PKM2 modulation regulates leukemia cell survival Authors: Nikolina Vrdoljak, Ekaterina N. Parfenova, Alessia Roma, Drake A. Mosca, Mark D. Minden, Paul A. Spagnuolo Published: 2026 ABSTRACT Acute myeloid leukemia (AML) cells exhibit aberrant metabolism defined by a shift away from oxidative phosphorylation and towards anaerobic glycolysis, favouring cell growth. Glycolytic enzymes are altered for this change to occur, including pyruvate kinase (PK), where the tetrameric and high activity M1 isoform (PKM1) is replaced with the predominantly dimeric and low activity M2 isoform (PKM2). Dimeric PKM2 produces less pyruvate and acts as a protein kinase in the nucleus, demonstrating divergent roles in both cell metabolism and as a transcriptional co-activator. In this study, the role of PKM2 in AML was defined, as PKM2 levels were elevated but PK enzymatic activity was reduced in AML cells compared to normal hematopoietic cells. Genetic and pharmacological studies show that decreasing and increasing PKM2 activity resulted in anti-AML effects both in vitro and in vivo. Indeed, these models show that inhibition and activation of PKM2 disrupt the native oligomeric state of the protein, resulting in reduced nuclear PKM2 accumulation and c-Myc expression, ultimately leading to cell death. Together, these results highlight the importance of PKM2 in AML, uncover the mechanisms by which both inhibition and activation cause AML cell death, and identify a novel modulator of PKM2 activity.Supplementary informationThe online version contains supplementary material available at 10.1186/s40170-026-00440-7. Introduction Pyruvate kinase (PK) is the enzyme catalyzing the final step of glycolytic metabolism, creating pyruvate. Two PK isoforms are encoded by the PKM gene, with isoform expression determined by mutually exclusive exon inclusion [1]. PKM1 is predominantly expressed in healthy cells with high energy requirements, while PKM2 is elevated in proliferating cells (i.e., cancer cells and embryonic stem cells) [2, 3]. Although structurally similar and only differing by 22 amino acids, catalytic activity varies between isoforms [4]. PKM1 is a constitutively active tetramer, while PKM2 is predominantly expressed as a low-activity dimer containing an allosteric binding pocket that enables cycling between dimeric and tetrameric conformations [4]. Small molecule or ligand (e.g., fructose-2,6-bisphophate (FBP) [4, 5] and serine [6]) binding, as well as post-translational modifications (e.g., phosphorylation [5] and oxidation [7]) regulate the structural conformation of PKM2. High activity, tetrameric PK (i.e., PKM1 or tetrameric PKM2) enables rapid pyruvate production that can be utilized for mitochondrial adenosine triphosphate (ATP) production. Alternatively, low activity, dimeric PK (i.e., PKM2) produces less pyruvate and promotes the accumulation of upstream intermediates that divert to biosynthetic pathways supporting anabolism [8]. As such, upregulation of PKM2 expression in malignancy contributes to increased cell proliferation [9]. Dimeric PK (i.e., exclusively PKM2) can translocate to the nucleus, where it acts as a protein kinase. Here, PKM2 phosphorylates proteins (e.g., signal transducer and activator of transcription 3 (STAT3) and histone H3) and acts as a transcriptional co-activator [10]. As a result, c-Myc activity increases and blocks myeloid cell differentiation to promote leukemogenesis [11, 12]. Indeed, PKM2-expressing cells are locked in an undifferentiated and proliferative state where the dominance of PKM2 supports pluripotency; proliferation halts in the absence of PKM2 [5]. Acute myeloid leukemia (AML) is defined by this impediment of myeloid cell differentiation [13]. A devastating disease where cells overexpress PKM2 [3], AML patients face poor prognosis [14]. Therefore, investigating druggable targets, including PKM2, is essential in improving patient outcomes. While studies pertaining to pharmacological modulation in AML remain sparse, inhibitors and activators of PKM2 (e.g., 3k, shikonin, ML-265, DASA-58) have been investigated, with promise shown in impairing growth of solid tumor models including esophageal squamous cell carcinoma, non-small cell lung carcinoma and ovarian cancer [15–19]. Paradoxically, increasing [5, 16, 20] or decreasing [15, 21–23] PKM2 enzymatic activity through pharmacological or genetic methods disrupts cancer metabolism and prevents tumor growth. In this study, the divergent roles of PKM2 (i.e., as a metabolic enzyme and protein kinase) could reconcile these seemingly contradictory observations, where up and downregulation of PKM2 activity result in similar phenotypes (i.e., cell death). The observations that both inhibition and activation impart cancer cell death suggest that the regulation and control of PKM2 levels predict cell fate. Homeostatic control of nutrients and cellular function is central to host survival. For example, blood glucose is maintained between 4 - 6 mM by numerous well-defined cellular and molecular pathways [24]. If untreated, exceeding or dropping below this range results in the same endpoint – death. Here, we extend this “Goldilocks effect”, where a biological factor must be within an optimal range, to PKM2 activity, and explore the functional role of PKM2 in AML. Genetic and pharmacological models were used to inhibit or activate PKM2 and determine how opposing cell responses result in the same cellular phenotype. Reduced or increased PKM2 activity paradoxically decreased nuclear translocation of dimeric PKM2 and decreased expression of c-Myc, leading to cell death. This modulation of PKM2 activity decreased AML cell growth in vitro and in vivo, indicating that homeostatic control of PKM2 determines cell fate. Methods Cell culture Ontario Cancer Institute acute myeloid leukemia-2 (OCI-AML2) cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM; Cytvia, Marlborough, MA, USA) supplemented with 10% fetal bovine serum (FBS; Cytvia). TEX cells were cultured in IMDM supplemented with 15% FBS (Sigma), 20 ng/mL stem cell factor (Peprotech, Rocky Hill, NJ, USA), 2 ng/mL interleukin 3 (Peptrotech) and 2 mM L-glutamine (Cytvia). IMDM for both cell lines were supplemented with an antibiotic solution containing 20 units of penicillin and 200 µg of streptomycin (per mL of media; Cytvia). All cells were maintained at 37 °C, 95% relative humidity and 5% CO2. Additional cell line information is included in the supplemental methods. The sources of primary, patient-derived AML and healthy peripheral blood mononuclear cells (PBMCs) are supplied in the supplemental methods, and AML patient cytogenetics are detailed in Table 1.Table 1Primary AML patient cytogeneticsPatient IDCytogenetics*AML 146,XY,inv(2)(p13q37)[19]/46,XY[1]AML 246,XX,t(9;11)(p22;q23)[20]AML 3unsuccessfulAML 4not doneAML 546,XX[20]AML 646,XY[20]AML 748,XX,+8+8[8]/49,idem,+8,+13,i(13)(q10)[7]/46,XX[5]AML 846,XY[20]AML 946,XY[20]AML 1046,XY[21]AML 1146,XYAML 1246,XY[20]AML 1346,XY[20]AML 1445 ~ 46,XX,t(8 ;16)(p11.2 ;p13.3),der(17 ;18)(q10 ;q10),der(19)t(1 ;19)(q23 ;q13.4)[cp18]AML 1546,XX[20]AML 16not doneAML 17not doneAML 1846, XY[23]* Reporting stylea. overall number of chromosomes identifiedb. sex chromosomesc. affected chromosomesd. type of abnormalities described in shorthande. chromosomal band locationf. in bracket, number of cells with given karyotype Table 1Primary AML patient cytogeneticsPatient IDCytogenetics*AML 146,XY,inv(2)(p13q37)[19]/46,XY[1]AML 246,XX,t(9;11)(p22;q23)[20]AML 3unsuccessfulAML 4not doneAML 546,XX[20]AML 646,XY[20]AML 748,XX,+8+8[8]/49,idem,+8,+13,i(13)(q10)[7]/46,XX[5]AML 846,XY[20]AML 946,XY[20]AML 1046,XY[21]AML 1146,XYAML 1246,XY[20]AML 1346,XY[20]AML 1445 ~ 46,XX,t(8 ;16)(p11.2 ;p13.3),der(17 ;18)(q10 ;q10),der(19)t(1 ;19)(q23 ;q13.4)[cp18]AML 1546,XX[20]AML 16not doneAML 17not doneAML 1846, XY[23]* Reporting stylea. overall number of chromosomes identifiedb. sex chromosomesc. affected chromosomesd. type of abnormalities described in shorthande. chromosomal band locationf. in bracket, number of cells with given karyotype Primary AML patient cytogenetics * Reporting style a. overall number of chromosomes identified b. sex chromosomes c. affected chromosomes d. type of abnormalities described in shorthand e. chromosomal band location f. in bracket, number of cells with given karyotype Immunoblotting To assess protein levels, total cell lysates were prepared as previously described [25, 26]. Methods are further outlined in supplemental methods and antibody information is provided in Table 2.Table 2Immunoblotting antibodiesAntibody TargetIsotypeMolecular Weight (kDa)SupplierDilutionHuman PKM2Rabbit604053; CST*1:1000 (milk)Human PKM1Rabbit607067; CST1:1000 (milk)Human c-MycRabbit57–655605; CST1:2000 (BSA)Human GAPDHRabbit37AB21054; Abcam1:1000 (milk)Human Histone H3Rabbit1712648; CST1:1000 (BSA)Human β-actinRabbit424967; CST1:1000 (milk)Human SDHARabbit7011998; CST1:1000 (BSA)*Cell Signaling Technology (CST) Table 2Immunoblotting antibodiesAntibody TargetIsotypeMolecular Weight (kDa)SupplierDilutionHuman PKM2Rabbit604053; CST*1:1000 (milk)Human PKM1Rabbit607067; CST1:1000 (milk)Human c-MycRabbit57–655605; CST1:2000 (BSA)Human GAPDHRabbit37AB21054; Abcam1:1000 (milk)Human Histone H3Rabbit1712648; CST1:1000 (BSA)Human β-actinRabbit424967; CST1:1000 (milk)Human SDHARabbit7011998; CST1:1000 (BSA)*Cell Signaling Technology (CST) Immunoblotting antibodies *Cell Signaling Technology (CST) Quantitative reverse transcription polymerase chain reaction Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was conducted as previously described [27] and is further outlined in the supplemental methods. Primer sequences (Invitrogen, Waltham, MA, USA) are provided in Table 3.Table 3qRT-PCR primer sequencesProteinPrimer Sequence (5’-3’)SourcePKM2Forward: ATTATTTGAGGAACTCCGCCGCCTInvitrogenReverse: ATTCCGGGTCACAGCAATGATGGPKM1Forward: CGAGCCTCAAGTCACTCCACInvitrogenReverse: GTGAGCAGACCTGCCAGACTGAPDHForward: CGACCACTTTGTCAAGCTCAInvitrogenReverse: AGGGGTCTACATGGCAACTG Table 3qRT-PCR primer sequencesProteinPrimer Sequence (5’-3’)SourcePKM2Forward: ATTATTTGAGGAACTCCGCCGCCTInvitrogenReverse: ATTCCGGGTCACAGCAATGATGGPKM1Forward: CGAGCCTCAAGTCACTCCACInvitrogenReverse: GTGAGCAGACCTGCCAGACTGAPDHForward: CGACCACTTTGTCAAGCTCAInvitrogenReverse: AGGGGTCTACATGGCAACTG qRT-PCR primer sequences Lentiviral mediated knockdown Lentiviral-mediated knockdown of PKM2 was conducted as previously described [25, 28], utilizing a hairpin-pLKO.1 vector with a puromycin-resistant gene and short hairpin ribonucleic acid (shRNA) sequences specific for PKM2 (Table 4; Horizon Discovery, Cambridge, UK). Methods are further outlined in supplementary methods.Table 4shRNA constructs for genetic knockdown of PKM2ManuscriptConstructHorizon Discovery shRNAClone IDMature Antisense Sequence5’ − 3’shRNA-21TRCN0000037610ATCTTGATGTTCTTTCCCTTCshRNA-23TRCN0000037612TTTAGTGTCTAGAGCCACAGC Table 4shRNA constructs for genetic knockdown of PKM2ManuscriptConstructHorizon Discovery shRNAClone IDMature Antisense Sequence5’ − 3’shRNA-21TRCN0000037610ATCTTGATGTTCTTTCCCTTCshRNA-23TRCN0000037612TTTAGTGTCTAGAGCCACAGC shRNA constructs for genetic knockdown of PKM2 Cell growth and viability Cell growth and viability were measured as previously described [25, 26, 29]. Cell growth was measured using Trypan-blue exclusion assays and colony formation assays. 7-Aminoactinomycin D (7-AAD; Cayman Chemical, Ann Arbor, MI, USA) and a Guava EasyCyte 8HT flow cytometer (Cytek Biosciences, Fremont, CA, USA) as well as 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt (MTS) dye and a SynergyTM 2 Multimode Microplate Reader (BioTek Instruments, Winooski, VT, USA) were used to assess cell viability. All methods are further described in supplemental methods. Pyruvate kinase activity assay PK activity was assessed with a lactate dehydrogenase (LDH)-coupled assay using cell lysate-based or recombinant enzyme-based methods as previously described [5, 15] which are further described in supplemental methods. Immunoprecipitation experiments Immunoprecipitation (IP) experiments were conducted similarly to previous studies [25], where a protein of interest is isolated following treatment with the compound of interest. In this study, IP of PKM2 was conducted according to a protocol provided by Cell Signaling Technology (CST, Danvers, MA, USA) using a magnetic Co-IP kit (Pierce; Thermo Fisher, Waltham, MA, USA) and is further described in the supplemental methods. Immunoblotting was used to analyze fraction purity. Following immunoprecipitation, the PKM2-rich fraction underwent acetone extraction followed by gas chromatography-mass spectrometry (GC-MS) analysis for PLB quantification, using a method derived from previous protocols [30] further described in supplemental methods. Cellular thermal shift assay A protocol similar to previously established methods [25] was conducted to determine the interaction between PKM2 and PLB and is further described in supplemental methods. PKM2 isoform cross-linking assay Disuccinimidyl suberate (DSS; Thermo Fisher) cross-linking assays were conducted as previously described [31] to determine the expression of dimeric and tetrameric PKM2 based on molecular weight and are further described in supplemental methods. Nuclear extraction Nuclear extraction was conducted using cell extraction buffer (Thermo Fisher; FNN0011) as per manufacturer’s instructions and is further described in supplemental methods. Immunoblotting was conducted following nuclear extraction. In vivo studies Xenograft and engraftment assays were conducted using non-obese diabetic severe combined immunodeficient gamma (NSG) mice (The Jackson Laboratory, Bar Harbor, ME, USA) as previously described and are further described in the supplemental methods [25, 26, 32]. Statistical analysis Statistics were analyzed using GraphPad 10.0 Prism, and results presented are the mean ± standard error of mean. The significance of the difference of values between groups was determined using an unpaired, one- or two-tailed Students t-test or one-way ANOVA paired with Tukey’s post-hoc test. Half-maximal inhibitory concentrations (IC50) values were calculated using non-linear regression (log[inhibitor] vs. response-variable slope). For in vivo studies, significance in engraftment levels and biochemical markers was determined using a Mann-Whitney U test. p < 0.05 was defined as statistically significant. Results PKM2 is elevated in AML cells Protein levels of PKM isoforms were first quantified in a panel of blood cancer cell lines. Immunoblotting suggests that PKM2 protein levels are elevated when compared to PKM1 (Fig. 1A, S4A). Next, qRT-PCR quantified microribonucleic acid (mRNA) in PBMCs from healthy donors and patient-derived AML cells. PKM2 mRNA was significantly higher in AML vs. PBMCs (p < 0.05; Fig. 1B). Notably, this differential isoform expression was accompanied with a significant decrease in overall PK enzymatic activity in patient-derived AML vs. PBMCs (p < 0.05; Fig. 1C). These results suggest that PKM2 levels differ in AML cells when compared to normal cells and may be associated with corresponding reductions in PK activity. These results mirror previous studies confirming that PKM2, but not PKM1, is overexpressed in cancer cells, resulting in low enzymatic (PK) activity, low pyruvate production, reduced glucose-derived (mitochondrial) ATP production, and an increased proportion of glucose-derived carbons shuttled to biosynthetic pathways (Fig. 1D) [33].Fig. 1PKM2 protein levels and activity are different in AML. (A) Immunoblots comparing pyruvate kinase M2 (PKM2) and M1 (PKM1) protein expression across a panel of hematological cancer cell lines. (B) qRT-PCR data demonstrating PKM2 micro ribonucleic acid (mRNA) levels in patient-derived AML cells (ID #1–6, where number represents patient ID) vs. peripheral blood mononuclear cells (PBMCs) from healthy donors (n = 5). (C) Baseline pyruvate kinase (PK) activity in AML patients (ID #1–3, 7, 8) vs. healthy donor PBMCs (n = 4), determined using a cell-based, lactate dehydrogenase (LDH)-coupled fluorometric PK activity assay. Cytogenetics for patient-derived AML samples are provided in Table 1. (D) Schematic representation of the role of PK in the glycolytic pathway, highlighting differential isoform expression in healthy vs. AML cells. Bolded terms indicate the predominant isoform in each respective model. PKM1, a constitutive tetramer highly expressed in healthy cells, supports high mitochondrial adenosine triphosphate (ATP) production. PKM2 primarily exists as a dimer with low catalytic activity, resulting in lower pyruvate production and increased flux to biosynthetic pathways such as pentose phosphate pathway (PPP) and serine synthesis pathway (SSP). PKM2 can cycle between dimeric and tetrameric conformations based on small molecule binding, allosteric binding and post-translational modifications. Created using BioRender.* p ≤ 0.05, unpaired, one-tailed Student’s t-test. All data is represented as the mean ± standard error of mean. G-6-P; glucose-6-phosphate, 3-PG; 3-phosphoglycerate, 2-PG; 2-phosphoglycerate, PEP; phosphoenolpyruvate PKM2 protein levels and activity are different in AML. (A) Immunoblots comparing pyruvate kinase M2 (PKM2) and M1 (PKM1) protein expression across a panel of hematological cancer cell lines. (B) qRT-PCR data demonstrating PKM2 micro ribonucleic acid (mRNA) levels in patient-derived AML cells (ID #1–6, where number represents patient ID) vs. peripheral blood mononuclear cells (PBMCs) from healthy donors (n = 5). (C) Baseline pyruvate kinase (PK) activity in AML patients (ID #1–3, 7, 8) vs. healthy donor PBMCs (n = 4), determined using a cell-based, lactate dehydrogenase (LDH)-coupled fluorometric PK activity assay. Cytogenetics for patient-derived AML samples are provided in Table 1. (D) Schematic representation of the role of PK in the glycolytic pathway, highlighting differential isoform expression in healthy vs. AML cells. Bolded terms indicate the predominant isoform in each respective model. PKM1, a constitutive tetramer highly expressed in healthy cells, supports high mitochondrial adenosine triphosphate (ATP) production. PKM2 primarily exists as a dimer with low catalytic activity, resulting in lower pyruvate production and increased flux to biosynthetic pathways such as pentose phosphate pathway (PPP) and serine synthesis pathway (SSP). PKM2 can cycle between dimeric and tetrameric conformations based on small molecule binding, allosteric binding and post-translational modifications. Created using BioRender.* p ≤ 0.05, unpaired, one-tailed Student’s t-test. All data is represented as the mean ± standard error of mean. G-6-P; glucose-6-phosphate, 3-PG; 3-phosphoglycerate, 2-PG; 2-phosphoglycerate, PEP; phosphoenolpyruvate Reduced PKM2 activity reduces AML growth in vitro and in vivo To determine the functional impact of reduced PKM2, lentiviral-mediated gene knockdown was used, where two unique shRNA sequences (denoted: shPKM2-21 and shPKM2-23) were transduced into TEX and OCI-AML2 cell lines. Genetic knockdown was confirmed by immunoblotting and activity assays, which showed decreased protein levels (F (2,3)=2058, p < 0.0001) and reduced PK activity (F(2,6)=195.6, p < 0.0001), respectively (Fig. 2A&B, S1A&B, S4B&C). All knockdown constructs were compared to cells transduced with scramble controls with endogenous PKM2 expression, which are hereafter referred to as transduced control, or TRC. Cell proliferation was significantly slower in shPKM2-21 and shPKM2-23 cells (F (2,6)=12.7, p < 0.01; Fig. 2C, S1C); cells with reduced PKM2 are hereafter referred to as shPKM2 cells. Similarly, shPKM2 cells had reduced colony forming abilities (day 5: F (2,6)=42.66, p < 0.001; Fig. 2D, S1D).Fig. 2Knockdown of PKM2 reduces leukemia cell growth in vitro and reduces engraftment while extending survival in vivo. (A) Immunoblot and densitometry analysis along with (B) a cell-based, lactate dehydrogenase (LDH)-coupled fluorometric pyruvate kinase (PK) activity assay in TEX, TRC and knockdown (shPKM2-21, shPKM2-23) cell lines. All experiments were normalized to TRC cells. (C) Cell proliferation and (D) colony formation in cells with reduced PKM2 expression. (E) Schematic depiction of the tail vein injection of non-obese diabetic severe combined immunodeficient gamma (NSG) mice with an equal number of TRC or shPKM2 cells. Created using BioRender. (F) Following 5 weeks, human (h)CD33+ and (h)CD45+ population was quantified via flow cytometry to determine engraftment potential of each construct. (G) Kaplan-Meyer curve demonstrates the difference in survival time of mice inoculated with TRC vs. shPKM2 cells (n = 5 mice/group). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. One-way ANOVA with Tukey’s post-hoc test (A,B); unpaired, two-tailed Student’s t-test (C,D); Mann-Whitney U test (F); log-rank mantel-cox test (G). All data is represented as the mean ± standard error of mean. CFU; colony forming unit Knockdown of PKM2 reduces leukemia cell growth in vitro and reduces engraftment while extending survival in vivo. (A) Immunoblot and densitometry analysis along with (B) a cell-based, lactate dehydrogenase (LDH)-coupled fluorometric pyruvate kinase (PK) activity assay in TEX, TRC and knockdown (shPKM2-21, shPKM2-23) cell lines. All experiments were normalized to TRC cells. (C) Cell proliferation and (D) colony formation in cells with reduced PKM2 expression. (E) Schematic depiction of the tail vein injection of non-obese diabetic severe combined immunodeficient gamma (NSG) mice with an equal number of TRC or shPKM2 cells. Created using BioRender. (F) Following 5 weeks, human (h)CD33+ and (h)CD45+ population was quantified via flow cytometry to determine engraftment potential of each construct. (G) Kaplan-Meyer curve demonstrates the difference in survival time of mice inoculated with TRC vs. shPKM2 cells (n = 5 mice/group). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. One-way ANOVA with Tukey’s post-hoc test (A,B); unpaired, two-tailed Student’s t-test (C,D); Mann-Whitney U test (F); log-rank mantel-cox test (G). All data is represented as the mean ± standard error of mean. CFU; colony forming unit To determine the significance of PKM2 in bone marrow repopulation, NSG mice were injected with an equal number of TRC or shPKM2 cells (Fig. 2E). Mice injected with shPKM2 cells had significantly less bone marrow engraftment compared to mice injected with TRC cells (~4 fold; p < 0.05 Mann-Whitney U = 0; Fig. 2F). In addition, mice inoculated with shPKM2 cells survived significantly longer compared to TRC cell-injected mice (median survival 42 days vs. 33 days, p < 0.05; Fig. 2G). These results show that reduced PKM2 expression and activity decrease AML cell growth. Increased PKM2 activity reduces AML growth in vitro and in vivo Identification of pharmacological activators of PKM2 Few pharmacological activators or inhibitors of PKM2 exist. Inhibitors (i.e., 3K, shikonin) modestly reduce solid tumor growth in esophageal small cell and ovarian carcinomas [17, 18], whereas activators (i.e., ML-265) reduce growth of lung cancer in vivo [19]. Despite promising results, the efficacy of ML-265 requires relatively high dosing (50 mg/kg, 2x/day), highlighting limitations due to feasibility and the need for additional PKM2-specific compounds. To identify specific modulators of PKM2 activity, an AML cell-based drug screen was conducted using TRC and shPKM2 cells (Fig. 3A, S2A). Here, we reasoned that compounds less cytotoxic towards shPKM2 cells require PKM2 for activity. Through this approach, 5-hydroxy-2-methyl-1,4-naphthoquinone (i.e., plumbagin; PLB), was identified as a PKM2 modulator (Fig. 3B). PLB’s molecular target is unknown, however the compound has been observed to alter cellular redox status, inhibit cell cycle proteins and impair cell signalling pathways (e.g., STAT3 and Wnt) in solid tumors [34, 35] and hematological malignancies [36, 37]. In this study, PLB’s cytotoxicity was significantly diminished in shPKM2 cells, indicating that PLB cytotoxicity is dependent on PKM2 (F (2,6) = 14.62, p < 0.01; Fig. 3C).Fig. 3A drug screen identifies PLB as a selective, pharmacological modulator of PKM2. (A) Screening of a drug library identified (B) 5-hydroxy-2-methyl-1,4-naphthoquinone, or plumbagin (PLB), as a modulator of pyruvate kinase M2 (PKM2). (C) Cell death following 3 µM PLB treatment highlights the difference in potency in TRC vs. knockdown (shPKM2) cells. (D) Recombinant PKM1 (left) and PKM2 (right) enzyme activity following one hour of treatment with PLB. (E) Schematic diagram depicting magnetic immunoprecipitation (IP) of PKM2 following incubation of TEX cells in the presence of PLB or control. Gas chromatography-mass spectrometry (GC-MS) detection of PLB determined an interaction effect. Created using BioRender. (F) GC-MS detection of PLB in PKM2-enriched fractions (top) compared to a PLB standard (bottom). (G) Quantification of PLB concentration within the PKM2-rich sample. (H) Schematic depiction of ligand binding to a protein increasing thermal stability. Created using BioRender. (I) Thermal stability at 68.5 °C of PKM2 (TEX) and (J) PKM1 (U87-MG) incubated in presence or absence of 5 µM PLB. *p ≤ 0.05, **p ≤ 0.01. One-way ANOVA with Tukey’s post-hoc test (C-D) or unpaired, two-tailed Student’s t-test (I-J). All data is represented as the mean ± standard error of mean. n.D.; non-detectable A drug screen identifies PLB as a selective, pharmacological modulator of PKM2. (A) Screening of a drug library identified (B) 5-hydroxy-2-methyl-1,4-naphthoquinone, or plumbagin (PLB), as a modulator of pyruvate kinase M2 (PKM2). (C) Cell death following 3 µM PLB treatment highlights the difference in potency in TRC vs. knockdown (shPKM2) cells. (D) Recombinant PKM1 (left) and PKM2 (right) enzyme activity following one hour of treatment with PLB. (E) Schematic diagram depicting magnetic immunoprecipitation (IP) of PKM2 following incubation of TEX cells in the presence of PLB or control. Gas chromatography-mass spectrometry (GC-MS) detection of PLB determined an interaction effect. Created using BioRender. (F) GC-MS detection of PLB in PKM2-enriched fractions (top) compared to a PLB standard (bottom). (G) Quantification of PLB concentration within the PKM2-rich sample. (H) Schematic depiction of ligand binding to a protein increasing thermal stability. Created using BioRender. (I) Thermal stability at 68.5 °C of PKM2 (TEX) and (J) PKM1 (U87-MG) incubated in presence or absence of 5 µM PLB. *p ≤ 0.05, **p ≤ 0.01. One-way ANOVA with Tukey’s post-hoc test (C-D) or unpaired, two-tailed Student’s t-test (I-J). All data is represented as the mean ± standard error of mean. n.D.; non-detectable A fluorescence-based assay using recombinant enzymes, in the presence or absence of PLB, determined effects on PK activity. While no significant changes were observed on PKM1, PKM2 activity increased significantly upon PLB exposure in a dose-dependent manner (F (3, 7) = 14.52, p < 0.01; Fig. 3D). To confirm a physical interaction, PLB was quantified in PKM2-rich fractions derived from PLB or control-treated cells (Fig. 3E); PKM2 predominance was confirmed via immunoblotting (Fig. S2B, S5). GC-MS confirmed PLB was present in the PKM2-enriched fraction (Fig. 3F&G, S2C&D). Finally, thermal shift assays, which measure changes in protein melting point as an indirect indicator of ligand or drug binding, were conducted (Fig. 3H). PLB increased PKM2 thermal stability of treated cells (p < 0.05; Fig. 3I, S2E, S6A&B). In contrast, PLB had no effect on PKM1 thermal stability (Fig. 3J, S7). This data supports PLB as a specific PKM2-targeting compound, with specific PKM2 activation properties. Pharmacological activation of PKM2 reduces AML cell survival in vitro A cell-based activity assay confirmed a dose-dependent increase in PK activity with PLB treatment (F (3,8)=4.671, p < 0.05; Fig. 4A, S2F), along with a corresponding dose-dependent decrease in leukemia cell death (IC50: 1.40–2.73 µM; Fig. 4B). Next, selectivity was tested in patient-derived AML cells and normal PBMCs. PLB elevated both PK and PKM2-specific activity in patient-derived AML cells with no change in PBMCs (p < 0.01; Fig. 4C&D). Selective cytotoxicity was confirmed as patient-derived AML cell viability decreased with increasing concentrations of PLB; no significant effects were noted in PBMCs (F (3,28) = 45.07, p < 0.0001; Fig. 4E). Finally, PLB (1 µM) decreased clonogenic growth of AML cells with no effects on PBMC colony formation (p < 0.001; Fig. 4F). Together, this data indicates that PLB is a selective anti-AML compound.Fig. 4PLB increases PK activity and selectively kills AML cells. (A) Changes in TEX cell pyruvate kinase (PK) activity following treatment with plumbagin (PLB). (B) Cell viability and IC50 values following 72-hour PLB treatment in a panel of hematological malignant cell lines. (C) Changes in PK activity following one hour treatment of patient-derived AML cells (red; ID #1, 2,4, 6, 8) and healthy PBMCs (blue; n = 5) samples. (D) Changes in PKM2-specific activity in patient-derived AML (red; ID #1–3, 9) and healthy peripheral blood mononuclear cells (PBMC, blue; n = 3). (E) Cell viability of primary AML cells (red; ID #3,5,6,8–14) and healthy PBMC (blue; n = 8) following 24-hour treatment of PLB. (F) Clonogenic growth of patient-derived AML samples (red; ID #1, 9–10, 15–17) and healthy PBMC (blue; n = 3) following incubation in the presence or absence of 1 µM PLB. Cytogenetics for patient-derived AML samples are provided in Table 1. # p = 0.05, *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001. Unpaired, two-tailed Student’s t-test (A); unpaired, two-tailed Student’s t-test (C-D, F); one-way ANOVA with Tukey’s post-hoc test (E). All data is represented as the mean ± standard error of mean. CFU; colony forming unit PLB increases PK activity and selectively kills AML cells. (A) Changes in TEX cell pyruvate kinase (PK) activity following treatment with plumbagin (PLB). (B) Cell viability and IC50 values following 72-hour PLB treatment in a panel of hematological malignant cell lines. (C) Changes in PK activity following one hour treatment of patient-derived AML cells (red; ID #1, 2,4, 6, 8) and healthy PBMCs (blue; n = 5) samples. (D) Changes in PKM2-specific activity in patient-derived AML (red; ID #1–3, 9) and healthy peripheral blood mononuclear cells (PBMC, blue; n = 3). (E) Cell viability of primary AML cells (red; ID #3,5,6,8–14) and healthy PBMC (blue; n = 8) following 24-hour treatment of PLB. (F) Clonogenic growth of patient-derived AML samples (red; ID #1, 9–10, 15–17) and healthy PBMC (blue; n = 3) following incubation in the presence or absence of 1 µM PLB. Cytogenetics for patient-derived AML samples are provided in Table 1. # p = 0.05, *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001. Unpaired, two-tailed Student’s t-test (A); unpaired, two-tailed Student’s t-test (C-D, F); one-way ANOVA with Tukey’s post-hoc test (E). All data is represented as the mean ± standard error of mean. CFU; colony forming unit Pharmacological activation of PKM2 imparts selective anti-AML activity in vivo To determine the effects of PKM2 activation in vivo, NSG engraftment mouse models were used (Fig. 5A). Preconditioned mice were intrafemorally injected (left femur) with patient-derived AML or normal cluster of differentiation (CD)34+ bone marrow (BM) cells, and after a BM aspirate confirmed engraftment, mice were randomized into control or treatment (PLB, 2.5 mg/kg; 3x/week) groups. Following five weeks of treatment, PLB reduced AML engraftment (right femur) by approximately 85% (p < 0.0001, Mann-Whitney U = 0; Fig. 5B). In contrast, mice injected with normal CD34+ BM cells undergoing the same treatment regimen experienced no significant change in BM engraftment (Fig. 5C). Whole blood was collected at the time of sacrifice via cardiac puncture and was analyzed by the Animal Health Lab at the University of Guelph. Assessment of body weight, bilirubin, creatine kinase (CK) and alanine aminotransferase (ALT), indicative of toxicity, showed no significant changes due to PLB treatment (Fig. 5D). Together, these results indicate that PLB is well-tolerated and selectively targets AML cells in vivo.Fig. 5PLB selectively reduces AML cells in vivo. (A) Schematic depiction of the intrafemoral injection of patient-derived AML cells (n = 9; due to loss of animal prior to endpoint analysis) or normal bone marrow (BM) cells (n = 10) into non-obese severe combined immunodeficient gamma (NSG) mice. Following an engraftment window, mice were treated three times per week for five weeks with a vehicle control or 2.5 mg/kg plumbagin (PLB) via intraperitoneal (i.P.) injection. At the conclusion of the treatment period, femoral BM was collected. The human (h) CD33+ and hCD45+ population was quantified via flow cytometry. Created using BioRender. (B) Engraftment for both AML patient-derived samples (ID #18) and (C) healthy donor BM. (D) Safety was assessed via the difference in mouse body weight and toxicity markers, assessed by the University of Guelph animal health Lab. Cytogenetics for patient-derived AML samples are provided in Table 1. n.S. non-significant. ****p ≤ 0.0001. Mann-Whitney U test. All data is represented as the mean ± standard error of mean. ALT; alanine aminotransferase, CK; creatine kinase PLB selectively reduces AML cells in vivo. (A) Schematic depiction of the intrafemoral injection of patient-derived AML cells (n = 9; due to loss of animal prior to endpoint analysis) or normal bone marrow (BM) cells (n = 10) into non-obese severe combined immunodeficient gamma (NSG) mice. Following an engraftment window, mice were treated three times per week for five weeks with a vehicle control or 2.5 mg/kg plumbagin (PLB) via intraperitoneal (i.P.) injection. At the conclusion of the treatment period, femoral BM was collected. The human (h) CD33+ and hCD45+ population was quantified via flow cytometry. Created using BioRender. (B) Engraftment for both AML patient-derived samples (ID #18) and (C) healthy donor BM. (D) Safety was assessed via the difference in mouse body weight and toxicity markers, assessed by the University of Guelph animal health Lab. Cytogenetics for patient-derived AML samples are provided in Table 1. n.S. non-significant. ****p ≤ 0.0001. Mann-Whitney U test. All data is represented as the mean ± standard error of mean. ALT; alanine aminotransferase, CK; creatine kinase Modulation of PKM2 disrupts nuclear translocation and c-Myc expression The same cell response (i.e., death) is observed following inhibition/loss [15] or activation of PKM2 [16, 20]. To reconcile the paradox of a similar cell fate with opposing activity levels, we examined the nuclear role of PKM2 following reduced (i.e., genetic loss) or elevated (i.e., pharmacologically induced) activity. Like PK activity, protein kinase function is directly linked to the oligomeric state, as only dimeric PK (i.e., PKM2) can translocate to the nucleus, act as a protein kinase, and upregulate oncogenes such as c-Myc. Thus, DSS crosslinking and immunoblotting assays were used to quantify expression of dimeric and tetrameric PKM2, which were distinguished by molecular weight. In cells with reduced PKM2 activity (i.e., shPKM2), dimeric and tetrameric PKM2 was decreased (shPKM2-21, p = 0.05; shPKM2-23, p < 0.001; Fig. 6A, S3A, S8A&B). Due to its activation effect, PLB treatment increased tetrameric PKM2 and increased the tetramer dimer ratio (p < 0.05, p < 0.01; Fig. 6B, S3B, S8C&D).Fig. 6Both activation and inhibition of PKM2 reduce nuclear PKM2 and c-myc expression. (A) Immunoblot following disuccinimidyl suberate (DSS) crosslinking showed expression of dimeric and tetrameric oligomeric states following genetic inhibition or (B) pharmacological activation of pyruvate kinase M2 (PKM2). (C) Collection of nuclear-enriched fractions followed by immunoblotting demonstrate nuclear PKM2 expression in genetic inhibition (D) and pharmacological activation models. (E) Immunoblots of whole cell lysates demonstrate expression of c-Myc following PKM2 genetic inhibition and (F) pharmacological activation models. (G) Schematic depiction of dimeric PKM2 translocation and protein kinase function within the nucleus. Genetic inhibition of PKM2 reduces PKM2 nuclear translocation. Activation of PKM2 causes tetramerization of PKM2 thereby inhibiting nuclear translocation. Both actions interrupt protein kinase function and effect transcription of oncogenes including c-Myc. Created using BioRender. # p = 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Unpaired, one-tailed Student’s t-test. All data is represented as the mean ± standard error of mean. PLB; plumbagin, shRNA; short hairpin RNA Both activation and inhibition of PKM2 reduce nuclear PKM2 and c-myc expression. (A) Immunoblot following disuccinimidyl suberate (DSS) crosslinking showed expression of dimeric and tetrameric oligomeric states following genetic inhibition or (B) pharmacological activation of pyruvate kinase M2 (PKM2). (C) Collection of nuclear-enriched fractions followed by immunoblotting demonstrate nuclear PKM2 expression in genetic inhibition (D) and pharmacological activation models. (E) Immunoblots of whole cell lysates demonstrate expression of c-Myc following PKM2 genetic inhibition and (F) pharmacological activation models. (G) Schematic depiction of dimeric PKM2 translocation and protein kinase function within the nucleus. Genetic inhibition of PKM2 reduces PKM2 nuclear translocation. Activation of PKM2 causes tetramerization of PKM2 thereby inhibiting nuclear translocation. Both actions interrupt protein kinase function and effect transcription of oncogenes including c-Myc. Created using BioRender. # p = 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Unpaired, one-tailed Student’s t-test. All data is represented as the mean ± standard error of mean. PLB; plumbagin, shRNA; short hairpin RNA Since only the dimeric configuration possesses nuclear activity, we reasoned that the impact of activation or inhibition on PKM2’s oligomeric state would impair nuclear accumulation. Thus, nuclear-enriched fractions were obtained from PLB or control treated cells, and PKM2 was quantified; nuclear protein predominance was confirmed via immunoblotting (Fig. S3G&H). Indeed, genetic loss resulting in decreased dimeric PKM2 led to reduced nuclear translocation (shPKM2-21, p < 0.001; shPKM2-23, p < 0.001; Fig. 6C, S3C, S9A&B). Similarly, tetramerization of PKM2 via pharmacological activation also decreased nuclear PKM2 accumulation (p < 0.05, p < 0.01; Fig. 6D, S3D, S9C&D). Thus, in both scenarios, increased or decreased PKM2 activity leads to reduced nuclear PKM2 accumulation. Finally, to determine the molecular consequence of reduced nuclear PKM2, we examined levels of c-Myc, which is a target of nuclear PKM2 that participates in leukemogenesis, as well as AML cell growth and survival [11, 38]. As expected, reduced nuclear PKM2 decreased c-Myc expression, with genetic inhibition leading to an approximate 50% decrease in c-Myc expression (shPKM2-21, p < 0.05; shPKM2-23, p < 0.05; Fig. 6E, S3E, S10A&B). Similarly, activation of PKM2 significantly reduced c-Myc expression (p < 0.05, p < 0.0001; Fig. 6F, S4F, S10C&D). Overall, the direct comparison of inhibition and activation point to nuclear kinase function as a converging point for cell death through modulation of PKM2 (Fig. 6G). Discussion Nutrient regulation is critical to health, where excess or deficiency leads to cell, or even host, death. Metabolites (e.g. nicotinamide adenine dinucleotide (NADH/NAD+), ATP) [39], ions (e.g., calcium) [40] and molecules (e.g., glucose [41]) are tightly regulated for cell survival and optimal function. In this study, we include PKM2 activity as a tightly controlled cellular process that regulates cancer cell survival. PKM2 exists as a dimer or tetramer, allowing the cell to maintain PK activity within a threshold to support survival. Inhibition of PK activity decreases AML cell growth in vitro and in vivo, while pharmacological PK activation, paradoxically, leads to the same anti-cancer phenotype. Mechanistically, the identical cellular consequence resulting from opposing levels of activity was attributed to the nuclear translocation of dimeric PKM2, offering unique insights into the non-metabolic role of PKM2. Nuclear translocation and protein kinase function of PKM2 are dependent on oligomeric state. As a tetramer, PK possesses high catalytic activity and generates pyruvate, but it lackss the ability to translocate into the nucleus. Alternatively, dimeric PKM2 possesses low catalytic activity, but is translocated to the nucleus by importin-α and importin-β [42, 43]. The nuclear role of PKM2 consists of transcriptional co-activator and protein kinase functions, resulting in the upregulation of oncogenes and transcriptional regulators such as c-Myc [44] and STAT3 [10]. This, in turn, enhances PKM2 expression by promoting alternative splicing favouring PKM2 over PKM1 to create a positive feedback loop [44]. c-Myc is a major oncogenic driver in AML and is overexpressed in patient-derived AML cells, promoting leukemia cell growth and blocking myeloid cell differentiation [12]; genetic inhibition of c-Myc induces cell cycle arrest and cellular differentiation [45]. Similarly, pharmacological inhibition of c-Myc promotes myeloid cell differentiation, downregulates oncogenic targets, suppresses leukemia cell growth, and induces apoptosis [46]. In this study, c-Myc expression is reduced due to impaired nuclear localization of PKM2, leading to anti-AML effects in vitro and in vivo. PKM2 is an anti-leukemia target, with both activation and inhibition acting as anti-cancer strategies. Increased PK activity in malignant cells as a result of increased PKM1 expression alters metabolism and promotes anti-cancer effects [47]. Similarly, pharmacological activation of PKM2 (i.e., increasing PK activity) using small molecule activators such as ML-265 modulate glycolytic metabolism and reduce tumor volume [19]. Interestingly, reduction of PK activity via genetic inhibition of PKM2 also alters metabolism and inhibits proliferation [22]. Similarly, pharmacological inhibition of PKM2 with 3k leads to the induction of autophagy and apoptosis, as well as reduced cancer cell proliferation [48]. The work presented here supports both sets of observations, providing evidence that links the opposing actions through the oligomeric state of the PKM2 protein. In this study, activation and inhibition alter proportion of dimeric of PKM2 expression, resulting in reduced nuclear accumulation. As a result, transcriptional co-activation is impaired, leading to AML cell death. Indeed, similar activities are seen with reactive oxygen species (ROS), where cells possess baseline ROS levels required to support cell signalling pathways (e.g., phosphoinositide 3-kinase (PI3K)-Akt, mitogen-activated protein (MAP) kinase, nuclear factor κ-light-chain-enhancer of activated B cells (NF- κB)) [49]. In AML, endogenous ROS are elevated to support aberrant cell signalling pathways, however excess ROS (i.e., following chemotherapy) leads to ROS-induced cell death [50]. Thus, like ROS – which is maintained in tight control and can promote leukemogenesis with elevated endogenous levels, PKM2 plays a critical role in regulating leukemia cell survival. In summary, activation or inhibition of PKM2, through reduced levels of dimeric protein, leads to selective leukemia cell death in vitro and in vivo. Understanding the mechanism by which similar cell fates arise from opposing cellular pathways, highlights the pivotal role of PKM2 governing cancer cell survival and supports PKM2 as a promising therapeutic target in AML. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1