Abstract
The FOXP1 transcription factor is expressed throughout B cell development until its extinction just prior to terminal differentiation. Foxp1 nulls die of cardiac defects at midgestation, but adult rescue via fetal liver transfer led to a strong pre–B cell block. To circumvent these limitations and to investigate FOXP1 function at later stages of B cell differentiation, we generated and analyzed floxed (F) Foxp1 alleles deleted at pro–B, transitional (T) 1, and mature B cell stages. Mb-1cre–mediated deletion of Foxp1F/F confirmed its requirement for pro–B to pre–B transition. Cd21- and Cd19cre deletion led to significant reduction of germinal center formation and a second block in differentiation at the T2/marginal zone precursor stage. T-dependent and -independent immunization of FOXP1 mutants led to reduction of Ag-specific IgM, whereas responses of class-switched Abs were unimpaired. Yet, unexpectedly, plasmablast and plasma cell numbers were significantly increased by in vitro BCR stimulation of Foxp1F/F splenic follicular B cells but rapidly lost, as they were highly prone to apoptosis. RNA sequencing, gene set enrichment analysis, and chromatin immunoprecipitation sequencing analyses revealed strong enrichment for signatures related to downregulation of immune responses, apoptosis, and germinal center biology, including direct activation of Bcl6 and downregulation of Aicda/AID, the primary effector of somatic hypermutation, and class-switch recombination. These observations support a role for FOXP1 as a direct transcriptional regulator at key steps underlying B cell development in the mouse.
Introduction
The human genome encodes over 100 forkhead box (FOX) winged-helix transcription factors. Many are essential for development and organogenesis in mice and humans (1–3), as well as critical to immune system function, for example, Foxn in immunodeficiency (4), Foxj1 in autoimmune inflammation (5), Foxp3 in development and function of regulatory T cells (6–8), and Foxo1 in B cell development (9).
The Foxp1 locus encodes multiple isoforms of a subfamily initially discovered in a murine B cell lymphoma (10) but was soon realized to be expressed in multiple tissues (11, 12). Foxp1 function was first documented in the mouse heart as an essential regulator of cardiac valve morphogenesis, myocyte proliferation, and maturation (13). However, it accumulates highest in hematopoietic cells (14), and recent studies have established essential functions for FOXP1 in a variety of hematopoietic contexts, including activation of CD4+ peripheral T cells (15) and repression of T follicular (FO) helper cell differentiation (16).
Within the B lineage, FOXP1 expression is modest in progenitors, gathers most highly in activated B cells (ABC) (17–20) and is extinguished in plasma cells (PC) (17, 19). We previously demonstrated an essential function for FOXP1 in early B cell development via reconstitution of RAG-deficient mice with Foxp1−/− E14.5 fetal liver cells (21). We observed a strong block in pro–B cell development and reduction in expression of key B cell transcription factors (including Pax5, E2a, and EBF), suggesting that FOXP1 sits within essential regulatory hierarchies. Additionally, we observed that FOXP1 directly repressed RAG1/2 expression via its recruitment to the common Erag enhancer (21). A distantly related paralogue, FOXO1, also functions as a direct Rag1/2 repressor at both early and late stages of B cell development (9). FOXO1 acts, in part, via activation of IL-7R expression, a signaling receptor required for proper pre-BCR and BCR formation (9, 22).
Expression of FOXP1 is deregulated in a variety of human B cell tumors. For example, in MALT lymphoma, recurrent translocation within the IgH enhancer leads to FOXP1 overexpression and is associated poor clinical outcome (19). Likewise, overexpression of FOXP1 in the ABC subset of diffuse large B cell lymphoma (DLBCL) is a hallmark of poor prognosis (17, 23).
Despite its clinical implications in lymphoma and its broad expression pattern across B cell development (14, 24), the function of FOXP1 in normal, mature B cell development remains poorly understood. In this study, we employ a B cell lineage–specific Cre recombination approach at multiple stages to confirm and extend earlier findings derived from RAG1-mediated Foxp1 knockout transfers as well as to provide new insights into transcriptional regulation of germinal center (GC) formation, post-GC differentiation, and isotype control.
Materials and Methods
Mice
Foxp1-floxed mice (15) were crossed with Cd19-Cre (25), mb1-Cre (26), or Cd21-Cre (27) mice for deletion of Foxp1 at varying stages of B cell development. Mice were bred and housed in the animal facility of The University of Texas and Burnham Institute for Medical Research. All experiments received approval from the respective institutional animal care and use committees. Mice of 8–10 wk of age were used for peripheral B cell analysis and those 4–6 wk of age were used for bone marrow (BM) cultures.
Flow cytometry
Single-cell suspensions (1 × 106) were first stained for 20 min at 4°C with biotinylated and/or Fc-blocking Abs in flow cytometry buffer (1% [v/v] FBS and 0.1% [w/v] azide in PBS), followed by incubation with a mixture of Abs conjugated to FITC, PE, peridinin chlorophyll protein complex/cyanine 5.5, PE/indotricarbocyanine, allophycocyanin, or allophycocyanin/indotricarbocyanine. Abs used were as follows: biotinylated anti-BP1 (6C3) and anti-IgD (11-26; both from eBioscience), anti-CD9 (KM08), anti-CD23 (B3B4), anti-CD86 (GL1), anti-IgG1 (A85-1), anti-IgG2a/IgG2b (R2-40), and anti-IgG3 (R40-82; all from BD Pharmingen); FITC-conjugated annexin V (556419), anti-CD2 (RM2-5), anti-CD69 (HI.2F3), anti-GL7 (GL7), anti-H2Db (KH95), anti-CD21 (7G6), anti-κ (187.1), anti-λ (R26-46), and anti–mouse IgG1 (A85-1; all from BD Pharmingen); anti-CD5 (53-7.3), anti-IgD (11–26), anti-μ (II/41), and anti-rabbit (11-4839-81; all from eBioscience); PE-conjugated anti-CD138 (281-2) and anti-CD43 (S7; both from BD Pharmingen); anti-CD23 (B3B4), anti-IgD (11–26), anti-CD3e (145-2C11), and anti-CD127 (A7R34; all from eBioscience); peridinin chlorophyll protein complex/cyanine 5.5–conjugated anti-B220 (RA3-6B2) and streptavidin (both from eBioscience); PE/indotricarbocyanine-conjugated anti-CD11b (M1/70; BD Pharmingen); anti-CD3 (145-2C11) and streptavidin (both from eBioscience); allophycocyanin-conjugated anti-CD19 (ID3) and anti-CD62L (MEL-14; both from BD Pharmingen); anti-CD127 (A7R34), anti-B220 (RA3-6B2), anti-IgM (II/41) and streptavidin (all from eBioscience); allophycocyanin/indotricarbocyanine-conjugated anti-CD19 (ID3; BD Pharmingen); and anti-B220 (RA3-6B2; eBioscience). Data were collected on a FACSCanto (BD Biosciences) and were analyzed with FlowJo software (Tree Star).
Semiquantitative and quantitative RT-PCR
BM pro–B cells were purified by initial depletion of CD25+, CD2+, IgM+, and CD11b+ cells with biotinylated Abs, followed by depletion with antibiotin beads (Miltenyi Biotec). B cells were further purified (85–90% pro–B) with B220-positive selection (Miltenyi Biotec). RNA was purified from cells with TRIzol LS according to the manufacturer’s instructions (Invitrogen). The SuperScript II First-Strand Synthesis Kit (Invitrogen) was used for reverse transcription. Serially diluted cDNA was used for 30 cycles of PCR with published primer pairs (28). Quantitative RT-PCR analysis employed RNA from purified B cells or from immunized mice as described (29) and normalized to expression of Gapdh (glyceraldehyde phosphate dehydrogenase).
Immunization and ELISA
For T cell–dependent immunization, mice were immunized i.p. with 100 μg of either NP (19)/keyhole limpet hemocyanin (KLH) or NP (24)/KLH precipitated in alum (Imject; Pierce), and serum was collected on days 0, 7, and 14 (or 21). For T-independent (TI)–2 immunization, mice were immunized with 10 μg TNP/Ficoll, and serum was collected on days 0, 7, and 21. ELISA for Ag-specific Ab was performed as previously described (29).
Cell culture
RBCs from splenic single-cell suspensions were lysed for 5 min at 4°C with ammonium/chloride/potassium buffer (150 mM NH4Cl, 1 mM KHCO3, and 0.1 mM Na2EDTA). B cells were purified by negative selection with CD43-labeled magnetic beads (Miltenyi Biotec). For CFSE cultures, B cells were labeled for 10 min at 20°C with 2.5 μM CFSE in PBS and washed and then were cultured for 3 d at a density of 2 × 106 cells/ml in a 96-well, flat-bottom plate in RPMI medium with 10% (v/v) FBS (Mediatech), penicillin/streptomycin (100 U/ml), l-glutamine (2 mM), and sodium pyruvate (1 mM) with goat anti–mouse IgM F(ab′)2 (The Jackson Laboratory) or rabbit Ab to mouse intact IgM (The Jackson Laboratory) at a concentration of 10, 1 or 0.1 μg/ml, with or without IL-4 (10 ng/ml; R&D Systems) or LPS (20 μg/ml; Sigma-Aldrich) plus IL-4 (10 ng/ml).
B cell progenitor retroviral infection
Bicistronic, retroviral plasmid backbones used in this work are pMIG (MSCV/IRES/GFP) and p–mouse stem cell virus–internal ribosomal entry site–Thy-1.1 (MIT) (MSCV/IRES/Thy1.1). In vitro cultures were assessed for IL-7Rα expression at 3 or 5 d postinfection with the above retroviruses, as previously described (9). Viral supernatants were generated from transfected Phoenix/eco cells as described (29). B cells were positively selected from BM with B220-labeled magnetic beads (Miltenyi Biotec). Cells were cultured for 48 h in pre–B cell culture medium consisting of Opti-MEM with 15% (v/v) FBS (Hyclone), penicillin/streptomycin (100 U/ml), l-glutamine (2 mM), and 2-ME (55 μM) with the addition of IL-7 (5 ng/ml; R&D Systems). Cells were resuspended at a density of 1 × 106 cells/ml in supernatants of cells transfected with a mouse stem cell–based retrovirus-encoding Bcl-xL, previously incubated with polybrene (8 μg/ml) and diluted 2-fold with OMEM. Cells were spun at 1000 g for 90 min at 30°C and then were incubated for 60 min at 37°C. Cells were then collected and resuspended in pre–B cell culture medium with IL-7 (5 ng/ml) and were cultured for an additional 48 h, then were superinfected with either empty virus (MIT) or MIT-Cre. Cells were returned to culture with IL-7 and were analyzed 3 and 5 d later by flow cytometry.
RNA isolation and RNA sequencing
Total RNA was extracted from FACS-sorted Cd43 splenocytes using TRIzol reagent (Invitrogen). Oligo(dT)-primed cDNA was prepared using SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen). For RNA sequencing (RNA-seq), RNA was isolated from CD19/creFoxp1floxflox (F/F) mice or inbred C57BL/6 controls. Library preparation employed the Nextera DNA Library Preparation Kit from Illumina (San Diego, CA). cDNA was analyzed by deep sequencing using Illumina sequencing technology. Data were analyzed using a high-throughput, next-generation sequencing analysis pipeline: FASTQ files were aligned to the mouse genome (mm9, National Center for Biotechnology Information Build 37) using TopHat2. Gene expression profiles for the individual samples were calculated with Cufflinks as reads per kilobase of transcript, per million mapped reads values, Cd19cre; Foxp1 samples were normalized to controls, and statistical analysis was performed using Cuffdiff. Gene ontology (GO) was used to determine significantly affected pathways and gene set enrichment analysis (GSEA) using ordered gene expression levels of Cd19creFoxp1 derived CD43 splenocytes and controls. Normalized enrichment score and false discovery rate q-values; false discovery rate ≤0.25 is considered significant.
Chromatin immunoprecipitation sequencing
Chromatin immunoprecipitation sequencing (ChIP-seq) assays were performed as described previously (30). ChIP pulldowns that were performed with either an IgG1 monoclonal FOXP1 Ab (12) or an anti-FOXP1 rabbit polyclonal Ab (21) were optimized by end point ChIP/PCR with a previously determined FOXP1 target, the enhancer of RAG (Erag) (23). Immunoprecipitation for ChIP-seq was performed with the anti-FOXP1 rabbit polyclonal Ab. DNA was analyzed by Illumina deep sequencing technology.
Biochemistry
For short-term stimulation, purified B cells at a density of 2 × 107 cells/ml were stimulated for various times with anti-IgM F(ab′)2 (10 μg/ml) in PBS. For overnight stimulation, cells at a density of 2 × 107 cells/ ml were cultured in RPMI medium with 10% (v/v) FBS (Mediatech), penicillin/streptomycin (100 U/ml), l-glutamine (2 mM), and sodium pyruvate (1 mM). Reactions were stopped by the addition of cold PBS, then cells were immediately pelleted and were lysed with radioimmunoprecipitation assay buffer (25 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1% (v/v) Nonidet P-40, 1% (w/v) sodium deoxycholate, and 0.1% (v/v) SDS) that included DNAse, protease, and phosphatase inhibitors.
Immunohistochemistry
Spleens were frozen in optimum-cutting temperature compound (Tissue-Tek), and sections 8 μm in thickness were stained with FITC-conjugated Ab to peanut agglutinin (PNA), with allophycocyanin-conjugated anti-B220 (pseudocolored red with Slidebrook) and blue with Moma-1 FITC–conjugated (Serotec MorphoSys, Oxford, U.K.). Cryostat sections included in cryostat-embedding medium (Bio-Optica) were fixed in cold acetone, washed with PBS (Sigma Chemical, St Louis, MO), and incubated for 45 min with the above Abs and were analyzed in a confocal microscope (FluoView for Olympus FV1000). Images acquired at ×20 and ×60 amplifications.
Statistics
Groups of three to eight mice were used for statistical analysis. The p values were calculated with Student t test or Welch t test.
Results
FOXP1 is essential for pro–B to pre–B transition
To confirm and extend our results obtained in Foxp1−/−/Rag1−/− fetal liver transfers (21), we crossed Foxp1F/F mice with the well-characterized deleter strain mb-1cre (26). Mb1-Cre targeting is restricted to the late Ly-6d+ common lymphocyte progenitor—considered the first stage of B cell development [Inlay et al. (31)]. Robust deletion was observed in both BM and spleen (Fig. 1A). Mb1Cre/Foxp1F/F mice displayed a reduction in B220+Cd43+Cd24+Bp-1+ late pro/early pre–B cells (Hardy fractions C and C′) in BM (Fig. 1B). Loss of these B220+Cd43+ cells was confirmed via analysis of MHC expression in the secreted IgM (sIgM)/B220+Cd19+ fraction (Fig. 1C). Reduction in the percentages of Cd25+ pre–B cells further corroborated a defect in the transition from pro–B to pre–B cell stage (Fig. 1D). Interestingly, Cd23+ transitional and FO B cells had strong reductions, whereas marginal zone (MZ) B cells were increased in the periphery (Fig. 1E, 1F).
FOXP1 is essential for pro–B to pre–B transition.
(A) Western blot analysis of Foxp1 protein levels in purified B220+ B cells isolated from Foxp1f/f/mb-1cre and control BM and spleen (Sp). β-Actin was analyzed as a loading control. (B) Foxp1f/f/mb-1cre mice display a significant reduction in B220+Cd43+Cd24+BP-1+ pro–B cells in the BM. (C) Confirmation of pro–B cell reduction in BM by analysis of MHC expression in the sIgM/B220+Cd19 fraction. Rag1−/− BM served as a negative control. (D) Mb1Cre/Foxp1FF mice display a significant reduction in BM percentages of Cd25+ pre–B cells. Mice were bred upon a LacZ/EGFP (ZEG) background (104) to allow mb-1cre deletion efficiency to be scored by GFP fluorescence as we have described previously (9, 25, 29). (E) Percentages of B220+Cd19+ B cells are reduced in Foxp1f/f/mb-1cre Sps. Middle panels, percentages of IgMloIgDhi within B220+Cd19+ cells are reduced in Foxp1f/f/mb-1cre Sps. Right panels, percentages of mature Cd19+IgD+ B cells are reduced within Foxp1f/f/mb-1cre lymph nodes and peritoneal cavities. (F) Loss of FOXP1 results in reduction of splenic B cell subtypes. Left panels, Mb1Cre/Foxp1FF mice show significant reduction in percentages of FO B cells based on expression of Cd21 and Cd23. Middle panels, reduction of immature (Cd23−) and mature (CD23+) B220+ cells in FOXP1 mutants. Right panels, Cd23+ and Cd23− cells were further subdivided based on Cd21 and IgM expression into FO (Cd23+Cd21+IgMlo), T2 (Cd23+Cd21+IgMhi), MZ (Cd23−Cd21+IgMhi), and T1 (Cd23−Cd21−IgMhi). (G) FOXP1 does not regulate pro–B to pre–B cell proliferation. CD19+ BM from FoxO1F/F (left panels) and Foxp1F/F (right panels) in vitro cultures were assessed for IL-7Rα expression at 3 or 5 d postinfection with Thy1+pMIT-Cre. Robust and equivalent deletion was achieved for both floxed alleles (Supplemental Fig. 1). The data shown are representative of a minimum of three replicate measurements.
FOXP1 is essential for pro–B to pre–B transition.
(A) Western blot analysis of Foxp1 protein levels in purified B220+ B cells isolated from Foxp1f/f/mb-1cre and control BM and spleen (Sp). β-Actin was analyzed as a loading control. (B) Foxp1f/f/mb-1cre mice display a significant reduction in B220+Cd43+Cd24+BP-1+ pro–B cells in the BM. (C) Confirmation of pro–B cell reduction in BM by analysis of MHC expression in the sIgM/B220+Cd19 fraction. Rag1−/− BM served as a negative control. (D) Mb1Cre/Foxp1FF mice display a significant reduction in BM percentages of Cd25+ pre–B cells. Mice were bred upon a LacZ/EGFP (ZEG) background (104) to allow mb-1cre deletion efficiency to be scored by GFP fluorescence as we have described previously (9, 25, 29). (E) Percentages of B220+Cd19+ B cells are reduced in Foxp1f/f/mb-1cre Sps. Middle panels, percentages of IgMloIgDhi within B220+Cd19+ cells are reduced in Foxp1f/f/mb-1cre Sps. Right panels, percentages of mature Cd19+IgD+ B cells are reduced within Foxp1f/f/mb-1cre lymph nodes and peritoneal cavities. (F) Loss of FOXP1 results in reduction of splenic B cell subtypes. Left panels, Mb1Cre/Foxp1FF mice show significant reduction in percentages of FO B cells based on expression of Cd21 and Cd23. Middle panels, reduction of immature (Cd23−) and mature (CD23+) B220+ cells in FOXP1 mutants. Right panels, Cd23+ and Cd23− cells were further subdivided based on Cd21 and IgM expression into FO (Cd23+Cd21+IgMlo), T2 (Cd23+Cd21+IgMhi), MZ (Cd23−Cd21+IgMhi), and T1 (Cd23−Cd21−IgMhi). (G) FOXP1 does not regulate pro–B to pre–B cell proliferation. CD19+ BM from FoxO1F/F (left panels) and Foxp1F/F (right panels) in vitro cultures were assessed for IL-7Rα expression at 3 or 5 d postinfection with Thy1+pMIT-Cre. Robust and equivalent deletion was achieved for both floxed alleles (Supplemental Fig. 1). The data shown are representative of a minimum of three replicate measurements.
Because the percentages of pro–B cells were reduced in mb1Cre/Foxp1F/F mice (Fig. 1B), we compared the effect of FOXP1 loss upon pro–B cell proliferation in response to IL-7 with that of FOXO1, a paralogue shown previously as essential for IL-7R–mediated proliferation (9). Foxo1- and Foxp1-floxed pro–B cells were purified, expanded in vitro with IL-7, and then infected with either MSCV/Thy1.1+ pMIT/Cre or empty retroviral control. FACS analyses at days 3 and 5 postinfection revealed robust deletion of cultured Foxp1- and Foxo1-floxed BM (Supplemental Fig. 1). However, we observed no reduction of IL-7R expression upon loss of FOXP1 (Fig. 1G).
FOXP1 and FOXO1 are both essential for direct repression of RAG1/RAG2 transcription (21, 26, 27). This defect alone could underlie their developmental blocks at pro–B cell stages. In turn, reduction of total numbers and percentages of peripheral B cells as well as reduced spleen sizes would be anticipated. But because FOXP1 is expendable for IL-7–mediated proliferation, FOXP1 and FOXO1 act, at least, in part, through additional, unshared mechanisms.
FOXP1 loss leads to reduction of mature B cells
To examine the consequences of FOXP1 loss in mature B cells, we crossed floxed Foxp1 mice with the Cd19-cre–deleter strain (25). Consistent with previous reports (32), we observed significant loss of FOXP1 in mature B cells (Fig. 2). Specifically, we noted that Cd19+Cd21+IgM+ recirculating B cells (Hardy fraction F) were significantly reduced in mutant BM compared with WT littermate controls (4.4 ± 1.3% versus 12.6 ± 2.2%; p < 0.0001) (Fig. 2A). Analyses shown in Fig. 2B indicated that spleens of conditional knockout (cKO) mice were significantly reduced in Cd19+IgM+Cd23+ FO B cells (32.7 ± 4.2% versus 46.7 ± 3.2%; p < 0.0001).
FOXP1 is required for maintenance of mature B cells in BM and spleen.
(A) BM of Cd19Cre/Foxp1F/F mice contain similar frequencies of mononucleated cells in the light scatter gate (top panel) but show significant loss of mature recirculating B cells (fraction F) relative to F/F controls when phenotyped as Cd19+Cd23+ (center panel; squares) or Cd19+Cd21+IgM+ (bottom panel; ovals) using flow cytometry. (B) Spleens of Cd19Cre/Foxp1F/F mice are significantly reduced in mature recirculating FO B cells relative to F/F controls whether phenotyped as Cd19+Cd23+ (center panel; squares) or IgM+Cd23+ (bottom panel; ovals). Surface IgM is reduced on B220+ B cells in Cd19Cre/Foxp1F/F BM (left) and spleen (right). (C) Frequency of Cd19+Cd23+ cells in WT (n = 5) and Cd19Cre/Foxp1F/F (n = 4) mice. Three independent experiments. Welch t test. ***p < 0.0001.
FOXP1 is required for maintenance of mature B cells in BM and spleen.
(A) BM of Cd19Cre/Foxp1F/F mice contain similar frequencies of mononucleated cells in the light scatter gate (top panel) but show significant loss of mature recirculating B cells (fraction F) relative to F/F controls when phenotyped as Cd19+Cd23+ (center panel; squares) or Cd19+Cd21+IgM+ (bottom panel; ovals) using flow cytometry. (B) Spleens of Cd19Cre/Foxp1F/F mice are significantly reduced in mature recirculating FO B cells relative to F/F controls whether phenotyped as Cd19+Cd23+ (center panel; squares) or IgM+Cd23+ (bottom panel; ovals). Surface IgM is reduced on B220+ B cells in Cd19Cre/Foxp1F/F BM (left) and spleen (right). (C) Frequency of Cd19+Cd23+ cells in WT (n = 5) and Cd19Cre/Foxp1F/F (n = 4) mice. Three independent experiments. Welch t test. ***p < 0.0001.
FOXP1 loss leads to stalling at transitional stage 2
To confirm this mature B cell phenotype, we bypassed the pro–B cell block by generating a Cd21Cre/Foxp1F/F allele. Cd21-cre–mediated deletion initiates primarily at the transitional 1 (T1) stage (27). Unlike mb1Cre/Foxp1F/F and Cd19Cre/Foxp1F/F mutant spleens, we observed no reduction in the levels of FO and MZ B (MZB) cells in the spleens, lymph nodes, and peritoneal cavities of Cd21-cre mutants (Fig. 3A–C). Upregulation of the activation marker Cd86/B7-2, a costimulatory molecule expressed on the surface of APCs (including B cells), was consistently observed in mutant spleens (Fig. 3D). Considering the importance of B7-2 in regulating both T and B cell function (33, 34), we readdress its deregulation in more detail below.
FOXP1 loss in developing transitional B cells leads to cell death.
Analysis of B cell subsets following FOXP1 reduction with Cd21-Cre. We observed no reduction in the percentages of B220+Cd23+ FO B cells (left) nor IgM+Cd21+ MZB cells (right) within mutant spleens (A) or lymph nodes (B) of Cd21Cre/Foxp1F/F mice. (C) Peritoneal cavities of Cd21Cre/Foxp1F/F mice (left) contain normal percentages of B220+Cd19+ FO B cells, normal percentages of Cd5+B220+Cd19+ B1 B cells (right) (D). Left, Cd21Cre/Foxp1F/F splenic (S) B cells express higher levels of the activation marker, CD86. Right, color-coded comparisons of S CD86 levels on individual mutants (cKO) with levels on litter-matched controls (WT). (E) T2 (B220+Cd21hiCd23hi) cells are decreased, and T1 (B220+Cd21hiCd21low) are increased in Cd21Cre/Foxp1F/F mice relative to controls. (F) Spleens have normal percentages of B220+Cd23-IgM+ and B220+Cd23−Cd21+ MZB cells but have fewer B220+Cd23+Cd21+ MZP cells. Mature, immature, and MZ (upper left, upper right, and lower left) and peritoneal (lower right) subset absolute numbers are normal, whereas absolute numbers of B220+IgMhiCd21hiCd23+ MZP/T2/MZPs in the spleens of Cd21Cre/Foxp1F/F mice are significantly decreased. *p ≤ 0.02.
FOXP1 loss in developing transitional B cells leads to cell death.
Analysis of B cell subsets following FOXP1 reduction with Cd21-Cre. We observed no reduction in the percentages of B220+Cd23+ FO B cells (left) nor IgM+Cd21+ MZB cells (right) within mutant spleens (A) or lymph nodes (B) of Cd21Cre/Foxp1F/F mice. (C) Peritoneal cavities of Cd21Cre/Foxp1F/F mice (left) contain normal percentages of B220+Cd19+ FO B cells, normal percentages of Cd5+B220+Cd19+ B1 B cells (right) (D). Left, Cd21Cre/Foxp1F/F splenic (S) B cells express higher levels of the activation marker, CD86. Right, color-coded comparisons of S CD86 levels on individual mutants (cKO) with levels on litter-matched controls (WT). (E) T2 (B220+Cd21hiCd23hi) cells are decreased, and T1 (B220+Cd21hiCd21low) are increased in Cd21Cre/Foxp1F/F mice relative to controls. (F) Spleens have normal percentages of B220+Cd23-IgM+ and B220+Cd23−Cd21+ MZB cells but have fewer B220+Cd23+Cd21+ MZP cells. Mature, immature, and MZ (upper left, upper right, and lower left) and peritoneal (lower right) subset absolute numbers are normal, whereas absolute numbers of B220+IgMhiCd21hiCd23+ MZP/T2/MZPs in the spleens of Cd21Cre/Foxp1F/F mice are significantly decreased. *p ≤ 0.02.
In addition, we noted a small, but consistent, increase in T1 and a decrease in T2 cells (Fig. 3E). On further inspection, splenic T2/MZ precursors (MZPs; phenotyped as B220+IgMhiCd21hiCd23+), previously identified as suppressive regulatory B cells (Bregs) primarily committed to MZB cell differentiation (35), were reduced significantly in absolute numbers (p < 0.002; Fig. 3F).
Accumulation of FOXP1-deficient FO B cells is selectively enhanced
One explanation for this apparent contradiction was that Foxp1-deficient mature B cells undergo higher turnover rates within the periphery. However, if these populations are rapidly replenished by transitional cells, as reflected by the lower (exhausted) numbers of MZP/T2 cells (Fig. 3F), normal numbers of mature B cells would, nonetheless, accumulate in the periphery. To test this hypothesis, BrdU treatment was administrated in drinking water over a period of 6 wk. As shown in Fig. 4A, the Cd21Cre/Foxp1F/F Foxp1–deficient splenic B cell pool underwent normal turnover rates, whereas the lymph node pool turned over modestly faster. However, FACS analysis of the indicated B cell subsets (Fig. 4B) revealed that mutant mature B cells (B220+Cd24lowIgMlow), but not transitional (B220+Cd24hiIgMhi) or the total B220+ B cell pool, accumulated more highly in the periphery of Cd21Cre/Foxp1F/F mutants relative to controls as judged by BrdU+ percentages (p ≤ 0.02; Fig. 4A, 4B). Therefore, cKO B cells persisted in the periphery and allowed for further examination of the consequence of Foxp1 absence in mature B cells.
FOXP1-deficient B cells undergo proliferative and survival defects.
(A) Absolute numbers of the FOXP1-deficient B cell pool indicate normal turnover rates in spleen (left), whereas the lymph node pool (right) turned over modestly (but not significantly) faster following administration of BrdU in drinking water over a period of 6 wk. (B) FACS analyses of the indicated subsets following BrdU administration as in (A). FOXP1-deficient peripheral mature B cells (B220+Cd24lowIgMlow) but neither transitional (B220+Cd24hiIgMhi) nor the total B220 B cell pool underwent higher turnover rates, as indicated by reduced BrdU uptake; (n = 3 WT and n = 4 cKO). *p ≤ 0.02. (C) Foxp1 cKO B cells suffer proliferative defects following anti-Cd40+ IL-4 stimulation. Mutant and WT CD43+ splenic FO B cells were analyzed for proliferation via CFSE dilution 4 d following culture with either anti-IgM(Fab’2), LPS alone, or in combination with IL-4 with or without anti-CD40. FOXP1-deficient B cells proliferated normally to all stimuli with the exception of Cd40+IL-4 treatment, which typically drives FO B cells to IgG1 and IgE. (D) FOXP1-deficient, stimulated [as in (C)] mature B cells are highly prone to apoptosis. Annexin V staining and FACS analyses were performed as described in 2Materials and Methods. Maximal apoptosis was observed following culture with anti-IgM; more modest apoptosis followed culture with anti-IgM+IL-4, treatments known to upregulate IgM and IgG3.
FOXP1-deficient B cells undergo proliferative and survival defects.
(A) Absolute numbers of the FOXP1-deficient B cell pool indicate normal turnover rates in spleen (left), whereas the lymph node pool (right) turned over modestly (but not significantly) faster following administration of BrdU in drinking water over a period of 6 wk. (B) FACS analyses of the indicated subsets following BrdU administration as in (A). FOXP1-deficient peripheral mature B cells (B220+Cd24lowIgMlow) but neither transitional (B220+Cd24hiIgMhi) nor the total B220 B cell pool underwent higher turnover rates, as indicated by reduced BrdU uptake; (n = 3 WT and n = 4 cKO). *p ≤ 0.02. (C) Foxp1 cKO B cells suffer proliferative defects following anti-Cd40+ IL-4 stimulation. Mutant and WT CD43+ splenic FO B cells were analyzed for proliferation via CFSE dilution 4 d following culture with either anti-IgM(Fab’2), LPS alone, or in combination with IL-4 with or without anti-CD40. FOXP1-deficient B cells proliferated normally to all stimuli with the exception of Cd40+IL-4 treatment, which typically drives FO B cells to IgG1 and IgE. (D) FOXP1-deficient, stimulated [as in (C)] mature B cells are highly prone to apoptosis. Annexin V staining and FACS analyses were performed as described in 2Materials and Methods. Maximal apoptosis was observed following culture with anti-IgM; more modest apoptosis followed culture with anti-IgM+IL-4, treatments known to upregulate IgM and IgG3.
FOXP1-deficient B cells proliferate normally, but present poor survival, mainly in the presence of anti-IgM
We tested whether mature Cd19Cre/Foxp1F/F mutant B cells would be susceptible to proliferative defects following BCR-mediated activation. Mutant and WT Cd43 negative splenic B cells were analyzed for proliferation via CFSE dilution 4 d following culture with either anti-IgM (Fab’2)+LPS alone or in combination with IL-4 with or without anti-Cd40. Foxp1-deficient B cells proliferated normally to these stimuli, with exception to an enhanced proliferation to anti-Cd40+IL-4 combination, which typically drives FO B cells to IgG1 and IgE (36, 37) (Fig. 4C, Supplemental Fig. 2). However, annexin V staining indicated that mutant B cells were highly prone to apoptosis after anti-IgM stimulation (Fig. 4D), and more modestly with anti-IgM+IL-4—events that direct IgM and IgG3 production. These data indicated that LPS stimulation induces normal proliferation and improves the survival of cKO cells.
FOXP1 loss results in defects in class-switch recombination, GC formation, and PC differentiation
The ultimate goal of class-switch recombination (CSR) is to generate Abs with high binding avidity. Prior to immunization, Cd19Cre/Foxp1F/F mutants and controls expressed relatively equivalent and modest levels of total IgG (IgG1, 2A, and 2B) on their surfaces (Supplemental Fig. 3). IgM, and to a lesser extent, IgG3, are the primary Ab classes secreted following immunization with a TI Ag. Immunization of Cd19Cre/Foxp1F/F mutants with a traditional TI-2 Ag (NP/Ficoll) led to a reduction in Ag-specific IgM and IgG3 (p < 0.001, NS; Fig. 5A). However, when mutants were immunized with either T-dependent Ags NP19/KLH (upper panel) or NP23/KLH, total cKO IgG were superior to controls (Fig. 5A), and Ag-specific IgM responses were reduced in mutants (Fig. 5A). Yet, in vitro, Foxp1 mutant FO B cells were fully capable of proliferating and expressing higher surface levels of IgG1 than controls (Supplemental Fig. 4).
FOXP1 loss results in defects in CSR, GC formation, and PC differentiation.
(A) Four top panels, Foxp1-deficient mice have reduced levels of serum IgM and IgG3 Abs as detected by ELISA following immunization with TI NP/Ficoll (left, IgG3 NS, IgM p < 0.001) or T-dependent (TD) NP (19)/KLH (right, IgM p = 0.008) (n = 4 per group). Bottom two panels, a parallel analysis was performed using a more highly conjugated NP (23)/KLH TD Ag for immunization. (B) Numbers of plasma-like cells generated in vitro following LPS or anti-Cd40+IL-4 stimulation were increased in Cd19Cre/Foxp1F/F mutants, as assessed by expression of the PC marker Cd138 (n = 4). (C) PB and PC generation in WT and Cd19Cre/Foxp1F/F mutant cells. Mutant B220+Cd138+ PB were in higher frequencies in the absence (∼4-fold) of stimulation (n = 3; WT, 2.76 ± 0.3, cKO 9.9 ± 1.34; p < 0.001) and PC frequency was increased in IgM+Cd40-stimulated FO B cells (∼10-fold, (n = 3; WT, 0.53 ± 0.08, cKO 4.85 ± 0.79; p < 0.001). (D) Cd19Cre/Foxp1F/F mutant spleens are reduced ∼2-fold in GL7+Fas+ GC B cells as assessed by FACS analysis (n = 3; WT, 0.54 ± 0.03, cKO 0.25 ± 0.10; p = 0.009). (E) Formation of GCs following NP/KLH immunization of FOXP1-deficient Cd19Cre/Foxp1F/F mice was severely disrupted (n = 4). Half-spleens were embedded in optimum cutting temperature (OCT) compound and then frozen at −80°C. Cryostat sections were immunostained with specific Abs as indicated. Upper panels, GCs (green) stained with PNA; no GCs were detected in cKO sections. Bottom panels, marginal sinus (blue) separating the MZ from FO B cells (red). FOXP1-deficient mice contained severely disrupted GCs by morphology. Magnification is with ×4 objective and ×10 lens objective for an original magnification ×40. Student t test was used for statistical analysis, where appropriate.
FOXP1 loss results in defects in CSR, GC formation, and PC differentiation.
(A) Four top panels, Foxp1-deficient mice have reduced levels of serum IgM and IgG3 Abs as detected by ELISA following immunization with TI NP/Ficoll (left, IgG3 NS, IgM p < 0.001) or T-dependent (TD) NP (19)/KLH (right, IgM p = 0.008) (n = 4 per group). Bottom two panels, a parallel analysis was performed using a more highly conjugated NP (23)/KLH TD Ag for immunization. (B) Numbers of plasma-like cells generated in vitro following LPS or anti-Cd40+IL-4 stimulation were increased in Cd19Cre/Foxp1F/F mutants, as assessed by expression of the PC marker Cd138 (n = 4). (C) PB and PC generation in WT and Cd19Cre/Foxp1F/F mutant cells. Mutant B220+Cd138+ PB were in higher frequencies in the absence (∼4-fold) of stimulation (n = 3; WT, 2.76 ± 0.3, cKO 9.9 ± 1.34; p < 0.001) and PC frequency was increased in IgM+Cd40-stimulated FO B cells (∼10-fold, (n = 3; WT, 0.53 ± 0.08, cKO 4.85 ± 0.79; p < 0.001). (D) Cd19Cre/Foxp1F/F mutant spleens are reduced ∼2-fold in GL7+Fas+ GC B cells as assessed by FACS analysis (n = 3; WT, 0.54 ± 0.03, cKO 0.25 ± 0.10; p = 0.009). (E) Formation of GCs following NP/KLH immunization of FOXP1-deficient Cd19Cre/Foxp1F/F mice was severely disrupted (n = 4). Half-spleens were embedded in optimum cutting temperature (OCT) compound and then frozen at −80°C. Cryostat sections were immunostained with specific Abs as indicated. Upper panels, GCs (green) stained with PNA; no GCs were detected in cKO sections. Bottom panels, marginal sinus (blue) separating the MZ from FO B cells (red). FOXP1-deficient mice contained severely disrupted GCs by morphology. Magnification is with ×4 objective and ×10 lens objective for an original magnification ×40. Student t test was used for statistical analysis, where appropriate.
These serum Ab deficiencies in IgM secretion led us to investigate the presumed downstream consequence, reduction in IgM-secreting PC. Staining with the classical PC marker Cd138 revealed a significant increase in numbers of putative PC following in vitro stimulation with agents (LPS+ and LPS+IL-4) known to specifically upregulate IgM (Fig. 5B). Further analysis revealed that B220+Cd138+ mutant plasmablast (PB) were significantly upregulated even in the absence of stimulation (Fig. 5C). But upon BCR activation with anti-IgM+Cd40, we observed not only increased PB, but also B220lowCd138+ PC. These results indicated that Foxp1 cKO IgM–bearing B cells differentiate unimpeded to PC.
The above deficiencies in CSR prompted us to more closely examination GC cells, the point through which incoming FO B cells mature and undergo CSR and somatic hypermutation (SHM). As shown in Fig. 5D, Cd19Cre/Foxp1F/F mutants suffered a ∼2-fold decrease in GL-7+Fas+ GC cells. Examination of splenic GC indicated that GC formation 14 d after NP/KLH immunization in FOXP1-deficient mice was severely disrupted (Fig. 5E), as assessed by morphology, by reduction in PNA+ GC metallophilic macrophages adjacent to the splenic MZ by reduction in sIgM expression and by disruption of the marginal sinus separating MZ from FO B cells (Fig. 5E).
These results suggested that in Foxp1 mutants, the Ag-independent (IgM) arm of the immune response stalls at the IgM+/IgG3+ PB to undergo apoptosis, whereas B cells entering the GC reaction (Ag-dependent phase) bypass this block to generate higher Ab production. That is, loss of Foxp1 leads to increased CSR.
Foxp1 dampens immune responses primarily by repression
The above data, along with the survival defects of Fig. 4, suggested that at least a significant portion of Foxp1-deficient GC B cells undergo apoptosis prior to the PB-to-PC transition, at which an excessive production of PC occurs. To determine the underlying transcriptional deregulation that might account for these consequences, we performed RNA-seq on isolated Cd19cre Cd43-negative GC mutant and WT mRNA. We observed 1841 genes significantly modulated as determined by Cufflinks analysis (p < 0.05). A total of 1709 genes were upregulated (repressed by FOXP1), and 132 were downregulated in cKO spleens (activated by FOXP1), as compared with controls (Fig. 6A).
FOXP1 regulates factors crucial to apoptosis, GC biology, PC development, and cytokine expression.
(A) RNA-seq expression analysis performed on Cd43− Cd19Cre/Foxp1F/F mutant and control splenocytes. There were 1841 genes significantly modulated as determined by Cufflinks analysis (p ≤ 0.05). A total of 1709 genes were upregulated (red; repressed by FOXP1), and 132 were downregulated (blue; activated by FOXP1), compared with controls. (B) DAVID Functional Annotation and GO analysis was performed on significantly modulated genes. Selected GO pathways are displayed. (C) GSEA was performed on gene sets derived from DAVID. The positively correlated enrichment plots indicated that FOXP1 repressed the immune response and the cell cycle response. (D) ChIP-seq for FOXP1 was performed on the human ABC DLBCL lines (TMD8, HBL1, and OCI-Ly10; all peak scores ≥10). Selected relevant and overlapping targets are shown, including FOXO3, BCL2A1, AICDA, IL-12A and IL-12B, IL-6R, BCL6, TLR8, PRDM1, CFD86, and CD80. (E) Heatmap of selected genes from splenic CD43-depleted B cell RNA-seq indicates significant alterations following loss of FOXP1 in Cd19Cre/Foxp1F/F (cKO) mice relative to controls (n = 3 or 4).
FOXP1 regulates factors crucial to apoptosis, GC biology, PC development, and cytokine expression.
(A) RNA-seq expression analysis performed on Cd43− Cd19Cre/Foxp1F/F mutant and control splenocytes. There were 1841 genes significantly modulated as determined by Cufflinks analysis (p ≤ 0.05). A total of 1709 genes were upregulated (red; repressed by FOXP1), and 132 were downregulated (blue; activated by FOXP1), compared with controls. (B) DAVID Functional Annotation and GO analysis was performed on significantly modulated genes. Selected GO pathways are displayed. (C) GSEA was performed on gene sets derived from DAVID. The positively correlated enrichment plots indicated that FOXP1 repressed the immune response and the cell cycle response. (D) ChIP-seq for FOXP1 was performed on the human ABC DLBCL lines (TMD8, HBL1, and OCI-Ly10; all peak scores ≥10). Selected relevant and overlapping targets are shown, including FOXO3, BCL2A1, AICDA, IL-12A and IL-12B, IL-6R, BCL6, TLR8, PRDM1, CFD86, and CD80. (E) Heatmap of selected genes from splenic CD43-depleted B cell RNA-seq indicates significant alterations following loss of FOXP1 in Cd19Cre/Foxp1F/F (cKO) mice relative to controls (n = 3 or 4).
DAVID Functional Annotation and GO analysis (38, 39) was performed on significantly modulated genes. Selected GO pathways, displayed in Fig. 6B, revealed strong enrichment for signatures related to apoptosis and immune responses, as well as other as-yet-to-be-examined mechanisms such as cell cycle control and wounding (Fig. 6B). For example, GSEA strongly correlated the immune response signature (p ≤ 1.08 × 10−18) with upregulated transcripts (Fig. 6C).
FOXP1 regulates factors crucial to apoptosis, GC biology, PC development, and cytokine expression
Although FOXP1 acts primarily as an immune response repressor, a highly significant and directly upregulated target was Bcl6, as determined by RNA-seq modulation and binding the promoter region in our previously generated ChIP-seq data in cell lines with high FOXP1 expression (23)(Figs. 6D, 6E). BCL-6 acts broadly as a “master regulator” to control a complex network of signaling pathways crucial to apoptosis and GC biology (40, 41) (readdressed in 23Discussion). BCL-6 targets deregulated in our analyses included TLR (TLR5 and TLR8), which function in the innate immune system to detect bacterial flagellin (42) or imidazoquinolines and viral ssRNA (43, 44), respectively. FOXP1 also targeted BCL2A1A, which acts downstream of BCL-6 to block caspase activation and prevents the release of proapoptotic cytochrome C from mitochondria (45).
Repressed by Foxp1 was another essential mediator of GC biology, Aicda, which encodes the activation-induced cytidine deaminase (AID), the primary effector of SHM and CSR (46–49) (Fig. 6D, 6E). AID, which also is indirectly regulated by BCL-6 via micro-RNAs (50), accumulates to highest levels in GC B cells (46, 49, 51), where it targets both IgH variable exons as well as “off-targets” distinguished by chromosomal translocations (48, 52). Because AID-mediated phosphorylation is required for double-stranded breaks generated during SHM and CSR, we suggest that the aberrant CSR observed in Fig. 5 owes, at least in part, to derepressed AID expression. Also identified were a number of directly repressed apoptotic factors, including FasL (53), as well as five of the nine membered CD300 family of paired activating and inhibitory receptors. These factors regulate apoptosis by binding to exposed phosphatidylserine and PE on the plasma membrane (54). Also directly downregulated was the apoptotic initiator FOXO3 (55).
We observed direct FOXP1-dependent repression of the costimulatory molecule CD86/B7-2 via both cell staining (Fig. 3E) as well as by ChIP-seq binding and RNA-seq analyses (Fig. 6D, 6E). B7-2 is known to be essential to the generation of normal Ab production and GC formation in vivo (56). Although the exact mechanism(s) that govern B7-2 expression on B cells is unknown, FOXP1 is clearly a negative regulator of both B7-2 as well as its associated CD3γ-chain (Fig. 6E).
A key transcription factor in the regulation of terminal PC differentiation repressed not only by FOXP1 (Fig. 6D, 6E), but also by BCL-6, is Prdm1/Blimp1 (57, 58). Because FOXP1 not only activated Bcl-6, but repressed Blimp1, one would predict strong activation of PC differentiation—exactly what we observed (Fig. 5C). In addition, FOXP1 repressed expression of both IL-6 and its receptor, IL-6R, which are both critical for terminal PC differentiation (Fig. 6D, 6E).
As shown in Fig. 6E, FOXP1 loss deregulated repression of 15 cytokines, including two members of the IL-12 family, the IL-12rb2R, IL-13r, IL-15, IL-17r, IL-18 (plus an inhibitory chain along with two of its receptors), and IL-1 plus four of its receptor chains. We found this highly unexpected, because few of these cytokines are known to be expressed in splenic FO B cells, the source from which our libraries were constructed. As recently reviewed by Turner et al. (59), these cytokine/receptor pairs constitute a broad range of specificities. For example, IL-12 is typically expressed in activated Th1 cells or macrophages where it stimulates type II IFN (IFN-γ) (60). Accordingly, upon FOXP1 loss, we observed activation of IFN-γ (Fig. 6E). IL-13R generally occupy T cells, whereas IL-15R resides on T and NK cells and IL-17R, on intestinal mucosa. IL-18 is a proinflammatory cytokine best known for augmenting NK activity via IFN-γ production, whereas IL-1 is produced by activated macrophages. Notably in this regard, FOXP1 repressed four members of the IL-1R family (IL-1R2, IL-1R1, ILRL2/Rrp2, and IL-1RL1), which are linked on chromosome 2q12. This observation raises the possibility that FOXP1 participates in long range–promoter/enhancer interactions (61).
These cytokine results have led us to speculate (detailed in 23Discussion) that FOXP1 might have evolved this repressive function to thwart the onset of inflammatory “cytokine storms.” The hypothesis is supported by the observed dependence on FOXP1 of IL-6, INF-γ, and IL-12A (Fig. 6E), whose L chain (IL-35) is synthesized under Blimp-1 control of PC expression (62, 63).
Discussion
Commitment to the B cell lineage followed by B cell proliferation and survival are highly orchestrated events. Each of the steps is dependent upon transcription regulation of VDJ recombination, CSR, or terminal PC differentiation. We show in this study that the Forkhead family member FOXP1 is a nonredundant key regulator at several points along this developmental pathway.
FOXP1 loss leads to a pro–B/pre–B block
We and others (21, 64) previously demonstrated that FOXP1 was required for B cell development. FOXP1 loss resulted in a block at the pro–B cell stage (21) by virtue of repression of key transcriptional regulators (e.g., E2a, EBF, and Pax5) and via direct repression of the Erag enhancer, which controls expression of RAG1/2. Our results were obtained via reconstitution of RAG-deficient mice with conventional Foxp1 null E14.5 fetal liver cells. In this study, we further probed this critical transition point by elimination of Foxp1 via mb-1cre (26), which promotes deletion at the late, Ly6d+ common lymphocyte progenitor, essentially the first stage of committed B cell development (31). In support of our earlier findings, we observed significant loss of pro– and pre–B cells in BM, as well as in the numbers of their descendants recirculating within the periphery as transitional, marginal, and FO B cells. That is, mb1-Cre/Foxp1F/F mice as compared with WT littermate controls retained ∼75% of their B220+ and Cd19+ total BM B cells, but only ∼33% of B cells beyond the pro–B to pre–B transition (Fig. 1A, 1B), thus confirming the previous block seen in Foxp1 null fetal livers (21).
Consistent with these observations, Rao et al. (65) showed that constitutive expression of the microRNA miR-34a, which directly targets FOXP1 mRNA, led to a block in B cell development at the pro–B to-pre–B cell transition, with concomitant reduction in mature B cells. However, miR-34 destruction of FOXP1 had no effect on pro–B cell numbers, even though miR-34 expression is 2-fold higher in pro–B than in pre–B cells. Pro–B cell depletion in our study was confirmed by intracellular IgM downregulation in the sIgM/B220+Cd19+ fraction (Fig. 1C). Although the source of these differences remain unclear, both studies suggest that a central role of FOXP1 is to promote output of the B cell developmental pathway.
Reduction of pro–B cells in our mb-1cre/Foxp1F/F mutants led us to determine whether the rate of pro–B cell proliferation in response to IL-7 was a contributing factor. In comparison with FOXO1, previously shown and verified in this study to be essential for IL-7Rα–mediated proliferation (65), we observed no reduction of IL-7Rα expression upon loss of FOXP1 (Fig. 1G). We found this curious given the previous observations of ours and others demonstrating that both forkhead box paralogues are essential for repressing Rag1 and Rag2 transcription (21, 26, 66). However, in CD4 T cells, FOXP1 functions as a repressor of IL-7Rα expression, and both FOXP1 and FOXO1 bind to the same fkh-binding site within the IL-7Rα enhancer (67). Although this competitive antagonism is present in TH cells, FOXP1 and FOXO1 act through nonredundant mechanisms but during pro–B to pre–B transition.
FOXP1 loss leads to reduction of mature B cells
A low input of cells from BM can still generate a peripheral B cell compartment, but with adverse characteristics (66). Following development in the BM, B cells are specified into functional subsets within the spleen, including mature FO and MZB cells as well as the transitional subsets T1, T2, and T3. Employing Cd19-cre, which strongly, but not exclusively, deletes at the mature B cell stage (25), we observed phenotypes consistent with mature B cell defects, including reduction of FO B cells in BM and spleen (Fig. 2A–C). Deletion with Cd19-Cre also revealed modest alterations in sIgM expression (Fig. 2A, 2B).
FOXP1 loss leads to stalling at transitional stage 2
To confirm that these effects did not result from BM progenitor populations that escaped Cd19cre/Foxp1F/F deletion, we eliminated Foxp1 with CD21-cre (27), which is expressed within the B cell lineage most strongly at the T2 transitional stage. As anticipated, the absolute numbers of pro- and pre–B were normal; yet, unexpectedly, lineages downstream of Cd21 deletion (FO and MZB cells) were spared as well (Fig. 3A–D). Notably in cKO spleens, we observed significant reduction (p ≤ 0.03) in absolute numbers of B220+IgMhiCd21hiCd23+ T2/MZP Bregs (Fig. 3E). T2/MZPs generally have been characterized as negative regulators that primarily reside in the spleen to control potentially damaging inflammatory responses (35). T2/MZPs were initially established as a population of bipotent precursors of both FO and MZB cells (67). More recent studies, however, suggest that they are predominantly progenitors of MZB cells (35). Amu et al. (68) reported that T2/MZP Bregs prevent and/or reverse lung inflammation via a not fully understood mechanism that reduces FOXP3+ T regulatory cells. We find those findings particularly curious given that FOXP3 and FOXP1 share high identity, form stable heterodimers (12), and their oligomerization leads to a phenotype in humans akin to XLAAD/IPEX autoimmune syndrome (69). Further studies on the role of FOXP1 in autoimmunity warrant consideration.
FOXP1 is required for mature B cell survival
In an attempt to understand how mature MZB and FO B cells in Cd21-Cre cKO mice were spared despite the loss in MZB/T2 precursors, we considered the possibility that the block was not so much failure of developmental transition but rather of T2 cell survival. More complex scenarios, such as a mild failure of survival of T1 cells and a severe failure of survival in T2 cells, with or without a failure of developmental transition, are also possible. But if our hypothesis was correct, BrdU-labeling experiments would predict a rapid loss of mature cells in the absence of influx from T2 progenitors while leaving other transitional B cells, and perhaps the total B cell pool, unscathed. Indeed, this was the case (Fig. 4A, 4B). In scanning the literature, we found a number of examples of T2 disruption [reviewed previously (70); but none without disruption of stages downstream].
Loss of FOXP1 results in apoptosis
The above results further predicted that mutant mature B cells would be more susceptible to proliferative defects. Indeed, mutant B cell pools underwent higher turnover rates as indicated by increased BrdU uptake (Fig. 4B). The underlying reason became quite clear: Foxp1-deficient B cells underwent rampant apoptosis, particularly following exposure to anti-IgM. Thus, we concluded that FOXP1 acts directly to repress apoptosis. Previously, we (23) and others (71) demonstrated in human B lymphoma that knockdown of FOXP1 led to apoptosis and upregulation of proapoptotic genes, including the BH3-only protein, BIK and TP53/p53. Also upregulated in both studies were several p53-regulatory proteins whose expression is correlated with bad prognosis (22).
Craig et al. (72) reported that the oncogenic activity of FOXP1 is repressed by p53-induced miR-34a, the same microRNA shown to directly block FOXP1-mediated pro–B/pre–B progression (65). In the present analyses, we found no evidence for a p53-dependent mechanism (Fig. 6A, 6B). Nonetheless, apoptosis/cell death GO categories achieved high significance (∼150 apoptosis-related transcripts), most of which were repressed by FOXP1. FOXP1 also targeted apoptotic BCL2 family members, such as BCL2A1A, which functions by releasing proapoptotic cytochrome C from mitochondria to block caspase activation (44). Additional apoptotic factors identified in our analyses as downregulated by FOXP1 include FasL (Fig. 6E), which participates in the extrinsic pathway of apoptosis by binding to its receptor FAS (73, 74). The FASL/FAS complex along with its adaptor, FADD, recruits procaspase-8 to form the death-inducing, signaling complex. Also strongly downregulated were five of the nine membered CD300 family of paired activating and inhibitory receptors. CD300 members regulate a broad array of immune cell processes, including apoptosis (54). Several members, notably CD300A, initiate the early stages of apoptosis by binding to exposed phosphatidylserine and PE on the outer leaflet of the plasma membrane, leading to their engulfment by phagocytes (54). Also downregulated was FOXO3, which functions as apoptotic initiator via upregulation of cell death genes, including Bim and Puma (75), or downregulation of antiapoptotic genes such as Flip (76).
FOXP1 regulates factors crucial to GC biology
Often referred to as the master regulator of GC biology (40, 41), BCL-6 primarily acts as a transcriptional repressor to modulate a complex network downstream of multiple signaling pathways crucial to GC biology. FOXP1 is now one of only two confirmed direct Bcl6 transactivators (77) (Fig. 6D, 6E) and, consequentially, several BCL-6 downstream targets, including TLR (TLR5 and TLR8), which function in the innate immune system to detect bacterial flagellin and imidazoquinolines (42, 44) or to detect viral ssRNA (43), respectively. That FOXP1 levels must be reduced for proper GC functionality (78), the timing of FOXP1 direct activation of Bcl6 must be restricted to a window after both B cell maturation and GC formation, but before activation of the GC B cell. In concert with such narrow timing window, an equally essential mediator upregulated in the GC found in this study to be directly repressed by FOXP1 is Aicda (Fig. 6D, 6E). Aicda encodes the primary effector of SHM and CSR, AID (46–49). AID is tightly regulated and only expressed in GC and ABC but is undetectable in naive B cells. Thus, the AID upregulation that results upon early loss of FOXP1 strongly contributes to both loss of GCs and aberrant CSR/Ig production (Fig. 5A, 5B, 5E).
FOXP1 repressed the expression of the B7-2/CD28 coactivation marker on GC B cells (Figs. 3E, 6D). Repression was also observed for its associated CD3-γ chain, which most typically is expressed with the TCR. We observed the anticipated consequence, based on earlier findings employing anti–B7-2 Ab inhibition (56, 79, 80)—disruption of both the B7-2/CD28 interaction that occurs between a T cell and a B cell in response to a T cell–dependent Ag as well as malformation of GCs (Fig. 5A, 5D, 5E). B7-2 is expressed at extremely low levels on resting B cells (81) and is a well-known activation marker on the B cell surface (82). Indeed, loss of FOXP1 clearly results in an ABC phenotype.
Because the major output of the GC leads to PC differentiation, it is critical to switch rapidly from the GC to the PC phenotype. Consistent with an earlier study (78), we observed downregulation of FOXP1 in GC (Fig. 5D), suggesting that FOXP1 reduction is essential for entering differentiation mode from GC to PC. This hypothesis is supported by the data discussed below.
FOXP1 regulates key transcription factors required for PB-to-PC transition
The central goal of terminal B cell differentiation, mass production of effective Abs, is orchestrated by a cascade of transcriptional regulators. These include IFN regulatory factor 4 (IRF4), which acts upstream of both the PC master regulator Prdm1/Blimp1. In addition to its central role in GC formation, BCL-6 represses Prdm1/Blimp1 to antagonize terminal differentiation to PC (57, 58). We show, in this study, that FOXP1 not only activates BCL-6, but represses Blimp1. Thus, if Blimp1 repression is alleviated by Foxp1 deletion, one would predict derepression of Blimp and robust activation of PC differentiation. Indeed, that is exactly what we observed (Fig. 5C). Reciprocally, BLIMP1 was shown to repress BCL-6 (57, 83) to sustain PC differentiation processivity.
BLIMP1 also inhibits PAX5, a central mediator for maintenance of B cell phenotype. In turn, PAX5 represses Xbp1, an essential regulator of the PC secretory phenotype. Van Keimpema et al. (84) employed transgenic FOXP1 overexpression to observe direct repression of Irf4, Prdm1, and Xbp1. Although our results reciprocally confirmed their finding of Prdm1 repression, we failed to detect significant modulation of Irf4 or Xbp1 in GC B cells. We did, however, confirm that in ABC/DLBCL lines that highly express FOXP1 direct downregulate Irf4 (23).
Also key to this pathway and repressed by FOXP1 are IL-6 and its receptor, IL-6R (Fig. 6D, 6E). By binding to IL-6R, IL-6 promotes maturation of B cells into Ab-secreting cells as well as survival and maintenance of long-lived PC and PB (85). We find it interesting in this context that IL-6 promotes commitment to T FO helper cells, a lineage previously shown to be dependent upon sustained expression of FOXP1 (16) via induction of BCL-6 (77). These data suggest a link between T and B cell GC responses that promote B cell proliferation and CSR phenotypes found to be disrupted by FOXP1 loss (Figs. 4A, 5A).
FOXP1 defends against a cytokine storm
As shown in Fig. 6E, FOXP1 significantly repressed expression of 15 cytokines, and/or their receptors or their regulatory chains. As detailed in 14Results, these cytokine/receptor pairs constitute a broad range of specificities, including several that stimulate IFN-γ (60, 86, 87). We also observed what appeared to be long distance operon control (61) in that FOXP1 corepressed four members of the IL-1R family clustered within 600 kbp on 2q12 (88).
Perhaps FOXP1 has evolved a protective role in B cells to repress this broad array of regulatory factors and their receptors to avoid inappropriate, if not cataclysmic, consequences. Severe clinical syndromes can be generated from overzealous host response to infection resulting in an activation cascade that autoamplifies cytokine production, i.e., the cytokine storm (89).
B cells can synthesize a few cytokines for multiple aspects of immunity, including lymphotoxins, which are essential for activation of secondary lymphoid organs, IL-6 and IFN-γ (both noted above as FOXP1 repressed targets), and TNF (90–92). B cells may regulate inflammatory immune responses, primarily through IL-10 and another FOXP1 target, IL-35. Although TH17 cells are the primary source of proinflammatory IL-17, CD138+B220+ PC are a major source of IL-17A during parasite infection (93). PC are also the major source of B cell–derived IL-35 in certain disease states (62). We find this significant in the present context as BLIMP-1, which is downregulated by FOXP1, controls the expression of EBI3, the H chain of IL-35 (62, 94).
Convergent analyses of FOXP1 B cell biology
During the preparation of this manuscript, Patzelt et al. (95) employed Cd19-cre–mediated deletion to address the function of Foxp1 in mature B cells. By employing both Cd19- and mb1-cre approaches, we confirmed their finding that Foxp1 loss resulted in significant reduction in mature FO B cells, but in MZB cells. However, by employing T1 stage deletion with Cd21-cre, we were able to bypass the pro–B cell block (which led to loss of FO B cells in the periphery and peritoneal cavities) to identify an additional, significant reduction in Bregs. As with Patzelt et al. (95), we found Foxp1-deficient B cells to be impaired in survival without significant consequence on proliferation (particularly following stimulation with anti-IgM). However, we observed defects in both T-independent and -dependent immunization, which we traced to faulty CSR and GC formation. Finally, both in this study and previously in human DLBCL (23), we confirmed the observation of Patzelt et al. (95) of defective expression of prosurvival BCL-2 family members, some of which we showed by ChIP-seq to be direct in human cell lines. In addition to survival, our RNA-seq, GSEA, and ChIP-seq analyses revealed critical FOXP1 target genes regulating GC biology, CSR, PB-to-PC transition, and cytokine production.
Long-term extensions
Employing multiple tissue- and stage-specific Cre-recombinases allowed us and Patzelt and colleagues (95) to identify multiple developmental transitions in which FOXP1 is essential. The failure to navigate these transitions appears to underlie numerous malignancies and autoimmunities. Initially characterized by germline deletion, in this study, we identified FOXP1 loss in the mouse as a requirement for pro–B to pre–B cell transition. In humans, this transition is often breached in acute lymphocytic leukemia, and FOXP1 loss or gain has been the culprit in several cases (19, 96–98). FOXP1 has been recognized as an oncogene in various mature B cell tumors, such as non–Hodgkin lymphomas (24, 99). With respect to the stall that we observed during T2 transition (Fig. 3E), human studies have implicated autoimmune outcomes, including expansions of circulating transitional cells in patients with systemic lupus erythematosus and immunodeficiency (100, 101). Finally, the third aberrant transition, dysregulation of PB-to-PC differentiation (Fig. 5C), is considered to be a principle oncogenic target of the more aggressive ABC subset of DLBCL (102), and FOXP1-deregulation has been strongly implicated (23, 84).
Nonetheless, a number of complexities remain to be considered, including the role of multiple FOXP1 isoforms and the wide range of cellular contexts, including GC and non-GC subtypes. That FOXP1/FOXP2 interaction is strongly associated with inferior DLBCL survival (103) underscores the need for further examination of FOXP1 binding partners and their impact on carcinogenesis. Understanding further subtleties between FOXP1 expression and function during disease will be critical in identifying additional FOXP pathway-based therapeutic strategies.
Acknowledgements
We thank K. Rajewsky (Harvard University) and M. Reth (Max-Planck Institute) for Cd21Cre and MB-1Cre mice, respectively. We thank June Harriss for excellent contribution to all aspects of the animal husbandry and Chhaya Das and Maya Ghosh for help in cell culture and molecular techniques. We thank members of the Rickert and Tucker laboratories for discussions and reading of the manuscript. RNA-seq was performed at the Next-Generation Sequencing core of the MD Anderson Cancer Center in Smithville, TX. Histology was performed at the histology core of the MD Anderson Cancer Center in Smithville, TX.
Footnotes
This work was supported by a Lymphoma Research Foundation Fellowship 300463 (to J.D.D.), National Institutes of Health (NIH) National Cancer Institute Fellowship F32CA110624 for production of the Foxp1 conditional knockout mice (to G.C.I.), NIH Grants AI059447 (to R.C.R.) and R01CA31534, Cancer Prevention Research Institute of Texas Grants RP100612 and RP120348, and the Marie Betzner Morrow Centennial endowment (to H.O.T.).
The sequences presented in this article have been submitted to the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo) under accession number GSE137463.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- ABC
activated B cell
- AID
activation-induced cytidine deaminase
- BM
bone marrow
- Breg
regulatory B cell
- ChIP-seq
chromatin immunoprecipitation sequencing
- CSR
class-switch recombination
- DLBCL
diffuse large B cell lymphoma
- F/F
floxflox
- FO
follicular
- GC
germinal center
- GO
gene ontology
- GSEA
gene set enrichment analysis
- KLH
keyhole limpet hemocyanin
- MIT
mouse stem cell virus–internal ribosomal entry site–Thy-1.1
- MZ
marginal zone
- MZP
MZ precursor
- PB
plasmablast
- PC
plasma cell
- PNA
peanut agglutinin
- RNA-seq
RNA sequencing
- SHM
somatic hypermutation
- sIgM
secreted IgM
- T1
transitional 1
- TI
T-independent
- WT
wild-type.
References
Disclosures
The authors have no financial conflicts of interest.