Enlibio,Your biological research partner
Xinqidi Biotech Co.,Ltd,Wuhan,China 2008-2022
R&D 14th year

Inhibition of the succinyl dehydrogenase complex in acute myeloid leukemia leads to a lactate-fuelle

Issuing time:2022-04-26 14:32


Metabolic programs can differ substantially across genetically distinct subtypes of acute myeloid leukemia (AML). These programs are not static entities but can change swiftly as a consequence of extracellular changes or in response to pathway-inhibiting drugs. Here, we uncover that AML patients with FLT3 internal tandem duplications (FLT3-ITD+) are characterized by a high expression of succinate-CoA ligases and high activity of mitochondrial electron transport chain (ETC) complex II, thereby driving high mitochondrial respiration activity linked to the Krebs cycle. While inhibition of ETC complex II enhances apoptosis in FLT3-ITD+ AML, cells also quickly adapt by importing lactate from the extracellular microenvironment. 13C3-labelled lactate metabolic flux analyses reveal that AML cells use lactate as a fuel for mitochondrial respiration. Inhibition of lactate transport by blocking Monocarboxylic Acid Transporter 1 (MCT1) strongly enhances sensitivity to ETC complex II inhibition in vitro as well as in vivo. Our study highlights a metabolic adaptability of cancer cells that can be exploited therapeutically.


Acute myeloid leukemia (AML) is a stem cell disease that occurs mostly in adults upon the abnormal transformation of hematopoietic stem/progenitor cells (HSPCs) in the bone marrow. Genetic aberrations lie at the heart of leukemia, and the majority have now been identified1,2. These and other studies uncovered the heterogeneity of the disease, whereby the combined set of mutations that give rise to AML can differ substantially between patients, and even within patients genetically distinct clones can co-exist3,4,5,6,7. As a result of the different mutational spectra, AML patients present a heterogeneous transcriptomic, epigenomic, and proteomic landscape, which leads to a variety of transcriptional networks, cell biological properties and drug sensitivities7,8. The complex and diverse mutational architecture of AML, which also evolves over time or as a consequence of treatment, creates one of the biggest challenges in AML therapy.

Metabolism has emerged as a targetable entity that might provide alternative means for cancer therapy. AML blast cells can be highly glycolytic, which correlates with poor prognosis, and inhibition of glycolysis enhances cytotoxicity of Cytarabine (Ara-C)9,10. Deletion of several glycolysis enzymes delayed leukemia progression11. In contrast, recent papers also indicated that in certain subtypes of AML leukemic stem cells (LSC) might in fact be particularly sensitive to OXPHOS inhibition12,13,14,15. Inhibition of the mitochondrial electron transport chain (ETC) complex I impaired proliferation and induced apoptosis in acute myeloid leukemia16, in particular in those AMLs characterized by high OXPHOS dependency16.

These studies have suggested that the heterogeneity in AML also extends to their metabolic profiles. A thorough understanding of how mutations that drive specific transcriptional programs might underlie such differences in metabolism is needed. Genetic abnormalities that involve internal tandem duplications or tyrosine kinase domain mutations in the FLT3 receptor gene (FLT3-ITD, FLT3-TKD) occur in ~30% of AML patients and are associated with poor prognosis and poor response to chemotherapy17,18. FLT3-ITD mutations drive distinct gene expression signatures19, FLT3-ITD+ AMLs are characterized by increased levels of reactive oxygen species (ROS) that lead to increased DNA double-strand breaks (DSBs)20 and inhibition of the FLT3-ITD imposes a specific dependency on glutaminolysis21.

In this study, we evaluate metabolic signatures in a large cohort of AML patients with FLT3-ITD by combining quantitative proteome analyses with functional metabolic approaches. We identify Electron Transport Chain (ETC) complex II as a specific targetable determinant of the OXPHOS state of AML. We find that under steady-state conditions cells can already utilize lactate as carbon source to drive the TCA cycle, and blocking ETC complex II makes the cells more sensitive to inhibition to MCT1-mediated exchange of lactate with the environment. Detailed insights into metabolic programs within specific genetic subtypes of AML will not only improve AML risk-stratification and allow tailored metabolic therapy approaches but will also increase our understanding of metabolic adaptation upon therapeutic interventions.


FLT3-ITD mutant AMLs display distinct OXPHOS-driven metabolic signatures compared to FLT3-wt AMLs

Previously, we identified that considerable metabolic heterogeneity exists between different subtypes of AML, whereby FLT3wt AMLs, with high expression of Pyruvate Dehydrogenase Kinase 1 (PDK1) displayed more glycolytic phenotypes and FLT3-ITD AMLs were more OXPHOS-driven22. In order to uncover the molecular mechanisms underlying the OXPHOS metabolism of FLT3-ITD AMLs we first characterized FLT3-ITD-specific protein expression profiles by analyzing our label-free quantitative proteome datasets of the stem/progenitor-enriched CD34+ fraction of FLT3-wt patients (n = 26) and FLT3-ITD+ patients (n = 15) (Supplementary Fig. 1a and Supplementary Data 1 and 2)7. Gene ontology (GO) analyses revealed that upregulated proteins in FLT3-ITD+ AML were significantly enriched for various metabolism-related GO terms, in particular those related to the respiratory electron transport chain, the Krebs cycle (tricarboxylic acid (TCA) cycle), and lipid metabolism (Supplementary Fig. 1b). Some of the overexpressed proteins in CD34+ blasts derived from AML patients with FLT3-ITD are linked to the GO terms NADH regeneration/oxidation and respiratory electron transport chain (Supplementary Fig. 1c). These included Succinate-CoA Ligases (SUCLG1/2) and several others that are involved in NADH regeneration, regulate mitochondrial respiration-dependent energy production, and control ETC activity and the TCA cycle (Fig. 1a and Supplementary Data 2).

Fig. 1: The OXPHOS state correlates with high expression of SDH and SUCLG proteins and ETC complex II activity in AML.
figure 1

a Protein expression levels (log10) of SUCLG1 and SUCLG2 in CD34+ FLT3-ITD+ (n = 15), FLT3-wt (n = 26) AML patients and in healthy peripheral blood mobilized stem cells (PBSC) (n = 6) (mean + /− SEM). b Forest plot for multivariable analysis for SDHA, SDHB, SUCLG1, and SUCLG2 as independent predictors of overall survival in AML patients. Cox hazard proportional model for overall survival using continuous values for gene expression signatures from the PRECOG database. Hazard ratios (HR) > 1 indicates that increasing values for continuous variable or the first factor for categorical variables has the poorer outcome. HR and their respective 95% confidence interval are indicated with blue circle and a line, respectively. Public datasets were described by the title of gene series expression (GSE) or by the code of the previous studies (TCGA: n = 121; GSE1427: n = 137; GSE10358 n = 182; GSE12417 n = 79; Bullinger: n = 163). c Mitochondria-driven energy flux through TCA-cycle and ETC complexes in FLT3-ITD+ AMLs. Functionally more active metabolic pathways are highlighted in green. d Basal OCR levels in CD34+-sorted FLT3-ITD+ (n = 6) and FLT3-wt (n = 13) AML primary samples. Each dot represents an independent patient sample, mean of technical quadruplicates. e Correlation of SUCLG2 and SDHB protein expression (log10) versus maximal OCR in CD34+ AML primary samples. f ETC complex II enzymatic activity levels in CD34+-sorted FLT3-ITD+ (n = 7) and FLT3-wt (n = 6) AML and healthy PBMSCs (n = 2). Each dot represents an individual sample (mean of technical duplicates) (mean + /− SEM). g Correlation of enzymatic ETC complex II activity compared to SDHA, SDHB, and SUCLG2 protein expression (log10) in CD34+ AML primary samples (n = 5). h ETC complex II activity in different protein amounts and (i) in 240 μg protein input of AML cells. Mean + /− SEM of independent replicates, in technical duplicates. j ETC complex I activity in (240 μg protein) AML cells (n = 1 or 2). k Western blot analysis of ETC subunits; complex I (NDUFB8,18 kD), II (SDHB, 29 kD), III (UQCRC2,48 kD), IV (COX II, 22 kD), and V (ATP5A, 54 kD) in AML cells. An equal amount of protein loaded per cell line and Actin is the loading control and the image is a representative of independent triplicates. e, g linear regression analysis, (a, d) Student’s t test (two-sided) or (f, i) one-way, (h) two-way ANOVA for multiple comparisons.

SUCLG and SDH (Succinate dehydrogenase, complex II of the ETC) proteins are functionally linked, whereby SUCLG proteins within the Krebs cycle generate succinate that in turn is metabolized by SDH to generate FADH 2 and consequently ubiquinone reduction. CD34+ AML blasts derived from FLT3-ITD+ patients expressed significantly higher levels of SUCLG1 and 2, although also a subset of FLT3-wt patients expressed high levels of ETC complex II proteins (Fig. 1a). We noted that the FLT3-wt subgroup with high levels of SUCLG2 was genetically distinct from the FLT3-wt group with low levels of SUCLG2. In our proteome, we observed that all NPMcyt and inv16 AML subtypes resided within the FLT3-wt/SUCLG2high group (Supplementary Fig. 1d). Pearson coefficients were calculated using the quantitative proteome comparing the FLT3-WT/SUCLG2high group with the FLT3-WT/SUCLG2low group, and the ranked list was used for GSEA which indicated significant enrichment for NPM1mutsignatures, inv16 signatures, and MLL-rearranged AML signatures (Supplementary Fig. 1e). Next, we independently confirmed these data using the MILE AML transcriptome dataset (GSE13159) and healthy HSPCs (GSE42519) and observed that indeed the inv16 and MLL-rearranged AML subtypes express significantly higher levels of SUCLG2 compared to other AML subgroups (Supplementary Fig. 1f). Finally, we correlated expression of genes associated with glycolysis, TCA cycle, and glutaminolysis-associated with overall survival. A cox hazard proportional model for overall survival using continuous values for gene expression signatures retrieved from the PRECOG database23 and a forest plot for multivariable analysis for SDHA, SDHB, SUCLG1, and SUCLG2 in shown in Fig. 1b, indicating that they act as independent predictors of overall survival in AML patients (see also Supplementary Data 1).

The OXPHOS state correlates with high expression of SDH and SUCLG proteins and ETC complex II activity in AML

The mitochondrial metabolic regulatory pathways that are potentially involved in FLT3-ITD+ AMLs which we wished to functionally study further in detail are highlighted in green in Fig. 1c. Oxygen consumption rates (OCR) were functionally assessed by Seahorse in primary AML cells, indicating that OCR was significantly higher in FLT3-ITD+ AMLs compared to FLT3-wt AMLs (Fig. 1d), which correlated with SUCLG2 or SDHB expression (Fig. 1e). These data are in line with our previous observations indicating FLT3-ITD+ cell lines display higher OCR compared to FLT3-wt cell lines22. We transduced healthy CB CD34+ cells with lentivectors overexpressing FLT3-ITD or MLL-AF9 and observed that OCR was increased in both conditions, while ECAR levels were slightly reduced in FLT3-ITD-transduced cells and slightly increased in MLL-AF9-transduced cells (Supplementary Fig. 2a). mRNA expression of SUCLG2 and SDHA were shown to be significantly higher in FLT3-ITD-transduced CB cells but not in MLL-AF9-transduced cells (Supplementary Fig. 2b). CD34+ stem/progenitor cells from FLT3-ITD+ AML patients displayed higher ETC complex II activity compared to FLT3-wt AMLs (Fig. 1f), which correlated with SDHA, SDHB, and SUCLG2 protein expression (Fig. 1g). The OXPHOS-driven cell lines MOLM13 and MV411 displayed higher ETC complex II activity compared to the FLT3-wt glycolytic line HL60 and FLT3-wt K562 cells (Fig. 1h, i). A similar trend was seen for ETC complex I activity, although this did not reach significance (Fig. 1j). Interestingly, the MLL-AF9-driven OXPHOS cell line THP1 displayed ETC complex I/II activity levels comparable to HL60 (Fig. 1h–i). At the protein expression level, higher levels of ETC complexes I, II, and IV were detected in FLT3-ITD+ AML cell lines compared to FLT3-wt cell lines (Fig. 1k). These data nicely align with our previous observations where in an individual patient FLT3-ITD+ and FLT3-wt subclones could prospectively be identified and isolated7, whereby The FLT3-ITD+ subclone also expressed higher levels of ETC complexes I, II, and IV22. Lentiviral overexpression of the FLT3-ITD in the FLT3-wt AML cell lines HL60 or TF1 was sufficient to increase ETC complex II activity (Supplementary Fig. 2c). Collectively, these data indicate that a higher OXPHOS state correlates with high expression of SDH and SUCLG proteins, which coincides with the higher OCR and ETC complex II activity seen in FLT3-ITD+ AMLs.

Inhibition of the mitochondrial succinyl dehydrogenase complex is a vulnerability in FLT3-ITD+ AML

With the aim to further improve personalized treatment strategies in AML based on distinct OXPHOS signatures, we evaluated the efficacy of ETC I (rotenone) and several ETC II inhibitors (thenoyltrifluoroacetone (TTFA) and 3-Nitropropionic acid (3-NPA)) in FLT3-ITD+ versus FLT3-wt AMLs and healthy CB-derived CD34+ cells. Within 24 h a strong increase in apoptosis was seen in the FLT3-ITD+ AML cell lines MV4–11 and MOLM13 upon incubation with ETC complex I/II inhibitors, while moderate effects were seen in FLT3-wt THP1 and HL60 cells (Fig. 2a). CB-derived CD34+ cells were also not sensitive to ETC complex II inhibition although a slight increase in apoptosis was seen upon ETC complex I inhibition (Fig. 2b). In primary CD34+ AML cells, the response to ETC complex II inhibition was heterogeneous (Fig. 2c), whereby FLT3-ITD+ AMLs were within the most sensitive group (Fig. 2d). These data are in line with previous observations indicating that FLT3-wt/PDK1high primary AML samples are not very sensitive to complex II inhibition22. Twenty-four hours of treatment with the PDK1 inhibitor DAP, to drive cells from glycolysis into an OXPHOS state, did not significantly reduce the viability of FLT3-ITD+ AMLs as compared to ETC complex II inhibition (Supplementary Fig. 2d), again in agreement with previously published data22. Cell viability across a panel of independent non-malignant CD34+cells from normal bone marrow (NBM) (n = 1), PBSC (n = 5), and CB (n = 5) was not affected by TTFA (Fig. 2e). To functionally show that TTFA inhibits ETC complex II activity we treated MOLM13, HL60, and HL60 cells transduced with FLT3-ITDs. The enhanced activity of ETC complex II in FLT3-ITD+ MOLM13 cells and in HL60 cells overexpressing FLT3-ITDs could be reduced by TTFA (Fig. 2f and Supplementary Fig. 2e). We also tested the potential apoptotic effects of ETC complex II inhibitors combined with the FLT3-ITD inhibitor AC-220 (quizartinib) in MOLM13 cells and showed that apoptosis was significantly further enhanced after co-inhibition (Supplementary Fig. 2f).

Fig. 2: Inhibition of the mitochondrial succinyl dehydrogenase complex is a vulnerability in FLT3-ITD+ AML.
figure 2

a (%) Annexin V+ DAPI+ AML cells (n = 2 FLT3-ITD+, n = 2 FLT3-wt) upon ETC complex II inhibitors (thenoyltrifluoroacetone 200 μM, as TTFA, and 3-nitropropionic acid 2 mM, as 3-NPA) and complex I inhibitor rotenone treatment (20 μM) for 24 h. Data presented as + /− SEM of independent measurements (each dot) measured in three technical replicates. FC = fold change in apoptosis compared to controls. For gating strategy, see Source Data. b (%) Annexin V+ DAPI+ healthy CD34+ CB cells upon ETC complex II inhibitors (TTFA 200 μM, 3-NPA 2 mM) and complex I inhibitor rotenone treatment (20 μM) (24 h). Data presented as mean + /− SEM of three independent measurements (each dot) measured in technical triplicates. c (%) DAPI viable CD34+ AML primary patient cells following 100 μM TTFA (48 h) on MS5 stromal co-culture. Each dot represents mean + /− SEM of technical replicates measured from each independent primary patient sample. d (%) DAPI viable CD34+ AML primary patient cells (n = 10 FLT3-ITD+, n = 8 FLT3-wt AMLs) following 100 μM TTFA treatment (48 h) on MS5 stromal co-culture. Each dot represents an independent patient sample, measured in technical triplicates (mean + /− SEM). e (%) DAPI viable CD34+ sorted healthy NBM (n = 1), CB (n = 5), and PBSC (n = 5) cells after 100 μM TTFA treatment (48 h). Each dot represents technical replicates for NBM, and mean of independent primary samples for CB and PBSC. Data was normalized to the control in each group (mean + /− SEM). f Colorimetric ETC complex II activity measurement in AML cells (n = 2) in the presence and absence of 200 μM TTFA for 24 h. Data were normalized to protein amounts. Each bar represents mean + /− SEM of two biological replicates measured in two technical replicates. g, h Cumulative cell growth of MOLM13 (g) and HL60 (h) cells transduced with shluc, shSUCLG1#1, shSUCGL1#2, shSUCLG2#1, shSUCLG2#2. i, j (%) Annexin V+ DAPI+ shluc, shSUCLG1#1, shSUCGL1#2, shSUCLG2#1, shSUCLG2#2 MOLM13 (i) and MV411 (j) cells in the presence and absence of TTFA treatment. a, c, d, f Student’s t test (two-sided) or (g, h, i, j) two-way ANOVA for multiple comparisons.

As an independent validation to evaluate SUCLG activity in AML subtypes, we used SUCLG1 and SUCLG2-directed shRNAs to downregulate their expression in FLT3-wt (HL60 and K562) and FLT3-ITD+AMLs (MOLM13 and MV411) (Supplementary Fig. 2g). Downregulation of SUCLG1 or SUCLG2 significantly reduced the growth of FLT3-ITD+ MOLM13 and MV411 cells (Fig. 2g and Supplementary Fig. 2h, respectively), while the proliferation of FLT3-wt HL60 and K562 cells were unaffected (Fig. 2h and Supplementary Fig. 2i, respectively). TTFA-induced apoptosis levels were significantly lower after knockdown of SUCLG1/2 s in FLT3-ITD+ AML cells (Fig. 2i, j).

ETC complex II inhibition results in decreased mitochondrial respiration which can be rescued by the addition of exogenous lactate

To investigate the metabolic consequences following 24 h of ETC complex II inhibition (Fig. 3a), we first measured mitochondrial respiration and glycolytic activity by Seahorse. The OCR was markedly reduced in FLT3-ITD+ AML cell lines MOLM13 and MV4–11 upon 24 h TTFA treatment, but not significantly in FLT3-wt cells (Fig. 3b). At earlier time points, 6 h TTFA treatment with 100 μM was already sufficient to reduce OCR levels while apoptosis was still not observed at that concentration (Supplementary Fig. 2j). ECAR was also reduced upon TTFA treatment (Fig. 3c).

Fig. 3: Inhibition of ETC complex II results in lower respiration and decreased lactate efflux in FLT3-ITD+ AML cells.
figure 3

a Schematic representation of the metabolic outcome of mitochondrial energy flux in FLT3-ITD+ AMLs after TTFA. Functionally more active metabolic pathways are highlighted in green (mean + /− SEM). b Effect of TTFA (100 μM, 24 h) on oxygen consumption rates (OCR) in AML cell lines and an AML primary patient. Data normalized to controls (mean + /− SEM of three biological replicates for cell lines and n = 1 AML primary sample (technical triplicates are shown). c Effect of TTFA (100 μM, 24 h) on extracellular consumption rate (ECAR) in AML cell lines (mean + /− SEM). d, e OCR levels (d) and ECAR levels (e) in shluc, shSUCLG1#1, shSUCGL1#2, shSUCLG2#1, shSUCLG2#2 MOLM13 and MV411 cells. f Spectrophotometric real-time measurement of lactate, glutamine, and glucose concentration in the culture medium of both treated/untreated (TTFA 100 μM, 32 h) MOLM13 cells. Data represent mean + /− SEM of three biological replicates, in four technical replicates. g Flux of lactate secreted to the media, corrected by cell and incubation time, measured in MOLM13 and HL60 cells upon ETC complex II inhibitor (TTFA 100 μM) at 10, 24, and 32 h (mean + /− SEM, four biological replicates (each dot) measured in four technical replicates). h (%) Annexin V+DAPI+ AML cells (n = 2 FLT3-ITD+, n = 2 FLT3-wt) upon TTFA (200 μM) with/without 2 mM lactate addition in the culture medium (24 h). Each bar represents mean + /− SEM of four biological replicates (measured in technical triplicates). Percentage apoptotic cell population difference is indicated above the bars. i (%) Annexin V+DAPI+ in AML FLT3-ITD+ CD34+ cells upon TTFA (200 μM) with/without 2 mM lactate addition in the culture medium for 24 h. Each bar represents mean + /− SEM of technical triplicates. j Fold change in DAPI- viable AML cell lines upon TTFA (200 μM) with/without 2 mM lactate or 2 mM Pyruvate addition in the culture medium (24 h) (mean + /− SEM, each dot shows independent biological replicates). k Extracellular lactate concentration with/without increasing concentrations of MCT1 inhibitor CHC (α-cyano-4-hydroxycinnamic acid 400, 800 μM) and AZD3965 (25, 50, 100, 200 nM) for 24 h in MOLM13 cells. Cell viability was comparable between groups. b, e, f, g, i, h, j, k Student’s t test (two-sided, Mann–Whitney), (c) one-way ANOVA Kruskal–Wallis test or (d) two-way ANOVA for multiple comparisons.

Downregulation of SUCLG1/2 in FLT3-ITD+ AML cell lines MOLM13 and MV4–11 also resulted in significantly lower OCR levels (Fig. 3d), but not in shSUCLG1/2 transduced FLT3-wt K562 cells (Supplementary Fig. 2k). Remarkably, the extracellular acidification rate (ECAR) was also reduced after SUCLG1/2 knockdown (Fig. 3e). ECAR activity levels remained unaffected after SUCLG1/2 knockdown in FLT3-wt K562 cells (Supplementary Fig. 2k).

Next, we measured glucose, glutamine, and lactate concentrations in the culture media of FLT3-ITD+ MOLM13 AML cells at 10, 24, and 32 h after ETC complex II inhibition. These studies revealed that, whereas glutamine consumption was only slightly reduced after 32 h of ETC complex II inhibition, glucose consumption was almost entirely abolished, corresponding to a strong reduction of lactate excretion (Fig. 3f). This was not simply due to a reduction in cell viability or proliferation because the specific flux of lactate secretion, calculated considering cell number and time, was comparably reduced (Fig. 3g). These reductions in extracellular lactate were not seen when ETC complex II activity was inhibited in HL60 cells (Fig. 3g).

Lactate was shown to fuel oxidative cancer cells instead of being exported as a waste product24. We then wondered if the FLT3-ITD+ cells might also consume lactate. Possibly, cells might rewire their metabolism and consume (more) lactate as an alternative carbon source in the non-truncated TCA-cycle branches in response to ETC complex II inhibition. We first questioned whether the exogenous addition of lactate would rescue the apoptotic phenotype induced upon ETC complex II inhibition. Indeed, a significant reduction in TTFA-induced apoptosis was observed when either MOLM13 or MV4–11 cells were supplied with 2 mM additional lactate (Fig. 3h). This was not observed in THP1 (MLL-AF9), HL60 or healthy CB CD34+ cells that were also intrinsically less sensitive to TTFA treatment (Fig. 3h). TTFA-induced apoptosis was also reduced in primary CD34+ FLT3-ITD+ AML cells in the presence of 2 mM exogenous lactate (Fig. 3i). We also tested whether pyruvate could act as an alternative energy substrate to drive TCA-cycle activity, and similar to lactate, the viability was increased upon the addition of exogenous pyruvate to MOLM13 or MV4–11 cells, but not HL60 cells (Fig. 3j). Lactate can be imported via the monocarboxylate transporter 1 (MCT1, encoded by SLC16A1) which is expressed both at the plasma membrane as well as the mitochondrial membrane where it primarily facilitates lactate influx fueling the TCA cycle25. We evaluated two independent MCT1 inhibitors (α-cyano-4-hydroxycinnamate (CHC) and AZD3965), and exposure of MOLM13 cells to these inhibitors resulted in elevated levels of extracellular lactate, indeed indicating that MOLM13 cells can actively consume lactate (Fig. 3k). MCT1, COX (cytochrome oxidase, ETC complex IV), and LDH (lactate dehydrogenase) co-localize in the mitochondrial membrane to form a mitochondrial lactate oxidation complex26. The lactate dehydrogenase B (LDHB) isoform reduces lactate to pyruvate in cells that utilize lactate as a fuel for oxidative metabolism27. We observed in primary FLT3-ITD+ blasts from AML patients that protein levels of MCT1 and LDHB were significantly positively correlated, opposite to what was seen in FLT3-wt AMLs (Supplementary Fig. 2l).

We then wished to address this further in real-time oxygen consumption measurements using Seahorse and Oroboros. As expected, inhibition of ETC complex II activity resulted in a reduction in OCR in FLT3-ITD+ MV4–11 and MOLM13 cells (Fig. 4a) while this was not observed in FLT3-wt HL60 cells (Supplementary Fig. 3a) when TTFA was injected alone. After the addition of increasing concentrations of exogenous lactate, OCR could be restored, but not when MCT1 transporters were inhibited (Fig. 4a). This effect on OCR was noted to a lesser extent in glycolytic HL60 cells (Supplementary Fig. 3a). Similar observations were done in a panel of two primary FLT3-ITD+ AML samples (Fig. 4b). These Seahorse data in which lactate was transported through the intact cell plasma membrane were then independently confirmed in Oroboros measurements where we initially permeabilized the plasma membrane to demonstrate lactate transport over the mitochondrial membrane in MOLM13 cells (Fig. 4c). First, ADP, malate, and lactate were injected to study the impact on mitochondrial oxygen consumption, after which TTFA and the MCT1 inhibitor were consecutively added to evaluate their effect on mitochondrial respiration. Finally, the ETC uncoupler FCCP was added to induce maximal respiration. Injection of TTFA reduced respiration in the absence of lactate, and respiration was also reduced when MCT1 was inhibited alone in the presence of lactate (Fig. 4c, see Supplementary Fig. 3b for real-time data). The TTFA-induced reduction in respiration was less pronounced when lactate was also present, which was not the case when lactate influx was blocked by MCT1 inhibitors CHC or AZD3965, resulting in a further significant reduction in mitochondrial oxygen consumption (Fig. 4c). The addition of lactate or the presence of CHC or AZD3965 did not show a significant impact on mitochondrial oxygen consumption of HL60 cells when TTFA was present (Supplementary Fig. 3c, d for representative real-time data). We also compared the impact of ADP, malate, and lactate on the mitochondrial oxygen consumption rate in permeabilized CD34+-sorted AML patient samples and showed that FLT3-ITD+ cells have higher basal and ADP-stimulated mitochondrial respiration compared to FLT3-wt (Supplementary Fig. 4e). FLT3-wt cells demonstrated lower respiration which was less impacted on by TTFA in comparison to its effects on ITD+ cells (Supplementary Fig. 3e). These data collectively show that AML cells have the capacity to restore mitochondrial respiration when lactate is shuttled through MCT1 transporters upon inhibition of the succinyl dehydrogenase complex.

Fig. 4: Lactate fluxes into TCA cycle and restores respiration activity of FLT3-ITD+ AMLs.
figure 4

a Relative real-time OCR in MV411 and MOLM13 cells. Either 600 μM TTFA or 600 μM TTFA with 600 μM CHC were injected at the start of the measurement and then 2 mM sodium lactate was injected every 25 min serially. Data normalized to the initial OCR at 100. Data show mean of four biological replicates, in quadruplicate repeats (mean + /− SEM). b Similar to Fig. 4a, real-time OCR in two FLT3-ITD+ AML primary samples after CD34+ sorting followed by 24 h MS5 culture. Data show two independent biological replicates measured in quadruplicates (mean + /− SEM). cReal-time high-resolution respirometry measurement in MOLM13 cells in the presence/absence of sequentially injected digitonin (2.5 μg/ml), malate (2 mM) with/without lactate (8 mM), ADP (1 mM), TTFA (200 μM), CHC (400 μM), (or AZD3965 (100 nM)), FCCP (1.5 μM) in MOLM13 (2.5 × 106 cells/ml). After digitonin, Malate and ADP are used at the beginning and FFCP is used at the end of all measurements as major respiration substrates. Additional injection conditions are indicated below the bar graphs. Injections followed every 3–5 min after respiration was stable. Bar graphs demonstrate end-point respiration levels after FCCP was injected according to each condition combination indicated below the x axis (n = 3 biological replicates, mean + /− SEM). For real-time representative data, see Supplementary Fig. 3b. d Internal relative abundance of TCA-cycle associated metabolites in MOLM13. Cells were grown with/without 200 μM TTFA and with/without 8 mM and 20 mM (high) labeled [U-13C3] lactate (6 h). Ion counts for each metabolite were normalized to internal recovery standards, and to the protein amount of the cell pellets (mean + /− SEM of independent triplicates). e (%) Isotopologue distributions (13C enrichment in each number of carbons) of TCA-cycle associated metabolites in MOLM13. Cells were grown with/without 200 μM TTFA and with 8 mM and 20 mM (high) labeled with [U-13C3] lactate for 6 h. f (%) total labeled molecules (Σm) obtained from the incubation with [U-13C3] lactate in the metabolites shown in (e). g m + 2 pools of malate and aspartate from [U-13C3] lactate in MOLM13 cells. Data are mean−/+SEM, n = 3 replicates. ag Student’s t test (two-sided) or (a) two-way ANOVA for multiple comparisons.

To formally prove that lactate can be utilized as a carbon source and to trace in detail how it is metabolized by leukemic cells we performed 13C3-labeled lactate-tracing studies. MOLM13 cells were grown in the presence of 8 mM or 20 mM uniformly stable isotope-labeled lactate (U-13C3 lactate) with or without TTFA for 6 h after which metabolites were extracted for gas chromatography-mass spectrometry (GC-MS) analyses (Supplementary Data 3 for metabolite list). We first analyzed total intracellular metabolite concentrations in control cells grown without labeled lactate and compared that to cells grown with U-13C3 lactate (8 mM), with TTFA alone, with TTFA + U-13C3 lactate (8 mM), or with TTFA + U-13C3 high lactate (20 mM). While the addition of labeled lactate did not result in increased intracellular lactate in the absence of the ETC complex II inhibitor, addition of TTFA resulted in a significant decrease in intracellular lactate when exogenous labeled lactate was not added (Fig. 4d). Furthermore, we observed that the succinate pool was significantly larger after TTFA treatment, validating the inhibiting effects of TTFA on ETC complex II activity, resulting in intracellular succinate accumulation (Fig. 4d). While main TCA-cycle metabolites citrate, glutamate, α-ketoglutarate (Fig. 4d), and others (Supplementary Fig. 3f) were not impacted by TTFA, we also noted significant increases in aspartate, fumarate, and malate upon ETC complex II inhibition (Fig. 4d). Intriguingly, however, while upon further addition of exogenous labeled lactate the effect of TTFA on succinate accumulation was further pronounced, the opposite was observed for aspartate and fumarate, suggesting that lactate processing is able to overcome the increase in fumarate and aspartate induced by ETC complex II inhibition (Fig. 4d).

Next, we investigated label incorporation in various TCA-cycle intermediates and confirmed that lactate can indeed act as a carbon source, with the highest label incorporation in citrate and its downstream metabolites under both untreated and treated conditions (Fig. 4e, f). Upon inhibition of ETC complex II activity, a significant accumulation of the 13C label was observed in succinate, in line with the threefold increase in succinate levels observed in TTFA treated cells and incubated with 13C3 lactate (Fig. 4e, f). Total pools of labeled carbon incorporation from 13C3 lactate in TCA-cycle intermediates were significantly enhanced when a higher concentration (20 mM) of exogenous lactate was added. Our mass isotopomer distribution analysis revealed that lactate was converted to pyruvate, after which it was further being metabolized through PDH (Pyruvate dehydrogenase, label incorporation indicated in purple in Supplementary Fig. 4) and PC (Pyruvate carboxylase, label incorporation indicated in green in Supplementary Fig. 4) to generate citrate that can be exported out of the mitochondria resulting in m + 3 and m + 5 labeling in the cytosol (Supplementary Fig. 4a). Inside the mitochondria, citrate can be metabolized via the oxidative metabolism pathway through mitochondrial IDH (Isocitrate dehydrogenase) to generate α-ketoglutarate (α-KG) and consecutively succinate (resulting in m + 2 and m + 4 labeling, Supplementary Fig. 4a). However, this α-KG can also be metabolized via the reductive metabolism pathway (Supplementary Fig. 4b), resulting in label incorporation at positions m + 2 and m + 4 of citrate. This m + 2/m + 4-labeled citrate can then be converted into oxaloacetate (m + 2) and in malate, aspartate, and fumarate through counter-clockwise reductive carboxylation reactions (Supplementary Fig. 4b). This counter-clockwise reaction will generate different labeling patterns in aspartate, malate, and fumarate depending on the m + 2 (metabolized) or m + 3 (not metabolized) labeled cytosolic citrate (Supplementary Fig. 4). In this regard, while total carbon (e.g., m + 1, m + 2, m + 3, m + 4, and m + 5) labeling of metabolites from U-13C3 lactate was comparable between control and treated cells (Supplementary Data 3), the total pools of m + 2 isotopologues of both malate and aspartate were significantly enriched (at ~18% and 22%, respectively) after inhibition of ETC complex II activity (Fig. 4g), indicating that this mechanism is favored in ETC complex II-inhibited cells. Although further studies are required to obtain a comprehensive detailed insight, this suggests that upon ETC complex II inhibition lactate is channeled into the TCA cycle predominantly via the cytosolic reductive carboxylation route rather than directly via the Krebs cycle (Supplementary Fig. 4b). Cytosolic reductive carboxylation rerouting was previously identified from labeled glutamine and glucose28,29 when SDH is truncated, and was also predicted by computer modeling-based simulations in SDH deficiency30. Taken together, we show that AML cells can utilize lactate as a carbon source at the steady-state (Supplementary Fig. 4) resulting in extensive incorporation of uniformly labeled 13C3-lactate in 40–60% of all main TCA-cycle intermediates, even when sufficient amounts of glucose and glutamine were present in the medium. When ETC complex II is inhibited, an increased exogenous pool of lactate is able to rescue respiration, and FLT3-ITD+ AML cells continue to actively metabolize and incorporate lactate, now more towards the reductive cytosolic branch of the truncated TCA cycle (Supplementary Fig. 4b).

Dual inhibition of MCT1 transporters and ETC complex II improved treatment efficacy in vitro and in vivo

Since inhibition of ETC complex II activity in FLT3-ITD+ AML resulted in an increasing shift towards lactate-driven OCR we wished to exploit this therapeutically by inhibiting both ETC complex II (using 3-NPA or TTFA) and MCT1 transporters to prevent lactate import (Fig. 5a). Inhibition of MCT1 (using CHC) together with ETC complex II resulted in a strong induction of apoptosis in FLT3-ITD+ cell lines, which was much less pronounced in FLT3-wt cell lines (Fig. 5b). We independently confirmed the therapeutic efficacy by combining ETC complex II inhibition with the second MCT1 inhibitor AZD3965 in MOLM13 cells and observed higher apoptosis upon co-inhibition (Fig. 5c). Also, primary AML samples were evaluated when grown on bone marrow stromal cells and the strongest decreases in cell viability upon co-inhibition of SDH and MCT1 were seen in primary FLT3-ITD+ AMLs, whereby healthy control CB-derived CD34+ cells were not sensitive to this combination of inhibitors (Fig. 5d). In primary FLT3-ITD+ AMLs, the extracellular lactate levels decreased upon TTFA treatment, which was not observed in FLT3-wt AMLs or in healthy CB and PB samples (Fig. 5e and Supplementary Fig. 5a). Finally, we also wished to confirm the efficacy of ETC complex II inhibition in glycolytic (Fig. 5f–i) and OXPHOS-dependent (Fig. 5j–n) leukemias in vivo and we also tested the combination of ETC complex II and lactate transporter inhibition in vivo in FLT3-ITD+ AML (Fig. 5j). We transplanted luciferase-transduced FLT3-ITD+ MV4–11 cells that are OXPHOS-driven (Fig. 1i, k) and after confirmation of engraftment (Supplementary Fig. 5b) mice were randomly assigned to four treatment groups: vehicle, 3-NPA, CHC or combination of both. Inhibition of ETC complex II did not impact on the leukemia development of glycolytic FLT3-wt HL60 cells (Fig. 5g–i). Although BM AML chimerism after 3-NPA treatment was comparable to controls in both HL60 and MV4–11 transplanted mice, in contrast, 3-NPA treatment decreased disease burden in peripheral blood and resulted in a reduced spleen size in OXPHOS-dependent FLT3-ITD+ MV4–11 cells (Fig. 5k–m).

Fig. 5: Combined inhibition of ETC complex II and lactate transporters impairs leukemogenesis in ITD+ AMLs in vitro and in vivo.
figure 5

a Schematic representation of the truncated TCA cycle and OXPHOS in FLT3-ITD+ AMLs after ETC complex II and MCT1 transporters inhibition. b (%) Annexin V+DAPI+ AML cells upon ETC complex II inhibitors (TTFA 200 μM, 3-NPA 2 mM) with/without 800 μM CHC for 24 h (mean + /− SEM of three biological replicates (each dot)). c (%) Annexin V+DAPI+ MOLM13 cells upon ETC complex II inhibitors (TTFA 100 μM and 3-NPA 2 mM) with/without 50 nM AZD3965 for 24 h (mean + /- SEM of three biological replicates). d % DAPI- CD34+ AML patient cells (n = 6 FLT3-ITD+, n = 10 FLT3-wt) and in CD34+ cord blood (CB) cells (n = 3) treated with TTFA 100 μM with/without 800 μM CHC for 48 h. Data normalized to controls. Each dot = independent patient, measured in technical triplicates (mean + /− SEM). e Extracellular lactate concentration [mM] in culture medium of cells screened in (d); CD34+ AML patient cells (n = 4 FLT3-ITD+, n = 7 FLT3-wt) and in healthy CD34+ cord blood (CB) cells (n = 3) and peripheral blood mobilized stem cells (PBSC) treated with TTFA 100 μM with/without 800 μM CHC (48 h) (mean + /− SEM). (f) FLT3-wt HL60-transplanted mice were treated daily intraperitoneally with 3-NPA (10 mg/kg), (n = 5), or vehicle (n = 5). g (%) human CD45+ cells in the peripheral blood (PB) and (h) in the spleen and (i) spleen weights at the time of sacrifice after 28 days of treatment (mean + /− SEM) (n = 5). j FLT3-ITD+ MV411-transplanted mice were treated intraperitoneally daily (for a period up to 28 days when the last mouse succumbed to leukemia) with vehicle, 3-NPA (40 mg/kg), CHC (100 mg/kg) or with both (n = 5 in each group). k (%) human CD45+ cells in the peripheral blood (PB) on day 14th of the treatment and (l) in spleen and (m) spleen weights at the time of sacrifice (mean + /− SEM) (n = 5). n Kaplan–Meier curve showing survival of mice transplanted with MV411 cells treated with the indicated therapy. The average survival in days are indicated at the bottom of the graph. P value (by log-rank test) indicate statistical significance for comparisons of 3-NPA + CHC treated mice with control mice (vehicle). ae, gi Student’s t test (two-sided) or (km) one-way ANOVA Kruskal–Wallis test for multiple comparisons.

We evaluated human mtDNA (mitochondrial DNA) content from the peripheral blood after confirmation of the engraftment by luciferase intensity measurement in both HL60 and MV411-transplanted mice. Indeed, peripheral blood human mtDNA levels were shown to significantly positively correlate with luciferase intensity in mice (Supplementary Fig. 5c). At the time of comparable engraftment, FLT3-ITD+ MV411-transplanted mice showed significantly higher mtDNA levels in peripheral blood compared to HL60-transplanted mice (Supplementary Fig. 5d). Co-treatment of 3-NPA and CHC significantly reduced bone marrow human mtDNA content at time of sacrifice validating reduced tumor burden in MV411-transplanted mice (Supplementary Fig. 5e). Mean luciferase intensity was also less pronounced at day 10 after combination treatment compared to the vehicle group (Supplementary Fig. 5f). The combination of 3-NPA and CHC treatment significantly reduced AML chimerism in PB (Fig. 5k), in the spleen (Fig. 5l) and reduced spleen weight (Fig. 5m), as well as blood leukocyte counts (Supplementary Fig. 5g) and leukemic cell counts per femur (Supplementary Fig. 5h). While monotherapies already slightly delayed leukemia development in mice transplanted with FLT3-ITD+ MV4–11, co-inhibition of lactate import and ETC complex II significantly further delayed leukemia onset compared to the vehicle group (Fig. 5n).


Succinyl-CoA ligases (SUCLGs) and succinate dehydrogenases (SDH, electron transport chain (ETC) complex II) are functionally connected proteins involved in cellular mitochondrial energy production. SUCLGs regulate the generation of succinate as part of the TCA cycle and SDH oxidizes succinate into fumarate to accomplish electron transfer to the terminal acceptor ubiquinone in the ETC31,32. Mutations in SDH have been identified in various solid cancer types including familial paraganglioma/pheochromocytoma where inactivation of SDH results in epigenetic and metabolic alterations resulting in a build-up of TCA-cycle intermediates33. In AML, mutations in SUCLG or SDH are uncommon, although SDHB mutations were described in a case of chronic lymphocytic leukemia34. In our current work we identify that succinyl ligases are upregulated and that the ETC complex II is highly active in OXPHOS-driven AMLs that are frequently FLT3-ITD+, which can be therapeutically exploited.

Besides an upregulation of Succinyl Ligases, we find other factors that might further support the OXPHOS state of FLT3-ITD+ AMLs. This includes CD36 (Supplementary Fig. 1a), which has been reported to be upregulated in chemotherapy-resistant cells, thereby driving fatty-acid oxidation35,36. By performing proteome studies in functionally defined leukemia-initiating cell populations it was shown that FLT3-ITD+ LSCs are enriched for OXPHOS signatures37 and the authors also identified both SUCLG1 and SUCLG2 to be higher expressed in the LSC sub-compartment compared to blast population and also compared to the healthy HSPC population. Baccelli et al. have shown that the metabolic landscape of AML is heterogeneous16, in line with our findings (current work and22), whereby FLT3-ITD+ AMLs are classified as OXPHOShigh. Sriskanthadevan et al. showed that in particular ETC complex II activity was enhanced in a panel of 11 primary AML samples compared to healthy counterparts38. It was also shown that treatment with venetoclax together with azacitidine disrupts ETC complex II-driven energy metabolism and eradicates persisting ROSLow LSCs39. The mitochondrial peptidase neurolysin, NLN, has been shown to be an essential protein to control oxidative phosphorylation in a subset of AML patients and was upregulated in FLT3-ITD+ AMLs40. Also, in our proteome dataset, we find NLN to be higher expressed in FLT3-ITD+ AMLs (Supplementary Data 2). It was proposed that NLN is associated with an increased formation of respiratory chain supercomplexes40, and although this requires further studies it is interesting to hypothesize that FLT3-ITD might impact on mitochondrial activity via similar mechanisms as well. Inhibition of the mitochondrial protease CLPP selectively kills AML cells by inhibition of oxidative phosphorylation and mitochondrial metabolism41 and we find CLPP to be higher expressed in FLT3-ITD+ AMLs (Supplementary Data 2). Together, these data indicate that the signaling networks downstream of FLT3-ITDs rewire metabolism at multiple levels that all facilitate a high OXPHOS state. We propose that factors such as STAT5 might play an important role, but which transcription factors downstream of FLT3-ITDs participate in these signaling networks is under current investigation, also in a larger cohort of AML patients at the subclonal level by performing transcriptome and chromatin accessibility studies, in line with our previous work7. However, we do not propose that an OXPHOS-driven metabolism is exclusive for FLT3-ITD+ AMLs. For instance, MLL-AF9+ THP1 AML cells demonstrated OXPHOS activities as high as FLT3-ITD+ cells22, and also within the cohort of FLT3-wt primary AML patients there were a few that displayed high OXPHOS. These data again highlight the heterogeneous landscape of AML and challenges ahead lie in the detailed characterization of how specific combinations of (epi)genetic alterations impact on the metabolome.

A potential pitfall that is frequently encountered with single-agent treatment strategies is drug resistance. By inhibiting specific pathways, tumor cells can swiftly rewire signaling, and also metabolic dependencies can easily be rerouted. Here, we uncover that AML cells can change to lactate-fueled respiration, most notably when ETC complex II activity is inhibited. This provides an attractive possibility for combination treatment strategies, and we indeed show that inhibition of the lactate importer MCT1 greatly enhances sensitivity to ETC complex II inhibition, both in vitro as well as in vivo in xenograft mouse models. These observations are somewhat similar to what was shown in pheochromocytoma and paraganglioma tumors where SDH loss induced a vulnerability to LDH and pyruvate carboxylase inhibition42. In AML, it was also suggested that inhibition of lactate metabolism might be an attractive strategy43. Also in line, Cardaci et al. showed that pyruvate carboxylation is an adaptive mechanism for the survival of SDH-deficient kidney cells and their tracing studies showed that pyruvate consumption is higher in SDH-deficient cells28. Similar to our observation, succinate was built up in the absence of SDH and the authors show that the increased consumption of extracellular pyruvate diverts glucose-derived carbons into aspartate biosynthesis, thereby sustaining cell growth28. In a parallel study, Lussey-Lepoutre et al. also show that loss of SDH activity increases the dependency of cells on pyruvate carboxylation to generate aspartate and drive cellular anabolism29.

Overall, our results are in line with these previous observations in different cell types, whereby pyruvate and lactate seem to be able to play an important role to support the TCA cycle when ETC complex II activity is deficient or low. While we have not extensively studied whether extracellular pyruvate can also rescue the ETC complex II inhibitor-induced phenotypes as we have seen upon lactate supplementation, our tracing studies clearly indicate that lactate is converted into pyruvate and consecutive further downstream TCA intermediates. Similar to functional observations above, elegant computer-based simulations from Zielinski et al. also predicted that ETC complex II deficient cells required maximal NADH and increased lactate oxidation to cross the mitochondrial inner membrane to support TCA-cycle activity and pyruvate was increasingly produced from lactate30. The mechanism behind this was indeed oxaloacetate production from increased pyruvate carboxylation and from malate generated from aspartate. An additional mechanism that was proposed in the same study was increased reductive carboxylation adaptive metabolism in SDH deficiency30. Our results from tracing studies suggest that lactate oxidation drives active cytosolic reductive carboxylation activity in malate-aspartate-fumarate and oxaloacetate upon ETC complex II inhibition. How these lactate-driven intermittent pathways are beneficial apart from fueling mitochondrial respiration in AML cells remains unknown and requires further studies. In conclusion, we identify how cells reroute their metabolism in response to ETC complex II inhibition and identify lactate-fueled respiration as a targetable vulnerability in AML.

Article classification: Biological abstract
Share to:
Add:Room A11-329, 1st Floor, No.1, SBI Venture Street, Optics Valley, East Lake
New Technology Development Zone, Wuhan, China.
Certificate NO.:U18Q28010569R0S