Critical roles of the E3 ubiquitin ligase FBW7 in B-cell response and the pathogenesis of experimental autoimmune arthritis

Mice

C57BL/6 mice (6–8 weeks) were purchased from Vital River Laboratory Animal Technology (Beijing, China). FBW7^f/f mice on a C57BL/6J background were kindly provided by Professor Ping Wang (Tongji University Medical School, Shanghai, China). The CD19-Cre and Cγ1-cre C57BL/6J mice were obtained from Jackson Laboratories. FBW7^f/f mice were crossed to CD19-Cre mice or Cγ1-Cre mice to generate FBW7-deficient mice at specific B-cell stages. The corresponding littermates (CD19-cre^+/− or Cγ1-cre^+/−) were used as controls. All mice were maintained in specific pathogen-free (SPF) conditions in the animal facility of East China Normal University (Shanghai, China), and all animal experiments conformed to the regulations drafted by the Association for Assessment and Accreditation of Laboratory Animal Care in Shanghai and were approved by the East China Normal University Center for Animal Research (M20150401). Transgenic animals were genotyped by PCR using total DNA from tail biopsies. Male and female mice were used for experiments. Primers are listed in Table S1.

Immunization

For T cell-independent antibody responses, mice aged 7–10 weeks were immunized by intraperitoneal (i.p.) injection of 50 μg NP-Ficoll (Biosearch Technologies, Novato, CA, USA) in PBS. For T cell-dependent antibody responses, mice aged 7–10 weeks were immunized by intraperitoneal (i.p.) injection of 100 μg alum-precipitated NP-CGG (Biosearch Technologies) and analysed 12–13 days later in the primary immunization. Two months later, mice were boosted with 20 μg soluble NP-CGG in PBS (i.p. injected) for secondary immunization.

Flow cytometry analysis

To prepare splenocytes, spleens were taken and cut in half with a razor blade. Typically, one half was used for frozen sections, and the other for enzyme-linked immunospot assay (ELISPOT) and flow cytometry (FCM) analysis. In some experiments, the whole spleen was used for FCM analysis or cell sorting. Single cell suspensions were prepared by passing the spleen through a 70-μM cell strainer (Falcon, Corning, NY, USA). After 5 min of 400 g centrifugation, erythrocytes were depleted by using red cell lysis solution (buffer containing 155 mM NH4Cl, 10 mM KHCO3, 1 mM EDTA). Then cells were washed and suspended in magnetic-activated cell sorting (MACS) buffer (PBS/2% BSA/200 mM EDTA). MACS buffer was used for washing cells between each staining step. Cells were stained with the indicated antibodies for 30 min on ice after blocking FcγR-mediated non-specific binding by incubation with anti-mouse CD16/CD32, according to the manufacturers' suggested protocols. Like splenocytes, bone marrow cells and peritoneal wash cells were stained with antibodies after removal of red blood cells. The flow cytometry staining procedure for cells cultured in vitro was the same as that for splenocytes except for the removal of erythrocytes. Cells were collected using a FACScalibur or BD LSRFortessa™ (BD Bioscience, San Diego, CA, USA) flow cytometer, and data were analysed using FlowJo software V10 (Tree Star Inc.). All antibodies are listed in Table S2.

Cell isolation and sorting

For mouse B-cell subsets: MZ B cells (B220⁺CD23^low/−CD21^high), FO B cells (B220⁺CD23⁺CD21⁻) and naive/mature B cells (B220⁺IgM^lowIgD⁺CD23⁺) were isolated from spleens of unimmunized C57BL/6 mice by fluorescence-activated cell sorting (FACS). Thirteen days after NP-CGG immunization, memory B cells (B220⁺CD138⁻CD38⁺IgG1⁺) and GC B (B220⁺CD138⁻CD38⁻IgG1⁺) cells in the spleens of C57BL/6 mice were sorted using a FACS Aria II. The purity of the sorted cells was confirmed by FCM analysis (>95%). All samples were subjected to real-time PCR. Generally, single-cell suspensions were prepared as described above and then labelled with antibodies. For Human B subsets: Human tonsils were obtained after routine tonsillectomy at the Sixth People's Hospital of Shanghai. Informed consent was obtained from the patients or there was an exemption from informed consent because residual material after diagnosis was used. All samples were fully anonymized. Tissue collection was approved by the Ethical and Scientific Committees of East China Normal University. Tissues were placed in cold phosphate-buffered saline (PBS) immediately after surgery, as previously described [40]. They were cut into pieces with surgical scissors before mechanical homogenization with a syringe plunger. The cell suspension was then passed through a 70-μm sterile cell strainer to remove debris and aggregated material. Mononuclear cells were isolated using Lymphoprep™ (Stemcell, Vancouver, Canada) density gradient centrifugation at 400 g for 30 min. To shorten the fluorescence-activated cell sorting time, enriched B cells were first obtained by magnetic depletion of non-B cells (T cells, NK cells, monocytes, dendritic cells, etc.), which were labelled with a cocktail of biotinylated CD2, CD14, CD16, CD3 and CD235a (glycophorin A) antibodies and subsequently with streptavidin-conjugated magnetic beads (Miltenyi Biotec, Germany). After consecutive washes with MACS buffer, the cell suspension was passed through a 40-μM sterile cell strainer to remove aggregates and then stained for multiparametric FACS. The B-cell populations were phenotypically defined as naive B cells (CD19⁺IgD⁺CD38⁻), pre-GC B cells (CD19⁺IgD⁺CD38⁺), GC B cells (CD19⁺IgD⁻CD38⁺CD44⁻), memory B cells (CD19⁺IgD⁻CD38⁻CD44⁺), centroblasts (CD19⁺IgD⁻CD38⁺CD44⁻CXCR4⁺) or centrocytes (CD19⁺IgD⁻CD38⁺CD44⁻CXCR4⁻) and were further sorted using a FACS Aria II (BD Bioscience) to >95% purity. Afterwards, cells were collected into the appropriate medium and kept on ice for subsequent experiments such as real-time PCR or Western blotting. Representative cell sorting strategies are shown in Figure S1. Antibody information is listed in Table S2.

Western blotting

Cells were harvested and lysed for 25 min in 1*RIPA buffer (CST, BSN, USA) containing a protease and phosphatase inhibitor cocktail. The protein concentration of samples was determined using BCA protein assay reagent kits (Beyotime, Shanghai, China) after which 25–50 μg protein was separated by standard SDS-PAGE on 12% or 15% gels and then transferred to PVDF membranes (Bio-Rad, CA, USA). These were subsequently blotted with specific primary antibodies followed by horseradish peroxidase (HRP)-conjugated secondary antibodies and visualized by Super Signal West Pico stable peroxide solution (Thermo Scientific). The antibodies used in this study are listed in Table S2.

Real-time PCR

Total RNA was extracted from the isolated B-cell subsets using RNA extraction kits (Gene Mark biolab, Taiwan, China), and reverse transcription was performed with ReverTra Ace (Toyobo, Osaka, Japan) according to the manufacturer's instructions. To examine the mRNA expression status of the designated genes, cDNA fragments were amplified by Hieff® qPCR SYBR Green Master Mix (YESAN, Shanghai, China). Assays were performed using ABI PRISM 7000 (Applied Biosystems, USA). All reactions were performed in duplicate. The mRNA level of each gene was first normalized to the expression level of β-actin by the 2^−ΔΔCt method. To facilitate comparison of the B-cell subsets, naive B cells were used as the control group, against which the data of the other groups were normalized. All data are reported as relative mRNA expression levels. All primer sequences used in this study are listed in Table S1.

In vitro B-cell stimulation

Naive B cells were sorted from unimmunized mouse spleens by magnetic depletion of labelled cells. Cells were labelled with a cocktail of biotin-conjugated antibodies against CD43 (Ly-48), CD4 (L3T4) and Ter-119, followed by streptavidin-coupled magnetic beads (Miltenyi Biotec). Untouched naive B cells (B220⁺% >95%) were adjusted to a density of 1 × 10⁶/ml and cultured in complete RPMI 1640 medium supplemented with 10% fetal bovine serum (Millipore, Billerica, MA, USA), 100 U/ml penicillin (Gibco, Carlsbad, CA, USA), 100 mg/ml streptomycin (Gibco), 2 mM L-glutamine (Gibco), 1 mM sodium pyruvate (Gibco) and non-essential amino acids solution (Gibco), 10 mm HEPES and 0·05 mM β-mercaptoethanol (Sigma) at 37°C in a humidified atmosphere containing 5% CO₂. For in vitro class-switching assays, cells were stimulated by 20 μg/ml LPS (Sigma) plus 25 ng/ml IL-4 (R&D) for the specific time described in the figure legends. Finally, cells were collected for FCM analysis or Western blotting. For B-cell proliferation assessment, purified naive B cells were first labelled by 2·5 μM 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) (Thermo Fisher Scientific, Ashville NC, USA) for 10 min at 37°C. After washing, the labelled cells were cultured under the specific conditions described above. Eventually, cells were collected and diluted fluorescence peaks of CFSE were analysed by flow cytometry using an Attune NxT Flow Cytometer (Thermo Scientific, USA). Viability of B cells was also measured by flow cytometry using a Zombie Fixable Viability Kit, gating living cells by Zombie-negativity.

ELISA and ELISPOT

Enzyme-linked immunosorbent assays (ELISA) for NP-specific Ig serum concentrations and ELISPOT for NP-specific antibody secreting cells (ASCs) were carried out as described [42, 43]. Blood samples were collected from the eye vein at different time-points, kept at room temperature (RT) for 2 h and centrifuged at 2900×g for 10 min, and repeated once. The supernatants were harvested and stored at −80°C before the ELISA. Briefly, 96-well plates (Corning, NY, USA) were coated with 50 μg/ml NP₅-BSA (for high-affinity NP-specific Ig titres) (Biosearch Technologies) or NP₂₃-BSA (for total NP-specific Ig titres) (Biosearch Technologies) in 0·1 M carbonate buffer at 4°C overnight. The plates were then blocked with PBS containing 10% (v/v) bovine serum for 2 h at 37°C. Dilutions of test serum collected at the indicated time-points were added for 2 h at 37°C, followed by detection using anti-IgM or anti-IgG1 coupled to HRP (Southern Biotechnology, Birmingham, AL, USA). 3,3′, 5,5′-tetramethylbenzidine TMB (Abcam, Cambridge, UK) was added subsequently, and the absorbance was measured at 450 nm after stopping the reaction with 1 m H₂SO₄ (Abcam). Each dilution of serum was duplicated. A mixture of sera from naïve C57BL/6J mice was used as a negative control in each plate, and antibody levels were shown as relative titres.

For ELISPOT assays: PVDF membranes were coated with 50 μg/ml NP₅-BSA or NP₂₃-BSA in PBS overnight at 4°C, and then blocked with PBS containing 10% bovine serum at 4°C overnight. Splenocytes (5 × 10⁵ cells/well) or BM cells (10⁶ cells/well) were added to 96-well plates and incubated on the PVDF membrane at 37°C, 5% CO2 for 5 h. After incubation, PVDF membranes were washed with running ultra-pure water to lyse cells. They were then washed once with PBS containing 50 mM EDTA, followed by PBS containing 0·1% Tween-20 twice and PBS once again. Finally, membranes were double-stained with alkaline phosphatase (AP)-conjugated anti-mouse IgM and HRP-conjugated anti-mouse IgG1 antibodies (Southern Biotechnology). HRP and AP activities were visualized using 3-aminoethyl carbazole (Sigma, St. Louis, USA) and naphthol AS-MX phosphate/Fast Blue BB (Sigma) respectively. Colour development was stopped by thoroughly rinsing both sides of the PVDF membrane with demineralized water. The membranes were dried in the dark at RT. The spots were counted using a dissecting microscope (Motic# SMZ-161-BLED, Xiamen, China). All samples were run in 3–5 replicates.

Collagen-induced arthritis

To induce arthritis, 9- to 11-week-old, FBW7^f/f-Cγ1^+/cre and Cγ1^+/cre mice were injected intradermally at the base of the tail with 100 µg (in 100 µL) chick collagen type II (Chondrex, Redmond, WA, USA), which had been dissolved overnight at 4°C in 0·01 M acetic acid (Sigma) and emulsified in an equal volume of Complete Freund's Adjuvant (CFA, Chondrex). CFA was prepared by grinding 100 mg heat-killed Mycobacterium tuberculosis (H37Ra; Difco Laboratories) in 20 ml Incomplete Freund's Adjuvant (IFA, Chondrex). Three weeks after primary immunization, a booster immunization with 100 μg of chick collagen type II (CII) emulsified in IFA was given at the base of the tail [25, 26, 41]. Mice were observed for the onset of arthritis and evaluated for the duration of onset and severity of inflammation, with a score of 0–4 for each paw, and a combined clinical score was determined as described by Brand et al [25]. Thirty-six days after the booster immunization with CII, the mice were killed following retro-orbital bleeding. Mice were scored three times per week for 8 weeks. Serum levels of anti-mouse CII antibody (Chondrex) were determined by ELISA.

Measurement of collagen type II-specific antibodies

ELISA plates (Corning#9018, NY, USA) were coated with 10 μg/mL of chick type II collagen dissolved in Tris buffer (0·05 M Tris, containing 0·2 M NaCl, pH 7·4), and blocked with 2% bovine serum albumin for 2 h. Thereafter, sera from CIA mice were loaded onto these type II collagen-coated 96-well plates and incubated for 2 h at RT. Pooled sera from arthritic mice served as a positive control and reference standard. A standard curve was created for each assay by including serial dilutions of the reference sample on each plate. The reference sample was arbitrarily assigned an antibody concentration of 2 AU/ml. Serum levels of IgG1, IgG2a, IgG2b or IgG2c antibodies against type II collagen were measured by incubation with HRP-conjugated goat anti-mouse isotype-specific (IgG1, IgG2a, IgG2b or IgG2c; Southern Biotechnology) detection Abs, followed by TMB substrate. Optical density was measured at 450 nm. The antibody concentrations for each serum sample were obtained by reference to the standard curve.

Polymerase chain reaction (PCR) amplification and sequencing of Ig V186.2

Genomic DNA from sorted splenic GC B (B220⁺GL7⁺Fas⁺) cells of NP–CGG immunized mice was extracted and frozen at −80°C until the PCR was carried out using the QIAamp DNA Micro Kit (Qiagen, Dusseldorf, Germany). PCR amplification and sequencing of variable heavy-chain region segments were carried out as previously described [42, 43]. Two sequential rounds of PCR were performed using high-fidelity PrimeSTAR DNA polymerase (Takara Bio, CA, USA). The two pairs of primers used are listed in Table S1. Primers corresponding to the V186.2 genomic DNA 5′ transcription start site and the intron JH2 sequence, respectively, were used for the first round of PCR (3 min 98℃ preincubation; 25 cycles of 98℃ 10 s, 55℃ 5 s, 72℃ 5 s; and 1 min at 72℃). 2 μl of this reaction mixture was reamplified for an additional 25 cycles (3 min 98°C preincubation; 25 cycles of 98°C 10 s, 51°C 5 s, 72°C 5 s; and 1 min at 72°C) using the second set of nested primers. The products were then visualized on 1·5% agarose gels, and a 300–400 bp band was detected and purified. After enzyme digestion and purification, the PCR products were cloned into the pcDNA3.1 vector. The recombinant plasmids were amplified by DH5α strain, extracted and sequenced. The chromatograms were identified by DNAStar software, and only high-quality sequences were considered Afterwards, sequences were blasted in the IgBlast database (https://www.ncbi.nlm.nih.gov/igblast/), which is linked to the ImMunoGeneTics V-QUEST online tool (IMGT).

Statistical analysis

Data on linear axes are depicted as mean ± standard deviation (S.D.) or mean ± standard error of the mean (S.E.M). The statistical tests applied are stated in the legend of each figure. Statistical analysis was performed using GraphPad Prism software (GraphPad, San Diego, CA, USA). Generally, for comparisons of two groups, parametric data were analysed using two-tailed, a two-tailed, unpaired Student's t-test distribution with a confidence interval of 95%. Significant p-values are indicated in the figures by *p < 0·05; **p < 0·01; and ***p < 0·001.

FBW7 expression in various B-cell subsets

To investigate the potential intrinsic role of FBW7 in B cells, we characterized different B-cell subpopulations isolated from human tonsil and mouse spleens by FACS. The gating strategies for these cells are shown in Figure S1a,c. FBW7 expression in the different human B-cell subsets at the protein level is shown in Figure 1a and Figure S1b and at the RNA level in mice by real-time PCR in Figure 1b. Our results indicate that the level of FBW7 protein expression is significantly greater in human GC B cells relative to naïve B cells (Figure 1a,b). Western blotting further confirmed that FBW7 was expressed in both human centroblasts and centrocytes (Figure S1b). Consistent with these results, the mRNA expression pattern of FBW7 in mouse B-cell subsets established by searching the public database ImmGen (www.immgen.org) also indicates that FBW7 is expressed in most B-cell subsets and strongly expressed in GC B cells (Figure 1c,d). In addition, it may be seen that levels of FBW7 mRNA in B1 cells of the peritoneal cavity (Figure 1c) and plasma cells (Figure 1d) are higher than in other B-cell subsets.

The role of FBW 7 in B-cell homeostasis and the TI antibody response

FBW7 is expressed in B cells over the course of their development (Figure 1c,d). To investigate its role in B-cell development and function, we crossed FBW7 mice carrying the loxp flox alleles (FBW7^f/f) to CD19-Cre transgenic mice, in which the earliest deletion of loxP-flanked regions starts at the pro-B-cell stage and continues throughout B-cell development [44]. First, we analysed B-cell development in the bone marrow (BM), as defined by expression of B220, IgM or CD43. Although there were no significant changes in the overall percentage (Figure S2a) and the number (data not shown) of BM B cells, the percentages of the mature recirculating B cells and mature B cells were significantly reduced in FBW7^f/f-CD19^+/cre mice relative to WT mice (Figure S2a). Correspondingly, peripheral B220⁺B cells in spleens did not decrease markedly in FBW7^f/f-CD19^+/cre mice (Figure S2b), but the number of B cells decreased sharply (Figure 2a). Likewise, FBW7^f/f-CD19^+/cre mice had lower proportions of mature B cells in the spleen (Figure S2b). Staining splenic B cells for CD21 and CD23 revealed significantly fewer marginal zone (MZ) B and follicular (FO) B cells (Figure 2a). It was also found that the density of surface expression of CD23 (MFI) was lower in B cells (Figure S2c). Furthermore, the proportions and numbers of innate-like B1a and B1b cells were both reduced in the peritoneal cavity, suggesting that FBW7 might be necessary for the maintenance of B1 cells (Figure 2b). Similarly, the number of B2 cells in the peritoneum of FBW7^f/f-CD19^+/cre mice was drastically reduced as well (Figure 2b). Consistent with this, FBW7^f/f-CD19^+/cre mice showed a decreased percentage of B1 cells in spleens, especially B1b cells (Figure S2d).

The subsequent effects of FBW7 deficiency in B cells, which as shown above resulted in a reduction in the number and proportions of functional B cells, were reflected in their functional immune responses. To analyse the impact of FBW7 deficiency on antibody responses, we immunized FBW7^f/f-CD19^+/cre and control mice with the T cell-independent antigen (Ag) NP-Ficoll, which induces immune responses dominated by Igλ⁺ NP-specific antibodies [45-49]. Both B1 and MZ B cells are considered to be major contributors to the TI-Ag antibody response. As expected, our results showed that the frequencies of IgM⁺ B cells and λ⁺ NP-specific B cells were both decreased in the spleens of FBW7^f/f-CD19^+/cre mice 7 days after immunization (Figure 2c,d). NP-Ficoll elicits robust IgM and IgG3 antibody production in a T cell-independent fashion [50]. Strikingly, NP-Ficoll immunization of FBW7^f/f-CD19^+/cre mice produced very low titres of IgG3 antibody compared to wild type, especially at day 14 after immunization (Figure 2e, right panel). We also observed lower IgM levels on day 7 after immunization but not thereafter (Figure 2e left panel). Taken together, these data show that humoral immune responses to the T-independent antigen NP-Ficoll were weakened in FBW7^f/f-CD19^+/cre mice, particularly the IgG3 response.

BCR signalling is important for B-cell homeostasis by delivering survival signals to the cells. The decrease of peripheral mature B cells and peritoneal B1 cells in FBW7^f/f-CD19^+/cre mice suggested that BCR signalling might be impaired. Thus, we purified splenic naïve B cells from FBW7^f/f-CD19^+/cre and control mice and labelled them with the division tracking dye CFSE. They were then stimulated with anti-IgM or cultured without stimulation. The survival rate of unstimulated B cells lacking FBW7 and WT B cells was similar (Figure S3a,b). However, under stimulation with anti-IgM, the number (Figure S3c) and viability (Figure S3d) of FBW7-deficient B cells decreased significantly with increasing culture duration. In the unstimulated state, the size of FBW7-deficient and control B cells was similar, but after anti-IgM stimulation, the size of the surviving viable FBW7-deficient B cells was smaller than the control group (Figure S3e). This further indicates that the absence of FBW7 affected the activation and growth of cells stimulated by anti-IgM. Consistently, the proliferation of FBW7-deficient B cells was impaired on anti-IgM stimulation (Figure S3f).

Taken together, these results confirm that FBW7 contributes to the maintenance of B-cell homeostasis, particularly regarding numbers of normal mature B cells and B1 cells. In vivo, FBW7 deficiency in mature B cells and B1 cells results in an impaired T cell-independent antibody response. Moreover, the results show that FBW7 is important for BCR-mediated B-cell proliferation and survival.

The role of FBW7 in the GC reaction and TD primary antibody responses

To elucidate the role of FBW7 in regulating GC formation and function, conditional knockout FBW7^f/f-Cγ1^+/cre mice were generated by crossing FBW7^f/f mice with Cγ1-cre mice [51]. Expression of Cre recombinase is induced by germline Cγ1 transcription after T cell-dependent antigen immunization, resulting in the Cre-mediated deletion of loxP-flanked genes in the majority of GC B cells [51]. B-cell development in the BM, spleen and the serum Ig isotypes appeared to be normal in naïve FBW7^f/f-Cγ1^+/cre mice (data not shown), in which any possible effects of FBW7 loss during early B-cell development could be circumvented. After immunization with the well-defined T cell-dependent Ag (4-hydroxy-3-nitrophenyl) acetyl (NP) conjugated with chicken γ-globulin (NP-CGG), the GC responses of FBW7^f/f-Cγ1^+/cre and Cγ1^+/cre mice were analysed on day 13 post-immunization. By flow cytometric analyses, we found that the proportion of splenic FBW7-null GC B cells and WT controls was similar (Figure 3a,b). Consistent with this, immunohistological analysis indicated no significant difference in the amount and size of GCs in spleens between FBW7^f/f-Cγ1^{+ /cre} and control mice (Figure S4a). T follicular helper (Tfh) cells provide an important signal for GC B survival and antibody production. Our results showed that the proportion of Tfh cells in GC was not significantly different in FBW7^f/f-Cγ1^{+ /cre} and WT mice (Figure S4b). To further investigate the impact of FBW7 deletion on GC B cells, centrocytes (CC) and centroblasts (CB) were distinguished by the expression of CXCR4 and CD86. We found that the ratio of CB to CC was increased by FBW7-deficiency (Figure 3c), suggesting that GC function may be compromised. Moreover, the levels of IgM⁻GC B cells and IgG1⁺GC B cells expressing class-switched BCRs were significantly reduced by FBW7 deficiency (Figure 3d,e and Figure S4c,d). These data indicate that Ig class-switching is impaired in FBW7-deficient GC B cells.

In vitro stimulation of naïve B cells with LPS plus IL-4 can be used to induce CSR. We found that 4 days after such stimulation, B cells derived from FBW7^f/f-CD19^{+ /cre} mice exhibited impaired CSR and significant reduction of IgG1⁺IgM⁻ class switched B cells relative to controls (Figure 3f and Figure S4e). However, FBW7-deficient B cells retained normal proliferative capacity and cell viability when stimulated by LPS plus IL-4, indicating that the CSR defect in FBW7^f/f-Cγ1^+/cre B cells does not result from a defect in their proliferative capacity (Figure S4f,g).

Primary antibody responses after 13 days of NP-CGG immunization include the production of high-affinity and total (including high- and low-affinity) anti-NP IgG1 and anti-NP IgM. The levels of antibody and antibody-secreting cells (ASCs) were tested using two antigens with different ratios of NP/BSA (NP5-BSA and NP23-BSA). ELISA experiments were performed to measure titres of NP-binding antibodies in serum. Whereas NP-specific IgM titres were only marginally affected in the FBW7^f/f-Cγ1^+/cre mice, the total (NP23-binding) and high-affinity (NP5-binding) NP-specific class-switched IgG1 antibodies were markedly impaired by FBW7-deficiency (Figure 4a). Likewise, FBW7-deficiency impaired the generation of class-switched total and high-affinity NP-specific ASCs (Figure 4b). In addition, we found that the ratio of NP5/NP23 IgG1 titres in FBW7^f/f-Cγ1^+/cre mice was significantly reduced compared to the control group, indicating impaired affinity maturation of antibody in these mice (Figure 4c). These data document that FBW7^f/f-Cγ1^{+ /cre} mice exhibit an impaired antibody response, particularly in their ability to produce high affinity (NP5-binding) class-switched (IgG1) antibodies (Figure 4a–c). Taken together, these results demonstrate that FBW 7 ablation in B cells at the GC stage has no significant impact on GC formation, but does affect two critical GC functions, namely CSR and antibody affinity maturation, which were severely impaired.

Another major characteristic of GC B cells is the somatic hypermutation (SHM) of Ig genes, which provides the substrate for affinity-driven clonal selection, leading to affinity maturation of antibodies. On day 13 after immunization with NP-CGG, the typical gene fragment V_H186·2 encoding high-affinity NP-specific antibodies dominates the BCR repertoire [52, 53]. To determine the role of FBW7 in regulating Ig gene SHM, the BCR mutation frequency in NP-CGG–immunized FBW7^f/f-Cγ1^+/cre and control mice were examined by genomic DNA sequencing. GC B cells were sorted for sequencing of VDJ segments of the V_HJ558 family of Ig H-chain genes. Our results showed that there was no significant clonal type shift of IgH-chain genes between wild-type GC B cells and FBW7-null GC B cells (Figure 5a). Ig mutation rates in VH186.2 clones and analogous clones exhibited no significant differences in GC B cells from FBW7 conditional knockout or WT mice (Table 1 and Figure 5b,c). Therefore, our data demonstrate that the SHM machinery is normal in FBW7^f/f-Cγ1^+/cre mice. Notably, in line with the reduced high-affinity NP-specific antibody titre and numbers of ASCs in immunized FBW7^f/f-Cγ1^+/cre mice (Figure 4a–c), we found a greatly reduced fraction of FBW7-deficient GC B-cell clones carrying the W33L mutation (Figure 4d). It has been reported that clones carrying this mutation can produce antibodies with an approximately 10-fold increased affinity for the NP antigen [54]. Taken together, all these results indicate that FBW7 is critical for affinity-based selection.

Cγ1 cre-mediated FBW7 gene deletions are mainly present in the GC B cells that usually account for 2–10% of total B cells. To further investigate the underlying molecular mechanisms responsible for the defective GC function in conditional FBW7-deficient mice, including CSR and affinity maturation, we performed a mechanistic study using B cells derived from FBW7^f/f-CD19^cre+ mice, in which all the B cells lack FBW7 expression. To this end, naive B cells were stimulated by LPS plus IL-4 in vitro for 4 days. We then measured activation-induced cytidine deaminase (AID) protein levels and found that they were similar in both FBW7-deficient B cells and controls. This suggests that the defective CSR in FBW7-deficient GC B cells was not the result of AID alteration (Figure S5a). The normal expression of AID, a key factor for both CSR and SHM, further explains why the SHM of BCRs can take place normally in FBW7-null GC B cells. Likewise, there were no significant differences between B cells from FBW7f/f-Cγ1^+/cre mice and controls in terms of other transcription factors essential for GC B-cell function, including interferon regulatory factor 4 (IRF4), Forkhead box O1 (FOXO1) and paired box 5 (PAX5) (Figure S5a). The transcriptional repressor B-cell lymphoma 6 protein (BCL6) is essential for the initiation and maintenance of the GC reaction. Mice lacking BCL6 cannot form GCs nor produce high-affinity antibodies. Multiple signalling pathways contribute to BCL6 downregulation. For these reasons, we also measured BCL6 expression at an earlier time-point and found that its expression was indeed down-regulated in FBW7-null B cells (Figure S5b).

Collectively, we conclude that deletion of FBW7 in GC B cells severely impairs GC function, leading to suppressed CSR and diminished antibody affinity maturation in primary antibody responses.

The role of FBW7 in a secondary antibody response

Another critical function of the GC reaction is the generation of memory B cells and long-lived antibody-secreting cells. To further study the role of FBW7 in the functional maturation of GC reaction, the memory response to secondary immunization with NP-CGG was evaluated. Following secondary immunization, sera from FBW7^f/f-Cγ1^+/cre mice and control mice were collected at different times and NP-specific antibodies were measured by ELISA. We found that the titres of NP-specific class-switched (IgG1) antibodies were lower in FBW7^f/f-Cγ1^+/cre mice than controls (Figure 6a). The deficiency in the secondary antibody response was particularly profound when NP-specific class-switched (IgG1) ASCs were measured in the spleen and bone marrow. Consistent with the data on NP-specific antibodies, the numbers of ASCs were significantly reduced in FBW7^f/f-Cγ1^+/cre mice, particularly the high-affinity (NP5-binding) ASCs (Figure 6b,c). In addition, the proportion of memory B cells with the B220⁺CD138⁻CD38⁺IgG1⁺ phenotype in the BM was significantly lower in FBW7^f/f-Cγ1^+/cre mice than WT control mice 4 weeks after primary immunization (Figure 6d). The frequency of this cell population in the spleen was also lower than in the control mice, although not to a significant extent. Thus, the deletion of FBW7 in GC B cells impairs GC functional maturation, leading to a diminished memory antibody response, and decreased long-lived ASCs.

FBW7 deficiency in GC B cells ameliorates the development and pathogenesis of CIA

To explore whether FBW7 is essential for the production of pathological antibodies that are indispensable for the pathogenesis of an autoimmune disorder, we employed the model of collagen-induced arthritis (CIA) in FBW7^f/f-Cγ1^+/cre mice. Our data show that FBW7-deficient mice exhibited a slower disease induction phase than the WT controls. Disease onset was delayed and disease severity significantly reduced in the FBW7-deficient mice (Figure 7a). In addition, disease incidence was also markedly lower in the FBW7^f/f-Cγ1^+/cre mice (Figure 7b). Joint destruction was significantly milder in FBW7 conditional knockout mice according to X-ray analysis of the inflamed paws during the onset of CIA (day 10 after the booster immunization with collagen) (Figure 7c). Consistent with this, levels of anti-CII autoantibodies in the sera of FBW7^f/f-Cγ1^+/cre mice were lower than WT controls, especially the IgG1 and IgG2a isotypes (Figure 7d). Thus, defects in antibody production in FBW7^f/f-Cγ1^+/cre mice ameliorate CIA, suggesting that FBW7 acts as a positive regulator of B-cell function in the CIA model.