Coordinated changes in glycosylation regulate the germinal centre through CD22

Germinal centres (GC) are sites of B-cell expansion and selection, which are essential for antibody affinity maturation. Compared to naive follicular B-cells, GC B-cells have several notable changes in their cell surface glycans. While these changes are routinely used to identify the GC, functional roles for these changes have yet to be ascribed. Detection of GCs by the antibody GL7 reflects a reduction in the glycan ligands for CD22, which is an inhibitory co-receptor of the B-cell receptor (BCR). To test a functional role for downregulated CD22 ligands in the GC, we generated a mouse model that maintains CD22 ligands on GC B-cells. With this model, we demonstrate that glycan remodeling is crucial for proper GC B-cell response, including plasma cell output and affinity maturation of antibodies. The defect we observe in this model is dependent on CD22, highlighting that coordinated downregulation of CD22 ligands on B cells plays a critical function in the GC. Collectively, our study uncovers a crucial role for glycan remodeling and CD22 in B-cell fitness in the GC.

Notch2-mediated plasticity between marginal zone and follicular B cells

Follicular B (FoB) and marginal zone B (MZB) cells are functionally and spatially distinct mature B cell populations in the spleen, originating from a Notch2-dependent fate decision after splenic influx of immature transitional B cells. In the B cell follicle, a Notch2-signal is provided by DLL-1-expressing fibroblasts. However, it is unclear whether FoB cells, which are in close contact with these DLL-1 expressing fibroblasts, can also differentiate to MZB cells if they receive a Notch2-signal. Here, we show induced Notch2IC-expression in FoB cells re-programs mature FoB cells into bona fide MZB cells as is evident from the surface phenotype, localization, immunological function and transcriptome of these cells. Furthermore, the lineage conversion from FoB to MZB cells occurs in immunocompetent wildtype mice. These findings demonstrate plasticity between mature FoB and MZB cells that can be driven by a singular signaling event, the activation of Notch2.

Deciphering the Role of Locally Disordered DNA Methylation on CLL Development In Vivo

Large-scale DNA methylation analysis of chronic lymphocytic leukemia (CLL) has identified a pervasive genome-wide level of discordance in local methylation state in leukemic cells compared to normal B cells. This is associated with variation in gene expression, increased clonal evolution and poorer clinical outcomes. We hypothesized that locally disordered methylation could lead to dysregulation of gene expression and hence contribute to cancer development and progression. To test this, we have engineered mouse lines with B-cell restricted homozygous or heterozygous knock-out of Dnmt3a by crossing Dnmt3a-floxed mice with CD19-Cre mice. Dnmt3a is a DNA methyltransferase, catalyzing the addition of a methyl group to CpG sequences in the DNA and thereby regulating gene expression. Although DNMT3A mutations are only rarely identified in CLL, RNA sequencing and protein expression analysis reveal dysregulation of DNMT3A. We confirmed partial or complete reduction in Dnmt3a protein levels in B cells from CD19-Cre;Dnmt3a heterozygous (Dnmt3a-het) and CD19-Cre;Dnmt3a homozygous mice (Dnmt3a-hom), respectively. These mice therefore provide a unique opportunity to study B cell restricted changes in locally discordant methylation over time. We first assessed the impact of Dnmt3a deletion on normal B cell development, prior to CLL development, by characterizing splenic B cell of CD19-Cre (control) or Dnmt3a-hom mice. Flow cytometry data using B220, CD21 and CD23 markers to identify B220+CD23+CD21- follicular B cells and B220+CD23+CD21high marginal zone B cells revealed elevated levels of follicular B cells (83.1% vs 87.6%, p=0.008) and reduced levels of marginal zone B cells (9.6% vs 4.1%, p=0.001) in Dnmt3a-hom mice in comparison to control mice (n=3 mice per group). These results indicate that mice with Dnmt3a deletion present with massive changes in their B cells, even prior to overt CLL development. We next monitored both Dnmt3a-het and Dnmt3a-hom cohorts over time for CLL development. We observed that 100% Dnmt3a-hom mice developed CLL-like disease by 7 months (n=23), characterized by CD5+B220+;Igk+ expression and evident within the blood, bone marrow (BM), spleen and peritoneum, suggesting a fundamental role of altered DNMT3A expression in generation of CLL. In comparison, 75% of Dnmt3a-het mice developed CLL-like disease by 18 months (n=12), with similar expansion of CD5+C220+ expansion in the BM and spleen. By RNA-sequencing analysis of normal splenic B cells from CD19-Cre and Dnmt3a-hom mice (n=3 mice, 10 weeks old), we detected substantial changes in gene expression, including 113 upregulated genes and 39 downregulated (p<0.05, FC>2). To explore the development of locally disordered methylation following transformation, CLL cells from Dnmt3a-hom mice (n=3) were subjected to reduced representation bisulfite sequencing (RRBS), a high-throughput technique to analyze genome wide methylation patterns. We found that murine CLL-like cells display locally disordered methylation, which was detected in all genomic features covered by this assay, indicating that disordered methylation is broadly affecting the murine CLL cells' epigenome. Additionally, we identified a set of differentially methylated regions (DMRs) between B cells from CD19-Cre vs CLL cells from Dnmt3a-hom (n = 2,839 DMRs), with a minimum difference of 0.2 and a minimum of 10 CpGs per DMR. Interestingly, gene ontology analysis demonstrated strong association with genes hypermethylated in TCL1 mouse model, linking this model with alternative murine models for CLL. In conclusion, we have studied B cell specific deletion of Dntm3a and showed the development of CLL in 100% of the case in Dnmt3a-hom mice. Our data suggest a fundamental role for Dnmt3a in CLL development through increased locally disordered methylation and changes in associated transcriptional signatures. This mouse model provides an exciting experimental model to undertake functional in vivo studies in order to elucidate the contribution of epigenetic changes on CLL development. Disclosures Neuberg: Pharmacyclics: Research Funding; Madrigal Pharmaceuticals: Equity Ownership; Celgene: Research Funding. Wu:Neon Therapeutics: Other: Member, Advisory Board; Pharmacyclics: Research Funding.

Induction of metabolic quiescence defines the transitional to follicular B cell switch

Transitional B cells must actively undergo selection for self-tolerance before maturing into their resting follicular B cell successors. We found that metabolic quiescence was acquired at the follicular B cell stage in both humans and mice. In follicular B cells, the expression of genes involved in ribosome biogenesis, aerobic respiration, and mammalian target of rapamycin complex 1 (mTORC1) signaling was reduced when compared to that in transitional B cells. Functional metabolism studies, profiling of whole-cell metabolites, and analysis of cell surface proteins in human B cells suggested that this transition was also associated with increased extracellular adenosine salvage. Follicular B cells increased the abundance of the cell surface ectonucleotidase CD73, which coincided with adenosine 5′-monophosphate–activated protein kinase (AMPK) activation. Differentiation to the follicular B cell stage in vitro correlated with surface acquisition of CD73 on human transitional B cells and was augmented with the AMPK agonist, AICAR. Last, individuals with gain-of-function PIK3CD (PI3Kδ) mutations and increased pS6 activation exhibited a near absence of circulating follicular B cells. Together, our data suggest that mTORC1 attenuation may be necessary for human follicular B cell development. These data identify a distinct metabolic switch during human B cell development at the transitional to follicular stages, which is characterized by an induction of extracellular adenosine salvage, AMPK activation, and the acquisition of metabolic quiescence.

KZF-L162R Mutation Affects Splenic Mature B Cell Development and Alters Expression of Aiolos Target Genes

Large-scale DNA sequencing efforts in chronic lymphocytic leukemia (CLL) have identified a broad array of putative cancer drivers arising from somatic mutations in this disease, but functional understanding of the impact of these genetic events on CLL onset and progression remains to be elucidated. One such example is mutation in the IKZF3 gene, encoding the zinc finger protein AIOLOS, mutated in ~2% of CLLs and associated with fludarabine-refractory disease. AIOLOS is a lymphoid-restricted transcription factor and a chromatin remodeler that plays an essential role in B cell development and maturation. In CLL, the IKZF3 mutation, also reported in few cases of diffuse large B cell lymphoma and mantle cell lymphoma,targets a highly conserved hotspot (L162R, homologous to murine L161R) that is localized in the 2nd zinc finger of the DNA-binding domain, required for DNA sequence recognition. Given the localization of this hotspot mutation, we hypothesized that it impacts the function of AIOLOS to drive CLL. To characterize the effects of the IKZF3-L162R mutation, we generated a knock-in mouse line that conditionally expresses the point mutation in a B cell lineage context through crossing with Cd19-cre mice, generating mouse lines carrying Ikzf3-L161R as either a heterozygous mutation (Ikzf3-L161RHet), homozygous mutation (Ikzf3-L161RHomo) or wild-type Ikzf3(Ikzf3WT). Given the established role of Aiolos in lymphoid differentiation, we first asked how the mutation impacts B cell development. By flow cytometry, using established markers to detect marrow pro-B, pre-B, transitional and mature B cell populations, or peritoneal B1a and B1b cell populations, no differences in the proportion of cells were observed between Ikzf3WTor Ikzf3-L161RHet. In the spleen, however, the average proportion of marginal zone B cells (B220+CD23+CD21high) was markedly reduced in heterozygousmice compared to wild type mice (6 mice/group: 4.9% vs. 11.5%, p=0.006), while the average proportion of follicular B cells (B220+CD23+CD21-) was increased (76% vs. 63%; p=0.003). Immunohistochemical staining of spleen sections confirmed that the marginal zone area was significantly reduced in Ikzf3-L161RHetmice (p=0.01). In addition, we noted a higher proliferation rate of B cells from Ikzf3-L161RHetmice when stimulated with LPS and IL-4 for 3 days (p=0.01), suggesting that the mutation confers a survival advantage to B cells. Similar analyses in Ikzf3-L161RHomomice are ongoing. By immunofluorescence and immunoprecipitation, neither Aiolos binding with its partners CHD4, SIN3 or HDAC1, nor its cellular distribution were impacted by the mutation. Of note, the total protein level of Aiolos was increased in Ikzf3-L161RHetmice (9 mice/group; p<0.05). Since the mutation localizes to a DNA binding domain, we hypothesized that it modifies the ability of Aiolos to control expression of its target genes. We therefore performed CHIP-seq in Ikzf3WTsplenic B cells, and identified Aiolos-associated high confidence peaks (fold change (FC) enrichment compared to input > 20) corresponding to DNA binding sites in the promoters of genes such as Rps19, Ogg1, Dusp2, Phf23 or Brfp1 and confident peaks (FC>10) in the anti-apoptotic gene Mcl1 and in genes involved in BCR signaling (i.e.Syk, Pi3kr1, Nfkbid), suggesting that their expression is under the control of Aiolos. Comparison of the expression by qPCR of these 8 genes in splenic B cells from the 3 mouse lines revealed Dusp2, Mcl1, Syk, Nfkbid and Phf23 to be upregulated in Ikzf3-L161RHomoB cells (p<0.05) but not in Ikzf3-L161RHetB cells. These findings suggest that the mutation directly impacts the expression level of Aiolos target genes. The upregulation of Mcl1 expression is particularly relevant in the context of CLL as dysregulation of anti-apoptotic signaling is characteristic of the disease. In conclusion, these data show that Aiolos mutation affects B cell subpopulation ontogeny, inducing a disproportionate abundance of follicular B cells endowed with high proliferative capacity. The mutation impacts Aiolos transcription capacity leading to upregulation of genes belonging to pathways cardinal to CLL development, including BCR signaling and apoptosis. Ongoing studies focus combining RNA-seq and CHIP-seq in mutant B cells, with the aim of identifying the breadth of differential expressed genes and dysregulated cellular pathways in mutant B cells in an unbiased manner. Disclosures Wu: Neon Therapeutics: Equity Ownership.