Main

The mitochondrial electron transport chain (ETC) transfers electrons from reducing equivalents (NADH and FADH₂) towards complexes I/II, or from peripheral metabolic enzymes that reduce coenzyme Q (ubiquinone, Q) to ubiquinol (QH₂) ¹. The electron transfer flavoproteins (ETFA and ETFB) accept electrons generated by the catabolism of branched-chain amino acids and fatty acids ^2–4. ETFs interact with ETF dehydrogenase (ETFDH or ETF-ubiquinone oxidoreductase) ⁵, which controls electron flow towards complex III (Supplementary Fig.1A). Loss-of-function mutations in ETF/ETFDH occur in metabolic diseases known as multiple acyl-CoA dehydrogenase deficiency (MADD) ^6,7. ETFDH is also essential for complex III activity in skeletal muscle ⁸. Surprisingly, we observed that in contrast to muscle, ETFDH is one of the most non-essential metabolic genes in cancer cells. Paradoxically, ETFDH downregulation leads to enhanced mitochondrial biogenesis and increased bioenergetic capacity of cancer cells. ETF insufficiency resulting from reduction in ETFDH levels triggers a tradeoff favored by cancer cells, whereby reduced flexibility in fuel utilization for oxidative phosphorylation is offset by retrograde signaling that remodels cancer cell bioenergetics to augment neoplastic growth.

ETFDH is dispensable in cancer cells and its abrogation accelerates tumor growth

ETFDH is essential in skeletal muscle cells ⁸. Intriguingly, a CRISPR-Cas9 knock out (KO) screen performed in a human acute lymphoblastic leukemia cell line (NALM6) ⁹ ranked ETFDH as a top non-essential metabolic gene that clustered with well-established tumor suppressors TP53 and RB1 (Supplementary Fig.1B). Mining DepMap also revealed that ETFDH is essential in only 1 out of 1150 cancer cell lines ¹⁰. This motivated us to examine the role of ETFDH in cancer. ETFDH mRNA levels are reduced in patient samples across different cancer types, including colon (Supplementary Fig.1C) and breast cancer (Supplementary Fig.1D). Immunohistochemistry (IHC) staining confirmed that ETFDH protein levels are significantly reduced in colorectal cancer vs. non-adjacent patient-derived tissues (NAT) (Fig.1A).

Figure 1:

We next abrogated the ETFDH gene using a CRISPR-Cas9 approach in a variety of cancer cell lines including HCT-116 (human colorectal cancer), NT2197 (murine HER2+ breast cancer), 4T1 (murine triple negative breast cancer cells), and NALM6 cells (Supplementary Fig.1E-H). We also investigated the effects of ETFDH loss in non-transformed normal murine mammary gland (NMuMG) cells (Supplementary Fig.1I), a parental immortalized mammary epithelial cell line that was transformed with an oncogenic ErbB2 variant to generate NT2197 cells ¹¹. Notwithstanding their high baseline proliferation rates, ETFDH ablation significantly increased proliferation across all tested cancer cell lines (Supplementary Fig.1E-H). This was not caused by the inadvertent effects of the CRISPR-Cas9, as rescuing ETFDH expression in HCT-116 and NT2197 cells reverted proliferation rates to levels comparable to ETFDH-proficient controls (Supplementary Fig.1E-F). In contrast, ETFDH abrogation did not affect proliferation of non-transformed NMuMG cells (Supplementary Fig.1I). ETFDH KO HCT-116 cells also formed more colonies in soft agar (surrogate measurement of cellular neoplastic potential ¹²) as compared to empty vector infected controls (WT EV), and ETFDH KO cells in which ETFDH was re-expressed (ETFDH Rescue) (Supplementary Fig.1J).

Based on these findings, we tested the impact of ETFDH loss on neoplastic growth in vivo. WT EV, ETFDH KO, or ETFDH Rescue HCT-116 cells were injected orthotopically in the caeca of SCID-BEIGE mice. Loss of ETFDH significantly increased the tumor growth rate and volume at endpoint relative to control and ETFDH Rescue cells (Fig.1B-C). ETFDH loss also increased NT2197 tumor growth and the final tumor volume as compared to ETFDH-proficient tumors following orthotopic mammary fat pad injections in SCID-BEIGE mice (Fig.1D-E). Altogether, these results show that ETFDH is an non-essential gene in cancer that is under-expressed across different cancer types, and that its disruption enhances neoplastic growth, while not affecting proliferation of non-transformed cells.

ETFDH abrogation results in ETF insufficiency that paradoxically increases bioenergetic capacity of cancer cells

In non-cancerous cells, ETFDH transfers electrons from catabolism of fatty acids and amino acids to the ETC (Supplementary Fig.2A). Palmitate oxidation (monitored by oxygen consumption) and ¹³C leucine labelling into citrate (citrate m+2) were strongly attenuated by ETFDH abrogation in HCT-116 cells (Supplementary Fig.2B-C), which confirmed that ETF plays a role in fatty and amino acid catabolism in malignant cells. Thus, ETFDH loss induces ETF insufficiency in cancer cells that mirrors defects observed in non-malignant tissues. Surprisingly, ETFDH loss in HCT- 116 cells increased oxygen consumption and extracellular acidification, relative to control WT EV and ETFDH Rescues (Fig.2A-B). ETFDH KO HCT-116 cells produced higher levels of ATP from oxidative phosphorylation (J ATP ox) as compared to ETFDH-proficient cells (Fig.2C-D). Consequently, ETFDH loss increased the bioenergetic capacity of HCT-116 cells (Fig.2E). Comparable results were observed in NT2197 cells (Supplementary Fig.2D-H). Collectively, these findings suggest that although ETF insufficiency limits the ability of cancer cells to utilize lipids and amino acids, it paradoxically upregulates their mitochondrial metabolism and bioenergetic capacity.

Figure 2:

Glutamine plays an important role in fueling mitochondrial metabolism in cancer cells ¹³. ETFDH abrogation increased glutamine uptake and glutamate production in both HCT-116 (Fig.2F) and NT2197 (Supplementary Fig.2I) cells. ETFDH loss increased ¹³C-glutamine incorporation in the forward direction (green) and the reverse direction (purple) of the citric acid cycle (CAC; Fig.2G-H, Supplementary Fig.2J), suggesting higher glutamine utilization in the CAC, as well as increased reductive glutamine metabolism. Accordingly, ETFDH KO HCT-116 cells exhibited greater sensitivity to glutamine deprivation than control WT EV cells (Fig.2I). These findings show that ETFDH loss elevates glutamine utilization in the CAC to support mitochondrial metabolism. Collectively, these data demonstrate that ETF insufficiency in cancer cells remodels mitochondrial metabolism while increasing the bioenergetic capacity.

ETFDH loss induces intracellular accumulation of amino acids, mTOR signaling and protein synthesis

Consistent with the established role of ETFDH in amino acid catabolism ², and reduction of leucine consumption upon ETFDH abrogation (Supplementary Fig.2C), steady state levels of most intracellular amino acids were increased in ETFDH KO vs. WT HCT-116 and NT2197 cells (Fig.3A). Moreover, ETFDH-deficient HCT-116 cells exhibited higher rates of protein synthesis relative to ETFDH-proficient cells (Fig.3B). Amino acids activate the mechanistic target of rapamycin (mTOR) which stimulates protein synthesis ¹⁴. Indeed, mTORC1 signaling was higher in ETFDH-deficient than ETFDH-proficient HCT-116 and NT2197 cells as illustrated by increased phosphorylation of ribosomal protein S6 kinases (S6K1/2; Thr389) and their substrate ribosomal protein S6 (rpS6; Ser240/244) (Fig.3C). Re-expression of ETFDH in ETFDH KO cells resulted in normalization of mTORC1 signaling to the levels observed in WT EV cells (Fig 3D). mTORC2 signaling was also augmented upon ETFDH abrogation in HCT-116 and NALM6 cells as evidenced by increased AKT phosphorylation (Ser473) (Supplementary Fig.3A-B). Despite bioenergetic rewiring, ETFDH loss did not exert a major effect on the activity of adenosine monophosphate-activated protein kinase (AMPK) as monitored by its phosphorylation (Thr172) or phosphorylation of its substrate acetyl-CoA carboxylase (ACC at S79) (Supplementary Fig.3A-B).

Figure 3:

Consistent with mTOR hyperactivation, ETFDH KO HCT-116 cells were more sensitive to the active site mTOR inhibitor Torin1 ¹⁵ as compared to their ETFDH-proficient counterparts (Supplementary Fig.3C). To further determine which mTOR complex is responsible for mediating the hyper-proliferative phenotype triggered by ETFDH loss, we selectively inactivated mTORC1 or mTORC2 by depleting RAPTOR or RICTOR, respectively using small hairpin RNA (shRNA) in WT and ETFDH KO HCT-116 cells (Supplementary Fig.3D). Depletion of RAPTOR, but not RICTOR strongly attenuated the proliferation of ETFDH KO cells (Supplementary Fig.3E-F). Moreover, the anti-proliferative effects of bisteric mTORC1-specific inhibitor BiS-35x ¹⁶ were stronger in ETFDH KO vs. ETFDH Rescue HCT-116 cells (Supplementary Fig.3G). Altogether, these data suggest that mTORC1, but not mTORC2 acts as a major mediator of hyper-proliferative phenotype triggered by ETFDH loss in cancer cells. Accordingly, the effects of amino acid deprivation on phospho-rpS6 (Ser240/244), and phospho-4E-BP1 (Ser65) were attenuated in ETFDH KO vs. WT control cells, which is consistent with the intracellular accumulation of amino acids upon ETFDH disruption (Fig.3E-G). In turn, serum stimulation (Stim) resulted in comparable induction in phosphorylation of rpS6 (Ser240/244) and 4E-BP1 (Ser65) in ETFDH-deficient and proficient cells (Fig.3E). Collectively, these findings suggest that ETFDH loss induces mTORC1 signaling at least in part by increasing intracellular amino acid levels.

Amino acid deprivation leads to general control nonderepressible 2 (GCN2)-dependent phosphorylation of the alpha subunit of eIF2 and subsequent induction of activating transcription factor 4 (ATF4) synthesis during the integrated stress response (ISR) ¹⁷. Consistent with an increase in intracellular amino acid, phosho-eIF2⍺ (S51) levels appeared to be modestly reduced in ETFDH-deficient vs. proficient HCT-116 cells under both serum starvation and amino acid depletion (Supplementary Fig.3H). However, the ETFDH status in the cells did not affect ATF4 levels at baseline or upon amino acid depletion (Supplementary Fig.3H). Altogether, these data demonstrate that ETFDH loss induces mTORC1 signaling and protein synthesis, which appears to be at least in part driven by intracellular accumulation of amino acids.

ETFDH loss diminishes 4E-BP1, but not 4E-BP2 protein levels

mTORC1 positively controls cell proliferation through the eukaryotic translation initiation factor 4E-binding proteins (4E-BPs; 4E-BP1-3 in mammals) ¹⁸. Of note, 4E-BP3 is expressed in a tissue-restricted manner, and does not appear to play a major role in regulating proliferation ¹⁸. mTORC1 phosphorylates 4E-BPs to promote eukaryotic translation initiation factor 4F (eIF4F) complex assembly by dissociating 4E-BPs from eIF4E, thereby stimulating cap-dependent translation initiation ¹⁹. Consistent with the increased mTORC1 activity, ETFDH-deficient cells exhibited increased phosphorylation of 4E-BP1 (Ser65), which was rescued upon re-expression of ETFDH in HCT-116 and NT2197 cells (Fig.4A). Unexpectedly, we observed that ETFDH loss coincided with a dramatic reduction in total 4E-BP1 protein levels in HCT-116, NT2197 (Fig.4A), NALM6 (Supplementary Fig.4A), and 4T1 cells (Supplementary Fig.4B). We also treated WT and ETFDH KO HCT-116 cell lysates with λ-phosphatase to exclude potential confounding effects of different 4E-BP1 phosphorylation states, which confirmed that disruption of ETFDH is paralleled by reduction in 4E-BP1 levels (Fig.4B). Notably, loss of ETFDH did not exert a major effect on 4E-BP2 protein abundance in HCT-116 (Fig.4A), NT2197 (Fig.4A), or 4T1 cells (Supplementary Fig.4B). In addition, 4E-BP1, but not 4E-BP2 protein levels were reduced in ETFDH-deficient HCT-116 as compared to WT tumors (Fig.4C-E). Strikingly, loss of ETFDH in non-transformed NMuMG cells did not exert a major effect on mTORC1 signaling (Supplementary Data Fig.4C) and failed to alter 4E-BP1 protein levels (Supplementary Fig.4C). Collectively, these findings indicate that ETF insufficiency is paralleled by a selective decrease in 4E-BP1 protein levels in cancer, but not non-transformed cells.

Figure 4:

Decrease in 4E-BP1 protein levels underpins the bioenergetic rewiring induced by ETF insufficiency in cancer cells

Considering that in mammals 4E-BPs play a major role in mediating the effects of mTORC1 on proliferation, neoplastic growth, mitochondrial functions and protein synthesis ^18,20,21, we next investigated whether the observed decrease in 4E-BP1 levels contributes to increased proliferation and tumor growth caused by ETFDH abrogation. To this end, we overexpressed 4E-BP1 protein in ETFDH-deficient NT2197 cells by infecting them with a retrovirus wherein EIF4EBP1 expression was driven by an exogenous promoter ¹⁸. Importantly, 4E-BP1 overexpression in ETFDH KO NT2197 cells decreased eIF4F complex assembly, as evidenced by m⁷-GTP pulldown (Fig.4F). 4E-BP1 overexpression also reduced oxygen consumption in ETFDH KO NT2197 cells, relative to empty vector infected cells (Fig.4G). Moreover, restoring 4E-BP1 protein levels decreased cell proliferation (Supplementary Fig.4D) and anchorage-independent growth (Supplementary Fig.4E) of ETFDH KO HCT-116 cells. Lastly, 4E-BP1 overexpression decreased growth and the final volume of tumors formed by NT2197 ETFDH KO cells in SCID-BEIGE mice (Fig.4H-I). Collectively, these data indicate that decreased 4E-BP1 protein levels mediate the pro-neoplastic effects of ETF insufficiency.

The mTORC1/4E-BP axis stimulates mitochondrial biogenesis and metabolism by altering the translation of nuclear-encoded mitochondrial mRNAs including mitochondrial transcription factor A (TFAM) ^21,22. ETFDH loss increased mitochondrial DNA levels (Fig.4J) and mass (Supplementary Fig.4F) which was paralleled by TFAM induction (Fig.4K). Collectively, these data show that the loss of 4E-BP1 plays a prominent role in mediating the effects of ETF insufficiency on bioenergetic reprogramming of cancer cells.

ETFDH loss decreases EIF4EBP1 transcription via BCL-6

We next set out to determine the level(s) at which 4E-BP1 protein levels are downregulated in ETF insufficient cancer cells. ETFDH loss did not have a major effect on 4E-BP1 protein synthesis or turnover (Supplementary Fig.4G-H), but it decreased the EIF4EBP1 mRNA levels in ETFDH-deficient vs. ETFDH-proficient HCT-116 (Fig.4L) and NT2197 cells (Fig.4M). EIF4EBP1 and ETFDH mRNA levels were also positively correlated in tumors isolated from colorectal adenocarcinoma patients (Supplementary Fig.4I). In turn, consistent with no apparent differences in 4E-BP2 protein levels, the levels of EIF4EBP2 mRNA were unaffected by the ETFDH status in both cell lines (Fig.4L-M). Notably, observed decrease in EIF4EBP1 mRNA levels was not caused by the effects of ETFDH loss on mRNA stability (Supplementary Fig.4J). Based on these data, we concluded that ETFDH abrogation is paralleled by suppression of EIF4EBP1 transcription and sought to identify transcription factors that may mediate the effects of ETFDH abrogation on suppression of EIF4EBP1 expression. ATF4, Snail and Slug have been shown to stimulate transcription of EIF4EBP1, but not EIF4EBP2 ^23,24. However, ATF4, Snail and Slug proteins remain unchanged upon ETFDH disruption (Supplementary Fig.3H, Supplementary Fig.5A). We also re-analyzed publicly available RNA-seq datasets, which revealed that EIF4EBP1, but not EIF4EBP2 expression was increased following STAT1 abrogation in MT4788 breast cancer cells ²⁵. Indeed, we observed that MT4788 STAT1 KO cells have increased 4E-BP1 protein levels relative to parental cells (Supplementary Fig.5B). Accordingly, ETFDH KO cells displayed increased STAT1 and downregulated 4E-BP1 protein levels (Supplementary Fig.5C). To determine the impact of STAT1 on EIF4EBP1 or EIF4EBP2 promoter occupancy, we performed chromatin immunoprecipitation qPCR (ChIP-qPCR) assay in control (WT EV), ETFDH KO, or ETFDH rescue HCT-116 cells (Supplementary Fig.5D-E). Binding of STAT1 was however comparable to IgG across all tested cell lines thereby suggesting that is unlikely that STAT1 directly suppresses EIF4EBP1 transcription in the context of ETFDH loss (Supplementary Fig.5D-E). Furthermore, reactive oxygen species (ROS) can also influence redox-responsive transcription ²⁶. Although loss of ETFDH induced ROS levels (Supplementary Fig.5F), 4E-BP1 protein abundance was not altered following treatment with the antioxidant N-acetyl-cysteine (NAC) (Supplementary Data Fig.5G).

Figure 5:

We next investigated the potential role of B-cell lymphoma 6 (BCL-6) ^27–29, which was previously suggested to suppress EIF4EBP1 transcription ³⁰. Notably, ETFDH abrogation in HCT- 116 and NT2197 cells resulted in a marked increase in BCL-6 protein levels, compared to ETFDH-proficient control cells (Fig.5A). These changes in BCL-6 protein levels were mTOR-sensitive (Fig.5B) and not accompanied by the alterations in BCL6 mRNA abundance (Fig.5C), which is consistent with recently reported mTOR-dependent translational activation of BCL-6 ^31–33. Indeed, BCL6 mRNA translation was increased in HCT-116 ETFDH KO relative to WT cells (Fig.5D). Taken together, these data point out that ETFDH disruption induces BCL-6 protein synthesis in an mTOR-dependent manner. ETF insufficiency increased BCL-6 binding to the EIF4EBP1, but not the EIF4EBP2 promoter (Fig.5E-F). BCL-6 silencing in NT2197 and 4T1 ETFDH KO cells increased 4E-BP1 protein levels (Fig.5G-H). Depletion of BCL-6 also attenuated proliferation of 4T1 ETFDH KO cells (Fig.5I). Collectively, these findings show that BCL-6 mediates the effects of ETF insufficiency on reduction in EIF4EBP1 expression and increased proliferation of cancer cells.

ETFDH acts as a haploinsufficient tumor suppressor

Data available in cBioPortal from 2683 samples across cancer types ³⁴ indicate that missense or truncating mutations in ETFDH are rare (0.3%) (Fig.6A). This suggests that ETFDH is not lost, but rather downregulated in human cancers. We observed that DNA methylation inhibitors, 5-azacytidine (5-aza) and 5-aza-2′-deoxycytidine (5-aza-2-deoxyC) increased ETFDH protein levels in HCT-116 cells (Fig.6B), thus suggesting that one of the potential mechanisms of downregulated ETFDH expression in neoplasia may stem from increased DNA methylation. Based on these observations, we developed a model that more closely reflects ETFDH downregulation, but not a complete loss, in human tumors using a method based on targeted poly(A) track insertion within endogenous genes ³⁵. The length of the poly(A) track directly correlates with the reduction in amount of protein produced, allowing for the investigation of gene-dosage effects on ensuing phenotypes ^35,36. Accordingly, mutants were designed by inserting poly(A) tracks of 12 and 18 adenosines [12A, equivalent to 4 consecutive AAA (Lys) codons (AAA)₄; 18A, equivalent to 6 consecutive AAA (Lys) codons (AAA)₆], as well as a control track (CTRL) [6 consecutive lysine AAG codons (AAG)₆] (Supplementary Fig.6A, see methods). Based on an Alphafold rendered structure (version 2) ^37,38, we chose residue 38 in an unstructured loop as an insertion site (Supplementary Fig.6B). Of note, (AAA)₆ and (AAG)₆ stretches are translated in the equal number of Lys residues, whereby (AAA), but not (AAG), stretches decrease ETFDH protein levels (Fig.6C). (AAG)₆ construct was thus used as a control to mitigate potential inadvertent effects of the insertion of poly-Lys on the ETFDH function. (AAA)₄ and (AAA)₆ stretches progressively reduced ETFDH protein abundance as compared to constructs harboring the (AAG)₆ track (Fig.6C). The reduction in ETFDH protein levels was mirrored by a gene-dose dependent increase in mTORC1 signaling, BCL-6 expression and reduction in 4E-BP1, but not 4E-BP2 protein levels (Fig.6C). Accordingly, Eif4ebp1 mRNA levels were reduced in ETFDH dose-dependent manner, while Eif4ebp2 mRNA levels remained unaffected (Fig.6D). Moreover, reduction in ETFDH protein levels correlated with increased mtDNA content (Fig.6E), oxygen consumption (Fig.6F), ATP production from OXPHOS (JATPox) (Fig.6G), bioenergetic capacity (Fig.6H), and cell proliferation (Fig.6I). Finally, ETFDH KO NT2197 tumors carrying (AAA)₆ insertions exhibit increased tumor growth and larger final tumor volumes relative to control tumors expressing (AAG)₆ insertions in SCID-BEIGE mice (Fig.6J-K). Altogether, these findings show that ETFDH gene-dosage influences neoplastic growth, thus suggesting that ETFDH may act as a haploinsufficient tumor suppressor.

Figure 6:

Discussion

Notwithstanding that ETFDH mutations play a detrimental role in MADD metabolic disorders, we show that ETF insufficiency caused by reduced ETFDH expression plays a major role in rewiring cancer metabolism and signaling to stimulate neoplastic growth. Reduction of ETFDH levels appears to be a near-universal feature of cancer cells and is likely a consequence of DNA hypermethylation (Fig.6B), an event that is frequently observed among tumor suppressor genes ³⁹. Indeed, ETFDH was found among the most under-expressed metabolic enzymes in a study that compiled cancer microarray datasets ⁴⁰. Moreover, recent studies highlighted crosstalk between aberrant mTOR signaling and epigenetic perturbations ^41–44, which further reinforces the plausibility that epigenetic reprograming in cancer cells may underpin the pro-neoplastic role of ETFDH downregulation and/or explain resulting phenotypes.

Intriguingly, although ETF insufficiency limits the flexibility of cancer cells in using fuels (e.g., lipids and amino acids) for OXPHOS, it triggers retrograde signalling that established feed-forward loop via mTORC1 which leads to increased BCL-6 levels and reduced EIF4EBP1 transcription (Fig.6L). ETF insufficiency thus orchestrates enhanced mTORC1 signaling with downregulation of 4E-BP1 protein levels to reprogram protein synthesis and ultimately increase mitochondrial biogenesis and functions thereby providing selective proliferative advantage to cancer cells (Fig.6L). To this end, ETF insufficiency allows cancer cells to trade of their metabolic flexibility for increased bioenergetic capacity and a re-wired signaling state that favors neoplastic growth. Importantly, while ETFDH plays an essential role in muscle, we did not observe a major effect of ETFDH disruption on proliferation, mTOR signaling or 4E-BP1 levels in non-transformed murine breast epithelial NMuMG cells. These observations suggest that ETF insufficiency causes remodeling of signaling and metabolic networks that are favourable to cancer cells.

Our findings are consistent with previously attributed tumor-suppressive effects of 4E-BP1 ^45–50. More recently, it has been demonstrated that under certain contexts 4E-BPs may play a beneficial role for tumor survival. For instance, 4E-BPs may facilitate tumor growth by regulating global and selective mRNA translation under periods of cellular stress ^51,52, including suppression of fatty acid biosynthesis under glucose deprivation ^53,54. Collectively, these findings demonstrate that 4E-BPs are likely to play more complex, context dependent roles during cancer progression than previously anticipated. Moreover, the role of different 4E-BP proteoforms in cancer was in general thought to be redundant. Our findings however demonstrate that in the context of ETF insufficiency selective decrease in 4E-BP1, but not 4E-BP2 levels plays a role in driving neoplastic growth thereby suggesting that 4E-BP1 and 4E-BP2 may play hitherto unappreciated non-overlapping roles in neoplasia.

In conclusion, we show that ETF insufficiency caused by reduced ETFDH expression causes retrograde signaling that remodels mitochondrial metabolism in a manner that drives neoplastic growth (Fig.6L), while either being neutral (NMuMG) or deleterious (muscle cells) in non-malignant cells. These findings explain how disruption of mitochondrial enzyme, rather than reducing cellular fitness, is common in cancer because it increases mitochondrial energy flux to drive neoplastic growth.

Methods

Cell lines and cell culture

HCT-116, NALM6, and HEK293T cells were obtained from American Type Culture Collection (ATCC) and verified using ATCC Cell line authentication service. NMuMG, NT2197, and MT4788 Parental and STAT1 KO cells were obtained by Dr. J. Ursini-Siegel’s lab, while 4T1 (4T1-526) cells were obtained from Dr. P.M. Siegel’s group. HEK293T and HCT-116 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin and supplemented with 2mM L-Glutamine to obtain a final concentration of 6mM L-Glutamine. NALM6 and 4T1 cells were cultured in RPMI-1640 supplemented with 10% FBS, 1% penicillin/streptomycin and 2mM L-Glutamine. NMuMG and NT2197 cells were cultured in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin, 2mM L-Glutamine, 10 mg/mL insulin, 20 mM HEPES, pH 7.5. NT2197 culture media was supplemented with puromycin (2 μg/mL) as previously described ¹¹. MT4788 were grown in DMEM, 2.5% FBS, mammary epithelial growth supplement (MEGS), 1% penicillin/streptomycin and gentamycin as previously described ²⁵. All cell lines were maintained at 37°C and 5% CO2 in a humidified environment and tested periodically for mycoplasma according to the manufacturer’s instructions.

Animals

6–8-week-old SCID-BEIGE mice were purchased from Charles River. Male SCID-BEIGE mice were used for intracaecal experiments, while female mice were used for mammary fat pad injections. All animal experiments were performed in accordance with the guidelines of the McGill University Animal Ethics Committee and the Canadian Council on Animal Care as approved by the facility animal care committee.

Xenograft Experiments

Luciferase-expressing HCT-116 cells (250,000 cells) were injected intracaecally in male (8-10 weeks old) SCID-BEIGE mice as previously described ⁵⁵. Tumor growth was monitored and assessed by luminescence (IVIS spectrum, PerkinElmer). Animals were sacrificed after 50 days. Quantification of signal intensity was performed with Living Image software (PerkinElmer). NT2197 mammary tumor cells (50,000 cells) were injected in mammary fat pads of 6-8-week female SCID-BEIGE mice. Tumor growth was monitored using calipers. Measurements were shown as once tumor volume reached 100∼200 mm³ and animals were sacrificed before tumors reached 1cm³. Tumors were excised, placed in 10% formalin, and embedded for immunohistochemistry.

Cell Proliferation Assay

Cells were counted using a Countess automated cell counter (Invitrogen) and trypan blue exclusion was used for determining cell viability. Cells (HCT-116, NT2197, NMuMG, 4T1, NALM-6, and MT4788) were seeded in 6-well dishes and counted to 3-5 days. For glutamine deprivation experiments, HCT-116 cells were seeded in full culture for 24 hours. The next day, cells were grown in media in the presence or absence of glutamine for 48 hours and then counted. For Torin1/BiS-35x treatment experiments, HCT-116 cells were seeded overnight. The next day, media was replaced with treatment media containing Torin1 (50nM, 100nM, 250nM, 500nM) or BiS-35X (1nM, 10nM, 100nM, 1000nM) or DMSO for 72 hours and then counted.

Immunohistochemistry

Immunohistochemical staining of colon human parafilm-embedded tissue array (TissueArray.com, CO992b) was performed as previously described ⁵⁶. Briefly, the tissue slide was baked for 55°C for 30 min followed by incubation in xylene (2× 5 min), 100% ethanol (2× 5 min), 95% ethanol (2× 5 min), 75% ethanol (1× 5min), and ddH₂O (2x 5min). The slide was submerged in 10mM sodium citrate buffer (pH 6.0), 0.05% Tween-20 and heat-induced epitope retrieval was carried out in a pressure cooker for 30 min. The slide was cooled for 20 min at room temperature, rinsed in ddH₂O for 5 min, and washed with 1xTBS + 0.3% Triton X-100 for 10 min, then 1xTBS for 5 min at room temperature. The slide was incubated in 3% H₂O₂ for 10 min at room temperature and washed with TBST (1xTBS + 0.1% Tween-20) for 5 min to inactivate peroxidases. A liquid blocker pen (PAP Pen, DAKO) was used to circle tissues on the slide. The TMA was blocked for 1 hour in room temperature in the presence of DAKO Protein Block-Serum Free (DAKO, Cat#X0909). The slide was incubated overnight at 4°C with the ETFDH antibody (1:500, ProteinTech 11109-1-AP). The next day, the slide was washed in TBST (3x 5min), and secondary incubation was performed for 30 min at room temperature with secondary antibody (Antibody background reducing diluent and HRP rabbit substrate (Cell Signaling, Cat#7074)). The slide was washed in TBST (3x 5min) and DAB (Di-amine-benzidine) peroxidase substrate (Vector, Cat# SK-4100) was applied to the slide for 5 min. Reaction was stopped by washing in ddH₂O for 5 min. The slide was counterstained with filtered hematoxylin for 20 sec and rinsed with ddH₂O for 5 min. The slide was placed in 75% ethanol (1 min), 95% ethanol (1 min), 100% ethanol (1 min) and xylene (2x 5min). The slide was mounted with a cytoseal at 60°C and dried in a fumehood overnight. Quantitation of sections was performed using Imagescope software (Aperio).

Western Blotting

Western blotting was performed as previously described ⁴⁴. Briefly, cells were washed twice with PBS and lysed with RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.5% sodium deoxycholate, 1 mM PMSF, 1 mM DTT, 1% NP40, 0.1% SDS), supplemented with 1X protease inhibitor (Roche) and 1x PhoSTOP (Roche). Lysates were cleared at 4°C (10min, 13,000rpm). Protein concentrations were determined using BCA kit (ThermoFisher). Lysates were mixed with 5x Laemmli buffer (60mM Tris-HCl (pH 6.8), 10% glycerol, 2% SDS, 5% 2-mercaptoethanol, 0.05% bromophenol blue) and boiled (95°C for 5min). Proteins were resolved by SDS-PAGE (Bio-Rad) and transferred via wet transfer apparatus (Bio-Rad) to nitrocellulose membranes (Amersham). Membranes were blocked in 5% skim milk (dissolved in TBST (0.1% Tween 20 in 1× TBS)) and incubated in primary antibodies overnight at 4°C. The next day, membranes were washed in TBST and incubated with HRP-conjugated secondary antibodies for 1 hour. Membranes were washed and applied with ECL (Bio-Rad) for 1min. Exposure was carried out on an Azure c300 (Azure Biosystems). Images were quantified by ImageJ and analysed using GraphPad Prism10. The list of primary antibodies is described in Supplementary Table 1.

For λ-phosphatase experiments, cells were lysed with RIPA and supplemented with 1X protease inhibitor, without 1x PhoSTOP. Protein samples were combined with NEBuffer for protein metallophosphatases (PMP), MnCl₂, and λ-protein phosphatase for 30min at 30°C according to the manufacturer’s instructions (New England Biolabs).

Cap-Binding Pull-Down Assay

Assay was carried out as previously described ¹⁸. Briefly, NT2197 cells were seeded overnight in 150mm plates. The next day, media was changed, and cells were washed with PBS and dissolved in lysis buffer containing 50mM MOPS/KOH (7.4), 100mM NaCl, 50 mM NaF, 2mM EDTA, 2mM EGTA, 1% NP40, 1% sodium deoxycholate, 7 mM β-mercaptoethanol, supplemented with 1X protease inhibitor (Roche) and 1x PhoSTOP (Roche). Lysates were cleared at 4°C (10min, 16,000g). Protein concentrations in samples were elucidated by BCA (ThermoFisher). Samples were equilibrated for 20 min rotating at 4°C with m⁷-GDP-agarose beads (Jena Bioscience), then washed in buffer containing 50mM MOPS/KOH (7.4), 100mM NaCl, 50 mM NaF, 0.5mM EDTA, 0.5 mM EGTA, 7 mM β-mercaptoethanol, 0.5 mM PMSF, 1mM Na₃VO₄ and 0.1mM GTP, by centrifugation (500g for 1min). Bound proteins were eluted from beads via boiling in loading buffer. Western blotting was performed on eluted proteins and input samples to assess for eIF4F complex formation using 4E-BP1, eIF4G1, eIF4E, and β-actin antibodies.

Flow Cytometry

NT2197 cells were seeded in 6-well dishes for 24 hours. The next day, cells were stained with 50nM MitoTracker Deep Red (ThermoFisher) for 15 minutes at 37°C in the dark. Cells were washed, counted, and 100,000 cells were dissolved in 500μl HBSS (Gibco). Samples were analyzed with the LSR Fortessa cytometer (Becton Dickinson, Mountain View, CA). Fluorescence intensity was detected by excitation at 644nm and acquisition on the 665/-A channel for MitoTracker Deep Red. ROS measurements were performed as previously described ⁵⁶. Briefly, cells were incubated with H₂DCFDA (Molecular Probes) for 30 minutes at 37°C. Cells were trypsinized, washed, and resuspended in 500μl HBSS. Samples were sorted on a BD FACS Canto II system. Fluorescence intensities were calculated by FlowJo (Tree Star, Inc.).

Mitochondrial DNA Quantitation

Quantitation of mitochondrial DNA was carried out as previously described ²¹. HCT-116 or NT2197 cells were seeded in 6-well dishes for 24 hours. Genomic and mitochondrial DNA were extracted using PureLink Genomic DNA Mini Kit (ThermoFisher) and quantified by qPCR using SensiFAST™ SYBR® Lo-ROX kit (Bioline).

Polysome-Profiling Assay

Experiments were carried out as previously described ⁵⁷. Briefly, HCT-116 cells were seeded overnight in 150mm plates. The next day, cells were washed in PBS containing 100 μg/ml cycloheximide and lysed in hypotonic buffer (5 mM Tris-HCl [pH 7.5], 2.5 mM MgCl₂, 1.5 mM KCl, 100 μg/ml cycloheximide, 2 mM DTT, 0.5% Triton X-100, and 0.5% sodium deoxycholate). Samples were loaded onto 10%-50% wt/vol sucrose density gradients (20 mM HEPES-KOH pH 7.6, 100 mM KCl, 5 mM MgCl₂) and spun on a SW40Ti rotor (Beckman Coulter) at 4°C (36,000 rpm for 2 hours). Samples were fractionated and recorded at OD 254nm using an ISCO fractionator (Teledyne ISCO). RNA from fractions and input was extracted using TRIzol (ThermoFisher) according to the manufacturer’s instructions. RNA from each fraction and input was isolated using TRIzol (Invitrogen) according to the manufacturer’s instructions. RT-qPCR was performed on fractions and input RNA (refer to section “RNA extraction and RT-qPCR” for more information). Primers are listed in Supplementary Table 2.

Puromycilation Assay

HCT-116 cells were grown in 10cm dishes for 24 hours. The next day, cells were treated with puromycin (10ug/ml) for 20 minutes. Cells treated with DMSO were used as negative controls. Cells were lysed and protein samples were quantified. Western blotting was performed (refer to “Western Blotting” section for more information) and membranes were incubated in anti-puromycin antibody (Millipore, Cat#MABE343). Quantification of bands was performed by ImageJ.

Stable Isotope Tracing

GC/MS quantification and stable isotope tracing analysis (SITA) was performed as previously described ⁵⁸. Briefly, NT2197 cells were seeded in 6-well dishes. Prior to ¹³C-glutamine SITA experiments, culture media was replaced with one whose composition contains 6mM unlabelled glutamine for 2 hours. Afterwards, media was replaced with media containing 6mM labelled ([U-13C])-glutamine media (Cambridge Isotope Laboratories, MA, USA; CLM-1822; L-glutamine ([U-13C5]), 99%) for 5, 15, 30, and 60 minutes. For ¹³C-leucine SITA experiments, culture media was replaced with one containing unlabelled leucine (0.105g/L) for 2 hours, then incubated in media containing labelled (0.105g/L) ([U-13C])-leucine media (Cambridge Isotope Laboratories, MA, USA; CLM-2262-H-0.1; L-leucine ([U-13C6]), 99%) for 24 hours. Both steady state and tracing samples were washed in cold saline solution (9g/L NaCl) and scraped off with 80% methanol. Samples were sonicated at 4°C for 10 minutes (high setting, 30 seconds on/30 seconds off cycles) to rupture cells. Lysates were centrifuged at 4°C (14000g,10 minutes), and supernatants were collected with addition of an internal standard (750ng myristic acid-D27). Samples were dried overnight at 4°C by speed-vac (Labconco). Dried pellets are resuspended in methoxyamine hydrochloride (10mg/ml), sonicated, and cleared for 10 minutes. Samples are heated 70°C for 30 minutes then applied with N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA) at 70°C for 1 hour. Derivatized samples are injected into an Agilent 5975C GC/MS (Agilent Technologies, CA, USA) using methods described previously ⁵⁸. ¹³C-tracer samples were run in parallel with unlabelled samples. MassHunter software (Agilent) was used to carry out mass isotopomer distribution analyses, with metabolites normalized to myristic acid-D27 and cell number. Isotopic distributions were corrected for naturally occurring isotopes using an in-house algorithm ⁵⁹.

RNA Extraction and RT-qPCR

HCT-116 and NT2197 cells were plated in 6-well dishes and extracted using TRIzol (Ambion) or Aurum Total RNA Mini Kit (Bio-Rad) following the manufacturers’ instructions. cDNA was synthesized from purified RNA using SensiFAST cDNA Synthesis Kit (Bioline), as per manufacturer’s protocol. RT-qPCR was performed using SensiFAST SYBR Lo-ROX kit (Bioline). Primer sequences are found in Supplementary Table 2.

Chromatin Immunopreciptation (ChIP)

Experiments were performed as previously described ⁴⁴. Briefly, HCT-116 cells were grown in 150mm dishes and fixed in 4% formaldehyde for 10 min, followed by centrifugation. Pellets were resuspended in ChIP-buffer (50 nM Tris (pH 8), 100 mM NaCl, 5 mM EDTA, 1X PMSF, 2 mM NaF, 0.25% NP-40, 0.25% Triton X-100, 0.25% Sodium Deoxycholate, 0.005% SDS) supplemented with 1x cOmplete protease inhibitor (Roche). Samples are sonicated using Sonic Dismembrator Model 500 (ThermoFisher) at 5 cycles at 20% power, 5 cycles at 25% power, 5 cycles at 30% power; each cycle is 10 sec. Lysates are spun at 4°C (14,000rpm, 10min) and protein concentration was measured from supernatants using the BCA kit (ThermoFisher). Samples were loaded onto Protein G Plus-Agarose Suspension Beads (Millipore) and pre-cleared at 4°C for 3 hours. Input samples were collected, and immunoprecipitation was performed with anti-BCL6 (Cell Signaling, #5650) or anti-IgG (Cell Signaling, #2729) on the remainder of the sample overnight at 4°C. The following day, samples were washed in three wash buffers (20 mM Tris (pH 8), 2 mM EDTA, 0.10% SDS, 1% Triton X-100 with 150/200/599mM for Wash1/2/3 buffers respectively), then a wash with LiCl buffer (10 mM Tris (pH 8), 1 mM EDTA, 0.25 M LiCl, 1% NP-40, 1% Sodium Deoxycholate), and two washes with TE buffer (10 mM Tris (pH 8), 1 mM EDTA). Samples were eluted in elution buffer (0.1 M NaHCO3, 1% SDS), and de-crossliked at 65°C overnight. The following day, proteinase K was applied to samples and heated at 42°C for 1 hour. DNA was purified and collected using DNA collection column (BioBasic). ChIP-qPCR was performed for EIF4EBP1 and EIF4EBP2 using sequences found in Supplementary Table 3.

Respirometry Assay

Oxygen consumption (OCR) and extracellular acidification rate (ECAR) were measured using Seahorse XFe24 and XFe96 analyzers (Agilent) with the Mito Stress Test (Agilent) and Palmitate Oxidation Assay (Agilent), according to the manufacturer’s instructions. Briefly, HCT-116 and NT2197 cells were seeded in Seahorse culture plates (XFe24: 50,000 cells or XFe96: 20,000 cells) and incubated at 37°C. Cells were washed twice and incubated in assay media. For mito stress test experiments, assay media is composed of 10mM glucose, 2mM glutamine, and 1mM sodium pyruvate. Injections include oligomycin (1µM), FCCP (1µM), rotenone/antimycin A (1µM), and monensin (20µM). ATP production from OXPHOS (ATPox) or glycolysis (ATPglyc) were quantified using algorithms presented previously ⁶⁰. For palmitate oxidation assay, cells were seeded in substrate-limited media (DMEM, 1% FBS, 0.5mM glucose, 1mM glutamine, 0.5mM carnitine). The next day, cells are washed with FAO Buffer (includes 0.5mM glucose, 0.5mM carnitine, and 5mM HEPES). Immediately prior to run, palmitate-BSA or BSA control is added to the cell media. Basal respiration was recorded. Values were normalized to cell counts.

Soft Agar/Colony Formation Assay

Experiments were conducted as previously shown ¹². Briefly, 6-well dishes are filled with a bottom layer solution (1:1; 1% noble agar and 2X culture media (2X DMEM, 20% FBS, 2% Pen/Strep/Glutamine)) per well. Dishes are incubated at room temperature for 30 min to allow solution to solidify. 5,000-10,000 HCT-116 and NT2197 cells are mixed 1:1 with 0.6% noble agar and 2X culture media. 1.5ml of this mixture is transferred on top of the solidified agar wells. Dishes are incubated at room temperature to allow the top layer to solidify. Cells are grown at 37°C until colony formation is present. Dishes were stained with nitroblue tetrazolium (Sigma), and colonies were counted.

Protein Stability Assay

HCT-116 cells were seeded in 10cm dishes. The next day, cells were treated with cycloheximide (CHX) (50µg/ml) or DMSO for 2 hours. Furthermore, cells were treated with the proteasomal inhibitor, MG-132 (10µM) or DMSO for 2 hours. Final time point for all conditions was 4 hours. Protein was extracted, quantified, and western blotting was performed for ETFDH, p21, 4E-BP1, or β-actin (refer to “Western Blotting” section for more information).

RNA Stability Assay

Assay was performed as previously described ⁶¹. Briefly, HCT-116 cells were seeded in 6-well dishes. The next day, cells were treated with actinomycinD (ActD) for 2,4,8 and 24 hours. RNA was extracted, cDNA was synthesized, and RT-qPCR was performed (refer to “RNA extraction and RT-qPCR” for more information).

Generation of ETFDH Knockout Cell Lines

To deplete endogenous ETFDH expression in HCT-116 and NALM6 cells, two guides RNA (gRNAs) targeting exon 6 and exon 9 of human ETFDH (against ETFDH sequences AGGTTGGCCGAATGCTAGGATGG and GATGTAGGGATACAAAAGGATGG) (refer to Supplementary Table 4 for gRNA sequences) were designed using CHOPCHOP (https://chopchop.cbu.uib.no/)⁶² and purchased with the appropriate overhangs to be cloned into LentiCRISPRv2(GFP) (Addgene #82416). Cloning was carried out as previously described ^63,64, with the following modifications: BsmBI-v2 (New England Biolabs) was used to digest the lentiCRISPRv2 plasmid; plasmid DNA was recovered using Zymoclean Gel DNA Recovery Kit (Zymo Research), and ligation was performed overnight at 4°C. Transformation of the ligated plasmid was carried out at 42°C in chemically competent Stbl3 cells (Fisher Scientific). Bacteria were plated on Lysogeny Broth (LB)-agar plates containing ampicillin (1 µg/mL) (BioBasic). Extraction of plasmid DNA was conducted using the QIAprep Spin Miniprep Kit (Qiagen) and samples were sequenced using primer for human U6 promoter (refer to Supplementary Table 5 for sequence). Nucleofection was used to deliver the validated lentiCRISPRv2 plasmids into HCT-116 and NALM6 cells. Transfections of the plasmids were carried out in Nucleofector cuvettes, according to the optimized protocols for NALM6 and HCT-116 cells (Lonza Bioscience), with 1×10⁶ cells used and 1X PBS was used instead of the commercial 4-D Nucleofector solution. Post-nucleofection, cells were allowed to grow for 72 hours at 37°C, 5% CO₂, prior to cell sorting (BD FACSAria™ Fusion Flow Cytometer (BD Biosciences)). Cells expressing GFP were selected and grown in 96-well plates containing conditioned media (45% filtered culture media, 45% fresh media, & 10% FBS). After two weeks, sorted clones were transferred into 24-well plates and later into 6-well plates to be collected and validated. Depletion of endogenous ETFDH protein levels in both HCT-116 and NALM-6 cells was initially confirmed by immunoblotting. Candidate clones with loss of ETFDH expression were kept for further validation. Moreover, the genomic DNA loci of candidate clones were sequenced and analyzed for insertion–deletion mutations (INDELs). Two sets of PCR primers were designed in regions flanking the CRISPR target sites (refer to Supplementary Table 5 for validation PCR sequences). Extraction of genomic DNA for each clone was carried out using the PureLink Genomic DNA Mini Kit (Fisher Scientific). Two rounds of polymerase chain reaction (PCR) on the extracted genomic DNA (35 PCR cycles with Set 1 followed by 45 PCR cycles with Set 2) were carried out and gel electrophoresis was utilized to separate the PCR products. The smaller PCR product for each clone was excised and purified using the Zymoclean Gel DNA Recovery Kit (Zymo Research). The products were sequenced, and chromatograms were analyzed for the presence of mutations causing a premature stop-codon or a frameshift. Insertion–deletion mutations (INDELs) in ETFDH were verified by PCR followed by sequencing, while the absence of ETFDH protein was confirmed by western blotting for HCT-116 cells and NALM6 cells.

To deplete ETFDH expression in NMuMG, NT2197 and 4T1 cells, two guides (gRNAs) targeting exon 2 of mouse Etfdh (against mEtfdh sequences 5’-ATTTTTATGCAGCGTATCACTGG-3’ and 5’-GAACATCTTGGAGCACACAGAGG-3’) were designed using CHOPCHOP and cloned in LentiCRISPRv2(GFP) (Addgene #82416) and sequenced as shown above. For lentiviral production, 2.5×10⁵ HEK293T cells were seeded overnight in a 6-cm dish (Sarstedt). The next day, cells were co-transfected with 4µg of respective lentiCRISPRV2 GFP plasmid containing either one of the gRNA inserts or empty-lentiCRISPRV2 GFP plasmid, 2.66µg of psPAX2 packaging plasmid (Addgene 35002), and 1.66 µg of pMD2.G plasmid (Addgene #12259) using the jetPRIME transfection reagent as described by the manufacturer’s protocol (Polyplus transfection). Growth media was changed after 24 hours. After 48 hours, viral supernatant was filtered (0.45 µm filter; (Frogga Bio)), mixed 1:1 with fresh culture media, and applied to target cells grown in 6-well plates along with 8 µg/mL polybrene (Sigma-Aldrich). Cells were re-transduced the following two days with viral supernatant. Cells were allowed to recover for 48 hours post-transduction. Cells were sorted into 96-well plates (Sarstedt) and grown to generate stable cell lines, as shown above. Depletion of mouse Etfdh protein levels was initially confirmed via immunoblotting for NT2197, 4T1, and NMuMG. The presence of INDELs leading to a premature stop codon or frameshift was determined by sequencing as shown above. Two sets of PCR primers were designed in the regions flanking the CRISPR target sites (refer to Supplementary Table 5).

Generation of ETFDH Rescue Cell Lines

To rescue human ETFDH expression in HCT-116 cells, ETFDH(NM_004453) Human Tagged ORF was cloned into pLenti-ETFDH-C-Myc-DDK-P2A-Puro plasmid according to the manufacturer’s instructions (Origene). The resulting plasmid, pLenti-ETFDH-C-Myc-DDK-P2A-Puro, was transfected with psPAX2, pMD2.G into HEK293T cells with JetPrime reagent (Polyplus) according to the manufacturer’s instructions. Viral supernatants were filtered (0.45µM) and applied to HCT-116 ETFDH KO cells with polybrene (6µg/mL). After 6 hours, cells were re-infected overnight. After 2 days, selection with puromycin (4µg/ml) was performed for 72 hours.

Uninfected cells were selected with puromycin to serve as negative controls. The re-introduction of ETFDH expression or control plasmid in the knockout cells was confirmed via western blotting. To rescue ETFDH expression in NT2197 ETFDH KO cells, a lentiviral plasmid conferring hygromycin B resistance and expressing mEtfdh with PAM site mutations or expressing an ORF-Stuffer was created using VectorBuilder (VB221206-1257xet and VB900123-2599cba respectively). Lentivirus production and transduction were carried out as previously mentioned for three consecutive days. After 48h of recovery, cells were selected with 1 mg/mL hygromycin B for 3 days. Cells were confirmed by western blotting.

Generation of Poly(A) Track ETFDH Variants

Mutants were designed by inserting poly(A) tracks of multiple adenosines: control track (CTRL) [6 consecutive lysine AAG codons (AAG)₆], 12 adenosines [12A, equivalent to 4 consecutive AAA (Lys) codons (AAA)₄; 18 adenosines [18A, equivalent to 6 consecutive AAA (Lys) codons (AAA)₆] within the rescue plasmid pLV[Exp]-Hygro-EF1A>mEtfdh. The polyA track insertion site was determined using AlphaFold (version 2) ^37,38 to minimize unintended effects caused by the poly-lysine stretches on the possible mitochondrial import sequence. The polyA tracks were inserted following residue 38, in an unstructured loop of the Etfdh protein. Site-directed mutagenesis was performed using QuikChange II site-directed mutagenesis kit, according to the manufacturer’s instructions (Agilent). Briefly, cycling parameters are followed as: Cycle 1: 95°C for 1min; Cycle 2: 95°C for 30 sec, 60°C for 30 sec, 68°C for 10min; Cycle 3: 68°C for 7min. The resulting plasmids were generated: pLV[Exp]-Hygro-EF1A>mEtfdh-AAG₆, pLV[Exp]-Hygro-EF1A>mEtfdh-AAA₄, pLV[Exp]-Hygro-EF1A>mEtfdh-AAA₆. To generate the poly(A) track Etfdh mutant lines in NT2197 cells (termed AAG₆, AAA₄, AAA₆), lentivirus production and transduction into NT2197 ETFDH KO cells was carried out as previously mentioned for three consecutive days. After 48h of recovery, cells were selected with 1 mg/mL hygromycin B for 3-4 days. ETFDH expression was confirmed by western blotting.

Generation of Cells Expressing EIF4EBP1

pBABE-puro-EV and pBABE-puro-4E-BP1 plasmids were used to overexpress 4E-BP1 in HCT-116 ETFDH KO cells. Retrovirus production and transduction was carried out similarly to lentivirus production, except 2.66 μg of pUMVC packaging plasmid (Addgene #8449) was utilized instead of psPAX2. Following three consecutive days of viral transduction, cells were allowed to recover for 48h prior to selection with 2 μg/mL puromycin (Sigma-Aldrich) for 3 days. To overexpress 4e-bp1 in NT2197 ETFDH KO cells, plasmids containing m4e-bp1 or ORF-Stuffer were generated by VectorBuilder (VB221206-1246sbx & VB900123-2599cba). Lentivirus production and transduction was carried out as previously mentioned for three consecutive days prior to a 48h recovery period. After this period, selection with hygromycin B (1 mg/mL) was conducted for 5-7 days. Validation was observed by western blotting.

Generation of Knockdown Models by shRNA/siRNA

Knockdown using shRNA was performed as previously described ²⁰. For BCL6 knockdown experiments, the following shRNA vectors were used: pLKO.1 Non-Target shRNA Control (Sigma, SHC002), BCL6 shRNA #1 (Sigma, TRCN0000084654), or BCL6 shRNA #2 (TRCN0000084655). For RAPTOR/RICTOR knockdown experiments, the following shRNA vectors were used: scramble shRNA (Addgene, Plasmid #1864), RAPTOR_1 shRNA (Addgene, Plasmid #1857), and RICTOR_1 shRNA (Addgene, Plasmid #1853). Briefly, HEK293T cells were transfected with psPAX2, pMD2.G or target pLKO.1 lentiviral shRNA vectors with JetPrime reagent (Polyplus) according to the manufacturer’s instructions. Viral supernatants were filtered (0.45µM) and transferred to target cells with polybrene (6µg/mL). After 6 hours, cells were re-infected overnight. After 2 days, selection with puromycin (4µg/ml) was performed for 72 hours. Uninfected cells were selected with puromycin to serve as negative controls. Cells were seeded and processed for protein extraction or cell proliferation. Validation was confirmed by western blotting for BCL6 knockdown and RAPTOR or RICTOR knockdown.

Knockdown using siRNA was performed as previously described ⁵⁸. Briefly, target cells were transfected for 24 hours with 50nM Silencer Negative Control No.1 (Thermofisher, AM4611) or BCL6 siRNA (Thermofisher, Cat#4390771) using JetPrime reagent (Polyplus) according to the manufacturer’s instructions. Cells were re-transfected the next day for 24 hours, then processed for protein extraction. Validation was confirmed by western blotting.

Expression Profiles from TCGA/GTEx Databases

Expression of ETFDH and EIF4EBP1 mRNA across both normal and tumor samples were obtained from The Cancer Genome Atlas (TCGA) and GTEx repositories. The expression data were normalized using the TMM (trimmed mean of M values) method and presented as log2 counts per million. Correlations between the genes were assessed using Pearson’s correlation coefficient.

Quantification and Statistical Analysis

Statistical analysis was performed using GraphPad Prism 10. Data are presented as mean +/- SD of independent experiments, unless stated otherwise. Technical replicates were averaged from 2-3 independent experiments. Details on data quantification, presentation, and statistical analysis are included in figure legends.

Data Availability

Plasmids generated in this study will be deposited to Addgene. The original data will be deposited and available at Mendeley Data. Metabolomics data is available at MetaboLights (MTBLS11048).

Supplementary figures

Supplementary Figure 1:

Supplementary Figure 2:

Supplementary Figure 3:

Supplementary Figure 4:

Supplementary Figure 5:

Supplementary Figure 6:

Acknowledgements

We are thankful to the members of all involved laboratories and Stephane Richard for helpful discussions. We thank Luc Choinière and Mariana Russo for assistance with metabolomic experiments and Christian Young for assistance with flow cytometry. This research was funded by the Terry Fox Foundation (TFF) Oncometabolism Team Grant (TFF-242122) to I.T., P.S, and M.P, and Canadian Institutes of Health Research (CIHR) (PJT-183843, PJT-479494) to I.T. Research in OL’s lab is supported by grants from the Swedish Research Council (2020-01665), Swedish Cancer Society (22 2186), the Cancer Research Funds of Radiumhemmet (231263) and the Wallenberg Academy Fellow Program. L-M.P. acknowledges the funds from Queen’s University. D.P. is supported by CIHR Postdoctoral Fellowship (MFE-171312), Cancer Research Society (CRS) The Next Generation of Scientists Award (NGS), and LDI Desjardins Fellowship Program. P.J. and H.K. are supported by Fonds de Recherche du Québec – Santé (FRQS) Fellowships. I.T. is supported by Canada Research Chair in Regulation of mRNA Translation and Metabolism. Metabolic analysis was performed at the Rosalind and Morris Goodman Cancer Research Centre’s Metabolomics Core Facility, which is supported by the Canada Foundation for Innovation, the Dr. John R. and Clara M. Fraser Memorial Trust, the Terry Fox Foundation (TFF Oncometabolism Team Grant; TFF-242122), and McGill University.

Additional information

Author Contributions

Conceptualization, D.P., M.P. and I.T.; Methodology, D.P., R.P., P.J., S.I., H.K., T.B., and S.D.; Formal Analysis, D.P., P.J., S.T., E.C., S.I., K.J.S. and T.B.; Investigation, D.P., R.P., P.J., S.T., E.C., V.S., S.I., S.M., L.Z., H.K., N.C., K.J.S., J.Z., J.V., F.A., and S.M.; Resources, E.G., M.T, A., O.L., L.-M.P., S.D., J.U.-S., P.M.S., M.P., and I.T.; Writing – Original Draft, D.P., M.P., and I.T.; Writing – Review & Editing, all authors; Supervision, M.P., and I.T.; Project Administration, M.P., and I.T.; Funding Acquisition, D.P., M.P., and I.T.

Material & Correspondence

Further information and requests for reagents may be directed upon reasonable request by the lead contact, Ivan Topisirovic (ivan.topisirovic@mcgill.ca).

Additional files

Supplemental Tables. Supplementary Table 1: List of Antibodies Supplementary Table 2: Primer sequences used for RT-qPCR Supplementary Table 3: Primer sequences used for ChIP-qPCR Supplementary Table 4: gRNA sequences used for generation of ETFDH KO cell lines Supplementary Table 5: Primer sequences used for sequencing