Inhibition of Mitochondrial Aconitase by Succination in Fumarate Hydratase Deficiency

1Altered metabolism is a key feature and hallmark of cancer cells. How this arises, and what steps link it to oncogenesis, still eludes us. One possible answer lies with “oncometabolites,” described as metabolites whose abnormal accumulation causes both metabolic and nonmetabolic (such as epigenetic) dysregulation and potential transformation to malignancy. Fumarate hydratase (FH) has been identified as a tumor suppressor because germline loss-of-function mutations are associated with the development of hereditary leiomyomatosis and renal cell cancer (HLRCC). FH has roles in both the mitochondria and cytosol, catalyzing the hydration of fumarate to malate. In mitochondria, FH is a key component of the Krebs cycle, essential for cellular energy production and macromolecular biosynthesis, whereas in the cytoplasm, FH metabolizes fumarate generated from arginine synthesis and the purine nucleotide cycle. Loss of FH activity results in accumulation of fumarate in cells. Elevated fumarate has been implicated in the development of FH-associated tumors through a number of pathways, e.g., by competitive inhibition of 2-oxoglutarate (2OG)-dependent oxygenases, including the hypoxia-inducible factor (HIF) hydroxylases, leading to stabilization of HIF and activation of oncogenic HIF-dependent pathways. However, there is increasing evidence that multiple independent pathways may have roles in FH-associated oncogenesis as a consequence of fumarate acting as an oncometabolite. In addition to being an allosteric inhibitor of the 2OG-dependent oxygenases similar to other oncometabolites, fumarate acts as an endogenous electrophile. It reacts spontaneously by a Michael addition reaction with free sulfhydryl groups to generate a thioether linkage with cysteine residues in proteins. This results in formation of S-(2-succino) cysteine (2SC), a process termed succination. This mechanism is distinct from succinylation of cysteine in which a thioester would be formed. Furthermore, 2SC immunohistochemistry is sufficiently sensitive and specific for use as a clinical biomarker of HLRCC.

2Significantly, succination of Kelch-like ECH-associated protein 1 (KEAP1) in FH-deficient cells leads to abrogation of its interaction with the transcription factor Nuclear factor erythroid 2-related factor 2 (NRF2) and activation of the potentially oncogenic NRF2-mediated antioxidant defense pathway. Furthermore, NRF2 activation has been shown recently to modulate cell metabolism possibly augmenting the cellular stress response. Elucidation of the functional consequences of KEAP1 succination prompted us to search for other 2SC targets that may contribute to the pathogenesis of FH-associated disease. Hence, we conducted a proteomic screen for 2SC in an Fh1-deficient (knockout [KO]) mouse embryonic fibroblast (MEF) cell line and in murine kidney tissue and fluid where Fh1 has been deleted from the kidney tubules. We identified 94 succinated proteins, including some that are succinated on functional cysteine residues. In particular, we investigated the succination of three key cysteines in the Krebs cycle enzyme, mitochondrial aconitate hydratase (Aconitase2, ACO2). We show here that fumarate-mediated succination of ACO2 impairs its enzymatic activity in a dose-dependent manner and that Fh1KO cells exhibit reduced aconitase activity. Our findings further highlight succination as a significant event that could target multiple cellular pathways in FH-associated pathogenesis.

3Previously using Fh1 MEFs, we confirmed by immunoblotting that accumulated intracellular fumarate results in high levels of 2SC in Fh1KO, but not Fh1 wild-type (WT), MEFs. To detect potential 2SC targets at low abundance, we performed mitochondrial and nuclear fractionations of Fh1KO MEFs (Figure S1A). To identify 2SC targets from biological tissue, we used cystic kidneys and aspirated kidney fluid from a 30-week-old Fh1KO mouse where Fh1 is conditionally deleted in the renal tubular epithelium causing the development of hyperplastic cysts. Protein extracts from mitochondrial, nuclear, and cytosolic fractions of Fh1KO MEFs and Fh1KO kidneys were separated by SDS-PAGE analyses and subjected to in-gel trypsin digestion and liquid chromatography tandem mass spectrometry (LC-MS/MS) analyses as described before. Combined proteomic analyses identified 4,095 proteins and 306,558 target peptide spectrum matches (PSMs) from Fh1KO MEFs (false discovery rate [FDR] 2.32%) and 3,569 proteins/226,606 PSMs from Fh1KO kidney tissue and fluid (FDR 1.96%). The MS/MS spectrum for each succination site was verified, and a total of 110 nonredundant 2SC sites were identified in 94 distinct proteins (Table 1). 2SC targets identified thus comprise proteins from diverse cellular pathways; but significantly, approximately 50% are metabolic processes (Figure S1B). Notably, ACO2, mitochondrial NFU1 iron-sulfur cluster scaffold homolog, Protein DJ-1, Peroxiredoxin-1, and Peroxiredoxin-3 are succinated on cysteine residues involved in their function. Also, the succination of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) at C149 was confirmed as reported previously by.

4To investigate the functional consequences of succination, we focused on ACO2 because of its role in the Krebs cycle, where it catalyzes the stereospecific isomerization of citrate to isocitrate via cis-aconitate. In particular, it is a mitochondrial oxidative stress sensor and requires an active [Fe4S4]2+ cluster, bound directly by three conserved cysteine residues, for catalysis. LC-MS/MS analyses of the tryptic peptide 379VGLIGS(2SC)TNSSYEDMGR395 derived from endogenous ACO2 in Fh1KO MEFs assigned succination to C385 unambiguously (Figure 1A). The tryptic peptide spanning C448 and C451 was detected as a mixture of two isomers (438DLGGIVLANA(PEC)GP(2SC)IGQWDR457 and 438DLGGIVLANA(2SC)GP(PEC)IGQWDR457) that are succinated at C451 and C448 (Figure 1B), respectively. Due to their identical mass and composition, the two succinated species could not be separated by LC-MS/MS, but measurement of resulting fragment ion masses in the MS/MS scan allowed identification of succination on both sites. Generally, we observed higher succination at C451 (∼90%) compared to C448 (∼10%) as determined by Mascot analysis.

5To determine if human ACO2 can be succinated on homologous cysteine residues, we transfected Fh1WT and Fh1KO MEFs with a V5-tagged ACO2 gene. LC-MS/MS analysis following V5 immunoprecipitation confirmed succination at all three cysteine residues (C385, C448, and C451) within the active site in ACO2 expressed in only Fh1KO MEFs. To complement the mouse data, we detected succination at C448 and C451 simultaneously from the same tryptic peptide, 438DLGGIVLANA(2SC)GP(2SC)IGQWDRK458 (Figure S1C).

6The three cysteine residues C385, 448, and 451 are crucial for iron-sulfur cluster binding in ACO2 (Figure 2A). To investigate if succination of ACO2 impairs its enzymatic activity, we preincubated pig heart ACO2 with fumarate and assayed its activity in vitro. One hour pre-exposure of ACO2 to increasing concentrations of sodium fumarate at pH 7.4 resulted in dose-dependent inhibition of its activity in the range of 1–50 mM fumarate, which parallels detection of 2SC by immunoblotting (Figure 2B). We then performed LC-MS/MS analyses of trypsin-digested ACO2 derived from assay mixtures. Succination was detected at five cysteine residues (C126, C385, C410, C451, and C592), and further, the levels of succination of the C385- and C448/C451-containing peptides increased with increasing fumarate concentration (Figures 2C, 2D, and S2A). We correlated ACO2 activity with succination in the range of 5–50 mM fumarate and obtained a negative linear correlation with succinated peptides containing 2SC385 and 2SC451/448 (Figure S2B). To relate the in vitro data to pathophysiological settings, we measured fumarate concentrations by capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS) in FH-deficient mouse kidneys and HLRCC tumors; these were estimated to contain 1.7 ± 0.4 mM and 3.4 ± 1.2 mM fumarate, respectively (Figures 2E and 2F). We calculated the fumarate concentration on the basis of tissue weight, but these values are likely an underestimate because the tissue is comprised of a heterogeneous population of cells, and no estimate was made of the aqueous volume of the tissue.

7When whole-cell lysates of Fh1WT and Fh1KO MEFs were compared to determine if succination impairs endogenous ACO2 activity, Fh1KO MEFs displayed significantly reduced aconitase activity (Figure 3A). To differentiate mitochondrial and cytosolic aconitase activity (ACO2 and ACO1, respectively), we utilized two cell lines derived from the Fh1KO MEFs, reconstituted with either full-length FH (Fh1KO+FH), or FH restricted to the cytosol by deleting the mitochondrial-targeting sequence (Fh1KO+FHcyt). Comparison of whole-cell lysates from the four MEF cell lines showed that aconitase activity is completely restored in Fh1KO+FH cells and only partially restored in Fh1KO+FHcyt, relative to that in Fh1KO cells (Figure 3A).

8Previously, we showed that despite having significantly reduced total cellular fumarate compared to Fh1KO, Fh1KO+FHcyt MEFs (as measured by 1H-nuclear magnetic resonance spectrometry) retain abnormal mitochondria morphology and impaired respiration. We redetermined fumarate levels in the four cell lines by CE-TOFMS (Figure 3B) and confirmed high levels of fumarate in Fh1KO MEFs (∼35 fmol/cell) and above-normal levels of fumarate in Fh1KO+FHcyt (∼10 fmol/cell) compared to Fh1WT (∼1.5 fmol/cell) and Fh1KO+FH (∼3 fmol/cell) MEFs. To relate these levels to molar concentrations, we performed cell volume measurements of the four MEF cell lines by atomic force microscopy. These analyses estimated the intracellular fumarate concentrations to be ∼6 mM for Fh1KO, ∼0.06 mM for Fh1WT, ∼0.14 mM for Fh1KO+FH, and ∼1.3 mM for Fh1KO+FHcyt MEFs (Figure S3). We postulated that fumarate may be accumulated in the mitochondria of Fh1KO+FHcyt MEFs, and consequently, ACO2 may be succinated in these cells. Therefore, we compared ACO2 succination in the four MEF cell lines by LC-MS/MS analyses of the 2SC385- and 2SC451/2SC448-containing tryptic peptides. Whereas no succination was detected at C385 or C451/C448 in Fh1WT or Fh1KO+FH cells, succination was detected in both Fh1KO and Fh1KO+FHcyt cells with that in Fh1KO being higher (Figure 3C). Following fractionation of the four cell lines into mitochondrial versus cytoplasmic portions, we performed immunoblotting of the derived protein extracts and confirmed the presence of 2SC in both the mitochondria and cytosol of Fh1KO MEFs and also in the mitochondria of Fh1KO+FHcyt MEFs (Figure S1A). Taken together, our data suggest that the partial restoration of aconitase activity in Fh1KO+FHcyt MEFs may be a combined effect of functional ACO1 activity and dysfunctional ACO2 due to succination in the mitochondria.

9To investigate if succination of aconitase might cause alteration to cellular metabolism, we first measured the levels of key Krebs cycle metabolites in Fh1WT and Fh1KO MEFs by CE-TOFMS. Consistent with FH being dysfunctional, levels of fumarate and succinate are significantly higher in Fh1KO, whereas that for malate is drastically lower (Figure 3D). Notably, we observed low levels of citrate and isocitrate in Fh1KO MEFs. We then cultured the cells in deuterium-labeled [d5]glutamine for 24 hr and analyzed them for label incorporation into these metabolites. We observed significant label incorporation in succinate (m+4) and fumarate (m+2), supporting the oxidative flux of the Krebs cycle. We also detected isocitrate m+2 but did not observe label enrichment in citrate (Figure 3E).

10Some aerobic glycolytic cancer cells display altered metabolism by utilizing the glutamine-dependent reductive mechanism to produce citrate, which can be used for lipogenesis and for anaplerosis of the Krebs Cycle. This pathway uses the NADP(+)-dependent isocitrate dehydrogenase (IDH)1 and 2 to reductively carboxylate 2OG to isocitrate and is considered to occur in both the mitochondria and cytosol. Our data suggest that in Fh1KO MEFs, 2OG can be converted to isocitrate by reversal of the IDH-catalyzed reaction, but isocitrate cannot be further metabolized to citrate due to impaired aconitase activity, possibly as a result of succination. Hence, succination of ACO2 may prevent Fh1KO MEFs from utilizing the reductive carboxylation pathway for citrate synthesis as adopted by some cancer cell lines. Furthermore, the absence of label in citrate suggests that both mitochondrial and cytosolic aconitase are potentially inactive in Fh1KO MEFs (Figure 3F).

11Here, we report the results of a proteomic screen to identify 2SC targets in FH deficiency. We describe the succination of three cysteine residues crucial for iron-sulfur cluster binding in the active site of the Krebs cycle enzyme mitochondrial Aconitase2 (ACO2) in Fh1KO cells, which exhibit reduced aconitase activity compared to Fh1WT cells. We have demonstrated that in vitro inhibition of ACO2 is a direct consequence of dose-dependent fumarate-mediated succination, particularly at ≥5 mM fumarate, equivalent to concentrations measured in FH-deficient tissues. Because tissue samples are a heterogeneous mix of both control and FH-deleted cells, precise determination of intracellular fumarate is difficult, and the actual concentrations in these FH-deficient cells could be significantly underestimated. Fumarate concentrations in subcellular compartments, e.g., mitochondria versus cytosol, could also be variable and differentially affect local protein succination. Additionally, compared to the relatively simple in vitro situation, the catalytic activity of aconitase in cells could be influenced by multiple components such as cosubstrate and cofactor availability. Interestingly, we did identify succination of proteins involved in iron-sulfur cluster assembly in our proteomic screen, potentially further hindering aconitase activity. Stable isotope tracer studies showed that Fh1KO MEFs do not utilize the reductive carboxylation mechanism for citrate synthesis, which may be a consequence of fumarate-dependent succination of ACO2, adding a further layer of complexity to the disruption of mitochondrial metabolism caused by FH deficiency.

12Cytosolic Aconitase1 (ACO1) also contains three iron-sulfur-binding cysteine residues and is a bifunctional enzyme that acts either as an iron response element (IRE)-binding protein to regulate iron uptake, sequestration, and utilization or as the cytosolic aconitase, depending on iron availability. Whether succination affects the IRE-binding ability of ACO1 and, by inference, iron homeostasis in FH-deficient cells is an interesting question that warrants future investigation.

132SC has been described in aging and diabetes, and its functional consequences have been reported for GAPDH and adiponectin in addition to the KEAP1/NRF2 pathway. Our proteomic screen for 2SC targets aims to expand our current knowledge of the extent of this modification and its cellular impacts. Despite the fact that our screen is biased toward abundant proteins, it is significant that whereas proteins encompassing diverse cellular pathways are targets for succination, around half are involved in metabolism. A few proteins including the iron-sulfur cluster assembly protein NFU1 and the thioredoxin-dependent peroxide reductases are succinated on critical cysteine residues, suggesting that succination may adversely affect function in these targets. The thioether adduct generated by fumarate modifications occurs nonenzymatically and is believed to be irreversible. However, it is conceivable that 2SC may influence signal transduction by targeting proteins that are cellular stress sensors such as KEAP1. Alternatively, 2SC may compete with other cysteine modifications such as S-nitrosylation and oxidation to sulfinic acid to indirectly target other cellular signaling events. Although the effects on individual proteins require closer investigations, our data provide evidence that succination is a significant posttranslational modification in FH deficiency and a potential key mechanism linking multiple pathways that may cause dysregulation of cell metabolism and contribute to oncogenesis.

14See also Extended Experimental Procedures.

15The aconitase assay is based on the protocol described in the Aconitase Assay Kit (Cayman Chemical) with modifications. NADPH production was followed by fluorescence (excitation 340 nm; emission of 465 nm) over 45 min at 37°C. Activation of pig heart aconitase and preparation of MEF cell lysates followed the manufacturer’s protocol.

16Cell fractionations were performed using Qproteome Mitochondria Isolation Kit (QIAGEN), or as previously described. Kidney samples were homogenized and sonicated in Urea-SDS buffer. Protein extracts were separated by SDS-PAGE and processed for trypsin digestion and LC-MS/MS analyses as previously described. Database searches were performed against SwissProt (06/2011) or International Protein Index (09/2012) database using Mascot or CPFP 1.3.0. For label-free quantitation of succinated peptides, samples were analyzed in three technical replicates. Relative quantitation was performed using Progenesis LC-MS v.4.0. Correlation analysis was performed using GraphPad Prism v.5 assuming Gaussian populations (Pearson) calculating two-tailed p values with a confidence interval of 95%. Tissue and cell samples for metabolite analysis by CE-TOFMS were prepared as described before.

17All procedures were conducted in line with American Association for Cancer Research guidelines and performed under UK Home Office regulations after approval by the Local Ethical Review Process at Oxford University. Anonymized human tumor and normal samples were collected with full ethical approval (MREC 05/Q1605/66) as approved by the Oxford Centre for Histopathology Research.

18Four mouse embryonic fibroblasts (MEFs) cell lines were used: Fh1+/+ (Fh1WT), Fh1−/− (Fh1KO), and isogenic Fh1KO MEFs, reconstituted with either full-length FH (Fh1KO+FH), or cytosolic-restricted FH by deleting the mitochondrial targeting sequence (Fh1KO+FHcyt). Human ACO2 gene was amplified from HEK293T cDNA libraries by PCR and ligated into pEF1/V5-HisA vector (Invitrogen). Transfection used FuGene®6 (Roche) following the manufacturer’s protocol. Cell culture was performed as previously described. Immunoprecipitation of V5-tagged ACO2 was performed according to the manufacturer’s protocol using V5 agarose beads (Invitrogen).Primary antibodies used were V5 (Invitrogen), FH (Autogen Bioclear), COXIV (Cell Signaling Technology), ACO1 (Cell Signaling Technology), ACO2 (Cell Signaling Technology) and 2SC and TUBB (Abcam). Immunoblotting was performed as previously described.Tissue and aspirated fluid from cystic kidneys were obtained from Fh1 fl/fl Ksp-cre (Fh1KO) as described previously.Pig heart aconitase (39.03 μg/ml) (Cayman) was activated by incubation at 4°C for 1 hr in 475 mM Tris-Cl, pH 7.4, containing 1.6 mM cysteine hydrochloride and 17 mM ferrous ammonium sulfate and assayed in a final total volume of 205 μl containing 256 mM Tris-Cl, pH 7.4; 0.4 mM cysteine hydrochloride; 4.25 μM ferrous ammonium sulfate; 0.5 mM sodium citrate; 0.25 mM NADP+ and 3.75 U of isocitrate dehydrogenase. For in vitro succination porcine ACO2 was incubated with sodium fumarate (0-150 mM final concentration) 15 min at 4°C prior to activation. Aconitase activity was determined by the rate of NADPH production by fluorescence (excitation 340 nm; emission of 465 nm) over 45 min 37°C in a Safire 2 plate reader (Tecan). Cell lysates for aconitase assay were prepared by harvesting the MEFs at 4°C in Homogenization Buffer (50 mM Tris-HCl, pH 7.4, 0.1 mM sodium citrate), followed by sonication and centrifugation at 14,000 rpm for 10 min at 4°C to clear the lysate.For proteomics identification of 2SC targets, protein extracts were separated by SDS-PAGE and excised sequentially into 6X 1cm-long gel slides before cutting to small pieces for further processing. hACO2 immunoprecipitated from Fh1WT and Fh1KO MEFs and porcine ACO2 derived from aconitase assay mixtures were separated by SDS-PAGE and gel bands corresponding to the size of ACO2 were excised. All gel slices were subjected to reductive alkylation and in-gel trypsin digestion as previously described.Peptides were separated on an Acquity nano UPLC system (Waters) supplemented with a 25 cm C18 column, 1.7 μm particle size (Waters) using a linear gradient from 3% buffer A (0.1% formic acid in water) to 40% buffer B (0.1% formic acid in acetonitrile) at a flow rate of 250 nl/min (approx. 7000 psi) from 0 to 60 or 90 min according to the complexity of the sample. Peptides were ionized and introduced to an LTQ Orbitrap Velos tandem mass spectrometer (Thermo Scientific) using an electrospray ionization (ESI) source. Collision induced dissociation (CID) was induced on the twenty most abundant ions per full MS scan using an isolation width of 1.5 Da. All fragmented precursor ions were actively excluded from repeated MS/MS analysis for 15 s. Raw data was converted to Mascot generic files using msconvert. Searches were performed using the SwissProt database (06/2011) with MASCOT and CPFP 1.3.0 with the following settings: Variable modifications: 2-succinyl (cysteine, +116.01 Da), pyridylethyl (cysteine, +105.06 Da), oxidation (methionine, +15.00 Da), peptide tolerance: ± 10 ppm, fragment tolerance: ± 0.5 Da.Atomic force microscopy was performed as previously described using a Bioscope Catalyst instrument (Bruker Nano GmbH). MEFs were plated overnight at low confluence in standard DMEM. Measurements were made using ScanAsyst-Fluid probes in PFQNM mode, regular SW (drive amplitude 300 nm) but special workspace (drive frequency 0.5 kHz). A minimum of 5 locations were scanned for each cell line and at least 1 high resolution image was captured.Cells were cultured in standard DMEM for 12 hr and then in medium containing 4 mM [d5]-Glutamine (Glutamine-2,3,3,4,4) for 24 hr prior to processing for isotope incorporation and metabolite analysis. Anionic and cationic metabolite profiling were performed as described previously. The automatic recalibration of each acquired spectrum was performed against reference masses of reference standards ([13C isotopic ion of protonated methanol dimer (2MeOH+H)]+, m/z 66.0632) and ([Hexakis (2,2- difluorothoxy)phosphazene +H] +, m/z 622.0290) and exact mass data were acquired as outlined in. Essentially all other details are as described previously.

19Four mouse embryonic fibroblasts (MEFs) cell lines were used: Fh1+/+ (Fh1WT), Fh1−/− (Fh1KO), and isogenic Fh1KO MEFs, reconstituted with either full-length FH (Fh1KO+FH), or cytosolic-restricted FH by deleting the mitochondrial targeting sequence (Fh1KO+FHcyt). Human ACO2 gene was amplified from HEK293T cDNA libraries by PCR and ligated into pEF1/V5-HisA vector (Invitrogen). Transfection used FuGene®6 (Roche) following the manufacturer’s protocol. Cell culture was performed as previously described. Immunoprecipitation of V5-tagged ACO2 was performed according to the manufacturer’s protocol using V5 agarose beads (Invitrogen).

20Primary antibodies used were V5 (Invitrogen), FH (Autogen Bioclear), COXIV (Cell Signaling Technology), ACO1 (Cell Signaling Technology), ACO2 (Cell Signaling Technology) and 2SC and TUBB (Abcam). Immunoblotting was performed as previously described.

21Tissue and aspirated fluid from cystic kidneys were obtained from Fh1 fl/fl Ksp-cre (Fh1KO) as described previously.

22Pig heart aconitase (39.03 μg/ml) (Cayman) was activated by incubation at 4°C for 1 hr in 475 mM Tris-Cl, pH 7.4, containing 1.6 mM cysteine hydrochloride and 17 mM ferrous ammonium sulfate and assayed in a final total volume of 205 μl containing 256 mM Tris-Cl, pH 7.4; 0.4 mM cysteine hydrochloride; 4.25 μM ferrous ammonium sulfate; 0.5 mM sodium citrate; 0.25 mM NADP+ and 3.75 U of isocitrate dehydrogenase. For in vitro succination porcine ACO2 was incubated with sodium fumarate (0-150 mM final concentration) 15 min at 4°C prior to activation. Aconitase activity was determined by the rate of NADPH production by fluorescence (excitation 340 nm; emission of 465 nm) over 45 min 37°C in a Safire 2 plate reader (Tecan). Cell lysates for aconitase assay were prepared by harvesting the MEFs at 4°C in Homogenization Buffer (50 mM Tris-HCl, pH 7.4, 0.1 mM sodium citrate), followed by sonication and centrifugation at 14,000 rpm for 10 min at 4°C to clear the lysate.

23For proteomics identification of 2SC targets, protein extracts were separated by SDS-PAGE and excised sequentially into 6X 1cm-long gel slides before cutting to small pieces for further processing. hACO2 immunoprecipitated from Fh1WT and Fh1KO MEFs and porcine ACO2 derived from aconitase assay mixtures were separated by SDS-PAGE and gel bands corresponding to the size of ACO2 were excised. All gel slices were subjected to reductive alkylation and in-gel trypsin digestion as previously described.

24Peptides were separated on an Acquity nano UPLC system (Waters) supplemented with a 25 cm C18 column, 1.7 μm particle size (Waters) using a linear gradient from 3% buffer A (0.1% formic acid in water) to 40% buffer B (0.1% formic acid in acetonitrile) at a flow rate of 250 nl/min (approx. 7000 psi) from 0 to 60 or 90 min according to the complexity of the sample. Peptides were ionized and introduced to an LTQ Orbitrap Velos tandem mass spectrometer (Thermo Scientific) using an electrospray ionization (ESI) source. Collision induced dissociation (CID) was induced on the twenty most abundant ions per full MS scan using an isolation width of 1.5 Da. All fragmented precursor ions were actively excluded from repeated MS/MS analysis for 15 s. Raw data was converted to Mascot generic files using msconvert. Searches were performed using the SwissProt database (06/2011) with MASCOT and CPFP 1.3.0 with the following settings: Variable modifications: 2-succinyl (cysteine, +116.01 Da), pyridylethyl (cysteine, +105.06 Da), oxidation (methionine, +15.00 Da), peptide tolerance: ± 10 ppm, fragment tolerance: ± 0.5 Da.

25Atomic force microscopy was performed as previously described using a Bioscope Catalyst instrument (Bruker Nano GmbH). MEFs were plated overnight at low confluence in standard DMEM. Measurements were made using ScanAsyst-Fluid probes in PFQNM mode, regular SW (drive amplitude 300 nm) but special workspace (drive frequency 0.5 kHz). A minimum of 5 locations were scanned for each cell line and at least 1 high resolution image was captured.

26Cells were cultured in standard DMEM for 12 hr and then in medium containing 4 mM [d5]-Glutamine (Glutamine-2,3,3,4,4) for 24 hr prior to processing for isotope incorporation and metabolite analysis. Anionic and cationic metabolite profiling were performed as described previously. The automatic recalibration of each acquired spectrum was performed against reference masses of reference standards ([13C isotopic ion of protonated methanol dimer (2MeOH+H)]+, m/z 66.0632) and ([Hexakis (2,2- difluorothoxy)phosphazene +H] +, m/z 622.0290) and exact mass data were acquired as outlined in. Essentially all other details are as described previously.