Inactivation of CREBBP expands the germinal center B cell compartment, down-regulates MHCII expression and promotes DLBCL growth

Significance

Genes encoding chromatin-modifying enzymes such as the histone acetyl-transferases (HATs) are often mutated in diffuse large B cell lymphoma (DLBCL), the most common lymphoma of adults. Here, we shed light on the tumor suppressive activity of HATs in human diffuse large B cell lymphoma (DLBCL) cell lines and in mice. Cell lines harboring an experimentally introduced patient mutation in the HAT CREBBP lose their MHCII expression and form tumors faster in subcutaneous and orthotopic xenograft models. Mice that lack Crebbp specifically in the germinal center B cell compartment also lose their MHCII expression in that compartment, and show hyperproliferation of germinal center B cells upon immunization, which predisposes them to MYC-driven lymphomagenesis. Our data implicate HATs as tumor suppressors in DLBCL.

Abstract

The genes encoding the histone acetyl-transferases (HATs) CREB binding protein (CREBBP) and EP300 are recurrently mutated in the activated B cell-like and germinal center (GC) B cell-like subtypes of diffuse large B cell lymphoma (DLBCL). Here, we introduced a patient mutation into a human DLBCL cell line using CRISPR and deleted Crebbp and Ep300 in the GC B cell compartment of mice. CREBBP-mutant DLBCL clones exhibited reduced histone H3 acetylation, expressed significantly less MHCII, and grew faster than wild-type clones in s.c. and orthotopic xenograft models. Mice lacking Crebbp in GC B cells exhibited hyperproliferation of their GC compartment upon immunization, had reduced MHCII surface expression on GC cells, and developed accelerated MYC-driven lymphomas. Ep300 inactivation reproduced some, but not all, consequences of Crebbp inactivation. MHCII deficiency phenocopied the effects of CREBBP loss in spontaneous and serial transplantation models of MYC-driven lymphomagenesis, supporting the idea that the mutational inactivation of CREBBP promotes immune evasion. Indeed, the depletion of CD4⁺ T cells greatly facilitated the engraftment of lymphoma cells in serial transplantation models. In summary, we provide evidence that both HATs are bona fide tumor suppressors that control MHCII expression and promote tumor immune control; mutational inactivation of CREBBP, but not of EP300, has additional cell-intrinsic engraftment and growth-promoting effects.

Perturbations of the epigenome due to mutations occurring in histone-modifying enzymes are emerging as a driving force in the pathogenesis of diffuse large B cell lymphoma (DLBCL) (1). The two main cell-of-origin subtypes of DLBCL, the activated B cell (ABC) and germinal center (GC) B cell-like (GCB) subtype, are both commonly affected by mutations in epigenetic modifiers (2). The most common recurrent somatic mutations in histone-modifying enzymes are loss-of-function mutations of the histone methyltransferase (HMT) KMT2D (also known as MLL2), gain-of-function mutations of the HMT EZH2, and loss-of-function mutations of the histone acetyltransferases CREB binding protein (CREBBP) and EP300. Together, these somatic mutations occur in ∼50% of DLBCL patients (2), and are believed to be founder events that predispose normal B cells to malignant transformation (1). DLBCL is characterized by extraordinary genetic diversity, both within and across tumors, which has been linked largely to the activity of activation-induced cytidine deaminase (AID), the enzyme responsible for introducing mutations in the Ig locus and, thereby, enabling the formation of high-affinity antibodies, and its effects on nonimmunoglobulin genes (3). More recently however, epigenetic heterogeneity has emerged as an equally important hallmark of DLBCL, with strong evidence for diversity in both DNA modifications (especially cytosine methylation) and the posttranslational modifications of histones. A higher degree of intertumoral and intratumoral heterogeneity has been associated directly with disease aggressiveness, likelihood of relapse, and inferior patient survival (4–6).

The contribution of mutations in histone-modifying enzymes to DLBCL initiation and progression is subject to intense investigation. Loss-of-function mutations in KMT2D disrupt histone H3 lysine K4 (H3K4) monomethylation and dimethylation and mostly affect gene enhancer regions, promoting the proliferation of GC B cells and preventing their terminal differentiation (7). KMT2D mutations occur in 23–32% of DLBCL patients (2, 8) and are even more common in follicular lymphoma (FL); in animal models, KMT2D loss synergizes with BCL2 to accelerate lymphomagenesis (7). Lymphomas from patients with gain-of-function EZH2 mutations show aberrant repression of GC-specific proliferation checkpoint genes, and mice engineered to express mutant EZH2 exhibit a massive expansion of GC B cells due to aberrant proliferation and differentiation blockade (9). Mutations in CREBBP and EP300 affect more than 30% of DLBCL and FL patients, and usually remove or inactivate the histone acetyl-transferase (HAT) coding domain of either gene (10); CREBBP in particular has been shown to function as part of an enhancer/superenhancer network that regulates GC/post-GC cell fate decisions, plasma cell differentiation, and antigen presentation by opposing the suppressive activities of BCL6/SMRT/HDAC3 complexes (11, 12).

Here, we have investigated the mutational status of CREBBP and EP300 in a panel of 11 DLBCL cell lines relative to their H3 acetylation. CRISPR technology was used to edit the CREBBP locus in a wild-type cell line, and CREBBP-mutant and wild-type DLBCL clones were subjected to xenotransplantation studies in s.c. and orthotopic models. We further generated mice with a Crebbp deletion specifically in the GC B cell compartment and assessed the contribution of Crebbp or MHCII loss, and CD4⁺ T cell depletion, to lymphomagenesis in spontaneous and serial transplantation models driven by the overexpression of MYC. All available evidence from the various models implicates the HATs as important tumor suppressors in DLBCL pathogenesis.

Results

The CREBBP and EP300 Genomic Loci Are Recurrently Mutated in DLBCL Cell Lines, Which Affects Histone H3 Acetylation and HLA Expression.

To determine the mutational status of a panel of 11 DLBCL cell lines, we performed targeted resequencing of the 31 exons each of the CREBBP and EP300 genomic loci (Fig. 1A). Only 2 of the 11 cell lines, U-2932 and OCI-Ly3, were found to be wild type for both alleles each of CREBBP and EP300, indicating that mutations targeting these loci were even more common in our panel of cell lines (>80%; Fig. 1A) than in previously examined DLBCL patient cohorts (2, 10). The two CREBBP/EP300–wild-type cell lines U-2932 and OCI-Ly3 expressed clearly detectable amounts of full-length CREBBP and EP300 (Fig. 1B). Of the remaining nine cell lines, three exhibited monoallelic or biallelic mutations or deletions of EP300, leading to the partial (SU-DHL-10) and complete (SU-DHL-2, RC-K8) loss, respectively, of EP300 expression (Fig. 1B). The acetylation of H3 on residues K18 and 27, and to some extent on K14, was reduced in the three EP300-mutant cell lines relative to the two wild-type cell lines (Fig. 1B), whereas the acetylation level of K9 did not mirror the other three residues (Fig. S1A). The remaining six DLBCL cell lines exhibited evidence of monoallelic (5/6) or biallelic (1/6) mutational inactivation of CREBBP by either truncating mutations leading to immature stop codons, or amino acid substitutions or chromosomal translocations that detectably affect CREBBP expression levels (Fig. 1 A and B). Reduced or prematurely truncated CREBBP expression coincided with strongly reduced H3K14, H3K18, and H3K27 acetylation in all CREBBP-mutant cell lines (Fig. 1B), with the notable exception of SU-DHL-6, which maintains high-level CREBBP expression due to a trisomy of chromosome 16. Interestingly, mutations in CREBBP and EP300 were mutually exclusive in our cell line panel as had been shown in primary DLBCL samples (2, 10), and the loss of just one of the total of four CREBBP/EP300 alleles was sufficient to produce a clear phenotype in terms of H3K14, H3K18, and H3K27 acetylation (Fig. 1 A and B). In contrast, the acetylation of P53, which has been attributed to HATs and their putative activity on nonhistone targets (10), was not affected by single-copy losses of CREBBP or EP300 (Fig. S1B). ABC- and GCB-DLBCL–derived cell lines were equally affected by CREBBP/EP300 mutational inactivation.

Fig. 1.

The mutational inactivation or deletion of CREBBP and EP300 affects histone H3 acetylation and HLA expression in DLBCL cell lines. (A) Targeted resequencing of CREBBP and EP300 (all 31 exons) in the panel of 11 indicated DLBCL cell lines. Chromosomal deletions and translocations are listed as reported in the literature. (B) Protein lysates from the DLBCL cell lines shown in A were subjected to immunoblot analysis using antibodies specific for CREBBP, EP300, α-tubulin, H3K14ac, H3K18ac, H3K27ac, and total H3. Blots shown are representative of two independent experiments. Color code in A and B: blue, wild type; black, mutant CREBBP; red: mutant EP300. Asterisks indicate truncated CREBBP protein. (C) U-2932 cells were treated with control siRNA or siRNAs specific for CREBBP, EP300, or both (pool); protein lysates were harvested after 72 and 96 h and subjected to immunoblot analysis using the indicated antibodies. A representative experiment is shown along with the densitometric quantification of four independent experiments. (D) Profile of histone acetylation at the HLA-DRA locus. ChIP of U-2932 cells was performed with H3K18ac, H3K27ac and control IgG antibody and was followed by PCRs of the eight regions indicated on the x axis. (E) Relative expression of HLA-DRA and HLA-DRB1 and of WEE1 as control, normalized to RPLP32, as assessed by qRT-PCR of U-2932 cells treated with siRNAs for 48 h as described in C. Data in C and E represent means + SD of four biological replicates. **P < 0.01; ***P < 0.001; ****P < 0.0001. n.d., not detectable.

Fig. S1.

The histone acetyl-transferases CREBBP and EP300 regulate H3K14, but not H3K9 or p53 acetylation. (A and B) Protein lysates of the indicated 11 DLBCL cell lines were subjected to immunoblot analysis using antibodies for TFIIH, total p53, ac-p53, H3K9ac, and α-tubulin. Blots shown are representative of two independent experiments. (C) U-2932 cells were electroporated with control siRNA or siRNAs specific for CREBBP, EP300, or both (pool); protein lysates were harvested 48 h after transfection and used for immunoblot analysis with antibodies specific for CREBBP, EP300, and α-tubulin. (D) Densitometric quantification of the knock down efficiency of CREBBP and EP300 of all four experiments shown in Fig. 1 C and E, as shown representatively for one experiment in C. Note that there is compensatory CREBBP expression due to EP300 depletion, but not vice versa. (E and F) U-2932 cells were treated with control siRNA or siRNAs specific for CREBBP, EP300, or both (pool); protein lysates were harvested after 72 and 96 h and subjected to immunoblot analysis using an antibody for H3K14ac. A representative experiment of two is shown in E, along with its densitometric quantification in F. P values were calculated using the Mann–Whitney test. *P < 0.05; **P < 0.01; ****P < 0.0001.

To assess in a more controlled setting whether CREBBP and EP300 expression affects H3 acetylation and histone modification-dependent target gene expression, we used RNA interference to silence both HATs in the wild-type cell line U-2932. The specific depletion of CREBBP or EP300, or of both HATs combined, reduced H3K14, K18, and K27 acetylation (Fig. 1C and Fig. S1 C–F). A genomic locus that is believed to be regulated by histone acetylation in B cells, especially on residues modified by CREBBP/EP300, is the HLA locus. Chromatin immunoprecipitation using H3K18-acetyl–specific and H3K27-acetyl–specific antibodies followed by qPCR confirmed that the CREBBP/EP300–wild-type cell line U-2932 indeed exhibits strong acetylation of K18 and K27 at regulatory elements upstream of the HLA-DRA transcription start site, but not at downstream regions of the gene or the MyoD control gene (Fig. 1D). The loss of CREBBP and EP300 expression significantly reduced HLA transcript levels, but did not affect an irrelevant control gene (Fig. 1E). The combined results indicate that the CREBBP and EP300 genomic loci are recurrently mutated in both ABC- and GCB-DLBCL cell lines, and that their gene products contribute to histone H3 acetylation and promote active transcription in DLBCL cells at a locus known to be specifically regulated by this epigenetic mechanism.

Genome Editing of CREBBP Mimicking a Patient Mutation Reduces Histone Acetylation and HLA Expression, but Does Not Affect in Vitro Growth Kinetics.

We next devised a strategy that allowed us to mutate one allele of the CREBBP locus in the wild-type U-2932 cell line and to thereby mimic a previously identified patient mutation, at position A4171T, leading to a premature truncation of CREBBP at amino acid K1323X and the loss of a functional HAT domain. We aimed to mutate only one allele because most patient mutations in CREBBP and EP300 are monoallelic. Genome editing was accomplished by CRISPR technology using guide RNAs that were designed to promote the excision of exon 23, one of 12 exons encoding the HAT domain of CREBBP (Fig. S2A). In 5 of >100 examined clones, the excision event had resulted in a frame shift leading to a premature (monoallelic) truncation of CREBBP. The resulting protein lacks the C terminus and more than half of the HAT domain (Fig. 2A and Fig. S2B). Whereas control, unedited clones expressed copious amounts of full-length CREBBP, the mutant clones exhibited a clear reduction of full-length CREBBP and instead featured an extra band representing the truncated protein (Fig. 2A). The expression of EP300 was not affected by genomic editing of the CREBBP locus, with the exception of one clone (Fig. 2A). As expected, H3 acetylation was reduced in the clones expressing full-length CREBBP from only one allele relative to the wild-type clones (Fig. 2A). Transcriptional profiling of five mutant and five wild-type clones by RNA sequencing revealed more than 100 strongly differentially regulated genes due to partial loss of CREBBP activity (Fig. 2B and Dataset S1, GEO accession no. GSE89879). The dominant biological process as determined by gene ontology analysis was “antigen processing and presentation,” as numerous MHCII genes, and the master regulator of MHCII expression, CIITA, were among the most differentially regulated genes (Fig. 2B). Loss of MHCII expression due to CREBBP loss was confirmed by qRT-PCR for HLA-DRA and HLA-DRB1 transcripts (Fig. 2C), confirming the results obtained by siRNA-mediated knock-down of CREBBP and EP300. Other interesting CREBBP target genes (down-regulated in mutant clones) include CREBBP itself and the surface receptors CD19, CD74, S1PR1, IFNAR2, and CD52. Surprisingly, almost half of dysregulated genes were up-regulated in the mutant clones despite the fact that CREBBP adds active (rather than repressive) histone marks (Fig. 2B and Dataset S1). We next examined the growth properties of the 10 clones under various culture conditions in vitro; all clones grew equally fast in standard cell culture media, irrespective of their CREBBP status (Fig. S2 C and D), and even under exposure to the histone deacetylase (HDAC) inhibitors sodium 4-phenylbutuyrate (PBA) or valproic acid sodium salt (VPA) (Fig. S2 E and F). Similarly, we could not detect differential susceptibility to HDAC inhibitors of our CREBBP/EP300 wild-type and mutant cell lines (Fig. S2 G and H). The combined results indicate that HAT inactivation has profound effects on global histone H3 acetylation and affects the expression of numerous genes, most notably MHCII genes, but does not confer obvious growth advantages in in vitro culture systems.

Fig. S2.

CRISPR-mediated genome editing of CREBBP. (A) Schematic of the strategy used for excising exon 23 of CREBBP with CRISPR/Cas9 technology. (B) PCR confirmation of the deletion. (C and D) Viability assessment of CREBBP wild-type and mutant clones. Five wild-type and five mutant clones were seeded at a density of 0.3 × 10⁶/mL. Cell number/mL is shown in C; the viability as assessed by Cell Titer Blue assay is shown in D. (E–H) To assess the sensitivity to HDAC inhibitors, CRISPR clones (E and F) and DLBCL cell lines (G and H) were seeded at a density of 0.4 × 10⁶/mL and treated with either PBA or VPA (Sigma-Aldrich) at the indicated final concentrations; viability was assessed by Cell Titer Blue assay. Data in E–H are represented as means of two independent experiments.

Fig. 2.

Genome editing of CREBBP in U-2932 cells reduces HLA expression. (A) Protein lysates from four CREBBP^+/+ and five CREBBP^+/− clones were subjected to immunoblot analysis with antibodies specific for CREBBP, EP300, α-tubulin, and H3K18ac. The asterisk indicates the truncated CREBBP protein band, which is only detectable in the CREBBP^+/− cell lines. (B) Transcriptional profiling of five CREBBP^+/+ and five CREBBP^+/− clones was performed by RNA sequencing. The top 100 most strongly differentially regulated genes (log₂ ratio ≥0.4 and ≤−0.4) are displayed in a heat map. Genes of the HLA family, as well as CREBBP and CIITA, are annotated. (C) Relative expression of HLA-DRA and HLA-DRB1 compared with RPLP0 as assessed by qRT-PCR of the clones shown in B, plus several additional wild-type clones. **P < 0.01.

Monoallelic CREBBP Inactivation Promotes s.c. and Orthotopic Growth of Xenografted DLBCL Cell Lines in Immunocompromised Mice.

We next s.c. transplanted a subset of CREBBP-mutant and wild-type clones onto the flanks of nod/scid/common gamma chain knockout (NSG) mice, and determined the tumor growth over time and at the study end point. CREBBP-mutant clones grew faster, and had reached a larger tumor weight and volume at the study endpoint than wild-type clones (Fig. 3A and Fig. S3A). The same effect was observed in a mouse strain expressing various human cytokines that have been knocked into the respective murine loci (termed MISTRG mice due to knock-in of M-CSF, IL-3, Sirp1a, thrombopoietin, GM-CSF in the Rag2^−/−IL2Rg^−/− background; ref. 13) (Fig. 3B and Fig. S3B). We next assessed whether U2932 clones would engraft in lymphoid organs of mice upon i.v. delivery; in NSG mice, this was only the case after s.c. passaging, as had been described (14). Interestingly, s.c. passaged CREBBP^+/− clones engrafted more readily than their CREBBP^+/+ counterparts in recipient bones (Fig. S3C), indicating that CREBBP loss may confer a growth advantage in this orthotopic but passage-dependent model. Most notably, however, we observed that both CREBBP–wild-type and –mutant clones engrafted in the bone marrow of MISTRG mice even without prior passaging (Fig. 3C), with involvement of the kidneys and ovaries observed in some, but not all mice (Fig. S3D); involvement of other lymphoid organs such as the spleen was not observed with U2932 cells. CREBBP^+/− clones engrafted more readily than CREBBP^+/+ clones in the bone marrow of MISTRG mice (Fig. 3C), suggesting that the loss of CREBBP confers a growth advantage in this orthotopic model. Interestingly, CREBBP–mutant clones exhibited lower HLA-DR expression even after extended growth in vivo (Fig. 3D). hCD20 was stained for comparison, and did not differ among mutant and wild-type clones (Fig. 3D). The combined results suggest that CREBBP inactivation confers a growth advantage in vivo despite comparable in vitro growth rates, and introduce a model of DLBCL xenotransplantation that supports orthotopic growth without prior conditioning of human lymphoma cells.

Fig. 3.

Growth of xenografted U-2932 CREBBP^+/+ and CREBBP^+/− clones in NSG and MISTRG mice. (A and B) Ten million cells of two CREBBP^+/+ and two CREBBP^+/− clones were s.c. transplanted onto the flanks of NSG (A) or MISTRG (B) mice. Tumor volumes were measured with calipers every other day from day 10 until day 29 after transplantation and at the study end point. The tumor weight was determined at the study endpoint. Each symbol represents one tumor. Results from two pooled experiments are shown for each mouse strain. (C and D) Ten million cells of three CREBBP^+/+ and three CREBBP^+/− clones were i.v. transplanted into MISTRG mice; mice were killed at 4 wk after transplantation and assessed with respect to their human tumor burden in the bone marrow (C), as well as the expression of hCD20 and HLA-DR on bone marrow-infiltrating tumor B cells (D). Pooled data from two studies are shown in C, and data from one of the two studies is shown in D. P values were calculated using the Mann–Whitney test. *P < 0.05; **P < 0.01; ****P < 0.0001.

Fig. S3.

Growth of xenografted U-2932 CREBBP^+/+ and CREBBP^+/− clones in NSG and MISTRG mice. (A and B) Ten million cells of two CREBBP^+/+ and two CREBBP^+/− clones were s.c. transplanted onto the flanks of NSG (A) or MISTRG (B) mice. Tumor volumes were measured with calipers every other day from day 10 until day 29 after transplantation for each mouse. Each line represents one mouse. Results from two pooled experiments are shown for each mouse strain. (C) Six million s.c. passaged cells of two CREBBP^+/+ and two CREBBP^+/− U2932 clones were i.v. transplanted into NSG mice; mice were killed at 4 wk after transplantation and assessed with respect to their human tumor burden in the bone marrow. (D) Representative images of H&E-stained sections of a tumor-bearing kidney and a healthy kidney (both from MISTRG mice) are shown. (Scale bars: 1,000 µm.) Images were acquired on a Leica microscope. Images are stitches of a certain number of acquired adjoining images of the sample.

GC B Cell Hyperproliferation and Loss of MHCII Expression upon Deletion of Crebbp.

We next sought to establish a mouse model that would allow the specific monoallelic and biallelic deletion of Crebbp and Ep300 in the GC B cell compartment. To this end, we crossed mice expressing Cre recombinase under the activation-induced cytidine deaminase promoter (AID-Cre) with mice harboring floxed Crebbp or Ep300 alleles. The composite strains were immunized i.p. with sheep red blood cells (SRBCs) to induce AID-Cre activity and GC formation. The expected excision of floxed alleles due to AID-Cre activity in the GC compartment was verified by the parallel immunization of AID-Cre × loxP-stop-loxP-RFP reporter mice (Fig. S4 A and B). Five consecutive rounds of SRBC immunization induced GC formation in the spleen and an increase in CD19⁺CD95⁺CD38^low GC B cells (Fig. 4 A and B). Interestingly, the deletion of one or both alleles of Crebbp led to hyperproliferation of the GC compartment relative to Crebbp-proficient animals; this effect was observed for all GC B cells (Fig. 4 A and B), irrespective of which of three alternative gating strategies was used to identify them (Fig. S4C), as well as for centroblasts and centrocytes that were differentiated based on their CD86 and CXCR4 expression (Fig. 4B and Fig. S4 C and D). The plasmablast and plasmacyte pools increased due to immunization but were not visibly affected by loss of Crebbp (Fig. S4 E and F). Strikingly, quite the opposite effects were observed upon deletion of one or both alleles of Ep300: Ep300-deficient mice generated lower numbers of GC B cells, centroblasts, and centrocytes upon immunization (Fig. 4 A and B and Fig. S4D). To obtain additional complementary information on the GC compartment of the same mice, we stained spleen sections cut at various depths for Ki67 and quantified GC numbers and size; this approach revealed that the size of individual GCs, but not their multiplicity, accounts for the increase in GC B cells in Crebbp-mutant mice, and the decrease in Ep300-mutant mice (Fig. 4C and Fig. S4G).

Fig. S4.

GC B cell hyperproliferation and loss of MHCII expression upon deletion of Crebbp. (A and B) AID^cre/wt and AID^cre/wt RFP^fl/wt mice were immunized once i.p. with 200 µL of 10% sheep red blood cells and killed 10 d afterward. The percentage of GC B cells of CD19⁺ live cells were assessed by flow cytometry. GC B cells were gated as singlets/live/CD19⁺/CD95⁺CD38^low. RFP⁺ GC B cells were gated as singlets/live/CD19⁺/CD95⁺CD38^low/RFP⁺. Representative FACS plots are shown in A; the quantification of RFP⁺ GC B cells after immunization is shown for two independent experiments in B. Each symbol represents one mouse. Horizontal lines indicate means. Statistical analysis was done by Mann–Whitney test. (C) Gating strategy for GC B cells, centroblasts (CB), and centrocytes (CC). GC B cells can be gated using three different strategies: (i) singlets/live/CD19⁺/CD95⁺CD38^low, (ii) singlets/live/CD19⁺B220⁺/CD95⁺PNA^high, and (iii) singlets/live/CD19⁺/B220⁺PNA^high. Centroblasts are Cxcr4^hiCD86^low; centrocytes are Cxcr4^lowCD86^low. (D) Littermates with either no (wt), one (fl/wt), or two (fl/fl) floxed alleles of Crebbp or Ep300, which additionally express Cre under the control of the AID promoter, were immunized four times at biweekly intervals with sheep red blood cells and killed 10 d after the last injection. Nonimmunized mice are shown as controls. The quantification of centrocytes (CC) in percentage of live CD19⁺ cells is shown for the animals included in Fig. 4B. (E) Gating strategy for plasmablasts (PB) and plasma cells (PC). Plasmablasts and plasma cells were gated as singlets/live/CD3⁻/CD138⁺B220⁺ and singlets/live/CD3⁻/CD138⁺B220⁻, respectively. (F) Quantification of the percentages of plasmablasts and plasma cells, of the mice shown in E and in Fig. 4B. (G) Representative image of a Ki67-stained spleen. Arrows indicate GCs. Spleens were formalin-fixed and embedded in paraffin, and three sections per spleen ∼100 µm apart were stained with a Ki67-specific antibody. Images were acquired on a Leica microscope. The image is a stitch of a certain number of acquired adjoining images of the sample. (H) MFI of MHCII expression of the centrocytes shown in D. In D, F, and H, each symbol represents one mouse. Pooled data from three to five independent cohorts are shown. (I) Four and three wild-type C57BL/6 mice were i.v. injected with 0.5 million Crebbp^fl/wt or Crebbp^wt/wt MYC-expressing lymphoma cells, respectively, that were harvested from the pooled inguinal and axillary lymph nodes of lymphoma-bearing donors. Mice were assessed with respect to their spleen weights and splenic B cell MHCII expression at the study endpoint (28 d after transplantation), relative to two nontransplanted controls. P values were calculated using the Mann–Whitney test. *P < 0.05; ***P < 0.001.

Fig. 4.

Deletion of Crebbp within the GC compartment promotes GC expansion, reduces MHCII expression, and accelerates MYC-driven lymphomagenesis. (A and B) Littermates with either no (wt), one (fl/wt), or two (fl/fl) floxed alleles of Crebbp or Ep300, which additionally express Cre under the control of the AID promoter, were immunized four times at biweekly intervals with sheep red blood cells and killed 10 d after the last injection. Representative flow cytometry plots of splenic GC B cells are shown in A, and the frequencies of CD95⁺CD38^low GC B cells, as well as of Cxcr4⁺ centroblasts (CB), are shown in percentage of all CD19⁺ live cells in B. Nonimmunized mice are included for comparison. Pooled data from three to five independent cohorts are shown. (C) The GC number and GC area per spleen area (arbitrary units) of the mice shown in B, as quantified based on Ki67 staining of three spleen sections per mouse. (D and E) Median fluorescence intensity (MFI) of MHCII surface expression on GC B cells and centroblasts of the mice shown in B and C; representative histograms are shown in D. (F) Mice of the three indicated genotypes were immunized at regular 14-d intervals and monitored for clinical symptoms and enlarged lymph nodes indicating lymphoma development. Moribund mice were killed, and their survival time since the first immunization was plotted. (G) Four and three wild-type C57BL/6 mice were i.v. injected with 0.5 million Crebbp^fl/wt or Crebbp^wt/wt MYC-expressing lymphoma cells, respectively, that were harvested from the pooled inguinal and axillary lymph nodes of lymphoma-bearing donors. Mice were assessed with respect to their lymph node weights (inguinal and axillary LNs are plotted separately for each recipient and share the same color code) and lymph node B cell MHCII expression at the study endpoint (28 d after transplantation), relative to two nontransplanted controls. P values were calculated using the Mann–Whitney test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

GC B cells lacking one or both alleles of either Ep300 or Crebbp further exhibited reduced MHCII expression relative to Crebbp/Ep300-proficient GC B cells; this was true for all GC B cells, and especially for centroblasts (Fig. 4 D and E and Fig. S4H). The loss of Ep300 mirrored the effects of Crebbp loss on MHCII expression (Fig. 4 D and E and Fig. S4H). Finally, to examine whether the loss of Crebbp would promote MYC-driven lymphomagenesis, we crossed AID-Cre × Crebbp^fl/wt mice with a mouse strain that expresses MYC under the control of the Ig heavy chain enhancer (Emu-MYC) (15). Interestingly, the loss of one Crebbp allele was sufficient to significantly accelerate the formation of MYC-driven nodal B cell lymphomas but was not sufficient to induce lymphomas in the absence of this second hit (Fig. 4F); furthermore, we found that upon i.v. transplantation of equal numbers of tumor cells from Crebbp-proficient or Crebbp-deficient lymphoma-bearing donors, the latter cells engrafted more readily in the lymph nodes and spleens of wild-type recipient mice, and retained their low MHCII expression relative to Crebbp-proficient cells (Fig. 4G and Fig. S4I). We conclude from the combined results that HATs control MHCII expression in both human and murine B cells and that the loss of Crebbp, but not of Ep300, additionally contributes to the immunization-induced expansion of the GC compartment; both phenomena appear to predispose to malignant transformation.

The Loss of MHCII Expression, or CD4⁺ T-Cell Depletion, Promotes MYC-Driven Lymphomagenesis.

To address whether the loss of MHCII expression that results from HAT inactivation promotes lymphomagenesis, we crossed mice lacking MHCII with Emu-MYC mice. MHCII^−/− mice developed MYC-driven tumors earlier than their MHCII^+/+ littermates; however, the fraction of mice that did not develop lymphomas at all in the time frame of observation (80 d) was comparable in the two groups (Fig. 5A). MHCII^−/− mice lack MHCII-restricted CD4⁺ T cells in all lymphoid tissues (Fig. S5A), which might explain their enhanced predisposition to MYC-driven lymphomagenesis. To dissect whether the loss of MHCII affects lymphomagenesis in immunocompetent mice with normal CD4⁺ T cell frequencies, we transplanted 100,000 MHCII^−/− or MHCII^+/+ MYC-expressing tumor cells pooled from the axillary and inguinal lymph nodes of two donors per genotype into wild-type C57BL6 recipients and recorded the tumor incidence. Whereas only 2 of 16 recipients of MHCII^+/+ cells developed lymphomas with this low dose of cells, 10 of 17 recipients of MHCII^−/− cells did (Fig. 5B). The differential engraftment of MHCII-proficient and MHCII-deficient cells was reflected in on average significantly higher spleen and lymph node weights of recipients of MHCII^−/− cells, and more efficient engraftment of tumor cells in bone marrow (Fig. 5C). Finally, to confirm that MHCII-restricted T cells indeed control lymphomagenesis in the serial transplantation model, we depleted CD4⁺ T cells starting at either 1 d before injection of a saturating dose of 1 million tumor cells, or once palpable tumors had formed in the inguinal and axillary lymph nodes on day 11 after transplantation. The depletion of CD4⁺ T cells was efficient in all examined organs (Fig. S5B) and strongly enhanced lymphoma cell engraftment and growth in lymph nodes, spleen, and bone marrow, especially if T cells were depleted from the start of the experiment (Fig. 5D). The combined results are consistent with the notion that the loss of MHCII expression confers a tumor cell survival benefit, especially during lymphoma initiation, indicating that immune surveillance driven by CD4⁺ T cells limits lymphomagenesis in immunocompetent animals.

Fig. 5.

Tumor cell-intrinsic MHCII loss and CD4⁺ T cell depletion both confer enhanced susceptibility to MYC-driven lymphomagenesis. (A) Mice of the indicated genotypes were immunized at regular 14-d intervals with sheep red blood cells and monitored for clinical symptoms and enlarged lymph nodes indicating lymphoma development. Moribund mice were killed, and their survival time since the first immunization was plotted. (B and C) Sixteen and 17 wild-type C57BL/6 mice were i.v. injected with 0.1 million MHCII^+/+ or MHCII^−/− MYC-expressing lymphoma cells, respectively, that were harvested from the pooled inguinal and axillary LNs of lymphoma-bearing donors. Mice were assessed with respect to lymphoma incidence (B) and spleen and LN weights (one data point per mouse, the mean of the four inguinal and axillary LNs is shown) and B cell frequencies in bone marrow at the study endpoint (C; 28 d after transplantation), relative to two nontransplanted controls. Pooled data from two studies are shown in B and C, except for the BM data which were from one study only. (D) Eighteen, 17, and 7 wild-type C57BL/6 mice were i.v. injected with 1 million splenic MHCII^+/+ MYC-expressing lymphoma cells, respectively, and assessed with respect to spleen and LN weights (one data point per mouse, the mean of two inguinal and axillary LNs is shown) and B cell frequencies in bone marrow at the study endpoint (28 d after transplantation), relative to five nontransplanted controls. Mice were additionally treated weekly with CD4-specific or isotype control antibody, starting either at 1 d before transplantation, or once palpable tumors had formed (at 11 d after transplantation). Pooled data from two studies are shown in D. P values were calculated using the Mann–Whitney test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Fig. S5.

Tumor cell-intrinsic MHCII loss and CD4⁺ T cell depletion both confer enhanced susceptibility to MYC-driven lymphomagenesis. (A) Mice of the indicated genotypes were assessed with respect to their CD4⁺ T cell compartment in the blood. (B) Eighteen, 17, and 7 wild-type C57BL/6 mice were i.v. injected with 1 million splenic MHCII^+/+ MYC-expressing lymphoma cells, respectively, and additionally treated weekly with CD4-specific or isotype control antibody, starting either at 1 d before transplantation, or once palpable tumors had formed (at 11 d after transplantation). Graphs show the CD4⁺ T cell depletion efficiency in the spleen, lymph nodes (LN), and bone marrow (BM); pooled data from two studies are shown. P values were calculated using the Mann–Whitney test. ****P < 0.0001.

Discussion

In this study, we have combined targeted resequencing of the genes encoding the HATs CREBBP and EP300 with experimental approaches using cell cultures and mouse models to shed light on the potential tumor suppressive activity of HATs in DLBCL. As expected from previous studies using patient cohorts (2, 8, 10), our panel of 11 DLBCL cell lines confirmed that inactivating mutations in CREBBP and EP300 are common, monoallelic, and mutually exclusive and typically compromise the enzymatic activity of both enzymes, leading to strongly reduced levels of histone H3 acetylation of the lysine residues K14, K18, and K27. The siRNA-mediated knockdown of CREBBP, alone and especially in combination with EP300, strongly reduced global H3 acetylation at the same three residues, but did not affect the acetylation of nonhistone proteins.

To model HAT function in DLBCL cell lines in a manner that would recapitulate the dosage effect of monoallelic mutational CREBBP inactivation and a possible dominant negative effect of truncated protein resulting from the introduction of premature stop codons, we introduced a typical patient mutation monoallelically into a CREBBP/EP300–wild-type cell line. The expression of truncated CREBBP resulted in reduced global H3K18 acetylation and large changes in gene expression. The dominant category of dysregulated genes contained the various alleles of MHCII genes involved in antigen presentation, as well as their master regulator, the transcription factor CIITA, which could be confirmed not only by targeted qRT-PCR in the human DLBCL cell line clones, but also in murine GC B cells upon Cre-mediated excision of Crebbp or Ep300. The dependence of MHCII expression on active histone marks added by HATs thus appears to be a universal feature of murine and human B cells, which could further be confirmed experimentally by chromatin immunoprecipitation of the HLA-DRA locus with acetyl-H3–specific antibodies. The transcriptional signature we identified here to be linked to CREBBP-mediated histone modification is consistent with previous reports showing that CREBBP regulates superenhancer networks with central roles in GC/post-GC cell fate decisions, including genes involved in signal transduction by the B cell receptor and CD40 receptor, transcriptional control of GC and plasma cell development, and antigen presentation (11, 12).

The reduced expression of MHCII is an obvious, but not the only, consequence of HAT inactivation in CREBBP-mutant DLBCL cell lines. Mutant clones grow faster when s.c. transplanted into NSG mice, and CREBBP-mutant tumors are larger and heavier at the study endpoint than wild-type tumors. The same was true in Rag2^−/−γc^−/− mice that express several human cytokines, allowing for efficient engraftment of various human immune cell compartments (13). The differential growth in both xenograft models is unlikely to be due to immune evasion caused by MHCII down-regulation, a mechanism that has been proposed to explain the lack of T cell activation and proliferation in CREBBP-mutant FL (16). In FL, CREBBP mutations are known to occur within the earliest inferable progenitor, result in decreased MHCII expression on tumor B cells, and correlate with decreased frequencies of tumor-infiltrating CD4⁺ helper T cells (16). As antitumor immunity driven by CD4⁺ T cells can be ruled out in the two immunocompromised xenograft models used here, CREEBP-mutant tumors must benefit from additional cell-intrinsic growth advantages conferred by loss of HAT activity in this setting. The same was observed in genetically manipulated mice, where the loss of one or both alleles of Crebbp led to hyperproliferation of the GC B cell compartment upon immunization. The dysregulated expansion of the GC compartment likely provides a setting in which additional mutations can accumulate and early clonal lesions can arise, especially given the genomic instability that is an inherent feature of GC B cells; indeed, Crebbp loss accelerated MYC-driven lymphomagenesis in both spontaneous and serial transplantation models. We hypothesize that CREBBP mutations confer two distinct benefits during DLBCL pathogenesis, one promoting unrestricted growth during the GC reaction, and the other promoting immune evasion. Our results showing enhanced engraftment of MHCII-negative, MYC-expressing lymphoma cells into immunocompetent mice provide experimental evidence for the concept that the loss of MHCII expression indeed promotes immune evasion in this setting. This result was phenocopied by CD4⁺ T cell depletion, indicating that MHCII-restricted T cells control early events of lymphomagenesis in immunocompetent hosts, either through direct cytotoxic effects on tumor B cells or by creating a cytokine milieu that is not conducive to tumor initiation and/or progression. Interestingly, only one of the two biological functions of CREBBP appears to be shared by EP300. Ep300 inactivation in the mouse GC compartment down-regulates MHCII expression to an extent that is comparable to Crebbp inactivation; in contrast, the hyperproliferation of the GC compartment upon immunization was not seen in Ep300 mutant mice, which rather exhibited a contraction of their GC B cells. Thus, EP300 and CREBBP have both shared and distinct activities, which, in turn, is reflected by the much more common mutational inactivation of CREBBP than of EP300 in DLBCL patients (2, 8, 10).

In addition to their role as bona fide tumor suppressors in DLBCL and mechanistic contribution to the control of the GC reaction and antigen presentation, wild-type HAT activity also confers clear and previously unknown benefits in terms of patient survival (17). Furthermore, HAT mutational inactivation likely predicts treatment responses to targeted therapies with epigenetically acting drugs, such as the HDAC3 inhibitors (11, 18). In summary, our results provide evidence for two complementary tumor suppressive mechanisms in GC B cells that depend on HAT enzymes, of which one controls proliferation in a cell-autonomous manner, and the other sensitizes malignant B cells to CD4⁺ T cell-mediated tumor control.

Materials and Methods

DLBCL Cell Lines and Targeted Resequencing of CREBBP and EP300.

The panel of DLBCL cell lines used here included five of GCB DLBCL subtype (SU-DHL-4, SU-DHL-6, SU-DHL-10, SU-DHL-16, and RC-K8) and six of ABC DLBCL subtype (U-2932, OCI-Ly3, OCI-Ly10, SU-DHL2, SU-DHL5, and RIVA). Culture conditions, transfections, CRISPR manipulations, RNA sequencing, ChIP-PCR and Western blotting techniques are all described in SI Materials and Methods.

Animal Experimentation.

The mouse strains B6.Cg-Crebbp^tm1Jvd/J, B6.129P2-Ep300^tm2Pkb/J, B6.Cg-Tg(IghMyc)22Bri/J, and B6;129P2-Aicd^tm1CreMnz/J were obtained from the Jackson Laboratories. MHCII^−/−, NOD/SCID/IL2Rγ^−/− (NSG), and M-CSF^h;IL-3/GM-CSF^h;hSIRPA^tg;TPO^h;Rag2⁻;γc⁻ (MISTRG) (13) mice were obtained from a local repository. Ep300^fl/fl and Crebbp^fl/fl mice were crossed to AID^cre animals and MYC-transgenic mice were crossed with MHCII^−/− mice and AID^cre × Crebbp^fl/fl mice to obtain composite strains. Six- to eight-week-old mice were immunized three to six times every second week with 200 µL of 10% sheep red blood cells (Innovative Research) and killed 10 d after the last immunization. All flow cytometry procedures are described in SI Materials and Methods. For xenotransplantation studies, CREBBP^+/− and CREBBP^+/+ U2932 clones (10 × 10⁶ cells in 150 µL of PBS) were injected s.c. into both flanks of 6- to 8-wk-old NSG or MISTRG mice, or i.v. into MISTRG mice in the orthotopic model. Once palpable tumors had formed in the s.c. model (∼40 mm³), the volume of the tumors was measured by calipers and calculated by using the formula (a² × b)/2, where a is the shorter and b the longer tumor dimension. For serial transplantation studies, 100,000–1 million tumor cells pooled from the axillary and inguinal lymph nodes, or from spleens of mice, were cryopreserved in FCS + 10% DMSO, and were subsequently injected i.v. in a volume of 100 µL. All animal studies were reviewed and approved by the Zürich Cantonal Veterinary Office (licenses 227/2015, 235/2015).

SI Materials and Methods

DLBCL Cell Culture, Transfection, and Cell Viability Assays.

Cell lines were maintained at 37 °C, 5% CO₂ in a humidified atmosphere in IMDM (RIVA, OCI-Ly3, OCI-Ly10) or RPMI (SU-DHL2 and SU-DHL5) supplemented with 10% or 20% (SU-DHL-4, SU-DHL-6, SU-DHL-10, SU-DHL-16, RC-K8, U-2932) heat-inactivated FBS and antibiotics. For the purpose of siRNA treatment, 0.5 × 10⁶ DLBCL cells were nucleoporated using the Amaxa Nucleofector II device such that the final concentration of siRNA was 100 µM. The following siRNAs were purchased from Qiagen: Hs EP300 7 FlexiTube siRNA catalog no. SI02626267, Hs CREBBP8 FlexiTube siRNA catalog no. SI02633099 and Control AllStars Negative Control siRNA catalog no. SI03650318. Cells were harvested 48 h, 72 h, or 96 h after transfection for protein or RNA extraction. For viability assays, cells were seeded at a density of 0.3 × 10⁶/mL. Fifty microliters of cell suspension were transferred into 96-well plates containing 50 μL of fresh medium in triplicates. Twenty microliters of CellTiter-Blue reagent (Promega) was added, and plates were incubated for 4 h at 37 °C, 5% CO₂ in a humidified atmosphere. Viability was subsequently assessed by measuring fluorescence at 560Ex/590Em using a SpectraMax M5 microplate reader.

CRISPR/Cas9 Manipulation of U-2932 Cells.

The PX458 plasmid used was obtained from Addgene (48138: pSpCas9(BB)-2A-GFP). Guide RNAs were designed using the Zheng Lab online tool (crispr.mit.edu/) and cloned into PX458. Guide RNA efficiency was tested by calcium phosphate transfection of HEK 293T cells, and the best gRNA combination was chosen. gRNA 1 fwd: CAC CGT GCA CTT ACC CTC ATG AC; rev: AAA CGT CAT GAG GGT AAG AAT GCA C; gRNA 2 fwd: CAC CGC AGC ATT CAG ATA GTT TGT, rev: AAA CAC AAA CTA TCT GAA TGC TGC. Half a million U-2932 cells were nucleoporated with 3 μg of the plasmid containing the desired gRNAs using the Amaxa Nucleofector II device, and 48 h after transfection, the cells were single cell sorted for GFP expression into preconditioned medium. Single-cell cultures were subsequently screened by gDNA isolation (adapted from the KAPA Mouse Genotyping Kit; KAPA Biosystems) and PCR-based verification of the deletion (fwd: TCA GGG TGA GTT GTT TCC CC, rev: TTA GAG AGT GCT GGC CAA CA).

RNA Sequencing and Data Analysis.

RNA from DLBCL cell lines as well as CRISPR/Cas9 edited U-2932 was isolated using the NucleoSpin RNA kit (Macherey-Nagel). RNA quality was assessed by Bioanalyzer 2100 followed by library preparation using the TruSeq RNA Sample Prep Kit v4 (Illumina). Sequencing was subsequently performed on the Illumina HiSeq 2500 instrument. RNA-seq reads were quality-checked with fastqc, which computes various quality metrics for the raw reads. RNA-seq reads were mapped to the GRCh38 reference human genome using STAR, and reads were counted according to Ensembl gene annotation using the featureCounts function in the Rsubread Bioconductor package. Statistical analysis of differential expression was conducted with the DESeq2 package.

Quantitative RT-PCR.

RNA was extracted using the Nucleospin RNA kit. One microgram of total RNA was reverse transcribed using SuperScriptIII reverse transcriptase (Invitrogen). For qRT-PCR, Lightcycler 480 Cyber Green Master I (Roche) was used followed by analysis on a Lightcycler 480 instrument. Samples were measured in duplicates. For all primer pairs, the efficiency was calculated by performing dilution series experiments. Target mRNA abundance was subsequently calculated relative to human RPLP0 or RPLP32. The primers used were as follows: RPLP0, fwd: CCA GCT CTG GAG AAA CTG CTG, rev: CAG CAG CTG GCA CCT TAT TGG; RPLP32, fwd: GAA GTT CCT GGT CCA CAA CG, rev: 5′-GCG ATC TCG GCA CAG TAA G; CREBBP, fwd: GTC CAG TTG CCA CAA GCA C, rev: CAT TCG GGA AGG AGA AAT GG; EP300, fwd: GCC AAG TAC TTC AGC TAC CCA GT, rev: GGC ATC AGT GCC TGT CGT AG; HLA-DRA, fwd: GCA CTG GGA GTT TGA TGC TC, rev: AGG GCA CAC ACC ACG TTC; HLA-DRB1, fwd: ACT ACG GGG TTG TGG AGA GC, rev: GAG CAG ACC AGG AGG TTG TG; WEE1, fwd: ATT GGC GGG CTC TGT TGA T, rev: GCC CAC GCA GAG AAA TAT CG.

Western Blotting.

Protein extracts were made in RIPA buffer (50 mM Tris⋅HCl, pH 8.0, 150 mM sodium chloride, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with 2 mM sodium orthovanadate, 15 mM sodium pyrophosphate, 10 mM sodium fluoride and 1× Complete protease inhibitor mixture (Roche). Protein extracts for CREBBP/EP300 were prepared in MNase buffer 1 (0.3 M sucrose, 15 mM Tris pH 7.5, 60 mM KCl, 15 mM NaCl, 5 mM MgCl₂, 0.1 mM EGTA, 0.5 mM DTT, 0.5% Nonidet P-40, and 1% sodium deoxycholate) + MNase buffer 2 (0.3 M sucrose, 85 mM Tris pH 7.5, 3 mM MgCl₂, 2 mM CaCl₂, and 10 U S7 MNase; Roche). The reaction was stopped by addition of EDTA with a final concentration of 5 mM. Protein concentrations were determined using BCA assay (Pierce) or Bradford assay (Bio-Rad) for MNase-isolated proteins, and equal amounts were separated by SDS/PAGE (4–20% Mini-PROTEAN TGX Precast Protein Gels; Bio-Rad) followed by transfer onto nitrocellulose membranes. Membranes were probed with antibodies against α-tubulin (DM1A; Sigma-Aldrich), p300 (N-15; Santa Cruz Biotechnology), CBP (A-22; Santa Cruz Biotechnology), TFIIH p89 (s-19; Santa Cruz Biotechnology), and p53 antibody (9282), Acetyl-p53 (Lys382), CBP (D6C5), Histone H3 XP (D1H2), Acetyl-Histone H3 Lys9 (C5B11), Acetyl-Histone H3 Lys27 XP (D5E4), Acetyl-Histone H3 Lys18 (D8Z5H), and Acetyl-Histone H3 Lys14 (D4B9) (all from Cell Signaling).

Flow Cytometric Analysis.

Cells isolated from mouse spleens, lymph nodes, blood, or bone marrow were treated with ACK red blood cell lysis buffer pH 7.2–7.4 (150 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂EDTA) and passed through a 40 µM cell strainer to produce single-cell suspensions. Cells were subsequently stained using the following fluorescent-labeled anti-mouse antibodies: PE-Cy7 conjugated anti-CD95/FAS (Jo2; Becton Dickinson), PerCP-Cy5.5 conjugated anti-CD86 (GL-1; BioLegend), Pacific Blue conjugated anti-I-A/I-E (M5/114.15.2; BioLegend), Purified Rat Anti-Mouse CD16/CD32 (93; BioLegend), Brilliant Violet 650 anti-CD45 (30-F11; BioLegend), Brilliant Violet 605 anti-CD4 (RM4-5; BioLegend), PE-Cy7 anti-TCR-beta chain (H57-597; BioLegend), APC anti-CD8a (53-6.7; BioLegend), FITC anti-CD19 (1D3; BioLegend), Alexa Fluor 700 anti-I-A/I-E (M5/114.15.2; BioLegend), FITC anti-CD4 (RM4-5; BioLegend), Pacific Blue anti-CD8a (53-6.7; BioLegend), PE anti-CD45 (30-F11; BioLegend), Fluorescein labeled Peanut Agglutinin (PNA; VectorLaboratories), and Fixable Viability Dye eFluor 780, APC conjugated anti-CD19 (eBio1D3), FITC conjugated anti-CD38 (90), PE conjugated anti-CXCR4 (2B11), eFluor 450 conjugated anti-CD45R/B220 (RA3-6B2), PE-Cy7 conjugated CD23 (B3B4), FITC conjugated anti-CD21/CD35 (eBio4E3), and PE conjugated anti-CD3e (145-2C11) all from ebio/Affymetrix. The fluorescent-labeled anti-human antibodies used were as follows: APC anti-CD45 (2D1), FITC anti-CD19 (HIB19), Pacific Blue anti-CD20 (HI47), PE anti-HLA-DR (LN3), and Hu FCR binding inhibitor purified all from ebio/Affymetrix. Data were acquired on CyAn ADP (Beckman Coulter) or LSR FORTESSA (BD) flow cytometers and analyzed with the FlowJo software package (TreeStar).

Treatment of Mice with Antibodies.

Anti-mouse CD4 (BE0119, Clone: YTS) and rat IgG2b isotype control (BE0090, Clone: LTF-2) antibodies were purchased from BioXCell. Mice were administered a 500-µg dose 1 d before i.v. injection of tumor cells, then once per week with a 250-µg dose until end of the study (28 d), or starting only once palpable tumors had formed.

Ki67 Staining of GCs.

CONFIRM anti-Ki-67 (30-9) Rabbit Monoclonal Primary Antibody (Roche) was used for staining of proliferating cells within GCs on formalin-fixed paraffin-embedded spleen sections using a Ventana automated slide stainer. This was provided as a service by the Laboratory for Animal Model Pathology of the Institut für Veterinärpathologie of the University of Zürich. Three sections ∼100 µm apart were assessed per mouse spleen. The images were acquired on a Leica microscope and ImageJ was used for area assessment. GCs were identified as Ki67-positive areas within blue areas. Background staining, mostly on the peripheries of the spleen sections, was not included.

ChIP-PCR.

ChIP was performed as described (19). One hundred million cells were fixed with 1% formaldehyde, lysed, and sonicated (Branson Sonicator; Branson), resulting in sheared chromatin of an average size of 200 bp. Two micrograms of antibodies H3K27ac (ab4729; Abcam), and H3K18ac (ab1191; Abcam), or control rabbit IgG (sc-2027; Santa Cruz Biotechnology) were added to 25 µg of precleared sample and immunoprecipitated overnight at 4 °C. The complexes were purified using protein A beads (Invitrogen) followed by elution from the beads and decross-linking. DNA was purified using PCR purification columns (Macherey Nagel) and quantified by qPCR. The following primers were used for qPCR: GAPDH, fwd: AGA AGG CTG GGG CTC ATT TG, rev: AGG GGC CAT CCA CAG TCT TC; MyoD, fwd: CCT CTT TCG GTC CCT CTT TC, rev: TTCCAAACCTCTCCAACACC; XL4, fwd: CAG AGA AAG GGA ACT GAA AGT CAT TT, rev: TTA TGA CAC TGT TTA GTC CTA GAA CAC TGA; −600 bp, fwd: ATG AGA TAC AAT GCC AGC CAT CC, rev: ACA GTT GGA GAG TTT GCG TAA GG; −300 bp, fwd: TGT CCC TTA CGC AAA CTC TCC, rev: ACA CAA GAT ACT CCG TTC ATT GG; WXY, fwd: GAT CTC TTG TGT CCT GGA CCC TTT GCA AGA ACC CT, rev: CCC AAT TAC TCT TTG GCC AAT CAG AAA AAT ATT TTG; +1,500 bp, fwd: CTC CGT CTC AAA CAA CCA AAC C, rev: ACC AAC ACC AAG GGA ATA ATG AAC; +3,500 bp, fwd: TTC CGC AAG TTC CAC TAT CTC C, rev: CGA GTT TCA CAC AAG CAT CAT AGG.

Statistics.

All statistical analyses were performed using GraphPad Prism software. Graphs represent means plus SD of at least two independent experiments and statistical analysis was performed using two-tailed Student’s t test for RT-PCR data, and using two-tailed Mann–Whitney test for in vivo studies. A log-rank test was performed for comparison of survival curves.

Data Availability.

The RNA-sequencing dataset is publicly available on the GEO Accession viewer with accession no. GSE89879. (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE89879).