EZH2 enables germinal centre formation through epigenetic silencing of CDKN1A and an Rb-E2F1 feedback loop

Abstract

The EZH2 histone methyltransferase is required for B cells to form germinal centers (GC). Here we show that EZH2 mediates GC formation through repression of cyclin-dependent kinase inhibitor CDKN1A (p21^Cip1). Deletion of Cdkn1a rescues the GC reaction in Ezh2 ^-/- mice. Using a 3D B cell follicular organoid system that mimics the GC reaction, we show that depletion of EZH2 suppresses G1 to S phase transition of GC B cells in a Cdkn1a-dependent manner. GC B cells of Cdkn1a ^-/- Ezh2 ^-/- mice have high levels of phospho-Rb, indicating that loss of Cdkn1a enables progression of cell cycle. Moreover, the transcription factor E2F1 induces EZH2 during the GC reaction. E2f1 ^-/- mice manifest impaired GC responses, which is rescued by restoring EZH2 expression, thus defining a positive feedback loop in which EZH2 controls GC B cell proliferation by suppressing CDKN1A, enabling cell cycle progression with a concomitant phosphorylation of Rb and release of E2F1.The histone methyltransferase EZH2 silences genes by generating H3K27me3 marks. Here the authors use a 3D GC organoid and show EZH2 mediates germinal centre (GC) formation through epigenetic silencing of CDKN1A and release of cell cycle checkpoints.

Fig. 1

In vivo depletion of Cdkn1a rescues GC formation in Ezh2 ^−/− mice. Ezh2 ^fl/fl, Cdkn1a ^−/−, Ezh2 ^fl/fl;Cγ1-cre and Ezh2 ^fl/fl;Cγ1-cre;Cdkn1a ^−/− littermate mice were immunized with SRBC to induce germinal center (GC) formation and were killed 10 days later. a Flow cytometry plot of one representative mouse spleen per group. The gated area shows the percentage of GC B cells (GL7 + FAS + ) within live B cells (B220 + DAPI-, see Supplementary Fig. 2A for gating strategy). b Average of GC B populations of each group of mice quantified by flow cytometry as in a (n = 7 mice per group). c Formalin fixed paraffin embedded splenic tissue was stained for PNA, Ki67, EZH2, and B220. One representative picture of three spleens analyzed per group is shown. d–f Quantification of PNA staining from c (n = 3 spleens per group). d “#GC/spleen section” is the count of all GC per spleen section. e “GC area/total spleen area” is the quantified area of each individual GC divided by the total area of the spleen section. f “Total GC area/total spleen area” is the sum of all GC quantified areas in a certain section divided by the total area of that spleen section. g Splenocytes were permeabilized and stained for GC B and EZH2 using a fluorochrome-conjugated anti EZH2 antibody. The gated area shows the percentage of GC B cells (GL7 + FAS + ) within live B cells (B220 + DAPI-, see Supplementary Fig. 2A for gating strategy) that are EZH2 positive. The flow plot shown is representative of three spleens analyzed per group. Values in b, d, e, f are shown as mean ± SEM. t test, *P < 0.05, **P < 0.01, ***P < 0.001. Results are representative of a total of four independent experiments performed with different cohorts of mice. See also Supplementary Fig. 2

Fig. 2

Cdkn1a ^−/− GC rescue is linked to the histone methyltransferase function of EZH2. Cdkn1a ^+/+ and Cdkn1a ^−/− mice were immunized with SRBC, treated daily with GSK503 (150 mg/kg/day) or vehicle and were killed 10 days after immunization. a Flow cytometry plot of one representative mouse spleen per group. The gated area shows the percentage of GC B cells (GL7 + FAS + ) within live B cells (B220 + DAPI-, see Supplementary Fig. 2A for gating strategy). b Average of GC B populations of each group of mice quantified by flow cytometry as in a (n = 5 mice per group). c Formalin fixed paraffin embedded splenic tissue was stained for PNA, Ki67, EZH2, and B220. One representative picture of three spleens analyzed per group is shown. d–f Quantification of PNA staining from c (n = 3 spleens per group, see Fig. 1d–f). g, h Splenocytes from 4 WT and 4 Cdkn1a ^−/− mice were permeabilized and stained for GC B cells (GL7 + FAS + B220⁺, see Supplementary Fig. 2A for gating strategy) and H3K27me3 g and EZH2 h using fluorochrome-conjugated anti H3K27me3 and anti EZH2 antibodies, respectively. The histograms depict the mean fluorescence intensity (MFI) ± SD of H3K27me3 and EZH2 per group of mice. Values in b, d, e, f are shown as mean ± SEM. t test, *P < 0.05, **P < 0.01, ***P < 0.001. Results are representative of a total of three independent experiments performed with different cohorts of mice. See also Supplementary Fig. 3

Fig. 3

Characterization of a 3D B cell follicular organoid to model the GC reaction. a Scheme of 3D B cell follicular organoid fabrication. Splenic B cells are co-encapsulated with 40LB stromal cells and IL4 into an RGD-presenting nanocomposite hydrogel that contain gelatin ionically cross-linked with synthetic silicate nanoparticles. b Representative fluorescence pictures of splenic GFP B cells in 3D organoid culture. c Flow cytometry plot of a 3D B cell follicular organoid. The gated area on the top shows the live B cells (B220 + DAPI−) and on the bottom plot, the organoid GC B cells (GL7 + FAS + ) within live B cells. d Average of GC B populations of organoids quantified as in c. e Average of percentage of proliferating organoid GC B populations, quantified as indicated in Supplementary Fig. 4B. f MFI of proliferation dye from e. g Flow cytometry plot of GC B cells stained with annexinV. h Average of percentage of apoptotic GC B cells quantified as in g. i RNA-seq profiles of organoid GC B cells after 4 and 6 days in culture (organoid GCB d4, n = 4 spleens, and d6, n = 3 spleens) were projected into the principal component space defined by GC B cells sorted from immunized mice (in vivo GCB, n = 6 mice), CD138 + plasma cells (in vivo PC, n = 6 mice) and FAS-GL7-IgD + B220 + naive B cells (NB, n = 3 mice). j, k Heat maps of gene expression level of GC B cells showed in i, represented as log2 ratio relative to mean naive B cells j and to mean plasma cells k. PC1/2 = principal component 1/2. l Splenocytes were permeabilized and stained for EZH2, and co-stained for GL7, FAS, IgD and B220 to identify GC B cells (FAS + GL7 + B220 + ) and naive B cells (FAS-GL7-IgD + B220 + ). m MFI of EZH2 (n = 3 per group) from l. n, o The indicated immunoglobulin variable loci were sequenced from organoid or in vivo GC B and naive B cells (n = 3 biological replicates per group). t test vs. naive B cells, *P < 0.05, **P < 0.01. Values in d–f, h, m–o are mean ± SEM. d–h, n = 3 organoids per time-point. Results shown in b–h are representative of 4 independent experiments. See also Supplementary Fig. 4

Fig. 4

CDKN1A repression by EZH2 is required for GC B cell cycle progression. a–f Organoids were generated using B cells isolated from Ezh2 ^fl/fl;Cγ1-cre, Cdkn1a ^−/−, Ezh2 ^fl/fl;Cγ1-cre;Cdkn1a ^−/− and Ezh2 ^fl/fl control mice (n = 3 mice per group) and were harvested for flow cytometry analysis after 4 days in culture. a Flow cytometry plots of representative organoids from each genotype. The gated area shows the percentage of organoid GC B cells (GL7 + FAS + ) within live B cells (B220 + DAPI−). b Average of percentage of organoid GC B populations quantified by flow cytometry as in a (n = 3 biological replicates per group). c Organoids received a BrdU pulse of 2 h before harvest. Cell cycle was analyzed by BrdU staining and 7AAD to measure DNA content. The representative gated area shows the percentage of organoid GC B cells (GL7 + FAS + B220 + DAPI−) that are in S phase (BrdU + ). d Average of percentage of organoid GC B populations in S phase of each group of genotype quantified by flow cytometry as in c (n = 3 biological replicates per group). e Organoid cells were permeabilized and stained for EZH2 and GC markers GL7, FAS, and B220 to identify organoid GC B cells. The flow cytometry plot shows one representative sample per genotype group. f MFI of EZH2 in organoid GC B cells (n = 3) quantified by flow cytometry as in e. g–m Organoids were generated using B cells isolated from 3 Cdkn1a ^+/+ and 3 Cdkn1a ^−/− mice. g Scheme of EZH2 inhibitor GSK343 and BrdU treatment ex vivo. h Organoid cells were permeabilized and stained for EZH2, H3K27me3 and GC markers GL7, FAS, and B220. The flow cytometry plot shows one representative sample per group. i MFI of H3K27me3 and EZH2 in organoid GC B cells quantified by flow cytometry as in h (n = 3 biological replicates per group). j–m Analysis of percentages of organoid GC B cells and percentages of these cells in S phase was performed, as described in a–d (n = 3 biological replicates per group). k, m show mean ± SEM. Values in b, d, f, i are mean ± SD. t test, *P < 0.05, **P < 0.01, ***P < 0.001. Results are representative of a total of three independent experiments

Fig. 5

EZH2 is required to enable Rb phosphorylation in GC B cells. a Formalin fixed paraffin embedded splenic tissue from mice shown in Fig. 1 was stained for phospho Rb Ser780 and PNA. One representative picture of three spleens analyzed per genotype group is shown. b, c Quantification of phospho Rb Ser780 staining from a (n = 3 spleens per group). b “pRb positive per GC area/total spleen area” is the quantified area of each individual GC divided by the total area of the spleen section. c “pRb positive in total GC area/total spleen area” is the sum of all GC quantified areas in a certain section divided by the total area of that spleen section. Values are shown as mean of triplicates ± SEM. t test, **P < 0.01. d Immunoblotting of whole cell lysates from naive B cells (FAS⁻ GL7⁻ IgD⁺ B220⁺) and GC B cells (FAS⁺ GL7⁺ B220⁺) sorted from 2 Cdkn1a ^−/−, 2 Ezh2 ^fl/fl;Cγ1-cre;Cdkn1a ^−/− double knockout (DKO) and 2 Ezh2 ^fl/fl control mice immunized with SRBC for 10 days. GAPDH was used as loading control. The faint EZH2 band in DKO GC B cells is due to the few residual GC B cells that were still EZH2 positive, consistent with incomplete CRE-mediated excision of Ezh2. Numbers on the left indicate molecular weight in kDa. See Supplementary Fig. 6A for uncropped scans of the western blots

Fig. 6

E2F1 induces the expression of EZH2 in GC B cells. a Expression level in FPKM of E2F genes in naive B and GC B cells from 4 human tonsils. Horizontal black lines represent mean. t test, ***P < 0.001. b RT-qPCR of E2f in GCBs (FAS + GL7 + B220 + ) and naive B cells (FAS-GL7-IgD + B220 + ) sorted from 7 WT mice. Horizontal black lines represent mean fold change mRNA levels normalized to Hprt1. t test, ***P < 0.001. c Immunoblotting of whole cell lysates from naive B and GC B cells from two human tonsil samples. Actin was used as loading control. The western blot shown is representative of a total of five tonsils analyzed. See Supplementary Fig. 6B for uncropped scans of the western blots. d Representative immunoblotting of whole cell lysates from OCI-Ly1 and OCI-Ly7 cells expressing shRNAs against E2F1, E2F2 or control. GAPDH was used as loading control. The experiment was repeated two more times with similar results. See Supplementary Fig. 6C for uncropped scans of the western blots. e Schematic representation of the genomic region corresponding to the promoter of EZH2. The black boxes represent a region of putative E2F binding sites. Two pairs of primers were design to flank the putative E2F binding sites. f E2F1 and control IgG qChIP was performed in the indicated cell lines using the primers described in e. As negative control qPCR was performed using primers for two regions in chromosomes 6 and 7 where no enrichment of transcription factors was found by ChIP-seq read density in OCI-Ly1 and OCI-Ly7 cells. As positive control qPCR was performed using primers flanking an E2F binding site at the promoter of CDK1. Fold enrichment was normalized to the input. Values are shown as mean of technical triplicates ± SEM. The experiment shown is representative of three independent ChIPs performed with OCI-Ly1 and OCI-Ly7, and two with WSU-DLCL2

Fig. 7

E2F1 is required for GC formation in an EZH2 dependent manner. a–f E2f1 ^−/− and control WT mice were immunized with SRBC to induce GC formation and were killed 10 days later. a Average of percentage of B cells (B220 + ) within live splenocytes (DAPI−) of each group of mice (n = 9 per group). b Flow cytometry plot of one representative mouse spleen per group. The gated area shows the percentage of GC B cells (GL7 + FAS + ) within live B cells (B220 + DAPI−, see Supplementary Fig. 2A for gating strategy). c Average of percentage of GC B populations of each group quantified by flow cytometry as in b (n = 9 per group). d Formalin fixed paraffin embedded splenic tissue was stained for PNA, Ki67, EZH2, phospho Rb Ser780 and B220. One representative picture of 5 spleens analyzed per group is shown. e, f Quantification of PNA staining from d (n = 5 spleens per group). e “#GC/spleen section” is the count of all GC per spleen section. f “GC area/total spleen area” is the quantified area of each individual GC divided by the total area of the spleen section. Results shown in a–f are representative of a total of three independent experiments performed with different cohorts of mice. g Bone marrow transplantation was performed using E2f1 ^−/− and WT donor mice. Each recipient group consisted of 8 mice. BM, bone marrow. h Gating strategy used to analyze the GFP positive GC B cells from transplanted mice. Representative flow cytometry plots showing the gating on GC B cells (GL7 + FAS + ) within live GFP B cells (B220 + DAPI-GFP + ) per mouse group. i Average of percentage of GFP positive GC B cells in each transplant group (n = 8) quantified by flow cytometry as shown in h. Values in a, c, e, f, i are shown as mean ± SEM. t test, **P < 0.01, ***P < 0.001. Results shown in a–f are representative of a total of four independent experiments