Advertisement

Ultrahigh resolution MS1/MS2-based reconstruction of metabolic networks in mammalian cells reveals changes for selenite and arsenite action

  • Teresa W.-M. Fan
    Correspondence
    For correspondence: Teresa W.-M. Fan
    Affiliations
    Center for Environmental and Systems Biochemistry (CESB), University of Kentucky, Lexington, Kentucky, USA

    Department of Toxicology and Cancer Biology, University of Kentucky, Lexington, Kentucky, USA

    Markey Cancer Center, University of Kentucky, Lexington, Kentucky, USA
    Search for articles by this author
  • Qiushi Sun
    Affiliations
    Center for Environmental and Systems Biochemistry (CESB), University of Kentucky, Lexington, Kentucky, USA
    Search for articles by this author
  • Richard M. Higashi
    Affiliations
    Center for Environmental and Systems Biochemistry (CESB), University of Kentucky, Lexington, Kentucky, USA

    Department of Toxicology and Cancer Biology, University of Kentucky, Lexington, Kentucky, USA

    Markey Cancer Center, University of Kentucky, Lexington, Kentucky, USA
    Search for articles by this author
Open AccessPublished:October 09, 2022DOI:https://doi.org/10.1016/j.jbc.2022.102586
      Metabolic networks are complex, intersecting, and composed of numerous enzyme-catalyzed biochemical reactions that transfer various molecular moieties among metabolites. Thus, robust reconstruction of metabolic networks requires metabolite moieties to be tracked, which cannot be readily achieved with mass spectrometry (MS) alone. We previously developed an Ion Chromatography-ultrahigh resolution-MS1/data independent-MS2 method to track the simultaneous incorporation of the heavy isotopes 13C and 15N into the moieties of purine/pyrimidine nucleotides in mammalian cells. Ultrahigh resolution-MS1 resolves and counts multiple tracer atoms in intact metabolites, while data independent-tandem MS (MS2) determines isotopic enrichment in their moieties without concern for the numerous mass isotopologue source ions to be fragmented. Together, they enabled rigorous MS-based reconstruction of metabolic networks at specific enzyme levels. We have expanded this approach to trace the labeled atom fate of [13C6]-glucose in 3D A549 spheroids in response to the anticancer agent selenite and that of [13C5,15N2]-glutamine in 2D BEAS-2B cells in response to arsenite transformation. We deduced altered activities of specific enzymes in the Krebs cycle, pentose phosphate pathway, gluconeogenesis, and UDP-GlcNAc synthesis pathways elicited by the stressors. These metabolic details help elucidate the resistance mechanism of 3D versus 2D A549 cultures to selenite and metabolic reprogramming that can mediate the transformation of BEAS-2B cells by arsenite.

      Keywords

      Abbreviations:

      BAsT (arsenite-transformed BEAS-2B cells), F6P (fructose-6-phosphate), GOT (glutamic-oxaloacetic transaminase), HBP (hexosamine biosynthesis pathway), IC-UHR-MS1/DI-MS2 (ion chromatography-ultrahigh resolution-MS1/data independent-MS2), ME (malic enzyme), MS (mass spectrometry), PC (pyruvate carboxylase), PDH (pyruvate dehydrogenase), PPP (pentose phosphate pathway), R5P (ribose-5-phosphate), Ru5P (ribulose-5-phosphate), S7P (sedoheptulose-7-phosphate), SIRM (stable isotope–resolved metabolomics), TALDO (transaldolase), TKT (transketolase)
      Metabolomics has been instrumental in accelerating the elucidation of metabolic reprogramming induced by disease states or drug treatment (
      • Schug Z.T.
      • Vande Voorde J.
      • Gottlieb E.
      The metabolic fate of acetate in cancer.
      ,
      • Sellers K.
      • Fox M.P.
      • Bousamra M.
      • Slone II, S.P.
      • Higashi R.M.
      • Miller D.M.
      • et al.
      Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
      ,
      • Fan T.W.
      • Higashi R.M.
      • Song H.
      • Daneshmandi S.
      • Mahan A.L.
      • Purdom M.S.
      • et al.
      Innate immune activation by checkpoint inhibition in human patient-derived lung cancer tissues.
      ,
      • van Deventer C.A.
      • Lindeque J.Z.
      • van Rensburg P.J.
      • Malan L.
      • van der Westhuizen F.H.
      • Louw R.
      Use of metabolomics to elucidate the metabolic perturbation associated with hypertension in a black South African male cohort: the SABPA study.
      ) and the discovery of metabolism-based biomarkers (
      • Zhang A.H.
      • Sun H.
      • Wang X.J.
      Recent advances in metabolomics in neurological disease, and future perspectives.
      ,
      • Trushina E.
      • Mielke M.M.
      Recent advances in the application of metabolomics to Alzheimer's disease.
      ,
      • Lindahl A.
      • Forshed J.
      • Nordstrom A.
      Overlap in serum metabolic profiles between non-related diseases: implications for LC-MS metabolomics biomarker discovery.
      ). As metabolite levels are governed by many factors including rates of synthesis and degradation, multiple input and output pathways, and exchange across compartments (
      • Fan T.W.
      • Lorkiewicz P.K.
      • Sellers K.
      • Moseley H.N.
      • Higashi R.M.
      • Lane A.N.
      Stable isotope-resolved metabolomics and applications for drug development.
      ), it has been challenging to reconstruct metabolic networks based on total metabolite profiles alone. Metabolic transformations through the network require numerous enzyme-catalyzed reactions that transfer the structural moiety among metabolites. Thus, the ability to track metabolite moiety will greatly reduce the ambiguities in metabolic network analysis. Stable isotope–resolved metabolomics (SIRM) fulfills this requirement by systematically tracking the transformations of individual tracer atoms from precursors to products using a combination of MS1 and NMR methods, which provides respectively the number and position of the tracer atoms in given metabolites. This approach has been successfully applied to determine altered metabolic activities by disease development and other perturbations in 2D/3D cell cultures (
      • Fan T.W.
      • Tan J.
      • McKinney M.M.
      • Lane A.N.
      Stable isotope resolved metabolomics analysis of ribonucleotide and RNA metabolism in human lung cancer cells.
      ,
      • Yang Y.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      Chloroformate derivatization for tracing the fate of amino acids in cells and tissues by multiple stable isotope resolved metabolomics (mSIRM).
      ,
      • Lane A.N.
      • Tan J.
      • Wang Y.
      • Yan J.
      • Higashi R.M.
      • Fan T.W.
      Probing the metabolic phenotype of breast cancer cells by multiple tracer stable isotope resolved metabolomics.
      ,
      • Fan T.W.
      • Yuan P.
      • Lane A.N.
      • Higashi R.M.
      • Wang Y.
      • Hamidi A.B.
      • et al.
      Stable isotope-resolved metabolomic analysis of lithium effects on glial-neuronal metabolism and interactions.
      ,
      • Fan T.
      • Bandura L.
      • Higashi R.
      • Lane A.
      Metabolomics-edited transcriptomics analysis of Se anticancer action in human lung cancer cells.
      ,
      • Ren J.G.
      • Seth P.
      • Clish C.B.
      • Lorkiewicz P.K.
      • Higashi R.M.
      • Lane A.N.
      • et al.
      Knockdown of malic enzyme 2 suppresses lung tumor growth, induces differentiation and impacts PI3K/AKT signaling.
      ,
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ), human tissues ex vivo (
      • Sellers K.
      • Fox M.P.
      • Bousamra M.
      • Slone II, S.P.
      • Higashi R.M.
      • Miller D.M.
      • et al.
      Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
      ,
      • Fan T.W.
      • Warmoes M.O.
      • Sun Q.
      • Song H.
      • Turchan-Cholewo J.
      • Martin J.T.
      • et al.
      Distinctly perturbed metabolic networks underlie differential tumor tissue damages induced by immune modulator beta-glucan in a two-case ex vivo non-small-cell lung cancer study.
      ,
      • Xie H.
      • Hanai J.I.
      • Ren J.G.
      • Kats L.
      • Burgess K.
      • Bhargava P.
      • et al.
      Targeting lactate dehydrogenase-A inhibits tumorigenesis and tumor progression in mouse models of lung cancer and impacts tumor-initiating cells.
      ), patient-derived xenograft mice in vivo (
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      • Yan J.
      Stable isotope resolved metabolomics of lung cancer in a SCID mouse model.
      ,
      • Sun R.C.
      • Fan T.W.
      • Deng P.
      • Higashi R.M.
      • Lane A.N.
      • Le A.T.
      • et al.
      Noninvasive liquid diet delivery of stable isotopes into mouse models for deep metabolic network tracing.
      ), and even human subjects in vivo (
      • Sellers K.
      • Fox M.P.
      • Bousamra M.
      • Slone II, S.P.
      • Higashi R.M.
      • Miller D.M.
      • et al.
      Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
      ,
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      • Farag M.A.
      • Gao H.
      • Bousamra M.
      • et al.
      Altered regulation of metabolic pathways in human lung cancer discerned by 13C stable isotope-resolved metabolomics (SIRM)).
      ).
      However, compared with mass spectrometry (MS), the moderate sensitivity of NMR limits the overall metabolite coverage. This limitation prompted us to develop an Ion Chromatography-Ultrahigh Resolution-MS1/data independent-MS2 (IC-UHR-MS1/DI-MS2) method to enable determination of tracer atom position(s) in metabolite moiety by MS with higher resolution and sensitivity than NMR. This in turn allows robust reconstruction of metabolic network responses to stressors at specific enzyme levels (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ). The UHR-MS1 step is capable of resolving the neutron mass difference among different tracer atoms (e.g., Δmass = 0.006995 amu between 13C and 15N) (
      • Yang Y.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      Chloroformate derivatization for tracing the fate of amino acids in cells and tissues by multiple stable isotope resolved metabolomics (mSIRM).
      ,
      • Higashi R.M.
      • Fan T.W.
      • Lorkiewicz P.K.
      • Moseley H.N.
      • Lane A.N.
      Stable isotope-labeled tracers for metabolic pathway elucidation by GC-MS and FT-MS.
      ). This capability enables multiplexing of biologically compatible tracer atoms such as 13C, 15N, and 2H in the same (e.g., [13C5,15N2]-Gln) or different substrates (e.g., [13C6]-glucose + [15N2]-Gln) to expand the metabolic pathway coverage while circumventing sample batch effects in multiplex SIRM studies (
      • Yang Y.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      Chloroformate derivatization for tracing the fate of amino acids in cells and tissues by multiple stable isotope resolved metabolomics (mSIRM).
      ,
      • Le A.
      • Lane A.N.
      • Hamaker M.
      • Bose S.
      • Gouw A.
      • Barbi J.
      • et al.
      Glucose-independent glutamine metabolism via TCA cycling for proliferation and survival in B cells.
      ).
      We have expanded the pathway reconstruction of purine/pyrimidine nucleotide synthesis to the reconstruction of metabolic networks consisting of the Krebs cycle, pentose phosphate pathway (PPP), gluconeogenesis, and UDP-GlcNAc synthesis pathways in 3D A549 spheroids and arsenite-transformed BEAS-2B cells. By tracing [13C6]-glucose or [13C5,15N2]-Gln transformations into the moiety of these pathway metabolites, we were able to deduce changes in specific enzyme activities induced by selenite in A549 spheroids or by arsenite in BEAS-2B cells. This information enabled us to surmise the resistance mechanism of 3D versus 2D A549 cultures to selenite and metabolic reprogramming that presumably mediates the transformation of BEAS-2B cells by arsenite.

      Results

      Isotope enrichment distributions of major metabolites from glycolysis, the Krebs cycle, PPP, gluconeogenesis, and UDP-GlcNAc metabolism were obtained from the UHR-MS1 and MS2 spectra in both [13C6]-glucose–traced A549 spheroids ± anticancer selenite treatment and [13C5,15N2]-Gln–traced BEAS-2B cells compared with arsenite transformated BEAS-2B cells. Example MS1 (A) and MS2 (B) spectra are shown for citrate in Fig. S1. Isotopologue concentrations were calculated from the peak area ratio of samples to calibration standard mixtures after natural abundance correction, followed by normalization to the sample protein concentration.

      The Krebs cycle

      The glycolytic product of [13C6]-Glc (13C3-pyruvate) enters the Krebs cycle either via 13C2-acetyl CoA produced from the pyruvate dehydrogenase (PDH) reaction or directly into 13C3-oxaloacetate via pyruvate carboxylase (PC) activity. After the first turn, the PDH-initiated Krebs cycle produces 13C2-isotopologues () of various intermediates, whereas PCB-initiated Krebs cycle generates 13C3-isotopologues () of citrate, cis-aconitate, malate, fumarate, and aspartate (
      • Sellers K.
      • Fox M.P.
      • Bousamra M.
      • Slone II, S.P.
      • Higashi R.M.
      • Miller D.M.
      • et al.
      Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
      ), and the malic enzyme (ME) reaction scrambles 13C in pyruvate leading to the synthesis of 13C1-metabolites () (Figs. S2A and 1A). It should be noted that this pathway scheme takes into account unlabeled carbon (●) that can come from preexisting pools of free metabolites as well as their precursors such as glycogen, proteins, and lipids.
      Figure thumbnail gr1
      Figure 113C and 15N isotopologue analysis of IC-UHR-MS1 and MS2 data shows blocked Krebs cycle by selenite in A549 spheroids and by arsenite transformation in BEAS-2B cells. A, A549 spheroids. B, BEAS-2B cells. Polar extracts were analyzed by IC-UHR-MS1 and DI-MS2. 13C and 15N atoms were traced from [13C6]-Glc (A) or [13C5,15N2]-Gln (B) into the Krebs cycle metabolites after first and second turns (enclosed in brackets). Due to space limitation, not all possible labeled products are shown. ●: 12C; : 14N; : 15N; , , : 13C from the first turn of the PDH, PC, and ME-mediated Krebs cycle reactions, respectively. The X-axis refers to the number of 13C and/or 15N atoms in each isotopologue. The Y-axis represents μmole or ion intensity normalized to amounts of total protein. Data shown are mean ± SEM (n = 3) for A549 spheroids and mean ± SEM (n = 2) for BEAS-2B cells. The boxes are color-coded to denote the contribution of the GLS (red)/GOT (blue) in (B), PDH (red) in (A), and PC- (green) and ME-mediated (light blue) Krebs cycle reactions in (A) and (B) to given isotopologues of metabolites. αKG, α-ketoglutarate; AcCoA, acetyl CoA; ACO, aconitase; GOT, glutamate-oxaloacetate transaminase; GLS, glutaminase; GLUD1, glutamate dehydrogenase; GSH, glutathione; IDH, isocitrate dehydrogenase; ME, malic enzyme; OAA, oxaloacetate; OGDH, α-ketoglutarate dehydrogenase; PC, pyruvate carboxylase; PDH, pyruvate dehydrogenase. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.005; ∗∗∗∗p < 0.001.
      In the [13C6]-Glc–traced A549 spheroids, we saw the occurrence of 13C2- (2, red box) and 13C3-citrate (3, green box), which are the respective products of PDH-initiated (canonical) and anaplerotic PC-initiated Krebs cycle (Figs. 1A-b and S1A). The presence of the 13C3-3,4,5-citrate species (3) in the MS2 data also points to PC activity (Figs. 1A-c and S1B). It is evident from the UHR-MS1 data that 13C2-citrate accumulated more than 13C3-citrate, indicating a higher activity of PDH-initiated than anaplerotic PC-initiated Krebs cycle. However, 13C2-malate (f) and -Asp (g) were comparable in levels to the 13C3-counterparts (Fig. 1A). This discrepancy can be accounted for by the contribution of a second turn canonical Krebs cycle activity to the 13C3 pools, which is consistent with the synthesis of 13C4-citrate (b), a specific product of the second turn. Although low in levels, 13C1-citrate and -Glu (i) were present, suggesting contribution from the ME reaction. Selenite induced the depletion of all 13C2-, 13C3-, and 13C1-isotopologues of the Krebs cycle intermediates in A549 spheroids, except for αKG (d), which showed enhanced buildup. These data are consistent respectively with inhibition of PDH, PC, and ME-mediated Krebs cycle activity, particularly at the α-ketoglutarate dehydrogenase (OGDH) step by selenite leading to the accumulation of all 13C-isotopologues of αKG. The 13C-labeling patterns of the MS2 fragments verified the selenite effect on PDH (2 or 13C2-1,2-Asp, h; 3 or 13C3-Glu-GSH, k) and PC (3 or 13C3-1,2,3-Asp, h) activity (Fig. 1A) while revealing inhibition of GSH synthesis by blocking the PDH-initiated Krebs cycle activity and Ser→Gly synthesis pathways (cf., Fig. S3). The latter is evidenced by the depletion of 13C3-Glu (k) and 13C2-Gly (l) moiety of GSH. This information could not be ascertained based on the MS1 data of GSH (j) alone (Fig. 1A).
      In [13C5,15N2]-Gln–traced BEAS-2B cells, the labeled Gln enters the Krebs cycle by first conversion to 13C5, 15N1-Glu (a) via glutaminase-catalyzed glutaminolysis and then to 13C5-αKG (b) via glutamic-oxaloacetic transaminase (GOT)-catalyzed transamination and/or glutamate dehydrogenase 1–catalyzed oxidative deamination. 13C5-αKG is further transformed to 13C4-succinate (d), -fumarate (e), -malate (f), and -citrate (h) via the Krebs cycle (Figs. S2B and 1B). 13C4-malate can be converted to 13C3-pyruvate (l) via the ME reaction, leading to the synthesis of 13C2- and 13C3-citrate, -succinate, -fumarate, -malate, and -Asp via, respectively, PDH- and PC-initiated Krebs cycle activities. Moreover, 13Cx,15N-Asp (j) can be produced via GOT-catalyzed transamination while 13Cx,15N-GSH (m) is synthesized from 13Cx,15N-Glu. Such pathway reconstruction was deduced from the presence of all expected 13C and 13C,15N-isotopologues of the glutaminolytic and Krebs cycle products based on the MS1 and MS2 data. Arsenite transformed cells (BAsT) showed depletion of all of these products except for the labeled GSH in terms of both Glu (n) and Gly (o) moieties (Fig. 1B). These data pointed to inhibition of the glutaminase and/or Krebs cycle activity but activation of GSH synthesis in BAsT versus control cells.
      In addition, detailed analysis of the 13C- and/or 15N-labeling patterns of both the parent metabolites (molecular ions in MS1) and fragments (in MS2) revealed differential arsenite effects on individual enzyme reactions. For example, the first two products of glutaminase (i.e., 13C5,15N-Glu in a and 13C5-αKG in b) showed arsenite-induced depletion, which suggests glutaminase inhibition by arsenite. However, from the MS2 data, we saw 13C3 (3)- and 13C4 (4)-C1 to C5-citrate (i) accumulated while the product 13C3 (3)- and 13C4 (4)-C1 to C4-αKG depleted (c), which points to additional block at the aconitase and/or isocitrate dehydrogenase steps. The former is consistent with the known inhibition of aconitase by arsenite (
      • Mohamed A.H.
      • Anderson L.E.
      Light activation of purified aconitase by washed thylakoid membranes of pea (pisum sativum L.).
      ). If this were the only effect of arsenite, we would expect the same trend for the MS1 data for citrate (h), which was not the case. The production of these fragments had a contribution from the ME (, light blue box) and/or PC (, green box) in addition to the glutaminase (, red box)-mediated pathways. The observed discrepancy between MS1 and MS2 data could be attributed to the confounding activation of the ME and PC-mediated pathways by arsenite, leading to the accumulation of the three citrate fragments. This interpretation could also apply to the discrepancy between MS1 (f) and MS2 (g) data of malate. The accumulation of 13C4-succinate (d) and depletion of the products 13C4-fumarate (e) are consistent with the inhibition of succinate dehydrogenase (SDH) based on the MS1 data, which was reported previously (
      • Jizhong C.
      • Huiqiong W.
      • Ruikun S.
      • Yongmu H.
      Effect of in vivo and in vitro treatment with arsenite on rat hepatic mitochondrial and microsomal enzymes.
      ). Moreover, the arsenite-induced accumulation in the 13C5, 15N1-Glu (n), and 13C2, 15N1-Gly moieties (o) of GSH argue for the activation of the GSH synthesis pathway while that in 15N1-Glu suggests enhanced GOT activity in addition. The former is consistent with arsenite-induced GSH accumulation and activation of GSH synthesis genes reported for lung epithelial cells (
      • Jizhong C.
      • Huiqiong W.
      • Ruikun S.
      • Yongmu H.
      Effect of in vivo and in vitro treatment with arsenite on rat hepatic mitochondrial and microsomal enzymes.
      ). Thus, by combining the MS1 and MS2 data, it is practical to translate changes in the complex 13C- and 15N-labeling patterns of the Krebs cycle metabolites into altered activity of specific enzymes, which would not be reliable based on either MS1 or MS2 data alone.

      The PPP and gluconeogenesis

      The PPP is a major route for glucose oxidation to produce ribose-5-phosphate (R5P) and NADPH, which are respectively the precursor to nucleotide synthesis and reductant for anabolic and antioxidant metabolism. In this pathway, [13C6]-Glc is converted to ribulose-5-phosphate (Ru5P) via hexokinase, G6P dehydrogenase, and 6-phosphogluconate dehydrogenase, which is then isomerized to R5P (oxidative branch) and epimerized to xylulose-5-phosphate, followed by the transketolase (TKT) and transaldolase (TALDO) reactions to respectively produce sedoheptulose-7-phosphate (S7P) + glyceraldehyde-3-phosphate and fructose-6-phosphate (F6P) and erythrose-4-phosphate (freely reversible nonoxidative branch), respectively (Fig. 2A). In [13C6]-Glc–traced A549 cells, we saw domination of fully 13C-labeled isotopologues of G6P (a), 6PG (b), Ru5P/R5P (c), and S7P (d) in the MS1 data (Fig. 2A). For S7P, the 13C2- and 13C5-isotopologues were also present and at higher levels than the 13C1- (absent) and 13C3-isotopologues. Based on the TKT and TALDO reaction mechanism (denoted by green arrows), the former two species can be produced directly by the forward TKT reaction and the latter two species by the reverse TALDO reaction. Thus, the observed scrambled 13C-labeling patterns of S7P is consistent with higher forward or oxidative PPP than reverse or nonoxidative PPP activity. Selenite treatment enhanced the levels of 13C2- and 13C5-S7P while reducing those of 13C1- and 13C3-S7P (d), which suggests a shift from nonoxidative to NADPH-generating oxidative PPP. This is consistent with the lack of depletion of 13C-6PG (b) and 13C-R5P+Ru5P (c) despite the large depletion of G6P (a) by selenite. Interestingly, selenite induced depletion of 13C5- and 13C6-F6P (e) but buildup of the 13C3-4,5,6 fragment of F6P (f). Together with the accumulation of 13C-labeled S7P, the former points to inhibition of TALDO activity by selenite while the latter could be attributed to enhanced gluconeogenesis by selenite (cf., Fig. S3).
      Figure thumbnail gr2
      Figure 213C isotopologue analysis of IC-UHR-MS1 and MS2 data shows enhanced oxidative PPP in response to selenite in A549 spheroids or to arsenite transformation in BEAS-2B cells. A, A549 spheroids. B, BEAS-2B cells. 13C atoms were traced from [13C6]-Glc (A) or [13C5,15N2]-Gln (B) into the PPP and gluconeogenic products. Brackets in (A) denote 13C products of the reverse transaldolase (TALDO) reaction, while green curves and arrows delineate the recombination of R5P moiety with X5P (TKT reaction) or F6P moiety with E4P (TALDO reaction) to generate sedoheptulose-7-phosphate (S7P). Due to space limitation, not all possible labeled products are shown. The same sets of extracts as in were analyzed by IC-UHR-MS1 and DI-MS2. The X-/Y-axes and number of replicates are as in . 6PG, 6-phosphogluconate; E4P, erythrose-4-phosphate; G6PD/PGD, glucose-6-phosphate/6-phosphogluconate dehydrogenase; GPI, glucose-6-phosphate isomerase; PPP, pentose phosphate pathway; R5P, ribose-5-phosphate; Ru5P, ribulose-5-phosphate; S7P, sedoheptulose-7-phosphate; TK, transketolase; X5P, xylulose 5-phosphate; all other abbreviations and symbols are as in . ∗, p < 0.05; ∗∗, p < 0.01.
      In [13C5,15N2]-Gln–traced BEAS-2B cells, very low levels of 13C incorporation were evident in some of the PPP products and their 13C scrambling patterns presumably resulted from a combination of gluconeogenic, TKT, and TALDO activities (Fig. 2B). The fully 13C-labeled isotopologues of G6P (a), R5P+Ru5P (c), and F6P (e) as well as 13C1-6PG (b) accumulated more in BAsT than control cells. Although most of these changes were at the detection limit and nonstatistically significant, they could reflect enhanced oxidative PPP activity in BAsT cells (cf., Fig. 2A). This would generate more NADPH to support reduction of GSSG to GSH (cf., Fig. 1B) for relieving oxidative stress induced by arsenite (
      • Jizhong C.
      • Huiqiong W.
      • Ruikun S.
      • Yongmu H.
      Effect of in vivo and in vitro treatment with arsenite on rat hepatic mitochondrial and microsomal enzymes.
      ).

      UDP-GlcNAc biosynthesis pathway

      UDP-GlcNAc is an activated form of GlcNAc needed for O- and N-linked protein glycosylation, which are important in regulating numerous cellular processes, such as protein targeting to organelles (
      • Sellers K.
      • Fox M.P.
      • Bousamra II, M.
      • Slone S.P.
      • Higashi R.M.
      • Miller D.M.
      • et al.
      Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
      ) and nutrient sensing (
      • Taylor R.P.
      • Geisler T.S.
      • Chambers J.H.
      • McClain D.A.
      Up-regulation of O-GlcNAc transferase with glucose deprivation in HepG2 cells is mediated by decreased hexosamine pathway flux.
      ,
      • Hanover J.A.
      • Krause M.W.
      • Love D.C.
      The hexosamine signaling pathway: O-GlcNAc cycling in feast or famine.
      ). UDP-GlcNAc has four biochemical moieties (Fig. S4) that are derived from several intersecting metabolic pathways (
      • Moseley H.N.
      • Lane A.N.
      • Belshoff A.C.
      • Higashi R.M.
      • Fan T.W.
      A novel deconvolution method for modeling UDP-N-acetyl-D-glucosamine biosynthetic pathways based on (13)C mass isotopologue profiles under non-steady-state conditions.
      ) (Fig. 3). The hexosamine moiety comes from glucose and the amido N of Gln via the hexosamine biosynthesis pathway (HBP), the acetyl group is donated from acetyl CoA generated from glucose, amino acids, or fatty acids, the ribose unit derives from glucose via the PPP, and the uracil ring is produced from pyrimidine biosynthesis using C and N sources such as glucose and Gln.
      Figure thumbnail gr3
      Figure 3Altered 13C and/or 15N incorporation into UTP/UDP-GlcNAc and their moieties in response to selenite in A549 spheroids or to arsenite transformation in BEAS-2B cells. A, A549 Spheroids. B, BEAS-2B cells. 13C and/or 15N atoms were traced from [13C6]-Glc (A) or [13C5,15N2]-Gln (B) into UDP-GlcNAc. ●: 12C; : 14N; : 15N; , , : 13C from the first turn of the PDH, PCB, and ME-mediated Krebs cycle reactions, respectively. The same sets of extracts as in were analyzed by IC-UHR-MS1 and DI-MS2. A-a, B-a and A-d, B-d: determined from MS1 of UTP/UDP-GlcNAc; A-b to c/f and B-b to c/f: determined from the MS2 of the ribose and uracil moieties of UTP and UDP-GlcNAc in A549 spheroid and BEAS-2B cells, respectively; A-e and B-e: determined from the MS2 of the GlcNAc moiety of UDP-GlcNAc in A549 spheroid and BEAS-2B cells, respectively. The X-/Y-axes and number of replicates are as in . Ac, acetyl; CP, carbamoyl phosphate; GlcNAc1 or 6P, N-acetylglucosamine 1 or 6-phosphate; GLS, glutaminase; HBP, hexosamine biosynthesis pathway; ME, malic enzyme; OMP, orotidine monophosphate; PC, pyruvate carboxylase; PDH, pyruvate dehydrogenase; Pyr, pyruvate. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.005; ∗∗∗∗p < 0.001; ∗∗∗∗∗p < 0.0005.
      From the UHR-MS1 spectra of UDP-GlcNAc, high intensity of the 13C6-, 13C8-, 13C11-, and 13C13-peaks were observed in [13C6]-Glc–traced A549 cells (Fig. 3A-d). The ambiguities in the labeled unit assignment for these isotopologues were resolved with the DI-MS2 data. We observed low enrichment of 13C1-3 (1–3) peaks in the uracil fragment of UDP-GlcNAc (Fig. 3A-f), which was akin to the corresponding pattern of the precursor UTP (Fig. 3A-c). In contrast, the glucosamine plus acetyl fragment showed high enrichment of the 13C6 (6) and 13C8 (8) species (Fig. 3A-e), as the case for the two in the MS1 data (Fig. 3A-d). These two species can be confidently assigned to 13C6-glucosamine- and 13C6-glucosamine- + 13C2-acetyl–bearing UDP-GlcNAc, respectively. Although we did not directly observe relevant fragments, we can justifiably assign two other abundant isotopologues (13C11 and 13C13) to, respectively, 13C6-glucosamine + 13C5-ribose- and 13C6-glucosamine- + 13C5-ribose + 13C2-acetyl–bearing UDP-GlcNAc, based on the prevalence of the 13C6-glucosamine and 13C6-glucosamine + 13C2-acetyl moieties (Fig. 3A-e) as well as 13C5-ribose in the UTP precursor (Fig. 3A-b). Selenite treatment enhanced the enrichment of the 13C6-GlcNAc fragment of UDP-GlcNAc (e) but reduced that of the 13C8-GlcNAc (e) and 13C1-3-uracil fragments of UDP-GlcNAc (f) as well as the 13C8, 11-15 isotopologues of intact UDP-GlcNAc (d) (Fig. 3A). These data are consistent with the block in the uracil synthesis plus reduced synthesis and/or incorporation of ribose into UTP and UDP-GlcNAc, as either or both processes are required for the synthesis of the 13C11-15-isotopologues. They also point to the maintenance of glucosamine synthesis but reduced acetyl incorporation into GlcNAc via the HBP. Again, such detailed deduction of selenite’s effect on the UDP-GlcNAc biosynthetic pathway would not be feasible without the combined MS1 and MS2 data.
      In [13C5,15N2]-Gln–traced BEAS-2B cells, UHR-MS1 data of UDP-GlcNAc showed isotopologues with single (15N) and dual (13C,15 N) tracer atoms (Fig. 3B-d). Together with the MS2 fragment data, the two most abundant 13C3,15N2 (C3N2) and 13C3,15N3 (C3N3) species in the MS1 data mainly consisted of 13C3,15N1- and 13C3,15N2-uracil (e) plus 15N1-glucosamine units (f), respectively, with minor contribution of the 13C1,15N1-uracil plus 13C2,15N1-glucosamine unit. This is consistent with the prominence of the 13C3,15N1- and 13C3,15N2-uracil fragment in the UTP precursor (b). As illustrated in the atom-resolved pathway scheme, these two most abundant species should be derived from the reaction sequence of glutaminase—first turn of the Krebs cycle (PDH, PC, ME-mediated)—pyrimidine synthesis. The MS2 fragment of GlcNAc showed dominant enrichment of the 15N1-species with some enrichment of the 13C2,15N1-species (f) (Fig. 3B). These data indicate high activity of HBP along with ME-mediated Krebs cycle reactions giving rise to 13C2-acetyl CoA for acetyl transfer to glucosamine (cf., Fig. S4). Arsenite transformation depleted the low-abundance 15N and 13C,15N-isotopologues of UDP-GlcNAc (d) and its precursor UTP (a), which primarily resulted from reduced 15N and/or 13C incorporation into the GlcNAc and uracil moieties (b,e,f) since little 13C enrichment was evident in the ribose unit of UTP (c) (Fig. 3B). Thus, chronic exposure to arsenite blocked both HBP and uracil biosynthesis in BEAS-2B cells. However, the enrichment of the most abundant 13C3,15N2-uracil fragment of UDP-GlcNAc and its precursor UTP was enhanced by arsenite (b,e). This species can be derived from 13C4,15N1-Asp (with loss of 13CO2) + 15N-carbamoyl phosphate (15N-CP). As the 13C3,15N1-uracil fragment of both UTP and UDP-GlcNAc was reduced in enrichment, it is plausible that enhanced enrichment of 13C3,15N2-uracil–bearing UDP-GlcNAc is driven by the formation and/or incorporation of 15N-CP at the expense of the 13C3,15N1 species in arsenite-transformed BEAS-2B cells. Such detailed deduction of pathway changes is made possible by the use of the dual tracer in combination with the ability to resolve label positions in UDP-GlcNAc moieties by the DI-MS2 method.

      Discussion

      We have applied a previously developed Ion chromatography-ultrahigh resolution Fourier transform MS1/DI-MS2 method (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ) for extensive and robust reconstruction of [13C6]-Glc or [13C5, 15N2]-Gln–fueled central metabolic networks in mammalian cells. This method met the needs for resolving dual tracer distribution in intact metabolites with ultra high-resolution MS1 while simultaneously acquiring positional labeling in metabolite moieties via DI-MS2. In this report, we illustrated how to rigorously reconstruct the Krebs cycle, PPP, gluconeogenesis, and UDP-GlcNAc synthesis pathway by utilizing the combination of UHR-MS1 with MS2 data. This approach enabled us to unambiguously discern in-cell–altered activity of specific enzymes induced by anticancer selenite treatment in lung adenocarcinoma A549 spheroids or by arsenite transformation in lung epithelial BEAS-2B cells.
      For A549 spheroids, we found that selenite’s ability to attenuate the Krebs cycle activity lies in the blockade of enzymes both in the canonical (OGDH) and anaplerotic (PC, ME) pathways (Fig. 1A). This is consistent with the suppression of the OGDH gene and PC protein but contrary to the overexpression of the ME gene in the 2D counterparts reported previously (
      • Fan T.
      • Bandura L.
      • Higashi R.
      • Lane A.
      Metabolomics-edited transcriptomics analysis of Se anticancer action in human lung cancer cells.
      ,
      • Bruntz R.C.
      • Belshoff A.C.
      • Zhang Y.
      • Macedo J.K.A.
      • Higashi R.M.
      • Lane A.N.
      • et al.
      Inhibition of anaplerotic glutaminolysis underlies selenite toxicity in human lung cancer.
      ). Another notable distinction of selenite’s effect is less inhibition of GSH synthesis in 3D (Fig. 1A) versus 2D A549 cells (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ), which should contribute to a better capacity of the spheroid culture for antioxidation. Our present data points to reduced synthesis (i.e., blocked GOT), rather than attenuated incorporation, of the precursor Glu as the cause for selenite’s inhibition of GSH synthesis in A549 spheroids. This is reasoned from the depletion of 13C-labeled Glu despite the buildup of its 13C-labeled αKG precursor. As for PPP, selenite-induced shift to the oxidative branch is expected to produce more NADPH to better sustain the reduction of GSSG to GSH, which is used to alleviate oxidative stress by detoxifying reactive oxygen species (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ). This shift can also maintain R5P production despite the block of the TALDO activity in the nonoxidative branch (Fig. 2A). These changes of the GSH and R5P synthesis pathways in 3D A549 spheroids presumably contribute to their better resistance to selenite toxicity than the 2D counterpart, as observed previously (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ). In addition, our combined MS1 and MS2 data revealed that subsequent R5P incorporation into UTP and the supply of acetyl CoA and/or its entry into HBP was blocked by selenite, leading to attenuated synthesis of UDP-GlcNAc. This, together with somewhat compromised Krebs cycle, could underlie the growth inhibition of A549 spheroids with prolonged selenite treatment (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ).
      Arsenite is known to impact various metabolic proteins that contain the sulfhydryl group (
      • Shen S.
      • Li X.-F.
      • Cullen W.R.
      • Weinfeld M.
      • Le X.C.
      Arsenic binding to proteins.
      ) (e.g., IκB kinase and glucose transporter) leading to different disease states including cancer (
      • Kapahi P.
      • Takahashi T.
      • Natoli G.
      • Adams S.R.
      • Chen Y.
      • Tsien R.Y.
      • et al.
      Inhibition of NF-kappa B activation by arsenite through reaction with a critical cysteine in the activation loop of I kappa B kinase.
      ,
      • Singh A.P.
      • Goel R.K.
      • Kaur T.
      Mechanisms pertaining to arsenic toxicity.
      ). However, the details of metabolic reprogramming in transformed epithelial cells induced by chronic, low-dose exposure to arsenite are still elusive. Our MS1- and MS2-based metabolic network reconstruction revealed the complex action of arsenite on the Krebs cycle, PPP, and antioxidation pathways in lung epithelial BEAS-2B cells, including blockade of aconitase, isocitrate dehydrogenase, SDH, and glutaminase but activation of ME/PC, GOT, and GSH synthesis activities. One important outcome of these reprogrammed events can be reactive oxygen species buildup but not in excess to avoid apoptosis while driving different carcinogenic events (
      • Singh A.P.
      • Goel R.K.
      • Kaur T.
      Mechanisms pertaining to arsenic toxicity.
      ). Moreover, despite the block of HBP and overall uracil synthesis, arsenite-transformed BEAS-2B cells largely maintained UDP-GlcNAc production by activating the CP synthesis and/or incorporation steps of the UDP-GlcNAc synthesis pathway. UDP-GlcNAc is the required substrate for O-GlcNAcylation of several oncogenic regulators that drive cancer development (
      • Hanover J.A.
      • Chen W.
      • Bond M.R.
      O-GlcNAc in cancer: an oncometabolism-fueled vicious cycle.
      ) and the maintenance of this oncometabolite pool is expected to be important to arsenite transformation of BEAS-2B cells.
      In conclusion, we applied an IC-UHR-MS1/DI-MS2 method to track changes in 13C/15N-labeling patterns of metabolites and their moieties in SIRM studies of A549 spheroids or BEAS-2B cells in response to selenite or arsenite transformation, respectively. This approach enabled robust reconstruction of the metabolic network consisting of the Krebs cycle, PPP, gluconeogenesis, and UDP-GlcNAc synthesis pathway to discern specific enzyme activities in the network altered by the treatments. In turn, this information helps elucidate the resistance mechanism of 3D versus 2D A549 cultures to selenite and metabolic reprogramming that can mediate the transformation of BEAS-2B cells by arsenite.

      Experimental procedures

      Materials

      All materials including the make-up solvent methanol for Ion chromatography, individual standards of metabolites used for quantification were obtained as described previously (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ).

      Preparation of calibration standard mixtures

      A mixture of 86 (Mix 1) and 81 (Mix 2) standards were prepared as two separate calibration standard mixtures as described previously (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ). The standard mixtures were aliquoted, lyophilized, and stored at −80 °C for long term use. When needed, lyophilized Mix 1 was dissolved in 120 μl 18 MΩ water, vortexed, and 50 μl was used to reconstitute with Mix 2 to form the final calibration standard mixture.

      IC-UHR-MS1 and DI-MS2

      Ion chromatography-ultrahigh resolution fourier transform MS

      Metabolites were separated on an IonPac AG11-HC-4 μm guard column (2 × 50 mm) coupled to an IonPac AS11-HC-4 μm RFIC&HPIC (2 × 250 mm) analytical column in a Dionex ICS5000+ system (Thermo Scientific) equipped with a dual pump, an eluent generator, an autosampler, and a detector/chromatography module. Conditions for chromatographic separations (i.e., KOH gradient) and ion suppressor and desolvation in the heated electrospray were as described previously (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ). MS data were acquired using the Xcalibur software. A batch of samples started with a 15 min blank (water) injection to check for contamination in the instrument, followed by two injections of calibration standard mixtures to ensure the stability of MS signals and another 15 min water injection to check for carryover on the IC column. Lyophilized cell extracts were freshly reconstituted in 20 μl 18 MΩ water plus 1 μM DSS (sodium trimethylsilylpropanesulfonate) and run in a random order. Each sample was followed by one or two 15 min injections of water blank to minimize carryover. The calibration standard mixture was run after every 6 to 8 cell extracts to track signal loss in the same batch of run. Each sample batch ended with an injection of the calibration standard mixture, followed by water to double check the normality of MS signals and sample carryover.

      DI-MS2 measurement for cell polar extracts

      DI-MS2 analysis was performed in between full MS1 scans for quantifying targeted fragment(s) of major metabolites in polar extracts, as described previously (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ). To achieve this, we set (1) the cycle time of no more than 2 to 3 s for acquiring 10 to 15 points across each chromatographic peak for reliable quantification of precursors and their isotopologues; (2) sufficient resolving power in full scan (500,000) and MS2 (60,000) modes to discriminate 13C from 15N-containing isotopologues of precursors and fragments; and (3) full isotopologue coverage for each metabolite in selecting the precursor mass range for MS2 scan (i.e., 280–440 with the isolation window of 200 m/z). Other conditions were as described previously (
      • Sun Q.
      • Fan T.W.
      • Lane A.N.
      • Higashi R.M.
      An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
      ).

      Data analysis and quantification

      We first established an in-house exact mass database for the precursors and fragment products based on the corresponding mass ion spectra acquired for individual metabolite standards. Several public metabolomics databases, including the Human Metabolome DataBase (
      • Wishart D.S.
      • Jewison T.
      • Guo A.C.
      • Wilson M.
      • Knox C.
      • Liu Y.
      • et al.
      HMDB 3.0--the human metabolome database in 2013.
      ), the Kyoto Encyclopedia of Genes and Genomes (
      • Kanehisa M.
      • Goto S.
      • Sato Y.
      • Kawashima M.
      • Furumichi M.
      • Tanabe M.
      Data, information, knowledge and principle: back to metabolism in KEGG.
      ), and METLIN (
      • Tautenhahn R.
      • Cho K.
      • Uritboonthai W.
      • Zhu Z.
      • Patti G.J.
      • Siuzdak G.
      An accelerated workflow for untargeted metabolomics using the METLIN database.
      ), and Mass Frontier were used to help interpret MS2 data for metabolite fragmentation patterns. This database was then incorporated into TraceFinder v3.3 (Thermo Scientific) for assigning and integrating the peak areas of precursor ions in MS1 spectra and fragment ions in MS2 spectra of targeted metabolites in cell extracts for further quantification. Precursors and fragments were assigned with mass accuracy set to 5 ppm. Assignments were curated before isotopic peak areas were corrected for natural abundance as previously described (
      • Moseley H.N.
      Correcting for the effects of natural abundance in stable isotope resolved metabolomics experiments involving ultra-high resolution mass spectrometry.
      ). Metabolites in samples were quantified from the corrected MS1 data by calibrating against the two calibration standard mixtures run before (Std 1) and after (Std 2) the samples. The response factor was calculated for each sandwiched sample run as follows:
      Response factor = (Area [Std 1] + (Area [Std 2] – Area [Std 1]) × nth run number/run number))/std concentration. The metabolite concentration was then calculated by dividing the corrected MS1 peak area with the response factor and normalized against the extract aliquot and amount of total protein. The fragment peak areas were similarly normalized.

      Preparation of 13C-labeled polar extracts of 3D A549 spheroids ± selenite

      A549 cells were grown to 90% confluence in 10-cm plates, followed by loading with magnetic nanoparticles (Nanoshuttle, N3D Biosciences) overnight at 37 °C/5% CO2, as described previously (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ). Cells were then detached and seeded into 6-well Costar-cell repellent plates (Corning, Inc) at 400,000 cells/well for spheroid formation. Spheroids were cultured for 4 days before medium change to [13C6]-Glc ± 10 μM Na2SeO3 and grown at 37 °C/5% CO2 for 24 h. Spheroids were harvested, rinsed twice with cold PBS, and then briefly with cold nanopure water before simultaneous quenching and extraction of polar metabolites in cold 70% methanol (
      • Fan T.
      • El-Amouri S.
      • Macedo J.
      • Wang Q.
      • Song H.
      • Cassel T.
      • et al.
      Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
      ). One-eighth of the polar fraction was aliquoted and lyophilized for IC-UHR-MS1/DI-MS2 analysis.

      Preparation of 13C-, 15N-labeled polar extracts of 2D BEAS-2B cells ± arsenite transformation

      Primary bronchial epithelial BEAS-2B cells (ATCC) were cultured under two conditions: (1) in Bronchial Epithelial Cell Growth Medium (BEGM, Lonza Corporation) as control; (2) in BEGM + 1 μM Na2AsO3 in 10-cm plates as transformed cells (BAsT). Cells were grown to 60 to 70% confluence before passaging to generate over 24 weeks. At week 24, 4 mM [13C5,15N2]-Gln was introduced to both groups and grown at 37 °C/5% CO2 for 24 h. Cells were then quenched with cold acetonitrile and extracted for polar metabolites in acetonitrile/water/chloroform (V/V 2:1.5:1) as described previously (
      • Fan T.W.
      • Warmoes M.O.
      • Sun Q.
      • Song H.
      • Turchan-Cholewo J.
      • Martin J.T.
      • et al.
      Distinctly perturbed metabolic networks underlie differential tumor tissue damages induced by immune modulator beta-glucan in a two-case ex vivo non-small-cell lung cancer study.
      ,
      • Fan T.W.-M.
      Sample preparation for metabolomics investigation.
      ). One-eighth of the polar fraction was aliquoted and lyophilized for IC-UHR-MS1/DI-MS2 analysis.

      Data availability

      All data acquired are available upon request.

      Supporting information

      This article contains supporting information (
      • Lloyd S.J.
      • Lauble H.
      • Prasad G.S.
      • Stout C.D.
      The mechanism of aconitase: 1.8 A resolution crystal structure of the S642a:citrate complex.
      ,
      • Sun Q.
      • Fan T.W.M.
      • Lane A.N.
      • Higashi R.M.
      Applications of chromatography- ultra high-resolution MS for stable isotope-resolved metabolomics (SIRM) reconstruction ofmetabolic networks.
      ).

      Conflict of interest

      The authors declare that they have no conflicts of interest with the contents of the article.

      Acknowledgments

      This work was supported by NCI P01CA163223-01A1 , 1U24DK097215-01A1 , 1R01CA118434-01A2 , 5R21ES025669-02 , 5P20GM121327 , 5P30ES026529 , and Shared Resource(s) of the University of Kentucky Markey Cancer Center P30CA177558 . We thank Dr Salim EI-Amouri for assistance in the A549 spheroid tracer experiment and polar extraction and Ms Yan Zhang for the BEAS-2B cells tracer experiment and polar extraction. We also thank Dr Marc O. Warmoes, Patrick Shepherd, and Travis Thompson for developing the TraceFinder curation method and R scripts for automatic natural abundance correction, quantification, and normalization and Dr A. Lane for comments on the article.

      Author contributions

      T. W.-M. F. and R. M. H. conceptualization; T. W.-M. F., Q. S., and R. M. H. methodology; T. W.-M. F., Q. S., and R. M. H. investigation; Q. S. and R. M. H. formal analysis; T. W.-M. F. writing–original draft; Q. S. and R. M. H. writing–review and editing; T. W.-M. F. and R. M. H. funding acquisition.

      Supporting information

      References

        • Schug Z.T.
        • Vande Voorde J.
        • Gottlieb E.
        The metabolic fate of acetate in cancer.
        Nat. Rev. Cancer. 2016; 16: 708-717
        • Sellers K.
        • Fox M.P.
        • Bousamra M.
        • Slone II, S.P.
        • Higashi R.M.
        • Miller D.M.
        • et al.
        Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
        J. Clin. Invest. 2015; 125: 687-698
        • Fan T.W.
        • Higashi R.M.
        • Song H.
        • Daneshmandi S.
        • Mahan A.L.
        • Purdom M.S.
        • et al.
        Innate immune activation by checkpoint inhibition in human patient-derived lung cancer tissues.
        Elife. 2021; 10e69578
        • van Deventer C.A.
        • Lindeque J.Z.
        • van Rensburg P.J.
        • Malan L.
        • van der Westhuizen F.H.
        • Louw R.
        Use of metabolomics to elucidate the metabolic perturbation associated with hypertension in a black South African male cohort: the SABPA study.
        J. Am. Soc. Hypertens. 2015; 9: 104-114
        • Zhang A.H.
        • Sun H.
        • Wang X.J.
        Recent advances in metabolomics in neurological disease, and future perspectives.
        Anal. Bioanal. Chem. 2013; 405: 8143-8150
        • Trushina E.
        • Mielke M.M.
        Recent advances in the application of metabolomics to Alzheimer's disease.
        Biochim. Biophys. Acta. 2014; 1842: 1232-1239
        • Lindahl A.
        • Forshed J.
        • Nordstrom A.
        Overlap in serum metabolic profiles between non-related diseases: implications for LC-MS metabolomics biomarker discovery.
        Biochem. Biophys. Res. Commun. 2016; 478: 1472-1477
        • Fan T.W.
        • Lorkiewicz P.K.
        • Sellers K.
        • Moseley H.N.
        • Higashi R.M.
        • Lane A.N.
        Stable isotope-resolved metabolomics and applications for drug development.
        Pharmacol. Ther. 2012; 133: 366-391
        • Fan T.W.
        • Tan J.
        • McKinney M.M.
        • Lane A.N.
        Stable isotope resolved metabolomics analysis of ribonucleotide and RNA metabolism in human lung cancer cells.
        Metabolomics. 2012; 8: 517-527
        • Yang Y.
        • Fan T.W.
        • Lane A.N.
        • Higashi R.M.
        Chloroformate derivatization for tracing the fate of amino acids in cells and tissues by multiple stable isotope resolved metabolomics (mSIRM).
        Anal. Chim. Acta. 2017; 976: 63-73
        • Lane A.N.
        • Tan J.
        • Wang Y.
        • Yan J.
        • Higashi R.M.
        • Fan T.W.
        Probing the metabolic phenotype of breast cancer cells by multiple tracer stable isotope resolved metabolomics.
        Metab. Eng. 2017; 43: 125-136
        • Fan T.W.
        • Yuan P.
        • Lane A.N.
        • Higashi R.M.
        • Wang Y.
        • Hamidi A.B.
        • et al.
        Stable isotope-resolved metabolomic analysis of lithium effects on glial-neuronal metabolism and interactions.
        Metabolomics. 2010; 6: 165-179
        • Fan T.
        • Bandura L.
        • Higashi R.
        • Lane A.
        Metabolomics-edited transcriptomics analysis of Se anticancer action in human lung cancer cells.
        Metabolomics. 2005; 1: 325-339
        • Ren J.G.
        • Seth P.
        • Clish C.B.
        • Lorkiewicz P.K.
        • Higashi R.M.
        • Lane A.N.
        • et al.
        Knockdown of malic enzyme 2 suppresses lung tumor growth, induces differentiation and impacts PI3K/AKT signaling.
        Sci. Rep. 2014; 4: 5414
        • Fan T.
        • El-Amouri S.
        • Macedo J.
        • Wang Q.
        • Song H.
        • Cassel T.
        • et al.
        Stable isotope-resolved metabolomics shows metabolic resistance to anti-cancer selenite in 3D spheroids versus 2D cell cultures.
        Metabolites. 2018; 8: 40
        • Fan T.W.
        • Warmoes M.O.
        • Sun Q.
        • Song H.
        • Turchan-Cholewo J.
        • Martin J.T.
        • et al.
        Distinctly perturbed metabolic networks underlie differential tumor tissue damages induced by immune modulator beta-glucan in a two-case ex vivo non-small-cell lung cancer study.
        Cold Spring Harb. Mol. Case Stud. 2016; 2: a000893
        • Xie H.
        • Hanai J.I.
        • Ren J.G.
        • Kats L.
        • Burgess K.
        • Bhargava P.
        • et al.
        Targeting lactate dehydrogenase-A inhibits tumorigenesis and tumor progression in mouse models of lung cancer and impacts tumor-initiating cells.
        Cell Metab. 2014; 19: 1-15
        • Fan T.W.
        • Lane A.N.
        • Higashi R.M.
        • Yan J.
        Stable isotope resolved metabolomics of lung cancer in a SCID mouse model.
        Metabolomics. 2011; 7: 257-269
        • Sun R.C.
        • Fan T.W.
        • Deng P.
        • Higashi R.M.
        • Lane A.N.
        • Le A.T.
        • et al.
        Noninvasive liquid diet delivery of stable isotopes into mouse models for deep metabolic network tracing.
        Nat. Commun. 2017; 8: 1646
        • Fan T.W.
        • Lane A.N.
        • Higashi R.M.
        • Farag M.A.
        • Gao H.
        • Bousamra M.
        • et al.
        Altered regulation of metabolic pathways in human lung cancer discerned by 13C stable isotope-resolved metabolomics (SIRM)).
        Mol. Cancer. 2009; 8: 41
        • Sun Q.
        • Fan T.W.
        • Lane A.N.
        • Higashi R.M.
        An ion chromatography-ultrahigh-resolution-MS(1)/data-independent high-resolution MS(2) method for stable isotope-resolved metabolomics reconstruction of central metabolic networks.
        Anal. Chem. 2021; 93: 2749-2757
        • Higashi R.M.
        • Fan T.W.
        • Lorkiewicz P.K.
        • Moseley H.N.
        • Lane A.N.
        Stable isotope-labeled tracers for metabolic pathway elucidation by GC-MS and FT-MS.
        Methods Mol. Biol. 2014; 1198: 147-167
        • Le A.
        • Lane A.N.
        • Hamaker M.
        • Bose S.
        • Gouw A.
        • Barbi J.
        • et al.
        Glucose-independent glutamine metabolism via TCA cycling for proliferation and survival in B cells.
        Cell Metab. 2012; 15: 110-121
        • Mohamed A.H.
        • Anderson L.E.
        Light activation of purified aconitase by washed thylakoid membranes of pea (pisum sativum L.).
        Plant Physiol. 1983; 71: 248-250
        • Jizhong C.
        • Huiqiong W.
        • Ruikun S.
        • Yongmu H.
        Effect of in vivo and in vitro treatment with arsenite on rat hepatic mitochondrial and microsomal enzymes.
        J. Tongji Med. Univ. 1996; 16: 155-159
        • Sellers K.
        • Fox M.P.
        • Bousamra II, M.
        • Slone S.P.
        • Higashi R.M.
        • Miller D.M.
        • et al.
        Pyruvate carboxylase is critical for non–small-cell lung cancer proliferation.
        J. Clin Invest. 2015; 125: 687-698
        • Taylor R.P.
        • Geisler T.S.
        • Chambers J.H.
        • McClain D.A.
        Up-regulation of O-GlcNAc transferase with glucose deprivation in HepG2 cells is mediated by decreased hexosamine pathway flux.
        J. Biol. Chem. 2009; 284: 3425-3432
        • Hanover J.A.
        • Krause M.W.
        • Love D.C.
        The hexosamine signaling pathway: O-GlcNAc cycling in feast or famine.
        Biochim. Biophys. Acta. 2010; 1800: 80-95
        • Moseley H.N.
        • Lane A.N.
        • Belshoff A.C.
        • Higashi R.M.
        • Fan T.W.
        A novel deconvolution method for modeling UDP-N-acetyl-D-glucosamine biosynthetic pathways based on (13)C mass isotopologue profiles under non-steady-state conditions.
        BMC Biol. 2011; 9: 37
        • Bruntz R.C.
        • Belshoff A.C.
        • Zhang Y.
        • Macedo J.K.A.
        • Higashi R.M.
        • Lane A.N.
        • et al.
        Inhibition of anaplerotic glutaminolysis underlies selenite toxicity in human lung cancer.
        Proteomics. 2019; 19e1800486
        • Shen S.
        • Li X.-F.
        • Cullen W.R.
        • Weinfeld M.
        • Le X.C.
        Arsenic binding to proteins.
        Chem. Rev. 2013; 113: 7769-7792
        • Kapahi P.
        • Takahashi T.
        • Natoli G.
        • Adams S.R.
        • Chen Y.
        • Tsien R.Y.
        • et al.
        Inhibition of NF-kappa B activation by arsenite through reaction with a critical cysteine in the activation loop of I kappa B kinase.
        J. Biol. Chem. 2000; 275: 36062-36066
        • Singh A.P.
        • Goel R.K.
        • Kaur T.
        Mechanisms pertaining to arsenic toxicity.
        Toxicol. Int. 2011; 18: 87-93
        • Hanover J.A.
        • Chen W.
        • Bond M.R.
        O-GlcNAc in cancer: an oncometabolism-fueled vicious cycle.
        J. Bioenerg. Biomembr. 2018; 50: 155-173
        • Wishart D.S.
        • Jewison T.
        • Guo A.C.
        • Wilson M.
        • Knox C.
        • Liu Y.
        • et al.
        HMDB 3.0--the human metabolome database in 2013.
        Nucleic Acids Res. 2013; 41: D801-D807
        • Kanehisa M.
        • Goto S.
        • Sato Y.
        • Kawashima M.
        • Furumichi M.
        • Tanabe M.
        Data, information, knowledge and principle: back to metabolism in KEGG.
        Nucleic Acids Res. 2014; 42: D199-D205
        • Tautenhahn R.
        • Cho K.
        • Uritboonthai W.
        • Zhu Z.
        • Patti G.J.
        • Siuzdak G.
        An accelerated workflow for untargeted metabolomics using the METLIN database.
        Nat. Biotechnol. 2012; 30: 826-828
        • Moseley H.N.
        Correcting for the effects of natural abundance in stable isotope resolved metabolomics experiments involving ultra-high resolution mass spectrometry.
        BMC Bioinformatics. 2010; 11: 1-6
        • Fan T.W.-M.
        Sample preparation for metabolomics investigation.
        in: Fan T.W.-M. Lane A.N. Higashi R.M. The Handbook of Metabolomics: Pathway and Flux Analysis, Methods in Pharmacology and Toxicology. Springer Science, New York, NY2012: 7-27
        • Lloyd S.J.
        • Lauble H.
        • Prasad G.S.
        • Stout C.D.
        The mechanism of aconitase: 1.8 A resolution crystal structure of the S642a:citrate complex.
        Protein Sci. 1999; 8: 2655-2662
        • Sun Q.
        • Fan T.W.M.
        • Lane A.N.
        • Higashi R.M.
        Applications of chromatography- ultra high-resolution MS for stable isotope-resolved metabolomics (SIRM) reconstruction ofmetabolic networks.
        Trends Analyt. Chem. 2020; 123: 115676