Increased Expression of CD27 on Activated Human Memory B Cells Correlates with Their Commitment to the Plasma Cell Lineage
http://www.100md.com
免疫学杂志 2005年第7期
Abstract
Plasma cells (PC) or Ig-secreting cells (ISC) are terminally differentiated B cells responsible for the production of protective Ig. ISC can be generated in vitro by culturing human B cells with the T cell-derived stimuli CD40L, IL-2, and IL-10. ISC have traditionally been identified by the increased expression of CD38, analogous to primary human PC, and the acquired ability to secrete Ig. By tracking the proliferation history of activated B cells, we previously reported that the differentiation of memory B cells into CD38+ B cells is IL-10 dependent, and increases in frequency with cell division. However, <50% of CD38+ cells secreted Ig, and there was a population of CD38– ISC. Thus, the PC phenotype of CD38+ cells generated in vitro did not correlate with PC function. To address this, we have examined cultures of activated memory B cells to accurately identify the phenotype of ISC generated in vitro. We found that CD27 is also up-regulated on memory B cells in an IL-10-dependent and division-dependent manner, and that ISC segregated into the CD27high subset of activated memory B cells irrespective of the acquired expression of CD38. The ISC generated in these cultures expressed elevated levels of the transcription factors Blimp-1 and X box-binding protein-1 and reduced levels of Pax-5, and exhibited selective migration toward CXCL12, similar to primary PC. We propose that the differentiation of memory B cells into PC involves a transitional stage characterized by a CD27highCD38– phenotype with the acquired ability to secrete high levels of Ig.
Introduction
One of the primary functions of mature B cells is the production of protective high affinity Ig following terminal differentiation into plasma cells (PC).3 In humans, PC can be identified by the up-regulation in expression of CD38 and the concomitant decrease in CD20 (CD38++CD20±) (1, 2). Cells with this phenotype have been detected in lymphoid tissues such as spleen (3), tonsils (4, 5), intestine (6, 7), bone marrow (BM), and peripheral blood (1, 5). Importantly, PC are overrepresented not only in malignancies (2, 8), but also autoimmune disorders such as systemic lupus erythematosus (9, 10). Therefore, delineating the mechanisms involved in generating these cells is necessary for a complete understanding of the humoral immune response as well as PC dyscrasias.
Ig-secreting cells (ISC) resembling PC can also be generated in vitro by culturing human germinal center or memory B cells with BM stromal cells and/or activated T cells (11, 12), a follicular dendritic-like cell line plus CD40L, IL-2, IL-4, and IL-10 (13, 14) or the T cell-derived stimuli CD40L, IL-2, and IL-10 (15, 16, 17). These cells have typically been identified by the increased expression of CD38, and the acquired ability to secrete high levels of Ig (11, 12, 13, 14, 15, 16, 17). Such cells generated in vitro are rapidly proliferating (17, 18, 19, 20), suggesting they are more akin to plasmablasts detected in vivo (21, 22), rather than mitotically inactive PC (23). By tracking the proliferation history of activated B cells, we have recently found that the differentiation of memory B cells into CD38-expressing cells is IL-10 dependent, and increases in frequency with cell division; that is, CD38+ B cells appear in cultures of memory B cells stimulated with CD40L and IL-10 once the cells have undergone three or more rounds of cell division (17, 20, 24). Furthermore, the survival requirements of the CD38+ B cells differed from CD38– B cells present in the same culture inasmuch that CD38+ B cells became independent of CD40L, and responsive to B-cell activating factor belonging to the TNF family (BAFF), while CD38– B cells required continual stimulation with CD40L for their persistence (11, 17, 25). However, we also found that only 40% of CD38+ cells secreted Ig, and that there was a population of CD38– ISC (17, 25). Thus, the "PC" phenotype of CD38+ cells generated in vitro did not correlate with PC function. To address this discrepancy, we have now examined in greater detail cultures of activated memory B cells in an attempt to accurately identify the phenotype of ISC generated in vitro. We found that CD27 is up-regulated on the surface of memory B cells in an IL-10-dependent and division-dependent manner, and that ISC segregated into the CD27high subset of activated memory B cells irrespective of the acquired expression of CD38. The CD27high ISC exhibited increased expression of the transcription factors Blimp-1 and X box-binding protein-1 (XBP-1), and reduced expression of Pax-5, as well as phenotypic and migratory characteristics similar to those of human primary PC. In contrast, nondifferentiated B cell blasts (i.e., CD38–CD27+) present in the same culture retained features of memory B cells. These results reveal the dynamic ability of human memory B cells to generate multiple cell fates following activation with T cell-dependent stimuli to yield effector cells, namely ISC, as well as sufficiently expand the pool of nondifferentiated memory B cells to increase the size of this population of effector cells that can then be rapidly activated following re-exposure to the immunizing Ag.
Materials and Methods
Reagents
FITC-conjugated anti-CD20 and PE-conjugated anti-CD19 and CD45 mAb were purchased from BD Biosciences; anti-CCR7, PE-conjugated isotype controls, CD21, CD27, CD80, CD86, CD95, and CXCR4 mAb, biotinylated anti-CD44, anti-hamster IgG, and streptavidin (SA)-conjugated to PerCp were purchased from BD Pharmingen. PE-anti-CD22, CD23, CD31, CD38, CD62L, CD138, HLA-DR mAb, biotinylated IgG1 isotype control, and anti-CD38 mAb, allophycocyanin-conjugated isotype control, anti-CD20 and anti-CD38 mAb, and SA-allophycocyanin were purchased from Caltag Laboratories; biotinylated anti-CXCR5 mAb was from R&D Systems; biotinylated anti-human CD27 mAb was from eBioscience; PE-conjugated anti-CD40 mAb (mAb89) was provided by J. Banchereau (Schering Plough Labs, Dardilly, France); PE-conjugated anti-CD39 (A1) mAb has been described previously (26). Purified and biotinylated F(ab')2 of goat anti-human IgM, IgG, or IgA polyclonal Ab were purchased from Southern Biotechnology. Recombinant human BAFF, and mouse anti-BAFF receptor (anti-BAFF-R; clone 9.1), anti-transmembrane activator of and CAML interactor (TACI; clone C4D7) and hamster anti-B cell maturation Ag (BCMA; clone C4E2.2) mAbs were provided by Dr. S. Kalled (Biogen Idec, Cambridge, MA; Ref. 25). Recombinant CD40L expressed as membranes in Sf21 insect cells infected with baculovirus vector containing human CD40L cDNA was provided by Dr. M. Kehry (Boehringer Ingleheim, Ridgefield, CT). IL-2 was purchased from Endogen; IL-10 was provided by Dr. R. de Waal Malefyt (DNAX Research Institute, Palo Alto, CA). CFSE was from Molecular Probes. Recombinant human CXCL12 and CCL21 were purchased from PeproTech; CXCL13 was from R&D Systems.
Generation of CD70 transfectants
A complementary DNA encoding human CD70 was amplified from RNA prepared from the CD70-expressing human B cell line JY by PCR using Pfu polymerase and the following primers (Sigma-Genosys): 5'-GCA TGC GGA TCC TTC CTT CCT TCT CGG CAG CG (BamH1 site underlined), and 3'-GCA TGC GCG GCC GCA ATC AGC AGC AGT GGT CAG GG (NotI site underlined). The resulting product was digested with BamH1 and NotI, and ligated into pcdef3, a derivative of pEF-BOS containing the neomycin-resistance gene (27). The mouse mastocytoma cell line P815 was transfected with CD70/pcdef3 by electroporation, and positive cells were selected initially in the presence of G418 (Bio-Rad), and subsequently by cell sorting.
Cells
Total human B cells (>98% CD19+) were isolated as previously described (17, 28). Memory B cells were isolated by sorting on a FACStarPLUS or FACSVantage (BD Biosciences) following labeling with FITC-anti-CD20 and PE-anti-CD27 mAb and collecting CD27+CD20+ B cells (17, 26).
CFSE labeling and B cell cultures
Primary cultures. Memory B cells were labeled with CFSE (29) and cultured in 48-well plates (4 x 105/ml; BD Labware) for 5 days with CD40L alone (at a predetermined optimal dilution of the membrane preparation; 1/250), or in the presence of IL-10 (100 U/ml), or IL-2 (50 U/ml) plus IL-10. The cells were harvested and expression of CD27 and CD38 was then determined by immunofluorescence and flow cytometric analysis.
Secondary cultures. CFSE-labeled memory B cells were cultured for 4 days with CD40L, IL-2, and IL-10. The cells were harvested, washed, and recultured (2 x 105/ml) with IL-2 and IL-10 in the absence or presence of CD40L (1/500 dilution) or BAFF (2.5 μg/ml) for an additional 4 days (15, 17, 25). In some experiments, the activated memory B cells were recultured with parental P815 or CD70/P815 cells in the presence of IL-2 and IL-10. Before culture, the P815 cells were fixed in 1% formaldehyde, prepared in PBS, for 20 min (30), washed three times with PBS, and cultured in medium for 1 h to remove the formaldehyde. The memory B cells and P815 cells were cultured at a ratio of 5:1.
At the completion of the primary or secondary cultures, a known number of CaliBRITE beads (BD Biosciences) were added to culture wells before harvesting, and the number of viable B cells was calculated as a function of the ratio of beads to live cells (29). All cultures were performed using RPMI 1640 containing L-glutamine (Invitrogen Life Technologies), 10% FCS (CSL), 10 mM HEPES (pH 7.4) (Sigma-Aldrich), 0.1 mM nonessential amino acid solution (Sigma-Aldrich), 1 mM sodium pyruvate (Sigma-Aldrich), 100 U/ml penicillin, 100 μg/ml streptomycin (Sigma-Aldrich), 100 μg/ml Normocin (InVivoGen), and 40 μg/ml apo-transferrin (Sigma-Aldrich) and were conducted at 37°C in a humidified atmosphere containing 5% CO2.
Immunofluorescent staining
For phenotypic analysis, cells were incubated on ice with PE-, allophycocyanin-, or biotinylated specific mAb, or the appropriate isotype control, followed by SA-PerCp and analyzed on a FACSCalibur using CellQuest software (BD Biosciences). Binding of BAFF to activated B cells, as well as assessment of expression of the different BAFF receptors (i.e., BAFF-R, TACF, and BCMA) were determined as described previously (25)
Analysis of Ig secretion
ELISPOT. CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 4 days and then washed and recultured under the same conditions for a further 4 days. The cells were harvested and labeled with PE-anti-CD27 mAb and allophycocyanin-anti-CD38 mAb. Different populations of activated B cells, defined by division history and surface phenotype, were sorted directly into ELISPOT plates (Multiscreen-HA plates; Millipore), and the frequency of cells secreting IgM, IgG, and IgA was determined (17).
ELISA. CFSE-labeled memory B cells were cultured for 5 days with CD40L, IL-2, and IL-10 and then labeled with PE-anti-CD27 and allophycocyanin-anti-CD38 mAbs. Different populations of activated B cells were isolated by cell sorting and then recultured (105 cells/500 μl/well) for a further 2 days with CD40L, IL-2, and IL-10, after which time supernatants were collected and the level of secreted Ig was determined (17).
Analysis of expression of transcription factors
Semiquantitative PCR was used to examine gene expression in different populations of activated human memory B cells. B cell subsets were isolated by cell sorting, total RNA was extracted (Qiagen RNeasy Kit; Qiagen) and then transcribed into cDNA using oligo-dT (Boehringer Mannheim) or random hexamers (Invitrogen Life Technologies) as primer and Superscript II RNase H– reverse transcriptase (Invitrogen Life Technologies). Resulting cDNA was then normalized for expression of the constitutively expressed housekeeping gene GAPDH (5'-CCA CCC ATG GCA AAT TCC ATG GCA, 3'-TCT AGA CGG CAG GTC AGG TCC ACC) and then used as a template for PCR using REDTaq (Sigma-Aldrich) (3). The following primers were used (Sigma-Genosys): Pax-5 5'-GCA TAG TGT CCA CTG GCT CC; Pax-5 3'-CCA GGA GTC GTT GTA CGA GG; BLIMP-1 5'-GAT GCG GAT ATG ACT CTG TGG; BLIMP-1 3'-CTC GGT TGC TTT AGA CTG CTC; XBP-1 5'-GCT CAG ACT GCC AGA GAT CG; XBP-1 3'-GTC CSG AAT GCC CAA CAG G; Bcl-6 5'-CTG ACA GCT GTA TCC AGT TCA CC; Bcl-6 3'-TCT TGG GGC ATC AGC ATC.
Expression of Blimp-1 and Pax-5 protein by activated human memory B cells was also determined. B cell subsets were isolated by cell sorting, and then lysed in ice cold lysis buffer (10 mM Tris-HCl (pH 7.8), 1% Nonidet P-40, 150 mM NaCl, and enzyme inhibitors). Cell lysates were electrophoresed through 12% acrylamide gels containing 0.1% SDS and transferred to polyvinylidene difluoride membranes (Millipore). Membranes were probed with Abs against Blimp-1 (31), Pax-5 (C-20; Santa Cruz Biotechnology) or SHP-2 (Santa Cruz Biotechnology) followed by HRP-conjugated anti-rat, anti-goat, and anti-rabbit Ig antiserum, respectively (all from Santa Cruz Biotechnology). The membranes were developed using ECL (Pierce) and autoradiography.
Chemotaxis assays
Migration assays of activated memory B cells were performed using 5-μm Costar Transwell plates (Corning). CXCL12 (100 ng/ml), CXCL13 (1000 ng/ml), or CCL21 (600 ng/ml) (32) were diluted in RPMI 1640 containing 0.5% BSA (chemotaxis medium) and added to wells of a 24-well plate in 600 μl. Chemotaxis medium was used as a control for basal cell migration. CFSE-labeled memory B cells were activated with CD40L, IL-2, and IL-10 for two lots of 4 days, then were washed, and were resuspended in chemotaxis medium, and 5 x 105 cells were added to the upper chamber of the Transwell in 100 μl. Plates were incubated for 4 h at 37°C. A known number of Calibrite beads was then added to each bottom well before harvesting, and the number of B cells that had migrated was calculated. The migrated population was stained with anti-CD27 and CD38 mAb to determine the proportions of different populations of activated memory B cells which were then used to calculate the absolute number of cells of each population that had migrated. Migration was calculated as the percentage of input cells.
Results
CD27 is up-regulated on activated human memory B cells in vitro in an IL-10-dependent and division-linked manner
Human PC express a higher level of CD27 than memory B cells (3, 6, 9, 14, 33). Therefore, it was of interest to investigate the expression of CD27 on memory B cells cultured under conditions that induce their proliferation and differentiation to CD38+ B cells, which contain a population of ISC. Expression of CD27 on CD40L-stimulated memory B cells did not greatly change with division (Fig. 1a, left panel), remaining at a level comparable to cells before culture (data not shown). Addition of IL-10 alone or in combination with IL-2 resulted in the appearance of a population of cells expressing an increased level of CD27 (Fig. 1a, middle and right panels). Notably, this population increased in frequency with cell division in a manner analogous to the appearance of the CD38+ cells (Fig. 1, a and b).
FIGURE 1. Up-regulation of CD27 on activated memory B cells is IL-10 dependent and increases with cell division. CFSE-labeled human splenic memory B cells were cultured with CD40L alone or with IL-10, or IL-2 and IL-10 for 5 days. After this time, the expression of CD27 (a) and CD38 (b) was determined in the context of cell division (i.e., vs CFSE). The values in all panels represent the mean percentage of cells (±SEM) expressing CD38 or CD27 from four independent experiments. c, Coexpression of CD38 and CD27 by memory B cells activated with CD40L, IL-2, and IL-10 was determined according to division history by gating on consecutive divisions. The percentages of cells with distinct phenotypes in divisions 3–7 is indicated.
It was next determined whether CD38+ B cells generated from memory B cells coexpressed elevated levels of CD27, or whether a CD27highCD38– population was also generated. Because the increase in expression of both CD27 and CD38 was division-linked (Fig. 1, a and b), this analysis was performed by assessing their expression on cells in different divisions. Cells that had undergone <2 divisions (i.e., division 0–2) expressed uniform levels of CD27 and CD38. However, after approximately the third division, a population of cells appeared that had up-regulated expression of these molecules (Fig. 1c). With each subsequent division, multiple subsets of cells could be resolved such that in later divisions cells were either CD38+CD27high, CD38±CD27high, CD38+CD27low, or CD38±CD27low (Fig. 1c). Thus, when memory B cells are stimulated with T cell help, they are capable of generating numerous populations of cells discernible by division history and the differential expression of CD27 and CD38.
Proliferating B cells expressing increased levels of CD27 are enriched for ISC
The ability of distinct populations of activated B cells, defined by division history and differential expression of CD27 and CD38, to produce Ig was next determined. For these studies, B cells were initially resolved into populations that had undergone the following: <3 divisions (population 1); several rounds of division and remained CD38– (population 2); or undergone the same number of cell divisions as population 2 yet had differentiated to become CD38+ (population 3; see Fig. 1b; Refs. 17 and 25). ISC increased in frequency in populations 2 (CD38–) and 3 (CD38+) of activated memory B cells, compared with population 1 (Fig. 2a) (17). However, these populations were heterogeneous as demonstrated by <40% of them secreting IgM, IgG, and IgA (Fig. 2a). Further heterogeneity was apparent from the finding that within the CD38– and CD38+ populations in the later divisions, there were cells that expressed different levels of CD27 (i.e., low or high; Fig. 1c). To determine whether the ISC in populations 2 and 3 were partitioned into subsets expressing different levels of CD27, population 1, population 2 CD27low, population 2 CD27high, population 3 CD27low, or population 3 CD27high were examined by ELISPOT. For both populations 2 and population 3, ISC were enriched in the CD27high subset, containing up to 5-fold more ISC than the CD27low subset (Fig. 2b). This analysis revealed that up to 80% of these cells were ISC, an 2-fold increase over the total population 2 or 3 (compare Fig. 2, a and b).
FIGURE 2. ISC are enriched in populations expressing increased levels of CD27. a and b, CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 4 days, washed, and then re-cultured for an additional 4 days under the same conditions. Activated B cells defined as population 1, 2, or 3 (a, see Fig. 1a), or population 1 (b), population 2 CD27low or CD27high, or population 3 CD27low or CD27high were sorted into ELISPOT plates precoated with Ig H chain specific Ab and the proportion of cells secreting IgM (), IgG (), or IgA () determined. The results presented in a and b are from the same experiment. c, CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 5 days. Cells defined according to division history and expression of CD27 and CD38 were sort-purified and recultured for an additional 2 days with CD40L, IL-2, and IL-10. The amount of Ig secreted was determined by Ig H chain specific immunoassays. The values represent the mean Ig production from each population of activated memory B cells from three independent experiments.
To extend these findings, the five different populations of activated human memory B cells were sort-purified and recultured for a further 2 days with CD40L, IL-2, and IL-10, and the amount of secreted Ig was determined. Consistent with the ELISPOT data, population 1 B cells secreted very little Ig, while population 3 CD27high B cells secreted the greatest amounts of Ig, exceeding that by the other populations by 2- to 10-fold (Fig. 2c). Population 2 CD27high B cells produced significant amounts of Ig, albeit less than those produced by population 3 CD27high B cells (Fig. 2c). Population 2 B cells undergo less proliferation and greater apoptosis than population 3 B cells (17, 20, 24, 25)—this is the likely explanation for the reduced levels of Ig detected in cultures of these cells following the secondary culture compared with population 3 CD27high B cells, and the apparent disparity between the data obtained for ELISPOT and ELISA analysis of these cells (Fig. 2c). Thus, up-regulation of CD27 on divided memory B cell blasts facilitates the reliable detection of the predominant populations of ISC present in these cultures, while coexpression of CD38 and increased CD27 identifies cells capable of secreting the greatest amounts of Ig.
Phenotypic characterization of plasmablasts generated in vitro
In addition to increasing expression of CD38 and CD27, human PC alter the expression of other molecules such that their phenotype is distinct from that of naive or memory B cells. Compared with mature B cells, primary PC down-regulate CD20, CD21, CD22, CD84, BAFF-R, CXCR5, CCR7, and, to a lesser extent, CD19, CD40, and HLA-DR (3, 32). In contrast, expression of CD31, CD39, CD44, CD49d, CD86, and CD95 is increased while CD45 and CXCR4 remain unchanged (3, 32). Similarly, BM PC acquire expression of CD138 (5). To extend the characterization of ISC generated from memory B cells in vitro, we determined the phenotype of activated cells corresponding to population 1, and populations 2 and 3 that were either CD27low or CD27high. The phenotype of the ISC (i.e., CD27high populations 2 and 3) bore a striking resemblance to each other as well as to in vivo-derived PC inasmuch that expression of CD19, CD20, CD21, CD22, CD40, CD84, CXCR5, HLA-DR, and BAFF-R were reduced compared with the nonsecreting cell in population 1 as well as the CD27low subsets of populations 2 and 3 (Fig. 3; Table I). Similarly, CD39, CD95 (Table I), CD86, CD44, CD49d, and CD62L (Fig. 3) were uniformly expressed on CD27high ISC in populations 2 and 3 while expression of these molecules on the CD27low B cells was heterogeneous, with bimodal expression of some molecules being detected. CD31 and BCMA were weakly induced on activated B cells in populations 2 and 3, while expression of CD45 and CXCR4 was similar on all populations (Fig. 3, Table I), consistent with the phenotype of primary PC (3, 5, 32). No induction of CD138 expression was detected on the activated splenic memory B cells (Table I), analogous to its absence from primary PC in human spleen (3) and some in peripheral blood and tonsil (5). Collectively, the phenotypes of CD27high cells in populations 2 and 3 are very similar, consistent with the functional similarities of these cells (Fig. 2b). Remarkably, the phenotype of population 1 B cells was similar to resting human memory B cells (Fig. 3, Table I; Refs. 3 and 26), suggesting that these cells remained nondifferentiated with the potential to expand the memory B cell pool.
FIGURE 3. Memory B cells alter their phenotype during in vitro differentiation. CFSE-labeled human memory B cells were cultured as described for Fig. 2a. The cells were incubated with anti-CD27 and CD38 mAbs, and the expression of the indicated molecule on cells corresponding to population 1, population 2 CD27low or CD27high, and population 3 CD27low or CD27high was assessed by flow cytometry. For each plot, the thick and thin histograms represent the fluorescence of cells incubated with the specific or control mAb, respectively. Fluorescence was measured on a log scale.
Table I. Memory B cells alter their phenotype during in vitro proliferation and differentiation to IGSa
Differential expression of transcription factors by distinct populations of activated B cells
Differentiation of mature B cells into ISC is accompanied by changes in expression of specific transcription factors. For instance, while Pax-5 is expressed by naive and memory B cells, its expression is extinguished in PC. Similarly, expression of Bcl-6 is restricted to germinal center B cells (34) while Blimp-1 and XBP-1 are induced at the PC stage (3, 23, 35, 36, 37). Therefore, expression of these transcription factors during the in vitro differentiation process that yields ISC from memory B cells was assessed. Analysis of activated memory B cells revealed that expression of Pax-5 was maintained in population 1 (Fig. 4a, lane 2), but decreased in cells that had undergone further proliferation (Fig. 4a, lane 3–6). Interestingly, while Pax-5 was absent from population 3 CD27high B cells, it remained detectable in the CD27low subset of population 3 (Fig. 4a, lane 5). It is possible that the heightened expression of Pax-5 by these cells prevents them from becoming efficient ISC (Fig. 2), because Pax-5 may counter the inductive effect of XBP-1 on the differentiation of these cells toward an ISC fate (23). In contrast to Pax-5, Blimp-1 (Fig. 4b) and XBP-1 (Fig. 4c) were low in population 1 (lane 2) and incrementally increased as the B cells underwent proliferation and differentiation to an effector ISC (lanes 3–6), such that CD27high cells within population 3 expressed the highest levels of these genes. Bcl-6 was induced in activated B cells, down-regulated in population 2 and again increased as the cells acquired expression of CD38 to become population 3 (Fig. 4d).
FIGURE 4. In vitro-generated ISC express transcription factors consistent with commitment to the PC lineage. Population 1, population 2 CD27low or population 2 CD27high, population 3 CD27low, or population 3 CD27high B cells were isolated by sorting. a–e, RNA was extracted from the sort-purified populations and transcribed into cDNA. The amounts of cDNA were then normalized for expression of GAPDH (bottom panel; e) and used as template to determine the relative expression levels of Pax-5 (a), Blimp-1 (b), XBP-1 (c), and Bcl-6 (d) by semiquantitative PCR. Molecular grade dH2O (lane 1) was used as a negative control. f–h, Unsorted (total B cells) or sort-purified populations of activated B cells were solubilized in lysis buffer, and expression of Pax-5 (f) and Blimp-1 (g) was determined by Western blotting using specific Ab. h, Membranes were also probed with anti-SHP-2 Ab to demonstrate similar protein loading.
The differential expression of Pax-5 and Blimp-1 by the functionally distinct subsets of activated B cells was also examined at the protein level. For these experiments, activated memory B cells were sorted into: population 1, population 2 CD27low and CD27high, as well as population 3 CD27high. Insufficient numbers of population 3 CD27low cells were collected, thereby precluding analysis of these cells. Pax-5 and Blimp-1 were expressed in unsorted activated memory B cells (Fig. 4, f and g, lane 1), consistent with previous reports (38, 39). However, analysis of these proteins in the populations of sorted B cells demonstrated that Pax-5 was dramatically reduced once the memory B cells became CD27high (Fig. 4f, lanes 2–5). In contrast, expression of Blimp-1 was low to absent in population 1 B cells (Fig. 4g, lane 2), but increased once the cells underwent substantial cell division (Fig. 4g, lane 3–5). These results confirm the differences observed at the mRNA level (Fig. 4, a and f; b and g). Thus, the differential expression of transcription factors by phenotypically distinct populations of activated human memory B cells correlated with differences in function, with the ISC populations expressing increased levels of Blimp-1 and XBP-1, and lower levels of Pax-5.
Effect of BAFF and CD70 on the generation of different populations of activated memory B cells
The number of B cells in populations 2 and 3 can be influenced by CD40L or BAFF in the secondary cultures. CD40L is critical for the persistence of population 2 B cells, and can increase population 3 B cells, while BAFF preferentially favors the survival of population 3 (17, 25). Interestingly, culture of activated memory B cells with CD27 ligand (CD70), another TNF superfamily molecule, can increase the proportion of CD38+ B cells generated from memory B cells (30). However, the effect of CD70 on the number of surviving B cells, or different subsets of activated B cells, has not been examined. Therefore, we investigated whether the different B cell populations preferentially responded to secondary stimulation through different BAFF receptors or CD27.
BAFF increased the survival of population 2 CD27high B cells, as well as both subsets of population 3, yet had no significant effect on population 1 nor the CD27low subset of population 2 (Fig. 5a). However, the greatest effect of BAFF was on population 3 CD27high B cells, where almost 5-fold more cells were generated compared with secondary cultures containing IL-2 and IL-10 alone (Fig. 5a). Consequently, in the presence of BAFF, this population of cells dominated the culture, comprising >50% of total cells, compared with cultures containing IL-2 and IL-10 alone, where the CD38+CD27high cells represented <30%. When activated memory B cells were cultured with CD70 transfectants, there was an increase in the number of population 1 B cells, as well as the CD27high cells in populations 2 and 3 (Fig. 5b). Consistent with the down-regulation in CD27 expression, there was a lesser effect on the CD27low B cells (Fig. 5b). The ability of CD70-expressing transfectants to increase the number of CD27high B cells (i.e., ISC) correlated with increased Ig production in these cultures (data not shown). Thus, both BAFF and CD70 are capable of increasing the generation of ISC from memory B cells.
FIGURE 5. BAFF and CD70 increase the recovery of Ig-secreting effector cells generated from memory B cells. CFSE-labeled memory B cells were initially cultured with CD40L, IL-2, and IL-10 for 4 days, harvested, washed, and then recultured with IL-2 and IL-10 in the absence () or presence of BAFF () (a), fixed untransfected P815 cells () (b), or fixed CD70-expressing P815 cells (CD70 Tf; ). After a further 4 days, the total number of cells corresponding to population 1, population 2 CD27low and CD27high, and population 3 CD27low and CD27high was calculated by multiplying total cell number by the frequency of each of these cell subsets. The values in a represent the mean ± SEM of three independent experiments; those in b are representative data from one of two separate experiments.
Altered responsiveness to lymphoid chemokines characterizes the generation of ISC
A characteristic of B cell development and differentiation is the alteration in responsiveness to chemokines. For instance, B cells increase responsiveness to the lymphoid chemokines CXCL12, CXCL13, and CCR7 ligands (CCL19, 21) as they develop from pro-B cells in the BM through to follicular B cells in the spleen (40). In contrast, PC lose responsiveness to CXCL13 and CCR7 ligands, yet retain the ability to respond to CXCL12 (32, 41, 42). Given the alteration in expression of receptors for these chemokines on the different populations of activated memory B cells, the chemotactic responses of these populations was next examined. Population 1 B cells exhibited chemotactic responses to CXCL12, CXCL13, and CCL21 (Fig. 6). As the activated memory B cells underwent proliferation and differentiation to ISC, their responsiveness to CXCL13 and CCL21 gradually declined such that population 3 CD27high B cells were almost unresponsive to these chemotactic ligands (Fig. 6). In contrast, activated B cells continued to respond to CXCL12, with an increased response being evident for population 3 CD27high B cells (Fig. 6). Thus, in vitro-generated ISC behave analogously to primary PC with respect to chemotactic responses.
FIGURE 6. In vitro-generated ISC retain responsiveness to CXCL12, but not CXCL13 or CCL21. CFSE-labeled memory B cells were cultured as described for Fig. 2a. Activated memory B cells (5 x 105/well) were loaded into the upper chamber of a Transwell, and medium alone, CXCL13 (1000 ng/ml), CXCL12 (100 ng/ml), or CCL21 (600 ng/ml) were added to the lower wells. Cells were allowed to migrate for 4 h at 37°C. Migrated cells were harvested from the lower wells and stained with anti-CD27 and CD38 mAb, enabling resolution of population 1, population 2 CD27low or CD27high, or population 3 CD27low or CD27high B cells. The different symbols (, ?) represent the percentage of cells migrating in two different experiments; the columns represents the mean of the individual values.
Discussion
The generation of ISC from precursor cells is a complex and highly regulated process. Here, we have tracked in detail the in vitro differentiation of human memory B cells into effector cells capable of secreting Ig. Different stages of differentiation could be resolved by concomitantly comparing the division history of activated B cells and surface expression of CD27 and CD38 (Figs. 1, 2, and 7). Although CD27 is constitutively expressed by human memory B cells (26, 28), it was up-regulated on some of them as they proliferated in response to CD40L and IL-10, but not CD40L alone (Fig. 1). This parallels our previous observations of division-linked acquisition of CD38 expression by memory B cells stimulated with CD40L and IL-10, and CD138 on stimulated murine B cells, and the subsequent generation of ISC (17, 25, 43). Our data also demonstrated that the frequency of CD38+CD27high B cells appearing in cultures of stimulated memory B cells increased with cell division (Fig. 1c). Furthermore, these cells secreted high Ig levels (Fig. 2) and expressed a transcriptional profile comparable to primary PC (Fig. 4). Thus, this phenotype reliably identifies ISC generated in vitro and allows comparison between human ISC generated in vivo and in vitro (Fig. 7).
FIGURE 7. Model of terminal B cell differentiation. In human lymphoid tissues, naive B cells become activated by Ag and costimulatory factors and form a germinal center where mutation of Ig V region genes and Ig isotype switching occurs. Upon exiting the germinal center, B cells differentiate into either memory B cells or early plasmablasts, identified by increased expression of CD27. Plasmablasts can also rise directly from memory B cells upon re-exposure to specific Ag. Up-regulation of CD38 characterizes further differentiated ("late") plasmablasts. These cells can migrate to the BM, gut, red pulp of spleen, or mucosal epithelium of tonsil under the direction of specific chemokines (CXCL12, CCL25, CCL28; see Ref. 59 ). Once in these microenvironmental niches, ISC compete for survival signals facilitating differentiation into long-lived PC (see Refs. 23 , 32 , and 59 ). Factors identified that support the survival of (i.e., CD40L, CD70, BAFF, IL-6, stromal cells), and the phenotypes and transcription factors expressed by, each stage of terminal B cell differentiation are indicated. Some memory B cells can also yield alternative phenotypic and functional states which may correspond to memory B cell precursors that replenish the memory B cell population (detailed in Discussion).
In addition to increased production of Ig, the CD38+CD27high B cells exhibited several other features of primary PC. First, similar to PC in spleen, tonsil, and PB, CD38+CD27high ISC generated in vitro exhibit reduced expression of CD20, CD21, CD22, CXCR5, and BAFF-R; sustained expression of CD19, CD39, CD40, CD45, CD86, CXCR4, and HLA-DR; and lacked expression of CD138 (Fig. 3, Table I; Refs. 3 , 5 , 32 , and 44). Second, CD38+CD27high ISC retained responsiveness to CXCL12, while responses to CXCL13 and CCL21 were greatly diminished (Fig. 6), akin to primary PC (32, 41, 42). Lastly, expression of Blimp-1 and XBP-1 was greatest, and Pax-5 extinguished, in in vitro CD38+CD27high ISC compared with other populations of activated B cells (Figs. 4 and 7). Similar to alterations in surface phenotype, these molecular changes recapitulate the differential expression of Pax-5, Blimp-1 and XBP-1 observed for mature B cells and primary PC (3, 5, 7, 36, 38), and complement our recent report of gene expression by ISC generated from mouse B cells in vitro in response to T cell-dependent stimuli (43). Similar observations regarding expression of transcription factors have been obtained for human B cells activated in vitro to differentiate toward ISC (38, 39), or in murine splenic B cells following immunization (45). However, a disadvantage of these previous studies was that they examined bulk populations of activated B cells, rather than distinct subsets, and did not correlate mRNA levels with protein (38, 39, 45). Consequently, it was unclear in which cells these transcriptional changes were occurring. Notably, the proliferative activity (17, 20), coupled with Bcl-6 expression and responsiveness to BAFF (Figs. 4 and 5; Ref. 25) also distinguished in vitro CD38+CD27high ISC from primary human PC. The latter population contains noncycling cells which lack expression of Bcl-6, and do not respond to BAFF (3, 24, 25, 32), revealing them to be terminally differentiated. In contrast, in vitro-derived CD38+CD27high ISC likely correspond to plasmablasts, such as those detected in reactive plasmacytosis, or following infection (21) (Fig. 7). The similarities between plasmablasts generated in vitro from activated human memory B cells and those detected in vivo with respect to cell surface phenotype, expression of transcription factors, proliferation and function are remarkable (compare Fig. 3, Table I to Ref. 21), and thus illustrates the validity of our in vitro culture system to examine the processes involved in this aspect of human B cell differentiation.
Another novel finding in our study was the identification of a CD38–CD27high population with a phenotype and capacity to secrete Ig that was identical with the CD38+CD27high B cells (Figs. 2 and 3). Despite these similarities, some important differences were found between these two populations of in vitro-derived ISC. First was their responsiveness to BAFF. This current study refined our previous analysis (25) by revealing that the greatest effect of BAFF was on the viability of CD38+CD27high ISC, while its effects on CD38–CD27high ISC were modest (Figs. 5 and 7). Second, CD38+ B cells proliferate at a greater rate than CD38– B cells (17, 20, 25), revealing CD38+CD27high ISC to be plasmablasts. A molecular explanation for this difference may be attributable to differential expression of Bcl-6. CD38– B cells (Population 2) expressed less Bcl-6 than CD38+ B cells (population 3; Fig. 4). Interestingly, Bcl-6 can regulate the cell cycle by controlling expression of genes such as c-myc (46). This is further exemplified by the finding that the frequency of proliferating cells in Bcl-6 transgenic mice is markedly enhanced compared with control mice (47), and that overexpression of dominant-negative Bcl-6 arrested proliferation of a human B cell line (46). Thus, increased expression of Bcl-6 by CD38+ B cells may provide them with the molecular machinery required for differentiation into rapidly dividing plasmablasts. An additional explanation for the differences between the CD38–CD27high and CD38+CD27high ISC may be that these cells arise from different progenitor memory B cells, namely IgM-expressing or Ig isotype switched memory cells (24, 28). The finding that 1) CD38–CD27high ISC produced a greater proportion of IgM than CD38+CD27high ISC (Fig. 2), 2) BAFF preferentially induces Ig secretion by isotype switched memory B cells (25), and 3) a greater frequency of isotype switched B cells enter their first cell division than IgM-memory B cells (20) would be consistent with the proposal that IgM-memory B cells preferentially yields CD38–CD27high ISC while CD38+CD27high ISC arise from switched memory B cells.
The coexpression of Blimp-1 and Bcl-6 by B cells that have differentiated to become ISC (CD27highCD38+; Fig. 4) was notable because these transcription factors cross-regulate the expression and function of each other (23, 46, 48). Kallies et al. (31) recently described the generation of a mouse line where GFP was introduced into the Blimp-1 locus. This study elegantly demonstrated that all GFP+ cells were ISC, irrespective of surface phenotype. Interestingly, the stage of PC maturation could be inferred from the relative expression of Blimp-1, with cells expressing intermediate levels of Blimp-1 (i.e., Blimp-1/GFPint) representing immature PC while Blimp-1/GFPhigh cells were mature PC (31). Importantly, a population of Blimp-1/GFPint cells also expressed Bcl-6 and were rapidly dividing, leading to the suggestion that they are plasmblasts (31). These data are consistent with our proposal that CD38+CD27high cells, that express Blimp-1 and Bcl-6, are plasmblasts (Fig. 7), and most likely represent the human counterpart of the murine Blimp-1/GFPint cells (31). Another recent study also found that murine B cells could be induced to differentiate into ISC that expressed both Bcl-6 and Blimp-1, but not Pax-5, when stimulated with IL-21 (49). Thus, because both IL-21 and IL-10 are potent growth and differentiation factors for human (IL-10, IL-21; Refs. 50, 51, 52) and murine B cells (IL-21 only; Refs. 49 and 51), it is likely that the molecular alterations induced by these cytokines are similar.
Our examination also identified two other populations of B cells that appear to have distinct functions and fates: the CD38–CD27low and CD38+CD27low cells that were nonsecreting cells. Interestingly, only 50% of PC identified in human or murine lymphoid tissues by a CD38++CD20± or CD138+B220± phenotype, respectively, cells secrete Ig (3, 53). Thus, it is possible that the CD38+CD27low nonsecreters in vitro (Fig. 2b) correspond to these cells in vivo. Alternatively, CD38+CD27low cells may contain precursors of CD38+CD27high cells and could therefore undergo further differentiation to this phenotype and function (Fig. 7). This indeed appears to occur because CD38+CD27low cells can yield CD38+CD27high cells, as well as produce large amounts of Ig, following isolation and subsequent culture (not shown). Similarly, the CD38–CD27low cells may correspond to a pool of memory B cell precursors that replenish the memory B cell population to prevent depletion of these cells following their differentiation to ISC (53, 54).
Our finding of a CD38– population of ISC suggests the existence of an analogous population in vivo. Indeed, it has been reported that 60–80% of human ISC expressed CD38 (55, 56, 57). More recently, a CD38– population of ISC, that appeared to be the precursor of CD38+ ISC, was identified in human tonsils (44). These CD38– ISC expressed CD19, CD27, CD45, HLA-DR, and Blimp-1 (44), analogous to ISC generated in vitro (Fig. 3, Table I) and in vivo (3, 4, 5, 7). Thus, CD38–CD27high ISC generated in vitro from human splenic memory B cells may correspond to the 20–40% of ISC detected ex vivo that were also CD38–. Our findings regarding human ISC are also consistent with the recent descriptions of murine PC that lack CD138 (31, 58). Taken together, these studies reveal the vast heterogeneity in different populations of ISC both in vivo and in vitro and infer that, in the absence of a definitive cell surface marker, ISC cannot by solely defined by phenotype. We propose that human ISC can be resolved by increased expression of CD27, while the coexpression of CD38 delineates the ISC population into an early stage of commitment to the ISC lineage (CD38–) and plasmablasts (CD38+) (Fig. 7). These findings will further our understanding of the complex regulation of the pathways B cells undergo during differentiation to an effector cell.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Drs. Marylin Kehry, Rene de Waal Malefyt, and Susan Kalled for generously providing reagents, the Australian Red Cross Blood Service for providing human spleens, Tara Macdonald and Joseph Webster for cell sorting, and Prof. Tony Basten for critical review of this manuscript.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by the National Health and Medical Research Council (NHMRC) of Australia. S.G.T. is the recipient of an RD Wright Biomedical Career Development Award from the NHMRC; P.D.H. is a Principal Research Fellow of the NHMRC; L.M.C. is a Senior Research Fellow of the NHMRC.
2 Address correspondence and reprint requests to Dr. Stuart Tangye, Centenary Institute of Cancer Medicine and Cell Biology, Locked Bag No. 6, Newtown 2042, New South Wales, Australia. E-mail address: s.tangye@centenary.usyd.edu.au
3 Abbreviations used in this paper: PC, plasma cell; BM, bone marrow; ISC, Ig-secreting cell; SA, streptavidin; BAFF, B-cell activating factor belonging to the TNF family; BAFF-R, BAFF receptor; BCMA, B cell maturation Ag; TACI, transmembrane activator of and CAML interactor; XBP-1, X box-binding protein-1.
Received for publication October 25, 2004. Accepted for publication January 19, 2005.
References
Terstappen, L. W., S. Johnsen, I. M. Segers-Nolten, M. R. Loken. 1990. Identification and characterization of plasma cells in normal human bone marrow by high-resolution flow cytometry. Blood 76:1739.
Leo, R., M. Boeker, D. Peest, R. Hein, R. Bartl, J. E. Gessner, J. Selbach, G. Wacker, H. Deicher. 1992. Multiparameter analyses of normal and malignant human plasma cells: CD38++, CD56+, CD54+, cIg+ is the common phenotype of myeloma cells. Ann. Hematol. 64:132.
Ellyard, J. E., D. T. Avery, T. G. Phan, P. D. Hodgkin, S. G. Tangye. 2004. Ag-selected Ig-secreting cells persist in human spleen as well as bone marrow. Blood 103:3805.
Merville, P., J. Dechanet, A. Desmouliere, I. Durand, O. de Bouteiller, P. Garrone, J. Banchereau, Y. J. Liu. 1996. Bcl-2+ tonsillar plasma cells are rescued from apoptosis by bone marrow fibroblasts. J. Exp. Med. 183:227
Medina, F., C. Segundo, A. Campos-Caro, I. Gonzalez-Garcia, J. A. Brieva. 2002. The heterogeneity shown by human plasma cells from tonsil, blood, and bone marrow reveals graded stages of increasing maturity, but local profiles of adhesion molecule expression. Blood 99:2154.
Farstad, I. N., H. Carlsen, H. C. Morton, P. Brandtzaeg. 2000. Immunoglobulin A cell distribution in the human small intestine: phenotypic and functional characteristics. Immunology 101:354.
Medina, F., C. Segundo, A. Campos-Caro, I. Salcedo, A. Garcia-Poley, J. A. Brieva. 2003. Isolation, maturational level, and functional capacity of human colon lamina propria plasma cells. Gut 52:383.
Kyle, R. A.. 1991. Plasma cell proliferative disorders. R. Hoffman, and E. J. Benz, and S. J. Shattil, and B. Furie, and H. J. Cohen, eds. Hematology. Basic Principles and Practice 1021. Churchill Livingstone, New York.
Odendahl, M., A. Jacobi, A. Hansen, E. Feist, F. Hiepe, G. R. Burmester, P. E. Lipsky, A. Radbruch, T. Dorner. 2000. Disturbed peripheral B lymphocyte homeostasis in systemic lupus erythematosus. J. Immunol. 165:5970
Jacobi, A. M., M. Odendahl, K. Reiter, A. Bruns, G. R. Burmester, A. Radbruch, G. Valet, P. E. Lipsky, T. Dorner. 2003. Correlation between circulating CD27high plasma cells and disease activity in patients with systemic lupus erythematosus. Arthritis Rheum. 48:1332.
Vernino, L., L. M. McAnally, J. Ramberg, P. E. Lipsky. 1992. Generation of nondividing high rate Ig-secreting plasma cells in cultures of human B cells stimulated with anti-CD3-activated T cells. J. Immunol. 148:404.
Merville, P., J. Dechanet, G. Grouard, I. Durand, J. Banchereau. 1995. T cell-induced B cell blasts differentiate into plasma cells when cultured on bone marrow stroma with IL-3 and IL-10. Int. Immunol. 7:635.
Choe, J., Y. S. Choi. 1998. IL-10 interrupts memory B cell expansion in the germinal center by inducing differentiation into plasma cells. Eur. J. Immunol. 28:508.
Jung, J., J. Choe, L. Li, Y. S. Choi. 2000. Regulation of CD27 expression in the course of germinal center B cell differentiation: the pivotal role of IL-10. Eur. J. Immunol. 30:2437.
Arpin, C., J. Dechanet, C. Van Kooten, P. Merville, G. Grouard, F. Briere, J. Banchereau, Y. J. Liu. 1995. Generation of memory B cells and plasma cells in vitro. Science 268:720.
Arpin, C., J. Banchereau, Y. J. Liu. 1997. Memory B cells are biased towards terminal differentiation: a strategy that may prevent repertoire freezing. J. Exp. Med. 186:931.
Tangye, S. G., D. T. Avery, P. D. Hodgkin. 2003. A division-linked mechanism for the rapid generation of Ig-secreting cells from human memory B cells. J. Immunol. 170:261.
Jelinek, D. F., P. E. Lipsky. 1983. The role of B cell proliferation in the generation of immunoglobulin-secreting cells in man. J. Immunol. 130:2597.
Jego, G., R. Bataille, C. Pellat-Deceunynck. 2001. Interleukin-6 is a growth factor for nonmalignant human plasmablasts. Blood 97:1817
Tangye, S. G., D. T. Avery, E. K. Deenick, P. D. Hodgkin. 2003. Intrinsic differences in the proliferation of naive and memory human B cells as a mechanism for enhanced secondary immune responses. J. Immunol. 170:686.(Danielle T. Avery, Julia )
Plasma cells (PC) or Ig-secreting cells (ISC) are terminally differentiated B cells responsible for the production of protective Ig. ISC can be generated in vitro by culturing human B cells with the T cell-derived stimuli CD40L, IL-2, and IL-10. ISC have traditionally been identified by the increased expression of CD38, analogous to primary human PC, and the acquired ability to secrete Ig. By tracking the proliferation history of activated B cells, we previously reported that the differentiation of memory B cells into CD38+ B cells is IL-10 dependent, and increases in frequency with cell division. However, <50% of CD38+ cells secreted Ig, and there was a population of CD38– ISC. Thus, the PC phenotype of CD38+ cells generated in vitro did not correlate with PC function. To address this, we have examined cultures of activated memory B cells to accurately identify the phenotype of ISC generated in vitro. We found that CD27 is also up-regulated on memory B cells in an IL-10-dependent and division-dependent manner, and that ISC segregated into the CD27high subset of activated memory B cells irrespective of the acquired expression of CD38. The ISC generated in these cultures expressed elevated levels of the transcription factors Blimp-1 and X box-binding protein-1 and reduced levels of Pax-5, and exhibited selective migration toward CXCL12, similar to primary PC. We propose that the differentiation of memory B cells into PC involves a transitional stage characterized by a CD27highCD38– phenotype with the acquired ability to secrete high levels of Ig.
Introduction
One of the primary functions of mature B cells is the production of protective high affinity Ig following terminal differentiation into plasma cells (PC).3 In humans, PC can be identified by the up-regulation in expression of CD38 and the concomitant decrease in CD20 (CD38++CD20±) (1, 2). Cells with this phenotype have been detected in lymphoid tissues such as spleen (3), tonsils (4, 5), intestine (6, 7), bone marrow (BM), and peripheral blood (1, 5). Importantly, PC are overrepresented not only in malignancies (2, 8), but also autoimmune disorders such as systemic lupus erythematosus (9, 10). Therefore, delineating the mechanisms involved in generating these cells is necessary for a complete understanding of the humoral immune response as well as PC dyscrasias.
Ig-secreting cells (ISC) resembling PC can also be generated in vitro by culturing human germinal center or memory B cells with BM stromal cells and/or activated T cells (11, 12), a follicular dendritic-like cell line plus CD40L, IL-2, IL-4, and IL-10 (13, 14) or the T cell-derived stimuli CD40L, IL-2, and IL-10 (15, 16, 17). These cells have typically been identified by the increased expression of CD38, and the acquired ability to secrete high levels of Ig (11, 12, 13, 14, 15, 16, 17). Such cells generated in vitro are rapidly proliferating (17, 18, 19, 20), suggesting they are more akin to plasmablasts detected in vivo (21, 22), rather than mitotically inactive PC (23). By tracking the proliferation history of activated B cells, we have recently found that the differentiation of memory B cells into CD38-expressing cells is IL-10 dependent, and increases in frequency with cell division; that is, CD38+ B cells appear in cultures of memory B cells stimulated with CD40L and IL-10 once the cells have undergone three or more rounds of cell division (17, 20, 24). Furthermore, the survival requirements of the CD38+ B cells differed from CD38– B cells present in the same culture inasmuch that CD38+ B cells became independent of CD40L, and responsive to B-cell activating factor belonging to the TNF family (BAFF), while CD38– B cells required continual stimulation with CD40L for their persistence (11, 17, 25). However, we also found that only 40% of CD38+ cells secreted Ig, and that there was a population of CD38– ISC (17, 25). Thus, the "PC" phenotype of CD38+ cells generated in vitro did not correlate with PC function. To address this discrepancy, we have now examined in greater detail cultures of activated memory B cells in an attempt to accurately identify the phenotype of ISC generated in vitro. We found that CD27 is up-regulated on the surface of memory B cells in an IL-10-dependent and division-dependent manner, and that ISC segregated into the CD27high subset of activated memory B cells irrespective of the acquired expression of CD38. The CD27high ISC exhibited increased expression of the transcription factors Blimp-1 and X box-binding protein-1 (XBP-1), and reduced expression of Pax-5, as well as phenotypic and migratory characteristics similar to those of human primary PC. In contrast, nondifferentiated B cell blasts (i.e., CD38–CD27+) present in the same culture retained features of memory B cells. These results reveal the dynamic ability of human memory B cells to generate multiple cell fates following activation with T cell-dependent stimuli to yield effector cells, namely ISC, as well as sufficiently expand the pool of nondifferentiated memory B cells to increase the size of this population of effector cells that can then be rapidly activated following re-exposure to the immunizing Ag.
Materials and Methods
Reagents
FITC-conjugated anti-CD20 and PE-conjugated anti-CD19 and CD45 mAb were purchased from BD Biosciences; anti-CCR7, PE-conjugated isotype controls, CD21, CD27, CD80, CD86, CD95, and CXCR4 mAb, biotinylated anti-CD44, anti-hamster IgG, and streptavidin (SA)-conjugated to PerCp were purchased from BD Pharmingen. PE-anti-CD22, CD23, CD31, CD38, CD62L, CD138, HLA-DR mAb, biotinylated IgG1 isotype control, and anti-CD38 mAb, allophycocyanin-conjugated isotype control, anti-CD20 and anti-CD38 mAb, and SA-allophycocyanin were purchased from Caltag Laboratories; biotinylated anti-CXCR5 mAb was from R&D Systems; biotinylated anti-human CD27 mAb was from eBioscience; PE-conjugated anti-CD40 mAb (mAb89) was provided by J. Banchereau (Schering Plough Labs, Dardilly, France); PE-conjugated anti-CD39 (A1) mAb has been described previously (26). Purified and biotinylated F(ab')2 of goat anti-human IgM, IgG, or IgA polyclonal Ab were purchased from Southern Biotechnology. Recombinant human BAFF, and mouse anti-BAFF receptor (anti-BAFF-R; clone 9.1), anti-transmembrane activator of and CAML interactor (TACI; clone C4D7) and hamster anti-B cell maturation Ag (BCMA; clone C4E2.2) mAbs were provided by Dr. S. Kalled (Biogen Idec, Cambridge, MA; Ref. 25). Recombinant CD40L expressed as membranes in Sf21 insect cells infected with baculovirus vector containing human CD40L cDNA was provided by Dr. M. Kehry (Boehringer Ingleheim, Ridgefield, CT). IL-2 was purchased from Endogen; IL-10 was provided by Dr. R. de Waal Malefyt (DNAX Research Institute, Palo Alto, CA). CFSE was from Molecular Probes. Recombinant human CXCL12 and CCL21 were purchased from PeproTech; CXCL13 was from R&D Systems.
Generation of CD70 transfectants
A complementary DNA encoding human CD70 was amplified from RNA prepared from the CD70-expressing human B cell line JY by PCR using Pfu polymerase and the following primers (Sigma-Genosys): 5'-GCA TGC GGA TCC TTC CTT CCT TCT CGG CAG CG (BamH1 site underlined), and 3'-GCA TGC GCG GCC GCA ATC AGC AGC AGT GGT CAG GG (NotI site underlined). The resulting product was digested with BamH1 and NotI, and ligated into pcdef3, a derivative of pEF-BOS containing the neomycin-resistance gene (27). The mouse mastocytoma cell line P815 was transfected with CD70/pcdef3 by electroporation, and positive cells were selected initially in the presence of G418 (Bio-Rad), and subsequently by cell sorting.
Cells
Total human B cells (>98% CD19+) were isolated as previously described (17, 28). Memory B cells were isolated by sorting on a FACStarPLUS or FACSVantage (BD Biosciences) following labeling with FITC-anti-CD20 and PE-anti-CD27 mAb and collecting CD27+CD20+ B cells (17, 26).
CFSE labeling and B cell cultures
Primary cultures. Memory B cells were labeled with CFSE (29) and cultured in 48-well plates (4 x 105/ml; BD Labware) for 5 days with CD40L alone (at a predetermined optimal dilution of the membrane preparation; 1/250), or in the presence of IL-10 (100 U/ml), or IL-2 (50 U/ml) plus IL-10. The cells were harvested and expression of CD27 and CD38 was then determined by immunofluorescence and flow cytometric analysis.
Secondary cultures. CFSE-labeled memory B cells were cultured for 4 days with CD40L, IL-2, and IL-10. The cells were harvested, washed, and recultured (2 x 105/ml) with IL-2 and IL-10 in the absence or presence of CD40L (1/500 dilution) or BAFF (2.5 μg/ml) for an additional 4 days (15, 17, 25). In some experiments, the activated memory B cells were recultured with parental P815 or CD70/P815 cells in the presence of IL-2 and IL-10. Before culture, the P815 cells were fixed in 1% formaldehyde, prepared in PBS, for 20 min (30), washed three times with PBS, and cultured in medium for 1 h to remove the formaldehyde. The memory B cells and P815 cells were cultured at a ratio of 5:1.
At the completion of the primary or secondary cultures, a known number of CaliBRITE beads (BD Biosciences) were added to culture wells before harvesting, and the number of viable B cells was calculated as a function of the ratio of beads to live cells (29). All cultures were performed using RPMI 1640 containing L-glutamine (Invitrogen Life Technologies), 10% FCS (CSL), 10 mM HEPES (pH 7.4) (Sigma-Aldrich), 0.1 mM nonessential amino acid solution (Sigma-Aldrich), 1 mM sodium pyruvate (Sigma-Aldrich), 100 U/ml penicillin, 100 μg/ml streptomycin (Sigma-Aldrich), 100 μg/ml Normocin (InVivoGen), and 40 μg/ml apo-transferrin (Sigma-Aldrich) and were conducted at 37°C in a humidified atmosphere containing 5% CO2.
Immunofluorescent staining
For phenotypic analysis, cells were incubated on ice with PE-, allophycocyanin-, or biotinylated specific mAb, or the appropriate isotype control, followed by SA-PerCp and analyzed on a FACSCalibur using CellQuest software (BD Biosciences). Binding of BAFF to activated B cells, as well as assessment of expression of the different BAFF receptors (i.e., BAFF-R, TACF, and BCMA) were determined as described previously (25)
Analysis of Ig secretion
ELISPOT. CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 4 days and then washed and recultured under the same conditions for a further 4 days. The cells were harvested and labeled with PE-anti-CD27 mAb and allophycocyanin-anti-CD38 mAb. Different populations of activated B cells, defined by division history and surface phenotype, were sorted directly into ELISPOT plates (Multiscreen-HA plates; Millipore), and the frequency of cells secreting IgM, IgG, and IgA was determined (17).
ELISA. CFSE-labeled memory B cells were cultured for 5 days with CD40L, IL-2, and IL-10 and then labeled with PE-anti-CD27 and allophycocyanin-anti-CD38 mAbs. Different populations of activated B cells were isolated by cell sorting and then recultured (105 cells/500 μl/well) for a further 2 days with CD40L, IL-2, and IL-10, after which time supernatants were collected and the level of secreted Ig was determined (17).
Analysis of expression of transcription factors
Semiquantitative PCR was used to examine gene expression in different populations of activated human memory B cells. B cell subsets were isolated by cell sorting, total RNA was extracted (Qiagen RNeasy Kit; Qiagen) and then transcribed into cDNA using oligo-dT (Boehringer Mannheim) or random hexamers (Invitrogen Life Technologies) as primer and Superscript II RNase H– reverse transcriptase (Invitrogen Life Technologies). Resulting cDNA was then normalized for expression of the constitutively expressed housekeeping gene GAPDH (5'-CCA CCC ATG GCA AAT TCC ATG GCA, 3'-TCT AGA CGG CAG GTC AGG TCC ACC) and then used as a template for PCR using REDTaq (Sigma-Aldrich) (3). The following primers were used (Sigma-Genosys): Pax-5 5'-GCA TAG TGT CCA CTG GCT CC; Pax-5 3'-CCA GGA GTC GTT GTA CGA GG; BLIMP-1 5'-GAT GCG GAT ATG ACT CTG TGG; BLIMP-1 3'-CTC GGT TGC TTT AGA CTG CTC; XBP-1 5'-GCT CAG ACT GCC AGA GAT CG; XBP-1 3'-GTC CSG AAT GCC CAA CAG G; Bcl-6 5'-CTG ACA GCT GTA TCC AGT TCA CC; Bcl-6 3'-TCT TGG GGC ATC AGC ATC.
Expression of Blimp-1 and Pax-5 protein by activated human memory B cells was also determined. B cell subsets were isolated by cell sorting, and then lysed in ice cold lysis buffer (10 mM Tris-HCl (pH 7.8), 1% Nonidet P-40, 150 mM NaCl, and enzyme inhibitors). Cell lysates were electrophoresed through 12% acrylamide gels containing 0.1% SDS and transferred to polyvinylidene difluoride membranes (Millipore). Membranes were probed with Abs against Blimp-1 (31), Pax-5 (C-20; Santa Cruz Biotechnology) or SHP-2 (Santa Cruz Biotechnology) followed by HRP-conjugated anti-rat, anti-goat, and anti-rabbit Ig antiserum, respectively (all from Santa Cruz Biotechnology). The membranes were developed using ECL (Pierce) and autoradiography.
Chemotaxis assays
Migration assays of activated memory B cells were performed using 5-μm Costar Transwell plates (Corning). CXCL12 (100 ng/ml), CXCL13 (1000 ng/ml), or CCL21 (600 ng/ml) (32) were diluted in RPMI 1640 containing 0.5% BSA (chemotaxis medium) and added to wells of a 24-well plate in 600 μl. Chemotaxis medium was used as a control for basal cell migration. CFSE-labeled memory B cells were activated with CD40L, IL-2, and IL-10 for two lots of 4 days, then were washed, and were resuspended in chemotaxis medium, and 5 x 105 cells were added to the upper chamber of the Transwell in 100 μl. Plates were incubated for 4 h at 37°C. A known number of Calibrite beads was then added to each bottom well before harvesting, and the number of B cells that had migrated was calculated. The migrated population was stained with anti-CD27 and CD38 mAb to determine the proportions of different populations of activated memory B cells which were then used to calculate the absolute number of cells of each population that had migrated. Migration was calculated as the percentage of input cells.
Results
CD27 is up-regulated on activated human memory B cells in vitro in an IL-10-dependent and division-linked manner
Human PC express a higher level of CD27 than memory B cells (3, 6, 9, 14, 33). Therefore, it was of interest to investigate the expression of CD27 on memory B cells cultured under conditions that induce their proliferation and differentiation to CD38+ B cells, which contain a population of ISC. Expression of CD27 on CD40L-stimulated memory B cells did not greatly change with division (Fig. 1a, left panel), remaining at a level comparable to cells before culture (data not shown). Addition of IL-10 alone or in combination with IL-2 resulted in the appearance of a population of cells expressing an increased level of CD27 (Fig. 1a, middle and right panels). Notably, this population increased in frequency with cell division in a manner analogous to the appearance of the CD38+ cells (Fig. 1, a and b).
FIGURE 1. Up-regulation of CD27 on activated memory B cells is IL-10 dependent and increases with cell division. CFSE-labeled human splenic memory B cells were cultured with CD40L alone or with IL-10, or IL-2 and IL-10 for 5 days. After this time, the expression of CD27 (a) and CD38 (b) was determined in the context of cell division (i.e., vs CFSE). The values in all panels represent the mean percentage of cells (±SEM) expressing CD38 or CD27 from four independent experiments. c, Coexpression of CD38 and CD27 by memory B cells activated with CD40L, IL-2, and IL-10 was determined according to division history by gating on consecutive divisions. The percentages of cells with distinct phenotypes in divisions 3–7 is indicated.
It was next determined whether CD38+ B cells generated from memory B cells coexpressed elevated levels of CD27, or whether a CD27highCD38– population was also generated. Because the increase in expression of both CD27 and CD38 was division-linked (Fig. 1, a and b), this analysis was performed by assessing their expression on cells in different divisions. Cells that had undergone <2 divisions (i.e., division 0–2) expressed uniform levels of CD27 and CD38. However, after approximately the third division, a population of cells appeared that had up-regulated expression of these molecules (Fig. 1c). With each subsequent division, multiple subsets of cells could be resolved such that in later divisions cells were either CD38+CD27high, CD38±CD27high, CD38+CD27low, or CD38±CD27low (Fig. 1c). Thus, when memory B cells are stimulated with T cell help, they are capable of generating numerous populations of cells discernible by division history and the differential expression of CD27 and CD38.
Proliferating B cells expressing increased levels of CD27 are enriched for ISC
The ability of distinct populations of activated B cells, defined by division history and differential expression of CD27 and CD38, to produce Ig was next determined. For these studies, B cells were initially resolved into populations that had undergone the following: <3 divisions (population 1); several rounds of division and remained CD38– (population 2); or undergone the same number of cell divisions as population 2 yet had differentiated to become CD38+ (population 3; see Fig. 1b; Refs. 17 and 25). ISC increased in frequency in populations 2 (CD38–) and 3 (CD38+) of activated memory B cells, compared with population 1 (Fig. 2a) (17). However, these populations were heterogeneous as demonstrated by <40% of them secreting IgM, IgG, and IgA (Fig. 2a). Further heterogeneity was apparent from the finding that within the CD38– and CD38+ populations in the later divisions, there were cells that expressed different levels of CD27 (i.e., low or high; Fig. 1c). To determine whether the ISC in populations 2 and 3 were partitioned into subsets expressing different levels of CD27, population 1, population 2 CD27low, population 2 CD27high, population 3 CD27low, or population 3 CD27high were examined by ELISPOT. For both populations 2 and population 3, ISC were enriched in the CD27high subset, containing up to 5-fold more ISC than the CD27low subset (Fig. 2b). This analysis revealed that up to 80% of these cells were ISC, an 2-fold increase over the total population 2 or 3 (compare Fig. 2, a and b).
FIGURE 2. ISC are enriched in populations expressing increased levels of CD27. a and b, CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 4 days, washed, and then re-cultured for an additional 4 days under the same conditions. Activated B cells defined as population 1, 2, or 3 (a, see Fig. 1a), or population 1 (b), population 2 CD27low or CD27high, or population 3 CD27low or CD27high were sorted into ELISPOT plates precoated with Ig H chain specific Ab and the proportion of cells secreting IgM (), IgG (), or IgA () determined. The results presented in a and b are from the same experiment. c, CFSE-labeled memory B cells were cultured with CD40L, IL-2, and IL-10 for 5 days. Cells defined according to division history and expression of CD27 and CD38 were sort-purified and recultured for an additional 2 days with CD40L, IL-2, and IL-10. The amount of Ig secreted was determined by Ig H chain specific immunoassays. The values represent the mean Ig production from each population of activated memory B cells from three independent experiments.
To extend these findings, the five different populations of activated human memory B cells were sort-purified and recultured for a further 2 days with CD40L, IL-2, and IL-10, and the amount of secreted Ig was determined. Consistent with the ELISPOT data, population 1 B cells secreted very little Ig, while population 3 CD27high B cells secreted the greatest amounts of Ig, exceeding that by the other populations by 2- to 10-fold (Fig. 2c). Population 2 CD27high B cells produced significant amounts of Ig, albeit less than those produced by population 3 CD27high B cells (Fig. 2c). Population 2 B cells undergo less proliferation and greater apoptosis than population 3 B cells (17, 20, 24, 25)—this is the likely explanation for the reduced levels of Ig detected in cultures of these cells following the secondary culture compared with population 3 CD27high B cells, and the apparent disparity between the data obtained for ELISPOT and ELISA analysis of these cells (Fig. 2c). Thus, up-regulation of CD27 on divided memory B cell blasts facilitates the reliable detection of the predominant populations of ISC present in these cultures, while coexpression of CD38 and increased CD27 identifies cells capable of secreting the greatest amounts of Ig.
Phenotypic characterization of plasmablasts generated in vitro
In addition to increasing expression of CD38 and CD27, human PC alter the expression of other molecules such that their phenotype is distinct from that of naive or memory B cells. Compared with mature B cells, primary PC down-regulate CD20, CD21, CD22, CD84, BAFF-R, CXCR5, CCR7, and, to a lesser extent, CD19, CD40, and HLA-DR (3, 32). In contrast, expression of CD31, CD39, CD44, CD49d, CD86, and CD95 is increased while CD45 and CXCR4 remain unchanged (3, 32). Similarly, BM PC acquire expression of CD138 (5). To extend the characterization of ISC generated from memory B cells in vitro, we determined the phenotype of activated cells corresponding to population 1, and populations 2 and 3 that were either CD27low or CD27high. The phenotype of the ISC (i.e., CD27high populations 2 and 3) bore a striking resemblance to each other as well as to in vivo-derived PC inasmuch that expression of CD19, CD20, CD21, CD22, CD40, CD84, CXCR5, HLA-DR, and BAFF-R were reduced compared with the nonsecreting cell in population 1 as well as the CD27low subsets of populations 2 and 3 (Fig. 3; Table I). Similarly, CD39, CD95 (Table I), CD86, CD44, CD49d, and CD62L (Fig. 3) were uniformly expressed on CD27high ISC in populations 2 and 3 while expression of these molecules on the CD27low B cells was heterogeneous, with bimodal expression of some molecules being detected. CD31 and BCMA were weakly induced on activated B cells in populations 2 and 3, while expression of CD45 and CXCR4 was similar on all populations (Fig. 3, Table I), consistent with the phenotype of primary PC (3, 5, 32). No induction of CD138 expression was detected on the activated splenic memory B cells (Table I), analogous to its absence from primary PC in human spleen (3) and some in peripheral blood and tonsil (5). Collectively, the phenotypes of CD27high cells in populations 2 and 3 are very similar, consistent with the functional similarities of these cells (Fig. 2b). Remarkably, the phenotype of population 1 B cells was similar to resting human memory B cells (Fig. 3, Table I; Refs. 3 and 26), suggesting that these cells remained nondifferentiated with the potential to expand the memory B cell pool.
FIGURE 3. Memory B cells alter their phenotype during in vitro differentiation. CFSE-labeled human memory B cells were cultured as described for Fig. 2a. The cells were incubated with anti-CD27 and CD38 mAbs, and the expression of the indicated molecule on cells corresponding to population 1, population 2 CD27low or CD27high, and population 3 CD27low or CD27high was assessed by flow cytometry. For each plot, the thick and thin histograms represent the fluorescence of cells incubated with the specific or control mAb, respectively. Fluorescence was measured on a log scale.
Table I. Memory B cells alter their phenotype during in vitro proliferation and differentiation to IGSa
Differential expression of transcription factors by distinct populations of activated B cells
Differentiation of mature B cells into ISC is accompanied by changes in expression of specific transcription factors. For instance, while Pax-5 is expressed by naive and memory B cells, its expression is extinguished in PC. Similarly, expression of Bcl-6 is restricted to germinal center B cells (34) while Blimp-1 and XBP-1 are induced at the PC stage (3, 23, 35, 36, 37). Therefore, expression of these transcription factors during the in vitro differentiation process that yields ISC from memory B cells was assessed. Analysis of activated memory B cells revealed that expression of Pax-5 was maintained in population 1 (Fig. 4a, lane 2), but decreased in cells that had undergone further proliferation (Fig. 4a, lane 3–6). Interestingly, while Pax-5 was absent from population 3 CD27high B cells, it remained detectable in the CD27low subset of population 3 (Fig. 4a, lane 5). It is possible that the heightened expression of Pax-5 by these cells prevents them from becoming efficient ISC (Fig. 2), because Pax-5 may counter the inductive effect of XBP-1 on the differentiation of these cells toward an ISC fate (23). In contrast to Pax-5, Blimp-1 (Fig. 4b) and XBP-1 (Fig. 4c) were low in population 1 (lane 2) and incrementally increased as the B cells underwent proliferation and differentiation to an effector ISC (lanes 3–6), such that CD27high cells within population 3 expressed the highest levels of these genes. Bcl-6 was induced in activated B cells, down-regulated in population 2 and again increased as the cells acquired expression of CD38 to become population 3 (Fig. 4d).
FIGURE 4. In vitro-generated ISC express transcription factors consistent with commitment to the PC lineage. Population 1, population 2 CD27low or population 2 CD27high, population 3 CD27low, or population 3 CD27high B cells were isolated by sorting. a–e, RNA was extracted from the sort-purified populations and transcribed into cDNA. The amounts of cDNA were then normalized for expression of GAPDH (bottom panel; e) and used as template to determine the relative expression levels of Pax-5 (a), Blimp-1 (b), XBP-1 (c), and Bcl-6 (d) by semiquantitative PCR. Molecular grade dH2O (lane 1) was used as a negative control. f–h, Unsorted (total B cells) or sort-purified populations of activated B cells were solubilized in lysis buffer, and expression of Pax-5 (f) and Blimp-1 (g) was determined by Western blotting using specific Ab. h, Membranes were also probed with anti-SHP-2 Ab to demonstrate similar protein loading.
The differential expression of Pax-5 and Blimp-1 by the functionally distinct subsets of activated B cells was also examined at the protein level. For these experiments, activated memory B cells were sorted into: population 1, population 2 CD27low and CD27high, as well as population 3 CD27high. Insufficient numbers of population 3 CD27low cells were collected, thereby precluding analysis of these cells. Pax-5 and Blimp-1 were expressed in unsorted activated memory B cells (Fig. 4, f and g, lane 1), consistent with previous reports (38, 39). However, analysis of these proteins in the populations of sorted B cells demonstrated that Pax-5 was dramatically reduced once the memory B cells became CD27high (Fig. 4f, lanes 2–5). In contrast, expression of Blimp-1 was low to absent in population 1 B cells (Fig. 4g, lane 2), but increased once the cells underwent substantial cell division (Fig. 4g, lane 3–5). These results confirm the differences observed at the mRNA level (Fig. 4, a and f; b and g). Thus, the differential expression of transcription factors by phenotypically distinct populations of activated human memory B cells correlated with differences in function, with the ISC populations expressing increased levels of Blimp-1 and XBP-1, and lower levels of Pax-5.
Effect of BAFF and CD70 on the generation of different populations of activated memory B cells
The number of B cells in populations 2 and 3 can be influenced by CD40L or BAFF in the secondary cultures. CD40L is critical for the persistence of population 2 B cells, and can increase population 3 B cells, while BAFF preferentially favors the survival of population 3 (17, 25). Interestingly, culture of activated memory B cells with CD27 ligand (CD70), another TNF superfamily molecule, can increase the proportion of CD38+ B cells generated from memory B cells (30). However, the effect of CD70 on the number of surviving B cells, or different subsets of activated B cells, has not been examined. Therefore, we investigated whether the different B cell populations preferentially responded to secondary stimulation through different BAFF receptors or CD27.
BAFF increased the survival of population 2 CD27high B cells, as well as both subsets of population 3, yet had no significant effect on population 1 nor the CD27low subset of population 2 (Fig. 5a). However, the greatest effect of BAFF was on population 3 CD27high B cells, where almost 5-fold more cells were generated compared with secondary cultures containing IL-2 and IL-10 alone (Fig. 5a). Consequently, in the presence of BAFF, this population of cells dominated the culture, comprising >50% of total cells, compared with cultures containing IL-2 and IL-10 alone, where the CD38+CD27high cells represented <30%. When activated memory B cells were cultured with CD70 transfectants, there was an increase in the number of population 1 B cells, as well as the CD27high cells in populations 2 and 3 (Fig. 5b). Consistent with the down-regulation in CD27 expression, there was a lesser effect on the CD27low B cells (Fig. 5b). The ability of CD70-expressing transfectants to increase the number of CD27high B cells (i.e., ISC) correlated with increased Ig production in these cultures (data not shown). Thus, both BAFF and CD70 are capable of increasing the generation of ISC from memory B cells.
FIGURE 5. BAFF and CD70 increase the recovery of Ig-secreting effector cells generated from memory B cells. CFSE-labeled memory B cells were initially cultured with CD40L, IL-2, and IL-10 for 4 days, harvested, washed, and then recultured with IL-2 and IL-10 in the absence () or presence of BAFF () (a), fixed untransfected P815 cells () (b), or fixed CD70-expressing P815 cells (CD70 Tf; ). After a further 4 days, the total number of cells corresponding to population 1, population 2 CD27low and CD27high, and population 3 CD27low and CD27high was calculated by multiplying total cell number by the frequency of each of these cell subsets. The values in a represent the mean ± SEM of three independent experiments; those in b are representative data from one of two separate experiments.
Altered responsiveness to lymphoid chemokines characterizes the generation of ISC
A characteristic of B cell development and differentiation is the alteration in responsiveness to chemokines. For instance, B cells increase responsiveness to the lymphoid chemokines CXCL12, CXCL13, and CCR7 ligands (CCL19, 21) as they develop from pro-B cells in the BM through to follicular B cells in the spleen (40). In contrast, PC lose responsiveness to CXCL13 and CCR7 ligands, yet retain the ability to respond to CXCL12 (32, 41, 42). Given the alteration in expression of receptors for these chemokines on the different populations of activated memory B cells, the chemotactic responses of these populations was next examined. Population 1 B cells exhibited chemotactic responses to CXCL12, CXCL13, and CCL21 (Fig. 6). As the activated memory B cells underwent proliferation and differentiation to ISC, their responsiveness to CXCL13 and CCL21 gradually declined such that population 3 CD27high B cells were almost unresponsive to these chemotactic ligands (Fig. 6). In contrast, activated B cells continued to respond to CXCL12, with an increased response being evident for population 3 CD27high B cells (Fig. 6). Thus, in vitro-generated ISC behave analogously to primary PC with respect to chemotactic responses.
FIGURE 6. In vitro-generated ISC retain responsiveness to CXCL12, but not CXCL13 or CCL21. CFSE-labeled memory B cells were cultured as described for Fig. 2a. Activated memory B cells (5 x 105/well) were loaded into the upper chamber of a Transwell, and medium alone, CXCL13 (1000 ng/ml), CXCL12 (100 ng/ml), or CCL21 (600 ng/ml) were added to the lower wells. Cells were allowed to migrate for 4 h at 37°C. Migrated cells were harvested from the lower wells and stained with anti-CD27 and CD38 mAb, enabling resolution of population 1, population 2 CD27low or CD27high, or population 3 CD27low or CD27high B cells. The different symbols (, ?) represent the percentage of cells migrating in two different experiments; the columns represents the mean of the individual values.
Discussion
The generation of ISC from precursor cells is a complex and highly regulated process. Here, we have tracked in detail the in vitro differentiation of human memory B cells into effector cells capable of secreting Ig. Different stages of differentiation could be resolved by concomitantly comparing the division history of activated B cells and surface expression of CD27 and CD38 (Figs. 1, 2, and 7). Although CD27 is constitutively expressed by human memory B cells (26, 28), it was up-regulated on some of them as they proliferated in response to CD40L and IL-10, but not CD40L alone (Fig. 1). This parallels our previous observations of division-linked acquisition of CD38 expression by memory B cells stimulated with CD40L and IL-10, and CD138 on stimulated murine B cells, and the subsequent generation of ISC (17, 25, 43). Our data also demonstrated that the frequency of CD38+CD27high B cells appearing in cultures of stimulated memory B cells increased with cell division (Fig. 1c). Furthermore, these cells secreted high Ig levels (Fig. 2) and expressed a transcriptional profile comparable to primary PC (Fig. 4). Thus, this phenotype reliably identifies ISC generated in vitro and allows comparison between human ISC generated in vivo and in vitro (Fig. 7).
FIGURE 7. Model of terminal B cell differentiation. In human lymphoid tissues, naive B cells become activated by Ag and costimulatory factors and form a germinal center where mutation of Ig V region genes and Ig isotype switching occurs. Upon exiting the germinal center, B cells differentiate into either memory B cells or early plasmablasts, identified by increased expression of CD27. Plasmablasts can also rise directly from memory B cells upon re-exposure to specific Ag. Up-regulation of CD38 characterizes further differentiated ("late") plasmablasts. These cells can migrate to the BM, gut, red pulp of spleen, or mucosal epithelium of tonsil under the direction of specific chemokines (CXCL12, CCL25, CCL28; see Ref. 59 ). Once in these microenvironmental niches, ISC compete for survival signals facilitating differentiation into long-lived PC (see Refs. 23 , 32 , and 59 ). Factors identified that support the survival of (i.e., CD40L, CD70, BAFF, IL-6, stromal cells), and the phenotypes and transcription factors expressed by, each stage of terminal B cell differentiation are indicated. Some memory B cells can also yield alternative phenotypic and functional states which may correspond to memory B cell precursors that replenish the memory B cell population (detailed in Discussion).
In addition to increased production of Ig, the CD38+CD27high B cells exhibited several other features of primary PC. First, similar to PC in spleen, tonsil, and PB, CD38+CD27high ISC generated in vitro exhibit reduced expression of CD20, CD21, CD22, CXCR5, and BAFF-R; sustained expression of CD19, CD39, CD40, CD45, CD86, CXCR4, and HLA-DR; and lacked expression of CD138 (Fig. 3, Table I; Refs. 3 , 5 , 32 , and 44). Second, CD38+CD27high ISC retained responsiveness to CXCL12, while responses to CXCL13 and CCL21 were greatly diminished (Fig. 6), akin to primary PC (32, 41, 42). Lastly, expression of Blimp-1 and XBP-1 was greatest, and Pax-5 extinguished, in in vitro CD38+CD27high ISC compared with other populations of activated B cells (Figs. 4 and 7). Similar to alterations in surface phenotype, these molecular changes recapitulate the differential expression of Pax-5, Blimp-1 and XBP-1 observed for mature B cells and primary PC (3, 5, 7, 36, 38), and complement our recent report of gene expression by ISC generated from mouse B cells in vitro in response to T cell-dependent stimuli (43). Similar observations regarding expression of transcription factors have been obtained for human B cells activated in vitro to differentiate toward ISC (38, 39), or in murine splenic B cells following immunization (45). However, a disadvantage of these previous studies was that they examined bulk populations of activated B cells, rather than distinct subsets, and did not correlate mRNA levels with protein (38, 39, 45). Consequently, it was unclear in which cells these transcriptional changes were occurring. Notably, the proliferative activity (17, 20), coupled with Bcl-6 expression and responsiveness to BAFF (Figs. 4 and 5; Ref. 25) also distinguished in vitro CD38+CD27high ISC from primary human PC. The latter population contains noncycling cells which lack expression of Bcl-6, and do not respond to BAFF (3, 24, 25, 32), revealing them to be terminally differentiated. In contrast, in vitro-derived CD38+CD27high ISC likely correspond to plasmablasts, such as those detected in reactive plasmacytosis, or following infection (21) (Fig. 7). The similarities between plasmablasts generated in vitro from activated human memory B cells and those detected in vivo with respect to cell surface phenotype, expression of transcription factors, proliferation and function are remarkable (compare Fig. 3, Table I to Ref. 21), and thus illustrates the validity of our in vitro culture system to examine the processes involved in this aspect of human B cell differentiation.
Another novel finding in our study was the identification of a CD38–CD27high population with a phenotype and capacity to secrete Ig that was identical with the CD38+CD27high B cells (Figs. 2 and 3). Despite these similarities, some important differences were found between these two populations of in vitro-derived ISC. First was their responsiveness to BAFF. This current study refined our previous analysis (25) by revealing that the greatest effect of BAFF was on the viability of CD38+CD27high ISC, while its effects on CD38–CD27high ISC were modest (Figs. 5 and 7). Second, CD38+ B cells proliferate at a greater rate than CD38– B cells (17, 20, 25), revealing CD38+CD27high ISC to be plasmablasts. A molecular explanation for this difference may be attributable to differential expression of Bcl-6. CD38– B cells (Population 2) expressed less Bcl-6 than CD38+ B cells (population 3; Fig. 4). Interestingly, Bcl-6 can regulate the cell cycle by controlling expression of genes such as c-myc (46). This is further exemplified by the finding that the frequency of proliferating cells in Bcl-6 transgenic mice is markedly enhanced compared with control mice (47), and that overexpression of dominant-negative Bcl-6 arrested proliferation of a human B cell line (46). Thus, increased expression of Bcl-6 by CD38+ B cells may provide them with the molecular machinery required for differentiation into rapidly dividing plasmablasts. An additional explanation for the differences between the CD38–CD27high and CD38+CD27high ISC may be that these cells arise from different progenitor memory B cells, namely IgM-expressing or Ig isotype switched memory cells (24, 28). The finding that 1) CD38–CD27high ISC produced a greater proportion of IgM than CD38+CD27high ISC (Fig. 2), 2) BAFF preferentially induces Ig secretion by isotype switched memory B cells (25), and 3) a greater frequency of isotype switched B cells enter their first cell division than IgM-memory B cells (20) would be consistent with the proposal that IgM-memory B cells preferentially yields CD38–CD27high ISC while CD38+CD27high ISC arise from switched memory B cells.
The coexpression of Blimp-1 and Bcl-6 by B cells that have differentiated to become ISC (CD27highCD38+; Fig. 4) was notable because these transcription factors cross-regulate the expression and function of each other (23, 46, 48). Kallies et al. (31) recently described the generation of a mouse line where GFP was introduced into the Blimp-1 locus. This study elegantly demonstrated that all GFP+ cells were ISC, irrespective of surface phenotype. Interestingly, the stage of PC maturation could be inferred from the relative expression of Blimp-1, with cells expressing intermediate levels of Blimp-1 (i.e., Blimp-1/GFPint) representing immature PC while Blimp-1/GFPhigh cells were mature PC (31). Importantly, a population of Blimp-1/GFPint cells also expressed Bcl-6 and were rapidly dividing, leading to the suggestion that they are plasmblasts (31). These data are consistent with our proposal that CD38+CD27high cells, that express Blimp-1 and Bcl-6, are plasmblasts (Fig. 7), and most likely represent the human counterpart of the murine Blimp-1/GFPint cells (31). Another recent study also found that murine B cells could be induced to differentiate into ISC that expressed both Bcl-6 and Blimp-1, but not Pax-5, when stimulated with IL-21 (49). Thus, because both IL-21 and IL-10 are potent growth and differentiation factors for human (IL-10, IL-21; Refs. 50, 51, 52) and murine B cells (IL-21 only; Refs. 49 and 51), it is likely that the molecular alterations induced by these cytokines are similar.
Our examination also identified two other populations of B cells that appear to have distinct functions and fates: the CD38–CD27low and CD38+CD27low cells that were nonsecreting cells. Interestingly, only 50% of PC identified in human or murine lymphoid tissues by a CD38++CD20± or CD138+B220± phenotype, respectively, cells secrete Ig (3, 53). Thus, it is possible that the CD38+CD27low nonsecreters in vitro (Fig. 2b) correspond to these cells in vivo. Alternatively, CD38+CD27low cells may contain precursors of CD38+CD27high cells and could therefore undergo further differentiation to this phenotype and function (Fig. 7). This indeed appears to occur because CD38+CD27low cells can yield CD38+CD27high cells, as well as produce large amounts of Ig, following isolation and subsequent culture (not shown). Similarly, the CD38–CD27low cells may correspond to a pool of memory B cell precursors that replenish the memory B cell population to prevent depletion of these cells following their differentiation to ISC (53, 54).
Our finding of a CD38– population of ISC suggests the existence of an analogous population in vivo. Indeed, it has been reported that 60–80% of human ISC expressed CD38 (55, 56, 57). More recently, a CD38– population of ISC, that appeared to be the precursor of CD38+ ISC, was identified in human tonsils (44). These CD38– ISC expressed CD19, CD27, CD45, HLA-DR, and Blimp-1 (44), analogous to ISC generated in vitro (Fig. 3, Table I) and in vivo (3, 4, 5, 7). Thus, CD38–CD27high ISC generated in vitro from human splenic memory B cells may correspond to the 20–40% of ISC detected ex vivo that were also CD38–. Our findings regarding human ISC are also consistent with the recent descriptions of murine PC that lack CD138 (31, 58). Taken together, these studies reveal the vast heterogeneity in different populations of ISC both in vivo and in vitro and infer that, in the absence of a definitive cell surface marker, ISC cannot by solely defined by phenotype. We propose that human ISC can be resolved by increased expression of CD27, while the coexpression of CD38 delineates the ISC population into an early stage of commitment to the ISC lineage (CD38–) and plasmablasts (CD38+) (Fig. 7). These findings will further our understanding of the complex regulation of the pathways B cells undergo during differentiation to an effector cell.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Drs. Marylin Kehry, Rene de Waal Malefyt, and Susan Kalled for generously providing reagents, the Australian Red Cross Blood Service for providing human spleens, Tara Macdonald and Joseph Webster for cell sorting, and Prof. Tony Basten for critical review of this manuscript.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by the National Health and Medical Research Council (NHMRC) of Australia. S.G.T. is the recipient of an RD Wright Biomedical Career Development Award from the NHMRC; P.D.H. is a Principal Research Fellow of the NHMRC; L.M.C. is a Senior Research Fellow of the NHMRC.
2 Address correspondence and reprint requests to Dr. Stuart Tangye, Centenary Institute of Cancer Medicine and Cell Biology, Locked Bag No. 6, Newtown 2042, New South Wales, Australia. E-mail address: s.tangye@centenary.usyd.edu.au
3 Abbreviations used in this paper: PC, plasma cell; BM, bone marrow; ISC, Ig-secreting cell; SA, streptavidin; BAFF, B-cell activating factor belonging to the TNF family; BAFF-R, BAFF receptor; BCMA, B cell maturation Ag; TACI, transmembrane activator of and CAML interactor; XBP-1, X box-binding protein-1.
Received for publication October 25, 2004. Accepted for publication January 19, 2005.
References
Terstappen, L. W., S. Johnsen, I. M. Segers-Nolten, M. R. Loken. 1990. Identification and characterization of plasma cells in normal human bone marrow by high-resolution flow cytometry. Blood 76:1739.
Leo, R., M. Boeker, D. Peest, R. Hein, R. Bartl, J. E. Gessner, J. Selbach, G. Wacker, H. Deicher. 1992. Multiparameter analyses of normal and malignant human plasma cells: CD38++, CD56+, CD54+, cIg+ is the common phenotype of myeloma cells. Ann. Hematol. 64:132.
Ellyard, J. E., D. T. Avery, T. G. Phan, P. D. Hodgkin, S. G. Tangye. 2004. Ag-selected Ig-secreting cells persist in human spleen as well as bone marrow. Blood 103:3805.
Merville, P., J. Dechanet, A. Desmouliere, I. Durand, O. de Bouteiller, P. Garrone, J. Banchereau, Y. J. Liu. 1996. Bcl-2+ tonsillar plasma cells are rescued from apoptosis by bone marrow fibroblasts. J. Exp. Med. 183:227
Medina, F., C. Segundo, A. Campos-Caro, I. Gonzalez-Garcia, J. A. Brieva. 2002. The heterogeneity shown by human plasma cells from tonsil, blood, and bone marrow reveals graded stages of increasing maturity, but local profiles of adhesion molecule expression. Blood 99:2154.
Farstad, I. N., H. Carlsen, H. C. Morton, P. Brandtzaeg. 2000. Immunoglobulin A cell distribution in the human small intestine: phenotypic and functional characteristics. Immunology 101:354.
Medina, F., C. Segundo, A. Campos-Caro, I. Salcedo, A. Garcia-Poley, J. A. Brieva. 2003. Isolation, maturational level, and functional capacity of human colon lamina propria plasma cells. Gut 52:383.
Kyle, R. A.. 1991. Plasma cell proliferative disorders. R. Hoffman, and E. J. Benz, and S. J. Shattil, and B. Furie, and H. J. Cohen, eds. Hematology. Basic Principles and Practice 1021. Churchill Livingstone, New York.
Odendahl, M., A. Jacobi, A. Hansen, E. Feist, F. Hiepe, G. R. Burmester, P. E. Lipsky, A. Radbruch, T. Dorner. 2000. Disturbed peripheral B lymphocyte homeostasis in systemic lupus erythematosus. J. Immunol. 165:5970
Jacobi, A. M., M. Odendahl, K. Reiter, A. Bruns, G. R. Burmester, A. Radbruch, G. Valet, P. E. Lipsky, T. Dorner. 2003. Correlation between circulating CD27high plasma cells and disease activity in patients with systemic lupus erythematosus. Arthritis Rheum. 48:1332.
Vernino, L., L. M. McAnally, J. Ramberg, P. E. Lipsky. 1992. Generation of nondividing high rate Ig-secreting plasma cells in cultures of human B cells stimulated with anti-CD3-activated T cells. J. Immunol. 148:404.
Merville, P., J. Dechanet, G. Grouard, I. Durand, J. Banchereau. 1995. T cell-induced B cell blasts differentiate into plasma cells when cultured on bone marrow stroma with IL-3 and IL-10. Int. Immunol. 7:635.
Choe, J., Y. S. Choi. 1998. IL-10 interrupts memory B cell expansion in the germinal center by inducing differentiation into plasma cells. Eur. J. Immunol. 28:508.
Jung, J., J. Choe, L. Li, Y. S. Choi. 2000. Regulation of CD27 expression in the course of germinal center B cell differentiation: the pivotal role of IL-10. Eur. J. Immunol. 30:2437.
Arpin, C., J. Dechanet, C. Van Kooten, P. Merville, G. Grouard, F. Briere, J. Banchereau, Y. J. Liu. 1995. Generation of memory B cells and plasma cells in vitro. Science 268:720.
Arpin, C., J. Banchereau, Y. J. Liu. 1997. Memory B cells are biased towards terminal differentiation: a strategy that may prevent repertoire freezing. J. Exp. Med. 186:931.
Tangye, S. G., D. T. Avery, P. D. Hodgkin. 2003. A division-linked mechanism for the rapid generation of Ig-secreting cells from human memory B cells. J. Immunol. 170:261.
Jelinek, D. F., P. E. Lipsky. 1983. The role of B cell proliferation in the generation of immunoglobulin-secreting cells in man. J. Immunol. 130:2597.
Jego, G., R. Bataille, C. Pellat-Deceunynck. 2001. Interleukin-6 is a growth factor for nonmalignant human plasmablasts. Blood 97:1817
Tangye, S. G., D. T. Avery, E. K. Deenick, P. D. Hodgkin. 2003. Intrinsic differences in the proliferation of naive and memory human B cells as a mechanism for enhanced secondary immune responses. J. Immunol. 170:686.(Danielle T. Avery, Julia )