire for Ag-specific T cells. Thus, this culture method provides a platform for studying the BCR and TCR repertoires within a single individual.

Blood samples were collected from healthy adult donors with informed consent in accordance with guidelines from the Duke Institutional Review Board committee. Mononuclear cells were isolated by Ficoll-Paque PLUS (GE Healthcare Life Sciences) density gradient centrifugation with SepMate-50 tubes (STEMCELL Technologies). Cells were cryopreserved in liquid nitrogen until use. For microbial Ag-specific T cell studies, blood samples were collected 2–5 wk after tetanus–diphtheria toxoid (Td) boost and/or influenza vaccination.

Cells were cryopreserved based on a previous protocol, with modifications (39). Briefly, cells were suspended in RPMI 1640 medium (Invitrogen) or neat FBS (FCS HyClone; Thermo) at a concentration of ≤2 × 10⁷ cells per ml. An equal volume of cooled freezing medium containing 20% DMSO (Sigma) and 80% FBS was added drop-wise to the cell suspension to a final concentration of 10% DMSO. Cells were aliquoted into cryovial tubes and placed in a prechilled freezing container (Nalgene Mr. Frosty; Sigma). Cryovials were stored at −80°C for 4–24 h and then were stored in liquid nitrogen until thawing for culture.

The following mouse mAbs specific for human surface Ags were used for flow cytometry and cell sorting in this study. Anti-human CD3 allophycocyanin (clone HIT3a), CD4 PE (clone A151A1), CD8 allophycocyanin-Cy7 (HIT8a), CD19 PE-Cy7 and allophycocyanin (HIB19), CD24 PE and BV510 (ML5), CD27 BV421 (M-T271), CD38 BV510 (HIT2), CD45 PE-Cy7 (HI30), CD80 PE (5D10), CD86 biotin (IT2.2), IgD FITC and allophycocyanin-Cy7 (IA6-2), IgM FITC and allophycocyanin-Cy7 (MHM-88), and mouse IgG1 isotype-control PE, PE-Cy7, and biotin (MOPC-21) were purchased from BioLegend (San Diego, CA). CD38 biotin (HIT2) was purchased from eBioscience (San Diego, CA). IgM PE-Cy5 (G20-127), IgG allophycocyanin (G18-145), and CD3 PE-Cy5 (UCHT1), as well as mouse IgG1 isotype-control allophycocyanin (MOPC-21), BV510 (X40), and V450 (MOPC-21), were purchased from BD Biosciences (San Jose, CA). MHCII FITC (TDR31.3) was purchased from LifeSpan BioSciences. Streptavidin–Pacific Orange was purchased from Invitrogen. Mouse IgG1k isotype control was purchased from Rockland.

Analysis of cell phenotypes and cell isolation were performed by flow cytometry. Briefly, cells were incubated with fluorochrome-conjugated mAbs specific for human surface Ags (listed above) and resuspended in PBS containing 2% FBS prior to analysis. Bound biotin-conjugated Abs were revealed by fluorochrome-labeled streptavidin. Doublets were excluded from our analysis and cell sorting by combination(s) of forward scatter (FSC)-A versus FSC-H, FSC-H versus FSC-W, and side scatter-H versus side scatter-W gatings. Dead cells were excluded by positive 7-aminoactinomycin D (7-AAD) staining (BD Biosciences). Labeled cells were analyzed on a BD FACSCanto after fixation (BD Cytofix) or sorted on a BD FACSAria using Diva software (BD Biosciences).

Human mature naive B cells were isolated from PBMCs by negative selection with the EasySep Human Naive B Cell Enrichment Kit, according to the manufacturer’s instructions (STEMCELL Technologies). The purity of mature naive B cells (CD19⁺CD27⁻IgM⁺IgD⁺), as determined by flow cytometry, was >94%.

Human B cells (1–6 × 10³) were plated in six-well plates or 10-cm tissue culture dishes (BD Falcon) to achieve input cell densities of ∼100 B cells per cm². These culture plates or dishes were preseeded overnight with CD154-expressing stromal cells (CD40L^low cell line, gift from David Baltimore) (38). B cells were cultured in R5 medium (RPMI 1640 with 5% human serum [Sigma], 55 μM 2-ME, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 10 mM HEPES, 1 mM sodium pyruvate, and 1% MEM nonessential amino acids [all from Invitrogen]), supplemented with recombinant human IL-2 (50 ng/ml), IL-4 (10 ng/ml), IL-21 (10 ng/ml), and BAFF (10 ng/ml) (all from PeproTech, Rocky Hill, NJ) for 8 d, unless indicated otherwise. The final volume of B cell cultures was 2 ml per well for six-well plates and 6 ml per dish for 10-cm dishes.

Cells were fed with fresh R5 medium containing cytokines on days 4 and 6 by aspirating half of the old medium without touching the bottom of the wells and replacing the same volume with prewarmed fresh medium containing cytokines. In some experiments when cultures were carried beyond 8 d, cells were split onto new feeder cells with fresh cytokines on day 8, and medium was changed on posttransfer days +4 and +6. At the end of culture, CD B cells were harvested, counted, aliquoted, and cryopreserved in liquid nitrogen until use. In experiments quantifying the kinetics of mature naive B cell proliferation, input cell numbers were optimized to facilitate accurate cell counts: input cell numbers on day 0 were 1 × 10⁴, 2.5 × 10³, and 1 × 10³ per well in six-well plates for 4-, 6-, and 8-d cultures, respectively; beyond 8 d, cells were split onto new feeders, and the input cell numbers on day 8 were 4 × 10⁴, 1 × 10⁴, 2.5 × 10³, and 1 × 10³ per well for 10-, 12-, 14-, and 16-d cultures, respectively.

Human PBMCs recovered from tetanus–diphtheria vaccinees were incubated with a combination of flow cytometry mAbs and tetanus toxoid (TT) conjugated with PE (TT-PE), which was generated using the R-Phycoerythrin Labeling Kit-NH2 (Dojindo). Single live B cells (7AAD⁻CD3⁻CD19⁺) were gated, from which IgG memory B cells were defined as CD27⁺CD24^highIgM⁻IgD⁻IgG⁺. IgG memory B cells that did (TT-PE⁺) or did not (TT-PE⁻) bind TT were sorted into CD cultures at ∼100 cells per cm². Following their activation and proliferation for 8 d, TT-PE⁺ and TT-PE⁻ CD B cells were harvested and frozen until use. A fraction of day-8 CD B cells was placed into new cultures (∼100 CD B cells per cm²) and allowed to expand for another 8 d; these day-16 CD B cells were harvested and cryopreserved until use.

PBMCs were thawed and labeled with CFSE (Invitrogen). T cells were isolated by negative selection with the EasySep Human T Cell Enrichment Kit (STEMCELL Technologies) from CFSE-labeled PBMCs. The purities of CD3⁺ T cells (as determined by flow cytometry) were >98% after enrichment. Frozen CD B cells were thawed and counted. T and CD B cells were suspended well before coculture.

Equal numbers (1 × 10⁴ each) of T and CD B cells per well were cocultured in 96-well U-bottom plates (Fisher Scientific) in 100 μl per well of R5 medium without exogenous cytokines. For alloreactive T cell–proliferation studies, the plates were incubated at 37°C in a 5% CO₂ humidified incubator for 5 d. For microbial Ag–specific T cell–proliferation studies, the plates were incubated for 7 d. TT from Clostridium tetani (List Biological Laboratories), recombinant influenza hemagglutinin (HA) (H3 A/Wisconsin/67/2005, kindly provided by S.C. Harrison), and recombinant Bacillus anthracis protective Ag (rPA; BEI Resources) were used in Ag-specific T cell–proliferation studies. T cells were treated with equal numbers of anti-CD3/CD28 Dynabeads (Invitrogen) as positive controls in alloreactive and microbial Ag–specific T cell–proliferation studies.

Analysis of CD B cell phenotypes and T cell proliferation was performed by flow cytometry. Briefly, cells were incubated with fluorochrome-conjugated mAbs specific for human surface Ags (listed above) in PBS containing 2% FBS. Bound biotin-conjugated mAbs were revealed using Streptavidin–Pacific Orange. Analysis was performed with FlowJo software (TreeStar, Ashland, OR).

CD B cells were phenotyped using flow cytometry. Absolute cell counts were performed with CaliBRITE beads, according to the manufacturer’s instructions (BD Biosciences), or with a hemocytometer and trypan blue exclusion of dead cells. CD B cells were incubated with mAbs for surface staining of CD19, CD45, CD3, MHCII, CD80, CD86, CD27, CD24, CD138, IgG, and IgM. Dead cells were excluded from analysis by 7-AAD staining.

T cell proliferation was assessed as described previously (24). Briefly, cells were washed and resuspended in R5 medium without exogenous cytokines. CFSE dissolved in DMSO was added at a final concentration of 5 μM. Cells were mixed well with CFSE and incubated in the dark at room temperature for 5 min. Cells were washed three times with warm medium to remove excess CFSE and were resuspended in culture medium. After culture, cell proliferation was assessed by flow cytometry. Briefly, cells were harvested and incubated with mAbs for CD3, CD4, CD8, CD19, and 7-AAD. Single, live (7AAD⁻)CD19⁻CD3⁺ leukocytes were gated as the T cell population. The frequency of T cell proliferation was determined by CFSE dilution.

TCR repertoires of CD4⁺ T cells were analyzed using the immunoSEQ Analyzer (Adaptive Biotechnologies, Seattle, WA). CFSE-labeled CD4⁺ T cells isolated from recent Td vaccinees (donors A and D) were cultured with anti-CD3/CD28 Dynabeads or with autologous CD B cells in the presence of TT (described above). After 7 d of culture, we sorted CFSE^dimCD4⁺ T cells and isolated genomic DNA from the sorted CD4⁺ T cells by phenol/chloroform extraction (40). Isolated genomic DNA was sent to Adaptive Biotechnologies, which performed amplification of rearranged TCRB genes using multiplex PCR, high-throughput sequencing for the identification of V, D, and J gene segments using the Illumina HiSeq platform, and characterization of the TCRβ repertoire using the immunoSEQ human TCRβ assay (41, 42).

All TCRβ sequence data sets are available at https://clients.adaptivebiotech.com/pub/9816555b-5673-4316-85e2-244acb293f0b.

Images of cultured mature naive B cells were taken with a Canon EOS 20D camera through the eyepiece lens of an Olympus CKX41 microscope at original magnification ×200.

Graphs were compiled and statistical analysis was performed using one-way or two-way ANOVA, followed by a multiple-comparison test, with GraphPad Prism software, version 6 (GraphPad, San Diego, CA). Results are presented as mean ± SD or mean ± SEM. Differences between TT-binding enriched and unenriched CD B cells in inducing T cell proliferation were considered significant at p < 0.05.

To generate large numbers of activated human B cells in vitro, we developed a B cell culture system in which B cell populations are expanded on feeder cells that express low levels of CD154 in medium containing the recombinant human cytokines IL-2, IL-4, IL-21, and BAFF (38).

To evaluate the proliferation of these CD B cells, mature naive B cells from frozen peripheral blood samples were introduced (day 0) and maintained in cultures for as long as 16 d. B cell numbers increased substantially in this culture system (CD culture system); starting with B cell densities of ∼100 cells per cm², we routinely observed clusters of B cells by day 4 of culture that became confluent by day 8 (Fig. 1A). To avoid overcrowding and promote continued proliferation, we split and transferred cultured cells (100 cells per cm²) into fresh cultures that contained fresh cytokines and feeder cells. These newly expanded populations formed B cell clusters as early as 2 d after transfer and continued to proliferate to confluence by posttransfer day 8 (Fig. 1A). This culture system is capable of supporting vigorous B cell proliferation for ≥16 d. Indeed, CD B cells are capable of continued proliferation for at least another week in fresh cultures (data not shown).

To quantify the proliferative capacity of human B cells in the CD culture system, mature naive B cells (CD19⁺CD27⁻CD24^highIgM⁺IgD⁺) from frozen human peripheral blood (Fig. 1B) were cultured, as described above, with the adjustment of input cell numbers (see 2Materials and Methods) to obtain accurate kinetics of B cell proliferation. With an input of 10³ B cells per well in culture, B cells expanded to ≥10⁶ cells per well after 8 d. Subsequently, cultured B cell populations were split and transferred to fresh cultures (see above) at 10³ B cells per well on day 8; these cultured B cells continued to proliferate and reached ≥10⁶ cells at 8 d after transfer (day 16). A similar expansion capacity was seen in cells cultured for 4, 6, 10, 12, and 14 d. CD cultures supported logarithmic expansion of purified mature naive B cells with ≥10³-fold increases by day 8 of culture and 10⁶-fold increases by day 16 (Fig. 1C).

In CD cultures, human mature naive B cells soon acquire an activated phenotype that promotes effective Ag presentation, and they eventually differentiate into Ab-secreting plasmablasts and plasmacytes. To characterize the activation and differentiation of CD B cells during culture, we observed the expression of MHCII, CD80, CD86, membrane IgG, CD27, and CD138 on CD B cells over 16 d of culture. Within 4 d, CD B cells exhibited an activated phenotype that manifested in increased expression of MHCII, CD80, and CD86 (Fig. 2A, 2B, upper panels). Elevated levels of MHCII, CD80, and CD86 were generally sustained through the 16-d culture period although CD80 expression declined somewhat by days 14–16 (Fig. 2A, 2B, upper panels).

Later, CD cultures supported IgM → IgG class-switch recombination and differentiation to plasmablasts/plasmacytes. In these cultures, IgM → IgG class-switch recombination first became obvious on day 8, with 15–20% of cultured naive mature B cells expressing membrane IgG; this increased gradually and peaked at 30–40% of CD B cells by day 14 (Fig. 2A, 2B, upper panels, Supplemental Fig. 1). Expression of CD27, a differentiation marker linked to the human B cell memory compartments (43, 44), accumulated slowly on CD B cells until day 12 and then sharply increased on days 14–16 (Fig. 2A, 2B, lower panels, Supplemental Fig. 1). CD138 expression, a marker of specialization for Ab secretion (45), came later, increasing abruptly at days 14–16. Thus, the CD culture system efficiently activates human mature naive B cells and induces B cell proliferation and differentiation.

Given that many human samples are routinely cryopreserved, we also cryopreserved CD B cells and then recultured them to test whether they retained their proliferative and differentiative ability after the freeze-thaw process. Frozen aliquots of day-8 CD B cells were thawed and cultured at 10³ cells per well for 6 d. Like harvested B cells from cryopreserved peripheral blood (Fig. 1C), the frozen day-8 CD B cells proliferated 2 × 10²-fold upon reculture and maintained high expression of activation markers (data not shown). Thus, CD B cells are readily cryopreserved and retain their activation status and ability to proliferate.

These findings indicate that the CD culture system supports the extensive proliferation of human mature naive B cells and upregulates MHCII, CD80, and CD86 on cultured B cells. Taken together, our results suggest that CD B cells may be capable of acting as potent APCs for autologous and heterologous T cells.

To test whether CD B cells function as APCs to induce Ag-specific T cell proliferation, we first examined their ability to elicit proliferation of allogeneic T cells in MLRs. T cells from five unrelated donors (demographic data in Table I) were cocultured with their own (autologous) CD B cells or with CD B cells from the other donors. CFSE-labeled T cells and unlabeled day-8 CD B cells (10⁴ cells each) were cocultured for 5 d, and T cell proliferation was measured by CFSE dilution at the end of coculture. T cells cultured alone did not proliferate (∼0% CFSE^dim), whereas introduction of anti-CD3/CD28 beads resulted in proliferation of most T cells through multiple divisions (64–89% CFSE^dim) (Fig. 3A, Table II). Autologous CD B cells did not induce T cell proliferation (∼1% CFSE^dim); in contrast, every CD B cell cohort induced strong proliferation in allogeneic T cells (26–67% CFSE^dim) (Fig. 3B, Table II). Of note, allogeneic T cell proliferation was observed in the CD4⁺ and CD8⁺ compartments (Fig. 3B, Table II).

Interestingly, the relative intensities of the allogeneic T cell responses corresponded with those found during treatment with anti-CD3/CD28 beads (Fig. 3B, Table II); for example, T cells from donor E had the highest proliferation rate (CFSE^dim 60 ± 5% [mean ± SD]) in response to allogeneic stimulation by CD B cells among all donors (32 ± 6%, 41 ± 5%, 46 ± 9%, and 47 ± 5% CFSE^dim for donors A, B, C, and D, respectively), as well as had the most vigorous T cell division in response to anti-CD3/CD28 treatment (89 ± 3%). Conversely, T cells from donor A had the lowest proliferation frequency in response to allogeneic stimulation and to anti-CD3/CD28 treatment. In contrast, CD B cells from all donors have a similar ability to induce allogeneic T cell proliferation (CFSE^dim 47 ± 8%, 46 ± 10%, 43 ± 14%, 50 ± 15%, and 40 ± 9% by CD B cells from donors A, B, C, D, and E, respectively). Collectively, in vitro–expanded CD B cells efficiently presented alloantigens to induce allogeneic T cell proliferation but did not activate T cell proliferation nonspecifically.

To determine the ability of CD B cells to process and present microbial Ags, we cocultured CD B cells with autologous T cells from recent vaccinees in the presence of priming and control (unexposed) vaccine Ags and determined T cell proliferation by CFSE dilution after coculture for 7 d. Day-8 CD B cells (originating from mature naive B cells) were cocultured with autologous T cells from donors recently immunized with Td vaccine (donors A and B, 2–5 wk postimmunization), trivalent influenza vaccine (donor C), or both (donor D). Corresponding protein Ags (TT or HA) or an irrelevant Ag (rPA) (10 μg/ml each) were added to individual cocultures (Fig. 4). In the absence of added Ag, little or no T cell proliferation was observed (<5% CFSE^dim), whereas cultures containing anti-CD3/CD28 beads supported vigorous (>80% CFSE^dim) T cell proliferation. Significantly, Ag-dependent autologous T cell proliferation correlated well with each donor’s recent vaccination history. TT triggered CD3⁺ T cell proliferation (≤45% CFSE^dim) in donors A, B, and D, and HA induced T cell proliferation (13–30% CFSE^dim) in all donors; however, the highest frequencies of CFSE^dim T cells were observed in donors C and D (22 and 30% CFSE^dim, respectively), who were recently immunized with influenza vaccine. In contrast, rPA did not induce T cell proliferation (≤5% CFSE^dim), with the single exception of donor D (15% CFSE^dim). On inquiry, we discovered that this individual is exposed to rPA as a result of his occupation. In all cases, T cell proliferation was most evident in CD4⁺ T cells: 64–90% of the CFSE^dim T cells were CD4⁺ (Fig. 4). CD B cells efficiently take up, process, and present protein Ags to autologous CD4⁺ T cells and, thereby, induce Ag-specific T cell activation and proliferation.

Ag-specific B cells are supremely efficient APCs for their cognate Ags (9). Consequently, we determined whether Ag-specific memory B cells might proliferate and differentiate in our culture system and whether cultured memory B cells present specific Ags more efficiently than do unselected CD B cells. IgG memory B cells (CD19⁺CD27⁺CD24^hiIgM⁻IgD⁻IgG⁺) from the peripheral blood of Td vaccinees were sorted based on their ability to avidly bind TT-PE. TT-binding (TT-PE⁺) and non–TT-binding (TT-PE⁻) IgG memory B cell populations were expanded separately in CD cultures.

A representative example of our sorting strategy to identify TT-specific memory B cells shows that ∼40% of peripheral blood B cells from donor A exhibit the CD27⁺CD24^high memory phenotype, of which ∼41% underwent class-switch recombination (IgM⁻IgD⁻); 57% of the class-switched memory B cells express surface IgG (Fig. 5A). Among circulating IgG memory B cells, ∼2% were TT-PE⁺ (Fig. 5A). The frequency of the memory phenotype in another donor (donor B) is ∼35% of circulating B cells, and ∼53% of these memory B cells are IgM⁻IgD⁻; IgG memory cells represent 35% of the class-switched memory pool. The TT-PE⁺ frequency among IgG memory B cells is ∼12%, which is higher than in donor A. Combining data from both donors, circulating B cells are composed of 38.4 ± 2.3% (mean ± SD) memory B cells; 47.4 ± 8.5% of these memory B cells are class-switched, and 46.5 ± 15.4% of them are IgG⁺. The frequency of TT-PE⁺ cells in the IgG memory B cell pool varies between donors (12% versus 2%). Overall, the average frequency of TT-PE⁺ IgG memory B cells is 0.5% among circulating CD19⁺ B cells in these two recent Td vaccinees.

Isolated memory B cells were seeded at ∼100 cells per cm² in CD cultures and allowed to expand for 8 d; subsequently, the proliferating cultured cells were reseeded in fresh cultures (∼100 cells per cm²) for another 8 d (total 16 d). In CD cultures, TT-PE⁺ and TT-PE⁻ IgG memory B cells proliferated comparably, with ∼10³-fold increases by day 8 and 2 × 10⁵-fold increases by day 16 over input cell numbers (Fig. 5B). We measured TT-binding enrichment by comparing the frequencies of positive TT-PE labeling on these CD B cells using flow cytometry. In donor A, after subtracting the PE-TT signal (∼1%) on cultured TT-PE⁻ memory B cells, we found that ∼8 and 6% of CD B cells from the TT-PE⁺ population were positively labeled with TT-PE after 8 and 16 d of culture, respectively (data not shown). In donor B, ∼30 and 15% of CD B cells from 8- and 16-d cultures, respectively, were positively labeled with TT-PE (data not shown), indicating a successful enrichment of TT-binding memory B cells resulting from cell sorting.

To evaluate whether the cultured cells derived from IgG memory B cells also acquire the APC phenotype in CD cultures, the expression levels of surface MHCII, CD80, and CD86 on day-8 and day-16 cultured IgG memory B cells were compared with ex vivo unselected IgG memory B cells. TT-PE⁺ and TT-PE⁻ IgG memory B cells became activated, increasing the expression of MHCII, CD80, and CD86 following 8 d in culture; these expression levels decreased by day 16 but remained higher than the baseline levels (Fig. 5C). Approximately 92 and 64% of these cells had elevated expression of MHCII by days 8 and 16, respectively; similar trends were seen with regard to the expression of CD80 (74 and 31% by days 8 and 16) and CD86 (92 and 61% by days 8 and 16) (Fig. 5C). The expression levels of MHCII, CD80, and CD86, measured as mean fluorescent index (MFI) by flow cytometry, decreased ∼45, 70, and 65%, respectively, in day-16 cultured cells compared with the levels in day-8 cells (Fig. 5C). Surface IgG expression in CD B cells was also assessed. After culture for 8 d, ∼75% of cultured IgG memory B cells retained surface IgG expression, and the frequency decreased to ∼35% on day 16; however, the total expression levels of surface IgG decreased greatly: we observed a 77 and 94% reduction in the MFI of IgG in day-8 and day-16 cultured IgG memory B cells, respectively, compared with the MFI in input cells (data not shown). The expression levels of surface MHCII, CD80, CD86, and IgG in CD B cells from sorted TT-PE⁺ and TT-PE⁻ IgG memory B cell populations cultured for the same duration were comparable (data not shown).

To evaluate the ability of CD B cells enriched for TT binding to induce T cell proliferation in response to cognate Ags, cultured IgG memory B cells were recovered and cocultured with equal numbers (10⁴) of CFSE-labeled autologous T cells from donors A or B, who had recently received a Td booster. Both day-8 and day-16 CD B cells from TT-PE⁺ and TT-PE⁻ IgG memory B cells were tested for their ability to induce T cell proliferation in the absence or presence of TT (seven tested concentrations, 5-fold serial dilutions from 10 μg/ml), HA (2 or 10 μg/ml), or the irrelevant Ag rPA (10 μg/ml) in the cocultures. After 7 d of coculture, total T cell proliferation was determined by CFSE dilution (Fig. 6).

Without the addition of microbial Ags, both CD B cell populations (enriched or not for TT binding) induced little or no autologous T cell proliferation (CFSE^dim ∼3 and ∼5% for TT-PE⁺ and TT-PE⁻ CD B cells, respectively, from donor A and CFSE^dim ∼4 and ∼3.5% from donor B). A coculture well using day-8 TT-PE⁻ CD B cells (CFSE^dim ∼30%) was an exception, which may have been due to the effect of a culture medium component on the T cells. With the addition of rPA, little or no T cell proliferation was induced by either CD B cell population (CFSE^dim ∼6 and ∼2% for donor A and donor B, respectively). With the addition of irrelevant Ag or no Ag, cells derived from IgG memory B cells in CD cultures induced little or no T cell proliferation, indicating that the CD B cells from the memory pool, similar to those from the naive mature pool, do not activate T cells nonspecifically (Figs. 3, 4, 6).

With the addition of TT Ag, T cell proliferation was more robust in the cocultures with CD B cells enriched for TT binding than in those with CD B cells not enriched for TT binding, indicating that Ag-driven T cell proliferation corresponded with the Ag specificity of CD B cells (Fig. 6, donors A and B). Using day-8 CD B cells from donor A, the population of CD B cells enriched for TT binding induced more T cell proliferation (∼20% more CFSE^dim T cells) compared with CD B cells not enriched for TT binding in response to 10 μg/ml TT (CFSE^dim ∼45 and ∼25%, respectively). By serially reducing TT concentrations, we observed ∼10–38% more CFSE^dim T cells in the cocultures with day-8 CD B cells enriched for TT binding than in those not enriched for TT binding in the presence of TT ≥ 0.016 μg/ml (p < 0.0001) (Fig. 6, donor A); the superior ability of TT-binding–enriched day-8 CD B cells to induce T cell proliferation lasted until the concentrations of TT were <0.016 μg/ml. Similarly, day-16 CD B cells enriched for TT binding induced more T cell proliferation (∼10–37% more CFSE^dim T cells) than did CD B cells not enriched for TT binding (p < 0.0001), and the advantage remained until TT was <0.4 μg/ml. Day-8 and day-16 TT-binding–enriched CD B cells from donor B also exhibited an enhanced ability to induce T cell proliferation, with ∼10–36% more CFSE^dim T cells observed at TT ≥ 0.08 μg/ml (p < 0.01 and p < 0.001 for days 8 and 16, respectively, for CD B cells enriched or not for TT binding).

Furthermore, the population of CD B cells enriched for TT binding induced measurable T cell proliferation with the addition of the noncognate Ag, HA, at 10 μg/ml (13–23% CFSE^dim T cells observed using day-8 CD B cells and 4–9% CFSE^dim T cells using day-16 CD B cells, both donors) and at 2 μg/ml (5–17 and 2–5% CFSE^dim T cells detected using day-8 and day-16 CD B cells, respectively) (Fig. 6), suggesting that these B cells can acquire Ags through a BCR-independent pathway; however, the HA-presenting ability of TT-PE⁺ enriched CD B cells was less efficient than their presenting ability for the TT cognate Ags (Fig. 6). In contrast, CD B cells not enriched for TT binding exhibited an Ag-presenting capability that was similar to that of TT-binding–enriched cells in the presence of 10 μg/ml HA (14–22 and 7–9% CFSE^dim T cells observed using day-8 and day-16 unenriched CD B cells, respectively) or 2 μg/ml HA (10–12 and 2–3% CFSE^dim T cells observed using day-8 and day-16 unenriched CD B cells, respectively) (Fig. 6). Taken together, we found comparable capacities for CD B cell populations, regardless of TT-binding enrichment, to present non-TT Ags (HA and rPA) to autologous T cells (p ≥ 0.59 and p ≥ 0.49 for donor A and B, respectively). From the above T cell–proliferation results for all tested Ags (Fig. 6), we conclude that the expression of TT-specific BCRs on TT-PE⁺ enriched CD B cells contributes to their superior TT presentation compared with TT-PE⁻ enriched CD B cells.

Taken together, these results indicate that cells expanded from IgG memory B cells in vitro can function as APCs that take up Ags through BCR-dependent and BCR-independent pathways. Furthermore, the BCR-dependent Ag-uptake pathway significantly enhances the Ag-presenting function of these CD B cells in inducing autologous T cell proliferation.

To demonstrate the usefulness of our CD culture system in analyzing the human TCR repertoire, we characterized TCR Vβ usage of TT-specific CD4⁺ T cells from two donors. We isolated genomic DNA from CFSE^dim CD4⁺ T cells that proliferated in response to TT presented by autologous CD B cells and then amplified TCRβ VDJ rearrangements by PCR. For comparison, we amplified TCRβ VDJ rearrangements from genomic DNA of unstimulated CD4⁺ T cells and CFSE^dim CD4⁺ T cells activated by anti-CD3/CD28 from the same donors. From the two unrelated donors (A and D), we obtained a total of 31,019, 41,265, and 8,791 productive TCRβ VDJ rearrangements by deep sequencing from freshly isolated CD4⁺ T cells, anti-CD3/CD28–activated CD4⁺ T cells, and TT-activated CD4⁺ T cells, respectively.

Vβ gene segment usage of freshly isolated CD4⁺ T cells was diverse in both donors (Fig. 7A, 7B; gray bars). As expected for unspecific expansion of CD4⁺ T cells, Vβ gene segment usage was virtually identical between freshly isolated CD4⁺ T cells and anti-CD3/CD28–stimulated CD4⁺ T cells in both donors (Fig. 7A, 7B; gray and blue bars). In contrast, Vβ usage of TT-activated CD4⁺ T cells was clearly distinct from freshly isolated CD4⁺ T cells and from unspecifically activated CD4⁺ T cells (Fig. 7A, 7B; red bars). In donor A, Vβ2-1, Vβ4-3, Vβ5-4, Vβ6-1, and Vβ19-1 gene segments were particularly frequent in TT-activated CD4⁺ T cells compared with resting or unspecifically activated CD4⁺ T cells (Fig. 7A). In donor D, Vβ2-1, Vβ5-1, Vβ18-1, and Vβ29-1 gene segments were particularly enriched in TT-activated CD4⁺ T cells (Fig. 7B).

In general, although Vβ usage by TT-activated CD4⁺ T cells differed between donors A and D, the common use of the Vβ2-1 gene segment may represent a general structural solution for the TCR of TT-specific CD4⁺ T cells (Fig. 7A, 7B, red bars). The frequency of Vβ2-1 in resting and unspecifically activated CD4⁺ T cells was 5.2 and 4.6%, respectively, for donor A and 8.8 and 6.6%, respectively, for donor D (Fig. 7A, 7B). In TT-activated cohorts, the frequency of Vβ2-1 rearrangements was doubled in both donors (9.9% in donor A and 14.1% in donor D) (Fig. 7A, 7B).

This structural selection is evident in the decreased diversity of independent TCRβ rearrangements recovered from TT-activated T cells (Fig. 7C). Although the 10 most common TCRβ rearrangements in resting and unspecifically activated CD4⁺ T cells constituted <6% (1.1–5.5%) of all TCRβ sequences, the top 10 rearrangements from TT-activated cells represented ≈25% (24% for donor D and 26.9% for donor A) (Fig. 7C). Rearrangements of Vβ2-1 are particularly represented (among the top 10 TCRβ sequences in both donors; in donor A, Vβ2-1 rearrangements are ranked third and eighth in abundance, whereas in donor D they are ranked first) (Supplemental Table I). The increased frequencies of a handful of TCRβ rearrangements are consistent with clonal proliferation and dominance in response to TT; in donor A, two Vβ2-1 rearrangements account for 56% of all Vβ2-1 rearrangements in the TT-activated cohort, whereas in donor D, a single rearrangement accounts for 47% of all Vβ2-1 rearrangements.

This Ag-specific structural selection is also evident in CDR3 sequences of the dominant top 10 TCRβ rearrangements. The two most frequent Vβ2-1 rearrangements from TT-activated T cells for donor A (third and eighth) share a virtually identical CDR3 amino acid sequence motif (ASRPGQPPYEQY and ASSGGQPPYEQY, respectively). This near identity implies convergent selection, presumably for a common peptide/MHCII epitope. We conclude that TCR sequences from TT-activated CD4⁺ T cells reflect the structural restriction inherent in Ag-specific receptors preferentially expanded in response to TT presented by autologous CD B cells.

The value of the CD culture system in evaluating T cell specificity and the TCR repertoire is demonstrated by the efficient and large yield of CD B cells that are capable of acting as APCs to induce T cell proliferation against alloantigens or pathogenic Ags. Expanded Ag-specific human CD4⁺ T cells in the CD culture system were subsequently analyzed for the use of TCR Vβ gene segments. Furthermore, we showed that human Ag-specific memory B cells can be expanded efficiently in vitro and function as highly effective APCs, which should allow them to serve as a valuable tool for studying the interaction between cognate T and B cells.

The usefulness of CD B cells includes evaluating a B cell repertoire that changes during exposure to Ags in chronic inflammatory conditions (46–49). Mapping the alterations in B cell repertoires during the course of disease may provide insights into the pathogenesis of these diseases and, subsequently, potential therapeutic strategies for them. For example, the existence of B cells that secrete broadly neutralizing Abs against HIV demonstrates coevolutionary changes in the B cell repertoire and viral variants (46) and suggests that the timing of humoral immune responses does not correspond well with the progression of mutations in HIV Ags. Furthermore, B cell repertoire studies in patients with autoimmune diseases revealed that autoreactive B cell clones are generated as a result of defective tolerance checkpoints, as well as persistent Ag stimulation (47–49). In our laboratory, we are currently studying human B cell repertoires using the CD culture method.

Along with humoral responses, the repertoire of T cells theoretically changes upon persistent exposure to Ags (14, 50). This is because T cells are interacting with B cells that can function as APCs during chronic inflammatory diseases (51–53). Cognate T and B cells interact and provide reciprocal help that is required for activation and differentiation of both cell types, which results in alterations in T and B cell populations (9, 52). In contrast to studies of B cell repertoire dynamics, relatively few studies of coevolutionary changes in human T cell repertoires and chronic pathogenic Ags have been reported (54, 55). Furthermore, studies directly determining TCR specificity have been rare (56), at least in part because sufficient numbers of autologous APCs are not frequently available. In this study, we developed and explored a method that provides abundant autologous APCs using CD B cells and could be used to study T cell repertoire progression.

In our CD culture system, cytokines and CD40:CD154 interaction between B cells and feeder cells support activation, vigorous expansion (Figs. 1, 5) (57–59), and APC function of human naive and memory B cells (Figs. 4–7) (6, 60, 61). Among the cytokines present in our culture medium (IL-2, IL-4, IL-21, and BAFF), IL-21 is crucial to induce robust B cell proliferation (Supplemental Fig. 2), an observation that is consistent with other studies (62–64). BAFF supports B cell survival (65), the differentiation of human memory B cells into Ab-secreting cells (66), and class-switch recombination in human B cells (67, 68). Because IL-4 and IL-21 differentially regulate class-switching to certain Ig isotypes (64, 69), one might consider the combinations of cytokines for CD cultures depending on their own purposes.

Although studies also used CD40-mediated activation to induce proliferation of human B cells (6, 60, 61, 63), our cultures supported more robust and sustained proliferation of B cells than did previous culture methods. In the CD culture system, B cells expanded by ∼10⁶-fold after 16 d of culture; the rate of expansion was stable over a 16-d culture period after an initial lag phase (Fig. 1C). Beginning at day 4 of culture, a nearly log increase in cell numbers was observed every 2 d; the population doubling time was ∼15 h in day 4–16 CD B cells (calculated using the logarithmic least-squares fitting technique [http://www.doubling-time.com/compute.php]). We believe that our CD culture may be the most efficient system for inducing primary human B cell division in vitro (6, 60, 61).

In the CD culture system, mature naive human B cells are activated and express elevated levels of MHCII, CD80, and CD86 as early as day 4 (Fig. 2A), undergo IgM → IgG class-switch recombination (days 8–16), and later differentiate into plasmablasts/plasmacytes at around day 14–16 after culture (Fig. 2B). By day 16, the frequency of CD138⁺ plasmablast/plasmacyte populations reached 10–30% in the CD culture system (Fig. 2B). We note that our estimate of plasmablast/plasmacyte frequencies in the CD culture system might vary, depending on the use of surface markers (e.g., CD27^highCD38^high) to define plasmablasts/plasmacytes (44, 62).

The observation that the CD culture system supports activation and extensive proliferation of memory B cells (Fig. 5B) is consistent with previous reports that human memory B cells can be activated and differentiate through a pathway that bypasses BCR signaling (62, 63). CD B cells originating from memory B cells exhibited slightly less cell expansion and more frequent CD138 expression compared with those originating from naive B cells, although neither difference was statistically significant (Figs. 1C, 5B). Furthermore, the elevated expression of MHCII, CD80, and CD86 on cultured memory B cells declined faster than in cultured naive B cells (Figs. 2A, 5C). These findings may be due to fundamental differences in the activation capacity and differentiation potential of naive and memory B cells (63, 70).

CD B cells from day-8 cultures displayed more efficient Ag-presenting functions than did CD B cells from day-16 cultures (p ≤ 0.0067 and p ≤ 0.0011 for TT-PE⁺ and TT-PE⁻ enriched CD B cells, respectively) (Fig. 6). This loss of function correlated with the higher levels of MHCII, CD80, and CD86 expressed by day-8 CD B cells (Fig. 5C). Although plasmacytes may retain surface expression of MHCII, CD80, and CD86 and function as APCs (71), the expression of mRNA encoding MHCII, CD80, and CD86 molecules is very low in plasmacytes (71, 72). Thus, plasmacytic differentiation of CD B cells is also likely to contribute to losses in APC function.

Potent alloresponses by T cells are induced by CD B cells from allogeneic donors (Fig. 3, Table II). We observed alloreactivity among CD4⁺ and CD8⁺ T cell subsets, indicating that the Ag-display functions of MHC class I and MHCII are normal in CD B cells (27). This suggests that CD B cells acquire Ags derived from dying cells and/or through autophagy in an MHCII-dependent pathway (27, 73); conversely, the Ags for MHC class I–dependent presentation can be acquired from the intracellular space for canonical presentation and/or from the extracellular space for cross-presentation (27, 74). Unlike allogeneic cocultures, little or no T cell proliferation was observed in the autologous T–B cocultures (Figs. 3, 4, 6, Table II). These responses may have contained some minor component of xenoactivation, because we did not isolate CD B cells from the CD154-expressing mouse stromal cells after culture. Nonetheless, as expected (75), xenoreactivity of T cells against the mouse stromal cell line appeared to be negligible in these conditions (Figs. 3, 4, 6, Table II).

CD B cells efficiently induced the proliferation of autologous T cells against microbial Ags when these cells were prepared from donors who had recently been vaccinated with (components of) the same microbial Ags (Fig. 4). Presumably, this reflects the increased numbers of specific T cells elicited by homologous vaccination (56, 76). CD B cells expanded from nonselected naive mature B cells are capable of acquiring Ags through BCR-independent fluid-phase endocytosis (77) and presenting these Ags to T lymphocytes. As expected (78, 79), the presentation of exogenous Ags (TT and HA) by CD B cells to T cells occurs primarily through the MHCII-dependent pathway that induces CD4⁺ T cell responses (Fig. 4). Interestingly, we observed T cell responses against rPA in the individual (donor D) who had a history of occupation-related exposure to rPA (Fig. 4); a low level of Abs against rPA also was detected in this individual’s plasma (data not shown). Together with the coculture results, this indicates the presence of humoral and cellular immune responses against rPA in donor D (80, 81).

It is appropriate to use CD B cells with sets of Ags to determine the Ag specificity of T cells and to evaluate postvaccine cellular responses. The identified Ag-responding T cells can be isolated for subsequent determination of the TCR repertoire (Fig. 7) (56). Alternatively, many studies used tetramers of MHC molecules conjugated with peptides and fluorochromes to separate the target T cells (30); however, some limitations of this method should be considered. First, tetramers allow for the identification of T cells specific for pathogens with a few immunodominant peptides, not against pathogens with a complex set of epitopes (30, 82). Second, tetramers are not primarily designed to measure the breadth of cellular immune responses induced by vaccination (83). Third, each individual needs an MHC test to find MHC-matched tetramers (if available) (84). In contrast, CD B cells would allow for the identification (and subsequent isolation) of T cells against pathogens using a complex set of Ags. Because the tested Ags would not be limited to peptide forms (Figs. 4, 6, 7), CD B cells would allow more Ags to be tested and would provide a suitable method for determining the intensity and breadth of vaccine responses. Moreover, an MHC test would not be necessary because T and B cells could be isolated from the same individuals (Figs. 4, 6, 7). In addition to specific T cell isolation, CD B cells could be valuable in defining the characteristics of Ag-specific memory T cells, such as their expression of surface molecules and effector molecules.

Using autologous CD B cells and CD4⁺ T cells, we characterized the TCR repertoire for TT-reactive CD4⁺ T cells in two unrelated donors. Analysis of 8800 rearranged TCRB gene sequences from TT-reactive CD4⁺ T cells showed that, in each donor, distributions of TCR Vβ usage of TT-activated CD4⁺ T cells was distinct from resting CD4⁺ T cells or from CD4⁺ T cells unspecifically activated by anti-CD3/CD28 (Fig. 7). Although TCR Vβ usage of TT-activated CD4⁺ T cells differed substantially between donors, presumably as the result of dissimilar HLA types and/or differences in vaccination and exposure histories (85), we noted selective expansion of small sets of Vβ rearrangements, such that the top 10 rearrangements in either donor constituted ≥25% of all TCRβ amplicands. This decreased diversity is consistent with TT-specific clonal expansion.

Although genetically diverse, human TT-specific TCRαβ may share some general structural(s) characteristics, because we observed overrepresentation of Vβ2-1 rearrangements in both donors (Fig. 7, Supplemental Table I). Frequencies of Vβ2-1 usage were doubled in the TT-activated cohorts of donor A and D (Fig. 7), and this increase was the consequence of clonal expansion. Interestingly, in donor A, two Vβ2-1 TCR rearrangements, ranked third and eighth among the top 10, shared a CDR3 amino acid sequence that was independently generated by distinct VDJ rearrangements (Supplemental Table I). We hypothesize that this convergence represents a common structural solution for a single peptide/MHCII epitope.

One unique feature of Ag-presenting B cells is that B cells uptake Ag in BCR-independent and -dependent manners (1, 9, 86). Although nonantigen-specific B cells uptake Ags in a BCR-independent manner, Ag-specific B cells uptake Ags more efficiently through their high-affinity BCRs (9). Our results (Fig. 6) indicate that TT-binding memory CD B cells could present TT more efficiently than TT nonbinding memory CD B cells, which is consistent with previous reports using EBV-transformed B cells (9). Our data show that HA presentation was comparable in cultured memory B cells with or without enrichment for TT binding (Fig. 6). In addition, the comparable expression levels of MHCII, CD80, CD86, and IgG on both CD B cell groups (data not shown) suggest that BCR-independent Ag-presenting function was similar in these cell groups. Therefore, we conclude that the superior ability of TT-binding–enriched CD B cells to induce T cell proliferation in the presence of TT was related to their harboring of TT-specific BCRs, which uptake TT more efficiently.

Consistent with our own and earlier observations (9), TT-binding memory CD B cells induced T cell proliferation more efficiently than did naive CD B cells in the presence of TT (Supplemental Fig. 3). We also noted that APC functions of naive CD B cells were generally less efficient compared with those of memory CD B cells (TT-PE⁺ or TT-PE⁻), regardless of Ag types (Supplemental Fig. 3), a finding also noted by other investigators (87). Further studies, such as transcriptome analysis of memory and naive CD B cells, would advance our understanding of the differential APC functions of these CD B cells.

Several studies showed that human memory B cells act as APCs and activate allogeneic T cells (87–89). In contrast, studies in human Ag-specific memory B cells are often limited because of low cell numbers (90, 91). To obtain sufficient numbers of specific B cells, EBV-transformed B cell lines are commonly used in studies of the interaction between human cognate T and B cells (9). However, there are potential T cell responses to the EBV-infected B cells (92, 93); thus, EBV-transformed B cells are not suitable for this type of study. Our culture system enables the expansion of Ag-specific human memory B cells in vitro without EBV transformation (Fig. 5B), thus offering the opportunity to use activated and culture-expanded primary cells to evaluate the interaction between cognate T and B cells.

Our results suggest the usefulness and immediate application of CD B cells for establishing T cell lines/clones and epitope mapping. Numerous CD B cells originating from naive or memory subsets would provide sufficient numbers of APCs without the requirement for leukapheresis or EBV transformation. Expanded CD B cells retain surface expression of MHCII and costimulatory molecules (CD80 and CD86), and importantly, CD B cells can serve as APCs to induce T cell proliferation against microbial Ags. Furthermore, samples from patients with infectious diseases are infrequently accompanied by leukapheresis and abundant numbers of APCs. Thus, the use of CD B cells as a source of autologous APCs for expanding Ag-specific T cells and studying their epitope specificity would be a useful strategy for the development of effective T cell–based vaccines and for revealing the complexity of coevolutionary changes in the T cell repertoire during chronic inflammation. In conclusion, this CD culture system provides a platform for studying human B and T cell repertoires.

We thank David Baltimore for providing the CD154-expressing cell line, Stephen Harrison for providing HA, the blood donors, Elizabeth Wong and Alexander Reynolds for editorial assistance, and Dongmei Liao and Xiaoe Liang for technical assistance.

The authors have no financial conflicts of interest.