Temporal fluxomics reveals oscillations in TCA cycle flux throughout the mammalian cell cycle
2017; Springer Nature; Volume: 13; Issue: 11 Linguagem: Inglês
10.15252/msb.20177763
ISSN1744-4292
AutoresEunyong Ahn, Praveen Kumar, Dzmitry Mukha, Amit Tzur, Tomer Shlomi,
Tópico(s)Adipose Tissue and Metabolism
ResumoArticle9 November 2017Open Access Transparent process Temporal fluxomics reveals oscillations in TCA cycle flux throughout the mammalian cell cycle Eunyong Ahn Eunyong Ahn Department of Computer Science, Technion, Haifa, Israel Search for more papers by this author Praveen Kumar Praveen Kumar Department of Biology, Technion, Haifa, Israel Search for more papers by this author Dzmitry Mukha Dzmitry Mukha Department of Biology, Technion, Haifa, Israel Search for more papers by this author Amit Tzur Amit Tzur Faculty of Life Sciences, Bar-Ilan University, Ramat Gan, Israel The Institute of Nanotechnology and Advanced Materials, Bar-Ilan University, Ramat Gan, Israel Search for more papers by this author Tomer Shlomi Corresponding Author Tomer Shlomi [email protected] orcid.org/0000-0003-3491-5156 Department of Computer Science, Technion, Haifa, Israel Department of Biology, Technion, Haifa, Israel Lokey Center for Life Science and Engineering, Technion, Haifa, Israel Search for more papers by this author Eunyong Ahn Eunyong Ahn Department of Computer Science, Technion, Haifa, Israel Search for more papers by this author Praveen Kumar Praveen Kumar Department of Biology, Technion, Haifa, Israel Search for more papers by this author Dzmitry Mukha Dzmitry Mukha Department of Biology, Technion, Haifa, Israel Search for more papers by this author Amit Tzur Amit Tzur Faculty of Life Sciences, Bar-Ilan University, Ramat Gan, Israel The Institute of Nanotechnology and Advanced Materials, Bar-Ilan University, Ramat Gan, Israel Search for more papers by this author Tomer Shlomi Corresponding Author Tomer Shlomi [email protected] orcid.org/0000-0003-3491-5156 Department of Computer Science, Technion, Haifa, Israel Department of Biology, Technion, Haifa, Israel Lokey Center for Life Science and Engineering, Technion, Haifa, Israel Search for more papers by this author Author Information Eunyong Ahn1,‡, Praveen Kumar2,‡, Dzmitry Mukha2, Amit Tzur3,4 and Tomer Shlomi *,1,2,5 1Department of Computer Science, Technion, Haifa, Israel 2Department of Biology, Technion, Haifa, Israel 3Faculty of Life Sciences, Bar-Ilan University, Ramat Gan, Israel 4The Institute of Nanotechnology and Advanced Materials, Bar-Ilan University, Ramat Gan, Israel 5Lokey Center for Life Science and Engineering, Technion, Haifa, Israel ‡These authors contributed equally to this work *Corresponding author. Tel: +972 77 887 1543; E-mail: [email protected] Molecular Systems Biology (2017)13:953https://doi.org/10.15252/msb.20177763 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Abstract Cellular metabolic demands change throughout the cell cycle. Nevertheless, a characterization of how metabolic fluxes adapt to the changing demands throughout the cell cycle is lacking. Here, we developed a temporal-fluxomics approach to derive a comprehensive and quantitative view of alterations in metabolic fluxes throughout the mammalian cell cycle. This is achieved by combining pulse-chase LC-MS-based isotope tracing in synchronized cell populations with computational deconvolution and metabolic flux modeling. We find that TCA cycle fluxes are rewired as cells progress through the cell cycle with complementary oscillations of glucose versus glutamine-derived fluxes: Oxidation of glucose-derived flux peaks in late G1 phase, while oxidative and reductive glutamine metabolism dominates S phase. These complementary flux oscillations maintain a constant production rate of reducing equivalents and oxidative phosphorylation flux throughout the cell cycle. The shift from glucose to glutamine oxidation in S phase plays an important role in cell cycle progression and cell proliferation. Synopsis Metabolic flux dynamics is quantified throughout the cell cycle via isotope tracing and computational modelling. This approach reveals oscillating metabolism of glucose versus glutamine in the Krebs cycle and highlights the importance of glutamine for S phase progression. A temporal-fluxomics approach is used for quantifying metabolic flux dynamics throughout the mammalian cell cycle. TCA cycle fluxes are rewired as cells progress through the cell cycle with complementary oscillations of glucose versus glutamine-derived fluxes. Glucose-derived oxidative-TCA-cycle flux peaks in late G1 phase while oxidative and reductive glutamine metabolism dominates S phase. The shift from glucose to glutamine oxidation in S phase plays an important role in cell cycle progression. Introduction Cell cycle progression is tightly interlinked with cellular metabolism (Kaplon et al, 2015). The availability of sufficient metabolic nutrients and intracellular energy status controls the ability of cells to enter and progress through cell cycle. The absence of glucose was first shown to arrest cells at the G1/S restriction point (Blagosklonny & Pardee, 2002). More recently, cellular energy status (ATP/AMP ratio) was found to regulate canonical cell cycle signaling pathways via AMP-activated protein kinase (AMPK; Banko et al, 2011). The mammalian target of rapamycin (mTOR) plays a central role in regulating cell cycle progression and growth, integrating stimuli of amino acid, energy, and oxygen availability (Fingar & Blenis, 2004; Cuyàs et al, 2014). Cell cycle progression is further controlled by intracellular metabolites affecting epigenetics: Nuclear acetyl-CoA levels, determined by nuclear ATP citrate lyase (ACL; Wellen et al, 2009) and pyruvate dehydrogenase (PDH; Sutendra et al, 2014), regulate the acetylation of histones and thus control cell cycle progression (Berger, 2007; Li et al, 2007). Additionally, several metabolic enzymes were shown to directly regulate the cell cycle machinery, including PFKFB3 and PKM2, controlling the activity of cyclins and cyclin-dependent kinase (CDK) inhibitors in the nucleus (Yalcin et al, 2009; Yang et al, 2011, 2012). Signaling pathways that coordinate cell cycle progression further regulate metabolic activity to support the changing metabolic demands throughout the cell cycle. The ubiquitin proteasome system, which tightly controls the concentration of cyclins, regulates the activity of two key enzymes in glucose and glutamine metabolism (Almeida et al, 2010; Tudzarova et al, 2011; Estevez-Garcia et al, 2014); the ubiquitin ligase anaphase-promoting complex/cyclosome (APC/C) and ligase Skp1/cullin/F-box protein (SCF) complex control glycolytic flux via PFKFB3, restricting its expression to late G1 and early S; APC/C also regulates glutaminolysis via glutaminase 1 (GLS1), whose expression is induced in S and G2/M. Cyclins and cyclin-CDK complexes were further suggested to regulate central metabolic activities, including glycolysis, lipogenesis, and mitochondrial activity (Hsieh et al, 2008; Bienvenu et al, 2010). Furthermore, central oncogenes and tumor suppressors that control proliferation, growth, and cell cycle can stimulate the expression of enzymes that mediate glycolysis and glutaminolysis (Levine & Puzio-Kuter, 2010). While the cell cycle machinery was found to regulate the concentration of key metabolic enzymes, an understanding of how the actual rate of metabolic reactions and pathway (i.e., metabolic flux) changes throughout the cell cycle is still fundamentally missing. Metabolic flux is a not a directly measurable quantity and is typically inferred using isotope tracing techniques coupled with computational metabolic flux analysis (MFA; Wiechert, 2002; Sauer, 2006). Isotope tracing coupled with MFA is commonly used to address problems in biotechnology and medicine and has recently become a central technique in studies of cancer cellular metabolism (Metallo et al, 2009; Duckwall et al, 2013). Applied to cell populations with cells at different phases of the cell cycle, this approach typically estimates the average flux throughout the cell cycle. Here, we present a temporal-fluxomics approach for quantifying cell cycle-dependent oscillations in metabolic flux, combining isotope tracing in synchronized cell populations with computational deconvolution and metabolic network modeling. We applied this approach to derive a first comprehensive and quantitative view of flux dynamics in central metabolism of proliferating cancer cells. The analysis adds a temporal dimension to our understanding of TCA cycle metabolism, showing major oscillations in the oxidation/reduction of glucose versus glutamine-derived fluxes as cells progress through the cell cycle. Results Cellular concentration of central metabolic intermediates oscillate throughout the cell cycle To study metabolic dynamics throughout cell cycle, we synchronized HeLa cells using double thymidine block and applied high-throughput LC-MS-based targeted metabolomics analysis to synchronized cell populations (> 106 cells per sample) in 3-h intervals for two complete cell cycles (see cell synchronization dynamics measured via propodium iodide staining/FACS analysis in Fig 1A; Appendix Fig S1; Materials and Methods). To obtain a reliable and accurate view of periodic metabolic oscillations and to overcome a potential perturbation of metabolism due to synchronization-induced growth arrest, we let synchronized cells complete one cell cycle before starting the LC-MS analysis (9 h after cells are released in G1/S). Measured metabolite abundances in the synchronized cells were normalized by total cell volume in each time point to determine metabolite concentrations. Figure 1. Detection of oscillations in metabolite concentrations throughout the cell cycle in HeLa cells revealed by LC-MS-based metabolomics of synchronized HeLa cells and computational deconvolution Synchronization dynamics of a population of HeLa cells within almost three complete cell cycles measured via PI staining followed by FACS analysis. The increase in cell number following each mitosis is shown by an overlaid curve (in orange, mean and s.d. of n = 3). Computational modeling of the synchronization loss, considering 11% cell–cell variation in doubling time, shows that the simulated fraction of the cells in G1, S, and G2/M in the synchronized cells throughout the cell cycle (straight lines) match experimental measurements (marked with asterisk). The measured average cell volume in the synchronized cell population (red, mean and s.d. of n = 3, v(t) in equation 6), the deconvoluted signal (in case of no synchronization loss, green, in equation 6), and the simulated average cell volume considering the loss in synchronization (black, matching the measured concentration data, equation 6). The measured concentration of CTP in synchronized cells shown in red (mean and s.d. of n = 5, ui(t) in equation 7), the deconvoluted concentration dynamics, in case of no synchronization loss (green, in equation 7), and the expected concentration dynamics based on the deconvoluted concentrations and considering the loss in synchronization, matching the measured concentrations (black, equation 7). The measured concentrations converge toward the steady-state concentrations measured in non-synchronized cells (horizontal blue line). Download figure Download PowerPoint As cell synchronization is gradually lost with time due to inherent non-genetic cell-to-cell variability (a phenomenon also known as “dispersion”), the distribution of cell cycle phases in the synchronized cell population becomes similar to that of non-synchronized cells after completing three rounds of replications (Fig 1A). To account for the loss of synchrony and to precisely quantify oscillations in metabolite levels, we employed “computational synchronization” (Bar-Joseph et al, 2008): We constructed a probabilistic model that describes the dynamics of the cell population loosing synchrony, assuming that each cell has its own “internal clock” which controls the cell cycle progression rate (see Synchronization loss model in Materials and Methods). The parameters of the model were estimated by fitting a simulation of how the synchronized cell population progresses through the different phases of the cell cycle with corresponding FACS measurements, finding that cell–cell variability in the rate of cell cycle progression through the cell cycle is 11% (Fig 1B, Materials and Methods). We used this model for computational deconvolution of measured metabolomics data, estimating metabolite concentration dynamics throughout the cell cycle, circumventing the impact of cell dispersion (Materials and Methods). Inferring the dynamics of metabolite concentrations throughout the cell cycle (rather than that of metabolite abundances) further required estimates of the dynamics of cell volume throughout the cell cycle. The latter was estimated based on deconvolution of total cell volume measurements performed in the synchronized cell population (Materials and Methods, Fig 1C). For example, the concentration of the nucleotide cytidine triphosphate (CTP) was found to oscillate throughout cell cycle, showing a ~50% increase in concentration in G1 phase versus G2/M phase (measured and deconvoluted concentrations shown in Fig 1D). As shown, the magnitude of the oscillation drops with time and converges to the steady-state concentration measured in non-synchronized cells. Our analysis reveals 57 metabolites whose concentrations significantly oscillate throughout the cell cycle (Fig 2, Dataset EV1, FDR-corrected P-value < 0.05, Materials and Methods). Oscillations in nearly 44% of these metabolites could be detected only when in silico synchronization via computational deconvolution was applied (i.e., equation 7 in Materials and Methods), emphasizing the strength of our pipeline. The median size of the observed oscillations is ~60% (difference between maximal and minimal concentration throughout the cell cycle); roughly one-quarter of these metabolites show concentration changes larger than twofold throughout cell cycle. A significantly high fraction of the metabolites peaks either in late G1 (~50% in the second half of G1, P-value < 10−5, compared with the expected fraction assuming that concentration peaks are uniformly distributed throughout the entire cell cycle) or early S (~35% in the first half of S, P-value < 0.002). Figure 2. Oscillation in metabolite concentrations throughout the cell cycle in HeLa cellsThe figure shows metabolites found to significantly oscillate throughout the cell cycle (concentrations normalized per metabolite, maximal concentration in red, minimal concentration in blue). The amplitude of the oscillations is shown on the right. Metabolites are color-coded according to metabolic pathways: energy/redox cofactors (purple), amino acids (blue), glycolytic metabolites (red), and TCA cycle metabolites (green). Download figure Download PowerPoint The oscillating metabolites include glycolytic and TCA cycle intermediates, nucleotides, amino acids, and energy and redox cofactors. Expectedly, the concentrations of the deoxynucleotides dCTP and dATP increase in S phase, when utilized for DNA replication. Intermediates in polyamine metabolism (5-methylthioadenosine, acetyl-putrescine, S-adenosyl-L-methionine, and S-adenosyl-L-homocysteine) show marked oscillations, in accordance with the known cell cycle-dependent activity of this pathway (Oredsson, 2003). Several glycolytic metabolites peak during G1/S transition, in accordance with reports of increased glycolytic flux at this cell cycle phase (Colombo et al, 2011; Tudzarova et al, 2011). Cellular ATP/ADP ratio and redox potential (NADH/NAD+) further show a ~50% increase in the G1/S transition (Appendix Fig S2). Intracellular concentration of non-essential amino acids synthesized from consumed glutamine, including glutamate, ornithine, proline, and aspartate peak in S phase, in accordance with a reported increase in glutamine dependence in S phase (Gaglio et al, 2009; Colombo et al, 2011). Intriguingly, we find that different TCA cycle metabolites peak in distinct cell cycle phases: Acetyl-CoA and citrate peak in G1/S, while malic acid and α-ketoglutarate peak in late S, suggesting that the TCA cycle is rewired as cells progress through the cell cycle. Time-resolved fluxomics reveals increased glycolytic flux into TCA cycle in G1/S transition To observe metabolic flux dynamics in TCA cycle and in branching pathways throughout the cell cycle, we performed pulse-chase isotopic tracing experiments in synchronized HeLa cells with [U-13C]-glucose and [U-13C]-glutamine (1-h feeding), every 3 h for two cell cycles (Fig 3A and B, Materials and Methods). Here, LC-MS was utilized to measure the mass-isotopomer distribution of metabolites (i.e., the fraction of each metabolite pool having zero, one, and two labeled carbon atoms) after 1 h of feeding with the isotopic tracers. Computational deconvolution was employed to analyze oscillations in metabolite isotopic labeling patterns while considering cell dispersion. The deconvolution approach is based on the observation that the measured fractional isotopic labeling of a metabolite in the synchronized cell population represents the average labeling in cells with distinct intrinsic times, weighted by the metabolite pool size in these cells (i.e., the measured isotopic labeling pattern is biased toward that of cells with an intrinsic time in which the metabolite pool size is larger than in others, Materials and Methods). Overall, we detected statistically significant oscillations in the isotopic labeling pattern of 21 metabolites when feeding isotopic glucose, and 16 metabolites when feeding isotopic glutamine (FDR-corrected P-value < 0.05, Figs 3 and 4, Dataset EV2). Figure 3. Oscillations in isotopic labeling of TCA cycle metabolites throughout the cell cycle from [U-13C]-glucose show induced glycolytic flux into TCA cycle in G1/S A. Experimental scheme for a series of pulse-chase isotope tracing experiments in synchronized cells. B. Atom tracing of TCA cycle metabolites from [U-13C]-glucose (blue) and [U-13C]-glutamine (red). C–E. Measured relative fraction of the m + 2 labeling of TCA cycle intermediates after feeding [U-13C]-glucose (red, mean and s.d. of n = 3), the deconvoluted signal (green), and the expected labeling dynamics considering the loss in synchronization (black, representing TCA cycle oxidation of glucose-derived acetyl-CoA). F. Oscillations in citrate m + 4 labeling after feeding [U-13C]-glutamine (mean and s.d. of n = 3; experimentally measured labeling dynamics shown in red, the deconvoluted signal in green, and the expected labeling dynamics considering the synchronization loss in black). G. Oscillations in the total citrate concentration throughout the cell cycle (mean and s.d. of n = 5; experimentally measured concentration dynamics shown in red, the deconvoluted signal in green, the expected concentration dynamics considering the synchronization loss in black, and the measured concentration in non-synchronized cells in blue). H. The measured lactate secretion flux in synchronized cells shown in red (mean and s.d. of n = 5, fi(t) in equation 9), the deconvoluted secretion flux dynamics, in case of no synchronization loss (green, in equation 10), and the expected secretion flux based on the deconvoluted fluxes and considering the loss in synchronization, matching the measured fluxes (black, equation 10). Download figure Download PowerPoint Figure 4. Oscillations in isotopic labeling of TCA cycle metabolites throughout the cell cycle from [U-13C]-glutamine show induced oxidative and reductive glutamine metabolism in S phase A, B. Oscillations in aspartate (A) and malate (B) concentrations throughout the cell cycle when feeding [U-13C]-glutamine (representing oxidative TCA cycle activity). C. Uniform malate m + 4 labeling throughout the cell cycle (combined with the increase in malate concentration in S phase representing increased oxidative TCA cycle flux in S phase). D. Oscillations in pyrimidines m + 3 labeling throughout the cell cycle when feeding [U-13C]-glutamine (representing de novo pyrimidine biosynthesis). E. Oscillations in lactate m + 3 labeling throughout the cell cycle when feeding [U-13C]-glutamine (representing malic enzyme activity). F. Oscillations in malate m + 3 throughout the cell cycle when feeding [U-13C]-glucose (representing pyruvate carboxylase activity). G. Oscillations in citrate m + 5 when feeding [U-13C]-glutamine throughout the cell cycle (representing reductive IDH flux). H. Oscillations in acetyl-CoA m + 2 when feeding [U-13C]-glucose, representing oxidative glucose metabolism. I. Oscillations in acetyl-CoA m + 2 when feeding [U-13C]-glutamine, representing reductive glutamine metabolism. Data information: (A, B) Measured metabolite concentrations are shown in red (showing mean and s.d. of n = 5), the deconvoluted signal in green, the expected concentration dynamics considering the synchronization loss in black, and the measured concentrations in non-synchronized cells in blue. (C–I) Measured fractional isotopic labeling are shown in red (showing mean and s.d. of n = 3), the deconvoluted signal in green, and the expected isotopic labeling dynamics considering the synchronization loss in black. Download figure Download PowerPoint The inferred oscillations in metabolite isotopic labeling and concentrations were used to computationally analyze metabolic flux dynamics throughout the cell cycle, utilizing a variant of kinetic flux profiling (KFP; Yuan et al, 2008; Fig 5, in units of nmole/μl-cells/h, i.e., mM/h, Materials and Methods). Specifically, given a metabolite whose isotopic labeling dynamics throughout the cell cycle was inferred as explained above, we search for the most likely transient production and consumption fluxes in each 1-h interval through the cell cycle, such that the simulated labeling kinetics of this metabolite (within the 1-h interval) would optimally match the experimental measurements (Materials and Methods, Appendix Figs S3–S8). The simulation of the isotopic labeling kinetics of a metabolite of interest within a 1-h time interval is performed via an ordinary differential equations (ODE) model, relying on the inferred concentration of the metabolite within this time interval (considering that a metabolite with a larger pool size would take more time to label, per unit of flux), as well as the isotopic labeling kinetics of intermediates that produce this metabolite. While KFP is typically applied to estimate fluxes under metabolic steady state (in which fluxes satisfy a stoichiometric mass-balance constraint), here, we constrain the difference between transient fluxes that produce and consume a certain metabolite according to the measured momentary change in the concentration of that metabolite. Figure 5. Complementary oscillations of glucose versus glutamine-derived fluxes in TCA cycle A–I. Oscillations in metabolic flux throughout the cell cycle (in mM/h), computed based on metabolic modeling of measured oscillations in metabolite concentrations and isotopic labeling (red and green marks represent optimal estimates of transient flux with 95% CI). Blue lines represent average fluxes inferred in a non-synchronized cell population. As shown, glucose-derived flux into TCA cycle peaks in late G1 phase, while oxidative and reductive glutamine metabolism dominates S phase. Download figure Download PowerPoint Oscillations in the isotopic labeling pattern of TCA cycle intermediates when feeding isotopic glucose suggest that glucose-derived flux into TCA cycle increases in G1 phase and then drops in S phase. The fractional labeling of the m + 2 form of the TCA cycle intermediates citrate, α-ketoglutarate, and malate drops in S phase (Fig 3C–E). Feeding cells with isotopic glutamine, we further observed a drop in citrate m + 4 produced from oxaloacetate via citrate synthase in S phase (Fig 3F). Combined with the drop in citrate concentration during S phase (Fig 3G), metabolic modeling reveals a ~2-fold decrease in glycolytic flux into TCA cycle as cells progress through S phase; citrate synthase flux drops from ~6 mM/h in G1/S phase to ~3 mM/h in late S phase (Fig 5A, Appendix Fig S3). TCA cycle oxidation of citrate via isocitrate dehydrogenase (IDH) shows similar flux dynamics, with ~2-fold drop in S phase (Fig 5C, Appendix Fig S4). To examine whether the increase in glucose-derived flux into TCA cycle in G1/S is associated with increased in glycolytic flux, we measured lactate concentrations in the culture media in the synchronized cell population followed by computational deconvolution (Fig 3H, Materials and Methods). We find a ~65% increase in lactate secretion rate in G1/S transition. Considering that the average lactate secretion rate throughout the cell cycle is two orders of magnitude higher than that of pathways that branch out from glycolysis, the observed oscillation in lactate secretion represents cell cycle-dependent changes in glycolytic flux: The average lactate secretion throughout the cell cycle is ~600 mM/h, while oxidative pentose-phosphate pathway (PPP) is ~3 mM/h, reductive PPP is ~3 mM/h, glycogenesis is ~0.3 mM/h, and serine biosynthesis is below 1 mM/h (Appendix Fig S8). Overall, our data show that the increase in glycolytic flux in G1/S phase co-occurs with the increased glucose-driven flux entering the TCA cycle. Notably, analyzing oscillations in glycolytic flux based on direct measurement of changes in glucose consumption throughout the cell cycle (rather than based on lactate secretion) was not possible due to technical difficulty in accurately quantifying glucose consumption by synchronized cells within 3-h time intervals (considering that the synchronized cell population consumes ~1% of the glucose in media within this short time period). Induced oxidative and reductive glutamine metabolism compensates for the decreased glycolytic flux into TCA cycle in S phase Glutamine feeds TCA cycle flux by producing glutamate, which is converted to α-ketoglutarate either via transamination or by glutamate dehydrogenase. As a first estimation of the cell cycle dynamics of glutamine-derived flux into the TCA cycle, we quantified cell cycle-dependent glutamate production from glutamine versus glutamate secretion to the media. Tracing the m + 5 labeling dynamics of glutamate when feeding [U-13C]-glutamine and glutamate concentration throughout the cell cycle suggests that glutamate production flux increases by 25% in S phase compared to G1 (Fig 5D). This is evident by a marked increase in glutamate concentration in S phase and similar m + 5 glutamate labeling kinetics throughout the cell cycle (Appendix Fig S4). Glutamate secretion rate to the culture medium shows a marked drop in S phase, suggesting increased availability of glutamate for feeding the TCA cycle flux in S phase (Fig 5D). The increased entry of glutamine-derived flux into the TCA cycle in S phase is followed by a ~40% increase in α-ketoglutarate oxidation (Fig 5F, Appendix Fig S5). This is evident by the marked increase in malate and aspartate concentration in S phase (Fig 4A and B) and barely altered m + 4 and m + 3 labeling kinetics of these metabolites throughout the cell cycle, respectively (Fig 4C and Appendix Fig S9). The increased glutamine-derived anaplerotic flux into the TCA cycle in S phase (via net production of the TCA cycle intermediate α-ketoglutarate) is balanced by oscillations in cataplerotic fluxes, consuming TCA cycle intermediates for biosynthetic and bioenergetics purposes: We find a ~70% increase in pyrimidine biosynthesis flux in S phase, consuming oxaloacetate from TCA cycle (transaminated to produce aspartate, Fig 5G). This is evident by a marked increase in the m + 3 labeling of pyrimidines in S phase (Fig 4D and Appendix Fig S10), while considering the oscillations in the labeling kinetics of carbamoyl-aspartate in pyrimidine biosynthesis and pyrimidine concentrations (Appendix Fig S6). Oscillations in the biosynthetic flux of pyrimidines as well as purines is further supported by an increased m + 5 labeling of pyrimidines and purines in S phase (i.e., having all five ribose carbons labeled) upon feeding with isotopic glucose (Dataset EV2). The malic enzyme flux (decarboxylating malate dehydrogenases) further shows a marked ~65% increase in S phase (Fig 5H, Appendix Fig S7), as evident by the increased lactate m + 3 labeling in S phase when feeding isotopic glutamine (Fig 4E). Consistently, an increased concentration of lactate m + 3 in the culture media is further observed in S phase (Appendix Fig S11). Notably, while glutamine-derived anaplerotic flux increases in S phase, there is no major change in glucose-derived anaplerotic flux through pyruvate carboxylase in S phase (Fig 5I, Appendix Fig S5). This is evident by the drop in malate and aspartate m + 3 in S phase when feeding isotopic glucose (Fig 4F and Appendix Fig S12) occurring while the concentration of malate and aspartate increases (Fig 4A and B). While the increased glutamine-derived flux into the TCA cycle in S phase supports an increase in α-ketoglutarate oxidation, we further detect a surprisingly high ~55% increase in the rate of α-ketoglutarate reduction in early S phase (Fig 5B, Appendix Fig S3). This is evident by the marked increase in m + 5 citrate in S phase when feeding isotopic glutamine (Fig 4G). Considering the major drop in glycolytic flux into the TCA cycle in S phase, the relative contribution of reductive IDH to citrate production increases from ~15% in G1 to ~24% in S phase. Cell cycle os
Referência(s)