Mature B cell differentiation involves a well-established transcription factor cascade. However, the temporal dynamics of cell signaling pathways regulating transcription factor network and coordinating cell proliferation and differentiation remain poorly defined. To gain insight into the molecular processes and extrinsic cues required for B cell differentiation, we set up a controlled primary culture system to differentiate human naive B cells into plasma cells (PCs). We identified T cell-produced IL-2 to be critically involved in ERK1/2-triggered PC differentiation. IL-2 drove activated B cell differentiation toward PC independently of its proliferation and survival functions. Indeed, IL-2 potentiated ERK activation and subsequent BACH2 and IRF8 downregulation, sustaining BLIMP1 expression, the master regulator for PC differentiation. Inhibition of the MAPK–ERK pathway, unlike STAT5 signaling, impaired IL-2–induced PC differentiation and rescued the expression profile of BACH2 and IRF8. These results identify IL-2 as a crucial early input in mature B cell fate commitment.
T cell-dependent immune response is initiated in germinal centers (GCs) after seeding by a small number of rapidly dividing Ag-responding B cells. These cells undergo a series of proliferation/selection steps to give rise to memory B cells or long-lived plasma cells (PCs) (1). The GC response initiates in the outer follicle where naive B cells (NBCs) encounter their specific Ags (2). Subsequently, activated B cells relocate to the B zone–T zone boundary where Ag-specific B and T cells interact with each other and form long-lived pairs. In this pairing, B and T cells upregulate BCL6, proliferate (3, 4), and acquire a centroblastic or a follicular Th (TFH) cell identity respectively before migrating inside the follicle (3–7). In GCs, B cells undergo terminal differentiation and selection, which depend tightly on the light zone microenvironment (3, 4). The gene regulatory network that governs transition between GC B cells and plasmablasts is well understood and heavily controlled by cytokines, particularly those that reduce BCL6 expression and induce BLIMP1, two mutually exclusive transcriptional regulators. BLIMP1 orchestrates PC differentiation by extinguishing the mature B cell gene expression program including BCL6, freeing factors like IRF4, and XBP1 (8, 9). Notably, in response to stimulation, B cells stochastically pursue various distinct fates. Some cells undergo differentiation and isotype switching depending on cell division and cytokine environment (10–12).
The T cell help provides distinct signals that are critical in determining B cell behavior. During initial T–B cognate interaction, T cell potentially induces a large variety of inputs including CD40 and CD80/CD86 engagement (13, 14). However, T cell-derived signals seem not to be mandatory, as T cell-independent B cell clonal expansion, GC formation, and acquisition of GC B cell phenotype may be observed in the absence of T cells (15). However, stable contacts of TFH cells with B cells are absolutely required to sustain GC maturation, centrocyte selection, and PC generation (5, 16). TFH cells are not statically providing a single stimulus to B cells, but instead at some point they promote high-affinity GC B cells to become either PCs or memory B cells. This could be an early or late input that might function with a concomitant CD40 signal. First, T cells drive NBC proliferation and inhibit PC differentiation (17). At a later stage, TFH cells promote a reinforcement of BCL6 repression signals in high-affinity B cells (18, 19). Besides CD40L, TFH cells produce IL-21, which acts directly on GC B cells maximizing BCL6 expression leading to cell survival and proliferation (20–23). In contrast, IL-21 induces PC differentiation through a STAT3-dependent BLIMP1 induction (21, 24, 25). Numerous other T cell-derived cytokines enhance PC differentiation such as IL-2 produced by memory CD4+ T cells (26). Recently in mice, ERKs have been shown to be necessary to trigger PC generation, mediating the cytokine-induced production of BLIMP1 (27). Altogether, these findings reinforce the well-known assumption that a combination of a BCR signal and T cell help is required to initiate PC differentiation (28).
To explore events that govern human NBC differentiation, we designed an in vitro two-step culture model combining BCR signal, TLR activation, and T cell help in the form of CD40L and cytokines. Unswitched human naive precursors differentiated into CD20+CD38+ and CD20loCD38hi cells characterized by distinct cell fates. We explored one by one factors used in our model and found that T cell-produced IL-2 was critical for human NBC commitment to PCs. IL-2 activated the ERK pathway at a threshold level that triggered, beyond the induction of cell cycle progression, PC generation. BACH2 and IRF8 expression was downregulated through the MEK–ERK signaling pathway in IL-2–primed B cells. Therefore, IL-2 reinforced the mutual repression between BCL6 and BLIMP1. This study highlights early events that could take place when cognate B and T cells meet and interact before seeding follicle, where crucial inputs may condition final GC B cell destiny.
Materials and Methods
Primary B cell purification
PBMCs from healthy volunteers were obtained from the Etablissement Français du Sang (Rennes, France) after Ficoll density centrifugation (Sigma-Aldrich, St. Louis, MO). NBCs were purified by negative selection using magnetic cell separation (Naive B Cell Isolation Kit II; Miltenyi Biotech, Bergisch Gladbach, Germany), following the manufacturer’s instructions, using the AutoMACS deplete-sensitive program. Purity of isolated CD19+CD27− NBCs was routinely >99%, and in some cases IgG+CD27− memory B cells were removed by FACS sorting. Tonsil-derived paired memory B cells and centroblasts were isolated as previously described (29).
Cell culture, immunophenotyping, and cell sorting
Purified human NBCs were cultured at 7.5 × 105 cells/ml in 24-well plates and stimulated during 4 d with 2 μg/ml F(ab′)2
Day 4-activated B cells were washed and cultured at 4 × 105
B and T cell coculture.
At day 0, CFSE-labeled NBCs were stimulated by CpG plus F(ab′)2 anti-human Ig (BCR). At day 1, autologous CD4+ T cells were purified by magnetic cell separation using CD4+ microbeads (Miltenyi Biotech) and added to B cells (ratio 1T:2B) together with 0.5 μg/ml anti-CD3 plus 0.5 μg/ml anti-CD28 stimulation (Sanquin, Amsterdam, The Netherlands) in the presence of a blocking anti–IL-2 or a control isotype. IL-2 plus CD40L stimulated B cells were used as control. At day 4, T cells were removed from the coculture by cell sorting based on the CD2 expression. In parallel, CD4+ T cells were activated or not with anti-CD3 and anti-CD28 stimulation for 1 d, and their supernatants were collected and added to B cells at day 2, in the presence of CD40L and blocking Abs to IL-2 or IL-2R(α+β+γ). At day 4, B cells were subsequently cultured in the presence of IL-2, IL-10, and IL-4 for two additional days.
Abs used in this study are summarized in Table I. For quantitative RT-PCR analyses and ELISA assays, B cells were isolated at day 6 by FACS sorting after DAPI (Sigma-Aldrich) staining to exclude dead cells. Sorts were performed on a BD FACSAria cell sorter (BD Biosciences, San Jose, CA). Differentiated and nondifferentiated B cells were sorted as CD20loCD38hi and CD20+CD38+ cells, respectively. After surface staining, the Cytofix/Cytoperm kit (BD Biosciences) was used for intracellular staining of IgM and IgG.
Apoptosis and proliferation assays
Apoptosis and proliferation were analyzed using a PE-conjugated anti-active caspase-3 apoptosis kit (BD Biosciences) and an FITC-conjugated anti-BrdU kit (BD Biosciences), respectively, according to the manufacturer’s instructions. For BrdU staining, B cells were incubated with BrdU (10 μM) during 45 min followed by CD38 staining. Cells were then fixed, permeabilized, and DNAse treated before subsequent staining and FACS analysis.
Ig secretion and Western blotting
For Ig secretion assay, differentiated and nondifferentiated B cells were isolated and reseeded with IL-2, IL-4, and IL-10 at 1 × 106 cells/ml for 18 h. IgG, IgA, and IgM secretion was assessed by ELISA using a goat anti-human Ig for coating and secondary alkaline phosphatase-coupled Abs specific for γ, α, or μ chain, respectively (all from Jackson ImmunoResearch Laboratories). SDS-PAGE and Western blotting were performed according to standard procedures and Abs listed in Table I. Detections were performed with HRP-conjugated secondary Ab (Bio-Rad) and enhanced chemiluminescent (ECL Plus) reagent (Amersham). GAPDH or β-actin on the same membrane served as loading control.
Quantitative RT-PCR analysis
HPRT1 as endogenous control. The 2 exp(−ΔΔCt) method was used to determine the relative expression of each gene.
CFSE-labeled NBCs from three donors were first stimulated at day 0 [CD40L, CpG, and F(ab′)2 anti-human Ig]. Then, IL-2 was added or not at day 0 plus 16 h leading for each donor to two parallel cell culture conditions: IL-2+ and IL-2− cells. Cells were removed at different time points for microarray analysis (day 0, day 0 + 16 h, day 0 + 22 h). At day 4, IL-2+ and IL-2− cells were separated into CSFEhi and CSFElo fractions (Fig. 4A). Extracted RNAs (RNeasy microkit; Qiagen) were hybridized onto Illumina Human HT-12 v4 Whole-Genome Gene Expression BeadChips according to standard Illumina protocols (Illumina, San Diego, CA). The 18 samples were randomly distributed on two beadchips. Data extraction and quality control were performed using BeadStudio.
Statistical analyses and functional interpretation
Data are available at the Gene Expression Omnibus database (accession no. GSE36975; http://www.ncbi.nlm.nih.gov/geo/). Microarray statistical analyses were performed using Partek Genomics Suite software, version 6.5 (Partek, St. Louis, MO) and the R language (www.R-project.org). First, prefiltering of the raw data was performed by filtering out all probes for which detection above background p value was >5% for all samples as well as all uncharacterized probes (associated to gene symbols starting by Cxorf, FLJ, KIAA, HS., LOC and MGC). Hence, by reducing the number of probes to 21,788, the power of statistical tests was improved. Data were log2 transformed without normalization. Any batch effect highlighted with PCA was corrected using the Partek batch-removal ANOVA-based method. Statistical analysis focused on the effect of stimulation 1 at the different time points. Hierarchical clustering analysis (HCA) was used to study overall dynamics of the genes throughout the time course. Probesets differentially expressed between different time points were identified using paired t test with p value <5%, restricting the analysis to probes with absolute log2 fold change >1.2 and mean of the linear-scale gene expression difference >150. HCA was used to highlight dynamics of these differentially expressed genes throughout the time course. Functional interpretation of these differentially expressed genes was performed using IPA (Ingenuity Systems; www.ingenuity.com), more specifically the canonical pathway analysis tool that helped visualization of gene expressions on pathways of interest.
Statistical analyses for different experiments were performed using GraphPad Prism software (GraphPad Software), and p values were calculated by two-tailed Student t test.
Experimental design to differentiate human unswitched NBC precursors into PCs
To explore T-dependent B cell terminal differentiation, we developed a two-step B cell culture model (Fig. 1A). All Abs used for this study are listed in Table I. CD19+CD27− NBCs were first cultured with F(ab′)2 anti-Ig(A+G+M), CD40L, CpG, and IL-2. At day 4, activated B cells were washed and then reseeded with IL-2, IL-4, and IL-10 in the absence of CD40L.
CD38 and CD20 expression was monitored by flow cytometry (Fig. 1B). During the activation phase, cells remained phenotypically homogeneous as a CD20+CD38+ population. In the second step, a CD20loCD38hi subpopulation appeared as soon as day 5 and increased progressively reaching 22 ± 13% (n = 19) of the total viable B cells at day 6. Unlike CD20+CD38+ cells, CD20loCD38hi cells concomitantly increased the expression of the CD138 PC marker (Fig. 1B). Although recently described IgG+CD27− memory B cells (30) were found at an extremely low percentage in our purified NBC starting populations (1.13 ± 1.04%, n = 3), we decided to apply our differentiation model on CD27− NBCs cleared for IgG+ cells. Both CD20loCD38hi and CD20+CD38+ subsets were generated without any significant difference in PC generation compared with the standard procedure ruling out the possibility that the CD20loCD38hi cells resulted from the differentiation of CD27−IgG+ memory B cells (data not shown). Therefore, in all further experiments, we used CD27− NBCs.
A 3-fold increase of viable B cell number was observed during the activation phase, whereas the differentiation phase correlated with a decrease of cell viability, as evaluated by trypan blue exclusion (data not shown). A dual BrdU and active caspase-3 staining was thus performed to assess apoptosis and proliferation in the course of PC generation (Fig. 1C). During the activation phase, a proliferation burst was distinguished with more than 40% of the cells in S-phase at day 3 while only 4% of cells showed positive staining for caspase-3 (Fig. 1Ci–ii). At day 6 within the differentiation phase, cell proliferation was almost restricted to the CD20loCD38hi subset with 34.4 ± 7.9% of BrdU+ cells compared with 4.4 ± 2.8% for CD20+CD38+ cells that instead increased their susceptibility for apoptosis (Fig. 1Ci–ii). Beyond day 6, cell viability decreased dramatically as cells arrest to proliferate (Fig. 1D). To ascertain whether proliferation and differentiation were connected together, we labeled B cells with CFSE at day 0. The CD20loCD38hi subset was exclusively constituted by cells that had diluted the CSFE (Fig. 1Ciii). By sorting both CSFElo and CSFEhi subsets at day 4 before further culture, we demonstrated that only cells that underwent three or more divisions gave rise to CD38hi cells (data not shown).
To test generated B cell subsets for Ab secretion, intracytoplasmic Ig production was assessed by flow cytometry. As shown in Fig. 1E (left and middle panel), CD20+CD38+ cells were mainly IgM positive (80.18 ± 12.28%, n = 6), whereas CD20loCD38hi cells expressed preferentially IgG (52.36 ± 7.71%, n = 6) reflecting a class-switching process among those cells. The upregulation in mean fluorescence intensity of intracytoplasmic IgM staining in CD20loCD38hi compared with CD20+CD38+ cells suggested that IgM expression was mainly cytoplasmic in CD20loCD38hi cells. Moreover, we detected only low levels of surface IgM or IgG on CD20loCD38hi cells (data not shown). These results were in agreement with the increase of IgM, IgA, and IgG secretion by CD20loCD38hi cells detected by ELISA (Fig. 1E, right panel) and ELISPOT assay (data not shown).
Recapitulation at the transcriptional level of B cell differentiation hallmarks initially described in mouse models
Cells were monitored from day 0 to day 6 by mRNA expression profiling for various molecules that are crucial in the transcriptional cascade that converts B cells into PCs (9). Gene expression levels of BACH2, PAX5, BCL6, PRDM1, IRF4, BIP, SPIB, SPI1, ETS1, MITF, POU2AF1, AICDA, and the spliced variant of XBP1 (XBP1s) at day 4 and day 6 were studied relative to their gene expression at day 0, arbitrary defined as 1. At day 6, CD20+CD38+ and CD20loCD38hi subsets were sorted and tested separately. The two populations presented a downregulation of MITF, a suppressor of PC differentiation (31). Unlike CD20loCD38hi cells, CD20+CD38+ cells did not modify the expression of PAX5, IRF4, and BACH2 and maintained BCL6 and AICDA expression compared with day 4 cells consistent with a blockage in the differentiation process of these cells (Fig. 2A, 2B). Their comparison with freshly isolated human centroblasts and memory B cells for BCL6 and AICDA expression supported the notion of uncommitted activated B cell state (Fig. 2C). In contrast, CD20loCD38hi cells upregulated PRDM1, IRF4, POU2AF1, BIP, and XBP1s and downregulated drastically AICDA. These cells completed their differentiation program as seen by the downregulation of the Ets family genes SPI1, ETS1, and SPIB that are repressors of the B cell differentiation program, along with a strike increase of CD27 expression (Fig. 2A, 2B) confirmed at the protein level (data not shown). Altogether, we conclude that our human B cell differentiation model mimics the molecular dynamic that takes place during in vivo GC B cell maturation. Thus, it is a potentially valuable tool for the analysis of molecular and temporal aspects of human NBCs differentiation into PCs.
NBCs primed for PC commitment require IL-2 at early time point
As previously described (10, 32), we confirmed that only highly proliferative B cells are primed for PC differentiation. We first compared the ability of several combinations of factors to promote proliferation during the first phase of our model. Both CpG and BCR cross-linking were required to induce the highest cell proliferation (Fig. 3A). Omitting either CD40L or IL-2 did not significantly affect cell division. However, at day 6 we found that CpG plus BCR alone or completed with CD40L did not allow plasmablast generation under the second stimulation condition, which included IL-2, IL-4, and IL-10. In fact, IL-2 was required in the activation phase to trigger PC differentiation (Fig. 3B).
In agreement with the demonstration that BCR triggering and CD40 signaling could synergize with TLR9 to induce B cell activation (33–35), CD40L plus CpG stimulation also triggered B cell proliferation, but to a lower extent than CpG plus BCR (Fig. 3A). Consequently, the number of cell divisions obtained with CD40L plus CpG was probably not sufficient to promote significantly plasmablast generation, even in the presence of IL-2 (Fig. 3B) (10).
Kinetic analysis revealed that IL-2 needed to be added in the first 48 h of the culture to trigger maximum PC generation (Fig. 3Ci). Dose-response experiment showed that 0.5 U/ml IL-2 was sufficient to trigger differentiation (Supplemental Fig. 1A). Whereas IL-2 receptor expression was undetectable on NBCs, the three chains of the IL-2R, IL-2Rα (CD25), β (CD122), and γ (CD132), were upregulated upon stimulation with CpG, CD40L, and BCR triggering at the mRNA and protein levels (Supplemental Fig. 1B, 1C). CD25 expression was further upregulated in IL-2–treated cells at day 4, whereas CD122 and CD132 expression was only weakly modulated by IL-2. Thus, IL-2 drives a positive feedback loop in B cells as IL-2 signaling increased the expression of high-affinity IL-2R, an effect previously described in T cells (36, 37). We next tested the role of IL-2 in the second step of the culture. In absence of IL-2 between day 4 and day 6, the absolute number of generated CD38hi cells only decreased by 20% on average (Supplemental Fig. 1D). Thus, IL-2 in the second step of differentiation is not as crucial as IL-2 in the first step of culture to trigger PC differentiation. IL-4 removal from the second culture condition did not modify the number of generated plasmablasts at day 6 (Supplemental Fig. 1D).
IL-2–mediated B cell differentiation was inhibited when we added in parallel anti–IL-2 or when IL-2R was blocked with anti–IL-2Rα alone and even more efficiently with the combination of blocking Abs against the three IL-2R chains (Fig. 3Cii). Because T cells are the major source of IL-2, we tested whether CD4+ cells sustained PC production in our model. Indeed, CD3- and CD28-activated autologous CD4+ T cells promoted plasmablast generation. This effect strikingly decreased in the presence of anti–IL-2 neutralizing Ab (Fig. 3Ciii). Furthermore, supernatant of activated CD4+ T cells was sufficient to trigger differentiation and was again blocked with anti–IL-2 or anti–IL-2R Abs (Fig. 3Civ). Finally, IL-2 was substitutable by IL-15 but not by IL-21 (data not shown).
IL-2 sustains the initial transcriptional burst and initiates PC commitment as early as day 4
To decipher signaling that sustains IL-2 effect, a microarray gene expression profiling (GEP) was applied on highly purified cell fractions obtained at different time points of the cell culture. To disconnect IL-2 signal effects from other stimuli and because IL-2R chains expression was induced only after initial activation of NBCs (Supplemental Fig. 1B), IL-2 was added 16 h after starting cell culture. Therefore, at point day 0 plus 16 h, cells were cultivated either in presence or absence of IL-2 (Fig. 4A). We generated lists of genes differentially expressed between cell subsets (gene list; GL). GL1 contained 2456 differentially expressed probes and characterized the initial transcriptional burst signature that takes place between day 0 and day 0 plus 16 h (Fig. 4A). The outcome of this initial burst at day 0 plus 22 h and day 4 was explored by unsupervised analysis that showed no clear and distinguishable effect of IL-2 (Fig. 4B). Nevertheless, a global enhancement of the mean intensity signal of the 2456 probesets was observed at day 0 plus 22 h for IL-2+ cells (Fig. 4C, left panel). Later on, at day 4 the comparison of the mean intensity values between the two conditions showed a restricted number of 13 distinct genes with a fold change (FC) ≥ 2, all of them belonging to the IL-2+ condition (Fig. 4C, right panel). Among these genes, seven (CCND2, PHGDH, MAPKAPK3, ATF5, DUSP5, CCL3, SLC7A5) were described elsewhere as preferentially expressed in plasmablasts or mature PCs compared with memory or activated B cells (38), whereas some of these seven plus two more genes (LTA, IL-2RA) were known as regulated by IL-2 (36, 39).
IL-2 enhances the expression of genes involved in the MAPK–ERK signaling pathway
GEP analysis at day 0 plus 22 h between IL-2+ and IL-2− cells (GL2) revealed 113 probes corresponding to 112 genes, all of them overexpressed in the IL-2+ condition (Supplemental Table I, left panel). By using Ingenuity software to identify molecular pathways specific to GL2, only MAPK–ERK and integrin signaling were significantly represented (p value <5%) (Fig. 5A). Both pathways are tightly connected, and 24 GL2 genes belonged directly or indirectly to the MAPK–ERK pathway (Fig. 5B). Notably, this pathway was further enriched by several molecules related to protein and metabolism, cellular development and movement including ERK itself after we built a specific IL-2 signature constituted of 334 probesets obtained after addition of GL2 to GL4 and subtraction of GL6 probesets (Supplemental Fig. 2).
We then investigated GEP at day 4 with GL3 and found 86 genes: 76 overexpressed by IL-2+ cells and 10 by IL-2− cells (Supplemental Table I, right panel). Functionally, the 76 former genes were connected mainly to cell growth and proliferation and cell-to-cell interaction (Ingenuity, data not shown). Several of these genes were described in previous GEP as related to preplasmablast, plasmablast, or even fully mature bone marrow PC compared with memory or activated B cells (http://amazonia.transcriptome.eu). For instance, the two genes with the highest FC ratio, 5.99 and 4.43 for GZMH and CCND2, respectively, were found upregulated in plasmablast. These findings are in accordance with the fact that B cells committed to PC differentiation are intensively cycling, what we demonstrated for the CD20loCD38hi cells generated beyond day 4 (see earlier). Furthermore, at least 8 of those 76 genes were described as upregulated by IL-2 (Supplemental Table I, right panel), Altogether, these results associated with the fact that IL-2 is required at the beginning of the cell culture to produce PCs suggest that IL-2 primes cells to PC differentiation by enhancing MAPK–ERK signaling, which induces as soon as day 4 the expression of genes involved in plasmablast generation.
Pharmacological inhibition of MEK–ERK signaling blocks PC differentiation
Considering that stimulation by IL-2 at day 2 is sufficient to trigger maximum PC differentiation and to disconnect IL-2 signal effects from other stimuli, IL-2 was added in the culture medium at day 2 for the rest of the experiments. We first compared signaling events at the protein level triggered by IL-2, IL-15, and IL-21 keeping in mind our earlier results where only the two first cytokines triggered PC differentiation. All three factors induced STAT5 phosphorylation, but only IL-2 and to a lesser extent IL-15 but not IL-21 activated ERK1/2 (Fig. 6A). Similar effects of IL-2 and IL-15 but not IL-21 on PC generation suggested a major role of the ERK1/2 pathway in cytokine-mediated PC differentiation. Therefore, we decided to use pharmacological inhibitors of ERK and STAT5 pathways at concentration and timing (day 2) where CFSE proliferation profiles were unaffected (data not shown). At day 4 and day 6, no difference in terms of cell viability was observed between all conditions. MEK inhibitors U0126 and PD184161 blocked the IL-2 property in a dose-dependent manner, whereas vehicle (DMSO) showed no effect on IL-2–mediated PC differentiation. In contrast, STAT5 inhibitor did not affect IL-2 effect (Fig. 6B). By Western blot we found that B cells primed by IL-2 for 24 h had an increased IL-2 phospho-ERK response compared with untreated cells. The MEK inhibitor U0126 at 0.5 μM blocked this phospho-ERK IL-2–induced response (Fig. 6C). In our model, inhibition of IL-2–induced PC generation is only partial in the presence of MEK pharmacological inhibitors, whereas the complete absence of IL-2 abolished B cell differentiation. However, differentiation inhibition was superior when higher dose of drugs were used, but proliferation was also affected, reflecting probably drug toxicity. We could thus not exclude an additional ERK-independent role for IL-2.
IL-2–induced ERK signaling leads primarily to a downregulation of BACH2, a transcriptional repressor of PRDM1
Generation of PCs is associated with the downregulation of B cell transcription factors, PAX5, BCL6, IRF8, BACH2, and SPIB, and the upregulation of PC transcription factors, IRF4, BLIMP1, and XBP1. To understand mechanisms by which IL-2–induced ERK1/2 activation triggers PC differentiation, we sought for genes whose expression was modified in the presence of IL-2 and counteracted by PD184161 MEK–ERK inhibitor. To this end, we separated the CSFElo and CFSEhi cell fractions at day 4 from three different culture conditions: without cytokine, IL-2 added at day 2, and IL-2 in the presence of PD184161 at a concentration capable of inhibiting differentiation by 50% (Supplemental Fig. 3). Relative gene expression analyses by qRT-PCR showed a significant decrease of BACH2, IRF8, PAX5, and SPIB in IL-2–primed CFSElo cells compared with the condition without IL-2 (Fig. 7A). This IL-2 effect was efficiently counteracted by PD184161. In contrast, in CFSEhi fractions—those cells are unable to differentiate further—IL-2 did not modify gene expression (Fig. 7B). We finally took the CSFEhi gene expression level as reference to compare gene expression levels between cell treatments in CFSElo fractions. This analysis revealed that IL-2 effect goes through a significant downregulation of BACH2 and IRF8. Unexpectedly, PRDM1, BCL6, and XBP1s expression was significantly modulated in the CSFElo fractions whatever the considered culture conditions (Fig. 7C).
We then determined BACH2 protein levels at day 4 and day 5 in highly purified CFSEhi and CFSElo cell fractions. We first confirmed an increase of IL-2–mediated ERK activation specifically in the CFSElo fraction of cells pretreated with IL-2 at day 4 (Fig. 8A). At this time point, BACH2 protein expression was not different between no-cytokine and IL-2–primed cells in CFSElo fractions. Furthermore, BACH2 and PAX5 were less abundant in CFSElo compared with CFSEhi fractions suggesting posttranscriptional BACH2 and PAX5 regulatory mechanisms linked to cell division. PAX5 is a transcriptional repressor of PRDM1 and XBP1. This could explain the upregulation at the mRNA level of PRDM1 and XBP1 specifically in CFSElo fractions whatever the culture conditions (Fig. 7C).
BLIMP1 expression was not detected at day 4 (data not shown). However, at day 5 BLIMP1 protein was specifically detected in the CFSElo fractions and was more abundant in IL-2–primed cells than in cells activated without IL-2 (Fig. 8B). In contrast, BACH2 was no more detectable in CFSElo IL-2–primed cells consistent with the results obtained at day 4 regarding BACH2 mRNA expression levels. PAX5 was further repressed in this CFSElo IL-2–primed fraction suggesting a possible BLIMP1-dependent repression of PAX5.
Since the 1990s, major contributions have been made concerning GC biology and T cell-dependent B cell maturation. Several mouse models, eventually associated with intravital imaging, allowed identification of key factors and specific cell dynamics that govern normal GC reaction and GC B cell fate. At the cellular level, parsing the signals that direct cellular differentiation versus cell division remains largely uncovered. Identical remarks can be made concerning mechanisms guiding GC B cells to memory B cell differentiation or PC commitment. Although rodent models represent crucial tools to approach fundamental biological questions, they present some limitations in cell signaling, and translation of the findings to human is sometimes difficult. To address these questions, we decided to develop an in vitro model of B cell differentiation using peripheral human NBCs collected from healthy blood donors. This differentiation process includes two successive steps of culture: expansion of activated B cells in a first phase followed by PC maturation. This second step hosts the emergence of cycling CD38hi plasmablasts that give rise to PCs identified by CD138 expression on the cell surface; meanwhile cells undergo Ig class-switch recombination. We confirmed the successful differentiation of NBCs toward PCs by microarray GEP and immunophenotyping along the differentiation process. These analyses showed that this model recapitulated functionally cellular transitions and pinpoint crucial molecular events characteristic of GC B cell differentiation. For instance, the BLIMP1 and BCL6 double-negative feedback loop is clearly involved in the emergence of CD38hi cells and associated with the extinction of the transcriptional program that establishes the B cell phenotype (8). The transcriptional cascade that sustains PC differentiation is highly dependent on soluble factors issued largely from T cells present in the GC microenvironment. CD40 signaling for instance may steer toward memory B cell pathway rather than PC pathway, although its effect might be dependent on the timing and the level of B cell activation (19, 40). We demonstrated that CD38hi cells appeared after CD40L withdrawal and exposure to IL-2, IL-4, and IL-10. Meanwhile, these cells continued to cycle whereas the undifferentiated CD20+CD38+ cells underwent apoptosis. This finding suggests that B cells steering toward the PC pathway result from precursors present in the human IgG-negative NBCs. The number of those precursors is highly variable from blood donor to blood donor; we noticed an unpredictable variability of PC generation throughout all the experiments although culture conditions are fixed and proliferation properties unaffected (data not shown). Finally, our data showed also that cells initiated for PC differentiation respond dramatically to mitogenic signals. Indeed, plasmablasts emerged only from cells with the lowest CSFE staining at day 4, which is consistent with previous data obtained with memory B cells (32). Interestingly, our study is, to our knowledge, the first to show that XBP1, BCL6, and PRDM1 expression levels are driven by the B cell proliferation, but that proliferation is not sufficient per se to trigger PC differentiation.
The intermingled relationships between differentiation and proliferation during GC reaction underline the presence of intense intracellular signaling systems that remain poorly defined. In this study, investigations of the initial mitogenic stimulus revealed that the combination of CpG and BCR activation was sufficient and indispensable to get optimal cell proliferation but not to steer cells to the PC pathway. We demonstrated the need of an early IL-2 signal to initiate the differentiation. Along with IL-4, IL-7, IL-9, IL-15, and IL-21, IL-2 shares the common cytokine receptor γ-chain. Using activated autologous CD4+ T cells or supernatant of those cells, we showed that both conditions were able to replace CD40L and IL-2 in the first mitogenic phase of our model to get PC differentiation. This differentiation was suppressed when we blocked IL-2 or its receptor. We found also that IL-2 was not substitutable by IL-21, in contrast to IL-15, suggesting that the IL-21 positive effect on PC differentiation may intervene later in this process and especially in the establishment of long-lived PCs (21, 41).
Our investigations concerning the impact of IL-2 on the initial transcriptional burst (Fig. 4) allowed us to underline a broad enhancement of gene expression in the presence of IL-2, in agreement with the fact that IL-2 is a pleiotropic cytokine with a broad array of actions. However, despite this large and unspecified transcriptional effect, we sought whether a signature of genes connected to PC differentiation may be detectable during the mitogenic phase, which could be consistent with descriptions made for T cells where IL-2 helps for specific differentiated states and their maintenance (42, 43). After 4 d of culture, we highlighted 13 genes issued from the initial transcriptional burst that were overexpressed in cells cultured with IL-2, among which 7 genes were already described in the plasmablastic differentiation. We confirmed those data by the analysis at day 4 of genes differentially expressed between cells cultivated with or without IL-2 (GL3; Fig. 4A). Taken together, our data demonstrated, besides the broad impact of IL-2 at the transcriptional level, the emergence of genes specifically related to the initiation of PC differentiation within highly cycling B precursors. The translation of this finding within in vivo GC reaction could correspond to the step when B cells encounter cognate T cells and start to proliferate while migrating to the follicle interior (3, 7). Concerning the small magnitude of the gene expression differences, this might reflect heterogeneity within the cell population and could fit with previous studies indicating that a minority of GC B cells exhibits PC properties (5, 44).
IL-2 is known to activate several signaling pathways including RAS–MAPK, JAK–STAT, and PI3K/Akt/p70 S6 kinase pathways (43). The GEP analysis between cells collected at time point day 0 plus 22 h with or without IL-2 revealed a specific and significant representation of the MAPK–ERK signaling pathway in IL-2+ cells. By using pharmacological inhibition with selective MEK antagonists, the specific upstream activator of ERK, we were able to inhibit IL-2–dependent initiation of PC differentiation without affecting proliferation demonstrating the requirement for a specific ERK signal to trigger PC production. Notably, although STAT5 factors are essential mediators of IL-2 signaling in T cells and are suspected to control IL-2–induced PRDM1 (45, 46), we found that STAT5 inhibitors used in a similar manner as MEK inhibitors did not affect IL-2–induced CD38hi cells. Our investigations at the protein level allowed deciphering of the IL-2 impact on ERKs signaling by showing increased ERK1/2 phosphorylation in response to IL-2 stimulation in cells cultivated with the mitogenic mixture that contained IL-2, the condition that leads to CD38hi cells. Altogether, this unexpected ERK enrollment in IL-2 initiation of PC generation matches perfectly with the recent study of Yasuda et al. (27), who demonstrated through mice models how ERK signaling plays a role in PC differentiation. Both studies raise questions about ERK implication because BCR signaling activates the ERK pathway and was described as a potent inhibitor of PC differentiation triggered by other receptors, such as TLR9 (47, 48). This discrepancy of ERK impact on PC generation compared with our data seems more related to differences between study design, timing, and type of molecules treatment. In the current study, BCR alone or associated with TLR9 signaling (CpG) was insufficient to drive PC generation (48). As suggested by Allman and Cancro (49), ERK might act like a huge integrator of several surface inputs, and some of them, like IL-2 in our model and/or high-affinity BCRs by transducing high-intensity signal (50), may push ERK-mediated signals to a threshold level that triggers beyond the induction of cell cycle progression to PC generation.
ERK-deficient B cells exhibit reduced BLIMP1 production after stimulation, which expression is required usually to generate PCs (27). In activated T cells, IL-2 induces BLIMP1, which in turn through a regulatory feedback loop inhibits IL-2 transcription (46). In our model using highly purified cell fractions, we found that IL-2–induced ERK signal acts on the PC differentiation associated with an early downregulation of IRF8 and BACH2 expression without any direct effect on PRDM1 at this time point. BACH2 is a PRDM1 inhibitor (51), its inhibition enhances BLIMP1 protein expression (52), and in BCR/ABL-positive cells, BACH2 expression was shown to be repressed via the MEK–ERK pathway (53). In contrast, IRF8 modulates the GC reaction by inducing the expression of BCL6 and AICDA in centroblasts (54). Taking into account all these findings, we propose to complete the previous scheme of PRDM1/BLIMP1 controlling PC differentiation (55) by adding the T cell-delivered IL-2 effect, which sustains indirectly the mitogenic-induced BLIMP1 expression by downregulating IRF8 and BACH2 via an ERK signal. This cytokine-mediated signaling is necessary early in B cell activation to allow the emergence of cells primed for PC differentiation among a highly proliferative cell population.
The authors have no financial conflicts of interest.
We thank Joelle Dulong and the Etablissement Français du Sang of Bretagne for providing the peripheral blood mononuclear cells, the “Institut Fédératif de Recherche-140” of Rennes University for the cell sorting core facility, and the Biogenouest Genomic Platform Rennes for bioinformatics advice.
S.L.G. performed research, analyzed data, and established the B cell differentiation model; G.C. designed experiments and performed research; C.D. performed research and analyzed cell signaling; D.R. performed microarray analyses; K.T. contributed to study design; and T.F. designed and supervised research, raised funds, and wrote the paper.
Microarray data presented in this article have been submitted to the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession no. GSE36975.
This work was supported by an internal grant from the Hematology and Immunology Laboratory, Pôle Cellules & Tissus, Centre Hospitalier Universitaire de Rennes. S.L.G. was supported by a research grant from La Ligue Contre le Cancer/Région Bretagne, and C.D. was supported by an internal grant from the Pôle Cellules & Tissus, Centre Hospitalier Universitaire de Rennes.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- fold change
- germinal center
- gene expression profiling
- gene list
- hierarchical clustering analysis
- naive B cell
- plasma cell
- quantitative RT-PCR
- follicular Th.
- Received January 24, 2012.
- Accepted April 25, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.