NF-κB factors control the induction of NFATc1 in B lymphocytes

NFATc1 ablation affects the formation of NF-κB complexes in splenic B cells

For the optimal induction of the Nfatc1 gene, splenic B cells have to be stimulated through their BCR by α-IgM whereas stimulation with α-CD40, LPS, or CpG – all of which induce NF-κB factors – exerts only a marginal effect on Nfatc1 induction 18. Compared to NFATc1 induction, α-IgM-mediated BCR signals exert a poor and transient effect on the induction of nuclear NF-κB complexes in splenic B cells whereas α-CD40, LPS, or CpG signals induce strongly and persistently nuclear NF-κB complexes (Supporting Information Fig. 1A and B). In EMSA supershift experiments we observed predominantly NF-κB1/p50 protein as component of faster migrating (p50/p50 homodimeric) NF-κB complexes and c-Rel as the main, albeit not sole component of slower migrating (c-Rel/p50 heterodimeric) NF-κB complexes upon α-IgM stimulation for 8 h (Fig. 1A).

At the RNA level, BCR stimulation by α-IgM led to a three- to fourfold increase in c-Rel transcripts whereas the transcripts of all other Rel members remained almost unaffected (Supporting Information Fig. 2). In contrast, α-CD40 stimulation of splenic B cells resulted in a four- to fivefold increase in RNA reads of Nfkb2 gene encoding p52 whereas only a slight increase was detected for RelB transcripts upon α-CD40 stimulation for 24 h (Supporting Information Fig. 2). Inactivation of the Nfatc1 gene in BM B cells from Nfatc1 ^fl/fl x mb1-cre mice did neither affect the transcripts of κB genes (results not shown) nor of occurrence of nuclear κB proteins (Fig. 1A).

Whereas NFATc1 ablation did not affect the expression of individual κB proteins, it influenced the formation of nuclear NF-κB complexes. In WT B cells that were strongly activated by “pulsed” α-IgM+α-CD40 or T + I treatment, the predominant formation of active, heterodimeric NF-κB complexes is observed, while more “inactive” p50/p50 homodimeric complexes are seen in Nfatc1^−/− B cells (compare lanes 5 + 6 with lanes 11 + 12 in Fig. 1B and see Supporting Information Fig. 3). These findings indicate that NFATc1 supports the formation of NF-κB complexes but does not markedly affect the expression of NF-κB members.

NFATc1 deficiency affects NF-κB complex formation in chicken DT40 B lymphoma cells

To investigate the effect of NFATc1 ablation on NF-κB in another type of B cells, we inactivated the NFATC1 gene in chicken DT40 B cells and studied the generation of nuclear NF-κB complexes. When we tested in EMSAs the occurrence of nuclear NF-κB complexes in NFATC1^−/− DT40 cells stimulated with α-IgM for 2 h, we observed a marked decrease in complex formation compared to WT DT40 cells. In particular, the formation of slowly migrating heterodimeric “activating” NF-κB complexes was found to be impaired (compare lanes 5–8 with lanes 1–4 in Fig. 2A).

Numerous signals leading to NF-κB activation in lymphocytes are mediated through the activation of protein kinase Iκκβ 19. Therefore, we investigated the effect of NFATC1 inactivation on the induction of Iκκβ activity in kinase assays using protein extracts from WT and NFATC1^−/− DT40 cells. While we did not see any difference in Iκκβ activity between WT and NFATC1^−/− cells stimulated with T + I, we observed a distinct weaker Iκκβ induction in NFATC1^−/− cells compared to WT cells stimulated through their BCR by the α−IgM Ab M4 for 10–30 min (or by a hybridoma supernatant for 10 min; Fig. 2B). This suggests that signaling events upstream of Iκκβ are controlled by NFATc1.

Nuclear NFATc1 complexes upon secondary BCR stimulation are not formed in Nfκb1^−/−/p50^−/− mice

Compared to the drastic effect of inactivation of Rela gene on embryonic development 20, inactivation of Nfkb1 gene exerted only a mild, if any effect on embryogenesis but affected strongly the activation and function of peripheral lymphocytes, in particular of B cells 21, 22. When we compared the NFATc1 induction by T + I in splenic B cells from Nfkb1^−/− mice with that in B cells from WT mice, we observed a mild decrease in formation of nuclear NFAT complexes upon primary induction by T + I for 0.5–24 h (Fig. 3A) but a very strong defect upon secondary stimulation (Fig. 3B). This might be due to a comparable amount of preformed cytosolic NFATc1 protein in splenic B cells from WT and p50-deficient mice, as detected in immunoblots of preparations from whole cellular proteins (Supporting Information Fig. 4). Upon secondary stimulation of B cells induced first by α-IgM for 24 h with T + I for 10 min to 4 h a drastic drop in NFATc1 levels was detected, both for the induction of nuclear NFATc1 complexes in EMSAs using nuclear protein from Nfkb1^−/− B cells and in western blots using whole cellular protein (Fig. 3B and C). This is in line with the strong decrease in P1 promoter-directed induction of NFATc1 RNA levels, as observed in RT-PCR assays. Already upon primary stimulation of splenic B by T + I for 0.5–6 h resulting in a three- to fourfold increase in P1-directed RNA transcripts in WT cells, a complete defect in P1-directed RNA induction was detected (Fig. 3D). And upon secondary stimulation—following the treatment of B cells by α-IgM for 24 h—a further decrease in P1-directed transcripts was observed (Fig. 3E). These findings demonstrate that p50 is necessary for the induction of P1 activity in B lymphocytes.

Ablation of c-Rel abolishes any NFATc1 induction

Similar to the inactivation of Nfkb1 gene, ablation of c-Rel did not lead to gross abnormalities in embryonal development and in differentiation of hematopoietic cells but in defects of B-cell function 23, 24. In particular, the growth of BCR-activated peripheral Rel^−/− B cells during the G1 phase of cell cycle and their promotion to progress from the G1 to S phase were strongly affected 24. Freshly prepared splenic B cells from c-Rel-deficient mice did not show any formation of nuclear NFATc1 complexes (Fig. 4A). The complexes that we observed in EMSAs using an NFAT consensus probe and nuclear protein from Rel^−/− B cells consisted mainly of NFATc2 but not NFATc1 (see lanes 5–8 in Fig. 4A). While in immunoblots with whole cell proteins one NFATc1 band was detected in nonstimulated c-Rel^−/− B cells, no NFATc1 induction was observed upon α-IgM or α-IgM+α-CD40 treatment in c-Rel^−/− B cells (Fig. 4B). By mapping of P1-directed NFATc1/α transcripts in RT-PCR assays, a similar suppression of NFATc1/α induction was detected at the transcriptional level (Fig. 4C).

Two composite κB/NFAT sites in the Nfatc1 P1 promoter

Within the 800 bp of Nfatc1 P1 promoter that are highly conserved between mouse and human, there are two composite κB/NFAT motifs located around the nucleotide positions –90 and –720, respectively (Fig. 5A). These motifs share the NFAT core binding motif ^T/_AGGAAA with numerous other sites to which both NFAT and NF-κB factors can bind, and with the sequence 5′-NGGRN(A/T)TTCC-3′ that was identified as optimal c-Rel-binding site 8. When these κB/NFAT sites were used as probes in EMSAs with nuclear protein from in vitro-induced GC B cells (iGCs) which—due to their stimulation by CD40 25—express high levels of NF-κB, we detected the formation of typical κB complexes that were indistinguishable from complexes generated with a κB consensus site (Fig. 5B and not shown). This conclusion is supported by EMSA competition experiments in which competition with a 50-fold molar excess of a κB consensus site, but not with a site mutated within the κB motif, led to disappearance of NF-κB complexes (Fig. 5C). However, in EMSAs with nuclear proteins from splenic B cells that contain less NF-κB proteins than iGCs, a very poor formation of p50 homodimers was observed with the two κB/NFAT sites, compared to a κB consensus site (compare lanes 1–8 with lanes 9–14 in Fig. 5D).

To demonstrate the binding of p50 and c-Rel to the P1 promoter in vivo, by using κB-specific Abs in ChIP (chromatin immune precipitation) assays, we precipitated P1 chromatin from unstimulated splenic B cells or cells stimulated by T + I for 4 h. In PCRs with P1-specific primers, we observed a weak binding of both p50 and c-Rel to the chromatin of P1 promoter in unstimulated B cells, but a marked six- to tenfold increase upon B-cell induction (Fig. 5E and F). This shows that BCR signals enhance the binding of p50 and c-Rel to the Nfatc1 P1 promoter.

To determine the functional relevance of factor binding to the κB/NFAT sites, we introduced point mutations into the sites (Fig. 5A) that abolished the binding of NF-κB and NFAT, and cloned the mutated promoter in front of a luciferase reporter gene. Cotransfections of P1-directed constructs with a vector expressing c-Rel into 293 HEK cells revealed a two- to threefold increase in luciferase activity upon T + I induction and a further increase by cotransfecting c-Rel. By using P1-luciferase constructs that were mutated in either the proximal κB/NFAT_-90 or the distal κB/NFAT_–720 motif, or in both motifs, we observed a strong decrease in luciferase induction, and c-Rel had no effect on the construct bearing mutations in both κB/NFAT sites (Fig. 6A).

Upon transfection into El4 thymoma cells, we also observed a marked decrease in the activity of P1 promoter mutated in both κB/NFAT motifs (at –720 and –90) compared to the nonmutated promoter, and cotransfection of c-Rel was without effect. However, c-Rel exerted only a minor, if any positive effect on the induction of WT promoter by T + I or I + cAMP signals which exert a strong stimulatory effect on P1 in El4 cells 26 (Fig. 6B). Interestingly, for the mutated P1 promoter, any T + I-mediated induction was abolished (Fig. 6C) while its induction by I + cAMP signals was weaker but not abolished (Fig. 6B). We assume that in El4 cells due to their high constitutive CN activity and NFATc1 levels 27 c-Rel exerts only a minor effect on P1 activity in these cells. A similar situation seems to exist in effector lymphocytes upon induction (see Discussion).

NFAT and/or NF-κB binding to composite κB/NFAT sites

In Figure 7A, we present a compilation of composite κB/NFAT motifs from numerous promoters that are active in lymphocytes. One of those elements, the κB-like site from the Myb promoter, was used in EMSAs of Figure 2A for the analysis of NF-κB/NFAT binding in nuclear protein from chicken DT40 cells showing strong NF-κB binding upon T + I induction, and weak NF-κB binding upon ablation of NFATc1 in DT40 cells. But when human NFATc1/αA was overexpressed in DT40 cells deficient for NFATc1 (Fig. 7B) instead of NF-κB binding, we observed the predominant binding of NFATc1 to the κB/NFAT probe. This is demonstrated in lane 4 of Figure 7B in which adding an Ab specific for NFATc1 resulted in a supershift of complexes generated with the Myb κB/NFAT site. In contrast, only traces of endogenous NF-κB1 were found in this complex (Fig. 7B).

In order to see whether like the κB/NFAT sites from the HIV-LTR, the Myb site favors the formation of NFAT dimers, we performed EMSAs with bacterially expressed GST-NFATc1-RSD protein containing the DNA-binding domain of human NFATc1. In these assays, the composite Myb κB/NFAT site formed complexes very similar to those with the HIV-1 LTR κB/NFAT site, whereas the complexes formed at a similar site from the c-IAP2 promoter resembled both the HIV-1 LTR κB/NFAT and the canonical distal NFAT (+AP-1) site from the Il2 promoter (Fig. 7D). These findings suggest that both sites, the Myb and c-IAP2 sites, can be bound by NFAT homodimers, as the κB/NFAT sites in the HIV-1 LTR 10.

Mice, isolation, and culture of B cells

Eight- to 12-week-old C57 BL/6 mice were used. Splenic B cells were isolated using a Miltenyi Biotec B-cell isolation kit (mouse; #130-090-862). A total of 5 × 10⁶ cells were cultured in 1 mL X-vivo 15 medium (Lonza) and stimulated with 10 μg/mL α-IgM (F(ab‘)2 fragment of goat anti-mouse IgM; Jackson ImmunoResearch Laboratories), 5 μg/mL α-CD40 (R&D Systems), 10 μg/mL LPS (Sigma-Aldrich), or TPA (T; 10 ng/mL), and ionomycin (I; 0.5 μM). In some assays, B cells were “pulsed” (p) for 30 min (or for 2 h) at 4°C with α-IgM, washed, maintained for 6 h at 37°C and, after adding α-CD40, cultured to 24 h. In vitro-induced GC B cells were generated by cultivating splenic B cells on 40LB 3T3 feeder cells for 4 days in the presence of IL-4 25. Chicken DT40 B cells were cultured at 39.5°C with 5% CO₂ using RPMI-1640 medium supplemented with 10% heat-inactivated FCS, 1% chicken serum, 2-mercaptoethanol (50 μM), and l-glutamine (2 mM). Human 293 HEK cells were transfected using a conventional Ca⁺⁺-phosphate DNA transfection protocol with luciferase reporter genes directed by a −1200 bp WT Nfatc1 P1 promoter fragment, or by fragments mutated in the κB/NFAT motifs (Fig. 6). The cells were left untreated or treated with T + I. Forty-eight hours post transfection luciferase activity was measured. EL4 thymoma cells were transfected by DEAE-dextran and induced as described earlier 26.

The B-cell stage-specific cre lines mb1-cre 36 and Cd23-cre 37 have been described previously. The following mice lines were used: two Nfatc1 ^fl/fl mouse lines (also designated as Nfat1 ^fl/fl mice: 38-40), and the Nfκb1^−/− 21 and Rel^−/− 23 lines. Animal experiments were performed according to project licenses (55.2–2531.01–53/10B), which are approved and controlled by the Regierung von Unterfranken, Würzburg.

EMSAs

Nuclear proteins from splenic B cells or DT40 cells, or bacterially expressed GST-NFATc1 spanning the RSD (aa 418–716) were prepared and used in EMSAs as described 26. For the detection of NFAT and NF-κB binding, the distal NFAT site from the murine Il2 promoter, a NF-κB consensus site or composite κB/NFAT sites were used as probes. In supershift EMSAs, the following Abs were used: NFATc1, BD Pharmingen (#556602); NFATc2, Santa Cruz Biotechnology (#sc-13034); NF-κB1/p50 (#sc-114X); RelA/p65 (#sc-109X); c-Rel (#sc-70X); RelB (#sc-226X); NF-κB2/p52 (#sc-298X).

Immunoblotting

Immunoblots were performed by fractionating whole protein extracts on PAGE-SDS gels followed by detection of NFATc1 using the 7A6 mAb (BD Pharmingen, #556602). For detecting NFATc1/αA (Fig. 3C, lower panel), a pAb raised against the NFATc1α-peptide (ImmunoGlobe) was used. As loading control, filters were stained by Ponceau red. Signals were developed using a chemiluminescence detection system (Thermo Fisher Scientific).

Cloning of the mutated Nfatc1 P1 promoter

The −1155/+275 mouse Nfatc1 P1 promoter was amplified by the primer pair (5′-AACCAAATGTGCCCCAGAAGGAAGG-3′,5′-CAGCGGCGAGCAGCAGGAGGAGGAG-3′) and cloned into the pGL3basic vector as a NheI/XhoI fragment. Mutations were introduced by PCR using the following primer pairs (proximal κB/NFAT site: 5^′GGGGGAGGTGTTTTAAAGCTTT AAAAAGGCAG AAGG-3^′, 5'-GCCTTTTTAAAGCTTTAAAACACCTCCCCCGGCTC-3^′; distal κB/NFAT site: 5- TGGGAATTTAATTACTCCCGCCTGCCAG-3^′,5^′-CGGGAGTATTAAATTCCCATCCCG CTAAA-3^′).

RT-PCR assays

RNA was extracted from splenic B cells using TRIzol reagent (Invitrogen), and RT-PCR assays were performed with primer sequences for detecting Nfatc1 P1 promoter-directed transcripts as described 41.

Chicken NFATC1 gene targeting and expression of human NFATC1/αA

NFATC1 targeting vectors containing segments from the chicken NFATC1 locus were constructed by replacing a ∼3.3 kb genomic fragment encoding exons 4 and 5 with drug-resistance gene cassettes. The targeting vectors were introduced into WT DT40 cells, and cloning of the targeted cells was performed by culturing of cells in the presence of blasticidin, histidinol D, or puromycin as described 42. Full-length cDNA encoding human NFATC1/αA was subcloned into the pcDNA3.1 vector between its BamHI-KpnI sites by PCR using cDNA of human BALM-14 cells as template. The vector was introduced into NFATC1^−/− DT40 cells by electroporation.

RNA-Seq assays

RNA from splenic B cells was extracted using Qiagen's RNeasy kit and Illumina's RNA purification beads. RNA-Seq libraries for Next Generation Sequencing were prepared using Illumina's TruSeq RNA Sample Prep Kit V2 following manufacturer's instructions. Barcoded cDNA libraries were sequenced in one lane on an Illumina HiSeq 2000 platform for 50 nucleotides (single end).

IKKß kinase assays

DT40 cells were lysed in TNT-buffer (200 mM NaCl, 20 mM Tris pH 8.0, 1% Triton X-100, protease inhibitor cocktail (Roche), 10 mM NaF, 10 mM β-glycerophosphate, 1 mM DTT). A total of 750–1000 μg of the resulting whole cell extract were subjected to immunoprecipitation with Abs specific for the different IKK subunits (anti-NEMO (IKKγ) Ab: SC-8330, anti-IKK1 Ab: SC-7218, and anti-IKK2 Ab SC-7607, 1 μg each sample). For the kinase reaction, 0.5 μg of a bacterial expressed GST-IκBα (1–53) fusion protein and 10 μCi of ³²P-labeled γATP were added per sample followed by incubation for 30 min at 30°C under agitation and the addition of 50 μL of a 50% GSH-sepharose slurry. The samples were washed once with PBS and were subsequently subjected to a SDS-PAGE. The proteins were transferred to a nitrocellulose membrane, and phosphorylation was visualized by autoradiography.

ChIP assays

ChIPs were performed as described previously (Bhattacharyya et al. 18). Chromatin complexes were immunoprecipitated using p50 (#sc-114X) or c-Rel (#sc-70X) specific Abs, or an unrelated Ab as control. Real-time PCRs were performed with the following P1-specific primers: Nfatc1p1(1 F)atgtaaaatcgcaggcttcc and Nfatc1p1(1 R)atttagcgggatgggaattt, Nfatc1p1(2 F)gcctcgcttcacactttttc and Nfatc1p1(2 R)gacgttctaactgcccaacc.

Statistical analysis

Statistical analysis was performed by using Graphpad prism 5.0. Unpaired t-test was performed to calculate the p value. A p value less than 0.05 was considered significant.