Abstract

   The growth and development of non-small cell lung cancer (NSCLC)
   primarily depends on glutamine. Both glutamine and essential amino
   acids (EAAs) have been reported to upregulate mTOR in NSCLC, which is a
   bioenergetics sensor involved in the regulation of cell growth, cell
   survival, and protein synthesis. Seen as novel concepts in cancer
   development, ASCT2 and LAT transporters allow glutamine and EAAs to
   enter proliferating tumors as well as send a regulatory signal to mTOR.
   Blocking or downregulating these glutamine transporters in order to
   inhibit glutamine uptake would be an excellent therapeutic target for
   treatment of NSCLC. This study aimed to validate the metabolic
   dysregulation of glutamine and its derivatives in NSCLC using cellular
   1H-NMR metabolomic approach while exploring the mechanism of
   delta-tocotrienol (δT) on glutamine transporters, and mTOR pathway.
   Cellular metabolomics analysis showed significant inhibition in the
   uptake of glutamine, its derivatives glutamate and glutathione, and
   some EAAs in both cell lines with δT treatment. Inhibition of glutamine
   transporters (ASCT2 and LAT1) and mTOR pathway proteins (P-mTOR and
   p-4EBP1) was evident in Western blot analysis in a dose-dependent
   manner. Our findings suggest that δT inhibits glutamine transporters,
   thus inhibiting glutamine uptake into proliferating cells, which
   results in the inhibition of cell proliferation and induction of
   apoptosis via downregulation of the mTOR pathway.

   Keywords: cancer, mTOR, vitamin E, SLC1A5, tocotrienols, apoptosis,
   cell growth, cell transporters, essential amino acids, ASCT2,
   glutaminolysis, alanine, glutathione, glutamate, lung, bio actives,
   nutraceuticals

1. Introduction

   Non-small cell lung cancer (NSCLC) presents itself as aggressive tumors
   arise from the airway epithelial cells (majority) and interior parts of
   the lungs [[28]1]. It remains one of the leading causes of
   disease-related mortalities in the world. The current therapeutic
   options for NSCLC, which include surgery, radiotherapy, and
   chemotherapy [[29]1], have slightly improved NSCLC survival rate at
   some developmental stages in both men and women. However, there has
   been a plateauing of the overall five-year survival rate, hovering
   ~12–18% between the years 1975 and 2011 [[30]2]. Also, several studies
   report that there is a high probability of reoccurrence and development
   of resistance to drug therapies in NSCLC after treatment with
   chemotherapeutic agents, surgical resection, and radiation therapy
   [[31]3]. This warrants efforts to identify novel therapeutic agents and
   targets for preventing and treating NSCLC.

   Research in nutrition-based modulation against diseases has opened up
   new horizons in cancer prevention, contributing to drug discovery and
   development processes for numerous chronic diseases, including cancer
   [[32]4,[33]5]. Most bioactive agents extracted from plants show minimum
   cell cytotoxicity while simultaneously targeting multiple signaling
   pathways involved in cell growth, apoptosis, invasion, angiogenesis,
   and metastasis in cancer cells [[34]6,[35]7]. Tocotrienols (α, β, γ,
   and δ), isomers of vitamin E, are found in vegetable oils, including
   rice bran oil and palm oil, wheat germ, barley, annatto, and certain
   other types of seeds, nuts, and grains [[36]8]. They exert biological
   effects including antiangiogenesis, antioxidant activities, and
   anticancer activities [[37]9,[38]10]. Our previous studies clearly
   demonstrated that delta-tocotrienol (δT) inhibits the proliferation and
   metastatic/invasion potential while concurrently inducing apoptosis in
   NSCLC cells, in a dose-dependent manner [[39]11]. We also identified
   some of the probable molecular targets of δT treatments on NSCLC
   [[40]11,[41]12,[42]13]. Therefore, δT is multitargeted and can be
   considered a valuable potential approach to further investigate for
   treatment of NSCLC.

   Metabolomics, a novel, versatile, and comprehensive approach, can
   provide unbiased information about metabolite concentrations, altered
   signaling pathways, and their interactions. Most current cancer
   metabolomics studies focus on finding diagnostic biomarkers and
   understanding fundamental mechanisms in cancer [[43]14]. Nonetheless,
   this approach could also be used effectively for identifying the
   efficacy of treatments [[44]15]. The NSCLC metabolome is a potentially
   informative reflection of the impact of the disease and its dynamics
   which could lead to promising developments in cancer research, strongly
   geared toward the discovery of new biomarkers of disease onset,
   progression, and effects of treatment regimens. Given that cancer
   cells, including NSCLC, show aberrant energy metabolism
   [[45]16,[46]17], it is of interest to investigate the changes in energy
   metabolism in NSCLC cells upon δT treatment, utilizing the global
   advantage of the metabolomic approach [[47]18].

   Glutamine plays a role as an indirect energy source in NSCLC, which
   produces ATP through glutamine-driven oxidative phosphorylation
   [[48]19]. Extra consumption of glutamine in tumors is used for
   generating metabolic precursors for uncontrolled cell proliferation.
   These precursors include elevated levels of nucleic acids, lipids, and
   proteins for cell proliferation [[49]20], as well as increased GSH
   production for cell death resistance [[50]21]. Current literature
   provides further evidence that glutamine in cancer facilitates exchange
   of EAAs (essential amino acids) with glutamine into proliferating cells
   via glutamine transporters, which induces mTOR (mammalian target of
   rapamycin) activation in NSCLC and other types of cancer
   [[51]22,[52]23]. Activated mTOR then promotes protein translation and
   cell growth via activation of its downstream genes such as S6k1 and
   4EBP1 [[53]24]. Alanine, serine, cysteine-preferring transporter 2
   (ASCT2), also known as (SLC1A5), and bidirectional L-type amino acid
   transporter 1 (LAT1) are the two primary transporters for glutamine
   uptake [[54]25,[55]26]. LAT1 enables transport of the EAAs to improve
   cancer cell growth via mTOR-induced translations, and ASCT2 sustains
   the cytoplasmic amino acid pool to drive LAT1 function [[56]27]. This
   collaboration of ASCT2 and LAT1 reduce apoptosis and enhance the energy
   production and cell growth via net delivery of glutamine inside the
   cell [[57]27].

   A recent study reported that A549 and H1229 lung cancer cells show
   glutamine dependency, and that deprivation of glutamine inhibits cell
   growth [[58]28]. Decreases in glutamine uptake, cell cycle progression,
   and mTORC1 pathway after inhibition of ASCT2 functionality by chemicals
   or shRNA in vitro was observed in prostate and pancreatic cancer cell
   lines [[59]29]. Also, inhibition of LAT1 using BCH
   (2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid) in H1395 lung cancer
   cell line reduced the cellular leucine uptake and consequently
   inhibited mTOR pathway activity, which finally reduced cell
   proliferation and viability [[60]30]. Induction of apoptosis was also
   reported in hepatoma, hybridoma, leukemia, myeloma, and fibroblast
   cells after glutamine deprivation [[61]31,[62]32]. Our preliminary
   metabolomics studies showed that δT treatments inhibited glutamine
   levels in A549 and H1299 cells. Also, in our previous studies,
   induction of apoptosis and inhibition of cell growth was observed in
   A549 and H1299 cells in a dose-dependent manner after δT treatments
   [[63]11,[64]33,[65]34,[66]35,[67]36]. Therefore, the aim of this study
   was to verify the metabolic dysregulation of glutamine and its
   derivatives upon δT treatment while investigating the effect of δT on
   the expression of glutamine transporters (ASCT2 and LAT1) and the mTOR
   pathway.

2. Results

2.1. δT Changes Metabolite Profiles in A549 and H1299 Cells

   To investigate the changes in metabolism and metabolites with δT
   intervention, supervised OPLS-DA analysis was performed using NMR
   spectral data acquired from intracellular cell lysate. The OPLS-DA
   score plot of cellular NMR metabolic profile resulting from 30 µM δT
   treated and control cells lines are shown in [68]Figure 1A. The OPLS-DA
   score plot exhibited clear separation between control and treatment
   groups in A549 cells and H1299 cells with δT treatment; the high Q2 and
   R2 values indicate a considerable difference in the cellular metabolic
   profile of treated cells compared to control cells while validating the
   model that we used for OPLS-DA analysis.

Figure 1.

   [69]Figure 1

   [70]Figure 1
   [71]Open in a new tab

   OPLS-DA analysis of metabolome of lung cancer cell lines after treating
   with/without δT for 72 h. (A) OPLS-DA Scores plot based on the cellular
   metabolic profiling of lung cancer cell lines, namely A549 (Top) and
   H1299 (Bottom); the 30 µM treatment (Yellow) and control (green) were
   generated using SIMCA+ software; the results indicated that cellular
   metabolic profiling of lung cancer cell lines was significantly changed
   after δT treatment for 72 h. (B) The S-Line plots of OPLS-DA analysis
   of A549 (top) and H1299 (Bottom) from treatment (30 µM) and control (0
   µM) cells. The key metabolites that changed significantly are marked on
   the S-Line plot and include (1) leucine, (2) glutamine, (3) glutamate,
   (4) glutathione, (5) lactate, (6) taurine, and (7) formate.

   To identify the metabolites represented in the NMR spectral regions
   (bins) that varied significantly between control and treatment groups,
   the corresponding loading S-Line plot from the OPLS-DA model was
   generated. [72]Figure 1B shows a representative S-Line plot
   corresponding to the score plot of [73]Figure 1. These bin numbers were
   further analyzed to identify the significant metabolites (using
   Chenomx) that contributed to the separation of the control and
   treatment groups seen in the OPLS-DA model. Based on the analysis of
   S-Line plot bin numbers, the key bin numbers responsible for the
   differences could be attributed to glutamine, glutamate and
   glutathione, and some amino acids in both cell lines.

2.2. Quantification of Metabolites Reveals That δT Alters Glutamine
Metabolism

   Chenomx 7.6 Suite NMR software was used to probe the metabolome
   profiles in the treatment and control groups. 1H-NMR spectra provided
   information on over 45 metabolites (both cell lines), including amino
   acids, intermediates of the tricarboxylic acid cycle (TCA), energy
   molecules, and nucleic acid associated molecules ([74]Table 1).

Table 1.

   List of metabolite concentrations determined using NMR in A549 (A)
   H1299 (B) cells. p-values less than 0.05 were considered statistically
   significant for univariate analysis. Treatment column indicates the
   samples with the 30 µM treatment of δT. All the concentrations are
   reported in µM.

   (A)
   Metabolite Name Mean ± SD (Control) Mean ± SD (Treatment) p-Value Fold
   Changes Control/Trt
   Amino Acids
   Aspartate 102.3 ± 11.9 55.9 ± 4.7 0.0016 1.8
   Glutamate 80.8 ± 7.9 48.7 ± 4.7 0.0019 1.7
   Leucine 33.7 ± 4.1 17 ± 3.7 0.0030 2.0
   Glycine 33.1 ± 1.2 20.4 ± 4.2 0.0035 1.6
   Alanine 31 ± 1.4 19.8 ± 3.9 0.0045 1.6
   Glutamine 99.9 ± 6.7 64.7 ± 13.3 0.0073 1.5
   Histidine 54 ± 8.4 85.9 ± 31.3 0.0815 0.6
   Asparagine 116.9 ± 16.2 54.5 ± 13.1 0.0033 2.1
   Taurine 90.3 ± 19.9 78.2 ± 26.8 0.2822 1.2
   Valine 23.8 ± 1.4 21.6 ± 6.3 0.2878 1.1
   Tryptophan 81.3 ± 15 72.7 ± 28.7 0.3340 1.1
   Proline 51.9 ± 49.3 63.7 ± 25.7 0.3659 0.8
   Lysine 41.6 ± 22.8 37.2 ± 6.1 0.4075 1.1
   Isoleucine 31.5 ± 9.9 30.6 ± 7 0.4499 1.0
   Methionine 5.8 ± 5.3 5.5 ± 3.4 0.4653 1.1
   Arginine nd nd
   Intermediate of TCA Cycle and Energy Metabolism
   Lactate 138.5 ± 5.6 99.9 ± 3.6 0.0003 1.4
   2-Oxoglutarate 43.6 ± 3.3 29.3 ± 4.7 0.0061 1.5
   AMP 32.1 ± 5 45 ± 1.7 0.0063 0.7
   Glutaric acid monomethyl ester 17.8 ± 6.4 34 ± 2.8 0.0077 0.5
   Malate 90.2 ± 10.7 48.7 ± 10.3 0.0111 1.9
   Succinate 9.3 ± 2.6 5.2 ± 2.8 0.0645 1.8
   Glucose 119.1 ± 53.4 187.3 ± 63.7 0.1139 0.6
   ADP 47.8 ± 8.3 40.8 ± 4.8 0.1370 1.2
   Citrate 42.4 ± 3.8 35.6 ± 11.6 0.1959 1.2
   NADH 38.4 ± 3.5 43.4 ± 16 0.3040 0.9
   NADPH 47 ± 6.3 51.3 ± 12.5 0.3118 0.9
   ATP 42.2 ± 5.4 42.9 ± 11.3 0.4653 1.0
   Nucleic acid Associataed Metabolites
   Uracil 98 ± 14.1 60.1 ± 24 0.0387 1.6
   UDP-N-Acetylglucosamine 6.9 ± 2.1 3.9 ± 3.4 0.1266 1.8
   Other
   Glutathione 69.6 ± 2.1 41.7 ± 6.7 0.0011 1.7
   Citrulline 81.9 ± 5.1 63.9 ± 13 0.0438 1.3
   Cystine 81.4 ± 6.3 58.4 ± 19 0.0582 1.4
   N-Acetylglucosamine 21.9 ± 9.3 12.8 ± 5.2 0.1065 1.7
   Formate 294.3 ± 68.5 312.8 ± 8.9 0.3334 0.9
   Fumarate 25 ± 3.2 27.7 ± 5 0.2363 0.9
   [75]Open in a new tab

   (B)
   Metabolite Name Mean ± SD (Control) Mean ± SD (Treatment) p-Value Fold
   Changes Control/Trt
   Amino Acids
   Aspartate 105.5 ± 3.5 77.4 ± 4.3 0.0010 1.4
   Glutamate 80.1 ± 5.7 49.3 ± 6.2 0.0033 1.6
   Leucine 31.8 ± 1.3 18.3 ± 0.8 <0.0001 1.7
   Glycine 28.2 ± 4.7 18.1 ± 3.2 0.0561 1.6
   Alanine 28.8 ± 2.2 18.2 ± 2.3 0.0044 1.6
   Glutamine 75.3 ± 5.1 53.7 ± 8.4 0.0177 1.4
   Histidine ND ND
   Asparagine 105 ± 21 84 ± 23.3 0.1986 1.3
   Taurine ND ND
   Valine 28.8 ± 4.9 21.7 ± 5.6 0.1706 1.3
   Tryptophan 36.8 ± 2 17.8 ± 11.4 0.0401 2.1
   Proline 90.2 ± 39.3 74.3 ± 34.9 0.3453 1.2
   Lysine 38.8 ± 11.3 19.4 ± 7.1 0.0547 2
   Isoleucine 37.2 ± 4.9 23.8 ± 2.7 0.0138 1.6
   Methionine 8.7 ± 0.8 6.7 ± 1.9 0.1247 1.3
   Arginine 43.8 ± 2.7 28.4 ± 6.6 0.0189 1.5
   Intermediate of TCA Cycle and Energy Metabolism
   Lactate 125.8 ± 7.3 122 ± 15.4 0.3857 1
   2-Oxoglutarate 32.5 ± 7.9 17.2 ± 1.5 0.0272 1.9
   AMP 27.5 ± 0.2 13.7 ± 2 0.0003 2
   Glutaric acid monomethyl ester 27.4 ± 0 20.6 ± 7.4 1.3
   Malate 130.9 ± 7.8 84.7 ± 9 0.0027 1.5
   Succinate 13.9 ± 1.7 5.3 ± 3.8 0.0215 2.6
   Glucose 196.4 ± 50.1 147.1 ± 19.4 0.1324 1.3
   ADP 33.6 ± 5.1 14.9 ± 7.7 0.0227 2.3
   Citrate 35.2 ± 0.8 25.6 ± 4.3 0.0183 1.4
   NADH 65.3 ± 11.7 43.7 ± 30.7 0.2024 1.5
   NADPH 48.6 ± 11.1 38.1 ± 23.5 0.2996 1.3
   ATP 43.5 ± 7.8 22.2 ± 5.5 0.0171 2
   Nucleic acid Associated Metabolites
   Uracil 88.5 ± 11.9 40.2 ± 16.3 0.0139 2.2
   UDP-N-Acetylglucosamine ND
   Other
   Glutathione 42.3 ± 4.5 28 ± 6.5 0.0319 1.5
   Citrulline 65.4 ± 20.6 53.4 ± 25.4 0.3156 1.2
   Cystine 61 ± 7.2 26.3 ± 14.1 0.0338 2.3
   N-Acetylglucosamine
   Fumarate
   Formate 354.5 ± 90.9 346.7 ± 41 0.4585 1
   Tyrosine 12.9 ± 0.6 67.8 ± 9.1 0.0134 0.2
   [76]Open in a new tab

   The table shows the detailed results including p-values, mean and
   standard deviation from the t-test for the groups (with or without 30
   µM of δT treatment) tested. Among the metabolites that were
   significantly different in concentration in the δT treated vs. control
   cells, we identified several metabolites from the glutamine metabolism
   and related pathways that were significantly decreased (p < 0.05) in
   the treatment group as compared to controls. In addition, we found that
   metabolites such as leucine and some essential amino acids had
   significantly lower concentrations in both cell lines after δT
   treatment. These essential amino acids include isoleucine, leucine,
   lysine, methionine, and tryptophan. Moreover, the metabolites related
   to cell proliferation such as 2-oxoglutarate, citrate, succinate,
   malate, aspartame, ATP, ADP, NADPH, and uracil significantly decreased
   (p < 0.05) in the treatment group as compared to controls ([77]Table
   1).

   Heatmap analysis from MetaboAnalyst 3.0 revealed that A549 and H1299
   cell lysates had similar changing trends in metabolites of δT treated
   groups versus control ([78]Figure 2A), which suggests that the
   supplement of δT impacts both cell lines in a similar manner. At the
   same time, our heatmap results also revealed that control and treatment
   groups supplemented with δT were clustered into two major groups (Green
   and Red groups at the top of the Heatmap) which suggest clear
   separation in two groups with their metabolites and also validates the
   separation in OPLS-DA analysis. The random forest importance plot
   identified 15 metabolites key in classifying the data with aspartame,
   alanine, leucine, glutamate glutathione, and glutamine having the most
   influence on classification ([79]Figure 2B).

Figure 2.

   [80]Figure 2

   [81]Figure 2

   [82]Figure 2
   [83]Open in a new tab

   Hierarchical clustering analysis of δT-altered metabolites (Heatmap)
   and contribution of metabolites in A549 and H1299. The metabolites,
   quantified with Chenomx software analysis of NMR spectra of A549 and
   H1299 cells after incubating with or without δT for 72 h, were used to
   generate the heat map (A) using Metaboanalyst software. Each column
   represents a sample, and each row represents the expression profile of
   metabolites. Blue color represents a decrease, and red color an
   increase. The very top row with green color indicates the control
   samples and red color row indicates the samples with the 30 µM
   treatment of δT. Random Forest (B) showed in bottom graphs identifies
   the significant features. The features are ranked by the mean decrease
   in classification accuracy when they are permuted.

   To further comprehend the biological relevance of the identified
   metabolites from Chenomx analysis, we performed pathway analysis using
   MetaboAnalyst 3.0 software [[84]25]. Some of the key altered pathways
   identified from pathway analysis include lysine biosynthesis, purine
   metabolism, alanine, aspartate and glutamate metabolism, glutamine and
   glutamate metabolism, citrate cycle (TCA cycle), and pyruvate
   metabolism for both cell lines ([85]Figure 3A).

Figure 3.

   [86]Figure 3
   [87]Open in a new tab

   The most predominant altered metabolic pathways (A) and top 25
   metabolites correlated with glutamine (B). Summary of the altered
   metabolism pathways (A) after treating with/without δT for 72 h, as
   analyzed using MetaboAnalyst 3.0. The size and color of each circle was
   based on pathway impact value and p-value, respectively. Circles,
   larger and higher along the Y axis, show higher impact of pathway on
   the organism. The top 25 metabolites, correlating with glutamine level
   (B) after treating with/without δT for 72 h. X-axis shows maximum
   correlation; pink color shows positive correlation whereas blue shows
   negative correlation.

   As random forest importance plot and pathway analysis indicate that
   glutamine-based metabolites play a significant contribution to
   glutamine metabolism and related pathways, correlation between other
   metabolites were assessed using Pearson correlation analysis to
   validate the relationship between glutamine and metabolites in other
   pathways. Interestingly, nearly 20 metabolites showed more than (>0.7)
   correlation with glutamine and metabolites belonging to the key
   impaired pathways identified from pathway analysis using MetaboAnalyst
   3.0 software. The metabolites in glutamine and glutamate metabolism
   include glutathione, glutamate, 2-oxoglutarate which show a 0.9, 0.7,
   and 0.6 correlation in A549 and 0.8, 0.8, and 0.8 correlation in H1299
   ([88]Figure 3B).

2.3. δT Inhibits Glutamine Transporters (LAT-1 and ASCT2) and the mTOR
Pathway in A549 and H1299 Cells

   Metabolomic analysis and subsequent quantification of metabolites using
   Chenomx NMR suite (Edmonton, AB, Canada) revealed the potent effect of
   δT on glutamine metabolism, downstream metabolites of glutamine and
   essential amino acids ([89]Figure 1 and [90]Figure 2, [91]Table 1).
   Current literature provides evidence that glutamine uptake and some
   essential amino acids, including leucine, are associated with the
   activation of the mTOR pathway [[92]37]. Thus, Western blot analysis
   was performed to investigate the effect of δT on the mTOR pathway and
   glutamine transporters. Upon intervention with δT (30 µM), the
   glutamine transporters (LAT-1 and ASCT2) and key mTOR pathway proteins
   (P-mTOR and p-4EBP-1) were found to be inhibited, relative to the
   untreated controls ([93]Figure 4).

Figure 4.

   [94]Figure 4

   [95]Figure 4
   [96]Open in a new tab

   δT inhibits glutamine transporters (LAT-1 and ASCT2) and the mTOR
   pathway in A549 and H1299 cells. (A) The expressions of LAT-1, ASCT2,
   p-mTOR, mTOR, p-4EBP-1, 4EBP1, and β-actin proteins were detected by
   Western blot analysis in A549 and H1299 after treating with 0 µM and 30
   µM concentrations of δT for 72 h. (B) The fate of glutamine uptake in
   A549 and H1299 involving metabolites (purple), associated key proteins
   (pink), and functions (orange). Glutamine in cancer facilitates
   exchanging of EAAs (essential amino acids) into proliferating cells via
   glutamine transporters (LAT1 and ASCT2), which induces mTOR activation
   in A549 and H1299. Activated mTOR then promotes protein translation and
   cell growth via activation of its downstream genes 4EBP1. The black
   arrows indicate pathway direction, while the red downward arrows
   indicate inhibition.

3. Discussion

   In this study, we used multivariate analysis of NMR spectra and NMR
   quantification data to observe differences in the intracellular
   metabolomes. We discovered clear differences in the intracellular
   metabolomes, and subsequently the contributing metabolites, of the
   control and δT treated cells using OPLS-DA and Heat map analysis
   ([97]Figure 1 and [98]Figure 2A). Also, we observed a minor difference
   in the results obtained through multivariate analysis of NMR spectra
   and NMR quantification variation in this analysis which is common in
   metabolomic data sets. This type of variation is well documented in
   several publications in the current literature [[99]6]. Most variations
   arise from the metabolites present in very low concentrations. In
   addition, metabolites whose resonances yield a very high number of
   overlapping peaks also suffer from variations in quantitation [[100]6].
   The two different methods were therefore used in conjunction to verify
   the data.

   Previously, using histone ELISA and ANNEXIN V stain-based flow
   cytometry analysis, we reported that the 10 to 30 μM range of δT was
   not necrotic to A549 and H1299 cells, and that it induced apoptosis in
   a dose-dependent manner [[101]11,[102]12]. Also, using MTS and
   clonogenic assays in the previous studies, we demonstrated that 30 μM
   of δT inhibited cell growth significantly in the A549 and H1299 cells
   lines [[103]12]. Other metabolomics investigations have also reported
   changes in metabolism after inducing apoptosis in different cancer
   types, namely leukemia cell lines [[104]38]. Our data suggests that
   metabolite changes in the control vs. δT treated lung cancer cell
   populations are a result of induction of apoptosis after δT treatment.

   The role of natural dietary components in cancer growth and progression
   has become a very popular subject with minimum effect or no effect on
   normal cells. Several cell culture studies showed that δT was not
   causing apparent impairment towards the noncancerous cell lines,
   although it significantly effects different cancer cell types,
   including lung cancer. For instance, Human Fetal Lung Fibroblast Cells
   treated with 100 µm or higher of δT did not show any toxic effect
   including induction of apoptosis and DNA damage [[105]18]. In another
   study, 10 μM DT3, a lower dose than our treatment, was determined to be
   nontoxic, and enhanced cell viability and proliferative potential in
   the human lung fibroblast cell lines MRC-5 and HFL1, as shown by WST-1
   and clonogenic assays [[106]39]. In addition, Immortal human pancreatic
   duct epithelial cell lines did not show any significant inhibitory
   effect on cell proliferation and cell cycle progression when they were
   incubated with δT [[107]40]. Similarly, normal human melanocytes
   treated with δT (5–20 μg/mL) for 24 h or 48 h did not affect cell
   growth at both time intervals [[108]41]. Preclinical and clinical
   evidence also supports the use of δT to reduce tumor growth with no
   effects on healthy humans or animals, making δT attractive compounds.
   No adverse effects were observed upon administration of 300 mg/kg dose
   of δT, in any tissues or organs of mice [[109]42]. In humans, δT can be
   safely administered at doses up to 1600 mg twice daily [[110]43]. In
   another study with osteopenic women, supplementation for 12 weeks did
   not affect body composition, physical activity, quality of life, or
   intake of macro- and micronutrients [[111]44]. All of the
   aforementioned studies used δT concentrations above 30 μM that we used
   for this study, and it is obvious that δT does not affect healthy cells
   including human fetal lung fibroblast cells. Therefore, a control arm
   of normal lung cells with expressed or unexpressed LAT1 and/or ASCT2
   were not included in our study design.

   Further, LAT1 or ASCT2 transporters with cancer is nowadays
   well-assessed [[112]9]. Overexpression of LAT1 is well described in
   many human cancers and it certainly relates to metabolic changes
   occurring in cancer development and progression [[113]45]. LAT-1 is
   expressed in cancers of most human tissues according to GENT database
   [[114]46], which suggests an important role of LAT-1 expression on
   cancer development. In contrast, it is poorly expressed or, in some
   cases, absent in most of the corresponding noncancer human tissues
   [[115]46]. In the immunohistochemistry analysis of the normal lung,
   LAT1 protein was identified only on granular regions in the cytoplasm
   of chondrocytes of the bronchial cartilage, serous cells of the
   bronchial glands, and alveolar macrophages within the normal lung,
   whereas the expression was zero for nonciliated bronchiolar epithelial
   cells (Clara cells), goblet cells of the bronchus, mucinous cells of
   the bronchial glands, and alveolar type I or type II cells [[116]47].
   In the same study, expression of LAT1 protein appeared in the cytoplasm
   of bronchial surface epithelial cells as a single nodular spot, which
   was considered to represent an intracellularly localized nonfunctional
   protein [[117]47]. ASCT2 transporters also are poorly expressed or, in
   some cases, absent in most of the corresponding noncancer human tissues
   according to GENT database [[118]46]. Hassanein et al. identified ASCT2
   transporters expressed in stage I NSCLC when compared to matched
   controls using shotgun proteomic analysis [[119]48]. In addition, ASCT2
   deficient mice showed regular functions such as normal B-cell
   development, proliferation, and antibody production [[120]49].
   Therefore, control arms of normal lung cells that are expressed or
   unexpressed (LAT1 and ASCT2) was also not included in our study design
   as there was a minimum expression and/or functionality observed for
   LAT1 and ASCT2 in other tissues and noncancerous tissues.

   A significant reduction of glutamine, glutamate, GSH and 2-oxoglutarate
   after treating with 30 µM of δT on NSCLC cell lines was observed
   ([121]Table 1). The key aberrant pathways identified using the pathway
   analysis tool include glutamate and glutamine, alanine, aspartate,
   glutathione metabolism, and the TCA cycle ([122]Figure 3). In addition,
   the metabolites identified from these pathways show a strong
   correlation with glutamine levels ([123]Figure 3B). Further, glutamine
   and its related metabolites were identified in the S-plot of OPLS-DA
   analysis and the Random Forest importance plot as the key players
   causing the separation, reflecting the differences in their metabolomic
   profiles ([124]Figure 1 and [125]Figure 2B). Glutamine deprivation has
   been shown to induce apoptosis in hepatoma, hybridoma, leukemia,
   myeloma, and fibroblast cells [[126]50]. In contrast, increased levels
   of glutamine were detected in lung cancer tissue especially in NSCLC
   when compared to other types of cancer, such as colon or stomach cancer
   [[127]47]. Glutamine dependency has been reported in H1299 and A549
   cells [[128]28]. Our findings strongly suggest the beneficial impact of
   δT on glutamine and related pathways in non-small cell lung cancer
   cells.

   Considering metabolism of glutamine ([129]Figure 5), one of its major
   roles in cancer cell proliferation is to replenish the TCA cycle
   intermediates removed by the process called glutaminolysis, and GSH
   synthesis [[130]30,[131]31]. In the process of glutaminolysis, the
   glutaminase enzyme (GLS1/2) catalyzes the conversion of glutamine to
   glutamic acid and the subsequent conversion of glutamate to
   α-ketoglutarate (2-oxoglutarate), catalyzed by glutamate dehydrogenase
   (GLUD) [[132]32]. Aminotransferase also catalyzes the reaction from
   glutamate and oxaloacetate to aspartate or alanine and α-ketoglutarate.
   In this study, a significant reduction of glutamine, glutamate, and TCA
   cycle intermediates after treating with 30 µM of δT was observed, which
   is an indicator of reduced energy metabolism ([133]Figure 5). In cancer
   cells, the enhanced production of 2-oxoglutarate and glutamate from
   glutamine metabolism can be observed, as it helps to maintain the
   citric acid cycle intermediate for energy production [[134]32]. Glucose
   and glutamine provide substrates for macromolecular synthesis supplying
   both ATP and carbon skeletons in cancer cells [[135]29]. This supports
   uncontrolled cell proliferation in cancer cells and requires a large
   number of macromolecules to create new biomass, including DNA,
   proteins, and lipids [[136]28]. Our data suggests that by decreasing
   the availability of glutamine, δT retards this process, thereby leading
   to inhibition of uncontrolled cell proliferation in A549 and H1299 as
   reported in our previous studies [[137]11,[138]12,[139]35].

Figure 5.

   [140]Figure 5
   [141]Open in a new tab

   Glutamine metabolism and the effect of δT on glutamine metabolism in
   A549 and H1299 cells. Glutamine mainly replenishes the TCA cycle
   intermediates and GSH synthesis in cancer cell proliferation. In the
   process, glutaminase enzymes (GLS1/2) catalyzes the conversion of
   glutamine to glutamic acid and the subsequent conversion of glutamate
   to α-ketoglutarate (α-kG), catalyzed by glutamate dehydrogenase (GLUD)
   and amino transferase. This process supports for uncontrolled cell
   proliferation in cancer cells and requires a large number of
   macromolecules to create new biomass, including DNA, proteins, and
   lipids. The black arrows indicate the pathway’s direction, while the
   red downward arrows indicate the inhibition of metabolites as an effect
   of δT treatment.

   Considering possible causes for the significant decrease in glutamine
   and its downstream metabolites, we hypothesized that it may be due to
   inhibition of glutamine transporters. We thus measured the protein
   levels of glutamine transporters, namely LAT1 and ASCT2, known to play
   a fundamental role in glutamine uptake process in normal cell
   physiology. LAT-1 facilitates glutamine efflux in exchange for the
   influx of leucine and other essential amino acids (EAA) across the cell
   membrane; similarly, ASCT2 mediates uptake of neutral amino acids
   including glutamine [[142]51]. Our observations from western blot
   analysis established that δT treatments inhibit the expression of LAT-1
   and ASCT2 ([143]Figure 4). We also quantified detectable EAA including
   leucine in cell lysates, the concentration of which were decreased
   significantly after treating NSCLC cells with δT by NMR analysis.
   Inhibition of EAA in A549 and H1299 cells upon δT treatment reflects
   function of LAT-1 which facilitate glutamine efflux in exchange for the
   influx of leucine and other essential amino acids (EAA). This supports
   the beneficial effects of δT on LAT1 transporters inside A549 and H1299
   cells. In addition to facilitating the transport of EAAs for protein
   synthesis, LAT1 and ASCT2 stimulate the growth of cancer cells via mTOR
   [[144]27,[145]52,[146]53]. In head and neck squamous cell carcinoma
   cell lines, inhibition of the LAT-1 transporter using an inhibitor
   lowered the levels of phosphorylation of mTOR and its downstream
   signaling molecules [[147]54]. Thus, if the inhibition of glutamine
   transporters and EAA uptake with δT treatment is valid, it is logical
   to expect inhibition or lower activation of mTOR pathway after treating
   with δT in NSCLC. Indeed, we observed lower activation of mTOR along
   with LAT-1 and ASCT2 after treating with δT, using Western blot
   analysis, which illustrates that inhibition of glutamine transporters
   affect the mTOR signaling pathway ([148]Figure 4).

   mTOR functions are mediated by two downstream proteins, the eukaryotic
   initiation factor 4E (eIF4E)-binding protein 1 (4E-BP1) and p70
   ribosomal S6 kinase 1 (p70S6K1, S6K1) ([149]Figure 4) [[150]55]. For
   further confirmation, we tested the expression levels of downstream
   genes of mTOR namely P-4E-BP1. We observed the similar inhibitory
   effect on mTOR downstream proteins 4E-BP1suggesting an inhibitory
   effect of glutamine transporters passing through mTOR to downstream
   pathway ([151]Figure 4). mTOR downstream proteins 4E-BP1 and S6K1
   regulate F-actin reorganization, focal adhesion formation, and tissue
   remodeling through the proteolytic digestion of extracellular matrix
   via upregulation of matrix metalloproteinase 9 (MMP-9) [[152]56].
   Interestingly, in our previous study, we observed that δT reduced cell
   migration, invasion and adhesion in a dose- and time-dependent manner,
   and inhibited MMP-9 expressions in NSCLC cells [[153]13,[154]34], which
   is an additional supporting inhibitory function of δT.

   Further, in the previous study, we demonstrated that δT induces
   apoptosis in a dose-dependent manner in NSCLC from Annexin based flow
   cytometry analysis and histone ELISA [[155]12]. The current literature
   also provides evidence to support the relationship between GSH and
   apoptosis. For instance, GSH depletion in cancer cells induces
   apoptosis in vitro and in vivo [[156]57]. Dalton TP et al. showed
   GSH-depleted knockout mouse of γ-GCS died from massive apoptotic cell
   death [[157]58]. Elevated levels of GSH are also associated with
   apoptotic resistant phenotypes in several models of apoptosis in
   previously reported studies [[158]59,[159]60], and GSH depletion by
   itself has been observed to either induce or stimulate apoptosis
   [[160]59,[161]61]. GSH quantification, after treating with δT in A549
   and H1299 cells, shows a clear decline in intercellular GSH levels in
   both cell lines ([162]Table 1). The results reveal there may also be a
   possible association between GSH levels and induction of apoptosis in
   NSCLC cells after treating with δT.

4. Materials and Methods

4.1. Cell Culture and Treatment with δ-T

   NSCLC cell lines A549 and H1299 were cultured in RPMI medium
   (Mediatech, Manassas, VA, USA) supplemented with 10% fetal bovine serum
   and 1% penicillin and streptomycin in 5% CO[2] at 37 °C. The culture
   medium was renewed every 2 to 3 days. Adherent cells were detached by
   incubation with trypsin-EDTA and centrifuged at 80× g. The treatment
   media was prepared by mixing δT (<0.01% DMSO as a vector) in the RPMI
   medium, whereas the control was treated only with RPMI media. Three δT
   solutions at concentrations of 10 μM, 20 μM, and 30 μM containing
   <0.01% DMSO were chosen as the treatment concentration based on our
   previous studies. δT was a gift from the American River Nutrition for
   this study.

4.2. Intracellular Metabolite Extraction and Determination

   We used a modified method which is explained in Saadat et al., 2018
   [[163]62]. In brief A549 and H1299 lines were seeded at a density of 2
   × 10^6 per 100-mm dish for 24 h, followed by replacement of media
   absent or supplemented with different δT concentrations (10 µM, 20 µM,
   and 30 µM) at 37 °C. Cells were then incubated for another 72 h before
   extracting metabolites. Before extracting intracellular metabolites,
   existing culture media was removed on ice followed by washing twice
   with ice-cold PBS. Two milliliters of ice-cold methanol was added while
   scraping with cell scrapers on ice. The Petri dish was shaken for 5 min
   at 4 °C and ice-cold methanol was transferred into Eppendorf tubes. The
   cell debris was removed by centrifugation and all the extraction
   solvents were readily removed before NMR analysis by a Speed Vac at
   room temperature. Subsequently, the intracellular metabolites powder
   was prepared by evaporating with methanol, and redissolving in 450 μL
   D[2]O containing 0.5 μM 2,2-Dimethyl-2-silapentane-5-sulfonic acid
   (DSS) as a spectral calibration standard and 10 μM imidazole as a pH
   indicator. An additional Petri dish was prepared for each
   treatment/control with the same conditions and cells collected from the
   additional petri dish were used for analyzing total protein. The total
   protein quantifications include control-A549 (1.283 mg), 30 µM-A549
   (1.099 mg), control-H1299 (1.325 mg), and 30 µM-H1299 (1.276 mg). The
   intracellular metabolite powder was redissolved in D[2]O and normalized
   based on the total protein contained in additional petri with
   corresponding treatment before performing NMR. We made sure to maintain
   the final concentration of internal standards at aforementioned levels.

4.3. 1H-NMR Spectroscopy

   High-resolution 1H-NMR spectra of intracellular metabolites were
   obtained on a Varian 600 spectrometer operating at 600 MHz after
   normalizing the samples by total protein concentrations using BCA
   Protein Assays (Thermos Fisher Scientific, Rockford, IL, USA). 1H-NMR
   spectra of intracellular extracts were acquired using a 6-kHz spectral
   width and 64 K data points. The acquisition time was 5.44 s and the
   relaxation delay was 14.56 s with 64 scans.

4.4. 1H-NMR Spectroscopy Processing

   After NMR analysis, Free Induction Decay (FID) files were obtained and
   processed using NMR processing software ACD (Advanced Chemistry
   Development, Inc. Toronto, ON, Canada). NMR spectra of all the samples
   were stacked and processed simultaneously. First, FID files were
   Fourier-transformed to visualize spectra followed by phasing, baseline
   correction and binning with the auto option of the software. After
   completing these steps, the full spectra, as a batch, were divided into
   1000 bins using the intelligent bucketing algorithm in ACD software,
   giving a numerical value for corresponding peaks, and converted into a
   data table. Intelligent bucketing in ACD is an algorithm that was
   designed to make decisions as to where a bucket division should be.
   Intelligent bucketing chooses integral divisions based on local minima
   and therefore avoids the reduction of data resolution, while aligning
   the spectra as a batch.

4.5. Quality Control

   Relative standard deviation (RSD) values were calculated for each
   treatment group separately and Technical variation within metabolomics
   datasets, recorded using one dimensional NMR maintained less than <8%
   (reported as the median spectral RSD)

4.6. Multivariate Data Analysis: OPLS-DA

   The processed, digitized NMR spectral data table from ACD software
   (version 10) was imported into the SIMCA (version 15) software
   (Sartorius Stadium Biotech, Germany for Multivariate data analysis
   (MVDA). The data table was transposed and labeled accordingly. The
   integrals corresponding to the spectral region from 4.5 to 6 ppm were
   excluded as this region contains water peaks and exchangeable protons.
   Spectral regions displaying no peaks, DMSO, and spectral regions of
   methanol to all the samples were also excluded from the dataset. PCA,
   OPLS-DA models were created by generating optimum number of principal
   components needed to fit the data, using the autofit option in the
   software. Each model’s characteristics are described by how well it
   fits the data and its ability to predict new data accurately. Thus the
   value for R2 describes how well the data fits the model while the value
   of Q2 relates to the models ability to predict unknown data correctly.
   These are calculated by the for the purpose of evaluating and
   validating the models generated. The following cutoff criteria are used
   for validating the models that were generated. For NMR metabolomic
   data, it is recommended that the model generated has a Q2 > 0.5, a
   value of R2 higher than Q2 with the difference between them being no
   greater than 0.3. These criteria were adhered to for all the models
   utilized for the investigation. Samples were identified and
   distinguished by their respective labels and colored for visual
   convenience. The data was subjected to Pareto-scaling prior to
   analysis. The Hotelling T^2 test (based on the 95% confidence interval)
   and DMOD-X test (based on the distance from the model plane) was used
   to remove any statistically extreme outliers while maintaining a
   minimum of 4 replicates in each group. Initially, unsupervised
   Principal Component Analysis (PCA) was performed to view the clustering
   effects in the samples ([164]Supplemental Materials). Subsequently,
   OPLS-DA, a supervised pattern recognition method, was performed to
   maximize the identification of variation between groups tested.

4.7. Metabolite Identification and Quantification from Chenomx NMR Suite

   The metabolites were identified using Chenomx NMR suite (Chenomx Inc.,
   Edmonton, AB, Canada). The fid files from the 1D 1H-NMR spectra were
   imported into the Chenomx software. This software has its own
   processing interface where spectra were Fourier-transformed and
   baseline corrected. Phasing was done using DSS reference peak at 0.0
   ppm, and the water peak was also deleted. The processed spectra were
   analyzed in the profiler module of the software. The 600 MHz library
   with the corresponding pH was selected. Identification and
   concentrations of different metabolites were calculated by fitting the
   set of peaks for those compounds in the sample spectrum. If the area
   was crowded with many peaks, then multiple metabolites were adjusted at
   one time to match the reference spectrum closest to the sample
   spectrum. The identified and quantified compounds were then exported
   into an excel sheet.

4.8. Additional Multivariate Data Analysis and Metabolic Pathway
Identification Using MetaboAnalyst 3.0 Software

   MetaboAnalyst 3.0 software, a web-based metabolomics data processing
   tool [[165]63], was used to statistically analyze the metabolites
   identified using Chenomx NMR suite. Quantified data from Chenomx NMR
   suite were scaled using range scaling algorithm. Clustering
   differences, heat maps, and a Random Forest analysis plot were
   generated. Further, the top 25 metabolites correlating with glutamine
   were identified using Pearson correlation analysis and the significant
   features were identified by Random Forest analysis. Additionally,
   quantified data from Chenomx NMR suite was transferred into an excel
   table which allowed us to perform a Student’s t-test and calculate fold
   changes. A p-value of less than 0.05 was considered to be statistically
   significant for univariate analysis.

   Metabolic pathway identification was performed with the pathway
   analysis option of Metaboanalyst 3.0 software. Briefly, the Homo
   Sapiens Pathway Library was selected as a reference, and the pathway
   analysis was performed to generate pathway analysis output on all
   matched pathways, based on the p-values from pathway enrichment
   analysis and pathway impact values from pathway topology analysis.

   Further, metabolites that were changing most significantly between the
   control and 30 µM treatment were traced back to their origin, and the
   pathways were interpreted for metabolism changes using current
   biochemistry.

4.9. Western Blot for Protein Expression Analysis

   One million cells of each of A549 and H1299 were seeded in 100-mm
   dishes and incubated for 24 h; then, the original media was replaced by
   media with/without δT and incubated for another 72 h. After 72 h
   incubation, cells were washed with ice-cold PBS and lysed in the cold
   1X cell lysis buffer (Cell Signaling Technology, Danvers, MA, USA) for
   30 min on ice with 1X protease inhibitor (Cell Signaling Technology,
   Danvers, MA, USA). The cell lysate was kept at −80 °C overnight before
   quantifying.

   Protein concentrations were estimated using Pierce BCA Protein Assay
   kit (Bio-Rad Laboratories, Hercules, CA, USA). Total cell lysates (40
   µg) were mixed with equal amounts of 6x laemmli buffer (Bio-Rad
   Laboratories, Hercules, CA, USA), followed by boiling at 100 °C for 5
   min. Samples were loaded on 10% SDS-polyacrylamide gel electrophoresis,
   and then the gel was electrophoretically transferred to a
   nitrocellulose membrane (Whatman, Clifton, NJ, USA) in transfer buffer
   (25 mM Tris, 190 mM glycine, 20% methanol) using a Bio-Rad Trans-Blot^®
   Turbo™ Transfer System (Hercules, CA, USA). The membranes were
   incubated for 1 h at room temperature with 5% BSA in 1x TBS buffer
   containing 0.1% Tween. After incubation, the membranes were incubated
   overnight at 4 °C with primary antibodies (1:1000). The following
   antibodies ASCT2, LAT-1, p-mTOR, mTOR, p-4EBP-1,4-EBP1, and B-actin
   (Cell Signaling Technology, Danvers, MA, USA) were used in the
   analysis. The membranes were washed three times with TBS-T and
   subsequently incubated with the secondary antibodies (1:5000)
   containing 2% BSA for 2 h at room temperature. The signal intensity was
   then measured by chemiluminescent imaging with ChemiDoc XRS (Bio-Rad
   Laboratories, Hercules, CA, USA).

5. Conclusions

   In this work, the anticancer effects of δT on NSCLC cell lines A549 and
   H1229 were investigated and confirmed by 1H-NMR metabolomics analysis.
   A closer look into the intracellular metabolome of NSCLC cells revealed
   significant and potentially beneficial alterations in glutamine
   concentrations and related metabolism upon treatment with δT. The data
   purports that δT exerts its action by inhibiting glutamine uptake into
   proliferating cells by inhibition of glutamine transporters, thereby
   resulting in inhibition of cell proliferation and induction of
   apoptosis via downregulation of the mTOR pathway ([166]Figure 4B and
   [167]Figure 5). Through this work, NMR-based cellular metabolomics
   helps provide possible opportunities for evaluating the therapeutic
   effect of phytochemicals and systemic changes in cancer metabolism.

Acknowledgments