Metabolic reprogramming by PRDM16 drives cytarabine resistance in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) patients with high PRDM16 expression frequently experience induction failure and have a poor prognosis. However, the molecular mechanisms underlying these clinical features remain elusive. We found that murine AML cells transformed by MLL::AF9 fusion and oncogenic short-isoform Prdm16 overexpression (hereafter, MF9/sPrdm16) exhibited resistance to cytarabine (AraC), but not to anthracycline, both in vitro and in vivo. Intriguingly, MF9/sPrdm16 cells displayed a gene expression signature of high oxidative phosphorylation (OxPHOS) and increased mitochondrial respiration. The inhibition of mitochondrial respiration with metformin or tigecycline abrogated AraC resistance in MF9/sPrdm16 cells via an energetic shift toward low OxPHOS status. Furthermore, sPrdm16 up-regulated Myc and the glutamine transporter Slc1a5, activating the TCA cycle and glutaminolysis. Of note, both OxPHOS and MYC-target gene signatures were significantly enriched in AML patient samples with high PRDM16 expression. Together, we showed that PRDM16 overexpression activates mitochondrial respiration through metabolic reprogramming via the MYC-SLC1A5-Glutaminolysis axis, thereby conferring AraC resistance on AML cells. These results suggest that targeting mitochondrial respiration might be a novel treatment strategy to overcome chemoresistance in AML patients with high PRDM16 expression.

Introduction

PRDM16 is one of the 16 members of PR/SET domain (PRDM) family,¹ which is critical for the function of hematopoietic stem cells (HSC).² In pediatric acute myeloid leukemia (AML), patients with high PRDM16 expression showed significantly worse overall survival (OS) and event-free survival (EFS) than those with low PRDM16 expression.³ Further analyses of clinical parameters have revealed that the percentage of patients who did not achieve complete remission (CR) after induction chemotherapy was significantly higher in those with high PRDM16 expression than in those with low PRDM16 expression.³ High PRDM16 expression is also observed in 31% of adult AML cases.⁴ Remarkably, high PRDM16 expression was frequent among adult AML patients who did not achieve CR after induction chemotherapy.⁴ Moreover, OS was significantly worse in those with high PRDM16 expression than in those with low PRDM16 expression.⁴ Several other studies have also demonstrated that high PRDM16 expression in adult AML predicts poor OS, relapse-free survival (RFS), and low CR rate after induction therapy.^5-7 These findings suggest that high PRDM16 expression is a pivotal clinical determinant of poor prognosis and induction failure in both pediatric and adult AML.

PRDM16 was originally isolated as a transcriptionally activated translocation partner gene in t(1;3)(p36;q21) AML.⁸ Analysis using Prdm16-deficient mice revealed that Prdm16 is a critical regulator of hematopoietic and neural stem cell maintenance, partly by regulating oxidative stress.⁹ Prdm16 is selectively expressed in the earliest HSC compartment and is crucial for the establishment and maintenance of the HSC pool.² Full-length PRDM16 (fPRDM16) is a zinc finger transcription factor which also possesses an N-terminal PR domain, homologous with a SET chromatin remodeling domain. Another isoform, short-isoform PRDM16 (sPRDM16), lacks the PR domain and this sPRDM16 was mainly detected in the t(1;3)(p36;q21)-positive AML and myelodysplastic syndromes (MDS).^10,11

Of note, elevated sPRDM16 expression due to promoter hypomethylation was frequently observed in cytogenetically normal (CN) AML.¹² The roles of these two distinct PRDM16 isoforms on myeloid leukemogenesis seem to be slightly different. Although both fPrdm16 and sPrdm16-transduced cells cause myeloid diseases in vivo, fPrdm16 requires longer disease latency than sPrdm16.¹³ In addition, histone H3 lysine 4 methyltransferase activity of the PR domain within fPrdm16 was shown to be required for specific suppression of MLL-rearranged leukemogenesis both in vitro and in vivo, indicating that fPRDM16 functions as a tumor suppressor in this context.¹⁴ On the other hand, a number of studies have demonstrated the apparent leukemogenic effect of sPRDM16. Overexpression of sPRDM16 blocked granulocyte colony-stimulating factor (G-CSF)-induced granulocytic differentiation in myeloid L-G3 cells, whereas fPRDM16 could not block the differentiation.¹⁰ Moreover, overexpression of sPrdm16 can co-operate with p53 loss or Hoxb4 overexpression to induce myeloid leukemia in vivo.^12,15 More recently, sPrdm16 was shown to transform megakaryocyte-erythroid progenitors into myeloid leukemia-initiating cells by activating myeloid gene regulatory networks.¹³ These studies indicate that sPRDM16 is a bona fide oncogene in myeloid leukemia. However, how this oncogene drives high-risk myeloid leukemia with chemotherapy resistance remains elusive.

Here, we performed cell viability, transcriptome, metabolome and in vivo analysis using murine leukemic cells and patient samples and explored response to chemotherapeutic agents used in standard induction therapy for AML in leukemic cells with high expression of sPRDM16. Our study uncovered unique sPRDM16-driven metabolic reprogramming, leading to chemoresistance of AML cells with high sPRDM16 expression.

Methods

Animals

C57BL/6JJmsSlc (CD45.2) or C57BL/6-Ly5.1 (CD45.1) mice were purchased from Japan SLC (Shizuoka, Japan) and Sankyo Laboratory Service Corporation (Tokyo, Japan), respectively. Flt3^ITD mice have been previously described.¹⁶ Mice were housed in a special pathogen-free environment at our institutional animal facility. All animal procedures were approved by the Institutional Animal Care Committees in Yokohama City University.

Patient samples and RNA-sequencing

We analyzed RNA-sequencing data obtained from 139 of all the pediatric AML patients enrolled in the Japanese Pediatric Leukemia/Lymphoma Study Group (JPLSG) AML-05 trial.¹⁷ All patients provided informed consent. These 139 patients were divided into two groups based on a PRDM16/ ABL1 expression ratio of 0.010 as the cutoff using RT-qPCR: 39 patients with high PRDM16 expression and 100 patients with low PRDM16 expression, as previously described.³ This research was conducted in accordance with the Declaration of Helsinki and approved by the institutional review boards of Yokohama City University, the participating institutes, and the ethical review board of the JPLSG AML-05 trial.

Results

High sPRDM16 expression drives cytarabine resistance in acute myeloid leukemia

To establish murine bone marrow (BM) and leukemic cells with high sPRDM16 expression, we transduced murine normal BM or MLL::AF9-transformed cells with either mock or sPrdm16-expressing retroviral vector to establish models for sPrdm16-low or high murine BM and MLL::AF9 cells (Figure 1A, B). We confirmed transforming capacity of sPrdm16 in normal BM cells in vitro as previously described (Online Supplementary Figure S1A-C).¹³ Of note, murine MLL::AF9 leukemia cells did not substantially change immunophenotype (Online Supplementary Figure S1D, E), cellular growth (Online Supplementary Figure S1F), and in vivo leukemogenicity (Online Supplementary Figure S1G, H) by ectopic sPrdm16 expression, suggesting non-tumor suppressive role of sPrdm16 in MLL::AF9 leukemia cells, which is in sharp contrast to fPRDM16.¹⁴ Using these cells as cellular models of low or high sPrdm16 expression, we first examined the sensitivity of these cells to either cytarabine (AraC) or daunorubicin by cell viability assays. Strikingly, MLL::AF9 leukemia cells with high sPrdm16 expression (MA9/sPrdm16 cells) showed significantly higher 50% inhibitory concentration (IC₅₀) against AraC compared to MLL::AF9 leukemia cells transduced with mock vector (MA9/control cells) (Figure 1C). In contrast, IC₅₀ against daunorubicin were comparable in these leukemia cells (Figure 1C). Importantly, murine normal BM cells with high sPrdm16 expression (WT/sPrdm16 cells) also showed higher IC₅₀ against AraC compared to murine normal BM cells transduced with mock vector (WT/control cells) (Online Supplementary Figure S2A). Consistent with cell viability assays, WT/sPrdm16 cells showed decreased sensitivity to AraC but not to daunorubicin compared to WT/control cells in colony forming unit (CFU) assays (Online Supplementary Figure S2B, C). To validate these findings in human leukemia cells, we transduced human AML cell lines, MOLM13 and THP-1, with either mock or sPRDM16-expressing retroviral vector. As expected, both MOLM13 and THP-1 cells with high sPRDM16 expression showed higher IC₅₀ against AraC compared to cells transduced with mock vectors (Figure 1D, E). Similar to murine leukemia cells, IC₅₀ against daunorubicin were comparable in these human leukemia cells (Figure 1D, E). AraC is an antimetabolic agent which is converted to cytarabine-5’-triphosphate after cellular uptake and incorporated into DNA during DNA synthesis, thereby inhibiting DNA polymerase and inducing cell cycle arrest in an S phase-specific manner.¹⁸ We, therefore, checked the percentage of S phase in MA9/sPrdm16 and MA9/control cells, hypothesizing that lower percentage of S phase might have conferred AraC resistance in MA9/ sPrdm16 cells. Intriguingly, however, MA9/sPrdm16 cells showed a significantly higher percentage of S/G2/M fraction compared to MA9/control cells (Figure 1F). Likewise, WT/ sPrdm16 cells also showed a higher percentage of S/G₂/M fraction compared to WT/control cells (Online Supplementary Figure S2D). Of note, the percentage of S/G₂/M fraction was higher in AraC-treated WT/sPrdm16 cells compared to vehicle-treated WT/sPrdm16 cells (Online Supplementary Figure S2E). These data suggest that sPrdm16 overexpression drives AraC resistance of MA9/sPrdm16 and WT/sPrdm16 cells independently of S phase. We further examined the effect of high sPrdm16 expression on drug sensitivity using a murine Flt3^ITD model.¹⁶ In line with the murine MLL::AF9 model, murine Flt3^ITD BM cells with high sPrdm16 expression (Flt3^ITD/sPrdm16) also showed relative resistance to AraC, but not to daunorubicin, compared to control Flt3^ITD cells (Flt3^ITD/control) in both cell viability and CFU assays (Online Supplementary Figure S2F-H). In addition, Flt3^ITD/sPrdm16 cells and AraC-treated Flt3^ITD/sPrdm16 cells showed a higher percentage of S/G2/M fraction compared to Flt3^ITD/control cells or vehicle-treated Flt3^ITD/control cells, respectively, again indicating that high sPrdm16 expression drives AraC resistance independently of S phase (Online Supplementary Figure S2 I-J).

Figure 1.sPRDM16 induces cytarabine resistance in acute myeloid leukemia cells in a cell cycle-independent manner. (A) Experimental schema for the establishment of MA9/control or MA9/sPrdm16 cells. (B) Protein expression levels of sPrdm16, detected by western blotting analysis, were up-regulated in MA9/sPrdm16 cells. (C) In vitro drug sensitivity assay for cytarabine (upper panel) or daunorubicin (lower panel) using MA9/control and MA9/sPrdm16 (N=3 for each group). (D, E) Drug sensitivity assay using human acute myeloid leukemia (AML) cell lines, MOLM-13 (D) and THP-1 (E), transduced with either mock or sPRDM16-expressing vector (N=3 for each group). (F) Cell cycle analysis of MA9/control and MA9/sPrdm16 cells (N=3 for each group). Representative flow cytometry images (left panels) and percentages of each phase of the cell cycle (right panel) are shown. MA9/sPrdm16 cells exhibited an increased S/G2/M phase, which typically exhibits high sensitivity to cytarabine. All data are represented as mean ± standard deviation. An unpaired Student t test was used to calculate P values. N.S.: not significant; *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001; *****P<0.00001.

High sPrdm16 expression activates oxidative phosphorylation

To explore the molecular basis of AraC resistance driven by sPrdm16 expression, we next performed RNA-sequence (RNA-seq) using MA9/control and MA9/sPrdm16 cells and compared their transcriptome. Notably, 7,658 genes (96.5% of differentially expressed genes) were up-regulated in MA9/sPrdm16 cells compared to MA9/control cells (Figure 2A). Gene Ontology (GO) analysis revealed that “cell cycle”-, “cell division”- or “DNA replication”-related genes were enriched in the up-regulated genes (Online Supplementary Figure S3A). Gene set enrichment analysis (GSEA) revealed that the gene signature related to “oxidative phosphorylation (OXPHOS)” was strongly enriched in MA9/ sPrdm16 cells compared to MA9/control cells (Figure 2B, C). We also performed RNA-seq using WT/control and WT/ sPrdm16 cells. Similar to RNA-seq data derived from an MLL::AF9 leukemia model, the majority of differentially expressed genes (1,015 genes, 74.0%) were up-regulated in WT/sPrdm16 cells compared to WT/control cells (Online Supplementary Figure S3B). GO analysis confirmed enrichment of “cell cycle”-, “cell division”- or “DNA replication”-related genes in relatively up-regulated genes in WT/sPrdm16 cells (Online Supplementary Figure S3C). Importantly, GSEA showed that the gene signature related to “OXPHOS” was also significantly enriched in WT/ sPrdm16 cells compared to WT/control cells (Online Supplementary Figure S3D, E), suggesting that high sPrdm16 expression drives OXPHOS-related transcriptional output. To confirm if mitochondrial respiration is elevated in cells with high sPrdm16 expression, we conducted the Mito Stress Test and the Glycolysis Stress Test using MA9/ control and MA9/sPrdm16 cells. Consistent with the GSEA results, MA9/sPrdm16 cells showed significantly higher basal and maximum mitochondrial respiration, and higher glycolytic function compared to MA9/control cells in the Mito Stress and the Glycolysis Stress Test, respectively (Figure 2D). To further evaluate mitochondrial reactive oxygen species (ROS) and membrane potential of these leukemic cells, we labeled cellular mitochondria with either MitoSOX, Mitotracker or tetramethylrhodamine methyl ester (TMRM) dye and quantified their fluorescence with flow cytometry. In line with the results of the GSEA and the Mito Stress Test, MA9/sPrdm16 cells exhibited significantly higher median fluorescence intensity (MFI) of MitoSOX and relative TMRM compared to MA9/control cells (Figure 2E). Likewise, the Mito Stress Test of WT/sPrdm16 cells showed significantly higher basal and maximum mitochondrial respiration and higher glycolytic function compared to WT/control cells (Online Supplementary Figure S3F). Taken together, these data suggest that high sPrdm16 expression in both murine normal and leukemic BM cells drive mitochondrial respiration to activate OXPHOS-related transcriptional output.

Pharmacological inhibition of mitochondrial respiration sensitizes MA9/sPrdm16 cells to cytarabine

A previous study has shown that AraC-resistant human AML cells critically require oxidative metabolism.¹⁹ To determine if activated mitochondrial respiration driven by high sPrdm16 expression is responsible for AraC resistance in AML cells with high sPrdm16, we examined the effect of metformin and tigecycline, electron transport chain complex I inhibitor and mitochondrial protein synthesis inhibitor, respectively, on AraC resistance induced by sPrdm16 overexpression. Cell viability assays using MA9/ control and MA9/sPrdm16 cells revealed that addition of 2 mM metformin or 10 mM tigecycline decreased the IC₅₀ against AraC in MA9/sPrdm16 cells to the same level as those in MA9/control (Figure 3A, B). IACS-010759, a selective mitochondrial complex I inhibitor, also sensitized MA9/sPrdm16 cells to AraC (Figure 3C). We next validated if metformin, tigecycline or IACS-010759 treatment reduces mitochondrial respiration in MA9/sPrdm16 leukemia cells. Treatment with either 1 mM metformin alone or a combination of AraC and metformin significantly reduced mitochondrial respiration (Online Supplementary Figure S4A). Similar results were also observed using tigecycline and IACS-010759 (Online Supplementary Figure S4B, C). Collectively, these data indicate that activated mitochondrial respiration driven by high sPrdm16 expression is directly responsible for AraC resistance in AML cells with high sPrdm16 and that pharmacological inhibition of this metabolic pathway can reverse their AraC resistance.

Figure 2.sPrdm16 activates oxidative phosphorylation. (A) Differentially expressed genes in MA9/sPrdm16 cells compared to MA9/ control cells. There were 7,658 up-regulated genes and 281 down-regulated genes. (B, C) Top 15 gene signatures significantly changed in gene set enrichment analysis (GSEA) using RNA-seq data from MA9/sPrdm16 versus MA9/control cells (B). The gene signature related to “OXIDATIVE PHOSPHORYLATION” was the most significantly enriched in MA9/sPrdm16 cells (B, C). (D) Mito Stress Test (left panel) and Glycolysis Stress Test (right panel) in MA9/control and MA9/sPrdm16 cells (N=3 for each group). (E) Mitochondrial reactive oxygen species (ROS) and relative membrane potential in MA9/ sPrdm16 cells were measured using MitoSOX and the ratio of tetramethylrhodamine methyl ester (TMRM) and mitotracker, respectively (N=3 for each group). All data are represented as mean ± standard deviation. An unpaired Student t test was used to calculate P values. **P<0.01; ***P<0.001; ***P<0.0001. MFI: mean fluorescence intensity.

Myc activation and Slc1a5-mediated glutaminolysis drive metabolic reprogramming in acute myeloid leukemia with high sPrdm16 expression

To explore the rationale for activated OXPHOS in AML cells with high sPrdm16 expression, we performed capillary electrophoresis time-of-flight mass spectrometry-based metabolome profiling of MA9/control and MA9/sPrdm16 leukemia cells. We found significantly higher amounts of glutamine and trends toward higher amounts of succinate and fumarate in MA9/sPrdm16 cells (Figure 4A). To further seek metabolic profiles of AraC-exposed leukemia cells, we next profiled the metabolic status of AraC-treated MA9/control and MA9/sPrdm16 leukemia cells. Notably, AraC-exposed MA9/sPrdm16 cells showed significantly higher amounts of glutamine, α-ketoglutarate, and malate, and trends toward higher succinate and fumarate compared to AraC-exposed MA9/control cells (Online Supplementary Figure S5A). These data underscore activated glutamine uptake, glutaminolysis and subsequent reprogramming of the tricarboxylic acid (TCA) cycle in non-treated and AraC-exposed MA9/sPrdm16 leukemia cells.

Figure 3.Pharmacological inhibition of mitochondrial respiration abrogated cytarabine resistance in MA9/sPrdm16 cells. (A, B) In vitro drug sensitivity assay for cytarabine with vehicle (left panel) or metformin (right panel), an electron transport chain complex I inhibitor (A), and with vehicle (left panel) or tigecycline (right panel), a mitochondrial protein synthesis inhibitor (B). Metformin and tigecycline abrogated cytarabine resistance in MA9/sPrdm16 cells (N=3 for each group). (C) In vitro drug sensitivity assay for cytarabine with vehicle (left panel) or IACS-010759 (right panel), a selective mitochondrial complex I inhibitor. IACS-010759 also abrogated cytarabine resistance in MA9/sPrdm16 cells (N=3 for each group). All data are represented as mean ± standard deviation. An unpaired Student t test was used to calculate P values. N.S.: not significant; ^*P<0.05; ^**P<0.01; ^***P<0.001; ^****P<0.0001.

Figure 4.sPrdm16 induces metabolic reprogramming through the activation of glutaminolysis via the Myc-Slc1a5 axis. (A) Metabolite concentrations in the TCA cycle of MA9/control or MA9/sPrdm16 cells were measured using capillary electrophoresis time-of-flight mass spectrometry-based metabolome profiling (N=6 for each group). (B) Intracellular glutamine (left panel) and glutamate uptake (right panel) levels, measured by Glutamine/Glutamate-Glo Assay, were higher in MA9/sPrdm16 cells compared to control (N=3 for each group). (C) Transcripts per million (TPM) values of Slc1a5 gene were up-regulated in MA9/sPrdm16 cells (N=3 for each group). (D) Protein expression levels of Slc1a5, detected by western blotting analysis, were also up-regulated in MA9/sPrdm16 cells. (E) The gene signatures related to “MYC-TARGET” were significantly enriched in MA9/sPrdm16 cells. (F) TPM values of Myc gene were up-regulated in MA9/sPrdm16 cells (N=3 for each group). (G) Protein expression levels of Myc, detected by western blotting analysis, were also up-regulated in MA9/sPrdm16 cells. All data are represented as mean ± standard deviation. An unpaired Student t test was used to calculate P values. NS: not significant; ^*P<0.05; ^**P<0.01; ^***P<0.001; ^*****P<0.00001.

Glutaminase is a mitochondrial enzyme crucial for glutaminolysis which deaminates glutamine to form glutamate.²⁰ To explore whether activated glutaminolysis is a functional hub driving AraC resistance in MA9/sPrdm16 leukemia cells, we treated MA9/control or MA9/sPrdm16 leukemia cells with various concentrations of AraC with BPTES, a glutaminase inhibitor, and performed cell viability assays. Addition of BPTES sensitized MA9/sPrdm16 cells to AraC to the same level as MA9/control cells, indicating that glutaminase-mediated glutaminolysis contributes to AraC resistance in MA9/sPrdm16 cells (Online Supplementary Figure S5B). To accurately quantify intracellular glutamine and glutamate levels, we applied a luminescence-based Glutamine/Glutamate-Glo Assay to MA9/control and MA9/ sPrdm16 leukemia cells. In agreement with the metabolome data, we found higher amounts of glutamine and glutamate in MA9/sPrdm16 cells compared to MA9/control cells, suggesting that glutamine uptake and subsequent glutamate synthesis via glutaminolysis are relatively active in MA9/ sPrdm16 cells (Figure 4B).

A recent study has shown that MYC-induced transcription of the neutral amino acid transporter SLC1A5 promotes intracellular uptake of glutamine and subsequent TCA cycle/OXPHOS activity in leukemia stem cells (LSC) of human AML for energy metabolism and survival.²¹ Therefore, we hypothesized that MA9/sPrdm16 leukemia cells also drive the Myc-Slc1a5 axis to promote glutamine uptake, fuel TCA cycle metabolites and subsequently activate OXPHOS. In line with this notion, we observed significant upregulation of Slc1a5 in MA9/sPrdm16 cells compared to MA9/control cells, at both transcriptional and protein levels (Figure 4C, D). Importantly, GSEA based on RNA-seq data of MA9/control and MA9/sPrdm16 cells revealed significant enrichment of “MYC-TARGET” gene signatures in MA9/sPrdm16 cells (Figure 4E). Moreover, Myc was significantly up-regulated in MA9/sPrdm16 cells compared to MA9/control cells, at both transcriptional and protein levels (Figure 4F, G). Of note, GSEA based on RNA-seq of WT/control and WT/sPrdm16 cells also showed significant enrichment of “MYC-TARGET” gene signatures in WT/sPrdm16 cells (Online Supplementary Figure S5C). The transcription of Slc1a5 and Myc were significantly higher in WT/sPrdm16 cells compared to WT/control cells (Online Supplementary Figure S5D). Next, to explore if MYC-SLC1A5 axis-mediated metabolic reprogramming is also relevant in human AML cells with high sPRDM16 expression, we checked MYC and SLC1A5 expression in MOLM13 and THP-1 cells transduced with either mock or sPRDM16-expressing retroviral vector. Consistent with murine data, immunoblot analysis showed higher MYC and SLC1A5 expression in MOLM13/sPRDM16 and THP-1/sPRDM16 cells compared to control cells (Online Supplementary Figure S5E). Glutamine/Glutamate-Glo Assay revealed higher amounts of glutamine and glutamate in MOLM13/sPRDM16 and THP-1/sPRDM16 cells compared to control cells (Online Supplementary Figure S5F, G). Above all, the Mito Stress Test confirmed significantly higher basal and maximal respiration in MOLM13/sPRDM16 and THP-1/ sPRDM16 cells compared to control cells (Online Supplementary Figure S5H, I). Together, these data suggest that high sPrdm16 expression in murine and human AML cells drives Myc-Slc1a5 axis to promote glutamine uptake, fuel TCA cycle metabolites, and subsequently activate OXPHOS, which is a critical determinant of sPrdm16-mediated AraC resistance.

Myc is a functional hub of sPrdm16-mediated metabolic reprogramming and cytarabine resistance

Next, we sought to examine if Myc is functional in sPrdm16-mediated metabolic reprogramming and AraC resistance. To this end, we knocked down Myc using retroviral shRNA vectors in MA9/sPrdm16 cells and tested for a cell viability assay (Figure 5A). Consistent with our hypothesis, knockdown of Myc significantly sensitized MA9/ sPrdm16 cells to AraC (Figure 5B). In addition, the Glutamine/Glutamate Glo assay detected significantly lower amounts of glutamine and glutamate in Myc knockeddown MA9/sPrdm16 cells compared to Scramble MA9/ sPrdm16 cells (Online Supplementary Figure S6A). The Mito Stress Test revealed reduced basal and maximum mitochondrial respiration upon Myc knockdown in MA9/ sPrdm16 cells (Online Supplementary Figure S6B). To further confirm the role of Myc in sPrdm16-mediated AraC resistance, we performed a cell viability assay using an Myc inhibitor, 10058-F4. In MA9/control cells, addition of 10058-F4 did not sensitize leukemia cells to AraC (Figure 5C). In sharp contrast, addition of 10058-F4 significantly sensitized MA9/sPrdm16 cells to AraC (Figure 5C). Furthermore, Synergy Finder detected a significantly high synergy score between AraC and 10058-F4 in MA9/sPrdm16 cells but not in MA9/control cells, indicating that Myc inhibition synergizes with AraC to kill leukemia cell specifically in MA9/sPrdm16 cells (Online Supplementary Figure S6C). We also conducted a reciprocal experiment to determine if MLL::AF9 leukemia cells with Myc overexpression mimic the phenotype of MF9/sPrdm16 cells such as resistance to AraC and high OXPHOS status (Figure 5D, E). As expected, MLL::AF9 leukemia cells with Myc overexpression (MA9/ Myc cells) showed significantly higher IC₅₀ against AraC compared to MA9/control cells, confirming that Myc over-expression drives AraC resistance in MLL::AF9 leukemia cells (Figure 5F). Myc-induced AraC resistance was also associated with high amounts of glutamine and glutamate (Online Supplementary Figure S6D), high basal and maximum mitochondrial respiration (Online Supplementary Figure S6E), and high mitochondrial ROS and relative membrane potential (Online Supplementary Figure S6F). Overall, these data demonstrate that Myc does, indeed, function as a cornerstone of sPrdm16-mediated metabolic reprogramming and AraC resistance.

sPrdm16 overexpression in acute myeloid leukemia leads to cytarabine resistance in vivo

To validate if AML cells with high sPrdm16 expression exert AraC resistance in vivo, we transplanted MA9/control or MA9/sPrdm16 leukemia cells into sub-lethally irradiated recipient mice, treated them with either vehicle or AraC via intraperitoneal injection, and compared survival. AraC treatment prolonged survival of MA9/control mice (Figure 6). On the other hand, AraC treatment failed to improve survival of MA9/sPrdm16 mice, suggesting that sPrdm16 overexpression directly drives AraC resistance in AML in vivo (Figure 6).

Figure 5.Myc is a key regulator of sPrdm16-mediated resistance to cytarabine. (A) Relative Myc expression in MA9/sPrdm16 cells with Myc knockdown (N=3 for each group). (B) In vitro drug sensitivity assay for cytarabine in MA9/sPrdm16 cells with Myc knockdown (N=3 for each group). (C) In vitro drug sensitivity assay for cytarabine with vehicle or 10058-F4, a Myc inhibitor, in MA9/ control or MA9/sPrdm16 cells (N=3 for each group). In contrast to MA9/control cells, the combination of cytarabine and 10058-F4 reduced the IC₅₀ value of cytarabine in MA9/sPrdm16 cells. (D) Relative Myc expression in MLL::AF9 leukemic cells transduced with Myc-expressing vector (MA9/Myc) (N=3 for each group). (E) Protein expression levels of Myc, detected by western blotting analysis, were also up-regulated in MA9/Myc cells. (F) In vitro drug sensitivity assay for cytarabine in MA9/Myc cells (N=3 for each group). MA9/Myc cells mimicked the phenotype of MA9/sPrdm16 cells as evidenced by the resistance to cytarabine. All data are represented as mean ± standard deviation. An unpaired Student t test was used to calculate P values. NS: not significant; ^*P<0.05; ^**P<0.01; ^***P<0.001; ^****P<0.0001.

Acute myeloid leukemia patient samples with high PRDM16 expression display activated “OXPHOS” and “MYC Targets” gene expression signatures

Finally, we aimed to clarify if AML patients with high PRDM16 expression also present evidence of MYC-mediated metabolic reprogramming. We analyzed RNA-seq data derived from 139 pediatric AML patient samples registered in the Japanese Pediatric Leukemia/Lymphoma Study Group (JPLSG) AML-05 clinical trial and classified into PRDM16 low and high groups (Figure 7A). Strikingly, GSEA analysis revealed that the gene signatures related to “OXPHOS” and “MYC Targets” were significantly enriched in PRDM16-high samples compared to PRDM16-low samples (Figure 7B, C). Taken together, these data indicate that MYC-mediated metabolic reprogramming is a common transcriptional hallmark in both murine and human AML cells with high PRDM16 expression, which is a functional determinant of PRDM16-mediated AraC resistance.

Discussion

PRDM16 is one of the 16 members of PR/SET domain (PRDM) family.¹ sPRDM16, which lacks the PR domain, is mainly detected in t(1;3)(p36;q21)-positive AML and MDS.^10-12 In addition, high sPRDM16 expression is also observed in cytogenetically normal (CN) AML.¹² Likewise, a gene encoding a short isoform of PRDM3 lacking the PR domain, known as Ecotropic Viral Integration site 1 (EVI1), is also highly expressed in high-risk MDS/AML with inv(3)/t(3;3).¹ Altogether, PRDM16 is one of the two PRDM family members highly expressed in a subset of MDS/AML patients. Notably, there is a growing body of evidence to suggest that high PRDM16 expression is a transcriptional determinant of poor OS, RFS, and low CR rate after induction therapy both in pediatric and adult AML.^3-7 Importantly, a number of studies have shown leukemogenic effects of sPRDM16.^10,12,13,15 Based on these observations, we hypothesized that PRDM16 functions as an oncogene, which drives chemoresistance in AML.

In this study, we have shown that overexpression of sPRDM16 drives resistance to AraC but not to daunorubicin, the two major chemotherapeutic agents used in standard induction therapy for AML, in murine and human leukemic cells in vitro and in vivo. AraC is an antimetabolic agent which is incorporated into DNA during DNA synthesis, inhibits DNA polymerase, and induces cell cycle arrest in an S phase-specific manner.¹⁸ Paradoxically, murine BM cells and MLL::AF9 leukemia cells with high sPrdm16 expression showed a higher percentage of S/G₂/M fraction compared to control cells, suggesting that sPrdm16 drives AraC resistance in an S phase-independent manner. GSEA based on RNA-seq data revealed that both MA9/sPrdm16 and WT/ sPrdm16 cells showed enrichment of OXPHOS-related gene signature. The Mito Stress Test exhibited higher basal and maximum mitochondrial respiration in both MA9/sPrdm16 and WT/sPrdm16 cells compared to control cells. Moreover, metabolome profiling demonstrated activated glutamine uptake, glutaminolysis and reprogramming of the TCA cycle in both untreated and AraC-exposed MA9/sPrdm16 leukemia cells. Together, these data strongly denote activated mitochondrial oxidative metabolism in these cells with high sPrdm16 expression. Previous studies have shown a direct link between mitochondrial oxidative metabolism and AraC-resistance in AML. The ectonucleotidase CD39 (ENTPD1) is reported to be up-regulated in AraC-resistant leukemic cells from both AML cell lines and patient samples in vivo and in vitro.²² High CD39 activity was shown to promote AraC resistance by enhancing mitochondrial activity and biogenesis through activation of cAMP-mediated adaptive mitochondrial stress response.²² More recently, expression profiling of circular RNA, together with functional assays, revealed that downregulation of circFAM193B enhances OXPHOS, inhibits the lipid peroxidation of LSC, and promotes AraC resistance in AML.²³ In fact, low circFAM193B expression was significantly associated with poor survival in AML patients.²³ Of note, attenuation of mitochondrial OXPHOS by either metformin or tigecycline dramatically reversed AraC resistance of leukemic cells with high sPrdm16 expression. Collectively, these data clearly indicate that PRDM16 drives AraC resistance via activated mitochondrial respiration.

Figure 6.MA9/sPrdm16 mice are resistant to cytarabine. Kaplan-Meier survival curve of MA9/control (left panel) and MA9/sPrdm16 (right panel) mice treated with either vehicle (N=10-11) or cytarabine (N=11). 1.0x10⁴ leukemia cells were transplanted to each sublethally irradiated (6.0 Gy) recipient mouse. MA9/sPrdm16 mice exhibited resistance to cytarabine. The log rank test was used in survival analysis. IP: intraperitoneal injection.

Molecular mechanisms by which cancer cells activate mitochondrial respiration have been explored. Mutations driving constitutive activation of phosphatidylinositol-3 kinase (PI3K)/AKT signaling are among the most common mutations in human cancer. Growth factor signaling through PI3K/AKT drives membrane localization of transporters required for the uptake of critical nutrients such as glucose, amino acids and iron.^24-26 Additionally, oncogenic transcription factors such as MYC and hypoxia inducible factor (HIF) can also activate nutrient acquisition by increasing the expression of genes involved in the cellular uptake of critical nutrients such as glucose and glutamine.^27,28 Activated cellular uptake of these nutrients can ultimately fuel the TCA cycle, leading to metabolic reprogramming resulting in upregulation of mitochondrial respiration in cancer cells. In agreement with this, human AML LSC utilize pro-survival mechanisms where oncogenic transcription factor MYC induces transcription of amino acid transporter SLC1A5 to promote cellular uptake of glutamine, fuel the TCA cycle, and activate OXPHOS activity.²¹ Here, we demonstrate that both murine BM cells and MLL::AF9 leukemia cells with high sPrdm16 expression showed enrichment of “MYC-TARGET” gene signatures, upregulation of Slc1a5 transcription and glutamine uptake. Addition of glutaminase inhibitor BPTES neutralized sPrdm16-mediated AraC resistance in MLL::AF9 leukemia cells. MYC-SLC1A5 axis-mediated metabolic reprogramming was also relevant in human AML cells with high sPRDM16 expression. Knockdown or pharmacological inhibition of Myc also reversed sPrdm16-mediated AraC resistance in murine MLL::AF9 leukemia cells. Most importantly, transcriptomic analysis of pediatric AML patient samples with high PRDM16 expression identified significant enrichment for OXPHOS and MYC target signatures, which is adherent to the murine data.

Figure 7.Primary acute myeloid leukemia patient samples with high PRDM16 expression exhibit gene expression signatures associated with “OXPHOS” and “MYC-Targets”. (A) Relative total PRDM16 expression (including both full-length and short isoform) in acute myeloid leukemia (AML) patients with low (left, N=100) or high (right, N=39) PRDM16 expression. (B) Gene signatures significantly changed in gene set enrichment assay (GSEA) using RNA-seq data from 139 pediatric AML patient samples registered in the Japanese Pediatric Leukemia/Lymphoma Study Group (JPLSG) AML-05 clinical trial. (C) The gene signatures related to “OXIDATIVE PHOSPHORYLATION” (left panel) and “MYC-TARGET” (middle and right panels) were significantly enriched in AML samples with high PRDM16 expression. (D) Proposed model of cytarabine resistance mechanisms in AML with high Prdm16 expression. PRDM16 drives cytarabine resistance through upregulation of mitochondrial respiration via upregulation of MYC and glutamine transporter SLC1A5, activation of glutamine uptake, and glutaminolysis. Therefore, targeting mitochondrial respiration might be a novel treatment strategy to overcome cytarabine resistance in AML patients with high PRDM16 expression. Statistical analyses were performed using the Mann-Whitney U test. ^****P<0.0001.

Although multiple functional studies have shown an oncogenic role of sPrdm16 in myeloid leukemia, the precise biological effect of sPrdm16 on hematopoietic cells is pleiotropic.^10,12,13,15 Nishikata et al. have shown that sPrdm16 blocks G-CSF-induced myeloid differentiation in murine myeloid L-G3 cells.¹⁰ sPrdm16-overexpressing murine p53^-/- BM cells, Hoxb4-co-expressing murine BM cells and murine megakaryocyte-erythroid progenitors (MEP) can develop AML in recipient mice in vivo.^12,13,15 sPrdm16-overexpressing murine p53^-/- BM cells also showed increased clonogenicity in CFU assays.¹² In this study, we confirmed MYC-SLC1A5 axis-mediated metabolic reprogramming as a molecular basis for sPRDM16-induced Ara-C resistance in multiple cell types, including WT murine BM cells, MA9 murine leukemia cells, human AML cell lines, and primary AML patient samples. On the other hand, growth curve analysis showed comparable cell growth in MA9/sPrdm16 cells compared to MA9/ control cells (Online Supplementary Figure 1F). In addition, the untreated MA9/sPrdm16 cell-bearing mice lived longer than the untreated MA9/control cell-bearing mice when we injected 1.0x10⁴ leukemia cells to sublethally irradiated (6.0 Gy) recipients (Figure 6). The difference in OS between these two groups diminished when we injected a greater number of leukemia cells (5.0x10⁴ cells) to sublethally irradiated (6.0 Gy) recipients (Online Supplementary Figure 1G). These data suggest that sPrdm16-induced Ara-C resistance driven by metabolic reprogramming may be a common biological effect shared by a wide spectrum of normal hematopoietic and myeloid leukemia cells, whereas effect of sPrdm16 on cell growth and leukemogenicity may be context dependent, such as cell type or genetic background of leukemia cells. Taken together, these data suggest that PRDM16 drives AraC resistance through MYC-induced upregulation of the glutamine transporter SLC1A5, activation of glutamine uptake, glutaminolysis and subsequent mitochondrial oxidative metabolism (Figure 7D). This is further supported by the recent study showing that EVI1, another PRDM family member, also triggers metabolic reprogramming toward activated glutaminolysis and OXPHOS status in AML for leukemia cell survival, sensitizing EVI1-high AML cells to L-asparaginase which depletes intracellular asparagine and glutamine.²⁹ Our study revealed MYC-glutaminolysis-mitochondrial OXPHOS pathway to be a fundamental basis for PRDM16-driven AraC resistance, providing a rationale for mitochondrial respiration-targeted therapies against PRDM16-mediated chemoresistance which should be further clarified in future clinical studies.

Footnotes

• Received December 24, 2024
• Accepted May 23, 2025

Correspondence

H. Kunimoto kunimoto@yokohama-cu.ac.jp

H. Nakajima hnakajim@yokohama-cu.ac.jp

Disclosures

HK is a recipient of research funding from Daiichi-Sankyo. HN is a recipient of research funding from Takeda, Astellas, Eisai, Pfizer and Novartis. All of the other authors have no conflicts of interest to disclose.

Contributions

JI, HK, YS, KM, IK, MFT and HN are responsible for study concept and methodology. JI, YS, AM, TK, KM, IK and NS are responsible for validation, formal analysis, investigation and data curation. YS, ST, MT, MTF, AY and TT are responsible for resources. JI and HK wrote the original draft. JI, HK and HN wrote, reviewed and edited the paper. JI, HK and YS are responsible for visualization. HK and HN are responsible for funding acquisition, and project supervision and administration.

Funding

This investigation was supported by the Japanese Society for the Promotion of Science (JSPS) KAKENHI (JP19K17864, JP21K08400, JP24K11565), the Yokohama Foundation for Advancement of Medical Science Research Grant, the Japanese Society of Hematology Research Grant, the Kowa Life Science Foundation Research Grant, the Koyanagi Foundation Research Grant, the Mother and Child Health Foundation Research Grant, the Naito Foundation Research Grant, the Astellas Foundation for Research on Metabolic Disorders Research Grant, the Uehara Memorial Foundation Research Grant, the Princess Takamatsu Cancer Research Fund, the Nippon Shinyaku Public Research Grant, the American Society of Hematology Global Research Award, the Takeda Science Foundation Medical Research Grant, the Kato Memorial Bioscience Foundation Research Grant, the Mochida Memorial Foundation for Medical and Pharmaceutical Research Grant, the Japan Research Foundation for Clinical Pharmacology Research Grant.

Acknowledgments

We thank Akiko Adachi, Mitsuyo Tamaki, Masahiro Yoshitomi, Hiromi Soeda for technical support. We thank Yokohama City University Advanced Medical Research Center for flow cytometry services. HK is supported by the Japanese Society for the Promotion of Science (JSPS) KAKENHI (JP19K17864, JP21K08400, JP24K11565), the Yokohama Foundation for Advancement of Medical Science Research Grant, the Japanese Society of Hematology Research Grant, the Kowa Life Science Foundation Research Grant, the Koyanagi Foundation Research Grant, the Mother and Child Health Foundation Research Grant, the Naito Foundation Research Grant, the Astellas Foundation for Research on Metabolic Disorders Research Grant, the Uehara Memorial Foundation Research Grant, the Princess Takamatsu Cancer Research Fund, the Nippon Shinyaku Public Research Grant, the American Society of Hematology Global Research Award, the Takeda Science Foundation Medical Research Grant, the Kato Memorial Bioscience Foundation Research Grant, the Mochida Memorial Foundation for Medical and Pharmaceutical Research Grant, the Japan Research Foundation for Clinical Pharmacology Research Grant. ST is supported by the JSPS KAKENHI (JP21K15870, JP24K11004). HN is supported by the JSPS KAKENHI (JP20H03714, JP23H02939, JP23K27630).