The 6133 cells can be transplanted into recipient mice to induce low penetrant leukemia with symptoms closely recapitulating human AMKL (Mercher et al., 2009). To find additional factors needed to transform 6133 cells fully, we have reported a fluorescent probe (E84) that can be used to sort live cells according to PRMT1 protein concentrations (Su et al., 2018; Figure 1—figure supplement 1). We sorted 6133 cells into two populations for bone marrow transplantation (BMT) according to E84 staining intensities (Figure 1A). All mice that received 6133 cells expressing higher levels of PRMT1 (6133/PRMT1 cells) developed leukemia and died rapidly, while 6133 cells expressing lower levels of PRMT1 did not develop leukemia (Figure 1B). Consistently, a higher percentage of leukemia cells were detected in bone marrow and peripheral blood in recipient mice transplanted with E84-high (i.e., density staining of E84) 6133 cells according to fluorescence-activated cell sorting (FACS) analysis (Figure 1C).

Given that E84-high cells can initiate leukemia, we introduced PRMT1 into 6133 cells (aka 6133/PRMT1 cells) using a lentivirus vector. Overexpression of PRMT1 rendered 6133/PRMT1 cells to increase in a cytokine-independent fashion in cell culture (Figure 1D), and recipient mice transplanted with 6133/PRMT1 cells developed leukemia and died within 25 days (Figure 1E, F). Although PRMT1-mediated methylation triggers the degradation of RBM15, PRMT1 overexpression does not affect the stability of the RBM15-MKL1 fusion (Figure 2—figure supplement 1A and B). The leukemic mice displayed splenomegaly. Intriguingly, the leukemia mice were paralyzed with observable spinal bleeding during dissection, although the bone density was still normal (data not shown). In a xenograft model with human RBM15-MKL1 leukemia cells, leukemia also caused spinal bleeding (Thiollier et al., 2012). Collectively, PRMT1 is essential for the initiation of overt AMKL expressing the RBM15-MKL1 fusion protein, and leukemia initiated by 6133/PRMT1 cells in mice resembles several characteristics of human AMKL.

MS023 was reported to be a potent inhibitor of Type-I PRMTs including PRMT1 (Eram et al., 2016; Su et al., 2021). Furthermore, MS023 has been tested safe on mice at 80 mg/kg of body weight (He et al., 2019; Su et al., 2021). The 6133/PRMT1 cells were intravenously injected into sub-lethally irradiated recipient mice. A week after BMT, we injected MS023 intraperitoneally every other day for a month. Notably, while the untreated group of mice exhibited rapid illness and developed moribund symptoms, such as severe weight loss and rear limb paralysis, within 30 days, the majority of mice treated with MS023 remained healthy and symptom-free. Treatment with MS023 significantly alleviated the leukemia burden, as shown by FACS analysis of the percentages of leukemia cells in bone marrow and peripheral blood (Figure 2D). We also validated that the bone marrow cells from MS023-treated mice had reduced global levels of arginine methylation (Figure 2—figure supplement 1C). Leukemia-associated splenomegaly was also alleviated in MS023-treated mice (Figure 2D). Kaplan–Meier curves showed that the MS023-treated group was effectively cured after 120 days (Figure 2C), with only residual leukemia cells detected in bone marrow and peripheral blood (Figure 2D, E). Expression of the c-MPLW515L mutant in 6133 cells can render the 6133 cells fully penetrant for leukemia (Mercher et al., 2009), and in vitro treatment with MS023 reduced their proliferation (Figure 2F). Subsequently, we transplanted 6133/cMPLW515L cells into congenic mice. Figure 2G–I illustrates that MS023 also cures leukemia, suggesting that PRMT1 could be a valid target for leukemia or myeloid proliferative diseases driven by c-MPLW515L. Collectively, these data demonstrate that PRMT1 is critical for sustaining Rbm15-MKL1-initiated leukemia in mice, and pharmacological targeting of PRMT1 represents an effective strategy for treating leukemia.

Given that the role of PRMT1 in metabolic regulation has been well documented, we then investigated whether PRMT1 is involved in transforming 6133 cells through metabolic reprogramming. To assess this, we conducted Seahorse assays to measure the changes in ECAR (extracellular acidification rate) in 6133/PRMT1 cells compared to the parental 6133 cells. The ECAR curve of 6133/PRMT1 cells was elevated compared to that of 6133 cells (Figure 3A). Interestingly, the addition of Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), which uncouples oxidative phosphorylation from the tricarboxylic acid (TCA) cycle, did not increase acidification levels. Similarly, antimycin did not cause a significant drop in acid concentration. Remarkably, we treated 6133 cells with MS023 overnight prior to Seahorse assays and observed a reduction in ECAR levels (Figure 3B). These findings suggest that PRMT1 is responsible for the observed increase in acidification. In this experiment, glycolysis contributes the most to acidification in the 6133 leukemia cells, as the addition of mitochondrial respiratory inhibitors only moderately reduces acidification levels according to the principles outlined (Divakaruni et al., 2014). Lactate dehydrogenase A (LDHA), the final key enzyme in glycolysis that converts pyruvate to lactate for NAD⁺ regeneration, was found to be influenced by PRMT1 overexpression. Specifically, PRMT1 overexpression stimulated the tyrosine phosphorylation of LDHA, thereby activating its enzymatic activity (Fan et al., 2011) despite an overall decrease in LDHA levels (Figure 3C). Consistently, when we directly measured the intracellular and extracellular lactate levels in 6133 and 6133/PRMT1 cells, we observed that 6133/PRMT1 cells not only released more lactate into the medium but also exhibited higher intracellular lactate levels (Figure 3D). We concluded that PRMT1 promotes cellular glycolysis and lactate production, which predominantly contributes to the observed increase in acidification.

The Seahorse assays have demonstrated that the cellular OCR (oxygen consumption rate) was reduced in 6133/PRMT1 cells. Additionally, it was observed that these cells had a limited capacity for reserve respiration, as the maximum respiration level was nearly identical to the basal OCAR level, regardless of the expression levels of PRMT1 (Figure 4A). Conversely, when the 6133/PRMT1 cells were pre-treated with MS023 overnight before the Seahorse assays, there was an increase in mitochondrial oxygen consumption compared to the non-treated controls (Figure 4B). These findings further support the notion that PRMT1 plays a role in mediating metabolic reprogramming by enhancing glycolysis and reducing mitochondrial oxygen consumption.

We demonstrated that PRMT1 increases the number of mitochondria in MEG-01 cells, a human AMKL cell line, by confocal microscopy (Zhang et al., 2015). In this experiment, FACS analysis with MitoTracker staining (Figure 4C) and real-time PCR analysis of mitochondrial DNA (Figure 4D) further validated that PRMT1 upregulated mitochondrial biogenesis in 6133 cells. H2DCFDA staining showed that PRMT1 elevated global ROS levels in 6133 cells (Figure 4F). Interestingly, when we performed Mito-Sox staining for ROS levels within mitochondria, we found reduced levels of ROS (Figure 4G). These data imply that most of the ROS detected by H2DCFA may be generated from the cytoplasm rather than from mitochondria, which aligns with the Seahorse data indicating that oxidation consumption in mitochondria is reduced (Figure 4A). Next, we conducted TMRE staining for mitochondrial membrane potential. 6133/PRMT1 cells exhibited reduced staining, suggesting that PRMT1 decreases membrane potential (Figure 4E). This result also cross-validated the Seahorse findings that PRMT1 reduced mitochondrial oxygen consumption.

Since the cytoplasmic ROS level significantly increased with PRMT1 expression, we measured the ratios of two redox pairs: glutathione and NADP⁺. Interestingly, activation of PRMT1 raised the ratios of both GSH/GSSG and NADPH/NADP⁺. Given the crucial roles of GSH and NADPH in biomass synthesis, the upregulation of PRMT1 may promote anabolism, consistent with PRMT1’s role in supporting cell proliferation.

To further probe the complexity of PRMT1-mediated metabolic reprogramming, we established a new 6133 cell line that can conditionally express PRMT1 upon adding doxycycline to the medium. We compared the metabolomic status of 6133 cells before and after 12 hr of PRMT1 induction. The cells were grown in standard RPMI 1640 medium with 10% fetal bovine serum, with high concentrations of glucose and essential amino acids for 12-hr growth. We repeated the metabolomic analysis five times. Principal component analysis indicated that PRMT1 caused a significant shift in metabolite profiles (Figure 5A, B, Figure 5—source data 1). Notably, PRMT1 activation led to increased ATP production and the accumulation of succinyl-CoA, alanine, serine, and short-chain fatty acids and depletion of SAM, aspartic acid, nicotinamide, succinyl-homoserine, oxidized glutathione (which is consistent with the increased ratio of GSH/GSSG shown in Figure 4H), and alpha-ketoglutarate (aka oxoglutaric acid) (Figure 5C). These findings suggest that the beta-oxidation of fatty acids may be impaired, leading to the accumulation of fatty acids such as docosadienoic acid, caproic acid, and suberic acid, while the levels of palmitoyl-carnitine are reduced. Furthermore, the reduced levels of alpha-ketoglutarate indicate potential alterations in the TCA cycle and glutaminolysis. Overall, our data highlight the profound metabolic changes induced by PRMT1, particularly in amino acid biosynthesis and one-carbon metabolism, as shown in Figure 5D generated by the metaboAnalyst software (Pang et al., 2024).

After determining that glycolysis was enhanced in PRMT1-overexpressing cells, we performed glucose colorimetric assays to measure glucose concentrations in the cell culture medium when the 6133 and 6133/PRMT1 cells grew exponentially. A more significant reduction of glucose in 6133/PRMT1 cells was observed compared to parental 6133 cells (Figure 6A), indicating that the elevated glucose consumption via glycolysis is due to PRMT1 upregulation. Next, we cultured the cells in both glucose-containing and glucose-depleted media. Cell viability assays showed that 6133/PRMT1 cells grew more slowly in a glucose-free medium, while parental cells were less affected (Figure 6B). Accordingly, we used 2-deoxy-D-glucose (2-DG), a glucose analog that competes with glucose in glycolysis. The proliferation of 6133/PRMT1 cells was notably inhibited by 2-DG, while the proliferation of 6133 cells was barely affected (Figure 6C). These results are consistent with Seahorse analysis indicating that 6133/PRMT1 cells rely on glycolysis for growth (Figure 3). Consistently, we also demonstrated that E84-high 6133 cells will die quicker than the E84-low 6133 cells in glucose-minus medium (Figure 6—figure supplement 1). Next, we tested whether blocking glucose could impede AMKL progression in mice. Four days after the transplantation of the 6133/PRMT1 cells, we administered 2-DG to the mice at a dosage of 0.5 g/kg body weight every other day for a month. 2-DG treatment significantly delayed disease progression (Figure 6D). Splenomegaly was alleviated, and the burden of leukemia cells in the bone marrow and peripheral blood was notably decreased in the 2-DG treated mice (Figure 6E, F). Taken together, the in vitro and in vivo data suggest that PRMT1-mediated metabolic reprogramming renders leukemia cells highly dependent on glucose (Figure 6—figure supplement 2).

Rapidly growing cancer cells need de novo synthesis of fatty acids to meet the demand for building cell membranes (Menendez and Lupu, 2007). De novo fatty acid synthesis uses acetyl-CoA generated from fatty acid oxidation. Inhibition of fatty acid oxidation has been shown to kill leukemia cells (Samudio et al., 2010). Carnitine-palmitoyl-transferase (CPT1A) catalyzes the rate-limiting step of fatty acid oxidation by transporting fatty acids across the mitochondrial outer membrane. RBM15 binds to the 3′ UTR of CPT1A mRNA (Zhang et al., 2015). Furthermore, PRMT1 regulates the stability of the RBM15 protein. We demonstrated that the expression of both isoforms of PRMT1 lowered the mRNA levels of CPT1A (Figure 7A). We then performed western blotting and confirmed that the protein level of CPT1A was correspondingly reduced in 6133 cell lines ectopically expressing PRMT1 isoforms (Figure 6B). The downregulation of CPT1A leads to reduced consumption of long-chain fatty acids. We consistently observed greater lipid accumulation in 6133/PRMT1 cells than in 6133 cells through FACS analysis with a lipid-binding fluorescent dye, that is BODIPY 493/503 (Figure 7C). Metabolomic analysis also indicated the accumulation of fatty acids such as suberic acid and caproic acid, alongside a decrease in palmitoyl-carnitine, which is a product of CPT1A (Figure 5—source data 1 for metabolomic data). We then treated leukemia cells with etomoxir, which inhibits CPT1A enzymatic activity. The 6133/PRMT1 cells were more sensitive to etomoxir than the 6133 cells (Figure 6D). Orlistat, a FASN inhibitor, also suppressed the proliferation of 6133/PRMT1 cells more effectively than that of 6133 cells (Figure 6E). While CPT1A is responsible for transporting long-chain fatty acids, short-chain fatty acids can be directly transported to mitochondria. We added acetate, propionate, and butyrate in the form of triglycerides to a glucose-free medium. Strikingly, all three forms of short-chain fatty acids supported the proliferation of 6133/PRMT1 cells better than the parental 6133 cells (Figure 6F). The data suggest that leukemia cells with elevated levels of PRMT1 expression can utilize short-chain fatty acids to compensate for their need for glucose.

We then used a lentivirus to express CPT1A in 6133/PRMT1 cells ectopically. CPT1A wild type and mutant (H473A) that does not have succinylation activity (Kurmi et al., 2018) dampens the proliferation rates of 6133/PRMT1 cells cultured under standard conditions. Intriguingly, the enzymatically inactive CPT1A mutant (G710E) inhibits the proliferation of the 6133/PRMT1 cells, suggesting that the 6133 cells still rely on fatty acid oxidation for de novo lipid synthesis necessary for proliferation (Figure 7G). Taken together, we demonstrated that although PRMT1 downregulates fatty acid oxidation as an energy source, PRMT1 upregulates using short-chain fatty acids as an alternative energy source for mitochondria.

6133 cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum 100 U/ml penicillin and 100 μg/ml streptomycin. The addition or withdrawal of mSCF (mouse stem cell factor) (10 ng/ml) in the culture was performed accordingly. The L-Lactate Assay Kit (Cayman, Ann Arbor, Michigan) was used to measure the intracellular and extracellular lactate levels of the cultured cells: approximately 1 × 10⁷ cells were cultured in 10 ml of fresh medium for 24 hr; the cell pellet and medium were collected and processed separately as instructed. The Glucose Colorimetric Assay Kit (Cayman, Ann Arbor, MI) was used to quantify the remaining glucose in the culture medium: 1.25 × 10⁵ cells were cultured in fresh medium overnight; the supernatant/medium was collected and processed as instructed. The fluorescent or colorimetric signals from the assays mentioned above were measured using a microplate reader (Biotek, Winooski, VT). The ratio of NADP/NADPH was assessed using a kit (Cat# MAK479, MilliporeSigma). A total of 6133 cells (0.1 million) grown in the exponential phase were harvested for the assay in a 96-well plate according to the manufacturer’s instructions. The ratio of glutathione GSH/GSSG was determined following the manufacturer’s instructions (Cat# MAK440, MilliporeSigma).

For lentivirus production as described (Su et al., 2021), viral vectors were co-transfected with the envelope vector pMD2.G and the packaging vector ps-PAX2 into 293T cells. Fresh or concentrated viruses were then used to infect the target 6133 cell lines. Stable cell lines were selected using GFP-based flow cytometry sorting with the BD FACSAria II system (BD, Franklin Lakes, NJ).

GFP-positive 6133 cells were intravenously transferred into 8- to 12-week-old sub-lethally irradiated (6 Gy) C57BL/6 mice, which was approved in the IACUC APN-10182 protocol at UAB. Disease progression was closely monitored on a daily basis. Moribund mice were sacrificed to further analyze GFP-positive donor leukemia cells in the peripheral blood, bone marrow, and spleen. For PRMT1 inhibitor treatment, 80 mg/kg body weight of inhibitor solution (16 mg/ml) was intraperitoneally injected into recipient mice every other day for 1 month, starting on the fourth day post-cell transfer. The PRMT1 inhibitor was dissolved in saline with 20% Captisol, 20% PEG-400, and 5% NMP (vol/vol). For 2-DG treatment, 0.25 g/kg body weight of 2-DG (saline solution) was intraperitoneally injected into recipient mice every other day for 1 month, beginning on the fifth day post-cell transfer.

FACS analysis was conducted using the BD LSRFortessa (BD, Franklin Lakes, NJ). MitoTracker DeepRed FM and TMRE stains (Molecular Probes, Eugene, OR) were employed to evaluate cellular mitochondria. H2DCFDA dye (Cat#D399 Thermo Fisher) was used to quantify cytoplasmic ROS levels. BODIPY 493/503 (4,4-Difluoro-1,3,5,7,8-Pentamethyl-4-Bora-3a,4a-Diaza-s-Indacene) (Thermo Fisher) was added at a final concentration of 1 μM and incubated with 6133 and 6133/PRMT1 cells during the exponential phase for 15 min at 37°C in the dark. After incubation, the cells were washed with phosphate-buffered saline (PBS) twice before being subjected to FACS analysis using the manufacturer-recommended spectrum filters.

Dr. Zheng’s laboratory at the University of Georgia synthesized E84, storing a 5 mM stock solution of E84 in dimethyl sulfoxide at –20°C. Exponentially growing cells were harvested and washed with ice-cold PBS. E84 was then incubated at a final concentration of 10 nM with 5 × 10⁵ cells in 100 μl of PBS on ice for 30 min. Afterward, the cells underwent two additional washes with PBS, and the labeled cells were immediately used for cell sorting and FACS analysis. The FACS parameters included a laser wavelength of 640 nm, along with a filter set of 650 LP + 670/14. Fluorescence measurements were collected using the allophycocyanin channel on the BD LSRFortessa machine (BD, Franklin Lakes, NJ). The FACS data were analyzed with FlowJo software, and cell sorting was performed on a BD Aria II sorter.

6133 cells were collected from culture and lysed in 1 ml of H-Lysis buffer (20 mM HEPES pH 7.9, 150 mM NaCl, 1 mM MgCl₂, 0.5% NP40, 10 mM NaF, 0.2 mM NaVO₄, 10 mM β-glycerol phosphate, and 5% glycerol) with freshly added dithiothreitol (1 mM), PMSF (100 μM), and a protease inhibitor cocktail (Roche, Branford, CT). The cells were incubated on ice for 30 min and sonicated using the Bioruptor Ultra-sonication system (Diagenode, Denville, NJ). SDS–PAGE sample buffer was added to the sonicated extracts and boiled. The samples were resolved by SDS–PAGE and transferred to PVDF membranes (Millipore, Billerica, MA). The membranes were blotted with antibodies and then visualized using the Immobilon Western Chemiluminescent reagent (Millipore) with the Bio-Rad ChemiDoc MP system (Bio-Rad, Hercules, CA). The antibodies used in this study include PRMT1 (Cat# 07404, Millipore), LDHA (Cat# MA5-17246, Invitrogen), p-LDHA (Cat# 8176, Cell Signaling), and CPT1a (Cat# 66039, Proteintech).

Total RNA was prepared using Direct-Zol RNAprep Kit (Zymo Research, Irvine, CA). cDNA was generated by the Verso cDNA synthesis Kit (Thermo Scientific, Walthum, MA) with random hexamer priming. Real-time PCR assays were performed with Absolute Blue qPCR SYBR Green Mix (Thermo Scientific) on a ViiA 7 system (Applied Biosystems, Waltham, MA). The relative quantity of gene expression was calculated by the ΔΔCt method. Housekeeping gene Actb (forward: GGC TGG CCG GGA CCT GAC AGA CTA C; reverse: GCA GTG GCC ATC TCC TGC TCG AAG TC) was used for normalization.

Cell viability was measured using the CellTiter-Glo Viability Assay Kit (Promega, Madison, WI). A total of 1000 6133 cells were seeded in 96-well plates (100 μl per well) with or without treatment. At the specified time after culture setup, 100 μl of CellTiter-Glo reagent was added to each well. The luminescent signal was recorded with a microplate reader (Biotek, Winooski, VT).

Cells were washed twice with ice-cold PBS before extracting metabolites in a 20% methanol solution (LC–MS grade methanol, Fisher Scientific) at −70°C. The tissue–methanol mixture underwent bead-beating for 45 s using a Tissue/cell disruptor (QIAGEN). The extracts were centrifuged for 5 min at 2000 × g to pellet insoluble material, and the supernatants were transferred to clean tubes. This extraction procedure was repeated two more times, and all three supernatants were pooled, dried using a Vacufuge (Eppendorf), and stored at −80°C until analysis. The methanol-insoluble protein pellet was solubilized in 0.2 M NaOH at 95°C for 20 min, and the total protein concentration was quantified using a Bio-Rad DC assay. On the day of metabolite analysis, the dried cell extracts were reconstituted in 70% acetonitrile to achieve a relative protein concentration of 1 μg/ml, and 4 μl of this reconstituted extract was injected for LC/MS-based untargeted metabolite profiling.

Cell extracts were analyzed using an LC/MS system that featured an Agilent Model 1290 Infinity II liquid chromatography platform coupled with an Agilent 6550 iFunnel time-of-flight mass spectrometer. The chromatography of metabolites utilized aqueous normal phase chromatography on a Diamond Hydride column (Microsolv). The mobile phases consisted of (A) 50% isopropanol with 0.025% acetic acid and (B) 90% acetonitrile containing 5 mM ammonium acetate. To minimize the interference of metal ions on chromatographic peak integrity and electrospray ionization, EDTA was introduced to the mobile phase at a final concentration of 6 μM. The mobile phase gradient was as follows: 0–1.0 min, 99% B; 1.0–15.0 min, decreasing to 20% B; 15.0–29.0 min, 0% B; and from 29.1 to 37 min, returning to 99% B. Raw data were analyzed with MassHunter Profinder 8.0 and MassProfiler Professional (MPP) 15.1 software (Agilent Technologies).

To determine the identities of differentially expressed metabolites (p < 0.05), LC/MS data were queried against an in-house annotated personal metabolite database, created using MassHunter PCDL Manager 8.0 (Agilent Technologies), based on monoisotopic neutral mass (<5 ppm mass accuracy) and chromatographic retention times of pure standards. A molecular formula generator (MFG) algorithm in MPP was utilized to generate and score empirical molecular formulas, considering monoisotopic mass accuracy, isotope abundance ratios, and spacing between isotope peaks. A tentative compound ID was assigned when the PCDL database and MFG scores matched for a given candidate molecule. Tentatively assigned molecules were confirmed based on matching LC retention times and/or MS/MS fragmentation spectra for pure molecular standards.

The metabolic activities of living 6133 cells and 6133/PRMT1 cells were measured using a Seahorse XF24 metabolic flux analyzer (Agilent Technologies, Santa Clara, CA), following the described procedures (Yu et al., 2013). Briefly, AMKL cells (1.5–2 × 10⁵) were seeded in 24-well plates designed for XF24 cell culture. The plates were pre-coated with Cell-Tak (Corning, Corning, NY) and incubated in XF base medium containing glucose (10 mM), l-glutamine (2 mM), and sodium pyruvate (1 mM) at 37°C for 1 hr. OCRs were measured at baseline and after the sequential addition of the mitochondrial inhibitor oligomycin (500 nM), the mitochondrial uncoupling compound FCCP (5 μM), and the respiratory chain inhibitor antimycin (1 μM) to the culture. The FCCP concentration must be determined by titration. Glycolytic activities were simultaneously assessed using the same instrument based on ECARs. All measurements were conducted according to the manufacturer’s protocols, and the data were normalized by cell numbers.

The figure legends provide all the experimental details. In the bar graphs, a two-tailed Student’s t-test was employed for significance testing, with p values less than 0.05 deemed significant. Quantitative data are presented as mean ± SEM. The R programming package and GraphPad Prism 6 were utilized for statistical analysis.