Loss of Pax5 exploits Sca1-BCR-ABL^p190 susceptibility to confer the metabolic shift essential for pB-ALL

Abstract

Preleukemic clones carrying BCR-ABL^p190 oncogenic lesions are found in neonatal cord blood, where the majority of preleukemic carriers do not convert into precursor B-cell acute lymphoblastic leukemia (pB-ALL). However, the critical question of how these preleukemic cells transform into pB-ALL remains undefined. Here we model a BCR-ABL^p190 preleukemic state and show that limiting BCR-ABL^p190 expression to hematopoietic stem/progenitor cells (HS/PC) in mice (Sca1-BCR-ABL^p190) causes pB-ALL at low penetrance, which resembles the human disease. pB-ALL blast cells were BCR-ABL-negative and transcriptionally similar to pro-B/pre-B cells, suggesting disease onset upon reduced Pax5 functionality. Consistent with this, double Sca1-BCR-ABL^p190+Pax5^+/− mice developed pB-ALL with shorter latencies, 90% incidence, and accumulation of genomic alterations in the remaining wildtype Pax5 allele. Mechanistically, the Pax5-deficient leukemic pro-B cells exhibited a metabolic switch towards increased glucose utilization and energy metabolism. Transcriptome analysis revealed that metabolic genes (IDH1, G6PC3, GAPDH, PGK1, MYC, ENO1, ACO1) were upregulated in Pax5-deficient leukemic cells, and a similar metabolic signature could be observed in human leukemia. Our studies unveil the first in vivo evidence that the combination between Sca1-BCR-ABL^p190 and metabolic reprogramming imposed by reduced Pax5 expression is sufficient for pB-ALL development. These findings might help to prevent conversion of BCR-ABL^p190 preleukemic cells.

Introduction

Although preleukemic clones carrying BCR-ABL^p190 oncogenic lesions are frequently found in neonatal cord blood (1,2), they often remain silent, since the majority of these carriers do not develop precursor B-cell acute lymphoblastic leukemia (pB-ALL). In addition, small fractions of normal B-cells in healthy adults carry silent BCR-ABL oncogenes (3). These findings suggest that BCR-ABL^p190 might promote leukemogenesis by creating an aberrant progenitor compartment that is susceptible to malignant transformation through accumulation of additional secondary hits, which act as drivers of leukemogenesis. Thus, BCR-ABL^p190 does not seem to be a dominant oncogene within the natural cellular hematopoietic stem/progenitor cell (HS/PC) compartment where the BCR-ABL^p190 oncogenic lesion takes place. This is further supported by both clinical data showing that BCR-ABL^p190-induced tumorigenesis is not reversible through the unique inactivation of the gene defect initiating leukemia development (4), and by murine in vivo data showing that suppression of BCR-ABL^p190 in leukemic mice carrying a tetracycline-repressible BCR-ABL^p190 transgene did not rescue the malignant phenotype (5). However, the mode of action that BCR-ABL^p190 exerts in the HS/PC compartment, remains difficult to demonstrate using available models of transgenic-driven BCR-ABL ALL (6,7), since the penetrance of most of the respective pB-ALL disease models is 100%. Thus, current available models are not able to mimic the human scenario, where the presence of the BCR-ABL transgene in HS/PCs does not necessarily lead to disease development, but mainly generates a susceptibility that is preserved in the cancer-initiating cells. In this work, we explored a novel mode of action of BCR-ABL^p190 in HS/PCs, utilizing mice with restricted expression of the BCR-ABL^p190 oncogene to the stem/progenitor cell compartment.

Materials and Methods

Generation of Sca1-BCR-ABL^p190 and Sca1-BCR-ABL^p190+Pax5^+/−

The heterozygous Pax5^+/− mice (8) have been described previously. Heterozygous Pax5^+/− mice were bred to Sca1-TK-IRES-BCR-ABL^p190 mice to generate compound heterozygotes. The Sca1-TK-IRES-BCR-ABL^p190 vector was generated as follows. The 9 kb EcoRI-EcoRI TK-IRES-BCR-ABL^p190 cassette was inserted into the ClaI site of the pLy6 vector (9), resulting in Sca1-TK-IRES-BCR-ABL^p190 vector. The transgene fragment was excised from its vector by restriction digestion with NotI, purified for injection (2 ng/ml) and injected into CBAxC57BL/6J fertilized eggs. Transgenic mice were identified by Southern analysis of tail snip DNA after EcoRI digestion. Human ABL cDNA was used for detection of the transgene. Two founder lines were obtained for the Sca1-BCR-ABL^p190 transgene with different positional integration of the transgenic construct in both lines. All animal work has been conducted according to relevant national and international guidelines and it has been approved by the Bioethics Committee of University of Salamanca and by the Bioethics Subcommittee of Consejo Superior de Investigaciones Cientificas (CSIC). For all genotypes both male and female mice of a mixed C57BL/6 × CBA background were included in the study. We systematically used littermates of the same breeding. Upon signs of disease, mice were sacrificed and subjected to standard necropsy procedures. All major organs were examined under the dissecting microscope. Tissue samples were taken from homogenous portions of the resected organ and fixed immediately after excision. Differences in Kaplan-Meier survival plots of transgenic and WT mice were analyzed using the log-rank (Mantel-Cox) test.

Real Time analysis (BCR-ABL^p190)

cDNA used in quantitative PCR studies was synthesized using reverse transcriptase (Access RT-PCR System; Promega, Madison, WI). 2 µl of second round amplified RNA was transcribed. Primers and probes used for quantitative PCR have been described previously (10). The probes were designed so that genomic DNA would not be detected during the PCR. The sequences of the specific primers and probes were as follows: BCR-ABL^p190, sense primer 5’-CCGCAAGACCGGGCAGAT-3’, antisense primer 5’-CAGATGCTACTGGCCGCTGA-3’ and probe 5’-TGGCCCAACGATGGCGAGGG-3’; c-Abl, sense primer 5’-CACTCTCAGCATCACTAAAGGTGAA-3’, antisense primer 5’-CGTTTGGGCTTCACACCATT-3’, and probe 5’-CCGGGTCTTGGGTTATAATCACAATG-3’.

Real-time PCR quantification (qRT-PCR) of Hk2, Ldha Slc2a3, Idh1 and Pgk1

We analyzed expression of Hk2, Ldha, Slc2a3, Idh1 and Pgk1 in both leukemic BCR-ABL^p190 and leukemic BCR-ABL^p190+Pax5^+/− cells by qRT-PCR as follows: cDNA was synthesized using reverse transcriptase (Access RT-PCR System; Promega, Madison, WI) and genomic DNA was removed by DNAase treatment (Roche, 04 716 728 001). Real-time PCR reactions were performed in an Eppendorf MasterCycler Realplex machine.

Commercially available assays for quantitative PCR from IDT (Integrated DNA Technologies) were used: Hk2 (Assay ID: Mm.PT.S832698746), Ldha (Assay ID: Mm.PT.58.29860774), Slc2a3 (Assay ID: Mm.PT.30464830), Idh1 (Assay ID: Mm.PT.58.5996441), Pgk1 (Assay ID: Mm.PT.39a.22214854), and RplpO (Assay ID: Mm.PT.58.43894205). Probes were specifically designed to prevent detection of genomic DNA by PCR. Measurement of RplpO gene product expression was used as an endogenous control. All samples were run in triplicate. The comparative CT Method (ΔΔCt) was used to calculate relative expression of the transcript of interest and a positive control.

Microarray data and enrichment analysis

Total RNA was isolated in two steps using TRIzol (Life Technologies) followed by Rneasy Mini-Kit (Qiagen) purification following the manufacturer’s RNA Clean-up protocol with the optional On-column DNase treatment. Integrity and quality of the RNA were verified by electrophoresis. Samples were analyzed using Affymetrix Mouse Gene 1.0 ST arrays.

The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (11) and are accessible through GEO Series accession number GSE85600. Differentially expressed genes were tested for enrichment using GSEA enrichment analysis from MSigDB (12) (http://www.broad.mit.edu/gsea/). Gene expression signatures, specifically up or down-regulated in human B-ALL gene sets (13,14), were assessed for their overlap with up or down-regulated genes within tumor specimen, using GSEA enrichment analysis. Moreover enrichment analysis for the specific normal murine B-cell stage signatures (15), BCR-ABL signature (16), BCR-ABL target genes (obtained from http://www.broad.mit.edu/gsea/) and for the specific Pax5 gene signatures described in (17–20) were carried out.

Cloning of Hek293T constructs

A cDNA encoding the CDS of the wildtype murine Pax5 was obtained from the Pax5 Plasmid (#35003) (21) (Addgene) via PCR using the Phusion High-Fidelity DNA Polymerase (Thermo Scientific). The mutant cDNAs for murine Pax5 (P80R, P80L, R38C, P32L, I54N, V26G) were created by site directed mutagenesis by PCR with the same polymerase. The obtained Pax5 sequences were cloned into a derivative of a bicistronic expression vector (pMC3) used previously for stable expression of genes in cell lines (22). Here, the vector was modified to encode the Hygromycin resistance gene as the second cistron (pMC3-PAX5.Hygro). The identity of the respective cDNAs was confirmed by Sanger sequencing.

Hek293T cells were transfected using the Attractene Transfection Reagent (Qiagen), according to manufacturer’s protocol. Briefly, 1.2 µg of each pMC3-PAX5.Hygro construct or the empty vector was diluted in 100 µl serum-free media, before 4.5 µl of Attractene was added. After an incubation period of 15 minutes the DNA/Attractene mix was added to Hek293T cells (6 × 10⁵) in suspension and plated on a 6-well plate. Cells harboring the plasmids were selected using 200 µg/ml Hygromycin B (Life Technologies).

Luciferase assay

Hek293T cells expressing pMC3-PAX5.Hygro^WT, pMC3-PAX5.Hygro^Empty or pMC3-PAX5.Hygro^mutant were transfected with 2 µg luc-CD19 construct (kindly provided by M. Busslinger) (23) and 100 ng pRL-TK Renilla luciferase plasmid DNA (Promega) using FuGene 6 (Roche Diagnostics). 48 h after transfection, cells were lysed and Firefly and Renilla luciferase activity measurement was performed using the Dual-Glo Luciferase Assay System (Promega), according to manufacturer’s protocol. All transfections were carried out in triplicate in at least 3 independent experiments. Firefly luciferase activity was normalized according to the corresponding Renilla luciferase activity.

Immunoblot analysis

B220⁺ BM cells of healthy Sca1-BCR-ABL^p190 and Sca1-BCR-ABL^p190+Pax5^+/− mice were FACS sorted, while Ba/F3 cells expressing BCR-ABL^p190 served as a positive control (24). Cells were lysed in RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.5% Sodiumdeoxycholate, 1% NP-40 substitute, 0.1% SDS), containing protease and phosphatase inhibitors (Roche Diagnostics). 20 µg of whole protein was separated on SDS-PAGE and transferred to Hybond-C Extra membranes (Amersham Biosciences). Immunoblotting was carried out using the following antibodies: c-ABL 1:1000 (Cell signaling, #2862), p-STAT5 1:1000 (Cell Signaling, mAb #9359), STAT5 1:1000 (mAb #9358, Cell Signaling) and beta-Actin clone AC-74 1:10000 (Sigma-Aldrich). Detection was carried out using anti-rabbit or anti-mouse horseradish peroxidase conjugates (Santa Cruz Biotechnology), respectively, with an ECL system (Thermo Scientific).

Western blot analysis of BCR-ABL^p190 expression and presence of p-CrkL (Tyr207) was carried out in control and leukemic Sca1-BCR-ABL^p190+Pax5^+/− cells cultured in the presence of IL7. As positive controls, we used the following cell lines: Ba/F3+BCR-ABL^p190, a Ba/F3 cell line expressing human BCR-ABL^p190 (24), and TOM-1 cells that were established in 1983 from the bone marrow of a 54-year-old woman with refractory Philadelphia chromosome-positive acute lymphoblastic leukemia (ALL) (25). All cells were maintained in RPMI with 10% FCS. Extracts were normalized for protein content by Bradford analysis (Bio-Rad Laboratories Inc., Melville, NY, USA). The following primary antibodies were used: BCR (Ab-2) (Oncogene Science), p-CrkL (Tyr207) (Cell Signaling) and beta-actin (I-19; Santa Cruz Biotechnology). Reactive bands were detected with an ECL system (Amersham).

Glucose, Lactate and ATP Measurements

Glucose and lactate levels were measured using the Amplex Red Glucose/Glucose Oxidase Assay Kit (Invitrogen) and the L-Lactate Assay Kit (Cayman Chemical), respectively, according to the manufacturers’ protocols. Glucose and lactate concentrations were measured in fresh and spent medium. Total ATP levels were measured using the ATP Bioluminescence Assay Kit CLS II (Roche) according to the manufacturer’s protocol. One million cells per mL were seeded in fresh medium. Relative levels of glucose consumed, lactate produced and total ATP are shown. To account for different rates of cell proliferation and death, all three parameters were normalized to number of viable cells as determined by applying Trypan blue exclusion.

Metabolic Assays using the XFe24 Flux Analyzer

Glycolytic profiles were monitored under basal conditions (glycolytic activity) and in response to acute addition of oligomycin (an inhibitor of ATP synthase). Glycolytic activity was measured as extracellular acidification rate (ECAR). Oligomycin suppresses mitochondrial ATP production, leading to increased compensatory glycolysis to meet the energy demands of the cell and to maintain energy homeostasis. This elevated rate of glycolysis is called glycolytic capacity. Extracellular acidification rate (ECAR) was measured using a Seahorse XFe24 Flux Analyzer (Seahorse Bioscience). Briefly, 1.5 ×10⁵ cells per well were plated using Cell-Tak (BD Biosciences). Prior to measurements, cells were incubated in XF Base Medium supplemented with 25 mM glucose, 1 mM sodium pyruvate and 2 mM GlutaMAX for 1 hr at 37°C (non-CO2 incubator) for pH stabilization. ECAR was measured at the resting stage (glycolytic activity in XF Base Medium supplemented with glucose, sodium pyruvate and GlutaMax) and in response to oligomycin (1 µM; glycolytic capacity). Oxygen consumption rate (OCR) was measured using a Seahorse XFe24 Flux Analyzer with the XF Cell Mito Stress Test Kit (Seahorse Bioscience) according to the manufacturer’s instructions. OCR was measured in response to oligomycin (1 µM). All parameters were normalized to number of viable cells as determined by applying Trypan blue exclusion.

Chromatin immunoprecipitation sequencing (ChIP-seq) analysis

ChIP-seq tracks for PAX5 antibodies in human B lymphocytes (ENCODE, Encyclopedia of DNA Elements, GM12878) are downloaded from:. http://genome.ucsc.edu/ENCODE/dataMatrix/encodeChipMatrixHuman.html. Gene models are shown in UCSC genome browser hg19.

Results

Limiting BCR-ABL^p190 expression to HS/PCs induces pB-ALL with reduced penetrance

To explore early cellular changes in BCR-ABL^p190 preleukemic HS/PCs that might take place in the development of BCR-ABL^p190-pB-ALL, we engineered mice with restricted oncogene expression to HS/PCs by placing the BCR-ABL^p190 cDNA under the control of the Sca1 (Stem cell antigen 1) promoter (10,26) (Sca1-BCR-ABL^p190) (Supplementary Fig. 1a). Insertion of the TK-IRES-BCR-ABL^p190 cDNA under the control of the mouse Ly-6E.1 promoter (9) yielded the plasmid Sca1-BCR-ABL^p190 (Supplementary Fig. 1a), which was used to drive Sca1-directed expression of BCR-ABL^p190 in C57BL/6 × CBA mice. Two founder lines were obtained for the Sca1-BCR-ABL^p190 transgene, which were viable, developed normally and were used for further studies. The BCR-ABL^p190 transcript was only detected in Sca1⁺Lin⁻ cells (Fig. 1a) and expression and signaling of the fusion protein was absent in healthy B220⁺ sorted bone marrow (BM) cells of BCR-ABL^p190 mice as shown by western blot and GSEA analysis (Fig. 1a–b). In order to further assess the developmental stage at which Sca1-BCR-ABL^p190 expression was downregulated, qRT-PCR detection of the transgene was carried out in HS/PC populations of the respective healthy mice. These data confirmed a drastic decrease of BCR-ABL^p190 expression, starting at the stage of common lymphoid progenitors, while the transgene could no longer be detected at the level of pro-B cells (Supplementary Fig. 1b).

Figure 1. Transient BCR-ABL^p190 expression in HS/PCs is sufficient to induce pB-ALL.

(a) BCR-ABL^p190 is absent in precursor B-cells of preleukemic Sca1-BCR-ABL^p190 and Sca1-BCR-ABL^p190+Pax5^+/− mice. Left panel: B220 positive cells from the bone marrow (BM) of preleukemic Sca1-BCR-ABL^p190 and Sca1-BCR-ABL^p190+Pax5^+/− mice were FACS sorted and lysates were subjected to immunoblot analysis for c-Abl, as well as p-Stat5, and total Stat5. Ba/F3+BCR-ABL^p190 cells serve as positive control. One representative experiment out of two biological replicates is shown. Right panel: Quantification of thymidine kinase (TK) and BCR-ABL^p190 expression in Sca1-BCR-ABL^p190 mice by real-time PCR in Sca1⁺Lin⁻ and Sca1⁻Lin⁺ cells. The percentage of TK and BCR-ABL^p190 transcripts with reference to beta-actin is shown.

(b) GSEA analysis showing BCR-ABL target genes are not enriched in preleukemic pro-B/pre-B cells of Sca1-BCR-ABL^p190 mice compared to pro-B/pre-B cells of wildtype (WT) mice age-matched (FDR=0.988).

(c) pB-ALL specific survival curve of Sca1-BCR-ABL^p190 mice (blue, n=36) compared to wildtype control mice (black, n=20) (log-rank P value 0.0523).

(d) Flow cytometric analysis of hematopoietic subsets in diseased Sca1-BCR-ABL^p190 mice. Representative plots of cell subsets from the BM and peripheral blood (PB) show accumulation of blast B-cells in Sca1-BCR-ABL^p190 mice (n=5) compared to control littermate WT mice age-matched (n=5).

(e) Gene set enrichment analysis of leukemic mice. GSEA identified significant enrichment in human BCR-ABL pB-ALL gene sets (extracted from (13,14)) in Sca1-BCR-ABL^p190 tumor-bearing bone marrows compared to B220⁺ bone marrow B cells from WT mice (GSEA FDR = 0.002 and FDR = 0.004).

(f) Percentage of HSC, CLP and LRP in 4-months old Sca1-BCR-ABL^p190 mice (n=9–11) compared to age-matched WT mice (n=4–7) analyzed by flow cytometry. Error bars represent the standard deviation. Unpaired t-test p-value is indicated in each case.

(g) Percentage of BM pro-B cells in 4-months old Sca1-BCR-ABL^p190 mice (n=5) compared to age-matched WT mice (n=5) analyzed by flow cytometry. Error bars represent the standard deviation. Unpaired t-test p-value is indicated in each case. Representative flow cytometry plots of pro-B cell subsets are shown on the bottom.

(h) Cell death susceptibility mediated by IL7 removal in Sca1-BCR-ABL^p190 pro-B cells. Sorted B220⁺ cells from bone marrow of young Sca1-BCR-ABL^p190 and WT mice were cultured under conditions to allow the isolation and expansion of a pure population of pro-B cells. Pro-B cells were cultured with or without IL7 for 24h. Induction of apoptosis was assessed by flow cytometry using Annexin V/PI staining. Data represent the means ± SD of normalized live cells from six independent experiments.

In an aging mouse colony, Sca1-BCR-ABL^p190 mice developed pB-ALL with low penetrance (13%), with earliest disease occurrence at 13 months (Fig 1c). The leukemia was characterized by blast infiltration, appearance of blast B-cells (CD19-B220+IgM-) in BM and peripheral blood (PB) and disruption of splenic architecture (Fig. 1d and Supplementary Fig. 1c). Transcriptome analysis revealed that these Sca1-BCR-ABL^p210 pB-ALLs were aligned with the pro-B/pre-B cell stage (12,15) (Supplementary Fig. 1d) and resembled the human BCR-ABL^p190-pB-ALL signature, despite the absence of BCR-ABL^p190 in the blast cells (Fig. 1e) (13,14). In accordance with this observation, BCR-ABL signaling was not enriched in leukemic Sca1-BCR-ABL^p190 cells (Supplementary Fig. 1e). Therefore, these data demonstrate that limited expression of the BCR-ABL^p190 oncogene in HS/PCs is capable of inducing pB-ALL. Moreover, the resulting pB-ALLs align with a differentiation stage comparable to human BCR-ABL^p190-pB-ALL, while being independent of a detectable BCR-ABL^p190 oncogene expression and activation of its downstream signaling.

Pax5-defective pro-B/pre-B cells are permissive for Sca1-BCR-ABL^p190 pB-ALL development

To elucidate the mechanism underlying this specific BCR-ABL^p190 mediated pB-ALL susceptibility, we analyzed the different B-cell developmental stages in healthy Sca1-BCR-ABL^p190 and WT littermates of the same breeding. Healthy Sca1-BCR-ABL^p190 mice showed reduced numbers of hematopoietic stem cells (HSC), common lymphoid progenitors (CLP) and lineage restricted progenitors (LRP) compared to WT controls, whereas the percentage of pro-B/pre-B cells in the BM was significantly increased (Fig. 1f–g, Supplementary Fig. 1f–g and Supplementary Table 1). Serial blood analysis revealed that although healthy Sca1-BCR-ABL^p190 mice exhibited higher B220⁺ cell numbers at 2 months of age, their B cells rapidly decrease and level off below 20% around 6 months after birth (Supplementary Fig. 2a–b). Murine precursor B-cells are dependent on intact IL7/IL7R signaling, thus we explored the response of healthy Sca1-BCR-ABL^p190 pro-B cells to IL7 withdrawal. In agreement with the lack of BCR-ABL^p190 signaling in the respective cells, our findings showed that these cells are similar sensitive to IL7 withdrawal as their WT counterparts (Fig. 1h). Therefore, these data suggest that BCR-ABL^p190 favors the appearance of an aberrant precursor B-cell compartment in the BM and that differentiation to mature B-cells is impaired without exposure of the precursor B-cells to the oncogenic BCR-ABL^p190 kinase. Thus, we hypothesized that the nature and function of the second hit must determine the specific target cell expansion. In addition to the absence of the B-cell marker CD19, Pax5 transcriptional activity was lost or markedly reduced in leukemic cells of Sca1-BCR-ABL^p190 pB-ALLs (Supplementary Table 2–3 and Supplementary Fig. 2c). These findings suggested Pax5 as a candidate for a second oncogenic hit, since as a master regulator of B-cell development and identity (8), PAX5 loss-of-function alterations occur in around 30% of human BCR-ABL^p190 B-ALL (27). To investigate the functional consequence of combined reduction of Pax5 dosage and restricted BCR-ABL^p190 expression in HS/PCs, we generated Sca1-BCR-ABL^p190+Pax5^+/− mice. In line with the results from Sca1-BCR-ABL^p190 only mice, these mice also displayed absence of the BCR-ABL^p190 fusion protein in healthy B220⁺ BM cells and inactive BCR-ABL^p190 downstream signaling (Fig. 1a and 2a). Sca1-BCR-ABL^p190+Pax5^+/− mice did not display a decrease in the total number of pro-B cells; however, the number of B220⁺ cells in the PB was reduced directly after birth and their IL7 sensitivity was increased, comparable to pro-B cells of Pax5^+/− mice (22) (Fig. 2b–c and Supplementary Fig. 2a–b). Compared to Sca1-BCR-ABL^p190 mice, Sca1-BCR-ABL^p190+Pax5^+/− mice developed pB-ALL with higher incidence (90%) and earlier disease onset (Fig. 2d). The same pB-ALL development could be observed, when cells from Sca1-BCR-ABL^p190+Pax5^+/− mice were transplanted into sub-lethally irradiated syngeneic mice, while no leukemia developed from Pax5^+/− wildtype pro-B cells (Supplementary Fig. 3a–c). Since the development of pB-ALL in Pax5 heterozygous mice was recently linked to infection exposure (22), Sca1-BCR-ABL^p190+Pax5^+/− cohorts were kept in a conventional as well as a specific pathogen free (SPF) facility, in order to address whether infection plays a role for leukemia initiation. Both cohorts developed pB-ALL at comparable rates (Supplementary Fig. 4a–c), which highlights the fact that in the Sca1-BCR-ABL^p190+Pax5^+/− model, Pax5 does not function as a susceptibility event in pB-ALL development.

Figure 2. Pax5-defective pro-B/pre-B cells are permissive for BCR-ABL^p190 pB-ALL development, as shown in Sca1-BCR-ABL^p190+Pax5^+/− mice.

(a) GSEA analysis showing BCR-ABL target genes are not enriched in preleukemic pro-B/pre-B cells of Sca1-BCR-ABL^p190+Pax5^+/− mice compared to pro-B/pre-B cells of WT mice age-matched (FDR=0.121).

(b) Percentage of BM pro-B cells in 4-months old Sca1-BCR-ABL^p190+Pax5^+/− mice (n=5) compared to age-matched Sca1-BCR-ABL^p190 (n=5) and WT mice (n=5) analyzed by flow cytometry. Error bars represent the standard deviation. Unpaired t-test p-value is indicated in each case. Representative flow cytometry plots of pro-B cell subsets are shown on the right.

(c) Cell death susceptibility mediated by IL7 removal in Sca1-BCR-ABL^p190+Pax5^+/− pro-B cells. Sorted B220⁺ cells from BM of young Sca1-BCR-ABL^p190+Pax5^+/−, Sca1-BCR-ABL^p190, Pax5^+/− and WT mice were cultured under conditions to allow the isolation and expansion of a pure population of pro-B cells. Pro-B cells were cultured with or without IL7 for 24h. Induction of apoptosis was assessed by flow cytometry using Annexin V/PI staining. Data represent the means ± SD of normalized live cells from six independent experiments.

(d) pB-ALL specific survival curve of Sca1-BCR-ABL^p190+Pax5^+/− mice (green, n=34) compared to Sca1-BCR-ABL^p190 (blue, n=36) and WT control mice (black, n=20) showing a significantly (log-rank P value < 0.0001) shortened life span.

(e) Relative expression of BCR-ABL^p190 in total BM of a leukemic WT mouse transplanted with Sca1-BCR-ABL^p190+Pax5^+/− proB cells, total bone marrow of preleukemic Sca1-BCR-ABL^p190+Pax5^+/− mice and TOM-1 cell line, respectively. A Ba/F3+BCR-ABL^p190 (Ba/F3^p190) cell line was used as a positive control. Error bars represent the mean +/− the standard deviation of 3 replicates.

(f) Flow cytometric analysis of hematopoietic subsets in diseased Sca1-BCR-ABL^p190+Pax5^+/− mice. Representative plots of cell subsets from the BM and PB show accumulation of blast B-cells in Sca1-BCR-ABL^p190+Pax5^+/− mice (n=27) compared to control littermate wildtype mice age-matched (n=5).

(g) WT and leukemic Sca1-BCR-ABL^p190+Pax5^+/− primary cells were cultured in media without IL7. Proliferation was measured using Trypan blue. Values represent the mean of 3 replicates with essentially identical datasets.

In addition, the blast population of the pB-ALLs from Sca1-BCR-ABL^p190+Pax5^+/− mice lacked BCR-ABL^p190 activity (Fig. 2e and Supplementary Fig. 5a). In line with these results, neither BCR-ABL^p190 protein nor downstream signaling was detected in leukemic Sca1-BCR-ABL^p190+Pax5^+/− cells (Supplementary Fig. 5b), which is further supported with resistance of the murine Sca1-BCR-ABL^p190+Pax5^+/− leukemic cells to Gleevec treatment (Supplementary Fig. 5c). The blast population was clonal, with a cell surface phenotype of CD19⁻B220⁺IgM⁻Mac1^+/− and accumulated in the BM, PB, lymph nodes and spleen (Fig. 2f and Supplementary Fig. 5d–e). Moreover, isolated leukemic pro-B cells were able to grow independent of IL7 (Fig. 2g). Sca1-BCR-ABL^p190+Pax5^+/− leukemic gene expression profiles clustered together with those of Sca1-BCR-ABL^p190 leukemias and classified Sca1-BCR-ABL^p190+Pax5^+/− pB-ALLs as pro-B cell stage (Supplementary Fig. 5f–g). These results suggest that even without active BCR-ABL^p190 expression, the pB-ALL cells of Sca1-BCR-ABL^p190+Pax5^+/− mice closely resemble the human BCR-ABL pB-ALL signature (Supplementary Fig. 5h). Moreover, they showed downregulation of genes important for B-cell development and differentiation, including Pax5, Ikzf1 and CD19 (Supplementary Table 4), which was confirmed by lost or markedly reduced Pax5 transcriptional activity in Sca1-BCR-ABL^p190+Pax5^+/− leukemic cells (Supplementary Fig. 5i). In order to rule out that the heterogeneity of the control B-cells could also contribute to the observed differential gene expression pattern, we characterized the global expression signature of purified wildtype pro-B/pre-B cells and compared to the expression signature of leukemic Sca1-BCR-ABL^p190+Pax5^+/− and Sca1-BCR-ABL^p190 cells. The analysis showed similar differential gene expression profile and similar GSEA enrichments than when it was compared to BM B220⁺ cells (Supplementary Fig. 5f (lower panel) and Supplementary Fig. 5j–k).

Taken together, the characteristics of Sca1-BCR-ABL^p190+Pax5^+/− pB-ALLs overlap to a large extent with the observations in Sca1-BCR-ABL^p190 leukemias. However, the increase in pB-ALL incidence and reduced latency without affecting the incidence of myeloid and non-hematopoietic tumors suggest that the time point and cellular compartment in which the Pax5 mutations take place facilitate and determine to a large extent the clonal evolution of Sca1-BCR-ABL^p190 mediated pB-ALL development.

Identification of Pax5 as a barrier of clonal selection in Sca1-BCR-ABL^p190 pB-ALL development by next generation sequencing

To validate these findings on a genomic level, as well as to further identify somatically acquired second hits leading to pB-ALL development, we performed whole exome sequencing of three Sca1-BCR-ABL^p190 and 13 Sca1-BCR-ABL^p190+Pax5^+/− BM tumor samples and their corresponding germline samples. The resulting somatic variants revealed that Jak3 and Pax5 genes were recurrently mutated in Sca1-BCR-ABL^p190+Pax5^+/− pB-ALLs, while Sca1-BCR-ABL^p190 only tumors displayed a heterogeneous mutational spectrum (Supplementary Fig. 6a–b and Supplementary Table 3). For the Pax5 gene, a variety of different Pax5 mutations were found in 46% of Sca1-BCR-ABL^p190+Pax5^+/− mice, which were all located in the gene’s DNA binding domain (Fig. 3a).

Figure 3. Mouse tumor exome sequencing data identify the second hit within the WT Pax5 gene.

(a) Schematic of Pax5 gene domains. Displayed are single nucleotide variants (SNVs) identified by whole exome sequencing (WES) in Sca1-BCR-ABL^p190+Pax5^+/− mice, which are all located in the gene’s DNA binding domain.

(b) Attenuated transcriptional activity of Pax5^mutant proteins. Relative luciferase activity is indicated for Pax5^WT as well as Pax5^mutant proteins, as measured by a Pax5-dependent reporter gene assay in Hek293T cells, which were stably transfected with the different Pax5 constructs. Bars show mean luciferase activity from three individual experiments with triplicate measurements.

(c) Upper panel: Whole genome analysis identifies a deletion spanning more than 27 kb in the Pax5 gene on Chromosome 4 in mouse W556 (marked in red). Lower panel: Pax5 deletion analysis on cDNA level reveals two additional Pax5 truncations in mice S650 and S314 (similar to that observed in mouse W556). Whole BM cDNA from a healthy Pax5^+/− mouse (WT) served as control.

(d) Backtracking of I54N in the murine Pax5 gene by Sanger and Amplicon sequencing, using genomic DNA of bimonthly bleedings of Sca1-BCR-ABL^p190+Pax5^+/− mouse W555. The leukemic clone is only detectable in the diseased mouse at around 5 months (sacrifice) of age (n=2).

Given that Pax5 p.P80R and p.V26G result in reduced PAX5 activity in humans (27,28) we performed reporter gene assays to further assess the functional consequences of the newly identified Pax5 variants. The Pax5-dependent reporter gene assay (23) showed significant reduction of all Pax5 variants in transcriptional activation compared to WT Pax5 (Fig. 3b). Moreover, whole genome sequencing revealed an additional wide-range deletion of Pax5 Exon 2–4 in one Sca1-BCR-ABL^p190+Pax5^+/− mouse (W556), while similar deletions were detected in two additional Sca1-BCR-ABL^p190+Pax5^+/− mice without Pax5 variant (S650 and S314) by PCR (Fig. 3c). Therefore, these data suggest that reduction/loss of wildtype Pax5 played a significantly role in disease development. In order to further monitor disease evolution over time, we performed deep sequencing of the Pax5 variant p.I54N with a read depth up to 2 million reads using bleedings of the corresponding mouse. Pax5 p.I54N was first detectable in the leukemic sample (Fig. 3d), while it was absent in the sample taken prior to disease development, emphasizing the relevance of aberrant Pax5 expression for disease progression. This indicates that during the early lifespans of mice, the pre-B cell population initiates polyclonal VDJ recombination events from which a monoclonal leukemia will emerge later in life. In agreement with this, polyclonal VDJ recombination events were evident in this preleukemic pre-B cell population from which monoclonal pre-B leukemia can emerge (Supplementary Fig. 7a–b). Taken together, these data show that in order to achieve clonal selection for BCR-ABL^p190 mediated pB-ALLs, elimination or significant reductions of wildtype Pax5 activity is critical for disease progression.

Reduced Pax5 activity drives the metabolic switch toward glucose utilization and energy metabolism in Sca1-BCR-ABL^p190 leukemia

To elucidate the underlying mechanism by which reduced Pax5 activity mediates disease progression, transcriptome analyses were performed, which showed enrichment in genes expressing mediators of glucose uptake and utilization, as well as energy metabolism in both Sca1-BCR-ABL^p190 and Sca1-BCR-ABL^p190+Pax5^+/− pB-ALL cells, when compared to WT cells (Fig. 4a and Supplementary Fig. 8–10). To investigate the clinical relevance of our findings and to determine whether human pB-ALLs share a similar metabolic signature, we studied gene expression data from human pB-ALLs as well as normal pro-B, pre-B-I, pre-B-II and immature B-cells (29,30). As expected, expression of PAX5 was downregulated in human pB-ALLs (Supplementary Fig. 11). In contrast, several metabolic genes (IDH1, G6PC3, GAPDH, PGK1, HK3, MYC, ENO1 and ACO1) were upregulated in both our mouse pB-ALL models (Fig. 4a) and human pB-ALL cells (29,30) (Supplementary Fig. 11). The metabolic genes identified play key roles in modulating glucose utilization and energy metabolism. Hexokinases (HKs) mediate the first committed step in most glucose metabolism pathways (31), and HK3 is a transcriptional target of the myeloid transcription factor CEBPA, which opposes B-cell transcription factors (32). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH), phosphoglycerate kinase (PGK1) and enolase 1 (ENO1) catalyze distinct steps of glycolysis (33). G6PC3 encodes the catalytic subunit of glucose-6-phosphatase that mediates the last step of gluconeogenic and glycogenolytic pathways (34). Isocitrate dehydrogenase 1 (IDH1) and acotinase 1 (ACO1) are essential enzymes in the tricarboxylic acid cycle (TCA cycle), which plays a critical role in energy metabolism (35). Furthermore, MYC-driven transformation leads to increased expression of glycolytic enzymes and glucose utilization (36). Notably, analysis of Chromatin Immunoprecipitation Sequencing (ChIP-seq) data obtained using human B lymphocytes (ENCODE GM12878) revealed that PAX5 binds to the promoter region of IDH1, G6PC3, GAPDH, PGK1, MYC1, ENO1 and ACO1 (Supplementary Fig. 12). PAX5 can function as a transcriptional activator as well as a repressor through interaction with Groucho family factors (37). These findings suggest that loss of PAX5 function may drive metabolic reprogramming during disease progression through transcriptional regulation of these metabolic genes.

Figure 4. Glycolytic energy production in Sca1-BCR-ABL^p190 negative leukemic pro-B cells.

(a) Gene expression changes in regulators of glucose uptake and energy supply in leukemic Sca1-BCR-ABL^p190 (n=2), Sca1-BCR-ABL^p190+Pax5^+/− (n=7) and WT BM B220⁺ cells (n=5) were plotted as heat map.

(b–g) Metabolic measurements were performed for (1) Pax5 WT pro-B, (2) Pax5 WT; Sca1-BCR-ABL^p190 pro-B (non-leukemia), (3) Pax5^+/− pro-B, (4) Pax5^+/−; Sca1-BCR-ABL^p190 pro-B (non-leukemic), and (5) Pax5^+/−; Sca1-BCR-ABL^p190 leukemia cells. For measurements of glucose consumption (n=6) (b), lactate production (n=6) (c) and total ATP levels (n=6) (d), cells were seeded in fresh medium. On day 3 following seeding of cells, cells and spent medium were harvested and processed for metabolic measurements.

(e–g) Oxygen consumption rate (OCR) was measured to monitor mitochondrial ATP turnover (n=3) (e). Extracellular acidification rates (ECAR) were measured at the resting state (with glucose; glycolytic activity) (n=6) (f) and in response to oligomycin (glycolytic capacity) (n=6) (g). All values are normalized to cell numbers and are shown as average amounts relative to average values of wildtype control (±SD). P-values are indicated as calculated by the Student’s T-Test.

To elucidate the functional significance of gene expression changes observed in our models, metabolic assays were performed to characterize Sca1-BCR-ABL^p190- and PAX5-mediated metabolic reprogramming in our models. Consistent with the metabolic signature observed, significant increases in levels of glucose consumed (~55-fold), lactate produced (~37-fold) and total ATP (~16-fold) were drastically increased in Sca1-BCR-ABL^p190+Pax5^+/− leukemic cells compared to their normal counterpart (Fig. 4b–d). Furthermore, increases in glycolytic activity (~5 to 6-fold), glycolytic capacity (~9 to 10-fold) and mitochondrial ATP production (~4-fold; Fig. 4e–g) were observed in Sca1-BCR-ABL^p190+Pax5^+/− leukemic cells. Metabolic measurements also revealed that glucose consumption (~1.7-fold), lactate production (~1.4-fold) and total ATP levels (~3-fold) were increased in non-leukemic Sca1-BCR-ABL^p190 pro-B cells carrying WT Pax5 (Fig. 4b–d). Similarly, increased glucose consumption (~1.3-fold), lactate production (~1.2-fold) and total ATP levels (~3-fold) were observed in Pax5-deficient pro-B cells when compared to sorted B220⁺ BM cells from WT control (Fig. 4b–d). These findings suggest that either Sca1-BCR-ABL^p190 alone or Pax5-deficiency resulted in modest, but significant increases in glucose utilization and energy metabolism. Interestingly, while Sca1-BCR-ABL1^p190 alone increased total ATP levels by ~3-fold, deficiency in Pax5 doubled the levels of total ATP (Sca1-BCR-ABL^p190 pro-B cells carrying WT Pax5: ~3-fold vs. Sca1-BCR-ABL^p190+Pax5^+/− pro-B cells: ~6-fold).

In addition, oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured to monitor mitochondrial function (ATP turnover) and glycolytic profiles, respectively. Compared to Pax5 WT pro-B cells, mitochondrial ATP production was increased by ~1.4-fold in non-leukemic Sca1-BCR-ABL^p190 pro-B cells. Interestingly, Pax5-deficiency resulted in additional increases in mitochondrial ATP production. In comparison to Pax5 WT pro-B cells, mitochondrial ATP production was increased by ~2.3 fold in Sca1-BCR-ABL^p190+Pax5+/− pro-B cells (Fig. 4e). Glycolytic profiles were monitored under basal conditions (glycolytic activity) and in response to acute addition of oligomycin (an inhibitor of ATP synthase). Oligomycin suppresses mitochondrial ATP production, leading to increased compensatory glycolysis to meet the energy demands of the cell (glycolytic capacity). Consistent with the above observations, glycolytic activities of both non-leukemic Sca1-BCR-ABL^p190 pro-B cells (~2-fold) and Pax5-deficient pro-B cells (~2 to 3-fold) were higher compared to WT pro-B cells (Fig. 4f). Furthermore, glycolytic capacities of Sca1-BCR-ABL^p190 pro-B cells (~2-fold) and Pax5-deficient pro-B cells (~4 to 5-fold) were increased (Fig. 4g). Significant energy supply to fuel macromolecular synthesis for proliferation is required during transformation of leukemia cells. Recently, it has been demonstrated that PAX5 functions as a metabolic barrier through transcriptional repression of GLUT1/3/6 (glucose transporters), HK2/3, PFKL, PGAM, PYGL and G6PD (glucose utilization) as PAX5 targets in pB-ALL (38). Here, we identified novel metabolic targets of PAX5 (IDH1, G6PC3, GAPDH, PGK1, MYC, ENO1 and ACO1) and observed a similar metabolic signature in human pB-ALLs. Furthermore, we demonstrated that Pax5-deficiency led to increased glucose utilization and energy metabolism, which is further enhanced in cells primed by the BCR-ABL^p190 oncogene at the stage of HS/PCs.

Discussion

The common belief in pB-ALL development is that the disease originates from pro-B/pre-B cells that critically depend on survival signals (39–41). However, the BCR-ABL^p190 oncogene is not able to confer self-renewal stem-cell-like properties to progenitors in mice (42). In agreement with this view, both, twin-based data, and the high frequency of preleukemic clones carrying BCR-ABL^p190 oncogenic lesions frequently found in neonatal cord blood, support that the BCR-ABL^p190 gene fusion creates a preleukemic clone remaining clinically silent until secondary mutational events give rise to a full blown leukemia (1,2). Common among these secondary events are alterations disrupting the PAX5 gene, occurring in around 1/3rd of pB-ALLs in general, and in almost every second case of the high-risk BCR-ABL-positive pB-ALL (27). Mostly, these mutations result in reduced Pax5 transcriptional activity, altering target gene expression (43), which highlights the fact that PAX5 deletions might contribute to leukemogenesis (27). In this setting, and as opposed to its role as susceptible gene in infection-exposure-driven pB-ALL (22), PAX5 deficiency seems to retain driver functions in established leukemia because it was previously shown that restoring endogenous Pax5 expression can trigger disease remission in mice (18). However, the critical question of how BCR-ABL^p190 and PAX5 loss contribute to pB-ALL leukemogenesis remained undefined. Here, we demonstrate that limited expression of the BCR-ABL^p190 oncogene within HS/PCs is capable of inducing pB-ALL, further suggesting that BCR-ABL^p190 favors the appearance of an aberrant precursor B-cell compartment in the BM. Significant reduction of Pax5 activity can be the critical factor in this experimental set up producing the selection pressure in favor of pB-ALL progression, since pB-ALL incidence drastically accelerated in Sca1-BCR-ABL^p190+Pax5^+/− mice with the accumulation of additional somatic Pax5 lesions in the remaining Pax5 WT allele. The emergence of these inactivating Pax5 variants could further be linked to disease onset, since amplicon sequencing of PB mononuclear cells taken at routine intervals confirmed their absence prior to first phenotypic signs of illness. This observation highlights the fact that Pax5 loss does not function as a susceptibility event in BCR-ABL^p190-pB-ALL development. In our model, the BCR-ABL^p190 oncogene primes HS/PCs for the acquisition of secondary Pax5 mutations/deletions which then act as the driving event in promoting clonal selection in pro-B/pre-B cells. Mechanistically, we show that the oncogenic capacity of reduced Pax5 transcriptional activity in the Sca1-BCR-ABL^p190 Pax5 deficient preleukemic clone is the result of a specific metabolic reprogramming characterized by elevated levels of glucose utilization and energy metabolism. The metabolic reprogramming is further shown by increased glycolytic capacity. Augmented glycolytic capacity indicates that Sca1-BCR-ABL^p190 Pax5-deficient ALL cells have an elevated ability to engage compensatory glycolysis to restore ATP levels when mitochondrial ATP production is inhibited. This metabolic rewiring is enhanced when Pax5 loss acts as a secondary event in the context of BCR-ABL^p190 susceptibility and is an early event recapitulated in healthy Sca1-BCR-ABL^p190 Pax5-deficient pro-B cells.

Transformation of leukemia cells imposes significant requirements for energy supply to support macromolecular synthesis for proliferation. Recent studies have identified GLUT1/3/6 (glucose transporters), HK2/3, PFKL, PGAM, PYGL and G6PD (glucose utilization) as PAX5 targets in pB-ALL (38). Using our mouse models, we identified novel PAX5 targets that are involved in glucose utilization as well as energy metabolism (IDH1, G6PC3, GAPDH, PGK1, MYC, ENO1 and ACO1) and showed that deficiency in Pax5 resulted in metabolic reprogramming towards glucose utilization and energy metabolism. Importantly, gene expression analysis showed that human pB-ALLs carry a similar metabolic signature, illustrating that metabolic reprogramming could be a vulnerability and potential therapeutic target in the treatment of pB-ALLs.

IKZF1 (IKAROS) is known to be a hallmark of Ph⁺ (44), as well as Ph-like ALL (45) and similar to PAX5, IKZF1 was shown to act as a metabolic gatekeeper, preventing malignant transformation (38). Although, we were not able to identify inactivating mutations/deletions in the here presented model, we observed downregulation of Ikzf1 in the expression profiles of Sca1-BCR-ABL^p190+Pax5^+/−pB-ALLs. Therefore, it will be interesting to further assess the pB-ALL incidence, mutational landscape and metabolic profile of Sca1-BCR-ABL^p190+Ikzf1^+/− mice as depicted here for Pax5.

Taken together, our results show that Pax5 is rate-limiting for BCR-ABL^p190 leukemogenesis. This supports a driver role for Pax5 deficiency in a preleukemic cell by determining to a large extent the clonal evolution of BCR-ABL^p190 pB-ALL development through metabolic reprogramming. Overall, the findings presented herein are important for encouraging novel interventions that might help to prevent conversion of BCR-ABL^p190 preleukemic cells.