Fig 2: Loss of Ebf1 expression does not occur through loss of heterozygosity in most EB tumors. (a) One EB tumor (4606) expressed B220 but not CD19, suggesting loss of Ebf1 expression and/or function in this tumor as CD19 is positively regulated by EBF1. Flow cytometric analysis of peripheral blood leukocytes (PBL) from aged controls and an affected EB mouse is shown. Tumor cells are B220lo/CD244+/CD19–. CD19 is a positive target and CD244 is a negative target of EBF1 transcriptional regulation. Histograms in this figure are representative of one affected EB, five Bcl-xLTg and five wild-type control mice analyzed. (b) High-throughput sequencing of Ebf1 genes was performed on paired normal tissue (tail or ear clip) and tumor (lymph node or spleen) DNA from eight different mice bearing EB tumors. The table lists copy number variations in the samples found by the control-FREEC application and verified by visual inspection of aligned reads using normalized counts. One tumor (EB54) displayed ~50% loss of coverage inside four regions in introns 7, 8 and 10. (c) Histograms depicting depth of sequence coverage for two regions with loss in EB54 tumor compared to tail DNA are shown. Boxes highlight the areas of sequence coverage loss. ChIP-seq data show monomethyl H3K4 density across Ebf1 in the CH12 murine B lymphoma cell line (Hardison PSU, UCSC accession wgEncodeEM002370). Boxes correspond to the locations of EB54 sequence coverage loss shown above. (d) Correlation of gene expression data for EBF1, TCF3, RUNX1, BCL2L1, BCL2 and MCL1 from N=177 high-risk pediatric B-cell-acute lymphoblastic leukemia (B-ALL) patient samples.8 The significance of each correlation is shown in the legend for each graph (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001, ns=not significant.) Data accessed from Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo), accession GSE11877.

Fig 3: Neoplastic cells display a pre-B-cell phenotype with aberrant expression of NK-cell/myeloid markers. (a) Neoplastic cells in the bone marrow, thymus and peripheral blood lymphocytes (PBL) display a B-cell phenotype. B cells from 8-month-old littermate control (shaded) or EB (black line) bone marrow, thymus and PBL are shown. B-cell percentages are low in control mice due to their age. (b) Peripheral blood leukocytes (PBL) of affected EB but not unaffected EB or Bcl-xLTg control mice contain a uniform population of enlarged (lymphoblastic) B progenitor cells (B220loIgM-). Unaffected EB mice do not display this population in the periphery (spleen). (c, d) Neoplastic cells in the spleen display an early B-cell progenitor phenotype with aberrant expression of the natural killer (NK) cell markers CD244 and NK1.1. Splenocytes from Ebf1+/– or EB mice were labeled with the indicated antibodies for flow cytometry. All panels in (b, c) represent lymphocyte-gated cells. Panels in (d) are additionally gated on CD19+ cells. Neoplastic cells do not display membrane-bound immunoglobulin. Notably, CD19+ lymphoma cells display NK-cell markers as well as B-cell markers. Histograms and density plots in this figure are representative of eight affected EB, four unaffected EB, five Bcl-xLTg and five Ebf1+/– control mice analyzed (see Table 1).

Fig 4: EB mice develop aggressive disease with neoplastic infiltrates in multiple organs. Compared to Bcl-xLTg control mice, affected EB mice have dense neoplastic round cell infiltrates within hematopoietic and lymphoid tissues including skull bone marrow (a, b) and spleen (c, d), as well as in the lungs (f, g) and kidneys (h, i). Insets show higher magnification. Within the skull (b) and some long bones (not shown), neoplastic cells fill and expand the medullary cavity. They also extend into the meninges and middle ear (b; black arrows). Neoplastic cells completely replace the hematopoietic tissue of the skull bones and are arranged in dense sheets among a scant fibrovascular stroma. Neoplastic cells are large, round to oval with scant to small amounts of eosinophilic homogenous cytoplasm and indistinct cell margins, while normal bone marrow displays a heterogenous population of cells (compare b and a insets). Nuclei are large, round to oval with vesicular chromatin and one to rarely two prominent central nucleoli. Anisokaryosis and anisocytosis are mild. Mitoses (b; open arrows) are frequent (8–10 per × 400 field). (d) Affected EB splenic white pulp (asterisks) is expanded by similar neoplastic cells yet maintains typical architecture. The red pulp contains hematopoietic cells (extramedullary hematopoiesis). Lungs (g) and kidneys (i) from EB mice have perivascular neoplastic infiltrates (black arrows), while Bcl-xLTg lungs (f) and kidneys (h) display limited infiltrates of small lymphocytes with a normal morphology. (e,j) Immunohistochemistry (IHC) shows neoplastic cells in the spleen (e) and mesenteric adipose tissue (j) which display intense nuclear staining with Ki-67 antibody. Images a–j are representative of three EB, three Ebf1+/– and two Bcl-xLTg mice analyzed. (k, l) IHC reveals numerous cells in control bone marrow (k, Ebf1+/-) display nuclear staining with EBF1 antibody, while neoplastic cells in EB bone marrow do not retain stain (l). (m) Immunoblot showing EBF1 expression in whole cell extracts from wild-type MACS-sorted splenic B cells (EBF1 staining control) compared to whole cell extracts from pre-B tumor cell isolates in four different EB mice and three different Eµ-Myc mice. GAPDH is included as a loading control.

Fig 5: Pre-B lymphoma/leukemia is characterized by loss of lineage-commitment transcription factors. (a) A population of B220loCD244+ cells is present in healthy EB mouse bone marrow but not in peripheral tissues or in Bcl-xLTg or wild-type mice. Additional staining (not shown) defined this population as late pre-B: B220loCD19+BP-1-CD25+IgM- (same phenotype as EB tumor cells, far right panel). Flow cytometric analyses of total bone marrow, splenocytes, peripheral blood leukocytes (PBL) and subiliac lymph nodes are shown. Fluorochromes used for B220 were PE-Cy7 and eFluor450. Contour plots in this figure are representative of eight affected EB, four unaffected EB, five Bcl-xLTg and five Ebf1+/– control mice analyzed. (b) RT–qPCR analysis of relative transcript levels from selected genes in sorted late pre-B cells (B220+IgM–BP-1–CD25+) from the bone marrow of five congenic wild-type mice, three Bcl-xLTg mice, B220loCD244+ (presumed tumor progenitor) cells from the bone marrow of three healthy EB bone marrow transfer recipients (Unaffected EB), and leukemic cells from four different EB mice (EB tumors). *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001. Primers sequences are listed in Supplementary Table S1.